You are on page 1of 393

2.

01 Introduction to Volume 2
LJ Barbour, Department of Chemistry and Polymer Science, University of Stellenbosch, Stellenbosch, South Africa
Ó 2017 Elsevier Inc. All rights reserved.

The vast array of supramolecular systems, as well as the concepts that control their assembly, are described in much detail in the
other volumes of this major reference work. Whatever we currently know about these systems has only been made possible by
the advent of advanced methods for quantitative analysis and atomic/molecular level imaging techniques. Most studies of supra-
molecular assemblies require the application of a number of complementary analytical methods, many of which are the remit of
specialists who have invested many years in building up their expertise. The objective of Volume 2 is to provide the reader with
a broad overview of the most important experimental and computational techniques that are critical to our understanding of
the molecular world. Each chapter has been prepared by an expert in the field and discusses the technique in the context of its histor-
ical development, recent innovations, and its application to the analysis of supramolecular assemblies.
Chapter 1 describes the application of thermal analysis to crystalline solids from the viewpoints of both physical and structural
chemists, and in particular with regard to inclusion compounds and solid-state phase transformations. Over the past century, X-ray
crystallography has been the go-to method for the structural analysis of crystalline solids. At first crystallography was used to analyze
molecular structure, thereby revealing much about structure and bonding with atomic-level precision. However, the technique is
now more routinely used to elucidate packing modes and the intermolecular interactions that govern packing, concepts that are
central to supramolecular chemistry. While single-crystal diffraction, as described in Chapter 2, is still the most convenient, accurate,
and precise tool for imaging three-dimensional solid-state structure, great strides have been made in the application of powder
diffraction methods (Chapter 3). Once regarded merely as a “fingerprinting” tool for crystalline phase identification, powder diffrac-
tion has matured into an advanced structural method to rival single-crystal diffraction. The two diffraction approaches are still
regarded as complementary, but the developments in powder diffraction are particularly important given that it is often not possible
to obtain samples as single crystals. Solid-state NMR (Chapter 4) is also often used as a complementary method to X-ray diffraction.
However, NMR can provide additional information that is generally inaccessible to X-ray crystallography; examples include solid-
state dynamics and transport, aggregation in amorphous phases, and interactions at interfaces. Methods for preparing new solid
materials have also been revolutionized in recent years. Mechanochemistry has been used through the ages to prepare various useful
substances, but it is only recently that we have begun to understand how and why the method works. Chapter 5 describes the appli-
cation of mechanochemistry to supramolecular systems, and also how we have gained new insight into the process by using inno-
vative milling methods coupled with advanced in situ structural analysis. The study of molecular receptors forms a large component
of supramolecular chemistry and there has been much interest in understanding the interactions between solids and gases, partic-
ularly with regard to the development of new sensors. In this context, Chapter 6 reviews the concepts and experimental methods
involved in gas/solid complexation and inclusion. Of the many properties exhibited by materials, magnetism is arguably one of the
most interesting and important. A review of the theory and measurement techniques is provided in Chapter 7.
The earlier-mentioned techniques are mainly concerned with bulk three-dimensional structure, but there is much scope for
supramolecular organization at interfaces. Chapter 8 discusses strategies for utilizing the concepts of supramolecular chemistry
in order to construct low-dimensional assemblies on surfaces. The concept of using supramolecular receptors as sensors can be
expanded by using mass sensitive techniques based on the piezoelectric effect, and Chapter 9 provides a comprehensive overview
of the theory and practice of quartz crystal microbalance devices. The thermodynamics of binding that occurs as a result of supra-
molecular recognition can be studied using various calorimetric methods. These techniques are particularly useful for supramolec-
ular systems in solution, and Chapter 10 describes isothermal titration calorimetry in terms of its advantages and limitations relative
to other calorimetric approaches. Continuing with the theme of binding phenomena, Chapter 11 outlines the use of supramolec-
ular systems as the stationery phase in chromatography, with particular emphasis on calixarenes. Neutron scattering encompasses
a range of different characterization techniques, the fundamentals of which are described in Chapter 12. Of these, small angle
neutron scattering warrants particular attention with regard to its contributions to our understanding of large supramolecular
assemblies in solution (Chapter 13). Experimental methods are often complemented by computational studies, and computer tech-
nology and theoretical approaches have advanced to the point where they can be applied to the study of large assemblies (Chapter
14). Signal transduction is an important aspect of the utility of supramolecular binding; Chapter 15 discusses electrochemical
sensing and its relevance to understanding supramolecular interactions, and in using these interactions to develop macroscopic
devices and machines. The notions of extraction and transport through membranes constitute an important aspect of the utility
of supramolecular binding, and these are described in Chapter 16. Finally, cooperativity is a key concept in supramolecular chem-
istry and Chapter 17 discusses treating the properties of a system as being influenced by the ensemble of interactions.
All of the chapters in this volume provide comprehensive reviews of the relevant techniques. However, it is inevitable that there
will be significant overlap with other chapters in this series because it is not possible to completely separate discussions of particular
supramolecular systems from the methods used to study them. It is hoped that the reader will find in this volume sufficient infor-
mation for a broad-based perspective of the methods that are commonly applied to the elucidation of supramolecular phenomena.

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12488-6 1


2.02 Thermal Analysis
LR Nassimbeni and NB Báthori, Cape Peninsula University of Technology, Cape Town, South Africa
Ó 2017 Elsevier Ltd. All rights reserved.

2.02.1 Introduction 3
2.02.2 Thermoptometry 3
2.02.3 Thermomechanometry 4
2.02.4 Thermogravimetry and DSC 5
2.02.5 Interpretation of TG and DSC Results 6
2.02.6 Application of Thermal Methods in Supramolecular Chemistry 8
2.02.6.1 Combined TG, DSC, and Spectroscopy 8
2.02.6.2 High Pressure DSC 9
2.02.6.3 Metal Organic Frameworks 9
2.02.6.4 Catalysis 10
2.02.6.5 Polymers 10
2.02.6.6 Complex Matrices 11
2.02.6.7 Pharma 11
2.02.6.8 Fast-Scan DSC 13
2.02.6.9 Purity Measurements of Pharmaceuticals 13
2.02.6.10 Cyclodextrins 14
2.02.6.11 Calixarenes and Resorcinarenes 14
2.02.6.12 Inorganic Hosts 16
2.02.6.13 Guest Exchange 16
2.02.7 Modulated Temperature Methods 18
References 20

2.02.1 Introduction

The study of supramolecular systems in the solid state comprises three main areas of enquiry, which are structure, thermal stability,
and dynamics. The major part of this article will deal with the molecular and crystal structure of various kinds of inclusion
compounds and how this relates to the macro-properties of such compounds. Thus it will explore the correlation between molec-
ular packing and thermal stability, kinetics of enclathration and decompositions, as well as the dynamics of guest exchange in host–
guest systems.
Supramolecular chemistry has been defined as the chemistry of the intermolecular bond 1 and organized polymolecular systems.
J.-M. Lehn described it as a “sort of molecular sociology,” and is the product of the phenomenon of molecular recognition. The sum
of the various interactions that occur between a given molecule and its neighboring molecules, their strengths and directions are an
expression of molecular recognition and have been summarized in various texts.2 There are a number of works that deal with
thermal analysis in general3 and with respect to its application to supramolecular chemistry, the reader is referred to the review
by Mary Ann White.4 Thermal analysis has proved useful in a wide range of applications, dealing with subjects which are totally
unrelated. For example, a recent review deals with the thermal analysis of meat and meat products,5 while another discusses the
analysis of plasters from Byzantine monuments in the Balkan region.6 There are several thermal analysis techniques which measure
a variety of properties. Brown lists 14 such techniques, but the most common are thermoptometry, thermomechanometry, thermal
gravimetry (TG), and differential scanning calorimetry (DSC).

2.02.2 Thermoptometry

Thermoptometry is a series of techniques that measures various optical properties of a sample while subjected to a controlled
temperature program. Thermal microscopy, also known as hot-stage microscopy (HSM), is the most common and is widely
used in following the decomposition of host–guest compounds. The usual procedure is to cover crystallites of the sample with
a drop of silicon oil and raise the temperature slowly, typically 5 C min 1. If the inclusion compound contains a volatile guest,
bubbles emanate from the crystallites and these can be correlated with the corresponding endotherm in the DSC trace. Similarly
the dissolution, recrystallization, and color changes of the sample can be recorded photographically and associated to the endo-
therms/exotherms in the DSC profile. The comparison of the temperatures at which the various changes occur is not always straight
forward. Differences may often occur due to the geometries of the sample holder in the DSC furnace and the hot stage of the

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12492-8 3


4 Thermal Analysis

microscope. The book by Wendlandt 3a contains excellent information on this and the related techniques of thermoluminescence
and thermal photometry.

2.02.3 Thermomechanometry

Thermomechanometry is a group of techniques which measure some mechanical property of a sample as a function of temperature.
In terms of supramolecular chemistry, the most important of these is thermodilatometry, which measures the thermal expansion of
solids and liquids. This has been employed in the measurement of the inclusion of liquid guests by various clays, which has been
extensively studied. The reaction of the Kaolinite group of minerals with polar guests has been summarized 7 and a volumetric
method of following the rate of intercalation of the layered host vanadyl phosphate by a series of aliphatic alcohols (C1 to C8)
has been devised. The kinetics showed a concomitant 10% decrease in the molar volume of the guest and yielded values of the
reaction rates and activation energies. The apparatus is simple and can be used for any enclathration reaction which is carried
out by suspending the solid in an insoluble liquid.8
Thermodilatometry is also employed in crystal structure determination at different temperatures. The technique involves the
measurement of unit cell parameters at various temperatures in order to check for phase transitions. For organic and organometallic
compounds the measurements are typically carried out between 300 and 100 K in 20 K intervals. Care should be taken to allow the
crystal to equilibrate to the new temperature while it is subjected to a stream of liquid nitrogen vapor for ca. 20 mins. The unit cell
parameters (a, b, c, a, b, g) are measured and the unit cell volume is mapped as a function of temperature. Abrupt changes in any of
the cell parameters signal a phase transformation, and crystal structure analysis is then performed to reveal the molecular packing
arrangements for both phases. The linear coefficients of thermal expansion are evaluated and the change in cell volume at the tran-
sition temperature gives an indication of whether the phase transition is first or second order according to Ehrenfest’s classification. 9
An example of this phase transformation is the change in crystal structure of the inclusion compound formed by 4,40 -bis(diphe-
nylhydroxymethyl)diphenyl with 4-picoline, in which the transition occurs between 235 and 240 K and the detailed changes in the
structure are described ( Fig. 1).10
Applying large temperature changes to crystalline materials can yield unusual and sometimes startling results. The metal organic
framework (MOF) {[Cd(HBTC)(BPP)]∙1.5DMF∙2H2O}n which contains mono-protonated 1,3,5-benzetricarboxylate (HBTC2 )
as the rigid part and 1,3-bis-(4-pyridyl)propane (BPP), when subjected to systematic changes in temperature from 100 to 260 K,
exhibits negative, zero, and positive linear thermal expansion, which was explained in terms of a combined stretching and tilting
mechanism. 11

Figure 1 Changes in cell parameters with temperature; the values are normalized to the LT structure as defined by the transformation matrix.
Reprinted from Nassimbeni, L. R.; Su, H.; Weber, E.; Skobridis, K. Cryst. Growth Des. 2004, 4, 85–88. Copyright 2016, with permission from
American Chemical Society.
Thermal Analysis 5

In a similar study, a nitromethane solvate of 18-crown-6 was analyzed in the range of 90–273 K and exhibited very large positive
thermal expansion in two axial directions and negative thermal expansion in the third. The negative coefficient of linear thermal
expansion of  129  10 6 K 1 was claimed by the authors as the highest value yet recorded for an organic inclusion compound. 12

2.02.4 Thermogravimetry and DSC

TG measures mass loss of a solid while it is subjected to a selected temperature program under a controlled atmosphere while DSC
measures the onset temperature and accompanying enthalpy change of a guest loss or phase transformation which occurs on heat-
ing. Fig. 2 shows a typical result for the curves obtained for an inclusion compound with a volatile guest which desorbs in a single
step. The mass loss is often reproducible to  1% and yields an accurate value of the host–guest ratio. This is an important aspect of
inclusion chemistry, where the host–guest stoichiometry may be variable, and the result allows one to assign sensible site occupancy
factors to the guest molecules in structure refinement.
In Fig. 2, endotherm A accompanies the loss of the guest. Its onset temperature is generally reproducible to  1 C, and the area
under the curve is a measure of the enthalpy change for the reaction.
heat
H$Gn ðb; sÞ / Hða; sÞ þ nG[ðgÞ (1)
The onset temperature, Ton, is a measure of the thermal stability of the inclusion compound, and a useful parameter of relative
stability when comparing host–guest systems is (Ton  Tb) or Ton/Tb, where Tb is the normal boiling point of the guest. 13 The second
endotherm B is associated with the melting point of the host compound. In the case where the entrapped guest is not volatile (e.g.,
dimethylsulfoxide, b.p. ¼ 189 C; nitrobenzene, b.p. ¼ 211 C), the TG and DSC curves are generally broader, as shown in Fig. 3, and
the onset temperature Ton is not well defined. In such cases, the peak temperature Tpeak is generally reported. The endotherm is due
to the release of the guest which dissolved the host, so that the area under the curve represents the enthalpy change of the guest loss
as well as its dissolution.
In order for the results of TG and DSC experiments to be reproducible, it is imperative that the instruments be calibrated. This has
to be carried out for the temperature, enthalpy, and specific heat. 3e,14
Differential thermal analysis (DTA) is the simplest technique in thermal analysis and measures the temperature difference
between the sample and the reference material while both are subjected to the same heating program. The advantage of DTA instru-
ments is that they can operate at high temperatures (z1200 C), while DSC instruments are more restricted to a maximum temper-
ature of z700 C, which is usually sufficient for organic compounds. The advantage of DSC is that the proportionality constant
relating the area under the curve to the enthalpy change is fixed, whereas with DTA it is temperature dependent. DTA instruments
therefore require several calibrations which cover the temperature range of the experiments being undertaken. When calibrating
a TG or DSC/DTA instrument, the following factors must be taken into account and kept constant in the subsequent experiments:
(a) The heating rate (often 10 C min 1)
(b) The kind and flow rate of the purging gas (e.g., dry N2 or dry He flowing at 40 mL min 1)
(c) The type of sample holder (aluminum or ceramic which may be vented or not)
(d) The particle size distribution of the sample.
The last condition is the most difficult to achieve. In the case of samples which are stable under ambient conditions, sieving is
strongly recommended and the sample selected in the narrow range of adjacent sieves. This, however, is not always possible in the
case of inclusion compounds, particularly if they hold volatile guests.

Figure 2 Schematic of TG and DSC results.


6 Thermal Analysis

Figure 3 TG and DSC curves for analyte with broader melting endotherm.

2.02.5 Interpretation of TG and DSC Results

Figs. 2 and 3 display decomposition of a sample in a single step. When decomposition occurs in multiple steps, this is particularly
useful, since it reveals the mechanism of decomposition of the compound under investigation. An example is shown in Fig. 4, in
which the Werner clathrate [Ni(NSC)2(isoquinoline)4],p-xylene decomposes in three distinct steps: step 1 represents the desorption
of the p-xylene guest, while steps 2 and 3 are associated with the loss of pairs of the isoquinoline bases.15
Further important information on the thermal decomposition may be obtained by employing the method of Flynn and Wall, 16
in which the mass-loss curves are recorded at fixed heating rates. The extent of reaction for each curve is defined by
a ¼ ðm0  mt Þ=ðm0  mN Þ (2)
in which m0 is initial mass, mt is mass at time t, mN is final mass. Linear plots of log b/b0 versus 1/T, where b is the heating rate, b0 is
the “standard” heating rate, and T is the temperature in K, can then be obtained for chosen values of the extent of reaction, yielding
activation energies for each decomposition step. The results for the decomposition of the Werner clathrate [Ni(NCS)2(isoquino-
line)4],p-xylene are shown in Fig. 5. The activation energies obtained for steps 1, 2, and 3 were 185, 97, and 112 kJ mol 1,
respectively.

Figure 4 Thermal decomposition of the Werner clathrate [Ni(NSC)2(isoquinoline)4],p-xylene.


Thermal Analysis 7

Figure 5 Activation energies for (A) step 1, (B) step 2, and (C) step 3 for the Werner clathrate [Ni(NCS)2(isoquinoline)4],p-xylene. Modified from
Wicht, M. M.; Báthori, N. B.; Nassimbeni, L. R. Dalton Trans. 2015, 44, 6863–6870. Copyright 2016, with permission from Royal Society of
Chemistry.

The slope of the log b/b0 versus 1/T curves yields the activation energy for a reaction at a selected value of the extent of reaction a:
Ea
slope ¼ 0:457 (3)
R
where Ea is the activation energy and R is the gas constant. Ideally the slopes of such plots, obtained at various values of a, should be
parallel, affording a single activation energy for the complete process ( Fig. 6A). Often, however, one obtains a series of nonparallel
lines (Fig. 6B), producing a range of activation energies which may also be indicative of a changing mechanism for the reaction and
a change in the rate law governing the decomposition.
This points to the disadvantage of the nonisothermal method of kinetic measurements, in that, although it is rapid and can be
typically completed in a few hours, it does not allow the derivation of the rate law pertaining to the decomposition reaction. In order
to derive the rate law, the decomposition reaction should be followed isothermally as a series of selected temperatures as shown in
Fig. 7.
The mass-loss versus time curves are recorded at fixed temperatures and converted to a-time curves. The data are then fitted to
one of the kinetic laws, which have been derived on models based on (a) nucleation, (b) contracting geometry, (c) autocatalysis,
(d) diffusion, or (e) order of reaction. 17
In practice, the following are the most common:
First order F1  lnð1  aÞ ¼ kt (4)

1
=
Contracting area R2 1  ð1  aÞ 2
¼ kt (5)

1 =
Contracting volume R3 1  ð1  aÞ 3
¼ kt (6)
In order to obtain consistent results, it is strongly recommended that the particle size distribution of the compound be narrow,
but as mentioned earlier, this may not be possible in the case of inclusion compounds containing volatile guests, as sieving will lead
to guest loss. In such cases, it is advisable to grow the host–guest compound under constant stirring in order to have a narrow
particle size distribution and to store it under mother liquor. The kinetic runs should be carried out on compound grown in the
same batch. A good model should hold over a wide range of a, the extent of reaction, typically from a ¼ 0.05–0.95, and have a corre-
lation coefficient of fit between experimental and calculated data which is > 0.99. In addition it should hold over the complete range

Figure 6 Semilogarithmic plots of the heating rate versus 1/T indicating constant mechanism (A) and changing mechanism (B) at different stages
of the reaction. Since only pure numbers can be converted into the corresponding logarithm, it follows that in any expression which contains the
logarithm of a given quantity, that quantity must be dimensionless. Hence in the semilogarithmic graph of heating rate versus reciprocal temperature,
the ordinate should be labeled logb/b0, where b0 is the “standard” heating rate, analogs to the standard state in thermodynamics. If the heating rate
b is measured in  C min 1, then b0 ¼ 1 C min 1.
8 Thermal Analysis

Figure 7 Isothermal decomposition curves carried out at fixed selected temperatures.

Figure 8 Semilogarithmic Arrhenius plot yielding values of the activation energy Ea and the frequency factor A.

of temperature T1 to T5 ( Fig. 7). In the event of more than one model fitting the data, one should obtain confirmation by comple-
mentary techniques such as microscopy or spectroscopy.
The rate constant k is derived from the results obtained at each temperature and can then be applied to the Arrhenius equation:
 
Ea
k ¼ Aexp (7)
RT
The semilogarithmic plot shown in Fig. 8 affords the activation energy Ea, the energy barrier to the reaction, and A, the frequency
factor, a measure of the recurrence of the situation that may lead to product formation. The application of this equation to hetero-
geneous reactions is controversial, but Galway and Brown have discussed this topic and have given theoretical justification for its
use.18
The thermal decomposition of sodium bicarbonate NaHCO3 was employed to study the kinetics under modulated temperature
conditions. 19 This allows one to determine the apparent values of the activation energy, Ea, at different extents a of the reaction, thus
revealing possible changes of reaction mechanism. The differential rate equation
 
da Ea
¼ Aexp  f ðaÞ (8)
dt RT
1
, derived from the contracting area model yielded activation energies varying from 98 to 104 kJ mol 1
=
where f ðaÞ ¼ 2ð1  aÞ 2

(Fig. 9).

2.02.6 Application of Thermal Methods in Supramolecular Chemistry


2.02.6.1 Combined TG, DSC, and Spectroscopy
An investigation of the acid–base complexes of picric acid with N-heterocyclic bases was carried out by IR, NMR, and thermoana-
lytical methods. Their crystal structures were elucidated and provided details of the nonbonded interactions. The optimized
Thermal Analysis 9

Figure 9 Mass-loss traces for the thermal decomposition of NaHCO3 (100  170 mesh, m0 ¼ 5.0 mg) in flowing N2 (80 cm3 min 1) under a modu-
lated temperature condition (period ¼ 5 min, amplitude ¼ 10 K) based on linear nonisothermal measurement at b ¼ 1 K min 1. Reprinted from
Koga, N.; Groshi, Y.; Yoshikawa, M.; Tatsuoka, T. J. Chem. Educ. 2014, 91, 239–245. Copyright 2016, with permission from American Chemical
Society.

structures were calculated in terms of density functional theory. TG and DSC studies were extended to evaluate the ignition delay
values over a fixed temperature range. 20
1-Methylcyclopropane is an ethylene inhibitor and thus shows the physiological responses of aging, color change, and ripening
of horticultural products. It forms an inclusion compound with the host a-cyclodextrin which decomposes in four distinct phases,
releasing the 1-methylcyclopropane between 90 C and 230 C. The kinetic parameters of this reaction have been evaluated and the
effect of differing host:guest ratios has been studied. 21
Thermal analysis can be combined with other techniques to reveal the mechanism involving sorption and phase transformations
in inclusion compounds. The techniques are particularly useful when they are related to structure and the mechanism of structural
changes which may occur when the compound under investigation is subjected to changes in temperature, pressure, or doses of
radiation.
The simultaneous analysis of a sample with DSC and Fourier transform infrared spectroscopy (FTIR) results in combining the
exothermic and endothermic thermal events of the sample with concurrent changes in the chemical and physical compositions. 22
The main advantage of this coupled technique is that the analysis is conducted on the same sample in real time. The cocrystal forma-
tion between theophylline and citric acid was followed by DSC-FTIR microspectroscopy and the formulation was examined under
different storage conditions.23 It was concluded that there were no changes in the thermal and the structural features of the physical
mixture of TP-CA when stored under 25 C and a relative humidity of 75% for a period of 90 days. When the same physical mixture
was stored under 55 C and a relative humidity of 75% for 1 day, the formation of cocrystalline material was observed. Comparing
these results to previous findings,24 it may be concluded that the cocrystal formation was induced by the higher temperature and
humidity.

2.02.6.2 High Pressure DSC


One class of compounds that is currently receiving attention is energetic materials. In a recent publication, high pressure differential
scanning calorimetry (HP-DSC) was carried out on trinitrotoluene (TNT), cyclotrimethylenetrinitramine, and ammonium dinitra-
mide, and the effect of pressure in the range of 0.1–14 MPa on vaporization, melting, and decomposition kinetics was studied. This
selection of energetic materials has either negative or positive oxygen balance, thus both high pressure inert and oxidative atmo-
spheres were applied. 25 The elevated pressure may have several effects on the thermal behavior of the selected material, such as
suppression of evaporation, influence on the melting temperature, or changes the kinetics of the thermal process. It was found
that the increase of pressure up to 5 MPa suppresses the evaporation process of TNT and the decomposition is more pronounced
at 319 C (Fig. 10) and increasing the melting temperature (Fig. 11).

2.02.6.3 Metal Organic Frameworks


MOFs are also a group of compounds that currently are at the center of interest of the scientific community. This is because they
constitute new kinds of porous materials that are useful for selective sorption and gas storage.
Takamizawa et al. 26 studied the absorption of methane by [Cu(II)2(bza)4(pyz)]n by single crystal diffraction in the temperature
range 90–298 K and in the pressure range 3.2–42 MPa. They correlated the structural parameters of host and guest atoms with the
changes in temperature and pressure.
10 Thermal Analysis

Figure 10 Thermal behavior of TNT at heating rate 10 K min 1 0.1 MPa (black curves) and 5 MPa (gray lines) pressure. Reprinted from
Muravyev, N. V.; Monogarov, K. A.; Bragin, A. A.; Fomenkov, I. V.; Pivkina, A. N. Thermochim. Acta 2016, 631, 1–7. Copyright 2016, with permission
from Elsevier.

Figure 11 Pressure influence on melting temperature for TNT. Reprinted from Muravyev, N. V.; Monogarov, K. A.; Bragin, A. A.; Fomenkov, I. V.;
Pivkina, A. N. Thermochim. Acta 2016, 631, 1–7. Copyright 2016, with permission from Elsevier.

A series of MOFs has been analyzed by TG and DSC, their thermal stability and molar heat capacities were measured, and the
results were correlated to the metal coordination in their secondary building units. 27 The kinetics of decomposition of the MOF
[Zn2(dbc)2(dabco)],L (dbc2  ¼ terephthalate, dabco ¼ 1,4-diazabicyclo[2.2.2]octane, L ¼ cyclohexane or benzene) has been dis-
cussed. The compound containing cyclohexane is the most stable, despite the fact that the kinetic diameters of these guests are
very similar. This is attributed to the flexibility of the cyclohexane that can change from chair to boat conformation.28 The related
MOFs [Zn4(DMF)(ur)2(ucd)4],5DMF , H2O and [Zn4(DMF)(ur)2(ucd)4], 4(ferrocene) (ndc2  ¼ 2,6-naphthalenedicarboxylate,
ur ¼ hexamethylenetetramine, DMF ¼ N,N0 -dimethylformamide) were studied by TG; the kinetic laws governing the decomposition
and the Arrhenius parameters established.

2.02.6.4 Catalysis
Thermal methods are important in the characterization of catalysts. The catalyst 4-dimethylaminopyridine is employed as a hyper-
nucleophilic acylation catalyst and its thermodynamic properties were analyzed by DSC, TG, adiabatic, and combustion calorim-
etry. Heat capacity measurements from 80 to 401 K allowed the accurate determination of the molar enthalpy and entropy of
fusion, while calorimetry yielded the values of DU and DH of combustion. 29
A special device, employing simultaneous TG/DSC, has been constructed, which is coupled to a mass spectrometer, and
measures the activity of a heterogeneous catalyst. This apparatus is a high-throughput screening instrument capable of screening
about 70 catalysts per day. 30

2.02.6.5 Polymers
The use of thermal analysis for the characterization of polymers has a long history, and DSC is used to measure glass transition
temperatures, the kinetics of crystallization, and the degree of crystallinity. 3a,b,c,e Recent work has employed thermal methods
for several interesting problems. Real time analysis of the self-assembly process of nanofibers of poly-3-hexylthiophene was carried
Thermal Analysis 11

out by DSC and TG. The study was augmented by UV-visible spectrometry, small angle X-ray scattering, and electron microscopy.
The morphologies of the products of the self-assembly carried out at different temperatures revealed the crystallization
parameters.31 An ion attachment mass spectrometer was combined with an infrared image surface unit in order to analyze poly(-
tetrafluoroethylene) by evolved gas analysis (EGA).32

2.02.6.6 Complex Matrices


Thermal analysis has been employed in the analysis of soil organic matter, a highly complex matrix. 33 An infrared gas analyser was
coupled to a TG/DSC apparatus to detect CO2 and H2O by EGA. Using pure cellulose and CaCO3, the CO2 from the evolved gas and
the DSC profiles were similar and analysis of diverse soil samples showed this to be a valuable method of soil characterization
(Fig. 12).

2.02.6.7 Pharma
Thermal analysis has a long tradition as an important characterization tool in pharmaceutical chemistry. A recent educational publi-
cation 34 exposes students to examination of an active pharmaceutical ingredient (API), acetaminophen, and its excipient a-lactose
hydrate. DSC, TG, and thermal microscopy are employed to determine transition temperatures, enthalpy of transition, and the stoi-
chiometry of the water of hydration.
Polymorphism, when substances exist in different crystalline forms, is important because of the difference in physical properties,
such as melting point or solubility. The energy differences between polymorphs are of the order of 1 kJ mol 1, and this reflects in
very small differences in their melting points. 35 Thus the application of DSC to differentiate between the solid crystalline forms via
obtaining their accurate melting points is mandatory. The two polymorphic forms of bezafibrate, 2-[4-[2-(4-chlorobenzamide)
ethyl]-phenoxy]-2-methylpropanoic acid, a lipid lowering drug have been reinvestigated and fully characterized by optical micros-
copy, FTIR, DSC, TG, variable temperature X-ray powder diffraction, and single crystal X-ray diffraction.36 There are several differ-
ences in the molecular conformations and in the crystal structures of the two polymorphs which are discussed in detail. The DSC
curve of the a form shows a melting endotherm at Ton ¼ 183.7 C (Fig. 13A). The b form shows a small endotherm at 160.7 C with
a heat of fusion of 4.3 kJ mol 1 (Fig. 13B) which represents the solid–solid phase change of b / a, followed by melting of form
a (Ton ¼ 183.0 C). Calculating the Gibbs free energy difference between the two polymorphs over temperature ranges by using
direct heat capacity measurements reflects their thermodynamic relationships. An energy versus temperature diagram for the forms
a and b was constructed and an enantiotropic relationship was described (Fig. 14).
Cocrystals, when the crystalline material is formed from more than one chemical entity, also exhibit polymorphism. The ther-
modynamic stability of the dimorphic cocrystal system of caffeine–glutaric acid (CA–GA, Forms I and II) has been investigated by
DSC, TG, HSM, single crystal X-ray structure elucidation, and variable temperature powder X-ray diffraction. 37 The relative stabil-
ities of the two forms revealed enantiotropic relationships with a transition temperature of 79 C.
Thermal analysis has also been applied in the characterization of hydrated multicomponent gels. 38 Stearoyl macrogol-32 glyc-
eride gels were used to investigate the effect of water content on the architecture of the gel using DSC, NMR relaxometry, and

Figure 12 Multiple results from various techniques in the analysis of selected soil samples. Reprinted from Fernandez, J. M.; Peltre, C.; Craine, J.
M.; Plante, A. F. Environ. Sci. Technol. 2012, 46, 8921–8927. Copyright 2016, with permission from American Chemical Society.
12 Thermal Analysis

Figure 13 DSC curves for form a (A) and b (B). Modified from Lemmerer, A.; Bathori, N. B.; Esterhuysen, C.; Bourne, S. A.; Caira, M. R. Cryst.
Growth Des. 2009, 9, 2646–2655. Copyright 2016, with permission from American Chemical Society.

Figure 14 Suggested schematic energy versus temperature diagram for form a and b of bezafibrate. Tp,b / a indicates the phase transition from
form b to a at constant pressure, mpb is the melting point of form b and mpa indicates the melting point of form a observed at 185 C. Reprinted
from Lemmerer, A.; Bathori, N. B.; Esterhuysen, C.; Bourne, S. A.; Caira, M. R. Cryst. Growth Des. 2009, 9, 2646–2655. Copyright 2016, with permis-
sionfrom American Chemical Society.

scanning electron microscopy (SEM). The diffusion behavior of water was correlated with the network of the freeze-dried gels
observed by SEM.
An interesting application of TG is the release of a corticosteroid loaded into SBA-15 silica. 39 Prednicarbate is used in the treat-
ment of atopic dermatitis, and its inclusion in nanostructured mesoporous silica was confirmed by its thermal profile obtained by
TG and DSC, as well as FTIR and X-ray diffraction analyses. The TG result after loading showed a mass loss of 5.48% between 105 C
and 850 C, indicating the presence of the adsorbed drug on the silica and demonstrated a decrease in the rate of the decomposition
reaction.
A typical use of drug carriers is to improve the solubility and dissolution of the API. Naringenin is a flavonoid which shows ther-
apeutic effects, is an anti-inflammatory, anticarcinogenic, and antitumor agent. However, it displays poor solubility in water, poor
dissolution, a low half-life, and fast clearance from the body. It was complexed with b-cyclodextrin and with phospholipids and the
results monitored by FTIR, the surface morphology by SEM, and the phase transition by DSC. 40
An aspect of drug formulation which is not fully studied is the interaction of the API or API complex with the various pharma-
ceutical excipients used in the final solid dosage formation. Fosinopril, an angiotensin-converting enzyme inhibitor, is poorly
soluble in water. It has been complexed with hydroxypropyl-b-cyclodextrin to improve bioavailability, and the complex mixed
with magnesium stearate, talc, and starch. 41 The results were monitored by TG, DSC, and FTIR and demonstrated the applicability
of thermal analysis for the fast screening of API–excipient interactions.
An unusual application of thermal analysis in the pharmaceutical field is the use of DSC measurements to determine the
phase diagram pertaining to two APIs. Fluocinolone acetonide is a corticosteroid used for the treatment of skin disorders and
inflammatory eye, ear, and nose disease. Acyclovir, a synthetic purine nucleoside, exhibits inhibitory activity against herpes
simplex virus. These APIs are used in combinations, 42 and the phase diagram was constructed by DSC measurements of
known mixtures. The position of the eutectic point was established via a Tamman plot, derived from the length of the
eutectic halts as a function of composition. The eutectic mixture of the APIs was then complexed with b-cyclodextrin in
a 1:1 ratio.43
Thermal Analysis 13

Figure 15 DSC profiles for carbamazepine Form III, the least stable form, at various heating rates. Modified from McGregor, C.; Saunders, M. H.;
Buckton, G.; Saklatvala, R. D. Thermochim. Acta 2004, 417, 231–237. Copyright 2016, with permission from Elsevier.

2.02.6.8 Fast-Scan DSC


Fast-scan DSC (or high-speed DSC, HyperDSCÔ) is a modified form of the conventional DSC method when accelerated heating
rates, typically  100 C min 1, are applied. 44 This type of analytical measurement obviously produces shorter instrumental analyt-
ical time. This may be advantageous under certain conditions but more importantly, the increase in scanning rates will visually
increase the size of the peak of interest and this essentially allows the use of smaller sample weights. The best use of fast-scan
DSC is when kinetically controlled processes (e.g., dehydration, decomposition, or recrystallization) are overlapping with events
of interest, for example, melting transition. One typical problem is the characterization of the least stable polymorph, which usually
has the lower melting point, when it undergoes recrystallization to form the more stable form.
The thermal behavior of Forms I and III of carbamazepine was studied using fast-scan DSC and was demonstrated that heating
rates of 250 C min1 altered the kinetics of the melting transition of Form III such that its recrystallization to the more stable Form I
was inhibited ( Fig. 15).45 This allowed direct measurement of the enthalpy of the melting endotherm for Form III from a single
transition.

2.02.6.9 Purity Measurements of Pharmaceuticals


Usually the purity of a compound can be verified by high performance liquid chromatography or gas chromatography. If the
compound has a sharp melting point which is clearly detectable (i.e., no decomposition or other overlapping thermal event occurs
in the range of the melting point), DSC as an alternative method can be applied. When an impurity is soluble in the melt but not in
the solid (i.e., forms an eutectic system), a depression in the melting point of the analyte will occur and the purity of the sample can
be estimated with the modified Van’t Hoff equation: 46
RT02
Tm ¼ T0  x2 (9)
DH0
where Tm is the sample temperature at equilibrium, T0 is the melting point of the pure substance, R is the gas constant, 6H0 is the
heat of fusion of pure substance, and x2 is the mole fraction of impurity in the liquid phase. When an impurity is present in
a substance, the shape of the melting endotherm is also altered: the higher percentage impurity will produce broader melting curves
(Fig. 16). The necessary experimental procedure is summarized by Laye47 and the calculation of purity based on the geometry of the
curve is discussed by Brown.3e
The sensitivity of the method was tested when the purity of 16 pharmaceutical reference standards were measured by DSC and
compared to purity values obtained from HPLC or other accepted analytical techniques. 48 The purity difference was calculated
between the DSC purity and certified reference purity and found to be less than 1.0% for 11 reference standards (betamethasone
dipropionate, propylparaben, bisacodyl, salicylic acid, caffeine, ethinyl estradiol, tolnaftate, loratadine, acetaminophen, labetalol
hydrochloride, and perphenazine) and was 1.0–2.0% for another four samples (methylparaben, azatadine maleate, aspirin, and
N-methyl desloratadine). Significant purity difference was observed only in one sample (4.4%) but this compound, a perphenazine
intermediate, has a much lower certified purity (94.8%) than usual. These findings were expected; hence the DSC method accuracy
decreases with decreasing sample purity. It was concluded that the DSC method is suitable for the analysis of high-purity analytes
(> 99 mol%) when the obtained purity is within 1% of the purity obtained from other traditional analytical techniques.
14 Thermal Analysis

Figure 16 Melting endotherms for substance with different levels of impurities (a < b < c mol% purity).

In summary, using DSC to analyze impurities has many advantages, such as that of no need of a corresponding reference stan-
dard, the required sample size is smaller, and the overall analysis time is shorter when compared to chromatographic methods.
However, the method is applicable only to relatively pure substances and cannot reveal the nature of the impurity.

2.02.6.10 Cyclodextrins
There is a continued interest in the inclusion chemistry of cyclodextrins and their derivatives. They are important host compounds
with a wide variety of uses in the food, cosmetic, and pharmaceutical industries. Their significance is highlighted by the fact that an
entire volume of Comprehensive Supramolecular Chemistry is dedicated to them. 49 Their molecular structures and properties have been
summarized50 and their applications reviewed.51 Szente49 has given an account of various analytical techniques for the analysis of
cyclodextrins, their derivatives, and inclusion compounds. The main thermal techniques include calorimetry, TG, DSC, and EGA.
The native cyclodextrins, a, b, and g, contain various amounts of water of crystallization which can be detected by TG. The water
molecules are generally released at < 100 C after which the cyclodextrin molecules tend to decompose, making the quantitative
detection of other guest molecules difficult. The stoichiometry of cyclodextrin inclusion compounds is dependent on their mode
of preparation. A salient example is the interaction of cypermethrin with b-cyclodextrin, which was investigated by different
methods of preparation: coprecipitation, suspension, kneading, and melting in solution.52 The results were analyzed by TG,
DSC, and X-ray diffraction. A calibration curve was obtained showing a linear relationship between the enthalpy of melting of cyper-
methrin and its percentage content in the complex. This could be used to measure the relative efficiencies of the different methods of
complex preparation.
The permethylated cyclodextrins, however, are more stable and their thermal analysis can often yield clear results. An example is
with the permethylated TRIMEB as heptakis(2,3,6-tri-O-methyl)-beta-cyclodextrin which formed inclusion complexes with insec-
ticides and herbicides. 53 The included guests, (a) fenitrothion, (b) fenthion, and (c) acetochlor, each form inclusion compounds
with a host:guest ratio of 1:1. Their structures were elucidated by single crystal X-ray diffraction and the complex with fenitrothion is
shown in Fig. 17.
All three complexes display two-step mass losses as shown in Fig. 18. The first step is due to the loss of the guest while the second
step is associated with host decomposition. The measured and calculated mass losses of the guests are in close agreement, with
differences of < 1%. The kinetics of thermal decomposition was followed by isothermal method for the TRIMEB–fenitrothion
compound, and the extent of reaction a versus time curves are shown in Fig. 19A. The curves fitted both the first order reaction
model, F1, and the three-dimensional diffusion model, D3, and the corresponding semilogarithmic plots for the Arrhenius equa-
tion are shown in Fig. 19B, yielding activation energies of 126  4 and 114  1 kJ mol 1 for the F1 and D3 models, respectively.

2.02.6.11 Calixarenes and Resorcinarenes


Calixarenes and resorcinarenes are important host compounds with a variety of uses. They still generate a great deal of interest in the
current literature, although most of it is directed at synthesis, structure, and selectivity of both neutral and charged guests. A recent
work describes the host–guest complexes of C-methylcalix[4]resorcinarene and C-undecylcalix[4]resorcinarene with a variety of
solvents. The inclusion compounds were characterized by TG and DSC as well as NMR and X-ray diffraction. The effect of water
on the solubility of the resorcinarenes was investigated in detail and reconciled with the structural results. 54
A series of ammonium-p-tert-butylcalix[6]arene salts were characterized by TG, DSC, and X-ray diffraction. Upon heating, the
solid phase underwent a reverse proton transfer reaction, allowing the ammonium cation to dissociate from the arene host as
an amine. 55 The decomposition temperature depended upon the chain length of the amine guest molecules and on their basicity.
Thermal Analysis 15

Figure 17 TRIMEB–fenitrothion complex with their host residue numbering. Reprinted from Cruickshank, D. L.; Roigier, N. M.; Maurel, V. J.;
de Rossi, R. H.; Bujan, E. I.; Bourne, S. A.; Caira, M. R. J. Incl. Phenom. Macrocycl. Chem. 2013, 75, 47–56. Copyright 2016, with permission from
Springer.

Figure 18 A schematic TG trace for the TRIMEB inclusion complexes. Reprinted from Cruickshank, D. L.; Roigier, N. M.; Maurel, V. J.; de Rossi, R. H.;
Bujan, E. I.; Bourne, S. A.; Caira, M. R. J. Incl. Phenom. Macrocycl. Chem. 2013, 75, 47–56. Copyright 2016, with permission from Springer.

Figure 19 Isothermal plots of a versus time (A) and Arrhenius plot obtained from the isothermal kinetic data (B) for TRIMEB–fenitrothion complex.
Reprinted from Cruickshank, D. L.; Roigier, N. M.; Maurel, V. J.; de Rossi, R. H.; Bujan, E. I.; Bourne, S. A.; Caira, M. R. J. Incl. Phenom. Macrocycl.
Chem. 2013, 75, 47–56. Copyright 2016, with permission from Springer.
16 Thermal Analysis

2.02.6.12 Inorganic Hosts


Inclusion compound formation by inorganic hosts continues to attract attention by specialist groups, and the resultant compounds
have been known and studied for many years. Several texts have reviewed this subject. 56,57 The reactivity of various organic guests in
layered aluminosilicates has been studied intensively, as have the intercalates of zirconium phosphates and phosphonates.
Aromatic amines have been intercalated into crystalline a-titanium hydrogen phosphate and characterized by IR spectroscopy,
X-ray diffraction, and TG.58 The host suffers an expansion of the interlayer distance, and by using calorimetric titration, the
exothermic enthalpy of intercalation was obtained.

2.02.6.13 Guest Exchange


Guest exchange in host–guest systems is a form of dynamic molecular recognition which is of fundamental importance in such
topics as the selective absorption of gases, heterogeneous catalysis and controlling the properties of crystalline materials. There
are two possible mechanisms which govern guest exchange in inclusion compounds:
(a) The host–guest system retains its structure throughout the exchange:
H$nG1 ðs; bÞ þ mG2 ðl=gÞ/H$mG2 ðs; bÞ þ nG1 ðl=gÞ (10)
The host behaves like a zeolite or MOF and the incoming guest, G2, replaces G1, while the host structure, the b-phase, is essentially
retained, with minor distortions. This is most readily achieved if there are strong similarities between the guests G1 and G2 with
respect to molecular size and polarity. 59
(b) The host–guest compound desorbs guest G1 to yield the apohost in its a-form, nonporous phase, which in turn forms a new
compound with G2, and is, in effect, a recrystallization.
H$nG1 ðs; bÞ/Hðs; aÞ þ nG1 ðl=gÞ (11)

Hðs; aÞ þ mG2 /H$mG2 ðs; bÞ (12)


These mechanisms are shown schematically in Fig. 20.
There are several analytical techniques which are employed to monitor guest exchange, including NMR, GC, HPLC, elemental
analysis, and various forms of thermal analysis. The apparatus for carrying out guest exchanges, pictured in Fig. 21, is simple and
readily available in most laboratories.60

Figure 20 Schematic representation of guest exchange: (A) zeolitic exchange and (B) recrystallization. Reprinted from: Ambombo Noa, F. M.;
Bourne, S. A.; Su, H.; Nassimbeni, L. R. Cryst. Growth Des. 2016, 16, 1636–1642. Copyright 2016, with permission from American Chemical Society.
Thermal Analysis 17

Figure 21 Schematic apparatus for guest exchange reaction. Reprinted from Nassimbeni, L. R.; Su, H. CrystEngComm 2013, 15, 7396–7401.
Copyright 2016, with permission from Royal Society of Chemistry.

The inclusion compound H ,nG1(s), in the form of single crystals or powder, is placed in a closed container which may be
evacuated. The vapor from G2 then displaces G1 and the reaction can be monitored by a suitable analytical technique. In
order to obtain kinetic reproducibility the particle size distribution of the H ,nG1 powder should be narrow and the temper-
ature of the experiment should remain constant, as this controls the vapor pressure of G2. Sieving of the starting compound
is recommended but it is not always possible if H ,nG1 is unstable and likely to decompose. The extent of reaction may be
followed by TG as shown in Fig. 22. Small quantities are removed from the host–guest sample at selected times and the
thermogram recorded.
This method works when G1 and G2 have sufficiently different boiling points to yield the steps of mass loss which can be distin-
guished (e.g., acetone, b.p. 56 C and dimethylsulfoxide, b.p. 189 C).
DSC can also be employed using the same procedure. Fig. 23 shows endotherms associated with the guest-release reactions,
together with the endotherm due to the host melt.
The areas under the endothermic peaks are proportional to the enthalpy of the reaction, but in practice quantitative results
are more difficult to obtain than for TG, being strongly dependent on particle size distribution. This method was employed in
the guest exchange reaction of the inclusion compounds formed by the host 9-(3-chlorophenyl)-9H-xanthen-9-ol with p-xylene
which was exposed to vapors of benzene and toluene. 61 The kinetics of such guest exchanges can be correlated to structure by
monitoring the changing structure in a single crystal. The apparatus is similar, but the method requires care in mounting
a single crystal in an open Lindeman tube as shown in Fig. 24. This apparatus was used to monitor the changing structure
of the inclusion compound formed by a xanthenol host containing chlorobenzene which was exposed to the vapors of
toluene.60
The different aspects of dynamic molecular recognition between host and guest molecules have been discussed by Gorbatchuk
et al. 62 They studied a variety of vaporous guest and their enclathration by the solid hosts 2,20 -bis(9-hydroxyfluorene-9-yl)biphenyl
and tert-butylcalix[4]arene. They obtained a series of isotherms which were analyzed in terms of host–guest cooperativity via the
Hill equation. They correlated the free energy of guest transfer with the molar refraction of the guest. The correlation was good
for the diol host but poor for the calixarene. It is pointed out that the vapor saturation methodology is well suited to the investi-
gation of structure–property relationships because the inclusion conditions are strictly comparable for all hosts and guests. We
believe that this mode of host–guest investigation is deserving of further attention.

Figure 22 TG as a method of following the kinetics of a guest exchange reaction, in which the outgoing and incoming guests, G1 and G2, have
significantly different boiling points. Reprinted from Nassimbeni, L. R.; Su, H. CrystEngComm 2013, 15, 7396–7401. Copyright 2016, with permission
from Royal Society of Chemistry.
18 Thermal Analysis

Figure 23 DSC is a suitable method of following a kinetics of a guest exchange reaction but can seldom be employed quantitatively. Reprinted
from Nassimbeni, L. R.; Su, H. CrystEngComm 2013, 15, 7396–7401. Copyright 2016, with permission from Royal Society of Chemistry.

Figure 24 Single crystal in open Lindemann tube employed to monitor the changing structure in a guest exchange reaction. Reprinted from Nas-
simbeni, L. R.; Su, H. CrystEngComm 2013, 15, 7396–7401. Copyright 2016, with permission from Royal Society of Chemistry.

2.02.7 Modulated Temperature Methods

Both modulated TG and modulated DSC employ periodic temperature perturbation, but the latter technique has gained greater
acceptance in practice. In conventional DSC, the sample and the reference are both subjected to the same linear temperature
program and differences in heat flow that flows to both the sample and reference material are detected. Its primary applications
are the analysis of the thermal stability, glass transitions, polymorph transformation, crystallization, or melting of the analyte.
In the case of temperature modulation DSC, the sample and the reference are both subjected to a simultaneous linear and typically
sinusoidal temperature program ( Fig. 25A).63 As an overall effect, the sample temperature is raised on average at a constant rate by
repeated heating and cooling in short time periods. (Fig. 25A presents an example of the temperature profile for a heating exper-
iment, but this approach of temperature modulation is also used for cooling and isothermal experiments.) The linear change in
temperature permits the measurement of heat flow while the modulated change allows the calculation of heat capacity. Therefore
the method is typically applied in analysis of complex systems or heat capacity measurements during kinetic processes.64
The complex theory of Modulated Temperature Differential Scanning Calorimetry (MTDSC) is described by Reading et al. 65 and
the mathematical background of the data manipulation is discussed by Wunderlich and coworkers.66 An excellent review was pub-
lished by Verdonck et al.63 in 1999 describing the MTDSC method and its applications. This work ends with a comprehensive bibli-
ography of the most important works published until that time on selected topics, such as the MTDSC technique and its application
in crystallinity, melting, glass transition, phase transition, thermal conductivity, and heat capacity measurements; and some selected
examples from different domains of chemistry (inorganic compounds, polymers, pharmaceutics, food science). A very recent review
by Knopp et al. was also published on the topic of MTDSC applications in drug development.64
In summary, the average heating rate, corresponding to the rate for a conventional DSC experiment, is called the underlying
heating rate ( Fig. 25B). The modulated heating rate is variable between a minimum and a maximum value; and determined by
the value of the underlying heating rate, the period, and the amplitude of the superimposed wave.
The total heat flow into the sample during a general DSC experiment can be described by the following equation:
dQ=dt ¼ Cs b þ f ðT; t Þ (13)
where Csb represents the heat capacity dependent component and is generally referred to as the reversing signal. The f(T,t) section
represents the kinetic component of the total heat flow and is regarded as the nonreversing signal. A conventional DSC is unable to
Thermal Analysis 19

Figure 25 (A) Temperature as a function of time for a typical DSC and MTDSC experiment. (B) Heating rate as a function of time for a typical DSC
and MTDSC experiment. Reprinted from Verdonck, E.; Schaap, K.; Thomas, L. C. Int. J. Pharm. 1999, 192, 3–20. Copyright 2016, with permission
from Elsevier.

separate these two contributions, thus the sum of the two components is determined. Using MTDSC the total heat flow can be
separated into the reversing and nonreversing components:
Qtotal ¼ Qreversing þ Qnonreversing (14)

Heat capacity, glass transition, and melting typically occur in the reversing heat flow signal, while the nonreversing
or kinetic component generally contains events of enthalpic relaxation, evaporation, crystallization, denaturation, or
decomposition.
An application of MTDSC is shown in Fig. 26 when quenched polyethylene terephthalate (PET) was analyzed. The total
heat flow which is equivalent to a conventional DSC signal exhibits a glass transition (Tg) and an exotherm of cold crystal-
lization and eventually a melting endotherm. Seemingly a base line separation can be found between the crystallization and
the melting endotherm. Once the total heat flow is separated into reversing and nonreversing components, the competition
between the endothermic melting, which appears in the reversing heat flow, and the exothermic (re)crystallization, as
exhibited in the nonreversing heat flow, is revealed and the overlap explains the seemingly nil effect on the total heat
flow in this region.
20 Thermal Analysis

Figure 26 Deconvoluted signals (total heat flow, reversing heat flow, nonreversing heat flow) for the MTDSC heating experiment of a quenched
PET sample and calculation of the initial crystallinity of a quenched PET sample; result from total heat flow: 50.6–35.1 J g 1 ¼ 15.5 J g 1; result
from reversing and nonreversing heat flow: 134.3–134.6 J g 1 z 0 J g 1. Reprinted from Verdonck, E.; Schaap, K.; Thomas, L. C. Int. J. Pharm.
1999, 192, 3–20. Copyright 2016, with permission from Elsevier.

References

1. Lehn, J.-M. Supramolecular Chemistry; VCH: Weinheim, 1995.


2. (a) Steed, J. W.; Atwood, J. L. Supramolecular Chemistry, 2nd ed.; Wiley: Chichester, 2009; (b) Desiraju, G. R.; Vittal, J. J.; Ramanan, A. Crystal Engineering; World Scientific
Publishing Co.: Singapore, 2011; (c) Steed, J. W.; Turner, D. R.; Wallace, K. J. Core Concepts in Supramolecular Chemistry and Nanotechnology; Wiley: Chichester, 2007.
3. (a) Wendlandt, W. W. Thermal Analysis, 3rd ed.; Wiley: New York, 1986; (b) Wunderlich, B. Thermal Analysis; Academic Press: London, 1990; (c) In Thermal Analysis:
Techniques and Applications; Charsley, E. L., Warrington, S. B., Eds.; Royal Society of Chemistry: Cambridge, 1992; (d) In Principles of Thermal Analysis and Calorimetry;
Haines, P. J., Ed.; Royal Society of Chemistry: Cambridge, 2002; (e) Brown, M. E. Introduction to Thermal AnalysisdTechniques and Applications, 2nd Edition; Kluwer
Academic Publishers: Dordrecht, 2001; (f) In Handbook of Thermal Analysis and Calorimetry Principles and Practice; Brown, M. E., Ed.; Elsevier: Amsterdam, 1998.
4. White, M.A. Comprehensive Supramolecular Chemistry; Atwood, J. L., Davies, J. E. D., MacNicol, D. D., Vogtle, F., Eds.; Vol. 8, Davies J. E. D., Ripmeester J.A., Eds.;
Pergamon: Oxford, 1996; Chapter 4.
5. Tamilmani, P.; Panday, M. C. J. Therm. Anal. Calorim. 2016, 123, 1899–1917.
6. Anastasion, M.; Hasapis, Th.; Zorba, T.; Pavlidon, E.; Chrissafis, K.; Parasekevopoulos, K. M. J. Therm. Anal. Calorim. 2006, 84, 27–32.
7. (a) Theng, B. K. G. The Chemistry of Clay-Organic Reactions; Hilger: London, 1974; (b) Thomas, J. M.; Theocharis, C. R. In Inclusion Compounds; Atwood, J. L.,
Davies, J. E. D., MacNicol, D. D., Eds.; Vol. 5; Oxford University Press: Oxford, 1991. Chapter 4.
8. Votinský, J.; Kalousová, J.; Benes, L.; Baudisová, I.; Zima, V. J. Incl. Phenom. Macrocycl. Chem. 1993, 15, 71–78.
9. Atkins, P.; de Paula, J. Physical Chemistry, 8th ed.; Oxford University Press: Oxford, 2006.
10. Nassimbeni, L. R.; Su, H.; Weber, E.; Skobridis, K. Cryst. Growth Des. 2004, 4, 85–88.
11. Lama, P.; Das, R. K.; Smith, V. J.; Barbour, L. J. Chem. Commun. 2014, 50, 6464–6467.
12. Engel, E. R.; Smith, V. J.; Bezuidenhout, C. X.; Barbour, L. J. Chem. Commun. 2014, 50, 4238–4241.
13. Barbour, L. J.; Caira, M. R.; Nassimbeni, L. R. J. Chem. Soc. Perkin Trans. 1993, 2, 1413–1414.
14. Chapter 4 by Gallagher, P. K.; Chapter 5 by Haines, P. J.; Reading, M.; Wilburn, F. W. In Handbook of Thermal Analysis and Calorimetry; Richardson, M. J.; Charsley, E. L.,
Eds.; Vol. 1: Principles and Practice; Brown, M. E., Ed., Elsevier: Amsterdam; 1998.
15. Wicht, M. M.; Báthori, N. B.; Nassimbeni, L. R. Dalton Trans. 2015, 44, 6863–6870.
16. Flynn, J. H.; Wall, L. A. Polym. Lett. 1966, 4, 323–328.
17. Galwey, A. K.; Brown, M.E. In Handbook of Thermal Analysis and Calorimetry; Brown, M. E., Ed.; Vol. 1: Principles and Practice; Elsevier: Amsterdam; 1998; Chapter 3.
18. (a) Brown, M. E.; Galwey, A. K. Anal. Chem. 1989, 61, 1136–1139; (b) Galwey, A. K.; Brown, M. E. Proc. Roy. Soc. Lond. A 1995, 450, 501–512.
19. Koga, N.; Groshi, Y.; Yoshikawa, M.; Tatsuoka, T. J. Chem. Educ. 2014, 91, 239–245.
20. Goel, N.; Singh, U. P. J. Phys. Chem. A 2013, 117, 10428–10437.
21. Loon Neoh, T.; Yamauchi, K.; Yoshii, H.; Furuta, T. J. Phys. Chem. B 2008, 112, 15914–15920.
22. Lin, S.-Y.; Wang, S.-L. Adv. Drug Deliv. Rev. 2012, 64, 461–478.
23. Lin, H.-L.; Hsu, P.-C.; Lin, S.-Y. Asian J. Pharm. Sci. 2013, 8, 19–27.
24. Hsu, P. C.; Lin, H. L.; Wang, S. L.; Lin, S.-Y. J. Solid State Chem. 2012, 192, 238–245.
25. Muravyev, N. V.; Monogarov, K. A.; Bragin, A. A.; Fomenkov, I. V.; Pivkina, A. N. Thermochim. Acta 2016, 631, 1–7.
26. Takamizawa, S.; Nakata, E.; Miyake, R. Dalton Trans. 2009, 1752–1760.
27. Mu, B.; Walton, K. S. J. Phys. Chem. C 2011, 115, 22748–22754.
28. Logvinenko, V. A.; Dybtsev, D. N.; Bolotov, V. A.; Fedin, V. P. J. Therm. Anal. Calorim. 2015, 121, 491–497.
29. Shi, Q.; Tan, Z.-C.; Di, Y.-Y.; Tong, B.; Li, Y.-S.; Wang, S.-X. J. Chem. Eng. Data 2007, 52, 941–947.
30. Loskyll, J.; Maier, W. F.; Stoewe, K. ACS Comb. Sci. 2012, 14, 600–604.
Thermal Analysis 21

31. Roy, D.; Shastri, B.; Mukhopadhyay, K. J. Phys. Chem. B 2012, 116, 7920–7925.
32. Kitahara, Y.; Takahashi, S.; Kuramoto, N.; Sala, M.; Tsugoshi, T.; Sablier, M.; Fujii, T. Anal. Chem. 2009, 81, 3153–3158.
33. Fernandez, J. M.; Peltre, C.; Craine, J. M.; Plante, A. F. Environ. Sci. Technol. 2012, 46, 8921–8927.
34. Riley, S. R. R. J. Chem. Educ. 2015, 92, 932–935.
35. Price, S. L. Acta Cryst. 2013, B69, 313–328.
36. Lemmerer, A.; Bathori, N. B.; Esterhuysen, C.; Bourne, S. A.; Caira, M. R. Cryst. Growth Des. 2009, 9, 2646–2655.
37. Vangala, V. R.; Chow, P. S.; Schreyer, M.; Lau, G.; Tan, R. B. H. Cryst. Growth Des. 2016, 16, 578–586.
38. Codoni, D.; Belton, P.; Qi, S. Mol. Pharm. 2015, 12, 2068–2079.
39. Neto, H. S.; de Avanjo, G. L. B.; dos Santos, L. L.; Cosentino, I. C.; de Sousa Carvalho, F. M.; do Rosario Matos, J. J. Therm. Anal. Calorim. 2016, 123, 2297–2305.
40. Semalty, A.; Tanwar, Y. S.; Semalty, M. J. Therm. Anal. Calorim. 2014, 115, 2471–2478.
41. Sbarcea, L.; Ledeti, I.; Dragan, L.; Kurunczi, L.; Fulias, A.; Udrescu, L. J. Therm. Anal. Calorim. 2015, 120, 981–990.
42. Numthavaj, P.; Thakkinstian, A.; Dejthevaporu, C.; Attia, J. BMC Neurol. 2011, 11, 1–10.
43. Marinescu, D.-C.; Pnicu, E.; Stanculescu, I.; Melter, V. Thermochim. Acta 2013, 560, 104–111.
44. Ford, J. L.; Mann., T. E. Adv. Drug Deliv. Rev. 2012, 64, 422–430.
45. McGregor, C.; Saunders, M. H.; Buckton, G.; Saklatvala, R. D. Thermochim. Acta 2004, 417, 231–237.
46. Giron, D.; Goldbronn, C. J. Therm. Anal. 1995, 44, 217–251.
47. Laye, P. G. In Principles of Thermal Analysis and Calorimetry; Haines, P. J., Ed.; Royal Society of Chemistry: Cambridge, 2002. Chapter 3.
48. Mathkar, S.; Kumar, S.; Bystol, A.; Olawoor, K.; Min, D.; Markovich, R.; Rustum, A. J. Pharm. Biomed. Anal. 2009, 49, 627–631.
49. Comprehensive Supramolecular Chemistry; Atwood, J. L., Davies, J. E. D., MacNicol, D. D., Vogtle, F., Eds.; Vol. 3; Pergamon: Oxford, 1996.
50. Steed, J. W.; Atwood, J. L. Supramol. Chem., 2nd ed.; Wiley: Chichester, 2009.
51. Szejtli, J.; Encyclopedia of Supramolecular Chemistry; Atwood, J. L., Steed, J. W., Eds.; vol. 1; Marcel Dekker: New York, 2004.
52. Orgovanyi, J.; Poppl, L.; Otta, K. H.; Lovas, G. A. J. Therm. Anal. Calorim. 2005, 81, 261–266.
53. Cruickshank, D. L.; Roigier, N. M.; Maurel, V. J.; de Rossi, R. H.; Bujan, E. I.; Bourne, S. A.; Caira, M. R. J. Incl. Phenom. Macrocycl. Chem. 2013, 75, 47–56.
54. Ziaja, P.; Krogul, A.; Pawlowski, T. S.; Litwinenko, G. Thermochim. Acta 2016, 623, 112–119.
55. Lazzarotto, M.; Nachtigall, F. F.; Schnitler, E.; Castellano, E. E. Thermochim. Acta 2005, 429, 111–117.
56. Inclusion Compounds; Atwood, J. L., Davies, J. E. D., MacNicol, D. D., Eds.; Vol. 5; Oxford University Press: Oxford, 1991.
57. Breck, D. W. Zeolite Molecular Sieves; Wiley: New York, 1974.
58. Nunes, L. M.; Airoldi, C. Thermochim. Acta 2005, 435, 118–123.
59. Ambombo Noa, F. M.; Bourne, S. A.; Su, H.; Nassimbeni, L. R. Cryst. Growth Des. 2016, 16, 1636–1642.
60. Nassimbeni, L. R.; Su, H. CrystEngComm 2013, 15, 7396–7401.
61. Ramon, G.; Coleman, A. W.; Nassimbeni, L. R.; Taljaard, B. Cryst. Growth Des. 2005, 5, 2331–2335.
62. Gorbatchuk, V. V.; Tsifarkin, A. G.; ANtipin, I. S.; Solomonov, B. N.; Konovalov, A. I.; Seidel, J.; Baitalov, F. J. Chem. Soc. Perkin Trans. 2000, 2, 2287–2294.
63. Verdonck, E.; Schaap, K.; Thomas, L. C. Int. J. Pharm. 1999, 192, 3–20.
64. Knopp, M. M.; Löbmann, K.; Elder, D. P.; Rades, T.; Holm, R. Eur. J. Pharm. Sci. 2016, 87, 164–173.
65. Reading, M.; Luget, A.; Wilson, R. Thermochim. Acta 1994, 238, 295–307.
66. Wunderlich, B.; Jin, Y.; Boller, A. Thermochim. Acta 1994, 238, 277–293.
2.03 Single-Crystal X-ray Diffraction
LJ Barbour, Department of Chemistry and Polymer Science, University of Stellenbosch, Stellenbosch, South Africa
Ó 2017 Elsevier Ltd. All rights reserved.

2.03.1 Introduction 23
2.03.2 Exploring the Submicroscopic World 23
2.03.3 Why X-rays? 24
2.03.4 Theory 25
2.03.4.1 What Is a Crystal? 25
2.03.4.2 Symmetry and Periodicity 25
2.03.4.3 X-ray Diffraction 26
2.03.4.4 The Reciprocal Lattice 27
2.03.4.5 Structure Factor Equation 28
2.03.5 Practical Crystallography 28
2.03.5.1 Growing and Mounting Single Crystals 29
2.03.5.2 Instrumentation 29
2.03.5.3 Data Collection 30
2.03.5.4 Data Processing 30
2.03.5.5 Structure Solution and Refinement 31
2.03.5.6 Crystal Structure Validation and Archiving 31
2.03.5.7 Structure Interpretation and Publication 31
2.03.5.7.1 Capped-Stick 32
2.03.5.7.2 Ball-and-Stick 32
2.03.5.7.3 Space Filling 32
2.03.5.7.4 Thermal Ellipsoid Plots 33
2.03.5.7.5 Mixed-Metaphor Illustrations 33
2.03.5.7.6 Packing Diagrams 37
2.03.5.7.7 The Use of Color 38
2.03.5.7.8 Hirshfeld Surface Analysis 38
2.03.5.8 Limitations of X-ray Crystallography 38
2.03.5.9 Crystallography and Supramolecular Chemistry 39
References 42

2.03.1 Introduction

It is easy to argue that single-crystal X-ray diffraction has revealed more about molecular structure, chemical bonding, and intermo-
lecular interactions than any other analytical technique known to science. Over the past century crystallography has had a vast
impact on the development of physics, chemistry, materials science, mineralogy, metallurgy, biology, pharmaceuticals, and, of
course, supramolecular chemistry. It would be impossible to cover all aspects of crystallography in just these few pages; instead,
this article aims merely to provide the reader with an appreciation for the power of the technique, particularly with regard to
the impact that it has had on the field of supramolecular chemistry. Following a superficial discussion of the background to crys-
tallography, a general overview is provided on instrumentation, limitations of the technique, and examples of supramolecular
systems that have been studied in the solid state. In short, reading this article will not turn you into a crystallographer!

2.03.2 Exploring the Submicroscopic World

Humans have evolved a set of senses that allow us to interact effectively with the world in which we live. Despite the vast range of
wavelengths that constitute the electromagnetic spectrum (Fig. 1), our eyes are only sensitive to the rather limited portion
comprising the aptly named “visible” spectrum. In a world bombarded with white-light radiation from the sun, our limited sensi-
tivity to the electromagnetic spectrum has proven quite adequate, at least for most of human history, to allow us to directly image
objects at a scale that is relevant to everyday survival. However, over the past century we have developed technologies to access
almost the entire electromagnetic spectrum, thus allowing us to indirectly perceive the “unseen” at scales ranging from the massive
(e.g., far-off galaxies) to the miniscule (molecules, atoms, and subatomic particles). We have consequently learned far more about
Nature in the past century than during all of the rest of human history combined. Rather than merely satisfying our curiosity, these
newfound abilities have opened up new opportunities for exploring our world, piquing our interest in the heretofore unseen even

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12493-X 23


24 Single-Crystal X-ray Diffraction

Figure 1 The electromagnetic spectrum. Image adapted from http://www.everythingmaths.co.za/science/grade-10/11-electromagnetic-radiation/11-


electromagnetic-radiation-03.cnxmlplus.

further. Imaging objects at the molecular and atomic level has revolutionized our understanding of the properties of materials and
the functioning of biological processes. Indeed, the ability to image the atomic-scale world has empowered our efforts to manip-
ulate that world in ways that were unimaginable not so long ago. Vastly extending the scale at which we can image objects has
provided us with power over the submicroscopic world, with clear consequences for enhancing our overall quality of life and
longevity.

2.03.3 Why X-rays?

When imaging an object using electromagnetic radiation a practical constraint is imposed by diffraction effects such that the reso-
lution is limited to approximately half the wavelength being used. In order to “see” atoms and molecules we need to satisfy two
important criteria: (i) the wavelength of the radiation needs to be in the range of about 0.1–2 Å (i.e., X-rays) and (ii) the objects
to be imaged must be arranged in a regular array akin to a three-dimensional diffraction grating.
It is beyond the scope of this article to describe the entire historical development of X-ray crystallography, from the discovery of
X-rays more than a century ago, through to the modern era where the determination of even highly complex crystal structures has
become a routine undertaking using the most advanced instrumentation and computing power. The interested reader is instead
referred to the authoritative book Early Days of X-ray Crystallography by Authier for detailed accounts of the most important early
developments.1 However, it is worth paying tribute to the critical discoveries that paved the way to modern crystallography.
Wilhelm Röntgen (Fig. 2A) is credited with the discovery of X-rays in 1895,2,3 an achievement for which he was awarded the first
Nobel Prize in Physics in 1901. Although X-rays had been observed before, Röntgen was the first to recognize that they constituted
a previously unknown form of radiation, and he was also the first to study them systematically. The discovery electrified the physics
community which, for almost two decades thereafter, was split into two factions. Some believed that X-rays were particles while
others believed that they were waves. However, Röntgen himself had failed to establish that X-rays could be reflected, refracted,
or diffracted, despite repeated attempts to do so. Assuming that X-rays were waves, Stark estimated from energetic measurements
that their wavelengths ranged from 0.1 to 100 Å.4

Figure 2 (A) Wilhelm Röntgen, (B) Max von Laue, (C) William Henry Bragg, and (D) William Lawrence Bragg. Images taken from https://en.
wikipedia.org/wiki/Wilhelm_R%C3%B6ntgen, https://en.wikipedia.org/wiki/Max_von_Laue, https://en.wikipedia.org/wiki/William_Henry_Bragg, and
https://en.wikipedia.org/wiki/William_Lawrence_Bragg.
Single-Crystal X-ray Diffraction 25

In a discussion that took place early in 1912, Peter Paul Ewald disclosed to Max von Laue (Fig. 2B) that crystals were believed to
consist of regular arrays of atoms or molecules, which he referred to as resonators.1 When considering the probable distances
between the particles, von Laue concluded that a crystal should act as a three-dimensional diffraction grating for X-rays and he
enlisted the help of Friedrich and Knipping to test his hypothesis using a crystal of zinc blende. In a landmark event for science,
exposure of the crystal to an X-ray beam produced a diffraction pattern, thus demonstrating conclusively the wave nature of X-
rays.5 Von Laue was awarded the Nobel Prize in Physics in 1914 “for his discovery of the diffraction of X-rays by crystals.”
After confirming von Laue’s experiment, William Henry Bragg (Fig. 2C) and his son William Lawrence (Fig. 2D) advanced the
concept by demonstrating that X-ray diffraction can be used to deduce the spatial arrangement of the particles within the crystal.
They determined the crystal structures of simple alkali halide salts6 and diamond.7 In order to explain the geometry of the diffrac-
tion pattern, William Lawrence Bragg proposed the well-known Bragg equation, which relates the angle of diffraction q to the wave-
length l of the X-rays and the interplanar spacings d of the crystal lattice:
n l ¼ 2dsinq
The Braggs were awarded the Nobel Prize in Physics “for their services in the analysis of crystal structure by means of X-rays.” The
pioneering work of Röntgen, von Laue, the Braggs, and many others paved the way for the development of X-ray crystallography
into the powerful atomic-level imaging technique that it is today, thereby revolutionizing our knowledge of atomic and molecular
structure, as well as intermolecular interactions. The impact of their discoveries on modern science is immeasurable.

2.03.4 Theory

Over the past century the approach to determining “small-molecule” crystal structures by means of single-crystal diffraction, as pio-
neered by the Braggs, has spawned a number of related diffraction techniques. Each of these employs highly specialized instrumen-
tation and methods of data interpretation, and is selected according to the nature of the structural problem in question. These
techniques include neutron diffraction, powder X-ray or neutron diffraction, macromolecular crystallography, electron diffraction,
small-angle X-ray or neutron scattering, wide-angle scattering, and fiber diffraction. Some of these methods are described elsewhere
in this volume and the current article deals specifically with small-molecule single-crystal X-ray diffraction. A “small molecule” can
loosely be defined as a molecule consisting of fewer than 1000 atoms and small-molecule crystallography is often also referred to as
chemical crystallography as distinct from macromolecular crystallography.
In order to determine a scientifically credible crystal structure the would-be crystallographer needs to understand what a crystal
is, how X-rays are diffracted, how the diffraction pattern relates to the crystal structure, and limitations of the technique. The
required practical skills include growing and mounting crystals, setting up and measuring the diffraction pattern, data processing,
structure solution and refinement, structure validation, and interpretation of the structure in order to extract chemically relevant
information. Many of these operations are sufficiently routine such that they can now be automated with the aid of robots and
advanced computer programs. However, many crystal structure determinations involve potential pitfalls that can still only be
avoided by informed human intervention, and any crystallographer would be wise not to entrust the entire process to a computer
program that lacks chemical intuition.

2.03.4.1 What Is a Crystal?


It is important to understand the difference between a crystal and its structure. A crystal can be defined as a solid whose structure is
characterized by a periodic repetition in three dimensions of a motif composed of atoms, molecules, or ions. A crystal structure thus
consists of a conceptually infinite set of coordinates that indicate the spatial arrangement of the constituent atoms relative to one
another. So a crystal structure is a model that allows us to make assertions about molecular structure, packing, and interactions in
the crystalline solid. Of course, in reality a crystal is neither infinite nor a perfect periodic arrangement of its contents. Crystals
typically possess defects (or dislocations) in their internal periodicity and they are obviously finite entities since the periodicity
terminates at the crystal boundaries (i.e., facets). A crystal structure is only concerned with the relative arrangement of atoms within
the bulk of the crystal and tells us little about its surface structure.

2.03.4.2 Symmetry and Periodicity


A point group is a set of symmetry operations that map an object or arrangement of objects onto itself, while keeping at least one
point unmoved. Therefore, the symmetry operations may entail rotations, reflections, inversions, or combinations of these.
However, symmetry operations involving translations are prohibited. Of the infinite number of possible three-dimensional point
groups, the requirement imposed by periodicity within a crystal dictates that only 32 are relevant to crystal packing.
Conceptually, a crystal consists of a three-dimensional tiling of an irreducible repeat unit. If one arbitrarily selects a point within
the crystal structure, it is possible to locate additional points with identical environments that repeat in three noncollinear direc-
tions. Together, these points constitute a lattice (note that a lattice is not a physical entity but rather a geometrical description of
periodicity). If we select a lattice point and then consider the three shortest noncollinear vectors to neighboring lattice points, these
vectors can be used as the basis for describing the three-dimensional periodicity of the crystal (i.e., the unit cell). Indeed, the unit cell
26 Single-Crystal X-ray Diffraction

is thus the smallest entity that can be seen as being representative of the bulk material (i.e., the density and stoichiometry of the unit
cell are the same as that of the material). These concepts are illustrated in two dimensions in Fig. 3.
We specify the geometry of a three-dimensional unit cell using six parameters that represent the lengths of three repeat distances
(a, b, and c) and the angles between these vectors (a between b and c, b between a and c, and g between a and b). The unit cell axes
are not necessarily equal in length or orthogonal to one another; three-dimensional tilings can be accommodated by seven different
sets of relationships between the unit cell axes, giving us the seven crystal systems listed in Table 1.
Every lattice can be assigned a point group symmetry, which is a set of symmetry operations that map a point lattice onto itself. A
unit cell with lattice points only at its eight corners is said to be primitive and is denoted by the symbol P. Although every lattice can
be described as primitive, it is preferable in some cases to select a centered lattice (i.e., as described by a unit cell containing addi-
tional lattice points) in order to take advantage of the presence of higher point group symmetry than the primitive lattice choice of
crystal system might imply. The combination of seven crystal systems with the concept of lattice centering yields 14 Bravais lattices.
When we combine the 14 Bravais lattices (i.e., lattice periodicity) with the 32 crystallographic point groups, we get 230 unique space
groups. A crystallographic space group is a set of geometrical symmetry operations that map a three-dimensional periodic arrange-
ment of identical objects onto itself. Although a good grasp of space group symmetry is central to understanding crystal packing, it is
beyond the scope of this article to provide an in-depth treatment of the subject. The reader is instead referred to more comprehen-
sive texts and Volume A of the International Tables for Crystallography.8 In addition to symmetry elements such as rotations, reflec-
tions, and inversions (and combinations thereof), space group symmetry also includes rotations and reflections combined with
fractional unit cell translations (i.e., screw axes and glide planes).
In order to reconstruct a model of a crystal structure, the minimum information that is required includes the unit cell parameters,
space group symmetry, and atomic coordinates. Owing to space group symmetry and lattice periodicity, we only need the atomic
coordinates of the asymmetric unit (i.e., the unique set of atoms that are not related to one another by symmetry or unit cell trans-
lations). The position of each atom within the unit cell is specified as a set of three coordinates corresponding to its position along
the three crystallographic axes, with values ranging from 0 to 1.

2.03.4.3 X-ray Diffraction


For a periodic three-dimensional arrangement of atoms it is possible to define families of parallel planes that pass through sites of
identical electron density. Fig. 4 illustrates this concept for a two-dimensional lattice. The orientations of each family of planes rela-
tive to the unit cell axes can be specified using three integers, which are termed the Miller indices (hkl), where h, k, and l indicate the
periodicity of the plane in the directions of a, b, and c, respectively. X-rays interact with the electrons in a crystal, and the electrons are

Figure 3 A two-dimensional periodic arrangement of identical objects (left). A set of dimensionless points (shown as black circles) with the same
two-dimensional periodicities can be superimposed on the tiling (middle) such that they define a lattice. A unit cell can be defined such that its edges
link the closest noncollinear lattice points (right). In this instance the placement of the unit cell is arbitrary provided that its geometry and orientation
are maintained. The total composition of each of the different unit cell selections (blue lines) is the same in each case, and the density of the unit cell
is equal to the density of the bulk.

Table 1 The seven crystal systems

Crystal System Lattice centering Point group symmetry at lattice points Unit cell axial relationships

Triclinic P 1 a b c; a b g 90 , 120


Monoclinic P, C 2/m a b c; a ¼ g ¼ 90 ; b 90 , 120
Orthorhombic P, C, I, F mmm a b c; a ¼ b ¼ g ¼ 90
Tetragonal P, I 4/mmm a ¼ b c; a ¼ b ¼ g ¼ 90
Cubic P, I, F m3m a ¼ b ¼ c; a ¼ b ¼ g ¼ 90
Hexagonal P 6/mmm a ¼ b c; a ¼ b ¼ 90 ; g ¼ 120
Trigonal R or P 3m a ¼ b ¼ c; a ¼ b ¼ g 90 , <120
Single-Crystal X-ray Diffraction 27

Figure 4 A two-dimensional lattice. Depending on the direction of the observer’s gaze across the lattice, different spacing of the lattice points can
be discerned. Photograph by the author.

distributed such that they form a three-dimensional diffraction grating. The geometric description of X-ray diffraction is often
simplified by describing the phenomenon as constructive interference of X-rays being reflected by adjacent parallel planes
(Fig. 5). For this reason, we refer to diffracted X-rays as “reflections,” but this is a misnomer.

2.03.4.4 The Reciprocal Lattice


For a given set of lattice planes a diffracted beam can potentially be detected providing that the crystal is oriented such that the Bragg
equation is satisfied for that set of planes. The collection of diffracted beams for all possible integer combinations of Miller indices
(hkl) constitute the diffraction pattern. In practice, the maximum and minimum values of h, k, and l are limited by the resolution of
the experimental configuration (i.e., the X-ray wavelength, the scattering power of the crystal, and the geometry of the instrument). It
is useful to consider each diffracted beam as a point in space, which is indeed how it is recorded by an imaging device such as
a photographic plate or a digital X-ray camera. Owing to the reciprocal relationship between the angle of diffraction and the
d-spacing of the planes giving rise to the spot, the diffraction pattern is reconstructed in what we call reciprocal space. Each spot
is characterized by its Miller indices and its recorded relative intensity. When plotted in reciprocal space the diffraction pattern
resembles a lattice, which can be assigned a unit cell that has a reciprocal relationship to the real-space unit cell of the crystal.

Figure 5 Simplification of X-ray diffraction by two adjacent parallel planes (hkl). The spacing between the planes is AC ¼ d, the wavelength of the X-
ray beam is l, and the angle between the incident beam and the planes is q. The two waves of the incident beam are in phase with one another and
travel the same distance until they reach AB. When the waves are scattered by electrons at points A and C, they will once again be in phase if the
additional distance (i.e., BC þ CD) traveled by the bottom wave is equal to an integral number of wavelengths. Since BC ¼ CD ¼ d sinq, the condition
for constructive interference is satisfied when nl ¼ 2d sinq (n is a positive integer) and a diffracted beam will be observed such that it appears to be
a reflection of the incident beam. The angle between the incident and the diffracted beams is 2q.
28 Single-Crystal X-ray Diffraction

Once the diffraction pattern has been recorded, we can derive the real-space unit cell parameters from those of the reciprocal cell.
Instead of regarding the reciprocal lattice as a set of dimensionless points, we can assign a weight to each point according to the
recorded relative intensity of the reflection. Using the Miller indices and intensities of the reflections, the point group of the diffrac-
tion pattern can be assessed. Friedel’s law states that reflections hkl and hkl should have the same intensities (exceptions exist
due to anomalous dispersion). Therefore, every diffraction pattern should appear to possess at least inversion symmetry in addition
to the point group symmetry of the space group of the crystal. Because of the Friedel relationship, it is not possible to determine
conclusively from the symmetry of the diffraction pattern whether or not the space group is centrosymmetric.

2.03.4.5 Structure Factor Equation


Although von Laue had suggested that each atom in the unit cell contributes to the intensity of each reflection, it was Ewald who
introduced (and named) the structure factor.9 The diffraction pattern is a Fourier transform of the electron density distribution of
a crystal; the structure factor Fhkl is a function of the amplitude and phase of the reflection produced by lattice plane hkl:
X
N
Fhkl ¼ fj e2piðhxj þkyj þlzj Þ
j¼1

jFhkl j2 is proportional to the intensity Ihkl of reflection hkl and the terms are summed over all atoms j ¼ 1, ., N within the unit cell.
For each atom j the structure factor equation comprises an electronic component fj, as well as a structural component exp(2pi
[hxj þ kyj þ lzj]) using the fractional atomic coordinates xj, yj, zj. The inverse Fourier transform relationship is expressed as:
1 X XX
rx;y;z ¼ F e2piðhxþkyþlzÞ
V h k l hkl

where rx,y,z is the electron density at unit cell coordinates x, y, z and V is the volume of the unit cell.
In order to determine the electron density at x, y, z it is necessary to determine both the amplitude and phase of each reflection.
However, in practice it is only possible to measure amplitudes, while the phase information is lost. This is referred to as the Phase
Problem.
Before the advent of advanced numerical methods for solving the phase problem, the phases of reflections could be approxi-
mated or calculated in a number of ways. In simple cases one could construct a hypothetical structure, which could then be tested
by calculating the diffraction pattern and comparing it to the observed pattern. The Patterson function provides a map of vectors
that occur midway between atoms, but this method is only effective for structures that contain “heavy” atoms. The third method
involves molecular replacement, which requires knowledge of a substantial part of the crystal structure (this method is typically
used in macromolecular crystallography). In 1985 Karle and Hauptman (Fig. 6) shared the Nobel Prize in Chemistry “for their
outstanding achievements in the development of direct methods for the determination of crystal structures.” They had introduced
a statistical approach to solving the phase problem,10 and this approach is now used routinely to solve crystal structures.

2.03.5 Practical Crystallography

Although the determination of a crystal structure using single-crystal diffraction has become a routine undertaking that can typically
be accomplished in the timeframe of minutes to hours, the process involves a number of steps that require theoretical and practical
expertise, and access to specialized instrumentation. This section provides a brief overview of the process and the equipment. Much

Figure 6 Jerome Karle (left) and Herbert Hauptman (right). Taken from https://www.nobelprize.org/nobel_prizes/chemistry/laureates/1985/.
Single-Crystal X-ray Diffraction 29

of the text that follows is based specifically on instrumentation and processing software supplied by Bruker AXS, but the procedure is
similar for other instrument suppliers.

2.03.5.1 Growing and Mounting Single Crystals


Our objective is to record a three-dimensional diffraction pattern for a single crystal. The term “single crystal” refers to a crystalline
solid consisting of a contiguous assembly of identical unit cells packed together in the same orientation. It has already been stated
that a typical crystal is not as perfect as our ultimate model of its structure might suggest. However, the technique of single-crystal
diffraction is somewhat resilient to lattice defects, twinning, and mosaicity. The most common method of growing crystals in a labo-
ratory is to prepare a saturated solution of the compound of interest, and then to allow the solvent to evaporate slowly. Crystals may
begin to form once the solution becomes supersaturated. Other common methods include sublimation, slow cooling of a saturated
solution, solvothermal synthesis, vapor diffusion, liquid–liquid diffusion, and gel growth.
Once the crystals have grown, it is necessary to inspect them under an optical microscope in order to identify and mount a suit-
able candidate for the diffraction experiment. A microscope equipped with cross-polarizers, a rotation stage, and transillumination
capability is preferable. Because crystals can become intergrown or otherwise adhered to one another, uniform rotation of polarized
light is usually a good indication of a single crystal (Fig. 7). Crystals with dimensions in the range 0.1–0.4 mm are generally consid-
ered suitable for diffraction. If the crystals are too large or intergrown, it is often possible to cleave a single fragment of the appro-
priate size using a sharp blade.
Once a suitable crystal has been identified it must be attached to a goniometer head which, in turn, is mounted onto the goni-
ometer of the diffractometer. The crystal is usually adhered to a glass fiber, a nylon loop, or a MiTeGen11 polymer mount by means
of a small quantity of immersion oil. The crystal mount material should not produce a diffraction pattern and it should be relatively
transparent to the X-ray beam. In a typical experiment, the surface tension of the immersion oil is sufficient to keep the crystal in
place until it can be placed in the cold stream of a cryostat fitted to the diffractometer, whereupon the oil will stiffen and immobilize
the crystal. It is important that the crystal should not change its orientation during data collection. For room- or high-temperature
data collection it is better practice to glue the crystal to a glass fiber using epoxy. Data collection at much higher temperatures may
require specialized heat-resistant adhesives.
If the crystal is not stable to the atmosphere during normal handling, it may be necessary to take special precautions to prevent
decomposition. For example, crystals may be air or moisture sensitive, or they may have lattice-included solvent that escapes, thus
causing the crystal to fracture. If coating the crystal in immersion oil fails to prevent decomposition, the crystal should be mounted
in a glass capillary either under inert atmosphere conditions, or with mother liquor. The presence of the glass capillary in the beam
usually gives rise to some background due to scattering of X-rays.

2.03.5.2 Instrumentation
In principle, the experimental configuration of a modern diffractometer is still much the same as that of the original apparatus used
by von Laue, Friedrich, and Knipping. A diffractometer consists of an X-ray source (including a monochromator and a collimator),
a goniometer for orienting the crystal in the beam, and a detector. The technology behind a conventional diffractometer has
advanced significantly in the past century. Instead of high-powered Coolidge tubes we now use relatively low-powered high-flux

Figure 7 Rotation of plane-polarized light by a crystal. When the polarizer and analyzer are oriented at right angles relative to one another no light
should pass through to the observer. However, if the crystal rotates the polarized light admitted by the polarizer then a small component of this light
can pass through the analyzer. This affect is not produced by hexagonal crystals viewed along the hexagonal axis, or by cubic crystals.
30 Single-Crystal X-ray Diffraction

microfocus sources, rotating anode tubes, or synchrotron facilities. Photographic film has been replaced by electronic cameras
equipped with CCD or CMOS sensors. Goniometers are automated and data collection, data reduction, and structure determination
are carried out using powerful computers that employ user-friendly interfaces and advanced algorithms.
A typical diffractometer is equipped with a liquid nitrogen cryostat to control the temperature of the crystal during data collec-
tion; it has become routine to determine crystal structures at 100 K in order to reduce thermal motion of the atoms and to thereby
sharpen the electron density map. Although far less common, helium cryostats are available to cool crystals to a few degrees above
absolute zero. Other in situ techniques for controlling the conditions under which data are recorded include environmental gas
cells12–14 and diamond anvil cells.15 Moreover, crystals can be irradiated with light to investigate the effects of photoirradiation
on the crystal structure.16,17

2.03.5.3 Data Collection


A crystal is mounted on the goniometer of the diffractometer with its unit cell in a random orientation. When the crystal is
irradiated in any particular orientation relative to the X-ray beam, only the lattice planes that satisfy the Bragg equation for
that orientation will give rise to diffracted rays. Depending on the position of the X-ray camera, only a few of these diffracted
rays will actually be recorded. In order to record as much of the reciprocal lattice as possible it is necessary to take many expo-
sures with the crystal in different orientations. Most diffractometers (Fig. 8) are built around automated goniometers with
several axes that intersect at the position of the crystal. The crystal is systematically rotated by a small angle about one of these
axes during the recording of each frame, and thus the reciprocal lattice is gradually swept through the plane of the detector.
After determining the positions and intensities of the spots on the successive frames, the control software reconstructs a virtual
reciprocal lattice.
After mounting and centering the crystal the first step is to determine the unit cell by recording only a subset of the reciprocal
lattice. This subset usually contains sufficient data to determine the reciprocal unit cell, from which the real-space unit cell can be
derived. The orientation of the real-space cell with respect to the coordinate system of the goniometer is also determined in the form
of an orientation matrix. The crystal system can be deduced from the unit cell parameters, and it is useful to know the likely Laue
symmetry of the diffraction pattern before formulating the data collection strategy. In principle it is only necessary to record the
unique portion of the reciprocal lattice. However, redundant data are useful for a number of reasons, including the application
of empirical absorption corrections, averaging symmetry-equivalent reflection intensities using multiple measurements, and veri-
fying the Laue class to which the crystal belongs.

2.03.5.4 Data Processing


A typical data collection protocol results in the recording of a large number of frames. Once the reciprocal lattice has been extracted
from these frames, several corrections need to be applied to compensate for absorption, Lorenz and polarization effects. Then the
diffraction data can be saved to a text file where each line of text represents a measured Bragg reflection. Each reflection is assigned
values of h, k, l, relative intensity, and the standard error of the intensity measurement. The next step involves a numerical evaluation
of the diffraction data and refined unit cell parameters to determine the probable space group based on systematic absences that
might be present in the diffraction record.

Figure 8 A Bruker SMART Apex diffractometer.


Single-Crystal X-ray Diffraction 31

2.03.5.5 Structure Solution and Refinement


It should be emphasized that the final crystal structure is only a model based on experimental data; the accuracy and precision of the
model are only as good as the original data, and our ability to interpret these data in a chemically sensible way. Structure solution
involves determining the phases of enough reflections to yield a good approximation of most of the nonhydrogen atomic coordi-
nates. These coordinates are then subjected to an iterative process of refinement until the model converges to agree as well as
possible with the available data, and cannot be improved by further refinement.
Several computer packages are available for structure solution and refinement. By far the most popular of these is the SHELX
program suite,18 the latest version of which is SHELX-2014.19 It is beyond the scope of this article to describe all of the capabilities
of SHELX and the reader is instead referred to the SHELX manual. The programs SHELXS and SHELXL are used to solve and refine
crystal structures, respectively. Both of these programs are controlled using instructions embedded in a text file and the results of the
calculations are also output in text format. The lack of a native graphical user interface (GUI) facilitates cross-platform implemen-
tation of the SHELX software. However, since SHELX instructions are parsed via text files the programs are amenable to the devel-
opment of third-party GUIs and several of these are in common use (e.g., WinGX,20 Olex2,21 ShelXle,22 and X-Seed23). Although
a GUI is a convenient way to interact with SHELX in an intuitive fashion, its use should not be seen as a substitute for being
completely familiar with the entire SHELX instruction set.
A successful structure solution usually yields all or most of the nonhydrogen atoms of the model. Subsequent refinement is used
to expand the structure to locate additional atoms (e.g., unanticipated solvent), to model disorder (if any), and to model thermal
motion of the atoms. Hydrogen atoms are usually placed in calculated positions according to predefined geometrical criteria, and
then refined using a riding model based on the positions of the parent atoms. Where calculated positions are not well defined,
hydrogen atoms can either be omitted from the model or refined independently based on difference electron density peaks that
might be present. It should be noted that the positions of hydrogen atoms are not determined reliably using X-rays because X-
rays are scattered by electrons. The single electron of a hydrogen atom is normally polarized in the direction of a single bond to
the parent atom. Therefore, an electron density peak for a hydrogen atom will seldom be a good indication of the position of
the atom’s nucleus. If the accurate determination of hydrogen atom positions is a critical aspect of the structure determination,
then neutron diffraction data are required.

2.03.5.6 Crystal Structure Validation and Archiving


When a crystal structure determination is finalized the results are archived in plain text format as a Crystallographic Information File
(CIF).24 Most journals require authors to submit CIFs for all reported crystal structures as supplementary information. A general
requirement is also that the CIF be submitted to the Cambridge Crystallographic Centre for archiving in the Cambridge Structural
Database (CSD).25 Many crystallographic pitfalls and shortcomings can be detected by suitable algorithms. For this reason, the
International Union of Crystallography provides a web-based service called checkCIF26 to examine a CIF for possible errors or other
troubling issues. The checkCIF routine returns a validation report in Adobe PDF format and many journals also require this report
for each CIF as additional supplementary information for review purposes. Alerts are flagged as Level A, B, or C in decreasing order
of significance. Authors are usually required to address serious alerts either by improving the quality of their data and/or their
model. If this is not possible then the author should at least explain why the issue cannot be addressed. An example of a Level
A alert that can be justified is “Structure Contains Solvent Accessible VOIDS of 441 Ang3.” Although this alert is intended as
a warning that a solvent molecule or a counterion has been mistakenly omitted from the model, it would also arise for a porous
crystal where such voids are feasible.
The CSD is an archive of published and deposited crystal structures that was established in 1965. As of 2016 the database
contains over 800,000 crystal structures, the oldest of which date back to 1923. The exponential growth of the CSD is shown in
Fig. 9. Using the search program ConQuest, licensed users of the CSD have access to all of the crystal structures in the database.
ConQuest features a GUI and employs sophisticated search criteria to interrogate the database. Other useful programs that form
part of the system include Mercury for visualization and analysis of structures, Mogul to assess molecular geometries, and IsoStar
to investigate interactions. All of these programs are particularly useful to supramolecular chemists.

2.03.5.7 Structure Interpretation and Publication


A completed crystal structure analysis should produce a file containing at least the unit cell parameters, the space group, and a list of
atomic coordinates for the atoms comprising the asymmetric unit. The CIF is considered the standard format for communicating
crystallographic data and includes additional experimental information such as crystal size, morphology and color, chemical
formula and weight, the temperature of the crystal during data collection, details about the instrument, unit cell determination
and final refinement parameters, atomic displacement parameters, etc. A number of computer programs are available to read
CIFs so that the user can interactively explore the structural model. The space group symmetry can be used to generate atoms
that are related to the asymmetric unit by symmetry and the unit cell parameters can then be used to expand the structure as far
as required into three dimensions. Apart from accessing precise details about molecular structure (e.g., intramolecular connectivity;
bond lengths; and bond, torsion, and dihedral angles), we can visualize how the molecules pack in three dimensions and we can
also assess intermolecular interactions.
32 Single-Crystal X-ray Diffraction

Figure 9 Growth of the Cambridge Structural Database.

The program Mercury, which forms part of the CSD distribution, is an invaluable tool for exploring a crystal structure and for
producing publication quality molecular graphics. A version of Mercury with reduced functionality is available free of charge to
researchers who are not licensed users of the CSD. Some of the other programs that are available include CrystalExplorer,27 Materials
Studio,28 Olex2,21 PLATON,29 WinGX,20 X-Seed,23 and several others.
The effective use of molecular graphics30 is central to describing crystal structures in journal articles, conference presentations,
and dissertations. Molecular graphics has benefitted substantially from advances in computing power such that it is now possible to
prepare esthetically compelling and informative renderings of structural models that are far superior to those produced by the pio-
neering vector graphics programs such as PLUTO31 and Oak Ridge Thermal Ellipsoid Plot (ORTEP)32 (Figs. 10 and 11).
The ultimate objective of X-ray crystallography is to produce a model of a crystal structure. Apart from facilitating accurate
measurements of intra- and intermolecular parameters, the model allows us to visualize aspects of a crystal structure to aid our
molecular-level understanding of molecular structure, crystal packing, and physical properties of the material. It is important to
keep in mind that the images are just metaphors. What a molecule actually looks like has no meaning because the concept of vision
is meaningless at the molecular level. There are four main styles (or metaphors) that can be used to represent atoms and molecules.

2.03.5.7.1 Capped-Stick
This style is called “capped-stick” because the bonds are represented as cylinders of a given radius, and each atom is represented as
a sphere of the same radius as the bonds. Therefore, the bonds look like sticks that are capped with hemispheres. Proportions are
important and a good value to use for capped-stick bond widths is 0.18 Å (Fig. 12).

2.03.5.7.2 Ball-and-Stick
The origin of the term “ball-and-stick” should be obvious. Atoms can either be represented as spheres of the same size, or the
different elements can be shown as spheres with radii that are proportional to their van der Waals radii (Fig. 13). A good guideline
is to assign the radii of the spheres to be about 0.2 times the atomic van der Waals radii (except hydrogen atoms, where the factor
should be about 0.15). Bond radii should be about 0.08–0.10 Å. Capped-stick and ball-and-stick figures are useful for showing
connectivity because their relatively uncluttered nature allows one to generally see where all the atoms are, provided that a suitable
projection is used. However, these figures can be deceptive in terms of showing the space that the molecules occupy.

2.03.5.7.3 Space Filling


Space-filling illustrations (Fig. 14) are generally not suitable for showing connectivity. However, they provide the observer with an
excellent sense of the space that a molecule occupies (i.e., its bulkiness) relative to other molecules, or relative to the overall packing
arrangement. An example of the deceptiveness of using ball-and-stick illustrations is shown in Fig. 15. The structure was described as
a stack of dimers that form channels with rectangular dimensions of 3.9  9.5 Å.36 This assumption was based on a ball-and-stick
plot similar to that shown in Fig. 15A. When the same dimer is shown in the space-filling metaphor (Fig. 15B), it can easily be seen
that there is no atomic-scale space between the two molecules, and that the two molecules are in van der Waals contact with one
another. Hence the apparent channel shown in Fig. 15A is nonexistent.
Fig. 16 shows two space-filling images illustrating the OeH/O hydrogen bonded ring in Dianin’s compound (Fig. 16A) and
the SeH/S hydrogen bonded ring of the thiol derivative of Dianin’s compound (Fig. 16B). The two molecular images are specif-
ically shown at the same scale because the objective is to indicate the relative sizes of the holes through the rings. The two separate
assemblies are also shown in approximately the same orientation because they are structurally relateddusing arbitrary orientations
Single-Crystal X-ray Diffraction 33

Figure 10 PLUTO vector graphic illustrations of supramolecular complexes involving (A) calix[6]arene and C60 and (B) calix[6]arene and C70.
Corresponding illustrations produced by POV-Ray33 and X-Seed are shown in (C) and (D), respectively.34

for each would distract the reader from the main objective of the images. Fig. 16 illustrates that it is necessary to consider the
purpose of the image and that the reader should not be distracted by unnecessary information or inconsistencies.

2.03.5.7.4 Thermal Ellipsoid Plots


The ORTEP32 program was first developed in 1965 by Carroll Johnson to illustrate anisotropic atomic displacement ellipsoids. The
original version of ORTEP produced vector graphic line drawings (see Fig. 11A) but vector graphics have generally been superseded
by raster graphics. Thermal displacement images are useful for illustrating the thermal motion experienced by atoms and, in general,
provide a qualitative and very visual assessment of the “quality” of the crystal structure analysis. Every atom experiences some degree
of anisotropic thermal motion, and structure refinement programs model this motion as an ellipsoid. In extreme cases, thermal
ellipsoids may indicate dynamic disorder as well. For example, at room temperature each of the four tert-butyl groups of p-tert-
butylcalix[4]arene is modeled as disordered over two positions. However, the thermal ellipsoid diagram shown in Fig. 17A indicates
that the tert-butyl groups are probably even more disordered than the final model suggests, owing to their substantial degree of
rotational freedom.
A dinuclear metal–organic ring with acetonitrile solvent and BF 4 anion is shown in Fig. 17B. The small and relatively spherical
thermal ellipsoids indicate that the atoms do not experience much thermal motion and thus imply that the structure determination
is very reliable. The larger thermal ellipsoids of the acetonitrile guest molecule indicate that the solvent is probably not held in place
strongly; this is generally to be expected for small volatile molecules such as solvents and should not be surprising.

2.03.5.7.5 Mixed-Metaphor Illustrations


The various metaphors described earlier can be mixed, as deemed appropriate, in order to illustrate a concept. Optimal use can be
made of the various advantages that the different metaphors offer. Fig. 18 shows solvent molecules confined within a channel
34 Single-Crystal X-ray Diffraction

Figure 11 (A) ORTEP illustration (30% probability) of a cocrystal between p-iodocalix[4]arene benzyl ether and C60.35 (B) A corresponding illustra-
tion prepared using POV-Ray and X-Seed.

formed when dinuclear metallocycles stack in columns. The metallocycles are shown in the capped-stick metaphor (i) to show the
connectivity and relative positions of the atoms and (ii) in order not to obscure the solvent molecules. The acetone solvent mole-
cules are represented in the space-filling metaphor to show how efficiently they occupy the channel (shown as a semitransparent
surface) and also to show the relatively weak van der Waals contacts that they make with one another.
Fig. 19A shows one of Cram’s cryptand molecules (capped stick) with encapsulated nitrobenzene (space filling). Fig. 19B shows
interdigitation of two p-tert-butylcalix[4]arene molecules to form a dimeric arrangement about an inversion center. Each molecule
inserts a tert-butyl group into the cavity of the other molecule. The two molecules are shown in different metaphors in order to take
advantage of the benefits of each; the capped-stick metaphor shows the molecular connectivity and the space-filled metaphor is
better for illustrating the bulk of the molecule.

Figure 12 Capped-stick representation of tryptophan.


Single-Crystal X-ray Diffraction 35

Figure 13 Ball-and-stick representation of tryptophan.

Figure 14 Space-filling representation of tryptophan.

Figure 15 Dimeric association of two L-shaped molecules in (A) ball-and-stick and (B) space-filling representations.

Figure 16 Perspective view along the guest-accessible channels of (A) Dianin’s compound and (B) its thiol derivative. The images show the relative
sizes of the OeH/O and SeH/S hydrogen bonded apertures.
36 Single-Crystal X-ray Diffraction

Figure 17 Thermal ellipsoid plots of (A) p-tert-butylcalix[4]arene37 and (B) a BF4 salt of a metallocyclic complex.38

Figure 18 Acetone guest molecules confined within channels that pass through stacked metallocycles.

Figure 19 (A) Acetonitrile confined within the molecular cavity of a cryptand.39 (B) An interdigitated dimeric arrangement of two p-tert-butylcalix[4]
arene molecules.40
Single-Crystal X-ray Diffraction 37

Figure 20 Bilayer packing of sublimed p-tert-butylcalix[4]arene. The layers repeat along the crystallographic b axis in abcd fashion (colored yellow,
green, blue, and red). The same molecules are shown projected along (A) the crystallographic c axis and (B) an arbitrary direction.37

2.03.5.7.6 Packing Diagrams


Packing diagrams convey a sense of continuity of the structure but such figures require some thought. A packing diagram shown
from an arbitrary direction can be far more confusing that one shown projected along a lattice vector. For example, Fig. 20 shows
two projections of the well-known abcd bilayer packing of p-tert-butylcalix[4]arene. Fig. 20A is a projection along the crystallo-
graphic c axis while Fig. 20B is a projection along an arbitrary direction. The bilayer packing is far more apparent in Fig. 20A.
When the user instructs a structure analysis program to extend the structure within a specified packing range, care should be
taken to check that the resulting packing diagram is complete. Most structure analysis programs merely generate symmetry-
equivalent instances of the asymmetric unit such that at least one of the atoms of the fragment fall within the boundaries of the
packing rage. This is done without considering whether or not complete molecules are generated. For example, one of the known
polymorphs of the hexahost compound hexa-(4-cyanophenoxy)benzene (Fig. 21A) crystallizes in the space group Pccn with half

Figure 21 Packing diagrams of the hexahost hexa-(4-cyanophenoxy)benzene (A) projected along the crystallographic b axis. (B) If care is not taken
the packing diagram can be incomplete at its extremities. Even if the molecules at the extremities are completed the packing diagram can be shown
(C) with or (D) without a border.
38 Single-Crystal X-ray Diffraction

a molecule in the asymmetric unit.41 The molecule is situated on an inversion center and specifying a packing range of [0.5 to 1.5,
0 to 1,  0.5 to 1.5] produces the packing diagram shown in Fig. 21B. The arrows indicate incomplete molecules at the extremities of
the packing diagram and it should be considered good practice to complete these molecules. It is a matter of preference whether or
not to enclose the packing diagram within a rectangular border (Fig. 21C and D). Truncating the diagram with a border conveys
a sense of continuity.

2.03.5.7.7 The Use of Color


CPK42,43 is an acronym for Corey, Pauling, and Koltun and originally refers to commercial plastic van der Waals space-filling models
of molecules with the different elements color coded (e.g., black for carbon, white for hydrogen, red for oxygen, blue for nitrogen,
etc.). Modern molecular graphics programs can produce high-quality images that mimic photographs of CPK models. Although
a particular color convention is not mandated, whenever a molecule is represented in an article using colors to distinguish atoms
types, a key should be included either in the figure or in the caption to indicate the assignment of colors to the different elements.

2.03.5.7.8 Hirshfeld Surface Analysis


Hirshfeld surface analysis is a useful tool for visualizing interactions in molecular crystals. CrystalExplorer27 is a computer package
that utilizes calculated Hirshfeld surfaces of molecules within a crystal structure to determine the intermolecular interactions
between particular molecules or for the crystal structure in its entirety. Hirshfeld surfaces are created by an extension of the weight
function describing an atom in a molecule to include the function of a molecule in a crystal. The isosurface generated from these
calculations, with a specified weight function w(r) ¼ 0.5, surrounds the molecule and by partitioning the electron density of the
molecular fragments, delineates the space occupied by a molecule in a crystal. Hirshfeld surfaces can provide information about
intermolecular interactions in the crystal as the surface is determined by both the enclosed molecule and its closest neighbors. Other
molecular surfaces used to visualize and quantify molecular geometries such as the fused sphere van der Waals (or CPK) and
smoothed Connolly surfaces do not have this advantage as they are designated only by the molecule itself. For quality data
regarding intermolecular interactions to be extracted from Hirshfeld surfaces, the only prerequisite is that the crystal structures
imported into the program are well-characterized with all hydrogen atoms located or placed accurately.

2.03.5.8 Limitations of X-ray Crystallography


The most important limitation of using single-crystal X-ray diffraction analysis is that the compound of interest must form crystals
of sufficient quality to produce a suitable diffraction pattern. The resulting analysis can only provide information about the inter-
actions that occur in the solid state and any speculation about how these features might be relevant to the solution or gas phase
should be treated with great caution. It has already been mentioned that hydrogen atoms scatter X-rays relatively weakly and their
electron densities are usually not spherically distributed about their nuclei. Therefore X-ray diffraction does not yield reliable posi-
tions for hydrogens atoms; if hydrogen atom coordinates are critical to the analysis then it is advisable to use neutron diffraction. In
cases where hydrogen atoms are expected to reside in well-defined positions (e.g., aromatic hydrogen atoms) they can be included
in the crystallographic model using a riding model (i.e., where the hydrogen atom coordinates and thermal parameters are tied to
those of their parent atoms). Use of a riding model is generally not possible for some hydrogen atoms (e.g., hydroxyl groups) but
methods of placing such atoms when they are involved in hydrogen bonding have been suggested.44,45
A solvate is a multicomponent crystal that includes solvent molecules as part of its structure. The solvent molecules reside in
interstitial sites between the major components of the crystal (i.e., the host molecules) and often experience a greater degree of
thermal displacement owing to their smaller size and loose confinement. In many cases solvent molecules are severely disordered
and cannot be modeled completely. In some cases, the solvent molecules are not present at full occupancy. When a crystal incor-
porates volatile content it is generally considered good practice to supplement its structure analysis with thermogravimetry. The
crystallographic model should agree with the stoichiometry deduced from thermogravimetric analysis.
It often occurs that a crystalline material is only available in polycrystalline (powder) form; that is, the individual particles of the
material are considered too small for single-crystal diffraction analysis. In such cases it is generally possible to record a powder dif-
fractogram. A powder diffractogram can also be simulated from single-crystal structural data and then compared to that recorded
experimentally in order to establish whether or not the single crystal and the polycrystalline material have the same structure. This is
particularly useful when a crystal becomes unsuitable for single-crystal diffraction, either owing to a large phase transition or guest
loss. These events usually cause sufficient rearrangement of the components of the crystal such that it does not survive intact. The
material may still retain the original shape of the crystal or it may fracture into separate small pieces. When a crystal becomes opaque
it is almost certainly a sign that it has become a polycrystalline material. However, if its measured powder diffractogram matches
that of a known single-crystal structure then its crystal structure has been established without the need for Rietveld refinement. It
should be noted that single-crystal diffraction analysis is most often carried out under cryogenic conditions while powder diffracto-
grams are usually recorded at room temperature. In by far the majority of cases crystal structures do not change substantially with
temperature. However, one should be aware that some crystals undergo thermally induced polymorphic phase transformations and
that the crystal structure determined at 100 K might not be the same as that at room temperature. In rare cases crystals undergo
anomalously large anisotropic thermal expansion46 and their powder patterns at vastly different temperatures might be significantly
different even in the absence of a phase transition.
Single-Crystal X-ray Diffraction 39

2.03.5.9 Crystallography and Supramolecular Chemistry


Our understanding of molecular structure and intermolecular interactions would still be extremely limited without the detailed
insights gained from X-ray crystallography over the past 70 years. It would be impossible to adequately describe in only a few pages
the impact that crystallography has had on the field of supramolecular chemistry. Therefore this section will mention only a few of
the landmark crystallographic studies of supramolecular systems, and it is only meant to be illustrative of the value of the technique
to the field. There is significant overlap between solid-state supramolecular chemistry and crystal engineering, and no attempt will
be made to distinguish between these two subdisciplines.
In his acclaimed book The Nature of the Chemical Bond47 Linus Pauling states that the first mention of the hydrogen bond was
made by Moore and Winmill.48 Pauling goes on to describe some of the earliest crystallographic verifications of hydrogen bonding,
citing structures involving [HF2] ions.49,50 The earliest studies of supramolecular systems involved investigations of inclusion
compounds. The history of inclusion compounds is traced back to 1811 when Davy observed a chlorine clathrate hydrate,51 the
preparation of which was later reported by Faraday.52 Other significant early preparations of inclusion compounds are summarized
in Table 2. From these reports, it is apparent that the nature of inclusion compounds was not well understood at the time and none
of the authors ventured to propose a structure.
The advent of X-ray structural studies of inclusion compounds is generally considered to be a milestone in the development of
the field. In 1927, Caspari64 determined the unit cell dimensions and space group for a quinol clathrate with methanol, but did not
attempt to determine atomic coordinates. Then, in 1945, Palin and Powell65 reported a preliminary structure of the host framework
of the sulfur dioxide adduct of quinol using X-ray techniques (Fig. 22). They suggested that the SO2 molecules are situated in the
cavities formed as a result of the hydrogen bond linkages between the quinol molecules. Two years later these authors published
details of an X-ray crystal structure determination showing conclusively that the sulfur dioxide molecules were indeed located in the
cavities at nonlinked distances from the quinol molecules.66 The following year Powell67 published a survey of known inclusion

Table 2 Chronology of important early studies of inclusion compounds

Year System studied

1841 Graphite intercalates53


1849 b-Quinol H2S clathrate54
1891 Cyclodextrin inclusion compounds55
1897 Nickel cyanide ammonia inclusion of benzene56
1906 Inclusion compounds of triphenylmethane57
1909 Tri-o-thymotide inclusion of benzene58
1914 Clathrates of Dianin’s compound59
1916 Inclusion compounds of cholic acids60
1935 Phenol clathrates61
1940 Urea inclusion compounds62
1946 Amylose inclusion compounds63

Figure 22 Host framework of the quinol/SO3 inclusion compound. Hexagons denote hydrogen bonds and each of the longer sloping lines connect-
ing different hexagons represents the oxygen–oxygen axis of a quinol molecule, for which the benzene ring has been omitted for clarity. Taken from
Palin, D. E.; Powell, H. M. Nature 1945, 156, 334–335.
40 Single-Crystal X-ray Diffraction

compounds and coined the term “clathrate” to describe compounds where one component is enclosed within the framework of
another.
Since the pioneering work of Powell, the study of inclusion compounds has become a well-established branch of chemistry. This
is mostly due to the increasing accessibility of X-ray crystallographic methods, because an understanding of the nature of inclusion
compounds is largely dependent on knowledge of the spatial arrangements of their components. Over the last 70 years, the number
of published studies of inclusion compounds has increased dramatically.
A growing number of types of inclusion compounds can be distinguished where diversity within a class is achieved by modifi-
cation of the host compound by means of chemical substitution, or by alteration of the guest. Since justice cannot be done to all the
known classes in one article, a few of the more widely studied host–guest systems are described briefly in an attempt to make the
unfamiliar reader aware of the diversity and current trends of the field of inclusion chemistry
Zeolites68 consist of a large group of porous tectosilicates with about 60 different framework topologies that occur naturally, or
have been synthesized. Each different framework is characterized by its unique system of channels and cavities giving rise to guest-
shape-specific properties. This, coupled with their high stability, makes these compounds extremely versatile for industrial
applications.
Intercalation compounds69 are formed by the insertion of mobile atomic or molecular guest species into a solid host layered
structure and are of interest as reversible battery electrodes, electrochromic displays, hydrogen storage devices, shape-selective
heterogeneous catalysts, electro-catalysts, and electronic conductors.
Hofmann inclusion compounds have the general formula M(NH3)2M0 (CN)4$2G (M ¼ Mn, Fe, Co, Ni, Cu, Zn, or Cd; M0 ¼ Ni,
Pd, or Pt; and G is a small aromatic molecule).70 The metal complex hosts can include five- or six-membered aromatic molecules
without bulky substituents. As shown in Fig. 23, the host molecules form planar layers containing the metal ions and the cyano
groups, but with the NH3 moieties protruding above and below the planes. The latter then define a void within which the guest
molecule is accommodated. Size selectivity is a characteristic of this host structure since the crystallographic c dimension is relatively
constant (a notable exception has M ¼ Mn, M0 ¼ Ni, and G ¼ biphenyl).
Werner clathrates71 consist of a wide range of inclusion compounds where the host component is a coordination compound
with the general formula MX2A4 (M ¼ Fe, Co, Ni, Cu, Zn, Cd, Mn, Hg, or Cr; X ¼ NCS, NCO, CN, NO   
3 , NO2 , CI , Br , or

I ; A is a neutral substituted pyridine; and G is a small aromatic molecule). Irregular octahedral coordination about the metal
atom and nonbonded repulsions in the densely populated central region result in a limited number of stable conformations of
the host molecules. Schaeffer72 has shown that these host structures are selective toward the isomers of disubstituted benzenes,
and this property has been utilized in the chromatographic separation of these isomers.73
Urea74 forms inclusion compounds with n-alkanes and n-alkenes, where n > 6 (substitution of the carbon chain is allowed to
a small degree). The urea molecules hydrogen bond to one another forming long helical chains. This produces hollow channels
(shown in cross-section in Fig. 24) in which the guests can be accommodated. Thiourea inclusion compounds are analogous to
those of urea but they have a larger channel diameter, thus allowing the inclusion of highly branched hydrocarbons. However,
n-paraffins fit too loosely in the channels and are not included.
Dianin’s compound59 is one of the most iconic hosts in the field of supramolecular chemistry. In its racemic form it is a good
example of a known clathrate host that has been structurally modified as a design strategy for new multimolecular host systems.
Systematic studies have shown that the cage geometry can be altered markedly by replacement of the heteroatom.75 Substitution
of the ring skeleton,76 modification of the substitution at C(2) and C(4),77 changing the hydrogen bond functionality,78 or combi-
nations of the above. Barrer and Shanson investigated guest-free sublimed Dianin’s compound and reported in 1976 that it behaves
like an organic zeolite with the ability to absorb a number of gases79 (one of the first reports of a porous molecular crystal). More

Figure 23 The structure of Hofmann’s benzene clathrate Ni(NH3)2Ni(CN)4$2C6H6. Large open circle, six-coordinate Ni; crossed large circle, square
planar Ni; solid circle, N; open circle, C; small open circle, H. Host hydrogen atoms are not shown. Taken from Iwamoto, T. In Inclusion Compounds;
Atwood, J. L.; Davies, J. E. D.; MacNicol, D. D., Eds.; Vol. 1, Academic Press: London, 1984.
Single-Crystal X-ray Diffraction 41

Figure 24 Van der Waals cross-section of the cavity in the urea channel compared with the size of n-octane (left), benzene (top), 3-methylheptane
(right), and 2,2,4-trimethylpentane (bottom). Taken from Schlenk, W. Justus. Liebigs Ann. Chem. 1949, 565, 204–240.

Figure 25 Carbon tetrachloride clathrates of (A) Dianin’s compound, (B) the thiol derivative of Dianin’s compound, and (C) a quasiracemate con-
sisting of the R enantiomer of Dianin’s compound and the S enantiomer of its thiol analog. Net polar ordering of the guest occurs only in the quasir-
acemate host. Taken from Jacobs, T.; Bredenkamp, M. W.; Neethling, P. H.; Rohwer, E. G.; Barbour, L. J. Chem. Commun. 2010, 46, 8341–8343.
42 Single-Crystal X-ray Diffraction

recent studies of Dianin’s compound reported salt formation by partial host-to-guest proton transfer80 and that a quasiracemate of
Dianin’s compound and its thiol derivative is capable of guest polar ordering (Fig. 25).81 The achiral compound 4-phenoxyphenol
forms a host structure analogous to that of racemic Dianin’s compound and has been shown to entrap CO2 molecules.82

Many other supramolecular systems have been investigated extensively by means of single-crystal X-ray diffraction. Among
these, the following are relevant examples: calixarenes,83–85 crown ethers,86 cyclodextrins,87 cryptophanes,88 rotaxanes,89 cate-
nanes,90 and cucurbiturils.91 Solid-state supramolecular chemistry has grown into a vast and highly active research area and it would
be impossible to provide a concise survey of the impact that single-crystal X-ray diffraction has had on the field. Some of the more
recent spectacular studies include Fujita’s “sponge” method92 for elucidating the structures of compounds that can only be obtained
in very small quantities, Stoddart’s molecular Borromean rings,93 Cooper’s assembly of porous organic molecular crystals,94 and the
entire vast and rapidly growing field of metal–organic frameworks.95

References

1. Authier, A. Early Days of X-Ray Crystallography; Oxford University Press: Oxford, 2013.
2. Röntgen, W. C. Sitzungsber. Der Wu¨rtzburger Physik -Medic. Gesellsch. 1985, 137, 132–141.
3. Stanton, A. Nature 1896, 53, 274–276.
4. Stark, J. Nature 1908, 77, 320.
5. Friedrich, W.; Knipping, P.; Laue, M. Sitzungsberichte Kgl. Bayer. Akad. Wiss. 1912, 303–322.
6. Bragg, W. H.; Bragg, W. L. Proc. R. Soc. Lond. A 1913, 88 (605), 428–438.
7. Bragg, W. H.; Bragg, W. L. Nature 1913, 91, 557.
8. Hahn, T., Ed. International Tables for Crystallography, Volume A: Space Group Symmetry, 5th edn.; Springer: Dordrecht, 2002.
9. Ewald, P. P. Phys. Zeit. 1914, 15, 399–401.
10. Karle, J.; Hauptman, H. Acta Crystallogr. 1950, 3, 181–187.
11. Thorne, R. E.; Stum, Z.; Kmetko, J.; O’Neill, K.; Gillilan, R. J. Appl. Crystallogr. 2003, 36, 1455–1460.
12. Yufit, D. S.; Howard, J. A. K. J. Appl. Crystallogr. 2005, 38, 583–586.
13. Warren, J. E.; Pritchard, R. G.; Abram, D.; et al. J. Appl. Crystallogr. 2009, 42, 457–460.
14. Cox, J. M.; Walton, I. M.; Benson, C. A.; Chen, Y.-S.; Benedict, J. B. J. Appl. Crystallogr. 2015, 48, 578–581.
15. Merrill, L.; Basset, W. A. Rev. Sci. Instrum. 1974, 45, 290–294.
16. Kaminski, R.; Jarzembska, K. N.; Kutyla, S. E.; Kaminski, M. J. Appl. Crystallogr. 2016, 49, 1383–1387.
17. Brayshaw, S. K.; Knight, J. W.; Raithby, P. R.; et al. J. Appl. Crystallogr. 2010, 43, 337–340.
18. Sheldrick, G. M. Acta Crystallogr. 2008, A64, 112–122.
19. http://shelx.uni-ac.gwdg.de/SHELX/.
20. Farrugia, L. J. J. Appl. Crystallogr. 2012, 45, 849–854.
21. Dolomanov, O. V.; Bourhis, L. J.; Gildea, R. J.; Howard, J. A. K.; Puschmann, H. J. Appl. Crystallogr. 2009, 42, 339–341.
22. Hübschle, C. B.; Sheldrick, G. M.; Dittrich, B. J. Appl. Crystallogr. 2011, 44, 1281–1284.
23. Barbour, L. J. J. Supramol. Chem. 2001, 1, 189–191.
24. http://www.iucr.org/resources/cif/documentation#docs.
25. https://www.ccdc.cam.ac.uk/.
26. http://checkcif.iucr.org/.
27. Spackman, M. A.; Jayatilaka, D. CrystEngComm 2009, 11, 19–32.
28. http://accelrys.com/products/collaborative-science/biovia-materials-studio/.
29. Spek, A. L. J. Appl. Crystallogr. 2003, 36, 7–13.
30. Atwood, J. L.; Barbour, L. J. Cryst. Growth Des. 2003, 3, 3–8.
31. Motherwell, W. D. S.; Clegg, W. PLUTO: Program for Plotting Molecular and Crystal Structures; University of Cambridge: England, 1978.
32. Burnett, M. N.; Johnson, C. K. ORTEP-III: Oak Ridge Thermal Ellipsoid Plot Program for Crystal Structure Illustrations; Oak Ridge National Laboratory Report ORNL-
6895; 1996.
33. http://www.povray.org/.
34. Atwood, J. L.; Barbour, L. J.; Raston, C. L.; Sudria, I. B. N. Angew. Chem. Int. Ed. 1998, 37, 981–983.
35. Barbour, L. J.; Orr, G. W.; Atwood, J. L. Chem. Commun. 1997, 1439–1440.
36. Hu, Z.-Q.; Chen, C.-F. Chem. Commun. 2005, 2445–2447.
37. Atwood, J. L.; Barbour, L. J.; Lloyd, G. O.; Thallapally, P. K. Chem. Commun. 2004, 922–923.
38. Dobrzanska, L.; Lloyd, G. O.; Raubenheimer, H. G.; Barbour, L. J. J. Am. Chem. Soc. 2005, 127, 13134–13135.
39. Yoon, J.; Knobler, C. B.; Maverick, E. F.; Cram, D. J. Chem. Commun. 1997, 1303–1304.
40. Brouwer, E. B.; Udachin, K. A.; Enright, G. D.; et al. Chem. Commun. 2001, 565–566.
41. Das, D.; Barbour, L. J. J. Am. Chem. Soc. 2008, 130, 14032–14033.
42. Corey, R. B.; Pauling, L. Rev. Sci. Instrum. 1953, 24, 621–627.
Single-Crystal X-ray Diffraction 43

43. Koltun, W. L. Space Filling Atomic Units and Connectors for Molecular Models. U.S. Patent 3170246, 1965.
44. Lusi, M.; Barbour, L. J. Cryst. Growth Des. 2011, 11, 5515–5521.
45. Woinska, M.; Jayatilaka, D.; Spackman, M. A.; et al. Acta Crystallogr. A Found. Adv. 2014, 70, 483–498.
46. Das, D.; Jacobs, T.; Barbour, L. J. Nat. Mater. 2010, 9, 36–39.
47. Pauling, L. The Nature of the Chemical Bond; Cornel University Press: Ithaca, 1948.
48. Moore, T. S.; Winmill, T. F. J. Chem. Soc. Trans. 1912, 101, 1635–1676.
49. Helmholz, L.; Rogers, M. T. J. Am. Chem. Soc. 1939, 61, 2590–2592.
50. Pauling, L. Z. Kristallogr. 1933, 85, 380–391.
51. Davey, H. Philos. Trans. R. Soc. Lond. 1811, 101, 1.
52. Faraday, M. Quart. J. Sci. 1823, 15, 71–74.
53. Schafhäutl, C. J. Prakt. Chem. 1841, 21, 129–157.
54. Wöhler, F. Ann. Chem. Liebigs. 1849, 69, 294–300.
55. Villiers, A. C. R. Hebd. Seances Acad. Sci. 1891, 112, 536–538.
56. Hoffmann, K. A.; Küspert, F. Z. Anorg. Allg. Chem. 1897, 15, 204–207.
57. Hartley, H.; Thomas, N. G. J. Chem. Soc. Trans. 1906, 89, 1013–1033.
58. Spallino, R.; Provenzal, G. Gazz. Chim. Ital. 1909, 39, 325–336.
59. Dianin, A. P. J. Soc. Phys. Chem. Russe. 1916, 46, 1310–1319.
60. Wieland, H.; Sorge, H. Z. Physiol. Chem. Hoppe-Seyler’s 1916, 97, 1–27.
61. Terres, E.; Vollmer, W. Z. Petroleum 1935, 31, 1.
62. Bengen, M. F. German Patent Application OZ123438, March 18, 1940.
63. Mikus, F. F.; Hixon, R. M.; Rundle, R. E. J. Am. Chem. Soc. 1946, 68, 1115–1123.
64. Caspari, W. A. J. Chem. Soc. 1927, 1093–1095.
65. Palin, D. E.; Powell, H. M. Nature 1945, 156, 334–335.
66. Palin, D. E.; Powell, H. M. J. Chem. Soc. 1947, 208–221.
67. Powell, H. M. J. Chem. Soc. 1948, 61–73.
68. Barrer, R. M. In Inclusion Compounds, vol. 1, Atwood, J. L., Davies, J. E. D., MacNicol, D. D., Eds.; Academic Press: London, 1984.
69. Schöllhorn, R. In Inclusion Compounds, vol. 1, Atwood, J. L., Davies, J. E. D., MacNicol, D. D., Eds.; Academic Press: London, 1984.
70. Iwamoto, T. In Inclusion Compounds, vol. 1, Atwood, J. L., Davies, J. E. D., MacNicol, D. D., Eds.; Academic Press: London, 1984.
71. Lipkowski, J. In Inclusion Compounds, vol. 1; Academic Press: London, 1984.
72. Schaeffer, W. D.; Dorsey, W. S.; Skinner, D. A.; Christian, C. G. J. Am. Chem. Soc. 1957, 79, 5870–5876.
73. Lipkowski, J.; Pawlowska, M.; Sybilska, D. J. Chromatogr. A 1979, 176, 43–53.
74. Takemoto, K.; Sonoda, N. In Inclusion Compounds, vol. 2, Atwood, J. L., Davies, J. E. D., MacNicol, D. D., Eds.; Academic Press: London, 1984.
75. MacNicol, D. D.; Mills, H. H.; Wilson, F. B. J. Chem. Soc. D 1969, 1332–1333.
76. Hardy, A. D. U.; McKendrick, J. J.; MacNicol, D. D. J. Chem. Soc. Chem. Commun. 1974, 972–973.
77. Hardy, A. D. U.; McKendrick, J. J.; MacNicol, D. D. J. Chem. Soc. Chem. Commun. 1976, 355–356.
78. Hardy, A. D. U.; McKendrick, J. J.; MacNicol, D. D.; Wilson, D. R. J. Chem. Soc. Perkin Trans. 2 1979, 729–734.
79. Barrer, R. M.; Shanson, V. H. J. Chem. Soc. Chem. Commun. 1976, 333–334.
80. Lloyd, G. O.; Bredenkamp, M. W.; Barbour, L. J. Chem. Commun. 2005, 4053–4055.
81. Jacobs, T.; Bredenkamp, M. W.; Neethling, P. H.; Rohwer, E. G.; Barbour, L. J. Chem. Commun. 2010, 46, 8341–8343.
82. Jacobs, T.; Smith, V. J.; Thomas, L. H.; Barbour, L. J. Chem. Commun. 2014, 50, 85–87.
83. Andreeti, G. D.; Ungaro, R.; Pochini, A. J. Chem. Soc. Chem. Commun. 1979, 1005–1007.
84. Atwood, J. L.; Barbour, L. J.; Jerga, A. Science 2002, 296, 2367–2369.
85. Atwood, J. L.; Barbour, L. J.; Jerga, A.; Schottel, B. L. Science 2002, 298, 1000–1002.
86. Hu, J.; Barbour, L. J.; Gokel, G. W. J. Am. Chem. Soc. 2001, 123, 9486–9487.
87. Caira, M. R.; de Vries, E. J. C.; Nassimbeni, L. R. Chem. Commun. 2003, 2058–2059.
88. Joseph, A. I.; El-Ayle, G.; Boutin, C.; Léonce, E.; Berthault, P.; Holman, K. T. Chem. Commun. 2014, 50, 15905–15908.
89. Miljanic, O. S.; Dichtel, W. R.; Khan, S. I.; et al. J. Am. Chem. Soc. 2007, 129, 8236–8246.
90. Frasconi, M.; Kikuchi, T.; Cao, D.; et al. J. Am. Chem. Soc. 2014, 136, 11011–11026.
91. Park, K.-M.; Whang, D.; Lee, E.; Heo, J.; Kim, K. Chem. Eur. J. 2002, 8, 498–508.
92. Inokuma, Y.; Yoshioka, S.; Ariyoshi, J.; et al. Nature 2013, 495, 461–466.
93. Chichak, K. S.; Cantrill, S. J.; Pease, A. R.; et al. Science 2004, 304, 1308–1312.
94. Jones, J. T. A.; Hasell, T.; Wu, X.; et al. Nature 2011, 474, 367–371.
95. Yaghi, O. M.; O’Keeffe, M.; Ockwig, N. W.; et al. Nature 2003, 423, 705–714.
2.04 X-ray Powder Diffraction
C Tedesco, University of Salerno, Fisciano, Italy
M Brunelli, Swiss Norwegian Beam Lines at the European Synchrotron Radiation Facility, Grenoble, France
Ó 2017 Elsevier Ltd. All rights reserved.

2.04.1 Overview 46
2.04.2 XRPD: An Historical Perspective 46
2.04.3 Structure Analysis From XRPD Data 47
2.04.3.1 Data Acquisition 47
2.04.3.2 Indexing 51
2.04.3.3 Space Group Determination 52
2.04.3.4 Line Fitting and Whole Powder Pattern Fitting 52
2.04.3.5 Structure Solution 53
2.04.3.5.1 Reciprocal Space Methods 53
2.04.3.5.2 Direct Space Methods 53
2.04.3.5.3 Dual Space Methods 54
2.04.3.5.4 Structure Refinement: The Rietveld Method 55
2.04.3.6 Available Software for XRPD 56
2.04.4 Pair Distribution Function 57
2.04.5 XRPD Uses 57
2.04.5.1 Phase Identification 57
2.04.5.2 Lattice Parameters Determination 58
2.04.5.2.1 Kinetic Studies 58
2.04.5.3 Crystal Structure Determination 58
2.04.5.4 Crystal Structure Refinement and Quantitative Phase Analysis 60
2.04.6 Examples of Application of XRPD in Supramolecular Chemistry 60
2.04.6.1 Structure Determination of MOFs 60
2.04.6.1.1 Structure Determination and Characterization of an MOF from Laboratory Data 61
2.04.6.1.2 Structure Determination of MOFs by HR-XRPD 61
2.04.6.2 Structure Determination of Covalent Organic Frameworks 61
2.04.6.3 Structure Determination of Fullerene-Based Compounds 62
2.04.6.4 Adsorption Studies 63
2.04.6.4.1 Adsorption Studies in MOFs 64
2.04.6.4.2 In Situ Adsorption Studies of Flexible MOFs 64
2.04.6.4.3 Gas Adsorption Studies in a Permanently Porous Molecular Organic Compound 64
2.04.6.5 Polymorphism and Solvatomorphism 65
2.04.6.5.1 Polymorphism and Solvatomorphism in a Pharmaceutical Compound 65
2.04.6.5.2 Solvatomorphism in a Calixarene-Based Compound 66
2.04.6.6 Complementarity With ss-NMR 66
2.04.6.7 Complementarity With IR and Raman Spectroscopy 68
2.04.6.8 Complementarity With DFT Calculations 69
Acknowledgments 70
References 70
Relevant Websites 73

Nomenclature
3Br-py 3-Bromo-pyridine MOFs Metal-organic frameworks
camph Camphorate NMR Nuclear magnetic resonance
DFT Density functional theory PDF Pair distribution function
ESRF European synchrotron radiation facility PSI Paul Scherrer Institute
HR-XRPD High-resolution X-ray powder diffraction py Pyridine
IP Image plate pyz Pyrazine
IR Infrared SC-XRD Single crystal X-ray diffraction

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12489-8 45


46 X-ray Powder Diffraction

SNBL Swiss Norwegian beamlines XRD X-ray diffraction


ss-NMR Solid-state nuclear magnetic resonance XRPD X-ray powder diffraction
TGA Thermogravimetric analysis

2.04.1 Overview

A century has passed since the first successful X-ray powder diffraction (XRPD) experiments by Debye and Scherrer in Göttingen
(Germany) in 1916 and Hull in Schenectady (United States) in 1917.1–4
With time XRPD revealed its power as a nondestructive analytical technique in various field as mineralogy, metallurgy, materials
science, forensic science, archeometry, biology, pharmaceutical industry, and, of course, supramolecular chemistry.
This article will focus on the most recent applications of XRPD in the field of supramolecular chemistry with the aim to outline
the advantages of the technique and its complementarity with other analytical tools.
A brief introduction on the salient aspects of the method will be followed by examples of its applications in the structure deter-
mination of metal-organic frameworks (MOFs), fullerenes, inclusion compounds, but also in the study of phase transitions, poly-
morphism, and solvatomorphism. The list of examples is by no means exhaustive, but it would serve as a base to discuss in details
how XRPD could be used to solve problems in supramolecular chemistry.
A basic knowledge of X-ray diffraction (XRD) is assumed, as the article is designed as a practical guide, theoretical aspects are just
outlined and references are made to authoritative contributions (books, reviews, and articles) for each subject.
Several introductory crystallography books describe the principle of XRPD5–7 and the excellent book by Dinnebier and Billinge
provides an exhaustive description of the most recent advances in the field. 8

2.04.2 XRPD: An Historical Perspective

As Hull detailed in his biography, he was prompted to use XRPD by discussing with W. L. Bragg about the lack of the atomic struc-
ture of elemental iron. 9 He knew that it was difficult to grow single crystals of suitable size and quality for that element and he
decided to try with a powdered sample.
Thus, from its origins the method configured as an alternative to single-crystal XRD in all that cases where it is difficult to obtain
single crystals.
Moreover, the XRPD data correspond to the bulk material and not to a single representative of the whole sample as in single-
crystal XRD. This makes XRPD the preferred techniques for the analysis of multiphase samples, as already pointed out by Hull early
in 1919, since XRPD patterns act as a fingerprint for a crystalline material. 10
Later on in the 1930s of the last century, following Hull considerations, Hanawalt proposed a phase identification method based
on the three strongest intensity lines in the pattern. 11,12 Nowadays search/match algorithms are implemented in X-ray analysis
software.
Also the presence of amorphous phases can be detected, it is possible to estimate the crystallinity of the sample and characterize
the amorphous content. 13
Short-range order in semi-crystalline or noncrystalline materials can be evidenced by extracting the pair distribution function
(PDF) from the experimental XRPD pattern. 14
The type of structural information derived from the diffraction pattern of a polycrystalline sample is summarized in Table 1. In
a certain sense the information content of an XRPD pattern is greater than single-crystal XRD data, the main problem is to extract
this structural information.
Single-crystal XRD data are distributed and spatially recorded in three dimensions, while powder diffraction data are projected
invariably in one dimension. Thus, the reflections do overlap systematically or randomly and a single diffraction peak may corre-
spond to multiple reflections. This makes difficult to assign the intensity to a certain reflection and this is particularly important for
molecular compounds, for which the crystal symmetry is usually low and the number of reflections with similar 2q angles increases.
All steps of structural analysis, from indexing to structure determination and refinement, are compromised by reflections overlap.
The structural refinement method, introduced by Rietveld in 1969, 15,16 overcomes this difficulty by considering as data directly
the intensities observed at each data point rather than the integrated intensities. Rietveld idea is at the base of the modern whole
powder pattern fitting methods.
In particular, Pawley in 1981 17 and Le Bail in 198818 devised two procedures for the extraction of integrated peak intensities by
means of whole powder pattern decomposition.
The advent of synchrotron radiation sources in the 1970s of the last century determined a drastic improvement in the quality of
the diffraction data from powder samples. Both spectral resolution (line separation) and peak-to-background ratio improved
dramatically, allowing the detection of very weak peaks at high 2q angles. Nowadays an instrumental resolution of
X-ray Powder Diffraction 47

Table 1 The type of structural information derived from the diffraction pattern of a polycrystalline
sample

Diffraction data Structural information

Peak position (2q) Lattice constants (indexing)


Intensities Crystal symmetry (space group determination)
Atom coordinates
Thermal factors
Occupancies (disorder)
Preferred orientation (texture analysis)
Peak width and shape Crystallite size
Microstrain
Scale factor in multiphase patterns Phase percentage weight fraction (quantitative phase
analysis)
Background modulation Short-distance order (local structure)
Amorphous content

Dd/d z 1.5  10–4 at 10 keV can be achieved at high-resolution XRPD beamlines, such as ID22 at the European Synchrotron Radi-
ation Facility (ESRF) (Grenoble, France).
Both advances in experimental setup and computing methods are responsible for the continuously increasing number of struc-
tures determined ab initio from powder X-ray data in the last decades.
The term ab initio is used to indicate that the structural analysis is based exclusively on XRPD data.
Frequently additional information from other experimental techniques, such as solid-state nuclear magnetic resonance (ss-
NMR), infrared (IR) spectroscopy, or electron microscopy, are necessary to determine the crystal structure in a conclusive way.
Combined use of XRPD data and solid-state density functional theory (DFT) calculations is also applied. Thus, a multitechniques
approach is often the key to the success of the structural analysis and the interplay between complementary approaches will be more
and more exploited in the future.

2.04.3 Structure Analysis From XRPD Data

The structure analysis from XRPD data consists in the following steps:
l Data acquisition
l Indexing
l Space group determination
l Whole powder pattern fitting
l Structure determination
l Structure refinement (Rietveld method)
A brief description of each step follows together with the main problems that could be encountered during each step. The para-
graph is written to assist as a workflow for the structural analysis by means of XRPD.

2.04.3.1 Data Acquisition


The choice of the experimental setup depends on the purposes of the structural analysis. The use of synchrotron radiation instead of
X-ray tubes can be justified in case of weakly diffracting samples for structure determination purposes or when subtle phase changes
must be detected.
In transmission (or Debye–Scherrer) geometry a capillary tube is filled with a fine powder, and in reflection (or Bragg–Brentano)
geometry a flat plate holds the powder sample. Transmission geometry is preferred for organic compounds, since the absorption
coefficient of the sample is low and the X-ray beam is effectively transmitted through the sample. Reflection geometry is indicated
for compounds that have a high absorption coefficient (i.e., are constituted by heavy atoms) in this way the X-ray beam does not
penetrate into the sample. The use of high energy synchrotron radiation with a high penetration power allows to use transmission
geometry also for these latter samples (although data should always be corrected to account for absorption effects).
Transmission geometry may be preferred also because it requires only very small sample amounts, although, if “zero-
background” flat plates (i.e., silicon crystal sample holders) are available, good results may be obtained also in reflection geometry
by spreading a thin layer of compound on the flat plate.
Most laboratory diffractometers display a transmission parafocusing geometry, also known as Bragg–Brentano geometry
( Fig. 1). The divergent beam coming out of the X-ray tube passes through Soller slits to control the size, then it is reflected by
the sample and refocused at the detector. The X-ray source and the detector belong to a circle, whose center is at the sample. Usually
48 X-ray Powder Diffraction

Figure 1 Schematic drawing of Bragg–Brentano geometry. The dotted circle is centered on the sample and represents the goniometer circle. The
diffracted beam is refocused at the detector. Source: Cockcroft, J.K.; Fitch, A.N. In Powder Diffraction: Theory and Practice; Dinnebier, R.E., Billinge,
S.J.L., Eds.; RSC Publishing: Cambridge, 2008.

the X-ray tube is fixed, while the sample moves by an angle u(¼ q) and the detector by 2q, alternatively the sample is fixed and the
X-ray tube and the detector move both by q. The detector may be a point scintillation detector or a linear position sensitive detector.
Bragg–Brentano diffractometers offer high-resolution and high beam intensity at the cost of very precise alignment requirements
and carefully prepared samples. A misplaced or a partially transparent sample, or a rough sample can lead to specimen displacement
errors.
Alternatively the Debye–Scherrer method can be used, the incident monochromatic X-ray beam passes through the sample and
the diffracted beam is recorded by the detector ( Fig. 2). Originally the detector was a strip of a photographic film positioned in
a cylindrical camera coaxial with the capillary. Nowadays image plate (IP) or curved position sensitive detectors are used, they cover
a rather extended 2q range, usually up to 120–140 degrees. In this way a whole powder pattern may be collected in a few seconds.
During the measurement the sample is rotated around the capillary axis to ensure a random distribution of the crystallographic
orientation of crystallites. In general, preferred orientation effects are reduced by using capillaries in a Debye–Scherrer geometry
(with the only notable exceptions of needle-like crystallites).
The main drawback of the method is the higher peak-to-background ratio due to the absence of any filter or monochromator on
the diffracted beam (if point detector are not used).
Moreover, with a divergent beam (as it is the case with most laboratory diffractometers) the peak resolution is lower with respect
to parafocusing Bragg–Brentano geometry. A smaller capillary diameter may improve the peak resolution, but will also determine
higher counting times. Moreover, if the peak resolution is sample dependent (i.e., the sample is poorly crystalline), the use of
smaller capillaries is only detrimental.

Figure 2 Schematic drawing of parafocusing a Debye–Scherrer diffractometer equipped with a curved monochromator crystal and a linear or
curved position sensitive detector. The sample is a glass capillary. Source: Cockcroft, J.K.; Fitch, A.N. In Powder Diffraction: Theory and Practice; Din-
nebier, R.E., Billinge, S.J.L., Eds.; RSC Publishing: Cambridge, 2008.
X-ray Powder Diffraction 49

Parallel beam geometry is exploited at synchrotron radiation sources, avoiding errors associated with sample displacement or
transparency or rough surfaces for flat plate samples or with imperfect focusing of the diffracted beam in transmission geometry.
Because peak resolution does not depend on the capillary diameter, large capillaries can be used to optimize the diffracted intensity.
The highest 2q resolution is achieved by means of an analyzer crystal between the sample and the point detector. The analyzer
crystal is usually a perfect Si or Ge 111 crystal, whose orientation with respect to the incident beam defines rigorously the 2q scat-
tering angle.
Multistage analyzer crystals plus detectors are used to overcome the decreased flux of X-ray photons received by the detector,
which would imply otherwise a longer acquisition time to achieve the desired peak-to-background ratio.
High-resolution X-ray powder diffraction (HR-XRPD) data are often required for structure solution and refinements, in partic-
ular for indexing the diffraction patterns of new materials with reasonable accuracy and confidence.
High-resolution instruments are available at most synchrotron radiation sources: ID22 at ERSF, BM31 at SNBL (ESRF), I11 at
Diamond, 11BM at Advanced Photon Source, and MS at Swiss Light Source.
Fig. 3 shows two XRPD patterns of a calixarene sample collected in transmission geometry at ID31 (now ID22) HR-XRPD beam-
line at ESRF and using a Bruker D8 diffractometer equipped with Ni-filtered CuKa radiation and a linear position sensitive detector.
The pattern collected at HR-XRPD beamline evidenced the presence of a second phase in the sample.
A parallel beam geometry may be realized also using X-ray tubes by means of Göbel mirrors. They are multilayered crystals,
whose stack is carefully controlled at the nm size during manufacture. Göbel mirrors are parabolically bent, so that a divergent
beam striking the mirror at different locations and angles yields a diffracted parallel beam. The intensity gain with respect to par-
afocusing geometry is useful for capillary measurements with pharmaceutical or biological compounds.
Radiation damage may be a major inconvenience using synchrotron radiation and data collection strategies must take it into
account, usually multiple scans are collected and the pattern checked for signs of radiation damage (as intensity decay, peak split-
ting, or peak shift to low theta values).
In this cases it is useful to expose a fresh part of the sample to the incoming beam once the signs of radiation damage appear.
The preparation of the sample is a very important issue, a grainy powder would give nonhomogeneous rings and unreliable
intensity values would be collected using a point detector, whereas an area detector would point out the problem immediately.
Preferred orientation of the crystallites while loading the sample in a flat plate sample holder can be avoided loading the sample
in a capillary tube and using transmission geometry.
For in situ studies at nonambient conditions the sample may be placed into furnaces or diamond anvil cells for high-pressure
studies or gas cells for adsorption/desorption experiments. All additional equipments will require a careful experimental setup
and will also influence the choice of the X-ray source and detector type.
The choice of the detector is dictated by the compromise between angular resolution and acquisition speed. Point detectors
equipped with analyzer crystals in parallel geometry are the best choice to achieve the best angular resolution, although they
may not be useful in case of in situ experiments, since a sample transformation may occur during a theta scan or involve a change
in the crystallites size.

Figure 3 XRPD patterns for a calixarene sample collected in transmission geometry (A) at ID31 (now ID22) HR-XRPD beamline at ESRF and
(B) using a Bruker D8 diffractometer with Ni-filtered CuKa radiation and a linear position sensitive detector. The arrow indicates the main peak of an
additional phase in the sample. Source: Tedesco, C. unpublished results.
50 X-ray Powder Diffraction

Curved position sensitive detectors (as the Mythen detector developed at Paul Scherrer Institute, PSI) and area detectors (as the
Frelon camera developed at ESRF or the Pilatus detector developed at PSI, etc.) allow to collect the whole powder pattern in a short
time with acceptable angular resolution.
Fast area detectors have an acquisition time down to a few tenth of milliseconds. They are the best choice for in situ experiments
with variable environmental conditions such as temperature, pressure, chemical potential, where fast solid-state reactions occur.
Vapor sorption XRPD experiment was performed at the Italian beamline BM08 at ESRF, which was equipped with a large IP
detector suitable for rapidly collecting high statistics XRPD patterns. 19 The experimental setup is shown in Fig. 4A. The slits allowed
to record only a narrow portion of the Debye rings. The IP translated behind the slits at a constant speed, so that the diffraction
patterns were recorded continuously as a function of time (Fig. 4B).
This experimental setup was used for studying the acetonitrile adsorption in a calixarene-based material by means of in situ
XRPD measurements. The crystalline powder in a capillary gas cell was exposed to acetonitrile vapors carried by a nitrogen flow,
with time the vapors condensed inside the capillary, dissolved the crystalline powder and formed single crystals of a new crystal
form.
The collected diffraction images clearly showed that the crystalline powder was progressively transforming into single crystals
( Fig. 5). It was then possible to remove the crystals from the capillary and analyze them by single-crystal XRD.20
Apart from the fundamental book by Klug and Alexander, 6 detailed accounts on more recent advances in the instrumentation
are provided by Cockcroft and Fitch.21 A detailed description on XRPD setups at nonambient conditions is given by Norby and
Schwarz.22

Figure 4 (A) The translating IP setup for in situ XRPD at the GILDA beamline at ESRF. S, sample; BS, beam stopper; G, goniometer head; S2, anti-
scatter slit; SH, rotating shutter; A, absorber; I0, ionization chamber; S1, S2, slits. (B) XRPD diffraction patterns recorded with the translating IP during
exposure to CHCl3 vapors (slit aperture 1 mm). Intensities change with time as a consequence of vapor adsorption. Source: (A) Meneghini, C.; Artioli,
G.; Balerna, A.; Gualtieri, A.F.; Norby, P.; Mobilio, S. J. Synchrotron Radiat. 2001, 8, 1162–1166;(B) Tedesco, C. unpublished results.
X-ray Powder Diffraction 51

Figure 5 (A) Diffraction image of a calixarene powder recorded at BM08-GILDA (ESRF) in transmission geometry using a Fuji Image Plate as
detector. (B) Diffraction image of the same powder after exposing the capillary tube to CH3CN vapors for 1 day. (C) Oscillation frame image of
a single-crystal formed inside the capillary tube (SNBL, ESRF). Source: Tedesco, C.; Erra, L.; Immediata, I.; Gaeta, C.; Brunelli, M.; Merlini, M.; Mene-
ghini, C.; Pattison, P.; Neri, P. Cryst. Growth Des. 2010, 10, 1527–1533.

2.04.3.2 Indexing
The peak positions in an XRPD pattern depend on the lattice parameters a, b, c, a, b, g, (or a*, b*, c*, a*, b*, g* in the reciprocal
space):
 
2 sin q 2 1
¼ 2 ¼ h2 a2 þ k2 b2 þ l2 c2 þ 2hka b cos g þ 2hla c cos b þ 2klb c cos a
l d
Indexing means to determine the correct lattice parameters by using the information on the peak positions in the experimental
pattern. This corresponds also to attribute to each peak three integer numbers h, k, l, that correspond to the Miller indices of the
crystallographic plane that generated the Bragg reflection.
Several procedures were devised for the automatic indexing of the peaks:
– Automatic exhaustive indexing is performed by the program TREOR. 23,24 It assumes first a cubic cell and the cell parameter is
increased until a threshold value is reached or a solution is found. Then tetragonal and hexagonal systems with two lattice
parameters are considered, and then orthorhombic (three lattice parameters), monoclinic (four lattice parameters), and triclinic
(six parameters) systems follow. The program KHOL25 offers a specific version of the method for searching monoclinic
systems.26
TAUP 27 is another trial-and-error method based on permutation of Miller index.
– Ito’s method tries to find crystallographic zones (i.e., origin containing reciprocal planes) assuming initially a triclinic system.
The intersection between two zones is a reciprocal axis and the angle between two zones is a reciprocal cell angle. In this way the
reciprocal lattice constants can be completely determined. Low symmetry patterns are efficiently indexed by this method
proposed by Ito 28 and extended by Visser.29
– The program DICVOL applies finite increments to the lengths of the cell parameters by means of a dichotomy method. 30,31 The
initial increments of the cell parameters are halved when a possible solution is found and the procedure iterated. The search is
applied considering first a cubic cell (one cell parameter) and then the symmetry is lowered until all six cell parameters are
varied. The procedure results to be rather sensitive to the quality of the data.
– Singular value decomposition method is iteratively applied in the indexing procedure proposed by Coelho. 32 It considers all the
observed peaks and not only a few of these (usually 20 peaks are considered in the already mentioned methods) and it results to
be less sensitive to zero-shift errors, to the presence of impurities, and to the lack of high d-spacing values.
– Monte Carlo and grid search methods are used by the program McMaille. 33 The peak positions are compared with those
calculated by a Monte Carlo or by an exhaustive grid search procedures.
– Whole-pattern fitting and a genetic algorithm are used in the indexing procedure described by Kariuki et al. 34 Whole pattern
fitting allows to avoid problems with peak overlap and zero-shift errors. The genetic algorithm is used as a global optimization
procedure to search the parameters space. The pattern R-factor, Rwp, obtained by a Le Bail fit,18 is used to discriminate among the
possible solutions.
Indexing programs generally provide a list of possible solutions, which are ranked on the base of a figure of merit, to assess the
quality of the combination of lattice parameters.
Usually it is advisable to use several indexing procedures and compare the obtained results, the correct cell is often among the
first results in the list.
The choice of the correct cell is aided by considering the experimental density (if available) or the calculated density (using the
cell volume obtained by the indexing procedure) and comparing it with those obtained for similar compounds.
It is also important to exclude the presence of impurities, which could determine spurious peaks in the XRPD pattern and lead to
the failure of the indexing procedure owing to a high number of unindexed peaks. On the other hand, if the presence and the iden-
tity of secondary phases are known, indexing the main unknown phase becomes feasible.
52 X-ray Powder Diffraction

Successful indexing of a mixture of two unknown phases is extremely uncommon, and feasible only if high-resolution powder
diffraction data are available. 35,36

2.04.3.3 Space Group Determination


The symmetry elements of a crystal are described by the space group, which is the group of the three-dimensional symmetry
elements of the crystal. Space group determination is based on the evaluation of the systematic absences in the diffracted intensities.
Unfortunately, due to peak overlap the estimation of the integrated intensities cannot be accurate. Thus, systematic absences are not
easily or reliably detected from XRPD data.
Programs as DASH 37 or EXPO38 are able to suggest possible extinction groups among which choose the correct space group.
Frequently the correct space group can be estimated only at the final stage of the Rietveld refinement by comparing the disagreement
indices Rp and Rwp.39

2.04.3.4 Line Fitting and Whole Powder Pattern Fitting


XRPD patterns arise by the convolution of instrumental and sample factors. Instrumental contributions include both effects related
to the geometry of the diffractometer (axial divergence, sample displacement, etc.) and to the shape of the X-ray emission spectrum
(the X-ray-source size, the angle of divergence of the incident beam, etc.). Sample contributions include particle size distribution,
nonhomogeneous strain, and textures.
The idea of pattern fitting is at the basis of the Rietveld method: in its beginning the method was applied to low resolution,
constant wavelength, neutron powder diffraction data, whose line shapes are well described as simple Gaussian functions consid-
ering the instrumental and sample contributions to the pattern as a whole.
With the application of the Rietveld method to XRPD data, it became evident that simple Gaussian functions would be inade-
quate and other functions were devised to achieve an empirical description of the line shape, as pseudo-Voigt (a linear combination
of Gaussian and Lorentzian functions) or Pearson VII functions.
A different approach to the problem of pattern fitting is proposed and applied by taking into account every contribution (i.e.,
instrument geometry, radiation source, sample size, etc.) to the pattern by considering instead fundamental parameters, that is,
sample and slit lengths, receiving slit and source widths, horizontal divergence, primary and secondary Soller slits angles, etc.
Indeed the use of a phenomenological description of the diffraction pattern is standard practice in Rietveld refinement, but
a fundamental parameters approach would probably become more and more used as the research in the field of line profile analysis
advances. 40,41
Fitting of single Bragg peaks is useful for several different purposes, that is, determination of peak positions for indexing
purposes, stress characterization, and determination of microstructural parameters.
Whole powder pattern fitting can be applied with or without a structural model, in the former case this leads to the Rietveld
method for structure refinement, 15,16 in the latter case this leads to whole pattern decomposition methods, as Pawley17 and Le
Bail18 methods.
Whole pattern decomposition methods maybe considered as variants of the Rietveld method in the absence of a structural
model.
Pawley, in 1981, showed that it was possible to determine the integrated intensities by applying a least-squares analysis of the
XRPD pattern, as in the case of Rietveld refinement, considering as variables the peak areas themselves. 17
His approach involves the least-squares fitting of the diffraction pattern by adjusting the background parameters, cell param-
eters, peak shape parameters, and integrated intensities. The method assumes that all the reflections have independent variable
intensities. Therefore, during a least-squares refinement, although the total area of a group of overlapping peaks can be well deter-
mined, the individual intensities can assume any value (including negative ones), which could lead to completely unrealistic
results. This problem was mitigated by constraining the individual intensities to be close to the “mean value” calculated for over-
lapping peaks.
Subsequently Le Bail, in 1988, 18 proposed to assign arbitrarily equal values to the integrated intensities of overlapping reflec-
tions, which are no longer treated as least-squares variables and are never refined. In this way each least-squares cycle is very fast
since the matrix remains small.
Initially the intensity values are treated as calculated values as if they had been derived from a structural model, then it is applied
the Rietveld procedure for partitioning the observed peak intensity among the hkl reflection that contribute to it.
The least-squares procedure will minimize the difference between the observed and calculated y(i) intensities and will provide
better “calculated” integrated intensities that will be used in the subsequent cycles until convergence is reached.
It must be noted that in case of severe overlapping (as it occurs systematically in high symmetry systems or as it may occur at high
angular values), both methods will tend to give erroneous values and care must be taken in using such data for structure solution
purposes.
Noteworthy, the final values of the pattern R factors in a Pawley 17 or Le Bail18 refinement, Rp or Rwp, will represent a lower limit
for the pattern R factors in subsequent refinement procedures.
Whole pattern decomposition methods maybe used to refine the cell parameters and to help in space group determination, but
mainly is used to extract the integrated intensities for structure solution purposes.
X-ray Powder Diffraction 53

Also in cases where the extraction of the integrated intensities is not needed for structure solution, whole pattern decomposition
methods are useful to establish the initial values of the profile parameters to be used.

2.04.3.5 Structure Solution


Structure solution methods aim to determine the atomic coordinates of a compound of interest by estimating the phases of the
observed structure factors. According to the working space, they can be divided into:
– Reciprocal space methods, such as Patterson methods, 42,43 Direct methods,44 and the maximum entropy methods.45 They are
able to extract the phase information embedded in the observed intensities, that is, the structure factors moduli. Data
completeness and high resolution (greater than 1.2 Å) are crucial for the success of the methods.46
– Direct space methods, such as Simulated Annealing 47 or genetic algorithm.48 They work on direct (real) space data, by exploring
the coordinate space in search of a solution that provides the best agreement with the observed intensities. These methods
require a previous knowledge of the structure connectivity.
– Dual space methods, such as charge flipping algorithm 49 and vive la difference (VLD) approach.50,51 They alternate iteratively
modifications of the current model in direct space (by modifying the electron density, i.e., the atomic coordinates) and reciprocal
space (by modifying the structure factors).
In principle the same methods that are used for single-crystal XRD data can be used also for XRPD data, but the intrinsically low
quality of the experimental data requires the application of specific algorithms. In particular, structure solution methods when
applied to XRPD have to cope with the rather low data-to-parameter ratio and with the general low reliability of the intensity values
due to a high degree of peak overlap.
An authoritative and detailed account of the methods for the structure determination from XRPD data is in the book by David
et al. 52 More recent overviews are provided by Caliandro et al., 53 and Harris.54 Several reviews are also available.47,55–58
In the following we will describe briefly the principles of the methods and the strategies devised to improve the chances of
a successful structure solution using XRPD data.

2.04.3.5.1 Reciprocal Space Methods


Reciprocal space methods use the intensities of individual reflections, which are extracted from the powder diffraction pattern.
If very high-quality data are collected from a sharply diffracting sample, accurate structure factors may be obtained from the inte-
grated intensities to near-atomic resolution. Indeed, in 1991 Cernik et al. 59 showed that it was possible to solve the crystal structure
of cimetidine from HR-XRPD data at 1.273 Å by Direct methods using the program SIR88,60 which was developed for SC-XRD.
As a matter of fact enhancing the quality of the extracted integrated intensities would improve the phasing process. In particular,
in case of overlapped reflections the results of the extraction procedure can be improved, if the starting values of the intensities are
close to the true values. Giacovazzo and his group exploited the use of additional information, gained during the phasing step, back
into the pattern decomposition step. As additional information they considered the presence of pseudo-translational symmetry, 61
the expected positivity of the Patterson function both in reciprocal62 and direct space,63 a well-located molecular fragment.64
The first step of Direct methods’ procedures is the calculation of normalized structure factors E
 
CFo;hkl 
Ehkl ¼ qffiffiffiffiffiffiffiffiffiffiffi
P 2ffi
tm j fj

where C is the scale factor and tm is the overall atomic displacement (temperature factor) as determined by a Wilson plot. 65,66
In the case of XRPD data the quality of the Wilson plot is often poor, giving usually a meaningless negative temperature factor.
The use of properly estimated overlapped reflections in the normalization process allows to correct this systematic error. 67
According to Sheldrick’s rule, 46 for Direct methods to succeed at least half the number of possible reflections between 1.1 and
1.2 Å resolution should have F > 4s(F).
Bricogne explained the rule on a structural basis by showing that the structure factors in that range contain fundamental stereo-
chemical information. 68 He concluded that a “stereochemically aware structure-factor statistics and likelihood functions could
significantly relax the data-resolution requirements.”
Indeed most devised strategies aim to include extra information in the structure solution process, as for example, a fragment of
known geometry. The phase information included in the fragment information is incorporated in the phasing process. 67,69
Methods based on the Patterson function are less demanding in terms of data-to-parameter ratio. 70 They are particularly useful
in the presence of heavy scatterers, because Patterson peaks correspond to interatomic vectors among heavy atoms and are readily
located as maxima in the Patterson map, either by visual inspection or by automated Patterson superposition techniques. Patterson
algorithms allow also to determine the rotational and translational parameters of a known structural fragment by comparison of the
observed Patterson map and the one calculated by including the known fragment in the unit cell.71–74

2.04.3.5.2 Direct Space Methods


Direct space methods do not require high-resolution data and/or a highly favorable data-to-parameter ratio. 75 Indeed, by making
use of the intensities observed at each data point, the number of data is increased and the intensity extraction procedure avoided.
54 X-ray Powder Diffraction

Figure 6 A schematic representation of the steps involved in a single Monte Carlo move for a molecule with just one degree of freedom, the
torsion angle between the two rings. CF, cost function. Source: Caliandro, R.; Giacovazzo, C.; Rizzi, R. In Powder Diffraction: Theory and Practice; Din-
nebier, R.E., Billinge S.J.L., Eds.; RSC Publishing: Cambridge, 2008.

Nevertheless to explore the coordinates direct space, it is desirable to reduce the number of degrees of freedom. A very effective
strategy is the use of internal coordinates (i.e., bond lengths, bond and torsion angles) instead of atomic coordinates to describe the
molecular structure. In the case of molecular compounds, bond lengths and bond angles have predictable values based on the
knowledge of the stereochemistry and also the values of some torsion angles can be predicted to some extent. Thus the number
of molecular degrees of freedom can be considerably reduced from 3N 6, where N is the number of atoms in the molecule. More-
over, there are cases where the translation and rotational degrees of freedom (also called external degrees of freedom) can be further
reduced by considering the crystal symmetry. For example, if the molecule possesses a crystallographic inversion center, the molec-
ular centroid has to be located in a special position.
In direct space methods trial crystal structures are generated and their XRPD pattern calculated and compared with the experi-
mental XRPD pattern by means of a figure of merit, called cost function, which is usually related to the disagreement indices Rp or
Rwp.
Direct space methods differ in the way the trial structures are generated.
Grid-search methods76–79 perform an exhaustive search in the direct space, but they are of limited use with complex structures,
when the number of degrees of freedom increases.
Therefore random search methods (mainly based on Monte Carlo method) are applied in order to generate trial structures in the
initial stage ( Fig. 6). The exploration of the direct space is performed either by simulated annealing37,38,80,81 or using evolutionary
algorithms.75,82–85
In a simulated annealing procedure, by analogy with the process of forming a solid by cooling from a melt, the temperature is
slowly decreased and the structural model is allowed to feature large values of the cost function until the best solution is found. If
a solution present a worse cost function it is accepted with a probability weight in order to allow a possible escape from local
minima. 47
The heart of the genetic algorithm is the breeding (i.e., systematic combination) of the members (i.e., structural models) of the
initial population by following the rules of genetics. 86,87 Two kinds of combinations are considered: the mating and the mutation
(Fig. 7). By mating selected genes (i.e., structural parameters) are interchanged between two members of the population; by muta-
tion just one structural parameter (chosen at random) is changed. Of course if two feasible structures are coupled, it is not warranted
that child structures are meaningful. The cost function is chosen in order to eliminate unfitting children structures by natural
selection.

2.04.3.5.3 Dual Space Methods


The charge flipping algorithm 49,88 makes use of the measured intensities and assign random phases to the structure factors to calcu-
late the electron density map. Since the electron density cannot be negative, all points of the electron density map that are below
a predetermined small positive value are flipped to make them positive. The modified electron density map is then used to calculate
new phases, which are in turn used to calculate a new electron density map from measured intensities. The reflection intensities are
X-ray Powder Diffraction 55

Figure 7 
Flow chart of the genetic algorithm for structure solution. Adapted from Cerný, R.; Favre-Nicolin, V. Z. Kristallogr. 2007, 222, 105–113.

calculated and compared with the measured ones. Convergence is reached when the calculated and measured intensities match,
then the phases will not change significantly.
The charge flipping algorithm, as implemented in Superflip, 89 was modified to address the problem of reflection overlap in
XRPD data. Each reflection is flagged as being single or a member of an overlap group. The repartitioning of the total intensity within
a group is performed periodically and is coupled with a second perturbation of the electron density map by means of histogram
matching.90 The electron density histogram used for the histogram matching is derived from the chemical composition or from
a related structure.
Charge flipping resulted very effective in solving structures that obey the Sheldrick’s rule and in particular those containing heavy
scatterers, as zeolites or MOFS.
Purely organic structures with no heavy atoms and low symmetry represent still a hard problem to solve, although a few guide-
lines for approaching such problems have been recently formulated. 91

2.04.3.5.4 Structure Refinement: The Rietveld Method


The Rietveld method 15,16 is a whole-pattern-fitting structure refinement method.92
It applies the least-squares method by finding the minimum of a function, which is the sum of the squares of the difference
between the individual intensities of the observed pattern and the calculated pattern.
X  2
wi yi;o  yi;c
i

where wi ¼ 1/yi,o and the sum is extended to each individual intensity in the pattern.
As a nonlinear least-squares refinement procedure, the Rietveld method requires that the starting model is close enough to the
correct structure. Sometimes the model may be incomplete and difference Fourier analysis can be used with Rietveld refinement in
order to complete the structural model. Usually maximum entropy reconstruction may provide easier interpretable Fourier maps. 93
Care must be taken that the scaling factor between the calculated pattern and the observed one is correct in order to obtain a reli-
able difference Fourier map. There are cases where automatic scaling could lead to erroneous results. 94
In the case of zeolites (and in general of framework materials) the low-angle reflection intensities are particularly sensitive to the
presence or absence of guest molecules, whereas the high-angle reflection intensities are determined primarily by host framework.
Thus, if the partial model consists only of the host atoms, the high-angle reflections are best suited for the determination of the scale
factor. The intensity differences at low angles reflect the incompleteness of the model and contain the information used to locate the
missing atoms in the difference Fourier. Thus, the scale factor must be kept fixed until all missing atoms have been added to the
model. 94
The amount of structural details that can be reliably refined depends on the data-to-parameter ratio. To avoid problem with
convergence during refinement cycles, geometric restraints based on standard molecular geometries are usually applied. Indeed
by describing the crystal structure using internal coordinates as structural parameters, the number of refinable parameters may
be limited by stereochemical considerations.
Inevitably the scale, the occupancy factors and the atomic displacement parameters (thermal factors) are highly correlated with
one another, and are more sensitive to the background correction than are the positional parameters. 95 Usually an overall thermal
factor is refined for all nonhydrogen atoms at least in the early stages of the refinement, while occupancy factors are usually con-
strained so that the sum for atoms located on the same site is equal to 1.
The fit between experimental and calculated powder diffraction patterns is assessed by means of the following disagreement
indices.The powder pattern R-factor Rp
P
jyi;o  yi;c j
Rp ¼ i P
i yi;o
56 X-ray Powder Diffraction

The weighted powder pattern R-factor Rwp


2  2 31=2
P
i wi yi;o  yi;c
6 P 2 7
Rwp ¼4 5
i yi;o

Rwp reflects directly the refinement progress, since the fraction contains exactly the quantity to be minimized.To allow a direct
comparison with single-crystal refinement data RF 2 and RF are calculated 92
P 2
i jFi;o  Fi;c j
2
RF 2 ¼ P 2
i Fi;o

P
jjFi;o j  jFi;c jj
RF ¼ i P
i jFi;o j

F2i,o and Fi,o are not observed quantities, but are derived from the observed powder pattern by applying the following equation
X  yc;hkl
Ihkl ¼ yi;o  yi;c P
i hkl yc;hkl
P
where hkl indicates the sum of all the Bragg reflections underneath the observed peak.
In this way the integrated intensity under each peak is partitioned among the contributing Bragg reflections. Since the contribu-
tion of each reflection to a peak depends on the structural model under consideration, the values RF 2 and RF are model-biased.
Finally as pointed out by Young “numerical criteria are very important, but numbers are blind,” 92 the visual inspection of the fit
between the observed and calculated pattern may be very effective to direct the refinement process and chemical sense should
always be used to distinguish wrong models with apparently good R factors.94
Still it is valid what stated by Stephens (adapting a phrase attributed to Leonardo da Vinci) “Rietveld refinement is never finished,
only abandoned.” 96 Sometimes it is difficult to abandon a Rietveld refinement and it is useful to remind that although the obtained
final structural parameters are not as accurate or precise as those from SC-XRD data, reliable structural details may be provided that
are of interest to chemists and supramolecular chemists, such as the molecular packing arrangement and identification of the inter-
molecular interactions.
A very interesting development, proposed by Schmidt, 97 allows to refine organic crystal structures from XRPD data with incorrect
lattice parameters. It is based on the similarity measure, developed by de Gelder et al.,98 by using the cross- and auto-correlation
functions of a simulated and an experimental powder pattern. The lattice parameters, translation and orientation parameters,
and selected internal degrees of freedom are optimized until the similarity measure reaches a maximum. Then, a Rietveld refinement
is carried out.

2.04.3.6 Available Software for XRPD


Several software packages for XRPD are available, among freely available programs the package EXPO is able to carry out all the steps
of the crystal structure determination process: indexing, space-group determination, integrated intensity estimation by Le Bail
method, 18 reciprocal- and direct-space phasing techniques, and Rietveld refinement.38
The FullProf Suite 99,100 is constituted by a set of freely available crystallographic programs (FullProf, WinPLOTR, EdPCR, GFou-
rier, etc.) mainly developed for Rietveld analysis of neutron (constant wavelength, time of flight, nuclear and magnetic scattering) or
XRPD data. The program can be also used for pattern decomposition using the Le Bail method.18
FOX (Free Objects for Crystallography) is a free, open-source program for the ab initio structure determination from XRPD data
by using global optimization algorithms. 80 It allows a versatile description of the crystal contents dealing with different fragments
(including polyhedra) and to build inorganic or hybrid organic/inorganic structures from them. It provides an automatic correction
for special positions and shared atoms between polyhedra, suitable for global optimization algorithms.
GSAS-II 101 follows the widely used Rietveld refinement program GSAS.102,103 GSAS-II is an open source Python project that
addresses all types of crystallographic studies, from simple materials through macromolecules, using both powder and single-
crystal diffraction and with both X-ray and neutron probes. GSAS-II is able to handle all the steps in diffraction analysis, such as
data reduction, peak analysis, indexing, Pawley fit,17 small-angle scattering fit, structure solution in addition to structure
refinement.103
DASH 37 provides a user-friendly GUI for solving crystal structures from XRPD data, it is optimized for molecular structures and
distributed by the Cambridge Crystallographic Data Center.
TOPAS 104 is a powerful suite of software written by Coelho for the analysis of powder diffraction data. It allows to perform
indexing, individual reflection fitting, Pawley fit,17 simulated annealing for structure solution, Rietveld refinement.15,16 TOPAS is
distributed by Bruker AXS, an academic version may be requested directly to the author.
For a full list of programs it is advisable to refer to the IUCr website: http://www.iucr.org/resources/other-directories/
software.
X-ray Powder Diffraction 57

2.04.4 Pair Distribution Function

The PDF represents another way to extract the maximum information from the experimental XRPD pattern.
PDF analysis allows to determine the different interatomic distances, which occur in a certain compound and it is particularly
useful for semi-crystalline or noncrystalline materials, where short-range order can be evidenced. 14
The PDF, or reduced G(r) function, gives the probability of finding pairs of atoms separated by a distance r and is obtained from
the total structure factor S(Q) via a sine Fourier transform (FT):
ð
2 Qmax
Gðr Þ ¼ Q ½SðQÞ  1 sin Qr dr
p Qmin

where Q is the magnitude of the scattering vector, Q ¼ 4psin q/l, and r is the interatomic distance. S(Q) is the experimental coherent
X-ray scattering intensity, after correcting the raw data for sample self-absorption, multiple and Compton scattering.
The PDF is related to the Patterson function, 42,43 which is derived from single-crystal diffraction data by a Fourier trans-
form of the reflection intensities. The Patterson map shows the lengths and the directions of the crystallographically averaged
interatomic vectors. The PDF is the Fourier transform of the scattered intensity from an isotropic sample (a powder or amor-
phous material) and gives a distribution of interatomic distances like the Patterson function, except that directional informa-
tion is lost and, because both diffuse and Bragg intensities are sampled in the PDF transform, the PDF is sensitive to
nonperiodic components of the structure. In this sense, the PDF can be thought of as an orientation averaged generalized
Patterson function.
Although the structures of disordered systems such as liquids and glasses have been studied for over 50 years via Fourier trans-
form of diffraction data into real-space, it is only over the past few decades that the use of the PDF method has become increasingly
popular for the structural studies of crystalline or partially crystalline systems. More recently, the PDF method has been gaining
interest for studying organic materials, especially nanocrystalline and amorphous organic compounds. However, the PDF analysis
of organic compounds is at present mainly restricted to the use of the PDF curve as “fingerprinting,” allowing to identify packing
patterns or polymorphs and to determine the ordering length by comparison with the local structures of crystalline, nanocrystalline,
and amorphous samples. Examples of applications can be found in pharmaceutical industry, where the role is to uniquely identify
different amorphous and nanostructured phases of polycrystalline pharmaceuticals compounds. 105,106
PDF method applied to molecular and pharmaceutical compounds has been essentially employed for phase identification and
quantification in crystallization, re-crystallization, and amorphization processes, for observing kinetic behavior, and for the char-
acterization of the local structure of amorphous phases that might occur. 107,108
Hitherto, a wider use of the PDF method in molecular chemistry, in terms of structure solution and refinement, has been
hampered by the fact that the fit of a crystal structure to the PDF data is not practically feasible for organic molecular compounds,
because of the limitations of the available software programs to deal with rigid groups of atoms (rigid bodies and/or molecular
fragments).
Recently, progresses toward structure solution and refinement of molecular organic crystals have been made by developing ad
hoc global optimization procedures. 109

2.04.5 XRPD Uses

XRPD found several applications in various fields, as metallurgy, archeometry, materials science, pharmaceutical industry, etc. In the
following we describe the most frequent and useful applications with particular reference to the field of supramolecular chemistry.

2.04.5.1 Phase Identification


Phase identification is probably the most frequent and routine application of XRPD in all fields (metallurgy, archeometry, materials
science, and pharmaceutical industry). 110
Both the position and the relative intensity of the peaks in the XRPD pattern depend on the crystal structure of the
analyzed crystalline phase. If a material contains more than one crystalline phase, the pattern for each phase will sum
to give the final XRPD pattern. Thus, the XRPD pattern of a crystalline phase may be used as a fingerprint for phase
identification.
Several search/match software is available, some is included in the diffractometers vendor suites. Cranswick in 2008 provided
a rather accurate list, 111 which is still valid at the time of writing.
This software rely on the extensive use of structural information from the main structural databases: the Cambridge Structural
Database (organic and organometallic compounds), 112 the Inorganic Crystal Structure Database (inorganic, mineral, and metal
compounds),113 and CrystMet (metal, alloys, and intermetallic compounds).114 The Crystallography Open Database provides
open access to crystal structure information and related software.115 The International Center for Diffraction Data collects experi-
mental and calculated XRPD patterns and provides them in the Powder Diffraction File series, which contains also data mining
software.116
58 X-ray Powder Diffraction

In the last 30 years single-crystal XRD data are collected at low temperature (usually at 100 K), while in most cases XRPD data are
measured at room temperature. Therefore, care must be taken in comparing structural data at different temperatures, that is, calcu-
lated XRPD patterns from single-crystal XRD data at low temperature and experimental XRPD data at room temperature.
Two XRPD patterns collected at different temperatures may appear rather different because of anisotropic thermal expansion,
although they refer to the same crystalline phase.
Indexing the experimental pattern can help distinguish between anisotropic thermal expansion of the crystalline phase and a real
phase change. Sometimes anisotropic thermal expansion can be also exploited to enhance the extraction of diffracted intensities
from a heavily overlapped XRPD pattern. 117
The method was validated by the solution of an organic compound, which was 50% larger than any other structure solved in
a comparable way. 117

2.04.5.2 Lattice Parameters Determination


Lattice parameters determination is particularly important in solid-state chemistry in order to identify new materials. Usually the
indexing procedure is followed by a whole pattern fitting method as Pawley 17 or Le Bail18 fit to obtain refined values for the lattice
parameters.
Following the change of the lattice parameters during a phase transformation or during a solid-state reaction may give insights in
the thermodynamic and/or kinetic aspects of the processes under study. XRPD is very well suited for time resolved studies as well as
for variable temperature experiments or high-pressure studies.

2.04.5.2.1 Kinetic Studies


A calixarene-based compound, namely 1,2-dimethoxy-p-tert-butylcalix[4]dihydroquinone ( Fig. 8), has a cubic structure
(a ¼ 36.412(4) Å) with 3D networked channels and hydrophobic cages. The host framework is based on weak intermolecular inter-
actions as hydrogen bonds and CH–p interactions (Fig. 9). The channels have minimum and maximum diameter of 3.9 and 8.5 Å
(excluding van der Waals radii) and are filled with easily removable water molecules. Noteworthy, the 3D supramolecular frame-
work is surprisingly robust, being preserved also after the removal of channel water molecules.118
The adsorption and the desorption of carbon tetrachloride, chloroform, and water molecules were studied by in situ XRPD and
TGA to obtain structural and kinetic information. 119
The translating imaging plate setup available on Italian BM08 beamline 19 allowed monitoring in situ the transformation
induced by guest uptake and release.
XRPD patterns recorded during adsorption and desorption processes made possible to relate the guest uptake and release to
a progressive change of cell parameter with retention of the cubic crystalline structure confirming the robustness of the porous
architecture.
In Fig. 10 XRPD patterns are shown for CCl4 adsorption and desorption processes, respectively. In particular, upon adsorption
the 200 reflection disappears, while the intensity of the 110 reflection increases. The cell parameters variation as obtained by Le Bail
fit18 upon guests adsorption and desorption are reported in Fig. 11.
The adsorption processes were described by a type I isotherm and time-dependent diffusivity coefficients were also calculated.

2.04.5.3 Crystal Structure Determination


The first unknown organic molecular structure (containing only light atoms) to be determined from laboratory XRPD data by Direct
methods was formylurea. 120
The first unknown crystal structure to be solved using a Monte Carlo method in the direct space was p-BrC6H4CH2CO2H. 121
Several other structures followed, here we report only some selected examples.
Probably the most impressive crystal structure determination using HR-XRPD data is due to Von Dreele et al. 122 A new variant of
the T(3)R(3) Zn-human insulin complex was obtained by mechanical grinding of a polycrystalline sample. The structure solution
was achieved by molecular replacement adapted for Rietveld refinement: 1630-atom protein was refined by combining 7981

Figure 8 1,2-Dimethoxy-p-tert-butylcalix[4]dihydroquinone. Source: Erra, L.; Tedesco, C.; Immediata, I.; Gregoli, L.; Gaeta, C.; Merlini, M.; Mene-
ghini, C.; Brunelli, M.; Fitch, A.N.; Neri, P. Langmuir 2012, 28, 8511–8517.
X-ray Powder Diffraction 59

Figure 9 Supramolecular host framework in the cubic porous form of 1,2-dimethoxy-p-tert-butylcalix[4]dihydroquinone with interconnected chan-
nels and cavities in yellow (probe radius 1.2 Å). Source: Erra, L.; Tedesco, C.; Immediata, I.; Gregoli, L.; Gaeta, C.; Merlini, M.; Meneghini, C.; Bru-
nelli, M.; Fitch, A.N.; Neri, P. Langmuir 2012, 28, 8511–8517.

Figure 10 XRPD patterns during CCl4 adsorption and desorption processes. Source: Erra, L.; Tedesco, C.; Immediata, I.; Gregoli, L.; Gaeta, C.;
Merlini, M.; Meneghini, C.; Brunelli, M.; Fitch, A.N.; Neri, P. Langmuir 2012, 28, 8511–8517.

stereochemical restraints with a 4800 individual intensity points (dmin ¼ 3.24 Å), yielding the residuals Rwp ¼ 3.73%, Rp ¼ 2.84%,
RF 2 ¼ 8:25%.
It resulted that the grinding induces a phase change with 9.5 degrees and 17.2 degrees rotations of the two T(3)R(3) complexes in
the crystal structure. It was also demonstrated that the material transforms back into the original T(3)R(3) crystal structure in a few
days.
Another impressive example of crystal structure determination using XRPD synchrotron data was reported by Ferey et al. in
2005. 123
The structure of MIL-101, a chromium terephthalate-based porous coordination network, with a giant unit cell (V ¼ 702,000 Å3)
and extra large pores was solved by a computational global optimization strategy developed by the group, called automated
assembly of secondary building units. The method combines inorganic clusters with organic ligands to form 3D periodic frame-
works. 124 A virtual library of candidate frameworks is produced, for each framework an XRPD pattern is simulated and compared
with the experimental one until a good match is obtained. Guest molecules are located from subsequent Fourier difference maps
and Rietveld refinement cycles.
In the last year XRPD helped to complete the set of crystal structures of the 20 aminoacids, by elucidating the structure of the
anhydrous crystal form of L-lysine. 125
60 X-ray Powder Diffraction

Figure 11 Cell parameter variation during (A) adsorption and (B) desorption for CCl4 (black squares), CHCl3 (open triangles), and water (circles).
Source: Erra, L.; Tedesco, C.; Immediata, I.; Gregoli, L.; Gaeta, C.; Merlini, M.; Meneghini, C.; Brunelli, M.; Fitch, A.N.; Neri, P. Langmuir 2012, 28,
8511–8517.

Indeed L-lysine crystallizes as a hydrate form and upon dehydration gives a microcrystalline powder. The difficulty to grow single
crystals of suitable quality explains why the structure elucidation of a member of a biologically important class of compounds
remained undetermined for so long.
The structure was solved from XRPD laboratory data by means of direct-space genetic algorithm, 48,56 as implemented in the
program EAGER.83–85 In total, 16 independent structure-solution calculations resulted in the same structure of highest quality.
Each calculation was run for 500 generations, in each calculation the population comprised 400 trial structures, with 40 mating
operations and 200 mutation operations per generation. A Rietveld refinement15,16 was performed starting from the best model
using the GSAS program.102
Thus, the structure determination from powder diffraction data configure as a feasible approach (though not always straightfor-
ward) to the structural elucidation of several compounds for which single crystals are not available.

2.04.5.4 Crystal Structure Refinement and Quantitative Phase Analysis


Quantitative analysis is one of the most important industrial applications of XRPD, in particular in quality control. Early attempts to
use the Rietveld method for quantitative phase analysis were made by Hill and Howard in 1987 126 and Bish and Howard in
1988.127
Usually the quantification of mixtures via the Rietveld method is restricted to known crystalline phases, amorphous phases may
be quantified by the addition of an internal standard to the sample. 128 Quantification of individual phases that have only partial or
unknown structures may be carried out by applying the partial or no known crystal structure method,129 which is implemented in
TOPAS.104
Of course the accuracy of the results is strongly dependent on the quality of the data, and careful sample preparation and data
acquisition are required for reliable quantifications. Nevertheless the accuracy may be limited by insufficient particle statistics,
preferred orientation, nonhomogeneous particle size distributions among phases, or highly different mass absorption coefficients
(microabsorption). 110,130

2.04.6 Examples of Application of XRPD in Supramolecular Chemistry

Some examples of the application of XRPD in the field of supramolecular chemistry are outlined in the following. They provide an
overview of experimental setup, data collection strategies, and solution methods with the aim to illustrate the advantages of the
method and also its limitations.

2.04.6.1 Structure Determination of MOFs


In MOFs inorganic and organic building blocks are linked by strong covalent bonds providing architecturally robust crystal struc-
tures with a typical porosity of greater than 50% of the crystal volume. In some cases only polycrystalline materials could be ob-
tained and XRPD helped in the structure elucidation either using synchrotron radiation or laboratory data.
X-ray Powder Diffraction 61

2.04.6.1.1 Structure Determination and Characterization of an MOF from Laboratory Data


A recent work carried out at the Inst. Lavoisier (Univ. de Versaille, France) and at the Inst. Charles Gerhardt (Univ. de Montpellier,
France) has provided a complex and complete structure determination and characterization of aluminum fumarate MOF A520. 131
MOF A520 is produced by the chemical company BASF for natural gas storage and deliver in automotive applications. 132 The
commercially available sample was resynthesized using a different recrystallization route, which enabled a better control of crystal-
lites growth, leading to a polycrystalline solid with improved crystallinity and catalytic activity, named MIL-53(Al)-FA (FA for fuma-
ric acid).
The crystal structure was analyzed through a combination of XRPD, DFT calculation, ss-NMR spectroscopy, IR spectroscopy, and
TGA.
In particular, the structural determination of MIL-53(Al)-FA compound was carried out from XRPD data collected on
a Bruker D8 Advance diffractometer in transmission geometry, equipped with a Ge(111) monochromator (Cu-Ka1 radiation)
and a LynxEye detector. Pattern indexing was performed by using DICVOL software 30,31 and a candidate cell was used for the
extraction of the integrated intensities extraction. The structure was solved by means of Direct methods, using the package
EXPO,38 leading to the determination of the position of Al atoms and most of the C and O atoms. Such partial structure
was optimized by DFT calculations, taking into account the presence of water in the pores as determined by TGA, and allow-
ing a full relaxation of both the atomic positions and the cell parameters. Finally, the DFT-optimized framework structure
was used as a starting point for Rietveld refinement, while additional water molecules were localized through successive
difference Fourier maps. The final refined allowed the analysis of the host–guest interactions in terms of distances between
the free water molecules and the framework.
In addition, variable temperature XRPD and TGA allowed to follow the thermal behavior of MIL-53(Al)-FA, which showed
a rigid character upon dehydration: no drastic shift of the Bragg peaks occurs upon the loss of free water molecules; only a small
contraction of the channel along the fumarate direction. The anhydrous compound shows a low degree of crystallinity, as evidenced
by peak broadening, that is reversible upon rehydration. Thus, upon dehydration a large strain effect occurs due to the rigid char-
acter of the fumarate linker.
To obtain further insight into the structure at the microscopic level, ss-NMR and Fourier Transform IR spectroscopy experiments
were carried. ss-NMR confirmed that the Al atoms are coordinated octahedrally with the octahedra sharing opposite corners, as
resulted from the structure determination, and IR spectroscopy versus temperature evidenced the differences between the commer-
cial and the optimized samples in terms of crystalline defects.
Scanning electron microscopy for morphologic investigation was performed, and porosity of the Al fumarate in its dehydrated
state was determined by nitrogen porosimetry, which gave a detailed description of BET areas.
Finally, in order to correlate the catalytic activity of MIL-53(Al)-FA for ethanol dehydration, high temperature (up to 250 C and
300 C), catalytic experiments were carried out on the dry compound exposed to ethanol-rich atmosphere and then identifying the
byproducts of the reaction.
This example shows how crystal structure elucidation of a compound is a prerequisite not only for understanding the chemical
properties, but also as a guide for further optimization of the synthetic conditions. Here the structure determination was achieved
combining XRPD and computational methods with ancillary techniques as TGA, IR spectroscopy, and ss-NMR.

2.04.6.1.2 Structure Determination of MOFs by HR-XRPD


When good quality XRPD data are available, structure solution of relatively large MOF structures is feasible using a combination of
Direct methods, electron density maps extractions, and difference Fourier maps. This approach was routinely and successfully
employed in the case CAU-1, 133 MIL-120 and MIL-124,134 MIL-118 in its different hydrate phases,135 MIL-53(Fe),136 and Ni-
STA-12,137 with data collected at the HR-XRPD beamline ID31 (now ID22) of the ESRF (Grenoble, France).
The high-angular resolution in reciprocal space provided by an HR-XRPD beamline helps unraveling overlapping peaks and,
thus, allows a reliable determination of the unit cell.
The high peak-to-background ratio, due to the use of a high intensity, monochromatic beam as delivered at synchrotron radi-
ation facilities, enhances the detection of weak peaks for a correct determination of space group symmetry and a more reliable inten-
sity extraction procedure.
Thus, the number and the good quality of the extracted integrated intensities allow to fulfill the Sheldrick’s rule 46 granting
success to reciprocal space methods.

2.04.6.2 Structure Determination of Covalent Organic Frameworks


Covalent organic frameworks (COFs) emerged as a new class of porous covalent organic compounds entirely made of light elements
(B, C, N, O, Si). They find applications as gas storage materials for energy applications, as solid supports for catalysis, and as opto-
electronic devices. The synthesis of COFs by design is made possible by the rigid geometry of the linkers, that allow to predict the
desired composition, pore size, and aperture.
XRPD is used to structurally characterize COFs, in detail the calculated XRPD pattern for a given COF is compared with the exper-
imental one, and if the agreement is satisfactory, this is used as a starting model for solving and then refining the crystal structure of
the target COF. 138
62 X-ray Powder Diffraction

Figure 12 (A B) Idealized ctn and bor topologies as the most symmetric possibilities for the assembly of tetrahedral and trigonal building units, respec-
tively. (C, D) Experimental XRPD patterns for COF-105 (black) and COF-108 (black), respectively, and calculated patterns for bor (blue) and ctn (red) topolo-
gies. Source: El-Kaderi, H.M.; Hunt, J.R.; Mendoza-Cortes, J.L.; Côté, A.P.; Taylor, R.E.; O’Keeffe, M.; Yaghi, O.M. Science 2007, 316, 268–272.

As an example COF105 and COF108 were synthesized as crystalline solids by condensation reactions of tetrahedral tetra(4-
dyhydroxyborylphenyl)methane or tetra(4-dyhydroxyborylphenyl)silane and by cocondensation of triangular 2,3,6,7,10,11-
hexahydroxytriphenylene. 139
ctn and bor topologies were expected, being the most symmetric possibilities for the assembly of tetrahedral and trigonal
building units, respectively. Thus, XRPD patterns were calculated considering both topologies, by comparing them with the exper-
imental data, a ctn topology may be assigned to COF-105 exhibits and a bor topology to COF-108 ( Fig. 12).
The model building was performed using the crystal building module Cerius2 140 and the calculation of simulated PXRD patterns
was performed by Powder CeLL.141
XRPD data were collected using a Bruker D8-Discover diffractometer in reflection geometry using parallel focusing Göbel mirrors
and a Vantec linear PSD detector. Zero-background sample holders were used.
Indexing was performed using TREOR 23 as implemented in the Powder-X software suite.142
To refine the cell parameters a model-biased Le Bail fit 18 was performed by means of the program GSAS program102 using the
atomic coordinates calculated by model building.
Background was corrected considering a six terms Chebyschev polynomial function, peak asymmetry, and polarization were also
taken into account. Unit cell parameters were refined followed by zero-shift. As final step unit cell parameters, peak asymmetry,
polarization, and zero-shift were refined.
The adopted parameters refinement sequence represents a standard procedure for performing a Le Bail fit. 18 Although a full Riet-
veld refinement was not performed, the available data unequivocally allow to assign the framework structure.
It must be underlined that the design principles of reticular chemistry are not only the base for the synthesis of the material but
also for the structure elucidation from XRPD data.

2.04.6.3 Structure Determination of Fullerene-Based Compounds


Fullerenes features as a test field for XRPD, as an example HR-XRPD data, were used by Stephens in 1994 to demonstrate that AC60
(A ¼ K, Rb, Cs) compounds undergo a thermal phase transition driven by the reversible formation and breaking of covalent
bonds. 143
X-ray Powder Diffraction 63

Figure 13 (A) Experimental (black), calculated (red), and difference XRPD patterns (blue) for PCBM. The higher 2q region is zoomed by seven
times; (B) detail of the molecular structure of PCBM showing the refined side chain conformation. Adapted from Casalegno, M.; Zanardi, S.; Frigerio,
F.; Po, R.; Carbonera C.; Marra, G.; Nicolini, T.; Raos, G.; Meille, S.V. Chem. Commun. 2013, 49, 4525–4527.

More recently XRPD was used by Meille and coworkers to solve the structure of solvent-free phenyl-C61-butyric acid methyl ester
(PCBM). 144 PCBM is a substituted fullerene widely used in solvent-free nanocrystalline films in plastic solar cells. In spite of its wide
application in optoelectronic devices, only the structures of the corresponding cocrystals from SC-XRD data were reported in the
literature.145
XRPD data were recorded at 298 K using a laboratory diffractometer, equipped with a secondary graphite monochromator and
a scintillation counter. The data collection range was 5.00–89.98 degrees (2q) and a step scan mode with a 0.02 degree step and
a counting time of 20 s were used with a Cu-sealed tube source. The quality of the diffraction pattern allowed to perform an accurate
structure analysis, using TOPAS. 104 As a starting model it was considered the PCBM molecule in the o-dichlorobenzene PCBM 1:1
cocrystal,145 a rigid body description was used for the entire molecule allowing to considerably reduce the parameters to be consid-
ered in a simulated annealing procedure. In particular, initially four side chain torsion angles were allowed to vary. Rietveld refine-
ment was applied until satisfactory final disagreement indices were obtained, the agreement between the observed and calculated
XRPD patterns can be appreciated in Fig. 13.
Thus, XRPD was successful in providing reliable atomic details of solvent-free PCBM paving the way to a better understanding of
charge mobility in electronics devices.

2.04.6.4 Adsorption Studies


Porous materials are used in catalysis, for purification and separation purposes as the capture of carbon dioxide or other toxic
gas or volatile compounds, for storage of fuels (hydrogen and methane). In situ XRPD analysis is an invaluable mean to char-
acterize the processes of adsorption and desorption, allowing to understand the molecular details of the host–guest recogni-
tion process.
64 X-ray Powder Diffraction

2.04.6.4.1 Adsorption Studies in MOFs


In the following we describe how it was possible to define:
The binding sites of CO2 and SO2 molecules in an MOF, called NOTT-300, were defined by performing in situ XRPD studies
during gas adsorption. 146
The solvated compound, indicated as NOTT-300-solvate, was prepared via solvothermal reaction. In situ HR-XRPD was collected
on Beamline I11 at Diamond Light Source by using multianalyzing-crystal (MACs) detectors and an in situ gas-handling system.
The powder pattern was indexed considering a body-centered tetragonal lattice and the independent unit cell parameters were
refined using TOPAS. 104
The systematic absence of 00l reflections with l s 4 indicate as possible tetragonal space groups the enantiomeric pair I4122 and
I4322. The structure was initially solved in space group I4122 by means of the charge flipping algorithm 49 as implemented in the
program Superflip.89 Subsequent analysis of the difference Fourier maps allowed to build the final model, which was refined by
Rietveld method15,16 using TOPAS.104
A total of 40 disordered water molecules per unit cell were included in the final structure refinement for NOTT-300-solvate.
During desolvation and SO2 loading there is a significant increase in some peak intensities, while there is no change in the peak
positions. This indicates that SO2 molecules enter into the material without any significant structural change in the host framework.
Guest molecules inside the framework were located by means of a Monte Carlo-based simulated annealing technique in which
the guest molecules are treated as rigid bodies. The final stage of the Rietveld refinement involved soft restraints to the C–C bond
lengths within the benzene rings. Rigid body refinement was applied to the guest molecules in the pore.
The obtained results agree with the results of the inelastic neutron scattering experiments combined with DFT modeling and
provide key insights into the dynamics of the NOTT-300 host upon SO2 inclusion.
In particular, they show that hydroxyl groups bind CO2 and SO2 through the formation of moderate/weak hydrogen bonds,
which are reinforced by weak supramolecular interactions with C–H atoms on the aromatic rings of the framework.

2.04.6.4.2 In Situ Adsorption Studies of Flexible MOFs


As MOFs find applications in technologies as carbon capture, gas storage, separation, sensing, and heat transformation, it is of prime
importance the identification of the different crystalline phases that might form in operando conditions, in order to understand and
possibly control the guest-induced transformation processes. By using intense X-ray beam from a synchrotron radiation source and
large area 2D detectors for fast data acquisitions, in situ adsorption diffraction experiments can be carried out to mimic operando
conditions. As an example, a team of researchers followed the adsorption of short of apolar linear alkanes, such as methane
CH4, ethane C2H6, propane C3H8, and butane C4H10 in the flexible MOF MIL-53(Fe). 147
HR-XRPD measurements were performed at the SNBL/ESRF (Grenoble, France) using an MAR345 imaging plate 2D detector,
with acquisition time of 30 s, while gas pressures up to 60 bar of alkane gases were applied to an evacuated MIL-53(Fe) sample.
In situ XRPD patterns collected during the adsorption of the linear alkanes (C1–C4) showed unambiguous changes in the crystalline
structure as a function of the gas adsorption into the channels. Diffraction patterns for adsorption isotherms at equilibrium allowed
the identification and indexing of the different main phases corresponding to different degrees of pore opening of the MIL-53(Fe),
according to the gas pressure: anhydrous very narrow pore, an intermediate form, narrow pore form, and a large pore form. Initial
guess crystal structures from previous analogous compounds were used as starting structures for Force Field energy minimization,
and successively Direct Space Monte Carlo simulations were carried out to probe the geometries of guest molecule. Thus, the appear-
ance of partial pore opening in the flexible MIL-53 solids seemed related to host–guest interactions and adequate fits between the
pore size and its content rather than to the chemical nature of the framework, for example, the number of interaction sites between
the hydrocarbon molecules and the MIL framework increases with the chain length of the alkane molecule. As a consequence, the
different affinity of the MIL-53(Fe) framework for the hydrocarbons, according to the chain length, could lead to applications in
separation.

2.04.6.4.3 Gas Adsorption Studies in a Permanently Porous Molecular Organic Compound


Organic molecular crystals with guest-occupied cavities are frequently observed, but the cavities tend to be unstable and collapse
upon guest removal.
The aforementioned calixarene-based compound, 1,2-dimethoxy-p-tert-butylcalix[4]dihydroquinone ( Fig. 8), represents a rare
exception providing a stable empty 3D supramolecular framework constituted exclusively by organic molecules linked by H-bonds
and CH–p interactions.118
In situ HR-XRPD gas adsorption measurements considered methane, 148 acetylene, and Ar149 as guest molecules (i) to probe the
affinity of the calixarene framework mainly constituted by aromatic moieties, (ii) to in situ recognize the binding sites, and (iii) to
characterize the host structural modifications upon guest adsorption.
Experiments were performed at the HR-XRPD ID31 (now ID22) beamline at ESRF, using a specifically designed gas-handling
system. 150 XRPD patterns were recorded during channel cleaning and gas loading. Both for the gas-loaded and evacuated samples
the cell parameters were evaluated by Le Bail refinement,18 confirming the cubic space group Pn3n. 
As reported in Fig. 14, while the evacuated sample shows a progressive decreasing of the lattice parameter with decreasing the
temperature, the methane-loaded sample shows an increase down to 223 K and then a decrease. This indicates that methane is able
to enter into the channels and this process is favored by decreasing the temperature down to 223 K.
X-ray Powder Diffraction 65

Figure 14 Cell parameter vs. temperature for an evacuated sample (black squares) and a methane-loaded sample (red circles). Source: Tedesco, C.;
Erra, L.; Brunelli, M.; Cipolletti, V.; Gaeta, C.; Fitch, A.N.; Atwood; J.L.; Neri, P. Chem. Eur. J. 2010, 42, 2371–2374.

HR-XRPD data were used to locate the gas molecules inside the host structure by analysis of difference Fourier maps, energy
calculations, and XRPD-based simulated annealing procedures using TOPAS. 104
Difference Fourier maps, considering only the apo-host as structural model, showed that the center of the channels and cages are
not occupied by any molecule, while higher electronic density can be detected near the wall of the channels. The configuration space
of the host–guest system was searched applying a Monte Carlo method to find the guest preferential sites. 151
The following assumptions were made:
1. The ratio between calixarene molecules and guest molecules corresponds to the value suggested by volumetric adsorption
measurements.
2. The host framework is rigid and does not change its structure.
3. P1 symmetry was applied to the guest molecules.
 space group, obtaining the possible sites.
The list of preferential sites was searched to find symmetry-related positions along Pn3n
Finally the host–guest structures were solved by simulated annealing and refined by Rietveld method 15,16 using the program
TOPAS.104
Noteworthy energy-driven simulations and Rietveld refinement gave similar results for the coordinates of the guest molecules.
Thus, we were able to locate confidently the guest molecules inside the channels and in case of methane and acetylene molecules we
demonstrated that CH–p bonds are responsible for the host–guest interactions.
These results demonstrate that adequate characterization of host–guest interactions by in situ HR-XRPD gas adsorption studies
may be achieved in spite of the lack of heavy scatterers, once a fairly accurate data acquisition is provided.

2.04.6.5 Polymorphism and Solvatomorphism


As different polymorphs and solvates have different chemical and mechanical properties, as thermal stability, solubility, process-
ability, bioavailability, etc., the exploration of a solid form landscape is an essential step during the drug development process.
Moreover, the appearance of an unexpected solid form after the approval of a drug is highly undesirable. XRPD is a invaluable
tool in the polymorph screening stage, but also a double-sided weapon in the patent litigation stage. A vivid account is given by
Bernstein in his book about polymorphism. 152
Here two examples are provided, the former in the field of pharmaceutical compounds and the latter in the field of calixarene
compounds.

2.04.6.5.1 Polymorphism and Solvatomorphism in a Pharmaceutical Compound


Benperidol is an antipsychotic drug, used for the treatment of schizophrenia and to control antisocial, and hypersexual behavior.
A recent polymorph screening 153 assessed that it exists in 3 polymorphic forms I–III, a dihydrate form DH, and 10 solvate forms
obtained from ethanol, methanol, acetonitrile, ethyl acetate, nitromethane, 1,4-dioxane, toluene/o-xylene, benzyl alcohol, carbon
tetrachloride, and chloroform (labeled as SEt1, 2SMe, SACN, SEtOAc, SNM, SDIOX, HH, SBenz, STCC, and SCLF, respectively).
XRPD, DTA/TG analysis, and IR spectroscopy were used to characterize the obtained crystal forms and also the products of the
desolvation process.
66 X-ray Powder Diffraction

The structures of HH, SBenz, and polymorph III were solved from XRPD laboratory data.
Samples were sealed in 0.5-mm borosilicate glass capillaries and mounted on a Bruker D8 Discover diffractometer equipped
with Göbel mirrors.
Indexing was achieved using DICVOL04 30 and SVD indexing algorithm32 (implemented in TOPAS104). The space group deter-
mination was carried out considering a statistical assessment of systematic absences. Density considerations allowed Z0 determina-
tion. Structure solution was achieved by simulated annealing using the package EXPO.38
As starting model of benperidol molecule the ones obtained in the SC-XRD analysis of polymorph I (for SBenz form) and SACN
(for HH form and polymorph III) were used, respectively.
The dihedral angles s1 6 (in Fig. 15) were allowed to vary. Also the position of the center of mass and the molecular orien-
tation were used as variables.
In the case of Sbenz form the initial geometry of benzyl alcohol molecule was taken from the CSD (refcode FEBCUL 154) and
dihedral angles sBenz1 and sBenz2 were considered as variables.
In the case of crystal form HH, oxygen atoms of water molecules were fixed in special positions, and the initial unit cell with
water molecules was taken from the CSD (refcode AMCHCA 155).
The final refinements were carried out by the Rietveld method, maintaining the rigid bodies introduced at the structure solution
stage. The background was modeled by a 20th-order polynomial Chebyshev function; peak profiles were described by the Pearson
VII function, and a common (refinable) isotropic thermal factor was attributed to all nonhydrogen atoms, while that of hydrogen
atoms was assumed to be 1.2 times higher.
The analysis of the solvate crystal structures and molecular properties indicated that the main reason for solvate formation has to
be ascribed to the inability of benperidol molecules to pack efficiently without solvent, whereas the presence of specific functional
groups in benperidol molecule enabled the formation of a wide range of stable solvate structures with various solvent molecules.

2.04.6.5.2 Solvatomorphism in a Calixarene-Based Compound


The peculiarity and specificity of the aforementioned calixarene-based porous framework prompted us to study the assembly prop-
erties of the calixarene building block.
1,2-Dimethoxy-p-tert-butylcalix[4]dihydroquinone ( Fig. 8) crystallizes from chloroform and anhydrous ethyl acetate giving
single crystals with the cubic porous structure featuring 3D networked channels filled with easily removable water molecules.
To understand the role of water molecules in obtaining the cubic porous framework, the compound was crystallized by adding
water to a CHCl3 solution. New triclinic single crystals were obtained at the interface. 20
Notably by using ethyl acetate with a high water content, together with chloroform as crystallization solvents, only microcrys-
talline powder samples could be obtained. By XRPD these resulted to be constituted by a two-phase mixture of cubic and triclinic
crystal form ( Figs. 3 and 16). Thus the formation of a cubic framework is favored only if the water content of the crystallization
solvents is kept as low as possible.
The powder mixture of cubic and triclinic phase was characterized by HR-XRPD at the beamline ID31 (now ID22) before and
after 3 days exposure to CCl4 vapors ( Fig. 16). The triclinic phase disappears and the sample can be indexed as a single cubic phase
(a ¼ 36.66847(15) Å). The lattice parameter is slightly larger than the pristine hydrate form (a ¼ 36.412(4) Å), indicating that CCl4
molecules could enter the channels and substitute the water molecules.
The same powder mixture after 3 days exposure to acetonitrile afforded single crystals ( Fig. 5), which were analyzed at BM01
showing the formation of a nonporous monoclinic chiral crystal structure.
Thus single-handed helices are obtained by supramolecular assembly of calixarene, acetonitrile, and water molecules ( Fig. 17).

2.04.6.6 Complementarity With ss-NMR


XRPD and ss-NMR spectroscopy complete each other in the structural characterization of solid-state materials. ss-NMR can give
structural details as the number of independent molecules in the asymmetric unit, the tautomeric form of the molecule, the presence
of disorder, the existence of specific interactions, and the values of specific interatomic distances.

Figure 15 Molecular structure of benperidol with the numbering of nonhydrogen atoms and labeling of flexible dihedral angles. Source: Berzins, A.;
Skarbulis, E.; Actins, A. Cryst. Growth Des. 2015, 15(5), 2337–2351.
X-ray Powder Diffraction 67

Figure 16 HR-XRPD patterns of a mixture of cubic and triclinic phase of 1,2-dimethoxy-p-tert-butylcalix[4]dihydroquinone (I) before, (II) after
1 day, and (III) 3 days exposure to CCl4 vapors. The patterns were recorded at HR-XRPD beamline ID31 (now ID22), ESRF (France). Source:
Tedesco, C.; Erra, L.; Immediata, I.; Gaeta, C.; Brunelli, M.; Merlini, M.; Meneghini, C.; Pattison, P.; Neri, P. Cryst. Growth Des. 2010, 10, 1527–1533.

Figure 17 Supramolecular helix of 1,2-dimethoxy-p-tert-butylcalix[4]dihydroquinone molecules. (A) View along the b-axis, which is the helix axis;
(B) view perpendicular to b-axis. Calixarene molecules are shown in green and blue, acetonitrile molecules in violet and cyan, and water molecules in
yellow and red. Source: Tedesco, C.; Erra, L.; Immediata, I.; Gaeta, C.; Brunelli, M.; Merlini, M.; Meneghini, C.; Pattison, P.; Neri, P. Cryst. Growth Des.
2010, 10, 1527–1533.

Recently ss-NMR is used to further validate the final structure obtained from a Rietveld refinement.
A combined XRPD and ss-NMR approach was successful in assessing the interactions, which drive the formation of a 1:1 coc-
rystal of indomethacin and nicotinamide. 156,157
XRPD pattern was measured at 294 K on a Bruker D8 instrument using Ge-monochromated CuKa1 radiation in transmission
mode. Data collection time was c.40 h.
The XRPD pattern was indexed using ITO’s method, 28,29 giving a monoclinic cell. To obtain a plausible density the unit cell must
contain four molecules of indomethacin and four molecules of nicotinamide (Z ¼ 4).
The Le Bail method 18 was used to obtain refined unit cell and profile parameters to be used in the subsequent structure solution
calculations. The space group was assumed to be P21/a on the base of the systematic absences.
The structure calculation was performed by means of the direct-space genetic algorithm 48,56 implemented in the program
EAGER.83,85 Indomethacin molecule was described with a total of 11 structural variables (3 translation parameters, 3 orientation
68 X-ray Powder Diffraction

parameters, and 5 torsion-angles). The nicotinamide was described by a total of seven structural variables (three position param-
eters, three orientation parameters, and one torsion-angle variable).
In total, 16 independent structure-solution calculations were performed, 12 resulted in the same structure of highest quality.
Each calculation was run for 100 generations, in each calculation the population comprised 100 trial structures, with 10 mating
operations and 50 mutation operations per generation. A Rietveld refinement 15,16 was performed starting from the best model
using the GSAS program.102
Bond lengths and bond angles were restrained to standard values, planarity restraints were applied to aromatic rings. An overall
isotropic displacement parameter (thermal factor) was refined for each molecule, with the value for the hydrogen atoms fixed at 1.2
times the value for the nonhydrogen atoms. The March Dollase function was used to account for preferred orientation. 158,159
In the final stages of the refinement a closer inspection of the agreement between the observed and calculated XRPD pattern
allowed to identify the presence of an impurity amount of indomethacin.
Thus, in the subsequent refinement cycles, the known crystal structure of pure indomethacin was added as a second phase. Inter-
estingly the presence of the impurity did not hamper the structure solution process.
Geometry optimization of the crystal structure determined from XRPD data resulted in only slightly shifted atomic positions.
This means that the final crystal structure lies very close to an energy minimum for the system.
Moreover, the DFT optimized structure was used in a Rietveld refinement calculation, where the atomic position was kept fixed
and an excellent fit of the experimental pattern was obtained.
The crystal structure determined by XRPD was used for the prediction of the ss-NMR spectra thanks to the Gauge Including
Projector Augmented Wave method.160–162
An excellent agreement was observed for observed and calculated 1H and 13C NMR chemical shifts and also for chemical shift
correlations for directly bonded CH, CH2, and CH3 moieties in two-dimensional 1H and 13C NMR spectra.
In this example the interplay between XRPD and ss-NMR is fully exploited and probably in the near future a more combined
approach of the two techniques would be the key to solve difficult problems in the structural characterization of materials.

2.04.6.7 Complementarity With IR and Raman Spectroscopy


IR and Raman spectroscopies are frequently used to confirm or complete the structural characterization of materials. In this example
the unusual CO2 adsorption behavior of a small-pore aluminum bisphosphonate MOF MIL-91(Al) was investigated using in situ
XRPD and IR spectroscopy. 163 As probed by microcalorimetry, CO2 adsorption appeared unusual with respect to CH4 and N2
adsorption: while in these latter cases no saturation is observed even at pressure of 50 bar, in the case of CO2 pore filling is achieved
at  15 bar. Moreover, the CO2 adsorption isotherm features an inflection at pressures below 0.5 bar. This was ascribed to a possible
change in the host structure as a consequence of the interaction with a polar molecule such as CO2. To prove such hypothesis in situ
XRPD measurements as a function of gas adsorption up to 30 bar were performed at the SNBL (ESRF) with an improved gas dosing
manifold system using an MAR345 imaging plate as detector. XRPD data at equilibrium condition were used to perform the struc-
ture analysis at each pressure. The cell parameters of CO2-loaded MIL-91(Al), obtained by Le Bail fit18 using the program Fullprof,99
differ only slightly from the activated MIL-91(Al) and the space group is also retained upon CO2 adsorption.
The crystal structure of the hydrated MIL-91(Al) was used as a starting model for the Rietveld refinements and the guest mole-
cules were subsequently located by analysis of the difference Fourier maps and further DFT calculations.
In spite of the modest quality of the XRPD patterns it was possible to determine a subtle local structural change in the host cor-
responding to a twist of 20 degrees of the organic linker in MIL-91(Al).
In situ IR spectroscopy measurements were carried out to confirm these structural findings and gain further insight into local
structural changes upon CO2 adsorption.
A sharp band present at 3688 cm 1 is characteristic of the bridging Al  OH  Al groups, upon CO2 adsorption, a new Al–OH
band appears at 3705 cm 1. The intensity of this latter band gradually increases with the CO2 loading at the expense of the band at
3688 cm 1.
The two bands can be related to a significant modification of the Al  OH  Al angle from 134 degrees to 163 degrees in passing
from the activated to the CO2-loaded structures, as evidenced by the XRPD data.
This multitechnique approach leads to the full characterization of the guest-induced flexibility of the MIL-91(Al) framework,
evidencing an interesting selectivity at low pressure which may prelude to a possible use of MIL-91(Al) for postcombustion CO2
capture.
The presence of disordered moieties and/or amorphous or liquid-like phases or the occurrence of surface phenomena in solid-
state materials constitutes a challenge for XRPD. In many cases the use of Raman spectroscopy can provide useful complementary
information.
To take full advantage of the complementarities of the two techniques in investigating solid-state transformations under non-
ambient conditions, at the Swiss Norwegian Beam Lines at ESRF it is available an experimental setup for simultaneous in situ
Raman/HR-XRPD experiments. 164
This allows the perfect synchronization of the two probes with the reaction coordinate and the elimination of possible bias
caused by different sample holders and environments. Environmental conditioning can be achieved in several ways, using a gas
blower for studies from room temperature to 700 K, a nitrogen cryostream for temperatures down to 100 K, a helium cryostat
X-ray Powder Diffraction 69

for temperature down to 5–10 K. A gas-pumping system is also available from vacuum to 30 bars. The experimental setup can be
further implemented with an UV lamp or laser to study photoreactivity.
The setup was initially tested on three solid-state transformations: (i) the kinetics of the fluorene-tetracyanoquinodimethane
solid-state synthesis, (ii) the thermal swelling and degradation of stearate-hydrotalcite, and (iii) the photoinduced (2 þ 2)-cycliza-
tion of (E)-furylidenoxindole.
One of the most recent applications in the field of supramolecular chemistry is related to the rational design of the solid-state
synthesis of materials based on polyaromatic molecular complexes. 165
The kinetic features of the solid-state reactions were fully elucidated by simultaneous in situ Raman/XRPD, by exploiting the
Avrami equation in isothermal and nonisothermal conditions rate constants, reaction orders, and activation energies were obtained.

2.04.6.8 Complementarity With DFT Calculations


Quantum chemical calculations, based on the DFT, enable geometry optimization of periodic objects and can be used in combi-
nation with powder diffraction.
Combined use of powder diffraction data and solid-state DFT calculations has often been applied in for both organic166–168 and
inorganic 169 structures investigations, replacing final Rietveld refinement by optimization of the crystal structure by energy mini-
mization at the DFT level.
Although the quality of an experimental powder diffraction pattern is an important factor for crystal structure solution and
refinement, structure solution process can be more forgiving to low quality of the experimental data with respect to the final struc-
ture refinement, when the powder diffraction pattern does not allow an accurate refinement of the crystal and molecular structures,
even though it may be good enough to solve the basic molecular packing scheme. In these circumstances one might want to verify or
need to complete the structural analysis by quantum-mechanical DFT. The use of dispersion-corrected DFT has been described by
van de Streek and Neumann, 170,171 where it served as a check on the accuracy of structures derived from single-crystal and powder-
diffraction studies, respectively. Furthermore, by lattice-energy minimization of the crystal structure, it was shown that DFT calcu-
lations can provide atomic coordinates of single-crystal quality, and help to distinguish between several possible space group
choices.172
A remarkable example of the interplay between XRPD and DFT methods is provided by Brammer and coworkers in a recent
article on an MOF zipper. 173
They reported on a 2D layered MOF, [Zn2(camph)2(py)2],2EtOH, camph ¼ (1R,3S)-(þ)-camphorate and py ¼ pyridine, which
consists of discrete Zn2(O2CR)4 paddlewheels connected by camphorate ligands in a 2D layered structure. Pyridine ligands occupy
the Zn axial sites and provide p–p interactions between adjacent layers. Ethanol molecules are located between the layers ( Fig. 18).
By thermal treatment at 150 C ethanol molecules and pyridine ligands are removed and Zn centers are zipped into a continuous
chain by slippage of carboxylate groups, resulting in a 3D nonporous structure [Zn(camph)]. The whole process is reversible, Zn
centers may be unzipped by soaking the 3D nonporous structure in ethanol and pyridine, obtaining the previous 2D layered
structure.

Figure 18 Zipping of noncovalently pillared MOF [Zn2(camph)2(py)2]$2EtOH into [Zn(camph)] and unzipping with pyridine, pyrazine, and 3-bromo-
pyridine. camph, (1R,3S)-(þ)-camphorate; py, pyridine; pyz, pyrazine; and 3-Brpy, 3-bromo-pyridine. Adapted from Smart, P.; Mason, C.A.; Loader,
J.R.; Meijer, A.J.H.M.; Florence, A.J.; Shankland, K.; Fletcher, A.J.; Thompson, S.P.; Brunelli, M.; Hill, A.H., Brammer, L. Chem. Eur. J. 2013, 19,
3552–3557.
70 X-ray Powder Diffraction

The removal of ethanol molecules and pyridine ligands was followed by in situ XRPD at beamline I11 at Diamond Light Source,
using a PSD detector with multiple Mythen2 modules. These studies indicated that crystallinity was retained during the thermal
treatment.
XRPD data collected at 150 C were used for solving the structure of [Zn(camph)], due to peak broadening the structure deter-
mination was achieved by using as a starting model the structure of the analogous compound [Mn(camph)]. 174
Accurate unit cell parameters were determined by Pawley refinement, using as starting values the unit cell parameters of the anal-
ogous Mn compound. The initial crystal structure was built up considering the refined unit cell parameters and the atomic coordi-
nates of [Mn(camph)] with Mn atoms substituted by Zn atoms. The structure optimization was performed by a periodic DFT
calculation using VASP175–177 and followed by Rietveld refinement 15,16 using TOPAS104 with an excellent final Rwp of 0.029.
The periodic DFT model was used also to predict the IR spectroscopy data and the excellent agreement with the experimental IR
spectroscopy data further validates the final model.
Interestingly the zipping and unzipping mechanism was exploited for the preparation of new materials. The zipped motif in
[Zn(camph)] may be unzipped upon addition of new ligands ( Fig. 18).
In situ XRPD experiments allowed to follow the conversion of [Zn(camph)] by uptake of pyrazine and 3-bromo-pyridine (3Br-
py) giving a covalently pillared MOF and a noncovalently pillared MOF, respectively.
The structures were solved from XRPD data collected at room temperature at beamline ID31 (now ID22) at ESRF using a multi-
analyzer crystal detector. The structures of analogous MOFs were used as starting models in simulated annealing procedures.
Finally the presence of the starting compound in the reaction products was always taken into account and Rietveld refinement
allowed to estimate also the yield of the solid-state reactions (which are 74% and 81%, respectively).
This study summarizes the state of the art of XRPD methods as applied in structure elucidation and materials characterization. It
constitutes a fine illustration of the extraordinary progresses made nowadays in supramolecular synthesis thanks to the interplay
between powerful experimental techniques and theoretical calculations allowing us to see the molecules in actions.
Such achievements have been crowned by the recent Nobel Prize in Chemistry 2016, awarded to Sauvage, Stoddart, and Feringa
for their development of molecular machines. Thus, we may think and dream that in the near future we would approach Nobel
Laureate Feynman’s 50-years old prophetic view, 178 in which he imagined what the properties of materials would be “if we could
arrange the atoms. the way we want them?”

Acknowledgments

C.T. is grateful to Prof. Emeritus Attilio Immirzi (University of Salerno) for 20 years of challenging discussions.
Both authors wishes to acknowledge the People Programme (Marie Curie Actions) of the European Union’s Seventh Framework Programme FP7/
2007–13 under REA grant agreement no. PIRSES-GA-2012-319011.

References

1. Debye, P.; Scherrer, P. Nachr. Ges. Wiss. Go¨ttingen, Math-Phys. Kl. 1916, 1–15.
2. Hull, A. W. Proc. Natl. Acad. Sci. U. S. A. 1917, 3, 470–473.
3. Hull, A. W. Phys. Rev. 1917, 10, 661–696.
4. Etter, M.; Dinnebier, R. E. Z. Anorg. Allg. Chem. 2014, 640 (15), 3015–3028.
5. Giacovazzo, C. In Fundamentals of Crystallography, IUCr Monographs on Crystallography, 3rd ed.; Oxford University Press: Oxford, 2011.
6. Klug, H. P.; Alexander, L. E. X-Ray Diffraction Procedures, 2nd ed.; John Wiley & Sons: New York, 1974.
7. Pecharsky, V. K.; Zavalij, P. Y. Fundamentals of Powder Diffraction and Structural Characterization of Materials, 2nd ed.; Springer: New York, 2009.
8. Dinnebier, R. E., Billinge, S. J. L., Eds. Powder Diffraction: Theory and Practice; RSC Publishing: Cambridge, 2008.
9. Hull, A. W. In Fifty Years of X-ray Diffraction; Ewald, P. P., Ed.; International Union of Crystallography: Utrecht, 1962; pp 582–587. Electronic format available at web site. http://
www.iucr.org/publ/50yearsofxraydiffraction.
10. Hull, A. W. J. Am. Chem. Soc. 1919, 41, 1168–1175.
11. Hanawalt, J. D.; Rinn, H. W. Ind. Eng. Chem. Anal. Ed. 1936, 8 (4), 244–247.
12. Hanawalt, J. D.; Rinn, H. W. Powder Diffr. 1986, 1, 2–6.
13. Kern, A.; Madsen, I. C.; Scarlett, N. V. Y. In Uniting Electron Crystallography and Powder Diffraction, NATO Science for Peace and Security Series B: Physics and Biophysics;
Kolb, U., Shankland, K., Meshi, L., Avilov, A., David, W. I. F., Eds.; Springer: Dordrecht, 2012.
14. Egami, T.; Billinge, S. J. L. Underneath the Bragg Peaks: Structure Analysis of Complex Materials In: Pergamon Material Series 7; Elsevier Ltd: Oxford, 2003.
15. Rietveld, H. M. Acta Crystallogr. 1967, 22, 151–152.
16. Rietveld, H. M. J. Appl. Crystallogr. 1969, 2, 65–71.
17. Pawley, G. S. J. Appl. Crystallogr. 1981, 14, 357–361.
18. Le Bail, A.; Duroy, H.; Fourquet, J. L. Mater. Res. Bull. 1988, 23, 447–452.
19. Meneghini, C.; Artioli, G.; Balerna, A.; Gualtieri, A. F.; Norby, P.; Mobilio, S. J. Synchrotron Radiat. 2001, 8, 1162–1166.
20. Tedesco, C.; Erra, L.; Immediata, I.; Gaeta, C.; Brunelli, M.; Merlini, M.; Meneghini, C.; Pattison, P.; Neri, P. Cryst. Growth Des. 2010, 10, 1527–1533.
21. Cockcroft, J. K.; Fitch, A. N. In Powder Diffraction: Theory and Practice; Dinnebier, R. E., Billinge, S. J. L., Eds.; RSC Publishing: Cambridge, 2008.
22. Norby, P.; Schwarz, U. In Powder Diffraction: Theory and Practice; Dinnebier, R. E., Billinge, S. J. L., Eds.; RSC Publishing: Cambridge, 2008.
23. Werner, P.-E.; Eriksson, L.; Westdahl, M. J. Appl. Crystallogr. 1985, 18, 367–370.
24. Altomare, A.; Giacovazzo, C.; Guagliardi, A.; Moliterni, A. G. G.; Rizzi, R.; Werner, P.-E. J. Appl. Crystallogr. 2000, 33, 1180–1186.
25. Kohlbeck, F.; Hörl, E. M. J. Appl. Crystallogr. 1976, 9, 28–33.
26. Kohlbeck, F.; Hörl, E. M. J. Appl. Crystallogr. 1978, 11, 60–61.
X-ray Powder Diffraction 71

27. Taupin, D. J. Appl. Crystallogr. 1973, 6, 380–385.


28. Visser, J. W. J. Appl. Crystallogr. 1969, 2, 89–95.
29. Ito, T. Nature 1949, 164, 755–756.
30. Boultif, A.; Louër, D. J. Appl. Crystallogr. 2004, 37, 724–731.
31. Boultif, A.; Louër, D. J. Appl. Crystallogr. 1991, 24, 987–993.
32. Coelho, A. A. J. Appl. Crystallogr. 2003, 36, 86–95.
33. Le Bail, A. Powder Diffr. 2004, 19, 249–254.
34. Kariuki, B. M.; Belmonte, S. A.; McMahon, M. I.; Johnston, R. L.; Harris, K. D. M.; Nelmes, R. J. J. Synchrotron Radiat. 1999, 6, 87–92.
35. Vaughan, G. B. M.; Mora, A. J.; Fitch, A. N.; Gates, P. N.; Muir, A. S. J. Chem. Soc. Dalton Trans. 1999, 79–84.
36. Dinnebier, R. E.; Olbrich, F.; Van Smaalen, S.; Stephens, P. W. Acta Crystallogr. B 1997, 53, 153–158.
37. David, W. I. F.; Shankland, K.; Cole, J.; Maginn, S.; Motherwell, W. D. S.; Taylor, R. DASH User Manual; Cambridge Crystallographic Data Centre: Cambridge, UK, 2001.
38. Altomare, A.; Cuocci, C.; Giacovazzo, C.; Moliterni, A.; Rizzi, R.; Corriero, N.; Falcicchio, A. J. Appl. Crystallogr. 2013, 46, 1231–1235.
39. Tedesco, C.; Dinnebier, R. E.; Olbrich, F.; van Smaalen, S. Acta Crystallogr. B 2001, 57, 673–679.
40. Scardi, P.; Leoni, M. In Theory and Applications, Diffraction Analysis of the Microstructure of Materials; Mittemeijer, E. J., Scardi, P., Eds.; Springer: Berlin, 2004.
41. Cheary, R. W.; Coelho, A. J. Appl. Crystallogr. 1992, 25, 109–121.
42. Patterson, A. L. Phys. Rev. 1934, 46, 372–376.
43. Patterson, A. L. Z. Kristallogr. 1935, 90, 517–542.
44. Giacovazzo, C. Direct Phasing in Crystallography: Fundamentals and Applications, IUCr Monographs on Crystallography; Oxford University Press: Oxford, 1998.
45. Bannister, C.; Bricogne, G.; Gilmore, C. J. In Bayesian Methods and Maximum Entropy; Skilling, J., Ed.; Kluwer: Cambridge, 1988; pp 225–232.
46. Sheldrick, G. M. Acta Crystallogr. A 1990, 46, 467–473.
47. 
Cerný, R.; Favre-Nicolin, V. Z. Kristallogr. 2007, 222, 105–113.
48. Harris, K. D. M.; Johnston, R. L.; Kariuki, B. M. Acta Crystallogr. A 1998, 54, 632–645.
49. Oszlányi, G.; Süto, A. Acta Crystallogr. A 2004, 60, 134–141.
50. Burla, M. C.; Caliandro, R.; Giacovazzo, C.; Polidori, G. Acta Crystallogr. A 2010, 66, 347–361.
51. Burla, M. C.; Carrozzini, B.; Cascarano, G. L.; Giacovazzo, C.; Polidori, G. J. Appl. Crystallogr. 2012, 45, 1287–1294.
52. David, W. I. F., Shankland, D., McCusker, L. B., Baerlocher, C., Eds. Structure Determination From Powder Diffraction Data; Oxford University Press: Oxford, UK, 2002.
53. Caliandro, R.; Giacovazzo, C.; Rizzi, R. In Powder Diffraction: Theory and Practice; Dinnebier, R. E., Billinge, S. J. L., Eds.; RSC Publishing: Cambridge, 2008.
54. Harris, K. D. M. In Advanced X-ray Crystallography; Rissanen, K., Ed.; Springer: Berlin, 2012.
55. Harris, K. D. M.; Cheung, E. Y. Chem. Soc. Rev. 2004, 33, 526–538.
56. Harris, K. D. M.; Habershon, S.; Cheung, E. Y.; Johnston, R. L. Z. Kristallogr. Cryst. Mater. 2004, 219, 838–846.
57. David, W. I. F.; Shankland, K. Acta Crystallogr. A 2008, 64, 52–64.
58. Shankland, K.; Spillman, M. J.; Kabova, E. A.; Edgeley, D. S.; Shankland, N. Acta Crystallogr. C 2013, 69, 1251–1259.
59. Cernik, R. J.; Cheetham, A. K.; Prout, C. K.; Watkin, D. J.; Wilkinson, A. P.; Willis, B. T. M. J. Appl. Crystallogr. 1991, 24, 222–226.
60. Burla, M. C.; Camalli, M.; Cascarano, G.; Giacovazzo, C.; Polidori, G.; Spagna, R.; Viterbo, D. J. Appl. Crystallogr. 1989, 22, 389–393.
61. Altomare, A.; Foadi, J.; Giacovazzo, C.; Guagliardi, A.; Moliterni, A. G. G. J. Appl. Crystallogr. 1996, 29, 674–681.
62. Carrozzini, B.; Giacovazzo, C.; Guagliardi, A.; Rizzi, R.; Burla, M. C.; Polidori, G. J. Appl. Crystallogr. 1997, 30, 92–97.
63. Altomare, A.; Foadi, J.; Giacovazzo, C.; Moliterni, A. G. G.; Burla, M. C.; Polidori, G. J. Appl. Crystallogr. 1997, 31, 74–77.
64. Altomare, A.; Giacovazzo, C.; Guagliardi, A.; Moliterni, A. G. G.; Rizzi, R. J. Appl. Crystallogr. 1999, 32, 963–967.
65. Wilson, A. J. C. Nature 1942, 150 (3796), 152.
66. Viterbo, D. In Fundamentals of Crystallography, IUCr Monographs on Crystallography; Giacovazzo, C., Ed., 3rd ed.; Oxford University Press: Oxford, 2011.
67. Gilmore, C. J.; Shankland, K.; Dong, W. In Structure Determination From Powder Diffraction Data; David, W. I. F., Shankland, D., McCusker, L. B., Baerlocher, C., Eds.; Oxford
University Press: Oxford, 2002.
68. Morris, R. J.; Bricogne, G. Acta Crystallogr. D 2003, 59, 615–617.
69. Gilmore, C. J.; Shankland, K.; Bricogne, G. Proc. R. Soc. Lond. A 1993, 442, 97–111; Gilmore, C. J.; Shankland, K.; Dong, W. In Structure Determination From Powder
Diffraction Data; David, W. I. F., Shankland, D., McCusker, L. B., Baerlocher, C., Eds.; Oxford University Press: Oxford, 2002.
70. Wilson, C. C. Acta Crystallogr. A 1989, 45, 833–839.
71. Rius, J.; Miravitlles, C. J. Appl. Crystallogr. 1987, 20, 261–264.
72. Rius, J.; Miravitlles, C. J. Appl. Crystallogr. 1988, 21, 224–227.
73. Wilson, C. C.; Tollin, P. J. Appl. Crystallogr. 1986, 19, 411–412.
74. Wilson, C. C.; Wadsworth, J. W. Acta Crystallogr. A 1990, 46, 258–262.
75. Immirzi, A.; Erra, L.; Tedesco, C. J. Appl. Crystallogr. 2008, 41, 784–790.
76. Immirzi, A. J. Appl. Crystallogr. 2007, 40, 1044–1049.
77. Masciocchi, N.; Bianchi, R.; Cairati, P.; Mezza, G.; Pilati, T.; Sironi, A. J. Appl. Crystallogr. 1994, 27, 426–429.
78. Chernyshev, V. V.; Schenk, H. Z. Kristallogr. 1998, 213, 1–3.
79. Mora, A. J.; Fitch, A. N. J. Solid State Chem. 1997, 134, 211–214.
80. Favre-Nicolin, V.; Cerný, R. Z. Kristallogr. 2004, 219, 847–856.
81. Coelho, A. A. J. Appl. Crystallogr. 2000, 33 (3 II), 899–908.
82. Immirzi, A.; Erra, L.; Tedesco, C. J. Appl. Crystallogr. 2009, 42, 810–814.
83. Tedesco, E.; Turner, G. W.; Harris, K. D. M.; Johnston, R. L.; Kariuki, B. M. Angew. Chem. Int. Ed. 2000, 39, 4488–4491.
84. Fujii, K.; Young, M. T.; Harris, K. D. M. J. Struct. Biol. 2011, 174, 461–467.
85. Martí-Rujas, J.; Meazza, L.; Lim, G. K.; Terraneo, G.; Pilati, T.; Harris, K. D. M.; Metrangolo, P.; Resnati, G. Angew. Chem. Int. Ed. 2013, 52, 13444–13448.
86. Immirzi, A.; Tedesco, C.; Erra, L. In Evolutionary Algorithm; Eisuke, K., Ed.; Intech: Rijeka, 2011.
87. Harris, K. D. M.; Johnston, R. L.; Kariuki, B. M. In Evolutionary Algorithms in Molecular Design; Clark, D. E., Ed.; Wiley-VCH Verlag GmbH: Weinheim, 2008.
88. Palatinus, L. Acta Crystallogr. B 2013, 69, 1–16.
89. Palatinus, L.; Chapuis, G. J. Appl. Crystallogr. 2007, 40, 786–790.
90. Zhang, K. Y. J.; Main, P. Acta Crystallogr. A 1990, 46, 41–46.
91. Jung, D. S.; Baerlocher, C.; McCusker, L. B.; Yoshinari, T.; Seebach, D. J. Appl. Crystallogr. 2014, 47, 1569–1576.
92. Young, R. A., Ed. The Rietveld Method; International Union of Crystallography and Oxford University Press: Oxford, 1993.
93. Kumazawa, S.; Kubota, Y.; Takata, M.; Sakata, M.; Ishibashi, Y. J. Appl. Crystallogr. 1993, 26, 453–457.
94. McCusker, L. B.; Von Dreele, R. B.; Cox, D. E.; Louër, D.; Scardi, P. J. Appl. Crystallogr. 1999, 32, 36–50.
95. Shankland, K. J. Res. Natl. Inst. Stand. Technol. 2004, 109 (1), 143–154.
96. Stephens, P. W. In Uniting Electron Crystallography and Powder Diffraction; Kolb, U., Shankland, K., Meshi, L., Avilov, A., David, W. I. F., Eds.; NATO Science for Peace and
Security Series B: Physics and Biophysics; Springer: Dordrecht, 2012.
72 X-ray Powder Diffraction

97. Habermehl, S.; Mörschel, P.; Eisenbrandt, P.; Hammer, S. M.; Schmidt, M. U. Acta Crystallogr. B 2014, 70 (2), 347–359.
98. de Gelder, R.; Wehrens, R.; Hageman, J. A. J. Comput. Chem. 2001, 22, 273–289.
;
99. Rodriguez-Carvajal, J. In 1990; p 127.
100. Roisnel, T.; Carvajal, J. R. Mater. Sci. Forum 2001, 378–381 (Parts 1&2), 118–123.
101. Toby, B. H.; Von Dreele, R. B. J. Appl. Crystallogr. 2013, 46 (2), 544–549.
102. Larson, A.C.; Von Dreele, R.B. General Structure Analysis System (GSAS), Los Alamos National Laboratory Report LAUR 86-748, 2000.
103. Toby, B. H. J. Appl. Crystallogr. 2001, 34, 210–213.
104. Coehlo, A.A. TOPAS, Bruker-AXS; TOPAS Academic, Coelho Software, Brisbane, Australia. http://www.topas-academic.net.
105. Sheth, A. R.; Bates, S.; Muller, F. X.; Grant, D. J. W. Cryst. Growth Des. 2005, 5, 571–578.
106. Billinge, S. J. L.; Dykhne, T.; Juhás, P.; Bozin, E.; Taylor, R.; Florence, A. J.; Shankland, K. CrystEngComm 2010, 12, 1366–1368.
107. Terban, M. W.; Cheung, E. Y.; Krolikowski, P.; Billinge, S. J. L. Cryst. Growth Des. 2016, 16 (1), 210–220.
108. Bordet, P.; Bytchkov, A.; Descamps, M.; Dudognon, E.; Elkaïm, E.; Martinetto, P.; Pagnoux, W.; Poulain, A.; Willart, J.-F. Cryst. Growth Des. 2016, 16 (8), 4547–4558.
109. Prill, D.; Juhás, P.; Billinge, S. J. L.; Schmidt, M. U. Acta Crystallogr. A 2016, 72, 62–72.
110. Whitfield, P.; Lyndon, M. In Principles and Applications of Powder Diffraction; Clearfield, A., Reibenspies, J., Bhuvanesh, N., Eds.; Wiley-Blackwell: Chichester, 2008.
111. Cranswick, L. M. D. In Powder Diffraction: Theory and Practice; Dinnebier, R. E., Billinge, S. J. L., Eds.; RSC Publishing: Cambridge, 2008.
112. Groom, C. R.; Bruno, I. J.; Lightfoot, M. P.; Ward, S. C. Acta Crystallogr. B 2016, 72, 171–179.
113. Hellenbrandt, M. Crystallogr. Rev. 2004, 10 (1), 17–22.
114. White, P. S.; Rodgers, J. R.; Le Page, Y. Acta Crystallogr. B 2002, 58, 343–348.
115. Grazulis, S.; Chateigner, D.; Downs, R. T.; Yokochi, A. T.; Quiros, M.; Lutterotti, L.; Manakova, E.; Butkus, J.; Moeck, P.; Le Bail, A. J. Appl. Crystallogr. 2009, 42, 726–729.
116. ICDD, PDF-4 þ Database, Newtown Square, PA, USA.
117. Brunelli, M.; Wright, J. P.; Vaughan, G. B. M.; Mora, A. J.; Fitch, A. N. Angew. Chem. Int. Ed. 2003, 42, 2029–2032.
118. Tedesco, C.; Immediata, I.; Gregoli, L.; Vitagliano, L.; Immirzi, A.; Neri, P. CrystEngComm 2005, 7, 449–453.
119. Erra, L.; Tedesco, C.; Immediata, I.; Gregoli, L.; Gaeta, C.; Merlini, M.; Meneghini, C.; Brunelli, M.; Fitch, A. N.; Neri, P. Langmuir 2012, 28, 8511–8517.
120. Lightfoot, P.; Tremayne, M.; Harris, K. D. M.; Bruce, P. G. J. Chem. Soc. Chem. Commun. 1992, 14, 1012–1013.
121. Harris, K. D. M.; Tremayne, M.; Lightfoot, P.; Bruce, P. G. J. Am. Chem. Soc. 1994, 116, 3543–3547.
122. Von Dreele, R. B.; Stephens, P. W.; Smith, G. D.; Blessing, R. H. Acta Crystallogr. D 2000, 56, 1549–1553.
123. Ferey, G.; Mellot-Draznieks, C.; Serre, C.; Millange, F.; Dutour, J.; Surble, S.; Margiolaki, I. Science 2005, 309, 2040–2042.
124. Mellot-Draznieks, C.; Newsam, J. M.; Gorman, A. M.; Freeman, C. M.; Ferey, G. Angew. Chem. Int. Ed. 2000, 39, 2270–2275.
125. Williams, P. A.; Hughes, C. E.; Harris, K. D. M. Angew. Chem. Int. Ed. 2015, 54, 3973–3977.
126. Hill, R. J.; Howard, C. J. J. Appl. Crystallogr. 1987, 20, 467–474.
127. Bish, D. L.; Howard, S. A. J. Appl. Crystallogr. 1988, 21, 86–91.
128. Madsen, I. C.; Scarlett, N. V. Y.; Kern, A. Z. Kristallogr. 2011, 226, 944–955.
129. Scarlett, V. Y.; Madsen, I. C. Powder Diffr. 2006, 21, 278–284.
130. Kaduk, J. A. In Principles and Applications of Powder Diffraction; Clearfield, A. H., Reibenspies, J., Bhuvanesh, N., Eds.; Wiley-Blackwell: Chichester, 2008.
131. Alvarez, E.; Guillou, N.; Martineau, C.; Bueken, B.; Van de Voorde, B.; Le Guillouzer, C.; Fabry, P.; Nouar, F.; Taulelle, F.; de Vos, D.; Chang, J.-S.; Ho Cho, K.; Ramsahye, N.;
Devic, T.; Daturi, M.; Maurin, G.; Serre, C. Angew. Chem. Int. Ed. 2015, 54, 3664–3668.
132. Gaab, M.; Trukhan, N.; Maurer, S.; Gummaraju, R.; Müller, U. Microporous Mesoporous Mater. 2012, 157, 131–136.
133. Ahnfeldt, T.; Guillou, N.; Gunzelmann, D.; Margiolaki, I.; Loiseau, T.; Férey, G.; Senker, J.; Stock, N. Angew. Chem. Int. Ed. 2009, 48 (28), 5163–5166.
134. Hajjar, R.; Volkringer, C.; Loiseau, T.; Guillou, N.; Marrot, J.; Férey, G.; Margiolaki, I.; Fink, G.; Morais, C.; Taulelle, F. Chem. Mater. 2011, 23 (1), 39–47.
135. Volkringer, C.; Loiseau, T.; Guillou, N.; Fèrey, G.; Haouas, M.; Taulelle, F.; Audebrand, N.; Margiolaki, I.; Popov, D. Cryst. Growth Des. 2009, 9 (6), 2927–2936.
136. DeCombarieu, G.; Morcrette, M.; Millange, F.; Guillou, N.; Cabana, J.; Grey, C. P.; Margiolaki, I.; Férey, G.; Tarascon, J.-M. Chem. Mater. 2009, 21 (8), 1602–1611.
137. Miller, S. R.; Pearce, G. M.; Wright, P. A.; Bonino, F.; Chavan, S.; Bordiga, S.; Margiolaki, I.; Guillou, N.; Férey, G.; Bourrelly, S.; Llewellyn, P. L. J. Am. Chem. Soc. 2008,
130 (47), 15967–15981.
138. Waller, P. J.; Gandara, F.; Yaghi, O. M. Acc. Chem. Res. 2015, 48 (12), 3053–3063.
139. El-Kaderi, H. M.; Hunt, J. R.; Mendoza-Cortes, J. L.; Côté, A. P.; Taylor, R. E.; O’Keeffe, M.; Yaghi, O. M. Science 2007, 316, 268–272.
140. Cerius2 Modeling Environment, 2000 Molecular Simulations Inc., San Diego, CA, USA.
141. Kraus, W.; Nolze, G. J. Appl. Crystallogr. 1996, 29, 301–303.
142. Dong, C. J. Appl. Crystallogr. 1999, 32, 838.
143. Stephens, P. W.; Bortel, G.; Faigel, G.; Tegze, M.; Jánossy, A.; Pekker, S.; Oszlanyi, G.; Forró, L. Nature 1994, 370 (6491), 636–639.
144. Casalegno, M.; Zanardi, S.; Frigerio, F.; Po, R.; Carbonera, C.; Marra, G.; Nicolini, T.; Raos, G.; Meille, S. V. Chem. Commun. 2013, 49, 4525–4527.
145. Rispens, M. T.; Meetsma, A.; Rittberger, R.; Brabec, C. J.; Sariciftci, N. S.; Hummelen, J. C. Chem. Commun. 2003, 2116–2118.
146. Yang, S.; Sun, J.; Ramirez-Cuesta, A. J.; Callear, S. K.; David, W. I. F.; Anderson, D. P.; Newby, R.; Blake, A. J.; Parker, J. E.; Tang, C. C.; Schröder, M. Nat. Chem. 2012, 4,
887–894.
147. Llewellyn, P. L.; Horcajada, P.; Maurin, G.; Devic, T.; Rosenbach, N.; Bourrelly, S.; Serre, C.; et al. J. Am. Chem. Soc. 2009, 131 (36), 13002–13008.
148. Tedesco, C.; Erra, L.; Brunelli, M.; Cipolletti, V.; Gaeta, C.; Fitch, A. N.; Atwood, J. L.; Neri, P. Chem. Eur. J. 2010, 42, 2371–2374.
149. Erra, L.; Tedesco, C.; Cipolletti, V. R.; Annunziata, L.; Gaeta, C.; Brunelli, M.; Fitch, A. N.; Knöfel, C.; Llewellyn, P. L.; Atwood, J. L.; Neri, P. Phys. Chem. Chem. Phys. 2012,
14, 311–317.
150. Brunelli, M.; Fitch, A. J. Synchrotron Radiat. 2003, 10, 337–339.
151. Sorption Locate Module, Materials Studio 4.3, Accelrys Inc. 2001, 9685 Scranton Road, San Diego, CA 92121-3752, USA.
152. Bernstein, J. Polymorphism in Molecular Crystals, IUCr Monographs on Crystallography; Oxford University Press: New York, 2002.
153. Berziņs, A.; Skarbulis, E.; Actiņs, A. Cryst. Growth Des. 2015, 15 (5), 2337–2351.
154. Kawahata, M.; Yamaguchi, K.; Ishikawa, T. Cryst. Growth Des. 2005, 5, 373–377.
155. Yamazaki, K.; Watanabe, A.; Moroi, R.; Sano, M. Acta Crystallogr. Sect. B 1981, 37, 1447–1449.
156. Dudenko, D. V.; Yates, J. R.; Harris, K. D. M.; Brown, S. P. CrystEngComm 2013, 15 (43), 8797–8807.
157. Dudenko, D. V.; Williams, P. A.; Hughes, C. E.; Antzutkin, O. N.; Velaga, S. P.; Brown, S. P.; Harris, K. D. M. J. Phys. Chem. C 2013, 117 (23), 12258–12265.
158. March, A. Z. Kristallogr. 1932, 81, 285–297.
159. Dollase, W. A. J. Appl. Crystallogr. 1986, 19, 267–272.
160. Pickard, C. J.; Mauri, F. Phys. Rev. B 2001, 63 (24), 2451011–2451013.
161. Yates, J. R.; Pickard, C. J.; Mauri, F. Phys. Rev. B 2007, 76, 024401.
162. Harris, R. K.; Hodgkinson, P.; Pickard, C. J.; Yates, J. R.; Zorin, V. Magn. Reson. Chem. 2007, 45, S174–S186.
163. Llewellyn, P. L.; Garcia-Rates, M.; Gaberová, L.; Miller, S. R.; Devic, T.; Lavalley, J.-C.; Bourrelly, S.; et al. J. Phys. Chem. C 2015, 119 (8), 4208–4216.
164. Boccaleri, E.; Carniato, F.; Croce, G.; Viterbo, D.; van Beek, W.; Emerich, H.; Milanesio, M. J. Appl. Crystallogr. 2007, 40, 684–693.
165. Palin, L.; Conterosito, E.; Caliandro, R.; Boccaleri, E.; Croce, G.; Kumar, S.; Van Beek, W.; Milanesio, M. CrystEngComm 2016, 18 (31), 5930–5939.
X-ray Powder Diffraction 73

166. Brooks, L.; Brunelli, M.; Pattison, P.; Jones, G. R.; Fitch, A. N. IUCrJ 2015, 2, 490–497.
167. Avila, E. E.; Mora, A. J.; Delgado, G. E.; Contreras, R. R.; Rincon, L.; Fitch, A. N.; Brunelli, M. Acta Crystallogr. B 2009, 65, 639–646.
168. Avila, A. E. E.; Mora, J.; Delgado, G. E.; Contreras, R. R.; Fitch, A. N.; Brunelli, M. Acta Crystallogr. B 2008, 64, 217–222.
169. Smrcok, L.; Brunelli, M.; Boca, M.; Kucharíka, M. J. Appl. Crystallogr. 2008, 41, 634–636.
170. van de Streek, J.; Neumann, M. A. Acta Crystallogr. B 2010, 66, 544–558.
171. van de Streek, J.; Neumann, M. A. Acta Crystallogr. B 2014, 70, 1020–1032.
172. Neumann, M. A.; Tedesco, C.; Destri, S.; Ferro, D. R.; Porzio, W. J. Appl. Crystallogr. 2002, 35, 296–303.
173. Smart, P.; Mason, C. A.; Loader, J. R.; Meijer, A. J. H. M.; Florence, A. J.; Shankland, K.; Fletcher, A. J.; Thompson, S. P.; Brunelli, M.; Hill, A. H.; Brammer, L. Chem. Eur. J.
2013, 19, 3552–3557.
174. Zhang, J.; Chen, S.; Valle, H.; Wong, M.; Austria, C.; Cruz, M.; Bu, X. J. Am. Chem. Soc. 2007, 129 (46), 14168–14169.
175. Kresse, G.; Hafner, J. Phys. Rev. B 1993, 47 (1), 558–561.
176. Kresse, G.; Hafner, J. Phys. Rev. B 1994, 49, 14251–14269.
177. Kresse, G. Phys. Rev. B 1996, 54 (16), 11169–11186.
178. Feynman, R. P. Eng. Sci. 1960, 23 (5), 22–36.

Relevant Websites

http://www.iucr.org/publ/50yearsofxraydiffractiondOn occasion of the 50th anniversary of Max von Laue’s discovery of the diffraction of X-rays by crystals a book collected the
personal reminiscences of the first crystallographers. Those vivid accounts are made freely available to the new generations of scientists by the IUCr.
http://www.iucr.org/resources/other-directories/softwaredA full list of crystallographic programs is available at the IUCr website, the list can be searched for XRPD software.
http://www.cristal.org/dArmel Le Bail’s site on Structure Determination from Powder Diffraction.

The websites of main crystallographic databases are reported in the following:


http://www.ccdc.cam.ac.ukdCambridge Crystallographic Data Center.
http://icsd.fiz-karlsruhe.dedInorganic Crystal Structure Database.
http://www.crystallography.net/cod/dCrystallography Open Database.
http://cds.dl.ac.uk/cgi-bin/news/disp?crystmetdCRYSTMET.
http://www.rcsb.org/pdbdProtein Data Bank.
2.05 Solid-State NMR of Supramolecular Materials
A Comotti, S Bracco, and P Sozzani, University of Milano-Bicocca, Milan, Italy
Ó 2017 Elsevier Ltd. All rights reserved.

2.05.1 Introduction 75
2.05.2 Structure Determination of Supramolecular Frameworks 75
2.05.3 Systems With Two or More Components 78
2.05.3.1 Host–Guest Supramolecular Adducts 78
2.05.3.2 Cocrystals 80
2.05.3.3 Polymers in Supramolecular Crystals 82
2.05.4 NMR of Gases in Porous Supramolecular Systems 87
2.05.4.1 Hyperpolarized 129Xe NMR 87
13
2.05.4.2 C and 1H NMR of Confined CO2, CH4, and H2 88
2.05.5 Molecular Rotors and Dynamics in Supramolecular Architectures 92
2.05.6 Conclusions 97
References 97

2.05.1 Introduction

Supramolecular materials comprise assemblies of molecular units that can form discrete entities or extended networks of both
organic and hybrid nature. Therefore, a diversity of spin-active nuclei, typical of organic and inorganic moieties, can be exploited
for the observation of topical parts of supramolecular materials by solid-state NMR spectroscopy. These multiple spin-active nuclei
1
H, 6Li, 7Li, 11B, 13C, 15N, 19F, 23Na, 29Si, and 31P and higher atomic mass atoms are distributed in the structure and, through their
diagnostic NMR resonances, behave as descriptors of their immediate surroundings.1–3 The interplay among abundant, for example,
hydrogen, phosphorous, and fluorine, and rare nuclei, for example, carbon or silicon, which label the molecular entities, allows the
performance of rich and informative 2-D homo- and heteronuclear correlated experiments describing their spatial relationships.
Thus, the response of solid-state NMR is extended to several observation points, which probe the individual phases or the individual
components in binary systems, such as host–guest compounds, and their reciprocal interactions.
Although supramolecular materials are generated by self-assembly processes or by reaction of properly designed multifunctional
molecules arranged in an ordered fashion, amorphous materials of great interest have been fabricated. 3 The solid-state NMR
response to amorphous phases and crystalline phases represents a further reason for completeness of such spectroscopy in charac-
terizing supramolecular materials. Solid-state NMR spectroscopy is intrinsically a supramolecular method because, unlike solution
NMR, it is sensitive not only to the electronic environment of the observed nuclei within the molecule thanks to through-bond
interactions but also to interactions occurring through the space by the exploitation of dipolar couplings. Indeed, several solid-
state NMR experiments are sensitive to the magnetization transport and distances between nuclei, thus enabling the precise location
of the next neighbors of an observed nucleus, irrespective of the homogeneous or hybrid nature of the interface.
Additionally, solid-state NMR of supramolecular systems can establish, by a quantitative response, the compound composition
and possible stoichiometry without the need of solubilizing the material and avoiding artifacts, which may be due to differential
solubility of the components as well as of impurities.
A favorable playground of solid-state multinuclear NMR is that of host–guest materials wherein two components interact inti-
mately: one behaves as the matrix that accommodates the guest. In molecular crystals, these materials are named “inclusion
compounds.” However, in some instances, the guests are partly or totally removed without a dramatic change of the host structure.
These permanently porous materials, which show open porosity, possess accessible apertures, mostly shaped as 1-D channels or
galleries, through which a guest can flow, paving the way for the direct observation of gases confined in a solid porous material.
In the following overview, we are depicting a series of significant results obtained by solid-state NMR in supramolecular chem-
istry along with the most robust NMR techniques useful in the field.

2.05.2 Structure Determination of Supramolecular Frameworks

Supramolecular arrangement of the constituents can be described by the measurement of interatomic distances among nonbonded
nuclei sitting in neighboring molecules in the structure through a few experiments such as cross polarization (CP) dynamics and
rotational-echo double resonance, transfer of population in double resonance (TRAPDOR), etc. 3 1H–13C 2-D heterocorrelated
spectra, performed with Lee–Goldburg decoupling and high spinning speed, showed generally high resolution in both hydrogen
and carbon domains.4–7 The interplay of 1H and 13C nuclei in the 2-D heteronuclear correlation NMR measurements enables
the detection of the spatial relationships (< 5 Å) between the spin-active nuclei. The magnetization transfer from the excited 1H

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12494-1 75


76 Solid-State NMR of Supramolecular Materials

to the rare 13C nuclei is guaranteed by nuclear dipolar interactions under the Hartman–Hahn conditions. The magnetization
buildup depends upon the duration of such communication conditions, called contact time or, simply, CP time, and the time
constant TCH depends on the dCH distance according to the following equation:8,9
   
hm0 2 p1:5 gC gH TCH sC
hdi6 ¼
4p 80
Metal–organic frameworks (MOFs) offer interesting applications of the aforementioned 2-D NMR technique because the
arrangement of the organic ligands may be complex and the association with the metal nodes is a key point for understanding
the overall structure. 10 Furthermore NMR, unlike XRD, is sensitive to both crystalline and amorphous phases, thus enabling a careful
check of the purity of the porous materials. By modulating CP times, it is possible to define the connectivity of the framework on
distinct length scales. As a case study, we consider an aluminum 1,4-naphthalenedicarboxylate Al(OH)(1,4-NDC) compound that
exhibits a three-dimensional framework composed of infinite chains of corner-sharing octahedral Al(OH)2O4 with 1,4-
naphthalenedicarboxylate ligands and forms two types of square-shaped channels (Fig. 1).11 In the 2-D 1H–13C heterocorrelated
(HETCOR) NMR spectrum with a contact time as short as 50 ms, the correlations between hydrogens and their covalently bonded
carbons are apparent. Due to the remarkable resolution achieved by the Lee–Goldburg decoupling, the three aromatic hydrogens
H7, H2, and H8 can be singly detected at dH ¼ 6.1, 7.0, and 9.4 ppm, respectively: a necessary condition for structural information
being achieved. At contact times as short as 0.5 ms, carbons not directly bonded to hydrogens and sitting at distances shorter than

Figure 1 (A) Structural motif of the nanoporous Al(OH)(1,4-NDC) compound. (B) 2-D 1H–13C Lee–Goldburg HETCOR NMR spectra of the
compound at different contact times. Adapted with permission from Comotti, A.; Bracco, S.; Sozzani, P.; Horike, S.; Matsuda, R.; Chen, J.;
Takata, M.; Kubota, Y.; Kitagawa, S. J. Am. Chem. Soc. 2008, 130, 13664–13672. Copyright (2008) American Chemical Society.
Solid-State NMR of Supramolecular Materials 77

2.3 Å are included in the correlation pattern, as well. Thus, the nonbonded correlations within naphthalene building blocks
emerged. Carboxylic carbons receive the magnetization from the aromatic hydrogens only at contact times longer than 1 ms, partic-
ularly from hydrogens placed at a distance of 2.8 Å center-to-center. At contact times of 5 ms, the correlations involve the OH groups
sitting on the oxygens of the inorganic Al–O–Al chains. Therefore, the 2-D NMR spectra depict the internuclear distance pattern in
the structure and demonstrated that the OH groups are in close relationship with the organic moieties. 2-D MAS NMR progressively
enlarged the observation domains with increasing contact times and elucidated the highly symmetrical crystal structure that presents
parallel and independent nanochannels without interpenetration and guests entrapped therein.
By combining 1H solid-state NMR spectroscopy and powder X-ray diffraction with density functional theory (DFT) chemical
shift calculations, it was possible to solve the supercell crystal structure of a substituted imidazolate MOF endowed with a cavity
of 7.5 Å, called SIM-1. 12 In the 1H MAS spectrum, two distinct resonances were detected for each imidazolate hydrogen with
a 1:1 intensity ratio, suggesting the presence of two distinct local environments within the asymmetric unit. 1H–13C INEPT and HET-
COR spectra with (decoupling using mind-boggling optimization) DUMBO homonuclear decoupling revealed the connectivity
between the three types of protons within a given imidazolate, while 2-D double-quantum (DQ)–single-quantum (SQ) proton
correlation spectrum (60 kHz spinning speed) depicted the correlations between protons belonging to the two distinct imidazolate
moieties (Fig. 2). The strong correlations between protons pertaining to two imidazolate units (highlighted in red) pointed out the
spatial proximity between the two structurally distinct linkers, providing a powerful geometrical constraint on the structure. The
atomic-scale geometry of SIM-1 was elucidated considering six possible basic crystal structures on which the DFT calculations of

Figure 2 (A) The sodalite topology (left), the framework of SIM-1 (middle), and the substituted imidazolate linker (right). (B) 1H NMR spectrum of
SIM-1: connectivity between protons of a given imidazolate linker is indicated schematically in blue and orange. (C) 2-D DQ–SQ 1H spectrum of SIM-
1, recorded with a rotor-synchronized BAck-to-BAck (BABA) scheme for excitation and reconversion blocks. Correlations between the protons of the
two different kinds of linkers are indicated in red, while connectivity between protons belonging to the same type of units is shown in blue and
orange. Adapted with permission from Baias, M.; Lesage, A.; Aguado, S.; Canivet, J.; Moizan-Basle, V.; Audebrand, N.; Farrusseng, D.; Emsley,
L. Angew. Chem. Int. Ed. 2015, 54, 5971–5976. Copyright (2015) John Wiley & Sons.
78 Solid-State NMR of Supramolecular Materials

Figure 3 13C CP MAS NMR spectra of the methylene regions of (A) MgCl2 2.75EtOH and (B) MgCl2 1.5EtOH compounds. Schematic representation
of the octahedral coordination of the MgII sites with a number of EtOH (L) ligands: n ¼ 1–4. (C) 13C CP MAS NMR spectra of the methylene regions
of pure compounds and their mixtures. Adapted with permission from Sozzani, P.; Bracco, S.; Comotti, A.; Simonutti, R.; Camurati, I. J. Am. Chem.
Soc. 2003, 125, 12881–12893. Copyright (2003) American Chemical Society.

1
H chemical shifts were carried out. Although none of the predicted structures matched the 1H NMR spectrum, a sum of two 1H
NMR spectra belonging to two distinct structures was in agreement with the experimental spectrum, suggesting the occurrence of
a supercell structure, which was confirmed by the comparison of the calculated PXRD pattern with the experimental one. Thus,
solid-state NMR contributes substantially to the solution of a complex crystal structure in the absence of single-crystal data.
Coordination adducts may comprise the association of metal cations with neutral molecules through electron-donating atoms,
that is, oxygen or nitrogen. Since such molecules do not participate in the overall charge balance, a dominant reason for the forma-
tion of a single stoichiometry is removed, and this entails the formation of a number of crystalline compounds and pseudo-
polymorphs. 13 This is an ideal playground for solid-state NMR, which could successfully distinguish and quantify mixtures of
crystalline compounds. For instance, ethanol molecules associate easily with MgCl2 and form crystalline adducts that can contain
up to six ethanol molecules per Mg2 þ ion. Lower ratios of ethanol-to-Mg lead to crystalline adducts that are elusive to XRD crys-
tallography because of the intrinsic difficulty of detecting an excess or deficiency of the organic ligand with respect to a certain
pseudo-polymorph. To remedy this, homonuclear and heteronuclear 2-D correlated solid-state NMR spectroscopy identified the
pure compounds with given numbers of ligands and the ethanol-to-magnesium coordination pattern. By varying systematically,
the composition of mixtures of ethanol with MgCl2, 13C MAS NMR spectra highlight the spectral profiles of the stoichiometric
compounds (MgCl2 $ 1.5EtOH and MgCl2 $ 2.75EtOH) and their variable proportions in the intermediate compositions (Fig. 3).
The multiplicity and internal ratios among the signals in the pure stoichiometric adducts enable the identification of the coordina-
tion sites in the asymmetric unit of the crystalline cell. Additionally, it enables the design of the adducts by crystal engineering start-
ing from an exact stoichiometry. As a matter of fact, the use of NMR is a special tool for crystallography to be exploited in difficult
cases, especially when large and low symmetry unit cells are examined, and is very attractive because it provides a number of
constraints that may be integrated into the structure hypothesis.14–19

2.05.3 Systems With Two or More Components


2.05.3.1 Host–Guest Supramolecular Adducts
Supramolecular crystals may contain two organic components in their structures that can be self-assembled by specific or nonspe-
cific interactions. In the latter case, they are classified as clathrates and one of two components can be recognized as the host because
it is mostly responsible for the overall stability of the structure. The juxtaposition, distances, and interactions between host and guest
are clearly identified by multinuclear solid-state NMR.
Solid-State NMR of Supramolecular Materials 79

Figure 4 (A) 2-D 1H–13C HETCOR NMR spectra of guanidinium chlorobenzenesulfonate-mesitylene compound with 0.5 and 5 ms. (B) Shielding
region about the aromatic ring and topology of the mesitylene methyl group located above the aromatic plane in the cage. (C) A methyl group of one
of the mesitylene guests enclosed in one of the three clip sites within the cage. (D) Inclusion compound: view of mesitylene guest embedded in the
cage (left) and view perpendicular to the channel axis illustrating the arrangement of mesitylene guests in the channel (right). Adapted with permis-
sion from Comotti, A.; Bracco, S.; Sozzani, P.; Hawxwell, S. M.; Hu, C.; Ward, M. D. Cryst. Growth. Des. 2009, 9, 2999–3002. Copyright (2009) Amer-
ican Chemical Society.

Organic crystalline host frameworks can exhibit a diverse range of guest habitats, which can exert considerable influence on the
properties of the confined guests and their corresponding inclusion compounds.20–27 Consequently, the characterization of the
local environment in which guest molecules reside is essential for understanding various aspects of their behavior under confine-
ment and any resulting properties. The extraordinary resolution of NMR spectroscopy enables detection of small differences in
chemical environments that arise from structural differences in polymorphs or guest locations and orientations in statically or
dynamically disordered structures.26–30 In a particular case, the unusual dual properties of identical guest molecules confined in
two crystallographically distinct host cavities within single crystals based on hexagonal frameworks comprising guanidinium
ions and organomonosulfonates are revealed by multinuclear 1H–13C HETCOR NMR spectroscopy ( Fig. 4).31
The environments of the two cavities differ substantially, as one is lined by the highly polar guanidinium sulfonate network,
while the other is defined by walls consisting of nonpolar aromatic groups. Interestingly, two separate 13C signals of equal intensity
were observed for the guest–mesitylene methyl groups. The effect of these different environments on the NMR properties of the
guest molecules is evident from chemical shift data and 2-D HETCOR spectra. The large magnetic susceptibility effect due to
ring currents of the aromatic hosts enables determination of the host–guest distances and suggests intermolecular CH/p interac-
tions. Also, 13C relaxation times are differentiated for the same guest in the two nanoscale environments, and, thus, the various
NMR parameters merge to depict the overall picture of the crystal structure.
The application of the principles of self-assembly and crystal engineering permitted the manipulation of polyconjugated light-
active molecules through new interactions and the shaping of specific nanoscale environments.32–34 Organized materials used as
hosts are of particular interest, since they can transfer the benefits of their peculiar topological properties by forming inclusion
compounds with the targeted guests. Oligothiophenes with long alkoxy moieties appended to their ends were encapsulated in infin-
ite nanochannels lined with aromatic groups of tris-o-phenylenedioxycyclotriphosphazene (TPP). The favorable arrangement can
be recorded by 2-D HETCOR MAS NMR spectra that highlight not only CH/p interactions in the structure but conformational
arrangements of the chains in the properly fitting channels. The 2-D cross signals are a clear demonstration of the intimate relation-
ships of the guest with the host aromatic groups, while the proton resonances shifted 2 ppm upfield for the guests indicate the
topology of the guest hydrogens above the planes of benzene rings at a distance of 2.5 Å ( Fig. 5).35–37 The conformational arrange-
ment of the alkoxy chains differs from that of the bulk: conformations contribute to additional upfield shifts in the 13C domain,
even reaching a shift of 4 ppm with respect to the bulk with all trans conformations. The large shift must be ascribed to the
80 Solid-State NMR of Supramolecular Materials

Figure 5 2-D 1H–13C HETCOR PMLG NMR spectra of (A) pure 4,40 -dipentoxy-2,20 -dithiophene and (B) inclusion compound with TPP. (C) A view of
the molecules in the monoclinic unit cell. (D) Schematic picture and (E) 1H MAS NMR spectrum of the inclusion compound. Adapted with permission
from Sozzani, P.; Comotti, A.; Bracco, S.; Simonutti, R. Angew. Chem. Int. Ed. 2004, 43, 2792–2797. Copyright (2004) John Wiley & Sons.

g-gauche shielding effect added with the same sign to the magnetic susceptibility.38 In the nanostructured environment, departure
from trans to gauche conformations is largely tolerated, since the guests are softly embraced within an organized environment.

2.05.3.2 Cocrystals
Molecules containing supramolecular synthons can self-assemble and form multicomponent crystals (cocrystals) with a specific
stoichiometry and geometry based on intermolecular interactions. 39 Single-crystal and powder X-ray diffractions are key methods
to determine the 3-D structure, but it may be difficult to grow single crystals or solve the crystal structure from PXRD. Advanced
multinuclear MAS NMR spectroscopy is a powerful tool for identifying the intermolecular interactions that direct the self-
assembly in the solid state and for demonstrating the intimate molecular proximity of the components in the cocrystal.40 In the
case of indomethacin–nicotinamide (IND–NIC), cocrystal internuclear distances were measured by 2-D 1H DQ and heteronuclear
MAS NMR spectra demonstrating the intimate relationships between the two components and the intermolecular hydrogen
bonding network (Fig. 6).41 In particular, 1H DQ spectrum directly showed the proton–proton proximities, while the 14N–1H het-
eronuclear spectra proved the close contact between the IND COOH and a NIC aromatic CH proton and, with a longer recoupling
time, between the NIC nitrogen and the IND COOH groups. These results unambiguously demonstrated the intimate relationships
of the IND and NIC molecules in the cocrystal via Narom/HOOC and CHarom/O¼ C hydrogen bonds, consistent with the forma-
tion of the most favorable heterosynthons (acid–amide, acid–pyridine, and hydroxyl–pyridine) over the homosynthons (acid–acid
and amide–amide), as suggested by Etter.42 Furthermore, solid-state NMR spectroscopy substantially contributes to the identifica-
tion of the correct crystal structure. In fact, 1H and 13C NMR chemical shifts and 2-D 1H–13C correlations of directly bonded CeH
moieties in the cocrystal were successfully reproduced by DFT calculations of the crystal structure proposed by PXRD.43
The recent development of in situ solid-state NMR spectroscopy has enabled the study of the crystallization processes from solu-
tion, particularly the evolution of different polymorphic forms as a function of time during crystallization. This challenging tech-
nique involves the preparation of a homogeneous solution inside the NMR rotors and the time resolution of the in situ monitoring
of the crystallization phenomenon by repeatedly recording high-resolution solid-state NMR spectra as a function of time. 44 Good
spectral resolution and high sensitivity of the signals are prerequisites for a successful experiment. Isotopic labeling and high
magnetic fields may contribute substantially to the identification and resolution of the structures. By this method, it was possible
to follow the hydrogen-bonded cocrystal formation between even-chain a,u-dihydroxyalkanes and urea in the 1:2 molar ratio. The
crystal structures contain double-stranded hydrogen-bonded ribbons of urea molecules that can be parallel or antiparallel and can
form either acute or obtuse angles between the axis of the dihydroxyalkyl chains and the positive direction of the urea strand
(Fig. 7). The use of 13C-labeled urea allowed unambiguous identification in the 13C NMR spectrum of the urea carbon signal
that is sensitive to the distinct polymorphs. In situ 13C solid-state NMR spectra of the crystallization process between
Solid-State NMR of Supramolecular Materials 81

Figure 6 Solid-state NMR spectra of an IND–NIC cocrystal: (A) 2-D 1H DQ MAS spectrum recorded using one rotor period of BABA recoupling. 2-D
14
N–1H HMQC spectra recorded using n ¼ 2 rotary resonance recoupling of 14N–1H heteronuclear dipolar couplings for (B) s ¼ 130 ms and (C)
s ¼ 670 ms. (D) 1H (SQ-DUMBO)–13C SQ refocused INEPT correlation spectrum, recorded with the insensitive nuclei enhanced by polarization trans-
fer (INEPT) spin-echo durations s ¼ 1.28 ms. Reproduced with permission from Maruyoshi, K.; Iuga, D.; Antzutkin, O. N.; Alhalaweh, A.; Velagad, S.
P.; Brown, S. P. Chem. Commun. 2012, 48, 10844–10846. Copyright (2012) Royal Society of Chemistry.
82 Solid-State NMR of Supramolecular Materials

Figure 7 (A) Schematic representation of the structure types for cocrystals with urea. (B) In situ 13C solid-state NMR spectra as a function of time
during the cocrystallization from methanol. Adapted with permission from Harris, K. D. M.; Hughes, C. E.; Williams, P. A. Solid State Nucl. Magn.
Reson. 2015, 65, 107–113. Copyright (2015) Elsevier Inc.

13
C-labeled urea and 1,10-dihydroxydecane revealed, in the early stages, the formation of a transient solid form with a parallel/acute
structure type (d ¼ 164.0 ppm), which transforms into an antiparallel/obtuse structure cocrystal phase (d ¼ 164.6 ppm) after a longer
time. Interestingly, in addition to the discovery of new polymorphs, this technique provides the time window during which the
polymorphs form during the crystallization process, enriching our knowledge of the crystal growth mechanisms and the complex
evolutions of crystal forms.

2.05.3.3 Polymers in Supramolecular Crystals


High molecular mass polymers embedded in nanochannels and nanosheets is a challenging research field since the self-assembly of
molecular components of vastly different masses and conformational flexibilities enables fabrication of innovative nanostructured
materials and nanocomposites endowed with unusual properties. 45 Flexible polymer chains can be efficiently entrapped and
aligned into crystalline structures by the action of supramolecular hosts that provide the stabilization energy through weak inter-
molecular interactions and favorable overall packing.46–50 The self-assembly phenomenon occurs even in the presence of consider-
able loss of entropy during the ordering of the macromolecules within the architecture. However, 13C T1 relaxation and 2H line
shape analysis show diffusional reorientation about the polymer axis, modeled as complete averaging for fast in-plane rotations
or librations of CeH bonds, which attenuates the expected entropy loss upon inclusion.2
High molecular mass polyethylene (PE) included in crystalline nanochannels of TPP exhibits upfield shifts in both the 1H and
13
C domains of about 2 ppm, indicating that the aromatic rings of TPP exert a diamagnetic susceptibility effect by directly facing the
polymer in the nanochannels ( Fig. 8).35,51,52 Such a shift indicates that the polymer nuclei sit at short distance from the channel
walls lined with aromatic rings and the occurrence of guest–host CH/p interactions. Strikingly, in a sample with excess PE, it is
possible to distinguish both the included and free components of PE. In the 2-D 1H–13C NMR spectrum, the included PE partic-
ipates in a complex correlation system due to mutual host–guest relationships, while free the PE cross signal remains isolated
because the entire population of the polymer chains is packed in the bulk phase.
Even rubbery polymers can be forced to form a crystalline phase. While the macromolecules in the low-Tg bulk polymer dynam-
ically explore multiple conformations, the macromolecules in the crystalline adducts are compelled by the host to assume an
extended chain conformation and align themselves in the parallel nanochannels. Quantitative 1H MAS NMR spectra recorded at
600 MHz with 35 kHz spinning speed revealed important features of the material. In particular, the complexity of samples obtained
by solvent-free mechanochemical treatment of excess 1,4-cis-polybutadiene (PB) and the host (TPP) can be disclosed: in fact, the
simultaneous identification of the crystalline, intercrystalline, and bulk amorphous phases was achieved ( Fig. 9).53 When the
amount of polymer corresponds to the complete filling of the channels, the 1H MAS NMR spectral profile is simplified and only
the upfield signals of CH and CH2 are detected. The ratio between host and guest components indicates that the polymer chains
are in the extended chain conformations inside the nanochannels. This information cannot be obtained by XRD analysis because
of the incommensurate property of the adduct and chain mobility.
Supramolecular self-assembly of the host molecule with selected blocks of triblock copolymers enabled the formation of
inclusion 2-D nanocrystals that connect consecutive copolymer chains. The selective inclusion of ethyleneoxide (EO) blocks
Solid-State NMR of Supramolecular Materials 83

Figure 8 (A) Schematic representation of the TPP molecule. (B) Porous crystal structure of TPP. (C) Modelling of in-plane rotation of a single
dCH2d group of PE in the nanochannel. 2D 1H–13C HETCOR NMR spectra of TPP/Pe adducts with (D) 1 ms and (E) 5 ms contact times. CH...p
interactions between the dCH2d moieties and p-system of TPP are highlighted in yellow. (F) Topology of guest hydrogens located at 2.5–2.7 Å
above the benzene plane. (G) Topology of guest hydrogens located at 2.5–2.7 Å above the plane of the benzene rings: the ring currents generate an
upfield shift of 2.2 ppm. Adapted with permission from Sozzani, P.; Comotti, A.; Bracco, S.; Simonutti, R. Chem. Commun. 2004, 40, 768–769. Copy-
right (2004) Royal Society of Chemistry.

in inclusion crystals and the phase segregation of PO blocks of poly(ethylene oxide-b-propylene oxide-b-ethylene oxide) EO–
PO–EO triblock copolymers provide an efficient route to create alternated crystalline lamellae and amorphous layers, forming
a well-organized material ( Fig. 10).54 The driving force for the fabrication of crystalline inclusion compounds with selected
EO segments is based on the establishment of cooperative noncovalent intermolecular interactions, while steric effects prevent
the formation of the inclusion crystal with the remaining PO blocks. The 2-D 1H–13C solid-state and fast-1H MAS NMR provide
direct evidence of the intimate interactions between the host and EO block and the topology of the block copolymer in the mate-
rial. On the contrary, the random copolymer exhibits a notable upfield shift for the protons of the methyl in the TPP/ran-
copolymer sample, unequivocally showing that methyl groups of isolated PO units are confined inside the channels. The large
magnetic susceptibility generated by the aromatic host nanochannels surrounding the included EO chains was interpreted by ab
initio calculations that carefully reproduce the chemical shifts associated with the effects of guest–host interactions. The theoret-
ical calculations enable the measurement of short intermolecular distances between the host and the target block, demonstrating
84 Solid-State NMR of Supramolecular Materials

Figure 9 (A) Schematic representation of the TPP/PB compound with an excess PB. 1H fast-MAS NMR (600 MHz; nr ¼ 35 kHz) spectra of TPP/PB
compounds: (B) with an excess PB and (C) with PB confined in the crystalline nanochannels. (D) 2-D 1H–13C and (E) 2-D 1H–31P HETCOR spectra of
TPP/PB compound. The intermolecular interactions are highlighted. Adapeted with permission from Bracco, S.; Comotti, A.; Valsesia, P.; Beretta, M.;
Sozzani, P. CrystEngComm 2010, 12, 2318–2321. Copyright (2010) Royal Society of Chemistry.

the existence of a diffuse network of multiple CH/p host–guest interactions that improve the robustness of the supramolecular
architecture.
Solid-state NMR can be used to study the conformation and dynamics of single macromolecules confined in highly regular one-
dimensional nanochannels of MOFs or PCPs. In particular, 2-D NMR experiments can provide a direct demonstration of the adduct
formation and the intimacy between the macromolecules and pore surfaces of the host. Interestingly, the supramolecular architec-
ture of MOFs with tuned channel sizes can modulate carrier mobility of confined polysilanes. 55 The extended host–guest interface
was strikingly demonstrated by the 2-D 1H–1H DQ CRAMPS NMR spectroscopy, performed at 600 MHz 1H Larmor frequency and
applying w-PMLG decoupling to achieve high resolution in the proton domain. This technique exploits the 1H–1H dipolar coupling
to probe through-space proximities between host and guest protons. In the 2-D 1H–1H spectrum of Fig. 11, the correlations between
Solid-State NMR of Supramolecular Materials 85

Figure 10 (A) Schematic representation of the poly(ethyleneoxide–propyleneoxide–ethyleneoxide) EO–PO–EO triblock copolymer. (B) Schematic
representation of the end A segments included in the nanochannels and the central B block excluded from the nanochannels. Fast-1H MAS (500 MHz,
nr ¼ 35 kHz) NMR spectra of (C) TPP/F108 and (F) TPP/random copolymer materials. The signals of the copolymer moieties confined in the nano-
channels of the inclusion crystals are indicated as EOin and POin, while those of nonincluded copolymer moieties are indicated as EOout and POout.
Adapted with permission from Bracco, S.; Comotti, A.; Ferretti, L.; Sozzani, P. J. Am. Chem. Soc. 2011, 133, 8982–8994. Copyright (2011) American
Chemical Society.

Figure 11 Schematic image for (A) nanochannel structures of MOF [Al(OH)L]n; (1a), L ¼ 2,6-naphthalenedicarboxylate; (1b) L ¼ 4,40 -biphenyldi-
carboxylate. (B) Molecular structure of polymethylpropylsilane. 2-D 1H–13C and 1H–29Si HETCOR spectra of MOF with polymethylpropylsilane PMPrS
(C,D). (E) 2-D 1H–1H DQ CRAMPS NMR (600 MHz, nr ¼ 13.6 kHz) spectrum of the nanocomposite. Host–guest correlations are indicated in orange.
Adapted with permission from Kitao, T.; Bracco, S.; Comotti, A.; Sozzani, P.; Naito, M.; Seki, S.; Uemura, T.; Kitagawa, S. J. Am. Chem. Soc. 2015,
137, 5231–5238. Copyright (2015) American Chemical Society.
86 Solid-State NMR of Supramolecular Materials

Figure 12 29Si MAS NMR spectra of polymethylpropylsilane in (A) narrow and (B) wide MOF channels. Conformational arrangement of the included
polymer as derived by molecular dynamics simulation in the (C) narrow and (D) wide channels. Adapted with permission from Kitao, T.; Bracco, S.;
Comotti, A.; Sozzani, P.; Naito, M.; Seki, S.; Uemura, T.; Kitagawa, S. J. Am. Chem. Soc. 2015, 137, 5231–5238. Copyright (2015) American Chemical
Society.

host Al(OH)(2,6-naphthalenedicarboxylate) with the narrow channels (8.5  8.5 Å2) and guest hydrogens emerge, providing
a direct observation of the polymer chains included in the host nanochannels. Moreover, fully relaxed 29Si MAS NMR experiments
detected the polymer main-chain nuclei as two signals resonating at about dSi ¼ 32 and 36 ppm (Fig. 12): The downfield signal
prevails when the polymethylpropylsilane is confined to the narrow channel, whereas the upfield signal prevails for the polymer in
the wider cavities (11.5  8.5 Å2) of the Al(OH)(4,40 -biphenyldicarboxylate). The backbone conformations adopted by the polymer
included in the two distinct-size pores are the origin of the 29Si chemical shift change. The stretched trans conformation prevails in
the narrow channel, which induced severe lateral confinement, thereby squeezing the polymer chain, while the extra room available
in the wider channels allows for the development of bulky main-chain conformations that explore frequent arrangements deviating
from trans. This conformational arrangement, determined by NMR, explains the increased carrier transport properties in the narrow
channels, which are thus manipulated by the supramolecular constraints.
The aforementioned examples pertain to polymers self-assembled in supramolecular adducts. Alternatively, it is possible to
generate confined polymer chains starting from monomers included into the host architectures and successive polymeriza-
tion.56–59 The resulting confined polymers and their fast dynamics within the crystalline environment have been addressed by
solid-state NMR since the late 1980s. 27,60–62 T1 relaxation measurements and 2H NMR spectroscopy, recorded at variable temper-
ature, have been the methods of choice to understand this unusual motional behavior (see in the succeeding text).
Recently, the use of MOFs has been successful in obtaining special features for vinyl polymers, such as polystyrene and polyme-
thylmethacrylate. 63,64 Typically, a single polymer chain can grow in each nanochannel, but it is possible to enhance the degree of
control over the resulting polymer morphology by multifunctional linkers (bearing divinyl and dicarboxylate groups) inserted in
the host walls (Fig. 13). These building blocks belong to the MOF supramolecular architecture and participate in the polymerization

Figure 13 (A) HRTEM image of polystyrene showing regular patterns with 4.9 Å periodicity attributed to regular polystyrene interchain spacing. (B)
Fast Fourier transform (FFT) of the HRTEM image reveals spots with d-spacings of 4.9, 3.6, and 2.4 Å, in agreement with a two-dimensional square
lattice. Adapted with permission from Distefano, G.; Suzuki, H.; Tsujimoto, M.; Isoda, S.; Bracco, S.; Comotti, A.; Sozzani, P.; Uemura, T.; Kitagawa,
S. Nat. Chem. 2013, 5, 335–341. Copyright (2013) Nature Publishing Group.
Solid-State NMR of Supramolecular Materials 87

process connecting two growing polymer chains in adjacent nanochannels. The direct evidence that the multifunctional linkers
copolymerized with styrene monomers resulting in a unitary structure was provided by 2-D 1H–13C HETCOR NMR spectra. Surpris-
ingly, the disassembly of MOFs by complexation of the metal nodes with EDTA produces a polystyrene that retains the crystalline
order, although atactic, thus constituting an imprint of the supramolecular structure.

2.05.4 NMR of Gases in Porous Supramolecular Systems


129
2.05.4.1 Hyperpolarized Xe NMR
NMR of diffusing atoms and molecules in supramolecular crystalline nanochannels is a challenging way to probe the size and shape
of the host cavities as well as the affinity of the gas to matrix walls. 65 In particular, 129Xe is a spin-1/2 nucleus and is particularly
suitable for exploring empty spaces in the crystalline materials. Its resonance value is sensitive to the size of the cavities, while the
chemical shift anisotropy (CSA) describes the cavity shape and is developed when the confined environment is comparable to the
Xe cross section (4.4 Å). Xe NMR spectroscopy also demonstrates the open accessibility of the channels and galleries. The sensitivity
of 129Xe NMR can be enhanced by orders of magnitude by the laser-assisted hyperpolarized technique (HP) that produces a signif-
icant xenon signal even at very low concentrations. Moreover, by the additional continuous flow technique, hyperpolarized Xe can
be continuously delivered to the sample and detected within a few milliseconds, allowing the demonstration of fast diffusion times
in samples endowed with permanent porosity.
The application of HP Xe NMR to molecular zeolites is a stimulating way to demonstrate directly the existence of empty cavities
in crystalline structures held together by soft interactions. The first example of such application in molecular zeolites is represented
by porous TPP crystals ( Fig. 14).66 The laser-polarized xenon gives rise to large anisotropic signals far from that for Xe in the gas
phase (at 0 ppm); the axially symmetrical CSA reflects the cylindrical geometry of the nanochannels with a cross section of about
5 Å. Indeed, within the limit of low dilution, Xe can rapidly diffuse along the channel axis but experiences the severe lateral
constraints of the channel walls.
Furthermore, a porous single crystal can be explored with xenon atoms whose chemical shift is diagnostic of the crystal orien-
tation. 67 When the nanochannels are aligned parallel to the main magnetic field, the Xe signal resonates downfield because xenon
experiences exclusively xenon–wall interactions (Fig. 14B). Conversely, when the nanochannels are placed perpendicular to the

Figure 14 (A) Continuous flow laser-polarized 129Xe NMR spectra of Xe/He mixtures at atmospheric pressure and room temperature of the porous
TPP crystalline powder. (B) Continuous flow laser-polarized 129Xe NMR spectra of Xe diffused into a single crystal inclined at different orientations
from 0 to 90 degree with respect to the magnetic field B0. Adapted with permission from Sozzani, P.; Comotti, A.; Simonutti, R.; Meersamnn, T.;
Logan, J. W.; Pines, A. Angew. Chem. Int. Ed. 2000, 39, 2695–2698. and Comotti, A.; Bracco, S.; Ferretti, L.; Mauri, M.; Simonutti, R.; Sozzani, P.
Chem. Commun. 2007, 43, 350–352. Copyright (2000 and 2007) John Wiley & Sons and Royal Society of Chemistry, respectively.
88 Solid-State NMR of Supramolecular Materials

main magnetic field, xenon can perceive Xe–Xe interactions that are very mild under the extreme Xe dilution; on increasing Xe pres-
sure, a downfield shift is observed due to the increase of Xe–Xe interactions.
In addition to the measurement of CSA, HP 129Xe NMR in a continuous flow can quantify polarized Xe uptake per unit time and,
consequently, diffusion coefficients in the materials. 68 In the case of crystals with narrow channels, Xe atoms are not permitted to
bypass one another and thus each atom has to wait for the advancement of the next atom in line to move on (single-file diffusion).
By CF HP 129Xe NMR spectroscopy, slow diffusive phenomena and diffusion times in the range of seconds could be measured. In
fact, the feed chamber supplies continuously polarized xenon that, after a series of 90 pulses, accumulates into the pores according
to the diffusion rates. The signal buildup can be followed with time and signal intensity dynamics recorded at variable temperature.
The perfect fit indicated that the model of the single-file diffusion, slower than the conventional mono-dimensional diffusion, was
operative under a wide range of conditions.
A similar approach has been applied to molecular crystals of biological origin that possess extended channels in their crystalline
structures. An extensive experimental and theoretical study provided significant insight into dipeptide crystals. 69,70 The main
interest derives from the peculiar feature that they can exist in a variety of structures with different pore sizes and helicities within
a given family. This resulted in a large diversity of chemical shift values observed as a function of the channel constriction. The
porous crystals of L-Ala-L-Val and L-Val-L-Ala were used to test single-file diffusion behavior in the narrow channels. Hyperpolarized
129
Xe NMR applying pulsed field gradient and 2-D exchange NMR could confirm single-file diffusion in molecular crystals, with
mobility factors of about 5  10 13 m2 s 1/2.71,72
HP 129Xe NMR enables the investigation of larger channels typically occurring in MOFs. However, in this case, the Xe line shape
does not develop an anisotropic signal because the gas atoms are less severely confined than in the previous cases. Al-based coor-
dination polymers are paradigmatic of the whole class of MOFs. 11 They constitute a complex case because in the crystal structure,
two types of channels with square-shaped cross sections are present (8  8 Å2 and 3  3 Å2). Xe can diffuse selectively into the large
nanochannels, while the small ones show no substantial uptake of xenon due to severe restrictions that prevent diffusion. The
hyperpolarized 129Xe NMR spectrum of the nanoporous compound, recorded at 295 K, revealed a narrow and symmetrical signal
at d ¼ 71.8 ppm (Fig. 15).
The absence of an anisotropic line shape showed that the channels were not narrow enough to impose their anisotropy on to the
Xe signal. From theoretical and experimental studies of Xe NMR in zeolites, it could be deduced that a chemical shift value of
72 ppm, under the extreme dilution limit, corresponded to a pore size of about 9 Å, which was consistent with the larger channels.
The compared evaluation of Xe chemical shifts and XRD analysis was a diagnostic tool to prove that the channels were free from
guests and from water molecules in the hygroscopic MOFs. Upon lowering the temperature, there was a downfield shift due to the
increased residence time of xenon on the internal surfaces. From the analysis of the two resonances belonging to the free gas and the
xenon exploring the nanochannels, the exchange dynamics of Xe between the confined space and the gas phase can be evaluated. In
the hyperpolarized 129Xe NMR spectrum at room temperature, the two signals resonate 5460 Hz apart, indicating that the exchange
time between the two states must be longer than 0.2 ms. To evaluate the exchange times on intermediate scales, the experiment of
HP 2-D Xe exchange NMR could be applied. 73 The experiment was performed at room temperature, with mixing times ranging from
1 to 50 ms. Already at mixing times as short as 15 ms cross peaks appear because free gas exchanges rapidly with xenon inside the
porous crystals, therefore the channels are open without restrictions and the apertures at the crystal surfaces are pervious.
The case presented is paradigmatic of the whole class of MOFs, as also highlighted in ZIFs and IRMOFs, which are imidazolate
frameworks and isoreticular MOFs, respectively. 74,75 The chemical shifts at room temperature were moderately moved downfield
with respect to the bulk gas, suggesting a limited confinement of the gas atoms in the lattice. By the application of HP exchange
spectroscopy (EXCSY) experiments, it was found that the xenon exchange between the adsorption sites and the free gas was
much slower than that among the adsorption sites within the IRMOF tunnels. The residence time in the crystal was long enough
and the pore nanospace large enough to allow for prolonged diffusion of polarized xenon in the galleries. Moreover, MOFs forming
cycles called molecular wheels, and assembled in nanotubes with open apertures, were addressed by 1-D and 2-D CG HP experi-
ments.76,77 Shape-persistent conjugated macrocycles self-assembled with cis-Pt(II) species, successfully fabricating porous solids. CF
HP 129Xe NMR was exploited to investigate the solid-state pores of the supramolecular complexes as a function of temperature.78

13
2.05.4.2 C and 1H NMR of Confined CO2, CH4, and H2
The direct observation of gas molecules diffused into nanocavities is a challenge for solid-state NMR spectroscopy. The stability of
the adducts with gases allowed easy manipulation of the samples and detection of the spectra of the gases in the crystals under
ambient conditions and in natural isotope abundance. Rare reports of methane observed by NMR spectroscopy have appeared
in the literature, with 13C NMR chemical shifts ranging from d ¼ 5 to 8 ppm: 79 in particular, methane stored in inorganic zeolites
with channel diameters of about 5 Å resonates at d ¼ 8 ppm. The resonance of methane stored in host TPP nanochannels appears,
as a sharp signal, at d ¼ 9.4 ppm relative to TMS, notably upfield to any resonance of methane (Fig. 16).80 The exceptional upfield
shift is explained by the large magnetic susceptibility effect of the aromatic matrix, showing unequivocally that methane fills narrow
crystal channels lined with aromatic groups. Even more impressively, the upfield shift was apparent in the 30 ppm total range of the
1
H NMR spectrum, which was notably resolved at the high spinning speed of 15 kHz: in this case, an intense signal of the stored
methane is detected at d¼2.0 ppm along with a weak signal for the gas-phase methane loaded in the sealed rotor at d¼0.3 ppm.
The quantitative analysis of the 1H NMR spectrum could give, independently, the amount of the gas absorbed in the organic zeolite,
confirming the high storage values obtained by the sorption isotherms. The tetrahedron of methane placed at the center of the
Solid-State NMR of Supramolecular Materials 89

Figure 15 (A) Continuous flow laser-polarized 129Xe NMR spectra of Al(OH)(1,4-NDC) MOF at variable temperatures: the peak at 0 ppm is due to
free Xe and the downfield peak to the xenon exploring the nanochannels. (B) Crystal structure of the MOF viewed along the channel axis. (C) Repre-
sentation of the shape of large and small nanochannels. Continuous flow hyperpolarized 2-D exchange hyperpolarized 129Xe NMR experiments
acquired at mixing times smix of (D) 1 ms and (E) 15 ms. Adapted with permission from Comotti, A.; Bracco, S.; Sozzani, P.; Horike, S.; Matsuda, R.;
Chen, J.; Takata, M.; Kubota, Y.; Kitagawa, S. J. Am. Chem. Soc. 2008, 130, 13664–13672. Copyright (2008) American Chemical Society.

channel behaves as a polydentate structure with a favorable topology for forming multiple interactions with the channel walls:
CH/p interactions can be established with two or even three host paddles at a time, resulting in a few stable configurations.
Some configurations are favored by symmetry, as three hydrogen atoms of a methane molecule are related by a C3-symmetry
element, as are the three host paddles at a channel cross section.
The 13C MAS NMR spectrum of the nanochannels containing CO2 shows, in addition to the matrix carbon atoms, a signal for
CO2 stored in the nanoporous crystals ( Fig. 16). However, to remove the overlapping of the CO2 signal with one of the carbon
atoms of the matrix, a nonquaternary suppression (NQS) experiment was applied in which suitable irradiation conditions sup-
pressed the protonated carbon atoms of the matrix. The filtered spectrum presents a neat signal at d ¼ 123.3 ppm, which was
assigned to the unprotonated carbon of CO2. The resonance of carbon dioxide stored in the nanochannels is found at the highest
field thus far reported in the literature: in particular, it appears more upfield (d¼ 133 ppm) than the signals for bulk solid CO2, CO2
associated with calixarenes and cucurbit[5]uril and free CO2 gas.81,82 The upfield shift denotes the tight proximity of the gas mole-
cules to the walls in the restricted aromatic environment and, in particular, reveals that the gas carbon atoms are located as closely as
possible to the “sticky” walls.
Quantum chemical calculations enable the determination of the through-space shielding effect generated by an aromatic moiety
on the methane molecule and relate the chemical shift displacements to the local structural features. A methane molecule was
placed at 240 specific positions with a distance ranging from 2.5 to 5.5 Å about the phenylenedioxyphosphate moiety. The geom-
etries of the phenylenedioxyphosphate and methane were optimized at the B3LYP/6-311G(d,p) level. Hartree-Fock Gauge Indepen-
dent Atomic Orbital HF-GIAO calculations of the NMR isotropic shielding value of the methane hydrogen in the distinct geometries
mapped the through-space shielding effects shows the resulting maps of the calculated proton NMR chemical shifts of methane
90 Solid-State NMR of Supramolecular Materials

Figure 16 13C and 1H MAS NMR spectra of CH4 confined in the van der Waals crystals (A,B). (C) 13C MAS NMR spectrum obtained by applying the
nonquaternary suppression filter and (D) single-pulse 13C MAS NMR spectrum of stored CO2. (E) A methane molecule at a particular site along the
nanochannel of TPP and the close-contact interactions with the aromatic units facing the channel walls. Adapted with permission from Sozzani, P.;
Bracco, S.; Comotti, A.; Ferretti, L.; Simonutti, R. Angew. Chem. Int. Ed. 2005, 44, 1816–1820. Copyright (2005) John Wiley and Sons.

relative to the isolated molecule ( Fig. 17). The aromatic ring generates the highest through-space magnetic shielding effect above
and below the planar ring, while a weaker and reverse effect is observed beside the ring. Thus, the contour map enables an accurate
determination of the distance of methane hydrogens from the aromatic rings. In particular, the experimental shielding shift
Dd ¼ 2.0 ppm of methane hydrogens is consistent with the center-to-center distance of 2.9 Å between guest hydrogen and the
aromatic ring. Additionally, the shielding surface (reported as Dd) in the XY plane and at a distance of 2.9 Å above the aromatic
plane reinforces the notion that the guest hydrogens lie above the center of the aromatic ring.
Dipeptide molecules can self-assemble from solution into crystals that can be totally freed from the guest solvent to yield porous
crystals. 83,84 The hydrophobic dipeptides bear alkyl residues that point toward the center of channel-like pores and modulate the
channel cross section. An interesting feature of this family of materials is that they can be of natural origin but also composed of
synthetic amino acids to form hybrid dipeptides. The diversification of the amino acid building blocks allows fine-tuning of channel

Figure 17 (A) Computed isoshielding surfaces perpendicular to the phenyledioxyphosphate plane (lateral view) and (B) shielding increment surfaces
of the phenyledioxyphosphate molecule (top view) computed at 2.9 Å above the aromatic plane. The color code reflects the chemical shift change Dd
of the methane hydrogen placed at a particular position relative to the aromatic ring, as compared with the chemical shift of an isolated molecule.
The phenylenedioxyphosphate molecule is represented with van der Waals radii (A) and ball and stick (B). 54 Adapted with permission from Bracco,
S.; Comotti, A.; Ferretti, L.; Sozzani, P. J. Am. Chem. Soc. 2011, 133, 8982–8994. Copyright (2011) American Chemical Society.
Solid-State NMR of Supramolecular Materials 91

Figure 18 (A) Schematic representation of CO2 molecules diffused in porous molecular crystals of L-norvalyl–L-valine (Nva–Val). (B) 2-D 1H–13C
MAS NMR of Nva–Val loaded with 13C-enriched CO2, as recorded at 240 K. In the hydrogen domain, the trace of CO2 resonance and, in the carbon
domain, the 1-D 13C CP MAS spectrum are reported (right). Adapted with permission from Yadav, V. N.; Comotti, A.; Sozzani, P.; Bracco, S.; Bonge-
Hansen, T.; Hennum, M.; Görbitz, C. H. Angew. Chem. Int. Ed. 2015, 54, 15684–15688. Copyright (2015) John Wiley & Sons.

cross sections by a systematic change of the side chains. This feature enables a proper fitting for small gas molecules such as CO2,
CH4, and H2.84
Direct observation of CO2 molecules inside the channels was provided by 1-D and 2-D 1H–13C MAS NMR experiments with CP
from hydrogen to carbon nuclei that reside at close distances for a residence time long enough to allow the magnetization transfer
(in the order of milliseconds). 83,85 Strikingly, 1H–13C CP experiments, performed at low temperature, are effective in highlighting
the CO2 carbon signal, even if it is depleted of hydrogens. CO2 undergoes anisotropic motions, since isotropic tumbling would
average out the dipolar interactions necessary for the magnetization transfer to occur efficiently. A further support to this conclusion
is the observation that the 13C MAS NMR spectra, recorded at relatively low spinning speed, showed the presence of CO2 rotational
echoes, which are residues of the CSA profile (not shown). Indeed, the 13C CP MAS NMR spectrum of a porous dipeptide loaded
with 13C-enriched CO2 (recorded at 240 K) exhibits an intense signal at d ¼ 125.6 ppm due to CO2 adsorbed in the channels
(Fig. 18). As CO2 does not possess hydrogens, the sole source of magnetization is that conveyed from host hydrogens to CO2
carbons in the surrounding space. Moreover, 2-D 1H–13C HETCOR experiments of Nva–Val/CO2 show cross signals, demonstrating
close-contact interactions of the host aliphatic and ammonium hydrogens with CO2 carbons. Such experiments prove unequivo-
cally the intimacy of the adsorbed gas and the porous solid at distances shorter than a few Å and the confinement of CO2 in the
nanochannels.
The combined use of soft interactions and metal–organic bonds provides a suitable tool to construct crystals made by hexameric
ring-shaped supramolecules. The self-assembly process implies the participation of counteranions to fabricate crystal structures con-
taining both cagelike and interstitial cavities. 86,87 The steric encumbrance of lateral moieties of organic ligands allows modulation
of the cage dimensions. The intense signal of CO2 in the carbon spectrum, as observed in Fig. 19, is a clear demonstration of CO2
captured by the crystalline structure and of the open porosity of the framework. To demonstrate the CO2 diffusion into the cages, the
2-D 1H–13C NMR spectrum with a contact time as short as 1 ms was performed in order to detect the specific interactions of CO2 at
shorter distances with the individual host ligand moieties protruding toward the center of the cage. Notably, CO2 strongly correlates
with host walls of the cages in addition to intercapsular voids, indicating that the CO2 molecules even enter into the intercapsular
cages.
13
C solid-state NMR spectroscopy was successfully applied to the study of CO2 adsorbed in a MOF Mg-MOF-74 or CPO-27-Mg
(Mg2dodbc), which exhibits exposed Mgþ 2 cation sites capable of efficiently capturing CO2. 88 Details of the specific host–gas inter-
actions and CO2 dynamics were elucidated by the analysis of 13C line shapes, collected in the temperature range from 200 to 380 K
of the samples loaded with 13C-enriched CO2. The presence of the CSA clearly proved the reduction of motional degrees of freedom
due to the confinement of CO2 in the pores. The axially symmetrical CSA is compatible with the axial molecular shape of CO2. A
careful analysis of the CO2 CSA tensors and the simulation of the narrowing of CO2 CSA signals with a fast motion regime > 106 Hz
suggested a uniaxial rotation at an angle q to the main molecular axis (Fig. 19). The rotation angles increase as the temperature is
lowered and are higher for lower loading; in fact, they vary from 56 C at the high temperature of 375 K for a 0.5 CO2/Mgþ 2-loaded
sample to 69 at the low temperature of 210 K for a 0.3 CO2/Mgþ 2-loaded sample, as shown in Fig. 20. From the measurement of
13
C spin–lattice (T1) relaxation times as a function of temperature and considering the Bloembergen–Purcell–Pound (BPP) theo-
ries, activation energies for CO2 rotational motion of 10 and 6 kJ mol 1 were calculated for 0.3 and 0.5 CO2/Mgþ 2-loaded samples:
as the loading increases, the activation energy increases, suggesting that the Mgþ 2/CO2 interactions weaken at higher loading.
H2 molecules were successfully confined inside the ellipsoidal-shaped cages of C70 fullerene. The 1H NMR spectrum recorded at
room temperature of the static powder sample shows, for the endohedral protons of entrapped H2, a narrow peak resonating at an
extremely upfield chemical shift (d ¼ 24.7 ppm), as compared with the d ¼ þ 7.4 ppm chemical shift of H2 in the gas phase. 89
Furthermore, from variable temperature 1H NMR spectra, it was also possible to calculate a CSA of 10.1 ppm. Both isotropic
and anisotropic chemical shifts are temperature-independent, suggesting the chemical shift interactions are dominated by external
92 Solid-State NMR of Supramolecular Materials

Figure 19 (A) The bis(pyrazolyl)methane ligand (L) (left) self-assembled with Agþ 1 ions and forms supramolecular structures organized at two
hierarchical levels: [AgL]6(PF6)6 metal–organic cyclic hexamers (middle) and their organization in 3-D porous architectures containing intracapsular
and intercapsular cavities. (B) The 2-D 1H–13C HETCOR NMR spectrum recorded at 250 K and a contact time of 5 ms of [Ag(L)]6(PF6)6 loaded with
enriched 13CO2. The 1H projection and the 1-D 13C CP MAS NMR spectrum with a 5 ms contact time are reported. The through-space correlations of
the matrix hydrogens with the CO2 carbons are indicated. (C) Schematic representation of the surfaces of the intracapsular and intercapsular cavities
explored by CO2. Adapted with permission from Bassanetti, I.; Mezzadri, F.; Comotti, A.; Sozzani, P.; Gennari, M.; Calestani, G.; Marchiò, L. J. Am.
Chem. Soc. 2012, 134, 9142–9145. and Bassanetti, I.; Comotti, A.; Sozzani, P.; Bracco, S.; Calestani, G.; Mezzadri, F.; Marchiò, L. J. Am. Chem. Soc.
2014, 136, 14883–14895. Copyright (2012 and 2014) American Chemical Society.

factors such as the electrons of the C70 cage. Surprisingly, the low-temperature NMR data indicate that endohedral H2 molecules can
be seen as probes to detect the behavior of the fullerene cages, specifically the alignment of the C70 long axis with the applied
magnetic field at low temperature, while at high temperature, T > 340 K, the isotropic molecular tumbling of fullerene averaged
out the H2 resonance.

2.05.5 Molecular Rotors and Dynamics in Supramolecular Architectures

The fabrication and investigation of molecular rotors is an intriguing research field, which is connected to the possibility of realizing
ultrafast rotating moieties and dipoles in the solid state. Materials containing molecular rotors may serve as switchable ferroelectrics
and can be modulated by external magnetic fields. 90,91 The supramolecular approach permits mounting of molecular rotors on
crystal surfaces, exploiting the formation of surface inclusion compounds.92
In a challenging example, molecules engineered as three segments comprising a shaft, a stopper, and a rotator can interact with
a porous crystal and insert the shaft into the bulk crystal, while the rotator lies on the surface. Solid-state NMR can distinguish
unequivocally the arrangement of the molecular segments on the surface and in the bulk crystal ( Fig. 21). The through-space rela-
tionships between hydrogens and carbons witnessed by cross resonances in 2-D HETCOR NMR spectra indicate that the long alkyl
chain (the shaft) belongs to the bulk inclusion compound, the stopper (carborane moiety) resides at the interface, and the rotator
head protrudes outside the crystal surface as expected by design because the rotator does not interact directly with the host, as shown
by the absence of cross NMR signals, even at longer contact times. Therefore, the NMR experiments described earlier were focused on
surfaces and promoted a clear understanding of the regular arrangement of rotor-bearing molecules on the crystal surfaces.
Molecular rotors themselves can be engineered to constitute the building blocks of porous supramolecular architectures. In this
way, they simultaneously sustain the structure and are exposed to the pores, through which they interact with the gas phase. Nano-
porous materials offer stimulating perspectives in building architectures endowed with ultrafast molecular rotors, which are easily
accessible by chemical species.93–100 Only in recent years have porous molecular crystals held together by weak interactions emerged
as a new field of interest in the panorama of materials with permanent porosity. This is likely due to the objective difficulties in
Solid-State NMR of Supramolecular Materials 93

Figure 20 13C CSA powder patterns of 13C-enriched CO2 in Mg-MOF-74 at variable temperatures. 13C spectra of samples loaded with CO2: (A) 0.3
CO2/Mg ratio and (B) 0.5 CO2/Mg ratio. Schematic representation of CO2 uniaxial rotation at the open Mgþ 2 site in Mg-MOF-74 (top). Adapted with
permission from Kong, X.; Scott, E.; Ding, W.; Mason, J. A.; Long, J. R.; Reimer, J. A. J. Am. Chem. Soc. 2012, 134, 14341–14344. Copyright (2012)
American Chemical Society.

preventing the tendency of molecules to aggregate into a close-packed crystal structure, thus canceling any significant porosity. In
fact, the absence of strong and directional covalent bonds is manifested in the tendency toward structural collapse. Indeed, until
recent years, the principle by which any molecular system would reach its maximum close packing was considered a taboo.
However, structural stability can be notably increased if molecular rigidity and appropriate shape factors reduce favorable packing.
Porous architectures can be induced by host–guest formation followed by mild guest removal. This process results in crystals with
permanent pores, which are generally accessible from the gas phase. Hydrogen bonds and soft interactions may contribute to stabi-
lizing low-density frameworks. 101 Indeed, rodlike molecules containing aromatic rings are suitable for engineering rotors appended
in their core, like a wheel mounted on an axle.
Rotors based on p-phenylene moieties appended in between ethynyl groups are the most suitable to keep the rotational energy
barrier as low as possible. Further extension of the rodlike molecules can be achieved by additional p-phenylenesulfonate units at
the ends of the molecule ( Fig. 22). Such rodlike organosulfonate molecules self-assemble with alkylammonium cations, forming
a porous architecture.102
Spin–lattice relaxation times of the carbon-13 nuclei describe thoroughly the motional behavior of each molecular fragment.
The maximum relaxation rate occurs when thermal agitation frequencies match the observation frequency according to the formula
sc uc  1. The data can be fitted by the Kubo–Tomita equation that describes the relaxation by the fluctuations of dipole interac-
tions: 103 The two peripheral p-phenylene rings show in the examined temperature range relatively long relaxation times of the order
of 20–40 s, indicating that they are far from being involved in rotation. On the contrary, the central ring finds the most efficient
relaxation at 250–260 K, with 13C T1 times as short as 250 ms, which is diagnostic of rotary motion with correlation times in
the nanosecond regime.
However, a method of choice that can be widely applied for a precise description of the motional trajectories of specific molec-
ular moieties in the solid state is spin-echo deuterium NMR spectroscopy: the experiments are carried out under static conditions to
94 Solid-State NMR of Supramolecular Materials

Figure 21 (A) Schematic representation of the inserted rotor into a TPP channel. (B) 2-D 1H–13C HETCOR NMR spectra of the TPP adduct with
shaft molecule recorded at variable cross polarization times a) 0.1, b) 0.5, c) 1.25, d) 2 ms and e) 1H MAS NMR spectrum recorded at 600 MHz and
spinning speed of 35 kHz of the TPP-d12 adduct (right). Intermolecular interactions of the aliphatic chains and carborane moieties with the host mole-
cules are highlighted. Adapted with permission from Kobr, L.; Zhao, K.; Shen, Y. Q.; Comotti, A.; Bracco, S.; Shoemaker, R. K.; Sozzani, P.; Clark, N.
A.; Price, J. C.; Rogers, C. T.; Michl, J. J. Am. Chem. Soc. 2012, 134, 10122–10131. Copyright (2012) American Chemical Society.

retain the precious anisotropic information, possibly averaged by internal material dynamics. 104 This method obviously requires
deuterium-labeled samples. In the particular instance of porous organodisulfonate-based salts, the dynamics of the labeled central
p-phenylene ring was studied and the 2H spin-echo spectra provided the mechanism of reorientation of the C–D vectors being sensi-
tive to motional averaging for frequencies that overcome the MHz regime.3 2H NMR line shape at variable temperature of C–D
bonds of the labeled central ring depicts a complex profile that can be simulated as a rapid reorientation about the rotation axis
due to a 180 degree flip mechanism.3 Extremely fast rotation frequency of 107 Hz has been observed at temperatures as low as
200 K (Fig. 22). At very low temperature, the 2H NMR spectral profile shows an anisotropic Pake pattern, typical of a slow C–D
reorientation regime (< 104 Hz). Thus, both spin–lattice relaxation times and deuterium profiles at a given temperature are theo-
retically associated to a correlation time (or frequency). The activation energy for the dynamical interconversion from one config-
uration to another can be evaluated by an Arrhenius plot, along with the rotational frequency extrapolated to infinite temperature.
In the present case, an energy value as low as 4.7 kcal mole was calculated and an attempt frequency of 1012, compatible with the
inertial mass of p-phenylene rotor. Such procedure has a very large applicable generality in the field of dynamics of supramolecular
solids. Moreover, the central ring is fully exposed to the crystalline channels and molecular rotor dynamics may be modulated by
interaction with gases flowing in. Indeed, the rotors were exposed to the interaction with I2 vapors at low pressure and the rotor
dynamics could be switched off and on by I2 absorption/desorption. The remarkable change of dynamics driven by the interaction
with gaseous species suggests the use of these molecular crystals in sensing and pollutant management.
Condensation of alkoxysilylated organic linkers, such as p-phenylenedisilyl- and p-diphenyldisilyl precursors,94–97 performed by
a template synthesis on surfactant micelles, offers a further strategy for retaining the order of a self-assembled organized suprastruc-
ture. After structure-directing agent removal, mesoporous materials with hexagonal periodicity called periodic mesoporous
Solid-State NMR of Supramolecular Materials 95

Figure 22 (A) Crystal structure of the permanently porous compound as viewed along the channel axis. (B) 2H NMR spectra of experimental and
simulated 2H NMR spectra of porous compound with deuterated central p-phenylene moiety as a function of temperature. The 2H NMR spectral
profiles were simulated considering a two-site 180 degree flip mechanism. The k is the 180 degree flip rate constant and is expressed in Hz. (C) 2H
NMR spectra of pristine-empty, iodine-loaded, and I2-freed sample. The exchange rates are reported. Adapted with permission from Comotti, A.;
Bracco, S.; Yamamoto, A.; Beretta, M.; Hirukawa, T.; Tohnai, N.; Miyata, M.; Sozzani, P. J. Am. Chem. Soc. 2014, 136, 618–621. Copyright (2014)
American Chemical Society.

organosilicas are formed. The organic linkers constitute the major part of the material and align in the pore walls, similarly to
smectic phases. Such linkers behave as molecular rotors in the crystalline-like pore walls and their dynamics was studied by 1H
and 13C T1 relaxation times as well as 2H NMR.
The prototypal building block to obtain such molecular rotors in mesoporous materials was 4,40 -bis-(triethoxysilyl)biphenyl. 94
13
C T1 relaxation times recorded at 75 MHz showed a maximum at about room temperature, indicating that at this temperature, the
rotors already respond in the 107 Hz range. Moreover, 1H T1 relaxation times at 30 MHz and 300 MHz enlarged the observation
window of the frequency spectrum and confirm that the sample response falls in fast motional frequencies. Analysis of the temper-
ature dependence of the relaxation rates yields, after reevaluation of temperature range, an apparent activation energy of
6.4 kcal mol 1, which corresponds to the energy value reported in the literature for a two-site 180 degree flip process of the p-phenyl
rings about their central axes.
The phenomenon was found to be common in the whole family of aromatic linkers such as p-phenylene, divinyl-p-phenylene,
and fluorinated analogues wherein fluorine atoms are substituted on the phenylene rotor. At room temperature, the 2H NMR spec-
trum of mesoporous material containing deutero-p-phenylene linkers shows a profile restricted to ¼ of the static Pake pattern
( Fig. 23). The line shape and width analysis at variable temperature indicates the motion mechanics of a rapid two-site 180 degree
flip reorientation of p-phenylene moieties about their para-axis and exchange rates k as high as 5.6 107 Hz at room temperature.
96 Solid-State NMR of Supramolecular Materials

Figure 23 (A) Schematic representation of the mesoporous material: each p-phenylene rotor pivots on two silicon atoms through C–Si bonds that
are inserted in inorganic siloxane layers. Rotors are aligned along the channel axes and exposed to the internal free volume of the empty channels.
(B) Deuterated disilyl-p-phenylene rotor. (C) 2H NMR spectra from 164 to 298 K of the mesoporous material compared with the simulated spectra.
(D) 2H NMR spectra of mesoporous material with C20, TEA, and (D) octadecylammonium chloride (OTMA) guests compared with the empty matrix.
The exchange rates (k in Hz) for each spectrum are given. Adapted with permission from Comotti, A.; Bracco, S.; Valsesia, P.; Beretta, M.;
Sozzani, P. Angew. Chem. Int. Ed. 2010, 49, 1760–1764. Copyright (2010) John Wiley & Sons.

The extremely high surface area and the open framework of the mesoporous material allow easy diffusion of chemical species
that, upon penetrating the nanochannels, interact directly with the aromatic groups on the pore walls and can affect their motion.
The collective dynamics of the entire population of rotors was fine-tuned by the active use of included molecules, thus enabling
the external regulation of the motional regime through weak intermolecular interactions produced at the extended interfaces. The
2
H NMR spectra recorded at variable temperatures for the sample with the nanochannels filled with octadecyltrimethylammo-
nium bromide (OTMA) and their simulations show mobility only above room temperature. At lower temperatures, the spectral
profiles display a “static” pattern typical of a slow exchange regime. The rotational speed of the p-phenylene moieties at any
temperature was reduced by three orders of magnitude with respect to that observed in the walls of the empty nanochannels.
This finding reveals unprecedented chemical control of the molecular rotor rates by soft interactions. Notably, the difference
of the energy barriers to rotation measured by NMR between empty and guest-filled samples was only 4 kcal mol 1, and this
value is consistent with weak interactions occurring at the extended interfaces between the rotor elements in the host walls
and the guest.
Control over the rotor dynamics was also achieved by a variety of guest molecules diffused into the nanochannels from solution
or from the melt. The screening of guests with varied polarity and molecular masses, such as n-eicosane (C20), tetraethylammonium
chloride (TEA), and water, showed the modulated response of the rotor dynamics in the host framework, which resembles the active
switching of molecular motion in engineered molecular machines.
The introduction of dipoles mounted onto the rotors is the final step toward the construction of materials responsive to an
external stimulus such as an applied electric field. This effect was obtained by substitution with electron-withdrawing atoms. 97 Fluo-
rine atoms asymmetrically substituted onto the rotor, called FOS1 and FOS2, did not hamper motion, as shown by solid-state NMR
relaxations (Fig. 24). This is a rare case of the combined presence of porosity and fast dipole reorientation in a same material.
Furthermore, open pores constitute avenues to convey chemical species, which actively modulate dipole dynamics by vapor diffu-
sion, for example, iodine reacted close to the rotor pivots and regulated motion like brakes.
Solid-State NMR of Supramolecular Materials 97

Figure 24 (A) Representation of the mesoporous organosilicas with 2-fluorodivinyl- and 2,3-difluoro-divinylbenzenesiloxane moieties. (B) 13C relax-
ation rates for C3 nucleus of organosilica with 2-fluorodivinylbenzenesiloxane moieties versus the reciprocal of temperature as fitted by nonlinear
mean-square analysis of the Kubo–Tomita equation (right). Adapted with permission from Bracco, S.; Beretta, M.; Cattaneo, A.; Comotti, A.; Falqui,
A.; Zhao, K.; Rogers, C.; Sozzani, P. Angew. Chem. Int. Ed. 2015, 54, 4773–4777. Copyright (2015) John Wiley & Sons.

2.05.6 Conclusions

Solid-state 1-D and 2-D NMRs in both the static and spinning modes are the most versatile methods to understand the structure and
nanoscale properties of supramolecular materials. Their potential complements and frequently overcomes those of other tech-
niques, especially when complex systems are under investigation.
This article was not intended as an operative guide to experimental execution, but as an orientation to find solutions and deter-
mine choices among NMR experiments or if they would be useful in the given case. The reported examples are not exhaustive but
representative of the synergy that is possibly achieved between the synthesis and spectroscopic observations. Sometimes, the imme-
diate execution of NMR spectra is an invaluable retrofit for the phase composition and purity of crystalline and amorphous mate-
rials. Finally, an in-depth knowledge of the spatial relationships, diffusion, and dynamics of the adducts is the desired target.

References

1. Ernst, R. R.; Bodenhausen, G.; Wokaun, A. Principles of Nuclear Magnetic Resonance in One and Two Dimensions; Oxford Science Publications: Oxford, 2004.
2. Duer, M. J. Solid-State NMR Spectroscopy Principles and Applications; Blackwell Science Ltd: Oxford, 2002.
3. Bovey, F. A.; Mirau, P. NMR of Polymers; Academic Press: San Diego, 1996.
4. Vega, A. J. J. Am. Chem. Soc. 1988, 110, 1049–1054.
5. van Rossum, B.-J.; Forster, H.; de Groot, H. J. M. J. Magn. Reson. 1997, 124, 516–519.
6. Vinogradov, E.; Madhu, P. K.; Vega, S. Chem. Phys. Lett. 1999, 314, 443–450.
7. Lesage, A.; Sakellariou, D.; Hediger, S.; Elena, B.; Charmont, P.; Steuernagel, S.; Emsley, L. J. Magn. Reson. 2003, 163, 105–113.
8. Pines, A.; Gibby, M. G.; Waugh, J. S. J. Chem. Phys. 1973, 59, 569–590.
9. Stejskal, E. O.; Memory, J. R. High Resolution NMR in the Solid State: Fundamentals of CP-MAS; Oxford University Press: New York, 1994.
10. Taulelle, F. Fundamental Principles of NMR Crystallography; Wiley: Oxford, 2009.
11. Comotti, A.; Bracco, S.; Sozzani, P.; Horike, S.; Matsuda, R.; Chen, J.; Takata, M.; Kubota, Y.; Kitagawa, S. J. Am. Chem. Soc. 2008, 130, 13664–13672.
12. Baias, M.; Lesage, A.; Aguado, S.; Canivet, J.; Moizan-Basle, V.; Audebrand, N.; Farrusseng, D.; Emsley, L. Angew. Chem. Int. Ed. 2015, 54, 5971–5976.
13. Sozzani, P.; Bracco, S.; Comotti, A.; Simonutti, R.; Camurati, I. J. Am. Chem. Soc. 2003, 125, 12881–12893.
14. Fyfe, C. A.; Brouwer, D. H.; Lewis, A. R.; Villaescusa, L. A.; Morris, R. E. J. Am. Chem. Soc. 2002, 124, 7770–7778.
15. Loiseau, T.; Mellot-Draznieks, C.; Sassoye, C.; Girard, S.; Guillou, N.; Huguenard, C.; Taulelle, F.; Ferey, G. J. Am. Chem. Soc. 2001, 123, 9642–9651.
16. Sozzani, P.; Comotti, A.; Simonutti, R.; Bracco, S.; Simonelli, A. In Strength from Motion in Crystals: The Example of Supramolecular Adducts in Strength From Weakness:
Structural Consequences of Weak Interactions in Molecular, Supermolecules, and Crystals; Domenicano, A., Hargittai, I., Eds.; Kluwer Academic Publishers: Dordrecht, 2002;
pp 319–333.
17. Comotti, A.; Simonutti, R.; Bracco, S.; Castellani, L.; Sozzani, P. Macromolecules 2001, 34, 4879–4885.
18. Comotti, A.; Gallazzi, M. C.; Simonutti, R.; Sozzani, P. Chem. Mater. 1998, 10, 3589–3596.
19. Comotti, A.; Simonutti, R.; Stramare, S.; Sozzani, P. Nanotechnology 1999, 10, 70–78.
20. Akira, H.; Yoshinori, T.; Hiroyasu, Y. Chem. Soc. Rev. 2009, 38, 875–882.
21. Liu, Y.; Hu, C.; Comotti, A.; Ward, M. D. Science 2011, 333, 436–440.
22. Holman, K. T.; Pivovar, A. M.; Ward, M. D. Science 2001, 294, 1907–1911.
23. Brown, M. E.; Hollingsworth, M. D. Nature 1995, 376, 323–327.
24. Miyata, M.; Tohnai, N.; Isaki, I. Acc. Chem. Res. 2007, 40, 694–702.
25. Dalgarno, S. J.; Thallapally, P. K.; Barbour, L. J.; Atwood, J. L. Chem. Soc. Rev. 2007, 36, 236–245.
26. Farina, M.; Di Silvestro, G.; Sozzani, P. In Comprehensive Supramolecular Chemistry, Macnicol, D.D.; Toda, F.; Bishop, R. (Ed.), Vol. 6, Pergamon, Oxford, UK, 1996; pp.
371–398.
27. Sozzani, P.; Bracco, S.; Comotti, A.; Simonutti, R. Adv. Polym. Sci. 2005, 181, 153–177.
28. Castellani, F.; van Rossum, B.; Diehl, A.; Schubert, M.; Rehbein, K.; Oschkinat, H. Nature 2002, 420, 98–102.
29. Copley, R. C. B.; Barnett, S. A.; Karamertzanis, P. G.; Harris, K. D. M.; Kariuki, B. M.; Xu, M. C.; Nickels, E. A.; Lancaster, R. W.; Price, S. L. Cryst. Growth. Des. 2008, 8,
3474–3481; Madine, J.; Jack, E.; Stockley, P. G.; Radford, S. E.; Serpell, L. C.; Middleton, D. J. Am. Chem. Soc. 2008, 130, 14990–15001.
30. Griesser, U. J.; Jetti, R. K. R.; Haddow, M. F.; Brehmer, T.; Apperley, D. C.; King, A.; Harris, R. K. Cryst. Growth. Des. 2008, 8, 44–56.
98 Solid-State NMR of Supramolecular Materials

31. Comotti, A.; Bracco, S.; Sozzani, P.; Hawxwell, S. M.; Hu, C.; Ward, M. D. Cryst. Growth. Des. 2009, 9, 2999–3002.
32. Sozzani, P.; Comotti, A.; Bracco, S.; Simonutti, R. Angew. Chem. Int. Ed. 2004, 43, 2792–2797.
33. Brustolon, M.; Barbon, A.; Bortolus, M.; Maniero, A. L.; Sozzani, P.; Comotti, A.; Simonutti, R. J. Am. Chem. Soc. 2004, 126, 15512–15519.
34. Barbon, A.; Bortolus, M.; Brustolon, M.; Comotti, A.; Maniero, A. L.; Segre, U.; Sozzani, P. J. Phys. Chem. B 2003, 107, 3325–3331.
35. Waugh, J. S.; Fessenden, R. W. J. Am. Chem. Soc. 1957, 79, 846–849.
36. Canceill, J.; Lacombe, L.; Collet, A. J. Am. Chem. Soc. 1986, 108, 4230–4232.
37. Rapp, A.; Schnell, I.; Sebastiani, D.; Brown, S. P.; Percec, V.; Spiess, H. W. J. Am. Chem. Soc. 2003, 125, 13284–13297.
38. Tonelli, A. NMR Spectroscopy and Polymer Microstructure: The Conformational Connection; Wiley: Oxford, 1989.
39. Aakeröy, C. B.; Chopade, P. D. In Supramolecular Chemistry: From Molecules to Nanomaterials; Steed, J. W., Gale, P. A., Eds.; John Wiley & Sons: New York, 2012.
40. Brown, S. P. Solid State Nucl. Magn. Reson. 2012, 41, 1–27.
41. Maruyoshi, K.; Iuga, D.; Antzutkin, O. N.; Alhalaweh, A.; Velagad, S. P.; Brown, S. P. Chem. Commun. 2012, 48, 10844–10846.
42. Etter, M. C. Acc. Chem. Res. 1990, 23, 120–126.
43. Dudenko, D. V.; Williams, P. A.; Hughes, C. E.; Antzutkin, O. N.; Velaga, S. P.; Brown, S. P.; Harris, K. D. M. J. Phys. Chem. C 2013, 117, 12258–12265.
44. Harris, K. D. M.; Hughes, C. E.; Williams, P. A. Solid State Nucl. Magn. Reson. 2015, 65, 107–113.
45. Isaacson, S. G.; Lionti, K.; Volksen, W.; Magbitang, T. P.; Matsuda, Y.; Dauskardt, R. H.; Dubois, G. Nat. Mater. 2016, 15, 294–298. http://dx.doi.org/10.1038/nmat4475.
46. Sozzani, P.; Bovey, F. A.; Schilling, F. C. Macromolecules 1991, 24, 6764–6768.
47. Schilling, F. C.; Amundson, K. R.; Sozzani, P. Macromolecules 1994, 27, 6498–6502.
48. Sozzani, P.; Comotti, A.; Bracco, S.; Simonutti, R. Chem. Commun. 2004, 40, 768–769.
49. Comotti, A.; Simonutti, R.; Catel, G.; Sozzani, P. Chem. Mater. 1999, 11, 1476–1483.
50. Becker, J.; Comotti, A.; Simonutti, R.; Sozzani, P.; Saalwaechter, K. J. Phys. Chem. B 2005, 109, 23285–23294.
51. Johnson, C. E.; Bovey, F. A. J. Chem. Phys. 1958, 29, 1012–1014.
52. Von Ragué Schleyer, P.; Maerker, C.; Dransfeld, A.; Jiao, H.; Van Eikema Hommes, N. J. R. J. Am. Chem. Soc. 1996, 118, 6317–6318.
53. Bracco, S.; Comotti, A.; Valsesia, P.; Beretta, M.; Sozzani, P. CrystEngComm 2010, 12, 2318–2321.
54. Bracco, S.; Comotti, A.; Ferretti, L.; Sozzani, P. J. Am. Chem. Soc. 2011, 133, 8982–8994.
55. Kitao, T.; Bracco, S.; Comotti, A.; Sozzani, P.; Naito, M.; Seki, S.; Uemura, T.; Kitagawa, S. J. Am. Chem. Soc. 2015, 137, 5231–5238.
56. Distefano, G.; Comotti, A.; Bracco, S.; Beretta, M.; Sozzani, P. Angew. Chem. Int. Ed. 2012, 51, 9258–9262.
57. Comotti, A.; Bracco, A.; Mauri, M.; Mottadelli, S.; Ben, T.; Qiu, S. L.; Sozzani, P. Angew. Chem. Int. Ed. 2012, 51, 10136–10140.
58. Comotti, A.; Bracco, S.; Beretta, M.; Perego, J.; Gemmi, M.; Sozzani, P. Chem. Eur. J. 2015, 21, 17994–18010.
59. Sozzani, P.; Bracco, S.; Comotti, A.; Valsesia, P.; Simonutti, R.; Sakamoto, Y.; Terasaki, O. Nat. Mater. 2006, 5, 545–551.
60. Sozzani, P.; Behling, R. W.; Schilling, F. C.; Bruckner, S.; Helfand, E.; Bovey, F. A.; Jelinski, L. W. Macromolecules 1989, 22, 3318–3322.
61. Schilling, F. C.; Sozzani, P.; Bovey, F. A. Macromolecules 1991, 24, 4369–4375.
62. Sozzani, P.; Bovey, F. A.; Schilling, F. C. Macromolecules 1989, 22, 4225–4230.
63. Uemura, T.; Horike, S.; Kitagawa, K.; Mizuno, M.; Endo, K.; Bracco, S.; Comotti, A.; Sozzani, P.; Nagaoka, M.; Kitagawa, S. J. Am. Chem. Soc. 2008, 130, 6781–6788.
64. Distefano, G.; Suzuki, H.; Tsujimoto, M.; Isoda, S.; Bracco, S.; Comotti, A.; Sozzani, P.; Uemura, T.; Kitagawa, S. Nat. Chem. 2013, 5, 335–341.
65. Sozzani, P.; Bracco, S.; Comotti, A. In Hyperpolarized Xenon-129 Magnetic Resonance: Concepts, Production, Techniques and Applications, New Developments in NMR No.
4; Meersmann, T.; Brunner, E. Eds., RSC, Cambridge, 2015; pp. 164–184, Chapter 9.
66. Sozzani, P.; Comotti, A.; Simonutti, R.; Meersamnn, T.; Logan, J. W.; Pines, A. Angew. Chem. Int. Ed. 2000, 39, 2695–2698.
67. Comotti, A.; Bracco, S.; Ferretti, L.; Mauri, M.; Simonutti, R.; Sozzani, P. Chem. Commun. 2007, 43, 350–352.
68. Meersmann, T.; Logan, J. W.; Simonutti, R.; Caldarelli, S.; Comotti, A.; Sozzani, P.; Kaiser, L.; Pines, A. J. Phys. Chem. A 2000, 104, 11665–11670.
69. Soldatov, D. V.; Moudrakovski, I. L.; Ripmeester, J. A. Angew. Chem. Int. Ed. 2004, 43, 6308–6311.
70. Moudrakovski, I.; Soldatov, D. V.; Ripmeester, J. A.; Sears, D. N.; Jameson, C. J. Proc. Natl. Acad. Sci. U. S. A. 2004, 101, 17924–17929.
71. Cheng, C.-Y.; Bowers, C. R. Chem. Phys. Chem. 2007, 8, 2077–2081.
72. Dvoyashkin, M.; Wang, A.; Vasenkov, S.; Bowers, C. R. J. Phys. Chem. Lett. 2013, 4, 3263–3267.
73. Jeener, J.; Meier, B. H.; Bachmann, P.; Ernst, R. R. J. Chem. Phys. 1979, 71, 4546–4553.
74. Springuel-Huet, M.-A.; Nossov, A.; Guenneau, F.; Gédeon, A. Chem. Commun. 2013, 49, 7403–7405.
75. Pawsey, S.; Moudrakovski, I.; Ripmeester, J.; Wang, L.-Q.; Exarhos, G. J.; Yaghi, O. M. J. Phys. Chem. C 2007, 111, 6060–6067.
76. Cheng, C.-Y.; Stamatatos, T. C.; Christou, G.; Bowers, C. R. J. Am. Chem. Soc. 2010, 132, 5387–5393.
77. Cheng, C.-Y.; Pfielsticker, J.; Bowers, C. R. J. Am. Chem. Soc. 2008, 130, 2390–2391.
78. Campbell, K.; Ooms, K. J.; Ferguson, M. J.; Stang, P. J.; Wasylishen, R. E.; Rykwinski, R. R. Can. J. Chem. 2011, 89, 1264–1276.
79. Lee, H.; Seo, Y.; Seo, Y.-T.; Moudrakovski, I. L.; Ripmeester, J. A. Angew. Chem. Int. Ed. 2003, 42, 5048–5051.
80. Sozzani, P.; Bracco, S.; Comotti, A.; Ferretti, L.; Simonutti, R. Angew. Chem. Int. Ed. 2005, 44, 1816–1820.
81. Miyahara, Y.; Abe, K.; Inazu, T. Angew. Chem. Int. Ed. 2002, 41, 3020–3023.
82. Beeler, A. J.; Orendt, A. M.; Grant, D. M.; Cutts, P. W.; Michl, J.; Zilm, K. W.; Downing, J. W.; Facelli, J. C.; Schindler, M. S.; Kytzelnigg, W. J. Am. Chem. Soc. 1984, 106,
7672–7676.
83. Yadav, V. N.; Comotti, A.; Sozzani, P.; Bracco, S.; Bonge-Hansen, T.; Hennum, M.; Görbitz, C. H. Angew. Chem. Int. Ed. 2015, 54, 15684–15688.
84. Sozzani, P.; Comotti, A.; Bracco, S.; Distefano, G. Chem. Commun. 2009, 45, 284–286.
85. Comotti, A.; Fraccarollo, A.; Bracco, S.; Beretta, M.; Distefano, G.; Cossi, M.; Marchese, L.; Riccardi, C.; Sozzani, P. CrystEngComm 2013, 15, 1503–1507.
86. Bassanetti, I.; Mezzadri, F.; Comotti, A.; Sozzani, P.; Gennari, M.; Calestani, G.; Marchiò, L. J. Am. Chem. Soc. 2012, 134, 9142–9145.
87. Bassanetti, I.; Comotti, A.; Sozzani, P.; Bracco, S.; Calestani, G.; Mezzadri, F.; Marchiò, L. J. Am. Chem. Soc. 2014, 136, 14883–14895.
88. Kong, X.; Scott, E.; Ding, W.; Mason, J. A.; Long, J. R.; Reimer, J. A. J. Am. Chem. Soc. 2012, 134, 14341–14344.
89. Mamone, S.; Concistrè, M.; Heinmaa, I.; Carravetta, M.; Kuprov, I.; Wall, G.; Denning, M.; Lei, X.; Chen, J. Y.-C.; Li, Y.; Murata, Y.; Turro, N. J.; Levitt, M. H. Chem. Phys.
Chem. 2013, 14, 3121–3130.
90. Kottas, G. S.; Clarke, L. I.; Horinek, D.; Michl, J. Chem. Rev. 2005, 105, 1281–1376.
91. Vogelsberg, C. S.; Garcia-Garibay, M. Chem. Soc. Rev. 2012, 41, 1892–1910.
92. Kobr, L.; Zhao, K.; Shen, Y. Q.; Comotti, A.; Bracco, S.; Shoemaker, R. K.; Sozzani, P.; Clark, N. A.; Price, J. C.; Rogers, C. T.; Michl, J. J. Am. Chem. Soc. 2012, 134,
10122–10131.
93. Comotti, A.; Bracco, S.; Ben, T.; Qiu, S.; Sozzani, P. Angew. Chem. Int. Ed. 2014, 53, 1043–1047.
94. Bracco, S.; Comotti, A.; Valsesia, P.; Chmelka, B. F.; Sozzani, P. Chem. Commun. 2008, 44, 4798–4800.
95. Vogelsberg, C. S.; Bracco, S.; Beretta, M.; Comotti, A.; Sozzani, P.; Garcia-Garibay, M. A. J. Phys. Chem. B 2012, 116, 1623–1632.
96. Comotti, A.; Bracco, S.; Valsesia, P.; Beretta, M.; Sozzani, P. Angew. Chem. Int. Ed. 2010, 49, 1760–1764.
97. Bracco, S.; Beretta, M.; Cattaneo, A.; Comotti, A.; Falqui, A.; Zhao, K.; Rogers, C.; Sozzani, P. Angew. Chem. Int. Ed. 2015, 54, 4773–4777.
98. Shustova, N. B.; Ong, T.-C.; Cozzolino, A. F.; Michaelis, V. K.; Griffin, R. G.; Dinca, M. J. Am. Chem. Soc. 2012, 134, 15061–15070.
99. Inukai, M.; Fukushima, T.; Hijikata, Y.; Ogiwara, N.; Horike, S.; Kitagawa, S. J. Am. Chem. Soc. 2015, 137, 12183–12186.
Solid-State NMR of Supramolecular Materials 99

100. Kolokolov, D. I.; Stepanov, A. G.; Guillerm, V.; Serre, C.; Frick, B.; Jobic, H. J. Phys. Chem. C 2012, 116, 12131–12136.
101. Mastalerz, M.; Oppel, I. M. Angew. Chem. Int. Ed. 2012, 51, 5252–5255.
102. Comotti, A.; Bracco, S.; Yamamoto, A.; Beretta, M.; Hirukawa, T.; Tohnai, N.; Miyata, M.; Sozzani, P. J. Am. Chem. Soc. 2014, 136, 618–621.
103. Kubo, R.; Tomita, K. J. Phys. Soc. Jpn. 1954, 9, 888–919.
104. Kristensen, J. H.; Hoatson, G. L.; Vold, R. L. Solid State Nucl. Magn. Reson. 1998, 13, 1–37.
2.06 Supramolecular Mechanochemistry
T Friscic and C Mottillo, McGill University, Montreal, QC, Canada
Ó 2017 Elsevier Ltd. All rights reserved.

2.06.1 Introduction 101


2.06.2 Historical Overview 102
2.06.3 Mechanochemical Activation of Solids 102
2.06.4 The Three-Step Mechanism of Mechanochemical Reactivity 105
2.06.4.1 Overview 105
2.06.4.2 Reactions Mediated by a Gas Phase 105
2.06.4.3 The Effect of Reactant Crystal Structure on the Solid-State Mechanochemical Reactivity 107
2.06.4.4 Reactant Diffusion on Surfaces versus Gas-Phase Diffusion 107
2.06.4.5 Mechanochemical Reactions Mediated by a Liquid Phase 107
2.06.4.6 Mechanochemical Reactions Mediated by an “Invisible” Eutectic Phase 109
2.06.4.7 Amorphous Solid Phase as a Reaction Intermediate 110
2.06.5 Atmospheric Effects: Air Moisture and Organic Vapors 113
2.06.5.1 Effect of Air Moisture on the Mechanochemical Assembly of Coordination Polymers 113
2.06.5.2 Catalytic Effect of Moisture in Mechanochemical Grinding Reactions 114
2.06.5.3 The Interaction of Mechanochemical Reactions With Carbon Dioxide 115
2.06.6 Templating, Kinetic, and Thermodynamic Effects in Supramolecular Mechanochemistry 115
2.06.6.1 Mechanochemical Equilibrium 115
2.06.6.2 Structure-Directing Effects and Seeding in Mechanochemical Reactions of Molecular Solids 116
2.06.6.3 Stepwise Reactions as a Result of Solid-State Supramolecular Competition 118
2.06.6.4 Stepwise Reactions in the Synthesis of Coordination Polymers and MOFs 119
2.06.7 In Situ Reaction Monitoring 121
2.06.8 Mechanochemical Cocrystal Synthesis 121
2.06.8.1 Synthesis of Three-Component Cocrystals 121
2.06.8.2 Systematic Studies of Structure-Templating Effects in Cocrystal Formation 123
2.06.8.3 Screening for Recognition Motifs of Biomolecules 124
2.06.8.4 Halogen-Bonded Cocrystallization 125
2.06.9 Combining Different Types of Molecular Self-Assembly 126
2.06.10 Cages and Interlocked Structures 126
2.06.10.1 Molecular and Supramolecular Cages 126
2.06.10.2 Mechanically Interlocked Supramolecular Structures 128
2.06.10.3 Mechanochemical Self-Sorting 129
2.06.11 Cocrystal–Cocrystal Reactions 130
2.06.12 Supramolecular Catalysis 131
2.06.13 Mechanochemical Synthesis of MOFs 132
2.06.13.1 Synthesis of MOFs by Neat Grinding 132
2.06.13.2 Synthesis of MOFs Using Liquid-Assisted Mechanochemistry 133
References 135

Nomenclature
bipy 4,40 -Dipyridyl LAG Liquid-assisted grinding
dabco 1,4-Diazabicyclo[2.2.2]octane ILAG Ion- and liquid-assisted grinding
dace 1,4-Diaminocyclohexane

2.06.1 Introduction

This text provides a brief, descriptive overview of the emergent area of supramolecular mechanochemistry, resulting from an inter-
play of mechanochemistry, that is, solid-state reactivity by milling, grinding, or shearing,1 with supramolecular chemistry, specif-
ically molecular recognition, self-assembly, and noncovalent interactions.2 Supramolecular mechanochemistry is a novel area,
especially when considering that mechanochemistry is traditionally associated with transformations and processing of inorganic
substances, such as metals, minerals, and ceramics. However, over the past three decades, the focus of research in mechanochemistry

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12500-4 101


102 Supramolecular Mechanochemistry

has broadened, and today, mechanochemical techniques are becoming mainstream in the synthesis of and screening for materials
generated through molecular self-assembly, specifically polymorphs, cocrystals, and coordination complexes. This change of focus
not only has brought forward a different range of industrially relevant targets for mechanochemical synthesis, such as pharmaceu-
tical cocrystals or microporous metal–organic frameworks (MOFs), but also has led to a qualitative change in the understanding of
mechanochemical reactions and their underlying mechanisms. Specifically, expanding the focus of mechanochemistry toward
organic and metal–organic materials has made the understanding of noncovalent interactions and processes that guide molecular
assembly in the solid state, such as polymorphic transformations or amorphization, central to understanding of mechanochemical
reactivity.
Consequently, supramolecular mechanochemistry is now emerging as a discipline that both targets the synthesis of new mate-
rials through manipulating noncovalent interactions and molecular self-assembly under mechanochemical conditions, and also
utilizes these concepts of supramolecular chemistry to understand solid-state reactivity. In order to highlight this interplay of
solid-state reactivity and supramolecular chemistry, we provide an overview of mechanochemical reaction mechanisms, interpreted
in the context of changes to solid-state molecular assembly, and of applications of mechanochemistry for the synthesis of self-
assembled targets based on hydrogen bonds, halogen bonds, or metal–ligand coordination bonds.

2.06.2 Historical Overview

Mechanochemistry encompasses chemical reactions induced and/or sustained by the use of mechanical force.1 The history of mech-
anochemical reactions is long, with the first written records of mechanochemical transformations of metal ores dating from 4th
century BC.3 Despite an early rise in research activity during 19th and early 20th century,4 the interest of chemists in mechanochem-
ical processes has been at best slight for most of the 20th century. However, the opening decade of the 21st century witnessed
a surprising revival of interest in mechanochemistry, which has been strongly impacted by the work of Toda, who recognized
the potential of grinding and milling for conducting environmentally friendly solvent-free chemistry.5 As a large fraction of the
extensive work reported by Toda’s group was focused on the formation of lattice inclusion compounds, which are based on
host–guest molecular recognition and hydrogen bonding, it is justified to say that Toda has also pioneered the development of
supramolecular concepts in solid-state grinding (or milling) reactions. In particular, his group’s seminal contribution from
1987,6 which focused on solid-state and solution-phase complexation of a variety of lattice hosts and guests, established a number
of phenomena that are central to modern mechanochemical supramolecular chemistry. Perhaps the most striking of these is the
ability to synthesize inclusion compounds that are not accessible by conventional solution-based routes. Specifically, the grinding
of the supramolecular host gem-bis(4-hydroxyphenyl)cyclohexane with 1,4-naphthoquinone was reported to readily yield a lattice
inclusion compound, whereas none was obtained from solution.6 Another early, highly significant contribution to supramolecular
chemistry under mechanochemical conditions was provided by Etter’s group,7 who demonstrated the formation of a Hoogsteen-
type hydrogen-bonded complex by milling together of 9-methyladenine and 1-methylthymine solids (Fig. 1A). The particular
significance of this work lies in the simultaneous demonstration that solid-state hydrogen-bonded complexes, that is, cocrystals,
can be obtained through mechanochemical techniques and that solid-state reactivity permits the formation of highly specific inter-
molecular recognition motifs. In that way, this work represents a pioneering precedent to the highly popular use of mechanochem-
ical techniques to screen for and synthesize pharmaceutical cocrystals. The interplay of mechanochemistry and supramolecular
chemistry in the context of developing new pharmaceutical solids has been addressed in recent reviews.12,13
An early demonstration of using mechanochemistry to conduct coordination-driven self-assembly reactions was provided by the
Steed group,8 who reported a mechanochemical ligand exchange reaction on a metal–organic cluster to form a coordination poly-
mer (Fig. 1B). The reaction, conducted by manual grinding in a mortar and pestle, involved dimeric copper(II) acetate hydrate and
1,3-bis(4-pyridyl)propane and led to the assembly of copper(II) acetate paddlewheel clusters into a one-dimensional (1-D) coor-
dination polymer. The formation of complexes between metal ions and macrocyclic crown ethers is one of the fundamental and
most extensively studied transformations of supramolecular chemistry. That complexation of metal ions with crown ethers can
be conducted mechanochemically, in the absence of bulk solvent, and was pioneered and established largely through contributions
from Braga’s group.9 In particular, Braga and coworkers have demonstrated the complexation of alkaline, ammonium, and transi-
tion and main group metal ions by simple grinding of corresponding binary salts in the presence of a crown ether. One particularly
interesting example is the in situ formation of the lead(II) hydrogensulfate complex with 18-crown-6 by three-component grinding
of PbSO4, a very poorly soluble substance (Ksp z 2  10 8 in water at 20 C), with sulfuric acid and the crown ether (Fig. 1C).10

2.06.3 Mechanochemical Activation of Solids

An often considered mechanism for mechanochemical activation of solids is through the direct cleavage of covalent bonds by
mechanical impact. This type of mechanical activation takes place in the formation of cracks, holes, and vacancies and is expected
to be significant for substrates with extended, polymeric covalent structures. In that context, Moore and Bielawski groups have pio-
neered the use of mechanically labile functional groups as “mechanophores” for the design of mechanoresponsive and self-healing
polymer materials.14,15 Direct cleavage of covalent bonds by milling is common upon mechanical treatment of inorganic materials,
specifically infinite covalent solids such as quartz, silica, or different sulfide minerals. The cleavage of strong covalent bonds leaves
Supramolecular Mechanochemistry 103

Figure 1 Early examples of different supramolecular processes conducted by milling or grinding mechanochemistry: (A) specific Hoogsteen-type
recognition of methylated adenine and thymine in mechanochemical milling, reported by Etter et al.7; (B) synthesis of a coordination polymer through
mechanochemical ligand exchange described by Belcher et al.8; and (C) synthesis of a metal crown ether complex, described by the Braga group.9,10
In designating mechanochemical reactions, we have adopted the symbol proposed by the Hanusa group.11

behind surface sites occupied by either charge-separated species or free radicals. Both types of sites represent “energy sinks” that
accumulate input mechanical energy and act as centers of reactivity. The high energy and character of these reactive sites have
been claimed to bear resemblance to extremely high-temperature plasma environments and are often addressed as the “magma–
plasma” model of mechanochemical reactivity.3 However, such models of mechanochemical reactivity are not very likely to be
of importance in supramolecular transformations taking place upon grinding or milling of soft molecular solids, where mechanical
impact is much more likely to lead to defects in crystal packing, and rupture or rearrangement of noncovalent intermolecular inter-
actions, including hydrogen bonds, halogen bonds, and p/p stacking interactions.
Therefore, for transformations of molecular solids, it might be more appropriate to consider the mechanical energy input in the
system as being stored in an “activated form” of the substrate after milling. The materials obtained by such mechanical activation
can exhibit very different properties from the starting bulk solid, including enhanced reactivity. This is due to two principal reasons.
First, milling and grinding generally induce a significant reduction in particle size. Depending on the milled material and the milling
assembly, particles can adopt a variety of sizes between one to several hundred nanometers, typically resulting in very high surface-
to-volume ratios. As a result, one of the principal measures of mechanical activation is the increase in the available surface area of
the sample. These considerations are not limited only to molecular solids, as the formation of nano-sized materials is a well-known
area of mechanochemistry of minerals and inorganic compounds.16 While mechanochemical particle comminution has been exten-
sively used to synthesize nanoparticles of metals, metal–metal oxide composites, and simple binary metal compounds,17 the same
principle can be applied to explain the reactivity of organic solids. This was demonstrated by the Blagden group, who systematically
monitored the reactivity in two binary mixtures of organic solids expected to yield a pharmaceutically relevant cocrystal.16 The two
mixtures that were explored consisted of equimolar amounts of urea and 2-methylbenzamide and of equimolar amounts of caffeine
with malonic acid. By premilling each reactant separately and separating fractions with well-defined particle size distribution, Ibra-
him and coworkers were able to demonstrate that whereas the cocrystal components normally did not react by simple physical
104 Supramolecular Mechanochemistry

contact, physical mixtures of premilled reactants slowly (on a time scale of hours to days) converted into the expected cocrystal
phase.18 The rate of conversion was in both cases inversely proportional to particle size: Mixtures of reactants containing particles
between 20 and 45 mm reacted the fastest, while particle size distribution between 180 and 250 mm led to very little conversion after
weeks of aging (Fig. 2A). In that way, the authors were capable to distinguish the activation of organic solids by particle size reduc-
tion from reactivity induced by mechanical stress. Importantly, this work also utilized thermal microscopy, surface area measure-
ments, and humidity sorption techniques to discount the potential participation of an intermediate liquid (melt or eutectic) or
amorphous phases in the mechanism of cocrystal formation. Consequently, it appears that the mechanical activation of the molec-
ular solids can sometimes be explained solely by the increase in the surface area of the sample and, therefore, improved molecular
diffusion.
Concomitant with the reduction in particle size, another important mechanism behind activation of solids by mechanochemical
milling is the introduction of defects and the destruction of crystalline order of the sample. This effect is, supposedly, most
pronounced on the surface (Fig. 2B).19 Indeed, combined infrared spectroscopy, X-ray diffraction, solid-state NMR, and electron
microscopy studies have revealed the formation of a disordered shell around the crystalline particle core upon mechanochemical
activation of a number of inorganic materials. Importantly, such surface amorphization is often difficult to observe for molecular
and metal–organic materials, as the resulting amorphous materials are likely to relax even more readily by crystallization.
For the often significantly softer molecular materials, mechanochemical activation by grinding or milling can readily lead to
complete amorphization. Thus, generated amorphous phases are typically significantly less kinetically stable than their inorganic
counterparts and are therefore significantly more difficult to directly observe. For example, amorphous intermediates in mech-
anochemical cocrystallization by milling have been trapped by the Rodriguez-Hornedo group, who studied the ball-milling coc-
rystallization (Fig. 3) of the active pharmaceutical ingredient (API) carbamazepine with the cocrystal former (coformer)
saccharin.20 Dry (or neat) milling of the two components at room temperature readily produced the (carbamazepi-
ne)$(saccharin) cocrystal,21 as established by a combination of powder X-ray diffraction and infrared spectroscopy. However,
if the mechanochemical treatment is conducted at temperatures close to the boiling point of liquid nitrogen (milling under cryo-
genic conditions, also known as cryomilling), the product is an amorphous, noncrystalline material that does not exhibit any
significant features in the powder diffraction pattern. Therefore, even at cryogenic conditions, the mechanical energy of milling
provided by a vibratory shaker mill can be sufficient to break down the crystalline order of the reacting substances and achieve
mixing at the molecular level. Aging of the cryomilled material at room temperature (between 22 C and 25 C) in dry air (0%
relative humidity) or in moist air (75% relative humidity) led to the spontaneous formation of the cocrystal. Such cocrystal
formation by aging of a milled mixture can be regarded as a means of relaxing the energy-rich amorphous phase. Consequently,
cryomilling enabled the process of mechanochemical cocrystallization to be separated into (i) the mechanically induced activa-
tion to form an amorphous phase and (ii) thermally induced recrystallization enabled after the reaction mixture has crossed its
glass transition temperature (Fig. 3). The formation of an amorphous phase in this system was subsequently evidenced by direct
monitoring, through recently developed techniques for real-time, in situ observation of mechanochemical ball-milling reaction
by synchrotron X-ray powder diffraction.22 Presumably, the same concept can be applied to explain at least some of the mech-
anochemical reactions conducted at or above room temperature, with the amorphous phase playing the role of a short-lived
intermediate.
The earlier considerations highlight the fact that even the weakest mechanical effects, associated with simple materials processing
procedures, such as manual grinding, tableting, or packaging, can be considered as sources of mechanical activation, particularly
when molecular solids are concerned. In such a scenario, the mild crystal structure deformation resulting from mechanical treat-
ment can provide sufficient activation for transformations that are then thermodynamically driven.

Figure 2 (A) Solid-state formation of the model hydrogen-bonded pharmaceutical cocrystal of 2-methylbenzamide and urea, with respect to time
and reactant particle size. (B) High-resolution TEM image of a mechanochemically synthesized nanoparticle of zinc stannate, Zn2SnO4, demonstrating
a crystalline core with a c. 2 nm thick amorphous shell. Adapted with permission from (a) Ibrahim, A. Y.; Forbes, R. T.; Blagden, N. CrystEngComm

2011, 13, 1141–1152; (b) Sepelák, V,; Becker, S. M.; Bergmann, I.; Indris, S.; Scheuermann, M.; Feldhoff, A.; Kübel, C.; Bruns, M.; Stürzl, N.; Ulrich,
A. S.; Ghafari, M.; Hahn, H.; Grey, C. P.; Becker, K.D.; Heitjans, P., J. Mater. Chem. 2012, 22, 3117–3126.
Supramolecular Mechanochemistry 105

Figure 3 Formation of the (carbamazepine)$(saccharin) pharmaceutical hydrogen-bonded cocrystal by milling the two components together at
room temperature or by a combination of cryomilling (milling at liquid nitrogen temperature) and aging at room temperature.20 The comparison of
the two procedures indicates that in some cases, mechanochemical activation can be explained by the transient formation of an amorphous phase
that further reacts to form the final product.

2.06.4 The Three-Step Mechanism of Mechanochemical Reactivity


2.06.4.1 Overview
Mechanochemical transformations by grinding or milling can be described in terms of an intuitively understood three-step process,
formulated by Kaupp.23,24 In this mechanistic view, the role of mechanical action is to first activate the solid reactants for a chemical
reaction, which is then followed by the nucleation and crystallization of the product phase. Continuous milling or grinding effec-
tively removes the product phase from the surface of unreacted material, in that way enabling and enhancing further reaction.
Consequently, the three-step reaction mechanism consists of three purely physical processes: (1) reactant activation/reaction, (2)
nucleation and growth of the product daughter phase, and (3) removal of the product phase by mechanical action (Fig. 4).
The three-step view of mechanochemical reactivity shows strong resemblance with a general model of solid-state reactivity
depicted by Paul and Curtin.25 It is clear that the three-step model is largely concerned with bulk phase and macroscopic transfor-
mations and is especially valuable because it allows the evaluation of reactant and product transport mechanisms. In that context,
the most important distinction between different mechanochemical transformations is in the nature of the mobile, “activated”
phase that enables the first reaction step. The following sections will each provide several illustrations of how mechanochemical
transformations can be mediated by the gas, liquid, or a highly mobile solid phase. In such an overview, it is important to note
that a mechanochemical reaction should not necessarily proceed through a single mechanism. Indeed, it is most likely that mech-
anochemical reactions in general take place through different combinations of mechanisms delineated in the succeeding text.

2.06.4.2 Reactions Mediated by a Gas Phase


Possibly the simplest mechanism through which reactant species can be transported in a mechanochemical reaction is through the
vapor phase. Such vapor-mediated transformations are expected to readily take place when one or more involved reactants exhibit
a noticeable vapor pressure. Among notable examples are mechanochemical syntheses of hydrogen-bonded and charge-transfer
cocrystals involving p-benzoquinone, extensively studied by Kuroda’s group.26 One of the best indications that a mechanochemical
process is actually mediated by a vapor phase is by observing reactivity between reactants that are physically separated. This was
observed in the cocrystallization of p-benzoquinone with 2,2ʹ-biphenol and with 4,4ʹ-biphenol (Fig. 5).
Manual grinding of p-benzoquinone with either of bisphenols results in the rapid formation of a highly colored cocrystal, held
together by O–H/O hydrogen bonds and by charge-transfer interactions between the electron-deficient and electron-rich
p-systems of the benzoquinone and the biphenol, respectively. Both cocrystals can also be readily obtained by vapor-phase diffu-
sion: aging of physically separated samples of both cocrystal components results in the formation of a layer of cocrystal on the
106 Supramolecular Mechanochemistry

Figure 4 Schematic representation of a general three-step mechanism for mechanochemical transformations.

Figure 5 Cocrystallization components used by Kuroda et al.26

surface of the biphenol reactant. The latter observation is a clear indication that cocrystal formation is a result of gas-phase mobility
of benzoquinone, illustrated by the high vapor pressure of c. 9  10 2 Torr.
The nonagitated intimate mixture of 2,2ʹ-biphenol and benzoquinone was found to slowly undergo incomplete conversion to
the cocrystal during a period of time lasting up to 60 hours. After that, the reaction could be again initiated by brief manual
grinding, providing evidence that the role of mechanical agitation in milling synthesis of the cocrystals is largely in enhancing
the mixing of reactants and removal of the product phase from reactant surface. It is also clear that conducting the cocrystallization
reactions by grinding substantially increases reaction rates and enables quantitative yields. The improvement in reactivity by
grinding was more significant in the case of 4,4ʹ-biphenol, which authors explained by the need to overcome the extended hydrogen
bonding network in that material. This observation tacitly acknowledges the fact that although the reaction can proceed via vapor-
phase diffusion of one of the reactants, mechanical agitation still remains an important factor in overcoming noncovalent
Supramolecular Mechanochemistry 107

intermolecular forces for the other reaction partner. It is reasonable to assume that the activation energies associated with such
processes are in the range associated with typical hydrogen bonding interactions and polymorph transformations of organic solids,
that is, around 20–100 kJ mol 1. Therefore, it appears justified to extend the classical definitions of mechanochemical reactivity and
mechanical activation of solids to include processes in which activation energies are considerably lower than the strength of a typical
covalent bond. The high vapor pressure of benzophenone at room temperature was used to explain the solid-state complexation of
this molecule with a “wheel-and-axle” diol host by Bond and coworkers.27 However, this work also observed irreproducibility of
quantitative reaction kinetics measurements using a variety of techniques, including powder X-ray diffraction and solid-state NMR,
suggesting potential influence of atmospheric factors. The complexation in the solid state of a xanthenol host with four different
guest molecules, 1-aminonaphthalene, 8-hydroxyquinoline, triethylenediamine, and acridine, was monitored by Jacobs and
coworkers28 who established first-order kinetics and postulated that these supramolecular complexation reactions proceed through
a vapor-phase diffusion mechanism.
The role of milling as a means of overcoming noncovalent interactions between molecules of solid reactants is further observed
in the form of a milling frequency threshold for mechanochemical cocrystallization of p-benzoquinone and racemic 2,2ʹ-binaph-
thol.26 In these experiments, ball milling of equimolar amounts of the two organic reactants led to very little or no cocrystal forma-
tion if the milling frequency was kept at 150 rpm or below. However, if the milling frequency was above 160 rpm, the formation of
the red hydrogen-bonded and charge-transfer cocrystal was readily observed.

2.06.4.3 The Effect of Reactant Crystal Structure on the Solid-State Mechanochemical Reactivity
The differences in mechanochemical reactivities of molecular solids are sometimes rationalized in the literature through differences
in their respective melting points. The underlying assumption of such rationalization is that the reactants in a mechanochemical
process must pass through a melt, either in the form of a eutectic or by reaching their melting point through temperature increase
under the impact and shearing effect of grinding media. Even though the relationship between mechanochemical reactivity and
melting point can sometimes appear to be clear, it would be incorrect to interpret it as a sign that more active reactants necessarily
undergo melting during mechanosynthesis. Instead, it is more likely that the enhanced reactivity and a lower melting point have
a common cause, such as a poorly connected crystal structure of the reacting material. This is also well illustrated in the study of
solvent-free cocrystallization of benzoquinone with isomeric 2,2ʹ- and 4,4ʹ-biphenol.26 The difference in reactivity between the
two isomeric biphenols was evident in solid-solid layering experiments in which a column of powdered benzoquinone was gently
layered with powdered biphenol (Fig. 6). The formation of the colored cocrystal was evident by the appearance of a dark red inter-
face between the two reactant solid phases. The development of the interface was considerably faster for 2,2ʹ-biphenol than for its
4,4ʹ-isomer. The difference in reactivity of the two isomeric biphenols was attributed to the existence of a more extensive two-
dimensional (2-D) network of O–H/O hydrogen bonds in 4,4ʹ-biphenol, as opposed to the 1-D O–H/O hydrogen-bonded
chains in 2,2ʹ-biphenol. The stabilization resulting from this topological difference in crystal structures also causes a much higher
melting point for the 4,4ʹ-isomer (284 C) than for the 2,2ʹ-diphenol (110 C). Because both cocrystallization reactions proceed
through the vapor phase rather than through an intermediate melt, it is safe to conclude that the apparent relationship between
reactivity and melting point differences is in this case caused by topological differences between crystal structures of alternative
reactants.

2.06.4.4 Reactant Diffusion on Surfaces versus Gas-Phase Diffusion


The described solid-state layering experiments are also a suitable method to evaluate the rates of solid-state mechanochemical reac-
tions that are mediated by a vapor phase and yield highly colored products. The first systematic studies of the kinetics of a solvent-
free reaction using this methodology were reported by Rastogi’s group, who explored the cocrystallization of picric acid with
different polycyclic aromatic hydrocarbons.29 Unexpectedly, this work revealed that simple gas-phase transfer of reactants is a rele-
vant part of the cocrystallization mechanism only for a very small number of investigated systems. For the majority of investigated
cocrystallization systems, the kinetic and thermodynamic data indicated that molecular diffusion occurs through a constrained 2-D
environment. For example, measurements of reaction kinetics revealed that activation energies for cocrystallization of solid picric
acid with solid a- or b-naphthol (Fig. 7) are 80 and 42 kJ mol 1, respectively.30 Both of these values are smaller than the reported
values of the respective sublimation enthalpies for a-naphthol (91 kJ mol 1) and b-naphthol (84 kJ mol 1), leading to the conclu-
sion that molecular diffusion across the surface of solid reactant particles is the most important process in the cocrystallization of
these molecular solids with picric acid. Similar conclusions were also made about the solvent-free cocrystallization of p-benzoqui-
none with a-naphthol.31 Presumably, this mechanism, which does not require any reactant amorphization, could also be relevant
in the studies of the Blagden group on the effect of reactant particle size on cocrystal formation between caffeine and malonic acid or
between 2-methylbenzamide and urea.18

2.06.4.5 Mechanochemical Reactions Mediated by a Liquid Phase


Whereas the preceding section has illustrated the potential of gas-phase and surface diffusion in enabling solvent-free mech-
anochemical reactions of molecular solids, the generation of transient liquid-phase melts has often been observed or postu-
lated as means through which mechanochemical reactions take place. It has been proposed that mechanochemical reactions
108 Supramolecular Mechanochemistry

Figure 6 (A) The solid-state reaction of benzoquinone (bottom layer) with 2,20 -biphenol in the top layer immediately after layering (left) after
5 minutes (middle) and after standing overnight (right); (B) analogous reaction of benzoquinone (bottom layer) and 4,40 -biphenol (top layer) immedi-
ately after layering (left), three days after layering (middle), and one week after layering (right). The colored charge-transfer cocrystal is observed as
the dark red phase between the original layers. The difference in reactivity between 2,20 -biphenol and 4,40 -biphenol is explained by the more robust
crystal structure for the latter, illustrated by the difference in the topology of hydrogen bonding networks for the pure solids: (C) 2,20 -Biphenol is
composed of one-dimensional zigzag hydrogen-bonded chains, while (D) 4,40 -biphenol is composed of two-dimensional hydrogen-bonded sheets.
Adapted with permission from Kuroda, R.; Higashiguchi, K.; Hasebe, S.; Imai, Y. CrystEngComm 2004, 6, 463–468.

Figure 7 Molecular schemes of picric acid, a-naphthol, and b-naphthol.

of abrasive inorganic substances may be mediated by a liquid phase formed by frictional heating during grinding or ball
milling. A mathematical model for such reactivity has been developed by Boldyrev and Urakaev,32 who describe localized
“hot spots,” that is, areas of z10 nm in size in which the temperature can exceed 1000 K for roughly 10 ns. However, such
extreme conditions might not be relevant in the context of supramolecular transformations by grinding, which typically involve
much softer organic and metal–organic substances: a recent study of mechanochemical Diels–Alder reactivity by the Mack and
Blair groups33 indicates that the mechanochemical milling environment is analogous to solution reactions taking place at
ca. 90 C.
Supramolecular Mechanochemistry 109

Figure 8 (A) Molecular diagrams of benzophenone and diphenylamine, molecular solids that react via an intermediate eutectic phase to produce
a cocrystal composed of hydrogen-bonded molecular complexes (benzophenone)$(diphenylamine); (B) the eutectic-mediated reaction of benzophe-
none and diphenylamine observed under a microscope: crystals of individual components (left), after mixing and after the cocrystal phase has nucle-
ated (middle) and during the growth of the cocrystal phase (right). The phase diagram for the benzophenone–diphenylamine system is shown under
(C), with indicated compositions of the submerged eutectic and the melting point of the cocrystal. Adapted with permission from Chadwick, K.;
Davey, R.; Cross, W. CrystEngComm 2007, 9, 732–734.

For reactions of soft molecular solids, the liquid phase adequate for mass transfer in a mechanochemical process can also be
a submerged eutectic, that is, a low-melting mixture of reactants whose melting point temperature is also below that of the final
reaction product. In the regime where submerged eutectic formation is possible, physical contact between reactant solids leads
to a melt from which the product is formed by nucleation and crystal growth. One of the best studied cases of such eutectic-
mediated reactions involves the formation of a hydrogen-bonded cocrystal between benzophenone and diphenylamine.34 Upon
mixing of the solid components, the interface quickly becomes colored due to the formation of a thin layer of the yellow eutectic
melt. Subsequently, the entire sample transforms into a yellow eutectic melt, from which the cocrystal of composition (benzophe-
none)$(diphenylamine) nucleates and grows (Fig. 8). Eventually, the entire melt solidifies into the yellow crystalline phase
(benzophenone)$(diphenylamine).
The formation of a submerged eutectic has been recognized as an important mechanism for the mechanochemical formation of
a variety of organic compounds. Outstanding examples of such eutectic-mediated reactions have been pointed out by Rothenberg
and coworkers35 and include aldol condensation reactions, Baeyer–Villiger oxidation with m-chloroperoxybenzoic acid, ferric chlo-
ride oxidation of naphthols, ring bromination of aromatics by N-bromosuccinimide, acid-catalyzed etherification of alcohols, and
condensation reactions between aldehydes and amines that lead to the formation of aldimines (Schiff bases). The latter is partic-
ularly important for the mechanochemical assembly of complex organic structures, such as cages36 or covalent organic frame-
works.37 In the grinding synthesis of bowl-shaped cyclotriveratrylene derivatives,35 which is one of the pioneering studies in
mechanochemical, solvent-free macrocycle synthesis, the sticky nature of the reaction mixture composed of the 4-allyloxy-3-
methoxybenzyl alcohol reactant and p-toluenesulfonic acid catalyst was interpreted as a sign of an intermediate eutectic phase.
In particular, the reaction mixture exhibited minimum viscosity, probably assessed by visual observation, roughly 5 minutes
into mixing. As the reaction mixture at this point contained less than 5% of the bowl-shaped product molecule and the amount
of water that could have been introduced in the system by the deliquescence of the acid catalyst was deemed too small to yield
such an effect, the reduction in viscosity was interpreted through the formation of an intermediate eutectic.35

2.06.4.6 Mechanochemical Reactions Mediated by an “Invisible” Eutectic Phase


Formation of an intermediate eutectic phase has also been clearly observed in the context of solvent-free ligand exchange chemistry.
For example, the solid-state mixing of the triphenylphosphine (PPh3) ligand with the organometallic complex FeCp(CO)2CH3 at
40 C results in the formation of an eutectic melt in which a two-step process takes place, consisting of the coordination of the
110 Supramolecular Mechanochemistry

Figure 9 The reaction between solid Mn(CO)4(PPh3)Br and solid triphenylphosphine: (A) the reaction scheme and (B)–(G) microscopy images of
the reaction at 42 C. Thermal microscopy reveals that the reaction between the touching crystals of the two reactants (on the left of each image)
proceeds through an intermediate eutectic melt phase although the temperature is 40 C below the melting point of any of the reactants. For compar-
ison, the right side of each image displays a separate crystal of triphenylphosphine (top) and the organometallic reactant (bottom). Adapted with
permission from Manzini, S. S.; Coville, N. J. Inorg. Chem. Commun. 2004, 7, 676–678.

phosphine ligand and ligand-induced methyl migration.38 Manually grinding the two reactants together at room temperature leads
to the same transformation sequence but, in contrast to the cocrystallization of benzophenone and diphenylamine, without any
apparent melting of the reactant mixture. Since it is reasonable to assume that manual grinding could produce local environments
of temperature exceeding 40 C, this mechanochemical ligand-induced methyl transfer process was explained through the rapid
formation and recrystallization of a transient eutectic melt. In other words, the intermediate eutectic during grinding appears “invis-
ible” as it forms in minute quantities and rapidly appears and disappears. So far, the best method for detecting such “invisible”
intermediates has been thermal microscopy of the interface between touching solid reactants. When comparing eutectic-
mediated cocrystallization and the described organometallic transformation, a tentative analogy can be drawn in the light that
both reactions (at least in the first step) involve the formation of a molecular adduct and that the energetics of these transformations
at the molecular level are similar: the ligand-induced methyl migration processes have been calculated to involve steps with acti-
vation energies between 30 and 80 kJ mol 1, very comparable to the energies associated with transformations of individual
hydrogen bonds.38 These analogies can be extrapolated to propose that although organometallic bond transformations by mech-
anochemistry have so far been scarcely explored,15,38 this field might be as rich as the recently developed areas of mechanochemical
transformations of cocrystals and coordination polymers.
A further example of an “invisible” intermediate eutectic is found is the solvent-free reaction of the manganese carbonyl bromide
complex Mn(CO)4(PPh3)Br with the PPh3 ligand (Fig. 9A).39 The ligand exchange transformation appears to proceed through
a solid-state mechanism: neither visual inspection of the bulk reacting mixture nor differential scanning calorimetry (DSC) studies
have indicated the formation of a eutectic melt. However, a thermal microscopy exploration revealed the melting of the interface
between touching reactant particles at a temperature as low as 39 C, that is, around 40 C lower than the melting point of any of the
reactants (Fig. 9B–G).

2.06.4.7 Amorphous Solid Phase as a Reaction Intermediate


There are a number of mechanochemical grinding reactions that proceed without any obvious indication of eutectic melt formation
or the presence of reagents with high vapor pressure. Whereas mass transfer in such reactions can be explained by postulating a very
short-lived eutectic phase, the formation of an amorphous solid phase is also a viable option.40 The principal challenge in detecting
and establishing the existence of an amorphous intermediate in a mechanochemical process lies in the lack of long-range order,
which often makes amorphous phases elusive to most solid-state analytic methods, such as powder X-ray diffraction, solid-state
NMR spectroscopy or Fourier transform infrared spectroscopy, and Raman spectroscopy. Indeed, the most valuable methods for
Supramolecular Mechanochemistry 111

identification of amorphous phases are considered to be thermal analysis (e.g., DSC) and surface area analysis. Overall, the iden-
tification and quantification of an amorphous intermediate in a mechanochemical reaction are best accomplished through a combi-
nation of different analytic techniques. The combination of powder X-ray diffraction and terahertz (THz) time-domain
spectroscopy41 was used for the systematic investigation of mechanochemical hydrogen bond-driven cocrystallization of phenazine
with mesaconic acid.42 Both techniques are sensitive to the changes in the supramolecular arrangement of molecules rather than to
changes to the chemical structure of individual molecules. Cocrystallization of the two components to form the (phenazine)(me-
saconic acid) cocrystal from solution or by mechanical grinding was first reported by Batchelor and coworkers, who also provided
the X-ray single-crystal structural analysis.43 The cocrystal is held together by R22(8) hydrogen-bonded synthons of O–H/N and
C–H/N interactions between molecules of phenazine and mesaconic acid, which alternate to form hydrogen-bonded chains
(Fig. 10). The investigation of the solid starting materials and of the cocrystal using THz spectroscopy revealed that the cocrystal
exhibits a notable absorption band around 1.2 THz, making it readily distinguishable from either of the reactant solids
(Fig. 10). Although the interpretation of THz absorption spectra of molecular solids remains a challenging task, the simulation
of THz vibration spectra using high-level calculations indicated that the 1.2 THz absorption peak is associated with a complex vibra-
tion of the hydrogen-bonded chains in the cocrystal, involving the wagging motions of phenazine and mesaconic acid units. The
1.2 THz absorption band of the cocrystal was subsequently used to construct a THz absorption calibration curve, based on different
mixtures of the cocrystal and the individual components, which was then used for quantitative monitoring of the mechanochemical
formation of the cocrystal. The analysis was done by milling the reactants for different amount of time and then recording powder
X-ray diffraction and THz spectra. Whereas powder X-ray diffraction analysis indicated the complete disappearance of reactants
within 60 minutes, THz spectroscopy pointed to a cocrystal yield not higher than 70% (Fig. 10). Moreover, milling beyond
60 minutes induced no significant change in the reaction mixture diffraction pattern, whereas THz spectroscopy indicated an actual
reduction in the reaction yield. Discrepancies between diffraction and spectroscopic measurements clearly demonstrated that
a mechanochemical reaction involving the two organic solids can contain c. 30% of the material in an amorphous form at the point
at which X-ray diffraction indicates the presence of only the cocrystal. Moreover, extended milling can lead to the further disappear-
ance of crystalline product by amorphization. The combination of THz spectroscopy, Raman spectroscopy, and powder X-ray
diffraction was also used to monitor the reaction kinetics of the neutralization reaction between L-tartaric acid and sodium
carbonate.44
An excellent opportunity to directly observe a signal attributable to the formation of an amorphous intermediate is provided by
fluorescence emission spectroscopy. For the gold(I) complex (C6F5Au)2(m-1,4-diisocyanobenzene), shown in Fig. 11, Ito and
coworkers have established a stark change in the wavelength of the fluorescence emission maximum upon manual mechanical treat-
ment in a mortar and pestle.45 Upon grinding, the well-structured fluorescence emission lines with a maximum positioned around
415 nm transform into a broadband with a maximum at 533 nm. In addition to the position of the emission maximum wave-
length, the quantum yield of luminescence was found to increase upon grinding from 0.09 to 0.19, concomitant with significant
shortening of the fluorescence lifetime. The inspection of the ground and not ground samples by powder X-ray diffraction shows

Figure 10 (A) The reaction of phenazine and mesaconic acid to give a hydrogen-bonded cocrystal based on O–H/N hydrogen bonds (yellow
dashed lines); (B) the time-dependent yield of the (phenazine)(mesaconic acid) cocrystal, as established by terahertz (THz) spectroscopy; and (C) THz
spectra of phenazine, mesaconic acid, and the resulting cocrystal. At 60 minute time, powder X-ray diffraction indicates a complete conversion of
reactants, suggesting the presence of c. 25% of the material in the form of an amorphous phase invisible to X-ray diffraction or THz spectroscopy.
Adapted with permission from Nguyen, K. L.; Friscic, T.; Day, G. M.; Gladden, L. F.; Jones, W. Nat. Mater. 2007, 6, 206–209.
112 Supramolecular Mechanochemistry

Figure 11 Mechanoluminescent behavior of (C6F5Au)2(m-1,4-diisocyanobenzene) upon manual grinding: (A) partially ground sample under UV light
(top) and ambient light (bottom), (B) fully ground sample in a mortar (top) and the same sample after adding a drop of chloroform to the center of
the mortar (bottom), and (C) sample ground in a mortar after exposure to solvent vapors (top) and the same sample after repeated grinding with
a pestle (bottom). The behavior represented in images (A)–(C) is a clear indication of the formation of a luminescent amorphous phase upon
mechanical treatment. (D) Overlay of powder X-ray diffraction patterns of (C6F5Au)2(m-1,4-diisocyanobenzene) before (top) and after (middle)
grinding, as well as after exposure of the ground material to vapors of dichloromethane (bottom). (E) Fragment of the crystal structure of
(C6F5Au)2(m-1,4-diisocyanobenzene). Its mechanoluminescence properties were explained as a result of the transformation of short 5.19 Å Au/Au
contacts in the crystal into aurophilic interactions in the amorphous solid. Adapted with permission from Ito, H.; Saito, T.; Oshima, N.; Kitamura, N.;
Ishizaka, S.; Hinatsu, Y.; Wakeshima, M.; Kato, M.; Tsuge, K.; Sawamura, M. J. Am. Chem. Soc. 2010, 130, 10044–10045.

evidence for the complete loss of crystalline structure by grinding, whereas Raman spectroscopy indicates significant changes in
the immediate environment of the isonitrile moiety bound to the gold(I) atom. Consistent with the behavior of molecular amor-
phous materials, exposure of the amorphous ground (C6F5Au)2(m-1,4-diisocyanobenzene) to a variety of organic solvent vapors
resulted in recrystallization to form a material structurally identical to the initial crystalline form. This set of physicochemical obser-
vations indicates that the broad fluorescence maximum at 533 nm is a characteristic signature of an amorphous phase of
(C6F5Au)2(m-1,4-diisocyanobenzene), demonstrating fluorescence emission spectroscopy as a highly powerful tool that enables
the direct observation of amorphous phases. The close separation of gold atoms in neighboring molecules in the crystal structure
of (C6F5Au)2(m-1,4-diisocyanobenzene) was used to rationalize the distinct change in fluorescence emission between crystalline
and amorphous forms of the coordination compound, through the mechanochemical formation and cleavage of gold/gold auro-
philic interactions, respectively.45
The potential of fluorescence emission spectroscopy for monitoring mechanochemical transformations is not limited to organ-
ometallic complexes, as demonstrated by Zhang and coworkers in the study of the fluorescent compound difluoroboron avoben-
zone.46 Smearing of difluoroboron avobenzone into the form of a thin film on a paper surface results in a strong green fluorescence
(Fig. 12). Agitation of this film, even by a process as gentle as rubbing with a cotton swab, immediately results in the appearance of
a yellow fluorescence trace that subsequently disappears either by aging at room temperature or by gentle heating. The yellow

Figure 12 The fluorescent properties of a layer of difluoroboron avobenzone smeared onto a paper surface (A) are modified by gentle scratching
(B) and can be reversed by aging or gentle heating (C). The process is repeatable many times, as the fluorescence of the difluoroboron avobenzone
layer after heating remains sensitive to mechanical scratching (D). Adapted with permission from reference Zhang, G.; Lu, J.; Sabat, M.; Fraser, C. L.
J. Am. Chem. Soc. 2010, 132, 2160–2162.
Supramolecular Mechanochemistry 113

fluorescence has been assigned to a mechanically induced amorphous phase whose thermal relaxation leads to recrystallization and
restoration of original optical properties. Besides illustrating fluorescence emission spectroscopy as an excellent method to study of
mechanochemical transformations of fluorogenic organic molecules, this work clearly demonstrated that the formation of a non-
crystalline amorphous phase can sometimes be induced by very gentle processing.
In addition to fluorescence emission, solid-state NMR spectroscopy also has the potential for monitoring amorphous interme-
diates in mechanochemical reactions as, sometimes, the amorphous phase can exhibit a distinct set of NMR signals than the asso-
ciated crystalline phase. However, solid-state NMR spectroscopy has, with few exceptions dealing with inorganic systems in which
amorphous phases are sufficiently long-lived to permit extensive analysis,19 remained largely unexplored as a tool to study amor-
phous intermediates in mechanosynthesis.

2.06.5 Atmospheric Effects: Air Moisture and Organic Vapors

One of the most frequently encountered problems in the interpretation and understanding of mechanochemical reactions conduct-
ed using a mortar and pestle is the lack of information on the atmosphere in which the reactions were conducted. This is surprising,
as profound effects of moisture have been observed in a number of reactions conducted by such open-air grinding.

2.06.5.1 Effect of Air Moisture on the Mechanochemical Assembly of Coordination Polymers


Braga and coworkers reported that the synthesis of coordination polymers by manual grinding of ZnCl2 and 1,4-diazabicyclo[2.2.2]
octane (dabco) is strongly influenced by atmospheric moisture, leading to an interesting example of a stepwise transformation in
mechanochemical coordination chemistry.47 The first step in the reaction is the formation of a crystalline material, characterized
by powder X-ray diffraction, which transforms to the 1-D zigzag coordination polymer [Zn(dabco)Cl2]n by further grinding or
heating. The final product [Zn(dabco)Cl2]n is held together by Zn/N coordination bonds (Fig. 13). The formation of the
intermediate phase was ascribed to the hygroscopic nature of dabco and ZnCl2. Thermogravimetric analysis indicated that the
intermediate is most likely a tetrahydrate of the final product, that is, Zn(dabco)Cl2$4H2O, that can be thermally dehydrated
into [Zn(dabco)Cl2]n. The active role of moisture in the appearance of this intermediate was confirmed by conducting the grinding
reaction in a dry atmosphere, with carefully dried reactants, which led to the immediate formation of [Zn(dabco)Cl2]n. It is,
therefore, clear that the course of a mechanochemical reaction can be seriously modified by the presence of moisture.47
In contrast, no intermediate was reported in the formation of an analogous zigzag 1-D coordination polymer by mortar-and-
pestle grinding of anhydrous ZnCl2 with bipy.48 Interestingly, the mechanochemical method provided only one of the three known
polymorphic forms of the [Zn(bipy)Cl2]n polymer.
Moisture from hydrated reactants or the lack thereof can also affect the progress of a reaction. The construction of an expected 2-
D sheet polymer from anhydrous CoCl2 and bipy was not possible by manual grinding. This sheet polymer [Co(bipy)Cl2]n was,
however, readily obtained by neat grinding of CoCl2$6H2O and bipy, suggesting that the water produced by desolvation of the
reagents can play an important role in achieving mechanochemical reactivity via neat grinding (Fig. 14).
Grinding of silver acetate, AgOAc, with dabco results in a coordination polymer with a metal/dabco ratio of 1:2.47 The product of
manual grinding was formulated as the water inclusion compound Ag(dabco)2(OAc)$5H2O and proved to be identical to single
crystals obtained from solution (Fig. 15). As both starting materials were anhydrous, the absorption of five equivalents of water
must have taken place directly from air. In a similar way as in the mechanochemical reaction of dabco and ZnCl2, the formation
of Ag(dabco)2(OAc)$5H2O is a clear example of how the surrounding atmosphere can interfere with the course of the

Figure 13 The formation of the [Zn(dabco)Cl2]n coordination polymer by manual grinding in air and grinding in a dry atmosphere.47
114 Supramolecular Mechanochemistry

Figure 14 Mechanochemical reactivity of bipy toward anhydrous CoCl2 and CoCl2,6H2O.48

Figure 15 Formation of a hydrated coordination polymer by neat manual grinding of silver(I) acetate and dabco in air. The water molecules and
acetate ions that act as guests in the lattice host self-assembled from cationic chains [Ag(dabco)2(H2O)þ]n are shown using the space-filling model.47

mechanosynthesis. As the included water molecules form coordination bonds to Agþ, therefore blocking metal coordination sites
that could potentially be taken by the dabco ligand, it appears that the metal-to-ligand ratio in the final product is actively deter-
mined by the absorption of moisture upon grinding.
Absorption of moisture into a product of mechanosynthesis was also observed in the mechanochemical reaction of AgOAc with
1,4-diaminocyclohexane (dace).49 Neat manual grinding of silver(I) acetate and dace in air provided a coordination polymer tenta-
tively characterized as Ag(dace)(OAc)$nH2O. The crystal structure of Ag(dace)(OAc)$nH2O is not yet known. However, recrystal-
lization from anhydrous methanol by slow evaporation or in a stream of argon gas resulted in the formation of two structurally
similar crystalline products: Ag(dace)(OAc)$3H2O and Ag(dace)(OAc)$½H2O$CH3OH, respectively.49 Both consist of coordina-
tion polymer chains in which silver ions are coordinated by two equatorial amino groups of two bridging dace ligands. A solvent
molecule (water in Ag(dace)(OAc)$3H2O or methanol in Ag(dace)(OAc)$½H2O$CH3OH) is coordinated to the silver ion. The dace
ligands around each Agþ center adopt a cisoid transformation, in which they are positioned exactly on top of each other and perpen-
dicular to the chain. The similarity of Ag(dace)þ chains in Ag(dace)(OAc)$3H2O and Ag(dace)(OAc)$½H2O$CH3OH suggests that
the product of mechanosynthesis, Ag(dace)(OAc)$nH2O, is also composed of cisoid chains. In contrast to grinding synthesis, coc-
rystallization of AgOAc and dace from a mixture of water and methanol provides a coordination polymer Ag(dace)(OAc)$4H2O.
The polymer is composed of Ag(dace)þ chains with dace ligands arranged in a transoid conformation (Fig. 16). Thus, mechano-
chemistry and solution reaction provide two different isomeric structures of the Ag(dace)(OAc) polymer.

2.06.5.2 Catalytic Effect of Moisture in Mechanochemical Grinding Reactions


The extensive work by Braga’s group on coordination polymer synthesis by grinding provides an illustration of how water vapor can
influence the course of mechanosynthesis by becoming stoichiometrically incorporated in the final product or an intermediate of
the synthesis. However, moisture can also assist in promoting mechanochemical reactions without becoming chemically incorpo-
rated in the final product. A potential mechanism for such an effect might involve water acting as a plasticiser, that is, as a substance
that can increase the mobility and lower the glass transition temperature of an amorphous intermediate in a mechanochemical
process. In that way, the presence of water could lead to a higher mobility of the reaction mixture and also faster separation of
the mechanochemical product by crystallization. However, the effect of moisture on the mechanosynthesis of organic molecules
and materials has not yet been extensively investigated. One of the few studies available led to very surprising results in the context
of a well-known, reversible organic reaction. In the synthesis of an aldimine Schiff base from 2-hydroxy-1-naphthaldehyde and 2-
aminobenzonitrile, Cincic and coworkers50 found difficulty in reproducing the reaction yield, and the measurements of reaction
kinetics were inconsistent between different experiments. In particular, the process appeared to depend on the time of the year
when the reaction was conducted, leading the authors to systematically explore whether air moisture could affect the aldimine
Supramolecular Mechanochemistry 115

Figure 16 Construction of solvated coordination polymers from silver(I) acetate and dace by mechanochemical and solution routes.49

condensation reaction.51 A latter study revealed that although the progress of this reversible reaction should be retarded by the pres-
ence of water, increased air moisture levels led to enhanced reaction rates and higher yields. For example, aging of a manually
ground mixture for 3 days at 98% relative humidity conditions led to the complete transformation of the reactants into the product
Schiff base. In contrast, if the ground reactant mixture was stored in dry air, noticeable amounts of the starting materials were still
observable after 18 days.
A much less mechanochemically reactive system of 2-hydroxy-1-naphthaldehyde with either 4-aminobenzonitrile or 4-
aminobenzoic acid exhibited little tendency to form a condensation product by grinding and aging in air for up to three months.
In contrast, if the ground mixtures are aged at 98% relative humidity, complete conversion is achieved within the same period. These
observations are all in apparent conflict with the predictions based on the Le Chatelier’s principle and solution-based reactivity. The
explanation is most likely to be found in phase separation effects and the potential of atmospheric moisture to enhance the mobility
of molecules on the solid particle surface. Nevertheless, these examples clearly demonstrate that solid-state reactivity can be strongly
influenced by the presence of air and that the outcomes of solid-state reactions cannot always be readily anticipated on the basis of
conventional understanding of solution-based chemistry. Indeed, the study described provides a very good explanation of the irre-
producibility and conflicting results of the systematic investigations of Schmeyers and coworkers52 and of Rothenberg and
coworkers35 on the Schiff base condensation of 4-methylaniline and 4-hydroxybenzaldehyde solids.

2.06.5.3 The Interaction of Mechanochemical Reactions With Carbon Dioxide


The effect of atmosphere on grinding reactions between solids is not limited to moisture only. An example of reactivity that is modi-
fied by the presence of carbon dioxide has been provided by the Bowmaker group, who described how manual grinding of solid
silver carbonate, Ag2CO3, with PPh3 in air and in the presence of liquid ethanol leads to the absorption of one equivalent of water
and one equivalent of CO2 gas, both administered through air, to yield the hydrogencarbonate salt of the bis(triphenylphosphino)
silver(I) cation. The carbon dioxide binding is reversible, as the ground material can again release CO2 upon thermal treatment.
Monitoring of the reaction through solid-state infrared spectroscopy revealed a two-step mechanism involving solid tetrakis(triphe-
nylphosphino)silver(I) carbonate as the intermediate. Besides showing how carbon dioxide from air could affect the course of
a grinding synthesis, this example points to potential applications of mechanochemistry in environmentally friendly
technologies.53

2.06.6 Templating, Kinetic, and Thermodynamic Effects in Supramolecular Mechanochemistry


2.06.6.1 Mechanochemical Equilibrium
Equilibration of mechanochemical reactions is an important topic for understanding of self-assembly processes under mechano-
chemical conditions but is still very poorly understood. A pioneering report in studying reaction equilibration during milling
has been provided by Schrader and Hoffmann (Fig. 17) who observed that extended milling of either calcite or aragonite form
of calcium carbonate (CaCO3) consistently leads to an identical solid-state mixture of the two polymorphs, containing 30% calcite
116 Supramolecular Mechanochemistry

Figure 17 (A) The schematic representation of one of the most famous cases of mechanochemical equilibrium: the mechanochemical equilibration
of calcite and aragonite forms of calcium carbonate (CaCO3) and (B) the mechanochemical equilibration of synthetic calcite (), natural aragonite (o),
or synthetic aragonite (l) into a 70:30 equilibrium mixture of aragonite and calcite, as reported by Schrader and Hoffmann. Adapted with permission
from Schrader, R.; Hoffmann, B. Z. Anorg. Allg. Chem. 1969, 369, 41–47.

and 70% aragonite.54 By demonstrating that the reaction system always reached the same end state, independent of the starting
position, this work represents early evidence for thermodynamic equilibration of mechanochemical reactions. It also highlights
the largely unexploited potential of mechanochemistry as a means to achieve the equilibration and self-assembly of inorganic
phases under mild conditions.
The observation of constant ratios of calcite and aragonite obtained upon milling of synthetic calcite, synthetic aragonite, or
naturally occurring aragonite led Schrader and Hoffmann to propose the existence of a mechanochemical equilibrium. Since
this original report, the mechanochemical equilibration of calcite and aragonite has become one of the most extensively studied
mechanochemical processes. In the context of molecular self-assembly, the possibility of a mechanochemical equilibrium in cova-
lent mechanosynthesis was systematically investigated only in 2011,55 using a base-catalyzed disulfide metathesis as a model reac-
tion. In solution, aromatic disulfides undergo exchange to form a thermodynamic equilibrium mixture consisting of the two
homodimeric reactants and a nonsymmetrical heterodimeric product. This base-catalyzed process leads to a statistical distribution
(i.e., 1:1:2 stoichiometric ratio) of the two homodimers and the heterodimer (Fig. 18A). Belenguer and coworkers have shown that
such base-catalyzed disulfide exchange was also possible by mechanochemical milling. However, upon base-catalyzed milling of
a 1:1 mixture of the bis(4-chlorophenyl)disulfide and bis(2-nitrophenyl)disulfide, the equilibrium mixture always yielded the het-
erodimer in 97% yield (Fig. 18B). The analogous mechanochemical reaction of bis(4-methylphenyl)disulfide and bis(2-
nitrophenyl)disulfide repeatedly gave a reaction mixture containing 40% of each homodimer and 20% of the heterodimer
(Fig. 18C). For both reactions, a steady composition of the reaction mixture was reached within 60 minute milling. Repeated reac-
tions with different initial ratios of homodimers and the heterodimer all led to identical product mixtures: even when the pure het-
erodimer alone was subjected to milling in the presence of a base, identical product mixtures were obtained.
The discrepancy between disulfide equilibration in solution and by milling was rationalized through arguments of reversible
thermodynamics. In particular, relative crystal energies of solid reactants and products, which are not relevant for reaction equilib-
riums in dilute solutions, could affect mechanochemical reaction equilibriums. This has been corroborated by high-level calcula-
tions that demonstrated that the heterodimer 4-chlorophenyl-2-nitrophenyldisulfide in the crystal is stabilized by 4.5 kJ mol 1
with respect to the corresponding crystalline homodimers, which is consistent with the observed mechanochemical equilibrium.
In contrast, the 4-methylphenyl-2-nitrophenyldisulfide in the crystal is 4.5 kJ mol 1 less stable than the corresponding crystalline
homodimers, which was again consistent with observed mechanochemical equilibrium (Fig. 18C).55

2.06.6.2 Structure-Directing Effects and Seeding in Mechanochemical Reactions of Molecular Solids


Whereas surface templating and seeding effects are well-known in the formation of crystalline products from melts or solution
media, they have been significantly less explored in the context of product assembly under mechanochemical milling conditions.
Indeed, the possibility of product structure templating under mechanochemical reaction conditions, that is, under continuous
mechanical impact and shearing, can be perceived as counterintuitive. Nevertheless, structure-directing effects were described by
Drebuschak and coworkers in the investigation of mechanically and thermally induced polymorphic transformations of the anti-
diabetic drug chlorpropamide at different temperatures (Fig. 19).56 Three polymorphic structures are known for chlorpropamide:
the room-temperature stable a-form, the high-temperature stable 3-form, and the 3ʹ-form that is obtained by a solid-state polymor-
phic transformation from the 3-form at low temperatures. As indicated by powder X-ray diffraction, milling of a-chlorpropamide led
to the formation of trace amounts of a yet unknown phase at room temperature, whereas cryogrinding produced no obvious change
Supramolecular Mechanochemistry 117

Figure 18 (A) Scheme of base-catalyzed aryl disulfide exchange reaction. Comparison of the aryl disulfide exchange equilibriums in solution (left)
and under milling conditions (right) for the reaction of (B) bis(2-nitrophenyl)disulfide with bis(4-chlorophenyl)disulfide and of (C) bis(2-nitrophenyl)
disulfide with bis(4-methylphenyl)disulfide.55

Figure 19 The molecular structure of chlorpropamide and an overview of milling and cryomilling transformations between the a-, 3-, and 3ʹ-forms
of this molecule, explored by room-temperature milling and cryomilling.56

to the material. On the other hand, room-temperature milling of the metastable 3-form chlorpropamide yielded only trace amounts
of the thermodynamically stable a-polymorph. However, cryomilling of 3-chlorpropamide resulted in a much more efficient trans-
formation to the a-form. The surprising observation that a mechanochemical transformation is enhanced by a reduction of temper-
ature has been explained by the low-temperature transformation of 3- to 3ʹ-chlorpropamide. The molecular conformation and unit
cell parameters of 3ʹ-chlorpropamide are very similar to those of the a-form, presumably facilitating the mechanochemical trans-
formation. This observation, reminiscent of the Hedvall effect in which a chemical transformation is facilitated at temperatures
118 Supramolecular Mechanochemistry

nearing a phase transformation, implies that the mechanochemical formation of a-chlorpropamide is facilitated by its nucleation
on the almost isostructural 3ʹ-form.56
Direct evidence of a heterogeneous seeding effect in mechanochemical transformations of molecular organic solids has been
offered by Cincic and coworkers, who conducted the mechanochemical condensation of solids 2-hydroxy-1-naphthaldehyde
and 2-aminobenzonitrile to form an aldimine product known to exist in four different polymorphs designated as forms I–IV.50
Dry grinding of the reactants at 20 C led to the formation of a solid crystalline phase, identified by powder X-ray diffraction as
the form II. However, grinding in the presence of added seed crystals of form III resulted in the formation of the same product mole-
cule but as the polymorphic form III. Consequently, the addition of seeds of form III into the covalent bond-forming reaction
mixture enabled control over the polymorphic composition of the product.

2.06.6.3 Stepwise Reactions as a Result of Solid-State Supramolecular Competition


The appearance of short-lived crystalline intermediates in mechanochemical cocrystallization was first reported by Cincic et al.57
who explored the synthesis of halogen-bonded cocrystals of liquid thiomorpholine and solid 1,4-diiodotetrafluorobenzene in
a stainless steel ball-milling assembly. Upon milling the two components in equimolar ratio for 30 minutes, the anticipated
halogen-bonded polymer cocrystal was obtained (Fig. 20). However, if the milling was interrupted after 5 or 20 minutes, powder
X-ray diffraction of the reaction mixture indicated the formation of a different crystalline phase that, upon further milling, trans-
formed into the 1:1 cocrystal. Subsequent screening of a variety of solution crystallization conditions led to the serendipitous
synthesis of another cocrystal composed of thiomorpholine and 1,4-diiodotetrafluorobenzene but in a respective 2:1 stoichio-
metric ratio. Comparison of the measured powder X-ray diffraction pattern of the mechanochemical intermediate to that simu-
lated for the new cocrystal revealed the reaction intermediate to be (thiomorpholine)2(1,4-diiodotetrafluorobenzene). The
same cocrystal could be obtained mechanochemically, simply by milling the two starting materials in the appropriate stoichio-
metric ratio. The formation of the cocrystal (thiomorpholine)2(1,4-diiodotetrafluorobenzene) as an intermediate in the synthesis
of (thiomorpholine)(1,4-diiodotetrafluorobenzene) was interpreted through kinetic arguments. Notably, the components in the
intermediate cocrystal are held together through strong I/N halogen bonds, whereas in the final product (thiomorpholine)(1,4-
diiodotetrafluorobenzene), the interactions between building blocks are a combination of strong I/N and weak I/S interactions.
Consequently, the mechanochemical reaction immediately led to the formation of a kinetic product based on strongest possible
supramolecular interactions, which subsequently reacted with another equivalent of 1,4-diiodotetrafluorobenzene through the
formation of additional weaker I/S halogen bonding interactions. Very similar behavior was observed (Fig. 20) for a related
halogen bond donor 1,2-diiodotetrafluorobenzene. However, in this case, the crystalline intermediate consisted of five-
membered supramolecular assemblies.
The formation of mechanochemical reaction intermediates in the synthesis of halogen-bonded cocrystals was interpreted as
a result of using a heteroditopic reaction component, in particular a halogen bond acceptor (thiomorpholine) displaying weak
sulfur-based and strong nitrogen-based bonding sites. Presumably, the supramolecular competition between these sites enables
the formation of kinetic intermediates. This mechanistic proposition was additionally supported by Karki and coworkers who
explored the mechanosynthesis of pharmaceutical model cocrystals involving a,u-aliphatic dicarboxylic acids with a heterodi-
topic molecule nicotinamide (vitamin B3).58 Nicotinamide provides two different sites for hydrogen bond-driven cocrystallization
with carboxylic acids: the pyridine moiety and the amide group. As a result, two varieties of cocrystals are possible for nicotinamide
with dicarboxylic acids, containing the respective components in a 2:1 and 1:1 stoichiometric ratio (Fig. 21). Mechanochemical
cocrystallization of nicotinamide and suberic acid in a 1:1 stoichiometric ratio provides the cocrystal (nicotinamide)(suberic acid)
as the ultimate product, without any intermediates observable by powder X-ray diffraction. The mechanochemical reaction of

Figure 20 The stepwise mechanochemical synthesis of halogen-bonded cocrystals, as reported by Cincic et al.57 The formation of reaction interme-
diates is explained by the supramolecular competition between the strong (N) and weak (S) halogen bonding sites on the thiomorpholine molecule.
Supramolecular Mechanochemistry 119

Figure 21 (A) Schematic representation of hydrogen-bonded structures in the cocrystals composed of nicotinamide and a dicarboxylic acid in a 1:1
(top) and 2:1 (bottom) respective ratio. Powder X-ray diffraction patterns for the mechanochemical reaction of nicotinamide and suberic acid in (B)
1:1 stoichiometric ratio and (C) 2:1 respective stoichiometric ratio. The X-ray reflections of most characteristic for the (nicotinamide)(suberic acid)
cocrystal are labeled with “*” and for the (nicotinamide)2(suberic acid) with “ ”.58
l

nicotinamide and suberic acid in a 2:1 stoichiometric ratio ultimately (after 60 minute milling) yields the expected
(nicotinamide)2(suberic acid) cocrystal. However, interrupting the mechanochemical milling after 5 minutes reveals the formation
only of the alternative cocrystal (nicotinamide)(suberic acid). After 20 minute milling, both the intermediate (nicotinamide)(suberic
acid) and the ultimate product (nicotinamide)2(suberic acid) are present (Fig. 21). The stepwise cocrystallization of nicotinamide and
suberic acid was explained by the competition of supramolecular synthons. The intermediate 1:1 cocrystal was held together by
a combination of the R22(7) pyridine–carboxylic acid heterodimeric motif (c. 42 kJ mol 1) with the strongest possible R22(8)
amide–carboxylic acid heterodimer (c. 60 kJ mol 1).58
Formation of the final product, which is the cocrystal whose composition is dictated by the stoichiometric composition of the
reaction mixture, involves the replacement of the 60 kJ mol 1 R22(8) amide–carboxylic acid heterodimer with a combination of two
weaker hydrogen-bonded motifs: the amide–amide R22(8) homodimer (c. 52 kJ mol 1) and another R22(7) carboxylic acid–pyridine
heterodimer (40 kJ mol 1). Consequently, although the formation of the final product (nicotinamide)2(suberic acid) is favorable
by considering the energies of synthesized supramolecular motifs, the formation of the intermediate (nicotinamide)(suberic acid)
could be kinetically favored by the strong acid–amide R22(8) heterodimeric synthon.58
The formation of reaction intermediates was also observed by Tumanov and coworkers in the self-assembly reactions of organic
solids involving proton transfer.59 In particular, the solid-state reaction of glycine and oxalic acid dihydrate in a 1:1 stoichiometric
ratio was initially found to yield bis(glycinium)oxalate. Further reaction led to the rearrangement of this product with remaining
oxalic acid dihydrate to yield the final product glycinium hydrogen oxalate.

2.06.6.4 Stepwise Reactions in the Synthesis of Coordination Polymers and MOFs


In the systems examined so far involving supramolecular and covalent transformations, the stepwise mechanochemical reaction
mechanism results from the competition of different reaction or binding sites. The synthesis of solvated coordination polymers
120 Supramolecular Mechanochemistry

and porous MOFs, however, provides another rationale for a stepwise reaction mechanism. As established by Friscic and Fábián, the
mechanochemical reaction of zinc oxide and fumaric acid can be greatly accelerated by the addition of different organic liquid
phases, whereas the addition of a stoichiometric amount of water leads to the selective and specific assembly of the hydrated coor-
dination polymers zinc fumarate tetrahydrate or zinc fumarate pentahydrate.60 A subsequent study revealed that the zinc fumarate
pentahydrate appears as an intermediate in the synthesis of the tetrahydrate form and subsequently rearranges (Fig. 22), in the pres-
ence of unreacted starting materials, to form the tetrahydrate.41 The stepwise mechanism can be explained through a mass action
effect in which the presence of a relatively large amount of liquid water reactant at the onset of the reaction leads to the formation of
the highest possible hydrate of the product. This initially formed pentahydrate furthermore acts as a “sink” for reactant water present
in the reaction mixture. The depletion of water by the highly hydrated intermediate triggers a change in reaction kinetics from a rapid
liquid-assisted process to a slower dry grinding reaction that was shown to be mediated by an amorphous phase.
The proposed mass action effect, leading to the formation of a highly solvated intermediate in the presence of a liquid reactant,
has also been observed in a different system involving a water-catalyzed reaction of copper(II) oxide with acetic acid. Milling of
copper(II) oxide in the presence of water leads to a slow reaction that is, however, accelerated by the addition of a few molar percent
of water. Such an accelerated reaction quantitatively yields copper(II) acetate monohydrate within 90 minutes of milling. If milling
is, however, interrupted after 15 minutes, powder X-ray diffraction analysis of the sample reveals the presence of a hydrogen-
bonded acetic acid solvate (Fig. 22) of copper(II) acetate.60 Again, the formation of this crystalline intermediate phase is explicable
through considering the relatively large stoichiometric excess of the liquid acetic acid reagent over the nascent copper(II) acetate in
the early stages of reaction.
The mass action effect of the liquid in the mechanochemical system can also be used to explain the appearance of reaction steps
in the mechanosynthesis of porous MOFs. Notably, the mechanochemical ion- and liquid-assisted grinding methodology (ILAG)
was used by Beldon and coworkers to conduct the synthesis of open zeolitic imidazolate frameworks (also known as ZIFs) through
milling of zinc oxide and a stoichiometric amount of substituted imidazole ligand in the presence of an organic liquid and a sub-
stoichiometric ammonium salt additive.62 The role of the organic liquid in such mechanochemical transformations is to act as
a space-filling agent or even a template for the synthesis of an open framework, whereas the salt accelerates the reaction through
a mechanism that is not yet completely clear but might involve proton transfer. When 2-methylimidazole was used as the ligand,
the expected porous sodalite (SOD) topology framework, known as ZIF-8, was readily obtained. However, with 2-ethylimidazole as
the ligand, the reaction yielded a previously not known nonporous framework with a nonzeolitic, quartz (qtz) topology. Interrupt-
ing the mechanochemical reaction at shorter grinding times revealed that the formation of the nonporous qtz framework takes place
through several intermediates (Fig. 23). The first reaction product, observed by powder X-ray diffraction of the sample milled for 5
or 10 minutes, was a mixture of previously reported63 porous frameworks with a zeolite r (RHO) and analcime (ANA) topologies.
After 20 minute milling, the reaction mixture was found to consist only of the ANA topology framework. The analysis of all three
structures observed in the mechanochemical reaction of ZnO and 2-ethylimidazole reveals that their density increases in the order
RHO < ANA < qtz, that is, in a sequence opposite to their appearance in the reaction mechanism. The relative densities of the
product frameworks suggest a reaction mechanism resembling the proposed solvent mass action effect. In such a scenario, the large
excess of the milling liquid with respect to the nascent zinc imidazolate leads first to the formation of a product framework with the
highest capacity for solvent inclusion. This highly porous initial product subsequently rearranges into less porous, higher density
forms. By analogy to zeolite structures, such order of framework transformations from lower to higher density is expected to be
thermodynamically driven, meaning that the sequence of reaction steps conforms to Ostwald’s rule of stages.62

Figure 22 Stepwise mechanochemical reaction mechanisms observed in the formation of solvated coordination compounds: (A) zinc fumarate and
(B) copper(II) acetate.40,61 Adapted with permission from Strobridge, F. C.; Judas, N.; Friscic, T. CrystEngComm 2010, 12, 2409–2418.
Supramolecular Mechanochemistry 121

Figure 23 The stepwise mechanism in the mechanosynthesis of zeolitic imidazolate frameworks. The densities of metal–organic frameworks
(MOFs) intermediate, and the final product is expressed as T/V, which is the number of tetrahedral sites (T, zinc sites) per unit volume (V).62

The examples of kinetic product formation in the synthesis of coordination compounds discussed so far appear to be strongly
related to the presence of a liquid phase that either binds to the metal ion or acts to stabilize a particular solvate, an inclusion
compound, or a porous structure. The mechanochemical reaction of cadmium chloride with the simple organic ligand cyanogua-
nidine offers an example in which stepwise reactivity may be related to topological aspects of reaction products.64 Due to the avail-
ability of two different metal binding sites on cyanoguanidine, milling of CdCl2 with this ligand is expected to yield two different
coordination frameworks of composition Cd(cyanoguanidine)Cl2 and Cd(cyanoguanidine)2Cl2. However, milling of the two
components in respective ratios 1:1 and 1:2 always gave the product Cd(cyanoguanidine)Cl2. Only after employing harsher milling
conditions, achieved by increasing the milling ball weight, was the 1:2 mixture of CdCl2 and (cyanoguanidine) successfully con-
verted to the predicted Cd(cyanoguanidine)2Cl2. Under such conditions, Cd(cyanoguanidine)Cl2 was observed as an intermediate
in the synthesis of Cd(cyanoguanidine)2Cl2. This stepwise mechanism was explained through differences in the crystal structures of
the two coordination polymers, which were determined directly from powder X-ray diffraction patterns of the mechanochemical
reaction mixtures. The polymer Cd(cyanoguanidine)Cl2 was found to be a three-dimensional (3-D) coordination polymer frame-
work, whereas Cd(cyanoguanidine)2Cl2 is composed of 2-D coordination polymers sheets connected through N–H/N hydrogen
bonds. Consequently, the Cd(cyanoguanidine)Cl2 structure can be regarded as a kinetic “trap” whose transformation into the final
reaction product requires the dismantling of a 3-D framework of coordination bonds to form a product of lesser dimensionality.64

2.06.7 In Situ Reaction Monitoring

One of the major drawbacks of mechanochemical milling and grinding chemistry is the inability of real-time observing the trans-
formation of reactants into products. As lamented by Drebuschak and coworkers, “it is a challenge to understand the processes
taking place in a powder sample during its grinding in a mill, or compacting, since one can neither measure local temperature, pres-
sure, shear stresses nor follow the changes in the diffraction patterns or vibrational spectra in situ.”56 Indeed, the ability of in situ
monitoring mechanochemical reactions has been largely limited to temperature and pressure measurements on mechanically
induced self-sustaining reactions or on monitoring pressure changes in inorganic reactions adsorbing or releasing a gas.30 Very
recently, however, the application of highly penetrating synchrotron X-ray radiation enabled Friscic and coworkers to conduct
a real-time in situ study of mechanochemical synthesis of porous MOFs.65–67 The excellent penetrating power of X-rays with a short
wavelength (0.14 Å) enabled the recording of the powder diffraction patterns of reaction mixtures as these have been milling, with
a time resolution in seconds.
These real-time structural studies have painted an unexpectedly active picture of mechanochemical milling reaction, involving
highly labile intermediate phases whose lifetime is less than a minute, dependent on milling conditions and environment, and
whose chemical and particle size evolution could be measured in real time and in situ (Fig. 24). For the mechanochemical synthesis
of the porous framework material ZIF-8 in situ, studies have enabled the quantification of product formation rates for reactions
conducted by liquid-assisted grinding (LAG) and by ILAG. Assuming Avrami–Erofeev kinetics, switching from LAG to ILAG by addi-
tion of a catalytic salt led to a c. 15-fold increase in the rate constant for product formation, specifically from k ¼ 0.00158(1) to
0.0212(9) s 1.65

2.06.8 Mechanochemical Cocrystal Synthesis


2.06.8.1 Synthesis of Three-Component Cocrystals
By avoiding the limitations of solution-based synthesis, mechanochemistry can circumvent some of the persistent challenges of
supramolecular chemistry, such as the synthesis of cocrystals containing more than two different constituents. Although supramo-
lecular designs for the construction of three-component (ternary) cocrystals can be devised (Scheme 1A), the synthesis of such mate-
rials from solution is often hampered in practice by differences in solubility between different cocrystal components or the
formation of alternative cocrystal or solvate phases.68 Thus, attempts to construct multicomponent cocrystals are likely to result
with cocrystal components either crystallizing as separate single-component solids or forming simpler products, such as a less
soluble binary cocrystal (Scheme 1B).
122 Supramolecular Mechanochemistry

Figure 24 Time-resolved in situ powder X-ray diffraction patterns for the mechanochemical synthesis of the MOF ZIF-8 using (A) liquid-assisted
grinding and (B) ion- and liquid-assisted grinding. (C) The comparison of reaction rates for the two mechanochemical approaches is illustrated by
time-dependent increase in the intensity of the characteristic (211) X-ray reflection of the product. Superimposed to each time-resolved diffractogram
is the simulated diffraction pattern for the ZIF-8 product. Adapted with permission from Friscic, T.; Halasz, I.; Beldon, P. J.; Belenguer, A. M.; Adams,
F.; Kimber, S. A. J.; Honkimäki, V.; Dinnebier, R. E. Nat. Chem. 2013, 5, 66–73.

Scheme 1 (a) Schematic representation of a design for the synthesis of a three-component (ternary) complex based on molecular building blocks
with complementary shapes and/or functionalities and (b) illustration of challenges and potential outcomes when the synthesis of such a ternary coc-
rystal is attempted using a solution-based approach.

By avoiding solubility effects, mechanochemical methods make it possible to explore designs for three-component molecular
recognition in the solid state, which are not readily accessible from solution. Pioneering studies68,69 have been conducted by Kur-
oda’s group, who used mechanochemical grinding to discover and develop a general family of three-component cocrystals held
together by a combination of strong O–H/O hydrogen bonds and charge-transfer interactions. The targeted cocrystals were
composed of the electron-deficient p-benzoquinone, the electron-rich bis-b-naphthol (BINOL), and an arene as building blocks.
The formation of cocrystals was readily recognized by the very intensive color of the product, arising from extensive charge-
transfer interactions. Grinding of p-benzoquinone with racemic BINOL and naphthalene provides the blue multicomponent coc-
rystal (rac-BINOL)2$(p-benzoquinone)$(naphthalene)2, which was also obtained from solution, based on a three-component
p-stacked motif of benzoquinone sandwiched between two rac-bis-b-naphthol molecules (Fig. 25A). The motifs form chains
through O–H/O hydrogen bonds. The stacking of the chains results in cavities that confine the naphthalene molecules by
C–H/p contacts. Grinding of the optically pure R-BINOL, however, with p-benzoquinone and naphthalene results in the forma-
tion of a red three-component cocrystal of composition (R-BINOL)2$(p-benzoquinone)2$(naphthalene)3. The same product was
also obtained by solution crystallization. The different colors of the three-component cocrystals of p-benzoquinone and naphtha-
lene with racemic (blue) and chiral (red) BINOL illustrate the ability to simply differentiate between chiral and racemic forms of
Supramolecular Mechanochemistry 123

Figure 25 (A) Reactivity of a mixture of benzoquinone (red) and naphthalene (yellow) toward enantiomerically pure (left, with omitted naphthalene
molecules) and racemic (right) BINOL (blue) in solution or in solid state and (B) difference in reactivity of a mixture of benzoquinone (red), anthra-
cene, and racemic BINOL (blue) upon mechanochemical grinding (left) and in solution (right). For clarity, anthracene molecules have been omitted,
and chromophoric supramolecular motifs are indicated by circles.69,70 Modified and adapted with permission from Friscic, T. Chem. Soc. Rev. 2012,
41, 3493–3510.

a compound through the color of a ground mixture. The color difference is explained by the distortion of the chromophoric three-
membered stacks, which, in the chiral cocrystal, assemble to form helical chains (Fig. 25A).69
Grinding of anthracene with p-benzoquinone and rac-BINOL leads to the formation of a red solid of composition (rac-
BINOL)2$(p-benzoquinone)2$(anthracene). From solution, the same components provided a different three-component solid,
the blue (rac-BINOL)2$(p-benzoquinone)$(anthracene)2. This cocrystal consisted of similar p-stacked and hydrogen-bonded chains
observed in the blue naphthalene analogue (Fig. 25B). However, since the red (rac-BINOL)2$(p-benzoquinone)2$(anthracene)
could only be obtained by a solid–solid reaction, its structural analysis via single-crystal X-ray diffraction was impossible. Structure
determination of the three-component cocrystal was successfully accomplished by Kuroda, Harris, and coworkers in a pioneering
report70 on structure determination using powder X-ray diffraction data. The red color of the cocrystal was explained by the absence
of the blue chromophore BINOL–benzoquinone–BINOL stacks. Instead, the structure contains of four-component BINOL–benzo-
quinone–benzoquinone–BINOL stacks connected into chains by O–H/O hydrogen bonds (Fig. 25B). The anthracene component
is placed in cavities between the chains and is held in the structure by extensive C–H/p bonding.
The systematic investigation of the relative efficiencies of neat grinding, LAG, and solution growth for the synthesis of ternary
complexes explored the inclusion compounds of a hydrogen-bonded host of caffeine and succinic acid.24,71 Although caffeine
and succinic acid do not form a binary hydrogen-bonded cocrystal, crystallization from dioxane readily yields a ternary cocrystal
of composition (caffeine)$(succinic acid)$(dioxane)0.66 in which dioxane guests are included in channels of a self-assembled
hydrogen-bonded host (caffeine)$(succinic acid) (Fig. 26A). A systematic exploration of 25 different guest molecules, including
dioxane, revealed the formation of a solid-state ternary complex with four potential guests when using solution-based methods.
In contrast, neat grinding led to the formation of the ternary complex in 15 cases, and LAG enabled the synthesis of 18 ternary solids,
clearly demonstrating the advantages of mechanochemical grinding in the assembly of complex hydrogen-bonded solids.71

2.06.8.2 Systematic Studies of Structure-Templating Effects in Cocrystal Formation


Because it is efficient and requires short reaction times (typically 20 minutes), LAG is an excellent method to systematically explore
factors underlying molecular recognition and templating in molecular solids. This potential of LAG was first systematically utilized
in exploring the structure-directing (templating) properties of guests in the formation of hydrogen-bonded host of caffeine and suc-
cinic acid.24,72 Notably, besides the previously described open framework (caffeine)$(succinic acid), the same two components also
form an alternative host of composition (caffeine)4$(succinic acid) (Fig. 26B).
Systematic LAG screening in the presence of potential guests (shown in Fig. 26C) carefully chosen to address differences in
molecular size, shape, or functionality revealed that the formation of the (caffeine)$(succinic acid) host was the most likely event
for guests with molecular size between 3.5 and 3.8 Å. The exception was haloform-type guests whose presence induced the forma-
tion of (caffeine)4$(succinic acid) (Fig. 27A). This mechanochemically established templating of the (caffeine)4$(succinic acid)
framework by haloforms was subsequently rationalized by crystal structure determination, which revealed close interactions
between the guest and the host framework (Fig. 27B). These interactions included the guest molecules as the donor of both
C–H/O hydrogen bonds and Cl/N (or Br/N) halogen bonding interactions. That both types of interactions are needed to
support the (caffeine)4$(succinic acid) host was established by mechanochemical screening experiments in which either the
haloform hydrogen atom was replaced with a similarly sized fluorine or the halogen atom was replaced with a methyl group of
similar size.71,72
124 Supramolecular Mechanochemistry

Figure 26 (A) Fragment of the (caffeine)$(succinic acid) host3 with guests omitted; (B) fragment of the (caffeine)4$(succinic acid) host with guests
omitted and (C) the library of potential guest molecules used in mechanochemical LAG screening of parameters that guide the formation of different
host structures based on caffeine and succinic acid.

Figure 27 (A) The mechanochemically established relationships between the guest structure and the preference for the formation of different
hydrogen-bonded host frameworks based on caffeine and succinic acid and (B) the hydrogen C–H/O and halogen Br/N bonds in the bromoform
inclusion compound of the (caffeine)4$(succinic acid) host framework.71,72

2.06.8.3 Screening for Recognition Motifs of Biomolecules


The described mechanochemical screening approach that was used to deduce which structural parameters of a guest molecule are
most relevant for the formation of inclusion host frameworks based on caffeine and succinic acid was subsequently applied to inves-
tigation of molecular recognition properties of steroids, leading to the unexpected discovery of a previously unknown self-assembly
motif of steroid sex hormones. The study, inspired by the pioneering investigations of steroid cocrystallization by Eger and
Norton,73 explored the propensity of steroids progesterone, pregnenolone, b-estradiol, and estrone toward cocrystallization with
arenes (Fig. 28A).74 Challenging each of the steroids with a library of 24 arene countermolecules (Fig. 28B) revealed that unlike
the other three steroids in the test, progesterone exhibited a remarkable ability for cocrystal formation with aromatic molecules.
Indeed, progesterone cocrystallization, observed through powder X-ray diffraction, took place regardless of the hydrogen bonding
functionalities on the arene. From all arenes in the test library, only perfluoronaphthalene and xanthines did not yield a cocrystal
with progesterone (Fig. 28B). Such mechanochemically established propensity of progesterone for cocrystallization with arenes was
rationalized through a combined structural and computational study involving single-crystal and powder X-ray diffraction structure
Supramolecular Mechanochemistry 125

Figure 28 (A) Explored steroids and (B) arene cocrystallization partners with indicated ability to form cocrystals with progesterone and b-estradiol.
(C) A progesterone–arene “sandwich” complex involving in the cocrystal with pyrene.74

determination, molecular shape analysis, and crystal structure prediction (CSP). Structural analysis and CSP computational exper-
iments revealed the persistent formation of three-component supramolecular assemblies in which the steroid fragments form
a “sandwich” around the arene (Fig. 28C). This type of molecular assembly revealed a new motif of steroid recognition that desig-
nated the a/p interaction. The proposed a/p interaction74 occurs between the a-surface of the steroid molecule and the p-system
of the arene as a result of highly complementary electrostatic molecular surface potentials. In contrast to the persistent a/p stacking
motif that was observed in cocrystals of progesterone, no such structural regularity was observed for the several prepared cocrystals
involving steroids pregnenolone, b-estradiol, and estrone. Specifically, mechanochemical screening for cocrystals of pregnenolone
yielded a cocrystal held together by hydrogen bonds, whereas the exploration of b-estradiol cocrystallization revealed a binary coc-
rystal with pyrene based on an open hydrogen-bonded framework of b-estradiol with molecules of pyrene included as guests.
In a subsequent study, Ardila-Fierro et al.75 conducted a systematic mechanochemical screen for cocrystal formation involving
b-estradiol and different arenes. As with progesterone, the screening revealed clear relationships between structure of the arene and
the ability to form a cocrystal. In particular, cocrystals were found preferably for arenes composed of three or four fused aromatic
rings, such as phenanthrene, pyrene, and different isometric phenantridines. Subsequent crystal structure analysis provided an
explanation of that trend, by revealing that the cocrystals are, in fact, host–guest lattice compounds based on a 3-D lattice host
of b-estradiol. The host framework is held by O–H/O hydrogen bonds between steroid molecules and exhibits channels with
a square cross section, occupied by arene molecules as guests. That the dimensions of the lattice host channels are largely deter-
mined by the geometry of the b-estradiol molecule provided an explanation for the observed preferences in cocrystallization
with arenes.75

2.06.8.4 Halogen-Bonded Cocrystallization


Mechanochemistry has found extensive application in the synthesis of molecular or ionic materials based on hydrogen bonds,
including cocrystals, salts, and salt cocrystals. However, in the context of halogen-bonded cocrystal systems, synthesis by milling
or grinding has been much less exploited. One of the first reports of mechanochemical formation of halogen-bonded structures
was provided by Cincic and coworkers,76 who utilized 1,4-dibromo- or 1,4-diiodoperfluorobenzene as halogen-bond donors, in
combination with oxa-, aza-, or thiaheterocycles as acceptors. Ball milling of different combinations of donors or acceptors in
a 1:1 stoichiometric ratio led to rapid, quantitative assembly of a set of eight isostructural halogen-bonded cocrystals. As the coc-
rystals were isostructural, the evaluation of their thermal stability (i.e., melting points) provided a simple entry to evaluate the rela-
tive strengths of different types of halogen bonding interactions. This approach revealed the following ordering of halogen bonds, in
the order of decreasing strength: I/N > Br/N > I/O z I/S > Br/O.
Subsequent study of the reaction of 1,4-diiodotetrafluorobenzene or 1,2-dioodotetrafluorobenzene with thiomorpholine
revealed a stepwise reaction mechanism, dictated by the kinetic formation of stronger N/I halogen bonds. This mechanism is
described in more detail in “Stepwise Reactions as a Result of Solid-State Supramolecular Competition” section. Further sets
126 Supramolecular Mechanochemistry

Figure 29 The design of a family of isostructural cocrystals through structural equivalence of aromatic C–H and N groups and halogen-bonded Br
and I donor atoms. The synthesis of the entire isostructural set required the use of mechanochemistry, as attempts to obtain the cocrystal of phena-
zine and bromopentafluorobenzene from solution were unsuccessful.78

of isostructural cocrystals based on 1,4-diiodotetrafluorobenzene or 1,4-diiodotetrafluorobenzene were also obtained by milling


with either acridine or phenazine as the halogen bond acceptor.77 In all cases, the differences in melting points of isostructural coc-
rystals indicated that I/N halogen bonding arrangements are more stable to corresponding Br/N ones. The reliable isostructur-
ality of cocrystals based on I/N and Br/N interactions led to a conclusion that the halogen-bonded iodine and bromine groups
are structurally equivalent. This concept was subsequently combined with the established structural equivalency of aromatic
nitrogen and C–H groups to deliberately design a family of four isostructural cocrystals, by cocrystallization of iodo- or bromopen-
tafluorobenzene as halogen bond donors with phenazine or acridine as halogen bond acceptors.78 The cocrystals were readily ob-
tained by milling the halogen bond acceptor and the halogen bond donor in a 1:1 stoichiometric ratio, and their isostructurality was
confirmed by powder X-ray diffraction. However, attempts to obtain the cocrystals from solution were successful only for combi-
nations of iodo- and bromopentafluorobenzene with acridine and for the combination of iodopentafluorobenzene with phena-
zine. The cocrystal of bromopentafluorobenzene and phenazine could not be obtained from solution, which was rationalized
due to low solubility or phenazine combined with the weakness of Br/N halogen bonds (Fig. 29). Mechanochemical milling
was also employed by Lapadula and coworkers to demonstrate the first designed synthesis of a halogen-bonded cocrystal involving
a coordination complex as a halogen bond acceptor.79

2.06.9 Combining Different Types of Molecular Self-Assembly

The ability to conduct multiple, orthogonal self-assembly processes provides an attractive route to synthesize increasingly complex
structures from simple building blocks. Solid-state mechanochemical reactions are readily applicable for such synthetic strategies, as
demonstrated by Friscic et al.80 who screened for lattice inclusion compounds of self-assembled metal–organic hosts by ball milling
of mononuclear nickel(II) or cobalt(II) dibenzoylmethanate complexes with isonicotinamide and a potential liquid guest. The
milling led to the formation of new inclusion compounds through a combination of three different types of self-assembly processes,
notably ligand exchange on the metal center, hydrogen bond-driven self-assembly of the resulting coordination complexes, and the
formation of an inclusion compound with the liquid guest (Fig. 30). Combinations of other types of self-assembly processes under
mechanochemical conditions have been explored by the James group, who combined the reversible aldimine condensation reac-
tion with coordination-driven self-assembly to produce a zinc complex of the popular salen ligand by a one-pot reaction of ZnO,
ethylenediamine, and salicylaldehyde.81 A similar process was also investigated by Cincic and coworkers who demonstrated the
one-pot synthesis of halogen-bonded cocrystals of a metal–organic complex by one-pot assembly of Schiff base ligand, its coordi-
nation to a metal salt, and the cocrystallization of the resulting complex with 1,4-diiodotetraflurobenzene.82

2.06.10 Cages and Interlocked Structures


2.06.10.1 Molecular and Supramolecular Cages
Mechanochemical reactions are highly versatile for the assembly of molecular or supramolecular cyclic or cage structures. Indeed,
one of the first reports of mechanochemical assembly of coordination bonds by manual grinding also revealed the fast, simple
Supramolecular Mechanochemistry 127

Figure 30 Example of mechanochemical process that combines different types of self-assembly in one pot. Adapted with permission from Friscic,

T.; Mestrovic, E.; Skalec- 
Samec, D.; Kaitner, B.; Fábián, L. Chem. Eur. J. 2009, 15, 12644–12652.

formation of open structures. Specifically, Orita and coworkers reported that manual grinding of cis-protected and kinetically inert
metal–organic building blocks, such as (ethylenediamine)palladium(II) nitrate or (ethylenediamine)platinum(II) nitrate with 4,4ʹ-
bipyridine (bipy), leads to the rapid (within minutes) synthesis of eight-membered cationic molecular squares (Fig. 31A).83 Impor-
tantly, the yield of the platinum-based molecular square after only 10 minutes of manual grinding in open air was comparable to
that obtained after four weeks at 100 C in solution. The mechanochemical synthesis was also readily expanded to the synthesis of
metal–organic bowls based on the tridentate ligand 1,3,5-tris(3-pyridyl)triazine (Fig. 31A). In this case also, solvent-free mechano-
chemical self-assembly led to the rapid formation of the palladium(II)-based molecular bowl in 90% yield after only 10 minutes of
grinding and 5 minutes of solvent-based work-up. For comparison, solution-based synthesis was reported to provide c. 56% conver-
sion within 4 hours. Copper(I) helicates could also be synthesized mechanochemically, with similar advantages: compared with the
five-hour solution-based procedure, the use of mechanochemistry enabled the complete conversion to a metal helicate in only
10 minutes.44
The described benefits of mechanochemical synthesis also expand to the chemistry of organic rings and cages. An early, highly
remarkable report of using manual grinding for the synthesis of a cage structure stems from the Atwood group,84 who obtained calix
[4]pyrogallolarene by grinding equimolar quantities of isovaleraldehyde and pyrogallol in the presence of p-toluenesulfonic acid as
a catalyst. The cyclic product calix[4]pyrogallolarene was obtained within 5 minute grinding, as a result of an eight-molecule (4 þ 4)
condensation reaction (Fig. 31B). A combination of solid-state NMR analysis, solution diffusion coefficient measurements, and
mass spectrometry revealed that the calix[4]pyrogallolarene product assembled into six-membered spherical hydrogen-bonded
assemblies, capable of enclosing a volume of > 1300 Å3. Consequently, only 5 minutes of mechanochemical treatment were suffi-
cient to induce the assembly of 48 reactant molecules into covalent- and hydrogen-bonded cages. In this case, the formation of
supramolecular cages is especially remarkable, as the calix[4]pyrogallolarene is also known to adopt an alternative type of structure
in the solid state, based on hydrogen-bonded molecular bilayers. The formation of cages was tentatively explained by templating of
the cage structure via water molecules, which either were generated by the reaction or have entered the reaction mixture from air.84
Importantly, the formation of the cage structure may also have been assisted by the containment of unreacted starting materials,
which has been suggested by diffusion studies of chemical shifts arising from residual reactants.

Figure 31 (A) Mechanochemical synthesis of a coordination molecular bowl (left) and a molecular square (right) demonstrated by Orita et al.83;
(B) mechanochemical synthesis and self-assembly of pyrogallol[4]arene capsules by Antesberger et al.84; and (C) a space-filling of one of the organic
cages mechanochemically prepared by Içli et al.85
128 Supramolecular Mechanochemistry

This pioneering work in mechanochemical synthesis of hydrogen-bonded supramolecular cage structures was extended to fully
covalent structures by Severin’s group. Specifically, Içli et al. have utilized mechanochemical ball milling to conduct the three-
component reaction of pentaerythritol with 4-formylphenylboronic acid and 1,3,5-trisaminomethyl-2,4,6-triethylbenzene.85 In
solution, under refluxing conditions and with water removal using a Dean-Stark trap, these reactants are known to undergo
a [6 þ 3 þ 2] reaction to form a rigid cage based on a combination of aldehyde–amine (aldimine) and boronic acid–alcohol (bor-
oxine) condensations. Depending on the choice of the solvent, the reaction proceeded in poor yield, which was in the range
between 56% when using ethanol as the solvent and 24% when the solvent was toluene. In contrast, milling of the three reaction
components led to facile, almost quantitative formation of the targeted molecular cage. The cage product (Fig. 31C) was obtained in
94% yield, after a solvent-based extraction and purification procedure, demonstrating remarkable efficiency compared with solvent-
based procedures. Moreover, application of the original solution-based approaches to a longer boronic acid linker based on
a biphenyl bridging unit led to poor yields (below 40%) of the targeted larger 11-component cage. In this case also, mechanochem-
ical synthesis by grinding led to significant improvement, giving the target product in a 71% isolated yield. Although the possible
role of autogenous templating by reactants has not been addressed in the syntheses of covalent cages, it is likely that the reduced
yield of the larger variant could result from reactants becoming trapped within its covalent skeleton.85

2.06.10.2 Mechanically Interlocked Supramolecular Structures


The described mechanosyntheses of coordination-based, hydrogen-bonded, and covalently bonded cages clearly demonstrate that,
despite the high concentration of reagents under mechanochemical milling conditions, intermolecular condensations can be driven
to selectively produce discrete, well-defined self-assembled structures of nanometer size. Such behavior can be rationalized by the
formation of supramolecular host–guest complexes capable of templating the assembly of more complex structures. This suggests
the possibility of creating mechanically bonded interlocked structures in a solvent-free environment, which was indeed demon-
strated by Koshkakaryan et al.86 who used mechanochemistry to obtain a cocrystal of an electron-rich bis(naphthalene) macrocyclic
ether with an electron-poor naphthalene diimide. Manual grinding rapidly led to a change in color, consistent with the formation of
a charge-transfer complex, and PXRD analysis that the resulting solid was identical to a threaded 1:1 pseudorotaxane complex of the
naphthalene diimide with the macrocycle, previously obtained from solution.
A mechanochemical approach to a rotaxane structure was provided by the Orita group, who described a solid-state route both to
synthesize alkylpyridinium molecular threads and to use them in the solid-state synthesis of rotaxanes.87 In particular, they reported
the synthesis of a tricationic molecular thread by nucleophilic substitution involving benzylic bromide and pyridine reactants,
achieved by grinding of a solid film that was obtained upon vacuum evaporation of a solution of the two reactants. Grinding of
such a solid film gave the molecular tricationic thread in 85% yield after 50 minutes (Fig. 32A). In contrast, simply grinding of
the solid reactants produces the thread in 60% yield after 90 minutes, whereas reaction in concentrated solution proceeded over
two days to reach a maximum yield of 85%. The described film grinding procedure was generally applicable to a variety of pyridine
and benzylic bromide reactants, with excellent yields also achievable without any mechanical treatment, simply by aging of the
dried film. The formation of the film in which the reactants are intimately mixed appears critical for the reactivity, as no reaction
improvement was observable in cases when the rapid evaporation of the reactant solution did not yield a film. The film grinding
technique was extended to a two-step synthesis of a rotaxane target by rapidly evaporating a solution of the pyridine, benzylic
bromide, and an electron-rich macrocyclic ether, followed by solvent-free aging of the resulting film in air (Fig. 32B). In this
way, a rotaxane was obtained in 85% yield, along with 8% of the noncomplexed molecular thread. Attempts to conduct an anal-
ogous reaction by simply grinding of the three solid reactants gave the desired rotaxane in a much lower yield of 11% yield. In solu-
tion, conversion to the rotaxane was c. 78% yield after three days, demonstrating that mechanochemical grinding not only does
permit but also can augment the assembly of threaded complexes.
Chiu’s group has reported a general approach to rotaxanes by a mixed approach that combines procedures of solution- and
solid-state chemistry.88 This achieves the formation of a starting threaded pseudorotaxane in solution, which is then subse-
quently converted into a rotaxane by a mechanochemical introduction of bulky end group onto the molecular thread
(Fig. 33A). The first report of this methodology described the synthesis of a [2]rotaxane by first threading of a simple molecular
thread, bis(4-formylphenylmethylene)ammonium cation, with an electron-rich macrocyclic ether. After solvent removal, the
desired pseudorotaxane was obtained as solid material, which was then manually ground with 1,8-diaminonaphthalene.
Grinding led to the capping of the thread with bulky naphthalene units, which enabled the formation and isolation of the desired
[2]rotaxane in 80% yield (Fig. 33B). In comparison, the same [2]rotaxane was obtained in 48% yield in solution, after a total of
24 hours. Similarly, one-pot grinding of the cationic thread, the macrocyclic ether, and the capping agent provided the [2]rotax-
ane in 49% yield along with a significant amount (44%) of the capped thread without the macrocyclic host.
That this by-product was indeed a result of a side reaction rather than of slippage of the macrocyclic ether from the pseudoro-
taxane complex was verified by monitoring the stability of the [2]rotaxane in solution.88 The described two-step sequence was
readily expanded to the mechanochemical synthesis of [4]rotaxanes based on a tris(ammonium)-branched molecular thread. In
this case, one-hour grinding led to [4]rotaxane target yields that were twice larger than those obtained after 24-hour synthesis in
solution (Fig. 33C). The two-step synthetic procedure was also adaptable to other capping strategies. Notably, mechanochemical
capping via a Diels–Alder reaction driven by dinitrogen extrusion (Fig. 33d) was applied for the synthesis of a yet smallest [2]rotax-
ane structure by Hsu et al.89 The methodology was subsequently expanded to the synthesis of both symmetrical and nonsymmet-
rical [2]rotaxanes.
Supramolecular Mechanochemistry 129

Figure 32 (A) The comparison of direct grinding and film grinding approaches to the molecular tricationic thread described by Orita et al. and
(B) the related synthesis of a [2]rotaxane.87 Adapted with permission from Friscic, T. Chem. Soc. Rev. 2012, 41, 3493–3510.

Kihara and coworkers have utilized mechanochemical grinding to synthesize polyrotaxanes based on polytetrahydrofuran
threads and permethylated cyclodextrin macrocycles.90 The permethylated nature of the cyclodextrin allowed the capping of
the polypseudorotaxane by acylating, in the presence of a dibutyltin dilaurate catalyst, the terminal alcohol groups of the
thread. The corresponding polypseudorotaxane complex was too labile for successful capping in solution. However, mecha-
nochemical grinding of the molecular thread, the cyclodextrin, and the acylating reagent provided the target polyrotaxane in
the maximum 24% yield after 30 minutes. By applying an acetyl-protected monohydroxycyclodextrin as the macrocyclic host,
a polyrotaxane was mechanochemically obtained in 43% yield. Subsequent hydrolysis of the acetyl-protecting groups yielded
a polyrotaxane decorated with hydroxyl groups, which gave rise to a topological gel upon cross-linking with diphenylmethane
diisocyanate.

2.06.10.3 Mechanochemical Self-Sorting


Mechanosynthesis of rotaxanes has very recently been extended to a five-component system that demonstrates self-sorting in
a solvent-free environment.91 In particular, the assembly of a dicationic molecular thread with two different types of macro-
cyclic ethers and two different molecular capping agent was conducted in two solvent-free steps to provide a nonsymmetrical
130 Supramolecular Mechanochemistry

Figure 33 (A) Scheme of the two-step solution- and grinding-based methodology for rotaxane synthesis developed by Chiu’s group,88 (B) the
synthesis of a [2]rotaxane using Chiu’s procedure,88 (C) the [4]rotaxane obtained using the same procedure,88 and (D) the “smallest rotaxane”
prepared by Hsu et al. in a two-step sequence involving a mechanochemical Diels–Alder as the end-capping process.89 Figure has been adapted with
permission from Friscic, T. Chem. Soc. Rev. 2012, 41, 3493–3510.

[3]rotaxane. This proof-of-principle work employed a molecular dicationic thread bearing terminal phenylacetylene and
propargyl groups, which was designed after careful exploration of supramolecular and Diels–Alder reactivity of simpler
systems. Heating of the molecular thread with a 24-atom and a 21-atom macrocyclic ether led to the self-assembly of
a [3]pseudorotaxane with the larger and smaller macrocycles selectively situated on the sterically more (phenylacetylene
group) and less (propargyl group) encumbered segments of the diammonium cation. Room-temperature grinding of this
[3]pseudorotaxane with tetrazine and 1,4-diphenyltetrazine readily produced a single-capped [3]pseudorotaxane, by a selec-
tive Diels–Alder coupling and dinitrogen extrusion involving the tetrazine and the terminal propargylic group of the molec-
ular thread. The final nonsymmetrical [3]rotaxane product was obtained in the final thermochemical step, by heating the
ground reactant mixture at 60 C, leading to a Diels–Alder reaction between 1,4-diphenyltetrazine and the phenylacetylene
moiety of the [3]pseudorotaxane.91 The aforementioned illustrated mechanochemical approaches to [2] and [3]rotaxanes
point to a general means of using mechanochemistry to assist rotaxane synthesis, in which the threading process that
happens in a melt or a solution is separated from the solvent-free capping, which is conducted mechanochemically and/or
thermally.

2.06.11 Cocrystal–Cocrystal Reactions

The mechanochemical environment provides a very unique opportunity, not readily accessible using solution-based techniques, to
conduct chemical and supramolecular transformations by using as reactants preassembled solid-state structures, such as cocrystals.
Such cocrystal–cocrystal reactivity was first explored in the form of mechanochemical reactions between enantiomeric cocrystals of
the model APIs caffeine or theophylline with either L- or D-forms of tartaric acid.92
Two outcomes of such reactivity have been observed: the formation of a centrosymmetrical cocrystal based on racemic tartaric
acid as the cocrystal former and the dismantling of both cocrystals with the formation of pure racemic tartaric acid in the crys-
talline state (Fig. 34A). In particular, grinding equimolar amounts of the (theophylline)2$(D-tartaric acid) and (the-
ophylline)2$(L-tartaric acid) cocrystals produced the centrosymmetrical cocrystal (theophylline)2$(DL-tartaric acid) (Fig. 34B),
which can formally be considered as a three-component cocrystal (theophylline)2$(D-tartaric acid)0.5$(L-tartaric acid)0.5. In
contrast, grinding of the equimolar mixture of (caffeine)$(D-tartaric acid) and (caffeine)$(L-tartaric acid) cocrystals led to the
formation of separated crystalline caffeine and DL-tartaric acid.93 A follow-up study addressed the cocrystal–cocrystal reactivity
of cocrystals involving theophylline and malic acid. In this case, the reactions led to the formation of the racemic cocrystal by
milling together the equimolar amounts of cocrystals of theophylline with L- and D-malic acids. However, the reaction was diffi-
cult to detect through the use of powder X-ray diffraction, as the enantiomerically pure and racemic forms of the cocrystals were
isostructural.93 A different pattern of reactivity, involving both the formation of a centrosymmetrical cocrystal and partial
dismantling of the cocrystal, was observed by Braga’s group, who explored kneading reactions between pyrazine cocrystals
with D- and L-tartaric acids.94 Kneading of equimolar amounts of chiral cocrystals (pyrazine)$(L-tartaric acid) and
(pyrazine)$(D-tartaric acid) led to the formation of centrosymmetrical (pyrazine)$(DL-tartaric acid)2 cocrystal along with libera-
tion of pyrazine (Fig. 34C). The reaction was affected by the rapid evaporation of the generated pyrazine, and if the grinding was
conducted in the presence of a large excess of pyrazine, no cocrystal degradation was observed, and the reaction led to a cocrystal
with composition (pyrazine)$(DL-tartaric acid).
Supramolecular Mechanochemistry 131

Figure 34 (A) Two types of mechanochemical reactions observed between enantiomerically related cocrystals; (B) the reaction between cocrystals
of theophylline with D- and L-forms of tartaric acid, leading to a racemic cocrystal92,93; and (C) the combination of mechanochemical cocrystal
dismantling and merging observed by Braga et al. for reactions of enantiomeric cocrystals involving pyrazine and tartaric acid.94 Figure has been
modified and adapted with permission from Friscic, T. Chem. Soc. Rev. 2012, 41, 3493–3510.

2.06.12 Supramolecular Catalysis

There are numerous examples of organometallic catalysis using mechanochemistry, and this area has recently been reviewed.95,96
Whereas most examples of organometallic catalysis by grinding are limited to palladium-catalyzed reactions such as Heck or Suzuki
couplings, Fulmer and coworkers have recently investigated the application of mechanochemistry to Sonogashira couplings
involving more complex combined copper(I) and palladium(0) catalysts.97 This work revealed an exciting opportunity, enabled
by the attrition forces associated with mechanochemical milling, to use the grinding equipment as an in situ source of the copper
cocatalyst. In particular, conducting the mechanochemical coupling reaction of acetylenes and aryl halides using grinding equip-
ment made from copper provided phenylacetylenes in yields comparable with those obtained in steel jars using a copper(I) cocat-
alyst. This approach was subsequently expanded to silver98 and nickel transformations,99 using as catalysts a sheet of silver foil or
nickel beads, respectively. The Suzuki coupling of phenylboronic acid with para-substituted bromobenzenes was used by Schneider
et al. as a model reaction to evaluate the solvent and energy efficiency of solution-based and mechanochemical reactions.100 The
latter work clearly demonstrated the superior efficiency of mechanochemical methods in conducting this important catalytic trans-
formation. Organocatalytic mechanochemical reactions,101 which rely on catalysis achieved by noncovalent interactions, have been
investigated extensively by the Bolm group. In particular, enantiomerically pure hydrogen bonding molecules, such as amino acids,
have been demonstrated as valuable catalysts in the solvent-free enantioselective synthesis using aldol condensations and glyoxal
rearranegements.101,102 The latter reactions are particularly illustrative of how concepts of supramolecular chemistry can be utilized
in mechanochemical organic synthesis, by combining the noncovalent structure-directing effects of basic chiral cinchona alkaloids
with reversible formation of C–S bonds using catalytic thiols to provide accelerated and enantioselective reactivity.
The MacGillivray group demonstrated how mechanochemistry can be combined with template-directed photochemical [2 þ 2]
photodimerization into an entirely solvent-free organocatalytic process for regiospecific photodimerization. In particular, Atkinson
et al. have utilized104 neat grinding to assemble pyridine-substituted olefin reactants and ditopic molecular templates into cocrystals
composed of four-membered hydrogen-bonded assemblies in which pairs of olefin reactant molecules are aligned for topochemical
[2 þ 2] photodimerization (Fig. 35A). Exposing the mechanochemically prepared cocrystals to ultraviolet (UV) radiation led to
anticipated regioselective photochemical synthesis of rctt-1,2,3,4-tetrasubstituted cyclobutanes. Subsequently, it was also shown
that such mechanochemical cocrystallization can be utilized to achieve catalytic photochemical [2 þ 2] photodimerization in the
solid state.103 Specifically, grinding together the reactant pyridine-substituted olefin with a substoichiometric (10–50 mol%)
amount of a resorcinol as the hydrogen bonding template, followed by irradiation of the resulting solid mixture to UV light resulted
in the formation of photoactive assemblies and template-controlled photocyclodimerization. Upon repeated grinding, however, the
resorcinol template reassembled with unreacted olefin, allowing the entire process to the repeated. In that way, through repeated
consecutive mechanochemical and photochemical steps, it was possible to regiospecifically obtain 99% yield of the rctt-cyclobutane
product, using 10 mol% of the catalyst over 80 hours. The catalytic behavior was explained through theoretical calculations, which
indicated that the supramolecular hydrogen-bonded assemblies of the resorcinol template with the reactant olefins are 4.8 kJ mol 1
more stable than the analogous assemblies of the cyclobutane product obtained after UV irradiation. The calculated energy differ-
ence provided an explanation of the driving force for the mechanochemical regeneration of the photoactive assemblies upon
grinding of the irradiated mixture. The development of such a designed approach to catalytic solid-state [2 þ 2] photodimerization
has, in a way, been predicted by Toda’s group6 who observed the stereospecific head-to-tail [2 þ 2] photodimerization of a chalcone
132 Supramolecular Mechanochemistry

Figure 35 (A) The catalytic mechano- and photochemical dimerization process developed by Sokolov et al.103 and (B) the catalytic mechanochem-
ical photodimerization of a chalcone reported by Toda.6 Figure adapted with permission from Friscic, T. Chem. Soc. Rev. 2012, 41, 3493–3510.

derivative (Fig. 35B) by mixing and irradiating in the presence of a wheel-and-axle diol host. The catalytic behavior was explained
through the transient formation of the solid-state photoactive inclusion compound between the chalcone and the host diol.

2.06.13 Mechanochemical Synthesis of MOFs


2.06.13.1 Synthesis of MOFs by Neat Grinding
The mechanochemical assembly of coordination bonds is a large, rapidly growing area that has so far been reviewed several
times.105,106 For this reason, this section focuses on mechanochemical strategies for the assembly of microporous MOFs. The first
reported entry of mechanochemistry into the synthesis of microporous MOFs was in 2006 by the James group, who conducted the
neat milling synthesis of the open copper(II) isonicotinate framework.107 The MOF was prepared quantitatively by 10 minute
milling of copper(II) acetate monohydrate (Cu(OAc)2$H2O) with isonicotinic acid. The product was obtained as a solvate, with
water and/or acetic acid by-products included in the framework pores. The open microporous framework was obtained by evacu-
ation of the crude product at 200 C. This work was subsequently expanded to the archetypal HKUST-1 framework, this time by
milling of Cu(OAc)2$H2O with 1,3,5-benzenetricarboxylic acid (trimesic acid) (Fig. 36A).108 The resulting mechanochemically
prepared MOFs have been shown to have identical or even superior properties compared with analogous materials prepared
from solution. Simple, neat grinding is a very successful approach for the synthesis of a wide range of MOFs, based on a diverse
selection of metals and linkers. For example, the Junk and Morsali groups have described the synthesis of a porous cadmium-
based MOF from cadmium acetate, 4,4ʹ-oxybis(benzoic acid) and 4,4ʹ-dipyridyl (bipy) (Fig. 36B),109 while Zhang et al. demon-
strated the synthesis of soluble coordination polymers starting from phenol ligands and zinc acetate.111 A so far unique example
of using a metal hydride as a precursor in the mechanosynthesis of a MOF is the formation of an yttrium-based framework by neat
milling of YH3 with trimesic acid (Fig. 36C).110
However, whereas solution-based (including solvothermal) synthesis of MOFs is highly dependent on using soluble metal salts,
such as chlorides, nitrates, or acetates, mechanochemistry provides a novel and also industrially attractive112 opportunity to utilize
not only readily accessible and safer but also significantly less soluble sulfate, oxide, hydroxide, or carbonate reagents. Several
groups have been pursuing the development of mechanochemical techniques for the efficient synthesis of MOFs from metal oxides
and metal carbonates.113,114 Whereas most of this work was achieved through the application of LAG and ILAG techniques, there
Supramolecular Mechanochemistry 133

Figure 36 Mechanochemical synthesis of various MOFs by neat grinding or by liquid-assisted grinding: (A) synthesis of HKUST-1 from copper
salts and trimesic acid,108 (B) synthesis of the pillared MOF material TMU-9 from cadmium acetate,109 and (C) solid-state synthesis of yttrium-based
framework material MIL-78.110

are also examples of 3-D open frameworks that were readily obtained by neat grinding. One of the earliest reported examples is the
synthesis of open MOFs from lanthanide(III) carbonates and trimesic acid.115 The mechanochemical reactions readily yielded 3-D
frameworks, and by using mixtures of carbonates of different lanthanide metals, it was also possible to obtain MOFs with different
metal nodes.
The neat grinding synthesis of MOFs is also adaptable for large-scale continuous manufacture of commercially relevant MOFs, as
demonstrated by the James group who described the synthesis of aluminum-based MOF Al(fumarate)(OH) from fumaric acid and
hydrated aluminum sulfate by twin screw extrusion.116 Neat milling is not an efficient approach for the synthesis of ZIFs from
a metal oxide, as reported by Beldon et al. who reported incomplete conversion of ZnO into the commercially relevant framework
ZIF-8 by ball milling with 2-methylimidazole.60 The result was subsequently corroborated by real-time reaction monitoring using
in situ synchrotron powder X-ray diffraction.65 Although the reaction did not reach completion within 90 min, in situ collected X-
ray diffractograms clearly indicated the formation of the SOD topology ZIF-8. More information on the mechanism of this and
potentially other neat grinding transformations was provided by the Tanaka group, who used high-resolution scanning electron
microscopy to establish the formation of core–shell nanostructures consisting of ZnO encapsulated by the product ZIF-8.117
Whereas mechanochemical treatment of MOFs in the presence of a liquid phase (i.e., under LAG or ILAG conditions) generally
leads to highly crystalline materials even upon extended milling periods, extended neat grinding was found to induce the loss of
crystallinity and formation of amorphous MOFs.118 The formation of such metal–organic glasses by milling or through melt
quenching has been studied by the Bennett and Cheetham groups,119 focusing largely but not exclusively120 on the mechanochem-
istry of ZIFs. Currently, such amorphous materials are still a rare and poorly understood family of materials,121 whose structural
characterization requires the use of highly advanced techniques, such as X-ray pairwise distribution function analysis122 and
solid-state NMR,123 which are often coupled with theoretical modeling approaches.

2.06.13.2 Synthesis of MOFs Using Liquid-Assisted Mechanochemistry


The addition of a small amount of liquid has been shown to dramatically enhance or accelerate the synthesis of coordination poly-
mers from oxides or carbonates. An early demonstration of such reactivity was provided by Adams and coworkers, who demon-
strated the assembly of coordination polymers by LAG or neat milling reaction of transition metal carbonates with protonated
hydrochloride salts of organic linkers, such as bipy.124 Variation of the liquid additive in LAG offers a means to direct the formation
of metal–organic structures, as demonstrated by Yuan et al., who explored the mechanochemical reactions of basic zinc carbonate
and terephthalic acid in the presence of different liquid additives.115 Milling with water led to the formation of a 1-D coordination
polymer with composition zinc terephthalate dihydrate, whereas milling in the presence of methanol produced the nonporous
monohydrate 3-D framework. In contrast to these nonporous materials, milling with N,N-dimethylformamide (DMF) led to the
formation of an open structure based on 2-D sheets with an open square grid (sql) topology. The sheets consisted of zinc carbox-
ylate paddlewheel secondary building units (SBUs) interconnected by terephthalate linkers. These mechanochemically obtained
134 Supramolecular Mechanochemistry

products were also found to be highly dynamic under mechanochemical conditions, readily interconverting upon milling in the
presence of an appropriate liquid (i.e., water or methanol or DMF). In contrast, attempts to achieve similar structural framework
interconversions by immersing the materials in the appropriate bulk solvent and stirring were largely unsuccessful, confirming
that MOFs can exhibit indicating structural flexibility under milling conditions. In the first example of mechanochemical synthesis
of an open coordination framework from a metal oxide precursor, Fábián and coworkers synthesized pillared MOFs by ball milling
of ZnO, H2fum, and either bipy or bpe as the pillaring ligand. While neat grinding did not result in any significant conversion, LAG
with DMF resulted in quantitative formation of the targeted 3-D pillared MOFs.61 Importantly, it was established that the reaction
progress was limited by the stoichiometric amount of liquid present. The minimum amount of liquid required for the complete
conversion of the reaction mixture was found to coincide to that calculated from the published structures of the corresponding
MOF with solvent included in the pores. If the amount of liquid present was not sufficient to completely fill the voids of the nascent
MOF, the reaction would not lead to completion. These observations highlight a dual role for the liquid additive in mechanochem-
ical synthesis of open framework structures: the liquid acts both as a means to accelerate the reaction and a space-filling or even
a structure-templating agent. Thus, mechanochemically prepared materials were readily desolvated by heating without structural
collapse, confirming their functional nature as microporous MOFs. Similarly, the structure of the guest-free pillared MOF based
on fumarate and bpy ligands was determined by Fujii and coworkers from powder X-ray diffraction data that were collected on
thermally desolvated material.125
The use of liquid-assisted mechanochemistry was also central in developing of a mechanochemical approach for the synthesis of
the archetypal material MOF-5 (IRMOF-1). The mechanosynthesis of MOF-5 was achieved by the milling reaction of terephthalic
acid with preassembled oxo-zinc carboxylate or amidate clusters containing the required Zn4O unit (Fig. 37).126 Whereas the neat
reaction using the readily accessible benzoate-based cluster proceeded poorly in the absence of a liquid additive, the addition of
a small amount of N,N-diethylformamide resulted in the rapid, quantitative transformation of reactants into the target MOF-5.
Upon evacuation, the product exhibited a very high surface area of 1831 m2 g 1. A similar strategy was applied by Uzarevic
et al.127 for the synthesis of zirconium-based UiO-66 and UiO-66-NH2 frameworks, which are based on 12-coordinated hexanu-
clear zirconium oxo-clusters as SBUs. The zirconium MOFs were readily assembled by liquid-assisted milling reaction of tereph-
thalic acid or 2-aminoterephthalic acid linker precursors with preassembled hexanuclear clusters coordinated with methacrylate
or benzoate ligands. Importantly, the UiO-66 framework could also be assembled in a multicomponent one-pot process, starting
from zirconium(IV) propoxide, terephthalic acid, and water. In all cases, the mechanochemical reactions provided rapid and easy
access to these MOFs, whose efficient synthesis in solution has so far been the target of much arduous investigation.128 Importantly,
the LAG synthesis of UiO-type frameworks could be readily conducted using methanol as the liquid additive, in that way again
enabling the complete removal of DMF from the synthetic procedure. The resulting materials exhibited both very high micropo-
rosity and catalytic activity that was on par with analogous materials prepared solvothermally.
An interesting example of a mechanochemical MOF synthesis was provided by Matoga and coworkers group,129 who demon-
strated the mechanochemical transformation of a neutral manganese isonicotinate open framework into a charged, proton-
conducting MOF by neat grinding or by LAG in the presence of ammonium thiocyanate. The mechanochemical reaction involves

Figure 37 Solid-state mechanochemical synthesis of the archetypal MOF-5 framework from preformed oxo-zinc precursors. Adapted with permis-
sion from Prochowicz, D.; Soko1owski, K.; Justyniak, I.; Kornowicz, A.; Fairen-Jimenez, D.; Friscic, T.; Lewinski, J. Chem. Commun. 2015, 51,
4032–4035.
Supramolecular Mechanochemistry 135

the attachment of isothiocyanate ligands onto and rearrangement of manganese-based framework nodes, resulting in a material that
is based on anionic 2-D open framework sheet material and ammonium (NHþ 4 ) cations, suitable for proton conduction.
An early approach to the mechanochemical synthesis of ZIFs was described by Adams and coworkers, who utilized manual
grinding of imidazolium halometallate salts with an external base, such as KOH.130
Whereas carbonates and hydroxides can be readily employed as precursors for the synthesis of MOFs by neat or LAG mechano-
chemistry, the use of metal oxide precursors is often more challenging. This is illustrated in the attempted syntheses of ZIFs by
milling or grinding of ZnO with different imidazole linker precursors, all leading to only partial product formation. However, mech-
anochemical reactivity of metal oxides was found to be enhanced by the addition of catalytic salt additives into the LAG reaction
mixture. This ILAG methodology131 provides a powerful technique to achieve rapid, room-temperature transformations of metal
oxides into diverse metal–organic materials, including microporous MOFs and metallodrugs (e.g., bismuth subsalicylate,132 the
active component of Pepto-Bismol). The first report of the ILAG113 technique addressed the formation of the popular pillared
MOF based on open square grid sheets of zinc terephthalate separated by dabco as the pillaring ligand. Regardless of the choice
of milling liquid, the targeted pillared MOF could not be synthesized by LAG of ZnO, terephthalic acid, and dabco. Instead, milling
consistently yielded mixtures of the dabco salt of terephthalic acid and unreacted zinc oxide. However, milling with a liquid additive
in the presence of a catalytic amount (z 5 mol% with respect to zinc) of nitrate ions, introduced in the form of NH4NO3 or an
alkaline metal nitrate salt, rapidly (i.e., within 30 minutes) produced the target MOF. Moreover, by using ammonium of alkaline
metal sulfate salts as catalytic salt additives, the reaction was readily steered toward the formation of an open isocompositional (i.e.,
polymorphic) framework based on hexagonal, Kagome-topology zinc terephthalate sheets pillared by dabco. Besides leading to
a topologically different product, sulfate salt additives were also significantly more active than nitrates as ILAG additives, inducing
MOF synthesis at amounts as low as 300 ppm with respect to zinc. Subsequently, the use of ILAG also enabled quantitative, room-
temperature mechanochemical formation of close-packed and microporous ZIFs using only stoichiometric amounts of reagents.60
In this case also, the selection of different catalytic salt additives led to the formation of topologically distinct frameworks.
Whereas the ability of ILAG to enable conversion of typically inert metal oxides into MOFs is of considerable value in the context
of clean, rapid manufacture of these novel microporous materials, it also represents a very exciting development in the context of
solid-state supramolecular chemistry, as it allows the transformations of normally inert and poorly soluble metal oxides under mild
conditions, more akin to those of molecular self-assembly than to high-temperature environments conventionally used to activate
metal oxides. Consequently, our view is that ILAG and, indeed, other recently developed mechanochemical techniques (e.g.,
seeding-assisted grinding51 or polymer-assisted grinding133) open completely new frontiers in the scope of solid-state supramolec-
ular chemistry and self-assembly.

References

1. James, S. L.; Adams, C. J.; Bolm, C.; Braga, D.; Collier, P.; Friscic, T.; Grepioni, F.; Harris, K. D. M.; Hyett, G.; Jones, W.; Krebs, A.; Mack, J.; Maini, L.; Orpen, A. G.;
Parkin, I. P.; Shearouse, W. C.; Steed, J. W.; Waddell, D. C. Chem. Soc. Rev. 2012, 41, 413–447.
2. Friscic, T. Chem. Soc. Rev. 2012, 41, 3493–3510.
3. Takacs, L. Chem. Soc. Rev. 2013, 42, 7649–7659.
4. Takacs, L. J. Therm. Anal. Calorim. 2007, 90, 81–84.
5. Tanaka, K.; Toda, F. Chem. Rev. 2000, 100, 1025–1074.
6. Toda, F.; Tanaka, K.; Sekikawa, A. Chem. Commun. 1987, 279–280.
7. Etter, M. C.; Reutzel, S. M.; Choo, C. G. J. Am. Chem. Soc. 1993, 115, 4411–4412.
8. Belcher, W. J.; Longstaff, C. A.; Neckening, M. R.; Steed, J. W. Chem. Commun. 2002, 1602–1603.
9. Braga, D.; Giaffreda, S. L.; Grepioni, F.; Pettersen, A.; Maini, L.; Curzi, M.; Polito, M. Dalton Trans. 2006, 1249–1263.
10. Braga, D.; D’Agostino, S.; Grepioni, F.; Gandolfi, M.; Rubini, K. Dalton Trans. 2011, 40, 4765–4777.
11. Rightmire, N. R.; Hanusa, T. P. Dalton Trans. 2016, 45, 2352–2362.
12. Delori, A.; Friscic, T.; Jones, W. CrystEngComm 2012, 14, 2350–2362.
13. Tan, D.; Loots, L.; Friscic, T. Chem. Commun. 2016, 52, 7760–7781.
14. Potisek, S. L.; Davis, D. A.; Sottos, N. R.; White, S. R.; Moore, J. S. J. Am. Chem. Soc. 2007, 129, 13808–13809.
15. Konda, S. S. M.; Brantley, J. N.; Varghese, B. T.; Wiggins, K. M.; Bielawski, C. W.; Makarov, D. E. J. Am. Chem. Soc. 2013, 135, 12722–12729.
16. Baláz, P. Mechanochemistry in Nanoscience and Minerals Engineering; Springer-Verlag: Berlin, 2008.
17. Baláz, P.; Achimovicová, M.; Baláz, M.; Billik, P.; Cherkezova-Zheleva, Z.; Criado, J. M.; Delogu, F.; Dutková, E.; Gaffet, E.; Gotor, F. J.; Kumar, R.; Mitov, I.; Rojac, T.;
Senna, M.; Streletskii, A.; Wieczorek-Ciurowa, K. Chem. Soc. Rev. 2013, 42, 7571–7637.
18. Ibrahim, A. Y.; Forbes, R. T.; Blagden, N. CrystEngComm 2011, 13, 1141–1152.

19. Sepelák, V.; Becker, S. M.; Bergmann, I.; Indris, S.; Scheuermann, M.; Feldhoff, A.; Kübel, C.; Bruns, M.; Stürzl, N.; Ulrich, A. S.; Ghafari, M.; Hahn, H.; Grey, C. P.;
Becker, K. D.; Heitjans, P. J. Mater. Chem. 2012, 22, 3117–3126.
20. Jayasankar, A.; Somwangthanaroj, A.; Shao, Z. J.; Rodríguez-Hornedo, N. Pharm. Res. 2006, 23, 2381–2392.
21. Almarsson, Ö.; Zaworotko, M. J. Chem. Commun. 2004, 1889–1896.
22. Halasz, I.; Puskaric, A.; Kimber, S. A. J.; Beldon, P. J.; Belenguer, A. M.; Adams, F.; Honkimäki, V.; Dinnebier, R. E.; Patel, B.; Jones, W.; Strukil, V.; Friscic, T. Angew. Chem.
Int. Ed. 2013, 52, 11538–11541.
23. Kaupp, G. CrystEngComm 2003, 5, 117–133.
24. Friscic, T.; Jones, W. Cryst. Growth Des. 2009, 9, 1621–1637.
25. Paul, I. C.; Curtin, D. Y. Acc. Chem. Res. 1973, 6, 217–225.
26. Kuroda, R.; Higashiguchi, K.; Hasebe, S.; Imai, Y. CrystEngComm 2004, 6, 463–468.
27. Bond, D. R.; Johnson, L.; Nassimbeni, L. R.; Toda, F. J. Solid State Chem. 1991, 92, 68–79.
28. Jacobs, A.; Nassimbeni, L. R.; Nohako, K. L.; Ramon, G.; Taljaard, J. H. New J. Chem. 2009, 33, 1960–1964.
136 Supramolecular Mechanochemistry

29. Rastogi, R. P.; Bassi, P. S.; Chadha, S. L. J. Phys. Chem. 1963, 67, 2569–2573.
30. Rastogi, R. P.; Singh, N. B. J. Phys. Chem. 1966, 70, 3315–3324.
31. Singh, N. B.; Singh, N. N.; Laidlaw, R. K. J. Solid State Chem. 1987, 71, 530–539.
32. Urakaev, F. K.; Boldyrev, V. V. Powder Technol. 2000, 107, 93–107.
33. McKissic, K. S.; Caruso, J. T.; Blair, R. G.; Mack, J. Green Chem. 2014, 16, 1628–1632.
34. Chadwick, K.; Davey, R.; Cross, W. CrystEngComm 2007, 9, 732–734.
35. Rothenberg, G.; Downie, A. P.; Raston, C. L.; Scott, J. L. J. Am. Chem. Soc. 2001, 123, 8701–8708.
36. Pascu, M.; Ruggi, A.; Scopelliti, R.; Severin, K. Chem. Commun. 2013, 49, 45–47.
37. Biswal, B. P.; Chandra, S.; Kandambeth, S.; Lukose, B.; Heine, T.; Banerjee, R. J. Am. Chem. Soc. 2013, 135, 5328–5331.
38. Bala, M. D.; Coville, N. J. J. Organomet. Chem. 2007, 692, 709–730.
39. Manzini, S. S.; Coville, N. J. Inorg. Chem. Commun. 2004, 7, 676–678.
40. Strobridge, F. C.; Judas, N.; Friscic, T. CrystEngComm 2010, 12, 2409–2418.
41. Heinz, A.; Strachan, C. J.; Gordon, K. C.; Rades, T. J. Pharm. Pharmacol. 2009, 61, 971–988.
42. Nguyen, K. L.; Friscic, T.; Day, G. M.; Gladden, L. F.; Jones, W. Nat. Mater. 2007, 6, 206–209.
43. Batchelor, E.; Klinowski, J.; Jones, W. J. Mater. Chem. 2000, 10, 839–848.
44. Liu, X.; Liu, G.; Zhao, H.; Zhang, Z.; Wei, Y.; Liu, M.; Wen, W.; Zhou, X. J. Phys. Chem. Solids 2011, 72, 1245–1250.
45. Ito, H.; Saito, T.; Oshima, N.; Kitamura, N.; Ishizaka, S.; Hinatsu, Y.; Wakeshima, M.; Kato, M.; Tsuge, K.; Sawamura, M. J. Am. Chem. Soc. 2010, 130, 10044–10045.
46. Zhang, G.; Lu, J.; Sabat, M.; Fraser, C. L. J. Am. Chem. Soc. 2010, 132, 2160–2162.
47. Braga, D.; Giaffreda, S. L.; Grepioni, F.; Polito, M. CrystEngComm 2004, 6, 458–462.
48. Adams, C. J.; Colquhoun, H. M.; Crawford, P. C.; Lusi, M.; Orpen, A. G. Angew. Chem. Int. Ed. 2007, 46, 1124–1128.
49. Braga, D.; Curzi, M.; Grepioni, F.; Polito, M. Chem. Commun. 2005, 2915–2917.
50. Cincic, D.; Brekalo, I.; Kaitner, B. Chem. Commun. 2012, 48, 11683–11685.
51. Cincic, D.; Brekalo, I.; Kaitner, B. Cryst. Growth Des. 2011, 11, 44–48.
52. Schmeyers, J.; Toda, F.; Boy, J.; Kaupp, G. J. Chem. Soc. Perkin Trans. 1998, 2, 989–993.
53. Bowmaker, G. A.; Effendy; Hanna, J. V.; Healy, P. C.; King, S. P.; Pettinari, C.; Skelton, B. W. Dalton Trans. 2011, 40, 7210–7218.
54. Schrader, R.; Hoffmann, B. Z. Anorg. Allg. Chem. 1969, 369, 41–47.
55. Belenguer, A. M.; Friscic, T.; Day, G. M.; Sanders, J. K. M. Chem. Sci. 2011, 2, 696–700.
56. Drebuschak, T. N.; Ogienko, A. A.; Boldyreva, E. V. CrystEngComm 2011, 13, 4405–4410.
57. Cincic, D.; Friscic, T.; Jones, W. J. Am. Chem. Soc. 2008, 130, 7524–7525.
58. Karki, S.; Friscic, T.; Jones, W. CrystEngComm 2009, 11, 470–481.
59. Tumanov, I. A.; Achkasov, A. F.; Boldyreva, E. V.; Boldyrev, V. V. CrystEngComm 2011, 13, 2213–2216.
60. Vives, G.; Mason, S. A.; Prince, P. D.; Junk, P. C.; Steed, J. W. Cryst. Growth Des. 2003, 3, 699–704.
61. Friscic, T.; Fábián, L. CrystEngComm 2009, 11, 743–745.
62. Beldon, P. J.; Fábián, L.; Stein, R. S.; Thirumurugan, A.; Cheetham, A. K.; Friscic, T. Angew. Chem. Int. Ed. 2010, 49, 9640–9643.
63. Park, K. S.; Ni, Z.; Côté, A. P.; Choi, J. Y.; Huang, R.; Uribe-Romo, F. J.; Chae, H. K.; O’Keeffe, M.; Yaghi, O. M. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 10186–10191.
64. Strukil, V.; Fábián, L.; Reid, D. G.; Duer, M. J.; Jackson, G. J.; Eckert-Maksic, M.; Friscic, T. Chem. Commun. 2010, 46, 9191–9193.
65. Friscic, T.; Halasz, I.; Beldon, P. J.; Belenguer, A. M.; Adams, F.; Kimber, S. A. J. Nat. Chem. 2013, 5, 66–73.
66. Halasz, I.; Kimber, S. A. J.; Beldon, P. J.; Belenguer, A. M.; Adams, F.; Honkimäki, V.; Nightingale, R. C.; Dinnebier, R. E.; Friscic, T. Nat. Prot. 2013, 8, 1718–1729.
67. Katsenis, A. D.; Puskaric, A.; Strukil, V.; Mottillo, C.; Julien, P. A.; Uzarevic, K.; Pham, M.-H.; Do, T.-O.; Kimber, S. A. J.; Lazic, P.; Magdysyuk, O.; Dinnebier, R. E.; Halasz, I.;
Friscic, T. Nat. Commun. 2015, 6, 6662.
68. Friscic, T.; Jones, W. Ann. Chim. Sci. Mater. 2009, 34, 415–428.
69. Imai, Y.; Tajima, N.; Sato, T.; Kuroda, R. Org. Lett. 2006, 8, 2941.
70. Cheung, E. Y.; Kitchin, S. J.; Harris, K. D. M.; Imai, Y.; Tajima, N.; Kuroda, R. J. Am. Chem. Soc. 2003, 125, 14658–14659.
71. Friscic, T.; Trask, A. V.; Jones, W.; Motherwell, W. D. S. Angew. Chem. Int. Ed. 2006, 45, 7546–7550.
72. Friscic, T.; Trask, A. V.; Motherwell, W. D. S.; Jones, W. Cryst. Growth Des. 2008, 8, 1605–1609.
73. Eger, C.; Norton, D. A. Nature 1965, 208, 997.
74. Friscic, T.; Lancaster, R. W.; Fábian, L.; Karamertzanis, P. G. Proc. Natl. Acad. Sci. U. S. A. 2010, 107, 13216.
75. Ardila-Fierro, K. J.; André, V.; Tan, D.; Duarte, M. T.; Lancaster, R. W.; Karamertzanis, P. G.; Friscic, T. Cryst. Growth Des. 2015, 15, 1492–1501.
76. Cincic, D.; Friscic, T.; Jones, W. Chem. Eur. J. 2008, 14, 747–753.
77. Cincic, D.; Friscic, T.; Jones, W. Chem. Mater. 2008, 20, 6623–6626.
78. Cincic, D.; Friscic, T.; Jones, W. New J. Chem. 2008, 32, 1776–1781.
79. Lapadula, G.; Judas, N.; Friscic, T.; Jones, W. Chem. Eur. J. 2010, 16, 7400–7403.
80. Friscic, T.; Mestrovic, E.; Skalec-Samec, D.; Kaitner, B.; Fábián, L. Chem. Eur. J. 2009, 15, 12644–12652.
81. Ferguson, M.; Giri, N.; Huang, X.; Apperley, D.; James, S. L. Green Chem. 2014, 16, 1374–1382.
82. Cincic, D.; Friscic, T. CrystEngComm 2014, 16, 10169–10172.
83. Orita, A.; Jiang, L.; Nakano, T.; Ma, N.; Otera, J. Chem. Commun. 2002, 1362–1363.
84. Antesberger, J.; Cave, G. W. V.; Ferrarelli, M. C.; Heaven, M. W.; Raston, C. L.; Atwood, J. L. Chem. Commun. 2005, 892–894.
85. Içli, B.; Christinat, N.; Tönnemann, J.; Schüttler, C.; Scopelliti, R.; Severin, K. J. Am. Chem. Soc. 2009, 131, 3154–3155.
86. Koshkakaryan, G.; Klivansky, L. M.; Cao, D.; Snauko, M.; Teat, S. J.; Struppe, J. O.; Liu, Y. J. Am. Chem. Soc. 2009, 131, 2078–2079.
87. Orita, A.; Okano, J.; Tawa, Y.; Jiang, L.; Otera, J. Angew. Chem. Int. Ed. 2004, 43, 3724–3728.
88. Hsueh, S.-Y.; Cheng, K.-W.; Lai, C.-C.; Chiu, S.-H. Angew. Chem. Int. Ed. 2008, 47, 4436–4439.
89. Hsu, C.-C.; Chen, N.-C.; Lai, C.-C.; Liu, Y.-H.; Peng, S.-M.; Chiu, S.-H. Angew. Chem. Int. Ed. 2008, 47, 7475–7478.
90. Kihara, N.; Hinoue, K.; Takata, T. Macromolecules 2005, 38, 223–226.
91. Chen, P.-N.; Lai, C.-C.; Chiu, S.-H. Org. Lett. 2011, 13, 4660–4663.
92. Friscic, T.; Fábián, L.; Burley, J. C.; Jones, W.; Motherwell, W. D. S. Chem. Commun. 2006, 5009–5011.
93. Friscic, T.; Fábián, L.; Burley, J. C.; Reid, D. G.; Duer, M. J.; Jones, W. Chem. Commun. 2008, 1644–1646.
94. Braga, D.; Grepioni, F.; Lampronti, G. I. CrystEngComm 2011, 13, 3122–3124.
95. Hernández, J. G.; Friscic, T. Tetrahedron Lett. 2015, 56, 4253–4265.
96. Rodríguez, B.; Bruckmann, A.; Rantanen, T.; Bolm, C. Adv. Synth. Catal. 2007, 349, 2213–2233.
97. Fulmer, D. A.; Shearouse, W. C.; Medonza, S. T.; Mack, J. Green Chem. 2009, 11, 1821–1825.
98. Chen, L.; Bovee, M. O.; Lemma, B. E.; Keithley, K. S. M.; Pilson, S. L.; Coleman, M. G.; Mack, J. Angew. Chem. Int. Ed. 2015, 54, 11084–11087.
99. Haley, R. A.; Zellner, A. R.; Krause, J. A.; Guan, H.; Mack, J. ACS Sust. Chem. Eng. 2016, 4, 2464–2469.
100. Schneider, F.; Szuppa, T.; Stolle, A.; Ondruschka, B.; Hopf, H. Green Chem. 2009, 11, 1894–1899.
Supramolecular Mechanochemistry 137

101. Bruckmann, A.; Krebs, A.; Bolm, C. Green Chem. 2008, 10, 1131–1141.
102. Schmitt, E.; Schiffers, I.; Bolm, C. Tetrahedron Lett. 2009, 50, 3185–3188.
103. Sokolov, A. N.; Bucar, D.-K.; Baltrusaitis, J.; Gu, S. X.; MacGillivray, L. R. Angew. Chem. Int. Ed. 2010, 49, 4273.
104. Atkinson, M. B. J.; Bucar, D.-K.; Sokolov, A. N.; Friscic, T.; Robinson, C. N.; Bilal, M. Y.; Sinada, N. G.; Chevannes, A.; MacGillivray, L. R. Chem. Commun. 2008,
5713–5715.
105. Lazuen Garay, A.; Pichon, A.; James, S. L. Chem. Soc. Rev. 2007, 36, 846–855.
106. Friscic, T. In Ball Milling Towards Green Synthesis; Stolle, A., Ranu, B., Eds.; RSC Publishing: Cambridge, United Kingdom, 2014.
107. Pichon, A.; Lazuen-Garay, A.; James, S. L. CrystEngComm 2006, 8, 211–214.
108. Yuan, W.; Garay, A. L.; Pichon, A.; Clowes, R.; Wood, C. D.; Cooper, A. I.; James, S. L. CrystEngComm 2010, 12, 4063–4065.
109. Masoomi, M. Y.; Morsali, A.; Junk, P. C. CrystEngComm 2015, 17, 686–692.
110. Singh, N. K.; Hardi, M.; Balema, V. P. Chem. Commun. 2013, 49, 972–974.
111. Zhang, P.; Li, H.; Veith, G. M.; Dai, S. Adv. Mater. 2015, 27, 234.
112. Czaja, A.; Leung, E.; Trukhan, N.; Müller, U. In Industrial MOF Synthesis in Metal-Organic Frameworks: Applications from Catalysis to Gas Storage; Farrusseng, D., Ed., 1st
ed.; John Wiley & Sons: Weinheim, Germany, 2011.
113. Friscic, T.; Reid, D. G.; Halasz, I.; Stein, R. S.; Dinnebier, R. E.; Duer, M. J. Angew. Chem. Int. Ed. 2010, 49, 712–715.
114. Yuan, W.; Friscic, T.; Apperley, D.; James, S. L. Angew. Chem. 2010, 49, 3916–3919.
115. Yuan, W.; O’Connor, J.; James, S. L. CrystEngComm 2010, 12, 3515–3517.
116. Crawford, D.; Casaban, J.; Haydon, R.; Giri, N.; McNally, T.; James, S. L. Chem. Sci. 2015, 6, 1645–1649.
117. Tanaka, S.; Kida, K.; Nagaoka, T.; Ota, T.; Miyake, Y. Chem. Commun. 2013, 49, 7884–7886.
118. Bennett, T. D.; Cheetham, A. K. Acc. Chem. Res. 2014, 47, 1555.
119. Bennett, T. D.; Cao, S.; Tan, J. C.; Keen, D. A.; Bithell, E. G.; Beldon, P. J.; Friscic, T.; Cheetham, A. K. J. Am. Chem. Soc. 2011, 133, 14546–14549.
120. Bennett, T. D.; Todorova, T. K.; Baxter, E. F.; Reid, D. G.; Gervais, C.; Bueken, B.; Van de Voorde, B.; De Vos, D.; Keen, D. A.; Mellot-Draznieks, C. Phys. Chem. Chem. Phys.
2016, 18, 2192–2201.
121. Bennett, T. D.; Cheetham, A. K. Acc. Chem. Res. 2014, 47, 1555–1562.
122. Cao, S.; Bennett, T. D.; Keen, D. A.; Goodwin, A. L.; Cheetham, A. K. Chem. Commun. 2012, 48, 7805–7807.
123. Thornton, A. W.; Jelfs, K. E.; Konstas, K.; Doherty, C. M.; Hill, A. J.; Cheetham, A. K.; Bennett, T. D. Chem. Commun. 2016, 52, 3750–3753.
124. Adams, C. J.; Kurawa, M. A.; Lusi, M.; Orpen, A. G. CrystEngComm 2008, 10, 1790–1795.
125. Fujii, K.; Garay, A. L.; Hill, J.; Sbircea, E.; Pan, Z.; Xu, M.; Apperley, D. C.; James, S. L.; Harris, K. D. M. Chem. Commun. 2010, 46, 7572–7574.
126. Prochowicz, D.; Soko1owski, K.; Justyniak, I.; Kornowicz, A.; Fairen-Jimenez, D.; Friscic, T.; Lewinski, J. Chem. Commun. 2015, 51, 4032–4035.
127. Uzarevic, K.; Wang, T. C.; Moon, S.-Y.; Fidelli, A. M.; Hupp, J. T.; Farha, O. K.; Friscic, T. Chem. Commun. 2016, 52, 2133–2136.
128. Katz, M. J.; Brown, Z. J.; Colón, Y. J.; Siu, P. W.; Scheidt, K. A.; Snurr, R. Q.; Hupp, J. T.; Farha, O. K. Chem. Commun. 2013, 49, 9449–9451.
129. Matoga, D.; Oszajca, M.; Molenda, M. Chem. Commun. 2015, 51, 7637–7640.
130. Adams, C. J.; Kurawa, M. A.; Orpen, A. G. Inorg. Chem. 2010, 49, 10475–10485.
131. Friscic, T. J. Mater. Chem. 2010, 20, 7599–7605.
132. André, V. M.; Hardeman, A.; Halasz, I.; Stein, R. S.; Jackson, G. J.; Reid, D. G.; Duer, M. J.; Curfs, C.; Duarte, M. T.; Friscic, T. Angew. Chem. Int. Ed. 2011, 50, 7858–7861.
133. Hasa, D.; Schneider Rauber, G.; Voinovich, D.; Jones, W. Angew. Chem. Int. Ed. 2015, 54, 7371–7375.
2.07 Gas/Solid Complexation and Inclusion
VV Gorbatchuk, AK Gatiatulin, and MA Ziganshin, Kazan Federal University, Kazan, Russia
Ó 2017 Elsevier Ltd. All rights reserved.

2.07.1 Introduction 139


2.07.2 The Objectives of Gas/Solid Complexation and Inclusion Studies 139
2.07.3 How Does It Work? 139
2.07.4 Advantages of Gas/Solid Complexation and Inclusion 141
2.07.5 Setbacks and Problems of Gas/Solid Complexation and Inclusion Studies 142
2.07.6 How to Bypass a Guest Inclusion Threshold in Binary Host–Guest Systems 144
2.07.7 Biomimetic Property: Simultaneous Inclusion with a Second Volatile Component May Help to
Overcome a Guest Inclusion Threshold 146
2.07.8 Avoiding Problems with a Guest Inclusion Threshold Using a Guest Exchange 147
2.07.9 What Selectivity May Be Expected from Gas/Solid Complexation and Inclusion? 148
2.07.10 Conclusion 149
Acknowledgments 149
References 149

2.07.1 Introduction

Gas–solid complexation and inclusion by supramolecular hosts have major technological, analytic, and theoretical importance.1–3
These processes can provide controlled conditions for preparation of inclusion compounds, solid crystalline complexes, and poly-
morphs using dry techniques, which exclude evaporation of excessive solvent and guest.4–9 Some supramolecular materials such as
metal–organic frameworks (MOFs) are designed especially for gas storage, where an inclusion or complexation is implied for flex-
ible structures. 3,10–12 Supramolecular hosts can be used as sensitive layers in various vapor and gas sensors.13–17 Gas–solid inclu-
sion/complexation provides a unique opportunity to determine the Gibbs formation energy of solid complexes/inclusion
compounds and to study structure–property relationships for their crystals prepared under comparable conditions.6–9,14,18,19
Hence, additional guidelines for optimization of corresponding preparation techniques can be found to reach a desired selectivity,
clathrate stability or inclusion reversibility, equilibrium conditions of gas or vapor inclusion experiments, and their reproducibility.

2.07.2 The Objectives of Gas/Solid Complexation and Inclusion Studies

Supramolecular hosts, such as calixarenes,4–8, 20,21 cyclodextrins,22–24 various cyclophanes,15,25,26 and nonmacrocyclic
cavitands3,4,11,27–29 have an ability to form complexes and inclusion compounds with vapors and gases, which are stable up to
temperatures much higher than the boiling point of the included guest. Not only the products of this gas–solid interaction are
formed normally because of reduced energy of cavity formation in the host phase, even compared with that of the pure guest
liquid,6,30 but also host–guest H-bonding and/or donor acceptor complexation may take place.18,30 Respectively, the inclusion
products may be classified as clathrates and coordinatoclathrates30 or just clathrates. In some cases, corresponding complexes
may be found both in the solid phase and in solution, as with inclusion complexes of cyclodextrins,31 guest–host capsules with
dimers of tetraureacalix[4]arene derivatives,32 and various cryptophanes.33,34 Another example of such complexation in both solu-
tion and the solid phase with a gas is a colored complex between NOx and tert-butylcalix[4]arene alkylated at the lower rim.35,36
Porous MOFs12 and cucurbiturils37,38 with a fixed gas–solid surface interface do not fit the definition of inclusion or clathrate-
forming hosts since they undergo reversible adsorption with coinciding adsorption and desorption isotherms.

2.07.3 How Does It Work?

Gas/solid complexation (inclusion) is extremely simple. It is an equilibration of solid host powder with a guest in the gaseous 39,40
or vapor4,6,8,9,18,29,41 phase. This procedure may be static6,8,9,18,29,41 or dynamic.42,43 The first is a batch process in a hermetically
closed system having two or three phases: gas þ solid, Fig. 1a and c, or gas þ liquid þ solid, Fig. 1b, respectively, without a direct
phase contact between the guest liquid and the host powder.29 Batch inclusion process may occur with, Fig. 1b and d, or without,
Fig. 1a, restriction on guest diffusion into the solid host. In the dynamic procedure, Fig. 1d, the guest vapor or gas is purged over the
host layer on some support using an external source of gaseous or vaporous guest.42,43 The batch process may be preferred if one
wants to reach an inclusion equilibrium at the expense of the longer inclusion kinetics. The dynamic inclusion is much faster, but it

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12499-0 139


140 Gas/Solid Complexation and Inclusion

Figure 1 Schemes of experimental cells for gas and vapor complexation/inclusion by solid host samples: (a) batch inclusion in 2-phase system
without restrictions on guest (gas) diffusion to the solid host; (b) batch inclusion in 3-phase system with restricted rate of guest vapor diffusion from
its liquid to the solid host; (c) batch inclusion in 2-phase system with restricted rate of guest (gas) diffusion to the solid host; (d) dynamic saturation
of solid host with guest gas or vapor.

requires much more complicated equipment, and control of the guest’s relative vapor pressure or thermodynamic activity may be
difficult, especially near its unity value or in the case of activity variation over the time of the experiment.
The specific experimental approach for studying gas–solid complexation/inclusion depends much on the thermodynamics of
the process. A feature peculiar to guest inclusion by solid host is a phase transition from an initial solid state of guest-free host
to a host–guest inclusion compound. The isotherm describing this transition is defined by the Gibbs phase rule. It has a sigmoidal
stepwise shape, curve a in Fig. 2, with zero uptake of guest A up to a threshold value of guest thermodynamic activity or relative
vapor pressure P/P0, where P is the partial vapor pressure of the guest in the studied system and P0 is its saturated vapor pressure. As
this threshold is reached, there is a cooperative increase of guest uptake at its fixed activity up to a value corresponding to the compo-
sition of a saturated clathrate. Then, the value A does not change until a noticeable multilayer adsorption of guest occurs on the
solid/air interface. The corresponding vapor sorption isotherm can be fitted by the Hill equation adjusted to “guest uptake A versus
relative vapor pressure P/P0” coordinates, Fig. 2a:6,8,9,41
h i
A ¼ SCðP=P0 ÞN 1 þ CðP=P0 ÞN (1)

where S is the guest content in the saturated clathrate, C is a sorption constant, and N is a cooperativity parameter. This equation is
a version of a more general mathematical expression used to describe cooperative processes in biological and other systems.44

Figure 2 The most common types of sorption isotherms for gas and vapor complexation/inclusion by solid host.
Gas/Solid Complexation and Inclusion 141

The fitting of a sorption isotherm with the Hill equation (1) is not just a formal description. The corresponding shape, curve a in
Fig. 2, implies a stable homogeneous state of initial host with the absence of pores larger than the guest molecules. The curve a is
observed for crystalline guest-free hosts after prolonged thermal treatment below the point of temperature-induced polymorphic
transitions45,46 but above the collapse of a possible metastable phase induced by guest inclusion and elimination.14,47–49 Such treat-
ment should be at a temperature high enough to reduce the proportion of amorphous phase in the host powder.
The sigmoidal shape a of a gas or vapor guest sorption isotherm fitted by Eq. 1 is an ideal case, while other isotherm shapes are
also possible in real gas/solid complexation/inclusion experiments, providing a cue for host preparation procedures. A rather rare
case is an isotherm b with two or more steps of guest inclusion with the formation of stable intermediate inclusion compounds,
Fig. 2.7,14,50 If the host is in a metastable polymorphic47 or in a more or less amorphous11 form, the observed sorption isotherm
may exhibit a significant guest uptake event up to a threshold value of its relative vapor pressure P/P0 corresponding to a phase
transition, curve c, or the apparent cooperativity of phase transition is not revealed, curve d.1 The last sorption isotherm shape
or, better, reproducible linear shape, is desirable for analytic detection of vapors.13 However, only isotherms a and b correspond
to an equilibrium process of guest inclusion in a binary host–guest system as both its initial and final states are at equilibrium.
Hence, all other types of inclusion isotherms for hosts with molecular packing should strongly depend on the host preparation
conditions and thus generally are not reproducible or can be reproducible only if these conditions are strictly repeated.

2.07.4 Advantages of Gas/Solid Complexation and Inclusion

Having a complex isotherm shape for guest vapor sorption by a solid host and specific requirements for the initial state of the host,
this type of clathrate preparation still has an edge over crystallization of host–guest inclusion compounds from liquid solutions. 2
Ordinary crystallization procedures using cooling of a hot solution or isothermal evaporation of solvent with subsequent drying
from excessive liquid are generally not equilibrium processes. In these procedures, the thermodynamic activities of the main
components and solvent impurities are not fixed, and even slight impurity, even if not included by the host, may have a strong
influence on the composition of the resulting crystals.51 Inclusion stoichiometry and thermal stability data determined for a result-
ing product may not always be comparable for clathrates of the same host with different guests.
Unlike crystallization from solution, guest inclusion from the gas (vapor) phase may be performed with both the initial state of
the host and the final state of the inclusion compound being at equilibrium. For an initial solid host (apohost), this may be
controlled using differential scanning calorimetry (DSC) 52 or simultaneous TG/DSC/MS experiments to check possible polymor-
phic transitions and irreversibly included solvent.14,47,48 In the final state, gas pressure P or relative vapor pressure of the guest P/P0
reaching the equilibrium value may be determined using just a pressure transducer40 or headspace GC analysis,18 respectively, and
the guest uptake A is calculated from mass balance. This experiment is also used for estimation of guest inclusion kinetics.5 If the
guest vapor is present in large excess, the equilibration process may be followed by the change of the host sample mass,4 and the
composition of resulting product may be analyzed using various experimental methods.29,48
Once the solid inclusion compound is prepared under equilibrium conditions, its stoichiometry S and other parameters become
eligible for analysis of structure–property relationships and predictions of host–guest inclusion behavior. The most important
parameter is the Gibbs inclusion energy DGc, which can be calculated from corresponding vapor sorption isotherm using the
equation: 6,7,9,18
ð1
DGc ¼ RT lnðP=P0 Þ dY (2)
0

where Y ¼ A/S is the extent of host saturation with the guest. The value of DGc corresponds to the molar Gibbs energy of guest
transfer from a standard state of its pure liquid or solid to a saturated inclusion compound with guest contents S, in mol of guest per
mol of host. Eq. 2 does not depend on the model used to describe the guest sorption isotherm but requires the isotherm to have
a saturation component. Besides, the guest saturated vapor pressure P0 should have a defined value at the experimental temperature.
For example, if the guest is a gas above its critical temperature, as in many gas sorption studies,39,40,52 one needs to use some
extrathermodynamic assumptions to proceed with the evaluation of DGc.
To calculate DGc directly from the fitting parameters of Eq. 1, the next equation may be used, which is a consequence of Eqs. 1
and 2:
DGc ¼ RT ln a0:5S ¼ RT ðlnCÞ=N (3)
where a0.5S is the guest activity (relative vapor pressure) at 50% saturation of the host by the guest (Y ¼ 0.5).
If guest inclusion occurs in two or more steps, Fig. 2, curve b, the Gibbs inclusion energy DGc is calculated as a weighted average
of its values DGc ðiÞ for the separate steps:7
X .X
DGc ¼ Si DGCðiÞ Si (4)

where Si is the guest mol number included in the ith step of clathrate formation.
Eq. 3 defines a physical meaning of parameters C and N in Eq. 2 for guest inclusion by solid host. Separately, these values are
formal except for an ideal case of phase transition, where the sorption isotherms of type a in Fig. 2 should have a vertical part with
142 Gas/Solid Complexation and Inclusion

an infinitely large value of the cooperativity parameter N / N, which corresponds to zero degrees of freedom according to the
Gibbs phase rule at constant temperature.
Determination of Gibbs inclusion energies DGc from vapor sorption data is a unique possibility for solid host–guest complexes/
inclusion compounds. The values DGc for various calixarenes6–9, 14,18,19 and diol host18 are in the range from  0.4 to  8.9 kJ/mol,
Table 1, which is low as compared with the values of Gibbs complexation energy, for example, with those of cyclodextrins in
aqueous solutions31 and of calixarenes in organic solvents as determined by NMR titration53 in a concentration scale. But the values
of Gibbs complexation energy for cyclodextrins in water become quite comparable to the DGc values given earlier when derived
from concentration to thermodynamic activity scale using very large limiting activity coefficients of hydrophobic solutes (guests)
in water solution54 if activity coefficients of hydrophilic host and hydrophilic inclusion complex have nearly the same values.55
The problem with the results of NMR titration experiments for host–guest complexation is in the association of polar guest mole-
cules in a nonpolar solvent, which is evident from a steep concentration dependence of their activity coefficients in such binary
liquid mixtures. 56 This association is practically invisible in NMR experiments when both solute (guest) and solvent molecules
do not have aromatic groups and are not H-bonded.57 As a result, when any aromatic compound (aromatic host) is dissolved
in this mixture, it becomes preferentially solvated by an already associated polar component (guest) of the binary solvent. Such
a phenomenon is discussed in the studies of solvatochromism in binary solvents.58 Still, the model of 1:1 host–guest complexation
may provide a good fitting for host–guest–solvent systems just because the properties of such moderately dilute solutions are well
described by the rather formal Wilson equation,56 giving a similar shape of concentration dependence for solute parameters.
Hence, the studies of gas–solid complexation and inclusion may be flexible enough to carry these processes out under equilib-
rium conditions so that determination of Gibbs inclusion energies DGc is possible. These values may be used to derive structure–
property relationships for crystalline host–guest inclusion compounds, 6,18 which may help to rationalize and to adjust experi-
mental procedures and methods for their preparation and study.

2.07.5 Setbacks and Problems of Gas/Solid Complexation and Inclusion Studies

A major problem, which should be accounted for in studies of gas/solid complexation and inclusion, is irreversibility of this process
producing strong hysteresis of sorption and desorption isotherms. 1,5,59 When inclusion compounds are formed, they may be
extremely stable as they are kept together by the same crystal cooperativity that prevents their formation below the threshold value
of guest activity. For calix[4]arene39 and thiacalix[4]arene,60 the onset release temperatures of some guests are more than 200 C
higher than their boiling points. This means the practical absence of desorption at standard temperature of 25 C for those guests,
which are liquid at this temperature. On the contrary, the inclusion process is rather fast14,59 and can reach an equilibrium if the
guest desorption is not needed to adjust the solid-phase composition to a current value of guest relative vapor pressure P/P0.
The influence of inclusion irreversibility may be especially important when using supramolecular hosts in vapor sensors and in
sorption studies with the lack or an equivalent amount of guest using a batch inclusion process, Fig. 1a–c. In the last case, the final
relative vapor pressure of guest P/P0 may not be that expected at equilibrium, if its initial value is above the inclusion threshold a0.5S
and decreases due to host–guest interaction. During equilibration, if guest inclusion is irreversible, some outer layer of the host has
initial contact with more concentrated vapor, thus forming inclusion compound with a composition corresponding to the satura-
tion part of the sorption isotherm, Fig. 3. Once the uptake of guest produces a decrease of its relative vapor pressure P/P0, this value
may fall below its threshold level a0.5S, thus making further inclusion impossible. In the absence of a noticeable desorption process,
the final P/P0 value will be determined by the composition of an inner layer of host sample that has interacted with less saturated
vapor. So the apparent isotherm will be shifted from its equilibrium position to the lower activity values, Fig. 3c.
This irreversibility factor would not have an effect if the inclusion isotherm were ideal, being vertical at the threshold activity of
guest a0.5S and having no transient parts below and above this value, Fig. 3a.3
The inclusion isotherms for real host–guest systems often have a slope lower than 90 near this point at P/P0 ¼ a0.5S, Fig. 3b.7 In
this case, if the relative vapor pressure of guest P/P0 is falling, its final value (curve c, Fig. 3) will be lower than that expected at
equilibrium, curve b, Fig. 3. In Fig. 3, point E is defined by the equilibrium values of guest uptake A and P/P0 determined at nearly
zero rates of sorption and desorption. At the falling P/P0 value, the apparent sorption isotherm will have point D with the same
average guest uptake A as the equilibrium point E, but its final P/P0 value will be equal to that of point F corresponding to the
last portion of guest included. As diffusion of guest vapor (gas) through a host powder is also a nonequilibrium process, such
an irreversibility effect may decrease reproducibility of the observed thermodynamic and kinetic data if the experimental conditions
are not sufficiently suitable.
To avoid these problems associated with inclusion irreversibility in a batch process, guest gas or vapor should be supplied to the
host sample at a lower rate than that of guest inclusion by the solid host, so that the guest P (or P/P0) value would only be increasing
during the equilibration process and not decreasing at any moment. This can be achieved using a two-chamber vessel with the guest
and host initially being in the different chambers, while diffusion of guest in gas phase to the solid host sample is restricted by
small diameter of connecting channels or tubes, Fig. 1b and c. Inclusion irreversibility does not influence the results of “guest
vapor þ solid host” interaction studies if the guest is taken in large excess compared with the host inclusion capacity, for example,
if a saturated inclusion compound is prepared. The dynamic inclusion processes, Fig. 1d, also do not experience such problems as
the guest pressure is kept constant.4,42,43
Gas/Solid Complexation and Inclusion 143

Table 1 Gibbs energies of guest inclusion DGc (kJ/mol) at 298 C determined from vapor sorption isotherms by solid calixarenes and diol host

tert-Butylcalix[4]arene 6,8,9,18
MeOH  1.2 Et3N 6.1 CH2Cl2 8.9
EtOH  2.3 n-Hexane 4.7 CHCl3 2.3
n-PrOH  2.5 n-Heptane 4.8 CCl4 3.9
Acetone  1.9 n-Octane 4.3 Benzene 7.3
MeCN  4.0 n-Nonane 2.4 Toluene 5.6
EtCN  5.2 c-Hexane 5.5 o-Xylene 2.7
PrCN  2.9 Pyridine 8.1

tert-Butylcalix[5]arene 7
EtOH  0.7 CH2Cl2 3.6 c-Hexane 2.4
EtCN  1.7 CHCl3 2.9 Benzene 4.6
CCl4 3.0 Toluene 2.8

tert-Butylcalix[6]arene a
EtCN  1.2 CH2Cl2 1.3 c-Hexane 1.0
CHCl3 2.8 Benzene 1.4
CCl4 2.0 Toluene 1.7

tert-Butylcalix[8]arene a
CH2Cl2 2.4 Pyridine 1.5
CHCl3 3.1 Benzene 7.1
CCl4 1.8 Toluene 5.5

Adamantylcalix[4]arene 14
MeCN  3.0 CCl4 1.9 n-Hexane 2.3
EtCN  1.6 Benzene 2.8 n-Heptane 2.0
PrCN  1.1 Toluene 3.4 c-Hexane 1.6
CHCl3  2.4 PhEt 2.2

tert-Butylthiacalix[4]arene 6
MeOH  0.4 Acetone 1.3 CH2Cl2 2.0
MeCN  1.9 CH3NO2 1.8 CHCl3 1.1
EtCN  1.1 CCl4 0.4

2,2-Bis(9-hydroxyfluoren-9-yl)biphenyl 18
MeOH  2.1 MeCN 2.6 CHCl3 1.8
EtOH  0.6 EtCN 2.5 CCl4 1.1
n-PrOH  0.8 PrCN 2.1 Pyridine 4.3
Acetone  1.9 1,4- 2.6 Benzene 1.3
Et3N  2.1 Dioxane

Calculated from experimental lnC and N values from Ref. ( 19) using Eq. 3.
a
144 Gas/Solid Complexation and Inclusion

Figure 3 Influence of the guest inclusion irreversibility on vapor sorption by a solid host at the rate of guest desorption being near zero: (a) ideal
isotherm defined by the Gibbs phase rule for a two-component three-phase system, (b) experimental isotherm for vapor sorption of tetrachloro-
methane by solid tert-butylcalix[5]arene 7 at relative vapor pressure P/P0 of guest increased along the host–guest equilibration time scale, (c) and
calculated isotherm for relative vapor pressure P/P0 of guest decreased during equilibration process. For each experimental point, a separate batch
equilibration was performed as shown in Fig. 1b.

In vapor sensor applications, the irreversibility of guest inclusion by solid hosts complicates the experiment if one tries to regen-
erate the sensitive host layer at a temperature that is much lower than the onset point of guest release from the inclusion
compound. 14 Guest desorption time under such conditions may be far beyond the patience margin of a researcher. For example,
heating of a quartz crystal microbalance (QCM) crystal with a layer of an orthogonal anthracene-bis-(resorcinol)tetraol at 140 C for
4 h is necessary for host regeneration.61,62 Such treatment gives guest vapor sorption isotherms of type a in Fig. 2, which confirms an
inclusion irreversibility. In this case, a possible option is to restrict the range of analytes to those that may still be removed
completely from the host under experimental conditions. For various hydrophobic hosts, such guests may be alcohols,13,14 having
the lowest affinity as estimated by absolute values of Gibbs inclusion energies DGc, Table 1.

2.07.6 How to Bypass a Guest Inclusion Threshold in Binary Host–Guest Systems

The specific isotherm shape of guest inclusion by a solid host, Figs. 2a and 4, makes this process selective in a way that is different to
the selectivity of complex formation in liquid solutions.14 In solution, a small variation in a guest molecular structure only produces
a change in the stability constant of the complex.31 This is quite different for guest inclusion by solid host with a phase transition. In
the latter case, a small change in guest molecular size can shift the inclusion threshold a0.5S above unity value with DGc > 0, Eq. 3,
which makes inclusion impossible at any guest activity P/P0,6 Fig. 4a–c. So, a studied host may include only one of two close guest

Figure 4 Stepwise selectivity by guest activity threshold required for inclusion by crystalline host: (a, b) two guests are included with the same
stoichiometry but with different Gibbs inclusion energy DGc < 0 defined by Eq. 3; (c) no inclusion. Dotted curves correspond to multilayer sorption
on the gas/solid interface.
Gas/Solid Complexation and Inclusion 145

homologues from the vapor phase.6,23,60 This behavior relates to that observed in crystallization of host–guest inclusion
compounds from solutions consisting of guests with more or less similar structure, where some are included and the others
are not.63
A possible way to overcome the guest inclusion threshold is to increase temperature and guest pressure P. Such a method was
used to include xenon and several other gases into tert-butylcalix[4]arene at 100 C and P ¼ 10–20 atm in a sealed quartz tube. 52 The
resulting clathrates are rather stable at room temperature. In this case, the guest is above its critical point of Tc ¼ 289.74 K at
P ¼ 57.7 atm.64 So, only its virtual relative vapor pressure P/P0 may be discussed. For example, some theoretical value of saturated
vapor pressure above guest critical point may be calculated by extrapolation using the Antoine equation parameters from the exper-
imental temperature range.65 Increasing temperature reduces the virtual P/P0 value at fixed pressure P, which should be increased to
compensate for this reduction. The temperature dependence of the guest inclusion threshold a0.5S has not been studied yet for such
hosts. This dependence may be rather complicated by the ability of the host to undergo polymorphic transitions induced by heating
and/or contact with the guest.
Another option to remove or reduce restrictions imposed by the Gibbs phase rule on the guest inclusion is to convert an initially
stable guest-free host to a metastable polymorph having significant free volume in its crystal structure. This can be achieved either by
heating the host above the point of its thermally induced polymorphic transition to a metastable form, Fig. 5a,5 or by guest inclu-
sion and release, Fig. 5b.3,47–49,66 The hosts with the first type of behavior show an endothermic peak corresponding to a polymor-
phic transition in their DSC curve, Fig. 5a.21,52 Polymorphs of the second type with a memory of previously included guest exhibit
an exothermic peak for host collapse to a stable polymorph, Fig. 5b.47–49,67 For hosts thus prepared, the observed guest sorption
isotherms are of types d and c in Fig. 2.5,47 But if the guest is large enough for the available free volume in the host phase, like acet-
ylene for a metastable polymorph of tert-butylcalix[4]arene,5 its vapor sorption isotherm acquires shape a, Fig. 2, expected for
a stable polymorph.
Some molecular hosts such as substituted calixresorcin[4]arenes, 68 Dianin’s compound,11,69 and cucurbit[6]uril37 can be
prepared in crystalline guest-free forms with stable porosity by removal of an initially included solvent, Fig. 5c. These hosts are

Figure 5 Schemes of preparation and thermal behavior in a simultaneous TG/DSC experiment for apohost (guest-free host) samples having free
volume: (a) thermally induced metastable polymorph, (b) metastable polymorph induced by guest inclusion and release, (c) host with permanent
porosity stable up to host thermal destruction, (d) host in an amorphous glass state.
146 Gas/Solid Complexation and Inclusion

able to sorb gases and vapors according to the isotherm c in Fig. 2.3,11,37 For such hosts, single-crystal-to-single-crystal transforma-
tion without a change in crystal packing is possible as a result of guest sorption.68
If a search of metastable polymorphs for vapor or gas inclusion is not successful for a given host, one may try to prepare it in an
amorphous form, Fig. 5d. This can be made by desorption of included guest from the host and further grinding,1 by cooling the
host melt,70 or, if the host does not melt without decomposition, by spray-drying used for a-cyclodextrin.22 The glassy thiacalix[4]
arene derivative with four AcOCH2CH2NHCOCH2Oe substituents on the lower rim and a partial cone configuration has a higher
inclusion capacity than its stable crystalline form.70 The use of this glassy calixarene helps to visualize an inclusion of guest vapors as
giving a nontransparent crystalline product. Being more cooperative, this process is more selective than crystallization of glassy poly-
mers in solvent vapors.
An amorphous state of a host is desirable for sensor applications because the absence of crystal cooperativity may reduce irre-
versibility of guest inclusion from the vapor phase, and the isotherms of type d in Fig. 2 may be expected. For dipeptides L-alanyl-L-
valine and L-valyl-L-alanine, the guest inclusion capacity of their thin amorphous films on the surface of a QCM sensor is higher in
most cases than that of their crystalline powders.71 Still, these thin amorphous films produce a reversible response to guest vapors if
their inclusion does not induce crystallization. Crystallization of the host film by guest inclusion detected by atomic force micros-
copy gives a dramatic decrease in sensor response to the same guest even if the host mass is returned to its initial guest-free value by
the regeneration procedure.71

2.07.7 Biomimetic Property: Simultaneous Inclusion with a Second Volatile Component May Help to Overcome
a Guest Inclusion Threshold

A guest activity threshold a0.5S that is too high for inclusion may be decreased not only by host conversion to metastable or amor-
phous form. Effective elimination of inclusion restrictions imposed by the Gibbs phase rule and low host–guest affinity is reached
using a third component with a higher value of such affinity and the capability of forming a ternary clathrate with the host and
a target guest. This is a current practice in the preparation of inclusion compounds with cyclodextrins, where more or less water
and/or organic solvent are added. 72,73 For this, the cyclodextrin and guest are mixed in aqueous or water-organic solutions, pastes,
or slurries, by suspension of hydrated host in liquid guest, or put together by damp mixing.72,73
Gas–solid complexation/inclusion with both guest and water sorbed by initially dry cyclodextrin from the vapor phase, Fig. 6,
can be useful in two ways. The first is a possibility for independent variation of guest and water activity, which makes it easier to
optimize inclusion conditions.23 The second is in a more simple thermodynamic picture of this inclusion process than, for example,
that of complexation in solution, where thermodynamic activity of the inclusion compound is not known. Equilibration of
a water þ guest vapor mixture with solid cyclodextrin in the absence of a liquid phase in contact with the solid host gives an equi-
librium, where activities of the host and inclusion compound are equal to unity, while the activity (relative vapor pressure (P/P0)) of
the guest can be determined experimentally.23
A typical vapor sorption isotherm expressed as guest uptake derived to unity value of guest relative vapor pressure A/(P/P0)
versus host hydration observed for benzene and b-cyclodextrin (bCD) 23 is given in Fig. 6. This isotherm shows a threshold hydra-
tion value being necessary for inclusion of a relatively hydrophobic guest by this hydrophilic host up to a maximum value limited
by competition with water for the inclusion sites in the host phase. Nearly the same results were obtained for suspensions of
hydrated cyclodextrins in liquid hydrophobic guest d-limonene.74 Such experimental data simplify the analysis of the role of water
in the complexing ability of cyclodextrins compared with the complexation parameters obtained for their liquid solutions, which
are usually studied in this context.31 This role has been an objective of research and various hypotheses for more than 45
years.31,75,76 Of no less theoretical importance is the possibility of comparing the hydration effect and the role of water for bCD
and for proteins studied under the same experimental conditions in ternary solid–gas systems.77,78

Figure 6 Scheme and isotherm of simultaneous vapor sorption of water and hydrophobic guest benzene by initially dry b-cyclodextrin. 23
Gas/Solid Complexation and Inclusion 147

2.07.8 Avoiding Problems with a Guest Inclusion Threshold Using a Guest Exchange

The simultaneous vapor sorption of two guest compounds (or guest and water) by a solid host is not a universal solution even for
cyclodextrins. For bCD and guests with an intermediate hydrophobicity, like propanols and propionitrile, inclusion compounds
with a significant amount of guest included can be prepared only by a batch process, Fig. 1b, with the lack of guest and water
in an optimal ratio.23 The saturation of prehydrated host with vapors of liquid guest taken in excess does not give inclusion of these
compounds. The same is valid for crystallization from water solutions according a procedure described earlier.79,80 In this case,
a good option is a solid-phase exchange of hydrophobic guest in its water-free clathrate with bCD for another guest from the vapor
phase, Fig. 7a. This procedure gives good results for the exchange of benzene in its anhydrous clathrate bCD∙0.9C6H6 for vapors of
aliphatic alcohols C1–C4, nitriles C2–C3, chloroform, and toluene, while significantly more hydrophobic cyclohexane and n-hexane
are included only in addition to encapsulated benzene.23 The inclusion capacity of bCD for n-propanol and propionitrile by this
technique is three times higher than at the optimal host–guest–water ratio. When a clathrate with tetrahydrofuran
bCD∙1.0THF∙1.0H2O is used for solid-phase exchange, THF cannot be replaced by vapors of benzene and n-hexane.24 This clath-
rate preparation method is equally successful when a second guest is more hydrophilic than an initially encapsulated one and does
not require a complicated optimization of the host–guest–water ratio as in the ordinary clathrate preparation methods for
cyclodextrins.
For calixarenes, a procedure of solid-phase guest exchange for vapors of a second guest is more simple, Fig. 7b, as it does not
include a step of clathrate dehydration. An effective strategy for this method is to prepare an initial saturated inclusion compound in
a binary host–guest system with the highest guest contents and the lowest onset temperature of guest release.48,60 For example, for
tert-butylthiacalix[4]arene, which includes few guests in a binary “guest vapor–solid host” system,6,48 its 1:2 (host–guest) clathrate
with 1,2-dichloroethane fits these requirements, while for thiacalix[4]arene, a 1:1 clathrate with pyridine is a good choice as an
initial material for guest exchange.60 Solid-phase exchange of guest in these clathrates for vapors of other guests gives a much wider
range of inclusion products than can be obtained in binary systems.48,60 In most cases, the exchange for a second guest gives more
stable inclusion compounds with the onset or/and differential thermogravimetry (DTG) peak temperatures of guest release higher
on 30–60 C than those of clathrates formed in binary systems. Such a method produces very stable inclusion products also with
gases as was found for saturation of 1:1 calix[4]arene∙acetone clathrate by CF4, C2F6 and several other similar guests.39 The method
of solid-phase guest exchange closely relates to a crystallization technique from guest–solvent mixtures where a solvent is not
included but is necessary for the formation of target inclusion compounds, as was shown for a tert-butylthiacalix[4]arene clathrate
with methanol.67
Experimental methods used to monitor guest exchange in host–guest systems, Fig. 7a and b, were reviewed in Ref. (29). These
are thermogravimetry (TG), DSC, gas–liquid chromatography, color change, conductivity and density measurements, powder X-ray
diffractometry, CP MAS NMR, and in situ crystallography. Very effective for these purposes is the simultaneous TG/DSC method
combined with mass spectrometric detection of evolved vapors and gases.23,24,48
Depending on the molecular structure of a second guest, its vapor can produce not only guest exchange in an initial inclusion
compound product but also guest-free host. For example, 1,2-dichloroethane is expelled from its inclusion compound with tert-

Figure 7 Schemes of solid-phase guest exchange (a, b) and expulsion (c) in inclusion compounds by vapors of a second guest for hydrophilic
(a) and hydrophobic (b, c) hosts.
148 Gas/Solid Complexation and Inclusion

butylthiacalix[4]arene almost completely by vapors of toluene and trichloroethylene and partially by water vapor, Fig. 7c.48 This is
not just a desorption process because the second guest plays an active role, and in its absence, the initial clathrate is quite stable. For
comparison, acetic and propionic acids mutually prevent each other from inclusion by this host suspended in their liquid mixture in
a certain concentration range while being included separately.81
This guest-expelling ability and solid-phase guest exchange may be used in sensor applications for regeneration of supramolec-
ular hosts in sensitive layers. For example, QCM sensors with a sensitive layer of adamantylcalix[4]arene can be effectively regen-
erated by air purging at 45 C for 2 min after saturation with guest vapor and subsequent solid-phase guest exchange for ethanol
vapor. 14 Otherwise, one would need prolonged heating at 220 C in vacuum for the same result, which procedure may be not
suitable.

2.07.9 What Selectivity May Be Expected from Gas/Solid Complexation and Inclusion?

This problem is relevant for applications of supramolecular hosts in sensors and for capture and separation of gases and vapors.
When applied to intricate molecular structures, a simple expectation is to find a better selectivity compared with that of ordinary
sorbents and liquid solvents. This was the objective of a special study, 82 which showed that the selectivity of a QCM sensor
response by thin layers of resorcin[4]arene and a-cyclodextrin for vapors of organic guests closely follows the variation of
vapor–liquid partition coefficients for the same analytes in liquid polymers under comparable conditions when derived to
the same units.
The data on gas solubility in liquid organic solvents 83 may be also used for comparison with results of the sorption studies
for supramolecular hosts. Ref. (83) compiles the thermodynamic data for 16 gases in 34 solvents. For example, the solubility
ratios for CO2/H2 and CO2/CH4 pairs of solutes in n-heptane are 17.5 and 3.48, while in acetone these ratios are equal to
61.8 and 10.1, respectively. These values are not much different from those usually observed for MOFs and a-cyclodex-
trin.22,84,85 The problem is in a moderate variation in energy of molecular interactions, which restricts the number of polymer
liquids with essentially different solvation selectivity for vapor sensor arrays to only three.86 The same is supposed for solid
hosts.82
Still, a special and much higher selectivity for guest vapors may be found for crystalline hosts with molecular packing. This selec-
tivity is not based on preferential binding related to a higher inclusion/sorption capacity and/or affinity for a target guest by a lock-
and-key principle. Irrespective of the varied external conditions, each host can have a restricted number of polymorphs with
different molecular packings and of stable saturated and intermediate inclusion compounds, the formation of which may be rather
selective. Such a concept was discussed in the study of guest separation by crystallization of inclusion compounds from a host solu-
tion in binary mixtures of liquid guests 4 and may be used for molecular recognition. For example, a solid-phase guest exchange may
be rather selective.23,48,60
In studies of guest vapor inclusion by solid calixarene A, an absolute selectivity for benzene was found. 50 Benzene can be recog-
nized both as a pure substance and in any mixtures with its close homologues and other compounds if its activity is high enough.
This molecular recognition is performed according to a logical procedure in Fig. 8. A count of guest inclusion steps in a batch QCM
sensor experiment is enough to determine whether or not benzene is present with a sufficient relative vapor pressure P/P0 > 0.60.
Only benzene is included in two steps by host A. Molecular recognition with a close or the same selectivity can also be achieved by
observation of guest-induced polymorphs, Fig. 5b.47–49,67

Figure 8 Scheme of molecular recognition of benzene vapor by solid calixarene A using the shape of the QCM sensor response and vapor sorption
isotherm in batch inclusion process, Fig. 1b, according to Ref. (50).
Gas/Solid Complexation and Inclusion 149

2.07.10 Conclusion

Gas/solid complexation and inclusion are a versatile method used for gas and vapor sensing, storage and separation, molecular
recognition, preparation of inclusion compounds, and the study of structure–property relationships for host–guest crystals. Specific
features of this method are linked to inclusion cooperativity caused by phase transition from initial solid host to inclusion
compound and by possible polymorphic transitions. Depending on the experimental technique used, this method may be
employed for structural modification of the host material for various applications and preparation procedures and determination
of quantitative thermodynamic parameters such as Gibbs inclusion energy. Gas/solid complexation and inclusion may provide
a genuine molecular recognition, which is not based on the ordinary lock-and-key concept but relates to nearly digital ability of
some supramolecular hosts to form a restricted number of guest-induced polymorphs, saturated and intermediated clathrates.
The ordinary selectivity estimated from different host affinities for various guests in inclusion and adsorption experiments does
not differ much from selectivity of liquid solvents by gas/liquid partition coefficients.
The results of gas/solid complexation and inclusion studies strongly depend on the initial equilibrium or nonequilibrium state
of the solid host, guest inclusion irreversibility, and the value of guest inclusion threshold by its activity. Being a major problem, the
inclusion irreversibility can cause poor reproducibility of observed results if not properly met. In this article, experimental proce-
dures are reviewed, which are used to reach a permanent, virtual, or metastable host porosity, maximal selectivity, or reversibility
of guest adsorption/inclusion, to enable inclusion of a target guest that cannot enter the host matrix under equilibrium conditions
in binary host–guest systems and to make host eligible for experimental measurements of the thermodynamic inclusion parameters.
These parameters provide more adequate descriptions of host–guest interactions compared with those for complexation in liquid
solutions, even when phase differences are taken into account.

Acknowledgments

The work was supported by Russian Government Program of Competitive Growth of Kazan Federal University (KFU) and by RFBR, grant no. 14-03-
01007.

References

1. Tian, J.; Thallapally, P. K.; Dalgarno, S. J.; McGrail, P. B.; Atwood, J. L. Angew. Chem. Int. Ed. 2009, 48 (30), 5492–5495.
2. Lusi, M.; Barbour, L. J. Chem. Commun. 2013, 49 (26), 2634.
3. Soldatov, D. V.; Ripmeester, J. A. Stud. Surf. Sci. Catal. 2005, 156, 37–54.
4. Barbour, L. J.; Caira, M. R.; le Roex, T.; Nassimbeni, L. R. J. Chem. Soc. Perkin Trans. 2002, 2 (12), 1973–1979.
5. Thallapally, P. K.; Dobrzanska, L.; Gingrich, T. R.; Wirsig, T. B.; Barbour, L. J.; Atwood, J. L. Angew. Chem. 2006, 118, 6656–6659.
6. Gorbatchuk, V. V.; Tsifarkin, A. G.; Antipin, A. S.; Solomonov, B. N.; Konovalov, A. I.; Lhotak, P.; Stibor, I. J. Phys. Chem. B 2002, 106, 5845–5851.
7. Ziganshin, M. A.; Yakimov, A. V.; Safina, G. D.; Solovieva, S. E.; Antipin, I. S.; Gorbatchuk, V. V. Org. Biomol. Chem. 2007, 5, 1472–1478.
8. Gorbatchuk, V. V.; Tsifarkin, A. G.; Antipin, I. S.; Solomonov, B. N.; Konovalov, A. I. J. Incl. Phenom. Macrocycl. Chem. 1999, 35, 389–396.
9. Gorbatchuk, V. V.; Tsifarkin, A. G.; Antipin, I. S.; Solomonov, B. N.; Konovalov, A. I. Mendeleev Commun. 1999, 9 (1), 11–13.
10. Li, B.; Zhang, Y.; Krishna, R.; Yao, K.; Han, Y.; Wu, Z.; Ma, D.; Shi, Z.; Pham, T.; Space, B.; Liu, J.; Thallapally, P. K.; Liu, J.; Chrzanowski, M.; Shengqian, M. J. Am. Chem.
Soc. 2014, 136 (24), 8654–8660.
11. Afonso, R.; Mendes, A.; Gales, L. J. Mater. Chem. 2012, 22, 1709–1723.
12. Li, J.-R.; Kuppler, R. J.; Zhou, H.-C. Chem. Soc. Rev. 2009, 38 (5), 1477–1504.
13. Pirondini, L.; Dalcanale, E. Chem. Soc. Rev. 2007, 36 (5), 695–706.
14. Yakimova, L. S.; Ziganshin, M. A.; Sidorov, V. A.; Kovalev, V. V.; Shokova, E. A.; Tafeenko, V. A.; Gorbatchuk, V. V. J. Phys. Chem. B 2008, 112, 15569–15575.
15. Dickert, F.; Sikorski, R. Mater. Sci. Eng. C 1999, 10, 39–46.
16. Gruber, T.; Fischer, C.; Seichter, W.; Bombicz, P.; Weber, E. CrystEngComm 2011, 13 (5), 1422–1431.
17. Korotcenkov, G. Handbook of Gas Sensor Materials, Properties, Advantages and Shortcomings for Applications Volume 2: New Trends and Technologies; Springer-Verlag: New
York, NY, 2014.
18. Gorbatchuk, V. V.; Tsifarkin, A. G.; Antipin, I. S.; Solomonov, B. N.; Konovalov, A. I.; Seidel, J.; Baitalov, F. J. Chem. Soc., Perkin Trans. 2 2000, 2287–2294.
19. Ziganshin, M. A.; Yakimov, A. V.; Konovalov, A. I.; Antipin, I. S.; Gorbatchuk, V. V. Russ. Chem. Bull. 2004, 53 (7), 1536–1543.
20. Herbert, S. A.; Janiak, A.; Thallapally, P. K.; Atwood, J. L.; Barbour, L. J. Chem. Commun. 2014, 50 (98), 15509–15512.
21. Ananchenko, G. S.; Moudrakovski, I. L.; Coleman, A. W.; Ripmeester, J. A. A. Angew. Chem. Int. Ed. 2008, 47 (30), 5616–5618.
22. Ho, T. M.; Howes, T.; Bhandari, B. R. Powder Technol. 2014, 259, 87–108.
23. Gorbatchuk, V. V.; Gatiatulin, A. K.; Ziganshin, M. A.; Gubaidullin, A. T.; Yakimova, L. S. J. Phys. Chem. B 2013, 117, 14544–14556.
24. Gatiatulin, A. K.; Ziganshin, M. A.; Gorbatchuk, V. V. J. Therm. Anal. Calorim. 2014, 118, 987–992.
25. Tian, J.; Thallapally, P. K.; McGrail, B. P. CrystEngComm 2012, 14 (6), 1909.
26. Barbour, L. J. Nat. Chem. 2015, 7 (2), 97–99.
27. Katzsch, F.; Gruber, T.; Weber, E. Cryst. Growth Des. 2015, 15 (10), 5047–5061.
28. Nassimbeni, L. R. Acc. Chem. Res. 2003, 36 (8), 631–637.
29. Nassimbeni, L. R.; Su, H. CrystEngComm 2013, 15 (37), 7396.
30. Weber, E. Shape and Symmetry in the Design of New Hosts. In Comprehensive Supramolecular Chemistry, Atwood, J. L.; Davies, J. E.D; MacNicol, D. D.; Vögtle, F., Eds.;
Elsevier Science: Oxford, 1996; vol. 6, pp 535–592.
31. Rekharsky, M. V.; Inoue, Y. Chem. Rev. 1998, 98, 1875–1917.
32. Ziganshin, M. A.; Yakimova, L. S.; Khayarov, K. R.; Gorbatchuk, V. V.; Vysotsky, M. O.; Bohmer, V. Chem. Commun. 2006, 3897–3899.
33. Hill, P. A.; Wei, Q.; Troxler, T.; Dmochowski, I. J.; Pennsyl, V.; Pennsyl, V., 2009, 15, 3069–3077.
150 Gas/Solid Complexation and Inclusion

34. Joseph, A. I.; Lapidus, S. H.; Kane, C. M.; Holman, K. T. Angew. Chem. Int. Ed. 2015, 54 (5), 1471–1475.
35. Zyryanov, G. V.; Kang, Y.; Stampp, S. P.; Rudkevich, D. M. Chem. Commun. 2002, 23, 2792–2793.
36. Khabibullin, A. A.; Safina, G. D.; Ziganshin, M. A.; Gorbatchuk, V. V. J. Therm. Anal. Calorim. 2012, 110 (3), 1309–1313.
37. Tian, J.; Liu, J.; Liu, J.; Thallapally, P. K. CrystEngComm 2013, 15 (8), 1528–1531.
38. Tian, J.; Ma, S.; Thallapally, P. K.; Fowler, D.; McGrail, B. P.; Atwood, J. L. Chem. Commun. 2011, 47 (27), 7626–7628.
39. Atwood, J. L.; Barbour, L. J.; Jerga, A. Science 2002, 296, 2367–2369.
40. Atwood, J. L.; Barbour, L. J.; Thallapally, K.; Wirsig, T. B. Chem. Commun. 2005, 51–53.
41. Gorbatchuk, V. V.; Tsifarkin, A. G.; Antipin, I. S.; Solomonov, B. N.; Konovalov, A. I. Mendeleev Commun. 1997, 7 (6), 215–217.
42. Kalchenko, V. I.; Koshets, I. A.; Matsas, E. P.; Kopylov, O. N.; Solovyov, A.; Kazantseva, Z. I.; Shirshov, Y. M. Mater. Sci. 2002, 20 (3), 73–88.
43. Schramm, U.; Roesky, C. E. O; Winter, S.; Rechenbach, T.; Boeker, P.; Schulze Lammers, P.; Weber, E.; Bargon, J. Sensors Actuators B Chem. 1999, 57 (1–3), 233–237.
44. Abundo, M.; Accardi, L.; Finazzi Agrò, A.; Mei, G.; Rosato, N. Biophys. Chem. 1996, 58 (3), 313–323.
45. Brouwer, E. B.; Enright, G. D.; Udachin, K. A.; Lang, S.; Ooms, K. J.; Halchuk, P. A.; Ripmeester, J. A. Chem. Commun. 2003, 1416–1417.
46. Atwood, J. L.; Barbour, L. J.; Jerga, A. Chem. Commun. 2002, 2952–2953.
47. Yakimov, A. V.; Ziganshin, M. A.; Gubaidullin, A. T.; Gorbatchuk, V. V. Org. Biomol. Chem. 2008, 6, 982–985.
48. Galyaltdinov, S. F.; Ziganshin, M. A.; Drapailo, A. B.; Gorbatchuk, V. V. J. Phys. Chem. B 2012, 116, 11379–11385.
49. Safina, G. D.; Gavrilova, O. M.; Ziganshin, M. A.; Stoikov, I. I.; Antipin, I. S.; Gorbatchuk, V. V. Mendeleev Commun. 2011, 21 (5), 291–292.
50. Safina, G. D.; Validova, L. R.; Ziganshin, M. A.; Stoikov, I. I.; Antipin, I. S.; Gorbatchuk, V. V. Sensors Actuators B Chem. 2010, 148 (1), 264–268.
51. Brehmer, T. H.; Weber, E.; Cano, A. H. J. Phys. Org. Chem. 2000, 13, 63–74.
52. Enright, G. D.; Udachin, K. A.; Moudrakovski, I. L.; Ripmeester, J. A. J. Am. Chem. Soc. 2003, 125 (33), 9896–9897.
53. Arduini, A.; Giorgi, G.; Pochini, A.; Secchi, A.; Ugozzoli, F. Tetrahedron 2001, 57 (12), 2411–2417.
54. Sherman, S. R.; Trampe, D. B.; Bush, D. M.; Schiller, M.; Eckert, C. A.; Dallas, A. J.; Li, J.; Carr, P. W. Ind. Eng. Chem. Res. 1996, 35, 1044–1058.
55. Grechin, A. G.; Buschmann, H.-J.; Schollmeyer, E. Angew. Chem. Int. Ed. 2007, 46 (34), 6499–6501.
56. Hirata, M.; Ohe, S.; Nagahama, K. Computer Aided Data Book of Vapor–Liquid Equilibria; Elsevier Science & Technology: New York, NY, 1975.
57. Reichardt, C.; Welton, T. Solvents and Solvent Effects in Organic Chemistry; Wiley-VCH Verlag: Weinheim, 2011.
58. Machado, V. G.; Stock, R. I.; Reichardt, C. Chem. Rev. 2014, 114 (20), 10429–10475.
59. Dewa, T.; Endo, K.; Aoyama, Y. J. Am. Chem. Soc. 1998, 120, 8933–8940.
60. Galyaltdinov, S. F.; Ziganshin, M. A.; Gubaidullin, A. T.; Vyshnevsky, S. G.; Kalchenko, O. I.; Gorbatchuk, V. V. CrystEngComm 2014, 16 (18), 3781–3787.
61. Matsuura, K.; Ariga, K.; Endo, K.; Aoyama, Y.; Okahata, Y. Chem. Eur. J. 2000, 6 (10), 1750–1756.
62. Naito, M.; Sasaki, Y.; Dewa, T.; Aoyama, Y.; Okahata, Y. J. Am. Chem. Soc. 2001, 123 (44), 11037–11041.
63. Weber, E.; Hens, T.; Gallardo, O.; Csoregh, I. J. Chem. Soc., Perkin Trans. 2 1996, 737–745.
64. Haynes, W. M., Ed. CRC Handbook of Chemistry and Physics, 92nd ed.; CRC Press: Boca Raton, FL, 2011; pp 4–123.
65. Theeuwes, F.; Bearman, R. J. J. Chem. Thermodyn. 1970, 2, 507–512.
66. Soldatov, D. V.; Enright, G. D.; Ripmeester, J. A. Cryst. Growth Des. 2004, 4 (6), 1185–1194.
67. Morohashi, N.; Nanbu, K.; Tonosaki, A.; Noji, S.; Hattori, T. CrystEngComm 2015, 17 (26), 4799–4808.
68. Kane, C. M.; Ugono, O.; Barbour, L. J.; Holman, K. T. Chem. Mater. 2015, 27, 7337–7354.
69. Barrer, R. M.; Shanson, V. H. J. Chem. Soc., Chem. Commun. 1976, (9), 333–334.
70. Gataullina, K. V.; Ziganshin, M. A.; Stoikov, I. I.; Gubaidullin, A. T.; Gorbatchuk, V. V. Phys. Chem. Chem. Phys. 2015, 17, 15887–15895.
71. Ziganshin, M. A.; Gubina, N. S.; Gerasimov, A. V.; Gorbatchuk, V. V.; Ziganshina, S. A.; Chuklanov, A. P.; Bukharaev, A. A. Phys. Chem. Chem. Phys. 2015, 17, 20168–
20177.
72. Del Valle, E. M. M. Process Biochem. 2004, 39, 1033–1046.
73. Hedges, R. A. Chem. Rev. 1998, 98, 2035–2044.
74. Yoshii, H.; Furuta, T.; Yasunishi, A.; Hirano, H. J. Biochem. 1994, 115, 1035–1037.
75. Liu, L.; Guo, Q. X. J. Incl. Phenom. 2002, 42, 1–14.
76. Crini, G. Chem. Rev. 2014, 114, 10940–10975.
77. Gorbatchuk, V. V.; Ziganshin, M. A.; Solomonov, B. N. Biophys. Chem. 1999, 81 (2), 107–123.
78. Mironov, N. A.; Breus, V. V.; Gorbatchuk, V. V.; Solomonov, B. N.; Haertlé, T. J. Agric. Food Chem. 2003, 51, 2665–2673.
79. Stezowski, J. J.; Jogun, K. H.; Eckle, E.; Bartels, K. Nature 1978, 274, 617–619.
80. Jogun, K. H.; Stezowsky, J. J. Nature 1979, 278, 667–668.
81. Morohashi, N.; Noji, S.; Nakayama, H.; Kudo, Y.; Tanaka, S.; Kabuto, C.; Hattori, T. Org. Lett. 2011, 13 (13), 3292–3295.
82. Grate, J. W.; Patrash, S. J.; Abraham, M. H.; Du, C. M. Anal. Chem. 1996, 68 (5), 913–917.
83. Wilhelm, E.; Battino, R. Chem. Rev. 1973, 73, 1–9.
84. Burd, S. D.; Ma, S.; Perman, J. A.; Sikora, B. J.; Snurr, R. Q.; Thallapally, P. K.; Tian, J.; Wojtas, L.; Zaworotko, M. J. J. Am. Chem. Soc. 2012, 134 (8), 3663–3666.
85. Sozzani, P.; Bracco, S.; Comotti, A.; Ferretti, L.; Simonutti, R. Angew. Chem. Int. Ed. 2005, 44 (12), 1816–1820.
86. Grate, J. W. Chem. Rev. 2000, 100 (7), 2627–2648.
2.08 Magnetic Measurements
EL Gavey, University of California, San Diego, La Jolla, California, USA
JM Rawson, The University of Windsor, Windsor, ON, Canada
Ó 2017 Elsevier Ltd. All rights reserved.

2.08.1 Introduction 152


2.08.2 The Magnetism of Spin-Only Ions 152
2.08.2.1 The Spin Quantum Number, S 152
2.08.2.2 The Zeeman Effect and the Brillouin Function 153
2.08.2.3 Magnetic Susceptibility: Curie and Curie–Weiss Behavior 154
2.08.2.4 Zero-Field Splitting, D 156
2.08.2.5 Summary 157
2.08.3 The Contribution of Orbital Angular Momentum 157
2.08.3.1 The Origin of Orbital Angular Momentum 157
2.08.3.2 The Spin–Orbit Coupling Constant, l 157
2.08.3.3 Spin–Orbit Coupling for Lanthanides 159
2.08.3.4 Spin–Orbit Coupling for Transition Metal Ions 161
2.08.3.4.1 Second-Order Spin–Orbit Coupling In “Spin-Only Ions” 162
2.08.3.4.2 First-Order Spin–Orbit Coupling 162
2.08.3.5 Summary 163
2.08.4 Exchange-Coupled Spins 163
2.08.4.1 Direct Exchange 164
2.08.4.2 Superexchange 164
2.08.4.2.1 Goodenough–Kanamori Rules 164
2.08.4.2.2 Correlation Between Bridge Angle And Exchange Interaction 165
2.08.4.3 Double Exchange 165
2.08.4.4 Dipolar Exchange 165
2.08.4.5 Spin Frustration 165
2.08.4.6 Magnetic Models for Exchange-Coupled Systems 166
2.08.4.6.1 Limiting High Temperature Limit 166
2.08.4.6.2 Limiting Low Temperature Limit 166
2.08.4.6.3 Quantitative Models for Exchange-Coupled Systems 167
2.08.4.7 Summary 168
2.08.5 Single-Molecule Magnets 169
2.08.5.1 Relaxation Processes 169
2.08.5.1.1 Thermal Relaxation 170
2.08.5.1.2 Quantum Tunneling Mechanism 170
2.08.5.1.3 Determining Relaxation Rates 170
2.08.5.2 Summary 171
2.08.6 Long-Range Magnetic Order 171
2.08.6.1 Ferromagnetism 173
2.08.6.1.1 Magnetic Characterization of a Ferromagnet 174
2.08.6.2 Ferrimagnetism 174
2.08.6.2.1 Magnetic Characterization of a Ferrimagnet 175
2.08.6.3 Antiferromagnetism 175
2.08.6.3.1 Magnetic Characterization of an Antiferromagnet 175
2.08.6.3.2 Field-Induced Transitions in Antiferromagnets 176
2.08.6.4 Canted Antiferromagnetism 176
2.08.6.4.1 Magnetic Characterization of a Canted Antiferromagnet 177
2.08.6.5 Summary 177
2.08.7 Magnetic Measurements 177
2.08.7.1 Sample Preparation 178
2.08.7.2 Preliminary Measurements 178
2.08.7.3 Advanced Magnetic Measurements 179
2.08.8 Conclusions 179
Acknowledgment 179
Further Reading 179

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12502-8 151


152 Magnetic Measurements

2.08.1 Introduction

A discussion of magnetic measurements inherently requires some understanding of the underlying theory of molecular magnetism
in order to understand and plan the experiments. In addition, measurements are only useful if the data can be interpreted and it is
clear that there is a strong symbiotic relationship between theory, experiment, and data analysis. The magnetism of paramagnetic
materials whether they be transition metals, lanthanides, or organic radicals is a complex topic and many chapters of magnetochem-
istry texts are devoted to developing a detailed understanding of the magnetism of isolated paramagnetic centers. Many provide
a discussion of exchange-coupled systems and/or long-range order. It is clear that this article cannot hope to cover all the theory
in adequate detail. Rather we aim to explore the basic concepts necessary to understand the experiments as well as the necessary
theory to interpret the observed data. In particular, we hope to provide sufficient insight into magnetism and magnetic measure-
ments that a reader will be better equipped to understand and interpret the types of measurements which are frequently encoun-
tered when reading research papers.

2.08.2 The Magnetism of Spin-Only Ions

We shall commence our study with a discussion of the so-called “spin-only” ions, focusing on (a) what is meant by “spin-only”;
(b) the effect of a magnetic field and temperature on the magnetization induced in a paramagnetic material; (c) how we analyze
magnetization versus field and the temperature dependence of the paramagnetic susceptibility, c, to extract information on the
spin, S, of simple spin-only ions.

2.08.2.1 The Spin Quantum Number, S


When we consider the magnetic properties of transition metal ions, we need to consider the dn configuration, the coordination
geometry, and the magnitude of the interelectron repulsion term (P) in relation to the crystal field (D). The d-orbital splittings
for common coordination geometries are shown in Fig. 1 with d-orbital energies given in terms of the octahedral crystal field split-
ting, Do. The magnitude of D depends upon the metal, its oxidation state, and the peripheral ligands (based on their positions in the
Spectrochemical Series). Introductory coordination chemistry lectures typically highlight that for first row transition metal ions the

Figure 1 Common coordination geometries and associated crystal field splittings in terms of Do.
Magnetic Measurements 153

magnitude of D and P are comparable, resulting in either “high spin” (D < P) or “low spin” (D > P) configurations depending upon
the electronic configuration of the metal, its coordination geometry, and ligand set. For lower symmetry cases such as square pyra-
midal, it is also possible to achieve intermediate spin states, for example, Fe(S2CNEt2)2Cl has S ¼ 3 2. Here the energies of the dxy
=

and dz2 are sufficiently low for the promotion energy D to be less than P. However the energy of the dx2–y2 orbital is too high, such
that the fifth electron of the d5 Fe3 þ ion is placed “spin down” in the dyz (or dxz) orbital. Each electron has a spin ms which can be
either þ½ (“spin up”) or ½ (“spin down”). The total spin, S, associated with a metal ion is equal to the sum of the individual spins:
X
S¼ ms (1)

2.08.2.2 The Zeeman Effect and the Brillouin Function


The total spin S is comprised of a series of microstates, MS, which range from þS to S in integer steps. Thus when S ¼ 1 the values of
MS comprise: þ 1, 0,  1. In general the number of possible microstates (the spin multiplicity) is defined by the value of 2S þ 1. For an
S ¼ 1 state, 2S þ 1 ¼ 3 and is described as a “(spin) triplet.” Conversely for S ¼ ½ the value of 2S þ 1 ¼ 2 and it is referred to as
a “doublet” state. In the absence of an applied magnetic field, the energies of the MS sublevels are degenerate (but see zero-field
splitting, section “ Zero-Field Splitting, D”) and so for the S ¼ ½ system the population of the MS ¼ ½ (“spin down”) and
MS ¼ þ½ levels (“spin up”) are identical so there is no net magnetization (the magnetic moments of the ½ and þ½ states cancel).
However the energies of the MS levels are affected when a magnetic field is applied according to:
EðMS Þ ¼ gbHMS (2)
where E(MS) is the energy of the MS microstate, g is the electron g-factor (which varies from sample to sample, but is typically close
to 2.0 for “spin-only” ions), b is a constant known as the Bohr magneton, and H is the magnitude of the applied field. It should be
noted that g can often be determined experimentally by electron paramagnetic resonance (EPR) spectroscopy and it is preferable to
measure g directly rather than use it as a “free variable.” The effect of the applied field (the “Zeeman effect”) is to split the degeneracy
of an S ¼ 1 state as shown in Fig. 2 (left). Note that the energy of the MS ¼  1 microstate lies lower than the MS ¼ 0 and MS ¼ þ 1
levels in an applied field. As a consequence a net magnetization is induced in the sample since the Boltzmann population of the
MS ¼  1 state is larger than the population of the MS ¼ 0 and MS ¼ þ 1 states. Since the magnetization follows a Boltzmann
distribution, then it is clear that the magnetization depends upon (a) the magnitude of the applied field H and (b) the temperature.
Thus the magnetization is largest in a large applied field and at low temperature when there is little thermal population of excited
states. The sample magnetization for a spin-only paramagnet has been shown to follow the Brillouin function (Eq. 3) and the
expected M versus H plot for an S ¼ 1 paramagnet is shown in Fig. 3 (left).
M ¼ NgbS$BS ðxÞ
where
BS ðxÞ ¼ ½ð2S þ 1Þ=2Scothfð2S þ 1Þx=2Sg  ð1=2SÞcot hfx=2Sg and x ¼ gbSH=kT (3)
We should note two important areas:
(a) In small applied fields the M versus H/T plot is essentially linear and the gradient is known as the magnetic susceptibility, c. This
term is central to our further discussion of magnetic materials.
(b) In large applied fields and at low temperature, only the lowest MS level is populated and the magnetization becomes saturated
at a value given by Msat.

Figure 2 The effect of a magnetic field on the MS levels of an S ¼ 1 ion; (left) in the absence of zero-field splitting (D ¼ 0); (right) including the effect
of zero-field splitting (D < 0).
154 Magnetic Measurements

Figure 3 (Left) The Brillouin function for an S ¼ 1 system highlighting the saturation magnetization at gS and the molar susceptibility, c; (right) Bril-
louin functions for S ¼ ½, S ¼ 3 2, S ¼ 5 2, and S ¼ 7 2 ions.
= = =

Although a SQUID magnetometer measures the induced magnetization in terms of emu (electromagnetic units), M versus H
plots are often presented in terms of the non-SI unit, the Bohr magneton (1b ¼ 5585 emu mol 1). In units of Bohr magnetons,
the saturation magnetization, Msat, for a spin-only ion is directly related to S according to:
Msat ¼ gSð 2SÞ (4)
where g is the g-factor which varies from compound to compound but, as previously mentioned, is typically close to 2.0 for spin-
only compounds. Thus a measure of the saturation magnetization permits S to be determined.
If we wish to measure c we should use a small applied field. However a small applied field gives a small observed magnetization,
so a compromise must be reached between ensuring the field is sufficiently small that we are in the linear region of the M versus H
plot and having a sufficiently large field that we observe a strong induced magnetization M in the sample. This can be achieved by
initially measuring M versus H at low temperature to identify the linear region and using a value of H near the end of the linear part of
the M versus H plot.
In physics the magnetic susceptibility reflects how the magnetization changes with a change in applied field, that is, c ¼ vM/vH.
However when we are working in a small applied field we have seen that the gradient is constant and we can write:
c ¼ ðM=HÞ (5)

2.08.2.3 Magnetic Susceptibility: Curie and Curie–Weiss Behavior


The magnetic susceptibility we observe, cobs, comprises both the sample susceptibility (c) and the susceptibility of the sample
holder (csh). To determine the molar susceptibility, cm (in emu mol 1), we measure the observed sample magnetization (M) in
a known applied magnetic field (H). We then scale by the molecular weight (MW in g mol 1) and sample weight (SW in g)
and make a correction for the susceptibility of the sample holder (Eq. 6). A gelatin capsule is often used as a sample holder
and, although each capsule is slightly different, many groups will have a standard estimate of the sample holder susceptibility.
cm ¼ ½ðM=HÞ  ðMW=SW Þ  csh (6)
In addition the molar susceptibility of the sample (cm) comprises a paramagnetic component (cp) arising from attraction of
the electrons into the magnetic field and a diamagnetic component (cd) arising from repulsion of electron pairs (core electrons,
electrons in bonds, and lone pairs). As we are interested in the paramagnetic component associated with the unpaired electrons
we must make a correction to the observed susceptibility. The sample diamagnetism (cd) can be estimated using Pascal’s
constants based on the average diamagnetic susceptibilities for each of the elements/groups present. Selected Pascal’s constants
for common ions and elements are presented in Table 1. For example, the sample diamagnetism of [Ni(H2O)6]Cl2 is
[ 12.8 þ (6   13) þ (2   23.4)]  10 6 emu mol 1 ¼  1.376  10 4 emu mol 1. Typically the diamagnetism of the sample
and sample holder is small in relation to the sample paramagnetism. As a consequence these corrections need only be approx-
imate. Since both the diamagnetism of the sample holder and the sample diamagnetism are both estimated then the correction is
inexact and small adjustments to this value are permitted. Experimentally it is beneficial to use a larger quantity of sample (where
possible) such that the sample susceptibility is much larger than the susceptibility of the sample holder and the correction for the
sample holder becomes less significant. Thus the molar paramagnetic susceptibility, cp, can be determined from Eq. (7):
cp ¼ cm  cd (7)
Magnetic Measurements 155

Table 1 Pascal’s diamagnetic corrections for selected metal ions and elements (10 6 emu mol 1)

Cations
Liþ 1 Naþ 6.8 Kþ  14.9 Rbþ 22.5 Csþ 35
NH4 þ 40 Mg2 þ 5 Fe2 þ  12.8 Co2 þ 12.8 Ni2 þ 12.8
Cu2 þ 12.8 Zn2 þ 15 Cd2 þ  24 Hg2 þ 40 Pb2 þ 32
Anions
F 9.1 Cl 23.4 Br  34.6 I 50.6 HO 12.0
CN 13.0 NCS 31.0 CO3 2  29.5 NO3  18.9 SO4 2 40.1
ClO3  30.2 ClO4  32.0 BrO3  38.8 IO3  51.4 IO4  51.9
Ligands
H2O 13 NH3 18 Py  49 en 46.5 bipy 105
Phen 128 acac 52 gly  37 PPh3 167

In most research papers the molar paramagnetic susceptibility, cp, is normally just written as c and reported in emu mol 1 and
we will use this nomenclature from now on. With the susceptibility, c, finally in hand we can examine its temperature dependence.
Since c is proportional to the magnetization, M, then (from Fig. 1) we see that as the temperature is lowered, M increases and so the
paramagnetism increases such that c f 1/T. This was originally experimentally observed by Pierre Curie and is embodied in the
Curie law:
c ¼ C=T (8)
1
where C is the Curie constant (emu K mol ) and T is the temperature (in K). We saw that different values of S will give different
values of both c and Msat and it can be shown that the Curie constant C is related to S via Eq. (9). Here the value of Nb2/3k  0.125
(or 1 8). From Eq. (8) we can see that a plot of 1/c versus T should be linear with gradient ¼ 1/C (Fig. 4) and this is a common
=

approach to determining S. Table 2 illustrates the values of C expected for different values of S for a range of spin-only ions,
assuming g ¼ 2.0. The following ions (geometry given in parentheses) are typical examples of systems which approximate closely to
“spin-only”: Cr3 þ (Oh); HS Mn2 þ (Oh,Td); HS Fe3 þ (Oh,Td); Co2 þ (Td); Ni2 þ (Oh); Cu2 þ (Oh). Many other ions are expected to

Figure 4 (Left) Temperature dependence of 1/c (top) and cT (bottom) for S ¼ ½, S ¼ 1, and S ¼ 5 2 Curie spins; (right) temperature dependence of
=

1/c (top) and cT (bottom) for a Curie–Weiss S ¼ ½ spin system with q ¼ þ20 K and q ¼ 20 K. Note q is the intercept on the x-axis in the 1/c versus
T plot.
156 Magnetic Measurements

Table 2 Curie constants, C (emu K mol 1), for common values of S (taking Nb2/3k ¼ 1 8 and g ¼ 2.0)
=

S C S C S C S C

0 0.000 1 1.000 2 3.001 3 6.000


½ 0.375 3/2 1.875 5/2 4.375 7/2 7.875

deviate (to a greater or lesser extent) from spin-only behavior and this will be discussed in section “The Contribution of Orbital
Angular Momentum.” The majority of organic radicals behave as spin-only systems.
Ng 2 b2 SðS þ 1Þ
C¼ (9)
3k
Reality is that not all compounds obey pure Curie behavior. However, many follow the phenomenological Curie–Weiss law (Eq.
10). In this case a plot of 1/c versus T still affords the gradient as 1/C but there is now a nonzero intercept on the x-axis. The intercept
is given by the Weiss constant, q (measured in K).
c ¼ C=ðT  qÞ (10)
There are many reasons why compounds deviate from pure Curie behavior. Among them include “zero-field splitting” (see later),
“spin–orbit coupling” (section “ The Contribution of Orbital Angular Momentum”), and “magnetic exchange interactions” (section
“Exchange-Coupled Spins”) or a combination thereof. It is therefore necessary to interpret the meaning of the Weiss constant with
caution. To determine which of these processes is most likely to contribute to q, we need to examine the electronic configuration
of the metal ion in its coordination geometry to check for residual spin–orbit coupling and to check if there are potential magnetic
exchange pathways. The magnitude of the Weiss constant, |q|, gives an indication of how strong the effects which give rise to deviation
from spin-only are. Zero-field splitting (ZFS) for “spin-only” ions is typically < 10 cm 1 whereas spin–orbit-coupling effects are
typically  100 cm 1. Exchange coupling can range from very small through to 103 cm 1. Clearly there is significant overlap between
the energy ranges associated with each of these factors and it is not always trivial to discriminate between these different effects. A
careful consideration of the ions involved in conjunction with the coordination geometry and structure is typically required, usually
backed up with quantitative models to determine the origin of the deviation from the Curie law. We will look at each of these in turn
during this article. Deviations from the Curie–Weiss behavior are common at low temperature and the values of C and q are derived
from a fit to the high temperature data in which 1/c versus T follows linear behavior (Fig. 4, top right).
An alternative method to examine the temperature dependence of c is to plot cT versus T. For a pure Curie paramagnet, a straight
line is observed with the value of cT ¼ C ( Fig. 4, bottom left). Systems which can be described as Curie–Weiss paramagnets deviate
upon cooling depending upon the sign of q. For positive q, the value of cT increases upon cooling whereas for q < 0, cT decreases
upon cooling. In the high temperature limit cT  C (Fig. 4, bottom right).

2.08.2.4 Zero-Field Splitting, D


In section “ The Spin Quantum Number, S” we examined the effect of the magnetic field on the energy of the MS sublevels (Eq. 2).
However, this neglected the possibility of ZFS which arises from magnetic anisotropy. For example, the metal may not be located in
perfectly octahedral or tetrahedral surroundings, for example, for a trans-ML4O2 geometry the local symmetry reduces to D4h which
manifests itself as differences in the magnetism between z and the xy plane. Even for seemingly symmetric MX6 coordination, the
presence of an asymmetric occupancy of a set of degenerate orbitals will give rise to a Jahn–Teller distortion, similarly lowering the
symmetry, typically from Oh to D4h. Even for ions which do not undergo Jahn–Teller distortions crystallographic studies often
reveal they are located on a crystallographic “general position” and inherently the crystal symmetry is lower than cubic which leads
to some small crystal anisotropy. Even for “spin-only” ions, there are second-order spin–orbit coupling effects which depend upon
the ratio l/D which also contribute to magnetic anisotropy (see section “Second-order spin–orbit coupling in ‘spin-only ions’”).
The presence of the spin–orbit coupling, together with structural distortions in the immediate coordination sphere, lead to small
differences for the energies of the MS states before the application of the magnetic field, H. This splitting of the MS microstates in
zero field is innovatively termed “zero-field splitting” (ZFS) and the magnitude of the splitting is determined by the ZFS parameter
D, which defines the magnetic anisotropy, that is, the difference between the magnetic “easy axis” z and the corresponding “xy”
plane. We modify Eq. (2) to take into account ZFS as follows:
EðMS Þ ¼ gbHMS þ DM2S (11)

Note that D can be positive or negative and this affects which of the MS states lies lowest in zero field. Schematically the effect of
ZFS is shown in Fig. 2 (right) for S ¼ 1 (e.g., octahedral Ni2 þ and tetrahedral V3 þ) with D < 0. We should note that for S ¼ 1 when
D > 0, then the MS ¼ 0 state lies lowest in energy whereas when D < 0 then the MS ¼  1 states lie lowest in energy, affording what is
referred to in molecular magnetism as a bistable ground state. The magnetic properties of molecules with large S and negative D
have attracted considerable attention as single-molecule magnets (SMMs) and will be discussed in more detail in section
“Single-Molecule Magnets.”
Magnetic Measurements 157

It is clear that a perturbation of energies of the MS levels will lead to changes in the populations of the different MS states and
hence will affect the observed magnetization and magnetic susceptibility. Indeed the anisotropy which induces ZFS also leads to an
anisotropy in the magnetism such that the magnetic susceptibility along the “z” axis (c||) is different from that in the xy plane (ct).
Equations to describe the magnetic susceptibilities of parallel and perpendicular components in the presence of ZFS have been
determined for a range of values of S ( Table 3). Commonly we do not examine oriented single crystals but observe the behavior
of a polycrystalline sample. Here the average susceptibility, hci, can be written as:
h i.
hci ¼ cjj þ 2ct 3 (12)

The temperature dependence of cT versus T for parallel and perpendicular susceptibilities and the average susceptibility for an
S ¼ 1 system are presented in Fig. 5 for positive and negative D. Even for |D| ¼ 5 K (a relatively large ZFS), the deviation from pure
spin-only behavior is only evident in the low temperature region and Curie–Weiss behavior is observed with |q| < |D|. While spin-
only ions should follow Curie law and exhibit a temperature independent value of cT, the presence of ZFS can lead to Curie–Weiss
behavior but will produce only small values of |q| (with |q| < |D|) and deviation from Curie behavior is only evident at low
temperature.

2.08.2.5 Summary
l Measurement of M versus H (at low temperature) up to the maximum field available permits an estimation of Msat from which
we can estimate S (assume g ¼ 2.0) and then g.
l A plot of 1/c versus T typically affords a linear relationship in the high temperature region. The gradient is 1/C. Assuming g  2.0
for “spin-only” ions we can determine S. Once we know the value of S (integer or half-integer) we can then derive a more
accurate estimate of g.
l Ideal Curie behavior is characteristic of a thermally well-isolated electronic ground state.
l The value of the Weiss constant q gives a measure of the deviation from idealized spin-only behavior which may reflect a range of
additional phenomena such as ZFS, spin–orbit coupling, or magnetic exchange coupling.

2.08.3 The Contribution of Orbital Angular Momentum

To this point we have only considered the contribution of the spin angular momentum, S, to the magnetic moment which, in many
cases, gives a good estimate of the magnetic moments of many transition metal ions. However, there are other contributions to the
magnetic moment and we now focus on the orbital angular momentum, L. In addition we need to be aware that the spin (S) and
orbital (L) angular momenta are not independent but couple with each other to afford a total angular momentum (J).

2.08.3.1 The Origin of Orbital Angular Momentum


Orbital angular momentum can be considered to arise from the “motion of electrons between degenerate, symmetry equivalent
orbitals which do not already contain electrons with the same spin.” Only motion between certain pairs of orbitals contributes
to the orbital angular momentum, depicted in the “magic pentagon” for d-orbitals ( Fig. 6, left). For example, movement of an elec-
tron from dxy to dx2 y2 contributes to orbital angular momentum about z. Conversely dxy and dz2 are not linked and so motion of an
electron between these two orbitals does not contribute to orbital angular momentum.
In order to understand the magnetism of particular metal ions, we need to consider the relative energies of the terms which
define their electronic configurations, that is, the magnitude of the spin orbit coupling constant (l) in relation to the crystal field
splitting (D) in relation to the thermal energy kT. In section “ The Spin–Orbit Coupling Constant, l,” we examine the factors
affecting l and then examine the magnetic properties of lanthanides in section “Spin–Orbit Coupling for Lanthanides,” where
l [ D and then 3d metal ions, where D [ l (section “Spin–Orbit Coupling for Transition Metal Ions”).

2.08.3.2 The Spin–Orbit Coupling Constant, l


To date we have identified that a magnetic moment arises from the electron spin (S) and its orbital angular momentum (L).
However S and L are not independent and interact with each other through the spin–orbit coupling constant, l, to generate the total
angular momentum, J. The spin–orbit coupling constant l is related to the one-electron spin-orbit coupling constant z by:
l ¼ z=2S (13)
For less than a half-filled shell l is positive and for a more than half-filled shell l is negative. The change in sign of l can be
ascribed to the fact that the “spin up” electrons contribute to both S and L for less than half-filled shells, whereas for more than
half-filled shells the “spin up” electrons contribute to S and the “spin down” electrons contribute to L. In addition z scales as Z4
and is small (101 cm 1) for light atoms (C/N/O), larger for first row transition metals ( 102 cm 1), and very large for lanthanides
158
Magnetic Measurements
Table 3 Equations for parallel and perpendicular susceptibilities taking into account a large axial zero-field splitting (D [ gbH) for spin-only ions with S ¼ 1, 3 2, 2, and 5 2
= =

S c|| ct

2Ngjj2 b2 2 expðD=kT Þ 2Ngt 2 b2 2kT ½1  expðD=kT Þ


1 $ $ $
3kT ½1 þ 2 expðD=kT Þ 3kT D ½1 þ 2 expðD=kT Þ
15Ngjj 2 b2 ½1 þ 9 expð2D=kT Þ b ½1 þ ð3kT =4DÞ½1  expð2D=kT Þ
2 2
15Ngt
32
=
$ $
48kT ½1 þ expð2D=kT Þ 4kT ½1 þ expð2D=kT Þ

6Ngjj 2 b2 ½2 expðD=kT Þ þ 8 expð4D=4kT Þ b ½ð6kT =DÞ½1  expðD=kT Þ þ ð4kT =3DÞ½expðD=kT Þ  expð4D=kT Þ
2 2
Ngt
2 $ $
3kT ½1 þ 2 expðD=kT Þ þ 2 expð4D=kT Þ 3kT ½1 þ 2 expðD=kT Þ þ 2 expð4D=kT Þ
35Ngjj2 b2 ½1 þ 9 expð2D=kT Þ þ 25expð6D=kT Þ b ½9 þ ð8kT =DÞ½1  expð2D=kT Þ þ ð5kT =2DÞ½expð2D=kT Þ  expð6D=kT Þ
2 2
35Ngt
52
=
$ $
48kT ½1 þ expð2D=kT Þ þ expð6D=kT Þ 48 ½1 þ expð2D=kT Þ þ expð6D=kT Þ
Magnetic Measurements 159

Figure 5 Temperature dependence of cT versus T for parallel and perpendicular susceptibilities and average susceptibility, hci, as well as Curie–
Weiss plot for an S ¼ 1 system with g|| ¼ gt ¼ 2.0; (left) D/k ¼ 5 K; (right) D/k ¼ þ5 K.

Figure 6 (Left) The “magic pentagon” illustrating which orbital combinations contribute to spin–orbit coupling; (right) orbital occupancy and term
symbols for (A) d 3, (B) low spin d 7, and (C) high spin d 7 configurations under Oh symmetry.

( 103 cm 1). Table 4 highlights selected values of z for 3d and 4f metal ions, illustrating the systematic increase as a function of Z.
Notably S increases and decreases across both the d- and f-block leading to a less evident dependence of l with Z.
For transition metals, we know that there is some degree of metal–ligand covalency in these complexes. If the unpaired electron
were 100% metal-based the free ion value of l would be appropriate. However, more ligand character (assuming C/N/O-based
donor atoms) will lead to a reduction in l which will depend on the degree of covalency. A phenomenological orbital reduction
factor (k) is often applied where k is typically  0.7–0.8 reflecting predominantly metal-based electrons. For example Co2 þ has
z ¼ 515 cm 1 such that for a high spin Co2 þ ðS ¼ 3 2Þ l   230 cm 1 since it has a more than half-filled shell (l < 0). Taking
=

into account orbital reduction, values of kl (often just cited as l) are  180 cm 1 for Co2 þ ions. For lanthanides the radially con-
tracted nature of the 4f-orbitals leads to negligible covalency and orbital reduction factors k are near unity.

2.08.3.3 Spin–Orbit Coupling for Lanthanides


For lanthanides l  103 cm 1 whereas D  102 cm 1, so we initially neglect the effect of the crystal field on the f-orbitals and antic-
ipate that the magnetic properties of the lanthanides reflect those of the free ion, that is, in which the seven f-orbitals are degenerate.
For the lanthanides the Ln3 þ oxidation state is prevalent with electronic configuration 6s04f n. Here we use Hund’s rules of

Table 4 One-electron spin–orbit coupling constants (z) (cm 1) for 3d metals ions M2 þ and M3 þ and lanthanide ions Ln3 þ

Ti V Cr Mn Fe Co Ni Cu

M2 þ 120 170 230 (300) 400 515 630 830


M3 þ 155 210 275 355 (460) (580) (715) (875)
Ce Pr Nd Pm Sm Eu Gd
Ln3 þ 643 800 900 1200 1415
Tb Dy Ho Er Tm Yb Lu
Ln3 þ 1620 1820 2080 2360 2800 2940
160 Magnetic Measurements

Figure 7 (Left) Filling of the f -orbitals to maximize S and L and determination of J for an f 9 configuration (Dy3 þ); (right) resultant energy level
diagram for the Dy3 þ ion arising from spin orbit coupling.

maximum multiplicity to determine S, L, and J for the free ion by adding the n electrons initially “spin up” to the schematic ( Fig. 7),
starting from the left then, for configurations with more than 7e, the additional “spin down” electrons are then added again
commencing from the left. We then derive S ¼ Sms and L ¼ Sml. For example, Dy3 þ ( f 9) has ðS ¼ 5 2Þ and L ¼ 5. Since S and L
=

are both vectors, the total angular momentum, J, can take values between |L þ S| and |L  S| in integer steps, depending whether
the spin and orbital components reinforce or oppose one another. For Dy3 þ these correspond to 15 2, 13 2, 11 2, 9 2, 7 2, and= = = = =
5 2. We use the term symbol nomenclature 2Sþ 1G to uniquely address the values of S, L, and J where G is a capital letter representing
=
J
the orbital angular momentum; S (L ¼ 0), P (L ¼ 1), D (L ¼ 2), F (L ¼ 3), G (L ¼ 4), H (L ¼ 5), I (L ¼ 6). Thus for Dy3 þ the term
symbols representing S, L, and the various values of J are 6H5/2, 6H7/2, 6H9/2, 6H11/2, 6H13/2 and 6H15/2. Hund’s rules tell us that
for a less than half-filled shell the lowest value of J lies lowest in energy, whereas for a more than half-filled shell the largest J
term lies lowest in energy. For Dy3 þ the 6H15/2 lies lowest in energy. In addition the remaining excited 6HJ states follow the Landé
interval rule in which adjacent terms with values of J and J þ 1 are separated by lJ(J þ 1), illustrated in Fig. 7. In the case of Dy3 þ the
ground term is 6H15/2 and the first excited term (6H13/2) will be separated by 15l/2. Since l ¼ 1820 cm 1 (Table 4) then the first
excited state (6H13/2) is 13,640 cm 1 higher in energy and will not be thermally populated even at room temperature
(kT  200 cm 1 at 300 K). To a good approximation the magnetism of the majority of lanthanides therefore reflects the spin ground
state, 2Sþ 1GJ.
We find that the magnetism of the lanthanide ions parallels that of the spin-only behavior (Eq. 9), except that we now utilize J
rather than S and the value of gJ may differ substantially from 2.0 (Eqs. 14 and 15).
NgJ2 b2 J ð J þ 1Þ
c¼ (14)
3kT

3 SðS þ 1Þ  LðL þ 1Þ
gJ ¼ þ (15)
2 2Jð J þ 1Þ
Note that if there is no orbital angular momentum (L ¼ 0), then J ¼ S and Eq. (15) gratifyingly reduces to g ¼ 2 and Eq. (14) is
then identical to the “spin-only” equation (Eq. 9). The term symbols for Ln3 þ ions and their corresponding values for gJ and C are
presented in Table 5 alongside typical observed values of the Curie constant, C, for simple lanthanide complexes. It is worth noting
that for the early lanthanides (4f n with n < 7), the spin and orbital contributions to the magnetic moment work against each other
such that the resultant magnetic moments (reflected in the Curie constant C) are small. For later lanthanides, the spin and orbital
contributions reinforce each other and give rise to much larger moments. With the exception of Eu3 þ and Sm3 þ, the majority of
lanthanide complexes exhibit room temperature cT values comparable with the Curie constant and show deviation below
 50 K associated with small Weiss constants (which originate from small crystal field splitting see later). For Eu3 þ the ground
term is 7F0 and although there are six unpaired electrons the spin and orbital moments in this case are equal and opposite
(C ¼ 0 emu K mol 1). However, the first excited term (7F1) lies just 1l above the 7F0 ground state. For Eu3 þ l is 236 cm 1, compa-
rable with kT at room temperature. Thus the magnetism of Eu3 þ is more complex arising from contributions from both the ground
state and several low-lying excited states. Such thermal population/depopulation effects of excited states lead to a marked deviation
from Curie behavior.
So far we have only considered the effect of spin–orbit coupling on the magnetic properties of the lanthanides. Typically at low
temperatures (< 50 K), the effect of the small crystal field (D) becomes evident. This crystal field splits the degeneracy of the MJ states
of the 2Sþ 1GJ ground term and we will use Dy3 þ by way of example. The ground state 6H15/2 term comprises a series of 16
Magnetic Measurements 161

Table 5 Term symbols and magnetic properties of lanthanide ions

Ion Configuration Term symbol gJ C (calc) emu K mol  1 C (obs) emu K mol  1

Ce3 þ 4f 1 2
F5/2 6/7 0.80 0.66–0.78
Pr3 þ 4f 2 3
H4 4/5 1.60 1.45–1.62
Nd3 þ 4f 3 4
I9/2 8/11 1.64 1.53–1.62
Pm3 þ 4f 4 5
I4 3/5 0.90
Sm3 þ 4f 5 6
H5/2 2/7 0.09 0.25–0.36
Eu3 þ 4f 6 7
F0 5 0.00 1.36–1.53
Gd3 þ 4f 7 8
S7/2 2 7.88 7.80–8.00
Tb3 þ 4f 8 7
F6 3/2 11.82 11.28–12.01
Dy3 þ 4f 9 6
H15/2 4/3 14.17 13.52–14.05
Ho3 þ 4f 10 5
I8 5/4 14.07 13.52–14.32
Er3 þ 4f 11 4
I15/2 6/5 11.48 11.04–11.52
Tm3 þ 4f 12 3
H6 7/6 7.15 6.30–7.22
Yb3 þ 4f 13 2
F7/2 8/7 2.57 2.31–3.00

(i.e., 2J þ 1) MJ states ranging from 15 2 to þ15 2 in integer steps, that is, 15 2, 13 2, 11 2, 9 2, 7 2, 5 2, 3 2, and 1 2.
= = = = = = = = = =

The presence of a crystal field splits the degeneracy of the MJ states into a series of Kramers doublets (KDs), each corresponding to
MJ. The relative energies of these MJ states depend upon the symmetry of the coordination environment and the magnitude of the
crystal field. The splittings of the 6H15/2 term for two different complexes, [Dy(Pc)2] and [Dy(W5O18)2]9 , are presented in Fig. 8.
Since the bonding in lanthanide complexes is largely electrostatic (nondirectional), both coordination numbers and geometries are
largely dictated by the ligand geometry and sterics, with the larger size of the lanthanides often supporting higher coordination
numbers of 8 or 9. Such Ln3 þ ions typically follow the Curie–Weiss behavior with Curie constants close to those predicted (Table 5)
and with Weiss constants reflecting successive depopulation of the MJ states and reflected in a decrease in cT upon cooling. As with
spin-only ions with bistable or KD ground states, such complexes can act as single ion magnets and are discussed in more detail in
section “Single-Molecule Magnets.”

2.08.3.4 Spin–Orbit Coupling for Transition Metal Ions


Let us now return to the effect of spin–orbit coupling on transition metal ions. Our approach to understanding the magnetism of
these ions is different to the lanthanide approach described in section “ Spin–Orbit Coupling for Lanthanides” for the following
reason: For lanthanides the contracted f-orbitals afforded small crystal field splittings (D) in relation to l, so the effect of spin–orbit
coupling was considered first, followed by the effect of D upon the 2Sþ 1GJ term, which was only evident at low temperature. For

Figure 8 (Left) The effect of a crystal field on the 6H15/2 ground state of two sandwich complexes of [Dy(Pc)2] and [Dy(W5O18)2]9 ; (right) struc-
tures of [Dy(Pc)2] (top) and [Dy(W5O18)2]9  (bottom).
162 Magnetic Measurements

transition metals, the crystal field splitting D is large in relation to l, so we examine crystal field effects first and then consider the
effect of l.
We will commence by considering the effect of the crystal field splitting on the electronic structure of a transition metal ion,
examining three different configurations; d3, low spin d7 and high spin d7 under octahedral (Oh) geometry. The octahedral crystal
field splits the d-orbital degeneracy to afford a t2g set and an eg set ( Fig. 1). For d3 under Oh symmetry, there are three coparallel
electrons in the t2g set (Fig. 6A) and there is only one way of drawing this configuration, that is, it is orbitally singly degenerate
and thus there is no orbital angular momentum hence, as previously discussed, it is described as “spin-only” (section “The Magne-
tism of Spin-Only Ions”). For low spin d7 there is now one-electron in the eg set which can either be in the dx2 y2 or dz2 orbital, that
is, it is orbitally doubly degenerate. However the magic pentagon (Fig. 6) reveals that there is no orbital angular momentum asso-
ciated with the movement of the electron between dx2 y2 and dz2 orbitals, so this too can be described as a “spin-only” ion.
Conversely, there are three energetically equivalent ways of representing the high spin d7 configuration with a “hole” (absence
of a “spin down” electron) in either the dxy, dyz, or dxz orbitals. This ion is therefore orbitally triply degenerate and we see from
the magic pentagon that motion of the e between these orbitals can give rise to orbital angular momentum.
In order to describe the spin and orbital contributions to the magnetism, we use the term symbol notation 2Sþ 1G to describe the
electronic structure in an analogous fashion to the lanthanides. However, to distinguish between gas phase “free ions” where the
d-orbitals (or f-orbitals for lanthanides) are degenerate and those in a complex where some orbital degeneracy is lost, we use
a second series of term symbols. For orbitally singly degenerate terms such as the octahedral d3 configuration ( Fig. 6A) we use
the symbol A (for some lower symmetry systems B is sometimes also used). Orbitally doubly degenerate ground states are described
as E terms, for example, low spin d7 (Fig. 6B). Finally, for orbitally triply degenerate states we use the label T, exemplified by the high
spin d7 ion (Fig. 6C).

2.08.3.4.1 Second-Order Spin–Orbit Coupling In “Spin-Only Ions”


In the last section we found that the d3 ion under octahedral coordination offered a 4A ground term in which spin–orbit coupling is
quenched. However this configuration offers a first excited state t22ge1g which is a 4T term and is approximately D higher in energy.
This excited state configuration exhibits spin–orbit coupling whose strength depends upon l. Since the ground state 4A and the
excited state 4T have the same spin multiplicity, they are allowed to mix. This mixing is inversely dependent upon the energy
gap between them. As a consequence there is a perturbation to the g-value (and hence susceptibility) of the ground term due to
mixing in the excited state which is proportional to kl/D:
g ¼ 2½1  ðakl=DÞ (16)
As we saw in section “ The Spin–Orbit Coupling Constant, l,” l is positive for less than half-filled shells such that spin–orbit
coupling affords g < 2.0 (and a consequent reduction in c), whereas for more than half-filled shells such second-order spin–orbit
coupling affords g > 2.0.

2.08.3.4.2 First-Order Spin–Orbit Coupling


For orbitally triply degenerate states such as the 4T term ( Fig. 6C) we note that we have “an asymmetric occupancy of a set of degen-
erate orbitals” and such octahedral complexes are known to undergo a small Jahn–Teller distortion (d) which is substantially
smaller than the crystal field splitting (D). In terms of perturbation theory it is more complex to determine whether to initially
consider the effect of d or l on the magnetic properties as they are often of comparable magnitude. As Kahn points out “efforts
to reproduce the details of the magnetic anisotropy for octahedral high spin cobalt(II) compounds have more or less failed.”
Two limiting approaches can easily be considered (d [ l and l [ d), but reality is typically somewhere between them.
On the one hand, as d tends to 0, we consider the 4T term to be affected by spin–orbit coupling. The T term can be considered
equivalent to Leff ¼ 1 and spin–orbit coupling within the 4T term gives rise to J ¼ 5 2, 3 2 and 1 2 ( Fig. 9, left). For octahedral d7 the
= = =

Figure 9 (Left) The effect of spin orbit coupling on the 4T term (l [ d); (center) the effect of a Jahn–Teller distortion on the 4T ground term and
subsequent effect of zero-field splitting on the ground 4A term of octahedral high spin Co2 þ (d [ l); (right) 1/c versus T data for a simple octahedral
Co2 þ complex, Co(pybtda)2Cl2, reflecting two regions corresponding to S ¼ 3 2 (T > 20 K) and S eff ¼ 1 2 (T < 20 K).
= =
Magnetic Measurements 163

J ¼ 1 2 ground state lies lowest. A subsequent perturbation due to a small distortion (d) reduces the symmetry from Oh to D4h and
=

introduces a ZFS (D). For the J ¼ 3 2 state this lifts the degeneracy of the MJ ¼ 3 2 and MJ ¼ 1 2 levels while the J ¼ 5 2 state will
= = = =

split into three KDs (MJ ¼ 5 2, 3 2, and 1 2). At low temperature only, the J ¼ 1 2 level is populated which behaves like an effec-
= = = =

tive doublet (Seff ¼ 1 2) with a large g-value (g ¼ 4.33). The separation of the MJ levels is small (a function of l and D) in relation to
=

thermal energy kT and so significant temperature dependence of cT is expected.


On the other hand, when d [ l then we consider the effect of the Jahn–Teller distortion next. This distortion (compression or
elongation) lowers the symmetry from Oh to D4h affording an orbitally singly degenerate 4A ground state (with an excited 4E term)
in which the orbital angular momentum has been formally quenched and we expect “spin-only” behavior with S ¼ 3 2. However the =

structural distortion also leads to an anisotropy in the magnetic moment and the 4A ground term can split into two KDs correspond-
ing to MS ¼ ½ and MS ¼ 3 2 separated by 2D due to ZFS ( Fig. 9, center). When d [ kT then we need to only consider the two KDs
=

arising from the 4A term. While the orbital contribution is effectively quenched by the Jahn–Teller distortion, the low-lying 4E term
permits a significant second-order spin–orbit coupling effect (the energy gap is now d rather than D) which is manifested in both D
and g. For a large axial elongation (dxz and dyz lower than dxy), then we reach the limit g|| ¼ 2.0 and gt ¼ 4.0 (hgi  3.3) and S ¼ 3 2. =

Such octahedral Co2 þ ions often approximate to Curie–Weiss behavior with S ¼ 3 2 but g > 2 in the high temperature region and
=

a large q value reflecting substantial magnetic anisotropy. At low temperature, ZFS of the 4A term becomes evident affording a KD
(MS ¼ ½) which can often be modelled as an effective Seff ¼ ½ ion with large g-value. It is common for such ions to fit the 1/c versus
T data to two distinct regions reflecting the high and low temperature limits, respectively (Fig. 9, right).

2.08.3.5 Summary
l The coupling of unquenched orbital angular momentum can lead to more complex magnetic behavior, depending upon the
magnitude of the spin–orbit coupling constant.
l For the lanthanides the large value of l leads to well-isolated 2Sþ 1GJ ground terms in the majority of cases in which essentially
Curie-like behavior (q  0 K) is followed from which the ground term can be confirmed. At low temperature, the effect of the
small crystal field splitting of the MJ states leads to depopulation effects leading to a deviation from Curie behavior.
l For transition metals in many cases the large crystal field fully quenches the orbital angular momentum, affording spin-only
ions. Where residual orbital angular momentum exists the similar magnitudes of l and the inherent Jahn–Teller distortion
(d) and ZFS (D) associated with these configurations makes interpretation more complex. A fit to Curie–Weiss behavior in the
high temperature region can be made assuming a spin-only moment with the orbital contribution leading to (i) a significant
deviation in g from 2.0 and (ii) a large Weiss constant, q. In the low temperature region, the ground state is typically a KD which
follows Curie–Weiss behavior with Seff ¼ ½, a large g-value and a small Weiss constant.

2.08.4 Exchange-Coupled Spins

So far we have considered systems where there is no communication between magnetic ions. In this section, we look at mechanisms
in which spins interact with each other, referred to as magnetic exchange and typically given the symbol J. When two spins interact,
they can either align coparallel (a ferromagnetic interaction) or antiparallel (an antiferromagnetic interaction). In fact when two
spins interact both possibilities exist, the sign of J determines which is most stable and the magnitude of J indicates the strength
of the interaction. If J > 0, then the ferromagnetic configuration is stabilized ( Fig. 10). If J < 0, then the relative energies of the
spin states are reversed and the antiferromagnetic configuration is more stable. Within a localized bonding approach it can be
shown that in the case of weak orbital overlap for spin-only systems, the magnetic exchange integral J approximates to:
2J ¼ 2K  4Sb (17)
The first term K is the potential exchange integral and is a reflection of Hund’s rules which indicate that electrons (in the same
region of space) prefer to align coparallel, whereas the second term reflects the extent of orbital overlap (S) between the wavefunc-
tions of the two interacting electrons and b is the resonance integral. We can quickly see that if the overlap integral S is zero

Figure 10 Crystal structure (left) and orbital interaction (center) for Cu2(OAc)4$2H2O; (right) antiferromagnetic (top right) and ferromagnetic (bottom
right) spin arrangements arising from coupling of two S ¼ ½ ions (J > 0).
164 Magnetic Measurements

(a nonbonding interaction between orthogonal orbitals), then 2J ¼ 2K and a ferromagnetic, coparallel spin alignment is preferred.
However the term 4Sb rapidly increases as S becomes nonzero such that in the absence of (near) orthogonality, J rapidly becomes
negative and antiferromagnetic coupling ensues.

2.08.4.1 Direct Exchange


The direct exchange mechanism relies on direct spatial overlap of orbitals containing unpaired electrons. Given the radially con-
tracted nature of the 3d and 4f orbitals, this is not typically a common mechanism for metal complexes, except where M.M sepa-
rations are less than 3.0 Å, such as Cu2(OAc)4$2H2O (Cu.Cu ¼ 2.6 Å). In this example, each square pyramidal Cu2 þ (d9) has an
unpaired electron in the dx2 y2 orbital which interact with each other through a weak face-to-face overlap ( Fig. 10) (a d-type
bonding interaction with two nodal planes) such that S > 0 and so the 4Sb term is significant and antiferromagnetic coupling
(J < 0) ensues. When J < 0 the ST ¼ 0 configuration is the ground state (Fig. 10).
Although direct exchange is considered rare in metal complexes, planar p-radicals are often considered to interact via a direct
exchange process (referred to as the McConnell I mechanism in organic parlance) where there is face-to-face interactions of the radi-
cals at distances less than or comparable to the sum of the van der Waals radii. Depending upon the relative molecular displace-
ments of the p systems, the orbital overlap varies such that the overlap integral can be zero (favoring a ferromagnetic
interaction) or nonzero, favoring an antiferromagnetic interaction. In the extreme the overlap integral is so large that the two elec-
trons can be considered to form a multi-center, 2e bond affording a diamagnetic ground state.

2.08.4.2 Superexchange
The structures of many metal salts reveal large distances between metal ions which clearly inhibit efficient orbital overlap of d-
orbitals, yet strong magnetic communication is often observed. This is explained through some degree of metal–ligand covalency,
that is, delocalization of unpaired electron density from the metal to the ligand. This then affords some orbital overlap in the
vicinity of the bridging ligands. For metals in octahedral coordination environments, electrons from the t2g set can be considered
of p-character since they delocalize into the ligand orbitals through a p-bonding interaction whereas the eg electrons are of s char-
acter. Generally efficient communication occurs for one or two atom bridges such as O2  or CN and weaker magnetic exchange is
observed for larger bridging units. The interpretation of the exchange coupling can be rationalized once more through a consider-
ation of the extent of orbital overlap (Eq. 17) but were initially expressed through a set of rules developed by Goodenough and
Kanamori.

2.08.4.2.1 Goodenough–Kanamori Rules


The Goodenough–Kanamori rules are a set of empirical rules which explain the observed magnetic behavior of many metal-oxo
bridged systems when the coupling is via a 90 or 180 M–O–M bridge:
(a) When the magnetic orbitals are arranged so as to afford a reasonable overlap integral (S > 0) the exchange is antiferromagnetic.
(b) When the magnetic orbitals are arranged in such a way as to have no overlap integral (S ¼ 0) then the exchange will be
ferromagnetic.
(c) Superexchange through the s framework tends to be more efficient than through the p framework.
For example, in Fig. 11 (left) a 180 Ni2 þ–O–Ni2 þ interaction between magnetic orbitals of dz2 character is expected to be anti-
ferromagnetic (Rule A), whereas a 90 Ni2 þ–O–Ni2 þ interaction between magnetic orbitals of dz2 character (Fig. 11, center) is ex-
pected to be ferromagnetic (Rule B) since both metals delocalize to the same oxygen atom but the p-orbitals on the O atom which
receive electron density from nickel are now different and mutually orthogonal (S ¼ 0). As a final example, a 180 Ni2 þ–O–Fe3 þ
interaction offers both antiferromagnetic contributions when the eg electrons of the Ni2 þ and the high spin Fe3 þ ion delocalize into
the same O orbital (S > 0) but are orthogonal when we consider the eg electrons of the Ni2 þ and the t2g electrons of the Fe3 þ ion.

Figure 11 Superexchange processes through: (left) 180 Ni–O–Ni interactions; (center) 90 Ni–O–Fe interactions; (right) 180 Ni–O–Fe(III)
interactions.
Magnetic Measurements 165

Figure 12 Double exchange mechanisms in mixed valence metal complexes afford a ferromagnetic alignment of spins.

Here we have competition between antiferro- and ferromagnetic interactions so the decision is less clear but Rule C suggests that the
antiferromagnetic interaction is preferred. These rules hold equally well for other bridging ligands such as CN and a large family of
Prussian blue salts Mx[M0 (CN)6] are known whose magnetic properties reflect these rules.

2.08.4.2.2 Correlation Between Bridge Angle And Exchange Interaction


In the last section, we saw that the exchange coupling between Ni2 þ ions could be antiferromagnetic (q ¼ 180 ) or ferromagnetic
(q ¼ 90 ), so clearly the exchange coupling is sensitive to the bridge angle. When the orbitals are rigorously orthogonal (S ¼ 0), then
a ferromagnetic interaction is expected (Eq. 17). However when S > 0 the antiferromagnetic term rapidly increases and dominates.
Work by Hatfield et al. found a linear correlation between the exchange coupling J and q for a series of CuII dimers, with ferromag-
netic interactions dominant from 90–97.5 and antiferromagnetic terms for angles greater than 97.5 . Similar trends have been
observed for other d-block metals.

2.08.4.3 Double Exchange


The double exchange mechanism is limited to mixed-oxidation state compounds in which electron delocalization occurs. This is
particularly relevant to extended lattice systems but also applicable to polynuclear metal complexes and organic systems comprising
both localized and conduction electrons. We will examine this in terms of a mixed valence V2 þ/V3 þ system in which the metals are
bridged via a m-5,6-dimethylbenzimidazolate ligand ( Fig. 12). Electron-transfer between metals within the V3 þ/V2 þ system affords
an equivalent V2 þ/V3 þ configuration. However to minimize electron–electron repulsion, the transferred electron should have the
same spin orientation as the other electrons on both metals, that is, the double exchange mechanism drives a ferromagnetic inter-
action. This mechanism is common in many mixed valence transition metal oxides.

2.08.4.4 Dipolar Exchange


Dipolar exchange is typically the weakest of the magnetic exchange interactions and approximates to:
m m
U ¼ 1 32 (18)
4pr
The r3 dependence of the dipole–dipole interaction means that for a dipole–dipole separation of 1 Å the interaction between
two ions with m ¼ 1mB the exchange interaction is of the order of 10 23 J or 1 K. With metal/metal distances in complexes typically
of the order of 3–6 Å or more then dipole–dipole interactions are only significant at low temperature, even for lanthanide ions
where m may be larger due to orbital contributions. Since the magnetic moments m1 and m2 have both magnitude and direction,
then the dipole–dipole interactions between spins can be ferro- or antiferromagnetic depending upon relative orientation.

2.08.4.5 Spin Frustration


In the last section we have seen that there are a range of mechanisms leading to ferromagnetic or antiferromagnetic exchange. Spin
frustration is a phenomenon which arises in certain spin topologies in the presence of antiferromagnetic coupling. These topologies
are based around geometries containing polygons with an odd number of spins, particularly triangular faces, and can incorporate
three-dimensional systems such as tetrahedra, as well as a range of spin topologies constructed from triangular or tetrahedral pla-
quettes. Fig. 13 (left) illustrates the frustration arising from antiferromagnetic coupling on a triangular geometry, in which the third
spin cannot simultaneously be antiferromagnetically coupled to the first two spins. The spin arrangement adopted in such frustrated
systems compromises the preferred antiparallel arrangement of spins between nearest neighbors. For triangles the spins are typically
all arranged 120 with respect to each other but may lead to magnetically degenerate ground states (Fig. 13, center and right). For
antiferromagnetically coupled spins on a tetrahedral plaquette, the spins are arranged at 109.5 . The presence of frustration is man-
ifested in a magnetically degenerate ground state which can lead to suppression of long-range order.
166 Magnetic Measurements

Figure 13 Spin frustration based on antiferromagnetic exchange on a triangular plaquette (left); degenerate spin ground states for the triangular
topology (center and right). In both cases the spins are oriented 120 with respect to each other.

2.08.4.6 Magnetic Models for Exchange-Coupled Systems


In sections “ Direct Exchange,” “Superexchange,” “Double Exchange,” and “Dipolar Exchange,” we examined the different mech-
anism which permit spins on separate metal centers to interact with each other. In this section we will examine qualitatively the
effect of ferromagnetic and antiferromagnetic interactions between spins. In order to do so we consider two limiting cases:
(i) the thermal energy kT [ exchange interaction, |J| and (ii) kT |J|.

2.08.4.6.1 Limiting High Temperature Limit


Here the thermal energy is high (kT [ |J|) and the magnetic behavior in the high temperature limit tends to the behavior of a set of
independent, non-interacting spins. For a compound comprising a series of spin-only ions, the magnetism is therefore defined by
Eq. (19), that is, the sum of the cT values for each individual spin.
Ng 2 b2 SSðS þ 1Þ
cT ¼ (19)
3k

2.08.4.6.2 Limiting Low Temperature Limit


In the low temperature regime we consider the effect of exchange coupling between spins to give a new total spin ground state, ST,
arising from the cooperative interaction between them. For two spins (SA and SB) interacting ferromagnetically then ST ¼ SA þ SB.
Conversely, if they interact antiferromagnetically then ST ¼ |SA  SB|. Fig. 14 illustrates the structures of two dodecanuclear
complexes; Ni12(chp)12(OAc)12(THF)6(H2O)6 [chp ¼ 6-chloropyridonate anion] and Mn12O12(OAc)16(H2O)4 (“Ni12” and
“Mn12”) and a cyclic decanuclear ferric complex [Fe(OMe)2(ClCH2COO)]10 (“Fe10”). In the Ni12 complex ferromagnetic interac-
tions between neighboring Ni2 þ ions (S ¼ 1) afford an ST ¼ 12 ground state, whereas the Fe10 complex (d5, S ¼ 5 2) exhibits anti- =

ferromagnetic interactions between nearest neighbors, generating an ST ¼ 0 ground state. The mixed valence Mn complex is more
complex; with the 8 Mn3 þ ions (d4, S ¼ 2) antiferromagnetically coupled to the central 4 Mn4 þ ions (d3, S ¼ 3 2) affording a net
=

ST ¼ 10 ground state.
The resultant susceptibility for the system in the low temperature limit is given by:
Ng 2 b2 ST ðST þ 1Þ
cT ¼ (20)
3k
Fig. 15 (left) illustrates the temperature dependence of the dodecanuclear nickel complex. At high temperature (kT [ |J|) the
magnetism reflects 12 independent Ni2 þ ions. Taking g ¼ 2 and Nb2/3k ¼ 1 8, Eq. (19) gives cT ¼ (1 8)(2)2[12  (1)(2)] ¼
= =

12 emu K mol 1. If the coupling between spins is ferromagnetic then at low temperature an ST ¼ 12 ground state is expected for which

Figure 14 Structures and spin arrangements for the dodecanuclear complexes Ni12(chp)12(OAc)12(THF)6(H2O)6 (left), Mn12O12(OAc)16(H2O)4
(center), and the decanuclear complex [Fe(OMe)2(ClCH2COO)]10 (right).
Magnetic Measurements 167

Figure 15 (Left) Temperature dependence of cT for the dodecanuclear complex, Ni12, and the decanuclear Fe10 complex; (right) low temperature M
versus H plots for Ni12 and Fe10.

cT ¼ 78 emu K mol 1 (Eq. 20, g ¼ 2). The cT versus T data for this Ni12 complex (Fig. 15) reflect a slightly higher value than that pre-
dicted in the high temperature limit, which arises from g > 2 and the presence of ferromagnetic interactions. The expected value of cT
for an ST ¼ 12 ground state is not quite reached at 1.8 K but M versus H studies at 0.15 K reveal saturation at 25.5 mB, consistent with
ST ¼ 12 and g ¼ 2.13 (Fig. 15). For the Fe10 system, the theoretical high temperature limit is expected to be 43.75 emu K mol 1,
decreasing to 0 emu K mol 1 for an ST ¼ 0 ground state. Experimentally cT is a little less than the high temperature limit consistent
with antiferromagnetic interactions but rapidly approaches zero at low temperature. M versus H data for Fe10 reflects a very low magne-
tization up to 6 T, consistent with an ST ¼ 0 ground state.
In section “ Magnetic Susceptibility: Curie and Curie–Weiss Behavior,” we saw that a negative value of the Weiss constant q
could arise from antiferromagnetic exchange (J < 0), whereas a positive q may reflect J > 0. In fact it can be shown that there is a rela-
tionship between q and J. According to mean field theory:
2zJSðS þ 1Þ
q¼ (21)
3k
where z ¼ number of nearest neighbor contacts. For example, in the Ni12 complex each Ni2 þ ion has two nearest neighbors (z ¼ 2).
Eq. (21) therefore provides an opportunity to assess the magnitude of the exchange coupling, J. A Curie–Weiss plot for Ni12 revealed
q ¼ þ 7 K, affording a first estimate of J/k as þ 2.6 K between neighboring Ni2 þ ions. These can be compared with exchange
couplings of þ 8 and þ 1.5 K for the two different exchange pathways in Ni12 derived from inelastic neutron scattering experiments
(data adapted to a H ^ 1^
^ ¼  2JS S2 Hamiltonian, see later).

2.08.4.6.3 Quantitative Models for Exchange-Coupled Systems


For discrete clusters, it is often possible to undertake quantitative analysis of the exchange coupling, J, by developing a magnetic
model which evaluates all the possible spin states ST and their relative energies. The magnetic behavior of the system is then deter-
mined by applying a Boltzmann distribution to the resultant set of spin states.
The process commences by examining the structure and identifying likely exchange pathways for magnetic communication
between the spins. For coordination complexes efficient magnetic exchange typically occurs through one or two atom bridges
between metal centers. We then write the spin Hamiltonian (H) ^ for the system which just defines which spins (S) ^ interact with
each other. A series of examples are given in Fig. 16 for a dimer, as well as triangular and linear arrangements of three spins. In
this text we use the H ^ ¼  2JS ^2 formulism in which a negative J implies antiferromagnetic interactions. It is important to note
^ 1S
the spin Hamiltonian used in papers as there is more than one definition with some researchers using H ^ ¼ JS ^2 which gives
^1S
similar behavior but the reported value of J is twice as large, whereas others use H^ ¼ þ JS ^2 which additionally means that antifer-
^1S
romagnetic interactions are reflected in positive values of J. As a consequence, particular care should be taken when comparing the
magnitude of J from different papers to take into account the convention used for the spin Hamiltonian.

Figure 16 Spin topologies for dinuclear and high symmetry trinuclear complexes.
168 Magnetic Measurements

For a dimer of S ¼ ½ ions, magnetic communication permits these spins to interact to generate two new spin states; a singlet,
ST ¼ 0 ([Y) and a triplet state, ST ¼ 1 ([[). In this case, the two configurations are separated by an energy 2J. (We will not cover
the derivation of all possible spin states and their relative energies here, but discussions can be found in the “Further Reading”
section.) If J < 0 then the singlet state lies lowest in energy whereas if J > 0 then the triplet is the ground state ( Fig. 10). We now
apply a Boltzmann distribution to these two configurations to evaluate the magnetism of the system:
Ng2 b2 SST ðST þ 1Þð2ST þ 1ÞexpðEðST Þ=kT Þ
c¼ (22)
3kT Sð2ST þ 1ÞexpðEðST Þ=kT Þ
In Eq. (22), the magnetic contribution from each state is highlighted in bold and is merely the Curie contribution arising from
that spin state. The rest of the expression represents the Boltzmann population of each state. For the copper dimer, there are just two
spin states (ST ¼ 0 and ST ¼ 1) and the susceptibility reduces to the Bleaney–Bowers expression:
Ng 2 b2 2
c¼ (23)
kT 3 þ expð2J=kT Þ
A fit to the temperature dependence of c permits an estimate of J. Since each individual spin, S, is a vector then the possible
spin states generated by coupling spins can be large. For example, a Cr3 þ (S ¼ 3 2) dimer has a series of spin states ranging
=

from |S1 þ S2|.|S1  S2| in integer steps, that is, 3, 2, 1, 0. The number of spin states and the equation for the susceptibility therefore
increases rapidly with the size of the cluster and the magnitude of S according to (2S þ 1)n. For a copper dimer (S ¼ ½ and n ¼ 2),
there are just four microstates (a singlet and a triplet), but for the Ni12 system there are 312 ¼ 531,441 states. Although the ST ¼ 12
configuration contains 25 microstates (2ST þ 1), it is clear that there are a great many other configurations which arise because of the
different ways in which individual spins can combine. For small clusters and small values of S, the derivation of such equations is
possible but often manually burdensome for larger clusters and several software packages are available for modeling clusters (MAG-
PACK, PHI). With extensive research into the magnetic properties of clusters, it is now less common that expressions for c need to be
derived ab initio for simple clusters and readers are encouraged to search the literature for appropriate expressions for the magnetic
susceptibility.
For discrete clusters, a finite number of spin states are formed and exact solutions can be derived. However for polymeric arrange-
ments of spins, exact solutions cannot be computed. Such polymeric arrangements include systems such as regular chains, alter-
nating chains, spin ladders, and (frustrated) zig-zag chains ( Fig. 17). For systems with two or more exchange couplings, the
spin ground state can often be sensitive to the ratio of J/J0 which is often set as a variable (a). Here approximate methods such
as a high temperature series expansion are typically used to estimate the behavior per spin. Caution should be used when modeling
such data as the equations may be specific to whether J is positive or negative or the magnitude of a. Reviews on such extended
systems, including expressions for their magnetic susceptibility, can be found in the bibliography. Notably the approximate
methods normally lead to some deviation at low temperature (T/J < x where x depends upon the expression given). Such low-
dimensional systems provide useful models upon which to test many examples of theoretical physics such as single chain magnets
and Haldane gapped phases.

2.08.4.7 Summary
l Paramagnetic centers can communicate with each other through several mechanisms. Superexchange via one or more bridging
ligands is the most prevalent for polynuclear metal complexes.
l Antiferromagnetic exchange leads to a preferred local antiparallel spin alignment (J < 0) and a reduction in cT on cooling
whereas ferromagnetic exchange affords coparallel spin alignment (J > 0) and an increase in cT upon cooling.
l For discrete clusters magnetic exchange leads to a set of molecular spin states, ST, formed from interaction of the individual
spins, Si.

Figure 17 Common one-dimensional and pseudo-one-dimensional polymeric arrangements of exchange-coupled spins; (A) regular chain,
(B) alternating chain, (C) spin ladder, (D) zig-zag chain.
Magnetic Measurements 169

l Preliminary magnetic studies on exchange-coupled clusters comprise an examination of cT versus T and M versus H at low
temperature.
l At high temperature (kT [ J), then the magnetic susceptibility is consistent with an ensemble of non-interacting spins.
l In the low temperature limit (kT J) only the spin ground state, ST, is populated.
l Expressions for the magnetic susceptibility of small-medium sized clusters permit quantitative estimates of J.
l Approximate methods can be employed to model the behavior of polymeric structures where an infinite number of configu-
rations exist.

2.08.5 Single-Molecule Magnets

In the last section we saw that exchange-coupled clusters could lead to large spin ground states, for example, ST ¼ 12 for “Ni12” and
ST ¼ 10 for “Mn12”. In section “ Zero-Field Splitting, D,” the effect of ZFS was described and led to a loss of degeneracy of the MS
sublevels of the ground state S. The same behavior can be expected for these larger clusters at low temperature when just their spin
ground state is populated. Thus for Mn12 with ST ¼ 10, the effect of ZFS is to lift the degeneracy of the  10,  9,  8 . 0 MS
sublevels. The ZFS parameter D   0.5 cm 1 such that the MS ¼  10 states of Mn12 lie lowest in energy (DM2S ¼ 50 cm1 ) in rela-
tion to the MS ¼ 0 state (DM2S ¼ 0 cm1 ). If cooled to low temperature in zero field, then the populations of the MS ¼  10 levels are
equally populated and there is no net sample magnetization. However if the system is cooled in an applied field, then the popu-
lations of the  10 levels are no longer equally populated and a magnetization is observed in the sample. More importantly if the
field is switched off, then magnetic relaxation (such that the population of the  10 states becomes equal) requires molecules to
2
progress over the energy  U. For integer values of ST this energy barrier corresponds to DST whereas for half-integer values
 2 barrier,
it corresponds to D ST  4 . For systems with large D and/or ST then this can provide a substantial energy barrier for thermal
1 =

relaxation and the sample retains its magnetization for a long period which can extend to months or longer. The energy of the
MS states of Mn12 is shown in Fig. 18A along with the zero-field energy barrier, DS2T. The presence of this energy barrier to relaxation
means that when the field is switched off, the magnetization of the molecule does not immediately reduce to zero and a hysteresis
loop is observed (Fig. 18C) which is reminiscent of the behavior of bulk ferro- and ferrimagnets and such compounds exhibiting
slow magnetic relaxation are termed “single-molecule magnets” (SMMs).
In contrast to the polynuclear metal complexes of the d-block, where a bistable ground state ( MS) arises from ZFS of the spin
ground state, ST, a range of single ion magnets based on lanthanide complexes are known in which a bistable ground state arises
from crystal field splitting of the 2Sþ 1GJ ground term ( Fig. 8). Later lanthanide ions (>f 7) have proved particularly attractive since
these give rise to large magnetic moments (section “Spin–Orbit Coupling for Lanthanides”). In addition Ln3 þ ions with odd
numbers of f-electrons afford noninteger J so all MJ states are bistable “Kramers doublets.” Ions with integer values of J will addi-
tionally exhibit an MJ ¼ 0 term. If this MJ ¼ 0 term is the ground state or is a low-lying excited state, then this may provide an efficient
mechanism for magnetic relaxation. In this context Dy3 þ (J ¼ 15 2) has attracted considerable attention as a building block for
=

single-molecule magnets.

2.08.5.1 Relaxation Processes


There are two common relaxation processes available to SMMs which can be considered to occur through thermal relaxation and/or
a quantum tunneling mechanism (QTM). At high temperature when kT is comparable to the energy barrier, U, then thermal relax-
ation is the dominant relaxation process. However at low temperature (kT U), then QTM is the major relaxation process. Both
relaxation processes follow first-order kinetics where the relaxation time is given by:
s ¼ s0 expðDE=kT Þ (24)

Figure 18 Energy level diagrams for S T ¼ 10 with D < 0 in (A) the absence and (B) the presence of an applied magnetic field; (C) the SMM Mn12
reveals a stepped hysteresis loop due to quantum tunneling.
170 Magnetic Measurements

However for thermal relaxation DE ¼ U and therefore s shows a strong temperature dependence whereas for QTM, relaxation is
expected to be temperature independent.

2.08.5.1.1 Thermal Relaxation


In zero field the energy barrier to thermal relaxation, DE is equal to U and relaxation occurs via the highest lying MS level of the ST
manifold. Clearly as the temperature increases the relaxation rate increases. In addition the energy barrier decreases in the presence
of an applied field since the energies of the MS states vary according to Eq. (11), that is, the highest lying state (MS ¼ 0 for integer
spin) is unaffected by H whereas the ground state shows a strong field dependence (Fig. 18B).

2.08.5.1.2 Quantum Tunneling Mechanism


At low temperature, thermal relaxation is very slow since very few molecules have sufficient thermal energy to overcome the thermal
energy barrier, U. In this temperature regime quantum tunneling is important. Tunneling is a transverse relaxation process between
energetically degenerate MS states which occurs through rather than over the classical thermal energy barrier. If the tunneling is via
the ground state MS levels, then DE ¼ 0 (Eq. 24) and relaxation is essentially temperature independent. Thermally assisted
quantum tunneling can also occur via excited degenerate MS states and referred to as Orbach and Raman processes. The rate of
relaxation depends on the transverse magnetic anisotropy term, E.
Since quantum tunneling requires energetically degenerate MS levels, then the effect of an applied field is to reduce quantum
tunneling since the field splits the degeneracy of the MS levels. However at certain applied fields different þMS and MS states
become degenerate and quantum tunneling can occur ( Fig. 18C) which is reflected in rapid decreases in sample magnetization,
producing “steps” in the experimental hysteresis loops (marked with arrows in Fig. 18C for the 2.8 K data). Clearly the rate at which
the measurements are undertaken (in relation to the QTM rate) will affect the shape of the hysteresis loop.

2.08.5.1.3 Determining Relaxation Rates


In selected examples, the relaxation rate is sufficiently slow that the relaxation time can be determined from a plot of sample magne-
tization versus time ( Fig. 19A). At elevated temperatures the rate of decay of sample magnetization via thermal relaxation increases.
However in many cases decay rates are too fast to be observed by dc magnetic susceptibility studies. In such cases zero-field ac
susceptibility measurements can be used.
In the ac experiment, a small oscillating ac field (often 2–5 Oe) of frequency u is applied to the sample and the magnetization is
measured “in-phase” and “out-of-phase” with respect to the ac field. Conceptually, at high temperature the paramagnet is able to
follow the applied field perfectly and is said to be “in-phase” with the oscillating applied field. Under these conditions the in-phase
susceptibility, c0 , is equal to the normal sample susceptibility, c. However at low temperature when sample relaxation is slow the
spin is unable to perfectly follow the alternating field. The observation of an out-of-phase signal, c00 , heralds the onset of slow
magnetic relaxation. For many SMMs a maximum in c00 is observed upon cooling which is found to be frequency dependent. At
high ac frequencies, the spin cannot relax sufficiently quickly to keep in phase with the ac field whereas lower ac frequencies provide
the spin more time to follow the field and out-of-phase components only become evident at low temperature. At the peak
maximum in a c00 versus T plot the relaxation time s ¼ 1/2pu. A plot of ln(s) versus 1/T for multiple frequencies permits the energy
barrier U to be determined. In Fig. 19C, the gradient ¼ U ¼ 38 K and ln(so) can be extrapolated from the intercept on the y-axis when
1/T ¼ 0.
For some systems, a well-defined maximum in c00 may not be apparent, due to the presence of QTM processes, which diminish
the effective energy barrier to relaxation, U. In such cases, the emerging maxima may be resolved by the application of a small static
dc magnetic field. The application of a static field serves to quench quantum tunneling by lifting the degeneracy of the KDs between
which tunneling occurs, while shifting the energies of the microstates as little as possible in order to minimize the difference

Figure 19 (A) The relaxation time s can be determined from an M versus t plot for slow relaxation processes (note log scale); (B) for rapid relaxa-
tion processes the onset of an out-of-phase signal in the ac data reflects the onset of slow relaxation; (C) a plot of ln(s) versus 1/T permits the
energy barrier DE to be determined.
Magnetic Measurements 171

Figure 20 (A) A classical Debye function for a single relaxation process illustrating the effect of the parameters, cS and cT as well as s and a on
the shape of the curve; (B) effect of the relative rates of relaxation r ¼ s2/s1 for a Cole–Cole plot involving two relaxation mechanisms (r ¼ 1 equates
to a single relaxation process).

between the effective energy barrier Ueff calculated in the presence of a static field and the true energy barrier U which occurs in zero
field. In some cases where c00 remains poorly resolved, then data can be derived from a “Cole–Cole” plot (in zero or an applied dc
field). Here c00 is plotted as a function of c0 for different frequencies, and u for a fixed temperature (where u is related to the ac
frequency f by u ¼ 2pf). For a single relaxation process the in-phase and out-of-phase susceptibilities follow:
 
ðc  c S Þ sinh½ð1  aÞlnðusc Þ
c0 ¼ cS þ T 1 (25a)
2 cosh½ð1  aÞlnðusc Þ þ cos½1 2 ð1  aÞp =

 
ðcT  cS Þ sin½1 2 ð1  aÞp
=
c00 ¼ 1 (25b)
2 cosh½ð1  aÞlnðusc Þ þ cos½1 2 ð1  aÞp
=

The experimental c0 versus c00 plot can be modeled using four parameters (cT, cS, a, and sc) where cS is the adiabatic suscep-
tibility, sc is the temperature-dependent relaxation time, and a is a measure of the dispersivity of relaxation times (0 < a < 1).
Notably a and sc change the width and position of the maximum in c00 . The temperature dependence of sc can then be used to
derive the energy barrier ( Fig. 19C). In some temperature regimes a single relaxation pathway dominates, for example, thermal
relaxation at high temperature or quantum tunneling at low temperature. However in some cases multiple relaxation processes
may be active. This can be reflected in either two maxima in the Cole–Cole plot or more frequently by a shoulder or kink in the
data. Here a fit to the data now corresponds to the sum of two components (Eq. 25a and 25b) with two sets of parameters a1
and a2, s1 and s2, etc. Fig. 20A illustrates the parameters which define the relaxation process for a single process. Fig. 20B shows
the effect of two relaxation processes on the Cole–Cole plot.

2.08.5.2 Summary
l Transition metal clusters exhibiting a negative ZFS (D < 0) and a non-zero spin ground state may exhibit slow magnetic
relaxation (single-molecule magnet behavior).
l Lanthanide complexes can also exhibit single-molecule magnet behavior which arises out of the crystal field splitting of the MJ
states of the 2Sþ 1GJ ground term. Notably this splitting is sensitive to the coordination geometry.
l Relaxation in single-molecule magnets can occur through thermal relaxation and/or quantum mechanical tunneling.
l The energy barrier to relaxation, U, and tunneling rate, so, can be determined from ac susceptibility measurements through the
maxima in c00 versus T or through modeling of Cole–Cole plots.

2.08.6 Long-Range Magnetic Order

Before commencing a discussion of the different types of long-range magnetic order, we need to briefly discuss what is meant by
long-range order and the requirements necessary to achieve long-range order. The transition from the paramagnetic (short-range
ordered) phase to the long-range ordered phase is analogous to a chemical transition such as a liquid–solid transition where the
molecules go from a disordered phase to an ordered phase. Such order–disorder transitions are marked by a discontinuity in
the physical properties and are typically associated with a l-type transition in heat capacity measurements. In order to determine
whether long-range magnetic order occurs we need to consider the dimensionality of the exchange coupling and the spin anisotropy
( Fig. 21). A simple way to consider the behavior is to start with a Heisenberg spin where there is no spin anisotropy (D ¼ 0). Long-
range three-dimensional order will therefore only occur when the exchange interactions propagate in all three dimensions.
172 Magnetic Measurements

Figure 21 Effect of spin and lattice dimensionality on magnetic ordering.

Long-range ordering is therefore dictated by the weakest of the exchange interactions. In this case the spins interact cooperatively
throughout the solid (the correlation length, z ¼ N, vide infra) but the spins have no preferred spin orientation and therefore there is
no resistance to spin reorientation (zero coercive field). If one degree of lattice dimensionality is removed then it can be compen-
sated by a decrease in spin dimensionality (Heisenberg to XY, D > 0) leading to a special transition known as Kosterlitz–Thouless
transition. Similarly if a further degree of lattice dimensionality is removed then this can be compensated by a further constraint in
the spin dimensionality (Ising spin, D < 0) affording a single chain magnet. If the system has fewer restraints on the spin orientation,
then long-range order cannot occur.
In the paramagnetic phase, at high temperature each spin is considered independent and the correlation length is zero (z ¼ 0).
Upon cooling the presence of exchange coupling permits each spin to interact cooperatively with its neighbors (z ¼ 1) and at still
lower temperatures next-nearest neighbors (z ¼ 2). Upon further cooling z continues to increase down to the magnetic ordering
temperature at which there is a divergence and z / N and the spins adopt an ordered arrangement throughout the solid. This
preferred spin orientation automatically infers some anisotropy and the preferred direction of alignment is known as the magnetic
“easy axis” with perpendicular spin orientations known as the magnetic “intermediate” and “hard” axes. Despite the presence of
long-range order, this does not mean that the magnetic structure is static and spins can be considered to precess about their easy
axis in a cooperative fashion known as a spin wave ( Fig. 22). As the temperature increases, the inclination of the spin with respect
to the easy axis (q) increases such that at T ¼ 0 K the spins are perfectly parallel with the easy axis but become orthogonal as T tends
to the ordering temperature. The magnetization parallel to the easy axis follows:
Measy ¼ gS cosðqÞ (26)

In this article, we will focus on the four most common types of magnetically ordered systems observed in molecule-based mate-
rials; ferromagnets, antiferromagnets, ferrimagnets, and canted antiferromagnets ( Fig. 23). In particular, we will focus on their
magnetic signatures and relevant experiments to help distinguish between different ordered phases. It is noteworthy that the
type of magnetic order is defined as the magnetic structure in zero field rather than field-induced transitions such as metamagnetic
behavior. As discussed earlier, the onset of long-range magnetic order corresponds to a disorder/order transition and the behavior
above and below the critical temperature (TC) are distinct, comparable with a discussion of the properties of water in the solid and
liquid states.

Figure 22 (Left) Cooperative precession of spins about their easy axis affords spin waves, viewed from the side (top) and parallel to the easy axis
(bottom); (right) temperature dependence of the sublattice magnetization below the critical temperature TC.
Magnetic Measurements 173

Figure 23 Common types of long-range magnetic order (in a ferrimagnet the “up” spins and “down” spins are of different magnitude).

2.08.6.1 Ferromagnetism
Below a critical temperature known as the Curie temperature for a ferromagnet, TC, the correlation length z ¼ N with all the spins
aligned coparallel. However in order to maximize the entropy of the system, the ferromagnet breaks up into a series of macroscopic
nanometer-sized domains of coparallel aligned spins. Domain wall boundaries occur at the interface between these domains of
coparallel alignment and comprise a series of many spins each displaced slightly ( 1 ) from its neighbor. The energy stored in
the distortion of the spins away from their preferred orientation depends on (a) the magnetic exchange field, He, which is related
to the energy of interaction between neighboring spins and (b) the anisotropy field, Ha, which reflects the energy to reorient the spin
away from its easy axis. At the microscopic level, He relates to the energy difference between coparallel ([[) and antiparallel ([Y)
spin alignment and is equal to S(S þ 1))J (2J for S ¼ ½) between two neighboring spins. For perfect Heisenberg spins D ¼ 0 and
Ha ¼ 0 but may be large for strongly anisotropic systems such as Ising spins (D < 0). Typical domain wall boundaries are 150–
200 spins long ( Fig. 24). Considerable energy may be stored in these domain wall boundaries and give rise to magnetic hysteresis.
It should be noted that the origin and manifestation of hysteresis in a bulk ferromagnet is different from a single-molecule magnet
(section “Single-Molecule Magnets”). Here the hysteresis comes from an energy barrier to spin reorientation arising from build-up
of energy at domain wall boundaries, whereas in an SMM it arises from an energy barrier to spin reversal between doubly degenerate
(KD) ground state configurations. In addition, hysteresis loops for bulk ferromagnets do not exhibit the “steps” originating from
quantum tunneling which are commonly observed for SMMs.
For a ferromagnetic material cooled to below TC in the absence of a magnetic field (“zero-field cooled” (ZFC)), the domains are
randomly oriented and there is no net magnetization. An applied field acts to orientate the domains with the applied field (working
to move the domain walls) and the magnetization increases with increasing field up to a saturation value, Msat, which occurs when
all domains align with the field. At this point Msat ¼ gS. When the field is switched off, the magnetic anisotropy keeps the spins
aligned but the thermal energy kT allows some mean displacement of the spins away from the easy axis leading to a remnant

Figure 24 (Top) Spin reorientation in the domain wall boundary (B) between domains A and C; (bottom left) hysteresis loop for a “hard” magnet and
(bottom right) a “soft” magnet.
174 Magnetic Measurements

magnetization (Mrem) which is less than Msat ( Fig. 24). When the field is applied in the opposite direction, work needs to be done to
reorient the domains in the opposite sense. The field required to return the sample magnetization to zero is known as the coercive
field (Hco) and reflects the magnetocrystalline anisotropy, Ha, but is also affected by the crystal shape (where the sample aspect ratio
also contributes to the sample anisotropy). The area bound within the hysteresis loop reflects the energy stored in the domain wall
boundaries. The shape of the hysteresis loop is temperature dependent which is a reflection of the thermal energy kT in the system.
As T increases toward TC, the thermal energy kT works in conjunction with the applied field to assist spin reorientation at domain
walls. Therefore as T increases the coercive field is smaller. Similarly as T increases the inclination of the spin from the easy axis
increases so the remnant magnetization gS$cos q (Fig. 22) decreases, that is, as T increases both Hco and Mrem decrease.

2.08.6.1.1 Magnetic Characterization of a Ferromagnet


In the paramagnetic regime above the critical temperature (known as the Curie temperature for a ferromagnet), ferromagnetic mate-
rials typically exhibit Curie–Weiss behavior with q > 0 such that cT increases upon cooling and a plot of 1/c versus T has a positive
intercept on the x-axis.
Below TC the magnetic susceptibility is more complex and depends on the applied field and a field-dependent magnetic response
is often considered diagnostic of the onset of long-range order. In a field-cooled (FC) experiment, we consider (i) the temperature
dependence of the magnetization ( Fig. 22) and (ii) the field dependence of the magnetization (Fig. 24) and their effects on the
magnetic susceptibility, c  M/H. Fig. 22 clearly reveals a rapid increase in magnetization on cooling below TC which saturates
out as T / 0 K. As a consequence c (¼ M/H) will increase rapidly on cooling below TC before saturating at low temperature
(Fig. 25). Fig. 24 clearly indicates M is approximately constant (ranging from Mrem to Msat) in the magnetically ordered regime
such that M/H is large for small fields and small when H is large. Thus the observed c is largest in a small applied field (Fig. 25).The
temperature dependence of cT initially shows an increase upon cooling but as c tends toward a saturation value then the value of cT
decreases upon cooling (Fig. 25).
In the ZFC experiment, the sample is cooled to base temperature in zero field and a (small) field applied. The system is in the
magnetically ordered regime but domains are randomly oriented so the magnetization is initially zero and increases nonlinearly up
to Msat as H is increased ( Fig. 24). Initially as the temperature increases, the small applied field plus thermal energy kT permits some
realignment of domains leading to an increase in M upon initial warming. As T further increases toward TC the inherent decrease in
spontaneous magnetization with temperature dominates and c decreases. The convergence of FC and ZFC data is typically used to
determine the ordering temperature TC.
Additional ac susceptibility studies should reveal the emergence of an out-of-phase signal at TC. Unlike SMMs where relaxation is
usually rapid leading to a frequency dependence of c00 ferromagnets should exhibit a frequency-independent behavior while heat
capacity studies should reveal a l-type transition associated with the phase transition from magnetically ordered to magnetically
disordered.

2.08.6.2 Ferrimagnetism
In a ferrimagnet, there are typically at least two different metals with magnetic moments on two (or more) magnetic sublattices
which are antiferromagnetically coupled to each other. The magnetization on sublattice A is gASA and that on sublattice B is
gBSB. Provided gASA s gBSB then a net magnetization is observed below the Curie temperature, TC:
Msat ¼ jgA SA  gB SB j (27)

Figure 25 (Left) c versus T for a ferromagnet with field-cooled data measured in 100, 500, and 1000 Oe and zero-field cooled data on warming in
100 Oe field; (right) corresponding cT data.
Magnetic Measurements 175

2.08.6.2.1 Magnetic Characterization of a Ferrimagnet


The magnetism of a ferrimagnetic material above TC is very different from a ferromagnetic material and this coupled with Msat in the
ordered regime discriminate it from a ferromagnet and antiferromagnet. For a ferromagnet 1/c versus T follows Curie–Weiss
behavior (Eq. 10) with q > 0, whereas the temperature dependence of 1/c is described by a Néel hyperbola:
1 Tq z
¼  (28)
c C T  q0
This temperature dependence is reflected in a local minimum in cT on cooling due to local antiferromagnetic interactions before
rising at lower temperatures as correlations between next-nearest neighbors become important.
Below TC a ferrimagnet typically exhibits similar hysteresis loops to a ferromagnet except that the saturation value is given by Eq.
(27). In this low temperature regime, the field and temperature dependence of a ferrimagnet is typically analogous to a ferromagnet
(Figs. 22 and 24).
As with ferromagnetic materials, additional ac susceptibility and heat capacity studies should confirm the phase transition from
magnetically ordered to magnetically disordered.

2.08.6.3 Antiferromagnetism
An antiferromagnet comprises a set of two sublattices with spins of equal magnitude which are aligned perfectly antiparallel, afford-
ing no net magnetic moment. The critical temperature at which ordering occurs is known as the Néel temperature, TN. In order to
understand the magnetism of an antiferromagnet below TN we will consider the magnetism parallel and perpendicular to the easy
axis and evaluate the average susceptibility based on parallel and perpendicular components (vide infra).

2.08.6.3.1 Magnetic Characterization of an Antiferromagnet


Above TN antiferromagnets exhibit Curie–Weiss behavior with q < 0 consistent with antiferromagnetic interactions. A maximum in
c is typically observed at TN. Caution in the interpretation should be taken here as short-range interactions within clusters also give
rise to a maximum in c, but the two can be distinguished by the form of the maximum and their behavior below the maximum;
long-range ordered systems typically exhibiting a “sharper” maximum before decreasing to  2/3 cmax ( Fig. 26), whereas short-
range interactions appear as broad, rounded maxima which tend to decrease toward c ¼ 0 at low temperature (for ST ¼ 0 ground
states).
In order to understand the magnetism below TN we note that there is an easy axis of spin alignment and therefore the magnetism
is inherently anisotropic. In the absence of an applied field the magnetization is zero but when a field is applied the magnetic
response is orientation-dependent and the magnetization data parallel and perpendicular to the easy axis differ. For single-
crystal studies, the divergence between c|| and ct heralds the onset of long-range antiferromagnetic order ( Fig. 26).

2.08.6.3.1.1 Magnetization Parallel to the Easy Axis


The “spin up” electrons align with the applied field but the strong exchange field, He, orients the electrons on the second sublattice
against the field. The net magnetization in zero field parallel to the easy axis is zero at T ¼ 0 K. Between T ¼ 0 and T ¼ TN the spins
precess around the easy axis at an angle q which increases as T increases. In the presence of an applied field, the magnetic field aligns
the “up spin” parallel to the easy axis. However the torque afforded by the applied field on the “spin down” electrons leads to a small
net magnetic moment which slowly increases with increasing field ( Fig. 27). As the temperature increases the precession increases
and the magnetization induced in the “spin down” sublattice increases such that the magnetization at 0 K is zero but increases upon
warming converging with ct at TN.

Figure 26 (Left) The difference between the maximum in c in short-range and long-range ordered systems; (right) the dependence of the magnetic
susceptibility parallel and perpendicular to the easy axis for an antiferromagnet with divergence at TN.
176 Magnetic Measurements

Figure 27 (Left) The effect of a small applied field on the moments associated with the A and B magnetic sublattices; (center) M versus H plot for
a field-induced spin-flop transition; (right) M versus H plot for a field-induced spin-flip (metamagnetic) transition (dotted lines represent the expected
M vs. H plot for a paramagnet).

2.08.6.3.1.2 Magnetization Perpendicular to the Easy Axis


The magnetic field is now perpendicular to the easy axis and works to move the spins away from coparallel alignment generating
a moment. The susceptibility perpendicular to the easy axis is therefore the same as it was just above TN and is temperature inde-
pendent ( Fig. 26). The susceptibility of a powder/polycrystalline sample is the weighted average of these values:
 
hci ¼ cjj þ 2ct 3 (29)

2.08.6.3.2 Field-Induced Transitions in Antiferromagnets


Two common field-induced transitions occur for antiferromagnets; a “spin-flop” transition (90 reorientation of both magnetic sub-
lattices) and a “spin-flip” (metamagnetic) transition (180 reorientation of one sublattice).
In a spin-flop phase the increase in the magnetic field stabilizes the “spin up” A sublattice but destabilizes the “spin down” B-
sublattice. Eventually the applied field is sufficiently large that at the spin-flop field, Hsf, both sublattices reorient themselves perpen-
dicular to the applied field, affording destabilization of the A sublattice whereas the B-sublattice is stabilized but both are still
aligned antiparallel. At fields above Hsf both sublattices can now begin to align with the applied field so are both further stabilized
( Fig. 27). The spin-flop field depends upon both the exchange field and the anisotropy field:
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
Hsf ¼ He Ha  H2a (30)

Spin-flop transitions are most likely to occur for systems with a low spin anisotropy (small Ha) which facilitates reorientation
into the hard plane. The spin-flop phase is reflected in a subtle change in gradient in the M versus H plot and often more easily
identified by plotting dM/dH versus H and typically occurs is in applied fields of several Tesla.
In a spin-flip or metamagnetic transition, the B-sublattice undergoes a 180 reorientation, leading to both sublattices aligning
coparallel with the applied field, akin to a ferromagnet. Such metamagnetic transitions are manifested in an abrupt field-
induced enhancement in susceptibility to a “ferromagnet-like” response with rapid saturation of the magnetization ( Fig. 27). Meta-
magnetic transitions are common in systems where there is strong magnetic anisotropy disfavoring reorientation into the hard
plane. In addition a weak antiferromagnetic exchange term is desirable. Such systems often comprise low-dimensional ferromag-
netically coupled systems (such as a layer-like motif) with weaker antiferromagnetic terms between layers. At a critical spin-flip field
the weaker antiferromagnetic exchange interactions are overcome and a 180 flip of the B-sublattice occurs. In such cases it is note-
worthy that (i) in the paramagnetic regime such systems may have a positive Weiss constant (q > 0) reflecting the presence of domi-
nant ferromagnetic interactions and (ii) it is important to measure the magnetic response in the low temperature region in the
vicinity of the transition temperature in a small applied field and/or undertake detailed field-dependent studies since c will be
very different with H < Hmeta and H > Hmeta.
The absence of a spontaneous magnetic moment in zero field means that antiferromagnets will not exhibit out-of-phase signals
in the ac susceptibility but will still exhibit a l-type peak in the heat capacity associated with the magnetic phase transition.

2.08.6.4 Canted Antiferromagnetism


A canted antiferromagnet comprises two antiferromagnetically coupled sublattices of equal magnitude. However the sublattices are
not perfectly antiparallel such that their two magnetic sublattices do not perfectly cancel affording a “spin canted” structure. Such
canted antiferromagnets are also referred to as “weak ferromagnets” due to the presence of a small spontaneous moment. Confus-
ingly the labeling of the ordering temperature depends upon the nomenclature employed; the descriptor “canted antiferromagnet”
Magnetic Measurements 177

typically uses TN whereas a “weak ferromagnet” uses TC. The origin of the spin canting is the presence of an additional antisymmetric
exchange term known as the Dzyaloshinski–Moriya interaction d. While the isotropic exchange term J is a dot-product
^ ¼  2JS
(H ^1 . S
^2), the Dzyaloshinski–Moriya term is a cross-product (dS ^1  S
^2). While the energy of the dot-product is minimized
when the spins are antiparallel (when J < 0), the cross-product is minimized when the spins are orthogonal. If the two sublattices are
related via an inversion center, then d ¼ 0 and there is no canting. On the other hand in a noncentrosymmetric structure d s 0 and
canting may be evident. The canting angle depends upon the ratio (d/J) such that sin(q) ¼ |d/J|. Typically q is only a few degrees and
often less than 1 .

2.08.6.4.1 Magnetic Characterization of a Canted Antiferromagnet


The behavior of a canted antiferromagnet above TN exhibits the Curie–Weiss behavior with the Weiss constant q < 0 in a 1/c versus T
plot, consistent with local antiferromagnetic interactions and cT decreases upon cooling. Below TN several scenarios can arise; in
some cases spin canting occurs simultaneously with magnetic ordering, but examples occur where the system initially orders as
a simple antiferromagnet but at low temperature (kT < d) then a further transition occurs to a canted state. We will focus on systems
in which ordering and canting occur simultaneously. In zero field, the magnetization parallel to the easy axis is zero but perpen-
dicular to the easy axis there is a small spontaneous moment given by:
MS ¼ gS$sinðqÞ (31)
This spontaneous magnetic moment is much smaller than the spontaneous magnetic moment in a ferromagnet (gS) and occurs
perpendicular to the easy axis of alignment. The overall magnetization comprises this small spontaneous moment plus a field-
dependent term which increases steadily with applied field in an analogous fashion to a regular antiferromagnet:
M ¼ MS þ cH (32)
The spontaneous moment MS can therefore be determined by plotting M versus H and extrapolating the linear component to
H ¼ 0 affording a nonzero intercept on the y-axis ( Fig. 28). The spontaneous moment is temperature dependent and follows the
saturation of the sublattice magnetization (Fig. 22). To evaluate the canting angle q, MS is determined at a series of temperatures
below TN and a plot of MS versus T generated and extrapolated to T ¼ 0 K. The canting angle can then be estimated from sin(q) ¼ MS/
gS. The presence of a spontaneous moment should afford the onset of a (frequency-independent) out-of-phase signal in the ac
susceptibility.

2.08.6.5 Summary
We end this section by introducing a flow chart ( Fig. 29) to assist identification of different types of magnetic response based on
preliminary studies of the behavior in the paramagnetic regime and below the critical temperature. While this should help assign the
type of long-range order in many cases, the types of magnetic order discussed here are not comprehensive and it is important to
check all observations are self-consistent. These observations clearly reveal that a preliminary set of experiments should include
M versus T in a small applied field coupled with M versus H at low temperature. These data can often successfully pinpoint the likely
type of magnetic order from which a set of more detailed experiments can be undertaken.

2.08.7 Magnetic Measurements

In the previous sections of this article, we have focused on an explanation of theory and appropriate expressions for analyzing
magnetic data. Here we briefly review practical aspects of magnetic measurements.

Figure 28 (Left) The temperature dependence of c for a weak ferromagnet; (right) M versus H plot for a canted antiferromagnet at different
temperatures.
178 Magnetic Measurements

Figure 29 Flow diagram to identify likely type of long-range order. FM, ferromagnet; FI, ferrimagnet; AFM, antiferromagnet; CAFM, canted antiferro-
magnet; MM, antiferromagnet with metamagnetic transition.

2.08.7.1 Sample Preparation


It is very important to avoid contamination of the sample with paramagnetic or ferromagnetic particles, for example, from a metal
spatula, when handling and preparing the sample. The use of plastic, teflon, or glass utensils is encouraged. The presence of small
quantities of a ferromagnetic particle can lead to anomalous behavior such as hysteresis or the presence of a spontaneous magne-
tization in zero field at low temperature. This is particularly the case for weakly paramagnetic samples such as organic radicals where
a ferromagnetic impurity can be mis-assigned as an intrinsic property of the system. The sample weight depends greatly on the
amount of sample available but, as we have seen ( Fig. 3), the magnetization induced in the sample varies significantly with S,
so for large molecules with S ¼ ½ then 50–75 mg might normally be considered appropriate whereas for molecules containing
many paramagnetic centers or with large S then  10 mg would be more than adequate. In this context many physicists examining
ceramics such as mixed metal oxides typically utilize 2–3 mg (careful explanation of the dilution effect may be required if collab-
orating with solid state physicists for magnetic measurements!). The sample is then packed in a gelatin (or Teflon) capsule. The
sample is typically immobilized with some cotton or a small quantity of grease so that it does not move during measurements.
If the sample is mounted in air, then incorporation of a pin-hole into the capsule is desirable so that oxygen trapped in the capsule
can be evacuated during the purge cycle as the sample is placed in the SQUID. The magnetism of O2 in the low temperature region is
complex but typically reveals a paramagnetic/antiferromagnetic phase transition around 43 K and a smaller transition around 25 K
reflecting the phase transition between a and b phases of O2 both of which are antiferromagnetic. These transitions are manifested
in abrupt changes in magnetization and data in the 20–50 K region should be carefully examined to ensure the response is intrinsic
to the sample. Oxygen contamination is normally minimized by purging the sample in the airlock. For badly contaminated
samples, it is possible to purge the airlock with the airlock valve open (seek the assistance of an experienced user!). If the signal
disappears it is O2 contamination. If the O2 signal rapidly reappears, then there is likely a leak in the magnetometer.

2.08.7.2 Preliminary Measurements


A normal set of experiments comprises (i) a magnetization versus field plot at base temperature (2 or 5 K) having cooled in zero
field, and (ii) a magnetization versus temperature plot (from base temperature to 300 K), measured in a small applied field ( 100–
500 G, using a larger field for smaller samples or those which are magnetically dilute). For preliminary studies, data may be
collected in temperature swept mode (temperature continues to change slowly during the course of the measurements). Data
are often collected using small increments in the low temperature region (1 K) and larger increments in the intermediate region
Magnetic Measurements 179

(10–100 K) and large steps in the high temperature region. This leads to slightly larger uncertainties on the measured temperature
but is considerably quicker than stabilizing at each temperature prior to making a measurement. This permits a Curie–Weiss anal-
ysis to determine C and q. In addition any discontinuities in M versus T may identify structural, magnetic, or field-induced phase
transitions. The M versus H data permits the saturation magnetization to be identified and may also be useful for identifying
magnetic phases where spontaneous moments occur in zero field (ferromagnetism, ferrimagnetism, and canted
antiferromagnetism).

2.08.7.3 Advanced Magnetic Measurements


According to mean field theory, magnetic ordering will not occur above |q|. For apparent discontinuities occurring below |q|, further
sets of dc measurements can be made using a range of small applied fields (e.g., 10, 100, 500, 1000 Oe) to probe field-dependent
behavior in regions where discontinuities look present. A careful consideration of q plus field dependence of the magnetization can
be used to assess potential types of magnetic response. A field-dependent response is typically a signature of a spontaneous
moment, heralding the onset of ferro- (q > 0), ferri-, or canted antiferromagnetic (q < 0) behavior. For an antiferromagnet c passes
through a maximum and typically drops to 2/3 cmax in the low temperature region.
Systems exhibiting a spontaneous moment should exhibit an out-of-phase component in the zero-field ac susceptibility data
below the transition temperature. This response should exhibit very little (no) frequency dependence. Conversely a clear
frequency-dependent response (over 1–2 order of magnitude changes in u) may point toward SMM-like behavior. If well-
resolved maxima in c00 are not detected in the zero-field ac data, then a small applied field (say up to 1500 Oe) can be applied
to switch off quantum tunneling effects. Fits to the Arrhenius plot or Cole–Cole expression can be implemented to probe relaxation
processes. With sufficient data in hand to identify the magnetic response, the literature is normally a convenient starting point to
identify additional experiments which can be undertaken to more fully elucidate the magnetic behavior. Many of these characteristic
responses have been discussed in the previous chapters of this text.

2.08.8 Conclusions

The magnetic response of paramagnetic materials is diverse. From a practical perspective an analysis of the spin state and geometry is
helpful for identifying the single ion behavior with configurations containing residual orbital angular momentum likely to afford
large ZFS (D) which is significant for both clusters (SMM behavior) and long-range order (Ising or XY spin). In addition determi-
nation of possible significant exchange interaction pathways is helpful to identify whether the system may act as a single ion,
a discrete cluster, or may exhibit some form of low-dimensional or long-range order. Experimentally a provisional set of measure-
ments will typically include (a) a 1/c versus T plot to determine the Curie constant C and Weiss constant q and (b) a M versus H plot
at base temperature to check for field-induced effects at low temperature. The diagnostic behavior of common types of the magnetic
system are described here but readers are additionally encouraged to expand their reading, for example, using the selected “Further
Reading” here which will provide other useful entrance level texts.

Acknowledgment

Much of this article was based on a number of standard texts and reviews on topics in magnetochemistry which are highlighted below. Diagrams have
been adapted from these texts or original research papers and the authors of these works are acknowledged.

Further Reading

1. Blundell, S. Magnetism in Condensed Matter; Oxford University Press: Oxford, 2003.


2. Carlin, R. L. Magnetochemistry; Springer-Verlag: Berlin, 1986.
3. Kahn, O. Molecular Magnetism; Wiley-VCH: Weinheim, 1993.
4. Mabbs, F. E.; Machin, D. J. Magnetism and Transition Metal Complexes; Chapman & Hall: London, 1973.
5. Crangle, J. Solid State Magnetism; Springer: Berlin, 1991.
6. In Molecular Magnetism: From Molecular Assemblies to the Devices; Coronado, E., Delhaes, P., Gatteschi, D., Miller, J. S., Eds.; NATO-ASI Series E, Vol. 321. Kluwer
Academic Press, Dordrecht, Nederlands, 1996.
7. Ramos, E.; Roman, J. E.; Cardona-Serra, S.; Clemente-Juan, J. M. Comput. Phys. Commun. 2010, 181, 1929.
8. Chilton, N. F.; Anderson, R. P.; Turner, L. D.; Soncini, A.; Murray, K. S. J. Comput. Chem. 2013, 34, 1164.
9. Sorace, L.; Benelli, C.; Gatteschi, D. Chem. Soc. Rev. 2011, 40, 3092.
10. Liddle, S. T.; van Slageren, J. Chem. Soc. Rev. 2015, 44, 6655.
11. Ishikawa, N.; Sugita, M.; Wernsdorfer, W. Angew. Chem. Int. Ed. Engl. 2005, 44, 2931.
12. AlDamen, M. A.; Cardona-Serra, S.; Clemente-Juan, J. M.; Coronado, E.; Gaita-Ariño, A.; Martí-Gastaldo, C.; Luis, F.; Montero, O. Inorg. Chem. 2009, 48, 3467.
13. Clark, E. R.; Anwar, M. U.; Leontowicz, B. J.; Beldjoudi, Y.; Hayward, J. J.; Chan, W. T. K.; Gavey, E. L.; Pilkington, M.; Zysman-Colman, E.; Rawson, J. M. Dalton Trans. 2014,
43, 12996.
14. Bleaney, B.; Bowers, K. D. Proc. Royal. Soc. A 1952, 214, 451.
15. Crawford, V. H.; Richardson, H. W.; Wasson, J. R.; Hodgson, D. J.; Hatfield, W. E. Inorg. Chem. 1976, 15, 2107.
180 Magnetic Measurements

16. Nanda, K. K.; Thompson, L. K.; Bridson, J. N.; Nag, K. J. Chem. Soc. Chem. Commun. 1994, 1337.
17. Halcrow, M. A.; Sun, J.-S.; Huffman, J. C.; Christou, G. Inorg. Chem. 1995, 34, 4167.
18. Andres, H.; Basler, R.; Blake, A. J.; Cadiou, C.; Chaboussant, G.; Grant, C. M.; Güdel, H.-U.; Murrie, M.; Parsons, S.; Paulsen, C.; Semadini, F.; Villar, V.; Wernsdorfer, W.;
Winpenny, R. E. P. Chem. Eur. J. 2002, 8, 4867.
19. Taft, K. L.; Delfs, C. D.; Papafthymiou, G. C.; Foner, S.; Gatteschi, D.; Lippard, S. J. J. Am. Chem. Soc. 1994, 116, 823.
20. Guo, Y.-N.; Xu, G.-F.; Guo, Y.; Tang, J. Dalton Trans. 2011, 40, 9953.
21. Gavey, E. L.; Beldjoudi, Y.; Rawson, J. M.; Stamatatos, T.; Pilkington, M. Chem. Commun. 2014, 50, 3741.
22. Yang, T.; Perkisas, T.; Hadermann, J.; Croft, M.; Ignatov, A.; Van Tendeloo, G.; Greenblatt, M. J. Mater. Chem. 2011, 21, 199.
23. Cheaib, K.; Martel, D.; Clément, N.; Eckes, F.; Kouaho, S.; Rogez, G.; Dagorne, S.; Kurmoo, M.; Chouad, S.; Welter, R. Dalton Trans. 2013, 42, 1406.
24. Khuntia, P.; Mahajan, A. V. J. Phys. Condens. Matter 2010, 22, 296002.
25. Shao, X.; Yamajia, Y.; Sugimoto, T. J. Mater. Chem. 2009, 19, 3688.
26. Lin, Q.-P.; Zhang, J.; Cao, X.-Y.; Yao, Y.-G.; Li, Z.-J.; Zhang, L.; Zhou, Z.-F. Cryst. Eng. Comm. 2010, 12, 2938.
27. Palacio, F.; Antorrena, G.; Castro, M.; Burriel, R.; Rawson, J.; Nicholas, J.; Smith, B.; Bricklebank, N.; Novoa, J.; Ritter, C. Phys. Rev. Lett. 1997, 79, 2336.
28. Gregory, S. Phys. Rev. Lett. 1978, 40, 723.
29. Lampropoulos, C.; Hill, S.; Christou, G. ChemPhysChem 2009, 10, 2397.
2.09 Two-Dimensional Supramolecular Chemistry on Surfaces
C Pfeiffer and NR Champness, University of Nottingham, Nottingham, United Kingdom
Ó 2017 Elsevier Ltd. All rights reserved.

2.09.1 Introduction 181


2.09.2 Two-Dimensional Arrays Assembled Using Hydrogen Bonding 181
2.09.3 Two-Dimensional Arrays Assembled Using van der Waals Interactions 191
2.09.4 Covalently Coupled Two-Dimensional Arrays: Planar Covalent Organic Frameworks 191
2.09.5 Complex Structures With Low Degrees of Symmetry 192
2.09.6 Conclusions 197
References 198

2.09.1 Introduction

One key premise of supramolecular chemistry is the construction of nanoscale structures through the processes of self-assembly.1
Self-assembly employs and exploits interactions between molecules such as hydrogen bonding and van der Waals interactions
instead of covalent interactions within molecules. Using supramolecular interactions, highly complex molecular structures with
targeted properties can be prepared without utilizing multistep synthetic pathways. Initially, solution-phase chemistry was the focus
of supramolecular research.2 More recently, developments have concentrated on solid-state chemistry in combination with crystal
engineering.3 This shift of attention from solution-phase to solid-state structures is due to advances in characterization techniques
that allow for precise identification of targeted supramolecular structures, in particular NMR and single crystal X-ray diffraction.
Surface-based supramolecular structures have been recently explored and have led to an increased interest in fashioning
low-dimensional, predominately two-dimensional, structures.4–10 Developments in surface-based supramolecular chemistry are
the focus of this article. Emphasis will be on the synthetic approaches to such structures, particularly on the design of arrays through
supramolecular interactions where molecules are specifically fashioned to control relative molecular organization. Hydrogen
bonding to form extended arrays that propagate in two dimensions will be the focus of this article; however, other supramolecular
interactions that will be highlighted include van der Waals interactions, metal–ligand coordination, and covalent coupling. Besides
synthetic approaches, this article will discuss detailed characterization of the resulting two-dimensional arrays and the level of
understanding that can be achieved using scanning probe microscopies, including scanning-tunnelling microscopy (STM) and
atomic force microscopy (AFM).
Surface-based chemistry has long been used to assemble arrays of molecules. Perhaps the most widely studied area is the devel-
opment of self-assembled monolayers (SAMs) of thiolate molecules adsorbed onto Au(111) substrates.11 The simple deposition of
molecules on surfaces, as seen with the Au–thiolate SAMs, leads to closed packed arrays with arrangements typically defined by van
der Waals interactions and simple geometric formations. Combining this with supramolecular chemistry, more complex and poten-
tially porous structures can be created by using intermolecular interactions, thus, logically expanding into the formation of robust
covalently coupled arrays.
The surface used as the environment to assemble a given supramolecular array defines a two-dimensional boundary upon which
the self-assembly process is developed. Suitable surfaces often studied include Au(111),12 Ag–Si(111)O3  O3R30 [Ag/Si(111)],13
and highly oriented pyrolytic graphite (HOPG).14 However, recent studies have started to explore other substrates including gra-
phene15 and boron-nitride (BN) monolayers.16 The choice of substrate is influenced by two factors: the tendency to adopt weak
interactions with the molecules used in the self-assembly and the requirements of the scanning probe microscopy utilized. Noting
the types of interactions involved in a system is significant in the choice of substrate. In the case of covalently coupled arrays, the
surface can play an active role in promoting coupling reactions and therefore the choice of substrate is extremely important. In many
instances, the determining feature has been dependent on the employed scanning probe microscopy. For instance, STM requires
conducting surfaces in order to produce images; therefore, the choice of substrates is restricted to (semi)conducting materials.
Recent improvements in resolution of AFM17 and the development of dynamic force microscopy (DFM)18 have led to a wider scope
for characterization and expanded the selection of substrates to insulating materials.
This article is subdivided into four sections, systems assembled using (i) hydrogen bonds, (ii) van der Waals interactions, (iii)
covalent bonds, and lastly a section (iv) on self-assembled systems with unusual ordering that illustrate the power of the approach
and particularly the molecular-level characterization of such systems.

2.09.2 Two-Dimensional Arrays Assembled Using Hydrogen Bonding

The use of hydrogen bonds to create supramolecular structures goes back to the origins of the field.19 An early example of surface-based
supramolecular chemistry is the formation of a unimolecular supramolecular structure mediated by hydrogen bonding reported by

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12501-6 181


182 Two-Dimensional Supramolecular Chemistry on Surfaces

Griessl et al.20 The structure is prepared via deposition of trimesic acid onto an HOPG substrate in ultra high vacuum (UHV) condi-
tions with subsequent imaging by STM, which reveals precise details of the self-assembled molecular structure. The intermolecular
carboxylic acid:carboxylic acid hydrogen bonds, which adopt the classic, R22(8) intermolecular arrangement,21 ensure that an open
structure is adopted in preference to any alternative close-packed arrangement. Importantly, this confirms that hydrogen-bonding
interactions are the dominant force in determining the molecular arrangement of these structures. The observed structure is the
expected honeycomb arrangement leading to a porous network structure; however, in addition to the honeycomb structure a secondary
so-called “flower” arrangement is also observed by STM measurements. This alternative self-assembled structure results from the adop-
tion of R33(12) supramolecular synthons formed by three carboxylic acid moieties from separate trimesic acid molecules. It is
proposed that the “flower” structure forms due to a higher molecular density on the surface, indicating that surface coverage influences
the final self-assembled structure, maximizing the energy gained through adsorbate–substrate interactions. Similarly, a study of an
elongated analogue of trimesic acid, 1,3,5-tris(4-carboxyphenyl)benzene, also deposited on HOPG, but in this case from a range of
alkyloic acids rather than by sublimation in UHV, also results in the formation of a honeycomb lattice at low temperatures, but
a more densely packed phase at higher temperatures.22 The authors suggested that the coadsorption of solvent molecules within
the honeycomb structure stabilizes the nominally porous structure at low temperatures, but upon elevation of the temperature the
weakly bound solvent molecules desorb, initiating the transition to the more densely packed and thermodynamically favored phase.
A simple example of a unimolecular self-assembled structure that uses hydrogen bonds is that of naphthalene-1,4:5,8-
tetracarboxylic diimide (NTCDI).23 NTCDI contains imide moieties at opposing ends of the rod-shaped molecule which can adopt
imide/imide hydrogen bonds. Analogous with carboxylic acid:carboxylic acid interactions an R22(8) intermolecular interaction is
observed. As a result of the divergent arrangement of the imide moieties chains of molecules are observed when NTCDI is deposited
onto a Ag/Si(111) surface (Fig. 1A). Imaging using DFM24–26 reveals submolecular details of the molecular arrangement. It is inter-
esting to note that features in the DFM images appear to coincide with where hydrogen bonds would be expected to be observed
(Fig. 1B); however, calculations indicate that these features do not arise as a result of the hydrogen bonds, but due to some other, as
yet undefined, phenomenon. The origin of such intermolecular features remains a highly debated topic.27
The full canvas of supramolecular design concepts can be brought to bear upon the construction of surface-based arrays and one
area that has attracted considerable attention is the use of DNA nucleobases, which have specific hydrogen-bonding capability. The
use of DNA nucleobases in supramolecular chemistry has been a persistent theme since the seminal work of Seeman28 and has been
developed by a number of research teams to significant effect29 including some surface-based studies.30,31 In addition to the studies
of self-assembly on surfaces of large DNA fragments, a particular focus for hydrogen-bonding studies has been the use of the

Figure 1 (A) STM image of three molecular chains of NTCDI adsorbed on Ag/Si(111)23; (B) constant height DFM image of NTCDI adsorbed on
Ag/Si(111) at 77 K (2.1  2.0 nm)24; (C) view of hydrogen-bonded chains in the single crystal X-ray structure of NTCDI.23 (A and C) Reproduced with
permission from the American Chemical Society; (B) reproduced with permission from the Nature Publishing Group.
Two-Dimensional Supramolecular Chemistry on Surfaces 183

individual DNA bases to create self-assembled structures on surfaces. The assembly of guanine and its derivatives on surfaces has
received particular attention.32–35 For example, the adsorption of guanine onto a Au(111) surface under UHV conditions leads to
the formation of guanine quartets as imaged by STM.32,35 The quartets form through Hoogsteen-style hydrogen bonding and are
associated through further NeH/N hydrogen bonds, leading to two-dimensional supramolecular arrays. This illustrates the
complexity of using DNA bases due to the variety of potential hydrogen-bonding motifs that can be adopted from classic
Watson–Crick pair formation, Hoogsteen interactions, reverse Watson–Crick, and reverse Hoogsteen arrangements (Fig. 2).
Only the latest examples are highlighted here due to the area being reviewed only recently.36
The hydrogen-bonding capability of DNA nucleobases can be exploited by attaching them as appendages to more complex mole-
cules, potentially leading to enhanced control over supramolecular structure. An example of such a strategy is given by the study of
a porphyrin molecule (tetra-TP) in which the porphyrin core is functionalized in each meso-position by a phenylthymine moiety
such that each thymine presents a hydrogen-bonding face exo to the porphyrin core.37 The molecule self-assembles on an HOPG
substrate to give rise to a two-dimensional structure wherein the molecules interact through R22(8) intermolecular thymine/thy-
mine hydrogen bonds (Fig. 3A). The asymmetric arrangement of the thymine groups appended to the tetra-TP molecules suggests
the potential for the molecules to adopt a chiral arrangement when adsorbed onto a surface and this is indeed observed. The almost
perfectly square two-dimensional unit cell observed for the tetra-TP structure suggests that all of the thymine groups within an
individual tetra-TP molecule adopt the same orientation with respect to the porphyrin, and that individual domains contain only
molecules of the same handedness. Overall the array of tetra-TP remains globally achiral by forming an equal area of domains
containing either right- or left-handed molecules. The formation, by prochiral molecules, of homochiral domains on surfaces con-
taining molecules of only a single handedness has been noted in many previous studies.38 Addition of a substituted adenine species,
9-propyladenine, to the assembly of tetra-TP produces a molecular network with a combination of disordered regions interspersed
with small domains of an ordered cocrystal structure containing both tetra-TP and 9-propyladenine. As anticipated, favorable
Watson–Crick thymine–adenine interactions are observed, but interestingly dimers of adenine are also formed. They achieve a high
surface coverage and form an additional NeH/N hydrogen bond to a further tetra-TP molecule (Fig. 3B). The complexity of the struc-
ture reveals the difficulties in designing such structures, even when relatively predictable hydrogen-bonding synthons are present.
A related approach has been used by González-Rodríguez and De Feyter39 who designed rod-shaped molecules decorated by
DNA bases at the molecule’s termini, either with guanine (G) and cytosine (C) or adenine (A) and uracil (U) (Fig. 4A). The
rods are designed such that opposing ends will form complementary hydrogen-bonding groups. Additionally, the authors function-
alized the rods with alkyl tails such that any residual space in the structure could be occupied such that no additional molecules are
required to fill space within the structure. Self-assembly on a HOPG substrate results in the formation of cyclic structures, as
confirmed by STM imaging (Fig. 4B), indicating the successful employment of the complementary hydrogen-bonding moieties
in forming heteromolecular hydrogen-bonding interactions (G:C or A:U). Despite the alkyl chain appendages filling the space
within the cyclic structures, the GC system is capable of adsorbing coronene with the alkyl chains assumed to vacate the space
due to preferential physisorption of the large, flat, polyaromatic guest onto the HOPG surface.
Moving beyond DNA systems it is possible to create nonnatural, hydrogen-bonding synthons for the generation of extended
two-dimensional arrays. For example, the triple hydrogen-bonding interaction between 2,6-di(acetylamino)pyridines and imide
groups has received extensive study in supramolecular chemistry19,40 and can also be exploited to form bimolecular structures
on surfaces.41 An early example of a bimolecular network was prepared from perylene-3,4,9,10-tetracarboxylicdiimide (PTCDI),
which bears two imide appendages at opposing ends of the rod, similar to NTCDI discussed above, and melamine (Fig. 5).42

Figure 2 Potential hydrogen-bonding interactions between thymine and adenine: (A) standard Watson–Crick, (B) reverse Watson–Crick, (C)
Hoogsteen, and (D) reverse Hoogsteen arrangements.
184 Two-Dimensional Supramolecular Chemistry on Surfaces

Figure 3 (A) Two-dimensional self-assembled network of tetra-TP adsorbed on HOPG liquid–solid interface, left: STM image. The inset shows
a high resolution, drift corrected STM image of the network with an individual 2D unit cell marked in red: scale bar ¼ 20 nm (inset ¼ 2 nm). Right:
Molecular model of the tetra-T-porphyrin network from molecular mechanics (MM) simulations. (B) 2D self-assembled network of tetra-TP and
9-propyladenine adsorbed on HOPG liquid–solid interface, left: STM image. The inset shows a high resolution, drift corrected STM image of the
network with an individual 2D unit cell marked in red: scale bar ¼ 20 nm (inset ¼ 1.6 nm). Right: Molecular model of the tetra-T-porphyrin/9-
propyladenine network from MM simulations.37 Reproduced with permission from the Royal Society of Chemistry.

Codeposition of the two building blocks onto a Ag/Si(111) surface under UHV conditions results in the formation of a honeycomb
network with triple hydrogen bonds formed between the PTCDI imide moieties and each side of the triangular melamine molecule.
The network was found to be commensurate with the underlying Ag/Si(111) surface, indicating an influence from the substrate in
the formation of the network.
Large pores are formed by the network, with a diameter of 2.4 nm, which are capable of hosting several large molecules as
initially demonstrated by the encapsulation of C6042 (Fig. 5). The entrapment of C60 by the PTCDI–melamine hexagonal network
leads to the formation of heptameric C60 clusters within the pores. The arrangement and properties of the C60 clusters were found to
differ from close-packed fullerene on the same surface, which in contrast to the heptamers does not align with the principle axes of
the underlying Ag/Si(111) substrate, thus demonstrating the templating and stabilizing effects of the network.42 Indeed, the ability
of the PTCDI–melamine family of arrays has been shown to host other fullerenes including C8443 and Lu@C82,44 and a range of
molecules discussed in more detail later.
The PTCDI–melamine network42 can also be prepared on a Au(111) surface,45 leading to an analogous honeycomb arrange-
ment to that observed on a Ag/Si(111) substrate. However, after annealing at higher temperatures, a parallelogram phase was
also observed on Au(111), which has the same stoichiometric ratio as the honeycomb structure, but is more dense.46 The parallelo-
gram phase is also capable of trapping guest molecules such as C60, but due to the restricted cavity size only two C60 molecules are
hosted, the size of the cluster being defined by the physical size of the network pores. Similar studies have demonstrated the entrap-
ment of Lu@C82 by the parallelogram phase, but in this case only single fullerenes are accommodated within each pore.44 The
parallelogram PTCDI–melamine phase is also able to trap two decanethiol molecules which sit parallel to the surface under the
UHV conditions used. The entrapment of a potentially more reactive molecule, such as a thiol, confirms that such networks
Two-Dimensional Supramolecular Chemistry on Surfaces 185

Figure 4 (A) Mixed guanine–cytosine and adenine–uracil hydrogen-bonding rods employed for self-assembly.39 (B) STM image of GC1 on HOPG
showing pairs of monomers and model of observed structure. (C) STM image of GC2 on HOPG and model of observed cyclic structure. (D) STM
image of AU2 on HOPG and model of observed cyclic structure.39 Reproduced with permission from Wiley-VCH Verlag GmbH and Co. KGaA.

have the potential for trapping a range of different species. It is interesting to note that adsorption of decanethiol onto the
PTCDI–melamine arrays, in UHV, leads to the destruction of the honeycomb phase, but not the parallelogram phase, suggesting
that the latter is preferentially stabilized by the thiol guest.47
The adsorption of thiol guests within the pores of PTCDI–melamine networks has been advanced significantly by Buck et al.48,49
It has been demonstrated that the PTCDI–melamine network can be assembled on a Au(111) surface from solution and used for
a variety of applications, particularly as a patterning tool (Fig. 6).48,49 Due to the solution-based preparative procedure it is possible
to apply traditional Au–thiolate SAM-synthesis. The adsorption of adamantanethiol within the pores of the network is demon-
strated with the thiolates sitting perpendicular to the surface, in contrast to related studies in UHV conditions.47 The clusters of
adsorbed thiols, or confined SAMs, in this example confirmed that adamantanethiol SAMs can be exploited to guide the deposition
of copper atoms within the pores using underpotential deposition. Copper metal was successfully deposited only on areas of the
surface that were not covered by the network, that is, those areas covered by thiols48 and a subsequent study demonstrated that the
PTCDI–melamine network acts as a diffusion barrier for copper atoms.49,50
The encapsulation of thiol-decorated molecules by the PTCDI–melamine network has been exploited further with examples that
demonstrate the versatility of the approach. Such studies have included the entrapment of thiol-functionalized porphyrins51 and
complex functionalized polyoxometalates (POM).52 The latter study reports the entrapment of a disulfide-functionalized
Keggin-type POM, [PW11O39Ge-{p-C6H4eCCeC6H4eNHC(O)(CH2)4{eCH(CH2)2SeSe}}]4 , into the PTCDI–melamine array
which has been adsorbed on a Au(111) substrate. Through formation of a AueS bond the functionalized POM can be trapped
specifically within the pores of the self-assembled array. STM studies indicate spatial resolution of the POMs that is reminiscent
of the underlying supramolecular array (Fig. 7).
186 Two-Dimensional Supramolecular Chemistry on Surfaces

Figure 5 (A) Schematic of the PTCDI–melamine junction, showing the nine hydrogen bonds that make up the structural node as dashed red lines.
(B) STM image of the PTCDI–melamine network on Ag/Si(111); inset, high resolution view of the Ag/Si(111) surface. (C) Schematic of the network
showing the registry with the hexagonal substrate. (D) STM image of fullerenes trapped within the pores of the hexagonal network, seen as bright
white features. (E) Schematic diagram of a C60 heptamer sitting within a pore.42 Reproduced with permission from the Nature Publishing Group.

Figure 6 (A) The adamantanethiol molecule. (B) Schematic of the network filling process. (C) STM image of the PTCDI–melamine network on
Au(111)/mica filled with adamantanethiol, scale bar ¼ 20 nm. Insets: higher resolution STM image and Fourier transforms, scale bar ¼ 5 nm.48 Repro-
duced with permission from the Nature Publishing Group.

Sublimation or solution-phase deposition can be employed to adsorb molecules onto hydrogen-bonded arrays but these
techniques are not always suitable for sensitive molecules and thus a further technique has been developed using electrospray depo-
sition.53,54 Electrospray techniques have been used to deposit Mn12O12(O2CCH3)16(H2O)4 clusters, a molecule of interest due to its
ability to act as a single molecule magnet, onto the PTCDI–melamine array on a Au(111) substrate.54 The softer deposition
Two-Dimensional Supramolecular Chemistry on Surfaces 187

Figure 7 A disulfide-functionalized POM (shown above with blue WO6 octahedra depicted with oxygen atoms at the vertices and metal cations
buried inside) and a STM image showing the POM trapped within the PTCDI–melamine network on a Au(111) substrate. POMs can be seen as bright
white features and the PTCDI–melamine array is indicated by the black dotted lines, scale bar ¼16 nm.52 Reproduced with permission from the
American Chemical Society.

technique is required as sublimation is not possible for Mn12O12(O2CCH3)16(H2O)4 due to its decomposition at elevated temper-
atures. In the absence of the hydrogen-bonded array the Mn12O12(O2CCH3)16(H2O)4 clusters assemble generating filamentary
structures on the surface (Fig. 8B). In contrast adsorption of Mn12O12(O2CCH3)16(H2O)4 clusters onto the PTCDI–melamine array
results in a low degree of ordering with some molecules accommodated within the pores of the network but with others positioned
on top of the array (Fig. 8C). The entrapment of the Mn12O12(O2CCH3)16(H2O)4 clusters in the array is not as efficient as that of
other species, including fullerenes, most probably due to a mismatch of dimensions between the cluster (1.6 nm) and pore
(2.5 nm). This study indicates that maximum organization of guest molecules by porous frameworks can be achieved by idealizing
size match between host and guest and by maximizing interactions between host and guest.
An advantage of the bimolecular approach to array formation is that each component can be altered, allowing for modification
of the resulting structure. Thus, for the PTCDI–melamine arrays the PTCDI molecule56,57 can be readily functionalized, leading to
manipulation of the framework and cavity dimensions within the array. Importantly, modification of the components not only
affects the sterics and dimensions of the building blocks, but can also affect their physical and chemical properties and tune their
solubility for studies at the surface–solution interface. By introducing steric bulk to the PTCDI molecules, the available space for
guest entrapment can be controlled which results in networks that can potentially trap individual molecules in a regular array.
Self-assembled arrays formed between melamine and PTCDI derivatives, and in some examples networks formed by the function-
alized PTCDI derivatives alone, are able to act as hosts to guest molecules.56,57 For example, self-assembly of thioadamantyl func-
tionalized PTCDI, (SAdam)2-PTCDI, and melamine leads to the anticipated honeycomb array.57 The thioadamantyl groups are
both bulky and rigid and therefore would be expected to provide an effective mechanism for inhibiting the space within the network
pores. However, upon formation of a (SAdam)2-PTCDI–melamine network some of the thioadamantyl groups are cleaved from the
PTCDI moieties.57 Consequently, a variety of distinct pore sizes and configurations are formed which can be visualized following
C60 adsorption onto the network and subsequent STM imaging. Individual C60 molecules are more readily imaged by STM than the
underlying thioadamantyl groups and, by employing this approach, it is feasible to identify the different orientations of the molec-
ular clusters within each pore. The number of C60 molecules adsorbed within a given pore is determined by the degree of thioada-
mantyl cleavage and clusters ranging from dimers to heptamers are observed. Considering the dimensions of the honeycomb pores
and the size of the thioadamantyl appendages, clusters of greater than five fullerenes can only be explained by cleavage of the thi-
oadamantyl groups, the heptamers corresponding to a pore where no thioadamantyl appendages remain. Further constriction of
pore size can be used to control the number of guest molecules, a principal demonstrated by the array formed by 1,7-bis(4-
benzoic acid)-3,4,9,10-perylenetetracarboxylic diimide.58 As with many disubstituted PTCDI species,41 Bz2-PTCDI assembles
into a unimolecular honeycomb array, without the need to add melamine, through the adoption of a tri-molecular R33(12)
188 Two-Dimensional Supramolecular Chemistry on Surfaces

Figure 8 (A) Schematic representation of Mn12O12(O2CCH3)16(H2O)4; STM images showing (B) the filamentary structures formed by molecular
aggregates comprised of individual Mn12O12(O2CCH3)16(H2O)4 molecules, scale bar ¼ 10 nm. (C) Mn12O12(O2CCH3)16(H2O)4 molecules deposited
onto a PTCDI–melamine network on a Au(111) surface, scale bar ¼ 20 nm.54 Reproduced with permission from the Nature Publishing Group.

hydrogen-bonded junction (see the succeeding text). The rigid protrusion of the phenylcarboxylate moieties into the framework
cavities leads to the encapsulation of single C60 molecules within each pore, which are in turn spaced in a regular fashion across
the surface.
The material employed as substrate for the self-assembly process clearly influences the self-assembly process and the resulting
framework structure and should not be considered an innocent agent in the synthetic strategy. The influence that the substrate can
have on supramolecular assembly is illustrated by the use of graphene or boron-nitride (BN) “nanomesh” monolayers grown on
Rh(111) crystals.55 Both systems exhibit moiré patterns that introduce distinct adsorption sites on the surface. The self-assembly of
PTCDI and related di-functionalized derivatives, 1,7-dipropylthio-perylene-3,4:9,10-tetracarboxydiimide (DP-PTCDI) and
1,7-di(butyl)-coronene-3,4:9,10-tetracarboxylic acid bisimide (DB-CTCDI), on graphene has been studied by STM (Fig. 9).55
Two distinct junctions formed by either two or three molecules are observed and result in different assemblies (Fig. 9C, 9D).
For example, unfunctionalized PTCDI forms rows up to 25 nm in length running parallel to the principal directions of the substrate
and adopts only simple dimeric hydrogen-bonding arrangements. The molecular arrangement on the graphene superstructure
differs significantly from that observed for a graphite substrate, on which three-dimensional islands are formed.60 The importance
of a commensurate match between molecular dimensions and the moiré periodicity is demonstrated by a comparison with adsorp-
tion of PTCDI on a BN “nanomesh” monolayer (Fig. 9). The BN monolayer is isoelectronic with graphene but has a slightly larger
periodicity of 3.2 nm, which is noncommensurate with the dimensions of PTCDI. On BN [on Rh(111)], individually isolated
PTCDI molecules are trapped in energy minima associated with the moiré pattern which clearly demonstrates the influence of
Two-Dimensional Supramolecular Chemistry on Surfaces 189

Figure 9 STM images acquired following deposition of (A) DP-PTCDI, (B) DB-CTCDI on a graphene monolayer formed on Rh(111). (C and D)
Diagram of junctions between DP-PTCDI dimers (C) and trimers (D) stabilized, respectively, by two and three C]O/NH hydrogen bonds between
neighboring molecules with dimer center–center spacing of d and trimer vertex to molecule center spacing r. (E) Placement of DP-PTCDI trimers and
dimers. (F) DB-CTCDI trimer junction analogous to (D) with vertex to molecule center spacing r and placement of DB-CTCDI trimer on the graphene
superstructure. (G and H) STM images of DB-CTCDI showing chirality of junctions and intramolecular details of molecules. The hexagons in (H)
highlight the chirality of the molecular arrangement.55 Reproduced with permission from Wiley-VCH Verlag GmbH and Co. KGaA.

the underlying substrate on supramolecular assembly. On graphene both DP-PTCDI and DB-CTCDI form linear rows that are built
using dimeric hydrogen bond junctions, and more complex arrays which include junctions where three molecules meet in
a hydrogen-bonded trimeric arrangement. In the case of DP-PTCDI the ratio of dimer:trimer junctions is 75:25 in comparison to
less than 1% of junctions being trimers for unfunctionalized PTCDI. In the case of DB-CTCDI the three molecule junctions domi-
nate the self-assembly process, with no linear dimers unambiguously identified. The array constructed from trimeric junctions
results in a honeycomb framework aligned with the graphene monolayer superstructure and encloses the areas of bright contrast
arising from the moiré pattern.
The intermolecular hydrogen-bonding synthon exploited in the PTCDI–melamine arrays can be readily incorporated into other
building blocks (e.g., thymine includes the same imide moiety seen in PTCDI) and thus many possible arrays can be targeted. Perhaps
the simplest example is the self-assembled array formed between cyanuric acid and melamine.61,62 Samori et al.63 have elegantly
demonstrated the self-assembly of a range of diimide molecules and melamine (Fig. 10) at the solution/solid interface on an
HOPG surface following deposition from a 1,2,4-trichlorobenzene/dimethylsulfoxide solution. The formation of hexagonal/honey-
comb networks is anticipated for such systems in an analogous fashion to the PTCDI–melamine systems, and such arrays are
observed, but detailed studies of concentration variation lead to the adoption of a range of structures. The nature of the structures
formed by combinations of the diimide molecules (labelled 1–4 in Fig. 10) with melamine ranges from open hexagonal networks
to close-packed structures and significantly varies with component concentration. Ultimately, a phase diagram of polymorphs is
generated for these systems which gives perspective to the complexity of the self-assembly processes involved (Fig. 10).
190 Two-Dimensional Supramolecular Chemistry on Surfaces

Figure 10 Diimide molecules used by Samorì et al.63 in combination with melamine (MEL) to afford self-assembled structures (top). The nature of
the structure is concentration dependent giving rise to a range of polymorphic products (bottom). Reproduced with permission from the American
Chemical Society.

Elongated analogues of PTCDI have also been employed in the synthesis of supramolecular arrays in combination with mela-
mine. In particular, a functionalized terrylene diimide has been used to assemble a honeycomb network, analogous to the PTCDI–
melamine array discussed in the preceding text.64 The terrylene diimide–melamine array is assembled on a graphene passivated
diamond (Fig. 11) and by using a gallium drop as a counter electrode the photocurrent of the array can be measured. It was found
that under irradiation at 710 nm an incident photon to current efficiency of 0.6% was observed. The study indicates the possible
application of such self-assembled supramolecular arrays for the fabrication of optoelectronic devices.

Figure 11 (A) Functionalized terrylene diimide employed in the assembly of (B) a terrylene diimide–melamine array on a graphene passivated dia-
mond surface. A gallium drop is used as a counter electrode to study the photoresponse of the supramolecular array.64 Reproduced with permission
from the Nature Publishing Group.
Two-Dimensional Supramolecular Chemistry on Surfaces 191

2.09.3 Two-Dimensional Arrays Assembled Using van der Waals Interactions

The role of van der Waals interactions in the creation of surface supramolecular frameworks cannot be underestimated. A large
number of studies have been reported using van der Waals interactions to create supramolecular structures and these have been
reviewed on a number of occasions.7–9,65,66 A particularly notable interaction that has been widely exploited is the strong adsorp-
tion between alkyl chains and HOPG. The role of the alkyl chain is clearly significant in the resulting assembly process, and inter-
digitation of alkyl chains from adjacent molecules is commonly observed and provides a method of control over the molecular
arrangement. A representative example of a self-assembled framework controlled by van der Waals interactions was investigated
by Schull et al.67 1,3,5-tris[(E)-2-(3,5-didecyloxyphenyl)-ethenyl]-benzene, a trimeric compound with six pendent decyl chains,
was deposited onto an HOPG substrate. Self-assembly through interdigitation of the alkyl chains causes the formation of a porous
network within which coronene and hexabenzocoronene could be accommodated. The coadsorption of coronene or hexabenzo-
coronene modified the nature of the resulting structure, in a similar fashion to solution-phase host–guest chemistry. Indeed, inter-
digitation of alkyl chains has been extensively used by De Feyter and coworkers in a number of elegant studies.7–9,65,66,68–73 One
area of particular focus of research has been the study of alkoxylated dehydrobenzo[12]annulenes.68 A representative study of
rhombic-shaped dehydrobenzo[12]annulenes assembled at a 1,2,4-trichlorobenzene/HOPG interface68 demonstrates that varia-
tion in the chain length of alkyl substituents results in five distinct network structures, three porous and two nonporous structures.68
Shorter alkyl chain substituents on the dehydrobenzo[12]annulene core favor formation of porous frameworks, whereas those with
longer alkyl chains tend towards the formation of nonporous arrangements. Concentration of the building blocks in solution also
plays a key role in the nature of the structures formed. Dilution of the dehydrobenzo[12]annulenes solutions leads to the transfor-
mation of nonporous structures into porous frameworks as a result of factors related to overall surface coverage.
As in the solid state, the subtle interplay of different intermolecular supramolecular interactions can lead to the formation of
a range of different structures with little energetic differences between the different arrangements, that is, polymorphs. A study
by De Feyter59 describes an approach to establish the range of polymorphs for a given system using building blocks that exploit
van der Waals interactions or hydrogen-bonding interactions. The formation of polymorphs in two-dimensional supramolecular
arrays, as with crystalline materials, is one of the most significant challenges in the research field as polymorphic arrangements
always provide a potential barrier to the preparation of well-defined, ordered arrays. De Feyter et al. approached the question of
polymorph screening by designing a mechanism to induce a solution flow across a substrate surface to generate a lateral concen-
tration gradient. This in situ generation of a gradient allows both discovery and separation of multiple polymorphs in a single exper-
iment. The authors describe three separate systems: hexadecyloxy-substituted dehydrobenzo[12]-annulene (DBA-OC16),
hexadecyl-substituted bis(dehydrobenzo[12]annulene) (bisDBA-C16), and 1,3,5-tris(4-carboxyphenyl)benzene (BTB). Their
approach demonstrates systems that exploit either van der Waals or hydrogen-bonding interactions. For each example, more
than one polymorph was found for each system by STM imaging of the respective samples; DBA-OC16dtwo polymorphs;
bisDBA-C16dfour polymorphs; BTBdtwo polymorphs (Fig. 12). The approach allows not only structural characterization of
the polymorphs observed, but also quantification of surface coverage of each arrangement, giving an approximate yield of the
respective polymorphic systems.

2.09.4 Covalently Coupled Two-Dimensional Arrays: Planar Covalent Organic Frameworks

The reversible formation, and hence possible structure correction of both hydrogen bonding and van der Waals interactions allows the
formation of well-ordered structures. However, to create robust two-dimensional structures, covalently-bonded frameworks are a more
attractive approach. Thus, a strategy that slows the formation of covalently coupled arrays is desirable. A number of approaches have
been used to target such systems, including aryl–aryl coupling74–77 and alkyne–alkyne coupling78 processes. Other approaches have
been taken into consideration in recent studies79 and the field has clear parallels to studies of covalent organic frameworks or COFs.80
An early approach to the development of covalently coupled arrays was the use of the intermolecular coupling between halo-aryl
species on Au substrates. Grill et al. developed their approach through the use of porphyrin building blocks with bromophenyl
“arms” that protrude from the meso-positions of the porphyrin.74 The number of bromophenyl arms can be varied such that a range
of building blocks with one, two, or four reactive appendages can be utilized. Reaction of mono-bromophenyl-substituted
porphyrin, by heating the building blocks on a Au(111) substrate, leads to dimer formation (Fig. 13A); trans-di-bromophenyl-
substituted porphyrin reacts to give chains (Fig. 13B) and tetra-bromophenyl-substituted porphyrin can react in all four positions
to give well-ordered two-dimensional grids of porphyrins (Fig. 13C). The approach was developed by Grill, Hecht, and coworkers
through the construction of covalently coupled arrays built from two different molecular components. Their strategy exploits the
difference in reactivity between bromophenyl and iodophenyl groups attached to a porphyrin core in a trans-arrangement leading
to different reactivity between the different edges of the porphyrins.75,76 Deposition of the dibromo-diiiodo-porphyrin (Br2I2TPP)
building block onto a Au(111) substrate and heating the sample results in the covalent coupling between adjacent porphyrin
species via the iodo-functionalized edges. This is due to the lower bond dissociation energy of the CeI bonds in comparison to
CeBr bonds. Preferential coupling of the two trans-arranged edges leads to the formation of one-dimensional chains of coupled
porphyrin building blocks (Fig. 14). Below 200 C only the iodo-functionalized arms react and leave the bromo substituents intact,
allowing the possibility of further reaction. However, following the formation of the one-dimensional chains of coupled porphy-
rins, functionalized with bromophenyl appendages, the reaction with a secondary species can be successfully achieved, as
192 Two-Dimensional Supramolecular Chemistry on Surfaces

Figure 12 (A) Molecular structures of DBA-OC16, bisDBA-C16, and BTB and representations of the polymorphs observed for each compound (B)
DBA-OC16; (C) bisDBA-C16; (D) BTB.59 Reproduced with permission from the Royal Society of Chemistry.

demonstrated via codeposition of dibromoterfluorene. Coupling of the bromo-appendages on the porphyrin chain and the terfluor-
ene molecules leads to decoration of the porphyrin chains with terfluorene groups. In some instances, cross-linking of adjacent
porphyrin chains is observed and a complex of two-dimensional array of coupled porphyrin chains and terfluorene moieties is
formed (Fig. 14). The simple strategy of hierarchical coupling, exploiting dissimilar chemical reactivity and thermal activation
barriers for the reactive components, causes the formation of a bimolecular array and opens up the possibility for the creation
of more complex covalently coupled arrays.
The strength of the strategy of using covalent coupling to form surface-based arrays is demonstrated by a recent study by Ama-
bilino and Raval.81 The study reports the coupling of polyaromatic molecules (pentacene, tetramesitylporphyrin, perylene) and
porphyrins (H2-porphyrin and Zn(II)diphenylporphyrin) through CeH bond activation reactions on a copper surface. It is demon-
strated that all of the molecules are capable of undergoing C–H activation and can be coupled on the chosen surface. The study
demonstrated not only the ability to homo-couple the molecules into extended arrays, for example, Zn(II)diphenylporphyrin
can be cross-coupled through the unsubstituted edges (i.e., those not functionalized by phenyl groups) to afford covalently coupled
one-dimensional chains of porphyrins, but also the hetero-coupling between the different building blocks. A large variety of
coupled structures are described (Fig. 15) with up to three different building blocks cross-coupled. STM studies provide sufficient
insight to allow characterization of each individual component of the covalently coupled structure. A related study forms one-
dimensional ladders using cross-coupled arrays of Co(II)diphenylporphyrin.82 Coadsorption of 1,3-phenylenebis(methylene)-
bis-imidazole allows coordination of the imidazole groups to the porphyrin-bound cobalt cation. As a result of the formation
of the Coeimidazole bond being reversibly formed it is possible to image the bis-imidazole molecule moving, or walking, from
one porphyrin to another along the covalently coupled chains.
These studies demonstrate the ability to prepare highly complex covalently coupled structures from multiple building blocks and
indicates the potential for creating bespoke covalent structures on surfaces.

2.09.5 Complex Structures With Low Degrees of Symmetry

One of the features that separates the study of surface-based supramolecular arrays from other areas of supramolecular chemistry is
the ability to image the resulting structures at the molecular and submolecular18,24 level. This allows a level of detailed
Two-Dimensional Supramolecular Chemistry on Surfaces 193

Figure 13 (A–C) Bromophenyl functionalized porphyrin building blocks with increasing numbers of functional moieties for subsequent polymeriza-
tion. STM images and representative models of (D) dimers prepared from mono-bromophenyl functionalized building-block BrTPP; (E) one-
dimensional chains prepared from trans-Br2TPP. The arrow indicates where two chains are held together by weak noncovalent interactions; (F)
two-dimensional sheets prepared from covalently coupled Br4TPP.74 Reproduced with permission from the Nature Publishing Group.

Figure 14 (A) A porphyrin building block with two bromophenyl and two iodophenyl appendages in a trans-arrangement. (B) Aryl–aryl coupling
leads to the formation of one-dimensional chains. (C) Subsequent reaction with dibromoterfluorene leads to the formation of a mixed, covalently
coupled framework structure. The STM images confirm the molecular arrangement within the covalently coupled structure.76 Reproduced with
permission from the Nature Publishing Group.
194 Two-Dimensional Supramolecular Chemistry on Surfaces

Figure 15 (A) Building blocks capable of undergoing CeH activation on a Cu(110) surface and (B) STM image demonstrating the covalent coupling
of three distinct components.81 Reproduced with permission from the American Chemical Society.

understanding that is extremely challenging, if not impossible, in other areas of supramolecular chemistry due to the methods of
characterization employed. The additional level of detail that can be achieved using STM, or AFM, has allowed characterization of
structures that would prove extremely challenging by any other technique.
One of the first examples of such a structure is that formed by terphenyl-3,300 ,5,500 -tetracarboxylic acid.83 Terphenyl-3,300 ,5,500 -
tetracarboxylic acid, adsorbed onto HOPG, leads to the formation of a two-dimensional hydrogen-bonded structure that utilizes
R22(8) intermolecular carboxylic acid:carboxylic acid interactions. STM imaging allows identification of the position of each mole-
cule in the framework and direct visualization of a nonordered structure. Indeed, the relative position of molecules within the array
is random and reminiscent of dynamically arrested systems such as glasses (Fig. 16). Intermolecular hydrogen bonding leads to the
formation of hexagonal junctions which are formed from three, four, five, or six molecules as a result of the dimensions of the mole-
cule. Consequently, the structure forms an extremely rare example of a random, entropically stabilized, rhombus tiling. It is impor-
tant to note that the detailed structure of this framework, and other related frameworks, can only be appreciated with a molecular
level understanding and that this can only be readily achieved by scanning probe microscopies. The degree of randomness of the
rhombus tiling can be evaluated by detailed analysis of each image and subsequent studies demonstrated that both solvent used for
deposition and temperature of experiment can affect how random the tiling structure is.84,85 It is important to note that the
rhombus tiling arrangement is only observed for molecules which have the appropriate dimensions, that is, those which are
rhombus shaped, and analogous molecules, such as quaterphenyl-3,3000 ,5,5000 -tetracarboxylic acid, form regular two-dimensional
structures.86
Two-Dimensional Supramolecular Chemistry on Surfaces 195

Figure 16 (A) STM image of a typical area of terphenyl-3,300 ,5,500 -tetracarboxylic acid network at the nonanoic acid/HOPG interface. The group of
three phenyl rings of the molecule backbone appear as bright features in the image. The hexagonal orientational order of the structure is indicated by
the group of blue dots in the lower right-hand corner of the image, marking the location of pores in the network. (B) Illustration of how the molecular
arrangement, and each molecule, maps onto a rhombus tiling. (C) Diagrams representing the five possible arrangements of terphenyl-3,300 ,5,500 -tetra-
carboxylic acid molecules around a network pore, accompanied by magnified STM image examples of each pore type. The locations of the magnified
regions are marked (A) by blue dashed squares.83 Reproduced with permission from the AAAS.

The rhombus tiling observed with terphenyl-3,300 ,5,500 -tetracarboxylic acid is related to a Penrose tiling87; however, the first
observation of a structure more closely related to such a tiling structure was reported for the self-assembled supramolecular structure
formed by ferrocenecarboxylic acid [Fc(COOH)] on a Au(111) substrate.88 Penrose tilings are related to quasicrystal structures in
that they exhibit long range, nonperiodic order, and unusual rotational symmetry. Fc(COOH) assembles through intermolecular
hydrogen bonding to form pentagonal arrangements of molecules (Fig. 17) with carboxylic acid:carboxylic acid OeH/O interac-
tions being additionally stabilized by CeH/O hydrogen bonds between adjacent molecules. Importantly the cyclic pentamer
observed for the surface-based array was found to be more stable than other potential hydrogen-bonding arrangements, such as
dimeric systems, by Density Functional Theory (DFT) calculations. It was found that a related compound ferroceneacetic acid
[Fc(CH2COOH)] does not form the pentagonal arrangements, adopting a more conventional dimeric arrangement. This can be
accounted for by the observation that the length of the additional methylene group, in comparison to Fc(COOH), prevents
interactions between the carboxyl oxygen atom of one carboxylic acid and the hydrogen atoms on an adjacent cyclopentadienyl
fragment. The pentagonal arrangements observed are related to subunits of the Penrose P1 tiling.87
The search for quasicrystalline structures using molecules has also led to the discovery of a two-dimensional framework built
from Europium centers linked through metal–ligand coordination bonds to the bridging ligand para-quaterphenyl-dicarbonitrile
linker p-NC-Ph4-p-CN.89 Self-assembly on a Au(111) substrate followed by STM imaging reveals the formation of a quasicrystalline
two-dimensional structure. It is clear from the observed structure that due to a combination of fourfold, fivefold, and sixfold
coordination of the Eu centers, a structure is formed which represents an irregular tiling of both triangles and squares (Fig. 18).
Finally, a highly unusual surface-based structure has been reported for the angular molecules 4,4-dibromo-1,1:3,1-terphenyl
(B3PB) and 4,4-dibromo-1,1:3,1:4,1-quaterphenyl (B4PB).90 Both B3PB and B4PB are functionalized on their termini with
bromo-aryl moieties which are instrumental in forming intermolecular halogen bonding interactions (Fig. 19). These interactions
196 Two-Dimensional Supramolecular Chemistry on Surfaces

Figure 17 (A) Schematic representation of the pentagonal arrangement formed by Fc(COOH) through intermolecular hydrogen bonding;
(B) pentagonal arrangements can be clearly seen in STM images of Fc(COOH) adsorbed on a Au(111) substrate.88 Reproduced with permission from
the Nature Publishing Group.

Figure 18 Dodecagonal motifs observed in the quasicrystalline structure formed by p-NC-Ph4-p-CN and Eu centers on a Au(111) substrate.89 Both
triangles and squares can be readily identified, the different colors showing the alternative arrangements and orientations of these features with respect
to the underlying surface. Reproduced with permission from the Nature Publishing Group.
Two-Dimensional Supramolecular Chemistry on Surfaces 197

Figure 19 (A) Molecular structures of B3PB and B4PB with dimensions. (B) High resolution STM images of B3PB (13  11 nm) and (C) B4PB
(33  29 nm) illustrating the formation of Serpinski triangles by self-assembly of the molecules through halogen bonding.90 Reproduced with permis-
sion from the Nature Publishing Group.

result in the formation of extended structures which share the same topological arrangement as Serpinski triangles which are fractal
structures. The formation of the fractal arrangements is truly remarkable and it is clear that the energy balance between such unusual
structures and other potential arrangements is very finely balanced. As with the rhombus and Penrose tilings, it is important to note
that a complete appreciation of the structures formed is only possible due to a molecular level understanding of the self-assembled
arrangement. This can only be readily achieved using scanning probe microscopies and it is perhaps for this reason that such
structures are found in a surface supramolecular frameworks.

2.09.6 Conclusions

It can be clearly seen from the systems described above that surface-based supramolecular chemistry has a rich future building upon
an understanding of the interplay between different intermolecular forces and through adsorbate–substrate interactions. The ability
to create well-ordered, two-dimensional frameworks shows great promise for the development of bespoke materials. Thus far
hydrogen-bonded frameworks have received a great deal of attention, as have those structures that rely on van der Waals interactions
to control framework formation, but increasingly the principles of two-dimensional framework formation are being applied to
covalently bonded structures. Traditional host–guest chemistry is also feasible using surface-based frameworks opening up
strategies of research for the preparation of nanoscale devices.
Although many of the approaches that are employed by supramolecular chemists in either the solution phase or solid state are
generally applicable to surface-based processes, significant differences are also evident. Notably, the surface does not play a passive
role in the two-dimensional self-assembly process. For molecules to adsorb on the surface there is inherently an interaction between
substrate and the molecule and this can lead to subtle differences even affecting the conformations of molecules91 and the way in
which they self-assemble. As the area of surface-based self-assembled frameworks continues to develop, it is important that the role
of the substrate becomes increasingly understood and ultimately exploited to control self-assembly. In no area is this more impor-
tant than in the formation of covalently bonded surface frameworks for which the substrate often plays an integral role in the forma-
tion of the covalent bond.
This article demonstrates that a number of successful strategies have already been developed for the synthesis of surface-based
framework structures and some of these approaches have led to the discovery of highly unusual structures based upon unusual tiling
processes, frameworks which are unlikely to be discovered using other approaches to supramolecular chemistry. Although it is clear
that further reliable pathways to robust frameworks still need to be developed, it is also clear that significant progress has already
been made in the field. Now that significant success in synthetic strategies has been achieved, the first few steps towards functional
materials are already underway.
198 Two-Dimensional Supramolecular Chemistry on Surfaces

References

1. Lehn, J. M. Proc. Natl. Acad. Sci. USA 2002, 99, 4763.


2. Reinhoudt, D. N.; Crego-Calama, M. Science 2002, 295, 2403.
3. Aakeröy, C.; Champness, N. R.; Janiak, C. CrystEngComm 2010, 12, 22.
4. Bartels, L. Nat. Chem. 2010, 2, 87.
5. Bonifazi, D.; Mohnani, S.; Llanes-Pallas, A. Chem. Eur. J. 2009, 15, 7004.
6. Plass, K. E.; Grzesiak, A. L.; Matzger, A. J. Acc. Chem. Res. 2007, 40, 287.
7. De Feyter, S.; Gesquière, A.; Abdel-Mottaleb, M. M.; Grim, P. C. M.; De Schryver, F. C. Acc. Chem. Res. 2000, 33, 520.
8. De Feyter, S.; De Schryver, F. C. Chem. Soc. Rev. 2003, 32, 139.
9. Kudernac, T.; Lei, S.; Elemans, J. A. A. W.; De Feyter, S. Chem. Soc. Rev. 2009, 38, 402.
10. Slater (née Phillips), A. G.; Beton, P. H.; Champness, N. R. Chem. Sci. 2011, 2, 1440–1448.
11. Vericat, C.; Vela, M. E.; Benitez, G.; Carro, P.; Salvarezza, R. C. Chem. Soc. Rev. 2010, 39, 1805.
12. Narasimhan, S.; Vanderbilt, D. Phys. Rev. Lett. 1992, 69, 1564.
13. Upward, M. D.; Moriarty, P.; Beton, P. H. Phys. Rev. B 1997, 56, R1704.
14. Ruben, M.; Lehn, J. M.; Muller, P. Chem. Soc. Rev. 2006, 35, 1056.
15. Coraux, J.; N’Diaye, A. T.; Busse, C.; Michely, T. Nano Lett. 2008, 8, 565; Sutter, P. W.; Flege, J.-I.; Sutter, E. A. Nat. Mater. 2008, 7, 406.
16. Berner, S.; Corso, M.; Widmer, R.; Groening, O.; Laskowski, R.; Blaha, P.; Schwarz, K.; Goriachko, A.; Over, H.; Gsell, S.; Schreck, M.; Sachdev, H.; Greber, T.; Osterwalder, J.
Angew. Chem. Int. Ed. 2007, 46, 5115.
17. Korolkov, V. V.; Svatek, S. A.; Allen, S.; Roberts, C. J.; Tendler, S. J. B.; Taniguchi, T.; Watanabe, K.; Champness, N. R.; Beton, P. H. Chem. Commun. 2014, 50, 8882–8885.
18. Gross, L.; Mohn, F.; Moll, N.; Liljeroth, P.; Meyer, G. Science 2009, 325, 1110–1114.
19. Whitesides, G. M.; Simanek, E. E.; Mathias, J. P.; Seto, C. T.; Chin, D. N.; Mammen, M.; Gordon, D. M. Acc. Chem. Res. 1995, 28, 37.
20. Griessl, S.; Lackinger, M.; Edelwirth, M.; Hietschold, M.; Heckl, W. M. Single Mol. 2002, 3, 25.
21. Etter, M. C. Acc. Chem. Res. 1990, 23, 120–126.
22. Gutzler, R.; Sirtl, T.; Dienstmaier, J. F.; Mahata, K.; Heckl, W. M.; Schmittel, M.; Lackinger, M. J. Am. Chem. Soc. 2010, 132, 5084.
23. Keeling, D. L.; Oxtoby, N. S.; Wilson, C.; Humphry, M. J.; Champness, N. R.; Beton, P. H. Nano Lett. 2003, 3, 9.
24. Sweetman, A. M.; Jarvis, S.; Sang, H.; Lekkas, I.; Rahe, P.; Wang, Y.; Wang, J.; Champness, N. R.; Kantorovich, L.; Moriarty, P. Nat. Commun. 2014, 5, 3931.
25. Sweetman, A. M.; Jarvis, S. P.; Champness, N. R.; Rahe, P.; Kantorovich, L.; Moriarty, P. Phys. Rev. B 2014, 90, 165425.
26. Jarvis, S. P.; Sweetman, A. M.; Lekkas, I.; Champness, N. R.; Kantorovich, L.; Moriarty, P. J. Phys. Condens. Matter 2015, 27, 054004.
27. Jarvis, S. P.; Rashid, M. A.; Sweetman, A.; Leaf, J.; Taylor, S.; Moriarty, P.; Dunn, J. Phys. Rev. B 2015, 92, 241405.
28. Seeman, N. C. Nature 2003, 421, 427.
29. Edwardson, T. G. W.; Lau, K. L.; Bousmail, D.; Sleiman, H. Nat. Chem. 2016, 8, 162–170; Mirkin, C. A.; Letsinger, R. L.; Mucic, R. C.; Storhoff, J. J. Nature 1996, 382,
607–609.
30. Winfree, E.; Liu, F.; Wenzler, L. A.; Seeman, N. C. Nature 1998, 394, 539.
31. He, Y.; Chen, Y.; Liu, H.; Ribbe, A. E.; Mao, C. J. Am. Chem. Soc. 2005, 127, 12202.
32. Otero, R.; Schock, M.; Molina, L. M.; Laegsgaard, E.; Stensgaard, I.; Hammer, B.; Besenbacher, F. Angew. Chem. Int. Ed. 2005, 44, 2270.
33. Ciesielski, A.; Perone, R.; Pieraccini, S.; Piero Spada, G.; Samorì, P. Chem. Commun. 2010, 46, 4493.
34. Ciesielski, A.; Haar, S.; El Garah, M.; Surin, M.; Masiero, S.; Samorì, P. L’Actual. Chim. 2015, 399, 31–36.
35. El Garah, M.; Perone, R. C.; Santana Bonilla, A.; Haar, S.; Campitiello, M.; Gutierrez, R.; Cuniberti, G.; Masiero, S.; Ciesielski, A.; Samorì, P. Chem. Commun. 2015, 51,
11677–11680.
36. Ciesielski, A.; El Garah, M.; Masiero, S.; Samorì, P. Small 2016, 12, 83–95.
37. Slater, A. G.; Hu, Y.; Yang, L.; Argent, S. P.; Lewis, W.; Blunt, M. O.; Champness, N. R. Chem. Sci. 2015, 6, 1562–1569.
38. Elemans, J. A. A. W.; De Cat, I.; Xu, H.; De Feyter, S. Chem. Soc. Rev. 2009, 38, 722–736.
39. Bilbao, N.; Destoop, I.; De Feyter, S.; González-Rodríguez, D. Angew. Chem. Int. Ed. 2016, 55, 659–663.
40. Zerkowski, J. A.; Mathias, J. P.; Whitesides, G. M. J. Am. Chem. Soc. 1994, 116, 4305.
41. Slater, A. G.; Perdigao, L. M. A.; Beton, P. H.; Champness, N. R. Acc. Chem. Res. 2014, 47, 3417–3427.
42. Theobald, J. A.; Oxtoby, N. S.; Phillips, M. A.; Champness, N. R.; Beton, P. H. Nature 2003, 424, 1029–1031.
43. Theobald, J. A.; Oxtoby, N. S.; Champness, N. R.; Beton, P. H.; Dennis, T. J. S. Langmuir 2005, 21, 2038.
44. Silly, F.; Shaw, A. Q.; Porfyrakis, K.; Warner, J. H.; Watt, A. A. R.; Castell, M. R.; Umemoto, H.; Akachi, T.; Shinohara, H.; Briggs, G. A. D. Chem. Commun. 2008, 4616.
45. Perdigão, L. M. A.; Perkins, E. W.; Ma, J.; Staniec, P. A.; Rogers, B. L.; Champness, N. R.; Beton, P. H. J. Phys. Chem. B 2006, 110, 12539.
46. Staniec, P. A.; Perdigão, L. M. A.; Saywell, A.; Champness, N. R.; Beton, P. H. ChemPhysChem 2007, 8, 2177.
47. Perdigão, L. M. A.; Staniec, P. A.; Champness, N. R.; Beton, P. H. Langmuir 2009, 25, 2278.
48. Madueno, R.; Räisänen, M. T.; Silien, C.; Buck, M. Nature 2008, 454, 618.
49. Silien, C.; Räisänen, M. T.; Buck, M. Small 2010, 6, 391.
50. Silien, C.; Räisänen, M. T.; Buck, M. Angew. Chem. Int. Ed. 2009, 48, 3349.
51. Lombana, A.; Battaglini, N.; Tsague-Kenfac, G.; Zrig, S.; Lang, P. Chem. Commun. 2016, 52, 5742–5745.
52. Lombana, A.; Rinfray, C.; Volatron, F.; Izzet, G.; Battaglini, N.; Alves, S.; Decorse, P.; Lang, P.; Proust, A. J. Phys. Chem. C 2016, 120, 2837–2845.
53. Saywell, A.; Magnano, G.; Satterley, C. J.; Perdigão, L. M. A.; Champness, N. R.; Beton, P. H.; O’Shea, J. N. J. Phys. Chem. C 2008, 112, 7706–7709.
54. Saywell, A.; Magnano, G.; Satterley, C. J.; Perdigão, L. M. A.; Britton, A. J.; Taleb, N.; Giménez-López, M. C.; Champness, N. R.; O’Shea, J. N.; Beton, P. H. Nat. Commun.
2010, 1, 75.
55. Pollard, A. J.; Perkins, E. W.; Smith, N. A.; Saywell, A.; Goretzki, G.; Phillips, A. G.; Argent, S. P.; Sachdev, H.; Müller, F.; Hüfner, S.; Gsell, S.; Fischer, M.; Schreck, M.;
Osterwalder, J.; Greber, T.; Berner, S.; Champness, N. R.; Beton, P. H. Angew. Chem. Int. Ed. 2010, 49, 1794–1799.
56. Perdigão, L. M. A.; Saywell, A.; Fontes, G. N.; Staniec, P. A.; Goretzki, G.; Phillips, A. G.; Champness, N. R.; Beton, P. H. Chem. Eur. J. 2008, 14, 7600.
57. Räisänen, M. T.; Slater (née Phillips), A. G.; Champness, N. R.; Buck, M. Chem. Sci. 2012, 3, 84–92.
58. Phillips, A. G.; Perdigão, L. M. A.; Beton, P. H.; Champness, N. R. Chem. Commun. 2010, 46, 2775.
59. Lee, S.-L.; Adisoejoso, J.; Fang, Y.; Tahara, K.; Tobe, Y.; Mali, K. S.; De Feyter, S. Nanoscale 2015, 7, 5344–5349.
60. Berdunov, N.; Pollard, A. J.; Beton, P. H. Appl. Phys. Lett. 2009, 94, 043110.
61. Perdigão, L. M. A.; Champness, N. R.; Beton, P. H. Chem. Commun. 2006, 5, 538.
62. Staniec, P. A.; Perdigão, L. M. A.; Rogers, B. L.; Champness, N. R.; Beton, P. H. J. J. Phys. Chem. C 2007, 111, 886–893.
63. Palma, C.-A.; Bjork, J.; Bonini, M.; Dyer, M. S.; Llanes-Pallas, A.; Bonifazi, D.; Persson, M.; Samorì, P. J. Am. Chem. Soc. 2009, 131, 13062.
64. Wieghold, S.; Li, J.; Simon, P.; Krause, M.; Avlasevich, Y.; Li, C.; Garrido, J. A.; Heiz, U.; Samori, P.; Mullen, K.; Esch, F.; Barth, J. V.; Palma, C.-A. Nat. Commun. 2016, 7,
10700.
Two-Dimensional Supramolecular Chemistry on Surfaces 199

65. Mali, K. S.; Adisoejoso, J.; Ghijsens, E.; De Cat, I.; De Feyter, S. Acc. Chem. Res. 2012, 45, 1309–1320.
66. Mali, K. S.; De Feyter, S. Phil. Trans. R. Soc. A 2013, 371, 20120304.
67. Schull, G.; Douillard, L.; Fiorini-Debuisschert, C.; Charra, F. Nano Lett. 2006, 6, 1360.
68. Tahara, K.; Okuhata, S.; Adisoejoso, J.; Lei, S.; Fujita, T.; De Feyter, S.; Tobe, Y. J. Am. Chem. Soc. 2009, 131, 17583.
69. Elemans, J. A. A. W.; Lei, S. B.; De Feyter, S. Angew. Chem. Int. Ed. 2009, 48, 7298.
70. Lei, S.; Surin, M.; Tahara, K.; Adisoejoso, J.; Lazzaroni, R.; Tobe, Y.; De Feyter, S. Nano Lett. 2008, 8, 2541.
71. Lei, S.; Tahara, K.; De Schryver, F. C.; Van der Auweraer, M.; Tobe, Y.; De Feyter, S. Angew. Chem. Int. Ed. 2008, 47, 2964.
72. Furukawa, S.; Uji-i, H.; Tahara, K.; Ichikawa, T.; Sonoda, M.; De Schryver, F. C.; Tobe, Y.; De Feyter, S. J. Am. Chem. Soc. 2006, 128, 3502.
73. Tahara, K.; Furukawa, S.; Uji-i, H.; Uchino, T.; Ichikawa, T.; Zhang, J.; Sonoda, M.; De Schryver, F. C.; De Feyter, S.; Tobe, Y. J. Am. Chem. Soc. 2006, 128, 16613.
74. Grill, L.; Dyer, M.; Lafferentz, L.; Persson, M.; Peters, M. V.; Hecht, S. Nat. Nanotechnol. 2007, 2, 687.
75. Lafferentz, L.; Eberhardt, V.; Dri, C.; Africh, C.; Comelli, G.; Esch, F.; Hecht, S.; Grill, L. Nat. Chem. 2012, 4, 215–220.
76. Champness, N. R. Nat. Chem. 2012, 4, 149–150.
77. Blunt, M. O.; Russell, J. C.; Champness, N. R.; Beton, P. H. Chem. Commun. 2010, 46, 7157–7159.
78. Klappenberger, F.; Zhang, Y.-Q.; Bjork, J.; Klyatskaya, S.; Ruben, M.; Barth, J. V. Acc. Chem. Res. 2015, 48, 2140–2150.
79. Clair, S.; Abel, M.; Porte, L. Chem. Commun. 2014, 50, 9627–9635.
80. Ding, S.-Y.; Wang, W. Chem. Soc. Rev. 2013, 42, 548–568.
81. Haq, S.; Hanke, F.; Sharp, J.; Persson, M.; Amabilino, D. B.; Raval, R. ACS Nano 2014, 8, 8856–8870.
82. Haq, S.; Wit, B.; Sang, H.; Floris, A.; Wang, Y.; Wang, J.; Perez-García, L.; Kantorovitch, L.; Amabilino, D. B.; Raval, R. Angew. Chem. Int. Ed. 2015, 54, 7101–7105.
83. Blunt, M. O.; Russell, J.; Giménez-López, M. C.; Garrahan, J. P.; Lin, X.; Schröder, M.; Champness, N. R.; Beton, P. H. Science 2008, 322, 1077–1081.
84. Stannard, A.; Russell, J. C.; Blunt, M. O.; Sallesiotis, C.; Gimenez-Lopez, M. C.; Taleb, N.; Schroder, M.; Champness, N. R.; Garrahan, J. P.; Beton, P. H. Nat. Chem. 2012, 4,
112–117.
85. Blunt, M. O.; Russell, J. C.; Gimenez-Lopez, M. C.; Taleb, N.; Lin, X.; Schröder, M.; Champness, N. R.; Beton, P. H. Nat. Chem. 2011, 3, 74–78.
86. Blunt, M. O.; Lin, X.; Gimenez-Lopez, M. C.; Schröder, M.; Champness, N. R.; Beton, P. H. Chem. Commun. 2008, 2304.
87. Penrose, R. Eureka 1978, 39, 16–32.
88. Wasio, N. A.; Quardokus, R. C.; Forrest, R. P.; Lent, C. S.; Corcelli, S. A.; Christie, J. A.; Henderson, K. W.; Kandel, S. A. Nature 2014, 507, 86–89.
89. Urgel, J. I.; Écija, D.; Lyu, G.; Zhang, R.; Palma, C.-A.; Auwärter, W.; Lin, N.; Barth, J. V. Nat. Chem. 2016, 8, 657–662.
90. Shang, J.; Wang, Y.; Chen, M.; Dai, J.; Zhou, X.; Kuttner, J.; Hilt, G.; Shao, X.; Gottfried, J. M.; Wu, K. Nat. Chem. 2015, 7, 389–393.
91. Jarvis, S. P.; Taylor, S.; Baran, J. D.; Champness, N. R.; Larsson, J. A.; Moriarty, P. Nat. Commun. 2015, 6, 8338.
2.10 Quartz Crystal Microbalances: Chemical Applications
FL Dickert, University of Vienna, Vienna, Austria
U Latif, COMSATS Institute of Information Technology, Lahore, Pakistan
Ó 2017 Elsevier Ltd. All rights reserved.
Dedicated to Prof. Adolf Neckel on the occasion of his 90th birthday.

2.10.1 Piezoelectric Effect 201


2.10.2 Quartz as Piezoelectric Substance 202
2.10.3 Properties of a QCM 203
2.10.4 QCM in Air 204
2.10.5 QCM in Liquids 206
2.10.6 Applications 207
2.10.6.1 Measuring Setup 207
2.10.6.2 Detection of Molecules 207
2.10.6.3 Detection of Cells 209
2.10.6.4 Detection of Complex Mixtures 210
References 211

2.10.1 Piezoelectric Effect

Piezoelectricity is defined as the generation of electric potential difference in some solid materials by applying a mechanical stress
along an appropriate direction.1 The piezoelectric effect was discovered by Pierre and Jacques Curie in 1880, and it can be observed
in those materials, which possess a polar axis without a center of symmetry.2,3 Polar axes are those axes of symmetry which cannot
be transferred to the opposite direction by symmetry operations. When a stress is applied on certain crystal faces of a piezoelectric
material, charge centers shift across the crystal and turn it into a dipole as shown in Fig. 1. The intensity and direction of charge
transfer depend upon the direction of compression. If the crystal is not short-circuited, this charge shifting leads to the production
of electric voltage. The whole effect is called the piezoelectric effect.
At first, this phenomenon was known by the name of pyroelectricity, but later on, Hankel suggested the name “piezoelectricity”
in 1881.1 Afterward, Lippmann said that the converse piezoelectric effect should also exist, and Curie verified the presence of the
converse effect in 1881, which is similar to the direct piezoelectric effect.1Thus, piezoelectricity can be reversed, which is called the
converse piezoelectric effect. The converse piezoelectric effect is the production of mechanical deformation in the material when an
electric field is applied. The mechanical deformation induced in the material is proportional to the strength of the applied electric
field, and the direction of the induced strain can be changed by reversing the polarity of the electric field.
The creation of an electric field in the direction of a polar axis of the crystal leads to a mechanical deformation, and by applying
an alternating field across the crystal, a mechanical oscillation can be stimulated whose frequency can be measured using appro-
priate devices. There are a lot of applications based on this phenomenon, such as the production and detection of sound, generation
of high voltages, very stable electronic oscillators, sensors,4 phase shifts for pulsed lasers, and ultra fine focusing of optical assem-
blies. A quartz crystal microbalance (QCM) operates on the basis of the converse piezoelectric effect because mechanical oscillations
are produced in the piezoelectric crystal by using an oscillating potential difference (AC voltages).

Figure 1 A description of the piezoelectric effect in a quartz crystal (generation of dipoles by deforming forces).

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12506-5 201


202 Quartz Crystal Microbalances: Chemical Applications

2.10.2 Quartz as Piezoelectric Substance

The frequency range of acoustic waves varies from 10 2 to 1012 Hz, whereas the frequency range of acoustic resonators is very
narrow, from 106 to 109 Hz.5,6
The absence of a center of symmetry in the crystals is a prerequisite for piezoelectricity. Although a large number of crystals fulfill
this requirement, the unique properties of quartz crystals led to their commercial significance. Crystalline quartz is a piezoelectric
material, which has low damping, extremely high mechanical stability, and low temperature dependence. Different resonator types
can be obtained by cutting a quartz crystal at a specific angle, which determines modes of mechanical vibrations. Thickness shear
mode (TSM), face shear mode, and flexural mode resonators (FPW) can be obtained from a quartz crystal having frequencies
ranging from 5  102 to 3  108 Hz. QCMs are AT-cut crystals, which operate in TSM mode,7,8 and can be prepared by a cutting
quartz crystal at an angle of 35 100 to the optical z-axis.
The orientation of the cut of a quartz crystal decides its piezoelectric coefficient and its temperature dependency, as shown in
Fig. 2A. The optimum orientations are those at which the crystal experiences minimal temperature dependencies in a practical appli-
cation (operating temperatures). In sensing applications, an AT-cut of 35 1000 would be more suitable because the ambient temper-
ature range is 25–45 C. But in electronics, the ambient temperature range is higher, which is why an AT-cut of 35 1500 would be
more preferable because it exhibits minimum temperature coefficients between 50 C and 70 C. The sensitivity of QCM sensors
depends on electrode diameters as shown in Fig. 2B. In order to explain this, Sauerbrey used a 14 MHz AT-cut quartz crystal
and deposited electrodes of different diameters. The figure shows the relationship between different electrode diameters and
maximum sensitivity.
SAW or FPW is more sensitive than TSM resonators, but TSM is still widely used in different applications due to its low cost and
robustness.9 AT-cut quartz crystals are most suitable for developing QCM sensors because they offer stable frequency (Df/f  10 8)
and  zero temperature coefficient. Disk-shaped resonators are utilized for developing QCM sensors having a fundamental
frequency in the range from 5 to 40 MHz, in extreme cases up to 150 MHz. The fundamental frequency and thickness of a QCM
are inversely proportional to each other. The fundamental frequency of QCM increases with a decrease in the thickness of the quartz
crystal and vice versa. A 5 MHz crystal is 330 mm thick, whereas the thickness of a 30 MHz crystal is 55 mm. Electrodes are coated on
an AT-cut quartz resonator to apply a potential difference, which will produce shear waves of opposite polarity at the electrodes with
shear displacement in plane with the crystal surface as shown in Fig. 3.
A rigid deposition of mass loaded on the quartz surface will increase the resonator thickness, which in turn decreases the
frequency of the quartz crystal. In this way, a QCM behaves like a microbalance.10
Quartz (SiO2) and lithium tantalate (LiTaO3) are widely used piezoelectric materials to fabricate mass-sensitive sensors, whereas
lithium niobate (LiNbO3) is not so common. Gallium orthophosphate (GaPO4) can be used at high temperatures, which works up
to 900 C, and a-quartz is stable up to 573 C, and then, its piezoelectricity disappears. The use of these materials in a particular field
depends on some other properties such as cost, temperature dependence, attenuation, and propagation velocity. An interesting
property of quartz and some other materials is its ability to provide temperature dependence selectivity by using optimal cut angle
and wave propagation direction. The first-order temperature effect can be minimized with proper selection. Quartz crystals are more
popular among all piezoelectric materials because these are cheap and can withstand thermal, chemical, and mechanical stress.
Moreover, quartz can easily be processed to obtain single crystals having very low temperature dependence (AT-cut), which are
frequently used in chemical sensing.11

Figure 2 (A) Temperature dependencies of an AT-cut quartz crystal resonance frequency with typical cutting angles used for different applications.
(B) QCM differential mass sensitivity as a function of electrode radius. Adapted from Sauerbrey, G. Z. Phys. 1959, 155 (2), 206–222 and reprinted
with permission from Springer.
Quartz Crystal Microbalances: Chemical Applications 203

Figure 3 Quartz crystal coated with metal (gold) electrodes. An applied potential difference will produce shear waves of opposite polarity and
frequency shifts of crystal oscillations after mass loading.

2.10.3 Properties of a QCM

The electric properties of a QCM can be represented by the Butterworth-van Dyke equivalent circuit as shown in Fig. 4.7,12 The
impedance associated with the shear motion of quartz is represented as the combination of L, C, and R, whereas parallel capacitance
originates due to the presence of the dielectric quartz material between two surface electrodes (parallel plate capacitor) and contri-
bution of wiring and the crystal holder. The parallel capacitance determines admittance away from resonance. Thus, the entire
network is a combination of parallel and series resonant circuits. The combination of L, C, and R in an equivalent circuit dominates
total impedance of the resonator at frequencies close to resonance, whereas parallel capacitance dominates total impedance away
from resonance.
The ratio of voltage and current amplitudes (the magnitude of impedance) and the phase angle between both quantities are then
computed for various frequencies and subsequently analyzed. Measurement of impedance and phase angle at frequencies
embracing the TSM sensor’s resonance frequency is now conveniently accomplished by network analyzers. The damping spectra
of a 10 MHz QCM were obtained by using a network analyzer to analyze electronic properties of the crystal resonator as shown
in Fig. 5. This figure describes electronic damping of the QCM and phase shift as a function of frequency.
Serial and parallel resonances occur at a frequency of 10 and 10.02 MHz, respectively. Such frequency behavior can be explained
by an equivalent circuit in which a capacitor and an inductor are connected in series and further with a capacitor in parallel. In the
case of damping, an ohmic resistor is connected in series with a capacitor and an inductor. Thus, the equivalent circuit is comprised
of two oscillators, one with serial frequency and the other with parallel frequency, and the impedance of this equivalent circuit can
be described by the following equation (neglecting R):
 
j u2 LC  1
Zq ¼
u C0 þ C  u2 LCC0
The change in frequency of a resonator caused by mass loading on its electrodes depends on the fundamental frequency of the
resonator as described by Sauerbrey, who gave the following equation:
Df ¼ Cm f02 Dm

A quartz crystal resonator can be regarded as a very sensitive microbalance, for example, the sensitivity of a 10 MHz QCM is
1 Hz ng 1 of mass loading. The detection of mass by using acoustic resonators is the best way to develop a label-free sensor in
comparison with other methods because every analyte has mass.13

Figure 4 Butterworth-van Dyke equivalent circuits of a QCM resonator.


204 Quartz Crystal Microbalances: Chemical Applications

Figure 5 Damping spectra and shift in phase angle of 10 MHz QCM recorded by a network analyzer.

Figure 6 Damping spectrum of a QCM with resonance peaks, compared with a 10 pF capacitor recorded by a network analyzer. Adapted from
Dickert, F. L.; Lieberzeit, P. A. In Piezoelectric Sensors; Janshoff, A., Steinem, C., Eds.; Springer Berlin Heidelberg: Berlin, Heidelberg, 2007; pp
173–210 and reprinted with permission from Springer.

The electronic properties of a QCM can be explained from its damping spectra by using a network analyzer as shown in Fig. 6.
The figure shows electronic damping as a function of frequency. The damping of a QCM decreases with an increase in frequency,
and its baseline resembles the frequency behavior of a capacitor (10 pF) because, in a dual electrode QCM, the electrodes behave as
a capacitor having a quartz crystal as a dielectric insulator between them. In the QCM spectrum, resonance spikes can be seen at 10,
30, 50, 70, and 90 MHz. In chemical sensing, two electrodes are formed by coating a metal onto the quartz crystal surface. By
applying an AC electric field of 30 V via metal electrodes, the crystal undergoes thickness shear stress in which crystal surfaces oscil-
late against each other as shown in Fig. 7A.
An AFM image of a metal electrode on a quartz crystal at an applied voltage of 0 and 30 V is shown in the figure, which depicts
the gold surface displacement of 3 nm (Fig. 7B and C) when voltage is switched on. Gold metal electrodes can be adhered onto
a quartz crystal by screen printing and sputtering or via gold evaporation to fabricate a QCM. For this purpose, a quartz crystal
is placed on a teflon block, which is further attached with a vacuum pump to hold the crystal in place tightly. Afterward, a required
electrode pattern is screen-printed on both sides of the quartz crystal by using gold paste. Then, it is heated in an oven at 400 C
for 5 h.
The surface roughness of the screen-printed gold electrode can be reduced by spinning the screen-printed quartz crystal or by
exposing it to dichloromethane atmosphere for 1 min as shown in Fig. 8A. The damping spectra of a gold-coated quartz crystal
also corroborate these findings (Fig. 8B).

2.10.4 QCM in Air

An AC electric field (1–20 V cm 1) is applied to a quartz crystal by coating metal electrodes onto it, which will generate oscillations
in the crystal. These metal electrodes experience shear movement at 180 degrees with respect to each other. A piezoelectric displace-
ment in the form of a standing wave is generated beneath metal electrodes, and some boundary problems of course exist such as
Quartz Crystal Microbalances: Chemical Applications 205

Figure 7 (A) The production of a shear wave in a QCM after applying a potential difference of 30 V. (B) The displacement in the QCM can be clearly
seen by contact mode AFM images at 0 and 30 V. (C) The displacement in the QCM was 3 nm as measured by AFM. Adapted from Dickert, F. L.;
Lieberzeit, P. A. In Piezoelectric Sensors; Janshoff, A., Steinem, C., Eds.; Springer Berlin Heidelberg: Berlin, Heidelberg, 2007; pp 173–210 and reprin-
ted with permission from Springer.

Figure 8 (A) Surface roughness measured by contact mode AFM of a screen-printed gold electrode: untreated gold electrode, spun at 6000 rpm,
and treated in a CH2Cl2 environment. (B) Damping spectra of a QCM having screen-printed gold electrodes: untreated gold electrode, spunat
6000 rpm, and treated in a CH2Cl2 environment. Adapted from Lieberzeit, P. A.; Glanznig, G.; Jenik, M.; Gazda-Miarecka, S.; Dickert, F. L.; Leidl, A.
Sensors 2005, 5 (12), 509–518 and reprinted with permission from MDPI.

cross talk, which shows that the electric field is not only confined to the gold electrodes. The fundamental frequency of the crystal
also depends on the thickness of the metal electrodes. The crystal will oscillate only when the metal is at least 50 nm thick to guar-
antee high electrode conductivity. It will cease to oscillate when the metal is too thick because of damping. The electric field applied
to the crystal will excite two modes, that is, transverse and longitudinal, but only the transverse standing wave needs to be consid-
ered if the thickness of crystal is in the range of 100–300 mm, and small side bands in the spectrum are due to torsional movements.
The piezoelectric displacement of a point on crystal surface is a Gaussian function of radial distance from the center of the electrode.
This displacement is dependent on the strength of the electric field and ranges from a few nanometer in water to tens of nanometer
in air or vacuum.
In the gas phase, a selective layer is deposited on one or two electrodes of the QCM for using it in chemical sensing (damping of
the crystal will also increase by coating layers on two electrodes). In this way, three interfaces exist: between crystal and metal,
206 Quartz Crystal Microbalances: Chemical Applications

between metal and selective layer, and between selective layer and analyte. These interfaces will generate reflection and refraction of
acoustic energy, and the whole structure must be treated as a multiple resonator. When a polymer layer (selective layer) is deposited
on the metal (Au) electrode of the QCM, then most important interface is that between the polymer and Au.
There are number of effects such as the hydrostatic effect (p), the frictional effect (x), and the sorption effect (m), which influence
the response of a piezoelectric sensor in the gas phase. The hydrostatic and frictional effects are regarded as nonspecific interferences,
whereas the sorption effect is the basis of chemical sensing for a QCM, which originates from interactions with an analyte of interest.
The hydrostatic effect is important when the QCM is exposed to liquid analytes. These effects can be described for an AT-cut quartz
crystal at 50 C as14
     
DF DF DF DF
 ¼ þ þ
F F p F x F m
   1=2
DF
  106 ¼ 1:35103 Ptorr þ 7:2103 pFrg hg þ 2:26 FDM
F
where F is the frequency of the crystal, P the gas pressure, rg the gas density, hg the gas viscosity, and M the mass in grams per unit
area.
The change in frequency of a 5 MHz crystal due to hydrostatic and frictional effects is  5 Hz when it is exposed from air (1 atm)
to vacuum ( 10 6 mmHg).

2.10.5 QCM in Liquids

In gas-phase analysis, the temperature dependence on the frequency signal of an AT-cut quartz crystal is very minimal (only a few Hz
in the case of quartz), and sensor response is mainly due to mass loading. However, liquids are viscous and have to be treated very
differently from gases.15 In the liquid phase, the temperature dependence of a dual electrode 10 MHz QCM (immersed in water) is
very high as shown in Fig. 9. The compensations of these temperature fluctuations should be done on the same quartz sheet other-
wise the cut of the crystal will vary.
The frequency shift as a function of temperature changes from 15 to 25 C is depicted in Fig. 9. An average frequency shift of
40 Hz was observed for every 1 C rise in temperature. This temperature dependence is very high in comparison with gas-phase
measurements. The change in temperature varies viscosity of liquids; as a result, a frequency shift of the quartz crystal is observed.
An oscillating quartz crystal interacts with the surrounding liquid and partly drags the molecules along with itself during oscillation.
Therefore, an oscillating quartz surface is decelerated due to the presence of viscous media in its surroundings. The higher the
viscosity of the liquid (in which the quartz is immersed), the larger would be the damping on the quartz and the negative frequency
shifts. One way to overcome this problem while using a QCM in chemical sensing is to coat two metal electrodes and to use one
electrode for sensing (by coating a sensitive layer) and other one as a reference. Edge effects can be eliminated by making one elec-
trode smaller than the other to reduce the effects of electric fields in case of ionic solutions.
In order to study the damping and phase shifts of a quartz crystal in liquid media, a dual electrode 10 MHz QCM was used and
immersed from air to liquid as shown in Fig. 10. The damping of the quartz crystal was increased after totally immersing it in
heptane. A change of almost 15 dB was observed in serial resonance, whereas a change of 25 dB occurred at parallel resonance

Figure 9 Frequency shift as a function of temperature changes from 15 C to 25 C of a dual electrode QCM immersed in water. Adapted from
Lieberzeit, P. A.; Glanznig, G.; Jenik, M.; Gazda-Miarecka, S.; Dickert, F. L.; Leidl, A. Sensors 2005, 5 (12), 509–518 and reprinted with permission
from MDPI.
Quartz Crystal Microbalances: Chemical Applications 207

Figure 10 Damping and phase shifts of a dual electrode 10 MHz QCM while immersing from air to liquid media.

by immersing the QCM in heptane. Similarly, a change of 20 degrees in the phase angle was also observed when a dual electrode
QCM was immersed totally in liquid media.

2.10.6 Applications

The applications of a QCM are based on the same physical principle (as discussed earlier), that is, the change in the resonance
frequency of the quartz resonator due to mass accumulation on the QCM surface. The objective of the QCM sensor is to quantify
the amount of specific analyte in a sample, which can be achieved by coating a recognition layer on the electrodes of the QCM. This
quantitative analysis can be achieved from experimental frequency shifts of QCMs due to mass loading of an analyte based on the
Sauerbrey equation, that is, there is a linear relationship between the mass and the frequency shifts obtained.

2.10.6.1 Measuring Setup


A dual electrode QCM of frequency 10 MHz having a diameter of 15.5 mm and a thickness of 0.168 mm can be used for mass-
sensitive measurements. These electrodes are favorably screen-printed onto a single AT-cut quartz crystal disk using a gold paste.
These mass-sensitive devices are capable of detecting analytes down to nanogram levels. Measurements can be carried out by placing
the dual electrode QCM in a microfluidic chamber (capable of accommodating 140 mL or less volume of analyte), which reduces
sensor response time by minimizing the analyte diffusion path. This microfluidic chamber is made of polydimethylsiloxane
(PDMS) layers, and quartz resonators are sandwiched in between. In this setup, the upper layer of the QCM is exposed to analytes,
whereas the bottom layer is filled with air in order to facilitate proper oscillations. The sensor response in terms of serial resonance
of the QCM can be monitored by customized setups or network analyzers. The interface electronic systems are beyond the scope of
this article, and the interested reader is referred to Arnau.16

2.10.6.2 Detection of Molecules


A highly efficient label-free QCM sensor system, having a molecular imprinted polymer as the sensitive layer, was designed to detect
endocrine-disrupting chemicals known as estradiols.17 The QCM sensor was very selective to estradiol and shows a very low sensi-
tivity to ethinylestradiol, estradiol benzoate, and bisphenol A. Similarly, ethanol and ethyl acetate can be differentiated by using
a QCM sensor, which was coated with imprinted polyurethane having optimized phloroglucinol as cross linker. An anthracene-
imprinted polyurethane layer shows sensor effects to chrysene and perylene and can easily differentiate between them. This
mass-sensitive QCM sensor exhibits higher sensor response to chrysene in comparison with perylene. A double-imprinted approach
was also followed by utilizing perylene and naphthalene for the fabrication of a QCM sensor in order to detect anthracene.18 A
QCM sensor was also used for the detection of the noble gas xenon by coating the electrodes with calixarene hosts and hydrophilic
polyester as shown in Fig. 11.19 Methylated calixarenes show higher sensor response toward xenon with good response time and
reversibility in comparison with other calixarenes and polyesters.
A double-imprinting approach was followed to fabricate a QCM sensor device for detecting atrazine.20 In the double-imprinting
approach, cavities were generated on nanoparticles by using natural antibodies, and afterward, an antibody replica was produced by
surface imprinting the polymer layer with these nanoparticles. The QCM sensor exhibits a higher sensor response to its templated
analyte and is much less sensitive to all other atrazine structural analogs. An array of QCMs were used for the detection of o-xylene,
m-xylene, and p-xylene.21 These QCMs were coated with cyclodextrins and calixarenes, enabling them to differentiate between
isomeric forms of xylene. The silylated cyclodextrin forms an elongated cone of host molecules, which is more suitable for p-xylene,
whereas the bulky nature of m-xylene hindered its accessibility to the available cavities in comparison with the nonsilylated
208 Quartz Crystal Microbalances: Chemical Applications

Figure 11 Detection of xenon using a dual electrode QCM by coating one electrode with a calixarene while other uncoated electrode is used as
a reference. Adapted from Hayden, O.; Latif, U.; Dickert, F. L. Aust. J. Chem. 2012, 64 (12), 1628–1632.

Figure 12 10 MHz QCM sensor responses to solvent vapors in air by using chloroform and tetrahydrofuran-templated polymers as sensor layers.
Adapted from Dickert, F. L.; Thierer, S. Adv. Mater. 1996, 8 (12), 987–990.

cyclodextrins. Different solvent vapors were also detected by designing the cavities in polyurethane via solvent imprinting. Here,
chloroform and tetrahydrofuran were used for polyurethane imprinting. A dual electrode QCM was also utilized by coating sensor
layers, which were polymerized from different solvents, for example, chloroform (CHCl3) and tetrahydrofuran (THF).22 These
imprinted polymers show different sensitivity patterns toward templated analytes. THF-imprinted polymers exhibit higher sensor
responses to its templated analyte THF in comparison with CHCl3 as shown in Fig. 12.
In the next step, a trielectrode QCM sensor array was fabricated by screen printing for detecting chrysene and perylene. This
sensor array reduces the cost and time of analysis. Two electrodes of sensor array were coated with imprinted layers templated
with anthracene and chrysene, whereas the third electrode was coated with a nonimprinted layer for reference. There is a structurally
minute difference between chrysene and perylene, but selectivity of the sensor device was achieved by following the molecular
imprinting approach as shown in Fig. 13.

Figure 13 10 MHz QCM sensor array response to chrysene and perylene by sensor layers having anthracene and chrysene as imprints. Adapted
from Dickert, F. L.; Tortschanoff, M.; Bulst, W. E.; Fischerauer, G. Anal. Chem. 1999, 71 (20), 4559–4563.
Quartz Crystal Microbalances: Chemical Applications 209

2.10.6.3 Detection of Cells


QCM resonators were used for detecting red blood cells by using an erythrocyte-imprinted polymer coating.23 The mass-sensitive
sensors were capable of differentiating between subgroups of red blood cells. Selective recognition sites for erythrocytes were
produced in polyvinylpyrrolidones by following surface imprinting technique. Subgroups of red blood cells A1 and A2 were selec-
tively recognized, which are inherited by alleles on the ABO locus as shown in Fig. 14. These are the mainly prevailing subgroups of
the ABO human blood group system. Different antigenic patterns are present on these subgroups, which were transferred to the
polymer by surface imprinting. In this way, a QCM sensor coated with this biomimetic material was successfully able to differentiate
between subgroups.
A dual electrode QCM yeast sensor was fabricated by coating one gold electrode with a yeast-imprinted polymer, whereas a non-
imprinted polymer was adhered to the second electrode.24 The sensitive layer exhibits pronounced sensor response toward the tem-
plated analyte, and the reference electrode remains almost insensitive as shown in Fig. 15.
The figure also shows an AFM image of the imprinted layer onto which a pattern has been produced by templating it with yeast
cells. Singlet and duplex yeast cells were also selectively recognized by using QCM sensors.25 Synchronization of Saccharomyces cer-
evisiae (yeast cells) was carried out by using N-hydroxyurea, which stops the cell division at the synthesis phase by inhibiting its DNA
replication after budding. At the synthesis phase, yeast cells exist in duplex form, which is a bit smaller than two single yeast cells. A
recognition pattern was introduced in the polyurethane layer by templating it with duplex yeast cells. The duplex yeast cell-
imprinted QCM sensor shows negative frequency shift (Sauerbrey effect) because the analyte can fit tightly into the cavities, whereas
an increase in frequency (anti-Sauerbrey effect) was observed toward single yeast cells. The single yeast cells can enter into the

Figure 14 Mass-sensitive effects of 10 MHz QCM to different blood groups coated with erythrocyte A1-imprinted polymer.

Figure 15 10 MHz QCM sensor response to yeast cells by yeast-imprinted sensor layer. Adapted from Hayden, O.; Dickert, F. L. Adv. Mater. 2001,
13 (19), 1480–1483.
210 Quartz Crystal Microbalances: Chemical Applications

Figure 16 HRV-imprinted 10 MHz QCM sensor response toward human rhinovirus (HRV) and foot-and-mouth disease virus (FMDV). Adapted from
Jenik, M.; Schirhagl, R.; Schirk, C.; Hayden, O.; Lieberzeit, P. A.; Blaas, D.; Paul, G.; Dickert, F. L. Anal. Chem. 2009, 81 (13), 5320–5326.

cavities generated by the duplex yeast cells but are not able to bind tightly to the surface, which becomes a basis for the anti-
Sauerbrey effect. In this way, a QCM sensor was able to differentiate between singlet and duplex yeast cells.26
Escherichia coli bacteria was detected by using a QCM as a transducer having a sensitive layer imprinted with bacteria cells.27
Reproducibility of the mass-sensitive sensor was ensured by using artificial rather than natural cells. These artificial cells or replicas
were prepared by using natural bacteria cells and following the double-imprinting approach. In this process, natural cells were used
to generate cavities on the polymer via surface imprinting, and then, these cavities were filled with PDMS. In this way, the properties
of natural bacteria cells were transferred to the polymer, and later on, these artificial cells were used for generating a bacteria recog-
nition layer via surface imprinting. The selectivity of the QCM sensor was successfully achieved by which different strains (W and B)
of E. coli can be easily detected. Human rhinovirus (HRV) and the foot-and-mouth disease virus (FMDV) were also detected by
mass-sensitive measurements as shown in Fig. 16. A higher sensor response was observed toward templated analyte HRV in compar-
ison with FMDV. HeLa cells and their organelles were also detected selectively by using a QCM sensor.

2.10.6.4 Detection of Complex Mixtures


A QCM sensor array was also designed for online monitoring of complex processes in composting. This mass-sensitive device was
able to monitor the degradation process of plants in which complex mixtures of analytes emanate during its composting. The com-
posting phases were characterized by fabricating two 10 MHz QCMs having four electrodes on each crystal plate as shown in Fig. 17.
Imprinted polymers were coated on these electrodes for selective recognition of water, limonene, estragole, a-pinene, b-pinene,
thymol, and eucalyptol in order to herbs. The polystyrene-coated mass-sensitive sensor array not only was capable of monitoring
composting processes but also was capable of distinguishing between isomers such as a-pinene and b-pinene.28

Figure 17 A QCM sensor array fabricated by using two 10 MHz QCMs having four electrodes on each crystal plate.
Quartz Crystal Microbalances: Chemical Applications 211

Degradation of lubricating oil is also a very complex process because it not only affects the additives but also oxidizes the base oil
to produce a large number of different acidic products. In order to monitor chemical changes in lubricating oil due to complex
oxidation processes, a QCM sensor device was fabricated by templating the sensitive layer with capric acid. The cavities developed
by capric acid were able to recognize those acidic components in used oil, which contained 5–15 carbon atoms. In this way,
a complex process of oil degradation can also be monitored by using mass-sensitive sensor devices.29

References

1. Ballato, A. In Proceedings of the 1996 IEEE Ultrasonics Symposium; 1996.


2. Mason, W. P. J. Acoust. Soc. Am. 1981, 70 (6), 1561–1566.
3. Katzir, S. Arch. Hist. Exact Sci. 2003, 57 (1), 61–91.
4. Benes, E.; et al. Sensors Actuat. A Phys. 1995, 48 (1), 1–21.
5. Janshoff, A.; Galla, H.-J.; Steinem, C. Angew. Chem. Int. Ed. 2000, 39 (22), 4004–4032.
6. Lucklum, R.; Hauptmann, P. Anal. Bioanal. Chem. 2005, 384 (3), 667–682.
7. Bottom, V. E. Introduction to Quartz Crystal Unit Design; Van Nostrand Reinhold: New York, 1982.
8. Hillman, A. R. J. Solid State Electrochem. 2011, 15 (7), 1647–1660.
9. Kaspar, M.; et al. Fresenius J. Anal. Chem. 2000, 366 (6), 602–610.
10. Ferreira, G. N. M.; Da-Silva, A.-C.; Tomé, B. Trends Biotechnol. 2009, 27 (12), 689–697.
11. Mecea, V. M. Anal. Lett. 2005, 38 (5), 753–767.
12. Wegener, J.; Janshoff, A.; Steinem, C. Cell Biochem. Biophys. 2001, 34 (1), 121–151.
13. Länge, K.; Rapp, B. E.; Rapp, M. Anal. Bioanal. Chem. 2008, 391 (5), 1509–1519.
14. Janata, J. In Principles of Chemical Sensors; Springer US: Boston, MA, 2009; pp 63–98.
15. Hossenlopp, J. M. Appl. Spectrosc. Rev. 2006, 41 (2), 151–164.
16. Arnau, A. Sensors 2008, 8 (1), 370.
17. Latif, U.; Qian, J.; Can, S.; Dickert, F. L. Sensors 2014, 14 (12), 23419.
18. Dickert, F. L.; Achatz, P.; Halikias, K. Fresenius J. Anal. Chem. 2001, 371 (1), 11–15.
19. Hayden, O.; Latif, U.; Dickert, F. L. Aust. J. Chem. 2011, 64 (12), 1628–1632.
20. Schirhagl, R.; Latif, U.; Dickert, F. L. J. Mater. Chem. 2011, 21 (38), 14594–14598.
21. Dickert, F. L.; Hayden, O.; Zenkel, M. E. Anal. Chem. 1999, 71 (7), 1338–1341.
22. Dickert, F. L.; Thierer, S. Adv. Mater. 1996, 8 (12), 987–990.
23. Seifner, A.; Lieberzeit, A.; Jungbauer, C.; Dickert, F. L. Anal. Chim. Acta 2009, 651 (2), 215–219.
24. Hayden, O.; Dickert, F. L. Adv. Mater. 2001, 13 (19), 1480–1483.
25. Seidler, K.; Polreichova, M.; Lieberzeit, P. A.; Dickert, F. L. Sensors 2009, 9 (10), 8146.
26. Latif, U.; Can, S.; Hayden, O.; Grillberger, P.; Dickert, F. L. Sens. Actuators B 2013, 176, 825–830.
27. Polreichova, M.; Latif, U.; Dickert, F. L. Aust. J. Chem. 2011, 64 (9), 1256–1260.
28. Iqbal, N.; Mustafa, G.; Rehman, A.; Biedermann, A.; Najafi, B.; Lieberzeit, P. A.; Dickert, F. L. Sensors 2010, 10 (7), 6361.
29. Lieberzeit, P. A.; Afzal, A.; Glanznig, G.; Dickert, F. L. Anal. Bioanal. Chem. 2007, 389 (2), 441–446.
2.11 Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry
G Arena, Università degli Studi di Catania, Catania, Italy
C Sgarlata, Università degli Studi di Catania, Catania, Italy
Ó 2017 Elsevier Ltd. All rights reserved.

2.11.1 Introduction 213


2.11.2 Obtaining the Thermodynamic Parameters 213
2.11.3 Brief Historical Notes 216
2.11.4 Getting Started: Advices and Caveats 217
2.11.5 Experiment Design 222
2.11.6 Is Determining Gibbs Energy Sufficient to Describe the Process? 226
2.11.7 Concluding Remarks 235
Acknowledgment 236
References 236

2.11.1 Introduction

“Supramolecular chemistry is the chemistry of the intermolecular bond, covering the structures and functions of the entities formed
by association of two or more chemical species” (Jean-Marie Lehn, Nobel Lecture, Abstract, Dec. 8, 1987).1 Several other definitions,
such as “chemistry beyond the molecule,” “the chemistry of noncovalent bond,” “lego chemistry,” have been added through the
years.2 The functionality of supramolecular systems has been emphasized in a recent editorial.3 As the field of supramolecular
chemistry expands and develops there is the need to have systems that are not only functional but also useful, that is, systems
that can function in “real environments.” Of all real environments the most important is water; water is not only the most abundant
solvent on earth but it also provides an environment for life and mediates, regulates, and controls many processes in nature.4 Thus,
it is not surprising that synthetic receptors that mimic natural systems in their ability to bind a given substrate are one of the most
investigated fields in supramolecular chemistry.4–7 While organic-soluble systems have offered insight into the forces involved in
binding, they do not account for the hydration that polar molecules experience in water which dramatically influences the
properties of the solvated species (reactants and products) and may result in strong desolvation and, hence, entropic benefits.
From this perspective water presents both challenges and opportunities. Attaining the desired solubility is perhaps the most impor-
tant challenge. Even when the system with the desired solubility is obtained researchers often face the problem of the protonation
state of each functional group: the form(s) (species) under which both the reactants leading to the supramolecular entity are present
in solution at a given pH must be precisely known before embarking on the investigation. The appropriate buffer that guarantees the
above form(s) must then be chosen. This is another challenging task since the buffer might reduce solubility and/or may itself
interact with the one of the reactants. Often there is the need to explore a relatively large interval of concentrations of the reactants;
this may be required either to optimize the interaction of the reacting units and/or to increase the signal/noise ratio. In such cases an
“inert” electrolyte must be used as background in order to keep the ionic strength constant; the choice of the “inert” salt may be
rather challenging as the cation, the anion, or both may interact with the receptor.
Yet, there are not only challenges but also opportunities. We have to understand and learn how to minimize or even exploit the
involvement of water in noncovalent processes. In this context, learning to harness the peculiar properties of water molecules when
included in hydrophobic cavities represents a particularly promising concept. Cavity water molecules cannot form stable hydrogen-
bonded networks with their few neighbors in a hydrophobic cavity of appropriate size and consequently their release to the bulk
restores their hydrogen-bonding ability ultimately providing a strong enthalpic driving force for the binding of the guest.8,9
Taken together these considerations emphasize a basic concept: the properties of supramolecular entities can best be understood
by considering the whole complex system in which they form and reside, including the solvent molecules that, particularly so in the
case of aqueous solution, have peculiar characteristics (e.g., high polarity, strong hydrogen-bonding ability, and specific feature of
the hydrophobic effect).

2.11.2 Obtaining the Thermodynamic Parameters

In the last couple of decades researchers have designed and built motors rotating in a single direction thanks to a series of confor-
mational changes,10 an artificial small-molecule robotic arm capable of selectively transporting a molecular cargo,11 linear artificial
molecular muscles;12 one of the most stimulating supramolecular machinery made the news in 2015.13 Cheng et al. reported the
development of an artificial molecular pump that, when supplied with redox energy from chemical reagents, triggers the accumu-
lation of tetracationic rings on a molecular dumbbell as part of an isolable, high energy kinetically stable entanglement; this

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.13716-3 213


214 Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry

rotaxane-based motor can pump macrocyclic rings onto a thread against a concentration gradient. Compared to nature’s systems,
that have taken billions of years to perfect, the above artificial pump is relatively simple. Nevertheless, such a system allows
molecules to flow “uphill” energetically using kinetic barriers, as many processes do in Nature. The authors have repeatedly
emphasized the notion that this is nonequilibrium chemistry and that chemists must develop a fundamental understanding of the
nonequilibrium dynamics of complex integrated systems based on compact simple components.13
This immediately prompts some considerations and raises crucial questions. Is moving away from equilibrium conditions at
odds with the determination of binding constants that per se requires the attainment of an equilibrium? The principle of equilibrium
thermodynamics can only be applied if and when equilibrium, a macroscopically time-invariant status, is reached. Supramolecular
systems often, though not always, do not meet this criterion; for example, molecular recognition, and much more so molecular
recognition in biological processes, occurs through a series of events in nonequilibrium conditions and often, though not always,
relies on kinetic selectivity. How do we reconcile these two aspects, that is the need of an equilibrium for the reaction to be inves-
tigated from the thermodynamic point of view and the necessity to have a sequence of steps that should hopefully move the system
away from equilibrium conditions?
Although many techniques, including calorimetry, are inadequate to describe the cascade of nonequilibrium events, they may
still be used to understand some of the steps involved in the cascade and thus shed light on the entire process.14 Though the entire
process cannot be assumed to be the summation of the individual steps determining its functioning, dissecting a ligand into smaller
fragments can provide a strategy for analyzing the role of key functional groups/steps in a supramolecular interaction.
Thus the characterization of the equilibrium (or the equilibria), eventually reached, from the thermodynamic point of view
becomes the starting point to study a binding process.
When a complex (a supramolecular species in this specific case) is formed in solution from two reagents A and B, the equilibrium
reaction may be expressed as
pA þ qB#A p Bq (1)

In Eq. (1) and following expressions, any electrical charges on the reagents or on the complexes are omitted for the sake of generality
and clarity. The thermodynamic stability constant, bGpq, is defined as
  
A p Bq A p Bq
bGpq ¼ ¼ G (2)
fA gp fBgq ½A p ½Bq
where {} and [] denote activity and concentration, respectively, and G is a quotient of activity coefficients. Experimentally,
measurements are often performed using solutions containing an excess of inert salt, that is, where G can be taken to be constant.
Under these conditions, stoichiometric stability constant, expressed in terms of concentration quotients, may be used
 
A p Bq
bpq ¼ p q (3)
½A  ½B
In so doing it is assumed that the quotient of activity coefficients, G, is constant over the conditions used for data collection. It
must be noted that such concentration constants are valid only under the conditions, including ionic strength, at which they were
determined whereas thermodynamic constants are dependent only upon temperature and pressure. The expressions (1)–(3) can be
extended to include three or more reagents.
In Eq. (3) A and B represent any two interacting species (e.g., guest and host, a ligand, and a macromolecule in biology, etc). For
the sake of simplicity, in the discussion that follows, it will be assumed that only 1:1 species are formed (i.e., p ¼ 1 and q ¼ 1) unless
differently specified.
A stability (association, binding) constant is related to the standard Gibbs energy change DG for the equilibrium by
DG ¼ RT lnb (4)
where R is the gas constant and T is the absolute temperature. It must be assumed that temperature, commonly 298 K, and pressure,
commonly 1 bar, are constant.
The standard Gibbs energy change term relates to the standard enthalpy change, DH , and the standard entropy change, DS ,
components as shown in Eq. (5).
DG ¼ DH  TDS (5)
The value of the stability constant (and hence DG ) governs the equilibrium described in Eq. (1), that is, the partition between
free and bound species, and gives a measure of the extent of advancement of the equilibrium reaction.
Several techniques may be used to determine the equilibrium (association or binding) constant of a supramolecular species;
these include UV–vis spectrophotometry, fluorescence spectrophotometry, NMR spectroscopy, and calorimetry. The applicability
of some of these techniques is limited by the physical characteristics of the compound(s) under study. For example, UV–vis
spectrophotometry requires an absorbing species and transparent solutions; cloudy solutions cannot be investigated using
this technique. Fluorescence spectrophotometry requires that a fluorescent unit be present within or attachable to the molecule
backbone. NMR requires a chemical shift change as a result of the formation of the supramolecular entity. Unlike other
techniques, calorimetry measures the heat absorbed or released by the reaction, which at constant pressure is equal to the
Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry 215

enthalpy of the reaction; the observable (heat) is at the same time the great advantage and Achilles’ heel of the technique. Heat is
a universal signal and as such does not require any labeling of the molecule(s); by contrast, precisely because it is a universal
reporter, heat cannot be linked directly with any specific molecule or part thereof as for example a chemical shift change in
NMR. Most importantly, calorimetry provides direct access to the stability constant, the stoichiometry, and the enthalpy of
binding in a single experiment.
The enthalpy of a reaction may also be determined indirectly by measuring the temperature dependence of the equilibrium
constant (van’t Hoff method). Under the assumption that only a 1:1 species forms, Eq. (4) becomes
DG ¼ RT ln K (6)
where K is the association constant for the AB species. In the van’t Hoff method, an expression relating the standard enthalpy change
to the equilibrium constant for a reaction is obtained by combining Eqs. (5), (6) and differentiating with respect to 1/T
d ln K
DH ¼ R  (7)
d T1

where DH denotes the van’t Hoff enthalpy. A plot of ln K versus 1/T should be linear with a slope equal to DH /R, provided
enthalpy does not change with temperature; it follows that Eq. (7) should be used only if the standard enthalpy is independent of
temperature, within experimental error. If DH changes significantly with temperature, implying that the heat capacity change (DCp)
differs from zero significantly, the integrated form of the van’t Hoff equation
 
K DHref  Tref DCp 1 1 DCp T
ln ¼  þ ln (8)
Kref R Tref T R Tref
may be conveniently used as suggested by Naghibi et al. where Tref is an arbitrarily chosen reference temperature (in K), Kref and
DHref are temporarily assigned values of the equilibrium constant and van’t Hoff enthalpy at that temperature, and DCp is the van’t
Hoff-derived heat capacity change.15 DCp is assumed to be nonzero and independent of temperature; the latter assumption is valid
due to the narrow temperature interval usually investigated (vide infra).
The validity of van’t Hoff enthalpies derived from the temperature variation of equilibrium constants has been extensively
debated over the past two decades.15–22 Studies on a variety of systems have shown that the linear van’t Hoff equation (7) is
frequently a poor approximation, even in cases where linear least squaring has yielded a correlation coefficient sufficiently close
to unity to appear to validate it.15,21,22 However, a nonzero DCp often leads to a curvature in the van’t Hoff plots. Even when
taking into account curvatures, data analysis may be nontrivial;16,18 it has also been pointed out that (i) estimates of DH and
DCp are highly correlated and neither parameter can be reliably estimated unless one can be constrained and (ii) small heat
capacity changes can bias the slope of van’t Hoff plots without producing curvature that is visible within the noise level of typical
data.17 This and other problems have induced some authors to question the validity of the van’t Hoff enthalpy.15,17,20–23 In any
case, the accuracy of the DH estimates is considerably lower than that for DH values obtained directly by calorimetry and discrep-
ancies remain “uncomfortably large.”24 Furthermore, in order to rival the accuracy of calorimetrically obtained data, the indirect
determination would imply the determination of a series of equilibrium constant over an appropriate temperature range, which
would require more time and material than the direct calorimetric determination. Unfortunately, often the DG change with
temperature is not large enough and, on the other hand, for practical reasons temperatures cannot be spaced over the interval
that would be needed to have association constants that significantly differ from each other. This is a serious obstacle in the field
of supramolecular chemistry owing to the well-known enthalpy–entropy compensation effect often observed for the formation
of supramolecular assemblies.25
Fig. 1 illustrates in a pictorial way the difficulty, often encountered in supramolecular chemistry, to extract enthalpy and heat
capacity values from the temperature dependence of the Gibbs energy of binding.26 In order to facilitate the reader we have
maintained the symbolism and units used in the paper from which figures and/or equations were reproduced throughout the
chapter.
Fig. 1 shows interesting features. DH and DS have similar dependence on the temperature. The reason for such a similarity may
be easily explained if we consider the equation expressing the Gibbs energy change
DG ¼ DH  TDS (9)
and the equations expressing the heat capacity change
DCp ¼ ðvDH=vT Þp (10)

DCp ¼ ðvTDS=vT Þp (11)

If we take the derivative of each of the two terms that contribute to the Gibbs energy change with respect to T
ðvDH=vT Þp ¼ DCp (12)

ðvTDS=vT Þp ¼ DCp þ DS (13)


216 Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry

Figure 1 Temperature profile of the Gibbs energy DG (solid line), the enthalpy DH (dashed line), the entropy TDS (dotted line), and the
affinity Ka (dashed–dotted line) for a macromolecule/ligand binding characterized by: (left) Ka (25 C) ¼ 109 M 1, DH (25 C) ¼ 5 kcal mol 1,
DCp ¼  0.5 kcal K 1 mol 1; (right) Ka (25 C) ¼ 109 M 1, DH (25 C) ¼ 5 kcal mol 1, DCp ¼ 0.5 kcal K 1 mol 1. Reprinted with permission
from Velazquez Campoy, A.; Freire, E. Biophys. Chem. 2005, 115, 115–124. Copyright 2005, Elsevier.

and take into account that for supramolecular systems usually DS values range from  30 to 30 cal K 1 mol 1 and DCp values range
from  200 to  500 cal K 1 mol 126 then the entropic change amounts to less than 10% of the heat capacity change and is
negligible with respect to DCp (Eq. 13) which leads to the conclusion that both DH and DS have nearly the same dependence factor
on temperature (i.e., DCp). Thus, although enthalpy and entropy exhibit a marked temperature dependency (determined by the heat
capacity change upon binding), the Gibbs energy change is almost flat and hardly changes (< 2 kcal mol 1) in the 10–40 C interval
owing to enthalpy–entropy compensation, making very difficult the determination of the binding enthalpy and, hence, of heat
capacity changes.
The above considerations, together with the availability of sensitive and automated calorimeters that make possible the accurate
determination of both enthalpies and equilibrium constants on the same sample for a wide variety of processes, favors the direct
(calorimetric) over the indirect (van’t Hoff) determination of enthalpy values provided that the calorimetric experiment is appro-
priately designed.

2.11.3 Brief Historical Notes

Skinner was the first to introduce a procedure based on batch solution calorimetry to determine an equilibrium constant;27
Benzinger was the first to describe “Equations to obtain, for equilibrium reactions, free-energy, heat, and entropy values from
two calorimetric measurements” more than half a century ago.28 The development of temperature rise, isoperibol (constant
environment) titration calorimeters by Christensen et al. in the mid-sixties,29 the simultaneous development of the so-called
entropy titration method for the “determination of DG (K), DH, and DS” by Hansen et al.30 as well as the introduction of calorim-
eters fit for the purpose into the market by Tronac greatly boosted the application of calorimetry for the simultaneous determination
of “reaction enthalpies, stoichiometries, and sometimes equilibrium constants for the reaction of interest.”31,32 The dewar vessel
was later scaled down to 3–4 mL once problems associated with heat leak modulus and delivery of small volumes of titrant
were solved.31,33 At the same time LKB introduced a constant environment temperature rise calorimeter, developed by Sunner
and Wadso34 (Model 8700) that could be employed also for titration calorimetry. Isoperibol titration calorimeters (e.g., Tronac
Model 450) generally produced a thermogram of millivolts versus time (or volume); millivolts had then to be converted into
heat by appropriate calibration. In addition to DH values, stability constants smaller than 104 could be determined.32 However,
the useful length of an isoperibol titration run could not extend beyond 30 min.
The first power compensation isothermal titration calorimeter was developed by the Brigham Young University research group
and introduced in the market by Tronac (Model 458).31 While the isoperibol titration calorimeter was simpler, less expensive, and
easier to operate, the isothermal titration calorimeter (i) directly produced the time derivative of the thermogram (dQ/dt vs. time)
that could be integrated over time to give the heat produced or absorbed during the chosen time interval, (ii) had a better sensitivity
than the isoperibol analogue, and (iii) could be used for runs extending into hours or days. However, owing to the compensation
system, the response of the system to a sudden change in the rate of heat production after any start or end of the titration was
a damped sinusoidal wave that would prevent the collection of usable data points for a period of 2–3 min.
Some 10 years later, thanks to the improvement of Peltier/Seebeck devices, heat conduction calorimeters were developed and
introduced by Thermometric and Calorimetry Science Corporation. Heat conduction calorimeters achieved fairly low detection
limits (tens of nanowatt) but did not allow continuous titrations. The introduction of power compensation calorimeters that
make use of smaller volumes and have fairly low detection limits by Microcal (now Malvern) and Calorimetry Science Corporation
(now TA Instruments) determined the surge in popularity for the study of chemical binding phenomena and the widespread use of
Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry 217

isothermal titration calorimetry (ITC). According to Wadso et al., an isothermal calorimeter is “a calorimeter designed for use in the
nanowatt (nW) range, under essentially isothermal conditions.”35 However, following the introduction of the power compensation
titration calorimeters the advantage of collecting many points, typical of the original power compensation and/or isoperibol
calorimeters, was lost; currently stepwise titrations in which 10–30 points are collected have become the routine method.36
Both Malvern and TA Instruments calorimeters use overflow cells; the observable is the heat rate, derived from the power supplied
by a control heater used to maintain a steady-state nearly zero temperature difference between the sample and the reference cell, the
only basic difference being the mode of compensation. The power from the control heater is compensated by a constant cooling
mechanism that is passive in the Malvern calorimeters and actively controlled in the TA Instruments calorimeters.37 Both types
of calorimeters can work with 1 or 0.2 mL reaction cells having a different design; the titrant is injected by a motor-driven syringe
whose needle also serves as a stirrer.

2.11.4 Getting Started: Advices and Caveats

Although a discussion of all preliminary operations and checks that need to be done in order to obtain accurate measurements of
the observable (heat values) with isothermal titration calorimeters is beyond the scope of the present chapter, few tips and relevant
references will be given which will guide the operator in the choice of appropriate conditions. In this section we will imagine that
isothermal titration calorimeters will be used though some of suggestions are of general applicability.
As illustrated in the Fig. 2, the calorimeter is basically composed of the calorimeter block and two cells that cannot be removed.
The calorimeter block as well as the two cells are encased within an insulated air-tight canister. The solution containing the receptor
(macromolecule) is placed in the sample cell while a solution containing the same components as the sample cell except the
receptor is placed in the reference cell. The sample cell may be overfilled or partially filled. In the first case as titrant is injected
into the reaction vessel, the sample solution overflows through the fill tube; in the second case there is a vapor space over the liquid
phase that may cause inconveniences (heat absorption or release) due to condensation or evaporation of the solution. The titrant is
injected stepwise by using a rotating syringe that is coaxially inserted through the long access tube and is operated by a stepper
motor. The tip of the needle of the syringe is flattened and slightly twisted and thus works also as a stirrer. Since the reference
cell has to match as closely as possible the thermal properties of the sample cell, a needle is also inserted in the reference cell during
the titration. Temperature differences between the reference cell and the sample cell are measured and a control heater, placed on
the sample cell, is used to maintain an approximately zero temperature difference between the sample and reference cells. The
power required to maintain this zero difference is used as the calorimeter signal and is monitored as a function of time; thus
the primary measurement is heat rate.
As the titrant is injected into the sample cell and the reaction occurs, heat is taken up or evolved. For an exothermic reaction the
temperature in the sample cell will increase and the feedback power will be diminished to maintain equal temperatures between the
two cells. The opposite is true for endothermic reactions; in this case the feedback circuit will increase the heat provided over time to
the sample cell to nearly zero the temperature difference. Accordingly, a negative and a positive power signal will be generated for
exothermic and endothermic reactions, respectively. Integration of the power curve over time will yield the heat released or
absorbed per each injection.
Preliminarily, the instrument must be equilibrated until the baseline (defined as “the value recorded for the calorimetric signal
when no thermal power is evolved in the reaction vessel”) reaches the slope and a peak-to-peak standard deviation (SD) set by the
researcher (usually < 0.1 mW h 1 and < 10 nW, respectively).37 The slope and the peak-to-peak SD of the straight lines obtained by

Figure 2 Cell design of an ITC instrument. Reprinted with permission from TA Instruments.
218 Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry

fitting the points collected over chosen intervals give an idea of the “drift” and the “noise” ( 2SD), respectively.38 “Noise” and
“drift” determine the acceptability limits set by a researcher and may also be indicative of problems; for example, a constant base-
line drift may indicate that a chemical is being absorbed or released by the cell walls and/or that the calorimetric system is hardly
reaching a steady baseline due to bad power compensation.
Then, the detection limit (DL) of the calorimeter should be determined; please note that DL is here defined as twice the SD of the
average heat per injection.37 This can easily be accomplished by titrating water into water and fitting the points to straight line, as
shown in Fig. 3.37,39
Reportedly, the DL for the run shown in Fig. 3 was  0.56 mJ/injection, that is, well below the limit ( 2 mJ/injection) recommen-
ded in the literature.37
Incidentally, time intervals between two consecutive injections should be sufficiently long to permit a complete return to
baseline. However, this may not fully guarantee that all the heat involved (absorbed/released) in the reaction is really detected
especially in the presence of slow reactions. This is particularly important for supramolecular systems where kinetic problems
may often be encountered. In order to avoid such a risk, experiments should never be run by using the same time intervals. The
binding of tetraethylammonium (NEt4 þ ) to the self-assembling, water-soluble tetrahedral [Ga4L6]12  host (L ¼ 1,5-bis(2,3-
dihydroxybenzamido)naphthalene) shown in the inset of Fig. 4 is illustrative of the above concept.40
NEt4 þ may bind to both the interior and the exterior of the host. 1H NMR studies had previously shown that the first step, that is,
the encapsulation process, can be relatively slow. Based on this information, the same ITC experiment was run with different time
intervals between guest addition, ranging from 300 to 1200 s. As shown in Fig. 4 complete equilibration of the guest encapsulation

Figure 3 Thermogram obtained by titrating water into water. (A) Typical run obtained with 20 injections at 300 s intervals (10 mL/injection). (B)
Linear fit of integrated areas. (C) Plot of residuals. Reprinted with permission from Sgarlata, C.; Zito, V.; Arena, G. Anal. Bioanal. Chem. 2013, 405,
1085–1094. Copyright 2013, Springer.
Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry 219

Figure 4 Titration of a 80 mM guest solution into a 1 mM solution of the host in 0.1 M aqueous KCl; inset: schematic and space filling model of
the [Ga4L6]12  framework. The inset is reprinted with permission from Sgarlata, C.; Mugridge, J. S.; Pluth, M. D.; Tiedemann, B. E. F.; Zito, V.;
Arena, G.; Raymond, K. N. J. Am. Chem. Soc. 2010, 132, 1005–1009. Copyright 2010, American Chemical Society.

process was achieved only at the longer time intervals; accordingly, the time interval between each of the first 8–9 additions was set
at 1200 s whereas a shorter time was employed for the remaining injections.
Calibration of the volume of titrant injected should be addressed next. Considering the weighing precision of a standard analyt-
ical balance, titrant volumes (i.e., mass) typical of individual injections cannot be determined accurately because of their small
size.37,41 The volume of the syringe can be determined by filling with water, weighing the syringe before and after dispensing water
by a known number of steps of the motor, and inferring the volume from the mass and density of the water dispensed. It is advisable
to work with full or nearly full syringes to minimize the error in the mass weighed. Calibration of the syringe via determination of
the heat per injection with a quantitative reaction has also been suggested.37
Calibration of the system may be an important source of systematic errors. Isothermal titration calorimeters are usually
calibrated electrically via an electrical calibration heater. Electrical calibration may introduce significant errors owing to failure to
transfer all heat to the calorimeter vessel because of heat generation in lead wires and because of a different distribution of heat
from the calibration heater than from a chemical process.35,37,38 It is estimated that errors due to electrical calibration are commonly
of the order of 5%.35 We have personally experienced errors resulting from electrical calibrations of the order of 4–5%.39 The above
considerations lead to the conclusion that the calorimetric system must be calibrated chemically to obtain a calibration factor that
renders measured changes in control power equal to the heat effects from a reaction. Several test reactions have been proposed and
critically discussed.38 We believe that the protonation of 2-amino-2-hydroxymethyl-propane-1,3-diol (tris(hydroxymethyl)amino-
methane TRIS) is a good test reaction provided that experimental conditions are appropriately chosen. In order to determine the
calibration factor, a series of TRIS protonation titrations may be run. Incidentally, a plot of the values obtained for each single
injection versus the number of injections should have a negligibly small slope. To estimate the calibration factor, the measured
enthalpy changes for test reactions (e.g., TRIS protonation) must be corrected for heat of dilution of the titrant to the ionic strength
of the solution contained in the reaction vessel. Blank titrations usually involve relatively small heats and, thus, introduce an
additional noise into the net reaction heat; in order to minimize this noise, we usually run a few blank titrations and then optimize
them to obtain the “best blank,” as illustrated in Fig. 5.
The heat values for the test reaction, corrected for the blank, may be fitted to obtain the experimentally determined heat value for
the test reaction; the first two injections may be thrown away since the reaction is practically quantitative. The ratio between the
literature (expected) and the experimentally determined value yields the calibration factor; all the data must be multiplied by
this factor to make measured changes in control power equal to the heat effects from a reaction.
All the considerations above imply that the volume of the cell be precisely known. The reaction vessel has an active region and an
overflow region; the effective volume depends on how much volume is displaced by the stirrer and how much of the overflow tube
is accessed by the reactant in the vessel during a reaction. The effective volume can be calculated only by determining the amount
of a reaction that occurs in the cell when it is overfilled with a reactant of a known concentration; for example, this may be
accomplished by determining the endpoint of a titration with a quantitative reaction or by using the dilution of NaCl.37,41
Last, but certainly not least, the purity (concentration) of both the titrant and the titrate needs to be accurately known; usually
fitted values are correlated and are very sensitive to small errors in concentrations.36,42,43 Such errors are deemed as “the first in line
of the factors that interfere with the reproducibility of calorimetric results.”14 If the reactants have ionizable protons, their
concentration may be checked pH-metrically. For example, a simple pH-metric titration and the appropriate design of the
experiment helped unveiling a subtle impurity that had led to inappropriate conclusions as to the number of protons released
by a p-tetrasulfonato-calix[4]arene in aqueous solution.44 The combination of thermogravimetric analysis with pH-metric results
can be conveniently used to determine the amount of solvent of crystallization contained in abiotic hosts.45,46 Water, and more
220 Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry

Figure 5 (A) “Blank” experiment. Titration of aqueous HCl (0.999 mM) into 10 mM NaCl. The cell volume was 0.988 mL and the injection volumes
were 2 mL for the first and 5 mL for subsequent injections; 16 injections at 300 s intervals. (B) Optimized “blank.” The red line represents the “best”
heat to be subtracted from the gross heat. The blank incremental heats are obtained by subtracting the (n  1)th from the nth value. Reprinted with
permission from Sgarlata, C.; Zito, V.; Arena, G. Anal. Bioanal. Chem. 2013, 405, 1085–1094. Copyright 2013, Springer.

in general solvent of crystallization, is often neglected when calculating the initial concentration of the reactant causing biased
results. The problem becomes even worse when water of crystallization is not determined preventively, for example, in a host
and this is used for calorimetric investigation in organic solvents; in such cases, results strongly correlated with the concentration
of the host (or guest) are obtained owing to the different polarities of water and the organic solvent which lead to misinterpretation
of the data.
Finally, the ionic strength of both titrant and titrate solutions deserves a few words. It should be realized that activity coefficients
are arbitrarily set equal to 1. While such an assumption often holds, one should avoid comparing data for supramolecular systems
obtained under conditions that are either not clearly defined or different; in addition, ion pairing might have a role in the total heat
detected in a calorimetric titration.42 Fig. 6 shows the impact that an inert background salt can have on a very simple calorimetric
titration such as the addition of HNO3 to water.
The figure shows that even the sign of the heat involved changes with the ionic strength of the solution; it is the perturbation of
the intrinsic solvent structure that rules the effects in polar media where ions are soluble and abundant. Controlling and knowing
the ionic strength in which the thermodynamic parameters for the binding processes have been determined is particularly important
when studying highly charged supramolecular systems. Moreover, the ionic strength may have a relevant impact on systems where
the electronic configuration determines the actual charge delocalization which may also have an impact on more remote parts of the
same molecular species.
Fig. 7 shows the thermograms obtained for the titration of 1,6-bis(pyridinium)-hexane dibromide (BPC6) into a p-sulfonato-
calix[4]arene (C4TS) in the presence and in the absence of an inert background salt.
Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry 221

Figure 6 Stepwise injection of 0.02 M HNO3 into water. (A) Plain water in the cell. (B) Both the syringe and the cell are 0.1 M in KNO3.

Figure 7 Effect of a background salt on the heat released when adding a 2 mM solution of 1,6-bis(pyridinium)-hexane dibromide (BPC6)
into a 0.3 mM solution of p-sulfonato-calix[4]arene (C4TS). Black curve: titration in which both the titrant and the titrate contain 0.1 M KNO3;
red curve: titration where neither the titrant nor the titrate contain KNO3; both solutions were buffered with phosphate (0.025 M) to have
a pH7.
222 Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry

The inclusion of BPC6 into C4TS had been investigated previously via 1H NMR.47 Regardless of the species produced by the
interaction of BPC6 with the host, understandably the heat produced in the presence of an inert salt (KNO3) is sizably different
from that released in a titration carried out under identical conditions but without inert electrolyte.
In summary, any spurious heat-generating process will contribute to the global detected heat and will likely lead to erroneous
conclusions. This further stresses that any examination among seemingly comparable data cannot prescind from the detailed
cognition of all the ionic partners present in solution.

2.11.5 Experiment Design

Although with titration calorimetry as well as other techniques (e.g., UV–vis, NMR, etc.) an observable (heat for calorimetry,
absorbance and chemical shift change for UV–vis and NMR, respectively) is monitored that is proportional to the increase in
the number of moles (calorimetry) or concentration (UV–vis, NMR) of the complex, ITC has the undeniable advantage over the
above techniques of yielding the association constant, the enthalpy of reaction, and the stoichiometry (n) of a reaction in a single
experiment, while other techniques would require a certain number of experiments and a number of assumptions to provide the
same parameters. Nevertheless, ITC requires a careful design of the experiment in order to provide accurate results.
The determinability and the accuracy of K, DH, and the stoichiometry factor n basically depend on the so-called Wiseman c value,
that is, “the product of the binding constant times the total macromolecule concentration.”48 The product between the association
(formation) constant and the concentration of the receptor is related to the degree of completion of the reaction at the equivalence
point as shown in Fig. 8.
Here, Kf and CR are the association constant and the concentration of the macrocycle in the reaction vessel. Regardless of the
way the heat produced in a reaction is plotted (heat per increment of titrant (A) or total heat (B), Fig. 8) the product KfCR
determines how rounded the isotherm is at the equivalence point. The data describing the curves can be deconvoluted to obtain
K, DH, and the stoichiometry factor n. Deconvolution, though, may be complicated by the inverse correlation between Kf and

Figure 8 Simulated titration curves for reactions with different values of KfCR plotted as heat per increment of titrant (A) and as total heat (B).
Note that the curves in (A) are the first derivatives of the curves in (B). Reprinted with permission from Hansen, L. D.; Fellingham, G. W.; Russell D.
J. Anal. Biochem. 2011, 409, 220–229. Copyright 2011, Elsevier.
Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry 223

DH. In any case, the points that affect mostly the enthalpy and the association constant values are those collected at the beginning
and around the equivalence point, respectively. The curves tend to have a sigmoidal trend as KfCR increases. With KfCR > 10 the
sigmoidal shape is well visible; the slope around the end point of such sigmoidal curves tends to vertical as c approaches infinity.
It must be emphasized that the applicability of the calorimetric method for the determination of Kf is based on the reaction being
sizably incomplete at the equivalence point. For a (nearly) quantitative reaction (Kf > 1000) the calorimetric curve drops sharply
as the reaction nears the equivalence point and there are not enough data points to determine Kf while n and DH can still be
determined (Fig. 8A).
The window, that is, the c interval within which one can safely operate to obtain accurate results, has been debated extensively in
the literature. Several authors have tackled this problem over the years; people willing to examine in depth the debate will find
appropriate articles, reviews, and book chapters in articles published in the last 10 years or so.26,36,42,43 While it is a commonly
shared view that the upper limit of the window should not significantly exceed 500 and must be kept below 1000, the lower limit
is still a matter of debate36,42,43 and it is believed that it may be extended well below the limit initially proposed by Wiseman et al.
(vide infra).42,43
Assuming that (i) only a 1:1 species forms for the reaction represented in Eq. (1) (i.e., p ¼ q ¼ 1), (ii) A and B are the titrant and the
titrate, respectively, at the end point we must have CT ¼ CR where CT and CR are the analytical concentrations of A (titrant) and B
(titrate) at the end point, respectively. By combining the equation expressing Kf for a 1:1 species as a function of aep (i.e., the fraction
of product formed) with the equation that gives the total heat produced at the end point (Qep) and imposing that (i) aep ¼ 0.956 (i.e.,
KfCR ¼ 500) and (ii) the detection limit (dQ) (see previous section) be less than 1% of Qep, Hansen et al. have obtained Eq. (14)

Kf jDHj < 4:72VR =dQ (14)

where DH and VR are the enthalpy change of the reaction and the reaction vessel volume, respectively.36 Eq. (14) links properties of
the calorimetric apparatus (VR and dQ) with quantities dictated by nature (Kf and DH) and shows that it is not the formation
constant value but the ratio Kf/jDHj that determines the upper limit of the Wiseman window; in turn, such a ratio depends on the
ratio between the volume of the cell and the detection limit (i.e., VR/dQ, mL mJ 1). Thus, ceteris paribus, simply running a reaction
with a 200 mL reaction vessel having a detection limit of 1 mJ reduces the upper limit accessible for Kf with a standard volume
reaction vessel (1 mL, dQ ¼ 2 mJ) by a factor of 2.5, that goes up to 5 if the two calorimeters have the same detection limit. Hansen
et al. suggest 100 to be the ideal value for the KfCR product; furthermore they indicate that this product should never be < 50 or
> 500, as values < 50 would determine large errors in DH and values over 500 would result in large errors in Kf.36 Moreover they
strongly suggest that at least five data points be taken around the end point since this is the minimum number needed to define
a sigmoid. It is also concluded that for reactions with very large Kf there is a border region beyond which the detection limit of the
calorimeter makes it impossible to study a reaction at low enough concentration to fall within the Kf window; in these cases only
n and DH may be determined while Kf cannot.
The lower boundary of the Wiseman window has been revisited by Turnbull et al.42 as well as by Tellinghuisen.49 To evaluate the
applicability of ITC for studying low affinity systems Turnbull et al. have examined experimentally the well-known 18-crown-6 with
potassium50 and barium37–39,51,52 system which have association constants in water equal to ca. 102 and 6  104, respectively. A
saturation fraction of the receptor equal to 80% (i.e., percentage of formation relative to 18-crown-6) was chosen for the tests in
line with other authors.39,53 It was shown that the values of DG for Kþ obtained from curve fitting, with the stoichiometry set
at n ¼ 1 for c < 5, had little variation over the range of c values tested, while DH values exhibited a clear trend toward larger negative
values as the 18-crown-6 concentration, and hence c, increased. In order to reach the high level of saturation (80%) chosen in these
experiments, large concentrations of the reactants (especially of the titrant) were requested (Fig. 9).
This implies that the deviation of activity coefficients from unity cannot be avoided. Furthermore, ion pairs may be present
(especially in the titrant) and this might imply a contribution to the observed heat changes in the dilution experiment resulting
from ion pair dissociation.54 The effect of errors in the titrant and the titrate concentrations on Kass and DH values was examined
by titrating Ba2 þ into 18-crown-6 (Fig. 10).
Fig. 10 shows that varying the titrant (BaCl2) or the titrate (18-crown-6) concentration by as much as  15% has little impact on
the calculated values for DG independently of the value chosen for c. Errors of the same magnitude in the titrant concentration have
little influence on the DH values for the low c value titration, while cause a significant change ( 1.5 kcal mol 1, i.e.,  20%) for the
high c value experiment (Fig. 10A). By contrast, errors in the titrate concentration have a marked effect on the DH values for the low
c value run while scarcely affect the DH values obtained for the high c experiment (Fig. 10B). The picture obtained at c ¼ 0.05 (not
shown in Fig. 10) was similar to that for c ¼ 0.5, except DH values showed even less variation with error in ligand concentration.
Incidentally, limiting the number of variables used to fit curves gives better results since it eliminates correlation of the variables.
Following a strategy similar to that adopted by Christensen et al.,55 simulated data containing a predetermined random error
were also used to investigate a wider range of Kass and DH values. Such theoretical data support the experimental results and allow
to conclude that the lower limit of the “window” of ITC titrations can be extended to much lower values of c for systems with low
binding association constants, provided that (i) a sufficient portion of the binding isotherm is used for analysis, (ii) the binding
stoichiometry is known (n is fixed), (iii) the concentrations of both ligand and receptor are known with accuracy, and (iv) there
is an adequate level of signal-to-noise in the data.
Later Tellinghuisen also examined the low c boundary of the “Wiseman window.”49 In line with Turnball et al.,42 he
emphasized that for low c values a great excess of (titrant) has to be added to the titrate (receptor) to ensure significant conversion
224 Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry

Figure 9 Representative ITC data for the interaction of KCl and 18-crown-6 showing (A) c ¼ 10 with 225 consecutive injections of 1.3 mL and (B)
c ¼ 0.01 with an initial injection of 2 mL followed by 10  5 mL and 24  8 mL. The corresponding heats of dilution are shown in (C) and (D), and the
integrated data, with best fitting lines in red, are displayed in (E) and (F) along with residuals of the fitting in blue (scaled-up by a factor of five). The
structure of the 18-crown-6/KCl complex is the inset in (C). Reprinted with permission from Turnbull, W. B.; Daranas, A. H. J. Am. Chem. Soc. 2003,
125, 14859–14866. Copyright 2003, American Chemical Society.

of the titrant to the complex, which implies that for the 1:1 reaction, X þ M $ MX, [X]0 » [M]0. Again, the symbols used in the
source reference are maintained for the sake of simplicity. Owing to large excess of titrant, [X] z [X]0 and hence the association
constant can be expressed as
K ½X 0 z½MX =½M (15)
43
If Eq. (15) is combined with the empirical equation previously introduced by the same author

Rm ¼ 6:4 c0:2 þ 13=c (16)

that sets the range of a titration, where Rm is the ratio [X]0/[M]0 in the titration vessel after the last (mth) injection, we immediately
realize that at low c values the conversion of the receptor to the complex is > 90%. Provided that this condition is satisfied, the
relative standard errors in both K and DH remain nearly constant as c decreases (Fig. 11).
Please note that h (i.e., [M]0DH ), the key parameter that governs the overall experiment sensitivity, is set to 0.1 cal L 1 and n is
frozen. Fig. 11 shows that under certain circumstances the relative standard errors in K and DH remain low over a large window
meaning that K can be reliably estimated by direct ITC at low values of c even when n is poorly known because the required freezing
of n in the analysis has no effect on the determined value of K. However, since high [X]0 are required, there are practical limitations
dictated by the solubility of both the titrant and the titrate as well as by possible large blank corrections, ion pairing,54 linked
equilibria, and change in activity coefficients.42 Perhaps we should remember the old maxim “Est modus in rebus, sunt certi
denique fines quos ultraque citraque nequit consistere rectum” (Quintus Horatius Flaccus, Roman poet) and avoid extremes as
much as possible, if conditions apply.
The considerations concerning the boundaries of the “Wiseman window” have led some authors to resort to competitive
displacement approaches.26,42,56 However, very often in the displacement approach it is assumed that a titrant displaces a ligand
Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry 225

Figure 10 Curve fitting BaCl2/18-crown-6 titration data with error in (A) the BaCl2 and (B) the 18-crown-6 concentrations ranging from 0 to
15% demonstrates that derivation of DG is not affected by inaccuracy in the ligand or receptor concentrations, whereas error in DH is depen-
dent on the value of c. The stoichiometry parameter, n, was fixed at 1.0 for c ¼ 0.5 but allowed to float freely for c ¼ 50. Error bars lie within the
bounds of the data point symbols. Reprinted with permission from Turnbull, W. B.; Daranas, A. H. J. Am. Chem. Soc. 2003, 125, 14859–14866.
Copyright 2003, American Chemical Society.

Figure 11 Relative standard errors in K and DH when n is frozen, as a function of c for h ¼ 0.1 cal L 1. Solid curves represent results when [X]0 in
the syringe is limited to 104 times [M]0; the fine and broad dashed curves are for this ratio dropped to 103 and 102, respectively. Cell volume is taken as
1.4 mL and total titrant volume 0.25 mL. Reprinted with permission from Tellinghuisen, J. Anal. Biochem. 2008, 373, 395–397. Copyright 2008, Elsevier.
226 Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry

from the macromolecule and that no ternary species of the type A–M–B is formed, where A, B, and M are the first ligand, the second
(competitive) ligand, and the macromolecule, respectively. While this assumption often holds for biomolecules, it may turn out to
be an oversimplification for supramolecular systems in which a metal ion is involved. In these cases, ternary species may be the rule
rather than the exception and the Gibbs energy term cannot be split into the enthalpic and entropic contributions by resorting only
to calorimetry since the simultaneous presence of multiple species with different degrees of protonation, often overlapping with
each other even within a narrow pH range, precludes a reliable fitting of the calorimetric curves.57,58

2.11.6 Is Determining Gibbs Energy Sufficient to Describe the Process?

There are countless examples showing that Gibbs energy tells just half the story; comparing two processes based on differences in
their Gibbs energy may be misleading since processes having fairly similar Gibbs energy values may have totally different enthalpic
and entropic contributions meaning that their binding mode is different. Thus, calorimetrically determined enthalpic and entropic
data are indispensable to describe a process in details. While they unveil subtle differences that are not expressed in the DG term
they may also indicate the right way to optimize interactions in supramolecular systems.
The optimization of the guest–host binding is mainly based on the tuning of geometric and functional group complementarity.
One of the reacting particles (usually the host) is modified in order to improve the binding affinity in a given direction; for example,
the enthalpic contribution may be enhanced by introducing a functional group that improves the directionality of hydrogen bonds
or establishes an additional hydrogen bond with the final goal of maximizing the mutual fit of the binding partners as well as their
affinity.
Using the widely encountered guanidinium–oxoanion binding motif,59 Schmidtchen et al. have shown that the efforts devoted to
the tailoring of the interactions between the binding partners lead to the expected improved affinity which, however, results from
different energetic reasons.60 The tetracarboxamide 1 (Fig. 12) was selected as the guanidinium ligand because it combines a high
density of hydrogen donor functions suitable for the oxoanion binding with functions that can be modified to incorporate 1 into poly-
modular hosts. It is expected that all hydrogen bond donors should converge onto the bound anionic guest, which would lead to
strong attractive forces and should result in a strong exothermic effect. By contrast the tetraallyl-substituted host 2, lacking the functions
present in 1, is expected to have a less exothermic effect. Compared to 1, the crowding due to the disposition of the carboxamide
anchor groups of 2 should impede their proper solvation, which causes an additional favorable enthalpic contribution to guest
binding owing to the reduced costs of desolvation. Thus, on the whole, the oxoanion binding to 2 is expected to be more exothermic
than the analogous interaction with 1. The data from NMR titration of guanidinium host 1 with dihydrogenphosphate in acetonitrile
were not consistent with an ordinary 1:1 binding model and also indicated that additional binding events were occurring.
Instead, ITC titrations revealed the subtleties underlying this supramolecular association. Several processes show up as the host/
guest ratio increases (Fig. 12A). The thermogram indicates that initially (host/guest ratio < 0.3, i.e., in the presence of excess guest)
higher order complexes form, though this portion of the curve could not be fit adequately. A further increase of the host/guest ratio
caused the formation of a low affinity and endothermic 1:2 (host/guest) complex that was taken over by an exothermic and stronger
1:1 adduct. Interestingly, titrating the oxoanion into host 1 (Fig. 12B) resulted in the initial formation of the 1:1 exothermic
complex that is taken over by the weaker and endothermic 1:2 species as more guest is added; higher order complexes are not
detected owing to both the moderate excess of guest added and to their low affinity. For the entire series of oxoanions, the 1:1

Figure 12 ITC traces of the titration of 1 into dihydrogenphosphate (1.53 mM, as a TBA salt) in acetonitrile at 293 K (A) or adding H2 PO4  into
host 1 solution at 0.69 mM (B). The lines represent the best fit to a two-independent-site model. The derived thermodynamic parameters for the 1:1
binding step are included in Fig. 13. Reprinted with permission from Jadhav, V. D.; Schmidtchen, F. P. Org. Lett. 2005, 7, 3311–3314. Copyright
2005, American Chemical Society.
Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry 227

Figure 13 Comparison of the binding energetics of 1:1 stoichiometric complexes of guest species 3-8 (as a TBA salt) to guanidinium hosts 1 and 2
in acetonitrile. Reprinted with permission from Jadhav, V. D.; Schmidtchen, F. P. Org. Lett. 2005, 7, 3311–3314. Copyright 2005, American Chemical
Society.

species formed with 1 was more stable than the analogous species with 2 (Fig. 13). Surprisingly the greater stability does not result
from the enthalpic contribution: the enhanced affinity observed is due to an overwhelming increase in the entropic component of
association, TDS, independently of the oxoanion.
This unexpected result cannot be due to the release of the supposedly well-structured solvent molecules to the bulk solvent. If
this were the case, the complexation of the guest featuring the most extended interface between the binding partners (i.e., guest 7,
Fig. 13) should have shown the largest entropic contribution since usually solvent release correlates with the interface area of host
and guest. Moreover, the greatest entropic differences between hosts 1 and 2 were found for the associations with the smallest
anions, 4 and 8. It must also be emphasized that if the increase in the entropic component of oxoanion association with 1 was
caused by the different solvation of the two guanidinium ligands, a more uniform absolute difference within the series should
have been observed. Based on these observations, Schmidtchen et al. concluded that the significant entropy differences originate
from variations in the number and stiffness of the mutual binding modes encompassing the partners rather than from desolvation.
Such a conclusion undermines the anticipation that the sheer number of attractive interactions leads to increased binding strength
thereby determining the affinity and clearly shows that molecular design based on geometric and functional complementarity
principles may misguide supramolecular constructions aimed at a unique host–guest binding mode.
Also Rekharsky et al. corroborate the view that geometric and functional complementarity, that is, a perfect shape/size matching
between the host and the guest, may not by itself provide the driving force to achieve an enhancement of the Gibbs energy.61 In the effort
to design host–guest systems that reach high levels of binding affinity in aqueous media, chemical functionalities were introduced into
a ferrocenic guest, which can develop significant noncovalent attractive forces between the guest and cucurbit[7]uril (CB[7]) (Fig. 14).
Since the replacement of the polar group of 1 with a charged functionality resulted in a 1000-fold increase of the binding
constant (Kass is 3  1012 and 3  109 for 2 and 1, respectively),62,63 it was reasoned that the introduction of a second charged group
in the guest, positioned ad hoc to interact with the other ring of the host’s carbonyls (3), would have further enhanced the binding
228 Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry

Figure 14 Chemical structures of the cucurbit[7]uril host and ferrocene guests: 1-hydroxymethylferrocene (1), 1-trimethylammoniomethylferrocene
(2), and 1,10 -bis(trimethylammoniomethyl)ferrocene (3). Reprinted with permission from Rekharsky, M. V.; Mori, T.; Yang, C.; Ko, Y. H.; Selvapalam,
N., et al. Proc. Natl. Acad. Sci. USA 2007, 104, 20737–20742. Copyright 2007, National Academy of Sciences, U.S.A.

affinity. The high association constant expected for 3 required sequential competition titration experiments involving a set of
reference guests exhibiting gradually increasing affinity; for 3 two sets of four step competition titrations had to be performed
the last of which for one set is illustrated in Fig. 15.
The two set of experiments yielded a binding constant (3  1015 M 1) that rivals that observed for the naturally occurring
avidin–biotin complex. As errors propagate when using a cascade of experiments, to obtain more accurate values the enthalpy
values were determined by titrations performed in the absence of competitive guest and the entropy values were calculated by
using the enthalpy values determined by direct titrations. The DH and TDS values obtained with the cascade of experiments
involving aminomethylcyclohexane1 þ in the last step (Fig. 15) were  90  1 and 2  2 kJ mol 1, respectively. The enthalpy
value for 3 is roughly the same as that obtained for 1 and 2 ( 90  2 and  90  1 kJ mol 1, respectively) while the TDS value
is higher than those determined for 1 and 2 ( 36  2 and  18  2 kJ mol 1, respectively). An enhanced affinity is obtained for
3 thanks to the introduction of a second positive charge into the guest but, unexpectedly, the boost in affinity results from
a more positive association entropy. The addition of one or two positive charges increases the TDS value while it does not
have a significant impact on the DH values. This is somehow surprising since these systems do not exhibit the usual
enthalpy–entropy compensation effect observed in supramolecular interactions.25,64–66 The DCp values for the interaction of
the three guests with CB[7] were also determined and all had negative values, which was deemed “reasonable for hydrophobic
interactions operating upon complexation.”61 The second-generation mining minima algorithm (M2) was employed to derive
a parameter, TDSconfig, that accounts for the loss of mobility of the host and guest due to the formation of the complex and
contributes to the total change in entropy. These results in combination with calorimetric evidences and X-ray crystal data
results clearly indicate that the size/shape matching between CB[7] and 3 cannot by itself provide the driving force required
to achieve the unusually high binding constant: “the release of structured water back to the bulk solvent upon binding is
an equally important factor.”
The last comment prompts a focus on the importance that the solvent molecules, and in particular water molecules, have on the
formation of the supramolecular entity due to the peculiar characteristics of water.
In this framework, Gibb et al. have focused on the “Hofmeister effect” determined by anions, that is, on the reasons why
highly solvated anions (i.e., kosmotropes like F and SO4 2 ) decrease the solubility of a protein and increase its fold while weakly
solvated anions (i.e., chaotropes like SCN and CIO4  ) cause the opposite.9 In this context, they have examined the binding of an
amphiphilic guest, adamantane carboxylate (AC), to the host system shown in Fig. 16.
The host is characterized by a water-soluble outer surface comprising eight carboxylic acids and a deep hydrophobic cavity (8 Å
wide  8 Å deep) that can accommodate guests as large as adamantanes. The same group had previously demonstrated that the
binding of the amphiphilic guest to host 1 at pHz9 leads to distinct 1:1 complexes in which the hydrophobic portion is inserted
into the pocket and the polar head group is located at the entrance of the binding site; such an arrangement inhibits dimerization
and capsule formation. It had also been found that the temperature change has a negligible effect on DG owing to an enthalpy–
entropy compensation which is the “trademark” of the hydrophobic effect.67
The binding of adamantane to 1 was investigated by ITC and 1H NMR both in the presence and in the absence of the sodium
salts of a series of anions, F, SO4 2 , AcO, Cl, Br, NO3  , CIO3  , I, SCN, and CIO4  , at pH 11.3 to ensure that the host had
good solubility and that the guest was in its deprotonated form. The calorimetric results are displayed in Fig. 17.
In the absence of any sodium salts the DG of inclusion is about 9 kcal mol 1 which is mostly due to the large
enthalpic contribution; entropy only slightly contributes to the Gibbs energy of binding. In the presence of kosmotropic salts
(e.g., F, SO4 2 ) as well as of mid Hofmeister series salts, DG increases by ca. 0.5 kcal mol 1 which is consistent with the
increase in hydrophobicity expected for these salts. By contrast, in the presence of chaotropic salts DG of inclusion decreases
monotonically to reach its minimum value for CIO4  that has a DG value of ca. 1.5 kcal mol 1 lower than that obtained for
the fluoride salt; these chaotropes have a salting-in effect similar to that observed in proteins. Interestingly the drop in the energy
Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry 229

Figure 15 Top: Competition ITC experiments on complexation of 3 with CB[7] in 73 mM solution of aminomethylcyclohexane HCl used as
competitor (curve fitting was performed by using the Single Set of Identical Sites Model). Bottom: computed conformation of CB[7]-3 complex from
the second-generation Mining Minima algorithm. Reprinted with permission from Rekharsky, M. V.; Mori, T.; Yang, C.; Ko, Y. H.; Selvapalam, N.,
et al. Proc. Natl. Acad. Sci. USA 2007, 104, 20737–20742. Copyright 2007, National Academy of Sciences, U.S.A.

Figure 16 (A) Chemical structure of host 1 and a schematic (blue bowl) representation of its overall topology. (B) Structure of adamantane
carboxylate (AC) guest. Reprinted with permission from Gibb, C. L. D.; Gibb, B. C., J. Am. Chem. Soc. 2011, 133, 7344–7347. Copyright 2011,
American Chemical Society.
230 Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry

Figure 17 ITC data for the binding of AC to 1 in 10 mM phosphate buffer (pH ¼ 11.3), in the absence and in the presence of various sodium salts.
Host ¼ 150 mM, salt ¼ 100 mM, guest titrant ¼ 1.5 mM. Reprinted with permission from Gibb, C. L. D.; Gibb, B. C., J. Am. Chem. Soc. 2011, 133,
7344–7347. Copyright 2011, American Chemical Society.

of binding results from a significant increase in entropy and a concomitant monotonic decrease in enthalpy that becomes even
unfavorable for CIO4  .
The 1H NMR investigation provides further insight into these effects. In the presence of kosmotropic and mid-Hofmeister series
salts, only slight signal broadening was observed at higher (600 mM) concentrations. By contrast, in the presence of chaotropes
(nitrate, chlorate, iodide, thiocyanate, and perchlorate salts) quite a few spectral changes were detected. Importantly, the benzal
protons that reliably report on guest inclusion shift progressively downfield as the concentration of the salt increases. Even more
importantly the addition of 1 equiv. of guest cancels out these downfield shifts: the spectrum becomes almost identical to that
obtained for the complex in the absence of salts thus showing that the chaotropic anions bind to the concave hydrophobic pocket
of 1. Based also on in silico studies showing that, upon inclusion, the largely solvated CIO4  partially loses some of the water
molecules of its hydration sphere, it is concluded that anions need to lose water molecules from their solvation shell to enter
the cavity and this determines their entropy contribution to the Gibbs energy and hence their different behavior. In other words,
the observed pattern results from the “plasticity” of the solvation sphere; strongly solvated kosmotropic anions are not prone to
lose molecules from their hydration sphere while the more weakly solvated chaotropic can partially lose water molecules to fit
into the concave deep pocket. Furthermore, a van’t Hoff plot for the binding of perchlorate to the host based on 1H NMR results
in conjunction with the results from an in silico study indicate that the ca. 4 water molecules hydrating the cavity form fewer
hydrogen bonds than the ones in the bulk and are readily lost upon inclusion of the anion. Computational studies also indicate
that the cation of the salts (sodium) has no influence on anion binding. Thus, the authors concluded that chaotropic anions
compete with the hydrophobic guest (AC) to occupy the hydrophobic concave pocket and “modulate the thermodynamics of
hydrophobic binding in a way that mirrors the Hofmeister effect.”9
Again it is demonstrated that basing the design of receptors on the use of directional, noncovalent interactions (e.g., ion–ion,
ion–dipole, and hydrogen bonds) between the guest and the host in order to achieve high binding affinities may not produce
the desired results since the desolvation of weakly solvated guests and the displacement of “unhappy” water molecules contained
within a hydrophobic cavity and possessing fewer hydrogen bonds than the bulk may play a significant role in host–guest
recognition.
The last consideration shifts our attention specifically to water molecules that are confined in a hydrophobic environment. Based
on the observation that the hydrophobic driving force can be increased by using suitably large concave receptors, Biederman et al.
have proposed an elegant strategy for the design of high-affinity receptors operating in aqueous solutions.68
The system proposed is based on the following key points (i) “natura abhorret vacuum” (nature abhors vacuum) and hence
cavities of appropriate size contain water molecule (ii) upon inclusion of an opportune guest the residual cavity water molecules
form fewer hydrogen bonds than the ones in the bulk and are consequently “high in energy,” and (iii) the further inclusion of an
appropriately chosen second guest causes the release of these “high in energy” water molecules that can thus form hydrogen-
bonded networks in the bulk. A previous study based on molecular dynamics (MDs) simulations and ITC experiments showed
that the release of high energy water from the cavity of cucurbit[n]uril (CBn) macrocycles results in a favorable enthalpic
contribution and mostly determines guest binding in aqueous solutions; the study also showed that the cavity size of CB7 turned
out to be the optimal compromise between the number of inner cavity water molecules to be released and their energetic
frustration.69 Building on the accumulated experience, the authors reasoned that a larger host (CB8) should have been able
to tightly include symmetric dicationic auxiliary guests (AG) in a 1:1 stoichiometry while still leaving room for high energy water
molecules; the entrance of a second guest (the analyte), determining the formation of a 1:1:1 ternary supramolecular complex,
would cause the release of residual cavity high energy water molecules provided that the guest as well as the analyte have
appropriate characteristics. Such a strategy is schematically illustrated in Fig. 18.
Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry 231

Figure 18 Schematic representation of 1:1:1 ternary complex formation with CB8 in aqueous solution. Bottom: changes in receptor selectivity for
analytes through auxiliary guest modification. Rearranged and reprinted with permission from Biedermann, F.; Vendruscolo, M.; Scherman, O. A.; De
Simone, A.; Nau, W. M. J. Am. Chem. Soc. 2013, 135, 14879–14888. Copyright 2013, American Chemical Society.

CB8 was known to include pairs of one electron-poor (e.g., methylviologen (MV)) and an electron-rich guest (e.g., naphthol);
however, the reasons for the inclusion of such pairs (represented in blue and red, respectively, in Fig. 18) remained puzzling since
charge transfer interactions did not seem to be energetically important in these ternary systems.70 Several pairs of AG and analytes
were investigated (Fig. 19).
The enormous amount of AG/analytes combinations reported by Biederman et al.68 (X in Fig. 19 can be a different substituent or
the same substituent occupying different positions) does not allow a detailed presentation in the limited space allotted here;
however, hopefully the examples selected show that calorimetry can be priceless especially when combined with techniques that
can either support and/or integrate calorimetric results.

Figure 19 Chemical structures and cartoon representations of (A) dicationic auxiliary guests for CB8 (halide counterions not shown) and (B)
electron-rich aromatic analytes. Rearranged and reprinted with permission from Biedermann, F.; Vendruscolo, M.; Scherman, O. A.; De Simone, A.;
Nau, W. M. J. Am. Chem. Soc. 2013, 135, 14879–14888. Copyright 2013, American Chemical Society.
232 Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry

MD simulations revealed the existence of water molecules inside the CB8$AG assembly and invariably showed that these
residual water molecules are characterized by an average hydrogen bonding number that is significantly reduced relative to
hydrogen bond count in the bulk and in the CB8 pocket before AG encapsulation. The computation study also indicated that
the transfer of these water molecules from the cavity to the bulk is energetically favorable. Interestingly, such a transfer is
energetically more favored for the inclusion of the analyte into the CB8$AG cavity than for the inclusion of the AG into CB8,
although the CB8$AG cavity contains, on the average, roughly less than half the number of water molecules contained in the
cavity of the uncomplexed CB8. If this model is correct, the analyte should possess electron-donating characteristics that stabilize
the positive charges of the auxiliary reagent; the analyte should also be small enough to fit within the cavity and at the same time
sufficiently large to displace the residual water molecules from the CB8$AG upon inclusion. Indeed, MD simulations indicated
that electron-rich analytes are preferentially included in the binary CB8$AG complex. MD simulations also indicated that both
small analytes (e.g., 1-phenol) and 1-naphthol (1-Np) were not able to fully displace water molecules from the cavity of the
preformed CB8$AG complex and thus these analytes should not have the same enthalpic contribution expected for ideal size
analytes that are able to fully displace residual water molecules from the cavity. Fig. 20 clearly shows that calorimetric evidences
match expectations originating from MD simulations.
Experimental calorimetric data do show that the inclusion of appropriately chosen aliphatic and aromatic analytes into the
preformed CB8$AG cavity is strongly favored by the enthalpic contribution (Fig. 20A); this is consistent with the high-energy water
release predicted by MD simulations as water molecules released to the bulk are involved in an energetically favorable hydrogen
bond network. On the other hand, the reduced degree of freedom resulting from the tight packing of the AG and the analyte within
the cavity accounts for the unfavorable entropy contribution. The calorimetric data collected for a series of derivatized naphthalenes,
indoles, and benzenes are shown in Fig. 20B. The inclusion of smaller analytes like benzene derivatives that do not fully displace high
energy water from the CB8$MV cavity results in lower enthalpic contribution compared to larger guests like naphthalene and indole
derivatives. Fig. 20B shows that the largest enthalpic driving forces for binding, that, in turn, determines the highest association
constant of the analyte with CB8$MV cavity, are observed for those analytes that are sufficiently large to displace all the water molecules
from the CB8$MV cavity. Interestingly, the data for the inclusion 1-Np fall in the benzene derivative region, that is, in the TDS –TDH
domain typical of small size analytes. Subtle differences arising just from the position of the substituent and indicated by the MD
simulation are readily validated by the experimentally obtained enthalpy values. Both 1-naphthol and 2-naphthol (2-Np) have

Figure 20 (A) ITC isotherm for titration of CB8$MV with 2-Np (top) and integrated heats for ternary complex formation of 1-Np (red) or 2-Np
(black) with CB8$MV (bottom). (B) Plot of DH and TDS values for the complexation of noncharged aromatic species with CB8$MV as determined
by ITC experiments in buffered aqueous solution at 298 K. (C, D) Representative MD snapshots of the CB8$MV$1-Np and CB8$MV$2-Np complexes,
respectively. All water molecules outside the CB8 cavity and the front CB8 atoms in the side view have been removed. Reprinted with permission
from Biedermann, F.; Vendruscolo, M.; Scherman, O. A.; De Simone, A.; Nau, W. M. J. Am. Chem. Soc. 2013, 135, 14879–14888. Copyright 2013,
American Chemical Society.
Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry 233

relatively large association constants with the binary complex formed by CB8 and MV (Kass ¼ 4  105 and 6.1  105 M 1, respectively).
Consistently with expectations, the inclusion of 1-Np as well as of 2-Np is strongly enthalpy driven while entropically unfavored.
However, the entrance of 1-Np into the CB8$MV cavity is accompanied by a lower enthalpy value relative to 2-Np (Fig. 20A),
consistently with MD results indicating that the high-energy water molecules occupying the CB8$MV cavity are not fully displaced
by the inclusion of 1-Np while are completely released upon 2-Np inclusion (Fig. 20C and D). Once again, this supports the view
that it is not the size but rather the high-energy water displaced by the analyte inclusion that determines the enthalpic contribution.
Even differences arising from the isotopic effect are nicely reported in detail by calorimetric titrations (Fig. 21). It is well known
that hydrogen bonding is stronger in D2O than in H2O. While the inclusion of MV into CB8 is weaker in D2O than in H2O by
a factor of 2, owing to a reduced enthalpic driving force for binding in D2O, the inclusion of indole into the CB8$MV preformed
cavity is 1.5 times stronger in D2O than in H2O; importantly, this increased stability stems from the more favorable enthalpy value
determined in D2O. The release of high energy water and the stronger bonding in bulk D2O account for such a difference.
The overall picture is further enriched by the data for 1:1 complexes of CB8$MBM and CB8$MNpM (Fig. 19); in these binary
CB8$AG complexes the portals are capped with the positively charged imidazolium units whose charges are stabilized by the
interaction with the portals themselves. Such an arrangement does not require any further interaction with an electron-rich unit
as observed for CB8 and the other dicationic AGs; as a result, only small analytes like acetone or tetrahydrofuran are included
(Fig. 18, bottom).
With the help of MD simulations, ITC provides evidence showing that the inclusion of an AG, like for example MV, changes the
well-solvated cavity of CB8 into a much poorly solvated pocket (Fig. 20) and that the further inclusion of a suitable analyte into this
preformed cavity determines the release of residual high energy water, which is the driving force for the formation of the ternary
CB8$MV$analyte supramolecular assembly. Notably, the receptor selectivity for analytes may be tuned through AG modification
as shown by the CB8$MBM and CB8$MNpM examples.
Perhaps the favorable contribution resulting from the release of water molecules is a more general phenomenon than usually
thought. Jacobson et al. investigated the inclusion of xenon into some cryptophanes by ITC in aqueous solution at 293 K;71 in
particular, xenon was reported to have a high binding association constant (Kass ¼ 42,000 M 1) with one of these host (tris-
(triazoleethylamine) cryptophane). Both the enthalpic and the entropic components contributed favorably to the Gibbs energy
(DH z  15 and TDS z 11 kJ mol 1, respectively) and were believed to result from “the release of one or more water molecules
from the cryptophane cavity” in addition to the dissolution of the clathrate water structure that surrounds xenon in solution
and to the noncovalent dispersion interactions between bound xenon and the host.
The examples reported above demonstrate that the release of high-energy water can determine not only the encapsulation of the
analytes shown in Fig. 19 but even of anionic guests9 and thus indicate that in aqueous solution the release of “frustrated” high-
energy water molecules can be more favorable than the use of direct host–guest interactions.
Some of the calorimetric results obtained counter the classical view of the hydrophobic effect and place these as well as other
systems in the family of events driven by the so-called nonclassical hydrophobic effect.72 As the family composed by such examples
is steadily increasing, perhaps the classical view of the hydrophobic effect needs to be revisited.8,73
Other strategies may be adopted to build tightly binding receptors whose association characteristics are not based on the direct
interaction between the guest and the receptor but rely on weak secondary interactions. Gunasekara et al. have reported on

Figure 21 ITC isotherm for titration of CB8$MV (34 mM) with indole (350 mM) in D2O (top) and integrated heats for ternary complex formation of
indole with CB8$MV in D2O and H2O (bottom). Reprinted with permission from Biedermann, F.; Vendruscolo, M.; Scherman, O. A.; De Simone, A.;
Nau, W. M. J. Am. Chem. Soc. 2013, 135, 14879–14888. Copyright 2013, American Chemical Society.
234 Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry

a supramolecular system where disengaged intramolecular interactions within a receptor are turned on by the weak interaction with
a guest that eliminates the “electrostatic frustration” between the charged groups of the receptor.74 Such cooperatively enhanced recep-
tors (CERs) basically exploit the positive cooperativity between intrahost interactions triggered by the interaction with a suitable guest
to reinforce their guest binding (Fig. 22). As shown in this figure, the two cholate units of 1 may fold in polar solvents with their hydro-
philic groups exposed to the bulk; note that 1 has a fluorescent label attached to it (moiety in purple). The terminal carboxylate of 1
(encircled in red) corresponds to the negatively charged groups of the CER (B, also in red) and the two cholates represent A and A0 ,
respectively.

Figure 22 Design of an electrostatically frustrated CER and its binding of an oppositely charged ligand to trigger intrahost A–A0 interactions. Rearranged
and reprinted with permission from Gunasekara, R. W.; Zhao, Y. J. Am. Chem. Soc. 2015, 137, 843  849. Copyright 2015, American Chemical Society.
Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry 235

Figure 23 ITC titration curves obtained at 298 K for the binding of 7 by (A) 3 and (B) 6. In a typical experiment, a 2–6 mM aqueous solution of the
guest in Millipore water was injected in equal steps of 10.0 mL into 1.42 mL of 0.05–0.2 mM solution of the host in Millipore water. The top panel
shows the raw calorimetric data. The area under each peak represents the amount of heat generated at each ejection and is plotted against the molar
ratio of the guest to the host. Reprinted with permission fromGunasekara, R. W.; Zhao, Y. J. Am. Chem. Soc. 2015, 137, 843–849. Copyright 2015,
American Chemical Society.

The interaction with a suitable positively charged guest (e.g., 7, Fig. 22) triggers the interaction between the two cholates of the
same arm rendering the supramolecular entity more stable. Three of such arms are attached to a 1,3,5,-triethynylbenzene unit to
generate the CER 6; this unit (purple moiety in 6) corresponds to the scaffold S of the scheme (also in purple). The analogue 6 is
used as a control receptor with its carboxylate groups intended to mimic the three carboxylates of 3. Both 3 and 6 possess three
carboxylates pointing in for guest binding; while in 6 they are rigidly arranged for binding, in 3 they bind the guest thanks to their
conformational flexibility. Fig. 23 shows the ITC titration curves for the interaction of 7 with CRC 3 and its mimicking analogue 6.
Both curves nicely fitted a 1:1 binding model yielding DG values equal to  7 and  6 kcal mol 1 for the binding of 7 with 3 and
6, respectively, as indicated by the continuous line in the bottom panels of the figure; the DH and TDS contributions to these DG
values were 10.5 and 17.5 kcal mol 1 for 3 and  35.6 and  29.6 kcal mol 1 for 6. However, while the thermogram obtained for
the binding of 6 has a monotonic trend, the curve obtained for the interaction with 3 indicates the occurrence of a relatively more
complex process. Noteworthy, the binding is enthalpy driven for 6 while is entropy driven for 3; furthermore, while the difference
between the Gibbs energy values for the two processes is equal to 1 kcal mol 1 the difference between the enthalpic and entropic
components for 3 and 6 is greater than 45 kcal mol 1. The proposed mechanism is further supported by ITC experiments
performed in the presence of organic solvents as well as large amounts of electrolytes. The addition of increasing amounts of
methanol, that is known to weaken the interaction between the organic cholate units in water, results in a sizable decrease of Kass.
The binding is also weakened by the presence of electrolytes. Since the presence of cations lowers the repulsion among the
carboxylates, the intrahost cholate–cholate interactions become “more fully engaged prior to the guest binding, destroying
the very basis of the cooperative enhancement.” This would be consistent with the authors’ earlier conclusion that repulsion
between the cholate carboxylates prevented the tight packing of cholates even though 3 was fully folded (Fig. 22).74

2.11.7 Concluding Remarks

The heat signal is an almost universal property of binding reactions including supramolecular processes. Since heat detection does
not require chemical labeling of the reacting partners, it is not surprising that calorimetry has attracted the attention of supramo-
lecular chemists worldwide; the advanced technological level, the high sensitivity, as well as the high sample throughput of modern
calorimeters make calorimetry priceless. In particular, ITC, mainly employed in its early stage for the study of biomolecular
interactions, has become the method of choice to study the binding thermodynamics of supramolecular systems.75 Calorimetry
is not immune from problems though. Its main advantage, that is, signal universality, may turn into a drawback: heat generated
or absorbed by any process occurring along with the reaction(s) under investigation contributes to the total heat, which may render
both experiment design and data interpretation all but trivial.
Since the detected signal is the result of many simultaneous processes occurring in solution it cannot, on its own, provide
information on the mechanism(s) of interaction. However, if the heat signal is combined with information coming from techniques
“reporting” on a specific change (e.g., NMR, spectroscopy etc),76 then calorimetry is invaluable since it gives direct access to the DH
and DG values, allows to derive also DS and Cp, and thus provides a better picture of a given process.
236 Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry

Considering the complex structures that can be designed and synthesized nowadays, calorimetry can be expected to have an even
more prominent role in supramolecular chemistry.77–84 Last, but not least, calorimetry has the unique ability to provide both
thermodynamic and kinetic data.85–87 Thus, stimulating applications in new directions may be foreseen in view of the recent
development of enzyme-mimicking supramolecular systems with unprecedented activity.88

Acknowledgment

University of Catania (FIR 2014, 9DD800, and 018B9A) is gratefully acknowledged for financial support.

References

1. Lehn, J. M. Angew. Chem. Int. Ed. 1988, 27, 89–112.


2. Steed, J. W.; Atwood, J. L.; Gale, P. A. In Supramolecular Chemistry: From Molecules to Nanomaterials; Steed, J. W., Gale, P. A., Eds.; Wiley: New York, 2012.
3. Davis, A. P.; Kubik, S.; Dalla Cort, A. Org. Biomol. Chem. 2015, 13, 2499–2500.
4. Oshovsky, G. V.; Reinhoudt, D. N.; Verboom, W. Angew. Chem. Int. Ed. 2007, 46, 2366–2393.
5. Laughrey, Z.; Gibb, B. C. Chem. Soc. Rev. 2011, 40, 363–386.
6. Schneider, H. J. Angew. Chem. Int. Ed. 2009, 48, 3924–3977.
7. Biros, S. M.; Rebek, J., Jr. Chem. Soc. Rev. 2007, 36, 93–104.
8. Biedermann, F.; Nau, W. M.; Schneider, H. J. Angew. Chem. Int. Ed. 2014, 53, 11158–11171.
9. Gibb, C. L. D.; Gibb, B. C. J. Am. Chem. Soc. 2011, 133, 7344–7347.
10. Kudernac, T.; Ruangsupapichat, N.; Parschau, M.; Maciá, B.; Katsonis, N.; Harutyunyan, S. R.; Ernst, K. H.; Feringa, B. L. Nature 2011, 479, 208–211.
11. Kassem, S.; Lee, A. T. L.; Leigh, D. A.; Markevicius, A.; Solà, J. Nat. Chem. 2016, 8, 138–143.
12. Liu, Y.; Flood, A. H.; Bonvallet, P. A.; Vignon, S. A.; Northrop, B. H.; et al. J. Am. Chem. Soc. 2005, 127, 9745–9759.
13. Cheng, C.; McGonigal, P. R.; Schneebeli, S. T.; Li, H.; Vermeulen, N. A.; et al. Nat. Nanotechnol. 2015, 10, 547–553.
14. Schmidtchen, F. P. In Analytical Methods in Supramolecular Chemistry; Schalley, C. A., Ed.; Wiley-VCH: Weinheim, 2007; pp 55–78.
15. Naghibi, H.; Tamura, A.; Sturtevant, J. M. Proc. Natl. Acad. Sci. U. S. A. 1995, 92, 5597–5599.
16. Horn, J. R.; Russell, D.; Lewis, E. A.; Murphy, K. P. Biochemistry 2001, 40, 1774–1778.
17. Chaires, J. B. Biophys. Chem. 1997, 64, 15–23.
18. Horn, J. R.; Brandts, J. F.; Murphy, K. P. Biochemistry 2002, 41, 7501–7507.
19. Weber, G. J. Phys. Chem. 1995, 99, 1052–1059.
20. Holtzer, A. J. Phys. Chem. 1995, 99, 13048–13049.
21. Liu, Y.; Sturtevant, J. M. Protein Sci. 1995, 4, 2559–2561.
22. Liu, Y.; Sturtevant, J. M. Biophys. Chem. 1997, 64, 121–126.
23. Arena, G.; Pappalardo, A.; Pappalardo, S.; Gattuso, G.; Notti, A.; Parisi, M. F.; Pisagatti, I.; Sgarlata, C. J. Therm. Anal. Calorim. 2015, 121, 1073–1079.
24. Mizoue, L. S.; Tellinghuisen, J. Biophys. Chem. 2004, 110, 15–24.
25. Inoue, Y.; Wada, T. In Advances in Supramolecular Chemistry, vol. 4, Gokel, G. W., Ed.; JAI Press: Greenwich, 1997; pp 55–96.
26. Velazquez Campoy, A.; Freire, E. Biophys. Chem. 2005, 115, 115–124.
27. Skinner, H. A.; Experimental Thermochemistry, vol.2; Interscience Publishers: New York, 1962.
28. Benzinger, T. H. Proc. Natl. Acad. Sci. U. S. A. 1956, 42, 109–113.
29. Christensen, J. J.; Izatt, R. M.; Hansen, L. D. Rev. Sci. Instrum. 1965, 36, 779–783.
30. Hansen, L. D.; Christensen, J. J.; Izatt, R. M. J. Chem. Soc. Chem. Commun. 1965, 3, 36–37.
31. Hansen, L. D.; Izatt, R. M.; Eatough, D. J.; Jensen, T. E.; Christensen, J. J. In Analytical Calorimetry; Porter, R. S., Johnson, J. F., Eds.; Plenum Press: New York and
London, 1974.
32. Izatt, R. M.; Hansen, L. D.; Eatough, D. J.; Jensen, T. E.; Christensen, J. J. In Analytical Calorimetry; Porter, R. S., Johnson, J. F., Eds.; Plenum Press: New York and London,
1974. and references therein.
33. Hansen, L. D.; Jensen, T. E.; Mayne, S.; Eatough, D. J.; Izatt, R. M.; Christensen, J. J. J. Chem. Thermodyn. 1975, 7, 919–926. and references therein.
34. Sunner, S.; Wadso, I. J. LKB Inst. 1966, 13, 1.
35. Wadsö, I.; Wadsö, L. J. Therm. Anal. Calorim. 2005, 82, 553–558.
36. Hansen, L. D.; Fellingham, G. W.; Russell, D. J. Anal. Biochem. 2011, 409, 220–229.
37. Demarse, N. A.; Quinn, C. F.; Eggett, D. L.; Russell, D. J.; Hansen, L. D. Anal. Biochem. 2011, 417, 247–255.
38. Wadsö, I.; Goldberg, R. N. Pure Appl. Chem. 2001, 73, 1625–1639.
39. Sgarlata, C.; Zito, V.; Arena, G. Anal. Bioanal. Chem. 2013, 405, 1085–1094.
40. Sgarlata, C.; Mugridge, J. S.; Pluth, M. D.; Tiedemann, B. E. F.; Zito, V.; Arena, G.; Raymond, K. N. J. Am. Chem. Soc. 2010, 132, 1005–1009.
41. Tellinghuisen, J. Anal. Biochem. 2007, 360, 47–55.
42. Turnbull, W. B.; Daranas, A. H. J. Am. Chem. Soc. 2003, 125, 14859–14866.
43. Tellinghuisen, J. J. Phys. Chem. B 2005, 109, 20027–20035.
44. Arena, G.; Calì, R.; Lombardo, G. G.; Rizzarelli, E.; Sciotto, D.; Ungaro, R.; Casnati, A. Supramol. Chem. 1992, 1, 19–24.
45. Arena, G.; Contino, A.; Lombardo, G. G.; Sciotto, D. Thermochim. Acta 1995, 264, 1–11.
46. Scharff, J. P.; Mahjoubi, M.; Perrin, R. New J. Chem. 1991, 15, 883–887.
47. Sgarlata, C.; Arena, G.; Fortuna, C. G.; Sciotto, D.; Bonaccorso, C. Supramol. Chem. 2016, 28, 544–550.
48. Wiseman, T.; Williston, S.; Brandts, J. F.; Lint, L. N. Anal. Biochem. 1989, 179, 131–137.
49. Tellinghuisen, J. Anal. Biochem. 2008, 373, 395–397.
50. Izatt, R. M.; Terry, R. E.; Haymore, B. L.; Hansen, L. D.; Dalley, N. K.; Ayondet, A. G.; Christensen, J. J. J. Am. Chem. Soc. 1976, 98, 7620–7626.
51. Briggner, L. E.; Wadsö, I. J. Biochem. Biophys. Methods 1991, 22, 101–118.
52. Buschmann, H. J.; Schollmeyer, E. Thermochim. Acta 1999, 333, 49–53.
53. Derenleau, D. A. J. Am. Chem. Soc. 1969, 91, 4044–4049.
54. De Robertis, A.; Rigano, C.; Sammartano, S.; Zerbinati, O. Thermochim. Acta 1987, 115, 241–248.
55. Christensen, J. J.; Wrathall, D. P.; Oscarson, J. O.; Izatt, R. M. Anal. Chem. 1968, 40, 1713–1717.
Modern Calorimetry: An Invaluable Tool in Supramolecular Chemistry 237

56. Zhang, Y. L.; Zhang, Z. Y. Anal. Biochem. 1998, 261, 139–148.


57. Arena, G.; Calì, R.; Cucinotta, V.; Musumeci, S.; Rizzarelli, E.; Sammartano, S. J. Chem. Soc., Dalton Trans. 1983, 1271–1278.
58. Arena, G.; Calì, R.; Cucinotta, V.; Musumeci, S.; Rizzarelli, E.; Sammartano, S. J. Chem. Soc., Dalton Trans. 1984, 1651–1657.
59. Best, M. D.; Tobey, S. L.; Anslyn, E. V. Coord. Chem. Rev. 2003, 240, 3–15.
60. Jadhav, V. D.; Schmidtchen, F. P. Org. Lett. 2005, 7, 3311–3314.
61. Rekharsky, M. V.; Mori, T.; Yang, C.; Ko, Y. H.; Selvapalam, N.; et al. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 20737–20742.
62. Jeon, W. S.; Moon, K.; Park, S. H.; Chun, H.; Ko, Y. H.; et al. J. Am. Chem. Soc. 2005, 127, 12984–12989.
63. Liu, S.; Ruspic, C.; Mukhopadhyay, P.; Chakrabarti, S.; Zavalij, P. Y.; Isaacs, L. J. Am. Chem. Soc. 2005, 127, 15959–15967.
64. Williams, D. H.; Stephens, E.; O’Brien, D. P.; Zhou, M. Angew. Chem. Int. Ed. 2004, 43, 6596–6616.
65. Rekharsky, M. V.; Inoue, Y. In Microcalorimetry in cyclodextrins and their complexes; Dodziuk, H., Ed.; Wiley-VCH: Weinheim, 2006; pp 199–230.
66. Bonaccorso, C.; Ciadamidaro, A.; Zito, V.; Sgarlata, C.; Sciotto, D.; Arena, G. Thermochim. Acta 2012, 530, 107–115.
67. Sun, H.; Gibb, C. L. D.; Gibb, B. C. Supramol. Chem. 2008, 20, 141–147.
68. Biedermann, F.; Vendruscolo, M.; Scherman, O. A.; De Simone, A.; Nau, W. M. J. Am. Chem. Soc. 2013, 135, 14879–14888.
69. Biedermann, F.; Uzunova, V. D.; Scherman, O. A.; Nau, W. M.; De Simone, A. J. Am. Chem. Soc. 2012, 134, 15318–15323.
70. Biedermann, F.; Scherman, O. A. J. Phys. Chem. B 2012, 116, 2842–2849.
71. Jacobson, D. R.; Khan, N. S.; Collé, R.; Fitzgerald, R.; Laureano-Pérez, L.; et al. Proc. Natl. Acad. Sci. U. S. A. 2011, 108, 10969–10973.
72. Smithrud, D. B.; Wyman, T. B.; Diederich, F. J. Am. Chem. Soc. 1991, 113, 5420–5426.
73. Sinanoglou, O. Int. J. Quantum Chem. 1980, 18, 381–392.
74. Gunasekara, R. W.; Zhao, Y. J. Am. Chem. Soc. 2015, 137, 843–849.
75. Blandamer, M. J. In Biocalorimetry: Applications of Calorimetry in the Biological Sciences; Ladbury, J. E., Chowdhry, B. Z., Eds.; Wiley: Chichester, 1998.
76. Sgarlata, C.; Raymond, K. N. Anal. Chem. 2016, 88, 6923–6929.
77. Ciardi, M.; Galàn, A.; Ballester, P. J. Am. Chem. Soc. 2015, 137, 2047–2055.
78. Guo, D. S.; Liu, Y. Acc. Chem. Res. 2014, 47, 1925–1934.
79. Corbellini, F.; Costanzo, L. D.; Crego-Calama, M.; Geremia, S.; Reinhoudt, D. N. J. Am. Chem. Soc. 2003, 125, 9946–9947.
80. Rebek, J., Jr. Acc. Chem. Res. 2009, 42, 1660–1668.
81. Linder, I.; Leisering, S.; Puttreddy, R.; Rades, N.; Rissanen, K.; Schalley, C. A. Chem. Eur. J. 2015, 21, 13035–13044.
82. Cresswell, A. L.; Piepenbrock, M. O. M.; Steed, J. W. Chem. Commun. 2010, 46, 2787–2789.
83. Bonaccorso, C.; Brancatelli, G.; Forte, G.; Arena, G.; Geremia, S.; Sciotto, D.; Sgarlata, C. RSC Adv. 2014, 4, 53575–53587.
84. Bonaccorso, C.; Sgarlata, C.; Grasso, G.; Zito, V.; Sciotto, D.; Arena, G. Chem. Commun. 2011, 47, 6117–6119.
85. Hansen, L. D.; Transtrum, M. K.; Quinn, C.; Demarse, N. Biochim. Biophys. Acta 2016, 1860, 957–966.
86. Williams, B. A.; Toone, E. J. J. Org. Chem. 1993, 58, 3507–3510.
87. Brown, C. J.; Toste, F. D.; Bergman, R. G.; Raymond, K. N. Chem. Rev. 2015, 115, 3012–3035.
88. Kaphan, D. M.; Levin, M. D.; Bergman, R. G.; Raymond, K. N.; Toste, F. D. Science 2015, 350, 1235–1238.
2.12 Chromatography in Supramolecular and Analytical Chemistry
of Calixarenes
O Kalchenko, Institute of Organic Chemistry NASU, Kyiv, Ukraine
J Lipkowski, Cardinal Stefan Wyszynski University in Warsaw, Warsaw, Poland
V Kalchenko, Institute of Organic Chemistry NASU, Kyiv, Ukraine
Ó 2017 Elsevier Ltd. All rights reserved.

2.12.1 Introduction 239


2.12.2 Chromatographic Analysis of Calixarenes 240
2.12.2.1 Separation of Homologues of tert-Butylcalix[n]arenes 240
2.12.2.2 Chromatographic Analysis of Water-Soluble Calixarenes 240
2.12.2.3 Separation of Chiral Calixarenes 241
2.12.3 Chromatographic Study of the Calixarene Host–Guest Complexation 243
2.12.3.1 Complexation of Aromatic Compounds 243
2.12.3.2 Complexation of Uracil and Adenine Derivatives 244
2.12.3.3 Complexation of Amino Acids 244
2.12.3.4 Calixarene Complexation With ADP and ATP 248
2.12.3.5 Complexation of Herbicides 248
2.12.4 Chromatographic Calixarene Phases 248
2.12.4.1 Synthesis of Calixarene Stationary Phases 248
2.12.4.2 Calixarene Phases in Gas Chromatography 249
2.12.4.3 Stationary Calixarene Phases in Liquid Chromatography 254
2.12.4.4 Calixarene Additives to the Mobile Phase 256
2.12.5 Conclusion 258
Acknowledgment 259
References 259

2.12.1 Introduction

Bowl-shaped macrocyclic compoundsdcalixarenes, thiacalixarenes, calixresorcinarenes, and calixpyrroles are widely used as a plat-
form for the design of specific molecular receptors capable of selective recognition of various cations, anions, and neutral molecules.
Such processes of molecular recognition are based on principles of supramolecular chemistry. The capability for molecular recog-
nition and formation of Host–Guest supramolecular complexes is the basis of application of macrocycles in chemistry, physics,
biology, material science, and nanotechnology.
Calixarene chemistry is closely associated with chromatography. David Gutsche was the first to use liquid chromatography for
separation, isolation, and identification of calixarene homologues containing different amounts of phenolic subunits in the mole-
cule. Later, calixarenes were utilized for developing the mobile and stationary phases for gas and reversed-phase high-performance
liquid chromatography (RP-HPLC). The RP-HPLC method was applied for examination of substrate–receptor interaction of calix-
arenes with organic molecules and biomolecules and for determination of stability constants of the supramolecular complexes
formed.
This article is dedicated to the application of the HPLC method in analytical and supramolecular chemistry of calixarenes. We
also considered the use of calixarenes as stationary phases for gas and liquid chromatography, and their application as selective
additives to the mobile phase in liquid chromatography.
The article consists of three sections. The first section presents an application of chromatography in the analytical chemistry of
calixarenes. Separation of tert-butylcalix[n]arene homologues, chromatographic behavior of functionalized calixarenes, and enan-
tioseparation of chiral calixarenes are described.
The second section is devoted to the application of the chromatography method in the supramolecular chemistry of calixarenes.
The use of the HPLC method for determination of stability constants of the calixarene complexes with amino acids, peptides,
adenine and uracil derivatives, adenosine diphosphate (ADP), adenosine triphosphate (ATP), herbicides are discussed.
The third section describes the preparation and application of calixarene containing solid phases in gas and liquid
chromatography and the use of calixarene additives to the mobile phases that improve separation of analytes by the RP-
HPLC method.

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.13799-0 239


240 Chromatography in Supramolecular and Analytical Chemistry of Calixarenes

2.12.2 Chromatographic Analysis of Calixarenes

The conditions of chromatographic analysis of calixarenes are dependent on the size of the macrocycle, its conformation, and on the
nature of the substituents at the upper and lower rim of the macrocyclic platform. These factors determine the solubility of calix-
arenes in the mobile phases and physical interaction with the material of the column support.

2.12.2.1 Separation of Homologues of tert-Butylcalix[n]arenes


The method of RP-HPLC in calixarene chemistry was used by David Gutsche 1 to identify homologues of tert-butylcalix[n]arenes
(1.1–1.9) (Fig. 1) formed by the cyclocondensation of tert-butylphenol with formaldehyde. Analysis was performed on a Spherisorb
C18 hydrophobic support in the flow of the mobile phase MeCN/MeOBut/CH2Cl2/CH3COOH. Later, similar conditions were used
for chromatographic study of the tert-butylcalix[n]arenes in Ref. 2 (Table 1).
Chromatographic characteristics tR and k0 increase with increasing number of phenol subunits in the tert-butylcalix[n]arenes
( Table 1).
In the work 3 tert-butylcalix[4,6,8]arenes (1.1), (1.3), and (1.5) were separated by supercritical fluid chromatography in carbon
dioxide flow on the phases of different polaritydC18, RP-18, CN, and Diol. The chromatographic characteristics of the analytes are
dependent on the nature of the modificator (methanol or chloroform) of carbon dioxide. The peaks obtained by supercritical fluid
chromatography were narrower and symmetrical compared to the peaks obtained by HPLC.

2.12.2.2 Chromatographic Analysis of Water-Soluble Calixarenes


The method of reversed-phase high-performance liquid chromatography was used to study a series of water-soluble calix[4,6]are-
nes, thiacalixarenes, and calix[4]resorcinarenes functionalized with carboxylic, sulfonous, and phosphonous acid moieties (LiChro-
sorb RP 18, MeCN/H2O or Separon SGX NH2, MeOH/MeCN/THF/H2O,3–6 Zorbax CN, H2O). 7,8 As was shown the retention

Figure 1 tert-Butylcalix[n]arenes (1.1–1.9).

Table 1 Retention times t R, retention factors k 0 , asymmetry coefficients K S of the tert-butylcalix[n]arenes


(Spherisorb ODS 1, MeCN/MeOBut/CH2Cl2/CH3COOH (84.6:10:4.5:0.9, v/v), 22 C)

tert-Butylcalix[n]arenes
(number of benzene rings) tR (min) k0 KS

1.1 (4) 4.86 0.53 1.00


1.2 (5) 8.01 1.44 1.12
1.3 (6) 7.84 1.40 1.34
1.4 (7) 9.71 1.80 1.46
1.5 (8) 10.32 1.93 1.57
1.6 (9) 11.84 2.56 1.70
1.7 (10) 12.81 2.68 1.86
1.8 (11) 18.82 4.41 2.10
1.9 (12) 23.26 5.80 2.50

Modified from Baudry, R.; Kalchenko, O.; Dumazet-Bonnamour, I.; Vocanson, F.; Lamartine, R. J. Chromatogr. Sci. 2003, 41,
157–163.
Chromatography in Supramolecular and Analytical Chemistry of Calixarenes 241

Figure 2 Calix[4]arene hydroxymethylphosphonous acids (1.10–1.13).

factors are significantly dependent on the type and nature of the macrocycle and the nature of substituents at the upper and
lower rim.
In the work 3 the chromatographic behavior of the water-soluble calixarene hydroxymethylphosphonous acids (1.10–1.13)
(Fig. 2) and their sodium salts was investigated. The analysis was performed on Zorbax CN column using the mobile phase
THF/H2O (90:10, v/v) for the acids and H2O/MeCN (99:1, v/v) for their sodium salts. The wavelength of the UV-detector was
254 nm. Retention times tR are within 3.34–6.55 min for the calixarene phosphonous acids and 3.14–12.37 min for their
sodium salts.

2.12.2.3 Separation of Chiral Calixarenes


Chiral calixarenes are promising compounds for the development of a variety of chiral materials and technologies: catalysts, chro-
matographic phase, nonlinear optic materials, enantiomerically pure pharmaceuticals, etc.9–13
HPLC has been used for analytical and preparative enantioseparation of inherently chiral calixarenes. The enantiomeric separa-
tion of asymmetrically alkylated inherently chiral calixarenes was performed on the chiral chromatographic supports Chiralpak
OP(þ) 14 or Sumipax OA-2000.15 Enantiomers of inherently chiral calixarene diphosphate (1.14) (Fig. 3) were successfully sepa-
rated on Chiralcel ODH column with hexane/isopropanol or heptane/isopropanol (95:5, v/v) mobile phases.16
Enantiomers of hexahomooxacalix[3]arene with various substituents at the upper rim were separated on chiral semipreparative
column Chiralpak AD with mobile phase hexane/methanol/isopropanol (90:6:4, v/v). 17
The high cost of the chiral columns stimulates development of more economical methods of the enantioseparation using chiral
solvating additives to the mobile phases. To separate inherently chiral calixarene phosphorus acids and their alkyl esters (1.15–
1.19) ( Fig. 4) a number of chiral additives such as D-()-tartaric acid, L-()-a-phenylethylamine, (1S, 2R)-(þ)-ephedrine hydro-
chloride, L-()-menthol, L-alanine forming diastereomeric salts, or hydrogen-bonded associates with P(O)(OH)2 or P(O)(OEt)2
groups were examined in the works.18,19
Diastereomeric salts formed by the acids with chiral L-()-a-phenylethylamine were effectively separated on the columns
Separon SGX C18 and Partisil 5 ODS 3. Neutral calixarene diphosphate (1.19) was separated on the LiChrosorb RP18 as diastereo-
meric associates with D-tartaric acid ( Table 2).
Semipreparative HPLC diastereoseparation of inherently chiral calix[4]arene (1S)-camphorsulfoester (1.20b) was performed on
achiral analytical column Zorbax CN (250  4.6 mm) using mobile phase hexane/THF (96:4, v/v), temperature 20 C, and wave-
length l ¼ 254 nm ( Fig. 5).20

Figure 3 Enantiomers of inherently chiral calix[4]arene diphosphate (1.14).


242 Chromatography in Supramolecular and Analytical Chemistry of Calixarenes

Figure 4 Inherently chiral calix[4]arenes (1.15–1.20).

Table 2 Separation of inherently chiral calix[4]arenes (MeCN/H2O, 86:14, v/v), 254 nm, 22 C

Retention time of Separation


Calixarene Column (chiral additive) stereoisomeres, tR (min) coefficient, RS

(1.15) Separon SGX C18 (L-()-a-phenylethylamine) 1.20 2.5


2.33
(1.16) Partisil 5 ODS 3 (L-()-a-phenylethylamine) 9.83 0.9
14.75
(1.17) Separon SGX C18 (L-()-a-phenylethylamine) 2.95 0.8
3.07
(1.18) Separon SGX C18 (L-()-a-phenylethylamine) 1.20 2.5
2.33
(1.19) LiChrosorb RP18 (D-tartaric acid) 6.17 0.6
6.83
(1.20) Zorbax CN 1.33 1.2
1.65

Modified from Kalchenko, O. I.; Tairov, M. O.; Vysotsky, M. O.; Lipkowski, J.; Kalchenko, V. I. Enantiomer 2000, 5, 385–390.

Figure 5 Diastereoseparation of inherently chiral calix[4]arene(1S)-camphorsulfoester (1.20b). Modified from Kalchenko, O. I.; Kalchenko, V. I.
Cromatografia in Chemistry of Calixarenes; Naukova Dumka: Kyiv, 2013.
Chromatography in Supramolecular and Analytical Chemistry of Calixarenes 243

2.12.3 Chromatographic Study of the Calixarene Host–Guest Complexation

Calixarenes are widely used for the design of highly selective complexants (receptors) for molecules and ions. As a rule to determine
the stability constants of the calixarene complexes the NMR, calorimetry, and UV–vis methods are used. However, the use of these
techniques is often limited by poor solubility of calixarenes. In this case, the RP-HPLC method can be used to determine the stability
constants in water 21 or water-organic22 solutions. The method is based on analysis of dependence of the retention time tR, and the
retention factor k0 of an analyte on the calixarene concentration in the mobile phase.21,22
Calixarene additives decrease the retention times of the analytes due to formation of the supramolecular Host–Guest complexes.
On the other hand, reversible calixarene sorption increases polarity of the chromatographic support surface.
In the chromatographic system for the analyte (S) and calixarene (CA) there is a balance between the mobile phase (m) and the
stationary phase (s):
KS KD KC
ðSÞs $ ðSÞm þ ðCA Þm $ ðCA  SÞm $ ðCA  SÞs ;

where (S)s is the analyte in the stationary phase, (S)m is the analyte in the mobile phase, KS is the distribution constant of the analyte
between the stationary and mobile phases, KD is the dissociation constant of the complex in the mobile phase, KC is the distribution
constant of the complex between stationary and mobile phases, (S)m þ (CA)m is the analyte þ calixarene in mobile phase, (CA–S)m
is the analyte–calixarene complex in mobile phase, (CA–S)s is the analyte–calixarene complex in stationary phase. Taking into
account such interactions a formula for calculation of the stability constant of the calixarene–analyte supramolecular complex was
proposed:
 
LA ¼ k00 1=k0  1 k00 ½CA ;

where k0 0 and k0 are the analyte retention factors before and after calixarene addition to the mobile phase, [CA] is the calixarene
concentration in the mobile phase. The linear dependence of 1/k0 on calixarene concentration in the mobile phase indicates 1:1
stoichiometry of the complex.

2.12.3.1 Complexation of Aromatic Compounds


The stability constant LA of the complexes of tert-butylcalix[n]arenes ( Fig. 1) with a series of aromatic compounds (m-xylene,
p-xylene, tert-butylphenol, pentachlorophenol, naphthalene, anthracene, fluoranthene) was determined by the HPLC method using
column Spherisorb ODS 1 and the mobile phase MeCN/MeOBu-t/CH2Cl2/CH3COOH (84.6:10:4.5:0.9, v/v).2
The LA values of the complexes (within 781–7672 M 1) are dependent on the calixarene size and nature of the aromatic Guest
molecule. The largest LA value (7672 M 1) was observed for the complex of calix[8]arene (1.5) with pentachlorophenol, and the
lowest (781 M 1) for the calix[10]arene (1.7) complex with anthracene. However, the LA value for the larger calix[12]arene (1.9)
complexed with anthracene is 1468 M 1. Increasing the size of the calixarene enhances the stability of their complexes with naph-
thalene and fluoranthene.
The HPLC complexation study of 5,17-bis-(N-tolylmethyleneimino)-25,27-dipropoxycalix[4]arene (2.1), 23 tetraphosphorylte-
trapropoxycalix[4]arene (2.2),24 and octakis(diethoxyphosphoryloxy)tert-butylcalix[8]arene (2.3)25 (Fig. 6) with 41 benzene

Figure 6 Functionalized calix[n]arenes (2.1–2.3).


244 Chromatography in Supramolecular and Analytical Chemistry of Calixarenes

derivatives was carried out. The determination was performed in the mobile phase MeCN/H2O (86:14, v/v) using Separon SGX C18
as the stationary phase.
The stability constants of the calixarene complexes range widely from 13 to 2795 M 1 24,25 depending on the nature, number,
and position of substituents in the Guest aromatic ring, as well as on size and substitution of the calixarene platform. The stability
constants are influenced by proton donor groups of the Guest molecules forming hydrogen bonds with oxygen atoms of phos-
phoryl groups of the calixarene Host. Octaphosphorylated calix[8]arene (2.3) is the most effective complexant among the investi-
gated compounds (2.1–2.3). The maximum LA value (2795 M 1) is observed for its complex with proton donative resorcinol.
Stability constants LA of the complexes of resorcinarenes (2.4–2.11) ( Fig. 7) with OH, OCH3, CHO, CCl3, CF3, CH3, C(CH3)3,
Cl, F, and NO2 substituted benzenes are within 17–863 M 1 25 (mobile phase MeCN/H2O, 86:14 v/v, column Separon SGX C18).
The stability constants of the complexes are similar to those for a-cyclodextrin.26,27

2.12.3.2 Complexation of Uracil and Adenine Derivatives


Calixarenes are utilized as platforms in the design of synthetic receptors 28,29 for recognition and binding of bio-active substrates
such as amino acids,30 dipeptides,31 proteins,32 choline and acetylcholine,33 vitamins,34 nucleic acids,35 and nucleotides.35 The
Host–Guest substrate–receptor interactions simulate the enzyme functions and is the basis for calixarene bio-activity.36
The stability constants of the complexes of di(tetra)propoxycalixarenes (2.12) and (2.13) or their upper rim dialkylphospho-
nous acid derivatives (2.2, 2.14–2.17) ( Fig. 8) with biologically important nucleic bases adenines and uracils were determined
in the mobile phase CH3OH/CH3CN/TGV/H2O (15:10:5:70, v/v) using Separon SGX NH2 column37 (Table 3).
The linear plots of 1/k0 ns the calixarene concentration in the mobile phase confirm 1:1 stoichiometry of the Host–Guest
inclusion complexes.
The upper rim phosphonyl groups in tetrapropoxycalixarenes (2.2, 2.15–2.17) spatially hinder the complexation process. The
opposite situation is observed in dipropoxycalixarene phosphonate (2.14) where phosphonyl groups at the upper rim increase
stability of the inclusion complexes.

2.12.3.3 Complexation of Amino Acids


To investigate the calixarene complexation with amino acids different methods such as UV–vis, fluorescence spectroscopy,38–41
liquid–liquid extraction, 42,43 membrane transport,44 mass spectrometry,45–47 and surface plasmon resonance48 were used.
Complexation of a series of water-soluble calix[4,6]arenes and thiacalix[4]arenes decorated at the upper rim by sulfonous, phos-
phonous, and phosphinous acid moieties (2.18–2.22) ( Fig. 9) with polar, nonpolar, positively or negatively charged, and aromatic
amino acids was investigated by the RP-HPLC method in water solution using Separon SGX CN and Zorbax CN supports7,21,49–51
(Table 4).
As shown in Table 4 the LA values of the complexes are dependent on the calixarene structure as well as on the amino acid
nature. The positively charged His, Arg, Lys, and aromatic amino acids Tyr, Trp, and Phe are most effectively bonded by the
calixarenes.

Figure 7 Functionalized calix[4]resorcinarenes (2.4–2.11).


Chromatography in Supramolecular and Analytical Chemistry of Calixarenes 245

Figure 8 Alkoxycalix[4]arenes (2.12, 2.13) and their upper rim phosphorylated derivatives (2.14–2.17).

Table 3 The stability constants (M 1) of the calix[4]arene (2.12–2.17, 2.2) complexes with adenine and uracil derivatives

Calixarene
N Substrate (2.12) (2.13) (2.14) (2.15) (2.16) (2.17) (2.2)

1 Adenine 14,000 12,260 21,500 10,500 2700 78,008 2300


2 9-Methyladenine 11,300 12,600 20,000 9500 2700 63,700 2000
3 5-Aminouracil 2650 3250 6600 3600 1550 4484 2850
4 5-Ethyluracil 7000 8400 13,650 6150 4800 16,011 8750
5 6-Amino-1-methyluracil 6800 9700 11,300 4700 2600 8472 6900
6 5-Methyluracil 8000 11,800 11,400 5650 3300 14,755 1300
7 1,3-Dimethyluracil 5300 5200 12,000 8200 2900 10,678 5300
8 6-Amino-1,3-dimethyluracil 5640 6200 12,300 7750 3400 11,247 4700
9 6-Methyluracil 6250 8860 14,450 5950 5100 7750 14,755
10 Uracil 6500 8900 12,300 4900 6800 6900 13,098
11 5-Chlorouracil 9200 12,000 11,100 5400 3500 1200 10,948
12 5-Nitrouracil 41,700 54,300 41,400 22,200 5150 4100 18,482

Calixarene

N Substrate (2.13) (2.15) (2.16) (2.2) (2.12) (2.14) (2.17)

1 Adenine 12,260 10,500 2700 2300 14,000 21,500 78,008


2 9-Methyladenine 12,600 9500 2700 2000 11,300 20,000 63,700
3 5-Aminouracil 3250 3600 1550 2850 2650 6600 4484
4 5-Ethyluracil 8400 6150 4800 8750 7000 13,650 16,011
5 6-Amino-1-methyluracil 9700 4700 2600 6900 6800 11,300 8472
6 5-Methyluracil 11,800 5650 3300 1300 8000 11,400 14,755
7 1,3-Dimethyluracil 5200 8200 2900 5300 5300 12,000 10,678
8 6-Amino-1,3-dimethyluracil 6200 7750 3400 4700 5640 12,300 11,247
9 6-Methyluracil 8860 5950 5100 7750 6250 14,450 14,755
10 Uracil 8900 4900 6800 6900 6500 12,300 13,098
11 5-Chlorouracil 12,000 5400 3500 1200 9200 11,100 10,948
12 5-Nitrouracil 54,300 22,200 5150 4100 41,700 41,400 18,482

Modified from Kalchenko, O., Marcinowicz, A.; Poznanski, J.; Cherenok, S.; Solovyov, A.; Zielenkiewicz, W.; Kalchenko, V. J. Phys. Org. Chem. 2005, 18, 578–585.

Stability constants and free Gibbs energies of the amino acid complexes with calix[4]arenes (1.10–1.12) ( Fig. 2) modified at the
upper rim by one or two fragments of hydroxymethyl phosphonous acids were determined in H2O/MeCN (99:1, v/v) solution on
Zorbax CN column52 (Table 5).
The calixarenes (1.10–1.12) are effective binders of the amino acids. Introduction of the second acid residue at the macrocyclic
skeleton enhances stability of the complexes. 53
246 Chromatography in Supramolecular and Analytical Chemistry of Calixarenes

Figure 9 Water-soluble (thia)calix[n]arene sulfonous and phosphonous acids (2.18–2.22).

Table 4 Stability constants LA (M 1) of the (thia)calix[4,6]arenes (2.19–2.22) complexes with the amino acids

Calixarenes
N Amino acids (2.19a) a (2.19b) a (2.20) b (2.21) b (2.22) b

1 Met 758 744 877 630 890


2 Leu 1164 941 719 495 533
3 Nor 1365 839 1398 663 743
4 Ala 1805 673 783 278 533
5 Gly 1949 578 170 544 578
6 Pro 1518 985 493 352 540
7 Ser 1447 564 1144 507 912
8 Cys 1307 888 971 545 608
9 His 2241 2267 1005 1410 1566
10 Arg 3031 3601 2425 6373 10,140
11 Lys 3134 2310 630 4240 8899
12 Asp 6585 4091 1944 1952 2496
13 Tyr 2041 1923 6713 2562 3973
14 Trp 2801 2460 5917 2057 1965
15 Phe 4157 933 6032 1619 1539

Separon SGX CN, 36 C, 254 nm.


a

Zorbax CN, 36 C, 254 nm.


b

Modified from Kalchenko, O.; Drapailo, A.; Shishkina, S.; Shishkin, O.; Kharchenko, S.; Gorbatchuk, V.; Kalchenko, V. Supramol. Chem. 2013, 25 (5), 263–268.
Chromatography in Supramolecular and Analytical Chemistry of Calixarenes 247

Table 5 Stability constants LA (M 1) and free Gibbs energies DG (kcal mol 1) of the complexes of calix[4]arene hydroxymethylphosphonous acids
(1.10–1.12) with the amino acids

Calixarenes
(1.10) (1.11) (1.12)
Amino acids LA DG LA DG LA DG

Gly 27,727 25,30 20,888 24,60 31,234  25,60


Ala 21,203 24,64 21,299 24,65 47,299  26,62
Pro 38,243 26,10 15,873 23,92 40,082  26,21
His 31,199 25,60 28,174 25,34 48,189  26,67
Lys 32,485 25,69 29,295 25,44 42,572  26,36
Asp 28,807 25,40 21,954 24,73 29,947  25,49
Arg 27,419 25,28 23,567 24,90 35,992  25,95
Phe 26,581 25,20 26,927 25,23 30,185  25,51
Trp 23,414 24,89 20,795 24,59 27,490  25,28
Tyr 26,885 25,23 17,546 24,17 20,983  24,61

Modified from Kalchenko O.; Cherenok, S. O.; Yuschenko, O. A.; Kalchenko, V. I. J. Incl. Phenom. 2013, 76 (1–2), 29–36.

It was shown that the calixarene tetrakis-methylenebisphosphonous acid (2.23) ( Fig. 10) is an effective complexant for amino
acids Gly (280 M 1), Pro (814 M 1), and Arg (2576 M 1) and the tetrapeptide Gly-Pro-Arg-Pro (3395 M 1) in H2O/MeOH,
50:50, v/v solution (Separon SGX CN column).54–56
The stability constants of calixarene (2.23) complexes correlate with hydrophobicity of the amino acids and tetrapeptide
( Fig. 11), indicating a role of hydrophobic interactions in the complexation.

Figure 10 Calix[4]arene tetrakis-methylenebisphosphonous acid (2.23).

Figure 11 Correlation of log LA with log P of Gly, Pro, Arg and Gly-Pro-Arg-Pro for the calix[4]arene (2.23) complexes (r ¼ 0.92). Modified from
Kalchenko, O. I.; Kalchenko, V. I. Cromatografia in Chemistry of Calixarenes; Naukova Dumka: Kyiv, 2013.
248 Chromatography in Supramolecular and Analytical Chemistry of Calixarenes

It should be noted that calixarene methylenebisphosphonous acid (2.23) has been studied with regard to its effect on fibrin
polymerization. It was hypothesized that the calixarene (2.23) blocks fibrin formation by combining with polymerization site
“A” (Aa17–19), which ordinarily initiates protofibril formation in a “knob-hole” manner. This suggestion was confirmed by the
HPLC experiment, which showed a Host–Guest inclusion complex of the calixarene with the synthetic peptide Gly-Pro-Arg-Pro,
an analog of site “A.” Further confirmation of the inhibitor acting at the initial step of the reaction was obtained by electron micros-
copy, no evidence of protofibril formation being evident. The calixarene also doubled both the prothrombin time and the activated
partial thromboplastin time in normal human blood plasma at concentrations of 7.13  10 5 M and 1.10  10 5 M, respectively.
Calix[n]arene sulfonous acids (2.18, 2.19) have been investigated for in vitro anticoagulant activity. 57 All the sulfonatocalix[n]
arenes showed antithrombin activity at concentrations in the micromolar range that were 10- to 50-fold higher than that of heparin.
The mechanism of action seems to be via their interaction with the serine protease inhibitor. These data demonstrate that calixarene
phosphonous acids and calixarene sulfonous acids are specific inhibitors of fibrin polymerization and blood coagulation and can be
used for the design of a new class of antithrombotic agents.
Calix[4]arenes bearing methylenebisphosphonous or hydroxymethylenephopshonous acid fragments displayed strong inhibi-
tion of some alkaline phosphatases and Yersinia protein tyrosine phosphatase. The mechanisms of the enzyme inhibition have been
discussed using a molecular docking approach by computational modeling of the inhibitors into active centers of the phosphatases.
The compounds are also selective modulators of calcium pumps in smooth muscle cells.58–60

2.12.3.4 Calixarene Complexation With ADP and ATP


The chromatographic investigation of calixarene diaminophosphonous acid (2.24) ( Fig. 12) complexation with ATP and ADP in
the mobile phase of H2O/MeCN (53:47, v/v) on Zorbax CN column was carried out.61
ATP forms a more stable calixarene complex (LA ¼ 5083 M 1) compared with ADP (LA ¼ 2938 M 1). The values are close to the
stability constants of ATP and ADP with calixarene trimethylammonium salt (2.25) determined by NMR in D2O solution (6700
and 2600 M 1, respectively). 35
It was shown that calix[4]arene aminophosphonous acid (2.24) hydrolyzes ATP. The velocity of the calixarene-dependent hydro-
lysis of ATP exceeds the velocity of spontaneous hydrolysis of ATP by at least 14–15 times. The data obtained can be a basis for
designing synthetic catalysts for ATP hydrolysis and also for subsequent investigation of both enzymatic and nonenzymatic ATP
hydrolysis reactions, processes of ATP-dependent Ca2 þ transporting in subcellular membrane structures. 61

2.12.3.5 Complexation of Herbicides


Complexation of calix[4]arenes (2.26–2.28) possessing two or four phosphonic acid moieties ( Fig. 13) with the herbicides 2,4-
dichlorophenoxyacetic acid (2,4-D) or atrazine (AT) in water (Separon SGX CN column) has been investigated by the RP-HPLC
method.
The stability constants of the 1:1 Host–Guest complexes in range 772–5077 M 1 (2,4-D) and 2513–6785 M 1 (AT) have been
determined. The constants are dependent on the conformation and stereochemical mobility of the calixarene skeleton, the number
of the dihydroxyphosphonyl groups at the upper rim, as well as the acid–base properties of the Guest molecules. 5

2.12.4 Chromatographic Calixarene Phases


2.12.4.1 Synthesis of Calixarene Stationary Phases
Calixarenes, thiacalixarenes, calixresorcinarenes, and calixpyrroles have been widely used to develop the stationary phases for anal-
ysis of organic or inorganic compounds by gas chromatography, liquid chromatography, and electrophoresis.62–67

Figure 12 Calix[4]arene aminophosphonous acid (2.24) and ammonium salt (2.25).


Chromatography in Supramolecular and Analytical Chemistry of Calixarenes 249

Figure 13 Calix[4]arene phosphonous acids (2.26–2.28).

The calixarenes bearing different functional groups at the upper and lower rim were utilized for this aim. For example, hydrox-
ycalix[n]arenes (n ¼ 4, 6, 8) were chemically bonded with silica gel surface modified by the 1,2-epoxypropanol (glycidol) moieties
giving phase (3.1) 68 (Fig. 14). The reaction takes place in toluene in the presence the strong base sodium hydride.
By a similar reaction the glycidylated silica gel was modified with the tert-butylcalix[6]arene-benzocrown-4-ether to form phase
(3.2) ( Fig. 15).
tert-Butyl-calix[4]arene bearing two moieties of acetyl chloride at the narrow rim was chemically bonded to the surface of silica
gel modified by g-(ethylenediamine)propyl groups to get the calixarene amide modified phase (3.3) ( Fig. 16).69
Stationary phase (4.4) was obtained in two stages by alkylation of aminopropylated silica gel with bromobutoxycalix[6]arene
hexasulfonic acid (sodium salt) and after that by silylation of the polar NH group formed with trimethylchlorosilane/hexamethyl-
disilazane mixture ( Fig. 17).70
Amide or hydroxamide derivatives of allylcalixarenes by the reaction with mercaptopropyltrimethoxysilane in the presence of
catalytic amounts of the cumene hydroperoxide were transformed into triethoxysilyl derivatives and condensed with activated silica
gel to obtain phase (3.5) ( Fig. 18).71–73
Similar allylcalix[n]arenes bearing ester or amide groups at the narrow rim were also linked to a silica gel surface in two steps. The
first step is the catalytic addition of triethoxysilane to the ethylene bond of the allylcalixarene in the presence of chloroplatinic acid.
The second stage is condensation of triethoxysilylcalixarene with activated silica gel surface to form phase (3.6) ( Fig. 19).74,75
Easy available lower rim allyloxycalix[n]arenes (n ¼ 4, 6, 8) and undecylenilcalix[4]resorcinarene were attached to Kromasil silica
gel surface by a similar two-stage method. 76,77 Now the Kromasil calixarene phases AI, AII, AIII, BI, BII, BIII and the resorcinarene
phase RES are produced by Synaptec GmbH under trademark Caltrex (Fig. 20).
The calixarene-based silica gel stationary phases are used for analysis of neutral organic molecules and ions by gas or liquid
chromatography.

2.12.4.2 Calixarene Phases in Gas Chromatography


Calixarene phases for gas chromatography were obtained by physical sorption of calixarenes (3.7) or their siloxane copolymers
( Fig. 21) to silica gel supports or to capillary columns.62

Figure 14 Synthesis of silica gel phases (3.1) modified with tert-butylcalix[n]arenes.


250 Chromatography in Supramolecular and Analytical Chemistry of Calixarenes

Figure 15 Synthesis of silica gel phases (3.2) modified with calix[6]arene-benzocrown-4-ether.

Figure 16 Synthesis of silica gel phase (3.3) modified with calix[4]arene amide.

Silanized Chromosorb W was modified by tert-butylcalix[4]arene (3.7a) and used for separation of alkanes, alkenes, haloge-
nated hydrocarbons, aromatic compounds, ethers, and alcohols. The inclusion calixarene complexes with these analytes were
formed in the separation process. 78
Stationary phases were prepared by the treatment of silanized Chromosorb W with a solution of tert-butylcalix[8]arene (3.7d) or
its octakis-methoxyethyl ether (3.7f) in tetrahydrofuran and used for separation of alcohols, chloroderivatives of hydrocarbons, and
aromatic compounds. 78 The best separation was obtained with a tert-butylcalix[8]arene (3.7d) phase, owing to the analyte inter-
actions with free OH groups of the calixarene.
tert-Butylcalix[n]arenes (3.7a–d) and the lower rim tetrasilylated tert-butylcalix[4]arene (3.7e) were precipitated on a silica
gel support from dichloromethane solution. The calixarene phases separate cyclic and acyclic alkanes and alkenes
Chromatography in Supramolecular and Analytical Chemistry of Calixarenes 251

Figure 17 Synthesis of silica gel phase (3.4) modified with calix[6]arene hexasulfonous acid.

Figure 18 Synthesis of silica gel phases (3.5) modified with calix[n]arene acetamides.

Figure 19 Synthesis of silica gel phases (3.6) modified with calix[n]arene acetic acid derivatives.
252 Chromatography in Supramolecular and Analytical Chemistry of Calixarenes

Figure 20 Chromatographic phases Caltrex.

Figure 21 Calix[n]arene modificators of gas chromatography columns (3.7).

(including isomeric forms), alkylbenzenes, polychloromethanes, alcohols, and ethers. 79 The highest selectivity was observed for
separation of alkylbenzenes and polychloromethanes. The total silanization of the calixarene hydroxyls decreases the selectivity
of separation.
The retention characteristics and separation selectivity of several aromatic analytes on the capillary column treated with solution
of alkylcalix[8]arene (3.7g) in polysiloxane SE-54 were investigated in Ref. 80. The modification of the upper rim of the calixarene
by adding a long alkyl chain improved its solubility in the polysiloxane. This calixarene containing phase strongly retained and
effectively separated naphthalene derivatives, quinoline, isoquinoline, and positional isomers of disubstituted benzene.
Efficiency, selectivity, polarity, and thermostability of the capillary columns modified with the lower rim substituted tert-
butylcalix[4]arenes (3.7h) and (3.7i) were investigated in Ref. 81. High selectivity of separation of the positional isomers of
aromatic compounds was shown.
Chromatography in Supramolecular and Analytical Chemistry of Calixarenes 253

Figure 22 tert-Butylcalix[4]arene-crown-5 polysiloxane (3.8) for modification of capillary columns.

Calixarene siloxane phases (3.7j,k) were used for separation of isomers of substituted benzenes and alicyclic compounds. 82
Thus, the tert-butyl groups of calix[4]arene (3.7j) have increased retention times, but did not increase the selectivity of analyte
separation.
Calix[8]arene (3.7g) dissolved in a polysiloxane phase was used for analysis of n-alkanes, regioisomers of substituted quinolines,
methyl- and chloronaphthalenes, and indole (number of theoretical plates was about 3000 m 1). 80
Regiomeric chlorophenols, dihydroxybenzenes, and xylenes were separated on the stationary polysiloxane phases treated with
0.5% solution of calix[4]arenes (3.7h,i) in dichloromethane. 81,83–87
Capillary column, modified by calix[4]arene-crown-5 polysiloxane (3.8) ( Fig. 22), showed high selectivity for the separation of
regioisomers of aromatic compounds.85 The number of theoretical plates was more than 3000 m 1 at operating temperature
310 C.
Capillary columns modified with a solution of pyridine-bridged calix[6]arenes (3.9a,b) in OV-1701 phase were used for sepa-
ration of isomers of monosubstituted phenols and other aromatic compounds under isothermal gas chromatography. The retention
time of the analytes was greater for asymmetric calix[6]arene (3.9b) as compared to its symmetric analog (3.9a) ( Fig. 23).88
Enantioseparation of amino acids was carried out on stationary polysiloxane phase (3.10) modified by the L-valine derivative of
calix[4]resorcinarene ( Fig. 24).89

Figure 23 The lower rim pyridine-bridged tert-butylcalix[6]arenes (3.9).

Figure 24 Chiral silica gel phase (3.10) modified with calix[4]resorcinarene L-valine derivative.
254 Chromatography in Supramolecular and Analytical Chemistry of Calixarenes

2.12.4.3 Stationary Calixarene Phases in Liquid Chromatography


Stationary phases with calixarenes, resorcinarenes, and calixpyrroles adsorbed or covalently bonded to the silica gel surface were
used in liquid chromatography to separate ions or neutral analytes. The separation of positional isomers of aminophenol, nitro-
aniline, and nitrophenol on a stationary phase modified by tert-butylcalix[6]arene (3.7s) was investigated in Ref. 86. While both
aminophenol and nitroaniline were successfully separated, attempts to separate nitrophenols failed.
A calixarene (3.7s) stationary phase was investigated for separation of polyaromatic compounds and nucleosides. 68 It was
shown that the calixarene phase is similar in properties to the C18 phase. The better selectivity for the calixarene (3.7c) phase
compared with the C18 phase was found for separation of sulfonamides and quinolines.70
Stationary phases based on tert-butylcalix[8]arene (3.7d) were investigated for separation of tricyclic neuroleptics 90,91 and water-
soluble vitamins.92 A phase modified with calixarenes (3.7d) was used for the separation of steroids.93
A calixarene (3.7a) phase was applied to the separation of mixtures of adenine, adenosine, cytosine, phenol, benzene, and
toluene. 94 The authors described a role of hydrophobic, p–p and dipole–dipole interactions between the analytes and calixarenes
in the separation. In the papers95,96 the tert-butylcalixarene (3.7b,d) phases were applied to the separation of regioisomers of nitro-
aniline and proline containing dipeptides. Chromatographic phases with chemically bonded calixarene acetic acid (3.11g–j)
(Fig. 25) were used for the separation of metal cations and organic compounds.
The phase based on tetraethyl ester of calix[4]arene tetraacetic acid (3.11a,c–f) demonstrated selectivity for the separation of
alkaline and alkaline earth metals. At the same time, the phase (3.11b) based on hexaethyl ester of calix[6]arene hexaacetic acid
was unsuitable for separation of these cations.
A calixarene amide (3.11c) stationary phase was used for separation of alkali, alkaline earth metals, and amino acids. 72,97,98
Precolumn concentration of trace amounts of lead in industrial wastewater was performed on a calixarene hydroxamate (3.11d)
phase.73 Characteristics and application of tert-butylcalix[4]arene phases are described in Ref. 99.
Phases based on calix[n]arene carboxylic acids (3.11g–j) were used for separation of isomeric methyluracils and estradiols. 95,96
These materials worked according to the reversed-phase mechanism, and their selectivity depends on the size of the calixarene
platform.
The Caltrex BII phase ( Fig. 20) showed higher selectivity for polycyclic aromatic hydrocarbons and fullerenes versus RP-18
phase.100–103
Calixarene-based supports (3.12a,b) ( Fig. 26) served for separation of regioisomers of disubstituted benzenes, polycyclic
aromatic hydrocarbons, purine and pyrimidine bases, and nucleosides (number of theoretical plates was 19,000 m 1).68
Stationary phases (3.13a,b) ( Fig. 27) containing calix[4]arene crown ethers in the 1.3-alternate conformation were used for
selective separation of alkali metal cations 3:13a : aKþ =Naþ ¼ 3:29; aKþ =Csþ ¼ 1:76; 3:13b : aCsþ =Naþ ¼ 2:66; aCsþ =Kþ ¼ 1:82.104
Calix[4]pyrrole (3.14, 3.15) ( Fig. 28) stationary phases were used for separation of fluorobenzenes, substituted amino acids,
nucleotides, oligonucleotides, and some anions.105 Polar groups in the analyte molecule have a decisive influence on the separation
selectivity. Selectivity separation of the fluorobenzenes is dependent on the number of fluorine atoms in the analyte molecule.

Figure 25 Silica gel phases (3.11) modified with acetic acid derivatives of calix[n]arenes.
Chromatography in Supramolecular and Analytical Chemistry of Calixarenes 255

Figure 26 Silica gel phases (3.12) modified with tert-butylcalix[4]arenes.

Figure 27 Silica gel phases (3.13) modified with calix[4]arene crown ethers.

Figure 28 Silica gel phases (3.14, 3.15) modified with calix[4]pyrroles.

Polycyclic compounds, isomeric nitroanilines, nucleosides, and nucleic bases were analyzed on phases Caltrex. 106 The elution
order of the analytes agreed with the order obtained on ODS support.107 The separation selectivity of sulfonamides in some cases
was higher compared with the C18 phase.
A detailed study of stationary phases Caltrex in the separation of cis- and trans-isomers of Doxepin and thioxanthenesdChlor-
prothixene, Clopentyksol, and Flupentyksol was described in paper. 108 The authors found that the selectivity of separation on these
phases is determined by the size and substitution manner of the calixarene skeleton. Advantages of the calixarene phases compared
with RP-C18 phase were demonstrated.
Chromatographic analysis of tricyclic neuroleptics (Promethazine, Promazine, Perazin, Levomepromazine, Chloropromazine,
Chlorprothixene, Clopentyksol, Fluphenazine, Flupentyksol) was performed on a Caltrex AIII phase. 109 The phase was more effec-
tive compared with Chromolith Performance RP-18 and LiChrospher phases.
256 Chromatography in Supramolecular and Analytical Chemistry of Calixarenes

A comparative study of Caltrex, Kromasil RP C18, Nucleosil C18, and Zorbax ODS phases was performed in the works. 109,110 To
rate the hydrophobic capacity and hydrophobic selectivity of the phases binary mixtures of aromatic hydrocarbons (toluene/ethyl-
benzene, anthracene/benzene, pentylbenzene/butylbenzene, anthracene/naphthalene, acenaphthene/naphthalene) were studied.
To determine the spatial selectivity, mixtures of triphenylene with o-terphenyl and biphenyl with o-terphenyl were examined. It
was found that Caltrex phases have lower hydrophobic capacity and higher spatial selectivity compared with C18 phases. The
authors separated the steroid mixture: Norethisterone, Norethisteronacetate, Chlomadinonacetate (Progestins), and testosterone
propionate (Androgen) on phases Caltrex. The elution order of steroids Norethisteron acetate and Chlomadinon acetate agreed
with their lipophilicity. The chromatographic analysis of Celecoxib on Caltrex AIII phase was investigated. The presence of Fenazon,
Metamizole, Phenazone, Methylprednisolone, Dexamethazone, Naproxen, Ibuprofen, Paracetamol, Caffeine, and degradation
products of Celecoxib in the sample does not interfere with the analysis. Analysis of Paracetamol, Caffeine, and acetyl salicylic
acid in a tablet on Caltrex phases is described in work 106.
A novel stationary phase was synthesized by grafting the calixarene imidazolium ionic liquid on mercaptopropyl-modified silica
gel. 111 To explore the retention supramolecular mechanism of the stationary phase, the linear solvation energy relationships equa-
tion was used as an effective mathematical model. In reversed-phase liquid chromatography, this phase presented specific chro-
matographic performance when evaluated using alkylbenzenes, polyaromatic hydrocarbons (PAHs), and phenols as solutes.
Due to the existing polar imidazolium functional groups, this stationary phase can also be used in hydrophilic interaction chroma-
tography. Six nucleosides and four ginsenosides were separated successfully in the hydrophilic mode. Furthermore, anions can be
separated on the column in anion exchange mode. Thus, this new material can be applied as a new kind of mixed-mode stationary
phase in liquid chromatography.
In paper 112 Intersil ODS-3 and Caltrex BIII were examined for separation of seven antihypertensive drugs. The results indicate
that the stationary phases used behave as reversed-phase packings and the hydrophobic interaction is the main factor having a role
in the retention of the antihypertensive drugs. The chemical structure and properties of the analytes were found to have an effective
role in their separation.
A new calix[4]arene derivatized chitosan stationary phase for HPLC was synthesized and characterized. 113 Its chromatographic
performance and retention mechanism were evaluated in reversed-phase mode compared with ODS using mono-substituted
benzenes, phenols, and nucleosides. The results showed that the calixarene stationary phase could provide various interactions
with analytes, such as hydrophobic, hydrophilic, p–p, and inclusion interactions.
The lower rim alanine-substituted tert-butylcalix[4]arene-bonded stationary phases were prepared and studied using PAHs,
monosubstituted benzenes, and disubstituted benzene isomers as analytes. 114 These new stationary phases exhibited a multiin-
teraction mode including a large array of interactions such as hydrophobicity, p–p, charge transfer, dipole–dipole, hydrogen
bonding, ion exchange, and inclusion complexation. They also showed stereoselectivities stronger than those of conventional
ODS phases.
A series of PAHs of different size and shape has been used to characterize the chromatographic behavior of the stationary phases
modified with five upper rim-substituted calix[4]arenes in the 1,3-alternate conformation. 115 The thermodynamic parameters
underlying the retention mechanisms revealed that the calix[4]arene columns exhibited variation in selectivity and retention of
PAHs caused by enthalpy and entropy effects. The observed divergences are due to differences in solute–stationary phase interac-
tions and originate in p–p and p–electron transfer specific to the analytes and the type of the calix[4]arene, as well as steric and
sorption phenomena.
Four proton diionisable 26-bis(carboxy-methoxy)calix[4]arene-crown-3, -crown-4, -crown-5, and -crown-6 in the cone confor-
mation were used to prepare bonded phases. 116 The clenbuterol levels in six samples of livestock meat (pork, pork casing, beef, beef
casing, mutton, and mutton casing) were analyzed on the phases with the limits of detection up to 0.06 mg mL 1.
The study of calixarene phases was also described in the papers 66,65,76,117–134.

2.12.4.4 Calixarene Additives to the Mobile Phase


Calixarenes are used as additives to the mobile phases in RP-HPLC analysis. Calixarene additives to the mobile phases decrease the
analyte retention 135 due to formation of Host–Guest supramolecular complexes and changing polarity of the chromatographic
support due to reversible sorption on its surface. Calix[4]arenes 3.16 (Fig. 29) and calix[4]resorcinarenes 3.17 (Fig. 30) were
used as additives to the mobile phases that improve the separation of uracil and benzene derivatives.136 Calixarenes (3.16) and
calixresorcinarenes (3.17) showed linear adsorption isotherms on support Separon SGX NH2 (Fig. 31) and Separon SGX C18,
respectively.
The linearity of the adsorption isotherms allowed the compounds (3.16a–e) and (3.17a) to be used as additives to improving
separation of uracil and benzene derivatives ( Tables 6 and 7).
Separation selectivity is dependent on substitution of the calixarene skeleton and the nature, number, and position of substit-
uents in the Guest molecules. The addition of phosphorylated calixresorcinarene (3.17a) to the mobile phase practically does not
affect separation of benzene/toluene (S ¼ 1.09); however it increases the separation of analytes bearing protonodonative groups:
phenol/p-aminophenol (S ¼ 1.48), p-cresol/p-aminophenol (S ¼ 1.64), o-phthalic/p-phthalic acid (S ¼ 1.43), and p-toluic/benzoic
acid (1.56) ( Table 7).
The selective separation of uracil derivatives is strongly dependent on the structure of the calixarene additive. For example, the
5-ethyluracyl/5-methyluracyl separation ratio S is 1.03 for dipropoxycalixarene (3.16e) and 3.15 for tetrapropoxycalixarene
Chromatography in Supramolecular and Analytical Chemistry of Calixarenes 257

Figure 29 Calix[4]arene additives (3.16) to mobile phases.

Figure 30 Calix[4]resorcinarene additives (3.17) to mobile phases.

Figure 31 Adsorption isotherms of calix[4]arenes (3.16a–e). Modified from Kalchenko, O. I.; Kalchenko, V. I. Cromatografia in Chemistry of Calixar-
enes; Naukova Dumka: Kyiv, 2013.

(3.16c). At the same time calixarene (4.16c) just slightly (S ¼ 1.06) increases the separation ratio 6-MeU/U. The introduction at
the calixarene upper rim of phosphoryl groups affects the supramolecular interactions with the Guest molecule. The separation
ratio S of 5-ethyluracil/5-methyluracil is 4 for diphosphonylcalixarene (3.16e) and 1.03 for unsubstituted calixarene (3.16a)
( Table 6). Water-soluble sulfonatocalix[6]arene was used as the additive to the mobile phase in separation of isomers of
substituted benzenes.
258 Chromatography in Supramolecular and Analytical Chemistry of Calixarenes

Table 6 Separation selectivity of the uracil derivatives induced by calix[4]arene additives (3.16a–e) to H2O/MeOH/MeCN/THF,
70:15:10:5, v/v mobile phase (Separon SGX NH2), 254 nm, 31 C

Selectivity without Selectivity in the presence Selectivities


Calixarene Analytes separated calixarene additive, a0 of calixarene additive, a1 ratio, a1/a0

3.16a 5-EtU/5-MeU 1.03 1.06 1.03


1,3-DiMeU/5-AmU 2.80 5.40 1.93
6-Am-1-MeU/5-AmU 2.24 3.46 1.55
6-Am-1,3-DiMeU/U 1.12 1.43 1.28
6-MeU/U 1.23 1.43 1.16
3.16b 5-EtU/5-MeU 1.24 1.44 1.16
1,3- DiMeU/5-AmU 1.76 2.78 1.58
6-Am-1-MeU/5-AmU 1.84 2.78 1.50
6-Am-1,3- DiMeU/U 1.23 1.40 1.14
6-MeU/U 1.23 1.73 1.41
3.16c 5-EtU/5-MeU 1.08 1.52 1.41
1,3-DiMeU/5-AmU 1.81 2.74 1.52
6-Am-1-MeU/5-AmU 2.66 6.21 2.33
6-Am-1,3-DiMeU/U 1.10 1.38 1.25
6-MeU/U 1.18 1.60 1.6
3.16d 5-EtU/5-MeU 1.00 1.44 1.44
1,3-DiMeU/5-AmU 1.98 2.65 1,34
6-Am-1-MeU/5-AmU 2.47 2.98 1.21
6-Am-1,3-DiMeU/U 1.17 2.25 1.92
6-MeU/U 1.00 1.58 1.58
3.16e 5-EtU/5-MeU 1.27 4.00 3.15
1,3-DiMeU/5-AmU 2.20 3.09 1.40
6-Am-1-MeU/5-AmU 1.88 3.64 1.94
6-Am-1,3-DiMeU/U 1.12 1.71 1.53
6-MeU/U 1.23 1.30 1.06

Modified from Kalchenko, O. I.; Cherenok, S. A.; Solovyov, A.; Kalchenko, V. I. Chromatographia 2009, 70, 717–721.

Table 7 Separation selectivity of benzene derivatives induced by calix[4]resorcinarene additive (3.17a) to MeCN/H2O,
86:14, v/v mobile phase (Separon SGX C18), 254 nm, 22 C

Selectivity before Selectivity after Selectivities ratio,


Analytes separated calixarene addition, a0 calixarene addition, a1 S ¼ a1/a0

p-Toluic/m-toluic acids 1.02 1.23 1.21


Benzene/p-cresol 1.08 1.25 1.16
o-Nitrobenzoic/p-nitrobenzoic acids 1.1 1.42 1.17
Benzene/toluene 1.31 1.43 1.09
o-Phthalic/p-phthalic acids 1.31 1.87 1.43
o-Bromotoluene/p-bromotoluene 1.32 1.75 1.33
p-Toluic/benzoic acids 1.36 2.12 1.56
Phenol/p-aminophenol 1.62 2.40 1.48
m-Aminobenzoic/p-aminobenzoic acids 1.67 1.81 1.08
p-Cresol/p-aminophenol 1.67 2.80 1.68

Modified from Kalchenko, O. I.; Cherenok, S. A.; Solovyov, A.; Kalchenko, V. I. Chromatographia 2009, 70, 717–721.

2.12.5 Conclusion

HPLC has been widely used in the study of calixarenes. Along with the classic scopedseparation and identificationdin recent years
HPLC has become an effective instrumental method in the supramolecular chemistry of calixarenes. HPLC is a highly sensitive,
expressive, and economical method for determination of binding constants of supramolecular complexes of calixarenes with
organic molecules or biomolecules in water and water-organic solutions. In this sense, HPLC is an alternative method to the classic
methodsdNMR, UV–vis, fluorescent spectroscopy, and microcalorimetry. HPLC was used for the determination of stability
constants for calixarene complexes with a variety of environmental pollutants (benzene derivatives, condensed aromatic
compounds, herbicides) and biologically important compounds (nucleic bases, amino acids, peptides, nucleotides).
The ability to recognize and form the supramolecular complexes with organic or inorganic analytes is the basis of application of
calixarenes functionalized with different receptor groups in the design of mobile and stationary chromatography phases.
Chromatography in Supramolecular and Analytical Chemistry of Calixarenes 259

Chromatographic properties of calixarene stationary phases are significantly different from those of conventional phases due to the
formation of Host–Guest inclusion complexes with analyte molecules. Thus, chromatographic properties of the phases are consid-
ered as a function of the size of the calixarene cavity, its architecture, and the nature of the substituents at the upper or lower rim of
the macrocycle.
Calixarene containing stationary phases are used for separation of different classes of organic compounds, environmentally
hazardous aromatic compounds, carboxylic acids, alkanes, alkenes, halogenated hydrocarbons, ethers, alcohols, adenine and uracil
derivatives, amino acids and peptides, vitamins, and pharmaceuticals. Calixarenes functionalized by appropriate binding groups
have been used for chromatographic separation of metal cations. Calixarene additives to mobile phases improve the selectivity
of analyte separation.
The results presented in this article indicate that HPLC is a valuable method in analytical and supramolecular chemistry of
calixarenes.

Acknowledgment

This work was supported by grants from State Foundation of Basic Research of Ukraine.

References

1. Gutsche, C. D. Calixarenes Revisited; The Royal Society of Chemistry: Cambridge, 1998, 233 p.
2. Baudry, R.; Kalchenko, O.; Dumazet-Bonnamour, I.; Vocanson, F.; Lamartine, R. J. Chromatogr. Sci. 2003, 41, 157–163.
3. Kalchenko, O. I.; Cherenok, S. O.; Rozhenko, O. B.; Yushchenko, O. A.; Kalchenko, V. I. J. Org. Pharm. Chem. 2012, 4 (40), 54–62.
4. Solovyov, A. V.; Cherenok, S. O.; Kalchenko, O. I.; Atamas, L. I.; Kazantseva, Z. I.; Koshets, I. A.; Tsymbal, I. F.; Kalchenko, V. I. J. Mol. Liq. 2011, 159, 117–123.
5. Kalchenko, O. I.; Solovyov, A. V.; Cherenok, S. A.; Starodub, N. F.; Kalchenko, V. I. J. Incl. Phenom. 2003, 46, 19–25.
6. Kalchenko, O. I.; Lipkowski, J.; Nowakowski, R.; Kalchenko, V. I.; Visotsky, M. A.; Markovsky, L. N. J. Chromatogr. Sci. 1997, 35, 49–52.
7. Cherenok, S. O.; Miroshnichenko, S. I.; Drapailo, A. B.; Kalchenko, O. I.; Rodik, R. V.; Boyko, V. I.; Matvieiev, Y. I.; Ruban, A. V.; Kalchenko, V. I. Phosphorus Sulfur Silicon
Relat. Elem. 2011, 186 (4), 878–883.
8. Drapailo, A. B.; Kharchenko, S. G.; Shishkina, S. V.; Shishkin, O. V.; Kalchenko, V. I. Phosphorus Sulfur Silicon Relat. Elem. 2011, 186 (4), 896–897.
9. Cherenok, S.; Kalchenko, V. I. Phosphorus-Containing Calixarenes. Topics in Heterocyclic Chemistry; Springer Publishing Company: Berlin 2009, Vol. 2, 229–273.
10. Kalchenko, V. Pure Appl. Chem. 2008, 80, 1449–1458.
11. Cherenok, S.; Dutasta, J.-P.; Kalchenko, V. Curr. Org. Chem. 2006, 10, 2307–2331.
12. Tairov, M.; Lalchenko, V. J. Org. Pharm. Chem. 2004, 2, 3–19.
13. Lalchenko, V. I.; Rodik, R. V.; Boklo, V. I. J. Org. Pharm. Chem. 2005, 3, 13–29.
14. Pappalardo, S.; Caccamese, S.; Giunta, L. Tetrahedron Lett. 1991, 32, 7747–7750.
15. Iwamoto, K.; Araki, K.; Shinkai, S. Tetrahedron 1991, 47, 4325–4342.
16. Gloede, J.; Keitel, I.; Costisella, B.; Kunath, A.; Schneider, M. Phosphorus Sulfur Silicon Relat. Elem. 1996, 117, 67–88.
17. Tsubaki, K.; Otsubo, T.; Kinoshita, T.; Kawada, M.; Fudji, K. Chem. Pharm. Bull. 2001, 49, 507–509.
18. Kalchenko, O. I.; Tairov, M. O.; Vysotsky, M. O.; Lipkowski, J.; Kalchenko, V. I. Enantiomer 2000, 5, 385–390.
19. Tairov, M. A.; Vysotsky, M. O.; Kalchenko, O. I.; Pirozhenko, V. V.; Kalchenko, V. I. J. Chem. Soc. Perkin Trans. 2002, 1, 1405–1411.
20. Boyko, V. I.; Yakovenko, A. V.; Matvieiev, Y. I.; Kalchenko, O. I.; Shishkin, O. V.; Shishkina, S. V.; Kalchenko, V. I. Tetrahedron 2008, 64, 7567–7573.
21. Kalchenko, O. I.; Da Silva, E.; Coleman, A. W. J. Incl. Phenom. 2002, 43, 305–310.
22. Lipkowski, J.; Kalchenko, O. I.; Slowikowska, J.; Kalchenko, V. I.; Lukin, O. V.; Markovsky, L. N.; Nowakowski, R. J. Phys. Org. Chem. 1998, 11, 426–435.
23. Kalchenko, O. I.; Solovyov, A. V.; Lipkowski, J.; Kalchenko, V. I. J. Chem. Res. 1999, 60–66.
24. Kalchenko, O. I.; Lipkowski, J.; Nowakowski, R.; Kalchenko, V. I.; Vysotsky, M. A.; Markovsky, L. N. J. Incl. Phenom. 1998, 23, 377–380.
25. Kalchenko, O. I.; Solovyov, A. V.; Lipkowski, J.; Kalchenko, V. I. J. Incl. Phenom. 1999, 34, 259–266.
26. Fujimura, K.; Ueda, T.; Kitagawa, M.; Takayanagi, H.; Ando, T. Anal. Chem. 1986, 58, 2668–2674.
27. Armstrong, D. W.; Nome, F.; Spino, L.; Golden, T. J. Am. Chem. Soc. 1986, 1084, 1418–1421.
28. Molenveld, P.; Engbersen, J. F. J.; Reinhoudt, D. N. Angew. Chem. Int. Ed. Engl. 1999, 38, 3189–3191.
29. Selkti, M.; Coleman, A. W.; Nicolis, I.; Douteau-Guével, N.; Villain, F.; Tomas, A.; de Rango, C. Chem. Commun. 2000, 161–162.
30. Antipin, I. S.; Stoikov, I. I.; Pinkhassik, E. M.; Fitseva, N. A. Tetrahedron Lett. 1997, 38, 5865–5868.
31. Zielenkiewicz, W.; Pietraszkiewicz, O.; Wszeliaka-Rylik, M.; Pietraszkiewicz, M.; Roux-Desgranges, G.; Roux, A. H.; Grolier, J.-P. E. J. Solut. Chem. 1998, 27, 121–134.
32. Hamuro, Y.; Calama, M. C.; Park, H. S.; Hamilton, A. D. Angew. Chem. Int. Ed. Engl. 1997, 36, 2680–2683.
33. Lehn, J.-M.; Meric, R.; Vigneron, J.-P.; Cerario, M.; Guilhem, J.; Pascard, C.; Asfari, Z.; Vicens, J. Supramol. Chem. 1995, 5, 97–103.
34. Kurihara, K.; Ohto, K.; Tanaka, T.; Aoyama, Y.; Kunitake, T. J. Am. Chem. Soc. 1991, 113, 444–450.
35. Shi, Y.; Schneider, H.-J. J. Chem. Soc. Perkin Trans. 1999, 2, 1797–1803.
36. Kalchenko, O.; Poznanski, J.; Marcinowicz, A.; Cherenok, S.; Solovyov, A.; Zielenkiewicz, W.; Kalchenko, V. J. Phys. Org. Chem. 2003, 16, 246–252.
37. Kalchenko, O.; Marcinowicz, A.; Poznanski, J.; Cherenok, S.; Solovyov, A.; Zielenkiewicz, W.; Kalchenko, V. J. Phys. Org. Chem. 2005, 18, 578–585.
38. Veklich, T. O.; Shkrabak, O. A.; Rodik, R. V.; Boiko, V. I.; Kalchenko, V. I.; Losterin, S. O. Ukr. Biokhim. Zh. 2010, 82, 6–17.
39. Sansone, F.; Barboso, S.; Casnati, A.; Sciotto, D.; Ungaro, R. Tetrahedron Lett. 1999, 40, 4741–4744.
40. Asharya, A.; Ramanujam, B.; Chinta, J. P.; Rao, C. P. J. Org. Chem. 2011, 76 (1), 127–137.
41. Li, W. Y.; Li, H.; Zhang, G. M.; Chao, J. B.; Li, X. L.; Shao, M. S.; Dong, C. J. Photochem. Photobiol. A Chem. 2008, 197 (2–3), 389–393.
42. Okada, Y.; Kasai, Y.; Nishimura, J. Tetrahedron Lett. 1995, 36, 555–558.
43. Mutihac, L.; Buschmann, H.-J.; Diacu, E. Desalination 2008, 148, 253–256.
44. Kim, L.; Hamdi, A.; Stancu, A. D.; Othman, A. B.; Vicens, J. J. Incl. Phenom. Macrocycl. Chem. 2010, 66, 55–59.
45. Durmaz, M.; Alpaydin, S.; Sirit, A.; Yilmaz, M. Tetrahedron Asymmetry 2007, 18, 900–905.
46. Castellano, R. K.; Nuckolls, C.; Rebek, J., Jr. J. Am. Chem. Soc. 1999, 121, 11156–11163.
47. Stone, M. M.; Franz, A. H.; Lebrilla, C. B. J. Am. Soc. Mass Spectrom. 2002, 13 (8), 964–974.
48. Chen, H.; Gu, L.; Yin, Y.; Koh, K.; Lee, J. Int. J. Mol. Sci. 2011, 12 (4), 2315–2324.
260 Chromatography in Supramolecular and Analytical Chemistry of Calixarenes

49. Kalchenko, O. I.; Perret, F.; Coleman, A. W. J. Chem. Soc. Perkin Trans. 2001, 2, 258–263.
50. Kalchenko, O.; Drapailo, A.; Shishkina, S.; Shishkin, O.; Kharchenko, S.; Gorbatchuk, V.; Kalchenko, V. Supramol. Chem. 2013, 25 (5), 263–268.
51. Kalchenko, O. I.; Cherenok, S. O.; Rodik, R. V.; Drapailo, A. B.; Miroshnichenko, S. I.; Kalchenko, V. I. Phosphorus Sulfur Silicon Relat. Elem. 2011, 186 (4), 898–902.
52. Kalchenko, O.; Cherenok, S. O.; Yuschenko, O. A.; Kalchenko, V. I. J. Incl. Phenom. 2013, 76 (1–2), 29–36.
53. Steed, J. W.; Atwood, J. L. Supramolecular Chemistry; John Wiley: Chichester/New York, NY/Weinheim/Brisbane/Singapore/Toronto, 2000, 990 p.
54. ;
Kalchenko, O. I.; Cherenok, S. O.; Rodik, R. V.; Drapailo, A. B.; Miroshnichenko, S. I.; Kalchenko, V. I. In 2010; p 89. Book of Abstracts.
55. Cherenok, S. O.; Yuschenko, O. A.; Gritsenko, P. G.; Lugovskoy, E. V.; Koshel, T. A.; Chernishov, V. I.; Koliesnik, I. O.; Kalchenko, O. I.; Komisarenko, S. V.; Kalchenko, V. I.
Phosphorus Sulfur Silicon Relat. Elem. 2011, 186, 964–965.
56. Lugovskoy, E. V.; Gritsenko, P. G.; Koshel, T. A.; Kolesnik, I. O.; Cherenok, S. O.; Kalchenko, O. I.; Kalchenko, V. I.; Komisarenko, S. V. FEBS J. 2011, 278, 1244–1251.
57. da Silva, E.; Ficheux, D.; Coleman, A. W. J. Inclusion Phenom. Macrocycl. Chem. 2005, 5, 201–206.
58. Cherenok, S. O.; Yushchenko, O. A.; Tanchuk, V. Y.; Mischenko, I. M.; Samus, N. V.; Ruban, O. V.; Matvieiev, Y. I.; Karpenko, J. A.; Kukhar, V. P.; Vovk, A. I.; Kalchenko, V. I.
ARKIVOC 2012, iv, 278–298.
59. Veklich, T. O.; Kosterin, S. O.; Rodik, R. V.; Cherenok, S. O.; Boĭko, V. I.; Kalchenko, V. I. Ukr. Biokhim. Zh. 2006, 78, 70–86.
60. Komisarenko, S. V.; Kosterin, S. O.; Lugovskoy, E. V.; Cherenok, S. O.; Yu Tanchuk, V.; Vovk, A. I.; Kalchenko, V. I. In Ligands: Synthesis, Characterization and Role in
Biotechnology; Gawriszewska, P., Stalmaszzczyk, P., Eds.; Nova Science Publishers, Inc: New York, NY, 2014; pp 67–116.
61. Shkrabak, O. A.; Kalchenko, O. I.; Rodik, R. V.; Veklich, T. O.; Kalchenko, V. I.; Kosterin, S. O. Ukr. Biochem. J. 2008, 80 (2), 90–100.
62. Mildbradt, R.; Bohmer, V. In Calixarenes 2001; Asfari, M.-Z., Böhmer, V., Harrowfield, J., Vicens, J., Eds.; Kluwer Academic Publishers: Dordrecht, 2001; pp 663–676.
63. Mokhtari, B.; Pourabdollah, K.; Dalali, N. Chromatographia 2011, 73, 829–847.
64. Mokhtari, B.; Pourabdollah, K.; Dalali, N. J. Incl. Phenom. 2011, 69, 1–55.
65. Sliwka-Kaszynska, M. Crit. Rev. Anal. Chem. 2007, 37, 211–224.
66. Meyer, R.; Jira, T. Curr. Anal. Chem. 2007, 3, 161–170.
67. Delahousse, G.; Lavendomme, R.; Jabin, I.; Agassea, V.; Cardinael, P. Curr. Org. Chem. 2015, 19, 2237–2249.
68. Hu, W.; Li, J. S.; Feng, Y. Q.; Da, S. L.; Chen, Y.-Y.; Xiao, X.-Z. Chromatographia 1998, 48, 245–250.
69. Xiao, X. Z.; Feng, Y. Q.; Da, S. L.; Zhang, Y. Chromatographia 1999, 49, 643–648.
70. Lee, Y. K.; Ryu, Y. K.; Ryu, J. W.; Kim, B. E.; Park, J. H. Chromatographia 1997, 46, 507–510.
71. Glennon, J. D.; Horne, E.; O’Connor, K.; Kearney, G.; Harris, S. J.; McKervey, M. A. Anal. Proc. 1994, 31, 33–35.
72. Glennon, J. D.; Horne, E.; Hall, K.; Cocker, D.; Kuhn, A.; Harris, S. J.; McKervey, M. A. J. Chromatogr. A 1996, 731, 47–55.
73. O’Connell, M. P.; Treacy, J.; Merly, C.; Smith, C. M. M.; Glennon, J. D. Anal. Lett. 1999, 32, 185–192.
74. Glennon, J. D.; O’Connor, K.; Srijaranai, S.; Manley, K.; Harris, S. J.; McKervey, M. A. Anal. Lett. 1993, 26, 153–159.
75. Joyce, T.; Grady, T.; Harris, S. J.; Diamond, D. Anal. Commun. 1998, 35, 123–125.
76. Schneider, C.; Menyes, U.; Jira, T. J. Sep. Sci. 2010, 33, 2930–2942.
77. Menyes, U.; Roth, U.; Troltzsch, C. WO/1997/027479: PCT/DE1997/000180, 1997.
78. Mnuk, P.; Feltl, L. J. Chromatogr. A 1995, 696, 101–112.
79. Mnuk, P.; Feltl, L.; Schurig, V. J. Chromatogr. A 1996, 732, 63–74.
80. Gross, B.; Jauch, J.; Schurig, V. J. Microcolumn Sep. 1999, 11, 313–317.
81. Lin, L.; Wu, C. Y.; Yan, Z. Q.; Yan, X. Q.; Su, X. L.; Han, H. M. Chromatographia 1998, 47, 689–694.
82. Lim, H. J.; Lee, H. S.; Kim, I. W.; Chang, S. H.; Moon, S. C.; Kim, B. E.; Park, J. H. Chromatographia 1998, 48, 422–426.
83. Lai, X. H.; Lin, L.; Wu, C. Y. Chromatographia 1999, 1650, 82–88.
84. Ye, H.; Lin, L.; Wu, C. Fenxi Huaxue 1999, 27, 1087–1090.
85. Xing, J.; Wu, C.-Y.; Li, T.; Zhong, Z.-L.; Chen, Y.-Y. Anal. Sci. 1999, 1715, 785–789.
86. Zhang, L. F.; Chen, L.; Lu, X. R.; Wu, C. Y.; Chen, Y. P. J. Chromatogr. A 1999, 840, 225–233.
87. Yu, X. D.; Lin, L.; Wu, C. Y. Chromatographia 1999, 49, 567–571.
88. Park, J. H.; Lim, H. J.; Lee, Y. K.; Park, J. K.; Kim, B. E.; Ryoo, J. J.; Lee, K.-P. J. High Resolut. Chromatogr. 1999, 222, 679–682.
89. Pheiffer, J.; Schurig, V. J. Chromatogr. A 1999, 840, 145–150.
90. Hashem, H.; Jira, T. Die Pharmazie 2005, 60, 186–192.
91. Hashem, H.; Jira, T. Comb. Chem. High Throughput Screen. 2007, 10, 387–396.
92. Li, L. S.; Da, S. L.; Feng, Y.-Q.; Liu, M. Talanta 2004, 64, 373–379.
93. Liu, M.; Li, L.-S.; Da, S.-L.; Feng, Y.-Q. Talanta 2005, 66, 479–486.
94. Gezici, O.; Tabakci, M.; Kara, H.; Yilmaz, M. J. Macromol. Sci., Part A: Pure Appl. Chem. 2006, 43, 221–231.
95. Gebauer, S.; Friebe, S.; Gubitz, G.; et al. J. Chromatogr. Sci. 1998, 36, 383–387.
96. Gebauer, S.; Friebe, S.; Scherer, C.; Gubitz, G.; Krauss, G.-J. J. Chromatogr. Sci. 1998, 36, 388–394.
97. Brindle, R.; Albert, K.; Harris, S. J.; Tröltzsch, C.; Horne, E.; Glennon, J. D. J. Chromatogr. A 1996, 731, 41–46.
98. Hutchinson, S.; Kearney, G. A.; Horne, E.; Lynch, B.; Glennon, J. D.; McKervey, M. A. Anal. Chim. Acta 1994, 291, 269–275.
99. Friebe, S.; Gebauer, S.; Krauss, G. J.; Gormar, G.; Kruger, J. J. Chromatogr. Sci. 1995, 33, 281–284.
100. Menyes, U.; Roth, U.; Jira, T. Separation of Polycyclic Aromatic Hydrocarbons by Using Calixarene-Modified Chromatographic Stationary Phases. Eur. Pat. Appl. EP
952134, 1999.
101. Menyes, U.; Roth, U. Process for the Separation of Fullerenes Using Chromatography. Eur. Pat. Appl. EP 952110, 1999.
102. Iwata, K.; Morigushi, S. (Showa Denko Kk, Japan) Jpn. Kokai Tokyo Koho 05264531 A 2, 1993.
103. Iwata, K.; Kimizuka, H.; Suzuki, H. (Showa Denko Kk, Japan) Jpn. Kokai Tokyo Koho 06058920 A 2, 1994.
104. Arena, G.; Casnati, A.; Contino, A.; Mirone, L.; Sciotto, D.; Ungaro, R. Chem. Commun. 1996, 19, 2277–2278.
105. Sessler, J. L.; Gale, P. A.; Genge, J. W. Chem. Eur. J. 1998, 4, 1095–1099.
106. Li, L. S.; Da, S. L.; Feng, Y.-Q.; Liu, M. J. Liq. Chromatogr. Relat. Technol. 2004, 27, 2167–2188.
107. Hashem, H. Chromatographia 2010, 71, 31–35.
108. Sokolieb, T.; Schonherr, J.; Menyes, U.; Roth, U.; Jira, T. J. Chromatogr. A 2003, 1021, 71–82.
109. Sokolieb, T.; Menyes, U.; Roth, U.; Jira, T. J. Chromatogr. A 2002, 948, 309–319.
110. Hashem, H.; Trundelberg, C.; Jira, T. Chromatographia 2010, 71, 91–94.
111. Hu, K.; Zhang, W.; Yang, H.; Cui, Y.; Zhang, J.; Zhao, W.; Yu, A.; Zhang, S. Talanta 2016, 152, 392–394.
112. Elhenawee, M.; Hashem, H.; Ibrahim, A. E. J. Liq. Chromatogr. 2014, 37 (1), 1–25.
113. Lu, J.; Zhang, W.; Zhang, Y.; Zhao, W.; Hu, K.; Yu, A.; Liu, P.; Wu, Y.; Zhang, S. J. Chromatogr. A 2014, 1350, 61–67.
114. Deng, Z.; Liu, J.; Hu, C.; Yang, L.; Du, H.; Hu, K.; Huang, Y.; Yang, X.; Jiang, Q.; Zhang, S. J. Sep. Sci. 2014, 37, 3268–3327.
115. Sliwka-Kaszynska, M.; Slebioda, M. J. Sep. Sci. 2014, 37, 543–550.
116. Mokhtari, B.; Pourabdollah, K. J. Sci. Food Agric. 2012, 92, 2679–2688.
117. Li, L.-S.; Liu, M.; Da, S.-L.; Feng, Y. Q. Chin. J. Anal. Chem. 2004, 32 (4), 511–515.
Chromatography in Supramolecular and Analytical Chemistry of Calixarenes 261

118. Liu, M.; Da, S.-L.; Feng, Y. Q.; Li, L.-S. Chem. J. Chin. Univ. 2004, 25, 1254–1256.
119. Li, L. S.; Da, S. L.; Feng, Y. Q.; Liu, M. Anal. Sci. 2004, 20, 561–564.
120. Xiao, Y.-X.; Xiao, X.-Z.; Feng, Y.-Q.; Wang, Z.-H.; Da, S.-L. J. Liq. Chromatogr. Relat. Technol. 2001, 24, 2925–2942.
121. Sokolieb, T.; Menyes, U.; Roth, U.; Jira, T. J. Chromatogr. A 2000, 898, 35–52.
122. Ding, C.; Qu, K.; Li, Y.; Hu, K.; Liu, H.; Ye, B.; Wu, Y.; dan Zhang, S. J. Chromatogr. A 2007, 1170, 73–81.
123. Bazylak, G.; Malak, A.; Ali, I.; Borowiak, T.; Dutkiewicz, G. Curr. Drug Discov. Technol. 2008, 5, 177–189.
124. Krauss, G. J.; Friebe, S.; Gebauer, S. J. Protein Chem. 1998, 17, 515–516.
125. Sessler, J. L.; Kim, S. K.; Gross, D. E.; Lee, C.-H.; Kim, J. S.; Lynch, V. M. J. Am. Chem. Soc. 2008, 30, 13162–13266.
126. Hu, K.; Zhao, W.; Wen, F.; Liu, J.; Zhao, X.; Xu, Z.; Niu, B.; Ye, B.; Wu, Y.; Zhang, S. Talanta 2011, 85 (1), 317–324.
127. Hu, K.; Yu, A.; Zhang, J.; Wen, F.; Liang, S.; Zhao, X.; Ye, B.; Wu, Y.; Zhang, S. J. Chromatogr. Sci. 2012, 50 (2), 123–130.
128. Delahousse, G.; Peulon-Agasse, V.; Debray, J. C.; Vaccaro, M.; Cravotto, G.; Jabin, I.; Cardinael, P. J. Chromatogr. A 2013, 1318, 207–216.
129. Chamseddin, C.; Jira, T. Chromatographia 2013, 76, 449–457.
130. Chamseddin, C.; Jira, T. Chromatographia 2014, 77, 1167–1183.
131. Hashem, H.; Ibrahim, A. E.; Elhenawee, M. J. Sep. Sci. 2014, 37, 2814–2824.
132. Erdemir, S.; Yilmaz, M. J. Sep. Sci. 2011, 34, 393–401.
133. Erdemir, S.; Yilmaz, M. J. Macromol. Sci., Part A: Pure Appl. Chem. 2012, 49, 1022–1029.
134. Durmaz, F.; Memon, F. N.; Memon, N. A.; Memon, S.; Memon, S.; Kara, H. Chromatographia 2013, 76, 909–919.
135. Kalchenko, O. I.; Lipkowski, J.; Kalchenko, V. I.; Vysotsky, M. A.; Markovsky, L. N. J. Chromatogr. Sci. 1998, 36, 269–273.
136. Kalchenko, O. I.; Cherenok, S. A.; Solovyov, A.; Kalchenko, V. I. Chromatographia 2009, 70, 717–721.
2.13 Neutron Scattering
V Cristiglio, GJ Cuello, and M Jiménez-Ruiz, Institute Laue-Langevin, Grenoble, France
Ó 2017 Elsevier Ltd. All rights reserved.

2.13.1 Introduction 263


2.13.2 Properties of Neutrons 264
2.13.2.1 Neutron Sources 264
2.13.3 Basis of Neutron Scattering 265
2.13.3.1 Scattering Cross Section 265
2.13.3.2 Scattering by a Single Nucleus 265
2.13.3.3 The Master Formula of the Scattering 266
2.13.3.4 Coherent and Incoherent Scattering 267
2.13.4 Neutron Diffraction 268
2.13.4.1 Nuclear Scattering by Crystals 269
2.13.4.1.1 Coherent Elastic Scattering 269
2.13.4.1.2 The Laue Method 270
2.13.4.1.3 Rotating Crystal 270
2.13.4.1.4 Powder Diffraction 270
2.13.4.1.5 Instruments 271
2.13.4.2 Total Scattering 272
2.13.4.2.1 Multiatomic Systems 273
2.13.4.2.2 Coordination Numbers 274
2.13.4.2.3 The First Difference Method 274
2.13.4.2.4 Instruments 276
2.13.5 Small-Angle Neutron Scattering (SANS) 276
2.13.6 Principle of SANS Technique 277
2.13.6.1 Scattering Length Density 277
2.13.6.2 Isolated Particles 278
2.13.6.3 Information From a Scattering Pattern 278
2.13.6.3.1 Guinier Regime 279
2.13.6.3.2 Intermediate Regime 279
2.13.6.3.3 Porod Regime 279
2.13.6.4 Interparticle Interference 279
2.13.6.5 Contrast Variation 280
2.13.6.6 SANS Instruments 280
2.13.6.6.1 Classical SANS SpectrometersdD11, D22, and D33 280
2.13.6.6.2 D16 Two-Axis Cold Neutron Diffractometer 281
2.13.7 Neutron Vibrational Spectroscopy 282
2.13.7.1 IN1-Lagrange Design 284
2.13.7.2 INS Example: Vibrational Study of an Oriented Clay 286
2.13.8 Conclusions 287
Acknowledgments 287
References 287

2.13.1 Introduction

Neutron scattering is one of the most powerful and versatile experimental techniques used to investigate the structure and dynamics
of condensed matter at the atomic scale (up to hundreds of nanometers) complementary to other structural probes, such as those
using photons (from visible light to synchrotron X-rays) or electrons (microscopy and diffraction), as well as to standard laboratory
measurements.
Fields of research covered by neutron scattering extend to a wide range of sciences, such as condensed matter physics, materials
science, chemistry, polymer science, biology, engineering, geology, etc. It is considered that neutron scattering is indispensable in
completing the fundamental understanding of the modern theory of matter and will play an important role in providing a founda-
tion for future technologies. The importance of neutrons is given by their unique properties; penetrating and uncharged, thus
penetrating matter deeply, they have a magnetic moment, wavelengths and energies similar to interatomic distances, and lattice
vibrations in condensed matter, respectively.

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12490-4 263


264 Neutron Scattering

The utility of neutron beams arises from the physical properties of the neutron itself, ranging from the ability to sensitively
observe atomic or molecular magnetism, to the ability to observe the details of atomic and molecular motions in both space
and time, to the ability to use atomic isotopic substitution to label particular regions of complex structures.
Together with synchrotrons, neutron facilities are considered the most useful microscopic probes of matter available to all the
international user communities. 1,2 The neutrons are produced in nuclear reactors or with spallation sources. Relevant instruments
concerning nanoporous and mesoporous materials investigation are powder neutron diffractometers (structure, phase analysis,
in situ studies at the atomic scale), small-angle scattering (SAS) machines (size and shape of large-scale structures) and inelastic
neutron spectrometers (intermolecular and intramolecular vibrations of the system).
This review is aimed to present a general overview of neutron scattering theory and techniques related to the multicomponent
molecular assembly studies. These techniques will be detailed presenting specific instruments currently available to the user
community at the Institut Laue-Langevin (ILL) in Grenoble, France, the most intense neutron research reactor in the world.
In the first section, we neutron properties and its production, with an overview on the neutron sources. Then, we continue with
the presentation of the neutron scattering theory. In “ Neutron Diffraction,” “Small-Angle Neutron Scattering (SANS),” and
“Neutron Vibrational Spectroscopy” sections we detail three neutron techniques: neutron diffraction, small-angle neutron
scattering (SANS), and neutron vibrational spectroscopy. We will make a detailed presentation of the theory related to these
techniques and then present significant examples of studies done in ILL instruments.

2.13.2 Properties of Neutrons

In this section, a general introduction of the relevant behavior of neutrons, neutron production, and the neutron sources currently
operational in the world will be presented.
For a general and complete theoretical description of the nuclear neutron scattering and the associated neutron techniques, we
refer the reader to these two particular references: a volume edited by Sköld and Price 3 and a review published recently by Schober.4
A neutron is an electrically neutral elementary subatomic particle with a mass close to that of the proton (mn ¼ 1.675  10 27 kg).
The neutron lifetime as a free particle is about 15 min, in spite of the fact that neutrons are stable when bound in an atomic nucleus.
Since neutrons are neutral particles, they interact with nuclei via the strong nuclear force, as opposed to X-rays, which are
scattered by the electron cloud. Neutron radiation penetrates deep into material, thus making it possible to study microscopic
properties of bulk samples. Such investigations cannot be performed by means of optical methods, X-ray scattering, or electron
microscopy. Moreover, neutron radiation is completely nondestructive; therefore, neutron-scanning is used to study even delicate
biological systems or archeological samples. The neutron scattering cross sections from nuclei of neighboring elements in the
periodic table may be substantially different, which makes it possible to see light nuclei in the presence of heavy ones. This behavior
is a great advantage over the X-ray scattering technique and leads to using isotopic substitution and to easily distinguishing the
neighboring elements. Particularly in the presence of hydrogen in a structure, the nuclei of hydrogen and deuterium are strong
scatterers for neutrons, whereas X-ray scattering is practically insensitive to both. This means that using neutrons, we can determine
the position of hydrogen in a crystal structure. Furthermore, the neutron-scattering lengths of hydrogen and deuterium have
opposite signs, which makes it possible to apply contrast variation. By changing the isotopic composition of the sample buffer
(by varying the amount of hydrogen and deuterium), it is possible to change the contribution of various components of the studied
object into the scattering. This contrast variation technique and the deuteration of the sample or the sample solution constitute one
of the most common methodologies used for soft matter and biochemical structure determination by neutron scattering.
Another very important feature of neutrons is that they have a magnetic moment, and therefore can be used to study microscopic
magnetic structure and magnetic fluctuations, which determine macroscopic parameters of matter. In this review, we will not cover
this topic, but an exhaustive review can be found in Ref. 5.
Neutron-scattering experiments use exclusively “thermal” neutrons, which means that the speed of the neutron are previously
reduced or moderate by inelastic interaction with a “thermalized” medium at suitable temperatures: 20 K using liquid hydrogen for
cold neutrons, 300 K for thermal neutrons, and 2873 K for hot neutrons.
The “window” of wavelength ranges from 0.5 to 20 Å, which are values comparable to interatomic distances; in addition, the
energy window is comparable to kinetic energies. These two features make neutron-scattering techniques suitable for the study
of the structure of matter at the atomic to nanometer scale, as well as for the study of atomic and molecular motion (diffusion,
rotations, vibrations, phonons, etc.).

2.13.2.1 Neutron Sources


Nowadays, there are two kinds of neutron sources designed to produce high-resolution/high-flux neutron beam: (1) by nuclear
fission reactor sources and (2) by spallation in large accelerator-based neutron sources.
With the first method, a high continuous flux of neutrons is produced in the core of a conventional fission reactor. The fission takes
place when a fissile nucleus captures a neutron and it decays splitting into smaller parts. The most used nucleus is 235U, although there
are a few others. In the fission reaction, a thermal neutron is absorbed by the 235U nucleus. The nucleus then passes into an excited state
and is divided into some fission fragments (semiheavy atoms) and 2–5 fast neutrons. On average, each reaction yields 2.5 neutrons,
which 1.5 are needed to sustain the chain reaction. Therefore, each reaction gives 1 useful neutron with a kinetic energy of about 2 MeV.
Neutron Scattering 265

In the second method, a target of heavy elements is bombarded with high-energy particles (typically protons). The incident
particle collides with individual nucleons inside the nucleus and the struck nucleons cause further collisions, forming a particle
cascade called the intranuclear cascade. Additional cascades are caused by the hadrons produced in the first stage, and this process
produces a very large number of neutrons. Spallation sources are typically pulsed, but can also be pseudocontinuous, depending on
the proton accelerator.
Because the produced neutrons are very fast, both types of sources are equipped by several kinds of moderators in order to reduce
their energies by many orders of magnitude to make the neutrons useful for research purposes. Afterward, they are transported to the
neutron-scattering instruments through neutron guide systems or beamtubes.
More detailed description of neutron production, moderation, beam optics, instrumentation, and detection may be found
elsewhere.6–10
Both types of neutron sources described previously are built as dedicated facilities to provide the international scientific
community with the brightest beams of neutrons possible, each of which hosts tens of instruments tailored for a particular type
of measurement and range of scientific applications.
At the time of writing, more than 20 neutron facilities are in operation worldwide, and 4 new neutron sources are currently under
constrictions. The most important being the reactor source ILL in Grenoble (France), 11 already mentioned in the introduction of
this article, and the spallation source ISIS Facility at the Rutherford Appleton Laboratory in Didcot, United Kingdom.12 The ILL
neutron beams are continuous, producing 1.5  1015 neutrons/s/cm2, with a thermal power of 58 MW, whereas those at ISIS are
produced in short bright pulses, 50 times per second.
Other sources are spread out in Europe and around the world; to have an exhaustive list of all the existing neutrons facilities
nowadays, visit the National Institute of Standards and Technology (NIST) web page. 13

2.13.3 Basis of Neutron Scattering

In this section, we intend to give a general overview of neutron scattering, defining the main formulas and using a theoretical
approach to understanding the interaction between neutrons and matter.
As with light, neutrons have the characteristic of wave-particle duality. The motion of a neutron is deduced from the laws of
propagation of the associated wave. We therefore support this duality to present the neutron-scattering theory in the following
sections.

2.13.3.1 Scattering Cross Section


In a neutron-scattering experiment, the measured quantity is the fraction of the neutrons with incident energy Ei that are scattered in
a solid angle dU with a final energy within an interval of energy dEf around Ef. This quantity is called the partial differential scattering
cross section and is defined as
d2 1 number of neutrons scattered per second in dU and dEf
¼ ; (1)
dUdEf F0 dUdEf
where F0 is the incident flux in (neutrons/cm2/s). If the energy of scattered neutrons is not analyzed and only neutrons scattered in
dU are counted, the differential scattering cross section is measured:
ds 1 number of neutrons scattered per second in dU
¼ : (2)
dU F0 dU
In the end, if all neutrons scattered in all directions are detected, the measured quantity is the total scattering cross section:
number of neutrons scattered per second
sscatt ¼ : (3)
f0

2.13.3.2 Scattering by a Single Nucleus


Let consider a collimated neutron beam incident on a single scattering center and traveling in the x-direction. This can be
represented by a plane wave Finc ¼ Aeikx with unit amplitude. Because the neutron sees the scattering center (a nucleus) as a point,
the scattering will be isotropic. The scattered neutron beam for a single scattering center in the origin can be represented as a spherical
wavefront of amplitude b/r as follows:
 
b ikx
Jscatt ¼ A e : (4)
r
Because we consider the scattering center rigidly fixed and the energy of the neutron is too small to change the internal state of the
nucleus, the scattering is elastic, that is, there is no change in the neutron’s energy, so the incident and scattered wave vectors both
have magnitude k.
266 Neutron Scattering

The constant b in Eq. (4) is the scattering length, and it represents the strength of the interaction between the neutron and the
nucleus. Here, b plays a central role in scattering process analysis, and detailed features of this are described in “Coherent and
Incoherent Scattering” section, later in this article.
The minus sign in the expression indicates that b is a positive number for repulsive interaction between neutron and nucleus.
Considering the neutron velocity n, the incident flux F0 is

F0 ¼ jJinc j2 v ¼ A2 v; (5)
while the number of neutrons passing through the area dS per unit time is
A2 b2
vdSjJscatt j2 ¼ vdS ¼ vA2 b2 dU: (6)
r2
From the definition of the differential cross section in Eq. (2), we can derive that
ds vA2 b2 dU
¼ ¼ b2 : (7)
dU vA2 dU
From Eq. (3), we can finally related the scattering length b and the total cross section sscatt as follows:
sscatt ¼ 4pb2 : (8)
 24
Cross sections are generally expressed in barns, where 1 barn ¼ 10 2
cm .

2.13.3.3 The Master Formula of the Scattering


Neutrons are uncharged particles, and they interact magnetically through its spin. Considering only systems that do not possess
unpaired electrons that create a magnetic moment, the only magnetic interaction that takes place is thus the interaction of the
neutron spin with those of atomic nuclei.
In the energy range of the neutrons used for condensed matter studies, the coreneutron interaction during a scattering event is
a very short range force (of the order of 10 13 cm), which is appropriate to describe it in a form of Dirac delta function, the Fermi
pseudopotential, as follows:
2pZ2 ! ! 
bd r  R j ; (9)
m
where m is the mass of the neutron, b is the scattering length introduced in Eq. (4), and r and R represent the instantaneous position
of the neutron and the nucleus j, respectively.
The interaction potential V, which represents the interaction between the neutron and the sample is obtained as the sum over all
the atoms in the sample, as follows:
2pZ 2 X  
Vð!
rÞ¼ bj d !r !
rj : (10)
m j

Now we go back to the principle of scattering introduced in “Scattering Cross Section” section earlier in this article. In
!  the intensity of neutrons scattered by matter (per incident neutron) is as a function of Q and E.
a neutron-scattering measurement,
The scattered intensity I Q ; E , expressed as the partial differential scattering cross section in Eq. ( 1), is known as the neutron-
scattering law and yields information about the structure and the dynamics of the sample as a function of the energy transfer Zu and
!
the momentum transfer Q defined by
Zu ¼ Ei  Ef
! ! ! : (11)
Q ¼ ki kf
In the case in which there is no energy exchange between the neutron and the sample, we talk about elastic scattering where
Zu ¼ 0, and the Q momentum transfer is given by Bragg’s law Q ¼ 4p/l sin 2q, where 2q is the scattering angle. In the other instance,
when we have a loss or a gain in the energy after the scattering event so that Ei sEf (by convention Zu is positive when neutron loses
energy), we talk about inelastic scattering. In Fig. 1, a sketch of the principle of elastic scattering and quasi-elastic is shown with
!
graphical representation of the vector Q .
Next, we deal with the scattering from two nuclei, labeled j and j 0 placed at fixed positions at time t. Using the definition of the
pseudopotential Vð! r Þ given in Eq. (10) and the Born approximation,14 that assumes the interaction with a scatterer point inde-
!
pendent of the scattering by the other scatterers and that the probability for a neutron with a wave-vector k i being scattered by
!
a potential Vð r Þ is proportional to
ð 2 ð 2
 i! i!
  
 e k i ,! ! Þe k f ,! !  ¼  ei!
Q ,!
r Vð!
r Þd!
r 
 ;
r r
 Vð r d r   (12)
Neutron Scattering 267

! !
Figure 1 Illustration of the scattering process with the incoming neutron beam (wave vector k i ) and the outcoming beam (wave vector k f ). The
! ! !
neutrons are scattered through a 2q angle with a scattering vector of Q ¼ k i  k f . In the elastic case (on the top), the neutrons do not exchange
energy during the scattering. In the inelastic case (on the bottom) neutrons can lose or gain energy after the scattering.

! 
we can generalize I Q ; E as follows:
!  ð þN !

d2 s k 1 X N
i Q $!
! r ðtÞ iut
r j0 ð0Þ i Q $!
I Q; E ¼ ¼ f b+
0 bj e e j
e dt; (13)
dUdEf ki 2pZ jj0 j N

where N is the total number of scatterers in the system, j represents a single atom in the system, and bj and !
r j ðt Þ are the
corresponding scattering length and position operator of that atom at time t, described by the Heisenberg time-dependent
representation, respectively.

2.13.3.4 Coherent and Incoherent Scattering


The neutron-scattering length b defined by Eqs. (4), (8) is the key parameter in scattering theory. Its value depends on the nature of
the incident quantum (neutrons in this case) and its interaction with the atoms present in the samples.
While the X-ray scattering lengths of atoms are proportional to the atomic number, neutron-scattering length varies irregularly
with the type of isotope in question, thus taking into account their relative nuclear spin state orientation (e.g., hydrogen nucleus has
b ¼ 3:7515 m, whereas deuterium has b ¼ 6:6715 m).
This characteristic determines significant differences in the scattering properties and leads to the use of this behavior to “mark”
a specific atomic species in the sample making an isotopic substitution for that. The possibility of varying the contrast by isotopic
substitution represents the major advantage of neutron scattering over other radiation techniques, and we will show some examples
of this method later in this article in “ Multiatomic systems” and “SANS Instruments” sections.
The imaginary part of b is related to the absorption cross section, which means that these nuclei strongly absorb neutrons. It is
generally very small in terms of thermal neutron energies, with some exceptions for such isotopes as 113Cd, 155Gd, and so on. 15 The
value of b is experimentally determined, and all the neutron-scattering length values for each isotope are listed in Ref. 16 and easily
accessible online at the “Neutron Scattering Lengths and Cross Sections” NIST web page.17 On the same web page, they report the
values of the scattering cross sections, which we will introduce next for all the isotopes as well. In this instance, b can be positive or
negative, which indicates in this case a phase shift of p radiants in the scattering.
Let us now make some further considerations of the expression bj0 bj in Eq. (13), considering the average over the distribution of
the isotopes and of the nuclear spin orientations.
Assuming that there is no correlation between values bj0 and bj, pertaining to two different sites j and j0 , we can write

bj0 bj ¼ bj0 bj ¼ b ¼ bcoh ;


2
(14)

where bcoh is defined as the coherent scattering length; meanwhile, considering the same site ðj0 ¼ jÞ, we have

bj0 bj ¼ bj bj ¼ b2 ¼ bincoh ; (15)


268 Neutron Scattering

where bincoh is defined as the incoherent scattering length. From Eq. (8), we can redefine the total cross section sscatt and two distinct
terms (coherent and incoherent cross section, respectively), as

sscatt ¼ 4pðbÞ2 ;
2
scoh ¼ 4pðbÞ ;
: (16)
 2
sincoh ¼ sscatt  scoh ¼ 4p ðbÞ2  b

In later sections of this article, we will specifically discuss the neutron techniques and the instruments most demanded in the
investigation of mesoporous systems and supramolecular assemblies that are present at the ILL neutron source in Grenoble.

2.13.4 Neutron Diffraction

The neutron diffraction technique, on single crystals, powders, or liquids, is widely used to determine the structure of materials in
various areas, from material science to biology. In principle, it is based in the same theoretical grounds than the better-known X-ray
diffraction technique. The main difference is the size of the interaction potential, which is comparable to the electronic cloud for
X-rays, which is almost punctual for neutrons (the Fermi pseudopotential). This difference in size makes a Q-dependent atomic
form factor appear in the X-ray interaction with atoms, while in the case of neutrons, such a factor is a constant known as the
scattering length.
Thus, for neutrons, the general expression for the partial differential cross section for nuclear scattering is
ð þN !

d2 s k 1 X N
! !!
¼ f bj0 bj ei Q $ r j0 ð0Þ ei Q $ r j ðt Þ eiut dt; (17)
dUdEf ki 2pZ jj0 N

where N is the total number of scatterers in the system, j represents a single atom in the system, and bj and ! r j ðt Þ are the
corresponding scattering length and position operator (this is the Heisenberg time-dependent position operator) of that atom at
time t, respectively. The brackets represent the thermal average of the position operators at temperature T. 18
The pairwise dynamical structure factor defined as follows:
!  ð

1 þN !! !!
Sjj0 Q ; u ¼ dt eiut ei Q $ r j ð0Þ ei Q $ r j0 ðtÞ (18)
2pZ N
which expresses an intrinsic property of the system, completely independent of the probe used to study it. It represents the
correlation between two atoms (j and j0 ) at times t and 0. Then, the partial differential cross section can be written as
d2 s k ! 
¼ f S Q; u ; (19)
dUdEf ki
! 
where S Q ; u is the dynamical structure factor for neutrons,
!  XN ! 
S Qu ¼ bj0 bj Sjj0 Q u : (20)
jj0

Replacing the scattering lengths by ones, we obtain a well-known dynamical structure factor that usually is derived from
numerical simulation:
!  X N ! 
Ssim Q ; u ¼ Sjj0 Q ; u ; (21)
jj0

which is independent of the probe.


 As shown in Ref. 18, this structure factor is the Fourier transform of the time-dependent (or Van
Hove) correlation function G ! r ;t :
!  ðð ! !  
1 
S Q; u ¼ d!r dt ei Q $ r ut G ! r ;t : (22)
2p
This correlation function is related to the probability of finding a particle at position ! r and time t, provided that the other
particle was at the origin at t ¼ 0.
!
On the other hand, by replacing the scattering lengths in Eq. (20) by the atomic form factors for X-rays, fj ðQ Þ, we obtain the
equivalent expression for the dynamical structure factor for X-rays:
!  X N    
! ! ! 
SX Q ; u ¼ f Q j0 f Q j Sjj0 Q ; u : (23)
jj0
Neutron Scattering 269

Here, we will focus in the neutron-scattering case using the dynamical structure factor defined in Eq. (20), which takes different
forms depending on the properties of the system under study. This structure factor can be split in a very useful way, as shown in
Ref. 18, separating the coherent and incoherent terms
!  s X
N !  s ! 
Scoh Q ; u ¼ coh Sjj0 Q ; u ¼ coh Ssim Q ; u (24)
4p jj0 4p

and
!  s X N ! 
inc
Sinc Q ; u ¼ Sjj0 Q ; u ; (25)
4p jj0
 
where scoh ¼ 4pb and sinc ¼ 4p b2  b .
2 2

2.13.4.1 Nuclear Scattering by Crystals


In this section, we evaluate the structure factor when the scattering system is a single crystal. We consider a Bravais crystal with one
!
atom per unit cell, whose sides are denoted by ! a 1; !
a 2 and !
a 3 . A lattice vector is given by ‘ ¼ ‘1 ! a 1 þ ‘2 ! a 2 þ ‘3 !a 3 , where ‘1, ‘2,
!  ! ! 
and ‘3 are integers, and the volume of the unit cell is y0 ¼ a 1 $ a 2  a 3 . The reciprocal lattice is defined as a lattice having the
! !  ! 
2p ! !  ! ! 
unit-cell vectors ! s 1 ¼ 2p
y0 a 2  a 3 ; s 2 ¼ y0 a 3  a 1 , and ! s 3 ¼ 2p
y0 a 1  a 2 , and a volume given by
! ! !  ð2pÞ3
s 1 $ s 2  s 3 ¼ v0 .
! !
Due to thermal motion, the nucleus ‘ is displaced from its equilibrium position ‘ by a vector ! u ‘ ðt Þ at !
r ‘ ðt Þ ¼ ‘ þ ! u ‘ ðt Þ. Then
the coherent and incoherent dynamical structure factors take the following forms:
!  s

N iut X N ! ! i!
i Q , ‘ Q ,!
! u ð0Þ
u ‘ ðt Þ i Q ,!
Scoh Q ; u ¼ coh e e e e 0
(26)
4p 2pZ ‘

and
!  s

inc N iut
! u ðtÞ i!
i Q ,! Q ,!
u 0 ð 0Þ
Sinc Q ; u ¼ e e 0
e : (27)
4p 2pZ
Now we can give a physical interpretation to these two contributions. The coherent part represents the correlations between any
atom at time 0 with any other atom at time t, while the incoherent part is the self-correlation between an atom at time 0 and the
same atom at time t.
Assuming that the interatomic forces in the crystal are harmonic (i.e., that the forces are linear functions of the displacements),
the displacements can be expressed as the sum of displacements due to a set of normal modes. This method allows for separating the
elastic and inelastic parts of the coherent and incoherent structure factors, which is known as phonon expansion. We do not describe
this expansion here; the interested reader can find the details in Ref. 18 or other textbooks.

2.13.4.1.1 Coherent Elastic Scattering


The first term in the phonon expansion of the coherent scattering represents the elastic scattering; that is, the interaction for which
the neutron does not exchange energy with the system (kf ¼ ki). In that case, the coherent dynamical structure factor (Eq. 26) can
easily be integrated over all final energies, which yields the following expression of the coherent static structure factor:

! s XN !!
Scohel Q ¼ coh N expð2W Þ ei Q , ‘ ; (28)
4p ‘
 2
where exp( 2 W) is the Debye–Waller factor, also defined as exp  Q$e Uei ; see Eq. (70). Using the properties of the Dirac delta
function, this expression can be rewritten as
  X ! 
ds ! s ð2pÞ3 !
¼ Scohel Q ¼ coh N expð2W Þ !sd Q s ; (29)
dU cohel 4p y0

where the sum extends over all vectors in the reciprocal space. This equation tells us that the scattering occurs only when
!
the momentum transfer vector equals a vector of the reciprocal space. In other words, whenever the vector Q matches one in the
reciprocal space, we should observe a peak in intensity. This condition is the same as Bragg’s law for X-ray scattering, and the
coherent elastic scattering of neutrons is simply Bragg scattering. If 2q denotes the scattering angle, Bragg’s law gives the relationship
between the scattering angle and the momentum transfer as Q ¼ 4p l sin q, where l is the wavelength of incident neutrons.
270 Neutron Scattering

The Debye–Waller factor can easily be calculated for a cubic crystal as 2W ¼ hui2 =3. But even for a noncubic crystal, when an
average over all normal modes is performed, this expression is approximately correct.
Now we will consider the case on a non-Bravais crystal, for which there are more than one atom per unit cell. In this case, the
! !
general expression for the position of an atom is r ! ðt Þ ¼ ‘ þ d þ u‘d ðt Þ. The average scattering length (over isotopes and nuclear
‘ d
spin) is now dependent on the position in the unit cell; thus, it must be denoted as bd . Also, the Debye–Waller factor depends on the
atomic position in the unit cell and takes the form Wd. Taking into account these changes, we use the same method in the calculus of
the coherent static structure factor, which this time yields to
  !
ds ð2pÞ3 X ! ! 2
¼ Scohel Q ¼ N d Q  s FN ð! s Þ ; (30)
dU coh el y0 !
s
P
Nc ! !
where FN ð!s Þ ¼ expðiQ , d ÞexpðWd Þ is the nuclear unit-cell structure factor and Nc is the number of atoms in the unit cell.
d
In order to interpret the experimental data properly, the mathematical expressions given by Eqs. (29), (30) should be translated
to real measurements of the coherent elastic scattering. In such a measurement, we determine the intensity of a Bragg peak (i.e., the
!
integrated number of neutrons as we pass through the condition Q ¼ ! s ). Owing to instrumental resolution and mosaic spread in
the crystal, the d-function in the theoretical cross section is spread out into a peak with finite width. We shall assume that the colli-
mation of the incident and scattered beams is always sufficiently relaxed for all the neutrons scattered into the Bragg peak to be
counted. The quantity actually measured is therefore the total cross section stot (i.e., integral of the partial differential cross section
with respect to U in all directions in space).
! !
When neutrons of wave vector k i are incident on a crystal, so that the angle j between k i and a specific reciprocal lattice vector
!s is fixed, the total cross section is given by

ð2pÞ3 2 2  
stot !
s
¼N FN ð!
s Þ d s2  2ki s cosj ; (31)
y0 p
! !
where p ¼ k i  !
s.

2.13.4.1.2 The Laue Method


In this case, we have neutrons in the fixed direction j and with a continuous range of wavelengths that are incident on a crystal of
!
a fixed orientation. For a general value of ki there is no Bragg scattering because the scattering vector Q does not equal the reciprocal
! ! !
vector s . But for the particular value ki ¼ s=ð2 cos jÞ, Q ¼ s , and the Bragg condition is satisfied. If the flux of incident neutrons is
F(l), then the intensity of the peak corresponding to ! s is
V l4  2
P¼N 2 FðlÞ FN ð!
s Þ ; (32)
y0 2 sin q

where V is the volume of the crystal and l is the wavelength of the scattered neutrons.

2.13.4.1.3 Rotating Crystal


!
In this method of measuring a Bragg peak, a monochromatic beam of neutrons of wave vector k i is incident on a crystal that can be
rotated.
The scattering angle 2q is set to satisfy the relation s ¼ 2ki sin q. The crystal is rotated about an axis perpendicular to the plane
! !
containing k i and k f so that the reciprocal lattice vector ! s remains in that plane. As the crystal is rotated (i.e., as the angle j
! !
between k i and s is varied), the vector !
! s matches Q and Bragg scattering occurs. The counting rate as a function of j is known
as a rocking curve. If F is the flux of incident neutrons, the integrated number of scattered neutrons per unit time in the Bragg peak is
V l2  2
P¼ 2 F FN ð!
s Þ ; (33)
y0 sin q

where l is the wavelength of the incident neutrons.

2.13.4.1.4 Powder Diffraction


!
A monochromatic beam of neutrons with wave vector k i is incident on a powder sample (i.e., a sample of many small single
!
crystals with random orientations). For a specified value of s (< 2ki), the wave vector k f of the scattered neutrons lies on a cone
!
known as a Debye–Scherrer cone ( Fig. 2). The axis of the cone is along k i , and only those microcrystals whose ! s vectors lie on
!
a cone with the same axis and semiangle j ¼ p/2  q contribute to scattering. The direction of k i is fixed and, for each microcrystal,
the vector !s points in any direction in space with equal probability. Then the total cross section for each cone is
V l3 X 2
stot s ðconeÞ ¼ FN ð!
s Þ ; (34)
y20 4 sin q !
s
Neutron Scattering 271

Figure 2 (A) The Debye–Scherrer cones in powder diffraction, intersecting with the plane of the detector. Three cones corresponding to three
diffraction planes are shown. (B) The signal observed in the detector corresponds to the projection of the Debye–Scherrer cones on the detector
surface (gray rectangle), which are represented by the arcs at the corresponding scattering angles.

where the sum over ! s is the sum over all reciprocal lattice vectors with the same value of s. If the neutron detector is at a distance r
from the target and has an effective diameter d, it intercepts a fraction d/(2pr sin 2q) of the neutrons in the cone. The counting rate is,
therefore,
V d l3  2
P¼F FN ð!
s Þ : (35)
2
y0 8pr sin 2q sin q

In the case of single crystals, the analysis of the structure is made through the positions of the Bragg spots, but the information of
the peak intensities is quite difficult to analyze. In powder diffraction experiments, there is a release of information due to the
projection of the tridimensional reciprocal space in only one dimension, but more quantitative analysis can be done using the
Rietveld method. 19 There are several programs used to analyze the powder diffraction data, but the most widely used is the FullProf
Suite, developed and maintained at ILL.20 The model used for describing the powder pattern profile is based on the following
expression:
X X  
yi calc ¼ S4 I !U Ti  T ! þ Bi ; (36)
4 ! 4h 4h
h
!  
where h labels the Bragg reflections, the subscript 4 labels the phases existing in the model. Here, U Ti  T ! is the value of the
ih
function U(x), normalized to unit area and selected to describe the peak shape at the position Ti due to the reflection centered at
19
T !. Finally, S4 is a scale factor and Bi is the background at Ti. In the Rietveld method, the functions I, U, and B are calculated
4h
using a model that depends on a series of parameters. The structural information is contained in the integrated intensities
(proportional to the nuclear unit-cell structure factor) and the peak positions (through the cell parameters and propagation vectors),
while the instrumental and microstructural effects are included in the peak shape function. If the set of model parameters is
!
b ¼ b1 ; b2 ; .; bp , the Rietveld method optimizes the c2function as follows:
X  !2
c2 ¼ ui yi; obs  yi; calc b ; (37)
i

where wi is the inverse of the variance associated to the observation i.

2.13.4.1.5 Instruments
There is a wide variety of instruments for neutron diffraction, each one adapted for a specific diffraction technique. For single-crystal
diffraction at the ILL, there are at least three worth mentioning: D3, 21,22 D9, and D10. They are all based in a continuous neutron
source, with a monochromatic incident beam. The first two use hot neutrons [i.e., neutrons of short wavelengths (< 1 Å)], while the
latter use thermal neutrons. The diffractometer D3 has the possibility of performing neutron diffraction with polarization analysis.
For powder diffraction, there are three two-axis instruments at ILL: D1B, 23 D2B,24,25 and D20,26 all working with a monochro-
matic beam of thermal neutrons. The diffractometer D2B is a very high resolution instrument, which is very well adapted to solve
positions and occupations in complex unit cells. The instruments D1B and D20 have large detectors in a bananalike configuration
that allows one to register the whole diffractogram (120 degrees and 150 degrees, respectively) in one shot (Fig. 3). This is very
272 Neutron Scattering

Figure 3 Scheme of the D20 two-axis diffractometer at ILL. The bananalike, position-sensitive detector has 1536 individual detectors covering an
angular range of 154 degrees.

useful for following the structural evolution as a function of applied conditions on the sample (temperature, pressure, electric or
magnetic field, etc.). In the specific case of temperature, the technique is known as thermodiffractometry.

2.13.4.2 Total Scattering


The ideal crystal does not exist in nature, and real-life materials have always a kind of disorder. The standard techniques for
crystalline materials can still be used if the degree of disorder remains small, but when the disorder is more and more important,
like in quasi-crystals, nanoparticles, adsorbed ions in surfaces and pores, glasses, amorphous systems and liquids, the technique of
total scattering becomes necessary. In this case, the total scattering signal is analyzed, not only the elastic coherent part.
Taking the expression for dynamical structure factor given by Eq. (20), we now consider the coherent part (Eq. 24). We assume
that the incoherent part can be properly subtracted from the total structure factor; thus, we will focus on the coherent part. Unlike
what was done before, this time we do not isolate the elastic part; rather, we shall consider the total signal (i.e., the integral of the
coherent dynamical structure factor over all energies):
! ð þN !  ð !!  
S Q ¼ duS Q ; u ¼ d! r ei Q , r G !
r ;0 : (38)
N

The previous integral defines the static structure factor as the Fourier transform of the Van Hove correlation function at time
zero. 18 In other words, the static structure factor gives information on the structure of the system “frozen” at time zero. This
correlation function at t ¼ 0 can be written as the sum of two terms: first, a delta function at the origin, corresponding to the
autocorrelation, plus the density fluctuations
     
G ! r ;0 ¼ d ! r þ rg ! r ; (39)

where r is the macroscopic density of the system and g ! r is the pair distribution function (PDF). Introducing this expression in
Eq. (38), the relationship between the static structure factor and the PDF is straightforward:27
! ð  
!!
S Q  1 ¼ r d! r g ! r  1 ei Q , r : (40)
V

Assuming an isotropic structure factor (i.e., the structure is independent of the direction in the reciprocal space), the angular part
of the previous integral can easily be solved, yielding the following relations between the isotropic structure factor and the 1D-PDF:
ð
4pr N
SðQÞ  1 ¼ r ½ g ðr Þ  1sinðQr Þdr (41)
Q 0
Neutron Scattering 273

Figure 4 The static structure factor (left) and the total correlation function (right) of Ge sulfide glasses, GexS1  xfor selected samples x ¼ 0.333,
0.400, and 0.467 (from bottom to top). 28 These are the result of combined experiments performed at reactor and spallation sources, on D4 and
SANDALS, respectively.

and
ðN
1
g ðr Þ  1 ¼ Q½SðQÞ  1sinðQr ÞdQ: (42)
2p2 rr 0

The PDF function contains more information than that contained in the elastic-coherent structure factor, but the main drawback
is that the three-dimensional (3D) structure has been collapsed in only one dimension. The situation is the same as in powder
diffraction, but here, we do not have the help of periodicity to simplify the problem. For disordered materials, the pairwise corre-
lation is rapidly lost as a function of distance, giving information about the short- and intermediate-range order.
As an example, the left side of Fig. 4 shows the structure factor for a series of Ge-rich sulfide glasses,28 where a strong dependence
of the first sharp diffraction peak is apparent. This peak is characteristic of glassy structures and is related to intermediate-range
order. The right side of Fig. 4 shows the corresponding total correlation functions, which are slightly modified PDFs: T(r) ¼ 4prrg(r),
where g(r) is the sinus Fourier transform of the structure factor (see Eq. 42). These functions present a sharp peak at about 2.3 Å,
corresponding to the first coordination shell. In fact, that peak is a combination of a Ge–S correlations (the nearest neighbors in
tetrahedral units), a Ge–X correlations, where X ¼ Ge or S, and short Ge–Ge second-neighbor contacts in edge-sharing tetrahedra.

2.13.4.2.1 Multiatomic Systems


For systems having n different atomic species, the structure factor can be generalized as follows:
X
n
SðQÞ ¼ cj cj0 bj bj0 Sjj0 ðQÞ; (43)
jj0

where Sjj0 ðQÞ represents the partial structure factors and cj is the atomic fraction for the species j. This is a linear system with n(n þ 1)/
2 unknowns that can always be written as a matrix equation. To solve it, n(n þ 1)/2 independent experiments are required, but we
cannot change the composition without changing the sample itself. The only choice is changing the scattering lengths using isotopic
substitution, anomalous scattering, or combining neutron and X-ray experiments.
As an example, for a binary system, we need three independent experiments, and the linear system is
0 1 0 2 2 10 1
SS1 ðQÞ cX bX1 c2Y b2Y1 2cX cY bX1 bY1 SXX ðQÞ
B C B CB C
@ SS2 ðQÞ A ¼ @ c2X b2X2 c2Y b2Y2 2cX cY bX2 bY2 A@ SYY ðQÞ A (44)
SS3 ðQÞ 2 2 2 2
cX bX3 cY bY3 2cX cY bX3 bY3 SXY ðQÞ

where X and Y stand for the two atomic species and S1, S2, and S3 for the three samples. Inverting this matrix, the three partial
structure factors, SXX(Q), SXY(Q), and SYY(Q), can be obtained as a function of the three experimental structure factors, SS1(Q),
SS2(Q), and SS3(Q).
274 Neutron Scattering

2.13.4.2.2 Coordination Numbers


When we need to calculate the coordination numbers (i.e., the number of atoms around a given atom), we use the radial distribu-
tion function or linear density, defined as
X
n X
n
Rðr Þ ¼ 4prr 2 g ðr Þ ¼ cj cj0 bj bj0 Rj j0 ðr Þ; (45)
j¼1 j0 ¼1

where Rj j0 ðr Þ ¼ 4prr 2 gj j0 ðr Þ.
The coordination numbers are different for various coordination shells but also must be distinguished among the chemical
species. Let us consider the number of particles of type b around a given particle a. This number can be calculated as follows:
ð ð
nba ¼ 4prb r 2 gab ðr Þdr ¼ cb Rab ðr Þdr; (46)

where rb ¼ rcb . Instead of the R(r) function, in Fig. 4, we use the total correlation function T(r) ¼ R(r)/r. The reason of this choice is
the fact that the peaks are more symmetrical in T(r) than in R(r), allowing the use of Gaussian functions in the fitting process.
For a given peak in the R(r) corresponding to a single correlation a–b, Eq. (45) can be written as
 
Rðr Þ ¼ ca cb ba bb Rab ðr Þ þ 1  dab ca cb ba bb Rba ðr Þ; (47)

where Rba ðr Þ ¼ Rab ðr Þ and dab is Kronecker’s delta. Integrating both sides over r, the left term is the integral of that peak, while the
right side gives the coordination numbers as follows:
 
ab ¼ ca ba bb nba þ 1  dab cb ba bb nab :
n (48)

ab ) is related
For heteronuclear correlations, a single experimental value (the peak integral n  to two unknowns: the number of b-type
atoms around an a-type atom (nba) and the number of a-type atoms around a b-type atom nab : n ab ¼ ca ba bb nba þ cb ba bb nab . In this
case, we need to know one of the two coordination numbers if we wish to obtain the other one.
For homo-nuclear correlations a–a, the situation is simpler because the experimental value n aa is related only to the naa
unknown as follows: n aa ¼ ca b2a naa .

2.13.4.2.3 The First Difference Method


For binary (or ternary) systems, the isotopic substitution is still affordable because only three (or six) independent experiments are
necessary to get the whole set of partial structure factors. But when four or more atomic species are present in the system, it is almost
impossible to have 10 or more independent experiments. In such cases, making at least one isotopic substitution, 29 we can obtain
partial information about the correlation functions.
Let us assume that we perform only one isotopic substitution, replacing the atomic species g by two of their isotopes, g1 and g2,
with scattering lengths given by bg1 and bg2 Then, for the isotope g1, we separate the terms, including the correlations where g is
present, as follows:
X
n X
n
g1
SðQÞ ¼ c2g b2g 1 Sgg ðQÞ þ 2cg bg1 cj bj Sjg ðQÞ þ cj cj0 bj bj0 Sj j0 ðQÞ (49)
jsg j;j0 sg

and the same applies to the second isotope g2:


X
n X
n
g2
SðQÞ ¼ c2g b2g2 Sgg ðQÞ þ 2cg bg 2 cj bj Sjg ðQÞ þ cj cj0 bj bj0 Sjj0 ðQÞ: (50)
jsg jj0 sg

Note that the atomic composition is exactly the same for both samples; the only change is the coherent scattering length of the
substituted isotope. Then, making the difference between the two previous expressions, we remove the last term:
   X
n
g1
SðQÞg2 SðQÞ ¼ c2g b2g1  b2g2 Sgg ðQÞ þ 2cg bg1  bg2 cj bj Sjg ðQÞ: (51)
jsg

Rearranging the terms, we can isolate the g–g correlation:


P
n
g1 SðQÞg2 SðQÞ
2 ca ba Sag ðQÞ
asg
  ¼ Sgg ðQÞ þ   ; (52)
c2g b2g1  b2g2 cg bg1 þ bg2

where the second term is usually negligible or can be neglected using distance constraints. It is worthwhile to point out the
denominator in the left part of the previous equation. This difference is a kind of measure of the contrast of the substitution: The
bigger this contrast is, the smaller are the errors in the structure factor determination.
Neutron Scattering 275

In some cases, substituting two atomic species (g and d), doing four independent experiments (with isotopes g1, g2, d1, and d2)
and applying the first difference method twice, we are able to isolate the single g–d correlation:
   
g1 g2 g1 g2
d1 SðQÞd1 SðQÞ  d2 SðQÞd2 SðQÞ
Sgd ðQÞ ¼   : (53)
2cg cd bg1  bg2 ðbd1  bd2 Þ

In fact, this procedure is called the second difference method because we use the difference of the two first differences to obtain the
sought structure factor. This time, the contrast factor is the product of the differences of the scattering lengths for each atomic species.
In order to illustrate these methods, we can mention a series of experiments on the hydration of ions in clay interlayers. Neutron
diffraction methods using isotopic substitution allow us to probe in detail the local environment of water molecules and cations in
the interlayer region of smectite clays. The ability to separate the ion-hydrogen and ion-oxygen PDFs enables us to distinguish
between binding to water molecules and binding to other oxygen atoms. In the case of the Nd- and Yb-montmorillonites, 30 the
data indicate a liquidlike, hydrogen-bonded interlayer water structure. Similar results were found for other lanthanides, such as
Sm31 and Eu,32 where we observed that the cations in the interlayer region of the clays are less hydrated than in aqueous solution,
showing their ability to stick to the clay surface.
Different experimental techniques, including neutron diffraction with isotopic substitution, extended X-ray absorption fine
structure spectroscopy, and quasi-elastic neutron scattering, in combination with molecular dynamics (MD) simulations, were
also applied to the study of the coordination of the Sm3 þ in the interlayer of hydrated synthetic montmorillonite and hectorite. 33
The neutron diffraction results indicate that not all oxygen atoms in the first coordination shell of the Sm3 þ belong to water mole-
cules, supporting the formation of the Sm3 þ inner-sphere complex. On the other hand, the other techniques suggest that the
adsorbed Sm3 þ cations form outer-sphere complexes with the clay surface. The hypothesis making all results compatible is that
there are different Sm species adsorbed in the clay interlayer: a part of Sm is in the Sm3 þ cationic form, forming outer-sphere adsorp-
tion complexes, and another part is hydrolyzed and present in the interlayer space as SmðOHÞ2þ , SmðOHÞþ 0
2 , or Sm(OH)3 species.
The latter are more hydrophobic than Sm3 þ cations and can be dehydrated and stick to the clay surface.
Similar studies can be performed with more complex systems, where a hydrophobic cation (NC4H12)þ was used to saturate the
interlayer space of nontronite NAu-1 ( Fig. 5) in order to reduce hydration of interlayer cations, which could hinder the effects
related to the clay–water interactions. The water content was low in order to reduce hydrogen bonding between water molecules,
and it was found that these molecules form strong hydrogen bonds with surface oxygen atoms of nontronite. Fig. 6 shows the first
difference real-space function centered at an arbitrary H atom (the Fourier transform of the expression corresponding to Eq. 52).

Figure 5 Schematic view of two nontronite clay layers. The layers are formed by tetrahedrally coordinated Si atoms, octahedrally coordinated Fe
atoms, and oxygen and hydrogen atoms, in an O–Si–H–O–Fe–O–H–Si–O series. Some of the Si atoms in the tetrahedral sheet are substituted with Al
atoms, creating a negative net charge of the layers. The interlayer region consists of the charge balancing cation and water molecules. 34
276 Neutron Scattering

Figure 6 The real-space difference function centered at an H atom of the bulk water. The first distances between H and other (X) atoms are shown
in the inset, and the corresponding peaks are identified with arrows. Peaks 1 and 2 correspond to intramolecular distances. 34

Figure 7 Scheme of the D4 two-axis diffractometer at ILL. The detector ensemble is made of nine detection banks, each one comprising 64
individual detectors. Due to collimation in front of each bank, the detector has to move around the sample to record a complete diffractogram. 35

This function represents the correlations between an H atom in the water molecule with any other chemical species in the system.
The results of this study confirm the assumption that surfaces of smectite clays with tetrahedral substitutions are hydrophilic.34

2.13.4.2.4 Instruments
The experiments given here as examples were all performed at ILL on the dedicated diffractometer for disordered materials and PDF
analysis, D4. 35 It has the same principle of functioning than other two-axis diffractometers (Fig. 7), but it uses short-wavelength
neutrons and consequently covers a wide Q range. This is a medium- to low-resolution instrument with high stability, which makes
it very well suited for the structural study of liquids and amorphous materials. As the instrument 7C236,37 at Laboratoire Léon
Brillouin (LLB), in Saclay, France, it has the simplicity of reactor-based instruments, making experimental corrections and analysis
simpler than instruments based in pulsed sources.
At the spallation sources, there are also several instruments dedicated to total scattering: namely, SANDALS, 38 NIMROD,39 and
GEM40 at ISIS (United Kingdom), and NO-MAD41 at the Spallation Neutron Source (SNS; United States). The main advantage of
these instruments is their large Q-range, which gives a better resolution in real space. Their drawback is a more complex data treat-
ment than in reactor sources.

2.13.5 Small-Angle Neutron Scattering (SANS)

Small-angle neutron scattering (SANS), as well as small-angle scattering using X-ray (SAXS) radiation, are methods used to probe
materials having a structural arrangement from approximately 1 nm up to a few hundred nanometers.
Neutron Scattering 277

Due to the nondestructive and magnetic aspects of neutrons, this technique is largely employed in soft matter for the structural
investigation of molecular self-assembly and interactions in complex fluids, polymers, colloids, surfactants, liquid crystals, protein
solutions, and generally biopolymers in hard condensed matter to investigate grain growth, nanocomposites, advanced ceramics,
porous catalytics, and other substances, and in magnetism studying flux lattices in superconductors, ferrofluids, and the relationship
between structural and magnetic domains and ordering.
From an SAS measurement, it is possible to extract a large number of information concerning the radius of gyration of the nano-
metric objects in the sample, the morphology, volume or mass, and fractal dimension, but also about how the objects are organized
in continuous media.
Because of the high amount of information contained in a single SAS pattern, interpretation of the experimental data is not
always straightforward. Often, different strategies in the preparation for and during the SAS experiment have to be adopted based
on the type of sample to be measured (solution, colloids, nanoparticles, etc.).
We already mentioned the possibility of changing the scattering profile using the isotopic substitution in “ Coherent and
Incoherent Scattering” section earlier in this article. In SANS, this methodology is largely adopted to distinguish different phases
or objects present in the system, particularly by recursing to contrast variation experiments or by using partially or entirely deuter-
ated samples.
Due to the complexity in the data interpretation, the SAS results are frequently supported by complementary measurements,
such as light scattering, ultraviolet (UV), transmission electron microscopy (TEM), etc.
In this section, we will present the principles of the SANS technique, introducing the formalism and the analysis of the scattering
pattern. Then, an overview of the SANS instruments at ILL will be shown, illustrating some representative examples in the nano-
porous/mesoporous media.
An extensive bibliography is available on this exciting technique; here, we are just providing a selection of articles that give the
general aspects of the technique and more technical details concerning the neutron instruments and data analysis.42–46

2.13.6 Principle of SANS Technique

In SANS, we will assume that the scattering is elastic (as discussed in “ The Master Formula of the Scattering” section earlier in this
article); that is, no exchange of energy occurs between the neutrons and the atoms in the sample.
The scattered intensity I(Q) is reported as a function of the scattering vector Q, which is related to the wavelength of the incident
radiation l and the scattering angle q by Q ¼ 4p sin(q)/l with a period d ¼ 2p/Q, where d is considered here as the characteristic
correlation distance between two objects in the sample. Following this formula, to measure structures that have distances of the
order of tens of nanometers and more, we need to work at small Q by increasing the neutron wavelength or by decreasing the scat-
tering angle. As it is very difficult to produce a highly intense neutron flux with a very long wavelength, it is necessary to use small
scattering angles (thus, the origin of the name of the technique).

2.13.6.1 Scattering Length Density


The relevant unit structure is not atoms, but molecules, atom aggregates, or structure inhomogeneities in nanometric scale. All the
atoms belong to these large unit structures generate in phase-scattered waves and their amplitude add up with each other. Conse-
quently, we speak of the coherent scattering length density of the object in question as a sum of all the coherent scattering lengths of the
constitutive atoms bi per molar volume V:
Xn
bi
rb ¼ : (54)
i
V

Taking the differential cross section ds/dU introduced in Eq. (2) for an atomic system, this was defined as a sum over discrete
atoms:
 2
1 ds ds 1 X i! Q ,!

r 
IðQÞ ¼ ðqÞ; ¼  e  ; (55)
V dU dU N  ! 
r

in which I(Q) is the measured scattering intensity. Since it is not possible to isolate the contributions of individual atoms to the
scattering, we can rewrite this equation as an integral over ! r as
ð 2
da 1  ! i!
Q ,! ! 
¼ rbð r Þe r
d r  ; (56)
dU N  V

where rbð!r Þ is the local scattering length density, and this is composed of an average term plus fluctuations around the average
rb þ drð!
r Þ.
For Q > 0, the average term rb  is null; hence,
ð 2
da 1 !! 
¼  drð! r Þei Q , r d!
r  : (57)
dU N V
278 Neutron Scattering

2.13.6.2 Isolated Particles


If we want to apply this formalism to the case of large objects (as large structures as nanometrical molecules, aggregates, and so on)
with an associated homogeneous rp in continuous media with a scattering length density rm, Eq. (57) gives
 2
1 2 ð !! 
ds  
¼ rp  rpm  ei Q , r d! r ; (58)
dU N  Vp 

where Vp is the average particle volume. Here, we are discussing the case of a very diluted system in which we assume that the
interactions between these particles are negligible. For N identical particles, the differential cross section becomes
 2  !  2
ds  
¼ V 2 N rp  rm FP ðQ Þ ;
dU
! 1 ð ! ! (59)
where Fp Q ¼ ei Q , r d!
r:
V V
!
Fp ðQ Þ is known as the particle form factor, and it is possible to extract its analytical form for particles with a well-defined geometry
(bulk or empty sphere, rods, cylinders, etc.).
For a spherical particle of radius R, it is possible to calculate analytically the corresponding form factor, starting from Eq. (59) as
follows, considering that the scattering is isotropic:
" #
 ! 2
Fp ðQ Þ ¼ 3 sinðQRÞ  QR cosðQRÞ : (60)
ðQRÞ3

2.13.6.3 Information From a Scattering Pattern


Fig. 8 shows a typical scattering intensity profile for a mono-dispersed system of particles in solution in highly diluted and more
concentrated solutions. For a dilute system, the SANS pattern exclusively represents the form factor. In this image, we compare the
model form factor function for an ideal system, with all the particles having exactly the same size, and a more realistic function,
considering the polydispersity (variations in the particle size) of the particles in the system.
We can identify three main regimes: Guinier regime, intermediate regime, and Porod regime. These are highlighted in Fig. 8 for better
comprehension. The same figure shows the difference of the scattering function I(Q), changing the particle concentration. The effect
of the particle correlation is visible in the low Q region, characterized by a plateau in the Guinier regime. In the following section, we
will treat in more detail the case of more concentrated solutions in which we have to consider the contribution of the structure
factor S(Q).

Figure 8 Log–log representation of the sphere-shaped particle form factors: ideal scattering function I(Q) for spheres of radius R ¼ 100 Å (dotted
gray); for spheres of radius R ¼ 100 Å smeared with a polydispersity function (red); and for high concentrated spheres, particles of radius R ¼ 100 Å
smeared with a polydispersity function (blue). In the plot, the three relevant Q domains are underlined: Guinier, intermediate, and Porod regimes.
Neutron Scattering 279

2.13.6.3.1 Guinier Regime


The Guinier approximation, applied to particles randomly oriented under the limit of QR  1, allows one to estimate the radius of
gyration RG by means of the definition of the scattered intensity as follows:
 
Q2 R2G
IðQÞ ¼ Ið0Þexp  ; (61)
3
with RG being
ð
1
R2G ¼ r 2 d!
r: (62)
Vp Vp
pffiffiffiffiffiffiffiffi
Considering, for example, a spherical particle with radius R, the gyration radius is RG ¼ R 3=5. From a scattering that measures
I(Q), R2G is obtained from the slope of the line plotting ln I(Q) versus Q2. The I(0) could be related to the molecular weight of the
particle when the intensity is given in absolute scale.

2.13.6.3.2 Intermediate Regime


Analyzing the intermediate regime, we are able to determine the shape and size of the object.
In general, the scattered intensity presents a trend of the IðQÞ  Qa decay in which parameter a is related to the shape of the
object (see Fig. 8). For example, for bulk objects, a ¼ 4 and for a wormlike chain, a ¼ 2.

2.13.6.3.3 Porod Regime


Porod approximation is valid at large Q values, such as QR [ 1.
Looking the profile trend of the scattering intensity, it is possible to detect correlations between different interfaces, and therefore
local interface roughness between two different phases, such as the particles and the media as two-phase system (shown in Fig. 8).
Considering a surface area Ap of the particle, Porod’s law predicts an average scattering function proportional to Q4 , written as
 2
rp  rm
IðQÞ ¼ 2p : (63)
Q4
In case of rough or diffuse interfaces, or when the two media are interpenetrated, the analysis is more complicated. The Fourier
transform of the scattering length density profile of the interface has to be taken into account.
More detailed examples of Porod’s law can be found in Ref. 43.

2.13.6.4 Interparticle Interference


As indicated in the previous section, in more concentrated systems, the contribution given by the interference between particles
becomes very important. This is the case, for example, in biological molecules or colloidal solutions.
To introduce this effect in the definition of the differential cross section, we assume we are dividing the system into Np cells, each
! !
one related to one particle. We define R i as the position of the center of the ith cell and d j as the position of the jth particle relative
to the cell center. The scattering intensity can be written as
* N 2 +

1 X !X !
Ni
i! i!
p
Q $ d j
IðQÞ ¼ e Q $ R i
b e  ; (64)
V  i¼1
ij
j¼1 

where bij is the scattering length for a particle j in cell i and Ni is the number of atoms in the cell.
Considering the form factor as
! XNi !!
Fi Q ¼ bi j ei Q $ d j ; (65)
j¼1

and that the form factor is the same for all particles in a system of identical particles, we finally obtain
Np  ! 2 !
IðQÞ ¼ FðQ Þ S Q : (66)
V
!
Thus, we have introduced function S Q , which is the interparticle structure factor, and it is expressed as
* N N +
! 1 X p X p !  R !
iQ, ! R0 i
S Q ¼ e i
: (67)
Np i¼1 i0 ¼1

It is noteworthy that when we have the uncorrelated particle, S(Q) ¼ 1, find again the expression of the scattering intensity that
was considered for isolated particles in “ Isolated Particles” section earlier in this article.
280 Neutron Scattering

2.13.6.5 Contrast Variation


The contrast variation technique is based on isotopic labeling of one or more components, which takes place most of the time with
a selective deuteration for biological systems. To highlight the particle contribution from the surrounding medium, the contrast is
given by the difference between the scattering densities of the two components, as expressed in Eq. (59). This case also can be
extended for multicomponent systems.
This procedure, although it requires accuracy in the selection and preparation of samples, allows one to change the behaviors of
the SANS-scattering profile and to obtain a more careful structural analysis.
Particularly in soft matter studies, the investigated sample contains a large quantity of hydrogen molecules, which can be
replaced with deuterium.

2.13.6.6 SANS Instruments


There are two typologies of instrument available for performing SANS measurements. The instruments are typically adapted to the
type of neutron production and the type of neutron source facility, described in “ Neutron Sources” section earlier in this article.
At the ILL reactor, the instruments work at fixed l and record the scattered intensity as a function of q. Due to their specific
geometrical characteristics, these instruments are generally placed in the end of a cold neutron beam guide. Other instruments,
mostly present at spallation sources, use the time-of-flight (TOF) mode, which uses a white polychromatic beam to vary l that is
working at a fixed q. An exception is the D33 instrument at ILL, which disposes of both types of technique.
As the SANS technique is one of the most popular, the user community can count on several SANS spectrometers being available
in almost all the neutron facilities.
There are four SANS instruments at ILL, three of them with similar geometry and with a classical SANS layout (D11, 47 D22,48
and D3349), and the D16 diffractometer,50 which has a different layout than the others. In the following section, a general layout of
these instruments will be described and some relevant scientific cases concerning nanoporous/mesoporous assemblies will be
reported.

2.13.6.6.1 Classical SANS SpectrometersdD11, D22, and D33


The classical layout of a SANS instrument at ILL is generally composed of a monochromator, a collimator, and a detector. A sketch of
the layout is shown in Fig. 9. A SANS instrument uses neutrons from a cold moderator and is situated at the end of the neutron
guide.
A mechanical velocity selector permits one to select the desired wavelength l, selecting the neutrons that have a given speed n
from the incident continuous beam coming out of the reactor, with Ol/l ¼ 10%.
The pinhole collimation system is made by circular diaphragms of strongly neutron-absorbing materials placed at the exit of the
guide and in front of the sample. It is possible to control the pinhole diameter and their distance (the collimation length), which is
typically of the order of 1–10 m.
The scattered neutrons are recorded by a two-dimensional (2D) position-sensitive gas detector with wire planes. The detector is
installed in a long tank under vacuum, and it can be moved into it to vary the sample-detector distance. Matching the sample-
detector distance together with the collimation length, we can select a desired Q-range. The length of the collimation can vary
from 1.5 m up to 40 m for D11 instrument, from 1.4 m up to 17 m on D22 and from 3 m up to 13 m on D33. An absorbing
beam stop is placed in the direct beam just before the detector to limit the number of neutrons from the direct beam.
Generally, these SANS spectrometers can cover approximately a Q range between 10 3 and 0.5 Å 1.
We invite the reader to refer to the literature47–49 for a more detailed description of the specificities for each instrument.
An example of SANS studies applied to protein solutions is shown in Fig. 10. SANS has been used to investigate the changes of
a model protein-cryoprotector solution composed by a mixture of lysozyme/sorbitol/water (D2O), under representative pharma-
ceutical processing conditions.51
The aim was to investigate the change in heterogeneity of the solutions, phase separation, protein interaction distance, hysteresis,
and possible irreversible changes in the scattering profiles. In Fig. 10, 2D-diffracted intensities following the cooling and heating
temperature curve are also demonstrated.

Figure 9 Schematic layout of a SANS spectrometer.


Neutron Scattering 281

Figure 10 (Left) 2D-diffracted intensities of low-concentrated water/sorbitol/lysozyme solution during cooling from 298 to 135 K and heating up to
298 K; (right) the corresponding SANS I(Q) versus Q curves at different temperatures. 51

The starlike pattern, visible at 274 K, and clearly evident at low temperatures could be attributed to the crystallization of the
freeze-concentrated solution and probably ice formation at lower temperatures.
As examples of mesoporous system investigations, two remarkable studies were done on a 2D-hexagonal silica nanostructure 52
and on mesoporous silica nanoparticle growth,53 both using time-resolved in situ SANS.
The first of these works (Ref. 52) focused on following the formation of the SBA-15 mesoporous material and the interaction
between the silica and organic micelles. SANS was used to provide direct experimental evidence for the three initial steps, from
shape transformation of the micelles from spherical to cylindrical to the precipitation of a 2D hexagonal phase.
In the second example (Ref. 53), SANS was used in the formation of small, mesoporous silica nanoparticles from the first stage
of the process. The ability of time-resolved SANS to resolve scattering contributions from both the nanoparticle and the micelle
populations gives important insight into these mechanisms and the factors governing their growth rates and final size.

2.13.6.6.2 D16 Two-Axis Cold Neutron Diffractometer


D16 is a highly versatile, cold-neutron, two-axis diffractometer at small momentum transfer Q. 50 It fills a gap in terms of Q-range
and resolution between classical SANS and an atomic resolution diffraction instrument that makes it a unique machine. Indeed,
D16 remains a unique instrument, not only with respect to other ILL instruments, but also worldwide, due to an excellent OQ
resolution. Fig. 11 illustrates a sketch of the D16 instrument, and Fig. 12 shows a photo of the instrument after the recent upgrade.
D16 applications range over a very large number of fields: biology (model and biological membrane multilayers), soft condensed
matter (colloids, liquid crystals, polymers, and surfactants), chemistry, and material science (clays, porous materials). The main
application of the instrument is the study of thin films in reflection geometry, as well as powders and liquids in diffraction mode.
The neutron beam is reflected either by a highly oriented, pyrolitic graphite monochromator (HOPG), with a mosaic of about
0.4 degrees, or by a rubidium, graphite-intercalated, crystal monochromator (Rb-GIC), with a mosaic of about 1.2 degrees, deliv-
ering 4.5 and 5.7 Å for the HOPG and 7.5 and 9.5 Å for the Rb-GIC as wavelengths with Ol/l ¼ 0.01 by two takeoff angles at 85
degrees and 115 degrees.
Both monochromators are composed of a series of crystals that are vertically disposeddnine for HOPG (which is also motorized
for a tunable focusing in the vertical direction, providing important flux gains) and three for the Rb-GIC. While the HOPG is regu-
1
larly used, offering a Q-range of 0:02  2:3 A with high neutron flux at the sample position and high resolution in the horizontal
plane, the Rb-GIC monochromator is an option that is more dedicated to low-resolution crystallography (structures of 20–200 Å),
which takes advantage in particular of the high detector resolution.
The vertical slit collimation geometry is one of the main instrument characteristics, obtained by means of two motorized pairs of
slits, one positioned after the monochromator and the other in the end of the vacuum tube, just before the sample position.
The Q-range is given by the combination of the selected l and the wide angular opening 2q of the detector around the sample
axis (which can be up to 120 degrees).
Another strength of the instrument lies in the high-resolution large area detector MILAND, which is unique. It is a high-pressure
3
He detector made of arrays of 320 cathodes  320 anodes, wired with 1-mm spacing and leading to a very high resolution of
1  1 mm2.
282 Neutron Scattering

Figure 11 Schematic representation of the D16 two-axis diffractometer at ILL. On the left is the sketch of the D16 layout, from the neutron guide
H52 to the detector position. The neutrons are reflected by a monochromator and collimated by two slits through a vacuum tube. The detector
moves around the sample covering an opening angle of 120 degrees. On the right is a 3D image of the D16 setup.

Figure 12 Photo of the D16 two-axis diffractometer at ILL, after the recent upgrade phase, finished in 2014.

All of the following features make D16 a very versatile instrument that can work in different geometry configurations: pinhole
SANS geometry, diffraction at wider angles, and a reflectivity mode for multilayered diffraction.
An example of SANS results obtained at D16 is shown in Fig. 13.
A study of wormlike micelle assembly in the 8-nm tubular nanopores of SBA-15 silica was carried out. 54 By adjusting the
surfactant-to-pore-wall interactions by coadsorption of a surface modifier (namely, the amino acid lysine), it is possible to observe
an evolution of equilibrium morphologies of the surfactant aggregates as a function of the level of surface modification q.
In Fig. 13, the 2D SANS scattering pattern and the respective intensity I(Q), taken at different degrees of surface functionalization
with lysine, are shown. In this study, the contrast-matching D2O/H2O technique was used to highlight the structure of the surfactant
aggregates in the pores.

2.13.7 Neutron Vibrational Spectroscopy

Vibrational spectroscopic techniques provide unique information about the high-energy atomic and molecular vibrations in
condensed matter systems. Among vibrational spectroscopies, inelastic neutron scattering (INS) presents very specific features
due to some characteristics of the neutron that have already been described in the introduction to this article. First, this technique
is mainly sensitive to vibrations involving hydrogen atoms because of their much higher incoherent cross section than with other
elements. As a consequence, INS is well suited for investigating hydrated compounds, as water vibrations can be observed on
Neutron Scattering 283

Figure 13 SANS 1D profiles I(Q) versus Q for SBA-15 samples at different degrees of surface modification with lysine showing three scenarios of
surfactant aggregation: (Left) Surface micelles adsorbed on the pore wall; (middle) rodlike micelles in the core of the pore; (right) wormlike micelles
in the solution outside the pores. The insets show the 2D scattering patterns and a 3D cartoon of the micelles. 54

a wide spectral range, even in zones where signals from the framework atoms would dominate the spectra in other vibrational
techniques. Second, neutrons probe the bulk properties of samples and can easily penetrate through container walls making
possible studies of unstable or conditioned materials. And third, neutrons can only properly be described as quantum mechan-
ical entities; thus, they exhibit both wavelike and particlelike properties, and for that reason, an inelastic scattering event results in
 1

a significant transfer of both energy (E; cm1 ) and momentum jQj A . The energy transfer (ET) is given by: ET ¼ Ei  Ef , where
the subscripts i and f refer to the incident and final values, respectively. The momentum transfer is given by: Q ¼ ki  kf, where
k ¼ 2p
l , k is the wave vector; and l is the wavelength of the neutron. INS spectroscopy spans a wide range of energies, from that of
the microwave region to the ultraviolet region of the electromagnetic spectrum. The information obtained by INS experiments is
complementary to that obtained by means of infrared (IR) and Raman spectroscopy, which yield information about lattice vibra-
tions, which in turn produce a variation of the dipole moment or the polarizability of the system, respectively. In addition in the
case of IR and Raman scattering processes, the momentum transfer is Q  0, and therefore both techniques are subjected to the
selection rule that only transitions at zero wave vector are observable. The INS method directly follows the motions of nuclei and,
since there are no specific selection rules, the INS spectra provide information about all vibrations of the system.
For the reasons explained previously, INS is a highly powerful technique for studying the lattice dynamics and atomic and
molecular vibrations in materials, especially those containing light elements as hydrogen. INS allows one to measure the general-
ized density of states (GDOS) of a powder sample, which is a particularly useful quantity for studying lattice dynamics. This is
particularly useful for a direct comparison with simulation data and is used as an important source of supplementary data for
computational techniques dealing with the dynamics of complex systems.
The observed quantity in a neutron experiment is proportional to the double differential cross section:
 2   2 
d2 s d s d s
¼ þ ; (68)
dUdE dUdE coh dUdE inc

which gives the number of neutrons scattered by the sample into a solid angle dU around U, with a final energy included between
E  dE and E þ dE. The total intensity and the relative proportion between coherent and incoherent depend on the specific chemical
composition of the sample.
In the INS, as the momentum transfer increases the strength of the elastic scattering decreases (with no energy exchange), and
most of the scattering takes place with an exchange of energy. For crystalline diffraction, the detected intensity has angular
variations; however, at high Q-values, these variations are no longer visible and the response is smooth and incoherentlike. The
dynamic response can then be calculated by treating the scattering as if it is incoherent, and this is called the incoherent approximation.
The form of the double differential scattering cross section can be rewritten in terms of a function S, which emphasizes the
dynamics of each individual atom in the sample. Conventionally called the scattering law, this is directly related to the observed
intensities when summed over all the atoms in the sample:
!   
4p ki d2 s
S Qu l ¼ ; (69)
sl kf dUdE l

where l refers to one specific atom of the molecule, ki and kf are the initial and final wave vectors, respectively, and the rest of the
variables are defined in the introduction of the chapter.
To the extent that kf is much smaller than ki, as for the indirect geometry spectrometer that will be described in this section, the
observed intensity is directly proportional to the GDOS, that is the hydrogen partial density of states in the case of hydrogenated
materials.
284 Neutron Scattering

The main significance of the scattering law is that it is the natural meeting point of experimental data and the dynamical
simulated models that have been developed for its understanding. The calculated relative intensity of the ith mode determined
!
at a momentum transfer Q and at an energy transfer ui, S, of an INS band is given by Ref. 55:
! ! 2n
!  Q $U i  2
S Q ; nui f e U
exp  Q$ ei s; (70)
n!
where ui is the ith mode at frequency u; n equals 1 for a fundamental, equals 2 for a first overtone or binary combination, equals 3
for a second overtone or ternary combination, etc.; Q is the momentum transfer defined here; Ui is the root mean square
displacement of the atoms in the mode; and s is the inelastic cross section of the atom. The exponential term in Eq. (70) is the
Debye–Waller factor, UTotal is the root mean square displacement of all atoms in all the modes, and its magnitude is determined in
part by the thermal motion of the molecule. Then, the intensities of the features in a neutron spectrum depend on the amplitude of
si
the vibrations of each atom involved in a particular mode weighted by their scattering power M , where si and Mi are the scattering
P si i

cross section and the mass of the atom i, respectively. The GDOS GðuÞ  Mi gi obtained from an INS experiment differs from the
i
real vibrational density of states (VDOS), which can be calculated by normal mode analysis and ab initio MD simulations.
The theoretical VDOS(u) can be calculated using the direct method on the basis of first-principles calculations. In that case,
P
VDOS can be calculated from the partial density of states, gi, of the different atomic species: VDOSðuÞ  gi
VDOS also can be obtained by means of ab initio MD simulations. In that case, the atoms are movedi according to classical
Newton’s mechanics, but the interatomic forces acting between the atoms are computed according to quantum mechanics, solving
from first principles the electronic structure for a given set of atomic positions, and then calculating the resulting forces on each
particle. In this case, the vibrational spectra can be extracted from MD trajectories via the time Fourier transform of the VACF velocity
autocorrelation function (VACF):
rffiffiffiffiffiffið N  
1 X ! v i ðt Þ!
v i ð0Þ
g ðuÞ ¼ eiut  !  dt; (71)
2p N i v i ð0Þ!v i ð0Þ
! ! 
v i ðtÞ v i ð0Þ
where the function ! !  is the VAF, calculated from a chosen initial time (t ¼ 0) for the MD trajectory and for each atom of
v i ð0Þ v i ð0Þ
the system i, along all three components of velocity ! iv . Density of states (DOS) was computed as the time-Fourier transform of the
velocity autocorrelation function (VACF) computed from the coordinates of each step of the trajectories.
Among the INS techniques, there are direct geometry chopper spectrometers (with fixed incident energy) and indirect geometry
spectrometers (with fixed final energy). A comparison of both kinds of spectrometer can be found in Ref. 56.
In the following section, we describe a type of indirect geometry spectrometer that combines the advantages given by the contin-
uous beam from a reactor neutron source and a large accepted solid angle for neutrons scattered by the sample. This new instrument
with the secondary spectrometer, named IN1-Lagrange, is installed at the hot neutron source of the high-flux ILL reactor.

2.13.7.1 IN1-Lagrange Design


The Lagrange project was launched with the goal to create a new spectrometer for studying lattice and molecular excitations in the
extended energy range of up to several hundred millielectron volts, which is typical for materials containing light chemical elements,
particularly hydrogen. Lagrange replaces the beryllium filter spectrometer placed on the hot neutron spectrometer IN1 at the ILL 57
and is dedicated to the study of molecular vibrations in complex systems, such as carbon nanotubes,58 zeolites,59 and hydrides.60
A number of neutron spectrometers suitable for VDOS measurement are available at existing neutron sources, both pulsed and
steady state. For the measurements of VDOS in a large range of energy transfers, typically indirect geometry TOF spectrometers are
exploited. Some examples of these kind of spectrometers are TOSCA 61 at ISIS (Oxford, Uinted Kingdom), which has become
a benchmark instrument in the matter, or NERA62 in the Joint Institute for Nuclear Research (JINR), in Dubna, Russia. Further
development of the TOSCA type of setup is represented by the VISION spectrometer63 at SNS at Oak Ridge, Tennessee, in the United
States. Instruments of this type use Bragg reflection to select a scattered neutron energy that is close to the wavelength cutoff edge of
a beryllium filter, which is used to suppress higher-order harmonics. They provide suitable energy resolution, which is monotonous
and slowly varied as a function of vibration frequency, in the whole range of measurements.
Lagrange (LArge GRaphite ANalyzer for Genuine Excitations) is placed on the hot neutron beamtube H8 at the ILL, and it shares
this position with the D4 diffractometer for liquids and amorphous systems and with the hot neutron, three-axis IN1 spectrometer.
Since Lagrange shares the primary spectrometer with these instruments, it benefits from the recently upgraded double-focusing
monochromator with a set of different reflecting planes:
l Cu(220) reflection, with an starting incident energy of < 27 meV ( 216 cm 1) of up to  500 meV ð 4000 cm1 Þ, used for the
high-flux configuration
l Cu(331) reflection, with an starting incident energy of  65 meV ð 520 cm1 Þ of up to about 1 V, used for the improved energy
resolution configuration at high energies
Neutron Scattering 285

Figure 14 (A) Lagrange design. (B) Trajectory through (Q, u) space of Lagrange. The inset shows the triangle scattering for the case of Lagrange,
where ki [kf and therefore Q z ki.

l Si(311) reflection, with an starting incident energy of  16:5 meV ð 132 cm1 Þ of up to  200 meV ð 1600 cm1 Þ; and the
Si(111) reflection, with an starting incident energy ranging from  5:5 meV ð 44 cm1 Þ up to  20 meV ð 160 cm1 Þ used for
the low-energy part of the spectra and that present the advantage of suppressed second-order contamination in the
monochromatic beam
With the reflections given here, this monochromator currently supplies more than twice the increased monochromatic flux in the
high neutron energy range available from the hot neutron source of the ILL reactor.
The Lagrange secondary spectrometer setup ( Fig. 14A) is based on the space focusing of neutrons scattered by the sample in
a very large solid angle, all of them registered with a relatively small single counter (a He3 gas detector). The focusing reflecting
surface of 1 m2 is built around the vertical sample detector axis from pyrolytic graphite crystals set to reflect neutrons with the fixed
average energy of 4.5 meV. The appropriately shaped cooled beryllium filter is installed right after the sample in order to remove
higher-order harmonics in the analyzer reflections. It is important that the absolute value of the accepted solid angle (about 2.5
steradians) stays among the highest of the most ambitious instrument projects at present. At the same time, it is realized in the rela-
tively small spectrometer volume, which is characteristic. The high-energy resolution of the new instrument is attained with a rela-
tively low grade of pyrolytic graphite. The carefully designed screen of boron-containing absorber is installed on the sample-detector
axis in order to suppress the intense elastic scattering from the sample. Further reduction of the instrument background, contam-
inated with the high-neutron-energy components, is achieved through a massive polyethylene shielding built around the whole
analyzer. Both the sample cryostat and the filter-cooling circuits are equipped with powerful closed-cycle helium refrigerators.
Note that there are no mutually moving parts within the secondary spectrometer, which is positioned as a whole around the mono-
chromator in order to record the inelastic scattering spectra, changing step by step the incident energy in a similar way as a typical
experiment on a three-axis spectrometer.
The measured line-widths with the 2.5-diiodothiphene sample are presented as points in Fig. 15, which also shows calculated
energy resolution curves with different analyzer and monochromator settings. The best expected energy resolution, of the order of
1% in the most informative energy range 100–200 meV, should be achieved with 120 and 100 Soller collimators that will be installed
before and after the monochromator in 2017. This ultimately high energy resolution will make Lagrange one of the top high-
resolution neutron spectrometers and will be available with the collimator changer that will be installed before the monochro-
mator. The changing over different resolution settings can be done very easily, leaving users the flexibility to optimize their measure-
ments with respect to resolution versus intensity choice.
It must be outlined here that with IN1-Lagrange, as opposed to the TOF spectrometers, one can concentrate on a selected energy
interval when detailed knowledge of the entire spectrum is not required. Thus, the available beam time can be used for improving
statistics in the most important spectral region of the measured GDOS or for detailed records of the effect of the sample environ-
ment conditions (variation of temperature, pressure, etc.). The price that is paid for the good resolution and simplicity of operation
of the low final energy instruments (such as Lagrange) is that they have a fixed trajectory in the (Q, u) space, as is shown in Fig. 14B.
Therefore, high energies always correspond to high momentum transfer values, and this will have two main consequences: (1) the
measured spectra will present the overtones described in Equation 70 and (2) the Debye–Waller factor will decrease the intensity at
high-energy transfer, even when the experiment is performed at a low temperature.
As an example of the information that can be obtained by means of INS, results obtained on a porous media are presented in the
following section.
286 Neutron Scattering

Figure 15 Calculated (lines) and measured (points) relative energy resolution of Lagrange with different primary spectrometer settings.
The black line corresponds to the Be-filter analyzer resolution with the Cu220 monochromator; the blue line shows the same in the case of
Lagrange; and the red line shows the Lagrange with the Cu331 monochromator. These calculations are made for the natural collimation of
the in-pile beam tube of IN1 and 20 Soller collimators in the monochromatic beam. Points are the peak widths divided by the peak energies,
as measured in the corresponding conditions with the 2.5-diiodothiophene sample. The broken line represents the best-expected future energy
resolution of Lagrange.

2.13.7.2 INS Example: Vibrational Study of an Oriented Clay


Swelling clay minerals are formed with two tetrahedral sheets (silica) sandwiching an octahedral sheet forming the 2:1 layer. Due to
isomorphic substitutions by less charged cations in either the tetrahedral or octahedral sheet, the solid layer bears a net negative
charge that is compensated for by interlayer cations that can control the hydration properties. This in turn is compensated for
by interlayer cations whose valence and hydration properties control both swelling and colloidal behavior. The vibrational prop-
erties of the dry, low-charge saponite were measured by INS along the direction parallel and perpendicular to the clay layers and
confronted with the VDOS obtained by normal mode analysis. 64 This analysis allowed in order to interpret the vibrational bands
of the dry saponite.
Fig. 16A shows the GDOS obtained directly from the neutron vibrational spectra of the dry, low-charge clay in two different
directions: parallel and perpendicular to the clay layer. The first characteristic that should be pointed out is the clear anisotropy
observed on the vibrational spectra, especially on the intermolecular, low-energy bands appearing below 400 cm 1 and the OH
stretching band of the structural hydroxyls that appears at 3700 cm 1. The presence of the stretching band in the perpendicular
direction, as well as the complete absence of it in the parallel one, confirm the 2D nature of our system.
The normal mode analysis was performed by means of Density Functional Theory calculation using Perdew-Burke-Ernzerhof
potential. Fig. 16B shows the calculated INS spectra that is decomposed in the different contributions, the fundamental vibrational
modes (n ¼ 1 in Eq. 70) and the two first overtones (n ¼ 2 and n ¼ 3, respectively). Taking into account the overtones on the
calculated INS spectra, it can be clearly seen that the vibrational spectra of the dry clay can be fully described:
l The low-energy bands (up to 500 cm1 ) correspond to the bending of the tetrahedral layer (Si–O–Si and Si–O–Al).
l The Mg3OH bending vibration observed at  650 cm1 . This mode has very low intensity on the IR spectra, but since this mode
involves a movement of very high amplitude of the hydrogens of the hydroxyl groups, this mode is well observed with INS
because the presence of H.
l The Si–O stretching is observed at  1000 cm1 at very low intensity for two reasons. First, the H atoms are not involved in the
movements; and second, the mode is probably masked with the signal of the overtones of the low-energy bands.
l The OH stretching region of saponite reflects its trioctahedral character with mainly Mg3OH units in the octahedral sheet,
absorbing near  3700 cm1 . This band is almost negligible for two reasons. First, at high energies, the mode has very low
intensity due to the Debye–Waller factor; and second, the calculated spectra shows an average in all directions, and the stretching
band shows a high anisotropy and only appears in the perpendicular direction. However, the experimental spectra has two
different components, parallel and perpendicular to the clay layers, and the OH stretching band is observed only in the
perpendicular direction, confirming the 2D character of the system.
The comparison of the experimental INS spectra with the calculated ones allows us to interpret the vibrational bands of the dry
saponite and validate the model used for the calculations. Since the interaction of the neutron is nuclear, we have been able to
measure all the vibrational bands, even the ones that are very weak on IR scattering.
Neutron Scattering 287

Figure 16 (A) Experimental results for low-charged dry clays for two orientations of the clays, parallel and perpendicular, with respect to the
neutron incidence and (B) INS spectra calculated using DMol3 in Materials Studio software 65 and generated using the aCLIMAX software.66

2.13.8 Conclusions

Neutron scattering is an extremely powerful technique in the study of structure properties from the atomic to nanometric scale, as
well as atomic and molecular vibrations. In this article, we have given a general overview on the basis theory of the neutron scat-
tering, and then focus on three neutron-scattering techniques, diffraction, SANS, and INS. Due to the wide varieties of application in
different scientific areas of the condensed matter in which the neutrons are involved, it was not possible to fully cover all the subjects
here. The choices of the neutron techniques and the examples of applications reported in this article have been related to the topics
that concern previous studies of mesoporous and molecular self-assembly systems; and all of this can be placed in relation with the
general purpose of the book. It is our earnest hope that this review will be useful as a base introduction to neutron scattering.

Acknowledgments

The authors are grateful to Isabelle Grillo for helpful comments and suggestions. The research leading to these results has received funding from the
People Programme (Marie Curie Actions) of the European Union’s Seventh Framework Programme FP7/2007-2013/under REA grant agreement
nPIRSES-GA-2012-319011.

References
1. Stirling, W. G.; Vettier, C. Neutron News 2010, 21 (1), 13–17. http://dx.doi.org/10.1080/10448630903409251.
2. Vogel, S. C. ISRN Mater. Sci. 2013, 2013, 302408. http://dx.doi.org/10.1155/2013/302408.
3. Price, D. L.; Sköld, K. Introduction to the neutron scattering in Methods in Experimental Physics. In Part A Neutron Scattering, Vol. 23, Sköld, K., Price, D. L., Eds.; Academic
Press: Orlando, 1986; pp 1–97.
4. Schober, H. J. Neutron Res. 2014, 17, 109–357. http://dx.doi.org/10.3233/JNR-140016.
5. Ballou, R.; Berk, N.; Brown, P.; Chatterji, T.; Ehlers, G.; Majkrzak, C.; Mezei, F.; O’Donovan, K.; Ouladdiaf, B.; Pappas, C.; Regnault, L.; Schweizer, J.; Wiedenmann, A. Neutron
Scattering From Magnetic Materials. In Chatterji, T., Ed.; Elsevier Science: Amsterdam, 2005.
6. Squires, G. L. Introduction to the Theory of Thermal Neutron Scattering, Reprint edition;; Dover Publications: Mineola, NY, 1978.
288 Neutron Scattering

7. Dobrzynski, L.; Blinowski, K. Neutrons and Solid State Physics in Ellis Horwood Series: Physics. In Cooper, M., Ed.; Ellis Horwood, 1994.
8. Dianoux, A. J.; Lander, G. Neutron Data Booklet; Institut Laue-Langevin: Grenoble, France, 2002.
9. Vogel, S. C.; Priesmeyer, H. G. Rev. Mineral. Geochem. 2006, 63, 27–57. http://dx.doi.org/10.2138/rmg.2006.63.2.
10. Vogel, S. C.; Carpenter, J. S. JOM J. Miner. Met. Mater. Soc. 2012, 64, 104–111. http://dx.doi.org/10.1007/s11837-011-0220-1.
11. Visit the home page at www.ill.fr.
12. Visit the home page at www.isis.rl.ac.uk.
13. http://www.neutron.anl.gov/facilities.html.
14. Messiah, A. Quantum Mechanics; Dover Publications: New York, 1999.
15. Mughabghab, M. D. S. F.; Holden, N. E. Neutron Cross Sections; Academic Press: New York, 1981 (Chapter 1).
16. Sears, V. F. Neutron News 1992, 3 (3), 29–37.
17. https://www.ncnr.nist.gov/resources/n-lengths/.
18. Squires, G. L. Introduction to the Theory of Thermal Neutron Scattering; Cambridge University Press: Cambridge, 1978.
19. Rietveld, H. M. J. Appl. Crystallogr. 1969, 2, 65–71. http://dx.doi.org/10.1107/S0021889869006558.
20. Rodríguez-Carvajal, J. Physica B 1993, 192 (1–2), 55–69. http://dx.doi.org/10.1016/0921-4526(93)90108-I.
21. Lelièvre-Berna, E.; Bourgeat-Lami, E.; Gilbert, Y.; Kernavanois, N.; Locatelli, J.; Mary, T.; Pastorello, G.; Petukhov, A.; Pujol, S.; Rouques, R.; Thomas, F.; Thomas, M.;
Tasset, F. Physica B 2005, 356 (1–4), 141–145. http://dx.doi.org/10.1016/j.physb. 2004.10.065.
22. Stunault, A.; Vial, S.; Pusztai, L.; Cuello, G. J.; Temleitner, L. J. Phys. Conf. Ser. 2016, 711, 012003.
23. Orench, I. P.; Clergeau, J.-F.; Martínez, S.; Olmos, M.; Fabelo, O.; Campo, J. J. J. Phys. Conf. Ser. 2014, 549 (1), 012003. http://dx.doi.org/10.1088/1742-6596/549/1/
012003.
24. Suard, E.; Hewat, A. W. Neutron News 2001, 12 (4), 30–33. http://dx.doi.org/10.1080/10448630108245006.
25. Hewat, A. W. Physica B 2006, 385–386, 979–984. http://dx.doi.org/10.1016/j.physb.2006.05.316.
26. Hansen, T. C.; Henry, P. F.; Fischer, H. E.; Torregrossa, J.; Convert, P. Meas. Sci. Technol. 2008, 19 (3), 034001. http://dx.doi.org/10.1088/0957-0233/19/3/034001.
27. Cuello, G. J. J. Phys. Condens. Matter 2008, 20 (24), 244109. http://dx.doi.org/10.1088/0953-8984/20/24/244109.
28. Bytchkov, A.; Cuello, G. J.; Kohara, S.; Benmore, C. J.; Price, D. L.; Bychkov, E. Phys. Chem. Chem. Phys. 2013, 15 (22), 8487–8494. http://dx.doi.org/10.1039/
C3CP50536G.
29. Sobolev, O.; Cuello, G. J.; Román-Ross, G.; Skipper, N. T.; Charlet, L. J. Phys. Chem. A 2007, 111 (24), 5123–5125. http://dx.doi.org/10.1021/jp072650w.
30. Pitteloud, C.; Powell, D. H.; González, M.Á.; Cuello, G. J. Colloids Surf. A Physicochem. Eng. Asp. 2003, 217 (1–3), 129–136. http://dx.doi.org/10.1016/s0927-7757(02)
00567-8.
31. Sobolev, O.; Charlet, L.; Cuello, G. J.; Gehin, A.; Brendle, J.; Geoffroy, N. J. Phys. Condens. Matter 2008, 20 (10), 104207. http://dx.doi.org/10.1088/0953-8984/20/10/
104207.
32. Sobolev, O.; Charlet, L.; Cuello, G. J.; Gehin, A.; Brendle, J. Radiochim. Acta 2008, 96 (9–11), 679–683. http://dx.doi.org/10.1524/ract.2008.1553.
33. Sobolev, O.; Cuello, G. J.; Scheinost, A. C.; Johnson, M. R.; Nikitenko, S.; Le Forestier, L.; Brendle, J.; Charlet, L. Phys. Status Solidi A 2011, 208 (10), 2293–2298. http://
dx.doi.org/10.1002/pssa.201000311.
34. Sobolev, O.; Buivin, F. F.; Kemner, E.; Russina, M.; Beuneu, B.; Cuello, G. J.; Charlet, L. Chem. Phys. 2010, 374 (1–3), 55–61. http://dx.doi.org/10.1016/j.chemphys.
2010.06.012.
35. Fischer, H. E.; Cuello, G. J.; Palleau, P.; Feltin, D.; Barnes, A. C.; Badyal, Y. S.; Simonson, J. M. Appl. Phys. A 2002, 74 (Suppl 1), S160–S162. http://dx.doi.org/10.1007/
s003390101087.
36. Ambroise, J.-P.; Bellissent-Funel, M. C.; Bellissent, R. Rev. Phys. Appl. 1984, 19 (9), 731–734. http://dx.doi.org/10.1051/rphysap:01984001909073100.
37. Cuello, G. J.; Darpentigny, J.; Hennet, L.; Cormier, L.; Dupont, J.; Beuneu, B. 7C2, the new neutron diffractometer for liquids and disordered materials at LLB. J. Phys. Conf.
Ser. 2016, 746, 012020.
38. Soper, A. K. In Proceedings of the Conference on Advanced Neutron Sources, Vol. 97 of IOP Conference Proceedings; Hyer, D. K., Ed.; Institute of Physics and Physical Society,
Bristol, 1988; p 353.
39. Bowron, D. T.; Soper, A. K.; Jones, K.; Ansell, S.; Birch, S.; Norris, J.; Perrott, L.; Riedel, D.; Rhodes, N. J.; Wakefield, S. R.; Botti, A.; Ricci, M.-A.; Grazzi, F.; Zoppi, M. Rev.
Sci. Instrum. 2010, 81, 033905. http://dx.doi.org/10.1063/1.3331655.
40. Hannon, A. C. Nucl. Inst. Methods Phys. Res. A 2005, 551, 88–107. http://dx.doi.org/10.1016/j.nima.2005.07.053.
41. Neuefeind, J.; Feygenson, M.; Carruth, J.; Hoffmann, R.; Chipley, K. K. Nucl. Inst. Methods Phys. Res. B 2012, 287, 68–75. http://dx.doi.org/10.1016/j.nimb.2012.05.037.
42. Guinier, A.; Fournet, G. A Small Angle Scattering of X-rays; Wiley: New York; Chapman and Hall: London, 1955.
43. Zemb, T., Lindner, P., Eds. Neutrons, X-rays and Light: Scattering Methods Applied to Soft Condensed Matter; The Netherlands, North Holland delta series: Amsterdam, 1991.
44. Hammouda, B. Probing Nanoscale StructuresdThe Sans Toolbox. https://www.ncnr.nist.gov/staff/hammouda/the_SANS_toolbox.pdf, 2015.
45. Grillo, I. In Soft Matter Characterization; Pecora, R., Borsali, R., Eds.; Springer Netherlands: Dordrecht, 2008; pp 723–782. http://dx.doi.org/10.1007/978-1-4020-
4465-6_13.
46. Cousin, F. EPJ Web Conf. 2015, 104, 01004. http://dx.doi.org/10.1051/epjconf/201510401004.
47. Lindner, P. A. P.; May, R. P. Physica B 1992, 180, 967–972. http://dx.doi.org/10.1016/0921-4526(92)90524-V.
48. http://www.ill.eu/instruments-support/instruments-groups/instruments/d22.
49. Dewhurst, C. D.; Grillo, I.; Honecker, D.; Bonnaud, M.; Jacques, M.; Amrouni, C.; Perillo-Marcone, A.; Manzin, R. Cubitt. J. Appl. Crystallogr. 2016, 49, 034007.
50. Giroud, V. C. B.; Didier, L.; Demé, B. Neutron News 2015, 26 (3), 22–24.
51. Khodadadi, S.; Clark, N. J.; McAuley, A.; Cristiglio, V.; Curtis, J. E.; Shalaev, E. Y.; Krueger, S. Soft Matter 2014, 10, 4056–4060. http://dx.doi.org/10.1039/C4SM00600C.
52. Impéror-Clerc, M.; Grillo, I.; Khodakov, A. Y.; Durand, D.; Zholobenko, V. L. Chem. Commun. 2007, 28 (8), 834836. http://dx.doi.org/10.1039/B611208K.
53. Hollamby, M. J.; Borisova, D.; Brown, P.; Eastoe, J.; Grillo, I.; Shchukin, D. Langmuir 2012, 28, 4425–4433. http://dx.doi.org/10.1021/la203097x.
54. Bharti, J. M. B.; Xue, M.; Cristiglio, V.; Findenegg, G. H. J. Am. Chem. Soc. 2012, 134 (36), 1475614759. http://dx.doi.org/10.1021/ja307534y.
55. Mitchell, P. C. H.; Parker, S. F.; Ramirez-Cuesta, A. J.; Tomkinson, J. Vibrational Spectroscopy With Neutrons, With Applications in Chemistry, Biology, Materials Science and
Catalysis; World Scientific: Singapore, 2005.
56. Parker, S. F.; Lennon, D.; Albers, P. Appl. Spectrosc. 2011, 65 (12), 1325. http://dx.doi.org/10.1366/11-06456.
57. http://www.ill.eu/instruments-support/instruments-groups/instruments/in1-taslagrange/characteristics/.
58. Rols, S.; Benes, Z.; Angleret, E.; Sauvajol, J.; Papanek, P.; Fischer, J.; Coddens, G.; Schober, H.; Dianoux, A. Phys. Rev. Lett. 2000, 85, 5222. http://dx.doi.org/10.1103/
PhysRevLett.85.5222.
59. Jobic, H. Curr. Opinion Solid State Mater. Sci. 2002, 6, 415–422.
60. Kolesnikov, A.; Antonov, V.; Fedotov, V.; Grosse, G.; Ivanov, A.; Wagner, F. Physica B 2002, 316, 158–161.
61. Parker, S. F. J. Neutron Res. 2002, 10, 173. http://dx.doi.org/10.1080/1023816021000020626.
62. http://flnp.jinr.ru/134/.
63. http://neutrons.ornl.gov/vision/.
64. Jiménez-Ruiz, M.; Ferrage, E.; Delville, A.; Michot, L. Submitted.
65. http://accelrys.com/products/materials-studio.
66. Ramirez-Cuesta, A. Comput. Phys. Commun. 2004, 157, 226–238. http://dx.doi.org/10.1016/S0010-4655(03)00520-4.
2.14 Small-Angle Neutron Scattering in Supramolecular Complexes
H Kumari, University of Cincinnati, Cincinnati, OH, United States
Ó 2017 Elsevier Ltd. All rights reserved.

2.14.1 Introduction 289


2.14.2 Introduction to SANS 289
2.14.2.1 Principles of Neutron Scattering 290
2.14.2.2 Differential Cross Section and Macroscopic Cross Section 290
2.14.2.3 Measurements 291
2.14.3 Analysis of SANS Data 291
2.14.3.1 Schulz Sphere Model 292
2.14.3.2 Core–Shell Sphere 293
2.14.3.2.1 Contrast Variation Method 294
2.14.3.3 Uniform Ellipsoid 294
2.14.3.4 Bimodal Schulz Sphere Model 295
2.14.3.5 Triaxial Ellipsoid 296
2.14.3.6 Cylinder Model 297
2.14.3.7 Core–Shell Cylinder 298
2.14.3.8 Hollow Cylinder 299
2.14.3.9 Fractal and Fractal Flexible Cylinder Models 300
2.14.4 Conclusion 301
References 301

2.14.1 Introduction

A variety of analytic methods are utilized to elucidate the structure and dynamics of supramolecular nanoassemblies. Single-crystal
X-ray diffraction (SCXRD), powder X-ray diffraction (PXRD), mass spectroscopy, and solid-state nuclear magnetic resonance (NMR)
spectroscopy are some of the traditional techniques that help elucidate the molecular structure, architecture, and intermolecular
interactions of supramolecular ensembles in solid state. The large supramolecular architectures often have solvent disorder within
their internal cavities, which makes them somewhat complex to solve; however, crystal engineering serves as an excellent tool to
determine and manipulate solid-state structures.
Solution chemistry of supramolecular complexes has been studied through several analytic techniques;1–5 however, NMR
spectroscopy has been one of the primary techniques since the last few decades.4,5 Chemical shifts, relaxation times, diffusion
coefficients/DOSY, spin–spin coupling, nuclear Overhauser effect, and correlation thereof are some of the conventional parameters
obtained from NMR spectroscopy.6 Diffusion coefficients, in particular, have been very instrumental in studying discrete nanoas-
semblies in solution.7 The discovery of high-resolution gradient-enhanced spectroscopy in the 1990s led to an improvement in
gradient performance for high-resolution NMR spectrometers that provided simultaneous diffusion coefficients for the entire
sets of peaks in an NMR spectrum.8,9 This technique enabled studying equilibrium conditions, diffusion of guest molecules, and
self-sorting mechanisms of hydrogen-bonded and nonparamagnetic nanoassemblies in solution. A major limitation of NMR tech-
nique has been the investigation of paramagnetic species and the presumption about the spherical architectures of nanoassemblies
in solution. The diffusion coefficient obtained through DOSY experiments is calculated using Stokes–Einstein equation wherein
diffusion of hypothetical spheres in solution is calculated.
The self-assembly of supramolecular nanoassemblies is directed primarily via noncovalent interactions that are often reversible.
The concept of intactness of solid-state geometries in solution is often questionable when we elucidate their solution properties.
Small-angle neutron scattering (SANS) is a new tool that has proved beneficial in addressing some of the limitations of NMR spec-
troscopy. Not only does it provide architectural (shape and size) information, but also it is suitable for analyzing both diamagnetic
and paramagnetic nanoassemblies in solution: superseding NMR limitations. This article focuses on discussing basics of SANS and
data analyses, which include varying types of mathematical models utilized to elucidate supramolecular nanoassemblies in solution
and soft matter (gel and polymers).

2.14.2 Introduction to SANS

Nuclear scattering is the interaction of an atomic nucleus with neutrons, whereas magnetic scattering involves the interaction of
unpaired electrons of an atom with a neutron. SANS is a technique that can be utilized for investigating geometric dimensions

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12491-6 289


290 Small-Angle Neutron Scattering in Supramolecular Complexes

of a species (nonmagnetic scattering) and exploring magnetic behavior in solution with polarized neutrons (magnetic scattering).
Our discussion here is focused on nonmagnetic neutron scattering.10,11

2.14.2.1 Principles of Neutron Scattering


SANS is based on a model of elastic scattering in which macroscopic entities are considered as a single nucleus that is assumed to
stay stationary at the origin of coordinates with no momentum exchange. The scattering depends on the interaction potential
between the neutron and the nucleus. This interaction potential is short-range and falls to zero at distances smaller than
10  10 15 m, which is shorter than the wavelength of the neutron (10  10 10 m), and thus, the nucleus acts as a point scatterer.
Thus, we can represent the beam of incident neutrons as
Ji ¼ eikz

where z is the distance from the nucleus in the propagation direction and k ¼ 2p/l is the wave number. The outgoing spherically
scattered wave is given by the wave function
b
Js ¼  eikr
r

where b is the nuclear scattering length and represents the interaction of the neutrons with the nucleus. The scattering lengths of the
nuclei vary across the periodic table. For example, scattering length for the hydrogen is  3.74  10 5 Å, whereas that for the
deuterium is 6.67  10 5 Å. In principle, replacement of hydrogen with deuterium is a useful way of changing the materials’
scattering length and enabling contrast variation to solve complex structures. For a 3-D assembly, the scattering wave vector would
be given as
X bi 
Js ¼  eikr eiqr
i
r

where the scattering vector q is equal to ki–ks, with ki and ks being the incoming and scattered neutron wave vectors.10,11

2.14.2.2 Differential Cross Section and Macroscopic Cross Section


The length scales we measure using neutrons are much larger than atomic dimensions, and thus, it is easier to think about macro-
scopic properties of the materials. To quantify a material’s scattering properties, we define a quantity called the scattering length
density given by the formulas
rðr Þ ¼ bi dðr  ri Þ
Pn
bi
r¼ i
V

where r is the scattering length density, bi the scattering length of atom i, and V is the volume containing n atoms. Scattering length
density is a valid description for a molecule and is a constant for a given sample:
 2
ds 1 XN 
iq,rq 
ðqÞ ¼  bi e 
dU N  i 

ð 2
dS N ds 1  
¼ ðqÞ ¼  rðr Þeiq,r dr 
dU V dU V V

Scattering can be represented by the Rayleigh–Gans equation given earlier, which states that small-angle scattering arises as
a result of inhomogeneities in scattering length densities of the solute and the solvent. The differential cross section, ds/dU, is
the directly measured quantity in a scattering experiment and is defined as the number of neutrons scattered per second into the
given solid angle dU in the given direction. Integrating the scattering length density distribution across the whole sample and
normalizing by the sample volume gives the macroscopic differential cross section, dS=dU. The integral term in the Fourier trans-
form of the scattering length density distribution and the differential cross section is proportional to the square of its amplitude, and
thus, all phase information is lost, and we cannot simply perform the inverse Fourier transform to get from the macroscopic cross
section back to the scattering length density distribution. The normalization of the differential cross section on an absolute scale
permits direct comparison of scattering from different samples.10,11
The macroscopic cross section has three components, those corresponding to coherent, incoherent, and absorption processes:
dS dS dSinc dS
ðqÞ ¼ coh ðqÞ þ þ abs
dU dU dU dU

Herein, the incoherent component is not q-dependent (q again is the scattering vector or difference between the incident and
scattered wave vectors) and contributes to the noise. Each atom has a coherent scattering length and an incoherent scattering length,
so the actual atomic composition accounts for the noise and incoherent scattering for a given system. On the other hand, the absorp-
tion component is usually small owing to the high penetration and low energy of cold neutrons and thus only slightly reduces the
Small-Angle Neutron Scattering in Supramolecular Complexes 291

Figure 1 Schematic of the contrast variation on a core–shell particulate system showing left, natural contrast (core, shell, and solvent); middle,
core of particulate (match SLD solvent ¼ SLD shell); and right, shell of particulate (match SLD solvent ¼ SLD core).

overall signal. It is the coherent scattering that represents scattering produced by the interference of the scattered waves and provides
the structural information of the sample.10,11
It is important when designing small-angle experiments to consider appropriate use of “contrast variation” to be able to solve struc-
tures. Contrast variation is usually achieved by substitution of hydrogen with deuterium since deuterium has a large coherent scattering
length density/SLD and much smaller incoherent scattering length than hydrogen. Additionally, in a contrast variation experiment, the
solvent and solute scattering length densities (SLDs) are varied to specifically match species in solution. Contrast variation is important
in designing experiments since it provides a difference in scattering length densities of the solvent and solute (Fig. 1).10,11
Employing the Rayleigh–Gans equation in such two-phase system at nonzero q values gives the macroscopic cross section, which
has two components. The first component is the square of difference in scattering length densities that has the information about
the material properties (density composition) and radiation properties (scattering lengths). While the second component is the
integral term that describes the spatial arrangement of the material.10,11

2.14.2.3 Measurements
SANS measurements are usually conducted on various neutron source beamlines 30 m SANS instruments at the NIST Center for
Neutron Research (NIST-NCNR) in Gaithersburg, MD, the United States,12 or at Oak Ridge National Laboratory, Knoxville,
Tennessee, the United States. For measurements, neutrons of specific wavelength (e.g., l ¼ 6 Å with full-width half-maximum
Dl/l ¼ 15%) are used and varying samples to detector distances are used to cover the overall q range. For example, samples to
detector distance of 1.3, 4.5, 6, and 13 m are used to cover the overall q range of 0.012 Å 1 < q < 0.152 Å 1 at NIST, where
q ¼ (4p/l) sin(q/2) is the magnitude of the scattering vector and q is the scattering angle. The sample to detector distances can
be varied, and choice of detector distances is dependent on the size of nanometric species in solution. For larger species, measure-
ments at longer sample–detector distances are performed to achieve scattering at much smaller angles.10–12
The measured raw data of the samples are corrected for background and empty cell scattering, followed by calibration with
detector-normalized detector sensitivity. The corrected data sets are then placed on an absolute scale using the direct beam flux
method, and we take the average over the annulus of 2-D data. Each annulus of 2-D data set corresponds to one data point in
reduced 1-D SANS data. The data are thus reduced and modeled employing q-resolution function and smearing effect for structure
solution using Igor Pro software provided by NIST. More recently, new SAS View software is available for SANS data reduction and
analyses. The SANS measurements are typically carried out at the temperature of 25 C using cells of 2 mm path length.10,11
However, temperature, shear, and measurement conditions can be varied by adding a temperature controller equipment or a rheom-
eter to SANS beamline, based on the type of experiment and sample employed.
SANS probes structure on a scale d, where
l ðwavelengthÞ

q ðscattering angleÞ

0.5 nm < l < 2 nm (cold neutrons)


0.1 < q < 10 (small angles)
1 nm < d < 300 nm

2.14.3 Analysis of SANS Data

Once the data reduction and normalization are performed, the scattering is present in the form of a macroscopic cross section that
requires model-dependent or model-independent analysis to extract useful information. The model-dependent analysis requires
building a mathematical model of the scattering length density distribution, while the model-independent analysis consists of direct
manipulations of the scattering data to yield useful information. A few examples of model-independent analysis include the scattering
invariant, Porod scattering, and Guinier analysis that are categorized based on high- or low-q limits of scattering. On the other hand, in
model-dependent analysis, the macroscopic cross section is divided into a contrast factor and an integral term. The contrast factor
describes the difference in scattering length densities between the phases, while the integral term contains information about the
292 Small-Angle Neutron Scattering in Supramolecular Complexes

spatial arrangement of material in the phases. This integral term is the function that is modeled. Additionally, the distribution of mate-
rial could be described in terms of the form factor, P(q), which represents the interference of neutron scattered from different parts of
the same object, and a structure factor, S(q), which represents the interference of neutrons scattered from different objects:10,11
dS N
ðqÞ ¼ ðr  r2 Þ2 Vp2 P ðqÞSðqÞ
dU V 1
For analysis of complex structures like metal-seamed organic nanocapsules (MONCs), we use contrast variation technique to
vary the scattering length densities through hydrogen–deuterium exchange. The hydrogen–deuterium exchange acts as a key advan-
tage of neutron scattering over other techniques.
Complex structures could have a distribution of sizes that could either dampen the high-q oscillations or cause smearing of the
scattering curve. The smearing effect is calculated by an integral over the appropriate size distribution that can make already existing
integral model computationally more intense, particularly if polydispersity of multiple dimensions are required and the particle is
anisotropically shaped. Likewise, the instrument resolution function, which depends on the geometry and the wavelength distribu-
tion, has similar effect on the scattering curve, and thus, correction of smearing is done before extracting the size distribution from
the fit.10,11
Herein, we would discuss some of the model-dependent analyses methods and supramolecular assemblies that have fitted well
to these models.

2.14.3.1 Schulz Sphere Model


Schulz sphere model calculates the scattering for a polydisperse population of spherical entities with uniform scattering length
density.13–15 The Schulz distribution obtained is a distribution of radii. The scattering intensity obtained for spherical entities is
normalized by the average particle volume. One can calculate the number density of particles based on the volume fraction and
polydisperse particle volume and can plot normalized probability distribution as a function of radius. The input variables in the
model include volume fraction, mean radius, polydispersity, SLD sphere, SLD solvent, and background. The returned value is scaled
to the units of cm 1 sr 1. The normalized Schulz distribution is given as
exp½ðz þ 1Þx
f ðRÞ ¼ ðZ þ 1Þzþ1 xz
Ravg Gðz þ qÞ

For other mathematical equations, refer to the model details on Igor Pro Software, provided by NIST.
Kumari et al. has studied several supramolecular C-alkylpyrogallol[4]arene-based nanoassemblies in solution, which were fitted
to Schulz sphere model. Pyrogallol[4]arenes are cyclized oligomers of 1,2,3-trihydroxybenzene. These macrocycles self-assemble
into nanocapsules, nanotubes, and bilayer-type motifs in solution. The copper-seamed C-alkylpyrogallol[4]arene/PgCnCu
(n ¼ alkyl tail length) hexameric nanocapsules consist of six pyrogallol[4]arene units and 24 metal centers (Fig. 2, right).16 A
hydrogen-bonded hexameric PgCn nanocapsule is constructed from 72 hydrogen bonds. The retroinsertion of metals occurs by
deprotonation of 48 hydroxyls, two from each pyrogallol unit to form 24-metallated metal-seamed organic nanocapsule/
MONC. Each metal center adopts an octahedral geometry with two water ligands and four phenoxy groups of adjacent
pyrogallol[4]arene units. The assembly adopts a near spheroidal or truncated cuboctahedron geometry in the solid state. SANS

Figure 2 Left: global fitting for copper-seamed C-hexylpyrogallol[4]arene hexamer at 1 wt% (dots) and 5 wt% (triangles) concentrations. Right: front
view of core structure of copper-seamed C-alkylpyrogallol[4]arene hexamer. Hydrogen bonds and alkyl tails have been removed for clarity. Color
codes: green, copper; red, oxygen; and gray, carbon.
Small-Angle Neutron Scattering in Supramolecular Complexes 293

study of PgCnCu hexamers in acetonic solution was conducted to investigate their geometry and stability in solution.17 A copper-
seamed C-propylpyrogallol[4]arene/PgC3Cu hexamer in solution was fitted to a Schulz sphere model and yielded a radius of 10 Å
in solution. The solution shape and size (10 Å) of PgC3Cu match closely with those of single-crystal X-ray dimension of
(11.11  0.01) Å. Similarly, the SANS data for PgC6Cu and PgC9Cu were fitted to Schulz sphere model and yielded radius of
16.6 and 18.6 Å, respectively, which match closely with solid-state geometries with (14.89  0.15) and (17.59  0.02) Å, respectively
(Fig. 2, left). For both PgC6Cu and PgC9Cu, SANS measurements were conducted at two concentrations of 1 and 5 wt%. The resul-
tant 2-D SANS data were globally fitted to Schulz sphere model.17 Global fitting of multiple data sets for a given system yielded
similar radius values with reasonable statistics demonstrating the monodisperse nature of these nanocapsular spheres in solution.
Therefore, a gradation in size of PgCnCu hexamers is observed as a function of alkyl tail length. Interestingly, this 24-metallated
hexamer is robust in solution, and its shape is unaffected by the change in alkyl tail lengths.17
Kumari et al. also investigated the self-assembly and stability of ferrocene-enclosed hydrogen-bonded C-methylpyrogallol[4]arene
dimers (PgC1 3 Fc) in solution.18 The dimer consists of two pyrogallol[4]arene units wherein the upper-rim hydroxyls interact through
hydrogen bonding via two water molecules. The internal cavity consists of one ferrocenium molecule, confirmed by NMR analyses. The
crystalline sample was analyzed in methanolic solution using SANS. The scattering data were fitted to Schulz sphere model yielding
a radius of (6.67  0.02) Å with O(c2/N) ¼ 1.34. The solution-phase radius value of PgC1 3 Fc matches closely with solid-state radius
value of 6.74 Å, demonstrating the stability of hydrogen-bonded pyrogallol[4]arene dimers in methanol. Most hydrogen-bonded PgC-
based nanocapsules are typically only stable in aprotic solvent (from NMR studies by Cohen et al.);19–24 however, the SANS results of
PgC1 3 Fc dimer provide confirmation about at least the short-term stability of hydrogen-bonded dimers in protic solvent.18
SANS has also been instrumental in probing structures of supramolecular assemblies that are otherwise difficult to crystallize.25
The lanthanide complexes of pyrogallol[4]arenes, namely, holmium-containing C-methylpyrogallol[4]arene/PgC1Ho and
holmium-containing C-propylpyrogallol[4]arene/PgC3Ho were obtained by reacting 1 equiv. of PgCn with 16 equiv. of holmium
nitrate and 8 equiv. of pyridine in acetonitrile–water (20:1) solvent mixture.25 The dark brown precipitate obtained from this reac-
tion was resistant to vapor diffusion, solvothermal, and solvent evaporation crystallization methods. SANS study of PgC1Ho and
PgC3Ho complexes in d6-DMSO yielded scattering curves that were fitted best to Schulz sphere model. PgC1Ho complex showed the
presence of (18.1  0.5) Å radius spheres, whereas PgC3Ho complex showed the presence of (18.2  0.9) Å radius spheres in DMSO.
The difference in shorter alkyl tail length is too small of a fraction of overall size of the capsule, and the radius obtained from SANS
Schulz sphere fit is orientationally averaged; hence, a remarkable difference in size is not observed. The radius obtained was used to
calculate the volume, which was then fitted to a variety of Platonic and Archimedean solid models to deduce its geometry in solu-
tion. The solution structure of these lanthanide complexes should adopt a truncated cuboctahedron structure that comprises 12
pyrogallol[4]arene units (squares) and 48 metal centers (8 hexagons and 6 octagon faces). This was the first reported metal-
containing pyrogallol[4]arene dodecameric nanoassembly in solution, which is much larger than a regular hexamer (10 Å radius).25

2.14.3.2 Core–Shell Sphere


The core–shell sphere model calculates the form factor for a monodisperse spherical particle with a core–shell structure, and the
resultant form factor is normalized by the total particle volume (Fig. 3).26 The input variables for this model include scale, core
radius, shell thickness, core SLD, shell SLD, solvent SLD, and background. The function calculated is
 
scale 3Vc ðrc  rs Þj1 ðqrc Þ 3Vs ðrs  rsolv Þj1 ðqrcs Þ 2
P ðqÞ ¼ þ þ bkg
Vs qrc qrs
where j1(x) ¼ (sin x  x cos x)/x2, rs ¼ rc þ t and Vi ¼ ð4p=3Þri3 .
The solid-state structure of hydrogen-bonded C-alkylpyrogallol[4]arene hexamers were elucidated in 2008 by Atwood et al.27
The hydrogen-bonded PgCn hexamers adopted a spherical geometry for alkyl tail lengths n ¼ 3, 6, and 9. At tail length n ¼ 11,
the hydrogen-bonded hexamers showed a spherical core structure of PgC0; however, the tail lengths were interdigitated to form
a more ellipsoidal geometry. For copper-seamed C-alkylpyrogallol[4]arene hexamers, solid structures were only obtained for up
to alkyl tail length 9.28 The longer hydrocarbon tail lengths yielded crystals that did not diffract well. The SANS solution study

Figure 3 Schematic showing the core–shell sphere geometry with shell thickness, core radius, SLD core, SLD shell, and SLD solvent parameters.
294 Small-Angle Neutron Scattering in Supramolecular Complexes

Figure 4 SANS fitting to core–shell sphere data for copper-seamed C-undecylpyrogallol[4]arene/PgC11Cu hexamer (dots) and copper-seamed
C-tridecylpyrogallol[4]arene/PgC13Cu hexamer (triangles) in d-chloroform.

of PgCnCu hexamers for longer tail lengths, n ¼ 11 and 13, was fitted to a core–shell sphere model wherein the scattering length
density of the PgC0Cu was held constant for the core and the scattering length density of the tails was modeled as shell
(Fig. 4).29 The SLD of solvent d-chloroform was held constant (SLD ¼ 3.16  10 6 Å2). The core–shell sphere model fitting for
PgC11Cu hexamer yielded a core radius of (8.4  0.63) Å and shell thickness of (13.7  0.78) Å, for a total radius of (22.1  1.0)
Å. Similarly, the core–shell sphere model fit for the PgC13Cu hexamer yielded a core radius of (10.0  0.41) Å and a shell thickness
of (12.0  0.53) Å with the total radius (22.0  0.64) Å in d-chloroform. The slight difference in core and shell sizes can be attributed
to different amounts of solvation around the alkyl tail lengths. The similarity in overall radius is owing to small difference in alkyl
tail length (undecyl- versus tridecyl-) and solvent-hydrocarbon tail interactions.29

2.14.3.2.1 Contrast Variation Method


In another study, Kumari et al. investigated the encapsulation of insulin within the C-methylresorcin[4]arene (RsC1) macrocycle
using SANS contrast variation method.30 Insulin acted as a biotemplate that allowed for paneling of macrocycle around it for
the construction of a larger nanocapsule. However, difficulty in crystallization of a protein molecule with water–acetonic mixture
was a challenge. SANS served as an excellent tool to study this nanoassembly in solution using contrast variation method. In
contrast variation method, the SLD of the shell was matched with that of the solvent such that only the core is visible in solution.
Thereafter, the contrast between the core and shell is gradually increased by virtue of increasing the difference in the SLD of solvent
and shell to make the shell visible. In this case, we expected insulin to act as core and RsC1 to act as shell. The SLD of the shell (RsC1)
was matched with that of the solvent. The SLD of the shell was manipulated by using a mixture of D2O and H2O. Specifically,
samples of RsC1 alone, insulin alone, and RsC1 with Insulin were studied in D2O alone, H2O alone, and D2O–H2O mixture
solvents. The SANS data of resulting samples were then fitted to a core–shell sphere model. The analyses yielded a shell thickness
of 7 Å, which corresponded to the bowl size of the RsC1 macrocycle, and a core with shell diameter of 32 Å, confirming the encap-
sulation of insulin monomer within the RsC1 macrocycle.30

2.14.3.3 Uniform Ellipsoid


The uniform ellipsoid model calculates the form factor of a monodisperse ellipsoid with uniform scattering length density
(Fig. 5).31 The form factor is normalized by the particle volume to cover all possible orientations of the ellipsoidal assembly.
The input variables for uniform ellipsoid model include the scale factor, rotation axis/Ra, Rb, SLD ellipsoid, SLD solvent, and inco-
herent background.31 The function is calculated as
ð1 h
scale   1=2 i
P ðqÞ ¼ ðrell  rsolv Þ2 f 2 qrb 1 þ x2 v2  1 dx þ bkg
Vell 0

ðsin z  z cos zÞ
f ðzÞ ¼ 3Vell
z3

4p 2
Vell ¼ ra r
3 b
ra

rb
Small-Angle Neutron Scattering in Supramolecular Complexes 295

Figure 5 Schematic showing the uniform ellipsoid geometry with major radius, minor radius, SLD core, and SLD solvent parameters.

The copper-seamed pyrogallol[4]arene hexamers are spherical in shape; however, it becomes increasingly difficult to obtain
the structural information for capsules with longer alkyl tails. Albeit difficult to crystallize, copper-seamed C-heptadecylpyro-
gallol[4]arene (PgC17Cu) was obtained as brown precipitate upon reaction of copper nitrate with PgC17. The longer alkyl tail
restricted its solubility in polar solvents, and the resultant precipitate only dissolved in deuterated o-xylene for SANS study.29
Scattering data were collected for 1 and 5 wt% PgC17Cu samples, and a variety of spherical and cylindrical models were tried
for SANS data fitting, including Schulz sphere, uniform cylinder, elliptical cylinder, and prolate core–shell models. The data
exclusively fitted to uniform ellipsoid model for both 1 and 5 wt%, separately and in global fitting. The fitting results suggested
the presence of equilibrium between the solvent and soluble ellipsoidal nanoassembly, rather than a mere aggregation of hex-
amers in solution. The dimensions of solution-phase ellipsoidal geometry of PgC17Cu were 115.4  0.2 Å (half length of the
axis of rotation) and 24.2  0.02 Å (half length of the minor axis). The rotation axis was larger than the minor axis, indicating
that the shape corresponds to a prolate ellipsoid with a limiting power-law slope approaching q 1 at low q.29 The ellipsoidal
geometry has a shorter axis diameter of 48 Å and is slightly larger than that observed for PgC13Cu hexamers (44 Å), indicating
that the alkyl tails reside or interdigitate along the longer axis of ellipsoid like those of a hydrogen-bonded PgC11 hexamers. In
addition, there is a possibility of the presence of more than one nanocapsule along the shorter axis, and the ratio of 1:5
between minor and major axes indicates an overall volume to accommodate  20 spheres in this solution-phase ellipsoidal
geometry.29

2.14.3.4 Bimodal Schulz Sphere Model


The bimodal Schulz sphere model calculates the scattering for a bidisperse population of spheres, each with a specific
uniform scattering length density.13–15 Each type of sphere has a distribution of radii that obeys Schulz distribution and
is polydisperse in nature. The intensity is normalized by the average particle volume, and the normalized probability distri-
bution can be plotted as a function of radius. In addition, the number density based on the volume fraction and polydisperse
particle volume can be calculated. The input variables include volume fraction 1 and 2, radius 1 and 2, polydispersity 1 and
2, SLD 1 and 2, SLD solvent, and background.13–15 Note that 1 and 2 correspond to two populations of spheres in a given
solvent system.
In 2012, Kumari et al. used bimodal Schulz sphere model to deduce the equilibrium phenomenon for metal-seamed pyrogallol
[4]arene hexameric and dimeric nanocapsules in solution.32 Unlike hexamers, dimers consist of two pyrogallol[4]arene units and
eight metal centers (Fig. 6). Each pyrogallol[4]arene unit acts a hemisphere with the lower rim forming the periphery of the capsule
and upper rim, consisting of hydroxyls, forming the coordination sites for metal coordination, which form an equatorial belt seam-
ing the two hemispheres of the dimeric nanocapsule. Each metal center is primarily pentacoordinated and adopts a square pyra-
midal geometry with four phenoxy coordination sites from pyrogallol[4]arene units and an external pyridine/DMSO ligand
(Figs. 6 and 7).
Albeit different in architecture, both dimers and hexamers consist of 1:4 ratio of macrocycle–metal. Selective isolation of
one species from a mixture of nanocapsules was difficult, and a deeper understanding of equilibrium shifts in solution was
necessary. SANS study was used as a tool to investigate the effect of temperature, metal type, and solvent on the relative
volume fractions of dimers and hexamers in solution. The in situ neutron scattering study was conducted at relatively lower
concentrations of 0.001 M to avoid precipitation/crystallization of product. Copper and nickel metals were chosen owing to
their ability to form both dimers and hexamers of pyrogallol[4]arenes. The solvents chosen were acetone and methanol, and
a range of temperatures ( 40 C, RT, and 50 C) were chosen based on the type of solvent utilized. The scattering intensity of
samples was measured at RT at day 0 and day 3. Two related syntheses were carried out to study equilibrium shifts: (a)
MONC synthesis was performed at RT, but the measurements were conducted at  20 C, RT, and 50 C; (b) synthesis and
SANS measurements were carried out at  20 C, RT, and 50 C. The resulting SANS data were fitted to a bimodal Schulz
sphere model wherein two populations of spheres with radius 7 and 10 Å, corresponding to dimers and hexamers, were ob-
tained. The relative volume fraction analyses of the two populations were calculated. The trends in volume fraction for day
0 indicated that despite the type of synthesis protocol, hexamers were predominant species at low temperatures, whereas
dimers were predominant species at high temperatures. The extended (3-day) stability study showed a shift in equilibrium
toward dimers, indicating that the dimers are thermodynamic products at all temperatures, whereas hexamers are the kinetic
296 Small-Angle Neutron Scattering in Supramolecular Complexes

Figure 6 Top view of a metal-seamed C-alkylpyrogallol[4]arene dimer.32 The metal centers have pyridine as external ligand. Hydrogen atoms and
alkyl tails have been removed for clarity. Color codes: blue, metal; green, nitrogen; red, oxygen; and gray, carbon.

products at low temperatures (Fig. 7). Temperature is hence a controlling parameter that could be manipulated to obtain
desired nanoassembly.32

2.14.3.5 Triaxial Ellipsoid


A triaxial ellipsoid model calculates the form factor of a triaxial ellipsoid with a uniform scattering length density, and the resulting
form factor is normalized by the particle volume (Fig. 8).31 The input variables include the scale factor, semiaxis A, semiaxis B and

Figure 7 Schematic showing the pyrogallol[4]arene/macrocyle (A), metal-seamed pyrogallol[4]arene dimer (B), and metal-seamed pyrogallol[4]
arene hexamer (C).
Small-Angle Neutron Scattering in Supramolecular Complexes 297

Figure 8 Schematic showing the triaxial ellipsoid geometry with three radius axes, SLD core, and SLD solvent parameters.

semiaxis C, SLD ellipsoid, SLD solvent, and incoherent background. The function calculates the form factor of an ellipsoid with all
three different semiaxes, with semiaxis A being the smallest and semiaxis C being the largest (Fig. 8). The form factor is calculated as
ð 1 ð 1 rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
h
px
px iffi
scale
P ðqÞ ¼ ðrell  rsolv Þ2 42 q a2 cos2 þ bsin2 ð1  y2 Þ þ c2 y2 $s$dxdy
2

Vel 0 0 2 2
 
sin x  x cos x 2
42 ðxÞ ¼ 9
x3

a2 þ b2 þ c2
R2g ¼
5
Kumari et al. has used triaxial ellipsoid model to investigate geometries of PgC17Cu hexamer in solution.29 However, the data
were exclusively fitted to a uniform ellipsoidal model. Triaxial elliptical model has been used to investigate other supramolecular
assemblies as well. Thus far, we have not found a triaxial ellipsoidal nanoassembly in solution.

2.14.3.6 Cylinder Model


The cylinder model calculates the form factor for a monodisperse right circular cylinder with uniform scattering length density, and
the resulting form factor is normalized by the particle volume (Fig. 9).26 The input variables for the cylinder model include scale,
radius, length, SLD cylinder, SLD solvent, and incoherent background. The function is calculated as
ð
scale p=2 2
P ðqÞ ¼ f ðq; aÞ sin a da
Vcyl 0


J1 ðqr sin aÞ
f ðq; aÞ ¼ 2 rcyl  rsolv Vcyl j0 ðqH cos aÞ
ðqr sin aÞ

Vcyl ¼ pr 2 L

sinðxÞ
j0 ðxÞ ¼
x
where J1(x) is the first-order Bessel function and alpha is the angle between the cylinder axis and the scattering vector, q. The
integral over alpha averages the form factor over all possible orientations of the cylinder with respect to q.26
The bowl shape of the macrocycles allows for the self-assembly of supramolecular nanoassemblies into spheres, tubes, and
helices.18,25,28,32–62 These geometries are often elucidated SCXRD studies. The mimicking of solid-state geometries becomes ques-
tionable for nonspherical architectures. An excellent example includes the study of hydrogen-bonded pyrogallol[4]arene nanotubes

Figure 9 Schematic showing the uniform cylinder geometry with core radius and length parameters.
298 Small-Angle Neutron Scattering in Supramolecular Complexes

in solution by Kumari et al.63 Pyrogallol[4]arenes were first shown to self-assemble as hydrogen-bonded nanotubes with pyrene as
exo-guest molecules in 2006 by Atwood and Dalgarno.64 PgC6 macrocycle was also shown to self-assemble into hydrogen-bonded
hexameric nanocapsules with enclosed pyrene guest molecules in the presence of acetonitrile or ethyl acetate.53 Crystallization of
PgC6 macrocycle in a mixture of acetonitrile and water yielded tetrameric hydrogen-bonded PgC6 3 pyrene nanotubes with pyrene
molecules located exo to the hydrogen-bonded tubular framework.64 The intriguing self-assembly process and nanotubular arrange-
ment were studied in acetone using SANS and diffusion NMR.63 The fitting of PgC6 3 pyrene nanotube to a cylindrical model
yielded poor statistics indicating the rupture of hydrogen-bonded tubular framework in solution. The data instead were fitted to
a Schulz sphere model with a uniform radius of 8.6 Å. The size of the PgC6 sphere in solution is slightly larger than that observed
for PgC3M (7 Å) dimeric nanoassembly, consistent with gradation in alkyl tail length (hexyl vs propyl).
Another excellent example is the SANS study of hydrogen-bonded ferrocene-enclosed C-methylpyrogallol[4]arene (PgC1 3 Fc)
nanotube.65 Unlike the tetrameric hydrogen-bonded PgC6 3 pyrene nanotube, the PgC1 3 Fc nanotube is hexameric and consists of
two layers of alternating 3PgC1 bowls enclosing an endo ferrocene guest molecule. In contrast to the PgC6 3 pyrene nanotubes
wherein the upper-rim hydroxyls form the inner core of the solid-state cylindrical moiety, the upper-rim hydroxyls form the outer
shell of the PgC1 3 Fc nanotube. The enclosed guest exists as ferrocenium ion, confirmed by NMR analyses. SANS study of this
nanotube in methanol shows a rearrangement of nanotubular framework to a dimeric nanocapsule. SANS data fitting to a cylinder
model gave poor statistics; however, fitting to a Schulz sphere model yielded a radius of 6.7 Å. The PgC1 dimer size of 6.7 Å is
slightly shorter than that observed for PgC3 dimer in solution (7 Å), consistent with gradation in tail length.65
Compared with the hydrogen-bonded PgC-based nanotubes, the iron-containing pyrogallol[4]arene nanotube was found stable
in solution. SANS study was conducted on PgC1Fe and PgC3Fe complexes in d6-DMSO. The data analyses suggested that the
nanoassembly fitted best to the cylinder model and yielded a radius of 7 Å and length of 124 Å. The metal-seamed nanotubes
were centrifuged from 0 to 2 min. The resultant nanoassembly again were fitted to a cylinder model yielding a radius of 7 Å;
however, centrifugation caused a reduction in the length of the nanotube, and it changed from 124 to 45 Å. In another experiment
of PgC1Fe, the molar ratio of pyridine to HNO3 was varied from 1:0 to 1:1 with an increment of 0.2. The SANS data for Py:HNO3
ratios of 1:0 and 1:0.2 were fitted to cylinders of radius 6.5–7 Å and lengths of 25 Å. For Py:HNO3 molar ratio of 1:0.4, 1:0.6, and
1:0.8, the data were fitted best to Schulz sphere model and yielded dimensions close to that of an hexamer (10 Å). At 1:1 molar ratio
of Py:HNO3, data were best fitted to a spherical dimer of 7 Å. Therefore, SANS study revealed pH and centrifugation time as control-
ling parameters for manipulating the geometry of PgC1Fe nanoassembly.
More recently, Kumari et al. performed a SANS study on carbon nanotubes.66 Carbon nanotubes may have single wall
(SWCNTs), double wall (DWCNTs), or multiwall (MWCNTs) with varying aspect ratios and lengths ranging from few microns
to millimeters. Their high stiffness, low density, and axial strength are useful properties that make them unique for investigating
thermal, mechanical, and electrical properties.67,68 Shortening the length of carbon nanotubes typically requires high-energy
sonication, toxic chemicals, or lengthy processing times.69–75 Controlled slicing of carbon nanotubes without compromising their
properties has been a challenging issue. Kumari et al. used SANS to study the lateral slicing of carbon nanotubes through thin-film
shearing using a vortex fluidic device (VFD).66 It was achieved by thin-film shearing under a pulsed Q-switch Nd:YAG laser at
1064 nm wherein a controllable mechanoenergy was generated within the liquid media at speed range of 2000–9000 rpm. The
rotational speed of 7500 rpm and q 45 on VFD and pulsed laser source of 1064 nm pulses of 5 ns at a repetition rate of 10 Hz
for 30 min provided optimal processing parameters for later slicing of CNTs. The CNT length was studied using atomic force micros-
copy (AFM) and SANS studies. SANS fitting to cylinder model for SWCNTs treated with laser alone and laser with VFD (confined
mode, 10 min, 30 min, and 60 min) yielded nanotubes with radius 0.7  0.2 nm, which is consistent with AFM results. Similarly,
SANS showed that DWCNTs were sliced down to 160 nm length and MWCNTs to 171 nm length, consistent with AFM results.66

2.14.3.7 Core–Shell Cylinder


The core–shell cylinder model calculates the form factor for a monodisperse, right circular cylinder with a core–shell scattering
length density profile (Fig. 10).76 The shell thickness, t, is considered uniform over the entire surface of the core, and the form factor

Figure 10 Schematic for core–shell cylinder showing SLD core, SLD solvent, SLD shell, radius of core, radius of core with shell, and the two
lengths of concentric cylinders.
Small-Angle Neutron Scattering in Supramolecular Complexes 299

is normalized by the total particle volume. The input variables include scale, core radius, shell thickness, core length, SLD core, SLD
shell, SLD solvent, and incoherent background.
The form factor, P(q), is calculated as
ð
scale p=2 2
P ðqÞ ¼ f ðq; aÞ sin a,da
Vshell 0

J1 ðqr sin aÞ J1 ½qðr þ t Þsin a


f ðq; aÞ ¼ 2ðrcore  rshell ÞVcore j0 ðqH cos aÞ þ 2ðrshell  rsolvent ÞVshell j0 ½qðH þ 1Þcos a
ðqr sin aÞ ½qðr þ t Þsin a

j0 ðxÞ ¼ sinðxÞ=x

Vcore ¼ pr 2 L

Vshell ¼ pðr þ t Þ2 Ltotal


where r is the radius of the core of the cylinder, J1(x) is the first-order Bessel function, alpha is the angle between cylinder axis and
scattering vector, q. The integral over alpha averages the form factor over all possible orientations of the cylinder.

2.14.3.8 Hollow Cylinder


The hollow-cylinder model calculates the form factor of a monodisperse, hollow, right circular cylinder/tube (Fig. 11).31 The hollow
cylinder has an internal SLD that matches that of the outer solvent environment. The form factor is normalized by the tubular mate-
rial volume only. The form factor can be calculated as
ð1 h  
 1=2  1=2 i sinðqHxÞ 2
P ðqÞ ¼ scale$ Vshell $ðDrÞ2 j2 q; Rshell 1  x2 ; Rcore 1  x2 dx
0 qHx
An excellent example of supramolecular complex that fits to hollow-cylinder models is the gallium-seamed pyrogallol[4]arene
complexes in solution.38 Gallium-seamed C-butylpyrogallol[4]arene/PgC4Ga hexameric nanocapsules are composed of six pyro-
gallol[4]arene units and 12 metal centers (Fig. 12).57
The resulting nanoassembly adopts a “rugby-ball”-shaped geometry in solid state, which is different from the typical “spher-
ical” copper- or nickel-seamed C-alkylpyrogallol[4]arene hexamers.57 The difference could be attributed to the difference in
oxidation state with Ga being in þ 3 oxidation state while Cu/Ni being in þ 2 oxidation state that allows for 24 Cu or Ni to
seam (PgCn)6Cu24 or (PgCn)6Ni24 hexamers versus only 12 to seam the (PgCn)6Ga12 hexameric nanoassembly. The PgCnGa hex-
amer (shortened notation for (PgCn)6Ga12) also has four water gates that could potentially allow for ion transport or insertion of
other metals within the framework (Fig. 12). Addition of zinc nitrate to a solution of preformed PgC4Ga hexamers causes a tran-
sition in solid-state geometry from the rugby ball to a near spheroid. This intriguing solid-state result incited interest in investi-
gating PgCnGa-based nanoassemblies in solution. SANS study of PgC4Ga and PgC4GaZn nanoassembly in acetone yielded
scattering curves that did not fit to expected ellipsoidal and spheroidal geometries. After testing a variety of model-dependant
models, PgC4Ga and PgC4GaZn were shown to fit to a hollow-cylinder model. To our surprise, in both cases, the dimensions
of hollow cylinders were different, indicating a difference in self-assembly of pyrogallol[4]arene units and metal centers in solu-
tion. The hollow cylindrical assembly of PgC4Ga has a length of 7 Å and core with shell radius of 17 Å. The length of 7 Å corre-
sponds to the length of individual pyrogallol[4]arene bowl with upper-rim hydroxyls directly facing up and interacting with
exposed solvent media. The orientation of bowls was confirmed by NMR analyses. The solution-phase hollow-cylinder geometry
of PgC4GaZn, on the other hand, has a length of 13 Å and a core with shell radius of 12 Å. The length of 13 Å corresponds to the
diameter of pyrogallol[4]arene bowl indicating a sideways orientation of bowls with its upper-rim hydroxyls forming the inner

Figure 11 Schematic for hollow-cylinder model showing SLD core, SLD solvent, SLD shell, length, radius of core, and radius of core with shell.
300 Small-Angle Neutron Scattering in Supramolecular Complexes

Figure 12 Front view of rugby-ball-shaped gallium-seamed C-butylpyrogallol[4]arene hexamer in solid state. For clarity, alkyl tails and hydrogen
atoms have been removed. Color codes: gray, carbon; red, oxygen; white, hydrogen; and turquoise, gallium.

core of the cylinder, confirmed by NMR analyses.38 The fluxional behavior of these supramolecular nanoassemblies in solution
was ascertained by combined SANS and NMR analyses. SANS played a crucial role in discerning the change in geometries of
supramolecular nanoassemblies in solution.

2.14.3.9 Fractal and Fractal Flexible Cylinder Models


The fractal model calculates the scattering from fractal-like aggregates that are built from the spherical building blocks following the
Teixeira reference.77 One can obtain the number density, aggregation number, and other parameters based on the volume fraction
and polydisperse particle volume from this model. The input variables for the fractal model include the volume fraction, block
radius, fractal dimension, correlation length, SLD block, SLD solvent, and incoherent background.
The scattered intensity is calculated as
IðqÞ ¼ P ðqÞSðqÞ þ bkgd

The form factor, P(q), is calculated as


P ðqÞ ¼ fVp Dr2 F ðqR0 Þ2
 
4
Vp ¼ pR30
3

3½sinðxÞ  x cosðxÞ
F ðxÞ ¼
x3

The structure factor, S(q), is the interference from the building blocks of fractal-like clusters and is calculated as
   
sin Df  1 tan1 ðqxÞ Df G Df  1
SðqÞ ¼ 1 þ   ðDf 1Þ=2
ðqR0 ÞDf 1 þ 1 q2 x2

The fractal flexible cylinder model plots scattering from a mass fractal object using the Teixeira model with a rigid cylinder as the
building block.77 The radius in the fractal S(q) is the Rg of the cylinder. As long as the fractal length scale/correlation length and the Rg
of the chain are very different so that the fractal knows no details of the cylinder, this construct is valid. The input variables for the
fractal flexible cylinder model include the volume fraction, fractal dimension, correlation length, SLD block, SLD solvent, contour
length, Kuhn length, cylinder radius, and incoherent background. This model combines scattering of fractal model and flexible cylinder
models. The block radius of the fractal model, in this case, corresponds to the radius of gyration of the flexible cylinder:
L
nb ¼
b

  bL
R2g ¼ a n2b
6
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

x 2
x 3 0:176=3
aðxÞ ¼ 1þ þ
3:12 8:67
Small-Angle Neutron Scattering in Supramolecular Complexes 301

The primary structure of supramolecular low-molecular-weight gel is reasonably understood; however, the 1-D hydrogen-
bonded framework that yields interconnected 3-D framework capable of immobilizing the fluid phase of the gel network is very
speculative. The limitation resides in the inability of analytic techniques to yield molecular-level information for extended gel
networks. The molecular structures of 1-D networks are traditionally elucidated through drying of the gel and the study of resultant
xerogels through PXRD technique. A major assumption in this technique is the holding of the gel network in the absence of solvent
that forms 90%–99% of the gel architecture. In situ SANS study has proved very useful in probing the supramolecular gel network at
varying length scales to address the higher-order supramolecular structure and methods by which gelator molecules are formed in
solution to form interconnected fibrous morphology. SANS, in particular, provides complimentary benefit along with other
traditional techniques (SAXS, WAXS, SEM, SCXRD, and PXRD) to elucidate solvent-enclosed gel networks. More recently, Kumari
et al. investigated bis(urea) gelators with perfluoroalkyl ponytails with their hydrocarbon analogues using SANS technique.66 It was
expected that insulating the hydrogen-bonded urea motif within the fluorous sheath would yield different gelation behavior despite
preserving its compatibility with a variety of solvents. SANS study shows discrete differences in thermal evolution and morphology
of fluorinated versus non-fluorinated gelators. Fitting to the fractal flexible cylinder model for SANS data yields a ratio of Kuhn and
Contour ratio that reflects on the rigidity of gel networks. SANS analyses reveal that the addition of perfluoroalkyl substituent to bis-
urea gelator makes it more transparent and stronger due to a reduction in interchain interaction in fluorous compounds. SANS study
shows an annealing effect exclusively for fluorinated gels, which is similar to that of Oswald ripening. The gel structure is intact in
DMSO solvent up to the Tgel of material. However, the fully reversible structure in DMSO collapses to spherical aggregates beyond
Tgel in DMF solvent, indicating that the solvent plays a crucial role in morphology of fluorinated gel network. In contrast, the non-
fluorinated gelator shows thermoreversible behavior in both solvents (DMSO and DMF).66
The local structure of fluorinated gelator is affected with the change in concentration for fluorinated gelator; however, fiber
rigidity and cylinder radius remain the same. SANS results indicate that the physical appearance of the gel could give inaccurate
estimate of its thermoreversibility and metastable structures. This SANS study addresses the gel maturation process and the effect
of temperature and concentration on the self-assembly (formation and alterations) of gel fibers in varying solvents. 66
In another study, the effect of thin-film shearing was studied on 3-D network of bis-urea-based fluorinated versus nonfluorinated
supramolecular gelators.78 Thin-film shearing was conducted using VFD. SANS study of VFD-treated fluorinated gelator showed
a complete disruption of gel fibers. The hydrocarbon analogue, on the other hand, was only partially broken owing to the resis-
tance.78 Again, the data were fitted to fractal flexible cylinder model to study the disruption phenomenon of gel networks.

2.14.4 Conclusion

This article covers some of the basic principles of SANS and few of the model-dependant SANS data analyses. We have restricted our
discussion to the models that have been employed to study supramolecular complexes in solution. A variety of spherical, tubular,
ellipsoidal, core–shell, and fractal models have been discussed, which can be fitted to supramolecular nanocapsules, nanotubes,
toroids, and gel-based assemblies. SANS has efficiently been used as an effective tool to discern solution- and soft-phase geometries
and some of the factors controlling their architectures. The difference in solution- and solid-phase geometries reflects the reversible
and flexible nature of non-covalently bonded interactions in self-assembly processes. Future studies using SANS would focus on
further understanding of the mechanism of self-assembly process that leads to nucleation and eventual crystallization of solid-
state supramolecular complexes.

References

1. Lehn, J.-M. Supramolecular Chemistry Concepts and Perspectives; VCH Weinheim, 1995. ISBN: 978-3-527-29311-7.
2. Steed, J. W.; Atwood, J. L. Supramolecular Chemistry, 2nd ed.; Wiley: Chichester, UK, 2009.
3. Atwood, J. L.; Lehn, J. M. Comprehensive Supramolecular Chemistry, 1st ed.; Pergamon: Oxford, England/New York, 1996.
4. Pons, M.; Millet, O. Prog. Nucl. Magn. Reson. Spectrosc. 2001, 38, 267–324.
5. Fielding, L. Tetrahedron 2000, 56, 6151–6170.
6. Hahn, E. L. Phys. Rev. 1950, 77, 297–298.
7. Stejskal, E. O.; Tanner, J. E. J. Chem. Phys. 1965, 42, 288.
8. Parella, T. Magn. Reson. Chem. 1996, 34, 329–347.
9. Stark, D. D.; Bradley, W. G. Magnetic Resonance Imaging, 3rd ed.; Mosby: St. Louis, MO, 1999.
10. Kumari, H. Solution, Magnetic and Insulin Encapsulation Study of Metal-Seamed Organic Nanocaspules. Ph.D. Thesis; University of Missouri-Columbia, 2011.
11. Jackson, A. J. Introduction to Small Angle Neutron Scattering and Neutron Reflectometry; NIST Centre for Neutrons: Gaithersburg, MD, 2008.
12. Hammouda, B. A Tutorial on Small-angle Neutron Scattering from Polymers. NIST Pub, June 1995; p 60.
13. Schulz, G. Z. Phys. Chem. 1939, B43, 25–46.
14. Schulz, G. V. Z. Phys. Chem. 1935, B30, 379–398.
15. Kotlarchyk, M.; Chen, S. H. J. Chem. Phys. 1983, 79, 2461–2469.
16. McKinlay, R. M.; Cave, G. W. V.; Atwood, J. L. Proc. Natl. Acad. Sci. U. S. A. 2005, 102, 5944–5948.
17. Kumari, H.; Kline, S. R.; Schuster, N. J.; Atwood, J. L. Chem. Commun. 2011, 47, 12298.
18. Mossine, A. V.; Kumari, H.; Fowler, D. A.; Maerz, A. K.; Kline, S. R.; Barnes, C. L.; Atwood, J. L. Israel J. Chem. 2011, 51, 840–842.
19. Avram, L.; Cohen, Y. J. Am. Chem. Soc. 2002, 124, 15148–15149.
20. Avram, L.; Cohen, Y. Org. Lett. 2003, 5, 3329–3332.
302 Small-Angle Neutron Scattering in Supramolecular Complexes

21. Evan-Salem, T.; Cohen, Y. Chem. Eur. J. 2007, 13, 7659–7663.


22. Cohen, Y.; Evan-Salem, T.; Avram, L. Supramol. Chem. 2008, 20, 71–79.
23. Slovak, S.; Evan-Salem, T.; Cohen, Y. Org. Lett. 2010, 12, 4864–4867.
24. Avram, L.; Cohen, Y.; Rebek, J., Jr. Chem. Commun. 2011, 47, 5368.
25. Kumari, H.; Kline, S. R.; Fowler, D. A.; Mossine, A. V.; Deakyne, C. A.; Atwood, J. L. Chem. Commun. 2014, 50, 109–111.
26. Guinier, A.; Fournet, G. Small-Angle Scattering of X-rays; John Wiley and Sons: New York, 1955.
27. Cave, G. W. V.; Dalgarno, S. J.; Antesberger, J.; Ferrarelli, M. C.; McKinlay, R. M.; Atwood, J. L. Supramol. Chem. 2008, 20, 157–159.
28. Kumari, H.; Mossine, A. V.; Schuster, N. J.; Barnes, C. L.; Deakyne, C. A.; Atwood, J. L. CrstEngComm 2014, 16, 3718–3721.
29. Kumari, H.; Kline, S. R.; Schuster, N. J.; Barnes, C. L.; Atwood, J. L. J. Am. Chem. Soc. 2011, 133, 18102–18105.
30. Kumari, H.; Kline, S. R.; Atwood, J. L. Chem. Commun. 2012, 48, 3599–3601.
31. Feigin, L. A.; Svergun, D. I. Structure Analysis by Small-Angle X-Ray and Neutron Scattering; Plenum: New York, 1987.
32. Kumari, H.; Mossine, A. V.; Kline, S. R.; Dennis, C. L.; Fowler, D. A.; Teat, S. J.; Barnes, C. L.; Deakyne, C. A.; Atwood, J. L. Angew. Chem. Int. Ed. 2012, 51, 1452–1454.
pii:S1452/1451-S1452/1203.
33. Drachnik, A. M.; Kumari, H.; Barnes, C. L.; Deakyne, C. A.; Atwood, J. L. CrystEngComm 2014, 16 (31), 7172–7175.
34. Maerz, A. K.; Fowler, D. A.; Mossine, A. V.; Mistry, M.; Kumari, H.; Barnes, C. L.; Deakyne, C. A.; Atwood, J. L. New J. Chem. 2011, 35, 784–787.
35. Kumari, H.; Dennis, C. L.; Mossine, A. V.; Deakyne, C. A.; Atwood, J. L. ACS Nano 2012, 6, 272–275.
36. Kumari, H.; Kline, S. R.; Atwood, J. L. Chem. Commun. 2012, 48, 3599–3601.
37. Kumari, H.; Kline, S. R.; Wycoff, W. G.; Atwood, J. L. Small 2012, 8, 3321–3325.
38. Kumari, H.; Kline, S. R.; Wycoff, W. G.; Paul, R. L.; Mossine, A. V.; Deakyne, C. A.; Atwood, J. L. Angew. Chem. Int. Ed. 2012, 51, 5086–5091. pii:S5086/5081-S5086/5031.
39. Mossine, A. V.; Kumari, H.; Fowler, D. A.; Shih, A.; Kline, S. R.; Barnes, C. L.; Atwood, J. L. Chem. Eur. J. 2012, 18, 10258–10260. pii:S10258/10251–S10258/10114.
40. Fowler, D. A.; Rathnayake, A. S.; Kennedy, S.; Kumari, H.; Beavers, C. M.; Teat, S. J.; Atwood, J. L. J. Am. Chem. Soc. 2013, 135, 12184–12187.
41. Kumari, H.; Dennis, C. L.; Mossine, A. V.; Deakyne, C. A.; Atwood, J. L. J. Am. Chem. Soc. 2013, 135, 7110–7113.
42. Kumari, H.; Erra, L.; Webb, A. C.; Bhatt, P.; Barnes, C. L.; Deakyne, C. A.; Adams, J. E.; Barbour, L. J.; Atwood, J. L. J. Am. Chem. Soc. 2013, 135, 16963–16967.
43. Kumari, H.; Good, A. J.; Smith, V. J.; Barbour, L. J.; Deakyne, C. A.; Atwood, J. L. Supramol. Chem. 2013, 25, 591–595.
44. Kumari, H.; Jin, P.; Deakyne, C. A.; Atwood, J. L. Curr. Org. Chem. 2013, 17, 1481–1488.
45. Kumari, H.; Zhang, J.; Erra, L.; Barbour, L. J.; Deakyne, C. A.; Atwood, J. L. CrystEngComm 2013, 15, 4045–4048.
46. Drachnik, A. M.; Kumari, H.; Barnes, C. L.; Deakyne, C. A.; Atwood, J. L. CrystEngComm 2014, 16, 7172–7175.
47. Jin, P.; Kumari, H.; Kennedy, S.; Barnes, C. L.; Teat, S. J.; Dalgarno, S. J.; Atwood, J. L. Chem. Commun. 2014, 50, 4508–4510.
48. Kumari, H.; Deakyne, C. A.; Atwood, J. L. Acc. Chem. Res. 2014, 47, 3080–3088.
49. Kumari, H.; Jin, P.; Teat, S. J.; Barnes, C. L.; Dalgarno, S. J.; Atwood, J. L. Angew. Chem. Int. Ed. 2014, 53, 13088–13092.
50. Kumari, H.; Jin, P.; Teat, S. J.; Barnes, C. L.; Dalgarno, S. J.; Atwood, J. L. J. Am. Chem. Soc. 2014, 136, 17002–17005.
51. Kumari, H.; Kline, S. R.; Atwood, J. L. Chem. Sci. 2014, 5, 2554.
52. Patil, R. S.; Mossine, A. V.; Kumari, H.; Barnes, C. L.; Atwood, J. L. Cryst. Growth Des. 2014, 14, 5212–5218.
53. Dalgarno, S. J.; Tucker, S. A.; Bassil, D. B.; Atwood, J. L. Science 2005, 309, 2037–2039.
54. Bassil, D. B.; Dalgarno, S. J.; Cave, G. W.; Atwood, J. L.; Tucker, S. A. J. Phys. Chem. B 2007, 111, 9088–9092.
55. Dalgarno, S. J.; Power, N. P.; Atwood, J. L. Chem. Commun. 2007, 33, 3447–3449.
56. McKinlay, R. M.; Dalgarno, S. J.; Nichols, P. J.; Papadopoulos, S.; Atwood, J. L.; Raston, C. L. Chem. Commun. (Camb.) 2007, 23, 2393–2395.
57. Dalgarno, S. J.; Power, N. P.; Atwood, J. L. Coord. Chem. Rev. 2008, 252, 825–841.
58. Dalgarno, S. J.; Power, N. P.; Warren, J. E.; Atwood, J. L. Chem. Commun. 2008, 1539–1541.
59. Jin, P.; Dalgarno, S. J.; Barnes, C.; Teat, S. J.; Atwood, J. L. J. Am. Chem. Soc. 2008, 130, 17262–17263.
60. Atwood, J. L.; Brechin, E. K.; Dalgarno, S. J.; Inglis, R.; Jones, L. F.; Mossine, A.; Paterson, M. J.; Power, N. P.; Teat, S. J. Chem. Commun. (Camb.) 2010, 46, 3484–3486.
61. Jin, P.; Dalgarno, S. J.; Atwood, J. L. Coord. Chem. Rev. 2010, 254, 1760–1768.
62. Lloyd, G. O.; Atwood, J. L.; Barbour, L. J. Chem. Commun. (Camb.) 2005, 14, 1845–1847.
63. Kumari, H.; Kline, S. R.; Wycoff, W.; Atwood, J. L. Small 2012, 8, 3321–3325.
64. Dalgarno, S. J.; Cave, G. W. V.; Atwood, J. L. Angew. Chem. Int. Ed. 2006, 45, 570–574.
65. Mossine, A. V.; Kumari, H.; Fowler, D. A.; Shih, A.; Kline, S. R.; Barnes, C. L.; Atwood, J. L. Chem. Eur. J. 2012, 18, 10258–10260.
66. Vimalanathan, K.; Gascooke, J. R.; Suarez-Martinez, I.; Marks, N. A.; Kumari, H.; Garvey, C. J.; Atwood, J. L.; Lawrance, W. D.; Raston, C. L. Sci. Rep. 2016, 6, 22865.
67. Wong, E. W. Science 1997, 277, 1971–1975.
68. Liu, L.; Yang, C.; Zhao, K.; Li, J.; Wu, H.-C. Nat. Commun. 2013, 4, 2989.
69. Minati, L.; Speranza, G.; Bernagozzi, I.; Torrengo, S.; Toniutti, L.; Rossi, B.; Ferrari, M.; Chiasera, A. J. Phys. Chem. C 2010, 114, 11068–11073.
70. Tran, M. Q.; Tridech, C.; Alfrey, A.; Bismarck, A.; Shaffer, M. S. P. Carbon 2007, 45, 2341–2350.
71. Wang, C.; Guo, S.; Pan, X.; Chen, W.; Bao, X. J. Mater. Chem. A 2008, 18, 5782.
72. Yudasaka, M.; Zhang, M.; Jabs, C.; Iijima, S. Appl. Phys. Mater. Sci. Process. 2000, 71, 449–451.
73. Zhang, J.; Zou, H.; Qing, Q.; Yang, Y.; Li, Q.; Liu, Z.; Guo, X.; Du, Z. J. Phys. Chem. B 2003, 107, 3712–3718.
74. Ziegler, K. J.; Gu, Z.; Shaver, J.; Chen, Z.; Flor, E. L.; Schmidt, D. J.; Chan, C.; Hauge, R. H.; Smalley, R. E. Nanotechnology 2005, 16, S539–S544.
75. Xie, J.; Ahmad, M. N.; Bai, H.; Li, H.; Yang, W. Sci. China: Chem. 2010, 53, 2026–2032.
76. Livsey, I. J. Chem. Soc., Faraday Trans. 2 1987, 83, 1445.
77. Teixeira, J. J. Appl. Cryst. 1988, 21, 781–785.
78. Kumari, H.; Kline, S. R.; Kennedy, S. R.; Garvey, C.; Raston, C. L.; Atwood, J. L.; Steed, J. W. Chem. Commun. 2016, 52, 4513–4516.
2.15 Computational Studies of Supramolecular Systems: Resorcinarenes
and Pyrogallolarenes
CA Deakyne and JE Adams, University of Missouri, Columbia, MO, United States
Ó 2017 Elsevier Ltd. All rights reserved.

2.15.1 Introduction 303


2.15.2 Background 304
2.15.3 Methodology 307
2.15.3.1 Quantum Chemical Calculations 307
2.15.3.2 MD Simulations 308
2.15.4 Subunit Components: Hydroxybenzenes 308
2.15.5 Macrocycles 310
2.15.5.1 Conformational Analyses 310
2.15.5.1.1 Calixarenes 310
2.15.5.1.2 Pyrogallol[4]arenes 312
2.15.5.1.3 Resorcin[4]arenes 314
2.15.5.2 Metal–Macrocycle Interactions 315
2.15.5.3 Macrocycle–Guest Interactions 318
2.15.5.4 Vase Versus Kite Structures 319
2.15.5.5 Exploiting Chirality 321
2.15.5.6 Organic Crystal Porosity 323
2.15.6 Hydrogen-Bonded Nanocapsules 324
2.15.6.1 Model Complexes 324
2.15.6.2 Nanocapsules 326
2.15.7 Metal-Seamed and Covalently Linked Nanocapsules 329
2.15.7.1 Metal-Seamed Nanocapsules 329
2.15.7.1.1 Zinc Model Complexes 329
2.15.7.1.2 Nanocapsules 332
2.15.7.2 Covalently Linked Nanocapsules 334
2.15.8 Conclusion 337
References 337

2.15.1 Introduction

Computational studies have made valuable contributions to our understanding of the properties of supramolecular assemblies and
their building blocks for several decades. In particular, it has long been recognized1–6 that computational modelingdelectronic
structure calculations and Monte Carlo and molecular dynamics (MD) simulations7–24daffords insight into the nature of the intra-
and intermolecular interactions stabilizing the assemblies, the “communication” between host and guest, the steric and magnetic
environments within the assemblies, the factors that govern templating of the assemblies, and the elementary atomic-level mech-
anisms by which guests enter and leave the assemblies. Furthermore, new criteria for complex design have been identified. Close
collaboration between computational and experimental groups allows immediate feedback that greatly facilitates both the
development of reliable theoretical models for the supramolecular assemblies and a more focused experimental strategy for the
construction and structural modification of the assemblies. Ultimately, a joint approach expedites the discovery of new materials
and methods for controlling their functions.
Collaborative studies pursued by the Deakyne (electronic structure calculations), Adams (MD simulations), and Atwood
(experimental) groups constitute just such a joint effort to understand, extend, and exploit the properties of the first isolated
metal-coordinated pyrogallol[4]arene dimer capsules, synthesized recently by the Atwood group.25,26 Our focus in this work has
been threefold: (1) the practical exploitation of the ability of macrocyclic compounds to both recognize and selectively bind guest
species, (2) the attractive prospect of assembling these compounds into superstructures with unique physical properties, and (3) the
synergistic interplay of theory and experiment in elucidating these properties and in optimizing the overall research strategy.
Because we have concentrated on computational studies of supramolecular systems constructed from pyrogallol (1,2,3-
trihydroxybenzene) and, to a lesser extent, resorcinol (1,3-dihydroxybenzene), the primary focus of this article will be the macro-
cycles, hydrogen-bonded and metal-seamed host–guest complexes, and one-dimensional (1D) and two-dimensional (2D) arrays
formed from these two subunit components. We will discuss the impact on host–guest interactions and on the communication
process between the host and guest of (1) the orientation and restricted motion of the guest; (2) the nature of hydroxyl group

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12496-5 303


304 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

substituents, R-group moieties, and metal ligands; and (3) the flexibility of both the macrocycles themselves and the assembled
capsules. The transmission of electronic information between the exterior and interior of an individual capsule and the use of
nascent capsules as templates for preparation of connected chains of capsules also will be addressed. For completeness and to high-
light additional applications of computational studies to problems in host–guest chemistry, we include here as well some related
results on supramolecular systems derived from phenol (hydroxybenzene).

2.15.2 Background

Macrocycles of the calixarene family consist of phenolic units linked by eCHR moieties that add to carbon sites lying ortho to
a hydroxyl substituent and 1,3 to each other on the ring. Here, R denotes an alkyl or an aryl group. Calixarene macrocycles and their
derivatives are denoted calix[n]arene, where n is the number of phenolic units. Values of n of 4, 5, 6, and 8 have been observed
depending on the calixarene family-based building block.27–30 Examples of the common building blocks, phenol (left), resorcinol
(middle) and pyrogallol (right), and the most common macrocycles, calix[4]arene (left), resorcin[4]arene (middle), and pyrogallol[4]
arene (right), can be found in Fig. 1. The hydroxyls form the smaller, lower rim of calix[4]arene and the larger, upper rim of both
resorcin[4]arene and pyrogallol[4]arene. When the macrocycles assume a chalice shape, these hydroxyl groups form a hydrogen-
bonded network. The eight-membered hydrogen-bonded ring at the lower rim of calix[4]arene limits its flexibility compared
with that of the other two polyarenes. For this reason, resorcinol- and pyrogallol-based macrocycles can capture a more sizeable
guest or multiple smaller guests.
Because the chalice shape of the cone conformer (also called a cavitand) maximizes the potential both for entrapment of a guest
and for nanocapsule formation, this conformer of the calixarene family of macrocycles has received particular attention. However,
these macrocycles can exist in a variety of conformations, defined by the relative orientation of the four aryl rings (Fig. 2).27,31–36 In
the partial cone conformation, one aryl ring has rotated roughly 180 degrees from its position in the cone conformation. In the boat
conformation, two opposing aryl rings have rotated about 90 degrees from their positions in the cone conformation. If the rings
instead rotate by about 180 degrees the 1,3-alternate (saddle) conformation is obtained. In the 1,2-alternate (diamond) conformation,
two adjacent aryl rings have rotated about 180 degrees from their positions in the cone conformation. If two aryl rings rotate by
about 90 degrees and one rotates by about 180 degrees, the chair conformation is obtained.
In addition to the conformation of the four pyrogallol rings, the stereoisomers of the resorcinarene and pyrogallolarene
macrocycles are defined by the relative configuration (rccc, rctt, rtct, or rcct) and individual configuration (axial or equatorial) of
the R-groups. In this notation, the r denotes a reference axial pendent R-group; moving in a consistent direction (clockwise or
counterclockwise), one then designates the remaining pendent groups as being oriented either cis (c) or trans (t) to the reference
group (Fig. 2).
The distribution of stereoisomers found in solution and the solid state depends on the R-group, solvent, solubility of the constit-
uents, and reaction time. The most prevalent forms of the resorcin[4]arenes are the all-axial rccc cone, rccc boat, and rctt chair,27,31–38
although an rctc 1,2-alternate39 and an rcct diastereomer believed to have a partial cone or 1,2-alternate conformation also have been

Figure 1 Calixarene family-based building blocks and macrocycles formed from them. Left, top, phenol and bottom, calix[4]arene; center, top,
resorcinol and bottom, resorcin[4]arene; right, top, pyrogallol and bottom, pyrogallol[4]arene.
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 305

Figure 2 From top to bottom and left to right, (A) possible resorcin[4]arene (and pyrogallol[4]arene) conformers, cone, boat, 1,3-alternate (saddle),
partial cone, chair, and 1,2-alternate (diamond) and (B) possible diastereomers, moving clockwise from the top left, rccc, rcct, rtct and rctt. Hydrogen
atoms have been removed for clarity. Color scheme, C, gray; O, red.

observed.40 On the other hand, the most prevalent forms of the pyrogallol[4]arenes are the all-axial rccc cone and rctt chair,41–46 with
the rccc boat also having been observed.47
One important factor in the stability of the rccc cone stereostructure of pyrogallol[4]arenes and resorcin[4]arenes is the cooper-
ative hydrogen-bonding network between the pyrogallol and resorcinol hydroxyl groups at the upper rim of the macrocycles.
Because this hydrogen-bonding network is disrupted in the other stereostructures, other factors, including CeH/O, OeH/p,
CeH/p, and p–p stacking interactions, yield larger influences on their stabilities. Reduced steric congestion at the lower rim of
the macrocycle and more favorable macrocycle–solvent interactions can also preferentially stabilize these other stereoisomers
with respect to the desired rccc cone. The different three-dimensional (3D) geometries and, consequently, the different complexa-
tion behaviors of these various stereoisomers afford additional opportunities for their use as molecular scaffolds and supramolec-
ular building blocks.
Dynamic supramolecular systems that can serve as metal-based catalysts,48 molecular motors,49 receptors,50 sensors,51 and
switches52 are of widespread interest. Among the earliest characterized resorcin[4]arene-based dynamic systems is the
quinoxaline-bridged resorcin[4]arene molecular cavitand synthesized by Cram and coworkers53–55 (Fig. 3). Molecular basket-
type containers of this type, formed from resorcin[4]arene derivatives and including a rigid bridge spanning upper rim hydroxyl

Figure 3 (A) Vase and (B) kite conformer of the quinoxaline-bridged resorcin[4]arene molecular cavitand introduced by Cram and coworkers.53–55
Hydrogen atoms have been removed for clarity. Color scheme, N, blue.
306 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

groups on adjacent resorcinols, exhibit either a closed vase or an open kite conformation.54,56–59 Rapid switching between these two
conformations can occur in solution,60,61 however, depending on the pH,62 the temperature,54,63 or the presence of Zn2 þ ions64:
lowering the temperature or the pH and adding Zn2 þ ions favor conversion to the kite conformer. (The repulsion of the bridging
moieties stemming from protonation of the container shifts the equilibrium toward the kite.) Host–guest complexes have been con-
structed from both conformers, but rigidifying the cavitand into the closed vase conformation and deepening the cavity have been
shown to strengthen guest binding and improve selectivity.60,65–68 Elusive reaction intermediates also have been confined and
observed within the cavity of suitably bridged resorcin[4]arene “molecular baskets.”69–71
Another attractive feature of resorcin[4]arenes, either as macrocycles or incorporated into the nanocapsules described in the suc-
ceeding text, is that chirality can be introduced through incorporation of stereogenic centers in the R-groups or of achiral subunits
with a restricted 3D arrangement that yields a chiral macrocyclic framework.39,72,73 Such resorcin[4]arenes have been investigated as
potential biomimetic receptors for guests such as amino acids and pharmaceuticals, with particular interest lying in the efficient
chiral recognition of the guest by the macrocyclic host. Frequently, such studies have been done in gas-phase systems, thereby elim-
inating perturbing effects from the solid-state or solution-phase medium.
Supramolecular nanocapsules are relatively new yet highly promising species with applications in several areas of chemical
research.74,75 Understanding and manipulating the enclosed chemical spaces associated with these nanocapsules foretell many
exciting advances in the areas of fuel storage, nanomedicine, magnetism and optics, organic nanotubes, and confined nanoscale
reactions. As molecular scaffolds, the rccc cone stereoisomers of both resorcin[4]arenes and pyrogallol[4]arenes self-assemble to
form homo- or heterocapsules stabilized by hydrogen bonds, covalent bonds, or metal–ligand coordination.6,25,26,42,76–79 Both
dimeric and hexameric nanocapsules have been synthesized from these two cavitands.74,79–96 Note, however, that the hydrogen-
bonded resorcin[4]arene hexameric nanocapsules require the presence of eight water molecules in the framework for stability,
whereas the pyrogallol[4]arene hexameric nanocapsules do not.81–84,86,88 The 24 intramolecular and 48 intermolecular hydrogen
bonds among the six pyrogallolarene macrocycles are sufficient to stabilize the latter hexamer. In the hydrogen-bonded dimers built
from the resorcin- and pyrogallolarenes, the two macrocycles are often bound together via interactions involving multiple solvent
molecules or other types of divergent synthons.84,87,89,91,93,97,98 The pyrogallol[4]arene-based metal–organic nanocapsules
(MONCs) are held together by eight metals arranged along the equator of the capsule (dimer)25,26,79 or by 24 metals arranged
in eight planar (M–O)3 arrays (hexamer)77,94,99–101 (Fig. 4). Direct coordination of a metal to the upper rim hydroxyl groups of
resorcin[4]arenes is uncommon, though, and when it does occur, it leads to coordination complexes involving one or two
cavitands.102–104 In the resorcin[4]arene-based metal-coordinated dimeric nanocapsules synthesized to date, one often finds that

Figure 4 Pyrogallol[4]arene MONCs, (A) hexamer formed from six pyrogallol[4]arene macrocycles and twenty-four metal ions, (B) hexamer metal
coordination, (C) dimer formed from two pyrogallol[4]arene macrocycles and eight metal ions, and (D) dimer metal coordination. Hydrogen atoms
have been removed for clarity. Color scheme, metal2 þ, teal. Adapted from Power, N. P.; Dalgarno, S. J.; Atwood, J. L. New J. Chem. 2007, 31, 17–
20; McKinlay, R. M.; Cave, G. W. V.; Atwood, J. L. Proc. Natl. Acad. Sci. U. S. A. 2005, 102, 5944–5948.
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 307

hydroxyl groups on adjacent resorcinols are bridged and the metal is ligated to a functionalized substituent at the 2-position of the
resorcinol.97,98,105–107 A second possible coordination site, one that has also been used in forming hydrogen-bonded and covalently
bound dimers, is a functionalized substituent that bridges a pair of hydroxyl groups.108–111
The possibility of constructing metal-seamed dimeric nanocapsules from the self-assembly of rctt chair stereostructures is much
less explored. Maerz and coworkers44,47 have shown, however, that following the usual protocol for capsule formation, in the pres-
ence of Zn2 þ, the C-phenyl-, C-naphthyl-, and C-3,4,5-trimethoxyphenylpyrogallol[4]arene macrocycles convert from the chair
conformer to the cone conformer to yield the zinc-seamed dimer. Castellano et al.112 have likewise reported a change from the
partial cone to the cone conformation of tetraurea calix[4]arene induced by intermolecular hydrogen bonding, and Frischmann
et al.113 have reported a change from planar to bowl-like tris(salphen) [3 þ 3] Schiff base macrocycles induced by the addition
of Cd2 þ. Nevertheless, the substitution pattern on the phenyl ring does appear to influence capsule synthesisdsome of the C-aryl-
pyrogallol[4]arene macrocycles examined could not be induced to undergo the conformational change.47 Here, as in many of the
other experimental studies described earlier, uncertainties remain as to why certain structures form readily whereas others do not,
why certain structures undergo conformational interconversions whereas others do not, and why certain structures constitute
building blocks for supramolecular complexes whereas others do not. Efficiently sorting out the various fundamental effects that
give rise to the wealth of behavior described to date requires additional input, just the sort of input afforded by computational
studies, to which we now turn our attention.

2.15.3 Methodology
2.15.3.1 Quantum Chemical Calculations
Recently, Waller et al.114 published a tutorial review of the quantum chemical protocol recommended to yield reliable results for
supramolecular assemblies. They discussed the need to compute (1) fully optimized geometries, (2) vibrational frequencies, (3)
single-point energies, (4) basis set superposition error (BSSE) corrections, (5) dispersion corrections, and (6) solvation effects.
They also stressed the need to calibrate the level of theory at which the quantum chemical calculations are performed. We expand
on these points in the succeeding text, recognizing that it is not appropriate here to delve into the fundamentals of modern
electronic structure theory or computational techniques, topics that are discussed in considerable detail in several excellent
monographs.115–119
Unless doing so is precluded by the system size, the system of interest always should be completely optimized if one is to obtain
reliable geometric and energetic properties. Normal-mode vibrational frequencies are then calculated for this optimized structure to
verify that the structure is indeed a minimum on the potential energy surface (no imaginary frequencies) rather than a transition
structure (exactly one imaginary frequency) or a higher-order saddle point (two or more imaginary frequencies). The vibrational
frequency calculation also provides the thermal correction terms used to convert the total energy evaluated in the calculation
(0 K, atoms at rest) to an enthalpy and a free energy at temperature T. At this point, in general, density functional theory (DFT)
is utilized to carry out these two steps of the calculational protocol (e.g., B3LYP,120,121 PBE0,122 BP86,123,124 M05-2X,125
M06,126 M06-2X,126 M06-L,127 and MPW1PW91128).
It has been established for some time that geometric parameters usually converge to their “exact” values at lower levels of theory
than do total energies. Thus, it is common to determine energies at higher levels of theory on geometries obtained at lower levels of
theory. The notation used to designate the calculation of a single-point energy is, for example, MP2/6-311G(d,p)//B3LYP/6-31G(d),
indicating that the energy has been determined at the MP2/6-311G(d,p) level of theory using a geometry optimized at the B3LYP/6-
31G(d) level. The use of these single-point energies improves the obtained thermochemical data, as does application of the BSSE
correction. The BSSE is the overestimation of the interaction energy between two subunits stemming from the use of finite basis sets.
(The basis functions centered on one of the subunits are effectively “stolen” to improve the description of the other subunits, and
vice versa, an appropriation leading to an inconsistent treatment of the basis set for each subunit as the intermolecular distance is
varied.) The counterpoise method is frequently applied to eliminate the BSSE.129
A number of studies have shown the importance of including electron correlation effects in the calculations when investigating
weak, noncovalent interactions.130–133 Not including these effects leads to both incorrect equilibrium structures and unreliable
energetics. The preferred way to incorporate electron correlation into the calculations is to use post-Hartree–Fock (HF) wave func-
tion theory (WFT) methods (e.g., MP2134 and CCSD(T)135–140). More commonly, however, the size of supramolecular systems
dictates the use of dispersion-corrected density functional theory (DFT-D) methods (e.g., B97-D,141 uB97X-D,142 PBE-D3,143,144
and TPSS-D3144,145) to estimate these effects.146
Hybrid methods, such as the ONIOM method,147,148 have been used to make post-HF WFT calculations more tractable for
supramolecular systems. In these methods, the system of interest is subdivided into two or three layers that are treated with progres-
sively lower levels of calculation. The components of the system most critical to the property under consideration are placed in the
highest layer; the remaining components are relegated to the lower layer(s). Thus, one of the most important components of the
hybrid approach is a calibration that establishes the assignment of the atoms of the system to the various layers.
Because many experimental studies on supramolecular systems are carried out in solution, solvent effects need to be taken into
account if one wishes to reproduce experimental data. Including explicit solvent molecules quickly becomes unwieldy in quantum
chemical calculations; thus, it is more common to include solvent effects implicitly using a self-consistent reaction field (SCRF)
308 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

model. In fact, it is often sufficient just to evaluate single-point energies in the presence of the implicit solvent, as it has been shown
that reoptimizing geometries in the presence of an implicit solvent usually has little effect on thermochemical trends.
The polarizable continuum model (PCM)149–152 is a frequently used SCRF approach to the modeling of solvation effects. In this
method, the solvent is treated as a continuous medium characterized by a solvent-specific dielectric constant within which a cavity
containing the solute is created. The IEF–PCM variant of this method considers this solute cavity to be formed by overlapping
nuclear-centered spheres of appropriate radii.153 The solvation free energy is then based on the bulk electrostatic interaction
between the solute and the SCRF of the solvent. The SMD method154 extends the IEF–PCM method and improves the solvation
free energy by adding a correction for short-range interactions (i.e., within the first solvation shell) between the solute and the
implicit solvent field based on the quantum mechanical charge density. This contribution is proportional to the degree to which
the surface area of the solute atoms is accessible to the solvent.
Perhaps the most important step in performing electronic structure calculations, although one that is too often overlooked, is the
calibration of the level of theory used in carrying out the calculations. Choosing a level of theory without prior calibration or justi-
fication makes it impossible to judge the quality of the results. The components varied in a well-crafted calibration study include the
calculational method (DFT or WFT), the basis set, and, if relevant, the effective core potential (ECP) on the metal atoms. Some
groups perform the calibration study by carrying out calculations at a variety of levels of theory and demonstrating that the trends
are independent of the choice of level. Others perform the study by carrying out a series of calculations on model systems for which
the results can be benchmarked against the results from “state-of-the-art” calculations. Still, others choose a calculational level that
has been calibrated for related systems and, if possible, show that the chosen level yields results that agree well with those obtained
from experiment. For more information, see the recent articles by Murphy and coworkers155,156 and by Tzeli and Rebek,157,158
wherein they have detailed the step-by-step procedures used to calibrate their work. Additional information is also provided in
the tutorial review by Waller et al.114
Unless stated otherwise, in all the quantum chemical calculations discussed in the succeeding text, structures were fully opti-
mized, and normal-mode vibrational frequencies were obtained for each equilibrium structure. Implementation of the other factors
described earlier will be addressed individually for each study.

2.15.3.2 MD Simulations
Of course, the quantum chemistry calculations discussed in the aforegoing section yield only static molecular structures, whereas
one also would like to probe the dynamics and thermal stability of various capsules and host–guest complexes. Ideally, these latter
calculations would proceed by determining the forces on each atom in the system at any instant of time from the sort of high-quality
quantum chemistry calculation discussed earlier, but in practice, such a strategy is prohibitive in systems as large as those considered
herein. One thus must retreat to the use of empirical force fields, such as the well-known AMBER force parameterization.159,160 The
general AMBER force field (GAFF) parameter set (augmented with, e.g., ff03 parameters should there be gaps in the GAFF
parameterization) provides an appropriate starting point, but in fact, no established force field will account for every bonding
environment encountered. In such situations, necessary parameters can be derived by fitting to ab initio results in accordance
with well-documented procedures.161,162 Often, though, a more appropriate strategy involves QM/MM simulations, wherein the
problematic bonding environments are assigned to a contiguous (semiempirical) quantum mechanical region of the simula-
tion.163–167 Readily available software suites such as AMBER168 (the package used in the present studies) or DL_POLY169 are easily
adapted to carry out these studies, with the latter software package providing additional force field functional forms and ensemble
choices. The interested reader is referred to several standard literature sources describing further details of the process by which one
carries out MD simulations corresponding to various thermodynamic states.118,119

2.15.4 Subunit Components: Hydroxybenzenes

Although resorcinol and pyrogallol, as well as phenol, readily form self-assembled macrocycles, formation of macrocycles from the
other di- and trihydroxybenzenes has not yet been observed. Because the acid-catalyzed synthesis of calix[4]arenes, resorcin[4]are-
nes, and pyrogallol[4]arenes has been suggested to proceed via electrophilic aromatic substitution,40 we used proton affinity (PA)
data, together with the activating and ortho, para-orienting influence of hydroxyl groups in this type of reaction, to probe the likeli-
hood that hydroxybenzene building blocks link to give macrocycles. In addition, we wished to improve our understanding of the
origin of the observed placement of the eCHR linkers on the hydroxybenzene rings and of the observed preference for the hydroxyl
groups being at the upper rim versus the lower rim of the macrocycles.
Experimentally, it has been shown that the mono-,170–173 di-,170,173 and trihydroxybenzenes174 preferentially protonate at
a carbon center rather than an oxygen center, as one might naively expect. The combined experimental and computational studies
of phenol170–173,175,176 and the dihydroxybenzenes170,173 have determined the magnitude of the PA, the preferred carbon
protonation site, and the relative proton affinities among the unique protonation sites. However, no such data were available
for the trihydroxybenzenes; hence, we began our investigations by carrying out quantum chemical calculations on
hydroxyquinol177 (1,2,4-trihydroxybenzene) and pyrogallol.178
The PA data were evaluated at a variety of levels of theory, directed by the studies on phenol and the dihydroxyben-
zenes170,175,176 and including the G4(MP2) level.179 The magnitudes of the calculated PAs varied from level to level as expected,
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 309

but the trends in the relative PAs did not.177 Consistent with the ortho, para-orienting influence of the hydroxyl groups, hydroxyqui-
nol is preferentially protonated at the C5 position, that is, at the carbon that lies ortho, meta, and para to the OH substituents (Fig. 5).
The second most favorable protonation site is C3, the site positioned ortho and meta to the hydroxyl groups. Although protonation at
the ring disrupts the ring aromaticity, it is favored over protonation at oxygen because electron density donated from the hydroxyl
groups to the ring stabilizes the resultant carbocation, whereas the reverse stabilizing electron donation does not occur. Specifically,
the PA of hydroxyquinol is 868.0  1.2 kJ mol 1, as determined by averaging the G4(MP2) PAs calculated from a series of isodesmic
reactions; the PA associated with protonation at C3 (rather than at C5) is some 40 kJ mol 1 lower.
Insight into the formation of macrocycles from hydroxybenzene subunits was gained by combining the PA data for pyrogallol
from our work178 with that for phenol and resorcinol as reported by Bouchoux and coworkers.170 On the basis of these data, we
proposed that if the two most favorable ring protonation sites lie ortho to a hydroxyl group, lie 1,3 with respect to each other, and
have PAs within 5–10 kJ mol 1 of each other, formation of macrocycles is feasible.177 For resorcinol and pyrogallol, locating the
eCHR linking groups at the two carbon sites with the highest preference for protonation places the hydroxyl groups at the upper rim
of the macrocycle. Linking the phenol subunits at the carbon sites that are ortho to the hydroxyl group, however, places the hydroxyl
groups at the lower rim of the macrocycle. Flipping the positions of the OH groups in calix[4]arene would necessitate connecting the
phenols at the 1,3 carbons that are meta to the OH. However, protonation at the carbons meta to the OH is some 55 kJ mol 1 less
stable than protonation at the carbons ortho to the OH.170 Even disregarding the first two criteria noted earlier for catechol (1,2-
dihydroxybenzene), hydroquinone (1,4-dihydroxybenzene), and hydroxyquinol, we find that the final criterion is not met for
any of the three subunits.170,177 The orientation of the relevant carbons with respect to the hydroxyl groups is ortho and para in resor-
cinol versus ortho and meta or meta and para in catechol and hydroquinone; the orientation is ortho, meta and para in pyrogallol
versus ortho, meta, and para and ortho, ortho, and para in hydroxyquinol. Overall, these results suggest that cyclization to form as
yet unknown calixarene-related macrocycles may be unfavorable for these particular subunits.177
Because dihydroxybenzene systems have been used as bioreductive anticancer agents and because their efficacy is related to their
standard reduction potentials E ,180 Liu et al.181 have determined E values for hydroquinone and catechol and for their mono- and
disubstituted eF, eCl, eOH, eNO2, eNH2, eCOOH, and eCN derivatives. The reduction potentials were evaluated computation-
ally, via quantum chemical calculations182 at the B3LYP/6-311 þ G(d,p) level of theory, for all of the species, and, experimentally,
via cyclic voltammetry (CV) measurements, for hydroquinone and catechol. The implicit water calculations required to determine
the DGsolv contribution to E were carried out with the conductor-like polarizable continuum model (CPCM)183,184. The calculated
E values for the latter two species reproduced the experimental values to within 0.003 V, with values of 0.693 versus 0.694 V for
hydroquinone and 0.806 versus 0.803 V for catechol.181 All possible substitution patterns were considered, but the trends were
found to be independent of the positions of the substituent(s). The eOH and eNH2 substituents lowered the reduction potential
of both dihydroxybenzenes, whereas the remaining five substituents raised it. The decrease in the potential was attributed to p–p

Figure 5 The most stable structure of 1,2,4-trihydroxybenzene (hydroxyquinol) indicating the numbering scheme for the protonation sites. Adapted
from Mayhan, C. M.; Kumari, H.; McClure, E. M.; Liebman, J. F.; Deakyne, C. A. J. Chem. Thermodyn. 2014, 73, 171–177.
310 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

interactions between the substituent and aryl ring that destabilize the anion formed in the first step of the overall oxidation process;
the increase was attributed to stabilizing p–p conjugating and electronegativity effects. They also observed a linear relationship
between E and MP2/6-311 þ G(d,p)//B3LYP//6-311 þ G(d,p) energy of the lowest occupied molecular orbital (LUMO) of the
reduced hydroxybenzene in aqueous solution. This relationship is significant in that it can be used to estimate E values for related
hydroxybenzene derivatives.
In other recent work, Shenghur et al.185 performed quantum chemical calculations,182 at the B3LYP/6-311þþG(d) level of
theory, to investigate the reaction of NO2 with phenol and substituted phenols to produce HONO. The study was undertaken
to help elucidate the contribution of ground sources to the concentration of HONO in the atmosphere. Humic acids in soil are
believed to be a key factor in HONO production, and phenols, which have been shown to convert NO2 to HONO,186 are key
components of humic acids. Thus, Shenghur et al.185 investigated the dark, gas-phase pathways and overall thermodynamics of
the earlier proton abstraction reaction for phenol and for a series of mono- (Eq. 1) and disubstituted phenols (eOH, eCH3,
eOCH3, eCO2H, eNHCH3, and eNH2). The following general, anhydrous pathway was considered: separated reactants / hydro-
gen-bonded reactant complex / transition structure / hydrogen-bonded product complex / separated products. Four reaction
mechanisms were identified, including hydrogen atom transfer and proton-coupled electron transfer187 mechanisms. Electron-
donating substituents located ortho and/or para to the hydroxyl group stabilize the phenoxy radical product of the proton abstrac-
tion reaction; hence, the dark, anhydrous reaction is fast and favorable for those phenols.185 The results suggest that NO2 should be
converted readily to HONO by both humic and fulvic acids.
RC6 H4 OH þ NO2 /RC6 H4 O, þ HONO (1)

2.15.5 Macrocycles

The calixarene, resorcinarene, and pyrogallolarene macrocycles, especially when found in a cup-shaped cone conformation, would
be intrinsically interesting due to their ability to sequester small molecules within their cavities through various nonbonding inter-
actions even if they found no use as nanocapsule precursors. Of course, the syntheses of these species need not necessarily yield
exclusively those conformers desired for guest binding, and thus, one would like to determine the reaction conditions that maxi-
mize the yield of the cone structures. With this goal in mind, various research groups have conducted numerous computational
studies focusing on assessments of the relative stabilities of the several calixarene, resorcinarene, and pyrogallolarene conformers;
of the strength of the binding between these host species and a series of guest molecules; and of the efficiency of chiral recognition in
suitably functionalized resorcinarenes. Also examined, via both quantum chemistry studies and MD simulations, is the effect that
metal binding exerts on conformational preference in these systems and the effect of a guest on the preference for the vase versus the
kite conformation of substituted resorcinarenes having deeper cavities.

2.15.5.1 Conformational Analyses


2.15.5.1.1 Calixarenes
The n ¼ 4, 5, and 6 calixarenes were the first of the three classes of macrocycles to be studied computationally to assess conforma-
tional stability.1,7,8,11,188–191 An overview of the experimental and computational work on calixarenes and thiacalixarenes, in the
latter of which a sulfur atom replaces the methylene bridging group, can be found in a recent article by Khedkar and coworkers.189
Investigations of host–guest complexes formed by calixarenes and thiacalixarenes also are summarized in the article.
For the calix[4]arenes, thiacalix[4]arenes, and calix[5]arenes, the most prevalent conformer observed experimentally is the cone,
whereas the most prevalent conformers found for the calix[6]arenes are boatlike (pinched cone) and 1,2,3-alternate-like struc-
tures.27 The first two of these conformers maximize the number of intramolecular hydrogen bonds (4–6); that number is reduced
by two in the last conformer. In contrast, p-tert-butylthiacalix[4]arene adopts a 1,3-alternate conformation both in the solid and in
solution.27 Difficulties in synthesis and selective derivatization have limited the experimental analysis of the n ¼ 5 systems. In
general, however, regardless of the value of n, the thiacalixarenes are found to be more conformationally flexible in solution
than are the corresponding calixarenes.27
In two of the earlier computational studies on calixarenes, Bernardino and coworkers investigated the cone, partial cone, 1,2-
alternate, and 1,3-alternate conformations of calix[4]arene1 and thiacalix[4]arene.188 Optimized structures were obtained at either
the B3LYP/6-31G(d)1 or B3LYP/6-31G(d,p)188 levels of theory, and B3LYP/6-31G(d,p)//B3LYP/6-31G(d) single-point energies
were subsequently computed.192,193 Both the relative energies and the electrostatic potentials of these conformers were evaluated,
the latter calculations being done with and without the hydroxyl groups at the lower rim to obtain information on the electrostatic
stabilization of charged species interacting with the macrocycles. They found that larger cations are likely to interact more strongly
with thiacalix[4]arene than with calix[4]arene, an observation consistent with the larger cavity volume in the thiacalix[4]arene cone
than in the calix[4]arene cone. Bernardino et al.1 also calculated a PA for calix[4]arene of 917.6 kJ mol 1 at the B3LYP/6-31G(d)
level of theory; however, it is important to note that they considered only oxygen protonation sites, whereas carbon protonation is
likely to be more favorable here, as it is in the case of phenol.170–173,175,176
The reported trend in (decreasing) stability found for the conformers of calix[4]arene is cone > partial cone > 1,3-alternate  1,2-
alternate1 (Table 1). The corresponding trend for thiacalix[4]arene, however, is cone > partial cone > 1,3-alternate > 1,2-alternate188
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 311

Table 1 Relative energies (DE ’s) of the calix[4,5]arene and thiacalix[4,5]arene conformers

DE (kJ mol 1)
Macrocycle Substituent Cone Partial cone 1,2-Alternate 1,3-Alternate
a
Calix[4]arene H 0.0 43.9 77.0 73.6
Hb 0.0 38.4 50.9 65.7
tert-Butylb 0.0 38.3 50.6 64.6
tert-Butylc 0.0 35.6 48.5 59.4
CH2Clb 0.0 42.5 51.7 64.3
NH2b 0.0 37.4 51.3 66.4
NO2b 0.0 35.2 46.7 60.1
SO3Hb 0.0 26.1 36.9 54.3
Thiacalix[4]arene Ha 0.0 42.3 65.3 51.0
Hb 0.0 38.0 55.6 51.1
tert-Butylb 0.0 38.0 55.4 55.6
Calix[5]arene Hb 0.0 37.9 18.1 63.9
tert-Butylb 0.0 38.7 19.7 65.2
tert-Butylc 0.0 40.6 21.3 63.9
CH2Clb 0.0 31.1 18.2 65.0
NH2b 0.0 36.5 64.4 62.7
NO2b 0.0 35.9 51.6 60.1
SO3Hb 0.0 23.4 17.7 12.0
Thiacalix[5]arene Hb 0.0 1.7 3.9 8.5
tert-Butylb 0.0 0.6 7.7 21.3
a
Refs. [1,188] B3LYP/6-31G(d,p) calculations.
b
Ref. [189] B3LYP/6-311G(d,p) calculations.
Ref. [191] B3LYP/6-31 þ G(d,p) calculations.
c

(Table 1). That the 1,3-alternate conformer is selectively stabilized (by 15–20 kJ mol 1) with respect to the cone conformer for thia-
calix[4]arene is consistent with the greater flexibility of the thia-substituted system, which has been observed experimentally.27
Somewhat later, Choe and coworkers190,191 reconsidered the preferred conformations of calix[n]arenes, where n ¼ 4, 5, and 6.
The cone, partial cone, 1,2-alternate, and 1,3-alternate conformations of p-tert-butylcalix[n]arene were again studied for n ¼ 4 and 5;
for n ¼ 6, the pinched cone, winged cone, partial cone, 1,2-alternate, 1,3-alternate, 1,4-alternate, 1,2,3-alternate, and 1,3,5-alternate
conformations were studied for both calix[6]arene and p-tert-butylcalix[6]arene. Each structure was optimized at the B3LYP/6-
31G(d,p) and B3LYP/6-31 þ G(d,p) calculational levels,193,194 but the authors did not report having calculated vibrational frequen-
cies. MPW1PW91/6-31G(d,p)//B3LYP/6-31G(d,p) single-point energies were computed for the n ¼ 6 systems.190 Arranged from
most stable to least stable, the four conformers of p-tert-butylcalix[4]arene have B3LYP/6-31 þ G(d,p) relative energies (DE’s) in
the following order: cone > partial cone > 1,2-alternate > 1,3-alternate (Table 1). For p-tert-butylcalix[5]arene, though, the order
is cone > 1,2-alternate > partial cone > 1,3-alternate.191
For the calix[6]arene, the MPW1PW91/6-31G(d,p)//B3LYP/6-31 þ G(d,p) DE values show the following trend (kJ mol 1):
pinched cone (0.0) > partial cone (45.2) > winged cone (61.1) z 1,2-alternate (61.9) z 1,2,3-alternate (64.4) > 1,4-alternate
(80.3) > 1,3-alternate (86.2) > 1,3,5-alternate (104.6).190 The MPW1PW91/6-31G(d,p)//B3LYP/6-31G(d,p) calculated trend is
different, though, for p-tert-butylcalix[6]arene: pinched cone (0.0) > 1,2-alternate (39.3) > winged cone (55.6) > 1,4-alternate
(68.6) z partial cone (69.5) > 1,2,3-alternate (83.7) > 1,3,5-alternate (99.2) > 1,3-alternate (105.9). The largest differences
observed upon addition of the p-tert-butyl groups to calix[6]arene are thus a destabilization of the partial cone and 1,2,3-
alternate conformers and a stabilization of the 1,2-alternate conformer.
In all the earlier systems, an analysis of their intramolecular hydrogen-bonding interactions reveals a direct correlation between
the number and types of hydrogen bonds and the relative conformer energies.190,191 This result is supported by the observation that
the calculated average O/O hydrogen bond lengths in the cone conformers of the calixarenes agree to within 0.15 Å with those
extracted from X-ray diffraction (XRD) data.195,196
Substituent effects on the preferred conformations of calix[4,5]arenes and their thia analogs also have been explored by Khedkar
et al.189 at the B3LYP/6-311G(d,p) level of theory.182 In that work, hydrogen was replaced by NH2, NO2, SO3H, tert-butyl, or CH2Cl
in the case of the calixarenes and by a tert-butyl group in the case of the thiacalixarenes. As usual, these replacements were made at
the ring position para to the hydroxyl group. The cone again is found to be the preferred conformer for all of the systems, although
the other conformers are much more competitive with respect to the cone conformer in the thiacalix[5]arenes (Table 1). For the
calix[4]arenes and thiacalix[4]arenes, with the single exception of SO3H, the substituents have little effect on the magnitudes of
or trends in the relative energies, despite the inclusion of both electron-donating and electron-withdrawing substituents in this
study. For example, the relative conformer stabilities for calix[4]arene and p-tert-butylcalix[4]arene follow the same orders noted
earlier, namely, cone > partial cone > 1,2-alternate > 1,3-alternate. On the other hand, the order for thiacalix[4]arene and p-tert-
butylthiacalix[4]arene is cone > partial cone > 1,3-alternate  1,2-alternate. For the calix[5]arenes, however, four different trends
312 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

are observed for the six macrocycles. The substituents that bind through a carbon atom exhibit the same trend as the parent system
(cone > 1,2-alternate > partial cone > 1,3-alternate), whereas the remaining three substituents yield three different trends. An
entirely different stability order is then observed for the thiacalix[5]arenes: cone z partial cone z 1,2-alternate  1,3-alternate.189
Several generalizations can be extracted from these studies summarized in Table 1. First, the results from the two higher levels of
theory (B3LYP/6-311G(d,p) and B3LYP/6-31 þ G(d,p)) agree within 5 kJ mol 1. The results from the lowest level of theory, though,
show poorer agreement; they deviate from those of the two higher levels by as much as 25 kJ mol 1. The 1,2-alternate conformation
in particular appears to be destabilized at the lowest level, a result sufficient to alter the conformer stability order in some cases.
Second, replacing the CH2 linker moiety with an S atom stabilizes the 1,3-alternate conformation relative to the other conforma-
tions. Third, the substituents impact the relative energies of the n ¼ 5 and 6 systems to a much greater extent than they do those of
the n ¼ 4 systems. Fourth, although the values collected in Table 1 are gas-phase DE values and not solution-phase DG values, the
results nonetheless predict a preference for the cone conformer of the calix[4]arenes, thiacalix[4]arenes, and calix[5]arenes and
demonstrate a much stronger competitiveness among all the conformers in the thiacalix[5]arenes.

2.15.5.1.2 Pyrogallol[4]arenes
The main stereoisomeric product observed experimentally for alkyl-substituted pyrogallol[4]arenes is the all-axial rccc
cone.41,43,45,74,78,79,84,94,197 Indeed, the preference for the cone increases as the alkyl group lengthens, a trend that has been attrib-
uted to steric effects.43 In contrast, Maerz et al.44 have demonstrated in a combined computational and experimental study that the
rctt chair is the most stable conformer of phenyl-substituted pyrogallol[4]arenes, a result disagreeing with conclusions from earlier
studies on resorcin[4]arenes27,31–33,35–37 that when the macrocycle forms, the chair is the kinetic product but the cone is the ther-
modynamic product, whether the R-group is an alkyl or an aryl group. The results of Maerz et al.44 for C-arylpyrogallol[4]arenes are
supported, though, by later quantum chemical studies carried out by Manzano et al.198 and by Thomas,199,200 both of whom also
studied C-alkyl-pyrogallol[4]arenes. Additional calculations on the alkyl-substituted pyrogallol[4]arenes have been reported by
Emamian.200,201
One piece of experimental evidence attesting to the greater thermodynamic stability of the rctt chair stereostructure of C-phenyl-
pyrogallol[4]arene compared with the rccc cone stereostructure was provided by refluxing the rctt chair form in DMSO for a week,
whereby the chair form was retained in the product isolated by recrystallization from the resultant solution.44 A second piece of
experimental evidence is that although both rctt chair and rccc boat forms of C-phenylpyrogallol[4]arene have been isolated
from methanol solution, the rccc cone form has not been observed.47 When the earlier refluxing experiment was repeated for C-phe-
nylresorcin[4]arene, on the other hand, the product isolated was the rccc cone stereostructure.44
Optimized structures of the pyrogallol[4]arenes examined by Maerz and coworkers (R ¼ C6H5),44 Thomas (R ¼ H and
C6H5),199,200 and Emamian (R ¼ CH3 and C2H5)200,201 were obtained using the 6-31G(d)44 or 6-31G(d,p) basis sets199–201 in
conjunction with the B3LYP and uB97X-D functionals.194 The uB97X-D calculations were carried out to account for dispersion
interactions within the species themselves, an approach required for a more balanced treatment of the attractive and repulsive forces
between aromatic rings.130–133 Single-point MP2 calculations with the 6-31G(d,p) basis set were performed on both the
B3LYP-44,199–201 and uB97X-D-optimized structures.199–201
To investigate the effect of solvation on the energetics of the pyrogallol[4]arenes, SMD calculations were implemented, and
single-point energies for the gas-phase-optimized structures were calculated at the levels described earlier.199–201 In particular,
the commonly used experimental solvents CH3CN, DMSO, and CH3OH were considered in this work.
The roles played by entropy, R-group, electron correlation/dispersion, and solvation effects in determining relative stability
trends were assessed by examining in detail the most stable configurations determined quantum mechanically for the cone, partial
cone, chair, boat, 1,3-alternate, and 1,2-alternate conformers of the pyrogallol[4]arene macrocycles.44,199–201 That is, the rccc cone,
rcct partial cone, rctt chair, rccc boat, rccc saddle, and rcct 1,2-alternate stereoisomers were studied. Each of the earlier factors makes
a significant contribution to the relative stabilities of these stereoisomers,44,199–201 as we illustrate for the rccc cone and rctt chair
stereostructures in Table 2. Thermodynamic data for both the C1 and Ci chair structures are tabulated there for R ¼ phenyl, because
the former structure is the more stable of the two except in CH3OH solvent.
We must caution that the good agreement between the B3LYP/6-31G(d,p) and MP2/6-31G(d,p)//B3LYP/6-31G(d,p) gas-phase
data in the table is somewhat misleading. In comparison with the other levels of calculation, the B3LYP/6-31G(d,p) calculations
often predict incorrect global minima for the cone (and other) diastereomers, whether the R-group is an alkyl or aryl
group.44,199–201 For example, the rtct cone stereoisomer is the global minimum when R ¼ phenyl, and the rcct cone stereoisomer
is the global minimum when R ¼ CH3. The data in the table have been adjusted accordingly. We also note that the agreement
between the uB97X-D/6-31G(d,p) and MP2/6-31G(d,p) data improves when the single-point energy calculation is based on the
uB97X-D equilibrium structure.199–201
In general, entropy, electron correlation/dispersion, and solvent effects stabilize the other stereostructures with respect to the rccc
cone stereostructure, the magnitude of the change depending on the R-group and solvent identities44,199–201 (Table 2). For example,
the stabilization of the chair conformer is pronounced for both the protic and aprotic implicit solvents, but the shift is largest for the
protic solvent CH3OH. Indeed, when R ¼ phenyl, the chair becomes the thermodynamically preferred conformer both in the gas
phase (once electron correlation/dispersion effects are taken into account) and in methanol solvent. Also, the free energy gaps
of 15 and 53 kJ mol 1 for R ¼ CH3 and C2H5, respectively, are consistent with the experimental observations that (1) the preference
for the cone conformer increases as the alkyl tail length increases43,202 and (2) the synthesis of C-alkylpyrogallol[4]arenes in meth-
anol solvent yields both the cone and chair isomers when R ¼ Me but only the cone isomer when R ¼ Et.202
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 313

Table 2 Relative thermochemical values for the rccc cone and rctt chair stereostructures of the pyrogallol[4]arene macrocyclesa

DG (DH) b DG (DH) b DG (DH) b DG DG


B3LYP c uB97X-D d MP2//B3LYP e B3LYP f CH3CN B3LYP g CH3OH
Stereostructure (kJ mol 1) (kJ mol 1) (kJ mol 1) (kJ mol 1) (kJ mol 1)

R¼H
Cone, C4 0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 0.0 0.0
Chair, Ci 42.6 (62.2) 60.0 (82.4) 47.1 (66.7) 28.3 23.2
R ¼ CH3
rccc cone, C4 0.0 (0.0)h 0.0 (0.0) 0.0 (0.0) 0.0 0.0
rctt chair, Ci 49.7 (43.2) 44.0 (59.3) 39.8 (33.3) 31.7 15.2
R ¼ C2H5
rccc cone, C2 0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 0.0 0.0
rctt chair, Ci 79.5 (92.8) 68.3 (81.6) 68.0 (81.4) 68.6 52.5
R ¼ C6H5
rccc cone, C4 0.0 (0.0)h 0.9 (0.0) 11.2 (0.0) 14.7 29.0
rctt chair, Ci 44.2 (70.9) 45.7 (75.2) 52.5 (68.8) 11.8 0.0
rctt chair, C1 1.5 (25.2) 0.0 (20.7) 0.0 (12.5) 0.0 2.1
a
Data for R ¼ H from Refs. [199,200 ], data for R ¼ CH3 and C2H5 from Refs. [200,201 ], and data for R ¼ C6H5 from Refs. [44,199,200 ].
b
Relative free energies and enthalpies (in parentheses).
c
B3LYP/6-31G(d,p) calculations.
d
uB97X-D/6-31G(d,p) calculations.
e
MP2/6-31G(d,p)//B3LYP/6-31G(d,p) calculations.
f
SMD-B3LYP/6-31G(d,p) single-point calculations in implicit CH3CN. There is little difference in the CH3CN and DMSO data; thus, only the former results
have been included.
g
SMD-B3LYP/6-31G(d,p) single-point calculations in implicit CH3OH.
h
Not the global minimum.

The macrocycles considered by Manzano and coworkers198 are C-methylpyrogallol[4]arene, C-fluoroethylpyrogallol[4]arene, C-


t-butylpyrogallol[4]arene, C-phenylpyrogallol[4]arene, C-tolylpyrogallol[4]arene, and C-p-fluorophenylpyrogallol[4]arene.
However, only rccc cone structures with C4 symmetry and rctt chair structures with Ci symmetry were examined, and only gas-
phase relative energy DE values were reported. The structure of each macrocycle was optimized at four different calculational
levels,182 B3LYP/6-311G(d,p), B3LYP/6-311þþG(d,p), B97-D/6-311G(d,p), and B97-D/6-311þþG(d,p), all with expanded basis
sets as compared with those used in the earlier studies. Note that the latter functional does incorporate a dispersion correction term.
The results of Manzano and colleagues198 provide further evidence that the calculations properly reproduce stability trends as
a function of the length of the alkyl R-group. Their results also confirm the importance of dispersion interactions in determining
stability trends. With the added stabilization of the Ci rctt chair relative to the C4 rccc cone upon expansion of the basis set, the chair
is determined to be the preferred stereostructure for the C-arylpyrogallol[4]arenes by 9–13 kJ mol 1 at the B97-D/6-311þþG(d,p)
level of calculation.
Electrostatic potential maps for C4 rccc C-methylpyrogallol[4]arene and C-fluoroethylpyrogallol[4]arene exhibit a large enough
difference in accumulated negative potential within the macrocycle cavity that Manzano et al. proposed a possibility of tuning the
cation binding strength by making a judicious choice of substituent on the methine bridges.198 To test this proposal, they evaluated
the B97-D/6-311G(d,p) BSSE-corrected binding energies between NH4 þ and two macrocycles. The resulting difference in the two
binding energies of roughly 30 kJ mol 1 (230 (CH3) versus 199 (C2H4F) kJ mol 1) suggests that such tuning may indeed be
feasible.
Ahn et al.203 have carried out mPW1PW91/6-311 þ G(d,p) quantum chemical calculations to examine the relative stabilities and
IR spectra of twelve stereoisomers of C-p-cyanophenylpyrogallol[4]arene. Each of the four R-group configurations (rccc, rctc, rctt, and
rcct) was combined with the three macrocycle conformers considered: chair, boat (denoted “table” in their article), and 1,2-
alternate. At this level of theory, the three most stable stereostructures are, from most to least stable (kJ mol 1), rtct chair
(0.0) > rcct 1,2-alternate (11.1)  rctt chair (11.8). Hydrogen-bonding and p–p stacking interactions are cited as the primary stabi-
lizing forces in these stereostructures, with the former dominating the latter. We note, though, that the strengths of the p–p stacking
interactions are likely to be underestimated and are perhaps even repulsive, in the absence of dispersion corrections. This observa-
tion helps rationalize first the z1.5 Å lengthening of the distance between the p-cyanophenyl rings in the calculated structures
compared with that found in the crystal structure204 and second the disagreement in the identities of the preferred calculated
and experimental (rctt chair) stereoisomers.
On a somewhat different yet relevant topic, Cazar and Torres205 studied the rotational profile around the R-C-C-C torsional
angle of the dimer precursors of C-methylpyrogallol[4]arene and C-phenylpyrogallol[4]arene in an attempt to gain insight into
the self-assembly of pyrogallol[4]arenes. The precursors consist of two pyrogallol rings substituted at C4 with eCHR moieties, where
R ¼ CH3 or C6H5. The two rings are linked together in the usual way through one of these eCHR groups. The B3LYP/6-311G(d,p)
energy was monitored as the R-C-C-C torsional angle, the angle that presumably controls macrocyclic assembly, was changed in 5
degree increments from 0 to 180 degrees. An R-C-C-C torsional angle of zero was apparently associated with the equilibrium struc-
ture of the dimer precursor, and single-point energies were calculated at each of the remaining angles. The relatively low-energy
314 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

barrier obtained when R is CH3 (35 kJ mol 1) suggests that rotation about the relevant CeC bond might occur as the macrocycle is
forming. In contrast, the extremely high rotational barrier obtained when R is C6H5 (2045 kJ mol 1) would seem to preclude rota-
tion about that bond during macrocycle assembly. Thus, achieving the thermodynamically preferred conformation without
breaking a CeC linker bond may be possible when R ¼ CH3 but not when R ¼ C6H5. We recognize that these energy barriers are
likely to have been overestimated inasmuch as the remaining geometric parameters were not reoptimized at each fixed torsional
angle. Nevertheless, the conclusions stated herein are likely to be valid given the magnitude of the energy gap between the two rota-
tion barriers.

2.15.5.1.3 Resorcin[4]arenes
The preferred stereostructure of the resorcin[4]arenes with R ¼ H and R ¼ C6H5 has been explored by determining the relative stabil-
ities of the rccc cone, rctt partial cone, rccc boat, rctt chair, rctt 1,2-alternate, and rccc saddle stereoisomers of these macrocycles. The
same computational protocol followed by Maerz et al.,44 Thomas,199,200 and Emamian200,201 for the pyrogallol[4]arenes was fol-
lowed by Drachnik206,207 for the resorcin[4]arenes.182 In addition, gas-phase uB97X-D/6-311 þ G(d,p) single-point energies were
calculated to assess the effect of basis set expansion on conformer stabilities, and SMD-MP2/6-31G(d,p)//uB97X-D/6-31G(d,p)
single-point energies were calculated for the R ¼ H macrocycle in implicit CH3CN, DMSO, and CH3OH solvents to assess the
combined impact of electron correlation and solvent effects on conformer stabilities.
The same trends observed for the pyrogallol[4]arenes are also observed for the resorcin[4]arenes, namely, that the other
conformers are stabilized with respect to the cone conformer when entropy, electron correlation/dispersion, basis set expansion,
and solvent effects are taken into account.206,207 To illustrate these trends, we give the relative enthalpies and free energies of the
cone and chair conformers of the R ¼ H and R ¼ C6H5 macrocycles in Table 3. Again, the calculations suggest that the rctt chair
stereoisomer is the thermodynamically preferred isomer when R ¼ C6H5. Given this general result, one possible rationalization
for the complete chair-to-cone conversion observed for C-phenylresorcin[4]arene in refluxing DMSO is that the activation barrier
for the conversion is lower for the resorcinarene than for the pyrogallolarene and that the resorcinarene cone conformer is less
soluble than the chair conformer in DMSO. Support for this interpretation comes from Cram and coworkers,36 who have noted
that the prevalence of the cone conformer of resorcin[4]arenes can stem from its lower solubility in the solvent used in the synthesis.
For example, the rccc cone stereoisomer of C-phenylresorcin[4]arene is found to be less soluble than the rctt chair stereoisomer in
95% ethanol.31,36,208
The importance of hydrogen bonding in stabilizing the possible conformers of resorcin[4]arenes has been investigated by Mäki-
nen and coworkers.209 They examined the preferred conformations of C-ethylresorcin[4]arene with the positions of the hydroxyl
group protons constrained so as to allow or eliminate formation of OeH/O hydrogen bonds between adjacent resorcinol rings.
Multiple arrangements of the protons also were considered when hydrogen bonding was allowed. Optimizations were performed
with the HF method in conjunction with a mixed basis set (MBS) comprising the 6-31þþG(d,p) basis set on the eight hydroxyl
groups and the 3-21G basis set on the remaining atoms.193 Single-point energies were obtained at the B3LYP/6-311þþG(d,p)//
HF/MBS level of theory.
The rccc configuration of the ethyl substituents was used for the cone, pinched cone, boat, and saddle conformations considered,
whereas the rctt configuration was used for the chair conformation, and the rcct configuration was used for the 1,2-alternate confor-
mation.209 In the absence of intramolecular hydrogen bonding, only the rccc boat, rcct 1,2-alternate, rctt chair, and rccc saddle stereo-
isomers were predicted to be stable, in that order from most stable to least stable. The preference for the boat conformer is believed
to stem from the minimization of repulsions between the OH groups and a relaxation of the tension otherwise present within the
molecular framework.

Table 3 Relative thermochemical values for the C4 rccc cone and Ci rctt chair stereostructures of the resorcin[4]arene macrocyclesa

DG (DH) b DG (DH) b DG (DH) b DG (DH) b DG DG


B3LYPc uB97X-D d uB97X-D e MP2//uB97X-D f CH3CN CH3OH
Stereostructure (kJ mol 1) (kJ mol 1) (kJ mol 1) (kJ mol 1) (kJ mol 1) (kJ mol 1)

R¼H
Cone, C4 0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 0.0 (0.0) 0.0g 0.0h
Chair, Ci 37.0 (50.3) 40.1 (51.9) 34.8 (46.6) 35.4 (46.5) 21.8 13.2
R ¼ C6H5
rccc cone, C4 20.9 (0.0) 34.1 (11.2) 27.5i 38.0j
rctt chair, Ci 0.0 (2.0) 0.0 (0.0) 0.0 0.0
a
Data from Refs. [206,207 ].
b
Relative free energies and enthalpies (in parentheses).
c
B3LYP/6-31G(d,p) calculations.
d
uB97X-D/6-31G(d,p) calculations.
e
uB97X-D/6-311 þ G(d,p) calculations.
f
MP2/6-31G(d,p)//uB97X-D/6-31G(d,p) calculations.
g
SMD-MP2/6-31G(d,p)//uB97X-D/6-31G(d,p) calculations in implicit CH3CN.
h
SMD-MP2/6-31G(d,p)//uB97X-D/6-31G(d,p) calculations in implicit CH3OH.
i
SMD-uB97X-D/6-31G(d,p) single-point calculations in implicit CH3CN.
j
SMD-uB97X-D/6-31G(d,p) single-point calculations in implicit CH3OH.
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 315

The C2v rccc boat stereostructure was then used as a starting structure for the hydrogen-bonded systems, but it converted into rccc
cone and pinched cone structures having the OeH/O hydrogen bonds arranged in different ways. The C4 rccc cone stereoisomer
with four circularly oriented OeH/O hydrogen bonds is the global minimum; the best pinched cone stereoisomer is found to lie
some 13 kJ mol 1 higher in energy. All of these stereoisomers are substantially more stable than those without hydrogen bonds,
however. In general, the relative energies depend on not only the nature of the conformation but also the number and orientations
of the hydrogen bonds.209
The conformational preferences of resorcin[4]arenes bearing substituents bound at the upper rim hydroxyl groups have also
been studied both computationally and experimentally. In their synthesis of a chiral resorcin[4]arene with OeMe groups replacing
the OH groups and alternating arabinitol-1 and pentyl substituents on the methine carbons, Sakhaii et al.39 introduced a new class
of calixsugars. Analysis of the 1H NMR spectrum of the product indicates the presence of a single, asymmetrical stereoisomer. Of the
68 possible stereostructures, 52 were used as starting structures in MD simulations to identify the structure with the lowest potential
energy (EP) and NOE distance restraint violation energy (ENOE). The Insight II and Discover 2.98 software, the CVFF force field
(Accelrys210), and an assumed dielectric constant of 3 ¼ 5 were used in running the simulations. From the 5200 stereostructures ob-
tained using a version of simulated annealing, Sakhaii and coworkers chose the best nine structures for further analysis.
The stereoisomer of the carbohydrate-containing resorcin[4]arene characterized by the lowest EP and ENOE values has an unusual
rctc configuration (r, sugar moiety) and a diamond conformation, with all four substituents bound in axial positions.39 The other
low-energy stereoisomers have chair, cone, or diamond conformations. Chemical shifts of the aromatic protons, because they differ-
entiate among the various ring conformations,31,39,208 were used to confirm the preference for the rctc diamond stereostructure.39
The resorcin[4]arene tetramethyl ether having OMe groups at the upper rim of alternating aryl rings was studied computationally
by Dvoráková and coworkers11 as a model for the O-alkylated resorcin[4]arene derivatives they synthesized.211 These derivatives
range from the parent resorcin[4]arene to the octaalkylated system. Cone, partial cone, chair, boat, diamond, and saddle minima
were identified on the B3LYP/6-31 þ G(d) potential energy surface of tetramethylresorcin[4]arene (R ¼ H).194 The global minimum
here is the cone conformer, the second most stable conformer, a partial cone, lying some 24 kJ mol 1 higher in energy.11 The cone is
stabilized by four strong MeeO/HeO hydrogen bonds, whereas the partial cone is stabilized by two strong OeH/OeMe
hydrogen bonds and OeH/p interactions. OeH/p and CeH/p interactions influence the stability of the other conformers
as well. The calculated results reproduce those obtained from proton NMR and NOE experiments, which indicate that only the
cone conformer is stable for the O-alkylated resorcin[4]arene derivatives at the temperatures considered (163–298 K).
Dvoráková and coworkers11 also investigated, both experimentally and computationally, the so-called “flip-flop” mechanism for
the interconversion of the cone conformer of the O-alkylated resorcin[4]arene derivatives. The mechanism discerned from the
B3LYP/6-31 þ G(d) calculations proceeds through a single-step conversion whereby a distorted chair transition structure is formed
by rotation of two adjacent aryl rings. The calculated barrier height associated with this mechanism is 40–45 kJ mol 1, in good
agreement with the activation free energy DGz value of 42 kJ mol 1 obtained from the temperature-dependent 1H NMR spectrum
(in CDCl2) of the tetraisopropyl derivative. Overall, the DGz value for the flip-flop inversion increases as the number of eOH groups
and accordingly the number of OeH/O hydrogen bonds in the macrocycle increases.11,28

2.15.5.2 Metal–Macrocycle Interactions


Cation–p interactions are recognized as being an important factor in the stabilization of biological212,213 and supramolecular
systems,214,215 especially when multiple aromatic binding sites are available.213,216–218 Thus, understanding the intrinsic role of
this interaction in stabilizing cation binding to the calixarene family of macrocycles is of interest. To eliminate competing interac-
tions that might stabilize metal–calix[4]arene complexes, Macias et al.219 constructed a model calix[4]arene system for which the
hydroxyl groups were replaced by H atoms. Using this dehydroxylated calix[4]arene model, they performed B3LYP/6-31G(d,p)
calculations on Liþ, Naþ, and Kþ complexes formed with the model host and focused in particular on the conformational stability
and metal selectivity of these complexes in the absence of electrostatic and hydrogen-bonding effects. The choice of this level of
calculation was justified by demonstrating good agreement for the Mþ–benzene and benzene–Mþ–benzene binding energies ob-
tained at this level with those measured experimentally.
Without the hydroxyl groups, the partial cone, boat, 1,3-alternate, and chair forms have essentially equal stabilities.219 That is,
the gas-phase energies lie within 2.0 kJ mol 1 of one another. (Note that the cone and 1,2-alternate forms are not minima on the
potential energy surface of the dehydroxylated calix[4]arene.) Binding a metal ion imparts a preference to the conformers, however,
with the particular preference depending on the identity of the metal and the number and orientation of the aryl rings that partic-
ipate in the binding. Liþ and Naþ preferentially bind to the 1,3-alternate conformer, whereas the larger Kþ preferentially binds to the
boat (flattened cone) conformer. In addition, the interaction strength decreases as the size of the ion increases (300 vs. 242 vs.
184 kJ mol 1). The complexes that exhibit multiple, weaker cation–p interactions involving several p-faces are generally found
to be more stable than those that exhibit a single strong cation–p interaction.
To examine the gas-phase interactions of the alkali metal ions Liþ, Naþ, Kþ, Csþ, and Rbþ with resorcin[4]arene and pyrogallol
[4]arene, Rozhenko et al.16 first optimized the cone and boat conformers of resorcin[4]arene (R ¼ H and Me) and pyrogallol[4]arene
(R ¼ Me).194,220 Only the all-axial configuration of the methyl groups was considered. The study was limited to these two
conformers because they were believed to be most important for metal complexation. Although a number of levels of theory
were considered, the reported minimizations were performed at the RI-BP86/SVP level,221 and RI-MP2/TZVP//RI-BP86/SVP
single-point energies were calculated.222 (The RI term in the notation indicates that the resolution of identity approximation
316 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

was used in the calculations.223–227) ECPs were used for Cs, Rb, and, in some cases, for K. Estimated BSSE corrections were found to
be small enough to be neglected. For the resorcin[4]arenes, the C4 rccc cone (0.0 kJ mol 1) and C2v boat (16–20 kJ mol 1) stereo-
structures yield similar energies, but the energy gap (120 kJ mol 1) is considerably larger for C-methylpyrogallol[4]arene.
Five types of alkali metal–arene complexes were identified, although not all of the five prove to be stable for every metal.16 The
metal ion lies within the cavity in the three complexes for which the macrocycle adopts the cone (either C4 or C1 symmetry) or
pinched cone (C2 symmetry) conformation. In the C1 rccc cone and Cs rccc boat Mþ–arene complexes, the metal interacts with
one aryl ring and two oxygen atoms. The remaining complex is the C2v rccc boat.
Two main contributing factors determine the relative stabilities of the Mþ–arene complexes. The first is the energy associated
with distorting the arene from its uncomplexed equilibrium geometry. The second is the strength of the cation–p (Kþ, Csþ, and
Rbþ) or cation–O (Liþ and Naþ) interaction energies. With the first factor favoring the cone conformation of the macrocycle
and the latter favoring the boat, perhaps unexpectedly, the C4 rccc cone Mþ–arene complex is generally not most favorable. For
the resorcin[4]arene complexes, cone, pinched cone, and boat conformers have about equal stability (within 10–15 kJ mol 1)
regardless of the nature of the metal. In contrast, with the exception of the Liþ complex, the cone and/or pinched cone conforma-
tions are preferred for the Mþ–pyrogallol[4]arene complexes. The Cs rccc boat complex is, on the other hand, most stable for Liþ–
pyrogallol[4]arene. For each of the three macrocycles, the metal binding energy decreases as the radius of the metal ion increases.16
Given the interest in storing molecular hydrogen, Posligua et al.228 have investigated the binding energy of H2 to Mþ–pyrogallol
[4]arene complexes. In this work, Mq þ ¼ Liþ, Naþ, Kþ, and Mg2 þ and R ¼ CH3 and C2H4F. Only the rccc cone stereoisomer was
examined for the macrocycles. With the metal placed inside the cavity and the C4 symmetry constraint relaxed, the geometries
of the metal–macrocycle complexes were fully optimized at the B3LYP/6-311G(d,p) level of theory.182 The H2 molecule was
then added to the complexes in an h2 orientation, and only the position of the metal and H2 was optimized at the B3LYP/6-
311G(d,p) and B3LYP/6-311G(d,p)/aug-cc-pVDZ (metal, aug-cc-pVDZ basis set229,230 and all other atoms, 6-311G(d,p) basis
set) levels. To obtain the BSSE-corrected H2 binding energies, single-point energies were calculated with the B97-D and MP2
methods. The latter calculations were performed using the ONIOM hybrid method, in which the H2, metal, and some neighboring
atoms were placed in the high-level (MP2) layer and the remaining atoms of the macrocycle were placed in the low-level (B3LYP)
layer. Both sets of binding energy calculations were carried out with both basis sets.
The Mg2 þ cation binds to C-methylpyrogallol[4]arene with a B3LYP/6-311G(d,p) strength of 967.5 kJ mol 1 and to C-fluoroe-
thylpyrogallol[4]arene with a strength of 919.8 kJ mol 1, binding strengths some four times as large as those for the singly charged
ions.228 The Mg2 þ [ Liþ > Naþ > Kþ trend in binding strengths correlates directly with the electron density transferred to the metal
and indirectly with the distance of the metal from the lower rim of the macrocycle. With the exception of the small Liþ ion, the
bound ions lie along the center line of the macrocycle (Fig. 6). Although, the calculated charge transfers and metal positions are
similar for the two R-groups (within 0.013 e and 0.013 Å, respectively), the weaker interaction found when R ¼ C2H4F rather
than when R ¼ CH3 is consistent for all of the metals, a result in agreement with the proposition of Manzano and coworkers.198

Figure 6 (A) Liþ–C-methylpyrogallol[4]arene complex. (B) Mg2 þ–C-methylpyrogallol[4]arene complex. (C) Mg2 þ–C-fluoroethylpyrogallol[4]arene
complex. Note the higher, off-center position of the Liþ cation in the former complex compared with the Mg2 þ position in the latter complexes. Color
scheme, F, aqua; Liþ, lilac; Mg2 þ, yellow. Adapted from Posligua, V.; Urbina, A. S.; Rincon, L.; Soetens, J. C.; Mendez, M. A.; Zambrano, C. H.;
Torres, F. J. Comput. Theor. Chem. 2015, 1073, 75–83.
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 317

Regardless of the level of calculation, the binding strength of H2 to the metal-functionalized pyrogallol[4]arenes follows the
order Mg2 þ [ Liþ > Naþ  Kþ, a trend similar to the order given earlier.228 Increasing the size of the basis set increases the magni-
tude of the interaction energy by  3 kJ mol 1, and including electron correlation/dispersion effects also tends to strengthen the
interaction. However, the B97-D calculations overemphasize the binding strength by some 5–20 kJ mol 1 compared with the
MP2 calculations for both basis sets. Similarly, Kocman and coworkers found that the B97-D method overemphasizes the binding
strength of H2 to a Liþ–coronene complex compared with that determined from of a state-of-the-art CCSD(T)/complete basis set
(CBS) calculation or from a quantum Monte Carlo calculation.231 Finally, the MP2 binding enthalpy of  17.6 kJ mol 1 for the H2–
Mg2 þ–C-fluoroethylpyrogallol[4]arene system suggests that the magnesium-functionalized C-fluoroethylpyrogallol[4]arene and C-
methylpyrogallol[4]arene macrocycles may be useful components of materials designed for storage of molecular hydrogen.228
Bakic et al.232 have synthesized three fluorescent p-tert-butylcalix[4]arene derivatives and tested their suitability as metal ion
sensors. Specifically, the OeH linkages at the lower rim of the macrocycle were replaced by four O-phenanthridine linkages, two
alternating O-phenanthridine linkages, or alternating O-phenanthridine and OeMe linkages. Spectrophotometric, potentiometric,
fluorimetric, and microcalorimetric titrations then were used to measure the stability constants and thermodynamic parameters
associated with binding an alkali metal to these derivatives in two different solvent systems (CH3CN/CH2Cl2 and CH3OH/
CH2Cl2). Quantum chemical calculations and MD simulations were performed to elucidate the structures of the Liþ-, Naþ-, Kþ-,
Rbþ-, and Csþ-calixarene complexes.
The experimental studies revealed that the strength of the alkali metal cation binding decreases with the radius of the cation.232
Indeed, the partially substituted phenanthridine derivatives have a strong affinity for only the Liþ and Naþ ions. For a given cation,
the interaction is strongest for the fully substituted phenanthridine derivative and the CH3CN/CH2Cl2 solvent system. Overall, the
standard free energy of complexation values ranged from 23 to 46 kJ mol 1. Both the substituted p-tert-butylcalix[4]arenes with
four O-phenanthridine linkages and with alternating O-phenanthridine and OeMe linkages showed substantial fluorescence
enhancement upon alkali metal ion binding and were deemed potential candidates for fluorimetric cation sensors.
The OPLS-AA force field233 as implemented in the GROMACS 4.6.5 package234 was used in the MD simulations, which were run
at 298 K for a total of 55 ns in an NPT (constant composition, pressure, and temperature, i.e., isobaric) ensemble.232 The simulations
were carried out for the two sensor candidates, with and without the cation present. The calixarene or cation–calixarene complex
was placed in a simulation box filled with either of two preequilibrated solvent mixtures, CH3CN/CH2Cl2 or CH3OH/CH2Cl2. For
the cationic systems, a ClO4  ion was added to the corner of the box to balance the charge. Each starting structure had the calixarene
in a flattened cone (boat) conformation and the metal ion positioned in the center of the lower rim cavity interacting with the host
ether oxygen and phenanthridine nitrogen atoms. This solute–solvent system was then briefly relaxed prior to collecting the
dynamics data. Almost immediately, in the presence or absence of a cation, a solvent molecule enters the upper cavity of the cal-
ixarene and induces a conformational change from the flattened cone to the cone. The guest solvent molecule is preferentially
a CH3CN or CH3OH molecule with its methyl group oriented toward the bottom of the cavity. However, the presence of a metal
ion can cause a reorientation of the solvent molecule such that the nitrile or hydroxyl group helps to stabilize the metal in the lower
cavity. The encapsulated solvent–metal interaction is accompanied by a reduction in the number of N atoms that coordinate to the
metal and an increase in substituent–solvent interactions. Regardless of the orientation of the solvent, however, its encapsulation
enhances the binding affinity of the metal, as measured by the number of cation–N interactions. (Note also that the cation remains
coordinated to all of the ether atoms throughout the simulations.) As was observed experimentally, there is an inverse correlation
found here between cation binding affinity and cation radius.
To gain more insight into how changing the cation and the total coordination number (the total number of oxygen atoms and
nitrogen atoms coordinated to the metal) affects the structure of the alkali metal–calixarene complex, quantum chemical calcula-
tions were carried out for the quadruply phenanthridine-substituted p-tert-butylcalix[4]arene.232 The potential energy surface of the
complex was first scanned using the PM6 semiempirical method235 by varying eight torsional angles and locating the resultant local
minima. These 450 local minima located were then optimized at the B3LYP/3-21G level of theory,182 thereby reducing the number
of unique minima to about 100. The most structurally and energetically relevant of those minima was finally reoptimized at the
B3LYP/6-31G(d) level of theory with an added dispersion correction term. Complexes with coordination numbers 4–8 were iden-
tified. The position of the cation in the lower cavity of the host is seen to depend not only on the size of the cation but also on the
total coordination number. That is, the metal ion shifts away from the centerline for complexes with total coordination numbers of
five and seven and shifts further from the plane of the oxygen atoms as the size of the ion increases. The orientation of the CH3CN
guest in the upper cavity of the host also varies with the size of the metal ion. As Bakic et al. suggest, knowing the importance of the
solvent in cation binding plus the structural trends detailed earlier should help in the targeted design of supramolecular
receptors.232
In a different approach to the study of the complexation ability of cavitands toward alkali metal ions (Naþ, Kþ, Rbþ, and Csþ),
Mattay and coworkers236 synthesized the first fluorescent cavitand crown ether, measured association constants in solution using
fluorescence titrations, and probed the structural features of the complexes using DFT calculations. To build the cavitand crown
ether, the nitrogen center of an aza-21-crown-7 ether with a dihydroxy coumarin chromophore was attached to one of the C2 centers
of a resorcin[4]arene (R ¼ C11H23) having the oxygen centers on adjacent resorcinol rings bridged by eCH2 moieties. For this
system, Kþ binds most strongly (K ¼ (1.1  0.2)  105), followed closely by Rbþ (K ¼ (6.3  0.5)  104). In each case, the alkali metal
forms a 1:1 complex with the cavitand crown ether. These association constants are an order of magnitude larger than those
measured for the N-benzyl-crown ether, an increase suggesting that the cavitand crown ether host forms a sandwich complex around
the metal ions.
318 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

The proposed cooperative binding of the alkali ions was validated by the DFT calculations, for which only Kþ and Csþ were
considered and the cavitand was simplified by using R ¼ CH3.236 (B3LYP functional in conjunction with the ECPs (on all atoms
but H) and corresponding basis sets of Stevens et al. (H, C, N, and O)237 or Preuss et al. (K and Cs)238 and polarization functions
were added to the basis sets for the C, N, and O atoms.194) The geometry optimizations were initiated from two starting structures
for each ion, one in which the cation was placed nearer to the aza-crown ether and one in which it was placed nearer to the resor-
cinarene. In each case, the crown ether covered the resorcinarene. Two minima of essentially equal energy (DE ¼ 8 kJ mol 1) were
located for the Kþ–cavitand crown ether complex, and one was located for the Csþ-cavitand crown ether complex (Fig. 7). All three
differed in total coordination number with respect to the crown ether oxygen and nitrogen atoms and the cavitand oxygen atoms,
from seven (lower-energy complex) to six for the potassium complexes to eight for the cesium complex. Although fewer resorcinar-
ene than crown ether oxygen centers participate in the binding, the two sets of coordinating centers form separate binding spheres to
cooperatively stabilize the cation-cavitand crown ether complexes.

2.15.5.3 Macrocycle–Guest Interactions


Because many studies have focused either on macrocycle–guest interactions directly or on aspects of nonbonded interactions that
are relevant to macrocycle–guest sequestration, we focus here on a subset of those that strike us as being particularly illustrative or
especially relevant to our own interests. We begin with the investigation of the complexation of deprotonated C-ethylresorcin[4]
arene and C-phenylresorcin[4]arene with a selection of mono-, di-, and oligosaccharides reported by Kalenius et al.,239 who
have employed mass spectrometric techniques and quantum chemical calculations. They were originally planning to study C-tet-
raethylpyrogallol[4]arene also but, despite trying numerous approaches, did not succeed in deprotonating that macrocycle. One
objective of the negative-polarization electrospray ionization (ESI) quadrupole ion trap and ESI Fourier transform ion cyclotron
resonance analyses conducted was a determination of the effect of the type and size of the saccharide on the thermodynamic
and kinetic stabilities of the complex formed. A second objective was determining whether conformational differences in the
two resorcinarenes affect complex formation. As these investigators have testified, the calculations were indispensible in explaining
the surprising experimental results obtained.
The two deprotonated resorcin[4]arenes formed 1:1 complexes with all 12 saccharides (xylose, ribose, fucose, quinovose, galac-
tose, mannose, glucose, cellobiose, cellotriose, cellotetraose, cellopentaose, and cellohexaose), an entirely unexpected result given
the size of the latter oligosaccharides.239 There was no difference found in the complexation behavior of the two resorcinarenes,
a result indicating that both macrocycles adopt the cone conformation in these complexes. Of the monosaccharides, the hexoses
formed the thermodynamically and kinetically most stable complexes; of the di- and oligosaccharides, biose and triose formed
the thermodynamically and kinetically most stable complexes. In fact, cellobiose and cellotriose are found to be even more favor-
ably bound than are the hexoses.
To gain insight into the formation of the complexes, the geometries of the complexes, the intermolecular interactions that stabi-
lize the complexes, and the relative stabilities of the complexes, quantum chemical calculations were performed with both DFT and
MP2 methods.194 Geometries were optimized with the B3LYP and PB86 functionals together with the SVP basis set.221 (We note
a potential caveat here, though, the authors did not report having calculated vibrational frequencies to confirm that the structures
they obtained correspond to energy minima.) The larger TZVP basis set222 was used to obtain RI-MP2226,227 single-point energies
with the PB86 geometries.
In the first step of the calculations, gas-phase-optimized geometries were obtained for glucose, cellobiose, cellotriose, cellote-
traose, and cellopentaose.239 As the number of glucose units increases in the polysaccharides, there is an increase in the curvature
of the saccharide skeleton such that the distance between the two ends of the molecules is reduced to 10–12 Å. The compatibility
between this distance and the diameter of the upper rim of the resorcinarenes found computationally ( 10 Å) explains the
observed facile complex formation for the larger polysaccharides. In the second step of the calculations, optimized geometries

Figure 7 Sandwich-like complex geometries calculated for the (A) Kþ- and (B) Csþ-cavitand crown ether complexes showing the closer contacts
between the metal and crown ether oxygens than between the metal and cavitand oxygens. Hydrogen atoms have been removed for clarity. Color
scheme, Kþ, lilac; Csþ, purple. Adapted from Stoll, I.; Eberhard, J.; Brodbeck, R.; Eisfeld, W.; Mattay, J. Chem. Eur. J. 2008, 14, 1155–1163.
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 319

were obtained for complexes of C-ethylresorcin[4]arene with glucose, cellobiose, and cellotriose. Inspection of the preferred
arrangements of the guest molecules in these complexes reveals that glucose participates in three hydrogen bonds with the macro-
cycle and cellobiose and cellotriose each participate in four hydrogen bonds. That the MP2/TZVP//BP86/SVP binding energies
are  121,  158, and  165 kJ mol 1 for the glucose, cellobiose, and cellotriose complexes, respectively, demonstrates the direct
correlation between the number of hydrogen-bonding interactions and stability.
In a computational investigation of factors that improve the uptake of small molecules by calixarene systems, Murphy et al.155
assessed the impact on guest-binding affinity of a transition metal bound to the lower rim tetraphenolic pocket of calix[4]arene. A
variety of first-row transition metals, with different oxidation and spin states, were considered in the study, in particular Mn3 þ
(quintet), Fe3 þ (quartet), Fe3 þ (sextet), Co2 þ (quartet), Co3 þ (quintet), Ni2 þ (singlet), Ni2 þ (triplet), Ni3 þ (quartet), Cu2 þ
(doublet), Cu3 þ (singlet), and Cu3 þ (triplet). The choice of metals is consistent with earlier computational work by these
investigators showing that calixarenes interact more strongly with late rather than early transition metals.156 The guests considered
were the important small molecules HCN, H2, H2O, H2S, NH3, N2, N2O, O2, CO2, and SO2. All possible binding modes for a guest
within the cavity were examined.
After an extensive calibration study in which the DFT functional (with and without correction for dispersion effects) and the
basis set were varied, geometry optimizations were performed using the B3LYP functional with GD3BJ240 empirical dispersion
in conjunction with the 6-31G(d,p) basis set for nonmetal atoms and the SDDALL basis set for the metal atoms.182 This choice
of computational level was further validated by evaluating modified B1 diagnostics,156,241 thereby verifying the single-reference
character of the wave function. The only change made in this approach for the energy calculations and implicit solvent (H2O) calcu-
lations was extension of the basis set to 6-311þþG(d,p) for the nonmetal atoms. Binding energies were corrected for BSSE.
After ascertaining that the enhancement of CO2 binding that occurs on complexation of a metal to calix[4]arene depends much
more strongly on the presence of a metal than on the type, oxidation state, or spin state of the metal, binding preferences toward the
earlier small molecules were determined for the Mn3 þ–calixarene and quartet Fe3 þ–calixarene complexes.155 In the absence of
a metal, H2 ( 6.3 kJ mol 1) is most weakly bound, and SO2 ( 49.3 kJ mol 1) is most strongly bound. In the presence of a metal,
H2 ( 7.5 kJ mol 1) remains most weakly bound, but SO2 ( 72 kJ mol 1) and NH3 ( 73 kJ mol 1) are now equally strongly
bound. (In each case, the SO2 is bound through the oxygen.) Overall, coordination of a metal causes binding affinity changes
that range from 1 kJ mol 1 (H2) to 45 kJ mol 1 (NH3), effectively increasing the binding selectivity and suggesting a potential
use for metal-coordinated calixarene complexes in detecting or separating gases.
An example of efforts to explore the use of cavitands, in this case a resorcin[4]arene, in drug development is provided by the work
of Patel et al.242 In this work, formation of an inclusion complex between p-sulfonato-resorcin[4]arene (substituted at C2 of the
resorcinol ring) and lamotrigine was investigated with respect to improving the solubility, toxicity, and bioavailability of this poorly
water-soluble drug. A number of analytic techniques were used to characterize the inclusion complex and to evaluate the stability
constant of the complex, including FT-IR spectroscopy, PXRD analysis, and differential scanning calorimetry. Molecular docking
studies and MD simulations were carried out as well to elucidate the intermolecular interactions that stabilize the inclusion
complex.
The molecular docking approach (Avogadro 1.1.0,243 Ghemical force field244) was used to identify the best arrangement of
a saddle-shaped p-sulfonato-resorcin[4]arene host and the lamotrigine guest in the complex.242 This arrangement was then supplied
as the starting structure for the MD simulation. The simulation was run with the YASARA software245 and the AMBER ff03 force
field246 at room temperature for a total of 10 ns. The equilibrated adduct then was placed in a simulation box filled with explicit
water molecules, and the total energy was minimized for 100 steps before the MD run was initiated. Because the structural integrity
of the complex was the focus of the study, the simulation was constrained to a constant volume and shape ensemble. A structure was
extracted from the MD trajectory at 2500 ps, and this equilibrated structure was used to analyze host–guest interactions.
The results of the experimental investigation show that a 1:1 inclusion complex is formed in aqueous solution with a stability
constant of 854, a value indicating that the lamotrigine is bound strongly enough to stabilize the complex but not so strongly as to
prevent complete release of the drug.242 Moreover, formation of the inclusion complex enhances the water solubility of the drug
and reduces the acute oral toxicity of the drug. The results of the computational investigation show that the inclusion complex
is stabilized by two hydrogen bonds between the amine group of lamotrigine and one of the p-sulfonato groups of the resorcin
[4]arene and by two p–p contacts between the triazine and phenyl groups of the guest and the upper rim aryl groups of the
host. Indeed, the sulfonato groups act as “supporter appendages” and are critical for maintaining the structural integrity of the
complex and sustaining guest entrapment within the host cavity.

2.15.5.4 Vase Versus Kite Structures


Another type of conformational flexibility is the vase-to-kite conversion exhibited by resorcin[4]arene cavitands suitably function-
alized at the upper rim hydroxyl groups or at the 2-position on the resorcinol ring. Dyker et al.247 synthesized a resorcin[4]arene
derivative with four isoquinolin-1-yl substituents attached to the upper rim because they envisaged these substituents creating
a narrow, electron-rich cavity that would selectively bind linear molecules with a positively polarized center. For this derivative,
adjoining resorcinols are connected via OeCH2eO linkages, R ¼ C5H11, and the isoquinolin-1-yl moieties are attached to the C2
center of the aromatic rings. Analysis of the XRD crystal structure of the cavitand revealed, as was expected, a vaselike conformer
with the bulky aromatic rings lying outside of the cavity and the N center lying above the cavity with the lone pair of the N posi-
tioned to interact with the guest. However, the placement of the aryl groups and distance between the N atoms suggested there
320 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

might be more flexibility in guest binding than anticipated. Indeed, the resorcin[4]arene derivative forms complexes not only with
acetonitrile and carbon disulfide but also with benzonitrile. Interestingly, analysis of the crystal structures of the three complexes
showed only benzonitrile interacts effectively with the nitrogen lone pairs. In order to understand why the isoquinolin-1-yl-
substituted resorcin[4]arene binds CS2 but not CO2, optimized structures were calculated for the resorcinare–CS2 and
resorcinarene–CO2 complexes at the TPSS-D3/def2-TZVP level of theory.248 The thermodynamic parameters associated with
ð3Þ
complex formation were evaluated at the TPSS  D3 þ Edisp =def 2  QZVP==TPSS  D3=def 2  TZVP level of theory.249 (The
ð 3Þ
Edisp term in the notation indicates that the total energy has been corrected for the three-body contribution to the dispersion
energy.144) The calculated interaction energy for the CO2 complex is  34.7 kJ mol 1, whereas that for the CS2 complex
is  59.4 kJ mol 1. The 25 kJ mol 1 difference primarily stems from the dispersion correction; without it, the relative energies
are reversed. The changes in free energy on complex formation are 10.0 kJ mol 1 for the CO2 complex and  8.4 kJ mol 1 for
the CS2 complex. This computational result indicating that the resorcinarene–CO2 complex is unbound but the resorcinarene–
CS2 complex is thermodynamically stable is in excellent agreement with the experiment.247
Diederich and coworkers57 have reported the synthesis of two resorcin[4]arene-based molecular baskets with different pendent
R-groups, the cavitand having been rigidified by the introduction of p-xylene bridges. For these two cavitands, R ¼ n-C6H13 or
R ¼ 3,5-(t-Bu)2PhCH2. Their first XRD structural analysis of a molecular basket revealed that the system incorporating the latter
R-groupdthe same observation presumably applies in the case of the former R-group as welldadopts a vase conformation having
a well-defined cavity within which guests may be encapsulated. With the use of both 1H NMR spectroscopy and isothermal titration
calorimetry, association constants Ka were measured for the R ¼ n-C6H13 host and a range of 24 cycloalkane and alicyclic heterocycle
guests. Among the driving forces for formation of the inclusion complexes are dispersion forces, optimal cavity filling, and CeH/p,
NeH/p, S/p, and CeO/C]O interactions. The six-membered heterocyclic guests are found to bind most strongly, with Ka
values ranging from 106 to 107, with favorable association enthalpies and entropies contributing to the strong, reversible binding.
The favorable association entropies were attributed to several factors, one of these factors being that guest encapsulation has little
effect on the conformational flexibility of the rigid, conformationally preorganized host. Because encapsulated guests are known to
prompt molecular baskets to adjust their cavity volumes so as to optimize the percentage of occupied space,60 MD simulations were
performed to assess the effect of the guest on the size and preorganization of the host cavity. In particular, packing coefficients, that
is, Vguest/Vhost ratios, were evaluated for each of the inclusion complexes formed between an R ¼ CH3 molecular basket and one of
the 24 guest species.57
Each of these 24 simulations, carried out using the MacroModel 9.7 software package, the OPLS force field, and the GB/SA solva-
tion model for CH3Cl,250 corresponded to an equilibrium system temperature of 300 K and a total time interval of 1 ns. (A 1.5 fs
time step was used throughout in these calculations.) Vhost then was calculated as the average volume of 1000 structures, with the
guest removed, extracted from each of the resulting trajectories. (The host and guest volumes were evaluated with the VOIDOO
program.251,252) The guest volumes were obtained on the basis of the structure optimized with the PM3 semiempirical
method.253,254 The strongly bound aliphatic heterocycles were found to yield packing coefficients of 63  9%, a range of values
that, though overlapping, generally lie beyond the range of preferred packing coefficients of 55  9% recommended by Mecozzi
and Rebek for apolar complexation.255 However, Mecozzi and Rebek have observed that the packing coefficient increases as the
number of host cavity–guest polar contacts increases, and Diederich and coworkers propose that the packing coefficient they found
results from the added NeH/p, S/p, and CeO/C]O polar interactions stabilizing these inclusion complexes.57
In an earlier study, Diederich and coworkers56 synthesized two resorcin[4]arene-based molecular baskets with alternating qui-
noxaline and diazaphthalimide bridging groups and either a pyrene or an anthracene moiety attached to the latter groups. Proton
NMR and fluorescence spectroscopy subsequently were used to determine the effect of acid addition, temperature, and guest
binding on the vase-to-kite conversion and to investigate p–p stacking interactions between the attached fluorophores, intramo-
lecular excimer formation, and [4 þ 4] photodimerization of the two anthracenes. MD simulations provided support for the unex-
pectedly large thermal conformational flexibility of both conformers of the two molecular baskets. Here, simulations were run for
only the vase conformers, and as in their earlier study, the n-C6H13 R-groups were replaced by CH3 groups. The computational
protocol described earlier was repeated in this work, but in this case, the MMFF94 force field256 and the MacroModel 9.5 software
package were utilized.257 An overlay of the 1000 vase structures extracted from the trajectories for each cavitand emphasized the
flexibility of the two vase conformers in solution, a result in sharp contrast with the perfect vase structures found in the solid
state.53–55,62,258,259 Consistent with this high host flexibility in solution, intramolecular excimer formation occurs for both the
vase and kite conformers, and no [4 þ 4] photodimerization of the anthracene pairs was observed.
Other conclusions drawn from the study are the following.56 Guest binding to the interior cavity (1) reduces the conformational
flexibility of the cavitands, (2) shifts the vase–kite equilibrium toward the vase form, (3) reduces the p–p stacking between the flu-
orophores, and (4) reduces intramolecular excimer formation. Changing the temperature and the pH likewise reversibly shifts the
vase–kite equilibrium.
For the resorcin[4]arene-based molecular basket synthesized by Ruiz-Botella et al.,58 the bridging groups are benzimidazoles
having ferrocenyl moieties appended to the imidazole’s C2 atom. The metal moieties were introduced in an attempt to construct
a “redox-switchable molecular gripper.” A key aspect of such a system is the ability to control the conformation and therefore
the trapping ability of the cavitand electrochemically. Thus, the effect of metal oxidation, both in the presence and in the absence
of four Cl counterions, on the structure of the cavitand was investigated using DFT calculations. As usual, the lower rim substit-
uents, in this case C11H23, were replaced by CH3 substituents for the calculations. To optimize the structures, the PBE-D3 method
was used in conjunction with a DZVP basis set260 for the valence orbitals and a plane wave basis set for the core orbitals.261 Each
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 321

Fe2 þ ion was assumed to be a singlet; each Fe3 þ ion was assumed to be a doublet. Single-point energies were determined for the
cavitands in implicit toluene solvent with the SMD method at the M06/6-31G(d,p)/SDD þ f level of theory.182 (The notation here
indicates that an f orbital262 was added to the SDD ECP and basis set for iron.263)
In agreement with the NMR analysis, the calculations show that the cavitand adopts a vase conformation in all cases (Fig. 8).58
However, oxidation of the metal causes the molecular basket to convert from a closed container to an open-ended container. That is,
the ferrocenyl moieties partially cover the internal cavity of the ferrocene-containing molecule but not that of the oxidized molecule.
Although adding the four chloride ions moves the ferrocenyl moieties closer together, the moieties nonetheless remain pointed
away from the cavity, and the container remains open-ended. The preference for the vase over the kite form is substantial here
regardless of whether the metals are reduced or oxidized (115 vs. 108 kJ mol 1 in solution, respectively).
With these encouraging computational results, the authors performed 1H NMR spectroscopic, ESI mass spectrometric, and cyclic
voltammetric analyses of host–guest binding properties. A series of tetraalkylammonium salts plus L-carnitine and choline were
selected as the guests. The guest binding energies in the 1:1 complexes are strong but size-dependent for the ferrocene-
containing host. Also, the tetraalkylammonium ions appear to bind as ion pairs stabilized by electrostatic or ion–p interactions.
Oxidation of the metals then causes the loss of the guest, making this a redox-switchable molecular cavitand with tunable trapping
properties (i.e., the desired “gripper”).58

2.15.5.5 Exploiting Chirality


With the goal of synthesizing artificial receptors that offer insight into enzyme activity and selectivity, Speranza and collaborators
have constructed a variety of chiral resorcin[4]arene-based host cavitands and have assessed these compounds’ enantioselectivity
toward a number of biomolecules.2,264–270 Their experimental studies in this area, along with the quantum chemical calculations
and MD simulations they have performed to aid them in interpreting their results, have been recently reviewed.72 In each case, gas-
phase measurements of the relevant properties of the resorcin[4]arene host–guest complexes were obtained, and thus, these
researchers could focus exclusively on the intrinsic factors that drive enantiodiscrimination processes. As examples of the types
of studies carried out, the work reported in two of their more recent publications will be summarized herein.267,268
To more fully understand the recognition capability of cyclochiral resorcin[4]arene receptors, Freschetti et al.268 looked at the
kinetics of exchange reactions (Eq. 2) between a neutral primary amine B (CH3CHWNH2, W ¼ H, CH3, and C2H5) and

Figure 8 (A) Optimized structure of the ferrocenyl form of the benzimidazole cavitand illustrating the closed cavity by the upper rim of the vase. (B)
Optimized structure of the oxidized form of the cavitand illustrating the wide-open cavity by the upper rim of the vase. (C) Optimized structure of the
oxidized form of the cavitand with four Cl ions to balance the charge illustrating the open cavity by the upper rim of the vase. Color code, Fe,
purple; Cl, green. Adapted from Ruiz-Botella, S.; Vidossich, P.; Ujaque, G.; Vicent, C.; Peris, E. Chem. Eur. J. 2015, 21, 10558–10565.
322 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

a proton-bound complex [M,H,G]þ of the cyclochiral rccc-2,8,14,20-tetra-n-decyl-4,10,16,22-tetra-O-methylresorcin[4]arene M


and 11 chiral bidentate or tridentate guests G. The cyclochirality of the host arises from the directionality (clockwise or counterclock-
wise) of the OeH/OeCH3 hydrogen bonds at the upper rim of the macrocycle. Using nano-ESI Fourier transform ion cyclotron
resonance mass spectrometry, they measured enantioselectivity ratios, defined as r ¼ khomo/khetero, to determine whether complexes
in which the host and guest have the same chirality (homochiral) react more or less quickly than those in which the host and guest
have opposite chirality (heterochiral). With the exception of 5-hydroxy-L-tryptophan, (1R,2R)-2-amino-1-phenyl-1,3-propanediol,
and L-norepinephrine, the homochiral complexes react more slowly than do their heterochiral homologues. That the reaction effi-
ciency depends on the PA of B indicates that the enantioselectivity originates from the relative stability of the transition structures
rather than from the relative stability of the proton-bound adducts for most guests.
½M,H,Gþ þ B/½M,H,Bþ þ G (2)
þ
To locate the proton (host or guest) and guest within the [M,H,G] aggregates and to determine the types of intermolecular
interactions stabilizing the aggregates, Fraschetti and coworkers268 then carried out a series of quantum chemical calculations on
complexes with rccc-2,8,14,20-tetramethyl-4,10,16,22-tetra-O-methylresorcin[4]arene as the host. RI-BP86/SVP optimizations
were carried out on 8–10 starting structures for each diastereomer.220,221 The number of XeH/O, XeH/N, and XeH/p inter-
actions was increased in each resulting adduct until the energy no longer stabilized. The lowest-energy structures were then reopti-
mized at the RI-BP86/TZVP level of theory, to expand the basis set, and then at the RI-B97-D/TZVP level of theory, to include
dispersion effects.222 The equilibrium structures of the homochiral and heterochiral proton-bound resorcin[4]arene-(1R,2R)-2-
amino-1-phenyl-1,3-propanediol complexes, determined to be the most enantioselective system (r ¼ 3.85), are given in Fig. 9.
For this guest, the preferred protonation site is the amino nitrogen, a result in agreement with ESI-CID results. The NHþ 3 moiety
lies at the center of the cavity in the homochiral complex but at a corner of the cavity in the heterochiral complex. Each complex
exhibits cation–p interactions between the guest NHþ 3 group and host aryl groups and two OeH/OeCH3 interactions between
guest aliphatic OH groups and host upper rim OCH3 groups. The joint experimental and computational results suggest that a combi-
nation of all three types of interactions is needed to drive host–guest complexation and to maximize host–guest interactions and,
thus, enantiodiscrimination.
Filippi et al.267 have measured the infrared multiple photon dissociation (IRMPD) spectra of proton-bound host–guest adducts
of the R, R, R, R- and S, S, S, S-enantiomers of a bis(diamido)-bridged resorcin[4]arene host with cytosine, cytidine, and cytarabine
guests. In this case, the macrocycle has eight upper rim OCH3 moieties, and each eC(O)NHCH(C6H5)CH(C6H5)NHC(O)e moiety
bridges the methine linkers of a given resorcinol subunit. The two subunits involved are oppositely facing.
Use of the IRMPD technique in conjunction with DFT calculations allowed these investigators to distinguish between the result-
ing diastereomeric noncovalent complexes. Such an achievement has proved difficult in the past because the structures and stabil-
ities of these types of complexes usually vary as a result of differences in weak effects, for example, charge transfer, conformational,
dispersion, and repulsion effects, rather than in the stronger electrostatic and hydrogen-bonding effects. ONIOM (B3LYP/6-31G(d)/
UFF)-calculated normal-mode vibrational frequencies were used to assign the vibrational modes associated with the features of the
IRMPD spectra of the diastereomers.194 The starting structures for the ONIOM minimizations were obtained from Monte Carlo
molecular mechanics docking calculations and constant-temperature MD simulations. Both N- and O-protonation of the guests
were examined for complexes in which the guest is accommodated inside the host cavity (at the lower rim of the boat-shaped mac-
rocycle) or outside the cavity. The preferred N- and O-protonation sites were determined by comparing proton affinities evaluated at
the B3LYP/6-311þþG(d,p) level of theory. Here again, the calculations brought clarity to the experimental work and aided in con-
firming that several isomeric complexes exist for all three systems in which the N-protonated guest is located either inside or outside
the cavity.

Figure 9 Equilibrium structures found for the (A) homochiral and (B) heterochiral proton-bound resorcin[4]arene-(1R,2R)-2-amino-1-phenyl-1,3-
propanediol complexes. Adapted from Fraschetti, C.; Letzel, M. C.; Paletta, M.; Mattay, J.; Speranza, M.; Filippi, A.; Aschi, M.; Rozhenko, A. B.
J. Mass Spectrom. 2012, 47, 72–78.
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 323

Szumna and collaborators73,271–274 also have reported multiple studies on the structures and properties of chiral resorcin[4]are-
nes. In one of the more recent publications by this group, electronic circular dichroism (ECD) measurements and TD-DFT calcu-
lations were utilized to assess the effect of self-assembly on the dynamic (stimuli-responsive) chirality of a set of iminoresorcin
[4]arenes synthesized by reacting tetraformylresorcin[4]arenes with (S)-amino acids.272 These cavitands, in which the resorcinol
C2 carbon is involved in the C]N bond of the imino substituent, exist as keto-enamine tautomers in solution and in the solid state.
When the (S)-iminoresorcin[4]arene is combined with the (R)-iminoresorcin[4]arene, the monomers self-assemble to form
a dimeric capsule.
The ECD spectra for all of the monomers constructed from (S)-amino acids have closely related patterns, indicating they all have
the same inherent chirality. To determine whether the (S)-iminoresorcin[4]arenes have a (P,S) or (M,S) configuration, a calculated
ECD spectrum was obtained for a simplified cavitand for which R ¼ CH3.182 The starting structure for the B3LYP/6-31G(d,p) opti-
mization of the cavitand geometry was taken from a crystal structure of a dimer capsule wherein the (S)-iminoresorcin[4]arene
unambiguously has a (P,S) configuration. The TD-DFT calculations (B3LYP/6-31G(d)) yielded an ECD spectrum for the model
(S)-iminoresorcin[4]arene that deviates significantly from the experimental spectra. Because most of the bands had opposite signs,
it was concluded that self-assembly induces a switch in the inherent chirality of the (S)-iminoresorcin[4]arenes. This conclusion was
validated when the starting structure for the optimization of the model cavitand geometry was taken from a crystal structure of one
of the (S)-iminoresorcin[4]arene monomers for which the configuration is unambiguously (M,S). In this instance, the resultant
ECD spectrum from the TD-DFT calculations on the (M,S) cavitand agrees well with the experimental spectrum. The TD-DFT calcu-
lations also revealed that the band associated with the HOMO / LUMO transition is the most reliable band for use in assigning the
inherent chirality of these systems.

2.15.5.6 Organic Crystal Porosity


Although quantum chemical calculations can yield accurate host–guest binding energies and optimized structures, they do not
directly elucidate the details of processes that are inherently dynamic. As an example, we cite the work of Atwood and coworkers,
who have shown that several small gas molecules may be sequestered by crystals of the low-density b0 polymorph of p-tert-
butylcalix[4]arene (TBC4) even though these crystals exhibit no channels that would permit passage of the gas molecules into
the solid matrix.275–285 We approached the mechanism of gas sequestration by first examining the stability of host–gas complexes
at various temperatures, anticipating that host cavity dynamics will modulate the binding and dynamics of the guest.286 All these
calculations were carried out using the GAFF force field159,160 as implemented within the AMBER MD suite,168 partial atomic
charges determined on the basis of computational chemistry calculations194 (HF 6-31G(d) and the RESP method287,288), and
the weighted histogram analysis method289,290 for the determination of potentials of mean force. (The specific simulation protocols
implemented in this work are detailed in the original publications.286,291)
These simulations indeed demonstrate that stable host–guest complexes will form at room temperature.286 At this temperature,
both CO2 and methanol are found to bind to TBC4 with a binding (Helmholtz) free energy of 14 kJ mol 1, and ethane is bound by
8.4 kJ mol 1. On the other hand, methane, ethylene, and acetylene molecules are bound by no more than 6.5 kJ mol 1, and
indeed, the host–guest complexes were found to dissociate over the nanosecond timescale of the simulations. These results revealed
the limitations of isolated host–guest binding simulations, though, in that gas sorption isotherms for acetylene, for example, clearly
indicate facile sequestration of the gas to higher densities than otherwise could be handled safely.281 Realizing that we required
a model that better reflects the environment within the b0 crystal, we then devised a simple tethered dimer model of the solid,
in which the upper rims of two hosts are oriented “face-to-face” but are offset relative to one another. (To prevent collapse of
this structure to the self-inclusion complex found in the high-density polymorph, the macrocycles are harmonically tethered at their
methine carbons to their equilibrium crystal positions.291) Using this model, we predict that acetylene and methane should be
sequestered successfully within the TBC4 crystal; when one of these molecules escapes from a host cavity, it collides with a tert-butyl
group of the second host and is deflected back into the cavity within which it was originally bound.
These last observations led to another question, however. If a molecule such as CO2 forms a stable complex with a TBC4 host
and if a simple model of the crystal suggests that more weakly bound species may be sequestered because a neighboring host
impedes dissociation of the complex, how do the gas molecules penetrate to the interior of the crystal? Binding selectivity cannot
be disregardeddTBC4 is known to sequester CO2 selectively when exposed to a mixture of CO2 and H2275dbut the binding ener-
gies alone292 do not clarify the mechanism of gas transport. A possible explanation is that, say, a trapped CO2 molecule simply
shuttles from host cavity to host cavity, but we find that doing so requires that the molecule surmount a potential energy barrier
of about 19 kJ mol 1 (which is 7.7 RT at room temperature) when the two hosts sit at their crystallographic positions and that
this barrier height is temperature-independent. The resolution to this quandary was suggested by the original synthetic work on
b0 TBC4 involving the loss of toluene from a toluene-TBC4 cocrystal in a two-step process that retains sample crystallinity.293 In
this process, the TBC4 bilayers are found to slide relative to one another as the toluene is lost. We conclude that this long-
wavelength phonon mode is characteristic of low-density TBC4 crystals and that as the host layers slide past one another, the energy
barrier for the loss of a guest decreases. The guests thus are relayed from host to host or are deposited within the interstitial space
until they are picked up by a neighboring host.291 In support of this mechanism, we note that rotation of the tert-butyl groups is
quite facile, and thus, their rotation reduces the friction between the host bilayers, that is, they act as molecular ball bearings in these
crystals.
324 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

2.15.6 Hydrogen-Bonded Nanocapsules


2.15.6.1 Model Complexes
Due to the large number of atoms in many (perhaps most) interesting supramolecular assemblies, high-level quantum chemical
calculations are not practical, and it is necessary to identify a reliable yet cost-effective calculational level at which to study these
systems. Often, to find the best balance between accuracy and required resources, calibration studies on small model molecular
assemblies are used to benchmark structural and energetic data from lower-level computations against those data from high-
level computations. The small models also provide a way to probe the intermolecular forces implicated in molecular recognition
and the factors controlling the assembly process.11,294–296 Also useful here are MD simulations, which, because they are based
primarily on semiempirical force fields, scale favorably with system size and afford the ability to include solvent molecules
explicitly.
In our first quantum chemical study of pyrogallol[4]arene nanocapsular assemblies, we utilized model systems to aid our exper-
imental collaborators in interpreting their single-crystal X-ray diffraction data (SCXRD) on hydrogen-bonded hexameric pyrogallol
[4]arene capsules297 (Fig. 10). Specifically, we examined hydrogen-bonded nanocapsular wall segments and a half nanocapsule to
assess electronic versus steric contributions to guest size effects.
Dalgarno and coworkers have investigated the interaction of molecular probes with hydrogen-bonded pyrogallol[4]arene-based
hexameric nanocapsules in an attempt to characterize the “inner phase” of the capsules and to delimit the guest size.297–299 To that
effect, they have examined encapsulation of pyrene butyric acid,298 1-(9-anthryl)-3-(4-dimethylaniline)propane (ADMA),299 benzo
[a]pyrene, and pentacene in the C-hexylpyrogallol[4]arene hexamer.297 In each case, the guest was found to be trapped within the
hexameric assembly but, except in the case of pyrene butyric acid, guest disorder and low capsule occupancy precluded locating the
guest within the nanocapsule per se by SCXRD analysis. For pyrene butyric acid, however, the SCXRD analysis revealed two guest
molecules lying parallel to opposite capsule walls.298 Space-filling representations of benzo[a]pyrene and pentacene with the hex-
amer suggested that these guests would lie in the center of the capsule rather than along a wall.297 To further address the issue of
guest placement within the nanocapsules and to interrogate host–guest interactions in these assemblies, electronic structure calcu-
lations194 were performed. The goals of the calculations were twofold: (1) to determine whether pyrene and benzo[a]pyrene are
positioned differently within a C-hexylpyrogallol[4]arene half nanocapsule and, if so, whether the difference in positioning results
primarily from steric interactions and (2) to verify that the low-to-medium levels of theory at which it was necessary to perform
most of the calculations can reproduce equilibrium structures obtained at higher levels of theory. In order to address the first point,
two sets of model complexes were studied for both pyrene and benzo[a]pyrene. B3LYP/cc-pVDZ calculations300 were performed for
complexes with a single pyrogallol molecule (C6H6O3/C16H10 and C6H6O3/C20H12) and with a “wall segment” composed of
three intra- and intermolecularly hydrogen-bonded pyrogallols ((C6H6O3)3/C16H10 and (C6H6O3)3/C20H12). In addition, the
geometry and position of each guest were optimized in a rigid half nanocapsule at the B3LYP/3-21G level of calculation. For the
half nanocapsule, the heavy atoms were fixed at the coordinates obtained from the XRD analysis, the hydroxyl hydrogens were
oriented to maximize hydrogen bonding, the R-groups were replaced with hydrogen atoms, and the alkyl groups were capped
with hydrogen atoms (Fig. 10). In order to address the second point, model complexes of pyrogallol with benzene
(C6H6O3/C6H6) were examined at the B3LYP/cc-pVDZ, MP2/cc-pVDZ, and MP2/aug-cc-pVDZ levels of calculation.229,230
Complexes of benzene with the wall segment were also investigated but only at the first level noted earlier.

Figure 10 C-hexylpyrogallol[4]arene hydrogen-bonded hexameric nanocapsule. The hexyl R-groups have been omitted for clarity. Adapted from
Dalgarno, S. J.; Szabo, T.; Siavosh-Haghighi, A.; Deakyne, C. A.; Adams, J. E.; Atwood, J. L. Chem. Commun. 2009, 1339–1341.
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 325

Two “edge-on” arrangements and one “face-to-face” arrangement of the benzene and pyrogallol were observed at each
calculational level.297 One edge-on arrangement has an OeH/p interaction, and the second has CeH/O (and possibly CeH/p)
interactions. In the face-to-face arrangement, the benzene/pyrogallol complex has a slipped parallel structure and an OeH/p inter-
action. Similar structures were found for the pyrogallol/benzo[a]pyrene and pyrogallol/pyrene complexes. However, for the latter
complexes, potential minima corresponding to the molecules being oriented face-to-face but experiencing no OeH/p interactions
were also observed. For these minima, the pyrene and benzo[a]pyrene are tilted somewhat to accommodate CeH/O interactions.
For all corresponding structures, the pyrogallol/benzo[a]pyrene binding energy is essentially equal to the pyrogallol/pyrene binding
energy.
Because p–p interactions are not properly represented at the B3LYP/cc-pVDZ level of theory, it is likely that the importance of
CeH/O interactions is overemphasized in these (and in the following) equilibrium structures. It is also likely that only edge-on
and face-to-face arrangements without OeH/p hydrogen bonds are relevant in the nanocapsule–guest systems, inasmuch as the
pyrogallol hydroxyl groups are expected to form intra- and intermolecular hydrogen bonds within the framework of the
nanocapsule.
Optimizing the geometry of the wall segment (B3LYP/cc-pVDZ and MP2/cc-pVDZ) yields a twisted configuration that
maximizes the intra- and intermolecular hydrogen bonding between the hydroxyl groups. When a complex is formed between
the wall segment and a guest molecule, the segment flattens but not to the extent observed in the nanocapsule (Fig. 11). The
only equilibrium structure located for the complex between benzene and the wall segment has the benzene oriented edge-on
with respect to the segment. In contrast, both pyrene and benzo[a]pyrene form edge-on and face-to-face structures with the wall
segment. Again, the latter structures are tilted somewhat to accommodate CeH/O interactions (Fig. 11), and there is little devi-
ation in the host–guest binding energies for the two guests.
At the start of the optimization process for the half-nanocapsule calculations, pyrene and benzo[a]pyrene were situated parallel
to the bottom of the half capsule at the same distance from the wall.297 Pyrene shifted position slightly to maximize CeH/O inter-
actions but stayed within the half capsule, whereas benzo[a]pyrene shifted position until it extended beyond the top of the half
capsule (Fig. 12). Indeed, the host–guest interactions were so unfavorable at the original benzo[a]pyrene position that the guest
molecule hinged by nearly 90 degrees until it was tilted sufficiently to relieve the steric strain and become planar again in the final,
optimized structure.
In summary, the small model system and half-nanocapsule calculations performed in this work indicate that (1) host–guest
equilibrium structures obtained at the B3LYP/cc-pVDZ level of calculation are representative of those obtained at higher levels,
(2) pyrene and benzo[a]pyrene show no significant difference in orientation or binding energy with respect to pyrogallol or the
wall fragment, and (3) the two guests are positioned differently in the host nanocapsule primarily for steric rather than electronic
reasons, in agreement with the conclusion drawn from the space-filling representations of the host–guest complexes. The results
also suggest that aromatic guests may be stabilized in the nanocapsule by p–p, CeH/p, and CeH/O interactions.297

Figure 11 Equilibrium structures obtained for the (A) benzene–wall segment, (B) pyrene–wall segment, and (C) benzo[a]pyrene–wall segment
complexes. Adapted from Dalgarno, S. J.; Szabo, T.; Siavosh-Haghighi, A.; Deakyne, C. A.; Adams, J. E.; Atwood, J. L. Chem. Commun. 2009,
1339–1341.
326 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

Figure 12 Views of (A) pyrene encapsulated within the half nanocapsule and (B) benzo[a]pyrene encapsulated within the half nanocapsule. The
guests are depicted in tubular form. Adapted from Dalgarno, S. J.; Szabo, T.; Siavosh-Haghighi, A.; Deakyne, C. A.; Adams, J. E.; Atwood, J. L.
Chem. Commun. 2009, 1339–1341.

2.15.6.2 Nanocapsules
Perhaps the nanocapsule most studied computationally is the hydrogen-bonded resorin[4]arene-based dimer first synthesized by
Rebek and coworkers301 (Fig. 13). Each cavitand of the dimer has pyrazine-2,3-dicarboxylic acid imide moieties bridging the
oxygen centers on adjacent resorcinol rings and C11H23 R-groups. In earlier computational work on this capsule, changes in guest
conformation,302–304 reactivity,305 mobility,305–307 and fluorescence upon encapsulation were investigated,158 as were the steric
and magnetic environments inside the capsule.308 In one of their most recent studies, Rebek and coworkers157 have examined the
effect of encapsulation on the structures and dimerization energies of the homo- and heterodimers of p-methylbenzoic acid,
p-ethylbenzoic acid, p-methylbenzamide, and p-ethylbenzamide. In this case, their work on competitive encapsulated
dimers309–311 has been extended to systems for which the cage cavity is not large enough to accommodate the dimers.
Equilibrium structures and interaction energies were obtained at three calculational levels, namely, M06-2X/6-31G(d,p),
uB97X-D/6-31G(d,p), and ONIOM M06-2X/6-31G(d,p)/PM6.182 The first two of these levels yielded similar structures (Fig. 13)
and the same energy trends, thereby lending credibility to these results. Although the calculated energy differences were smaller,
sometimes even leading to changed relative energies, for the third level compared with the other two, the 30  speedup achieved
in the calculations nonetheless makes the ONIOM approach worth pursuing.
The dimerization energies for the free p-(m)ethyl-substituted compounds follow the trend (benzoic acid)2 > benzamide–
benzoic acid > (benzamide)2.157 This trend is reversed for the encapsulated compounds as a result of the compression caused by
the relatively small cavity of this hydrogen-bonded dimer. Indeed, the interactions between the monomers and the interior walls
of the capsule are the primary stabilizing force in these host–guest complexes; the stability contributed by the weak monomer–
monomer hydrogen bonds is < 10% of that contributed by the monomer–host interactions. The computational results thus explain
why the host–homodimer and host–heterodimer systems under consideration have not been observed experimentally and suggest
a possible approach for separating competitive monomers or dimers of similar size.

Figure 13 Optimized structures calculated for (A) the hydrogen-bonded dimer synthesized by Heinz et al.301 and (B) the dimer with the (p-methyl-
benzoic acid)2 guest. The guest is shown in ball-and-stick representation. Note the distortion of the capsule caused by guest encapsulation. Adapted
from Tzeli, D.; Petsalakis, I. D.; Theodorakopoulos, G.; Ajami, D.; Rebek, J., Jr. Theor. Chem. Acc. 2014, 133, 1–14.
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 327

Ayhan et al.312 used EPR and 1H NMR spectroscopy, ESI-MS, and DFT calculations to investigate host–guest interactions for
a water-soluble variant of the benzimidazolone cavitand introduced by Ebbing and coworkers.313 In this variant, the substituents
on the methine bridges are pyridinium moieties314,315 (Fig. 14). Because the search for cages that efficiently sequester nitroxide
radicals is ongoing, seven nitroxide radical guests were considered. Only four of the guests formed inclusion complexes, di-tert-
butylaminoxyl (DTBN), 2,2,6,6-tetramethylpiperidin-1-oxyl (TEMPO), 4-oxo-2,2,6,6-tetramethylpiperidin-1-oxyl (TEMPONE),
and bKCTO (1,4,14,17-tetraoxa-9-aza-21-oxotetraspiro[4.2.1.2.4.2.3.2] tetracos-9-yl-9-oxyl), and those are the guests that will be
discussed in the succeeding text.
The purpose of the investigation was threefold312: (1) to probe the inner metric dimensions of the capsule, (2) to determine
whether cavitand–guest (1:1) or capsule–guest (2:1) inclusion complexes are preferred, and (3) to evaluate binding constants
for the inclusion complexes. The experimental results suggest that the 2:1 complex is strongly preferred for DTBN, TEMPO, and
TEMPONE, but the 1:1 complex is more competitive for bKCTO at low concentrations. (We note that the macrocycles exist as
both the vase and kite conformers in the absence of a nitroxide guest, but guest encapsulation shifts the distribution in favor of
the vase form.) The binding constants Ka determined from EPR titrations range from 5  106 to 7.9  107 for the 2:1 complexes,
with DTBN most strongly bound and bKCTO least strongly bound. On the other hand, the Ka values for the 1:1 complexes range
from 8  102 to 9.3  103, with bKCTO most strongly bound and TEMPO least strongly bound.
B3LYP/6-31G(d) structures determined in the presence of implicit water182 (CPCM183,184) were obtained to help explain the
EPR, 1H NMR, and ESI-MS results and to support the conclusions drawn from those results.312 When a cavitand/TEMPO complex
was examined, the calculations revealed that the 1:1 complex is not stable inasmuch as the guest exits the cavitand. In contrast, two
stable isomeric forms were identified for each of the four 2:1 complexes. The two forms are essentially equal in energy for each guest
(DE 2 kJ mol 1) and differ only slightly with respect to guest position near the center of the capsule. In each complex, the guest’s
NO group is oriented perpendicular to the C4 axis of the capsule and points toward the polar seam of the capsule. With widths of 6.7
and 7.9 Å, respectively, DTBN and TEMPO cause very little distortion in the capsule when entrapped (Fig. 14). However, the TEM-
PONE width of 8.1 Å, which is comparable to the width of the capsule itself, and its added carbonyl functionality cause one of the
aromatic panels at the center of the capsule to bend slightly from its “ideal” position upon encapsulation of this guest. This bending
is much more severe for bKCTO, with its banana-like shape, greater width (9.3 Å), and a height (16.9 Å) comparable to that of the
capsule, yet the host’s flexibility allows accommodation of this guest (Fig. 14) as well. The optimized structure for the capsule/
bKCTO inclusion complex shows that only 12 of the 16 possible hydrogen bonds between the benzimidazolones remain in this
complex, a result consistent with its lower Ka value. Moreover, the extent of the capsule framework contortion required in forming
the 2:1 complex provides a reasonable explanation for the greater competitiveness of the 1:1 complex for the bKCTO guest. Once

Figure 14 Side-on views of the equilibrium structures determined for (A) the hydrogen-bonded dimer capsule formed from the water-soluble
variant of the benzimidazolone cavitand, (B) the DTBN-dimer assembly, and (C) the bKCTO-dimer assembly. Hydrogen atoms have been removed for
clarity. Note the disruption of the capsule wall caused by encapsulation of bKCTO. Adapted from Ayhan, M. M.; Casano, G.; Karoui, H.; Rockenbauer,
A.; Monnier, V.; Hardy, M.; Tordo, P.; Bardelang, D.; Ouari, O. Chem. Eur. J. 2015, 21, 16404–16410.
328 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

again, this work highlights how capsule flexibility can extend the range of encapsulated guests by maximizing monomer–monomer
interactions while minimizing guest steric strain.
Octa acid316 (Fig. 15) is another resorcin[4]arene-based deep cavitand that has received considerable attention.317–319 In one of
their latest studies on the dimeric capsule formed from this cavitand, Choudhury et al.317 reported 1H NMR experiments and MD
simulations on host–guest complexes for which the guest was a phenyl-substituted hydrocarbon. The guests were taken from sets of
alkanes (six), alkenes (six), and alkynes (four) of different chain lengths. These complementary techniques were combined to estab-
lish the preferred conformation of the guest within the confined space of the host dimer and to follow the movements that the guest
molecule undergoes in attaining those conformations.
To prepare for the MD simulations (GROMACS program320,321 and OPLS-AA force field233) on the capsule–guest complexes, the
following computational procedure was pursued. After minimization of the structure of the octa acid capsule itself using the Merck
molecular force field,256 the structure was further equilibrated via a 40 ns all-atom MD run in explicit water solvent. The structures of
the guests were minimized using the MM2 force field322 prior to docking them in the equilibrated octa acid using the AutoDock
Vina program.323 To obtain the starting structures of the capsule–guest complexes utilized in the 40 ns all-atom MD runs in explicit
water, the most stable systems found from the docking procedure were reoptimized with the OPLS-AA force field.
If the extended hydrocarbon chain is longer than 15 Å, the chain is found to adopt a folded form that maximizes CeH/O,
CeH/p, and p/p interactions with the host.317 The probability that this folding process will occur depends on both the chain
length and the presence of multiple bonds. The rigid multiple bonds of the alkene and alkyne guests constrain the phenyl group to
one cavity as the remainder of the alkyl chain folds in the second cavity. With their greater flexibility, the alkane guests tumble freely
and assume unusual conformations not normally accessible in solution. The unexpected controls on guest binding found in this
systematic study will enhance our understanding of guest behavior in confined spaces.
In a second study by this same group, in order to mimic the photoinduced geometric isomerization of olefins observed in
biological systems, Samanta et al.319 investigated the singlet excited-state behavior of 1,4-diphenyl-1,3-butadiene (DPB) and
1,4-ditolyl-1,3-butadiene (DTB) in the hydrophobic, confined environment of the octa acid capsule. Capsule–guest assemblies
were formed from each of the three possible isomers: trans,trans; trans,cis; and cis,cis for both dienes (1H, COSY, and DOSY NMR
analyses). Competition experiments established the following stability order for the isomeric forms within OA: cic,-
cis > cis,trans  trans,trans. Before the photochemical studies were carried out, MD simulations were run to determine whether there
is a preference for one of the two possible conformers, cisoid or transoid, of the dienes when confined.
MD simulations were performed for the three isomers of both DPB and DTB following the protocol described earlier. The last
MD run was performed for 20 ns, and capsule–guest complexes extracted from the trajectories were bundled to determine the “most
representative” structure for each complex. Unlike the preference for the transoid conformer in the equilibrium mixture of the two
conformers in solution, regardless of the isomeric form, the MD simulations indicate that the cisoid conformer is preferred within
the octa acid capsule. This prediction was bolstered by the facile [4 þ 2] addition observed from the reaction of the entrapped trans,-
trans-DPB or trans,trans-DTB with singlet oxygen. The geometric isomerization of the dienes caused by irradiation (> 280 nm) of the
octa acid diene assembly produced primarily the cis,trans isomer (70–80%), a result that was independent of the isomeric form of

Figure 15 Representation of octa acid.


Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 329

the guest prior to irradiation. However, the yield of the cis,cis isomer was improved significantly (up to  70%) by selective irradi-
ation of the assemblies.319
With the interest in halogen bonds that has developed in recent years, Tero et al.324 synthesized resorcin[4]arene-based capsules
designed to exploit halogen bonds in the construction of 2D networks with porous structures. Halogen bonding, an analog of
hydrogen bonding, is an intermolecular interaction between an electrophilic halogen atom and a Lewis base. C-methylresorcin
[4]arene was modified by replacing one OeH group on each resorcinol with an OeMe moiety and the second OeH group with
an Oepyridine moiety. Both 3-pyridyl and 4-pyridyl groups were examined as potential halogen bond acceptors to evaluate the
effect of node angle on the resultant structure. The halogen bond donors were bromopentafluorobenzene, iodopentafluoroben-
zene, 1,3-dibromotetrafluorobenzene, 1,4-dibromotetrafluorobenzene, and 1,4-diiodotetrafluorobenzene. The resorcinarene aryl
halide linker assemblies were studied in both solid state and solution via X-ray crystallography, NMR spectroscopy, and ESI-MS.
DFT calculations were carried out to help interpret the complex 19F NMR spectra and to provide a better understanding of the prop-
erties of these assemblies in solution.
The crystal structure determination for the 1,4-diiodotetrafluorobenzene bivalent linker with the 3-pyridyl resorcinarene deriv-
ative revealed 2D linear arrays of resorcinarenes linked by chains of halogen bonds. The analogous determination for the 4-pyridyl
resorcinarene derivative, however, revealed a 2D interlocked tangled grid structure. In each case and in solution, the resorcin[4]arene
exists as the rccc boat stereoisomer. Although all five linker molecules halogen bond to the resorcinarene derivatives in solution, 2D
networks were not observed. To clarify the relationship between the change in the 19F chemical shift and the linking mode (mono-
valent vs bivalent), chemical shifts were calculated with the gauge including projector augmented wave approach on the basis of
structures optimized using ultrasoft PBE pseudopotentials.325 Because the linker can participate in a number of different interac-
tions in solution, depending on the concentrations of the interacting species, complexes of the following type were examined:
solvent–linker (with 1, 2, or 4 solvent molecules), linker–resorcinarene, and resorcinarene–linker–resorcinarene. In these
complexes, the solvent is chloroform, the linker is either 1,4-dibromo- or 1,4-diiodotetrafluorobenzene, and the resorcinarene is
modeled as the halogen bond-accepting pyridine ring. For purposes of calibration, the calculated 19F signals for the linker–solvent
complexes were compared with the signal for the free linker. The upfield shifts in the signals for the linker–solvent complexes are
consistent with the observed upfield shifts in the signals for these linkers in deuterated chloroform relative to their signals in the
noncoordinating solvent pentane. The calculated downfield shifts in the signals for the linker–resorcinarene complexes and upfield
shifts for the resorcinarene–linker–resorcinarene complexes explain the concentration dependence of the observed 19F chemical
shifts. The calculated chemical shift trends also indicate that 1,3-dibromotetrafluorobenzene forms only 1:1 complexes with the
resorcin[4]arene derivatives in solution.324

2.15.7 Metal-Seamed and Covalently Linked Nanocapsules


2.15.7.1 Metal-Seamed Nanocapsules
Having set the stage by examining not only the properties, conformations, and guest binding properties of the constituent moieties
but also related work on the structure and properties of selected nanocapsule systems, we turn now to a particular interest of ours,
namely, the structure and properties of rigid nanocapsules formed by seaming macrocycles together with metal ions. The following
discussion will begin with a consideration of Zn2 þ-seamed capsules, to be followed with a summary of similar work involving other
metal ions. Many metals have yet to be investigated in connection with nanocapsule formation; thus, such systems are likely to offer
many future opportunities for research and provide building blocks for novel materials that display unique functionality and struc-
tural features.

2.15.7.1.1 Zinc Model Complexes


Small zinc complexes have been used to date in a number of investigations aimed at modeling the catalytic site of zinc-containing
enzymes.326–332 As one example, a zinc complex with three histidines and a pyrogallol has been used to model the active site of the
matrix metalloproteinase MMP-1. In this work, Li et al.329 performed molecular modeling and docking studies to determine the
potency and selectivity of pyrogallol and myricetin as possible inhibitors of MMP-1 and MMP-3. They explained the higher potency
of myricetin in terms of its stronger binding to the two enzymes, whereas the greater selectivity of both inhibitors for MMP-1 stems
from the differences in (1) the hydroxyl group that binds to the Zn, (2) the number of hydrogen bonds formed between the inhib-
itor and active site residues, and (3) the strength of the interactions between the inhibitor and enzyme. It is unclear in these docking
studies, however, whether pyrogallol binds to MMP-1 in a monodentate or bidentate manner. A B3LYP/6-31G(d) optimization194
of the small zinc model complex clarified this issue, indicating that pyrogallol is indeed a monodentate ligand.
In our work, we examined relevant mononuclear and polynuclear zinc model complexes prior to tackling the metal-seamed
pyrogallol[4]arene-based nanoassemblies. The objectives of these quantum chemical calculations were to calibrate the computa-
tional approach, to quantify Zn–ligand interaction strengths, and to screen divergent ligands as possible tethers for the
nanoassemblies.

2.15.7.1.1.1 Calibration
Although computational studies on 4-, 5-, and 6-coordinate Zn complexes, including several calibration studies,333–336 have
appeared in the literature, none of the complexes considered when we began our work had reproduced the Zn coordination sphere
330 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

in the zinc-seamed pyrogallol[4]arene dimeric nanocapsules. Thus, we investigated the structures and energetics of small mononu-
clear zinc model systems to assess the reliability of various basis sets (LanL2DZ, cc-pVXZ, 6-31G(d), 6-311 þ G(d,p), and B2333) and
methods (B3LYP, M05-2X, M06-L, uB97X-D, and MP2). The zinc complexes used for these comparisons are Zn(OH)2(ROH)mLn337
and/or Zn(X)2Ln,338 where n ¼ 0–2; m ¼ 2–4; R is Me or H; X is deprotonated (Z)-ethene-1,2-diol, 1,2,3-trihydroxy-cis-1,3-
butadiene, or pyrogallol; and L is NH3, C5H5N, (CH3)2SO, (CH3)2NCHO, or CH3OH (Fig. 16). The calculational levels chosen
were based on the results of previous calibration studies.333–336 Some common conclusions drawn from the analyses of the results
of the two studies are the following. (1) Within a given method, all basis sets yield similar core structures, and the relative energies
follow similar trends. (2) At a given level of calculation, similar binding energies are obtained regardless of the basis set/method
combination used to optimize the geometry. (3) The complexes are stabilized by both conventional and unconventional hydrogen
bonds. (4) The ZneL and ZneO bond lengths and OeZneO(L) bond angles observed experimentally for the zinc-seamed pyro-
gallol[4]arene capsules are reproduced well by the model complexes. (5) NH3 is a viable alternative to C5H5N as a capsule ligand,
and the combination of deprotonated (Z)-ethene-1,2-diol (C2O2H3 ) and deprotonated 1,2,3-trihydroxy-cis-1,3-butadiene
(C4O3H4 2 ) is a viable alternative to deprotonated pyrogallol (C6O3H4 2 ) as the capsule backbone.
To obtain further support for our conclusion that the Zn(C2H3O2)2L model complexes replicate the Zn coordination environ-
ment in zinc-seamed pyrogallol[4]arene nanocapsules, polynuclear Zn complexes were assembled from 2, 3, 4, 6, or 8 zinc centers
and NH3 molecules together with sufficient C4O3H4 2  and C2O2H3  ions to complete the Zn coordination sphere and set up the
O/H/O hydrogen-bonding networks.338 As the number of zincs is increased, each model complex curves naturally such that the
shape of the 4-zinc model mirrors that of a half capsule and the shape of the 8-zinc model is a closed ring (Fig. 17), with a diameter
only 0.2 Å larger than that found for the [Zn8(C-propylpyrogallol[4]arene)2(pyridine)8 3 pyridine] dimer.26 However, adding
a second NH3 to the zinc centers, making them 6-coordinate, causes the structures to flatten out. These results suggest that penta-
coordination of zinc may be crucial to seaming the cone-shaped pyrogallol[4]arenes used in the synthesis of the dimers (Fig. 17).
A second piece of evidence supporting the use of the Zn(C2H3O2)2L model complexes is that the average s5 value for the zincs
in the Zn8(C4O3H4)8(NH3)8 model system (0.40  0.04), the average s5 value for the zincs in [Zn8(C-propylpyrogallol[4]
arene)2(pyridine)8 3 pyridine] (0.42  0.03), and the s5 value for the zinc in the mononuclear Zn(C2O2H3)2(NH3) model system
(0.41) are in excellent agreement.338 Computing s5 values affords a convenient way of comparing the metal coordination
environment of model complexes with those of the actual systems without having to analyze individual bond lengths and

Figure 16 Examples of the Zn(OH)2(ROH)mLn and Zn(X)2Ln model complexes investigated. (A) Zn(OH)2(H2O)2C6H5, (B) Zn(OH)2(H2O)2NH3, (C)
Zn(C2H3O2)2C6H5, and (D) Zn(C2H3O2)2NH3. Color scheme, Zn, teal. Adapted from Mayhan, C. M.; Szabo, T. J.; Adams, J. E.; Deakyne, C. A. Comput.
Theor. Chem. 2012, 984, 19–35; Mayhan, C. M.; Szabo, T. J.; Adams, J. E.; Deakyne, C. A. Struct. Chem. 2013, 24, 2089–2099.
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 331

Figure 17 Top views of the 5-coordinate 4-Zn model half capsule (top), 6-coordinate 4-Zn model flattened structure (bottom right), and 5-
coordinate 8-Zn model closed ring (bottom left). Adapted from Mayhan, C. M.; Szabo, T. J.; Adams, J. E.; Deakyne, C. A. Struct. Chem. 2013, 24,
2089–2099.

bond angles.339 The s5 index varies from zero to one as the coordination geometry of a 5-coordinate metal varies from square
pyramidal to trigonal pyramidal.

2.15.7.1.1.2 Zn–L Bond Strengths


Because it is often more expedient to introduce new external metal ligands by replacing existing ones, it is important to know the
relative strengths of relevant metal–ligand interactions. We have started to obtain this information by evaluating the strength of the
interaction between Zn and the series of ligands L noted earlier, chosen because they are typical solvents used in the experiments on
the zinc-seamed pyrogallol[4]arene capsules. Our MP2/B2-PP//M05-2X//B2-PP calculations333 on the Zn(C2H3O2)2L model
complexes yielded binding enthalpies in the order (kJ mol 1)338 C5H5N (96.6) ¼ (CH3)2SO (96.5) > (CH3)2NCHO (73.2) >
CH3OH (57.6). On the basis of these computations, the solvents and bases used in capsule formation were subsequently changed
to avoid replacement of a desired ligand, to improve the introduction of a desired ligand, and to enhance crystallization. Also, the
small magnitude of the binding affinity of CH3OH relative to those of the other solvents clarifies why zinc-seamed pyrogallol[4]
arene nanocapsules with methanol as the external ligand have not been observed experimentally.

2.15.7.1.1.3 Tethering
The magnetic properties and robust nature of metal-seamed pyrogallol[4]arene-based nanocapsules make them ideal candidates for
building blocks in supermolecular constructions. The ring of metals in the capsule “belt” and the relatively facile ligand substitu-
tion25,26 allow for systematic variation of the number, position, and nature of multifunctional ligands and, therefore, for geometric
and dimensional control of the arrays. As such, use of these metal-seamed capsules in the design of structurally diverse and sophis-
ticated multicapsule arrays has been explored to a limited extent. In an early study by the Atwood group, two distinct 1-D arrays
were constructed from preformed Cu2 þ nanocapsules.340 In one system, the capsules are directly linked such that Cu2 þ centers
from adjacent capsules form a [CueO]2 four-membered ring. The second system consists of Cu2 þ-seamed nanocapsules linked
by 4,40 -bipyridyl (bpy) ligands. These results motivated our combined computational and experimental investigations of zinc-
seamed pyrogallol[4]arene-based arrays.341,342
With the success of the models cited earlier in reproducing the Zn coordination environment in the capsules,337,338 the compu-
tational components of the earlier investigations were carried out using the mononuclear model systems. The first of the two studies
was a proof-of-concept study designed to determine whether quantum chemical calculations could be predictive with respect to
332 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

identifying viable tethering ligands. In the second study, the computational approach was applied to 17 additional divergent
ligands, with N, S, or O ligating atoms, to identify ligands likely to lead to interesting tethering motifs (Fig. 18).
With the data for the copper-seamed dimers in mind, we selected a bpy ligand to initiate our computational assessment of zinc-
seamed dimers as possible building blocks in the synthesis of multicapsule arrays.341 The Zn–bpy binding energy for
Zn(C2O2H3)2bpy was calculated to be 95.7 kJ mol 1 at the MP2/B2-PP//M05-2X//B2-PP level of theory, and the falloff in binding
energy was < 4 kJ mol 1 when the bpy was linked to a second zinc complex, (Zn(C2O2H3)2)2bpy. Given these results, using bpy as
a ligand to tether the capsules was recommended, a modification that proved successful.
The earlier study further spurred our interest in developing criteria to guide experimentalists in choosing divergent ligands likely
to link zinc-seamed dimeric nanocapsules in novel 2D arrangements.342 Two energetic criteria were used in assessing the ligands,
namely, that the Zneligand (ZneL) bond dissociation enthalpies in Zn(C2O2H3)2L and (Zn(C2O2H3)2)2L must be  80 kJ mol 1
and that the difference in the first and second ZneL bond dissociation enthalpies must be 5 kJ mol 1. In addition, the closest
contact distance obtained upon formation of the Zn(C2O2H3)2L complex was used as a geometric criterion to screen the divergent
ligands. For ligands that bind through their long axis, this distance must be  10.5 Å, but for ligands that bind through their short
axis, this distance can be as short as z8 Å (Fig. 18). Two of the six prime candidates identified for experimental study were tested,
and two new tethered systems were synthesized, yielding motifs different from one another and from that of the original system.

2.15.7.1.2 Nanocapsules
As for the capsules themselves, the Atwood group has synthesized MONCs composed of pyrogallol-based macrocycles seamed
together by Cu2 þ, Zn2 þ, Co2 þ, Ni2 þ, or Ga3 þ metal centers.25,26,99,343–345 The capsules have been constructed with R ¼ alkyl or
aryl groups on the eCHR linker moieties that connect the pyrogallols of the macrocycles. Most of these dimeric and hexameric
MONCs are essentially spherical in the solid state, the exceptions being the Ga3 þ-containing hexamers, which have a distorted rugby
ball shape. In solution, both the Ga3 þ-seamed and mixed Ga3 þ/Zn2 þ-seamed nanoassemblies rearrange to toroidal
architectures.346
Thus far, our computational studies have focused on the zinc-seamed pyrogallol[4]arene dimeric capsules. To form these
capsules, eight zincs stitch the two cone-shaped macrocycles together with concomitant deprotonation of 16 hydroxyls
(Fig. 19). The loss of two out of three protons on each pyrogallol results in a bridging hydrogen between outer oxygens on neigh-
boring pyrogallols of a given macrocycle. Each outer oxygen is bound to a single Zn center, whereas each medial oxygen is bridged
between two zinc centers. The resulting metal centers are 5-coordinate, where oxygen atoms from the two pyrogallol-based macro-
cycles occupy four of the coordination sites (two apiece), and the fifth coordination site is occupied by an exo equatorial ligand,
typically one that survives from the reactant zinc complex. (The capsules are usually synthesized by a 1:4:16 ratio of zinc complex
to macrocycle to pyridine that initially yields an exo pyridine.) However, facile substitution of the exo ligand has been observed in
solution.25 That the capsule is also stable in the absence of equatorial ligands is clear, given that these ligands are stripped in MALDI-
TOF analyses.26

Figure 18 Examples of the (Zn(C2O2H3)2)2L systems investigated, where L is a divergent ligand. (A) L ¼ bpy and (B) L ¼ 2,6-naphthyridine. Adapted
from Mossine, A. V.; Mayhan, C. M.; Fowler, D. A.; Teat, S. J.; Deakyne, C. A.; Atwood, J. L. Chem. Sci. 2014, 5, 2297–2303; Mayhan, C. M.; Drach-
nik, A. M.; Mossine, A. V.; Kumari, H.; Fowler, D. A.; Barnes, C. L.; Teat, S. J.; Adams, J. E.; Atwood, J. L.; Deakyne, C. A. J. Phys. Chem. C 2016,
120, 13159–13168.
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 333

Figure 19 Representation of [Zn8(C-phenylpyrogallol[4] arene)2 3 pyridineHþ]. Guest atoms and bridging hydrogens are shown as van der Waals
spheres. External ligands have been removed for clarity.

Guided by our earlier calibration studies on the zinc model complexes,337,338 we have carried out an extensive investigation to
determine the levels of theory to be used in our research on the structures and energetics of the zinc-seamed pyrogallol[4]arene
nanocapsules.347 In this investigation, the DFT method, ECP on Zn, and basis set were varied to identify levels of theory that are
reasonable compromises between reliability and required computational resources. The PBE0/LANL2DZ calculational level was
chosen to perform geometry optimizations on the capsules by comparing and contrasting the structural results obtained at this level
of calculation with those obtained from higher-level calculations and experiment. (We note that the PBE0 functional is also desig-
nated PBE1PBE.) In our first reported computational work on the zinc-seamed MONCs, PBE0/LANL2DZ equilibrium structures
were obtained for PgC0ZnNH3, PgC3ZnNH3, PgC0Zn 3 (benzene), PgC0Zn 3 (pyridine), PgC0Zn 3 (pyridineHþ),
PgC0ZnNH3 3 (benzene), and PgC0ZnNH3 3 (pyridineHþ).348 In the PgCnZnNH3 notation, for example, the n designates the
length of the alkyl chain of the R-group, and the NH3 indicates that the exo ligand is ammonia. In our selection of capsules to
examine, the use of benzene as a guest stemmed from an interest in its binding, which primarily derives from dispersion interac-
tions; the use of protonated pyridine and pyridine as guests was in response to experimental results revealing their presence within
some of the capsules.25,26
The later work348 was part of a combined experimental and quantum chemical investigation, in which the goals of the exper-
imental component were the synthesis and characterization of a mixed Zn/Ni dimer MONC starting from a preformed zinc-seamed
C-hexylpyrogallol[4]arene dimer. Such mixed dimers are of particular interest as a result of their potential to exhibit advantageous
magnetic properties. The goals of the computational component were (1) to ascertain how the choice of the R-group, the nature of
the guest, and the presence (or absence) of exo ligands impacts the zinc coordination environment in and metric dimensions of the
MONC and (2) to rationalize the experimental results.
Because their similar electron density makes it difficult to distinguish between Zn and Ni centers in a crystal structure, the identity
of the metal(s) in the SCXRD structure of the PgC6M product of the metal-exchange study was ascertained by comparing and con-
trasting s5 values. That is, the s5 values of the metal centers in a PgCnZn capsule (0.40–0.42)341 are quite distinct from those in
a PgCnNi capsule (0.34).349 As a result, the periphery of a zinc-containing capsule has sharp edges, whereas that of a nickel-
containing capsule is smoother and more circular. The analysis showed that, in fact, no metal exchange occurred. The product
MONC is a PgC6Zn capsule uniquely ligated by two water and six pyridine external ligands rather than by the usual eight pyridine
ligands.25,26 These results demonstrate the challenges faced in synthesizing mixed-metal dimers with this approach.
The results of the computational structural analysis of the seven capsules testify to the difficulty of removing zinc from the pre-
formed capsule.348 No significant weakening of the ZneO bonds in this robust, flexible MONC was found despite the changes in
diameter and the distortions in the framework caused by adding guests or ligands to the assembly. In general, the length of the alkyl
tail is found to have little effect on the capsular metric dimensions, but the nature of the guest and the presence of ligands produce
a noticeable perturbation of the dimensions. The smallest observed diameter, 9.46  0.26 Å, and distorted framework (Fig. 20) of
334 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

Figure 20 Front views of the optimized structures of (A) PgC0ZnNH3, (B) PgC0Zn 3 (pyridine), (C) PgC0Zn 3 (pyridineHþ), and (D)
PgC0ZnNH3 3 (pyridine). Guest atoms are shown as van der Waals spheres. Adapted from Kumari, H.; Mayhan, C. M.; Barnes, C. L.; Deakyne, C. A.;
Atwood, J. L. CrystEngComm. 2016, 18, 4909–4913.

the PgC0Zn 3 (pyridine) are attributed to a strong Zn2 þeN interaction between the Lewis acid host and the Lewis base guest.
Exchanging the pyridine with pyridineHþ causes the diameter to expand to 9.73  0.10 Å and removes the buckling (Fig. 20). Add-
ing NH3 ligands to the PgC0Zn 3 (pyridineHþ) capsule expands the diameter even further, to 9.99  0.07 Å. The trends are similar,
albeit smaller, for the PgC0Zn 3 (benzene) capsule. It is only when all three components of the capsular assembly are included that
the calculated capsular diameters reproduce the experimentally observed diameters (e.g., 9.92  0.08 Å, this work).
The diastereoselective encapsulation of chiral guests in the dissymmetric space of a resorcin[4]arene-based dimeric nanocapsule
has been investigated by circular dichroism (CD) spectroscopy and quantum chemical calculations.107 The resorcinarene cavitands
studied by Tsunoda et al. have the oxygen centers on adjacent resorcinols bridged by a CH2 moiety, R ¼ C11H23, and a eC6H4eC5N-
H3eC5NH4-substituted biphenyl moiety attached to each resorcinol at the C2 position. Each of the four pairs of biphenyl groups,
one biphenyl from each cavitand, coordinates a Agþ ion to build the capsule (Fig. 21). The chiral guests are 4,40 -diacetoxy-6,60 -
dimethyl-[1,10 -biphenyl]-2,20 -dicarboxylate compounds with ester R-groups chosen to tune the steric interactions between the
host and guest. The dibenzyl 4,40 -diacetoxy-6,60 -dimethyl-[1,10 -biphenyl]-2,20 -dicarboxylate guest used in the quantum chemical
calculations also is shown in Fig. 21.
The quantum chemical calculations were performed to establish the absolute configuration of the host capsule in the host–
guest assemblies. The conformational search350 on the host-dibenzyl 4,40 -diacetoxy-6,60 -dimethyl-[1,10 -biphenyl]-2,20 -
dicarboxylate assembly was initiated from geometries produced from a low-mode search.350 After optimization of the resulting
structures using the MMFF force field and the GB/SA solvation model for CH3Cl,351 the most stable structure identified for the
(P)-host-(S)-guest complex was reoptimized at the ONIOM (M06-2X/6-31G/UFF) level of theory (Fig. 21). Finally, excitation
and CD spectra were evaluated for the (P)-host-(S)-guest complex minus the Agþ ions (due to resource limitations), the
biphenyl-substituted resorcin[4]arene cavitand, and the isolated guest employing the ZINDO/S Hamiltonian. On the basis of
the three calculated CD spectra, CD signals in the 250–400 nm range were assigned to the capsule. That a negative value for
the molar CD D3 324 nm indicates P helicity of the capsule and a positive value for D3 324 nm indicates M helicity emerged from
a combination of exciton coupling theory352–354 and the quantum chemical results. Overall, it was found that a diastereoselec-
tivity of the dissymmetric capsule as high as 98% can be realized by encapsulation of chiral guests with electron-donating ester
substituents, for example, aryl substituents. That is, selectivity is enhanced when the ester substituent can participate in p/p
stacking interactions with the electron-deficient biphenyl groups.107

2.15.7.2 Covalently Linked Nanocapsules


We would not wish to imply that metal ions afford the only way of linking macrocycles into nanocapsules. Covalent linkages
also have been explored. To enlarge the framework of and introduce photoresponsive properties into a covalently bonded
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 335

Figure 21 Minimized structures for (A) the isolated dibenzyl 4,40 -diacetoxy-6,60 -dimethyl-[1,10 -biphenyl]-2,20 -dicarboxylate guest, (B) the biphenyl-
substituted resorcin[4]arene cavitand, and (C) the silver-seamed dimeric capsule–guest assembly. Adapted from Tsunoda, Y.; Fukuta, K.; Imamura,
T.; Sekiya, R.; Furuyama, T.; Kobayashi, N.; Haino, T. Angew. Chem. Int. Ed. 2014, 53, 7243–7247.

molecular capsule, Mattay and coworkers355 designed suitably derivatized resorcin[4]arene-based cavitands. Their resorcinarene
derivative (R ¼ n-C4H9) has four OeCH2eO linkages connecting adjacent resorcinol rings, and their dimeric capsule has four
C2eOeC6H4eN]NeC6H4eOeC2 linkages connecting the two cavitands. The NMR spectra in CDCl3 indicated that the azoben-
zene subunits are arranged symmetrically in the dimer and that there is no change in the symmetry of the resorcinarene cavitands
upon formation of the dimer. That each of the azobenzene linkers is in the trans configuration was demonstrated by comparing the
absorption spectra of the bound and unbound linkers. Irradiation of this rigid dimer in benzene solution with 365 nm light con-
verted the all-trans isomer to the all-cis isomer, an isomerization that was reversed by irradiation with 560 nm light. Thus, this cova-
lently bonded molecular capsule has a “switchable wall” that could be exploited in, for example, light-driven chemical reactions.
To further explore the changes in the shape, cavity size, and flexibility of the dimeric capsule caused by the trans to cis conversion
of the azobenzene units, RI-B97-D/TZVP-optimized geometries and relative free energies were obtained for local minima of the all-
trans, the trans,trans,trans,cis, the trans,trans,cis,cis, the trans,cis,cis,cis, and the all-cis forms of the dimer (Fig. 22).356 A preliminary MD
relaxation (MM2 force field322) of the initial geometries was performed before the structure energies were minimized. There is
a smooth decrease found in the stability of the molecular capsule as the number of cis isomers increases, as shown by the following
trend in relative free energies (kJ mol 1): all-trans (0.0) > trans,trans,trans,cis (88.3) > trans,trans,cis,cis (147.3) > trans,cis,cis,cis
(199.2) > all-cis (259.4) forms. (Recall here that larger relative free energy values imply lower relative stability.) The substantial
difference in the stability of the all-trans structure compared with the other structures is consistent with the experimental data.
The all-trans form is twisted to effect CeH/p and p/p stacking interactions among the aromatic rings of the azobenzene
linkers; the all-cis structure is hinged around the N]N double bonds for the same reason. The gradual transformation from a twisted
structure to a folded structure with the conversion to cis N]N bonds causes an increase in the flexibility of the capsule and, in some
cases, a notable increase in the size of the capsule cavity (Fig. 22).355
To examine guest binding within and egress from the covalently bound capsule with a long cylindrical cavity recently synthe-
sized by the Rebek group,357 Tzeli et al.358 performed DFT calculations on the encapsulation of the two long-chain alkanes n-
C18H38 and n-C20H42 and the rigid dibutylstilbene. The capsule comprises two resorcin[4]arene-based deep cavitands connected
through the nitrogen centers of 2,5-dimethyl-1,4-benzenediamine. Benzimidazoles link the oxygen atoms on neighboring resor-
cinols in the cavitand, and the R-group here is C11H23.
The geometries of the capsule and capsule–guest assemblies were optimized at the M06-2X/6-31G(d,p) level of theory, and
BSSE-corrected binding energies were computed at the same level of theory.182 The equilibrium structures of the capsule and
capsule-n-C20H42 assemblies are shown in Fig. 23. Guest encapsulation leads to a more symmetrical (less twisted) but more
336 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

Figure 22 Calculated structures for the (A) all-trans, (B) trans,trans,cis,cis, and (C) all-cis forms of the azobenzene-linked dimer. Adapted from Sahu,
S. N.; Rozhenko, A. B.; Eberhard, J.; Mattay, J. J. Photochem. Photobiol. A 2016, (in press).

outwardly bowed arrangement of the diamine spacers of the capsule. The greater coiling of the chain required to fit n-C20H42 in the
available space lowers its binding energy ( 71.1 kJ mol 1) compared with that of n-C18H38 ( 103.3 kJ mol 1). The binding
energy of dibutylstilbene ( 81.2 kJ mol 1) lies between those of the other two guests. (We note that binding energies also were
calculated at the B3LYP/6-31G(d) level of theory and predict the guests to be unbound.) In each case, the encapsulated guest is
distorted compared with its unbound configuration, but the distortion is most severe for n-C20H42.
The lowest-energy pathway and associated activation barrier for guest extraction were determined at the ONIOM (M06-2X/6-
31G(d,p)/PM6) level of theory. Guest extractions through an opening between adjoining spacer groups of the capsule and, for
the long-chain guests, along the long axis of the capsule were considered. The carbon atoms were extracted one by one until the
guest was completely removed from the inside of the capsule. At each step in the process, the resulting structure was fully optimized.
When the guest exits the capsule between the spacer groups, the capsule walls separate to create a larger opening or gate to facilitate

Figure 23 Minimized structure of the capsule (A) compared with that of the capsule-n-C20H42 assembly (B) showing the bowing of the capsule and
coiling of the guest that occurs on guest encapsulation. Adapted from Tzeli, D.; Petsalakis, I. D.; Theodorakopoulos, G.; Rebek, J. Chem. Phys. Lett.
2015, 633, 99–104.
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 337

guest egress. Concomitantly, the walls opposite to the gate move closer together. The size of the gate depends on the nature of the
guest and the number of carbons extracted, with the largest gates observed for dibutylstilbene. At each point along the pathway,
including when the last carbon has exited the capsule, the guest interacts with the inside and/or outside of the capsule via multiple
CeH/N and dispersion interactions. The activation barriers for this extraction pathway are 63.2, 103.8, and 110.9 kJ mol 1 for n-
C18H38, n-C20H42, and dibutylstilbene, respectively. That dibutylstilbene has the highest barrier is consistent with its having the
slowest observed exchange rate.
Albeit smaller, capsule openings also appear when the guest is extracted along the long axis of the capsule. As one might expect,
this extraction pathway is much less favorable than extraction between the spacer groups. Indeed, the barrier heights are now all
> 210 kJ mol 1.358

2.15.8 Conclusion

In conclusion, we have presented numerous examples testifying to the advantages of pursuing a synergistic experimental–theoretical
approach to the synthesis and characterization of supramolecular systems, focusing primarily on work elucidating the properties of
nanocapsules assembled from macrocyclic building blocks. In some cases, the theoretical work, encompassing both quantum
chemistry studies and MD simulations, elucidated otherwise ambiguous experimental findings or picked apart the multiple effects
that determine the final structure and composition of the system. In other cases, the calculations have provided direction to the
experiments, thereby leading to improved synthetic strategies and an enhanced ability to link particular desired properties with
prospective synthetic targets. One would not exaggerate in saying that computational work (encompassing both quantum chemistry
calculations and MD simulations) has become a standard tool for probing supramolecular systems, as necessary to the full charac-
terization of these systems as are routine analysis techniques.

References

1. Bernardino, R. J.; Cabral, B. J. C. J. Phys. Chem. A 1999, 103, 9080–9085.


2. Botta, B.; Delle Monache, G.; Salvatore, P.; Gasparrini, F.; Villani, C.; Botta, M.; Corelli, F.; Tafi, A.; Gacs-Baitz, E.; Santini, A.; Carvalho, C. F.; Misiti, D. J. Org. Chem. 1997,
62, 932–938.
3. Conn, M. M.; Rebek, J., Jr. Chem. Rev. 1997, 97, 1647–1668.
4. Nakamura, K.; Sheu, C.; Keating, A. E.; Houk, K. N.; Sherman, J. C.; Chapman, R. G.; Jorgensen, W. L. J. Am. Chem. Soc. 1997, 119, 4321–4322.
5. Timmerman, P.; Nierop, K. G. A.; Brinks, E. A.; Verboom, W.; van Veggel, F. C. J. M.; van Hoorn, W. P.; Reinhoudt, D. N. Chem. Eur. J. 1995, 1, 132–143.
6. Valdes, C.; Spitz, U. P.; Toledo, L. M.; Kubik, S. W.; Rebek, J., Jr. J. Am. Chem. Soc. 1995, 117, 12733–12745.
7. Aleman, C.; Casanovas, J. J. Phys. Chem. A 2005, 109, 8049–8054.
8. Aleman, C.; Zanuy, D.; Casanovas, J. J. Org. Chem. 2006, 71, 6952–6957.
9. Botta, B.; Delle Monache, G.; Zappia, G.; Misiti, D.; Baratto, M. C.; Pogni, R.; Gacs-Baitz, E.; Botta, M.; Corelli, F.; Manetti, F.; Tafi, A. J. Org. Chem. 2002, 67, 1178–1183.
10. Datta, A.; Pati, S. K. Chem. Eur. J. 2005, 11, 4961–4969.
11. Dvoráková, H.; Stursa, J.; Cajan, M.; Moravcova, J. Eur. J. Org. Chem. 2006, 2006, 4519–4527.
12. Houk, K. N.; Menzer, S.; Newton, S. P.; Raymo, F. M.; Stoddart, J. F.; Williams, D. J. J. Am. Chem. Soc. 1999, 121, 1479–1487.
13. Houk, K. N.; Nakamura, K.; Sheu, C.; Keating, A. E. Science 1996, 273, 627–629.
14. Kunsági-Máté, S.; Bakonyi, S.; Kollar, L.; Desbat, B. J. Incl. Phenom. Macrocycl. Chem. 2009, 64, 283–288.
15. Nakamura, K.; Houk, K. N. J. Am. Chem. Soc. 1995, 117, 1853–1854.
16. Rozhenko, A. B.; Schoeller, W. W.; Letzel, M. C.; Decker, B.; Agena, C.; Mattay, J. J. Mol. Struct. THEOCHEM 2005, 732, 7–20.
17. Scherlis, D. A.; Marzari, N. J. Am. Chem. Soc. 2005, 127, 3207–3212.
18. Sheu, C.; Houk, K. N. J. Am. Chem. Soc. 1996, 118, 8056–8070.
19. Zanuy, D.; Casanovas, J.; Aleman, C. J. Phys. Chem. B 2006, 110, 9876–9881.
20. Ghoufi, A.; Bonal, C.; Morel, J. P.; Morel-Desrosiers, N.; Malfreyt, P. J. Phys. Chem. B 2004, 108, 5095–5104.
21. Kunsági-Máté, S.; Szabo, K.; Bitter, I.; Nagy, G.; Kollar, L. J. Phys. Chem. A 2005, 109, 5237–5242.
22. Kunsági-Máté, S.; Szabo, K.; Desbat, B.; Bruneel, J.-L.; Bitter, I.; Kollar, L. J. Phys. Chem. B 2007, 111, 7218–7223.
23. Lipkowitz, K. B.; Pearl, G. J. Org. Chem. 1993, 58, 6729–6736.
24. Tolpekina, T. V.; Den Otter, W. K.; Briels, W. J. J. Phys. Chem. B 2003, 107, 14476–14485.
25. Power, N.; Dalgarno, S. J.; Atwood, J. L. Angew. Chem. Int. Ed. 2007, 46, 8601–8604.
26. Power, N. P.; Dalgarno, S. J.; Atwood, J. L. New J. Chem. 2007, 31, 17–20.
27. Böhmer, V. Angew. Chem. Int. Ed. 1995, 34, 713–745.
28. Gutsche, C. D.; Bauer, L. J. J. Am. Chem. Soc. 1985, 107, 6052–6059.
29. Mandolini, L., Ungaro, R., Eds. Calixarenes in Action; Imperial College Press: London, 2000.
30. Konishi, H.; Nakamura, T.; Ohata, K.; Kobayashi, K.; Morikawa, O. Tetrahedron Lett. 1996, 37, 7383–7386.
31. Högberg, A. G. S. J. Am. Chem. Soc. 1980, 102, 6046–6050.
32. Miao, S.; Adams, R. D.; Guo, D.-S.; Zhang, Q.-F. J. Mol. Struct. 2003, 659, 119–128.
33. Middel, O.; Verboom, W.; Hulst, R.; Kooijman, H.; Spek, A. L.; Reinhoudt, D. N. J. Org. Chem. 1998, 63, 8259–8265.
34. Prosvirkin, A. V.; Kazakova, E. K.; Gubaidullin, A. T.; Litvinov, I. A.; Gruner, M.; Habicher, W. D.; Konovalov, A. I. Russ. Chem. Bull. 2005, 54, 2550–2557.
35. Timmerman, P.; Verboom, W.; Reinhoudt, D. N. Tetrahedron 1996, 52, 2663–2704.
36. Tunstad, L. M.; Tucker, J. A.; Dalcanale, E.; Weiser, J.; Bryant, J. A.; Sherman, J. C.; Helgeson, R. C.; Knobler, C. B.; Cram, D. J. J. Org. Chem. 1989, 54, 1305–1312.
37. Gomez-Benitez, V.; Toscano, R. A.; Morales-Morales, D. J. Incl. Phenom. Macrocycl. Chem. 2004, 50, 199–202.
38. Ma, B.-Q.; Zhang, Y.; Coppens, P. Cryst. Growth Des. 2001, 1, 271–275.
39. Sakhaii, P.; Verdier, L.; Ikegami, T.; Griesinger, C. Helv. Chim. Acta 2002, 85, 3895–3908.
40. Weinelt, F.; Schneider, H. J. J. Org. Chem. 1991, 56, 5527–5535.
338 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

41. Antesberger, J.; Cave, G. W. V.; Ferrarelli, M. C.; Heaven, M. W.; Raston, C. L.; Atwood, J. L. Chem. Commun. 2005, 892–894.
42. Deakyne, C. A.; Fowler, D. A.; Atwood, J. L. Chem. Met. Phenolates 2014, 1, 515–547.
43. Funck, M.; Guest, D. P.; Cave, G. W. V. Tetrahedron Lett. 2010, 51, 6399–6402.
44. Maerz, A. K.; Thomas, H. M.; Power, N. P.; Deakyne, C. A.; Atwood, J. L. Chem. Commun. 2010, 46, 1235–1237.
45. Roberts, B. A.; Cave, G. W. V.; Raston, C. L.; Scott, J. L. Green Chem. 2001, 3, 280–284.
46. Yan, C.; Chen, W.; Chen, J.; Jiang, T.; Yao, Y. Tetrahedron 2007, 63, 9614–9620.
47. Maerz, A.K., Synthesis and characterization of host-guest complexes: Metal-organic nanocapsules using aryl-substituted pyrogallol[4]arenes, PhD, University of
Missouri, 2011.
48. Frischmann, P. D.; MacLachlan, M. J. Chem. Soc. Rev. 2013, 42, 871–890.
49. Balzani, V.; Credi, A.; Venturi, M., Molecular machines based on rotaxanes and catenanes, From Non-Covalent Assemblies to Molecular Machines. pp. 159–206, 2011.
50. Diederich, F., Stang, P. J., Tykwinski, R. R., Eds. Modern Supramolecular Chemistry; Strategies for Macrocycle Synthesis; Wiley-VCH: Weinheim, 2008.
51. Ramamurthy, V.; Schanze, K. S., Eds.; Molecular and Supramolecular Photochemistry, vol. 7; Optical Sensors and Switches; 2001. 519 p.
52. Feringa, B. L., Ed. Molecular Switches; Wiley-VCH: Weinheim, 2001.
53. Moran, J. R.; Karbach, S.; Cram, D. J. J. Am. Chem. Soc. 1982, 104, 5826–5828.
54. Moran, J. R.; Ericson, J. L.; Dalcanale, E.; Bryant, J. A.; Knobler, C. B.; Cram, D. J. J. Am. Chem. Soc. 1991, 113, 5707–5714.
55. Cram, D. J.; Choi, H. J.; Bryant, J. A.; Knobler, C. B. J. Am. Chem. Soc. 1992, 114, 7748–7765.
56. Shirtcliff, L. D.; Xu, H.; Diederich, F. Eur. J. Org. Chem. 2010, 2010, 846–855. S846/841-S846/849.
57. Hornung, J.; Fankhauser, D.; Shirtcliff, L. D.; Praetorius, A.; Schweizer, W. B.; Diederich, F. Chem. Eur. J. 2011, 17, 12362–12371. S12362/12361–S12362/12331.
58. Ruiz-Botella, S.; Vidossich, P.; Ujaque, G.; Vicent, C.; Peris, E. Chem. Eur. J. 2015, 21, 10558–10565.
59. Janosi, T. Z.; Makkai, G.; Kegl, T.; Matyus, P.; Kollar, L.; Erostyak, J. J. Fluoresc. 2016, 26, 679–688.
60. Gottschalk, T.; Jarowski, P. D.; Diederich, F. Tetrahedron 2008, 64, 8307–8317.
61. Gottschalk, T.; Jaun, B.; Diederich, F. Angew. Chem. Int. Ed. 2007, 46, 260–264.
62. Azov, V. A.; Beeby, A.; Cacciarini, M.; Cheetham, A. G.; Diederich, F.; Frei, M.; Gimzewski, J. K.; Gramlich, V.; Hecht, B.; Jaun, B.; Latychevskaia, T.; Lieb, A.; Lill, Y.;
Marotti, F.; Schlegel, A.; Schlittler, R. R.; Skinner, P. J.; Seiler, P.; Yamakoshi, Y. Adv. Funct. Mater. 2006, 16, 147–156.
63. Roncucci, P.; Pirondini, L.; Paderni, G.; Massera, C.; Dalcanale, E.; Azov, V. A.; Diederich, F. Chem. Eur. J. 2006, 12, 4775–4784.
64. Frei, M.; Marotti, F.; Diederich, F. Chem. Commun. 2004, 1362–1363.
65. Hooley, R. J.; Van Anda, H. J.; Rebek, J., Jr. J. Am. Chem. Soc. 2006, 128, 3894–3895.
66. Mann, E.; Rebek, J. Tetrahedron 2008, 64, 8484–8487.
67. Purse, B. W.; Butterfield, S. M.; Ballester, P.; Shivanyuk, A.; Rebek, J., Jr. J. Org. Chem. 2008, 73, 6480–6488.
68. Rudkevich, D. M.; Rebek, J., Jr. Eur. J. Org. Chem. 1999, 1999, 1991–2005.
69. Iwasawa, T.; Hooley, R. J.; Rebek, J., Jr. Science 2007, 317, 493–496.
70. Purse, B. W.; Rebek, J., Jr. Proc. Natl. Acad. Sci. 2005, 102, 10777–10782.
71. Restorp, P.; Rebek, J., Jr. J. Am. Chem. Soc. 2008, 130, 11850–11851.
72. Fraschetti, C.; Letzel, M. C.; Filippi, A.; Speranza, M.; Mattay, J. Beilstein J. Org. Chem. 2012, 8, 539–550.
73. Szumna, A. Chem. Soc. Rev. 2010, 39, 4274–4285.
74. Dalgarno, S. J.; Power, N. P.; Atwood, J. L. Org. Nanostruct. 2008, 317–346.
75. Cholewa, P. P.; Dalgarno, S. J. CrystEngComm 2014, 16, 3655–3666.
76. Cram, D. J., Cram, J. M., Eds.; Container Molecules and Their Guests; Royal Society of Chemistry: Cambridge, 1997.
77. McKinlay, R. M.; Thallapally, P. K.; Cave, G. W. V.; Atwood, J. L. Angew. Chem. Int. Ed. 2005, 44, 5733–5736.
78. Shivanyuk, A.; Friese, J. C.; Doering, S.; Rebek, J., Jr. J. Org. Chem. 2003, 68, 6489–6496.
79. Dalgarno, S. J.; Power, N. P.; Atwood, J. L. Coord. Chem. Rev. 2008, 252, 825–841.
80. Avram, L.; Cohen, Y. J. Am. Chem. Soc. 2002, 124, 15148–15149.
81. Avram, L.; Cohen, Y. Org. Lett. 2002, 4, 4365–4368.
82. Avram, L.; Cohen, Y. Org. Lett. 2003, 5, 3329–3332.
83. Avram, L.; Cohen, Y. J. Am. Chem. Soc. 2004, 126, 11556–11563.
84. Avram, L.; Cohen, Y.; Rebek, J., Jr. Chem. Commun. 2011, 47, 5368–5375.
85. Cave, G. W. V.; Dalgarno, S. J.; Antesberger, J.; Ferrarelli, M. C.; McKinlay, R. M.; Atwood, J. L. Supramol. Chem. 2008, 20, 157–159.
86. Gerkensmeier, T.; Iwanek, W.; Agena, C.; Frohlich, R.; Kotila, S.; Nather, C.; Mattay, J. Eur. J. Org. Chem. 1999, 1999, 2257–2262.
87. Kikuchi, Y.; Tanaka, Y.; Sutarto, S.; Kobayashi, K.; Toi, H.; Aoyama, Y. J. Am. Chem. Soc. 1992, 114, 10302–10306.
88. MacGillivray, L. R.; Atwood, J. L. Nature 1997, 389, 469–472.
89. Murayama, K.; Aoki, K. Chem. Commun. 1998, 607–608.
90. Palmer, L. C.; Rebek, J., Jr. Org. Lett. 2005, 7, 787–789.
91. Rose, K. N.; Barbour, L. J.; Orr, G. W.; Atwood, J. L. Chem. Commun. 1998, 407–408.
92. Shivanyuk, A.; Rebek, J., Jr. Proc. Natl. Acad. Sci. U. S. A. 2001, 98, 7662–7665.
93. Shivanyuk, A.; Rissanen, K.; Kolehmainen, E. Chem. Commun. 2000, 1107–1108.
94. Jin, P.; Dalgarno, S. J.; Atwood, J. L. Coord. Chem. Rev. 2010, 254, 1760–1768.
95. Kumari, H.; Deakyne, C. A.; Atwood, J. L. Acc. Chem. Res. 2014, 47, 3080–3088.
96. Kumari, H.; Jin, P.; Deakyne, C. A.; Atwood, J. L. Curr. Org. Chem. 2013, 17, 1481–1488.
97. Kobayashi, K.; Yamanaka, M. Chem. Soc. Rev. 2015, 44, 449–466.
98. Yamanaka, M.; Kobayashi, K. Org. Chem. 2013, 2, 276–289.
99. Dalgarno, S. J.; Power, N. P.; Warren, J. E.; Atwood, J. L. Chem. Commun. 2008, 1539–1541.
100. Jin, P.; Dalgarno, S. J.; Warren, J. E.; Teat, S. J.; Atwood, J. L. Chem. Commun. 2009, 3348–3350.
101. McKinlay, R. M.; Cave, G. W. V.; Atwood, J. L. Proc. Natl. Acad. Sci. U. S. A. 2005, 102, 5944–5948.
102. Munakata, M.; Wu, L. P.; Kuroda-Sowa, T.; Maekawa, M.; Suenaga, Y.; Sugimoto, K.; Ino, I. J. Chem. Soc. Dalton Trans. 1999, 373–378.
103. Solari, E.; Lesueur, W.; Klose, A.; Schenk, K.; Floriani, C.; Chiesi-Villa, A.; Rizzoli, C. Chem. Commun. 1996, 807–808.
104. Sorrell, T. N.; Pigge, F. C.; White, P. S. Inorg. Chem. 1994, 33, 632–635.
105. Fox, O. D.; Dalley, N. K.; Harrison, R. G. Inorg. Chem. 1999, 38, 5860–5863.
106. Yamanaka, M.; Toyoda, N.; Kobayashi, K. J. Am. Chem. Soc. 2009, 131, 9880–9881.
107. Tsunoda, Y.; Fukuta, K.; Imamura, T.; Sekiya, R.; Furuyama, T.; Kobayashi, N.; Haino, T. Angew. Chem. Int. Ed. 2014, 53, 7243–7247.
108. Pirondini, L.; Bertolini, F.; Cantadori, B.; Ugozzoli, F.; Massera, C.; Dalcanale, E. Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 4911–4915.
109. Bibal, B.; Tinant, B.; Declercq, J.-P.; Dutasta, J.-P. Chem. Commun. 2002, 432–433.
110. Eisler, D. J.; Puddephatt, R. J. Inorg. Chem. 2006, 45, 7295–7305.
111. Pinalli, R.; Cristini, V.; Sottili, V.; Geremia, S.; Campagnolo, M.; Caneschi, A.; Dalcanale, E. J. Am. Chem. Soc. 2004, 126, 6516–6517.
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 339

112. Castellano, R. K.; Rudkevich, D. M.; Rebek, J., Jr. J. Am. Chem. Soc. 1996, 118, 10002–10003.
113. Frischmann, P. D.; Facey, G. A.; Ghi, P. Y.; Gallant, A. J.; Bryce, D. L.; Lelj, F.; MacLachlan, M. J. J. Am. Chem. Soc. 2010, 132, 3893–3908.
114. Waller, M. P.; Kruse, H.; Mueck-Lichtenfeld, C.; Grimme, S. Chem. Soc. Rev. 2012, 41, 3119–3128.
115. Cramer, C. J. Essentials of Computational Chemistry: Theories and Models, 2nd ed.; Wiley: Chichester, 2004.
116. Jensen, F. Introduction to Computational Chemistry, 2nd ed.; Wiley: Chichester, 2006.
117. Koch, W.; Holthausen, M. C. A Chemist’s Guide to Density Functional Theory, 2nd ed.; Wiley: Chichester, 2001.
118. Allen, M. P.; Tildesley, D. J. Computer Simulation of Liquids; Oxford University Press: New York, NY, 1987.
119. Frenkel, D.; Smit, B. Understanding Molecular Simulation: From Algorithms to Applications, 2nd ed.; Academic Press: San Diego, CA, 2002.
120. Becke, A. D. J. Chem. Phys. 1993, 98, 5648–5652.
121. Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B Condens. Matter 1988, 37, 785–789.
122. Adamo, C.; Barone, V. J. Chem. Phys. 1999, 110, 6158–6170.
123. Becke, A. D. Phys. Rev. A Gen. Phys. 1988, 38, 3098–3100.
124. Perdew, J. P. Phys. Rev. B 1986, 33, 8822–8824.
125. Zhao, Y.; Truhlar, D. G. J. Chem. Theory Comput. 2006, 2, 1009–1018.
126. Zhao, Y.; Truhlar, D. G. Theor. Chem. Acc. 2008, 120, 215–241.
127. Zhao, Y.; Truhlar, D. G. J. Chem. Phys. 2006, 125, 194101/194101–194101/194118.
128. Adamo, C.; Barone, V. J. Chem. Phys. 1998, 108, 664–675.
129. Boys, S. F.; Bernardi, F. Mol. Phys. 1970, 19, 553–566.
130. Grimme, S. Angew. Chem. Int. Ed. 2008, 47, 3430–3434.
131. Kim, K. S.; Tarakeshwar, P.; Lee, J. Y. Chem. Rev. 2000, 100, 4145–4185.
132. Sinnokrot, M. O.; Valeev, E. F.; Sherrill, C. D. J. Am. Chem. Soc. 2002, 124, 10887–10893.
133. Tsuzuki, S.; Honda, K.; Uchimaru, T.; Mikami, M.; Tanabe, K. J. Am. Chem. Soc. 2002, 124, 104–112.
134. Møller, C.; Plesset, M. S. Phys. Rev. 1934, 46, 618–622.
135. Lee, T. J.; Rice, J. E. Chem. Phys. Lett. 1988, 150, 406–415.
136. Piecuch, P.; Paldus, J. Int. J. Quantum Chem. 1989, 36, 429–453.
137. Purvis, G. D., III; Bartlett, R. J. J. Chem. Phys. 1982, 76, 1910–1918.
138. Raghavachari, K.; Trucks, G. W.; Pople, J. A.; Head-Gordon, M. Chem. Phys. Lett. 1989, 157, 479–483.
139. Scuseria, G. E.; Janssen, C. L.; Schaefer, H. F., III J. Chem. Phys. 1988, 89, 7382–7387.
140. Scuseria, G. E.; Scheiner, A. C.; Lee, T. J.; Rice, J. E.; Schaefer, H. F., III J. Chem. Phys. 1987, 86, 2881–2890.
141. Grimme, S. J. Comput. Chem. 2006, 27, 1787–1799.
142. Chai, J.-D.; Head-Gordon, M. Phys. Chem. Chem. Phys. 2008, 10, 6615–6620.
143. Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77, 3865–3868.
144. Grimme, S.; Antony, J.; Ehrlich, S.; Krieg, H. J. Chem. Phys. 2010, 132, 154104/154101–154104/154119.
145. Tao, J.; Perdew, J. P.; Staroverov, V. N.; Scuseria, G. E. Phys. Rev. Lett. 2003, 91, 146401/146401–146401/146404.
146. For More Information on Basis Sets, WFT Methods, and DFT Methods Available, See for Example: msg.ameslab.gov and Gaussian.inc.
147. Dapprich, S.; Komiromi, I.; Byun, K. S.; Morokuma, K.; Frisch, M. J. J. Mol. Struct. THEOCHEM 1999, 461–462, 1–21.
148. Vreven, T.; Byun, K. S.; Komaromi, I.; Dapprich, S.; Montgomery, J. A., Jr.; Morokuma, K.; Frisch, M. J. J. Chem. Theory Comput. 2006, 2, 815–826.
149. Miertus, S.; Scrocco, E.; Tomasi, J. Chem. Phys. 1981, 55, 117–129.
150. Miertus, S.; Tomasi, J. Chem. Phys. 1982, 65, 239–245.
151. Pascual-Ahuir, J. L.; Silla, E.; Tunon, I. J. Comput. Chem. 1994, 15, 1127–1138.
152. Tomasi, J.; Mennucci, B.; Cammi, R. Chem. Rev. 2005, 105, 2999–3093.
153. Scalmani, G.; Frisch, M. J. J. Chem. Phys. 2010, 132, 114110/114111–114110/114115.
154. Marenich, A. V.; Cramer, C. J.; Truhlar, D. G. J. Phys. Chem. B 2009, 113, 6378–6396.
155. Murphy, P.; Dalgarno, S. J.; Paterson, M. J. J. Phys. Chem. A 2016, 120, 824–839.
156. Murphy, P.; McKinlay, R. G.; Dalgarno, S. J.; Paterson, M. J. J. Phys. Chem. A 2015, 119, 5804–5815.
157. Tzeli, D.; Petsalakis, I. D.; Theodorakopoulos, G.; Ajami, D.; Rebek, J., Jr. Theor. Chem. Acc. 2014, 133, 1–14.
158. Tzeli, D.; Theodorakopoulos, G.; Petsalakis, I. D.; Ajami, D.; Rebek, J. J. Am. Chem. Soc. 2012, 134, 4346–4354.
159. Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A. J. Comput. Chem. 2004, 26, 114.
160. Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A. J. Comput. Chem. 2004, 25, 1157–1174.
161. Alavi, S.; Afagh, N. A.; Ripmeester, J. A.; Thompson, D. L. Chem. Eur. J. 2006, 12, 5231–5237.
162. Hoops, S. C.; Anderson, K. W.; Merz, K. M., Jr. J. Am. Chem. Soc. 1991, 113, 8262–8270.
163. Estiu, G.; Suarez, D.; Merz, K. M., Jr. J. Comput. Chem. 2006, 27, 1240–1262.
164. Fatmi, M. Q.; Hofer, T. S.; Randolf, B. R.; Rode, B. M. J. Chem. Phys. 2005, 123, 054514/054511–054514/054518.
165. Park, H.; Brothers, E. N.; Merz, K. M., Jr. J. Am. Chem. Soc. 2005, 127, 4232–4241.
166. Shah, S. A. A.; Hofer, T. S.; Fatmi, M. Q.; Randolf, B. R.; Rode, B. M. Chem. Phys. Lett. 2006, 426, 301–305.
167. Wang, B.; Merz, K. M., Jr. J. Chem. Theory Comput. 2006, 2, 209–215.
168. Case, D. A.; Darden, T. A.; Cheatham, T. E.; Simmerling, C. L.; Wang, J.; Duke, R. E.; Luo, R.; Merz, K. M.; Pearlman, D. A.; Crowley, M.; Walker, R. C.; Zhang, W.; Wang, B.;
Hayik, S.; Roitberg, A.; Seabra, G.; Wong, K. F.; Paesani, F.; Wu, X.; Brozell, S.; Tsui, V.; Gohlke, H.; Yang, L.; Tan, C.; Mongan, J.; Hornak, V.; Cui, G.; Beroza, P.;
Mathews, D. H.; Schafmeister, C.; Ross, W. S.; Kollman, P. A. AMBER 9; University of California: San Francisco, CA, 2006.
169. Smith, W.; Yong, C. W.; Rodger, P. M. Mol. Simul. 2002, 28, 385–471.
170. Bouchoux, G.; Defaye, D.; McMahon, T.; Likholyot, A.; Mo, O.; Yanez, M. Chem. Eur. J. 2002, 8, 2900–2909.
171. Hunter, E. P. L.; Lias, S. G. J. Phys. Chem. Ref. Data 1998, 27, 413–656.
172. Kuck, D. Chem. Phenols 2003, 1, 259–332.
173. Olah, G. A.; Mo, Y. K. J. Org. Chem. 1973, 38, 353–366.
174. Olah, G. A.; Mo, Y. K. J. Am. Chem. Soc. 1972, 94, 5341–5349.
175. DeFrees, D. J.; McIver, R. T., Jr.; Hehre, W. J. J. Am. Chem. Soc. 1977, 99, 3853–3854.
176. Olasz, A.; Mignon, P.; De Proft, F.; Veszpremi, T.; Geerlings, P. Chem. Phys. Lett. 2005, 407, 504–509.
177. Mayhan, C. M.; Kumari, H.; McClure, E. M.; Liebman, J. F.; Deakyne, C. A. J. Chem. Thermodyn. 2014, 73, 171–177.
178. Kumari, H.; Mayhan, C. M.; Deakyne, C. A., Proton Affinity and Gas-Phase Basicity of Pyrogallol and Phloroglucinol, Manuscript in Preparation, 2016.
179. Curtiss, L. A.; Redfern, P. C.; Raghavachari, K. J. Chem. Phys. 2007, 127, 124105/124101–124105/124108.
180. Fu, Y.; Liu, L.; Yu, H.-Z.; Wang, Y.-M.; Guo, Q.-X. J. Am. Chem. Soc. 2005, 127, 7227–7234.
181. Liu, T.; Liu, M.-M.; Zheng, X.-W.; Du, C.-Y.; Cui, X.-Y.; Wang, L.; Han, L.-L.; Yu, Z.-Y. Tetrahedron 2014, 70, 9033–9040.
340 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

182. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Scalmani, G.; Barone, V.; Mennucci, B.; Petersson, G. A.; Nakatsuji, H.;
Caricato, M.; Li, X.; Hratchian, H. P.; Izmaylov, A. F.; Bloino, J.; Zheng, G.; Sonnenberg, J. L.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.;
Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Vreven, T.; Montgomery, J. A., Jr.; Peralta, J. E.; Ogliaro, F.; Bearpark, M. J.; Heyd, J.; Brothers, E. N.; Kudin, K. N.;
Staroverov, V. N.; Kobayashi, R.; Normand, J.; Raghavachari, K.; Rendell, A. P.; Burant, J. C.; Iyengar, S. S.; Tomasi, J.; Cossi, M.; Rega, N.; Millam, N. J.; Klene, M.;
Knox, J. E.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Martin, R. L.;
Morokuma, K.; Zakrzewski, V. G.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Dapprich, S.; Daniels, A. D.; Farkas, Ö.; Foresman, J. B.; Ortiz, J. V.; Cioslowski, J.; Fox, D. J.
Gaussian 09, Revision E.01; Gaussian, Inc: Wallingford, CT, 2009.
183. Barone, V.; Cossi, M. J. Phys. Chem. A 1998, 102, 1995–2001.
184. Cossi, M.; Rega, N.; Scalmani, G.; Barone, V. J. Comput. Chem. 2003, 24, 669–681.
185. Shenghur, A.; Weber, K. H.; Nguyen, N. D.; Sontising, W.; Tao, F.-M. J. Phys. Chem. A 2014, 118, 11002–11014.
186. George, C.; Strekowski, R. S.; Kleffmann, J.; Stemmler, K.; Ammann, M. Faraday Discuss. 2005, 130, 195–210.
187. Mayer, J. M.; Hrovat, D. A.; Thomas, J. L.; Borden, W. T. J. Am. Chem. Soc. 2002, 124, 11142–11147.
188. Bernardino, R. J.; Costa Cabral, B. J. J. Mol. Struct. THEOCHEM 2001, 549, 253–260.
189. Khedkar, J. K.; Pinjari, R. V.; Gejji, S. P. J. Phys. Chem. A 2011, 115, 10624–10637.
190. Kim, K.; Choe, J.-I. Bull. Korean Chem. Soc. 2009, 30, 837–845.
191. Kim, K.; Park, S. J.; Choe, J.-I. Bull. Korean Chem. Soc. 2008, 29, 1893–1897.
192. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.; Johnson, B. G.; Robb, M. A.; Cheeseman, J. R.; Keith, T.; Petersson, G. A.; Montgomery, J. A.; Raghavachari, K.;
Al-Laham, M. A.; Zakrzewski, V. G.; Ortiz, J. V.; Foresman, J. B.; Cioslowski, J.; Stefanov, B. B.; Nanayakkara, A.; Challacombe, M.; Peng, C. Y.; Ayala, P. Y.; Chen, W.;
Wong, M. W.; Andres, J. L.; Replogle, E. S.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Binkley, J. S.; DeFrees, D. J.; Baker, J.; Stewart, J. P.; Head-Gordon, M.; Gonzalez, C.;
Pople, J. A. Gaussian 94; Gaussian, Inc: Pittsburgh, PA, 1995.
193. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. A.; Stratmann, R. E.; Burant, J. C.;
Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo, C.;
Clifford, S.; Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.;
Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.;
Nanayakkara, A.; Gonzalez, C.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.; Gonzalez, C.; Head-Gordon, M.; Replogle, E. S.;
Pople, J. A. Gaussian 98; Gaussian, Inc: Pittsburgh, PA, 1998.
194. Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, J. A.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.;
Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.;
Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.;
Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.;
Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.;
Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.;
Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03; Gaussian, Inc: Wallingford, CT, 2004.
195. Cambridge Structure Database; Cambridge Crystallographic Data Centre: Cambridge, 2008.
196. Atwood, J. L.; Barbour, L. J.; Heaven, M. W.; Raston, C. L. Angew. Chem. Int. Ed. 2003, 42, 3254–3257.
197. Morikawa, O.; Iyama, E.; Oikawa, T.; Kobayashi, K.; Konishi, H. J. Phys. Org. Chem. 2006, 19, 214–218.
198. Manzano, S.; Zambrano, C. H.; Mendez, M. A.; Dueno, E. E.; Cazar, R. A.; Torres, F. J. Mol. Simul. 2014, 40, 327–334.
199. Thomas, H.M., Computational studies of three dhemical systems, PhD, University of Missouri, 2011.
200. Thomas, H. M.; Emamian, S.; Ellis, L. T.; Smith, P. N.; Deakyne, C. A.; Adams, J. E. Conformational Stability and Interconversion in Pyrogallol[4]Arenes, submitted to J. Phys.
Chem. A, 2016.
201. Emamian, S., Computational studies of five macromolecular systems, PhD, University of Missouri, 2015.
202. Mossine, A.V., Understanding the self-assembly process and behavior of metallopyrogallol[4]arene nanocapsules, PhD, University of Missouri, 2013.
203. Ahn, S.; Park, T. J.; Choe, J.-I. Bull. Korean Chem. Soc. 2014, 35, 1323–1328.
204. Jeon, Y. J.; Kim, H.; Kim, K.; Lee, M.; Han, C.; Kim, K.-J. Bull. Korean Chem. Soc. 2008, 29, 2284–2286.
205. Cazar, R. A.; Torres, F. J. Univ. Sci. (Pontif. Univ. Javeriana, Fac. Cienc.) 2014, 19, 133–137, 135.
206. Drachnik, A.M., A two-fold approach towards understanding properties of two self-assembled macrocycles: Resorcin[4]arene and pyrogallol[4]arene, PhD, University of
Missouri, 2015.
207. Drachnik, A. M.; Youmans, K. N.; Deakyne, C. A.; Adams, J. E. Conformational Stability and Interconversion in Resorcin[4]Arenes, Manuscript in Preparation, 2016.
208. Högberg, A. G. S. J. Org. Chem. 1980, 45, 4498–4500.
209. Mäkinen, M.; Kalenius, E.; Neitola, R.; Rissanen, K.; Vainiotalo, P. Rapid Commun. Mass Spectrom. 2008, 22, 1377–1383.
210. Dauber-Osguthorpe, P.; Roberts, V. A.; Osguthorpe, D. J.; Wolff, J.; Genest, M.; Hagler, A. T. Proteins Struct. Funct. Genet. 1988, 4, 31–47.
211. Stursa, J.; Dvoráková, H.; Smidrkal, J.; Petrícková, H.; Moravcová, J. Tetrahedron Lett. 2004, 45, 2043–2046.
212. Dougherty, D. A. Science 1996, 271, 163–168.
213. Ma, J. C.; Dougherty, D. A. Chem. Rev. 1997, 97, 1303–1324.
214. Gokel, G. W.; Barbour, L. J.; De Wall, S. L.; Meadows, E. S. Coord. Chem. Rev. 2001, 222, 127–154.
215. Ikeda, A.; Shinkai, S. Chem. Rev. 1997, 97, 1713–1734.
216. Lhotak, P.; Nakamura, R.; Shinkai, S. Supramol. Chem. 1997, 8, 333–344.
217. Mueller-Dethlefs, K.; Hobza, P. Chem. Rev. 2000, 100, 143–167.
218. Murayama, K.; Aoki, K. Chem. Commun. 1997, 119–120.
219. Macias, A. T.; Norton, J. E.; Evanseck, J. D. J. Am. Chem. Soc. 2003, 125, 2351–2360.
220. Ahlrichs, R.; Baer, M.; Haeser, M.; Horn, H.; Koelmel, C. Chem. Phys. Lett. 1989, 162, 165–169.
221. Schäfer, A.; Horn, H.; Ahlrichs, R. J. Chem. Phys. 1992, 97, 2571–2577.
222. Schäfer, A.; Huber, C.; Ahlrichs, R. J. Chem. Phys. 1994, 100, 5829–5835.
223. Dunlap, B. I.; Connolly, J. W. D.; Sabin, J. R. J. Chem. Phys. 1979, 71, 3396–3402.
224. Eichkorn, K.; Treutler, O.; Oehm, H.; Haeser, M.; Ahlrichs, R. Chem. Phys. Lett. 1995, 240, 283–290.
225. Vahtras, O.; Almloef, J.; Feyereisen, M. W. Chem. Phys. Lett. 1993, 213, 514–518.
226. Weigend, F.; Haser, M. Theor. Chem. Acc. 1997, 97, 331–340.
227. Weigend, F.; Haser, M.; Patzelt, H.; Ahlrichs, R. Chem. Phys. Lett. 1998, 294, 143–152.
228. Posligua, V.; Urbina, A. S.; Rincon, L.; Soetens, J. C.; Mendez, M. A.; Zambrano, C. H.; Torres, F. J. Comput. Theor. Chem. 2015, 1073, 75–83.
229. Kendall, R. A.; Dunning, T. H., Jr.; Harrison, R. J. J. Chem. Phys. 1992, 96, 6796–6806.
230. Woon, D. E.; Dunning, T. H., Jr. J. Chem. Phys. 1993, 98, 1358–1371.
231. Kocman, M.; Jurecka, P.; Dubecky, M.; Otyepka, M.; Cho, Y.; Kim, K. S. Phys. Chem. Chem. Phys. 2015, 17, 6423–6432.
Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes 341

232. Bakic, M. T.; Jadresko, D.; Hrenar, T.; Horvat, G.; Pozar, J.; Galic, N.; Sokol, V.; Tomas, R.; Alihodzic, S.; Zinic, M.; Frkanec, L.; Tomisic, V. RSC Adv. 2015, 5,
23900–23914.
233. Jorgensen, W. L.; Maxwell, D. S.; Tirado-Rives, J. J. Am. Chem. Soc. 1996, 118, 11225–11236.
234. Hess, B.; Kutzner, C.; van der Spoel, D.; Lindahl, E. J. Chem. Theory Comput. 2008, 4, 435–447.
235. Stewart, J. J. P. J. Mol. Model. 2007, 13, 1173–1213.
236. Stoll, I.; Eberhard, J.; Brodbeck, R.; Eisfeld, W.; Mattay, J. Chem. Eur. J. 2008, 14, 1155–1163.
237. Stevens, W. J.; Krauss, M.; Basch, H.; Jasien, P. G. Can. J. Chem. 1992, 70, 612–630.
238. Bergner, A.; Dolg, M.; Kuechle, W.; Stoll, H.; Preuss, H. Mol. Phys. 1993, 80, 1431–1441.
239. Kalenius, E.; Kekalainen, T.; Neitola, R.; Beyeh, K.; Rissanen, K.; Vainiotalo, P. Chem. Eur. J. 2008, 14, 5220–5228.
240. Grimme, S.; Ehrlich, S.; Goerigk, L. J. Comput. Chem. 2011, 32, 1456–1465.
241. Schultz, N. E.; Zhao, Y.; Truhlar, D. G. J. Phys. Chem. A 2005, 109, 11127–11143.
242. Patel, M. B.; Valand, N. N.; Modi, N. R.; Joshi, K. V.; Harikrishnan, U.; Kumar, S. P.; Jasrai, Y. T.; Menon, S. K. RSC Adv. 2013, 3, 15971–15981.
243. Hanwell, M. D.; Curtis, D. E.; Lonie, D. C.; Vandermeersch, T.; Zurek, E.; Hutchison, G. R. J. Cheminform. 2012, 4, 17.
244. Acton, A.; Banck, M.; Bréfort, J.; Cruz, M.; Curtis, D.; Hassinen, T.; Heikkilä, V.; Hutchison, G.; Huuskonen, J.; Jensen, J.; Liboska, R.; Rowley, C. Ghemical Force Field, 2006.
245. Krieger, E.; Darden, T.; Nabuurs, S. B.; Finkelstein, A.; Vriend, G. Proteins Struct. Funct. Bioinf. 2004, 57, 678–683.
246. Duan, Y.; Wu, C.; Chowdhury, S.; Lee, M. C.; Xiong, G.; Zhang, W.; Yang, R.; Cieplak, P.; Luo, R.; Lee, T.; Caldwell, J.; Wang, J.; Kollman, P. J. Comput. Chem. 2003, 24,
1999–2012.
247. Dyker, G.; Dietz, C.; Grimme, S.; Oppel, I. M.; Richter, M. Tetrahedron 2015, 71, 5830–5834.
248. Weigend, F.; Ahlrichs, R. Phys. Chem. Chem. Phys. 2005, 7, 3297–3305.
249. Weigend, F.; Furche, F.; Ahlrichs, R. J. Chem. Phys. 2003, 119, 12753–12762.
250. MacroModel, V., Schrödinger LLC, New York, 2009.
251. Gottschalk, T., PhD, Eidgençssische Technische Hochschule Zürich, 2008.
252. Kleywegt, G. J. VOIDOO, version 3.3.4; Uppsala Software Factory: Uppsala, 2007.
253. Stewart, J. J. P. J. Comput. Chem. 1989, 10, 221–264.
254. Stewart, J. J. P. J. Comput. Chem. 1989, 10, 209–220.
255. Mecozzi, S.; Rebek, J., Jr. Chem. Eur. J. 1998, 4, 1016–1022.
256. Halgren, T. A. J. Comput. Chem. 1996, 17, 490–519.
257. MacroModel;, Schrödinger LLC, New York, NY, USA, 2007.
258. Azov, V. A.; Jaun, B.; Diederich, F. Helv. Chim. Acta 2004, 87, 449–462.
259. Skinner, P. J.; Cheetham, A. G.; Beeby, A.; Gramlich, V.; Diederich, F. Helv. Chim. Acta 2001, 84, 2146–2153.
260. VandeVondele, J.; Hutter, J. J. Chem. Phys. 2007, 127, 114105/114101–114105/114109.
261. VandeVondele, J.; Krack, M.; Mohamed, F.; Parrinello, M.; Chassaing, T.; Hutter, J. Comput. Phys. Commun. 2005, 167, 103–128.
262. Ehlers, A. W.; Boehme, M.; Dapprich, S.; Gobbi, A.; Hoellwarth, A.; Jonas, V.; Koehler, K. F.; Stegmann, R.; Veldkamp, A.; et al. Chem. Phys. Lett. 1993, 208, 111–114.
263. Andrae, D.; Haeussermann, U.; Dolg, M.; Stoll, H.; Preuss, H. Theor. Chim. Acta 1990, 77, 123–141.
264. Botta, B.; Botta, M.; Filippi, A.; Tafi, A.; Delle Monache, G.; Speranza, M. J. Am. Chem. Soc. 2002, 124, 7658–7659.
265. Botta, B.; Fraschetti, C.; Novara, F. R.; Tafi, A.; Sacco, F.; Mannina, L.; Sobolev, A. P.; Mattay, J.; Letzel, M. C.; Speranza, M. Org. Biomol. Chem. 2009, 7, 1798–1806.
266. Botta, B.; Tafi, A.; Caporuscio, F.; Botta, M.; Nevola, L.; D’Acquarica, I.; Fraschetti, C.; Speranza, M. Chem. Eur. J. 2008, 14, 3585–3595.
267. Filippi, A.; Fraschetti, C.; Piccirillo, S.; Rondino, F.; Botta, B.; D’Acquarica, I.; Calcaterra, A.; Speranza, M. Chem. Eur. J. 2012, 18, 8320–8328. S8320/8321–S8320/8329.
268. Fraschetti, C.; Letzel, M. C.; Paletta, M.; Mattay, J.; Speranza, M.; Filippi, A.; Aschi, M.; Rozhenko, A. B. J. Mass Spectrom. 2012, 47, 72–78.
269. Speranza, M.; D’Acquarica, I.; Fraschetti, C.; Botta, B.; Tafi, A.; Bellucci, L.; Zappia, G. Int. J. Mass Spectrom. 2010, 291, 84–89.
270. Tafi, A.; Botta, B.; Botta, M.; Delle Monache, G.; Filippi, A.; Speranza, M. Chem. Eur. J. 2004, 10, 4126–4135.
271. Grajda, M.; Wierzbicki, M.; Cmoch, P.; Szumna, A. J. Org. Chem. 2013, 78, 11597–11601.
272. Jedrzejewska, H.; Kwit, M.; Szumna, A. Chem. Commun. 2015, 51, 13799–13801.
273. Jedrzejewska, H.; Wierzbicki, M.; Cmoch, P.; Rissanen, K.; Szumna, A. Angew. Chem. Int. Ed. 2014, 53, 13760–13764.
274. Kuberski, B.; Pecul, M.; Szumna, A. Eur. J. Org. Chem. 2008, 2008, 3069–3078.
275. Atwood, J. L.; Barbour, L. J.; Jerga, A. Angew. Chem. Int. Ed. 2004, 43, 2948–2950.
276. Atwood, J. L.; Barbour, L. J.; Jerga, A. Method of Separating and Storing Volatile Gases, (The Curators of the University of Missouri Office of Technology and Special Projects,
USA), Application: WO 2006, 32 p.
277. Atwood, J. L.; Barbour, L. J.; Jerga, A.; Schottel, B. L. Science 2002, 298, 1000–1002.
278. Atwood, J. L.; Barbour, L. J.; Thallapally, P. K.; Wirsig, T. B. Chem. Commun. 2005, 51–53.
279. Dalgarno, S. J.; Thallapally, P. K.; Barbour, L. J.; Atwood, J. L. Chem. Soc. Rev. 2007, 36, 236–245.
280. Thallapally, P. K.; Dalgarno, S. J.; Atwood, J. L. J. Am. Chem. Soc. 2006, 128, 15060–15061.
281. Thallapally, P. K.; Dobrzanska, L.; Gingrich, T. R.; Wirsig, T. B.; Barbour, L. J.; Atwood, J. L. Angew. Chem. Int. Ed. 2006, 45, 6506–6509.
282. Thallapally, P. K.; Kirby, K. A.; Atwood, J. L. New J. Chem. 2007, 31, 628–630.
283. Thallapally, P. K.; Lloyd, G. O.; Wirsig, T. B.; Bredenkamp, M. W.; Atwood, J. L.; Barbour, L. J. Chem. Commun. 2005, 5272–5274.
284. Thallapally, P. K.; McGrail, B. P.; Atwood, J. L. Chem. Commun. 2007, 1521–1523.
285. Thallapally, P. K.; Wirsig, T. B.; Barbour, L. J.; Atwood, J. L. Chem. Commun. 2005, 4420–4422.
286. Adams, J. E.; Cox, J. R.; Christiano, A. J.; Deakyne, C. A. J. Phys. Chem. A 2008, 112, 6829–6839.
287. Bayly, C. I.; Cieplak, P.; Cornell, W.; Kollman, P. A. J. Phys. Chem. 1993, 97, 10269–10280.
288. Cornell, W. D.; Cieplak, P.; Bayly, C. I.; Gould, I. R.; Merz, K. M., Jr.; Ferguson, D. M.; Spellmeyer, D. C.; Fox, T.; Caldwell, J. W.; Kollman, P. A. J. Am. Chem. Soc. 1995,
117, 5179–5197.
289. Grossfield, A., An implementation of WHAM: the weighted histogram analysis method, version 2.0.2, 2008.
290. Trzesniak, D.; Kunz, A.-P. E.; van Gunsteren, W. F. ChemPhysChem 2007, 8, 162–169.
291. Breite, M. D.; Cox, J. R.; Adams, J. E. J. Am. Chem. Soc. 2010, 132, 10996–10997.
292. Daschbach, J. L.; Thallapally, P. K.; Atwood, J. L.; McGrail, B. P.; Dang, L. X. J. Chem. Phys. 2007, 127, 104703/104701–04703/104704.
293. Atwood, J. L.; Barbour, L. J.; Jerga, A. Chem. Commun. 2002, 2952–2953.
294. Kimura, E. Acc. Chem. Res. 2001, 34, 171–179.
295. Makowska-Grzyska, M. M.; Jeppson, P. C.; Allred, R. A.; Arif, A. M.; Berreau, L. M. Inorg. Chem. 2002, 41, 4872–4887.
296. Salter, M. H., Jr.; Reibenspies, J. H.; Jones, S. B.; Hancock, R. D. Inorg. Chem. 2005, 44, 2791–2797.
297. Dalgarno, S. J.; Szabo, T.; Siavosh-Haghighi, A.; Deakyne, C. A.; Adams, J. E.; Atwood, J. L. Chem. Commun. 2009, 1339–1341.
298. Dalgarno, S. J.; Tucker, S. A.; Bassil, D. B.; Atwood, J. L. Science 2005, 309, 2037–2039.
299. Dalgarno, S. J.; Bassil, D. B.; Tucker, S. A.; Atwood, J. L. Angew. Chem. Int. Ed. 2006, 45, 7019–7022.
300. Dunning, T. H., Jr. J. Chem. Phys. 1989, 90, 1007–1023.
342 Computational Studies of Supramolecular Systems: Resorcinarenes and Pyrogallolarenes

301. Heinz, T.; Rudkevich, D. M.; Rebek, J., Jr. Nature 1998, 394, 764–766.
302. Wang, J.; Yang, Z.; Wang, X.; Zhang, J.; Cao, W. J. Mol. Struct. THEOCHEM 2006, 772, 39–44.
303. Wang, J.; Yang, Z.; Wang, X.; Zhang, J.; Cao, W. J. Mol. Struct. THEOCHEM 2007, 819, 153–159.
304. Hayashida, O.; Sebo, L.; Rebek, J., Jr. J. Org. Chem. 2002, 67, 8291–8298.
305. Körner, S. K.; Tucci, F. C.; Rudkevich, D. M.; Heinz, T.; Rebek, J., Jr. Chem. Eur. J. 2000, 6, 187–195.
306. Albunia, A. R.; Gaeta, C.; Neri, P.; Grassi, A.; Milano, G. J. Phys. Chem. B 2006, 110, 19207–19214.
307. Shivanyuk, A.; Rebek, J., Jr. J. Am. Chem. Soc. 2002, 124, 12074–12075.
308. Ajami, D.; Iwasawa, T.; Rebek, J., Jr. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 8934–8936.
309. Tzeli, D.; Petsalakis, I. D.; Theodorakopoulos, G.; Ajami, D.; Jiang, W.; Rebek, J., Jr. Chem. Phys. Lett. 2012, 548, 55–59.
310. Tzeli, D.; Petsalakis, I. D.; Theodorakopoulos, G.; Ajami, D.; Rebek, J. Int. J. Quantum Chem. 2013, 113, 734–739.
311. Tzeli, D.; Theodorakopoulos, G.; Petsalakis, I. D.; Ajami, D.; Rebek, J. J. Am. Chem. Soc. 2011, 133, 16977–16985.
312. Ayhan, M. M.; Casano, G.; Karoui, H.; Rockenbauer, A.; Monnier, V.; Hardy, M.; Tordo, P.; Bardelang, D.; Ouari, O. Chem. Eur. J. 2015, 21, 16404–16410.
313. Ebbing, M. H. K.; Villa, M.-J.; Valpuesta, J.-M.; Prados, P.; De Mendoza, J. Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 4962–4966.
314. Zhang, K.-D.; Ajami, D.; Gavette, J. V.; Rebek, J. J. Am. Chem. Soc. 2014, 136, 5264–5266.
315. Zhang, K.-D.; Ajami, D.; Rebek, J. J. Am. Chem. Soc. 2013, 135, 18064–18066.
316. Gibb, C. L. D.; Gibb, B. C. J. Am. Chem. Soc. 2004, 126, 11408–11409.
317. Choudhury, R.; Barman, A.; Prabhakar, R.; Ramamurthy, V. J. Phys. Chem. B 2013, 117, 398–407.
318. Kulasekharan, R.; Choudhury, R.; Prabhakar, R.; Ramamurthy, V. Chem. Commun. 2011, 47, 2841–2843.
319. Samanta, S. R.; Choudhury, R.; Ramamurthy, V. J. Phys. Chem. A 2014, 118, 10554–10562.
320. Lindahl, E.; Hess, B.; van der Spoel, D. J. Mol. Model. 2001, 7, 306–317.
321. Van Der Spoel, D.; Lindahl, E.; Hess, B.; Groenhof, G.; Mark, A. E.; Berendsen, H. J. C. J. Comput. Chem. 2005, 26, 1701–1718.
322. Allinger, N. L. J. Am. Chem. Soc. 1977, 99, 8127–8134.
323. Trott, O.; Olson, A. J. J. Comput. Chem. 2010, 31, 455–461.
324. Tero, T.-R.; Salorinne, K.; Malola, S.; Hakkinen, H.; Nissinen, M. CrystEngComm 2015, 17, 8231–8241.
325. Giannozzi, P.; Baroni, S.; Bonini, N.; Calandra, M.; Car, R.; Cavazzoni, C.; Ceresoli, D.; Chiarotti Guido, L.; Cococcioni, M.; Dabo, I.; Dal Corso, A.; De Gironcoli, S.; Fabris, S.;
Fratesi, G.; Gebauer, R.; Gerstmann, U.; Gougoussis, C.; Kokalj, A.; Lazzeri, M.; Martin-Samos, L.; Marzari, N.; Mauri, F.; Mazzarello, R.; Paolini, S.; Pasquarello, A.;
Paulatto, L.; Sbraccia, C.; Scandolo, S.; Sclauzero, G.; Seitsonen Ari, P.; Smogunov, A.; Umari, P.; Wentzcovitch Renata, M. J. Phys. Condens. Matter 2009, 21, 395502.
326. Bunn, S. E.; Liu, C. T.; Lu, Z.-L.; Neverov, A. A.; Brown, R. S. J. Am. Chem. Soc. 2007, 129, 16238–16248.
327. Chen, S.-L.; Fang, W.-H.; Himo, F. J. Inorg. Biochem. 2009, 103, 274–281.
328. Jahn, B. O.; Eger, W. A.; Anders, E. Z. Naturforsch. B 2010, 65, 425–432.
329. Li, D.; Zheng, Q.; Fang, X.; Ji, H.; Yang, J.; Zhang, H. Polymer 2008, 49, 3346–3351.
330. Liao, R.-Z.; Yu, J.-G.; Himo, F. Inorg. Chem. 2010, 49, 6883–6888.
331. Parkin, G. Chem. Rev. 2004, 104, 699–767.
332. Vahrenkamp, H. Acc. Chem. Res. 1999, 32, 589–596.
333. Amin, E. A.; Truhlar, D. G. J. Chem. Theory Comput. 2008, 4, 75–85.
334. Frison, G.; Ohanessian, G. J. Comput. Chem. 2008, 29, 416–433.
335. Rayón, V. M.; Valdés, H.; Diaz, N.; Suárez, D. J. Chem. Theory Comput. 2008, 4, 243–256.
336. Sorkin, A.; Truhlar, D. G.; Amin, E. A. J. Chem. Theory Comput. 2009, 5, 1254–1265.
337. Mayhan, C. M.; Szabo, T. J.; Adams, J. E.; Deakyne, C. A. Comput. Theor. Chem. 2012, 984, 19–35.
338. Mayhan, C. M.; Szabo, T. J.; Adams, J. E.; Deakyne, C. A. Struct. Chem. 2013, 24, 2089–2099.
339. Addison, A. W.; Rao, T. N.; Reedijk, J.; Van Rijn, J.; Verschoor, G. C. J. Chem. Soc. Dalton Trans. 1984, 1349–1356.
340. Fowler, D. A.; Mossine, A. V.; Beavers, C. M.; Teat, S. J.; Dalgarno, S. J.; Atwood, J. L. J. Am. Chem. Soc. 2011, 133, 11069–11071.
341. Mossine, A. V.; Mayhan, C. M.; Fowler, D. A.; Teat, S. J.; Deakyne, C. A.; Atwood, J. L. Chem. Sci. 2014, 5, 2297–2303.
342. Mayhan, C. M.; Drachnik, A. M.; Mossine, A. V.; Kumari, H.; Fowler, D. A.; Barnes, C. L.; Teat, S. J.; Adams, J. E.; Atwood, J. L.; Deakyne, C. A. J. Phys. Chem. C 2016, 120,
13159–13168.
343. Atwood, J. L.; Brechin, E. K.; Dalgarno, S. J.; Inglis, R.; Jones, L. F.; Mossine, A.; Paterson, M. J.; Power, N. P.; Teat, S. J. Chem. Commun. 2010, 46, 3484–3486.
344. McKinlay, R. M.; Thallapally, P. K.; Atwood, J. L. Chem. Commun. 2006, 2956–2958.
345. Rathnayake, A. S.; Feaster, K. A.; White, J.; Barnes, C. L.; Teat, S. J.; Atwood, J. L. Cryst. Growth Des. 2016, 16, 3562–3564.
346. Kumari, H.; Kline, S. R.; Wycoff, W. G.; Paul, R. L.; Mossine, A. V.; Deakyne, C. A.; Atwood, J. L. Angew. Chem. Int. Ed. 2012, 51, 5086–5091.
347. Mayhan, C.M., Computational studies of zinc-seamed pyrogallol[4]arene nanocapsules and model systems, PhD, University of Missouri, 2014.
348. Kumari, H.; Mayhan, C. M.; Barnes, C. L.; Deakyne, C. A.; Atwood, J. L. CrystEngComm 2016, 18, 4909–4913.
349. Kumari, H.; Mossine, A. V.; Kline, S. R.; Dennis, C. L.; Fowler, D. A.; Teat, S. J.; Barnes, C. L.; Deakyne, C. A.; Atwood, J. L. Angew. Chem. Int. Ed. 2012, 51, 1452–1454.
350. Mohamadi, F.; Richards, N. G. J.; Guida, W. C.; Liskamp, R.; Lipton, M.; Caufield, C.; Chang, G.; Hendrickson, T.; Still, W. C. J. Comput. Chem. 1990, 11, 440–467.
351. Qiu, D.; Shenkin, P. S.; Hollinger, F. P.; Still, W. C. J. Phys. Chem. A 1997, 101, 3005–3014.
352. Brown, H. C.; Okamoto, Y. J. Am. Chem. Soc. 1958, 80, 4979–4987.
353. Okamoto, Y.; Brown, H. C. J. Org. Chem. 1957, 22, 485–494.
354. Chiu, S.-H.; Liao, K.-S.; Su, J.-K. Tetrahedron Lett. 2004, 45, 213–216.
355. Sahu, S. N.; Rozhenko, A. B.; Eberhard, J.; Mattay, J. J. Photochem. Photobiol. A 2016. Ahead of Print. http://dx.doi.org/10.1016/j.jphotochem.2016.01.008.
356. Furche, F.; Ahlrichs, R.; Haettig, C.; Klopper, W.; Sierka, M.; Weigend, F. Wiley Interdiscip. Rev. Comput. Mol. Sci. 2014, 4, 91–100.
357. Asadi, A.; Ajami, D.; Rebek, J. Chem. Sci. 2013, 4, 1212–1215.
358. Tzeli, D.; Petsalakis, I. D.; Theodorakopoulos, G.; Rebek, J. Chem. Phys. Lett. 2015, 633, 99–104.
2.16 Electrochemically Controlled Supramolecular Switches and Machines
G Ragazzon, M Baroncini, and P Ceroni, University of Bologna, Bologna, Italy
A Credi and M Venturi, University of Bologna, Bologna, Italy; and Institute of Organic Synthesis and Photoreactivity, National
Research Council, Bologna, Italy
Ó 2017 Elsevier Ltd. All rights reserved.

2.16.1 Introduction 344


2.16.2 Electrochemical Analysis of Supramolecular Systems 344
2.16.2.1 Molecular Encapsulation of Electroactive Units 344
2.16.2.2 Redox-Controlled Supramolecular Switching 345
2.16.2.3 Electroactive Species on a Solid Support 346
2.16.2.4 Electroactive Nonequilibrium Systems 346
2.16.3 Self-Assembling Systems 346
2.16.3.1 Hydrogen and Halogen-Bonding Interactions 346
2.16.3.2 Metal Ion Coordination 347
2.16.3.3 Hydrophobic Interactions 350
2.16.3.4 Charge–Transfer Interactions 351
2.16.3.5 Cation Radicals Dimerization 353
2.16.3.6 Systems Working on Surfaces 353
2.16.4 Mechanically Interlocked Molecules 354
2.16.4.1 Systems Based on Rotaxanes 356
2.16.4.2 Systems Based on Catenanes 358
2.16.4.3 Systems Working on Surfaces 359
2.16.5 Nonequilibrium Systems 362
2.16.5.1 Investigation of Molecular Machines at the Single-Molecule Level 364
2.16.6 Concluding Remarks 366
Acknowledgments 366
References 366

Nomenclature
A, D (and their combinations) Acceptor, donor K Equilibrium constant
AD2 Bis(adamantyl)-rhodamine LFSE Ligand field stabilization energy
BIQ 8,80 -Diphenyl-3,30 -biisoquinoline moiety LUMO Lowest unoccupied molecular orbital
C, E (and their combinations) Chemical and MV2 D N,N0 -Dimethyl-4,40 -bipyridinium (methyl viologen
electrochemical steps of a reaction sequence or paraquat)
CB[n] Cucurbit[n]uril NDI Naphthalene diimide
CD Cyclodextrin PMI Pyrometallic diimide
CT Charge transfer PY 3,5-Dimethylpyridinium moiety, see Fig. 27
DMF Dimethylformamide SAM Self-assembled monolayer
DOB Dioxybenzene SCE Saturated calomel electrode
DON Dioxynaphthalene STM Scanning tunneling microscope
E1/2 Halfwave reduction potential TCQ Tetrachloro-p-quinone
FcCAL Ferrocene carboxylate anion TTF Tetrathiafulvalene
G Guest V2 D Viologen moiety
H Host Vn Functionalized viologen moieties, see Fig. 24

Units of Measurement
A Ampere m Meter
Degree Angle mol Mole
J Joule N Newton
K Kelvin degree s Second
M Molar concentration V Volt

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12498-9 343


344 Electrochemically Controlled Supramolecular Switches and Machines

2.16.1 Introduction

Supramolecular chemistry has been defined as “the chemistry beyond the molecule, bearing on organized entities of higher
complexity that result from the association of two or more chemical species held together by intermolecular forces.” 1 However,
as the field developed, it became clear that such a definition would be limiting, as it is strictly based on the type of bond that holds
the different components together. Indeed also in many systems held together by covalent or mechanical bonds, the components
retain their individual properties, therefore it seemed fair to discriminate between molecular and supramolecular entities on the
basis of intercomponent interactions rather than the type of bond that holds together the different units.2,3 Irrespective of defini-
tions, it is possible to say that in supramolecular chemistry the emerging of new functions shifted the focus from molecules to
molecular assemblies or multicomponent structures.
Macroscopic devices and machines are objects that perform an elaborated task, thanks to the complementary role and action
of the components that constitute them. On this basis, the idea that the concepts of “device” and “machine” could be applied at
the molecular level 4 has been proposed in different laboratories.5–7 In this analogy, the molecular components are assembled in
supramolecular species that exhibit novel and elaborated functions that cannot be performed by each component alone. This
attractive idea has been implemented in the last decades, leading to the establishment of a lively research area.8 Indeed, the
molecular bottom-up construction of nanoscale devices and machines has become one of the most stimulating challenges of
nanoscience.
Key areas of chemistry that contributed to the development of this field have been synthetic chemistry, along with substantial
progresses in the characterization of complex chemical systems, not only in solution, but also at interfaces. Electrochemistry is surely
among the most powerful tools for characterizing a supramolecular system; moreover, it can also be used to cause changes on it.
Indeed as for macroscopic devices, energy is needed to put the device at work, and electronic inputs are among the most widely
employed ones. This dual role is conveniently described in terms of “reading” (characterizing) and “writing” (operating) processes. 9
On top of this, electrochemistry offers convenient ways to interface molecular-level devices and machines with the macroscopic
world through the modification of the electrodes. For these reasons, electrochemistry has played a central role in supramolecular
chemistry, generating a significant amount of literature on the topic, promoting the role of supramolecular chemistry in several
fields, includingdbut not limited todenergy conversion and storage, communication and information technology, smart materials
and diagnostics.
The scope of this article is to present an overview on how supramolecular chemists have been exploiting the potential of elec-
trochemistry, both as an actuating stimulus and as an analytical tool. To begin with, a general discussion of the role of electrochem-
istry from the perspective of molecular recognition is presented, with a focus on the electrochemical switching of supramolecular
interactions. Examples of supramolecular systems that exploit electrochemical stimuli will then be introduced. Finally, a specific
section will be dedicated to highlight the properties of electrochemically operated systems that work away from thermodynamic
equilibrium. We are aware that given the huge body of research on the topic, the present article cannot be comprehensive, and
our apologies go to the colleagues who may feel that their work has been omitted. Examples from the literature were selected to
support the logical development of the article. A considerable body of work, including books10–17 and reviews,18–25 is available
for the readers who are interested to further extend their knowledge of the principles and implementations of electrochemistry
in supramolecular systems.

2.16.2 Electrochemical Analysis of Supramolecular Systems

Electrochemical techniques are intrinsically related to addition or removal of electrons from the investigated system. There-
fore for molecular species they reflect the properties of highest occupied molecular orbital and lowest unoccupied molec-
ular orbital (LUMO), whose energy is correlated to the reduction and oxidation potential of the molecule. Along with
redox potentials, other kinetic and thermodynamic parameters can be obtained, not only for the electron-transfer process
itself, but also for chemical reactions that are coupled with the electron-transfer process. Having access to such quantities is
particularly useful to characterize the systems and to understand how their properties are affected by the interaction
between the components.
Finally it is worth highlighting that an electronic stimulus can be due to an electron-transfer between an electrode and the target
species, but can also occur upon interaction with another chemical compound, or upon photonic stimulation. In other words, elec-
tronic stimulation is very versatile, as it can be indirectly (though easily) induced by other most commonly used stimuli. On the
other hand, orthogonal (i.e., fully independent) switching of different electroactive units in the same system is thermodynamically
not allowed, whereas chemical and photochemical stimuli offer this opportunity: to exploit multiple electrochemical switches
within the same system, sequential switching upon progressive potential increment (or decrease) is an available option, as will
be illustrated in selected examples.

2.16.2.1 Molecular Encapsulation of Electroactive Units


Supramolecular interactions include hydrogen bonding, dipole–dipole interactions, Van der Waals forces, metal–ligand coordina-
tion, and p–p interactions that are largely electrostatic in nature. This feature is at the basis of the dual role of electrochemical
Electrochemically Controlled Supramolecular Switches and Machines 345

techniques as a tool to recognize the state of the system and to influence it. Indeed, in the latter case an electronic input should
directly affect the supramolecular interaction, whereas in the former case a specific supramolecular event should influence the elec-
trochemical response of the system. Among the different supramolecular arrangements, molecular encapsulation is of particular
interest 20,24,26: when a guest is included inside a larger host, it becomes segregated from the environment and its interaction
with other species present in solutiondor with the solvent itselfdis hampered, if not completely prevented. Similar effects can
also be obtained by covalent attachment of large substituents like dendritic branches or polymeric structures. Electrochemistry is
thus very useful in the investigation of these systems, even if a word of caution has to be considered: often electrochemical tech-
niques exclude the use of solvents with low dielectric constant, and require the addition of a supporting electrolyte. Both these
factors can affect the extent of the interaction appreciably at study, either by direct interaction with the species or changing the prop-
erties of the environment.

2.16.2.2 Redox-Controlled Supramolecular Switching


When an electrochemical input is used to operate a supramolecular device, it is implied that oxidation or reduction of the electro-
active species affects the interaction between the two components, inducing a net change in the population of the states of the
device. It is helpful to visualize this process considering a host–guest system in which an electrochemically switchable guest shows
a very different thermodynamic affinity for its host in its reduced state, compared to the oxidized one. Thus, if the two molecules are
associated in the oxidized state, reduction of the guest induces the disassembly, or vice versa: a redox input induces a switching event
at the molecular level. This process is described by the square scheme in Fig. 1, which shows two electron-transfer processes coupled
with two homogeneous chemical equilibria (ECEC mechanism). Guest G and Host H associate according to the equilibrium
constant KH$G, and reduction of the guest G occurs at a specific formal potential (usually approximated with the corresponding
 
halfwave potential) depending on whether G is free ðEF 0 Þ or complexed (EC0 ), to afford G, that is involved in its binding equilib-
rium with H according to the equilibrium constant KH$G . All the electrochemically switchable systems can be described by an
analogue scheme, suitably adapted to the specific case: for this reason it is of interest to discuss some relevant features of the reaction
network reported in Fig. 1.
In order for the switching to be effective, that is, to obtain an appreciable change in the distribution of complexed and uncom-
plexed species, the equilibrium constants for the reduced and oxidized complex need to be sufficiently different and cannot be both
too high or too low. Interestingly, as the square scheme is a thermodynamic cycle, the difference in the binding affinities is directly
reflected in the difference between the formal potentials of the two redox couples (free and complexed guest, Eq. 1).
0 0
F ðE E Þ
KH,G
¼ e RT
F C
(1)
KH,G
In other words molecular recognition is directly reflected in a shift of the halfwave potential that corresponds to the difference in
the chemical binding constants of the two redox states. When coming down to laboratory practice, cyclic voltammetry is the most
commonly used technique to investigate electroactive systems. In a typical experiment, a difference of 15 mV in the redox potentials
of two reversible couples is a sufficient difference to be discriminated, allowing to detect a variation in the relative energies of about
1.4 kJ mol 1 and a ratio KH$G =KH$G equal to 0.56.
Despite the simplicity of the thermodynamics, the kinetic features of the system strongly affect the voltammetric response and
need to be considered. More specifically, the kinetics of the chemical reactions coupled with the electron-transfer determine the
chemical reversibility of the system. The electrochemical reversibility of the electron-transfer process is another element whose
knowledge is necessary to investigate the switching process, besides the sole characterization of the initial and final states. In partic-
ular, molecular encapsulation slows down the heterogeneous electron-transfer kinetics, and decreases the apparent diffusion coef-
ficient of the electroactive species. Although these requirements might appear as a complication to the analysis of the systems at
study, the strong influence of the kinetic on the voltammetric response can also be exploited to gain further insights on the dynamic
behavior of the system by changing the potential scan rate and rationalizing the observed changes. A detailed description of these
methodologies is beyond the scope of this article and can be found elsewhere. 11,15,18

Figure 1 Square scheme mechanism describing an electrochemically switchable host–guest system. Horizontal and vertical processes are the
chemical equilibria and the electron-transfer processes, respectively.
346 Electrochemically Controlled Supramolecular Switches and Machines

2.16.2.3 Electroactive Species on a Solid Support


The importance of surface science is increasingly being recognized, particularly in relation to material science and catalysis.
Chemical approaches to modulate the superficial properties include the formation of self-assembled monolayers (SAMs), Lang-
muir monolayers, and Langmuir–Blodgett multilayer films. 27,28 When an electroactive unit is involved in a recognition process
at the interface (i.e., either the host or guest is bound to the electrode surface), electrochemical techniques are suitable to char-
acterize the system and probe the self-assembly process29 or to switch the host–guest interaction by changing the potential
applied to the electrode.11,15,18,21,30,31 Beside the importance of fundamental studies, the incorporation of electroactive units
on surfaces and interfaces is important to interface nanoscale systems with the macroscopic world that is a fundamental step in
the transition from nanoscience to nanotechnology. Smart surfaces are particularly attractive for analytical purposes, and inter-
faces enable the ordering of molecular devices to make them work coherently and with spatial addressability. Explored possi-
bilities include the incorporation into polymers or onto surfaces, and immobilization into membranes and porous
materials.32–40

2.16.2.4 Electroactive Nonequilibrium Systems


In recent years, increasing attention has been received by supramolecular devices that operate away from thermodynamic equilib-
rium. 41 Among the systems that are not at thermal equilibrium, two distinct classes can be identified: kinetically trapped systems
and dissipative ones. In the first case the system is slowly evolving toward its equilibrium, but its current state has a sufficiently long
lifetime to be characterized and eventually react. On the contrary, dissipative systems exploit an external source of energy to stay
away from thermodynamic equilibrium and perform a function or maintain a particular structure. Although both kinds of systems
have a higher level of sophistication when compared to thermodynamic switches, dissipative systems are by far the most compli-
cated ones and, yet ubiquitous in nature, represent a largely unexplored field. Indeed, the high interest for these systems is justified
by the fact that biomolecular machines are exquisite examples of dissipative devices. Conceptually, in dissipative systems an
external source of energy is exploited to travel a cyclic reaction pattern with a directional bias, under constant environmental condi-
tions. When a system is suitably designed, an external energy source, like a chemical fuel or a light stimulus, can drive the circle
unidirectionally; in the case of electrochemical driving, a single redox couple at a fixed potential cannot drive a cycle of reactions
unidirectionally, as this would violate the second principle of thermodynamics. When an electrochemical driving force is applied,
the dissipative system has to be part of an electric circuit and exploit the electron flow in a useful manner. As it will be highlighted
with specific examples, this implies that the molecular entity has either to diffuse between the two electrodes, or to be in close prox-
imity with them.

2.16.3 Self-Assembling Systems

Host–guest complexes likely constitute the most common type of self-assembling systems. The larger and structurally more elab-
orated molecule is termed “host,” and the smaller and simpler component is termed “guest.” Pseudorotaxanes constitute a particular
class of host–guest systems, in which the host is a macrocycle and the guest has an elongated shape that protrudes through the mac-
rocycle. Redox reactions can enhance or decrease the binding strength between host and guest.
A convenient classification of host–guest systems is based on the nature of the main interaction that controls the self-assembly
and is affected by the electrochemical reaction, although it has to be reminded that recognition processes are usually the outcome of
the interplay between multiple driving forces.

2.16.3.1 Hydrogen and Halogen-Bonding Interactions


Redox modulation of hydrogen-bonding interactions is known since at least more than 20 years. 42 In this period of time, several
examples of redox-controlled hydrogen-bonded complexes have been developed, in particular flavins or arylamides in combination
with diaminopyridines or diaminotriazines,43 o-quinones in combination with arylureas,44,45 metallocenes combined with
dicarboxylic acids,46 or nitrobenzene combined with diarylurea47 proved to be reliable redox-switchable couples. This latter
nitrobenzene–diarylurea couple will be illustrated more in detail. Nitroaniline 1 (Fig. 2) is a moderate hydrogen-bond acceptor,
as the NeO bond is not very polar in character. On the contrary its radical anion localizes the excess charge on the oxygen atoms,
which thus become strong electron donors. Indeed the association constant of 1 with 1,3-diphenylurea 2 in dimethylformamide
(DMF) goes from 2.3  10 3 to 7.8  104 M 1. These values were obtained by simulation of the voltammetric pattern obtained
at different scan rates. Addition of 1,3-diphenylurea to a solution of 1 causes a shift of the reduction peak to less negative potentials,
in agreement with the rapid formation of a supramolecular complex between 1 and 2 (Fig. 2).
The introduction of an electron donating amino group in para-position has a modest effect on the hydrogen-bonding
ability of the oxidized form of the molecule, whereas makes the reduction process more difficult and induces the localization
of the excess negative charge on the oxygen atoms, thus increasing the difference in the hydrogen-bond acceptor abilities
between the oxidized and reduced forms of 1. Comparison of the affinity with several substrates revealed that to form a suffi-
ciently stable complex, 1 needs at least to form a hydrogen bond with one strong hydrogen-bond donor (e.g., an amine
Electrochemically Controlled Supramolecular Switches and Machines 347

Figure 2 (Left) Square scheme describing the redox-switching of an hydrogen-bonded array. Potentials are reported versus ferrocene. (Right)
Cyclic voltammetry of nitroaniline (1 mM) in DMF; the arrow indicates increasing concentrations of diphenylurea (0, 0.5, 1, 10 mM). Adapted from
Bu, J.; Lilienthal, N. D.; Woods, J. E.; et al. J. Am. Chem. Soc. 2005, 127, 6423, with permission. Copyright 2005 American Chemical Society.

directly attached to an electron-withdrawing group). When janus molecule 1,4-dinitrobenzene is used, 48 its dianion can form
1:2 complexes with 2, representing one of the rare examples where redox-controlled recognition involves more than two
species.
Systems where the redox-active moiety is not directly involved in the recognition process but is instead conjugated to the
hydrogen-bonding donor/acceptor have also been investigated. In these systems, the effect of the redox reaction on the binding
equilibrium is less pronounced; as an example, 49 hydrogen bonding forming 3 includes a tetrathiafulvalene (TTF) moiety that
undergoes two successive reductions (Fig. 3). Since the TTF unit is conjugated with the imide hydrogen bonding forming pattern
(acceptor–donor–acceptor), the electron density in the imide moiety is decreased, and 3 becomes a less efficient host for 4. Indeed
a þ 30 mV E1/2 shift for the TTF0/þ couple upon guest addition in NBu4PF6/CH2Cl2 reveals a threefold decrease in the binding
constant upon reduction.
Halogen bond has emerged in recent years as a newly recognized supramolecular interaction, 50 occurring in the solid state as
well as in solution. Its name comes from the analogy with hydrogen bond that has similar directional constraints and in some
selected cases even a similar strength. Electrochemical modulation of halogen-bonding has been recently achieved.44 Addition
of increasing amounts of iodo-perfluorohexane to tetrachloro-p-quinone (TCQ) induces a shift to less negative potentials in the
second reduction of TCQ, indicating that a favorable recognition process occurs between iodo-perfluorohexane and TCQ2  (but
not with TCQ). Analogue experiments with other halo-perfluorocarbons showeddas expectedda tendency in the electrochemical
shift in accord with the halogen bond donor strength (I > Br > Cl): the stronger the halogen bond donor, the larger the potential
shift. The same concept was applied for the sensing of chloride and bromide anions in acetonitrile and water/acetonitrile
mixtures.51

2.16.3.2 Metal Ion Coordination


Along with organic moieties, metal ions can also be used as electroactive units in supramolecular assemblies. Their switching mech-
anism is typically based on different preferred coordination geometries or environments (hard or soft ligands) in their accessible

Figure 3 Hydrogen-bond recognition pattern for the recognition process between 3 and 4.
348 Electrochemically Controlled Supramolecular Switches and Machines

redox states. Therefore in an effective redox-switching process involving a metal center (guest), the metal changes its coordination
sphere, and this event can be correlated with the translocation of the ion itself from one site to another, a conformational change
involving the ligands (host), or both. Only a limited number of metal ions can undergo a reversible reduction/oxidation between
states that require different coordination preferences; for this reason most of the examples reported to date rely on the Cu(II)/Cu(I)
and Fe(III)/Fe(II) redox couples.
Cu (II) has a d9 electronic configuration, and can adapt to coordination numbers of 4, 5, or 6 in order to maximize the ligand
field stabilization energy (LFSE); on top of it, Jahn–Teller effects may contribute to distort the coordination geometry. On the other
hand, Cu(I) is a d10 metal ion, and prefers a tetrahedral geometry in order to minimize the steric repulsion between the ligands.
Fe(III) is a high-spin d5 metal center, and therefore it does not require a specific coordination geometry, because the established
interactions are mainly electrostatic in character. On the contrary, Fe(II) is a low-spin d6 cation that takes advantage at most of
LFSE when it is surrounded by an octahedral coordination sphere.
From a supramolecular point of view, an interesting consequence of a metal-based redox reaction is the translocation to
a different coordination site within the same host. An example of this behavior is represented in Fig. 4.52
In this system, a flexible ditopic host can accommodate copper in two tetracoordinating compartments: a tetramine and a bis-
bipyridine. In acetonitrile solution, Cu(II) is more stable in the tetramine environment, within a square planar coordination. Upon
reduction to Cu(I) with ascorbic acid, relocation of the metal to the softer bipyridine compartment occurs, inducing a conforma-
tional rearrangement of the bipyridines to adopt a tetrahedral coordination geometry. Addition of a chemical oxidant like H2O2
restores the initial condition, as demonstrated by absorption spectroscopy measurements.
In a more elaborated example, the redox chemistry of copper was exploited by Schmittel and coworkers to put in communica-
tion two different switches. 53 The system is composed by two structurally similar switches, each one consisting of three arms
(Fig. 5).
One arm bears a Zn-porphyrin moiety and the second arm incorporates a phenanthroline unit; the last branch ends with an
azabipyridine or azaterpyridine unit, depending on the derivative. This latter arm plays a key role, as it can coordinate to the
Zn-porphyrin, ordin the presence of a suitable metal cationdcan chelate it to form a coordination complex in a synergistic
action with the phenanthroline unit of the second arm. Depending on its coordination mode, the central arm can switch
between two states: the azabipyridine unit forms a tetracoordinated complex with phenanthroline, whereas azaterpyridine
forms a pentacoordinated one. Therefore Cu(I) has a higher affinity for the former complex, whereas Cu(II) prefers the latter
coordination environment; accordingly, when the two switches are present in the same solution, Cu(I) is coordinated by the
azabipyridine. Upon chemical oxidation with tris(4-bromophenyl)ammoniumyl hexachloroantimonate Cu(II) dissociates
from its ligands and forms a pentacoordinated complex on a different switch: in this process the two switches are toggled
simultaneously, as the central arm rotates in both switches, in order to change its coordination mode. Addition of a chemical
reductant (decamethylferrocene) restores the initial situation, providing an example of reversible communication between two
molecular switches.
Building on this system, catalytic reactions were alternatively switched 54 or three switches could be put in communication.55
Copper ions have been extensively used in the development of helicates assembly and disassembly; the interested reader can refer
to a number of reviews on the topic.56
The Fe(III)/Fe(II) couple has also been widely exploited to achieve the redox-induced metal translocation.
In a typical example ligand 53  ( Fig. 6) is used as a host.57 53  bears a phenolic unit that has been functionalized with different
polydentate units at its ortho positions. On one side a tertiary amine nitrogen and two phenolate oxygens constitute a binding envi-
ronment with hard characteristics; on the other side, along with a tertiary amine, two pyridine nitrogens form a softer coordination
pocket.
Given these characteristics, when collidine is present in solution (thus guaranteeing the deprotonation of the phenolic units),
Fe(III), added as perchlorate salt, is taken by the harder compartment, giving rise to a six-coordinated complex. On the other
hand, when Fe(ClO4)2 is added, the bivalent metal prefers the softer coordination environment provided by the pyridines. Trans-
location could also be induced electrochemically. More specifically, cyclic voltammetry was used to investigate the system in NBu4-
ClO4/CH3CN solution. The voltammetric pattern reported in Fig. 7 shows cathodic and anodic peaks that are both chemically

Figure 4 Copper ion translocation following the reduction of the metal center. Reproduced from Amendola, V.; Fabbrizzi, L.; Mangano, C.;
Pallavicini, P. Acc. Chem. Res. 2001, 34, 488, with permission. Copyright 2001 American Chemical Society.
Electrochemically Controlled Supramolecular Switches and Machines 349

Figure 5 Coupled motion of two different mechanical switches mediate by copper ion redox changes. In both switches, the azapyridine-containing
arm can either coordinate zinc or copper. In the presence of copper ion, one switch coordinates copper selectively, depending on the oxidation state
of copper, therefore, in the switch that is not binding copper the azapyridine arm coordinates to the zinc porphyrine. Adapted from Pramanik, S.; De,
S.; Schmittel, M. Angew. Chem. Int. Ed. Engl. 2014, 53, 4709, with permission.

Figure 6 Translocation motion of an iron ion within a multitopic ligand with two separate binding pockets. The left pocket has a harder character in
comparison with the right one. Fe(III) ions are preferably coordinated by the harder compartment, whereas the softer Fe(II) ions prefer to be located
in the softer pocket. S indicates solvent (CH3CN) molecules that complete the coordination sphere of the metal ions.

irreversible. This observation is due to the fact that electrochemical oxidation is quickly followed by metal translocation. As the
potential is scanned back to  0.8 V, reduction of the metal center takes place and the initial situation is quickly restored. Notably,
even as the scan rate increases, it was not possible to observe a reversible peak at a potential corresponding to the oxidation of Fe(II)
to Fe(III): this behavior indicates that the metal translocation induced by oxidation occurs on a timescale faster than 10 ms. A tenta-
tive explanation of the fast-switching kinetics is based on the role of the central phenolic oxygen that is coordinated to the metal in
both its coordination sites and might therefore remain coordinated to the iron also in the transition state, thus lowering the switch-
ing barrier.
In selected cases the metal oxidation/reduction is not accompanied by a metal translocation:anion translocation occurring to
satisfy the coordination requirements of the redox-active metal has also been reported, 58 both intramolecularly and
intermolecularly.59
350 Electrochemically Controlled Supramolecular Switches and Machines

Figure 7 Cyclic voltammogram of a CH3CN/NBu4ClO4 solution of 5 (4  10 3 M), containing one equivalent of Fe(ClO4)2 and collidine. Working
electrode: platinum disk (5 mm diameter); scan rate: 0.1 V s 1; reference electrode: AgNO3/Ag in CH3CN. Reproduced from In Analytical Methods in
Supramolecular Chemistry; Schalley, C. A., Ed.; Wiley-VCH: Weinheim, 2012, with permission.

2.16.3.3 Hydrophobic Interactions


Supramolecular chemistry in water largely involves the engineering of hydrophobic interactions. For this reason, relatively small
and rigid cavities in which water cannot optimize its interaction with the neighboring water molecules are suitable hosts in aqueous
environment. Cyclodextrins (CDs) 60 and cucurbit[n]urils (CB[n])61 have such characteristics, and have been widely studied. CDs
can be considered as truncated cones, with hydroxyl groups on its portals and several chiral centers, as CDs are macrocycles consti-
tuted of sugar units. On the other hand, CB[n]s are barrel-shaped symmetric macrocycles without chiral centers and with
a pronounced negative charge on their portals, as a consequence of the carbonyl groups that surround the portals.
CDs and CB[n]s include a number of macrocycles with different dimensions. Commonly used CDs have 6, 7, or 8 sugars, cor-
responding to a, b, and g CDs, whereas CB[7] and CB[8] are the most used CB[n]s; this variety allows the inclusion of guests with
different sizes, as well as the inclusion of multiple guests simultaneously. The formation of inclusion complexes is usually driven by
hydrophobic interactions, but other interactions have a synergistic effect. Indeed electrochemical switching would not be possible if
only hydrophobic interactions were taken into account. CDs typically form host–guest adducts with association constant that range
from 103 to 105 M 1; CB[n]s can form much more stable inclusion complexes. In particular when the guests bear strategically
located positive charges, the inclusion of the guest is accompanied by the establishment of strong ion–dipole interactions: as a result
the association constant for these types of complexes can reach values as high as 1015 M 1 that is comparable with the highest
known association constant for naturally occurring host–guest complexes (i.e., the avidin–biotin pair 62).
The electrochemical studies performed on the host–guest pair formed by b-CD and ferrocene carboxylate anions (FcCA) are
presented here: in phosphate buffer (pH 9.2), addition of an excess of b-CD to FcCA induces a shift to more positive potentials
for the oxidation of FcCA, with a concomitant decrease in the recorded current. Moreover, on increasing the scan rate, a decrease in
the current value is also observed, contrary to what is usually observed for a reversible redox couple. These findings have been ratio-
nalized according to the scheme reported in Fig. 8.
FcCA and b-CD form an inclusion complex with a binding constant of 2200  100 M 1 at 20 C: this fact explains the shift
toward more positive potentials of the FcCA oxidation. On the contrary the oxidized zwitterionic form FcþCA has little or no

Figure 8 Chemical–electrochemical (CE) mechanism for the one-electron oxidation of FcCA in water in the presence of b-CD (schematically repre-
sented with the blue truncated cone).
Electrochemically Controlled Supramolecular Switches and Machines 351

affinity for the b-CD cavity. The decrease in the current intensity is explained by the decreased diffusion coefficient of the adduct in
comparison with the free guest, implying a slower diffusion-controlled supply of complex to the electrode surface, where electron-
transfer takes place. Finally, the decrease in the current intensities at faster scan rates is explained in light of a chemical–electrochem-
ical (CE) mechanism, that is, only FcCA is reduced under the present conditions, and [b-CD$FcCA] simply behaves as a source of
FcCA, that is formed upon dissociation of the complex. For this reason, at fast scan rates the dissociation reaction cannot keep up
with the fast oxidation of the free guest, and an overall decrease in the current is observed.
Also N,N0 -dimethyl-4,40 -bipyridinium (methyl viologen or paraquat, MV2 þ 63) is able to form inclusion complexes with b-CD.
In this case MV2 þ is not complexed by the host in its doubly charged form. On the contrary the reduced forms show an increased
affinity for the host cavity. This behavior is rationalized in terms of the reduced solubility of the guest, corresponding to a less effec-
tive solvation as the charge is lost. Analogously with the previously discussed case of FcCA included in the b-CD cavity, oxidation
of the complex is not observed, and complex dissociation has to occur before oxidation of the MV guest can take place.
In its doubly charged form MV2 þ forms inclusion complexes in aqueous media with CB[7], 64 with an association constant of
10 M 1. Kaifer and coworkers highlighted that the value of the association constant decreases with increasing ionic strength of the
5

medium: this effect is common to the complexation of many charged host–guest systems in nonaqueous solvents and is rational-
ized in terms of ion pairing effects. Reduction of MV2 þ only moderately decreases its affinity for the cavity, as evidenced by a negative
shift of z30 mV in the MV2 þ/MVþ potential in the presence of CB[7]. Concomitantly, a decrease in the current levels in compar-
ison with the free guest reflects the larger size of the adduct and thus its decreased diffusivity (Fig. 9).
Differently from what has been observed in the case of CDs complexes, voltammetric data reveal that reduction of the guest can
occur also inside the CB[7] cavity, and does not need to be preceded by a dissociation reaction. Indeed at increasing scan ratesdup
to 1.0 V s 1dthe peak potentials are not shifted and the current intensity increases at increasing scan rate.

2.16.3.4 Charge–Transfer Interactions


Charge–transfer (CT) interactions, also called donor–acceptor, are widely used in supramolecular chemistry. Binding constants for
a CT pair depend not only upon the ability of the components to donate and receive electron density, but also on other parameters,
like, for example, steric effects imparted by the substituents, complementary charge distribution, and shape matching. The sole
charge–transfer interaction is usually quite weak, with association constants ranging typically from 100 to 102 M 1 for structurally
simple D–A pairs. 65 However in combination with other favorable interactions like, for example, hydrogen-bonding or -hydro-
phobic interactions, much higher association constants are obtained; therefore CT complexes have been extensively used to engineer
the formation of supramolecular species with elaborated structures and functionalities.66 Molecular encapsulation is also beneficial
to increment the value of the association constant between a donor and an acceptor: the host can form a segregated environment for
the formation of a CT complex,67 or be involved directly in the CT interaction; solvophobic interactions usually act in a synergic
manner, favoring the donor–acceptor interaction. A large number of pseudorotaxane-type complexes exploit CT interactions. As

Figure 9 Cyclic voltammetric response at different scan rates (arrows indicate the increasing scan rates) on a glassy carbon electrode (0.072 cm2)
of 1.0 mM MV2 þ in 0.1 M NaCl (a) in the absence and (b) in the presence of 1 equiv. CB[7]. Scan rates: 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.8, and
1.0 V s 1. Reproduced from In Analytical Methods in Supramolecular Chemistry; Schalley, C. A., Ed.; Wiley-VCH: Weinheim, 2012, with permission.
352 Electrochemically Controlled Supramolecular Switches and Machines

Figure 10 Tetracationic cyclophane 64 þ constituted by two viologen units, connected by phenyl rings.

the electron density is directly affected by redox reactions, electrochemistry is a very convenient tool to control the assembly/disas-
sembly of donor–acceptor complexes.
In this context, the most studied cyclophane receptor is likely 64 þ ( Fig. 10), which is constituted by two electron acceptor viol-
ogen units connected in order to give a rigid box, that proved to be a suitable host for a wide variety of electron-rich guests.68
From the electrochemical perspective, 64 þ undergoes two bielectronic reductions. In the first process each viologen unit is con-
verted to the corresponding monocationic radical: when an electron-rich molecule is included in the cavity, the halfwave potential
for this process is shifted to more negative values. The second process corresponds to the conversion of the viologen units to their
neutral counterparts. 69 Typically upon the first reduction process, the CT interaction is lost and the electron donor unit dethreads
from the cyclophane cavity: when this happens the second reduction process is not affected by the presence of the guest.
For example when TTF is added to an acetonitrile solution of 64 þ, the reduction waves of the cyclophane change from  0.28
and  0.72 V versus SCE (saturated calomel electrode) to  0.32 and  0.72 V versus SCE. 70 As anticipated, the first reduction wave
is shifted to more negative potentials, that is, it is more difficult to reduce 64 þ when it is engaged in interactions with the electron
donor TTF. On the contrary, the second reduction process occurs at the same potential value as for the free host, because TTF deth-
readed from the cyclophane as a consequence of the host reduction. TTF is also electroactive; indeed it can undergo two successive
reversible oxidation processes. Oxidation of TTF can also be used to disrupt the CT interaction responsible for complex formation.
When TTF is included in 64 þ the first oxidation potential shifts from þ 0.32 to þ 0.39 V versus SCE (Fig. 11). On the contrary the
second oxidation process is not affected by the presence of 64 þ, and occurs at the same potential of the free TTF. Therefore the deth-
reading process can be controlled in a dual mode (oxidative and reductive). The same behavior is also observed in structurally
related pseudorotaxanes.70a,71
Notably, the double oxidation of TTF affords TTF2 þ, a doubly charged cation with moderate electron acceptor properties. 72 This
characteristic was exploited to obtain an electrochemically controlled three-pole switch.43d In acetonitrile solution, when both the
electron poor cyclophane 64 þ and the electron-rich crown-ether 7 (Fig. 12) are present,73 TTF in its neutral form is encircled by 64 þ.
Upon the first oxidation process, TTF dethreads and is mainly present in solution as a free guest, instead upon its second oxidation it
prefers to be encircled by 7.

Figure 11 Cyclic voltammetric patter of 0.5 mM TTF alone (a) and in the presence of equimolar 64 þ (b). Conditions: glassy carbon electrode
(0.08 cm2), acetonitrile solution, tetraethylammonium hexafluorophosphate 0.05 M, scan rate 50 mV s 1. Current intensities are corrected to take
into account differences in the diffusion coefficients. Adapted from Asakawa, M.; Ashton, P. R.; Balzani, V.; et al. Chem. Eur. J. 1997, 3, 1992, with
permission.
Electrochemically Controlled Supramolecular Switches and Machines 353

Figure 12 A three-pole supramolecular switch based on a host–guest system architecture is obtained exploiting the different affinities of TTF for
64 þ and 7, depending on its redox state.

2.16.3.5 Cation Radicals Dimerization


Radical cations can exhibit the tendency to dimerize, either with another cation radical, or with its neutral counterpart, to form
respectively a radical or mixed valence dimer. In the case of viologen-based compounds, the dimerization of their radical cations
is reversible, and it is known to occur in water at high concentration (about 1 M) since at least more than 50 years. 74 In recent years
this process has been exploited as a recognition motif in the construction of supramolecular devices,75 and host–guest architecture,
including supramolecular polymers.76 In particular, Stoddart and coworkers demonstrated the association of a functionalized viol-
ogen (82 þ, Fig. 13) with cyclophane 64 þ in acetonitrile solution at a concentration of 1 mM in both components.77 When 64 þ and
82 þ are mixed in solution, a single three-electron reduction process is observed, at a potential that is less negative in comparison
with the isolated components, suggesting the formation of an inclusion complex. The fact that three electrons are involved at the
same potential was tentatively explained with the formation of the doubly reduced 62 þ that rapidly transfers an electron to 82 þ to
form the radical cation dimer; subsequently a third electron is injected in viologen unit of 64 þ that becomes doubly charged after
the electron-transfer to 82 þ. Differently from the isolated components, the second reduction process of 64 þ appears to be split in
two one-electron processes, suggesting that the radical cation dimer involves a single viologen moiety of 62 þ. Moreover, these one-
electron processes occur at a potential that is more negative than in the separated components. Control experiments with dumbbell-
shaped compounds excluded the dimer formation outside the cavity, that is, formation of the radical cation dimer is allowed only
within the threaded structure. Spectroelectrochemistry clearly supports the formation of the inclusion complex; indeed a broad
band in the near infrared region appears in the presence of the radical cation dimer.
Cooperative effects can also be effective in lowering the required concentration to achieve molecular recognition based on
radical cation dimerization. Indeed, monomers with multiple branches were shown to polymerize upon reduction. 76 Also TTF
radical cation can form stable dimers with another TTFþ radical, or with the neutral TTF. In analogy with viologen, electrochemical
responsive polymers based on TTF were also developed.78 As will be highlighted in section “Mechanically Interlocked Molecules,”
this recognition motif demonstrated to be useful in the design of switchable interlocked molecules.

2.16.3.6 Systems Working on Surfaces


The recognition motifs described in the previous sections can be used to fabricate surfaces with electrochemically controllable
binding properties. In principle it can be achieved by functionalizing either the host or the guest with a proper linker (that should
not affect the self-assembling abilities of the functionalized moiety) that bears a functional group capable of binding to the surface,
for example, thiols are particularly suitable to bind to gold surfaces. For example, the previously described three-pole switch (see
“ Charge–Transfer Interactions” section) was adapted to operate on a surface. A TTF derivative bearing an ending disulfide group
(Fig. 14) was self-assembled on a gold electrode to give an electroactive monolayer.79 In analogy with what was previously observed
in solution, 9 in its neutral form is able to form a supramolecular complex with cyclophane 64 þ, whereas upon oxidation to its
TTF2 þ analogue, crown-ether 7 becomes the preferred guest. Therefore, the surface properties of the electrode can be electrochem-
ically controlled.
Interestingly, in a conceptually related system, 80 composed by a surface-bound TTF unit and 64 þ freely moving in solution, it
was observed that the oxidation of the TTF units occurs at a more positive potential with increasing surface coverage, indicating that
the surface features directly affect the recognition interaction. In a further evolution of this system,81 64 þ was incorporated in
354 Electrochemically Controlled Supramolecular Switches and Machines

Figure 13 Cyclic voltammetry of 1 mM 82 þ (a), 1 mM 64 þ (b), and an equimolar solution (1 mM) of both components (c). Conditions: glassy
carbon electrode (0.018 cm2), acetonitrile solution, tetraethylammonium hexafluorophosphate 0.1 M, scan rate 200 mV s 1. Adapted from Trabolsi,
A.; Khashab, N.; Fahrenbach, A. C.; et al. Nat. Chem. 2010, 2, 42.

Figure 14 Functionalized TTF derivative, bearing an ending disulfide group, suitable for the functionalization of gold electrodes.

a polymer: in the presence of nanoparticles functionalized with TTF of dioxynaphthalene (DON), the polymer undergoes aggrega-
tion with one type of nanoparticle or the other one, depending on the oxidation state of TTF. Indeed when TTF is in its neutral form,
the TTF-functionalized nanoparticles are preferably trapped by the 64 þ functionalized polymer; instead, when TTF is oxidized the
nanoparticles bearing DON moieties on their surface are selectively trapped.
Switchable surfaces can also be obtained upon functionalization of the solid support with a nonelectroactive host (guest) and
addition of an electroactive guest (host) in solution. An example of this strategy is provided by the assembly of ferrocenyl-decorated
dendrimers, interacting with the electrode surface where b-cyclodextrin derivatives (b-CD) are adsorbed ( Fig. 15).82 Since ferrocene
strongly interacts with b-CD, a stable monolayer at the b-CD functionalized electrode is obtained, due to multivalent interactions.
Instead, when ferrocene is oxidized to ferrocinium, the dendrimers desorb from the b-CD SAM. This process was studied in detail by
voltammetric techniques and electrochemical impedance spectroscopy, gaining information on the adsorption and desorption
kinetics and mechanism, as well as on diffusion coefficients and conformation of the ferrocene dendrimer adsorbed at the electrode
surface.82

2.16.4 Mechanically Interlocked Molecules

A new class of systems that became synthetically accessible in the last decades is constituted by mechanically interlocked molecules.
These systems are made of multiple components that are not covalently bound, but cannot be set apart without breaking a covalent
bond. Two main categories can be identified: rotaxanes and catenanes.
Rotaxanes 83 are minimally composed of a macrocycle that encircles a dumbbell-shaped molecule (“axle”), bearing two bulky
ends that prevent the slippage of the macrocycle away from the axle (Fig. 16A). Catenanes are minimally composed of two inter-
locked macrocycles (Fig. 16B), therefore they are topologically different from rotaxanes. Important features of rotaxanes and cat-
enanes originate from the presence of complementary recognition sites on the two components: the properties of these systems
are particularly interesting when multiple recognition sites (stations) are present on one of the two components, and the affinity
of the complementary recognition site on the other component can be switched between the two stations. In this situation,
large-amplitude controlled motion of one component with respect to the other can be achieved (Fig. 16).
Electrochemically Controlled Supramolecular Switches and Machines 355

Figure 15 (A) Molecular structures of the ferrocene-terminated dendrimer, b-cyclodextrin, and the b-cyclodextrin derivative used to functionalize
the electrode. (B) Schematic representation of the self-assembly of the dendrimer-b-CD complex, its adsorption on the b-CD modified surface, and
the desorption upon electrochemical oxidation of ferrocene. Reproduced from In Organic Electrochemistry, 5th ed.; Hammerich, O.; Speiser, B., Eds.;
CRC Press: Boca Raton, FL, 2016, with permission.

Figure 16 Relative movements associated to bistable rotaxanes (A) and catenanes (B). S1 and S 1 represent opposite stimuli that are required to
induce the motion. Pirouetting of the ring component in rotaxanes is also possible.

The control of motion at the nanoscale is a difficult task, and represents a challenge in nanoscience and nanotechnology: inter-
locked molecules provided a fruitful platform to approach it. This challenge is of particular interest, because the control of motion at
the nanoscale constitutes a premise for the construction of artificial molecular machines and motors, which are molecular entities
capable of performing a predetermined task at the molecular level: nature exploits the action of such systems and it is expected that
these artificial devices will find an application in material science, energy conversion, information technology, and medicine. 84
356 Electrochemically Controlled Supramolecular Switches and Machines

A thorough discussion on the basic concepts of molecular machines can be found elsewhere.8,85 Here only two basic concepts will
be highlighted:
(1) molecular machines require energy to operate: chemical, photonic, or electrochemical inputs are suitable energy input. In this
article the discussion will be limited to electrochemically driven machines;
(2) for a machine to perform work on its environment in a repetitive manner, thermodynamic control is not sufficient. Indeed, if
any work is performed in a switching process, it will be canceled in the back-switching process. Hence, kinetic control is needed
to perform the back-switching process without canceling the work that was initially performed. Electrochemically powered
systems that show kinetic control will be discussed in section “ Nonequilibrium Systems.”

2.16.4.1 Systems Based on Rotaxanes


Rotaxanes bearing two different recognition sites in their axle component can behave as molecular shuttles, 86 when the affinity of
the ring for the two stations can be controlled, inducing the shuttling of the ring from on station to the other.
Three examples of electrochemically driven shuttles are discussed that differ in the nature of the interactions stabilizing the
structure.
Rotaxane 10 87 (Fig. 17) consists of a benzylic amide macrocycledhaving hydrogen-bond donor characteristicsdthreaded onto
a molecular axle with two potential sites both having hydrogen-bonding acceptor characteristics: a succinamide and a naphthali-
mide site.
In the more stable co-conformation the ring is engaged in hydrogen-bonding interactions with the succinamide site,
however in DMF, upon reduction the naphthalimide becomes a stronger hydrogen-bond acceptor and the ring shuttles to
the naphthalimide station. The occurrence of this process was verified with a combination of techniques, comparing the
properties of the free axle with those of the rotaxane. Spectroelectrochemistry supported the switching process, indeed
the absorption maxima of the naphthalimide radical anion is shifted from 422 to 415 nm, as it is known to occur when
the radical anion is engaged in hydrogen-bonding interactions. Cyclic voltammetry evidenced that in the rotaxane the reox-
idation of the radical anion to the neutral form occurs at a potential that is 0.48 V more positive than in the axle. This value
corresponds to a stabilization interaction of 11.2 kcal mol 1, corresponding to four strong (or three extremely strong)
hydrogen bonds that lead to a quantitative (K > 1500) translocation of the ring onto the naphthalimide station. Molecular
modeling excluded that these effects could be the result of a different conformation of the rotaxane, not involving the
displacement of the ring. Reduction of the naphthalimide unit can also be achieved photochemically, in the presence of
an electron donor.
The second system presented is the rotaxane [11l Cu]þ ( Fig. 18) that is capable of inducing a switching process as a consequence
of the different coordination environments preferred by Cu(I) and Cu(II) (see “Metal Ion Coordination” section). Sauvage and
coworkers extensively developed the chemistry of copper-based interlocked systems88–90: the present example91 is the latest inves-
tigation performed by the group on this class of compounds.
The ring component includes a 8,80 -diphenyl-3,30 -biisoquinoline (BIQ) unit that acts as a bidentate chelating ligand for copper.
The axle includes a phenanthroline and a terpyridine units: when copper is present as Cu(I), a tetrahedral coordination geometry is
preferred, therefore the metal ion is coordinated by BIQ and phenanthroline units. Upon oxidation to Cu(II), the macrocycle encir-
cles the terpyridine unit. Operation and monitor of the system are performed using cyclic voltammetry. From the previous studies
performed in the group it is known that the Cu(II)/Cu(I) couple originates a reversible wave at a potential that is strongly affected by
the copper coordination geometry. In a tetrahedral environment, the reaction occurs at potentials above þ 0.4 V versus SCE, instead,
when Cu is pentacoordinated it occurs at z0.0 V versus SCE. In the cyclic voltammogram at low scan rates, an irreversible cathodic

Figure 17 Molecular structure of [2]rotaxane shuttle 10. The shuttling is electrochemically induced taking advantage of the enhanced hydrogen-
bond acceptor abilities of naphthalimide radical anion.
Electrochemically Controlled Supramolecular Switches and Machines 357

Figure 18 [2]rotaxane [11lCu]þ, thanks to the nonhindering nature of the 8,80 -diphenyl-3,30 -biisoquinoline ligand switching between the two
stations occurs on a timescale that is four orders of magnitude fasted than structurally related rotaxanes. Reproduced from Durola, F.; Sauvage, J.-P.
Angew. Chem. Int. Ed. 2007, 46, 3537.

peak is observed at zþ 0.4 V versus SCE, and an anodic wave is present at z 0.0 V versus SCE. This observations are consistent with
shuttling being fast in comparison with the scan rate: Cu(I) is oxidized at zþ 0.4 V versus SCE and shuttling quickly follows, there-
fore the reduction reaction involves a pentacoordinated Cu(II), thus occurring at z0.0 V versus SCE. Increasing the scan rate the
anodic peak decreases its intensity, and the reversible wave of the cathodic peak appears, in accordance with the fact that the time-
scale of the shuttling and that of the potential sweep become comparable. In contrast with the previously reported systems, shuttling
occurs on a timescale that is at least four orders of magnitude faster than that of a similar rotaxane where the ring component bears
a phenanthroline unit in place of the BIQ unit. This remarkable difference is rationalized in terms of the nonhindering nature of
BIQ, coupled with its ability to rotate around the CeC bond connecting the two quinoline units, that allows a stepwise decomplex-
ation. The same strategy has been applied to control the pirouetting motion of the ring component around the axle, simply locating
the two recognition sites on the ring rather than on the axle. 92
The last example is based on the viologen radical cation dimerization that was introduced in section “ Cation Radicals Dimer-
ization”: Stoddart and coworkers exploited this recognition motif also to build interlocked structures.75c,77,93 Axles bearing a viol-
ogen unit and an electron-rich DON or TTF unit proved to be suitable for building [2]rotaxanes that can be switched upon
reduction. Since 64 þ is electron poor, it initially resides on the electron-rich station; however, under reductive conditions the macro-
cycle shuttles on the viologen unit in order to form the radical dimer. In the [2]rotaxane including a TTF unit, the second oxidation
wave of TTF is also shifted to more positive values, because 64 þ remains in the proximity of TTF2 þ, as the macrocycle does not have
any other energetically more favorable position available. Indeed this is in sharp contrast with the corresponding pseudorotaxane,
where the first oxidation of TTF induces its dethreading. With these premises a tristable [2]rotaxaneda three-pole switchdwas
synthesized.
The rotaxane ( Fig. 19) includes three different stations in the axle: DON, TTF, and viologen. As for the previous cases under
neutral conditions TTF is encircled by 64 þ, and reductive conditions induce the shuttling of the macrocycle on the reduced viologen,
as radical dimerization occurs. Under oxidative conditions DON becomes a better electron-donor with respect to the TTF-based
station, and the macrocycle locates on the DON station.

Figure 19 The [2]rotaxane developer by Stoddart and coworkers that behaves as a three-pole switch. Charge transfer and radical dimerization inter-
actions are exploited in the operation of the system.
358 Electrochemically Controlled Supramolecular Switches and Machines

2.16.4.2 Systems Based on Catenanes


In analogy with rotaxanes, catenanes can be switched between multiple co-conformations if one of the rings has at least two recog-
nition sites.
Since the examples presented in the rotaxane section were based on hydrogen bond, metal coordination, and radical cations
dimerization, here examples based on the electrochemical modulation of CT interactions will be presented.
Catenanes 122 þ–172 þ ( Fig. 20)94 are composed of (a) an electron acceptor ring containing two different recognition units;
a viologen (V2 þ) moiety and a naphthalene or pyrometallic diimide (NDI or PMI, respectively) with electron-acceptor character-
istics. The other ring that completes the catenane structure has instead electron-rich characteristics, indeed it can include two
DON or dioxybenzene (DOB) units, thus being symmetric, or can include a TTF unit along with another electron donor like
DON or DOB. In the starting state of all these systems, the better electron acceptor, namely V2 þ, is engaged in charge–transfer inter-
actions. When different electron donors are included in the structure, TTF is positioned inside the electron poor ring, as it is the
better electron donor. This starting conformation can be electrochemically switched, and cyclic voltammetry is a convenient tech-
nique to monitor and operate these systems.
Both V2 þ and imides can undergo two successive reductions, therefore the cyclic voltammogram of 122 þ shows four reversible
monoelectronic reductions. The comparison with model compounds, in combination with spectroelectrochemical experi-
ments, 92b,95 revealed that (i) the first reduction process involves the V2 þ unit engaged in CT interactions inside the cavity of the
donor ring, (ii) also the second reduction process is viologen-based, and shows that such a unit is still engaged in CT interactions,
despite the previous reduction. Most likely this second reduction induces the translocation of PMI inside the electron-donor ring,
then (iii) the third and fourth reductions involve the PMI units: in the former PMI is inside the electron-donor ring, whereas little
information is known about the position of PMI in the latter process.
When crown-ethers with stronger electron-donating abilities (e.g., 132 þ, 142 þ) are used, V2 þ is more stabilized, thus it is more
difficult to reduce; therefore its second reduction overlaps with the first process involving PMI.
Unfortunately it is difficult to determine unambiguously the position of the electron-rich macrocycle after the second and third
reductions. 96 The conformation changes coupled with the four-electron reduction processes for the PMI-containing catenanes
122 þ142 þ are summarized in Fig. 21.
When NDI is included in the electron-poor macrocycle in place of PMI, the four reduction processes are alternatively V2 þ or NDI
based. Indeed the extended aromatic unit of NDI in comparison with PMI renders its reduction more favored. Despite this differ-
ence, the co-conformational changes associated with the stepwise four-electron reduction of the catenane are not markedly different
from what was observed before. Upon the first reduction process that affords Vþ, the electron-rich ring remains around the viologen
unit, then the first reduction of NDI follows, but the ring position is not affected. Upon the third and fourth reductions the position
of the ring is difficult to disclose. 96
The fully desymmetrized catenanes 142 þ and 172 þ, which include a TTF unit, can be studied fruitfully also under oxidative
conditions. Three oxidation processes are observed, the first two being the successive oxidations of TTF, 73 and the third one is
assigned to the DON unit oxidation.

Figure 20 Schematic structure and component units of the desymmetrized catenanes 122 þ–172 þ. Adapted from Cao, D.; Amelia, M.; Klivansky, L.
M.; et al. J. Am. Chem. Soc. 2010, 132, 1110.
Electrochemically Controlled Supramolecular Switches and Machines 359

Figure 21 Schematic representation of the conformational changes associated with the reduction processes for catenanes 122 þ–142 þ. Vertical and
horizontal processes represent the PMI-centered and viologen-centered reductions, respectively. In the upper right part of the scheme, red arrows
refer to the behavior of 122 þ, while green arrows describe the behavior of 132 þ and 142 þ. The dotted ellipses indicate that there is little information
about the donor ring location. Adapted from Cao, D.; Amelia, M.; Klivansky, L. M.; et al. J. Am. Chem. Soc. 2010, 132, 1110, with permission. Copy-
right 2010 American Chemical Society.

Since TTF is a better electron donor with respect to DON, it is initially located inside the cavity of the electron-poor ring, and is
engaged in strong charge-transfer interactions; this information can be extracted from the value of the potential at which the first
oxidation of the catenane occurs that is shifted to more positive values in comparison with that of the free electron-rich macrocycle.
Upon TTF oxidation, DON becomes a better electron donor, therefore circumrotation occurs, and DON becomes engaged in
donor–acceptor interactions ( Fig. 22).
Upon reduction of the obtained state, the system goes back to its initial state, and the opposite circumrotation occurs. It is worth
mentioning that the switching process is not directional, as control on the switching kinetic is missing. Rotatory motors based on
catenanes have been reported, 97 but do not rely on electrochemically switchable units.
Reconsidering the same systems from a different perspective, it is interesting to highlight the fact that these catenanes have access
to several states, characterized by distinct optical properties. This feature may open new possibilities for the development of
molecular-scale electronic devices. The most relevant example that builds on this property has been reported by Stoddart and cow-
orkers. 93a This system is a homo [2]catenane composed of two 64 þ units, and has six experimentally accessible redox states, namely
0, 2þ, 4 þ, 6þ, and 8þ. The strength of this work is also the diversity of techniques that are used to investigate the system. Indeed,
besides electrochemical techniques, also electron paramagnetic resonance, nuclear magnetic resonance, X-ray crystallography and
superconducting quantum interference device magnetometry have been used in the investigation, showing the variety of tools avail-
able to chemists when approaching such systems.

2.16.4.3 Systems Working on Surfaces


The simplest way to interface an interlocked molecule with an interface is probably to form a SAM, and to this aim SAMs on metallic
gold have been extensively investigated. 98,99 Intriguingly, in some of these systems the surface has a dual role of stopper and
electrode.100
One example of this is represented by rotaxane 184 þ ( Fig. 23), consisting of a molecular axle that includes an electron-rich dii-
minobenzene moiety and a bulky adamantane unit, that serves as a stopper for the electron-poor macrocycle 64 þ, which is origi-
nally located on the diiminobenzene unit.100b
360 Electrochemically Controlled Supramolecular Switches and Machines

Figure 22 Schematic representation of the conformational changes associated with the reduction processes for catenanes 152 þ–172 þ. Vertical and
horizontal processes represent the NDI-centered and viologen-centered reductions, respectively. The dotted ellipses indicate that there is little informa-
tion about the donor ring location. Adapted from Cao, D.; Amelia, M.; Klivansky, L. M.; et al. J. Am. Chem. Soc. 2010, 132, 1110, with permission.
Copyright 2010 American Chemical Society.

Figure 23 Electrochemically driven ring shuttling in surface-anchored rotaxane 184 þ.


Electrochemically Controlled Supramolecular Switches and Machines 361

When a negative potential of  0.53 V versus SCE is applied to the electrode, the positively charged macrocycle is reduced and
moves toward the electrode. The favorable interaction that drives the ring in proximity of the surface is electrostatic attraction, that
increases as the electrode potential becomes more negative, and is not balanced any more by CT interactions between the station
and the ring. Successive oxidation of the reduced macrocycle restores the initial situation.
In the study of this system, double step chronoamperometry and impendance measurements have been used to determine the
shuttling rate constants, as well as the position of the macrocycle. In phosphate buffer at 298 K, the reduced macrocycle shuttles
toward the electrode with a rate constant of 320 s 1, whereas upon back oxidation the cyclophane moves to the diiminobenzene
station with a rate constant of 80 s 1. Detailed studies on the translocation rates were also performed in the following studies. 100c
This process was exploited to modulate the hydrophilicity of the surface that was evidenced by a change in contact angle; indeed,
when the macrocycle is in its stable form, the contact angle is z55 degrees, whereas it becomes z105 degrees under reductive
conditions.
Other works on rotaxanes immobilized at the electrode surface are based on the b-CDdferrocene, 100a metal coordination,98a
and hydrogen-bonding101 recognition motifs.
As evidenced in section “ Systems Working on Surfaces,” immobilization of the electroactive device on the surface of
a nanoparticle is a possibility that has also been explored, both in the case of self-assembling and interlocked systems. Indeed,
in close analogy with the pseudorotaxanes attached to a metal nanoparticles described in section “Systems Working on
Surfaces,” Stoddart and coworkers also reported the functionalization of nanoparticles with catenanes.80 However, one of
the first examples of nanoparticles functionalized with electrochemically switchable mechanically interlocked molecules
was provided by Fitzmaurice and coworkers.102 It (see Fig. 24) is composed by titanium dioxide nanoparticles with diameter
of 6 nm, to which a tripodal viologen-based [2]rotaxane is adsorbed, because of the chelation of Ti4 þ sites by peripheral phos-
phonic acids. The rigid tripodal structure was designed to orient the rotaxane normal to the surface, and to spatially separate
the viologen units of different rotaxanes, to avoid the dimerization of viologen cation radicals. The switchable rotaxane is
composed of a crown-ether that can reside on two differently functionalized viologen: V1 and V2 (see Fig. 24). V1 is the
preferred station for the macrocycle among the two, because V2 is methylated, thus it is more electron-rich and its pyridinium
rings are not coplanar in its reduced state, thus this site is a worse station for the crown-ether in comparison with V1 and it
undergoes reduction at a considerably more negative potential in comparison with it. The position of the ring upon electro-
chemical reduction of the viologen units was assigned by comparison of the reduction potentials between the nanoparticle-
attached [2]rotaxane and the corresponding axle component adsorbed on the nanoparticle. In acetonitrile, as the potential
becomes more negative, V1 is doubly reduced, and then V2 undergoes two reductive processes. Except for the second reduction
of V2, all the processes are shifted to more negative values when the crown is interlocked. On the basis of these data, it follows
that the macrocycle is initially located on V1, and remains located on V1 also upon its first reduction; however when V1 is
doubly reduced the macrocycle shuttles to V2; as V2 is reduced the crown moves away from it, but its location along the
axle cannot be clearly assigned.
Surfaces can be very convenient to interface molecular-level devices with the macroscopic world. A practical example has been
provided by Stoddart and coworkers, which could bend a microcantilever by electrochemically actuating palindromic [3]rotaxanes
bound to it. 103 The system is depicted in Fig. 25: the two macrocycles reside far from one another and encircle the TTF units. The
cantilever is coated with a gold film, and the macrocycles of the [3]rotaxane bear a side chain terminated with a thiol group, that
binds to the gold surface. Since the gold film of the microcantilever is used as counterelectrode, when a positive potential is applied,
TTFs are oxidized. The macrocycles concomitantly shuttle to DON station and as a result a bending of the microcantilever occurs, as
evidenced by the deflection of an incident laser beam. It was estimated that a force of about 650 pN was exerted on the cantilever by
every single rotaxane, and it was thanks to the cooperative action of these molecular muscles that a movement at the microscopic
level could be obtained.

Figure 24 Molecular structure of the TiO2 nanoparticle-bound [2]rotaxane studied in the group of Fitzmaurice. V1 and V2 indicate the viologen
stations that serve as recognition sites for the crown-ether macrocycle. Reproduced from Long, B.; Nikitin, K.; Fitzmaurice, D. J. Am. Chem. Soc.
2003, 125, 15490, with permission. Copyright 2003 American Chemical Society.
362 Electrochemically Controlled Supramolecular Switches and Machines

Figure 25 Cartoon representation of a palindromic [3]rotaxane capable of bending a cantilever upon oxidation. Yellow, green, and red stations
represent TTF, TTF2 þ, and DON units, respectively.

2.16.5 Nonequilibrium Systems

In the field of supramolecular chemistry, the control on the kinetic properties of the systems is a present challenge; indeed new
opportunities arise when the kinetics of the system is under control. As anticipated in the “ Introduction” section, two different
ways of keeping the system away from thermal equilibriumdunder kinetically controlled conditionsdcan be distinguished. On
the one hand in kinetically trapped system the kinetic barriers hamper the system from reaching its equilibrium distribution; on
the other hand in dissipative systems the continuous input of energy obliges the system to find a stationary state that is controlled
by the kinetic features of the system. In this section, three examples of electrochemically operated nonequilibrium systems will be
illustrated, following an increasing level of sophistication. Finally, molecular machines operated at the single-molecule level will be
discussed.
The first example 104 describes a system that is kinetically trapped, and couples the electrochemical control on thermodynamics
with an independent control on the kinetics, dictated by structural requirements. The system exploits a calix[6]arene that is a macro-
cycle with two different rims (Fig. 26): the “upper” rim has three phenylurea units, and the “lower” rim instead bears six phenolic
oxygens.
Calix[6]arene are suitable host for alkylviologen salts, 105 with binding constants in apolar solvents of about 107 M 1; this
affinity can be completely canceled when the viologen unit is monoreduced, as it is clearly evident from the voltammetric pattern
of a 1:1 mixture of the two components, that shows an irreversible anodic peak in correspondence of the first viologen reduction.
The length of the chains attached to the electroactive unit does not affect the thermodynamic affinity of the host–guest complex. The
threading (and therefore dethreading) kinetics show two important features: (i) in strongly apolar solvents like dichloromethane or
toluene the threading occurs only through the upper rim: this finding can be explained considering that the ion pairs involving the
viologen salt need to be broken, and phenylurea moieties lower the activation barrier of this necessary step; (ii) the length of the
alkyl chain has huge effects on the threading kinetics that can be easily controlled over eight orders of magnitude. Because of this
features, when an asymmetric axle is added to a solution of host, only one of the two possible isomers is formed, in a kinetically
trapped state. Since the thermodynamic stability of the two isomers is in principle identical, heating the solution for a sufficient time

Figure 26 Molecular structures of calix[6]arene host and their alkylviologen salts guest. When multiple symmetric guests are present in solution,
kinetic self-sorting occurs (left). Alternatively, if an asymmetric guest is present in solution, only one isomer is kinetically formed. Reproduced from
Arduini, A.; Bussolati, R.; Credi, A.; et al. J. Am. Chem. Soc. 2013, 135, 9924, with permission. Copyright 2013 American Chemical Society.
Electrochemically Controlled Supramolecular Switches and Machines 363

leads to the population of both isomeric states. This characteristic can be exploited to obtain the kinetic self-sorting also at the inter-
molecular level, because when symmetric axles with chains of different length are mixed with a calix[6]arene host, only one species
is formed: on a short timescale, the faster threading process is the preferred one.
A system exploiting an analogue dual control on thermodynamics and kinetics is a [2]rotaxane that displays a photoinduced
memory effect. 106
The second example 107 is still related to formation of kinetically trapped states: in this case the electrochemical stimulus controls
not only the thermodynamic stability of the system but also the kinetic barriers. The system is composed of the cyclophane 64 þ and
the axle reported in Fig. 27.
The axle has a viologen unit that becomes a favorable station for 64 þ under reductive condition, because of the radical cation
dimerization. The viologen is substituted on one side with a long alkyl chain, terminated with a 2,6-diisopropylphenyl unit that acts
as a stopper for 64 þ, and on the other side with a 3,5-dimethylpyridinium (PY). The PY unit is positively charged, and is therefore
engaged in repulsive electrostatic interactions with the positively charged macrocycle during the threading or dethreading process. It
is important to note that this interaction depends on the oxidation state of the macrocycle. Reduction of 64 þ to 62 þ decreases the
positive charge on the macrocycle, therefore threading/dethreading of 62 þ is kinetically favored, whereas threading/dethreading of
64 þ is kinetically impeded. As a consequence of this, when the previously formed pseudorotaxane is back-oxidized, before deth-
reading of 64 þ through PY can occur, the macrocycle shuttles over the isopropylphenyl unit and locates along the alkyl chain, in
an energetically demanding state. The isopropylphenyl unit was suitably engineered in order to have a shuttling barrier intermediate
between the threading barrier of 62 þ and 64 þ through the PY unit, therefore when a second reduction cycle is performed, the quick-
est way to form a radical dimer is to uptake a second 62 þ from solution (requiring the shuttling of 62 þ over PY), instead of back-
shuttling of the trapped 62 þ over the isopropylphenyl pseudostopper. Upon successive oxidation the newly uptaken macrocycle
moves along the alkyl chain. Overall two 64 þ are pumped in a kinetically trapped high energy state.
To achieve this result, the strategy was initially proved in a nonstoppered pseudorotaxane, 108 and then a single macrocycle was
translocated in an energetically demanding state.109
The last example does not concern a kinetically trapped system but rather a system where spatial control on self-assembly is
achieved through a dissipative process. 110 As mentioned in the introductory section, to obtain an electrochemically driven dissipa-
tive self-assembling system, two redox couples are required. In other words the system needs to be part of an electric circuit, and
a recognition process needs to be affected by this electron flow. Thanks to a bipotentiostat Huskens and coworkers could maintain
two micropatterned electrodes at fixed potentials (see Fig. 28).
The electrodes are kept at þ 0.134 and þ 0.534 V versus Ag/AgCl in water at pH 9.1. In solution ferrocene carboxylate anion
(FcCA) is present: the potential of the redox couple FcþCA/FcCA is intermediate between these potential values at which the

Figure 27 Molecular structures and operation sequence of a molecular pump system, based on a pseudorotaxane architecture. PY indicates the
3,5-dimethylpyridinium unit that acts as a pseudostopper and allows the threading of 62 þ, but not the dethreading of 64 þ, as a consequence of the
increased electrostatic repulsion between PY and the cyclophane.
364 Electrochemically Controlled Supramolecular Switches and Machines

Figure 28 Schematic representation of the strategy used to obtain spatial control over a self-assembly process under dissipative conditions. Blue
cavities represent b-CDs forming a monolayer; the gray dots connected by a red dot represent bis(adamantyl)-rhodamine (AD2) guest. When no
current is flowing through the electrodes an equilibrium state is reached and the surface is uniformly covered (A). Instead if the two electrodes are
kept at a fixed potentials, a gradient of the competitive guest ferrocene carboxylate anion (FcCA) is obtained and consequently the distribution of
[AD2$b-CD] complexes shows an opposite gradient (B). V versus Ag/AgCl. Adapted from Krabbenborg, S. O.; Veerbeek, J.; Huskens, J. Chem. Eur. J.
2015, 21, 9638, with permission.

electrodes are kept, therefore oxidation of FcCA occurs at the anode and reduction of FcþCA occurs at the cathode. The surface of
the electrode array is not innocent, indeed it is functionalized with a b-CD monolayer, that is a suitable host for FcþCA, but not for
FcCA; moreover, a competitive guest namely bis(adamantyl)-rhodamine (AD2) is present in solution. Close to the anode a local
high concentration of FcþCA is present, because of its formation at the electrode surface, on the contrary a locally higher concen-
tration of FcCA is formed in proximity of the cathode. For this reason the amount of host–guest complex formed by b-CD and AD2
is higher close to the cathode, whereas they are less concentrated in the anode proximity, as it reflects the competitive guest concen-
tration. For this reason, when a stationary current flows, a concentration gradient of AD2 bound to the surface is obtained that is
visualized observing the fluorescence signal of the competitive rhodamine guest moiety. When the solvent is removed, the gradient
of available b-CDs can be backfilled with another fluorescent guest, thus obtaining an opposite gradient, as the newly added guest
occupies the b-CDs that were not previously occupied by AD2.

2.16.5.1 Investigation of Molecular Machines at the Single-Molecule Level


So far only the behavior of ensemble of molecules has been discussed. Observation and operation of single molecular machines has
also been performed by several groups. Although fluorescence microscopy has also been used to investigate single molecular
machines, 111 scanning probe microscopy has been employed in the vast majority of these studies112 that include catenanes,113 pol-
yrotaxanes,114 wheelbarrows,115 gears,116 rack-and-pinion devices,117 and nano-vehicles.118 In this section two specific examples of
molecular machines operated by scanning tunneling microscopy will be illustrated. In both cases, inelastic electron energy transfer
through the excited state of the molecular machine provides the energy for motion. In this framework, such systems can be consid-
ered in a nonequilibrium systems, as it is the dissipation of the excitation energy that induces the motion.
The first system is surface-mounted rotatory motor that undergoes rotation upon electrochemical stimulation. The present
system 119 belongs to a family of compounds developed by Rapenne and coworkers,120–122 and is a piano-stool ruthenium
complex, anchored to an Au(111) surface through thioether groups (Fig. 29). The piano-stool structure decouples the rotor part
from the surface that is free to rotate at 80 K, but locked at 4 K. The rotor part bears four ferrocene units and a truncated arm
that provides desymmetrization.
Computational insights support the utmost importance of the truncated arm for the unidirectional functioning of the rotor, as
well as for its observation. Indeed it desymmetrizes the ground state rotational energy landscape; interestingly, this asymmetry is
preserved in the excited state, with potential energy surfaces that differ in the position of the energy maxima and in the direction
of the sawtooth shape depending on the excited orbital. Luckily, the accessible excited states are located in different parts of the
molecules, therefore applying a suitable potential in a specific location by means of a scanning tunneling microscope (STM) tip
it is possible to produce predominantly a specific excited state, therefore rotation can occur, and its direction is dictated by the direc-
tion of the sawtooth-shaped energy profile of the predominantly populated state. In this type of system, the quantum yields of
Electrochemically Controlled Supramolecular Switches and Machines 365

Figure 29 (A) Molecular structure of the rotatory motor capable of performing unidirectional rotation under electrochemical stimulation. The prefer-
ential direction of the rotation can be controlled externally. (B) Calculated potential energy surface associated to the rotation of the motor in the
ground state (blue) and in an excited state obtained mixing LUMOþ3 to LUMOþ6 excited states (green). Adapted from Perera, U. G. E.; Ample, F.;
Kersell, H.; et al. Nat. Nanotechnol. 2013, 8, 46, with permission. Copyright 2013 Macmillan Publishers Ltd: Nature Nanotechnology.

rotation are extremely low, the highest value being 4.14  10 9. In a different system, the directional motion of a molecular rotor
could also be achieved using a chiral tip. 123
The second example of molecular machine operated at the single-molecule level is a so-called “nanocar” ( Fig. 30) that takes
advantage of a motor-module developed by Feringa, which is known to undergo unidirectional rotation around the overcrowded
alkene moiety upon light and thermal stimulation, with a direction that is dictated by the chiral features of the molecule. As elec-
tronic and vibrational excitations can also be obtained using a STM tip to which a suitable potential is applied, it was anticipated
that the motor modules could be operated also on a solid support at the single-molecule level, using a STM tip both to image and
operate the nanocar.

Figure 30 Molecular structure of the meso-(R,S–R,S) isomer of the electrically driven nanocar. The blue arrows indicate the direction of the light-
driven rotation for each motor-module. The pink arrow highlights the rotational freedom around the central axle. Only the nanocars that land in the
meso-(R,S–R,S) conformation show directional motion.
366 Electrochemically Controlled Supramolecular Switches and Machines

Individual machines are imaged at 7 K after deposition on a Cu(111) surface by sublimation: the imaging conditions are suffi-
ciently mild (scanning potential < 60 mV) that the position of the machine is not affected after several imaging cycles. Instead, when
a voltage pulse of 600 mV is applied, electronic and vibrational excitation of the molecule is obtained. A following scan under mild
conditions reveals the position of the motor after energy dissipation. Ten excitation steps resulted in a 6 nm motion across the
surface. As a result of the rotational freedom around the bis-alkyne CeC single bond of the frame that is locked upon deposition,
not all nanocars move directionally; indeed, the four motor units need to operate in a synergic manner to afford directional motion.
The same requirement (all motor units rotating in the same direction) was used to perform control experiments in which random
motion is stimulated upon STM excitation.
As pointed out by the authors, to obtain the motor rotation, the energy dissipated in the isomerization step needs to be greater
than the adsorption energy. In the present system, this energy is lower than in other aromatic systems, owing to the strained helical
system that prevent the aromatic moieties from aligning parallel to the surface.

2.16.6 Concluding Remarks

In this article, electrochemically operated supramolecular devices have been described. Electrochemical techniques served also as
powerful monitoring tools, providing fingerprints to disclose the state of the molecules and to gain information on the spatial orga-
nization of redox-active sites, the energetics of the interactions involving such electroactive units and the kinetic features of the
switching processes, as well as the electron-transfer itself. In this perspective electrochemistry is a powerful tool to “read” the state
of the machine and operated“write”dit.
The described examples highlight the versatility of electrochemical techniques, as well as the importance of fundamental studies
to appreciate and develop the chemistry at the basis of electrochemically driven molecular devices and machines. These studies also
evidence that in order to obtain any technological application it is of utmost importance to interface these molecular entities with
the macroscopic world and make them work coherently. To this regard the latest achievementsdwhich were partially described also
in this articledsuggest that in a not too distant future such systems could indeed be at the basis of new technological findings.
Chemists have made considerable steps in dealing with electrochemically driven systems of increasing complexity, however it
has to be recalled that electrochemistry is only one of the several aspects that are involved in the development of new systems
that are engineered at the molecular level. Collaborative efforts from many disciplines toward a common goal would be surely bene-
ficial for our society. Indeed to translate molecular devices and machines into practically useful products demands that people
belonging to different disciplines ranging from mathematics to biology work together and learn a common language.

Acknowledgments

Financial support from Ministero dell’Università e della Ricerca (PRIN 2010CX2TLM “InfoChem”), and Università di Bologna (Finanziamenti di
Ateneo per la ricerca di base, SLaMM Project) is gratefully acknowledged.

References

1. Lehn, J.-M. Supramolecular Chemistry: Concepts and Perspectives; Wiley-VCH: Weinheim, 1995.
2. Balzani, V.; Scandola, F. Supramolecular Photochemistry; Horwood: Chichester, 1991.
3. Marcaccio, M.; Paolucci, F.; Roffia, S. In Trends in Molecular Electrochemistry; Pombeiro, A. J. L., Amatore, C., Eds.; Dekker: New York, NY, 2004; p 223.
4. Balzani, V.; Credi, A.; Venturi, M. Chem. Eur. J. 2002, 8, 5524.
5. Joachim, C.; Launay, J. P. Nouv. J. Chim. 1984, 8, 723.
6. Balzani, V.; Moggi, L.; Scandola, F. In Supramolecular Photochemistry; Balzani, V., Ed.; Reidel: Dordrecht, 1987; p 1.
7. Lehn, J.-M. Angew. Chem. Int. Ed. Engl. 1990, 29, 1304.
8. Balzani, V.; Credi, A.; Venturi, M. Molecular Devices and MachinesdConcepts and Perspectives for the Nanoworld, 2nd ed.; Wiley-VCH: Weinheim, 2008.
9. Ceroni, P.; Credi, A.; Venturi, M. Chem. Soc. Rev. 2014, 43, 4068.
10. Bard, A. J.; Faulkner, L. R. Electrochemical Methods. Fundamentals and Applications, 2nd ed.; Wiley: Hoboken, NJ, 2001.
11. Kaifer, A. E.; Gómez-Kaifer, M. Supramolecular Electrochemistry; Wiley-VCH: Weinheim, 1999.
12. Balzani, V., Ed. Electron Transfer in Chemistry; Wiley-VCH: Weinheim, 2001.
13. Hodes, G., Ed. Electrochemistry of Nanomaterials; Wiley-VCH: Weinheim, 2001.
14. Willner, I., Katz, E., Eds. Bioelectronics; Wiley-VCH: Weinheim, 2005.
15. Ceroni, P., Credi, A., Venturi, M., Eds. Electrochemistry of Functional Supramolecular Systems; Wiley: Hoboken, NJ, 2010.
16. Hammerich, O., Speiser, B., Eds. Organic Electrochemistry, 5th ed.; CRC Press: Boca Raton, FL, 2016.
17. Schalley, C. A., Ed. Analytical Methods in Supramolecular Chemistry; Wiley-VCH: Weinheim, 2012.
18. Boulas, P. L.; Gómez-Kaifer, M.; Echegoyen, L. Angew. Chem. Int. Ed. 1998, 37, 216.
19. Venturi, M.; Credi, A.; Balzani, V. Coord. Chem. Rev. 1999, 185–186, 233.
20. Cardona, C. M.; Mendoza, S.; Kaifer, A. E. Chem. Soc. Rev. 2000, 29, 37.
21. Cooke, G. Angew. Chem. Int. Ed. 2003, 42, 4860.
22. Credi, A.; Ferrer-Ribera, B.; Venturi, M. Electrochim. Acta 2004, 49, 3865.
23. Nijhuis, J.; Ravoo, B. J.; Huskens, J.; Reinhoudt, D. N. Coord. Chem. Rev. 2007, 251, 1761.
24. Gadde, S.; Batchelor, E. K.; Kaifer, A. E. Aust. J. Chem. 2010, 63, 184.
Electrochemically Controlled Supramolecular Switches and Machines 367

25. Credi, A.; Semeraro, M.; Silvi, S.; Venturi, M. Antioxid. Redox Signal. 2011, 14, 1119.
26. Gorman, C. B.; Smith, J. C. Acc. Chem. Res. 2001, 34, 60.
27. Ulman, A. Characterization of Organic Thin Films; Butterworth-Heinemann: Boston, MA, 1995.
28. Petty, M. C. Langmuir-Blodgett FilmsdAn Introduction; Cambridge University Press: Cambridge, 1996.
29. For an early representative example, see: Zhang, L.; Godínez, L. A.; Lu, T.; Gokel, G. W.; Kaifer, A. E. Angew. Chem. Int. Ed. Engl. 1995, 34, 235.
30. Ludden, M. J. W.; Reinhoudt, D. N.; Huskens, J. Chem. Soc. Rev. 2006, 35, 1122.
31. Bunker, B. C.; Huber, D. L.; Kushmerick, J. G.; et al. Langmuir 2007, 23, 31.
32. Allara, D. L. Nature 2005, 437, 638.
33. Bayly, S. R.; Gray, T. M.; Chmielewski, M. J.; Davis, J. J.; Beer, P. D. Chem. Commun. 2007, 2234.
34. Biscarini, F.; Cavallini, M.; Kshirsagar, R.; et al. Proc. Natl. Acad. Sci. U. S. A. 2006, 103, 17650.
35. Clemente-León, M.; Credi, A.; Martínez-Díaz, M.-V.; Mingotaud, C.; Stoddart, J. F. Adv. Mater. 2006, 18, 1291.
36. Delonno, E.; Tseng, H.-R.; Harvey, D. D.; Stoddart, J. F.; Heath, J. R. J. Phys. Chem. B 2006, 110, 7609.
37. Feng, M.; Gao, L.; Du, S.; et al. Adv. Funct. Mater. 2007, 17, 770.
38. Pease, A. R.; Jeppesen, J. O.; Stoddart, J. F.; et al. Acc. Chem. Res. 2001, 34, 433.
39. Ruben, M.; Payer, D.; Landa, A.; et al. J. Am. Chem. Soc. 2006, 128, 15644.
40. Shipway, N.; Willner, I. Acc. Chem. Res. 2001, 34, 421.
41. (a) Elio, M.; Otto, S. Nat. Nanotechnol. 2015, 10, 111; (b) Epstein, I. R.; Xu, B. Nat. Nanotechnol. 2016, 11, 312.
42. Breinlinger, E.; Niemz, A.; Rotello, V. M. J. Am. Chem. Soc. 1995, 117, 5379.
43. (a) Breinlinger, E.; Rotello, V. M. J. Am. Chem. Soc. 1997, 119, 1165; (b) Boal, A. K.; Rotello, V. M. J. Am. Chem. Soc. 1999, 121, 4914; (c) Kajiki, T.; Moriya, H.; Hoshino, K.;
et al. J. Org. Chem. 1999, 64, 9679; (d) Borgel, C.; Boyd, A. S. F.; Cooke, G.; et al. Chem. Commun. 2001, 1954; (e) Cooke, G.; deCremiers, H. A.; Duclairoir, F. M. A.; et al.
Tetrahedron 2003, 59, 3341; (f) Ge, Y.; Lilienthal, R. R.; Smith, D. K. J. Am. Chem. Soc. 1996, 118, 3976; (g) Deans, R.; Niemz, E.; Breinlinger, E.; Rotello, V. M. J. Am.
Chem. Soc. 1997, 119, 10863.
44. Groni, S.; Maby-Raud, T.; Fave, C.; Branca, M.; Schöllhorn, B. Chem. Commun. 2014, 50, 14616.
45. (a) Ge, Y.; Smith, D. K. Anal. Chem. 2000, 72, 1860; (b) Ge, Y.; Miller, L.; Ouimet, T.; Smith, D. K. J. Org. Chem. 2000, 65, 8831; (c) Cooke, G.; Sindelar, V.; Rotello, V. M.
Chem. Commun. 2003, 752; (d) Carroll, J. B.; Gray, M.; Cooke, G.; Rotello, V. M. Chem. Commun. 2004, 442.
46. (a) Carr, J. D.; Lambert, L.; Hursthouse, M. B.; Malik, K. M. A.; Tucker, J. H. R. Chem. Commun. 1997, 1649; (b) Carr, J. D.; Coles, S. J.; Hursthouse, M. B.; et al. Angew.
Chem. Int. Ed. 2000, 39, 3296; (c) Westwood, J.; Coles, S. J.; Collinson, S. R.; et al. Organometallics 2004, 23, 946.
47. Bu, J.; Lilienthal, N. D.; Woods, J. E.; et al. J. Am. Chem. Soc. 2005, 127, 6423.
48. Chan-Leonor, C.; Martin, S. L.; Smith, D. K. J. Org. Chem. 2005, 70, 10817.
49. Canevet, D.; Sallé, M.; Zhang, G.; Zhang, D.; Zhu, D. Chem. Commun. 2009, 2245.
50. (a) Gilday, L. C.; Robinson, S. W.; Barendt, T. A.; et al. Chem. Rev. 2015, 115, 7118; (b) Cavallo, G.; Metrangolo, P.; Milani, R.; et al. Chem. Rev. 2016, 116, 2478.
51. Lim, J. Y. L.; Cunningham, M. J.; Davis, J. J.; Beer, P. D. Chem. Commun. 2015, 51, 14640.
52. Amendola, V.; Fabbrizzi, L.; Mangano, C.; Pallavicini, P. Acc. Chem. Res. 2001, 34, 488.
53. Pramanik, S.; De, S.; Schmittel, M. Angew. Chem. Int. Ed. 2014, 53, 4709.
54. De, S.; Pramanik, S.; Schmittel, M. Angew. Chem. Int. Ed. 2014, 53, 14255.
55. Pramanik, S.; De, S.; Schmittel, M. Chem. Commun. 2014, 50, 13254.
56. (a) Boiocchi, M.; Fabbrizzi, L. Chem. Soc. Rev. 2014, 43, 1835; (b) Elhabiri, M.; Albrecht-Gary, A. Coord. Chem. Rev. 2008, 252, 1079.
57. Belle, P.; Pierre, J.-L.; Saint-Aman, E. New J. Chem. 1998, 22, 1399.
58. (a) De Santis, G.; Fabbrizzi, L.; Iacopino, D.; Pallavicini, P.; Perotti, A.; Poggi, A. Inorg. Chem. 1997, 36, 827; (b) Fabbrizzi, L.; Gatti, F.; Pallavicini, P.; Zambarbieri, E. Chem.
Eur. J. 1999, 5, 682; (c) Amendola, V.; Colasson, B.; Fabbrizzi, L.; Rodriguez Douton, M. Chem. Eur. J. 2007, 13, 4988.
59. Curiel, D.; Beer, P. D.; Tárraga, A.; Molina, P. Chem. Eur. J. 2009, 15, 7534.
60. (a) Rekharsky, M. V.; Inoue, Y. J. Am. Chem. Soc. 2002, 124, 813; (b) . In Cyclodextrins and Their Complexes: Chemistry, Analytical Methods, Applications; Dodziuk, H., Ed.;
Wiley-VCH: Weinheim, 2006.
61. (a) Lagona, J.; Mukhopadhyay, P.; Chakrabartri, S.; Isaacs, L. Angew. Chem. Int. Ed. 2005, 44, 4844; (b) Lee, J. W.; Samal, S.; Selvapalam, N.; Kim, H.-J.; Kim, K. Acc.
Chem. Res. 2003, 36, 621.
62. Rekharsky, M. V.; Mori, T.; Yang, C.; et al. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 20737.
63. Monk, P. M. S. The ViologensdPhysicochemical Properties, Synthesis and Applications of the Salts of 4,40 -Bipyridine; John Wiley & Sons: Chichester, 1998.
64. (a) Kim, H.-J.; Jeon, W. S.; Ko, Y. H.; Kim, K. Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 5007; (b) Ong, W.; Gómez-Kaifer, M. E.; Kaifer, A. E. Org. Lett. 2002, 4, 1791.
65. Das, A.; Ghosh, S. Angew. Chem. Int. Ed. 2014, 53, 2038.
66. Kumar, M.; Rao, K. V.; George, S. Phys. Chem. Chem. Phys. 2014, 16, 1300.
67. Ko, Y. H.; Kim, E.; Hwang, I.; Kim, K. Chem. Commun. 2007, 1305.
68. Amabilino, D. B.; Stoddart, J. F. Chem. Rev. 1995, 95, 2725.
69. Anelli, P. L.; Ashton, P. R.; Ballardini, R.; et al. J. Am. Chem. Soc. 1992, 114, 193.
70. (a) Asakawa, M.; Ashton, P. R.; Balzani, V.; et al. Chem. Eur. J. 1997, 3, 1992; (b) Devonport, W.; Blower, M. A.; Bryce, M. R.; Goldenberg, L. M. J. Org. Chem. 1997,
62, 885.
71. (a) Asakawa, M.; Ashton, P. R.; Balzani, V.; et al. Eur. J. Org. Chem. 1999, 985; (b) Balzani, V.; Credi, A.; Mattersteig, G.; et al. J. Org. Chem. 2000, 65, 1924.
72. Nielsen, M. B.; Lomholt, C.; Becher, J. Chem. Soc. Rev. 2000, 29, 153.
73. Ashton, P. R.; Balzani, V.; Becher, J.; et al. J. Am. Chem. Soc. 1999, 121, 3951.
74. Kosower, E.; Cotter, J. L. J. Am. Chem. Soc. 1964, 86, 5524.
75. (a) Iordache, A.; Oltean, M.; Milet, A.; et al. J. Am. Chem. Soc. 2012, 134, 2653; (b) Iordache, A.; Kannappan, R.; Metay, E.; et al. Org. Biomol. Chem. 2013, 11, 4383; (c)
Iehl, J.; Frasconi, M.; Jacquot de Rouville, H.; et al. Chem. Sci. 2013, 4, 1462.
76. Zhou, C.; Tian, J.; Wang, J.; et al. Polym. Chem. 2014, 5, 341.
77. Trabolsi, A.; Khashab, N.; Fahrenbach, A. C.; et al. Nat. Chem. 2010, 2, 42.
78. Chen, L.; Zhang, S.; Wang, H.; et al. Tetrahedron 2014, 70, 4778.
79. Bryce, M. R.; Cooke, G.; Duclaroir, F. M. A.; et al. J. Mater. Chem. 2003, 13, 2111.
80. Klajn, R.; Fang, L.; Coskun, A.; et al. J. Am. Chem. Soc. 2009, 131, 4233.
81. Klajn, R.; Olson, M. A.; Wesson, P. J.; et al. Nat. Chem. 2009, 1, 733.
82. (a) Nijhuis, C. A.; Boukamp, B. A.; Ravoo, B. J.; Huskens, J.; Reinhoudt, D. N. J. Phys. Chem. C 2007, 111, 9799; (b) Nijhuis, C. A.; Dolatowska, K. A.; Ravoo, B. J.;
Huskens, J.; Reinhoudt, D. N. Chem. Eur. J. 2007, 13, 69.
83. Sauvage, J.-P., Dietrich-Buchecker, C. O., Eds. Molecular Catenanes, Rotaxanes and Knots; Wiley-VCH: Weinheim, 1999.
84. (a) Browne, W. R.; Feringa, B. L. Nat. Nanotechnol. 2006, 1, 25; (b) Kay, E. R.; Leigh, D. A.; Zerbetto, F. Angew. Chem. Int. Ed. 2007, 46, 72.
85. Erbas-Cakmak, S.; Leigh, D. A.; McTernan, C. T.; Nussbaumer, A. L. Chem. Rev. 2015, 115, 10081.
86. Anelli, P. L.; Spencer, N.; Stoddart, J. F. J. Am. Chem. Soc. 1991, 51, 5131.
368 Electrochemically Controlled Supramolecular Switches and Machines

87. Brower, A. M.; Frochot, C.; Gatti, F. G.; et al. Science 2001, 291, 2124.
88. Gaviña, P.; Sauvage, J.-P. Tetrahedron Lett. 1997, 38, 3521.
89. Collin, J.-P.; Gaviña, P.; Sauvage, J.-P. New J. Chem. 1997, 21, 525.
90. Armaroli, N.; Balzani, V.; Collin, J.-P.; et al. J. Am. Chem. Soc. 1999, 121, 4397.
91. Durola, F.; Sauvage, J.-P. Angew. Chem. Int. Ed. 2007, 46, 3537.
92. (a) Raehm, L.; Kern, J.-M.; Sauvage, J.-P. Chem. Eur. J. 1999, 5, 3310; (b) Kern, J.-M.; Raehm, L.; Sauvage, J.-P.; Divisia-Blohorn, B.; Vidal, P.-L. Inorg. Chem. 2000, 39,
1555; (c) Poleschak, I.; Kern, J.-M.; Sauvage, J.-P. Chem. Commun. 2004, 474; (d) Létinois-Halbes, U.; Hanss, D.; Beierle, J. M.; Collin, J.-P.; Sauvage, J.-P. Org. Lett.
2005, 7, 5753.
93. (a) Barnes, J. C.; Fahrenbach, A. C.; Cao, D.; et al. Science 2013, 339, 429; (b) Barin, G.; Frasconi, M.; Dyar, S. M.; et al. J. Am. Chem. Soc. 2013, 135, 2466; (c) Zhu, Z. X.;
Fahrenbach, A. C.; Li, H.; et al. J. Am. Chem. Soc. 2012, 134, 11709; (d) Barnes, J. C.; Fahrenbach, A. C.; Dyar, S. M.; et al. Proc. Natl. Acad. Sci. U. S. A. 2012, 109,
11546; (e) Li, H.; Fahrenbach, A. C.; Coskun, A.; et al. Angew. Chem. Int. Ed. 2011, 50, 6782.
94. Cao, D.; Amelia, M.; Klivansky, L. M.; et al. J. Am. Chem. Soc. 2010, 132, 1110.
95. Hamilton, D. G.; Davies, J. E.; Prodi, L.; Sanders, J. K. M. Chem. Eur. J. 1998, 4, 608.
96. Gosztola, D.; Niemczyk, M. P.; Svec, W.; Lukas, A. S.; Wasielewski, M. R. J. Phys. Chem. A 2000, 104, 6545.
97. Hernández, J. V.; Kay, E. R.; Leigh, D. A. Science 2004, 306, 1532.
98. (a) Weber, N.; Hamann, C.; Kern, J.-M.; Sauvage, J.-P. Inorg. Chem. 2003, 42, 6780; (b) Whelan, F.; Gatti, C. M.; Leigh, D. A.; et al. J. Phys. Chem. B 2006, 110, 17076; (c)
Nikitin, K.; Lestini, E.; Lazzari, M.; Altobello, S.; Fitzmaurice, D. Langmuir 2007, 23, 12147.
99. (a) Tseng, H.-R.; Wu, D.; Fang, N.; Zhang, X.; Stoddart, J. F. ChemPhysChem 2004, 5, 111; (b) Jang, S. S.; Jang, Y. H.; Kim, Y.-H.; et al. J. Am. Chem. Soc. 2005, 127,
1563; (c) Huang, T. J.; Tseng, H.-R.; Sha, L.; et al. Nano Lett. 2004, 4, 2065; (d) Jang, S. S.; Jang, Y. H.; Kim, Y.-H.; et al. J. Am. Chem. Soc. 2005, 127, 14804; (e) Guo, X.;
Zhou, Y.; Feng, M.; et al. Adv. Funct. Mater. 2007, 17, 763.
100. (a) Willner, I.; Pardo-Yssar, V.; Katz, E.; Ranjit, K. T. J. Electroanal. Chem. 2001, 497, 172; (b) Katz, E.; Lioubashevsky, O.; Willner, I. J. Am. Chem. Soc. 2004, 126, 15520;
(c) Katz, E.; Baron, R.; Willner, I.; Richke, N.; Levine, R. D. ChemPhysChem 2005, 6, 2179.
101. Long, B.; Nikitin, K.; Fitzmaurice, D. J. Am. Chem. Soc. 2003, 125, 15490.
102. (a) Cecchet, F.; Rudolf, P.; Rapino, S.; et al. J. Phys. Chem. B 2004, 108, 15192; (b) Fioravanti, G.; Haraszkiewicz, N.; Kay, E. R.; et al. J. Am. Chem. Soc. 2008, 130, 2593.
103. Juluri, B. K.; Kumar, A. S.; Liu, Y.; et al. ACS Nano 2009, 3, 291.
104. Arduini, A.; Bussolati, R.; Credi, A.; et al. J. Am. Chem. Soc. 2013, 135, 9924.
105. Credi, A.; Dumas, S.; Silvi, S.; et al. J. Org. Chem. 2004, 69, 5881.
106. Avellini, T.; Li, H.; Coskun, A.; et al. Angew. Chem. Int. Ed. 2012, 51, 1611.
107. Cheng, C.; McGonigal, P. R.; Schneebeli, S. T.; et al. Nat. Nanotechnol. 2015, 10, 547.
108. Li, H.; Cheng, C.; McGonigal, P. R.; et al. J. Am. Chem. Soc. 2013, 135, 18609.
109. Cheng, C.; McGonigal, P. R.; Liu, W.; et al. J. Am. Chem. Soc. 2014, 136, 14702.
110. Krabbenborg, S. O.; Veerbeek, J.; Huskens, J. Chem. Eur. J. 2015, 21, 9638.
111. Nishimura, D.; Takashima, Y.; Aoki, H.; et al. Angew. Chem. Int. Ed. 2008, 47, 6077.
112. (a) Grill, L. J. Phys. Condens. Matter 2008, 20, 053001; (b) Lussis, P.; Svaldo-Lanero, T.; Bertocco, A.; et al. Nat. Nanotechnol. 2011, 6, 553.
113. Payer, D.; Rauschenbach, S.; Malinowski, N.; et al. J. Am. Chem. Soc. 2007, 129, 15662.
114. Shigekawa, H.; Miyake, K.; Sumaoka, J.; Harada, A.; Komiyama, M. J. Am. Chem. Soc. 2000, 122, 5411.
115. Grill, L.; Rieder, K.-H.; Moresco, F.; et al. Surf. Sci. 2005, 584, L153.
116. Manzano, C.; Soe, W.-H.; Wong, H. S.; et al. Nat. Mater. 2009, 8, 576.
117. Chiaravallotti, F.; Gross, L.; Rieder, K.-H.; et al. Nat. Mater. 2007, 6, 30.
118. (a) Shirai, Y.; Morin, J.-F.; Sasaki, T.; Guerrero, J. M.; Tour, J. M. Chem. Soc. Rev. 2006, 35, 1043; (b) Shirai, Y.; Osgood, A. J.; Zhao, Y. M.; et al. J. Am. Chem. Soc. 2006,
128, 4854; (c) Sasaki, T.; Tour, J. M. Org. Lett. 2008, 10, 897; (d) Sasaki, T.; Osgood, A. J.; Alemany, L. B.; Kelly, K. F.; Tour, J. M. Nano Lett. 2008, 10, 229; (e) Vives, G.;
Kang, J. H.; Kelly, K. F.; Tour, J. M. Org. Lett. 2009, 11, 5602.
119. Perera, U. G. E.; Ample, F.; Kersell, H.; et al. Nat. Nanotechnol. 2013, 8, 46.
120. Rapenne, G. Org. Biomol. Chem. 2005, 3, 1165.
121. Carella, A.; Coudret, C.; Guirado, G.; et al. Dalton Trans. 2007, 177.
122. Vives, G.; Gonzales, A.; Jaud, J.; Launay, J.-P.; Rapenne, G. Chem. Eur. J. 2007, 13, 5622.
123. Tiernery, H. L.; Murpy, J.; Jewell, A. D.; et al. Nat. Nanotechnol. 2011, 6, 625.
2.17 Extraction and Transport
L Mutihac, University of Bucharest, Bucharest, Romania
Ó 2017 Elsevier Ltd. All rights reserved.

2.17.1 Introduction 369


2.17.1.1 Survey on Macrocyclic Receptors Involved in Separation Processes 369
2.17.2 Liquid–Liquid Extraction 370
2.17.3 Transport Through Liquid Membrane 373
2.17.3.1 Transport Through Synthetic Ion Channels 373
2.17.3.1.1 Anion Transport 373
2.17.3.1.2 Cation Transport 374
2.17.3.2 Transport via Synthetic Mobile Carrier 375
2.17.4 Concluding Remarks 377
References 377

2.17.1 Introduction

The emerging and development of various macrocyclic receptors as hosts in supramolecular chemistry have entailed their involve-
ment in analytic applications and particularly in separation chemistry. Consequently, several studies are published on molecular
recognition and separation of various compounds like organic, inorganic, or biological ones by synthetic macrocyclic receptors,
which currently are active areas of research.1–8 Recently, a large number of macrocyclic receptors capable to interact and bind specif-
ically targeted guests have been synthesized and successfully employed in many fields as selective extractants in solvent extraction or
transporters through membranes with a view to the separation of compounds.9–15 Moreover, the challenging topical research field
of transmembrane ion transport of biological compounds based on synthetic receptors constitutes another interesting active area of
research in supramolecular chemistry.16–20
Based on remarkable host–guest properties of macrocyclic receptors toward chemical and biologically relevant compounds, the
present overview is focused on specific supramolecular aspects involved in analytic applications such as extraction (liquid–liquid
extraction) and transport through liquid membrane of a large variety of compounds involving representative macrocyclic receptors
as extractants and carriers. The transmembrane ion transport based on these synthetic ionophores is briefly reported. Due to a very
large number of remarkable reports on these topics, some references to the representative ones are pointed out only.

2.17.1.1 Survey on Macrocyclic Receptors Involved in Separation Processes


Along with the excellent host–guest properties toward various compounds, macrocyclic receptors (e.g., crown ethers, cryptands,
cyclodextrins, calixarenes, calix[4]pyrroles, cucurbiturils, and their derivatives) have become important receptors in analytic appli-
cations such as sensing 6 and separations (extraction, transport through membrane, chromatography, capillary electrophoresis,
micellar electrokinetic chromatography, and so forth)7,9,15 or bioanalytical applications2,4,7,12 like drug delivery, medical diagnos-
tics, imaging and radiotherapy, or molecular imprinted polymers as well. Renowned research groups study the binding properties of
supramolecular receptors toward various neutral and charged inorganic or organic species along with their potential applications as
carriers through membrane transport or ionic channel structures in transmembrane ion transport. Thus, these compounds can be
used as building blocks in monomeric channel compounds, one of the most attractive strategies for building synthetic ion channels
being self-assembly.16,17,21 The binding properties of macrocyclic receptors toward different chemical and biochemical compounds
with respect to noncovalent interactions like hydrophobic effects, p–p stacking, hydrogen bonding, metal coordination, and van
der Waals forces and cation–p and anion–p interactions involved in supramolecular complexes have been studied. As such, their
relevance in biological processes such as signal transduction or normal function of cellular/organelle structure and immune
response has also been highlighted.22,23
Apart from their properties as building blocks for the preparation of stimulus-responsive material, calix[4]pyrroles are attractive
molecular containers capable of recognizing and storing various guests. They have been used as selective extractants for anions, ion
pairs, and carriers through membranes for transporting ions and ion pairs. 24 More recently, the synthesis of ditopic calix[4]pyrrole
derivatives allows them to be employed as efficient extractants and carriers for both anions and cations.25
The ability of calixarenes to selectively bind a wide variety of chemical and biological compounds is largely due to facile func-
tionalization of lower and/or upper rim, as well as a consequence of a large number of different size cavities. Among these prop-
erties, calixarenes as supramolecular receptors are used in sensing and self-assembly, catalysis, drug delivery, nanoscience, and
separation chemistry due to their ability to form reversible complexes with both neutral and charged compounds. 13,26,27 Besides,
the calixarenes, having the property to exist in different conformers, allow making ion channels. In this respect, Gokel and

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.12505-3 369


370 Extraction and Transport

coworkers studied a series of calix[4]arenes substituted with pendant crown ethers pointing out that both 1,3-alt and 1,2-alt
conformers exhibit activity, whereas the cone conformer is inactive.28
The ability of calix[n]arenes to be used as storing material or guest transport was highlighted by Atwood and coworkers. 29 As
such, these compounds are employed for gas separation and transport of water molecules through hydrophobic crystals30 and allow
storing gases within cavities in seemingly nonporous p-tert-butylcalix[4]arene crystals.31,32
Calixcrowns receptors, due to their properties, are widely employed in host–guest chemistry and separation science. 33,34 Much
attention was devoted to the synthesis of chiral calixarenic receptors and their use in enantiomeric separation of chiral chemical and
biological compounds, an important task in pharmaceutical science. The cyclodextrins (CDs) and their derivatives with hydro-
phobic cavities are well known to form inclusion complexes with a large number of compounds via noncovalent interactions
that lead to supramolecular assemblies.35–37 In this respect, they are intensively studied as macrocyclic receptors for molecular
recognition in host–guest chemistry with several important applications in separation science especially due to their chiral proper-
ties.38,39 Moreover, involving cyclodextrins in the synthesis of ion channels has become fashionable in recent years.40 A remarkable
feature of cyclodextrin derivatives is their potential use in enantioseparation of chiral compounds as extractants or carriers through
liquid membranes.41
Considering the most remarkable property of crown ethers and cryptands to selectively recognize a large range of compounds,
both chemical and biological ones, through controlled interactions according to size, shape, and structure, render them competitive
candidates in selective extraction of different species from mixtures and in transport through liquid membrane as specific trans-
porters and confer them particularly useful applications in separation science. Consequently, they were extensively studied in recog-
nition, separation, purification, catalysis, nanoscience, sensing, and transport. 3,42–44
Recently, the cucurbit[n]urils (CB[n], n ¼ 5–8, 10, 14), because of their remarkable recognition ability, have attracted much
interest as selective receptors for various organic, inorganic, and biological guests with potential applications in sensing, catalysis,
advanced materials, drug delivery, and construction of supramolecular nanoarchitectures.45–47 Their structure characterized by
a rigidity and a varying hydrophobic nonpolarizable cavity and two oxygen-crowned portals leads to the ability of forming stable
host–guest complexes by ion–dipole, dipole–dipole, hydrogen-bonding, and hydrophobic interactions. Among other applications,
supramolecular compounds like rotaxanes, dendrimers, polyrotaxanes, and rotaxane-based molecular switches using cucurbit[n]
urils as molecular beads were synthesized.47–49 A few attempts have been made to use these receptors in separation applications.
Even more recently, some new members of cucurbiturils, namely, hemicucurbit[n]urils (n ¼ 6, 12), were synthesized. 50 It was
emphasized that hemicucurbiturils have the ability to form complexes with anions, cations, and small molecules. Yet, a few studies
only were reported so far concerning the molecular recognition and separation of biological compounds by hemicucurbiturils.
The present contribution focuses on recent studies on applications of some representative macrocyclic receptors in separation
processes, namely, liquid–liquid extraction and transport through liquid membranes.

2.17.2 Liquid–Liquid Extraction

Liquid–liquid extraction (or solvent extraction) as a separation technique is based on the transfer of a species from one liquid phase,
usually an aqueous phase, to another liquid phase, usually an organic one. Because of its simplicity, this technique is one of the
most widespread in laboratory and industrial separation processes. Moreover, this technique is known for many applications in
sample preparation for the chromatographic analysis in order to achieve sample simplification or analyte concentrationdissues
that are discussed largely in dedicated books. 51,52
By using both calix[4]azacrowns (1,3-[ethylene-bis-aminocarbonylmethoxy)]-p-tert-butylcalix[4]arene) 1 ( Fig. 1) and (1,3-
[propylene-bis-aminocarbonylmethoxy)]-p-tert-butylcalix[4]arene) 2 (Fig. 1), the native and methyl ester amino acids were
extracted from aqueous phase (pH ¼ 5.5 in buffer, MES/NaOH system) to organic phase (CHCl3) as ion pairs in the presence of
picrate or tropaeolin 00 as counterions.53

Figure 1 Chemical structures of calix[4]azacrowns 1 and 2.


Extraction and Transport 371

Both receptors 1 and 2 exhibited high extractability toward amino acids under study as ion pairs in the presence of picrate
ion from 82% (L-tryptophan methyl ester) to 99% for L-serine methyl ester by using receptor 1 and from 89% (L-alanine) to
98% (L-phenylalanine methyl ester) by receptor 2. It was observed that the extraction yields obtained with calix[4]azacrown
(2) are larger than those obtained by calix[4]azacrown (1) for the same amino acids with one order of magnitude. In this case,
the structure of the receptor is responsible for these results. The calix[4]azacrown (2) has an additional methylene group in the
bridging chain, which slightly influences the calixarene conformation. By comparing these results with those obtained for the extrac-
tion of alkaline earth and transition metal picrates from water to dichloromethane with the same receptors, the values of extraction
yields of the alkaline earths are lower. 54 It was also suggested that the extraction depends on the structure of the amino acid, the
structure of calixarenes, and the nature of the anion used as counterion. Liquid–liquid extraction performed with both calix[4]aza-
crowns in the presence of tropaeolin 00 also showed good extraction abilities toward amino acid methyl esters, but lower yields of
extractability in the presence of tropaeolin were observed. A molecular dynamic study of liquid–liquid extraction of cesium picrate
salt with calix[4]crown derivative at the water–chloroform interface highlighted the important role of solvation effects and of small
amounts of water on the extraction.55
Novel lower-rim calix[4]arene diethylene glycol amide derivatives were synthesized and used in complexation of alkali and alka-
line earth metal cations in two solvents with different solvation and hydrogen-bonding properties (methanol and acetonitrile). 56
The abilities of these receptors to extract alkali and alkaline earth metal cations as metal picrates from aqueous phase to dichloro-
methane were investigated and the results indicated good extractability of receptors toward metal cations.
The new azocalix[4]arene amide and some of their telomers were synthesized and characterized by spectroscopic methods and
elemental analysis. 57 Their metal extraction abilities were investigated comparatively. It was observed that telomere structures of
azocalix[n]arene showed higher extraction rate than their monomers. The authors recommended these compounds, mostly
telomers to be employed in the field of ion-selective electrode.
Along the same topics on calixarene functionalization, the synthesis of pyrimidylthioamide-derivatized calix[4]arene and their
selective extraction toward heavy metals (Cu2 þ, Co2 þ, Ni2 þ, Pb2 þ, Hg2 þ, and Hg2 þ) were investigated. The compound with pyr-
imidylthioamide groups on the lower rim selectively extracted Hg2 þ and Pb2 þ. 58
The liquid–liquid extraction of native aromatic amino acids and derivatives from aqueous phase (pH ¼ 5.5) to chloroform was
carried out with a series of calixarenes in the cone conformation variously substituted by acid or amido functions, glycolic chains,
and hydroxyl groups ( Fig. 2, 3–7).59 Very large differences in extraction percentage between native and derivatives amino acids were
noticed. The extractability is essentially controlled by the structure of both amino acid and calix[4]arene derivative; these receptors
can be used without counterion. In the case of amino acids, particularly their hydrophobicity influenced the extraction process.
Diamido-calix[4]arene 3 provided the highest affinity toward L-phenylalanine methyl ester (21%), and the highest L-tryptophan
methyl ester extraction was observed for tetraamido 7 (65%) and diamido 3 (61%) calix[4]arene derivative. The functional groups
introduced in the calixarene structure play an important role in binding of amino acid, mainly the OH groups, glycolic chains, and
amido functions known for their ability to form oxygen-cation, hydrogen-bonded, and electrostatic interactions.
The liquid–liquid extractions of a series of amino acid methyl esters as ion pairs in the presence of picrate anion and tropaeolin
00 with functionalized calix[4]arene (5,11,17,23-tetrakis(N-methylpiperazino)-25,26,27,28-tetrahydroxycalix[4]arene) empha-
sized the strong influence of the nature of the anion used as counterion. High yields were obtained for picrate anion. 60
The extraction behavior of some nucleobases by the calix[6]arene carboxylic acid derivative with tert-octyl groups at the upper
rim in combination with di-(2-ethylhexyl)phosphoric acid (D2EHPA) as a synergistic reagent in an isooctane system was investi-
gated. 61 It was found that D2EHPA exhibited a synergistic effect by enhancing the efficiency of adenine and cytosine extraction. The
complex formation between adenine and one calix[6]arene and two D2EHPA molecules was studied by Job’s method.

Figure 2 Chemical structures of compounds 3–7.


372 Extraction and Transport

Figure 3 Chemical structures of modified cyclodextrin 8 and meso-octamethylcalix[4]pyrrole 9.

A chiral extraction system combining hydrophobic L-iso-butyl tartrate in organic phase and hydrophilic hydroxypropyl-b-cyclo-
dextrin for enantioselective partition of 2-phenylpropionic acid enantiomers on the basis of biphasic recognition was studied by
Tang and coworkers. 62 High separation factors were obtained at low pH concerning the enantioseparation of 2-phenylpropionic
acid enantiomers, and an increase of the separation factor from 1.08 to 1.69 by using the combined system as compared with
the monophasic system was also reported.
Liquid–liquid extraction by chiral heptakis (2,3,6-tri-O-acetyl)-b-cyclodextrin ( Fig. 3, 8) for emphasizing its ability for enantio-
meric discrimination of some native aromatic amino acids (tryptophan, phenylalanine, and tyrosine) and methyl ester derivatives
(tryptophan methyl ester hydrochloride, phenylalanine methyl ester hydrochloride, and tyrosine methyl ester hydrochloride) was
studied.41 The receptor provided a higher affinity toward tryptophan methyl ester; obviously the hydrophobicity of amino acid
could be responsible for this behavior. The enantioselectivity discrimination of L-enantiomer over D-enantiomer by extraction
was relatively low and the chiral receptor showed much the same enantioselectivity toward amino acids, preferentially to L-amino
acids. Circular dichroism measurements were performed to reveal the enantiomeric discrimination.
The same receptor was employed in liquid–liquid extraction of dopamine by means of an ion-pairing mechanism in the pres-
ence of picrate and tropaeolin 00, and the results proved that functionalized cyclodextrin is a good extractant for dopamine, espe-
cially in the presence of picrate anion. 63
An interesting study on the ability of meso-octamethylcalix[4]pyrrole ( Fig. 3, 9) to be used as ion-pair receptor for cesium chlo-
ride and cesium bromide in nitrobenzene solution was investigated by Sessler and coworkers by means of solvent extraction.24,64
The stoichiometry of the ion-paired formation was 1:1:1 cesium–calix[4]pyrrole–halide complex. By the synthesis of ditopic
receptor calix[4]crown-5 strapped calix[4]pyrrole as receptor for both CsF and KF ion pairs, characterized by 1H NMR measurements
and X-ray diffraction, the selective extraction of KF over CsF from an aqueous phase to nitrobenzene was achieved.25 A well-
documented review about the applications of supramolecular anion recognition was recently reported by Gale and coworkers.4
New receptors like hemicucurbit[n]uril, n ¼ 6,12 ( Fig. 4, 10, 11), were investigated as potential receptors for molecular recogni-
tion of some amino acids. In this respect, a series of native amino acid and their derivatives were extracted from an aqueous phase to
a chloroform phase as ion pairs in the presence of tropaeolin 00.65 Both receptors exhibited extractability toward amino acids with

Figure 4 Chemical structures of hemicucurbit[n]uril, n ¼ 6, 12, 10, 11.


Extraction and Transport 373

good affinity toward hydrophobic amino acid derivatives (L-phenylalanine, L-leucine, and L-valine). The structural properties of
amino acids, the structure of receptor, the pH, and the thermodynamic equilibrium influenced the extraction efficiency.

2.17.3 Transport Through Liquid Membrane

In chemistry and supramolecular chemistry, liquid membranes are frequently used both to evaluate molecular recognition and
transport properties of receptors and to separate chemical and biochemical species. The transport of species through a membrane
is achieved by the receptors, namely, transporters or carriers, with the ability to selectively bind the target compound, diffuse the
formed complex across the membrane, and release the compound on the other side of the membrane. This is the transport through
membrane by mobile carriers. In this case, there is a high selectivity of transporters but fluxes are influenced by the rate of diffu-
sion. 21 The transporters can be classified into two types: channel and mobile carriers. Concerning channels, they could be mono-
meric channels or self-assembled pores.21 It was pointed out that the transport rate through channels is always higher than the
transport rate achieved by mobile carriers. One of the most attractive strategies for building synthetic ion channels is self-
assembly.66 Ion transport through biological membranes is of most importance in both cell function and drug delivery.
Considerable effort has been devoted by supramolecular chemists in the challenging topics of synthetic ion channel design and
synthesis of molecules capable of acting as ion channels. There are well-known research groups working on the topics of synthetic
ion channels like the teams of Gokel, 5,16,67 Fyles,18,19 Matile,17,68 Smith,21 Gale,4,10 and Sessler.24,25
The study of membrane transport is of current interest for supramolecular chemistry community. It is known that in living
systems, a very important issue concerns the control of transmembrane ion transport. Hereafter, the study of the membrane trans-
port will proceed by classification on the basis of the transporters mechanism (channel or mobile carrier).

2.17.3.1 Transport Through Synthetic Ion Channels


Back in 1982, Tabushi and coworkers 69 reported the first synthetic ion channels based on a modified b-cyclodextrin used in Co2 þ
transport. A large number of macrocyclic receptors such as calixarenes,70 calix[4]pyrroles,24 cyclodextrins,40,71 crown ethers,72–75
cucurbiturils, 76 and resorcin[4]arene,75 together with peptides nanotubes,77 oligophenyl barrel-stave structures,68,78 and oligoester
bolaamphiphiles79,80 as carriers or ionic channel structures in synthetic membrane transport, have been intensively studied by
remarkable research groups. In living cells, the natural ion channels are large protein complexes. They are responsible for miscel-
laneous fundamental cellular functions like chemical signaling, osmotic homeostasis, control of transmembrane potentials, pH
control, and transepithelial and ion transport.81 The malfunction of ion channels leads to a number of diseases such as Dent’s
disease, myotonia, epilepsy, cystic fibrosis, and other genetic diseases.82 Drug targets in common diseases are another feature of
ion channel function.
The first crystal structure of a potassium ion channel isolated from Streptomyces lividans was reported by MacKinnon and
coworkers in 1998, 83,84 and the mechanism of water permeability through biological membranes by aquaporin water channels
was reported in 2002 by Agre and coworkers.85 The discovery of the aquaporins is of great importance to normal physiology
and to pathophysiology of clinical disorders.86
Recent studies in this field have produced new and interesting information, thereby enriching our knowledge of the mechanisms
of ion transport through synthetic ion channel, along with providing a better understanding of medical disorders caused by mal-
function of biological ion transport. 87
The synthesis of synthetic ion channels, the control of ionic selectivity, and biological functions of channels remain a challenge
of modern synthetic organic chemistry and supramolecular chemistry as well. 19,66 The importance of anion–p interactions for the
transmembrane transport of anions was emphasized by Matile and coworkers.88 According to the reported data, there are several
systems that combine the molecular recognition of the macrocycles with the membrane-spanning substituents that allow ions to
traverse a phospholipid bilayer.89 An important issue in the building of ion channels is given by the relationship between function
and structure of the receptors used.

2.17.3.1.1 Anion Transport


The transport of anions like bicarbonate, phosphate, and the most abundant anion in physiological solutions, chloride, through
lipid bilayer has a fundamental role in biological systems. As such, much attention was devoted to synthesis of anion transporters
either as synthetic channel or as mobile carrier. 10 The structure of cyclodextrins permits them to span a lipid bilayer providing pores
for anion transport. Interesting contributions in transport activity of cyclodextrin channel were carried out by Gin and coworkers
who synthesized two cyclodextrin-based ion channels able to transport anions and cations depending on the pH.90,91
Even though the results obtained by employing metal–organic frameworks for self-assembly into pore-like structures showed
low selectivity for anions, Tecilla and coworkers employed a stable 4 þ 4 metallacycle based on porphyrins with carboxylic acid
groups as an ion channel. 92 By using calix[4]arene in the 1,3-alt conformation functionalized with spermidine (Fig. 5, 12), the trans-
port of halide anions was achieved, whereas that for cations or oxygenated anions was not.93
Along the same topics, Davis and coworkers 70 highlighted the property of calix[4]arene tetrabutylamide 1,3-alt to form Cl
channels in both vesicle and cellular systems showing a significant Cl =SO42 transport selectivity. Moreover, it was reported
that minor changes in the primary structure of calix[4]arene amide lead to large differences in the chloride transport properties.94,95
374 Extraction and Transport

Figure 5 Structure of compound 12.

Based on solid-state crystal structures of partial-cone calix[4]arene and its para-substituted, the self-assembly and the Cl ion trans-
port could be controlled by the conformation of the side chain on the inverted arene of the partial cone.
Other synthetic ion transporters developed by Gokel and coworkers are the amphiphilic heptapeptides that are able to select
Cl > 10-fold over Kþ and showed voltage-dependent gating. 96,97 Isophthalamides and their pyridine analogs are also able to trans-
port Cl ions across phospholipid bilayer membrane and exhibit channel activity.98
There are many other interesting small molecules with the ability to act as synthetic ionophores in transmembrane ion transport.
Valuable studies on rigid-rod molecules in biomembrane showed that, by appending aromatic electron acceptors to the rigid rod, it
is possible to form higher order channels by self-assembling into a barrel-shaped channel. 68
Considerable effort was dedicated to highlighting the functions of ion channels. In this respect, by using the voltage-clamp tech-
nique, Fyles and coworkers studied the ion channel activity of a series of oligoester bolaamphiphiles in diphytanoyl phosphatidyl-
choline planar bilayers. 80,99

2.17.3.1.2 Cation Transport


A large number of compounds have the ability to form cation channels. 99,100,72 The most encountered biological cations are alkali
metal cations, Naþ and Kþ. Starting with the synthesis of cone-4-tert-butylcalix[4]arenetetra(diethylamide) (Fig. 6, 13) that is able to
extract Naþ from an aqueous phase with a selectivity factor of 68 over Kþ, many studies have been dedicated to artificial ion channel
design.101
Hydraphiles as synthetic ionophores mimicking the properties of protein channels have been inserted in the phospholipid bila-
yers forming unimolecular pores that span the bilayer. 67,102 They comprise crown polyesters as amphiphilic headgroups and as
a polar central element and accomplish the transport of Naþ and Kþ. Moreover, these synthetic ion channels are toxic to various
bacteria and yeast; hence, they can be used in chemotherapy.67 The important role of cation–p interactions in channel protein-
mediated ion transport was documented by Gokel and coworkers.103 Calix[n]arenes along with crown ethers, cyclodextrins, and
cucurbit[n]uril are other common building blocks in monomeric channel compounds.
Conjugated calix[4]arene-cholic acids were reported as synthetic ionophores, which are able to transport Naþ and Hþ across
a vesicle membrane by means of a unimolecular mechanism. 104 It was observed that the compounds characterized by the 1,3-alter-
nate conformation of calix[4]arene derivatives were more active than compounds in cone conformation of calix[4]arene derivatives
with respect to Hþ and Naþ transport rate. The molecular length of the channel and the ability to span the membrane seem to be the
responsible factors to account for. The studies reported by Barboiu and coworkers74 referred to a number of urea-linked crown

Figure 6 Chemical structures of compounds 13 and 14.


Extraction and Transport 375

Figure 7 Chemical structures of cucurbit[n]uril, n ¼ 5, 6, 15, and 16.

ethers as active self-assembled structures that can act as ion channels in bilayer membranes (Fig. 6, 14). In this case, the compounds
differ by the size of crown ethers and the ureido alkyl tail as well.
In order to mimic the ionic activity of KcSA channels, Barboiu and coworkers developed an artificial ion channel of H-bonded
hexyl-benzoureido-15-crown-5-ether, which favored Kþ cations with respect to Naþ cations. 105 A new class of resorcin[4]arenes
containing phenoxyalkyl ether, phenoxypolyether, or phenoxycrown ether moieties as potassium-selective ion channel over other
alkali metal cations has been advanced by Beer and coworkers.106 Novel artificial ion channels based on CB[n], n ¼ 5, 6 (Fig. 7, 15),
which are able to transport proton and alkali metal ion through lipid membrane, were reported by Kim and coworkers.107 They
observed the channel behavior during bilayer patch-clamp experiments. The mechanism of channel formation was not elucidated.
Calix[4]arenes in the 1,3-alternate conformation were employed to build several molecular architectures, such as molecular tubes
for cation and anion transport or containers for fixation of gases. Moreover, in the presence of Naþ, calix[4]arenes in the 1,3-alternate
conformation bearing four guanosine units form a self-assembled nanotube. 108 A series of symmetric and asymmetrical calix[4]
tubes featuring alkyl or phenyl substituents at the upper rim proving selectivity for potassium complexation over all metal cations
from group I have been reported.109 The first examples of N, C-linked peptidocalix[4]arenes by incorporating the calix[4]arene
amino acid into a pseudopeptide sequence highlighting pelicular self-assembling properties in apolar media were reported by
Ungaro and coworkers.110 Calix[4]arenes in 1,3-alternate conformation (Fig. 7, 16) have been used for the construction of extended
calix[4]tubes having the ability to act as cation receptors along the tube.111

2.17.3.2 Transport via Synthetic Mobile Carrier


The mobile carriers are small organic molecules that shuttle ions from one side to the other of either a biological or artificial
membrane. These ionophores have to bind selectively and reversibly the ions or molecules and form a lipophilic complex ion trans-
porter that diffuses through a bilayer membrane. A large number of supramolecular structures have been developed to be used both
in the molecular recognition of relevant chemical and biological compounds and as synthetic ion transporters through a membrane.
Several studies on the selectivity and the mechanism of transport of biological compounds by means of synthetic small molecules
have been published.
Potential applications of calix[4]pyrroles as carriers across lipid bilayer membranes have been reported by the Sessler 11 and
Gale4 groups. Calix[4]pyrroles are attractive systems and, similarly to calixarenes, they are facile to functionalize, thus becoming
useful receptors for recognizing, storing, and transporting different guests. Consequently, the parent octamethylcalix[4]pyrrole 9
(Fig. 3) carrier selectively transports CsCl through 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine vesicles via a symport mech-
anism (chloride and cation cotransport).11 By replacement of the b-pyrrole protons from compound 9 by fluorine substituents
(octafluorocalix[4]pyrrole 17, Fig. 8), a change was effected in the transport mechanism from CsCl symport to Cl/NO3 anti-
port.112 It was found that carrier 17 transports also chloride/bicarbonate antiport.
Gale and coworkers 113 synthesized a bis-1,2,3-triazole strapped calix[4]pyrroles (Fig. 8, 18,19,20). These new receptors with two
triazole groups display anion binding affinities and lipid bilayer transport properties. A similar switch was observed in the transport
mechanism for strapped calix[4]pyrroles 18, 19, and 20. Thus, the cesium chloride symport transport for 18 is replaced by a Cl/
NO3 antiport mechanism with receptors 19 and 20. The efficiency of chloride transport through a lipid bilayer depends on the
length of the alkyl chain from bis-triazole strapped calix[4]pyrroles.
The metal cation binding ability of cone-p-tert-butylhexahomotrioxacalix[3]arene tris(acetic acid) ( Fig. 9, 21) toward Naþ, Kþ,
Ag , and Pb2 þ together with the transport through liquid membrane of Cr3 þ, Fe3 þ, Co2 þ, and Cu2 þ using carrier 21 was reported
þ

by Cragg and coworkers.114


By means of 1H NMR and computational analysis used to characterize the binding affinities of receptor 21 toward metal cations,
it was suggested that the encapsulation of a cation in the cavity is possible through the phenoxy and carbonyl oxygen atoms. The
active transport of Cr3 þ, Fe3 þ, Co2 þ, and Cu2 þ through a chloroformic membrane was achieved. 114
376 Extraction and Transport

Figure 8 Chemical structures of compound 17 and strapped calix[4]pyrroles 18–20.

Figure 9 Chemical structure of compound 21.

Jin and coworkers 115 studied the alkali metal ion transport by a series of p-tert-butylcalix[n]arene (n ¼ 4, 5, 6, 7, 8) esters across
a soybean phospholipid bilayer membrane by using a voltage-clamp method. The relationship between the ion transport selectivity
of the calix[n]arene esters and the size of the alkali metal ions was established. Moreover, the same authors performed the
membrane transport of neurotransmitters such as acetylcholine, carbachol (carbamylcholine), and choline across a phospholipid
bilayer by a calix[6]arene esters.116
As mentioned earlier, the calix[n]arenes and their derivatives are suitable compounds to act as carriers in transport through liquid
membranes of relevant biological compounds like amines, amino acids, peptides, acetylcholine, and alkali metal ions aiming their
separation.117–119 Starting from the first transport of some derivative amino acids through a liquid membrane by using calix[6]arene
as carrier by Chang and coworkers, 120 the applications of calixarenes in separation science have been maturing. By means of a new
calix[4]arene with chiral pendant groups, a selective transport of some amino acid derivatives and Z-amino acid carboxylates into
CH2Cl2 was carried out.121
Some biogenic amines and amino acid methyl esters were transported through liquid membrane by p-tert-butylcalix[4]arene
(n ¼ 4, 6, 8) in the presence of counterions like picrate 122 or tropaeolin 00 ([4-40 -(anilinophenylazo)benzenesulfonic acid]) as
ion pairs.123 The transport driving force was the pH gradient between the source and the receiving phase. The results revealed
the influence of the structure of calix[n]arenes, pH, the nature of the counterion employed, the structure of amino acids, and the
membrane solvent on the transport experiments.
There is a recent review about the properties of functionalized calix[n]arenes to recognition, which act as extractants in solvent
extraction and as transporters through liquid membranes of various biological amino compounds. 14 The selective active transport
through a chloroform liquid membrane assisted by pH gradient of aromatic amino acid native and derivatives (L-tryptophan methyl
ester, L-phenylalanine methyl ester, and L-tyrosine methyl ester) by using a series of calix[4]arenes substituted by acid and amido
functions, glycolic chains, and hydroxyl groups (Fig. 2, 3–7) as carriers has been performed by Mutihac and coworkers.124 The
receptors having diacid and tetraamido functions showed better transport yields, suggesting the possibility of aromatic amino
acid methyl ester separation. The sequence of the decreasing transport yields is L-TrpOMe > L-PheOMe [ L-TyrOMe for all receptors,
but the high transport yields were obtained with diacid receptor 3 and tetraamido receptor 7. It is worth to mention that the func-
tional group attached to calix[4]arene profoundly enhanced the transport abilities of calix[4]arenes.
Extraction and Transport 377

The ability of calix[6]arene carboxylic acid derivatives to act as carriers through a liquid membrane for aromatic amino acids,
adenine, and catecholamine was studied by Oshima and coworkers. 125 The transport through the membrane was controlled by
changing the pH between the source and the receiving aqueous phases. Bulk liquid membrane transport of native and aromatic
amino acid methyl esters by using heptakis (2,3,6-tri-O-acetyl)-b-cyclodextrin (8) (Fig. 3) as carrier provided the highest affinity
of this receptor toward L-tryptophan methyl ester and less toward L-tyrosine.126 The active transport through the chloroform liquid
membrane assisted by the pH gradient of aromatic amino acid native and methyl esters by using heptakis (2,3,6-tri-O-acetyl)-
b-cyclodextrin as carrier showed the following sequence of decreasing transport yields of amino acids: L-TrpOMe > L-
PheOMe > L-TyrOMe. The transport yields are lower, differing significantly for the same aromatic amino acids in native form; these
differences may be attributed to the hydrophobicity of the amino acids. The separation of enantiomers, especially in the pharma-
ceutical field, is of current interest for chemistry and supramolecular chemistry research groups. Continuing their studies, Mutihac
and coworkers performed transport through a bulk liquid membrane by chiral heptakis (2,3,6-tri-O-acetyl)-b-cyclodextrin (8),
highlighting its ability to enantiomerically discriminate between some native (tryptophan, phenylalanine, and tyrosine) and methyl
ester derivatives (tryptophan methyl ester hydrochloride, phenylalanine methyl ester hydrochloride, and tyrosine methyl ester
hydrochloride).41 The receptor exhibited carrier ability, especially for the L-amino acid forms. Circular dichroism was carried out
to point out the enantiomeric discrimination.
Considerable effort has been devoted to designing ion-pair receptors able to simultaneously recognize and bind cationic and
anionic guests, as either ion pairs or cobound salts. In this respect, Sessler, Gale, Lee, and coworkers 127 studied the oligoether-
strapped calix[4]pyrrole ion-pair receptor, which has one anion binding site and two possible cation binding sites, which are
used as an effective carrier in a liposomal membrane for several chloride anion salts of group 1 cations by both symport
(chloride–cation cotransport) and antiport (nitrate for chloride exchange) mechanisms. It was observed that this transport is
different compared with the transport of octamethylcalix[4]pyrrole, which acts as carrier for cesium chloride. A well-
documented review about these compounds was reported by Sessler and Kim.11
The selective active transport through the liquid membrane assisted by the pH gradient of some amino acid acids as ion pairs in
the presence of picrate as counterion by using receptor 1 (1,3-[ethylene-bis-aminocarbonylmethoxy)]-p-tert-butylcalix[4]arene)
( Fig. 1) as carrier showed the following sequence of the decreasing transport yields of amino acids: L-Ala (43%) > L-SerOMe
(42%) > L-ValOMe (33%) z L-TrpOMe (33%) > L-PheOMe (24%) > L-LeuOMe (12%). Moreover, the hydrophobicity of the amino
acid appeared as an important parameter in all transport experiments.53
Chiral aza-15-crown-5 ethers bearing phenyl and phenoxymethyl moieties exhibited strong binding properties toward amino
acid esters and their potassium and sodium salts, but poor enantiomeric discrimination for the enantiomers of amino acids was
achieved. The transport experiments of amino acids in their zwitterionic forms or as potassium and sodium salts by chiral azacrown
ether derivatives were carried out with modest selectivity. 128
The ability of hemicucurbit[6]uril and hemicucurbit[12]uril ( Fig. 4) to act as carriers through the chloroform liquid membrane
for a series of amino acids as ion pairs in the presence of tropaeolin 00 as counterion with a view to their separation was investi-
gated. The results suggest that these receptors show affinities toward amino acids undertaking an active transport from the source
phase to the receiving phase with good yields.65

2.17.4 Concluding Remarks

Some meaningful aspects of macrocyclic receptors studied in the context of analytic applications such as liquid–liquid extraction
and transport through liquid membrane of a large variety of chemical and biochemical compounds as selectively extractants or
carriers have been presented in this contribution. Apart from their ability to form interesting complexes by specific interactions,
they exhibit separation features too. To date, the involvement of these receptors as synthetic ionophores in transmembrane ion
transport is of great interest in supramolecular chemistry and some of the related aspects have also briefly been reviewed.
The development of a large number of synthetic receptors having the ability to bind specifically targeted compounds has opened
new and useful applications in biochemistry, pharmaceuticals, and medicine, as well as chemistry. Special interest in macrocyclic
receptors stems from their ability to act as chiral receptors in chiral separation of various chemical and biological compounds. Some
aspects of enantioseparation based on chiral receptors have also been summarized.

References

1. Atwood, J. L., Davies, J. E. D., MacNicol, D. D., Voögtle, F., Eds. Comprehensive Supramolecular Chemistry, Vols. 1–11; Elsevier: Oxford, 1996.
2. Smith, B. D., Ed. Synthetic Receptors for Biomolecules, Design Principles and Applications; The Royal Society of Chemistry: Cambridge, 2015.
3. Lehn, J.-M. Supramolecular Chemistry, Concepts and Perspectives; VCH: Weinheim, 1995.
4. Busschaert, N.; Caltagirone, C.; Van Rossom, W.; Gale, P. A. Chem. Rev. 2015, 115 (15), 8038–8155.
5. Gokel, G. W.; Daschbach, M. M. Coord. Chem. Rev. 2008, 252 (8–9), 886–902.
6. You, L.; Zha, D.; Anslyn, E. V. Chem. Rev. 2015, 115 (15), 7840–7892.
7. Schneider, H.-J., Ed. Supramolecular Systems in Biomedical Fields; Royal Society of Chemistry: Cambridge, 2013.
8. Lee, T.-C.; Kalenius, E.; Lazar, A. I.; Assaf, K. I.; Kuhnert, N.; Grün, C. H.; Jänis, J.; Scherman, O. A.; Nau, W. M. Nat. Chem. 2013, 5, 376–382.
9. Lumetta, G. L., Rogers, R. D., Gopalan, A. S., Eds.. Calixarenes for Separations; ACS Symposium Series, Vol. 757; American Chemical Society: Washington, DC, 2000.
378 Extraction and Transport

10. Wenzel, M.; Hiscock, J.; Gale, P. Chem. Soc. Rev. 2012, 41, 480–520.
11. Kim, D. S.; Sessler, J. L. Chem. Soc. Rev. 2015, 44 (2), 532–546.
12. Hayashita, T.; Takagi, M. Molecular recognition: receptors for cationic guests. In Comprehensive Supramolecular Chemistry, Vol. 1, Gokel, G. W., Ed.; Oxford: Elsevier, 1996;
pp 635–669.
13. Gutsche, C. D. Calixarenes Revisited; The Royal Society of Chemistry: Cambridge, 1998.
14. Mutihac, L.; Lee, H. J.; Kim, J. S.; Vicens, J. Chem. Soc. Rev. 2011, 40 (5), 2777–2796.
15. Schalley, C., Ed. Analytical Methods in Supramolecular Chemistry; Wiley-VCH: Weinheim, 2007.
16. Gokel, G. W.; Negin, S. Acc. Chem. Res. 2013, 46 (12), 2824–2833.
17. Matile, S.; Jentzsch, A. V.; Montenegro, J.; Fin, A. Chem. Soc. Rev. 2011, 40 (5), 2453–2474.
18. Fyles, T. M.; van Straaten-Nijenhuis, W. F. Ion channel models. In Comprehensive Supramolecular Chemistry, Vol. 10, Reinhoudt, D. N., Ed.; Oxford: Elsevier, 1996;
pp 53–77.
19. Chui, J. K. W.; Fyles, T. M. Chem. Soc. Rev. 2012, 41 (1), 148–175.
20. Gale, P. A. Acc. Chem. Res. 2011, 44 (3), 216–226.
21. McNally, B. A.; Leevy, W. M.; Smith, B. D. Supramol. Chem. 2007, 19 (1–2), 29–37.
22. Schneider, H.-J. Angew. Chem. Int. Ed. 2009, 48 (22), 3924–3977.
23. Schneider, H.-J. Int. J. Mol. Sci. 2015, 16 (4), 6694–6717.
24. Custelcean, R.; Delmau, L. H.; Moyer, B. A.; Sessler, J. L.; Cho, W.-S.; Gross, D.; Bates, G. W.; Brooks, S. J.; Light, M. E.; Gale, P. A. Angew. Chem. Int. Ed. 2005, 44 (17),
2537–2542.
25. Kim, S. K.; Lynch, V. M.; Young, N. J.; Hay, B. P.; Lee, C.-H.; Kim, J. S.; Moyer, B. A.; Sessler, J. L. J. Am. Chem. Soc. 2012, 134 (51), 20837–20843.
26. Vicens, J., Böhmer, V., Eds. Calixarenes: A Versatile Class of Macrocyclic Compounds; Kluwer Academic Publishers: Dordrecht, 1991.
27. Baldini, L.; Sansone, F.; Casnati, A.; Ungaro, R. In Supramolecular Chemistry: From Molecules to Nanomaterials; Steed, J. W., Gale, P. A., Eds.; Wiley: New York, NY, 2012;
pp 863–878.
28. de Mendoza, J.; Cuevas, F.; Prados, P.; Meadows, E. S.; Gokel, G. W. Angew. Chem. Int. Ed. 1998, 37 (11), 1534–1537.
29. Atwood, J. L.; Barbour, L. J.; Jerga, A.; Brandi, L.; Schottel, B. L. Science 2002, 298 (5595), 1000–1002.
30. Thallapally, P. K.; Lloyd, G. O.; Atwood, J. L.; Barbour, L. J. Angew. Chem. Int. Ed. 2005, 44 (25), 3848–3851.
31. Dobrzanska, L.; Lloyd, G. O.; Raubenheimer, H. G.; Barbour, L. J. J. Am. Chem. Soc. 2006, 128 (3), 698–699.
32. Dobrzanska, L.; Lloyd, G. O.; Esterhuysen, C.; Barbour, L. J. Angew. Chem. Int. Ed. 2006, 45 (35), 5856–5859.
33. Oueslati, I. Tetrahedron 2007, 63 (44), 10840–10851.
34. Vicens, J. J. Incl. Phenom. Macrocycl. Chem. 2006, 55 (1–2), 193–196.
35. Rekharsky, M. V.; Inoue, Y. Chem. Rev. 1998, 98 (5), 1875–1918.
36. Nepogodiev, S. A.; Stoddart, J. F. Chem. Rev. 1998, 98 (5), 1959–1976.
37. Stoddart, J. F. Angew. Chem. Int. Ed. Engl. 1992, 31 (7), 846–848.
38. Másson, M. J.; Karlsson, F.; Valdimarsdóttir, M.; Magnúsdóttir, K.; Loftsson, T. J. Incl. Phenom. Macrocycl. Chem. 2007, 57 (1–4), 481–487.
39. Schneiderman, E.; Stalcup, A. M. J. Chromatogr. B 2000, 745 (1), 83–102.
40. Chui, J. K. W.; Fyles, T. M. Org. Biomol. Chem. 2014, 12 (22), 3622–3634.
41. Stancu, A. D.; Hillebrand, M.; Tablet, C.; Mutihac, L. J. Incl. Phenom. Macrocycl. Chem. 2014, 78 (1), 71–76.
42. Gokel, G. W., Ed. Comprehensive Supramolecular Chemistry, Vol. 1; Elsevier: Oxford, 1996.
43. Zolotov, Y. A., Ed. Macrocyclic Compounds in Analytical Chemistry; Wiley: New York, NY, 1997.
44. Mutihac, L.; Buschmann, H.-J.; Jansen, K.; Wego, A. Mater. Sci. Eng. C 2001, 18 (1–2), 259–264.
45. Assaf, K. I.; Nau, W. M. Chem. Soc. Rev. 2015, 44 (2), 394–418.
46. Lee, J. W.; Samal, S.; Selvapapm, N.; Kim, H.-J.; Kim, K. Acc. Chem. Res. 2003, 36 (8), 621–630.
47. Lagona, J.; Mukhopadhyay, P.; Chakrabarti, S.; Isaacs, L. Angew. Chem. Int. Ed. 2005, 44 (31), 4844–4870.
48. Buschmann, H.-J.; Cleve, E.; Mutihac, L.; Schollmeyer, E. Microchem. J. 2000, 64 (1), 99–103.
49. Buschmann, H.-J.; Mutihac, L.; Schollmeyer, E. J. Incl. Phenom. Macrocycl. Chem. 2006, 56 (3), 363–368.
50. Miyahara, Y.; Goto, K.; Oka, M.; Inazu, T. Angew. Chem. Int. Ed. 2004, 43 (38), 5019–5022.
51. Stephan, H.; Juran, S.; Antonioli, B.; Gloe, K.; Gloe, K. In Analytical Methods in Supramolecular Chemistry; Schalley, C. A., Ed.; Wiley-VCH: Weinheim, 2007; pp 391–418
(Chapter 4).
52. Moldoveanu, S.; David, V. Modern Sample Preparation for Chromatography; Elsevier: Amsterdam, 2015; pp 131–162.
53. Enache, V. I.; Mutihac, L.; Othman, A. B..; Vicens, J. J. Incl. Phenom. Macrocycl. Chem. 2011, 71 (3–4), 537–543.
54. Kim, J. S.; Quang, D. T. Chem. Rev. 2007, 107 (9), 3780–3799.
55. Sieffert, N.; Chaumont, A.; Wipff, G. J. Phys. Chem. 2009, 113 (24), 10610–10622.
56. Sviben, I.; Galic, N.; Tomisic, V.; Frkanec, L. New J. Chem. 2015, 39 (8), 6099–6107.
57. Karakus, Ö.Ö.; Delogoz, H. Supramol. Chem. 2015, 27 (1–2), 110–122.
58. Akkuş, G. U.; Al, E.; Korcan, S. E. Supramol. Chem. 2015, 27 (7–8), 522–526.
59. Hamdi, A.; Souane, R.; Kim, L.; Abidi, R.; Mutihac, L.; Vicens, J. J. Incl. Phenom. Macrocycl. Chem. 2009, 64 (1), 95–100.
60. Mutihac, L.; Varduca Enache, I.; Diacu, E.; Mutihac, R.-C.; Vicens, J. Supramol. Chem. 2014, 26 (7–8), 521–525.
61. Shimojo, K.; Goto, M. Sep. Purif. Technol. 2005, 44 (2), 175–180.
62. Tang, K.; Song, L.; Liu, Y.; Miao, J. Chem. Eng. J. 2012, 180, 293–298.
63. Diacu, E.; Mutihac, L.; Ruse, E.; Ceausescu, M. M. J. Incl. Phenom. Macrocycl. Chem. 2011, 71 (3–4), 339–342.
64. Wintergerst, M. P.; Levitskaia, T. G.; Moyer, B. A.; Sessler, J. L.; Delmau, L. H. J. Am. Chem. Soc. 2008, 130 (12), 4129–4139.
65. Cucolea, I. E.; Buschmann, H.-J.; Mutihac, L. Supramol. Chem. 2015. http://dx.doi.org/10.1080/10610278.2015.1121267.
66. Sisson, A. L.; Raza Shah, M.; Bhosale, S.; Matile, S. Chem. Soc. Rev. 2006, 35 (12), 1269–1286.
67. Gokel, G. W.; Carasel, I. A. Chem. Soc. Rev. 2007, 36 (2), 378–389.
68. Sakai, N.; Mareda, J.; Matile, S. Acc. Chem. Res. 2005, 38 (2), 79–87.
69. Tabushi, I.; Kuroda, Y.; Yokota, K. Tetrahedron Lett. 1982, 23 (44), 4601–4604.
70. Sidorov, V.; Kotch, F. W.; Abdrakhmanova, G.; Mizani, R.; Fettinger, J. C.; Davis, J. T. J. Am. Chem. Soc. 2002, 124 (10), 2267–2278.
71. Pregel, M. J.; Jullien, L.; Canceille, J.; Lacombe, L.; Lehn, J.-M. J. Chem. Soc., Perkin Trans. 2 1995, (3), 417–426.
72. Murillo, O.; Watanabe, S.; Nakano, A.; Gokel, G. W. J. Am. Chem. Soc. 1995, 117 (6), 7665–7679.
73. Voyer, N.; Potvin, L.; Rousseau, E. J. Chem. Soc., Perkin Trans. 2 1997, (8), 1469–1471.
74. Cazacu, A.; Tong, C.; Van der Lee, A.; Fyles, T. M.; Barboiu, M. J. Am. Chem. Soc. 2006, 128 (29), 9541–9548.
75. Yoshino, N.; Satake, A.; Kokuke, Y. Angew. Chem. Int. Ed. 2001, 40 (2), 457–459.
76. Jeon, Y. J.; Kim, H.; Jon, S.; Selvapalam, N.; Hyun Oh, D.; Seo, I.; Park, C.-S.; Jung, S. R.; Koh, D.-S.; Kim, K. J. Am. Chem. Soc. 2004, 124 (47), 15944–15945.
77. Ghadiri, M. R.; Granja, J. R.; Buehler, L. K. Nature 1994, 369 (6478), 301–304.
Extraction and Transport 379

78. Matile, S. Chem. Soc. Rev. 2001, 30 (3), 158–167.


79. Fyles, T. M.; Hu, C.; Knoy, R. Org. Lett. 2001, 3 (9), 1335–1337.
80. Moszynski, J. M.; Fyles, T. M. J. Am. Chem. Soc. 2012, 134 (38), 15937–15945.
81. Hübner, C. A.; Jentsch, T. J. Hum. Mol. Genet. 2002, 11 (20), 2435–2445.
82. Welsh, M. J.; Smith, A. E. Cell 1993, 73 (7), 1251–1254.
83. MacKinnon, R. Angew. Chem. Int. Ed. 2004, 43 (33), 4265–4277.
84. Doyle, D. A.; Cabral, J. M.; Pfuetzner, R. A.; Kuo, A.; Gulbis, J. M.; Cohen, S. L.; Chait, B. T.; MacKinnon, R. Science 1998, 280 (5360), 69–77.
85. Agre, P.; King, L. S.; Yasui, M.; Guggino, W. B.; Ottersen, O. P.; Fujiyoshi, Y.; Engel, A.; Nielsen, S. J. Physiol. 2002, 542 (1), 3–16.
86. Agre, P. Angew. Chem. Int. Ed. 2004, 43 (33), 4278–4290.
87. Barboiu, M. Angew. Chem. Int. Ed. 2012, 51 (47), 11674–11676.
88. Dawson, R. E.; Hennig, A.; Weimann, D. F.; Emery, D.; Ravikumar, V.; Montenegro, J.; Takeuchi, T.; Gabutti, S.; Mayor, M.; Mareda, J.; Schalley, C. A.; Matile, S. Nat. Chem.
2010, 2 (7), 533–538.
89. Cragg, P. J.; Iqbal, K. S. J. Dalton Trans. 2007, (1), 26–32.
90. Madhavan, N.; Robert, E. C.; Gin, M. S. Angew. Chem. Int. Ed. 2005, 44 (46), 7584–7587.
91. Madhavan, N.; Gin, M. S. ChemBioChem 2007, 8 (15), 1834–1840.
92. Boccalon, M.; Iengo, E.; Tecilla, P. J. Am. Chem. Soc. 2012, 134 (50), 20310–20313.
93. Izzo, I.; Licen, S.; Maulucci, N.; Autore, G.; Marzocco, S.; Tecilla, P.; De Riccardis, F. Chem. Commun. 2008, 26, 2986–2988.
94. Seganish, J. L.; Santacroce, P. V.; Salimian, K. J.; Fettinger, J. C.; Zavalij, P.; Davis, J. T. Angew. Chem. Int. Ed. 2006, 45 (20), 3334–3338.
95. Okunola, O. A.; Seganish, J. L.; Salimian, K. J.; Zavalij, P. Y.; Davis, J. T. Tetrahedron 2007, 63 (44), 10743–10750.
96. Pajewski, R.; Ferdani, R.; Pajewska, J.; Li, R.; Gokel, G. W. J. Am. Chem. Soc. 2005, 127 (51), 18281–18295.
97. Cook, G. A.; Pajewski, R.; Aburi, M.; Smith, P. E.; Prakash, O.; Tomich, J. M.; Gokel, G. W. J. Am. Chem. Soc. 2006, 128 (5), 1633–1638.
98. Yamnitz, C. R.; Negin, S.; Carasel, I. A.; Winter, R. K.; Gokel, G. W. Chem. Commun. 2010, 46 (16), 2838–2840.
99. Chui, J. K. W.; Fyles, T. M.; Luong, H. Beilstein J. Org. Chem. 2011, 7, 1562–1569.
100. Gokel, G. W.; Mukhopadhyay, A. Chem. Soc. Rev. 2001, 30 (5), 274–286.
101. Arduini, A.; Ghidini, E.; Pochini, A.; Ungaro, R.; Andreetti, D. G.; Calestani, G.; Ugozzoli, F. J. Incl. Phenom. Macrocycl. Chem. 1988, 6 (2), 119–134.
102. Gokel, G. W. Chem. Commun. 2000, 1, 1–9.
103. Weber, M. E.; Elliot, E. K.; Gokel, G. W. Org. Biomol. Chem. 2006, 4 (1), 83–89.
104. Maulucci, N.; De Riccardis, F.; Botta, C. B.; Casapullo, A.; Cressina, E.; Fregonese, M.; Tecilla, P.; Izzo, I. Chem. Commun. 2005, 10, 1354–1356.
105. Gilles, A.; Barboiu, M. J. Am. Chem. Soc. 2016, 138 (1), 426–432.
106. Wright, A. J.; Mattews, S. E.; Fischer, W. B.; Beer, P. D. Chem. Eur. J. 2001, 7 (16), 3474–3481.
107. Jeon, Y. J.; Kim, H.; Jon, S.; Selvapalam, N.; Oh, D. H.; Seo, I.; Park, C.-S.; Jung, S. R.; Koh, D.-S.; Kim, K. J. Am. Chem. Soc. 2004, 126 (49), 15944–15945.
108. Sidorov, V.; Kotch, F. W.; El-Kouedi, M.; Davis, J. T. Chem. Commun. 2000, 23, 2369–2370.
109. Matthews, S. E.; Schmitt, P.; Felix, V.; Drew, M. G. B; Beer, P. D. J. Am. Chem. Soc. 2002, 124 (7), 1341–1353.
110. Sansone, F.; Baldini, L.; Casnati, A.; Chierici, E.; Faimani, G.; Ugozzolli, F.; Ungaro, R. J. Am. Chem. Soc. 2004, 126 (20), 6204–6205.
111. Pulpoka, B.; Baklouti, L.; Kim, J. S.; Vicens, J. In Calixarenes in the Nanoworld; Vicens, J., Harrowfield, J., Eds.; Springer: Dordrecht, 2007; pp 135–149 (Chapter 7).
112. Gale, P. A.; Tong, C. C.; Haynes, C. J. E.; Adeosun, O.; Gross, D. E.; Karnas, E.; Sedenberg, E. M.; Quesada, R.; Sessler, J. L. J. Am. Chem. Soc. 2010, 132 (10),
3240–3241.
113. Yano, M.; Tong, C. C.; Light, M. E.; Schmidtchen, F. P.; Gale, P. A. Org. Biomol. Chem. 2010, 8 (9), 4356–4363.
114. Griffiths, K.; Sharma, K.; Marcos, P. M.; Ascenso, J. R.; Nind, J.; Cottet, K.; Cragg, P. J. Supramol. Chem. 2015, 27 (3), 167–173.
115. Jin, T.; Kinjo, M.; Kobayashi, Y.; Hirata, H. J. Chem. Soc., Faraday Trans. 1998, 94 (20), 3135–3140.
116. Jin, T. Chem. Commun. 1999, 20, 2129–2130.
117. Mutihac, L.; Buschmann, H.-J.; Mutihac, R.-C.; Schollmeyer, E. J. Incl. Phenom. Macrocycl. Chem. 2005, 51 (1–2), 1–10.
118. Izatt, S. R.; Hawkins, R. T.; Christensen, J. J.; Izatt, R. M. J. Am. Chem. Soc. 1985, 107 (1), 63–66.
119. Ludwig, R. Fresenius J. Anal. Chem. 2000, 367 (2), 103–128.
120. Chang, S.-K.; Son, H.-J.; Hwang, H.-S. Bull. Korean Chem. Soc. 1990, 11 (5), 364–365.
121. Okada, Y.; Kasai, Y.; Nishimura, J. Tetrahedron Lett. 1995, 36 (4), 555–558.
122. Mutihac, L.; Buschmann, H.-J.; Diacu, E. Desalination 2002, 148 (1–3), 253–256.
123. Mutihac, L.; Mutihac, R. J. Incl. Phenom. Macrocycl. Chem. 2007, 59 (1–2), 177–181.
124. Kim, L.; Hamdi, A.; Stancu, A. D.; Souane, R.; Mutihac, L.; Vicens, J. J. Incl. Phenom. Macrocycl. Chem. 2010, 66 (1), 55–59.
125. Oshima, T.; Inoue, K.; Furusaki, S.; Goto, M. J. Membr. Sci. 2003, 217 (1–2), 87–97.
126. Kim, L.; Stancu, A. D.; Diacu, E.; Buschmann, H.-J.; Mutihac, L. Supramol. Chem. 2009, 21 (1–2), 131–134.
127. Park, I.-W.; Yoo, J.; Kim, B.; Adhikari, S.; Kim, S. K.; Yeon, Y.; Haynes, C. J. E; Sutton, J. L.; Tong, C. C.; Lynch, V. M.; Sessler, J. L.; Gale, P. A.; Lee, C.-H. Chem. Eur. J.
2012, 18 (9), 2514–2523.
128. Kazaylek, M.; Kocakaya, S. O.; Oral, E. V.; Colak, M.; Karakaplan, M. Supramol. Chem. 2014, 26 (5–6), 363–372.
2.18 Cooperativity
T Nabeshima and T Nakamura, University of Tsukuba, Tsukuba, Japan
Ó 2017 Elsevier Ltd. All rights reserved.

2.18.1 Introduction 381


2.18.1.1 Concept: Importance of Cooperativity 381
2.18.1.2 Models of Cooperativity 382
2.18.1.3 Experimental Examples of Homotropic Allosteric Cooperativity 384
2.18.2 Macrocyclic and Macropolycyclic Receptors Exhibiting Allosteric Cooperativity 388
2.18.2.1 Introduction 388
2.18.2.2 Macrocyclic Receptors 388
2.18.2.3 Macrobicyclic and Macropolycyclic Receptors 390
2.18.3 Cooperative Supramolecular Systems 392
2.18.3.1 Introduction 392
2.18.3.2 Templated Synthesis and Cooperativity 393
2.18.3.2.1 Ion as a preorganization effector 393
2.18.3.2.2 Cooperativity in Multivalent Receptors 394
2.18.4 Supramolecular Cooperative Catalytic Systems 398
2.18.5 Summary and Outlook 402
References 402

Nomenclature
EM (Reference) effective molarity Kai Apparent stepwise binding constant for ith binding
EMi Stepwise effective molarity for ith binding Ki Stepwise binding constant for ith binding
K (Reference) binding constant a Cooperative parameter

2.18.1 Introduction
2.18.1.1 Concept: Importance of Cooperativity
Combination and organization of many simple small units at the molecular level are important and inevitable to create
complicated structures and sophisticated cooperative molecular functions. However, if there is no interaction with each unit, the
properties of the large molecule formed by the connection of the small units would be simple accumulation of the properties of
the initial small individual units (molecular weight is such an example of simple summation). On the other hand, if the small
starting materials as a building unit interact with each other in the accumulated structure, the properties and functions should
be cooperative, that is, different from those expected from just the summation of the individual units. In addition, external stimuli
are often utilized to modulate the functions and properties of molecules and molecular assemblies. One of the most efficient and
useful strategies to construct highly functional molecules and supramolecules is to change the structure coupled with a certain
function by using through-bond and through-space intramolecular interactions as well as intermolecular noncovalent interactions.
Cooperativity is a key concept not only in biological systems1 but also in conventional molecular and supramolecular systems,2
because intermolecular and intramolecular communications result in significant increase or decrease of molecular functions. One of
the most famous cooperative events in biology is allostery in hemoglobin oxygen binding.3 As oxygen molecules are bound to the
hemoglobin up to the O2 saturation, the O2 binding affinity increases. Structural change of the subunits upon the O2 binding
modulates the intersubunit interactions to result in this allosteric regulation.
The term “allostery” comes from Greek words, allos (other in English) and stereos (solid, hard). When an effector like O2 in the
case of hemoglobin binds to the allosteric site of the molecular system, structural change occurs to cause change in the activity of the
functional sites. A simple allosteric mechanism for molecular recognition is illustrated in Fig. 1. In general, the allosteric functional
molecules have a binding site for an effector and a controllable active site that is apart from the effector-binding site. The allosteric
phenomenon is a kind of cooperative event. The effector binding causes the conformational change of the receptor and then induces
the structural change of the binding site.
Studies on the cooperative phenomena are closely related to the science of molecular information. The cooperative events are
regulated by receiving and releasing information at the molecular level. The information is transported by a certain substance such
as a molecule, an ion, an electron, a photon, etc. In addition, the properties of the substances such as combination, order, or chirality
can govern the events.

Comprehensive Supramolecular Chemistry II, Volume 2 http://dx.doi.org/10.1016/B978-0-12-409547-2.13800-4 381


382 Cooperativity

Figure 1 Allosteric effect in molecular recognition.

In the allosteric systems, the change in the structural and chemical/physical properties (acidity and basicity, reactivity, binding
affinity, redox potential, photophysical properties, magnetic properties) of the molecular systems occurs to eventually achieve
the change in their functions. Noncovalent interactions such as classical and nonclassical hydrogen bonding, pi stacking, the
hydrophobic effect, coordination bonding, cation–pi interaction, anion–pi interaction, halogen bonding, charge transfer interac-
tion, CH–pi interaction, etc. play a key role in controlling these structural changes. It should be stressed here again that functions
based on cooperativity are not arose from a simple sum of individual functions where all the constitutional functional units are
independent, that is, the molecular information of one unit does not move to the other.
As seen in the hemoglobin oxygen binding, the cooperative regulation plays a very critical role to maintain life. Inspired by such
fascinating biological systems, many chemists have designed and synthesized a variety of host molecules and assembled supramo-
lecular systems with cooperative functions.
There is a similar term “synergy” to “cooperativity.”4 Actually, the concept of “cooperativity” includes “synergy.” When the
function is enhanced, the positive cooperativity is called “synergistic.”5 On the other hand, a negative cooperative effect is named
“interfering.” When there is no effect, it is “additive (noncooperative).” (See “Models of Cooperativity” section for the definition of
positive/negative cooperativity).
The events with high cooperativity at the molecular level may be considered to be one of the simplest examples of the “Complex
System,” which has recently attracted considerable attention from a wide range of scientific fields, that is, mathematics, chemistry,
physics, materials science, biology, meteorology, and even social sciences. An accepted definition of “Complex System” is “At first,
a complex system consists of many components that interact with each other.”6 According to M. Mitchell, the definition of the
complex system is “a system in which large networks of components with no central control and simple rules of operation give
rise to complex collective behavior, sophisticated information processing, and adaptation via learning or evolution.”6
The complex system is often said to exist between chaos and order. However, chemists can understand this confusing and
eccentric statement, when they consider it in terms of state function changes. That is, chaos and order are directly related to large
positive entropic change and to large negative enthalpic change, respectively. Hence, these metastable states of complex systems
such as alive systems exist between the two extreme states and are usually controlled by external stimuli to maintain homeostasis.
In the complex systems, catastrophic changes in structure and functions may take place and sometimes result in biological
evolution, if a severe stimulus is given to the system. Similarly, all-or-none regulation of molecular functions in artificial supramo-
lecular systems is one of the consequences in the cooperative molecular events.

2.18.1.2 Models of Cooperativity


Although there have been many studies on cooperativity in supramolecular chemistry, the definition and quantification of
cooperativity are still the subject of discussion.7–9 Hunter and Anderson presented a tutorial account where two types of coopera-
tivity, that is, allosteric and chelate cooperativity, are introduced.7 Ercolani and Schiaffino proposed to categorize and generalize
cooperativity in three classes; allosteric, chelate, and interannular cooperativity.8 Although their detailed discussion should be
referred to the original papers, simple definition of each cooperativity together with the related terms is discussed in this article
(vide infra).
Firstly, the allosteric cooperativity is a phenomenon caused by the interaction of two or more binding sites. As mentioned in
“Concept: Importance of Cooperativity” section, conformational change of the allosteric functional molecules is important.
When molecular recognition is the function of the system, binding properties of the host are changed upon complexation of the
first guest, effector. Noteworthy is that the target guest itself interacts with the host independently in a different site, although
the binding affinity to the guest is indirectly but efficiently affected by the effector. If the binding constant with the guest is enhanced
upon the effector binding, this event is called positive allostery. In the case of the decrease in the affinity, allostery is negative. When
the guest and the effectors are same and different, we call the allostery homotropic and heterotropic, respectively. Thus, the system
in Fig. 1 is a case of the heterotropic allostery.
In order to understand the cooperativity of guest binding in terms of thermodynamics, the reference system for 1:1 complexation
is described at first and then a homotropic cooperative system bearing two equivalent binding sites for guests is discussed as the
simplest cooperative case. For the 1:1 reference system composed of a one-site host H and a guest G (Fig. 2), the binding constant
K is simply defined by Eq. (1).
Cooperativity 383

Figure 2 1:1 reference system.

Figure 3 A homotropic allosteric system with two guest-binding sites.

½H $ G
K¼ (1)
½H½G
For a homotropic allosteric system with two guest-binding sites (Fig. 3), two microscopic binding constants K1 and K2 are
considered to evaluate the cooperativity by comparing the reference value K with them. When the statistical factors in the equilib-
rium are taken into account, Eqs. (2), (3) indicate the relation of the binding constants to the concentrations of the chemical species
in equilibrium.
½H$G
2K1 ¼ (2)
½H½G

1 ½H$G2 
K2 ¼ (3)
2 ½H$G½G
The stoichiometric stepwise binding constants Ka1 (step 1, formation of 1:1 complex form free host) and Ka2 (step 2, formation
of 1:2 complex from 1:1) would be experimentally determined instead of microscopic constants K1 and K2. The relationship
between these apparent binding constants and K1, K2 is shown here (Eqs. 4 and 5).
Ka1 ¼ 2 K1 (4)

Ka2 ¼ 1=2 K2 (5)


In the allosteric systems, cooperative parameter, a, reflects the degree of cooperativity (a > 1, positively cooperative; a ¼ 1,
noncooperative; a < 1, negatively cooperative). For the system in Fig. 3, the a is written as the ratio of the two microscopic binding
constants (Eq. 6)
K2
a¼ (6)
K1
In the case of positive allosteric cooperativity (a > 1, K2 > K1), the second binding is stronger than the first one. In other words,
the first guest enhances the affinity of the other binding site in the 1:1 host–guest complex H$G. Meanwhile, in the case of negative
cooperativity (a < 1, K2 < K1), the second binding is weaker than the first one, and the 1:1 host–guest complex H$G is a favored
species over the fully occupied 1:2 complex H$G2. As for the case of noncooperativity (a ¼ 1), the microscopic binding constants
are the same as that of single site reference host, thus K1 ¼ K2 ¼ K. Alternatively, using the stoichiometric stepwise binding constants,
Ka2/Ka1 > 0.25 and < 0.25 mean positive and negative cooperativity due to the statistical factor, respectively, and Ka2 ¼ 0.25 $ Ka1
indicates noncooperativity.
When an allosteric host bearing n identical independent sites (n > 2) noncooperatively interacts with monodentate ligands, the
statistical factor leads to the relationship between Ka1 and Kan, that is, Ka1/Kan ¼ n2. More detailed discussion on statistical factors has
been reported elsewhere.7,8,10,11
In contrast to allosteric cooperativity, two different cooperativity types, chelate cooperativity and interannular cooperativity for
the chelating molecular recognition systems, operate in a closing molecular assembling way (Figs. 4 and 5). These cooperative
phenomena arise, even when the first binding does not affect intrinsic binding affinity to the second one, that is, even in the absence
of allosteric cooperativity (a ¼ 1). The theoretical details on the chelate and interannular cooperativity are not discussed here, but
briefly demonstrated later.
384 Cooperativity

Figure 4 Chelate cooperativity.

Figure 5 Interannular cooperativity.

The chelate cooperativity is applied to the closed systems in which complexation of a host (H) with a guest bearing a linked
binding moieties (G) gives an annulated product. Even in the case of a host having two binding sites, many possible products other
than the annulated product are expected. To avoid this complexity, it is assumed that the complexation occurs under the conditions
where a large excess amount of the guest exists compared to the host. Then, we can consider only four states for the host to
quantitatively examine the equilibrium, a guest-free host (H), a 1:1 complex after the first intermolecular binding (open-H$G),
an annulated product after the subsequent intramolecular cyclization (closed-H$G), and a 1:2 complex (H$G2) (Fig. 4). Even in
the case of K1 ¼ K2 ¼ K (i.e., a ¼ 1, no allosteric cooperativity), the chelate cooperativity is expected in the presence of the equilibrium
between open-H$G and closed-H$G. A strong chelate cooperativity is observed in the preference of the intramolecularly closed
product (closed-H$G) to the intermolecular linear product (H$G2). This cooperative behavior is related to the effective molarity
(EM), which is defined as Eq. (7).
1 ½closed  H$G
K$EM ¼ (7)
2 ½open  H$G
The interannular cooperativity is also applied to a closed system similar to chelate cooperativity, but in the interannular case two
or more intramolecular binding interactions operate on the system. As the simplest example, we discuss the complexation of an
excess amount of a divalent guest (G) with a tetravalent host (H) in which two bidentate ligating moieties are connected to
each other by a freely rotatable linker (Fig. 5). The EM is expressed in an analogous way as in Fig. 4 for the formation of annulated
products. However, the effective molarities EM are different for the first guest binding and the following second one, because the
restriction of rotation after the first binding makes the second binding easier (i.e., EM1 < EM2). This difference in effective molarities
from the reference value constitutes interannular cooperativity.
As mentioned, the theoretical details on these cooperativity are not discussed here and should be referred to the original papers
and reviews.7–11 Instead, in this article we survey a variety of cooperative molecular and supramolecular systems to understand the
concept of the cooperativity and the strategies of how to design and synthesize the highly cooperative systems.

2.18.1.3 Experimental Examples of Homotropic Allosteric Cooperativity


Before introducing a series of cooperative supramolecules, a few examples of the experimental determination of the allosteric param-
eter a are explained. Here, we deal with the case of homotropic allosteric hosts bearing two identical binding sites (Fig. 3). To
determine the allostericity, the stepwise binding constants K1 and K2 (or Ka1 and Ka2) are needed, which can be calculated from
Cooperativity 385

Figure 6 A macrocyclic receptor 1 with two calix[4]arene moieties that shows homotropic negative allosteric cooperativity toward alkali metal ions Mþ.18

concentrations of the individual components, [H], [G], [H$G], and [H$G2] (Eqs. 2–5). As in the case of the single binding constant K
for 1:1 complexation, various experimental methods are available to determine the stepwise binding constants12: UV–vis absorption
and fluorescence,13 NMR,14 ITC,15 electrochemistry,16 extraction and transportation experiments,17 and so on. Basically, any experi-
mental method that measures physical properties related to molarity is applicable to appropriate binding models.
Two examples of NMR titration experiments for the determination of stepwise binding constants are presented here. A macro-
cyclic receptor 1 bearing two calix[4]arene moieties shown in Fig. 6 works as a two-site receptor for alkali metal ions.18 The four
ethereal oxygen atoms at the calixarene rim together with the pendant ester groups contribute to the chelating complexation of
guests. The NMR titration was performed where portions of solution containing an alkali metal salt were incrementally added
to host 1. Fig. 7 shows the 1H NMR spectra of the titration experiment with NaPF6. Three sets of signals are clearly discerned during
the course of titration, that is, signals assigned to 1, 1$Na, and 1$Na2, respectively. When the in-and-out exchanges of guests are slow
on the NMR timescale, the signals of the guest-free host, 1:1 complex, and 1:2 complexes are observed separately. Thus, the two

Figure 7 1H NMR spectra of the titration experiment with NaPF6 against host 1 (300 MHz, toluene-d8/CD3CN ¼ 6:5, [1]0 ¼ 5.0  10 4 M). The
signals of 1, 1$Na, and 1$Na2 are indicated by filled squares, filled circles, and open circles, respectively. Modified from Nabeshima, T.; Saiki, T.; Sumi-
tomo, K.; Akine, S. Tetrahedron Lett. 2004, 45, 6761–6763. Copyright Elsevier.
386 Cooperativity

stepwise binding constants are determined from the integral ratios of each proton. The following relationships exist for the
concentrations of each species, the total concentration of the host, [1]0, and that of the guest [Na]0 (Eqs. 8–10).

10 ¼ ½1 þ ½1$Na þ ½1$Na2  (8)

½Na0 ¼ ½Na þ ½1$Na þ 2½1$Na2  (9)

½1 : ½1$Na : ½1$Na2  ¼ a : b : c (10)


Here, [1]0 and [Na]0 are given from the experimental settings, and the ratios a:b:c are determined from the integral ratio of the
corresponding NMR signals. The number of unknowns is 4 ([1], [Na], [1$Na], and [1$Na2]), and the number of simultaneous
equations is also 4 (the ratio in Eq. 10 is equivalent to two independent equations). Thus, the values of the unknowns are obtained
algebraically. Minimally, one NMR spectrum at a certain guest concentration is enough to calculate the binding constants. After the
individual concentrations were determined, the binding constants K1 and K2 are calculated from Eqs. (2), (3) described in “Models
of Cooperativity” section.
The values of binding constants and cooperativity parameter a of the host 1 toward Liþ and Naþ are summarized in Table 1.
Note that the apparent binding constants Ka1 and Ka2, which include the statistical factors, are obtained in the titration experiments.
Negative cooperativity is observed for both guests, and cooperativity parameters a are a ¼ 0.15 for Liþ and a < 0.2 for Naþ (Ka1 for
Naþ is too high to be determined by the NMR titration method). The origin of the negative cooperativity is discussed to be the
combination of electrostatic repulsion between the two alkali metal cations in the 1:2 complex 1$M2 and conformational change
in 1 induced by the first binding of ion.
In the situations where the guest exchanges are fast on the NMR timescale, the concentrations of each species are not directly
determined from one NMR spectrum, but titrations are necessary to track the spectral change depending on the host/guest ratios.
A box-shaped supramolecular host 2 shown in Fig. 8 encapsulates one molecule of pentacene, which is a promising organic
semiconducting material but is notorious for its low solubility, utilizing p–p interactions.19 Meanwhile it behaves as a two-site
receptor for smaller acenes (naphthalene to tetracene), which are captured between each pair of parallel terpyridine-PtII moieties.
Fig. 9A shows NMR spectra of a titration experiment where portions of solution of anthracene (G) are incrementally added to the
host 2. As the result of the fast guest exchanges compared to the NMR timescale, only one set of NMR signals assigned to the host 2 is
observed, which are the averages of those of three species, guest-free 2, 2$G, and 2$G2. A chemical shift of an observed signal of 2 at
a certain concentration, d(obs) [ppm], has the relationship with the genuine chemical shift of individual species (d(2), d(2$G), and
d(2$G2)) as shown here (Eqs. 11–13).

Table 1 Binding constants of host 1 in the complexation of Liþ and Naþ (1H NMR, toluene-d8/CD3CN (6:5),
[1] ¼ 5.0  10 4 M)18

Guest Ka1/M 1 Ka2/M 1 a ¼ 4(Ka2/Ka1)


þa
Li 7300  280 270  9 0.15
Naþb > 1  105 5600  270 <0.2
a
Added as ClO4  salt.
b
Added as PF6  salt.

Figure 8 A box-shaped supramolecular receptor 2 with two pairs of terpyridine-PtII moieties that shows homotropic positive allosteric cooperativity
in the recognition of aromatic molecules.19
Cooperativity 387

Figure 9 (A) 1H NMR spectra of the titration experiment of anthracene (G) against host 2 (600 MHz, CD3CN, [2]0 ¼ 5  10 4 M). (B) A least squares
fitting to determine the binding constants Ka1 and Ka2 of the host 2 and anthracene (G) (data of 1H NMR signal i in (A)). Modified from Supporting
Information of Yamaki, Y.; Nakamura, T.; Suzuki, S.; Yamamura, M.; Minoura, M.; Nabeshima, T. Eur. J. Org. Chem. 2016, 2016, 1678–1683.

Table 2 Binding constants and cooperative parameters of host 2 in the recognition of naphthalene and
anthracene (1H NMR, CD3CN)19

Guest log(Ka1)/log(M 1) log(Ka2)/log(M 1) a ¼ 4(Ka2/Ka1)

Naphthalene 2.5(2) 3.3(2) 25


Anthracene 2.5(2) 3.7(2) 60

½2 ½2$G ½2$G2 


dðobsÞ ¼ dð2Þþ dð2$GÞþ dð2$G2 Þ (11)
½20 ½20 ½20

20 ¼ ½2 þ ½2$G þ ½2$G2  (12)

½G0 ¼ ½G þ ½2$G þ 2½2$G2  (13)

Here, [2]0 and [G]0 are given from the experimental settings, and d(obs) and d(2) are determined from the NMR signals.
However, d(2$G) and d(2$G2) are not always obtained directly from a NMR spectrum since the population of 2$G or 2$G2 does
not reach 100% during the titration experiments. Thus, there are more unknown parameters than the number of equations, and
a least square fitting utilizing many data sets is necessary. In this case, one can set four variables, d(2$G), d(2$G2), Ka1, and Ka2
(see Eqs. 2–5 for the latter two) in the fitting procedure. For the actual calculation, computer programs such as TitrationFit20 or
SPECFIT21 are useful, which are specifically developed for this purpose. Detailed explanation on the analyses using spreadsheet
programs can be found in contributions by Hirose.14,22 Fig. 9B shows the result of the least square fitting to the NMR data of
Fig. 9A. Only the signal of i is used for the fitting in Fig. 9B, but incorporation of the other signals in consideration would add
reliability.
Table 2 summarizes the apparent stepwise binding constants and cooperativity parameter a of the host 2 in the recognition of
naphthalene and anthracene. Strong positive cooperativity was observed for both guests (a ¼ 25 for naphthalene and a ¼ 60 for
anthracene, respectively). The origin of the positive cooperativity is explained as follows: The first bound aromatic guest sets
the two bis(terpyridine-PtII) moieties in a parallel conformation with the distance suitable to bind the second aromatic guest via
p–p interaction.
As seen from the two examples, the determination of stepwise binding constants and cooperative parameters reveals the nature
of the guest-binding events. Conformational changes upon the recognition, distances between the binding sites, intermolecular
interactions between hosts and guests, desolvation upon the guest binding, and many other factors are involved in the cooperative
parameter a. As supramolecular chemistry is the science of multiple components, the study on cooperativity is indispensable for
deep understanding of the system.
388 Cooperativity

2.18.2 Macrocyclic and Macropolycyclic Receptors Exhibiting Allosteric Cooperativity


2.18.2.1 Introduction
In “Experimental Examples of Homotropic Allosteric Cooperativity” section, examples of homotropic receptors with multiple
binding sites for one kind of molecules are presented. Utilization of an effector different from a bound guest greatly expands
the functions of the receptors, since it enables the transformation of molecular information. Many artificial heterotopic receptors
with chemical specificity have been reported, and there have been a lot of excellent books and reviews focused on this
subject.2,4,23–26 In this article, we present several major works in the last few decades.
Macrocyclic molecules represented by crown ethers27 and macropolycyclic ones represented by cryptands28 have been utilized
for the recognition of small molecules for many years.29 These molecules possess defined inner spaces where multivalent interac-
tions are exerted. Here, if a regulating unit that causes the conformational change responding to the effector binding is introduced to
the macrocyclic framework, the receptor can attain allostericity.
Here, macrocyclic and macropolycyclic receptors with allostericity are introduced. Firstly, cooperative macrocyclic receptors that
recognize metal cations (such as the cases with the binding of alkali metals by crown ethers) are introduced. Next, macrocycles that
allosterically recognize small organic molecules are explained. Then, macropolycyclic receptors are introduced.
As an important category of the cooperative hosts, ion-pair receptors that recognize both a cation and an anion coincidently are
widely studied.30,31In those cases, a cation is seen as an effector and an anion as a guest, or vice versa. Strong recognition can be
achieved utilizing favorable electrostatic interaction in the ion-pair, and various receptors with sophisticated cooperative design
have been reported. However, they will be covered in another chapter of this book series, thus are not introduced here.

2.18.2.2 Macrocyclic Receptors


One of the earliest examples of synthetic macrocyclic receptors with allosteric regulation was reported by Rebek, Jr. and coworkers.32
A crown-ether derivative 3 in Fig. 10 is synthesized by linking the 2,20 -bipyridyl (bpy) group to an oligoethylene glycol chain. Here,
the bpy group is the allosteric regulating unit responsible to a receptor. The binding constant K of 3 with sodium ion Naþ was eval-
uated before and after the complexation of a metal effector with the bpy group. The constant K was determined by an extraction
technique between aqueous phase and chloroform phase using alkali metal salts of picrates. Under the condition of 1:1 binding
between 3 and Naþ, the binding constant K is expressed as Eq. (14).32
F
K¼ n    2 (14)
Kd ð1  F Þ ½Naþ  0  F 30 VCHCl3 =VH2 O

Here, F is the fraction of complexed 3, [Naþ]0 and [3]0 are the total concentrations of Naþ and 3, respectively. VCHCl3 is the
volume of chloroform and VH2 O is that of aqueous phase. Kd is the distribution constant for alkali and ammonium picrate which
was separately determined.
The binding constant K of the effector-free 3 is 5.3 times larger than that of 3$W(CO)4 complex. Thus, negative allosteric
cooperativity was observed in this case. One explanation for this trend is as follows: The conformation of bipyridyl in the
effector-free 3 is relatively flexible, while that of 3$W(CO)4 is fixed after complexation. As a result, the effector restricts the freedom
of the guest-binding cavity, and Naþ does not match the cavity very much, which results in weaker binding.
Nabeshima and coworkers reported a series of allosteric receptors called pseudomacrocycles whose macrocyclic frameworks are
formed in response to a metal ion as an effector.2 The ionophore 4 is the earliest example (Fig. 11).33 4 takes a linear form in the
absence of metal ions. In the presence of Cuþ as an effector, the terminal two bipyridyl groups complex with Cuþ intramolecularly,
and a macrocyclic structure with a defined cavity surrounded by oxygen atoms are formed. As a result, the selectivity for Kþ is greatly
enhanced after complexation. The ion recognition ability was examined by ion-transport experiments through a liquid membrane.
In the absence of Cuþ, the transport selectivity for Kþ over Naþ is 2.5, while the selectivity rises to 10 after the complexation with
Cuþ. In this case, the rates of the transport become smaller after complexation. This decrease is probably due to the electrostatic
repulsion between Cuþ and an alkali metal ion.
A pseudocrown ether 5 with a chiral (R)-binaphthyl moiety exhibits the transfer of chiral information to the bis(bpy) CuI moiety
upon the complexation with Naþ (Fig. 12).34 The complex 5$CuI has a pair of diastereomeric isomers depending the helicity of
bis(bpy)-CuI unit, P or M. In the absence of Naþ, the equilibrium between P and M isomers is not largely biased, with the ratio
M:P ¼ 1:0.74. Meanwhile, upon the complexation with Naþ, the crown-ether Naþ unit is conformationally fixed, which results

Figure 10 A macrocyclic bipyridyl-incorporated oligoethylene glycol 3 that shows negative heterotropic cooperativity in the recognition of Naþ.32
Cooperativity 389

Figure 11 A pseudocrown ether 4, whose ion-transport selectivity is enhanced upon complexation with Cuþ effector.33

Figure 12 Transfer of chiral information from binaphthyl to bis(bpy)-CuI complex in the presence of Naþ effector.34

in the strong preference for the P isomer (M:P ¼ 1:10). The equilibria were investigated in detail by the combination of NMR,
UV–vis, and CD titration experiments. This is different from an example of typical allosteric systems where the binding constant
is affected by an effector (Fig. 1), but the chirality of the focused unit as properties of the substances is regulated upon the interaction
with Naþ as an effector. Thus, this can be also considered to be a kind of a cooperative event.
Other ionophores that employ similar strategies of pseudomacrocycles include a linear oligoethylene glycol podand with two
salicylaldoxime moieties, to which transition metals such as NiII, CuII, and ZnII bind as effectors.35 A divalent alkaline earth metal
390 Cooperativity

Figure 13 A pseudocyclophane host 6 that recognizes dansylamide 7 upon the complexation with an effector.36

ion Ba2 þ is strongly captured in their cavity, due to the phenoxy oxygen atoms of the salicylaldoxime moieties, which have partial
negative charges.
As seen earlier, allosteric receptors with an oligoethylene glycol are suitable for the recognition of metal cations. The same
strategy of pseudomacrocylic hosts can be developed to the recognition of small organic molecules. One of the early examples is
the pseudocyclophane host 6 that offers a hydrophobic cavity in the presence of metal ions (CuII or ZnII) (Fig. 13).36 The binding
of dansylamide 7 to the metal complexes of 6 was confirmed by NMR and fluorescence experiments, while the binding was
not detected with 6 alone. Increased hydrophobic effect and Nþ-arene interactions are considered to contribute to the binding.
Formation of the ternary complex is applied to the detection of metal ions utilizing fluorescent properties of dansylamide 7.
The allosteric regulations of the other molecular receptors are also studied by investigating binding constants of guests in the presence
and absence of the effectors. A Ca2 þ-responsive pseudocyclophane that recognizes 6-(toluidino)-2-naphthalenesulfonate (TNS)
fluorophore was reported by Deshayes and coworkers,37 and its binding constant was determined by fluorescence titration experiments
and Benesi–Hildebrand analysis.38 A bis(bpy) podand with an asymmetric center at oligoethylene glycol units shows enantiomeric
recognition toward (R) and (S) phenylethylammonium, whose recognition behavior was studied by calorimetric titrations in the
presence and absence of CuI.39 Cu2 þ-responsive recognition of flavin mononucleotide (FMN), which is an important biological
cofactor, is achieved by a bis(bpy) cyclophane with an ammonium anchor, and its allosteric regulation was studied by solvent extraction
and transport techniques.40

2.18.2.3 Macrobicyclic and Macropolycyclic Receptors


Macrobicyclic and macropolycyclic receptors can exhibit stronger and more selective binding than macromonocyclic ones because
of their defined cavity and accumulated interaction units. The concept to construct the receptor scaffold in the presence of the
effector can be also applied to compounds belonging to such a category. The pseudocryptand 8 is the first report to synthesize
such metal-responsive macrobicyclic receptor (Fig. 14).41 However, this attempt was not so successful because the affinity of alkali
metal ions is not changed by the formation of bis(bpy) complex with the effector. One of the reasons can be ascribed to the macro-
cyclic unit of 8 present even in its effector-free form.
Molecular design realizing complete on/off conversion of macrocyclic frameworks achieves excellent allosteric receptors with clear
switching of affinity (Fig. 15).42 A tripodal receptor 9 with three terminal bpy groups is acyclic in the absence of the effectors. In the
presence of FeII as an effector, a helical pseudocryptand 9$FeII is formed through the intramolecular complexation of the three bpy

Figure 14 Formation of pseudocryptand of 8 upon complexation with CuI.41


Cooperativity 391

Figure 15 A tripodal receptor 9 forming a helical pseudocryptand upon the complexation with FeII.42

Table 3 Binding constants K (M 1) of the receptors with alkali metal ions (1H NMR, CD3CN, 298 K)42

Guest
Host Naþ Kþ Rbþ Csþ

9 650 610 280 120


9$FeII –a 110 630 4700
a
Too small to be determined.

groups. Table 3 summarizes the binding constants between 9 and alkali metal ions in the presence and absence of FeII. A clear trend is
observed in terms of the size of the ions. That is, negative allostery is observed for the recognition of Naþ and Kþ, while positive
allostery is observed for the recognition of Rbþ and Csþ. Remarkably, the binding of Csþ is increased up to 39-fold. Here, the fixation
of the host structure results in the enhancement of selectivity toward a larger alkali metal ion, as seen in the case of Fig. 11.
A tripodal receptor which possesses two amide-hydroxamic acid moieties at each chain was synthesized by Shanzer
and coworkers.43 Its stepwise binding constants with FeII K1–K3 were determined by UV–vis and CD titration experiments, in
combination with the characterization by ESI-MS measurements. The binding behavior (positive or negative allostery) is dependent
on the length and substituents of receptors.
The expansion and contraction of the macrocyclic framework is also effective for the regulation of molecular recognition.
Schneider and coworkers synthesized macrocyclic and macrobicyclic polyazacyclophanes whose cavity is contracted upon
the complexation with metal effectors such as ZnII and CuII.44 In the contracted cavity, an aromatic guest such as
5-dimethylaminonaphthalene-2-sulfonic acid (DNSA) is strongly captured, thus positive allosteric regulation is achieved.
As a macropolycyclic receptor, Nolte, Rowan, and coworkers reported porphyrin-based macrocyclic receptors utilizing its
aromatic surfaces and metal center.45 The receptor 10 shown in Fig. 16 possesses two cavities on both sides of the porphyrin.46

Figure 16 Double-cavity porphyrin receptor 10 showing highly negative homotropic binding of viologen 11.46
392 Cooperativity

Stepwise binding constants of viologen 11 by receptor 10 in CHCl3/CH3CN ¼ 4:1 solution were evaluated by UV–vis and
fluorescence titrations. The values Ka1 ¼ 7  107 M 1 and Ka2 ¼ 5  104 M 1 were obtained, respectively, from which the cooperative
parameter a is calculated to be 0.003. Thus, highly negative allosteric binding is observed in this case. Electrostatic repulsion of the
second guest with the firstly bound guest and/or the conformational change (pinching of the vacant cavity after first binding, etc.) is
considered to be the reason for the negative cooperativity. Allosteric regulations of a Zn-porphyrin-based receptor with one-side
cavity were also reported, which utilizes pyridine derivatives as positive heterotropic effectors to bind to Zn center.47–51

2.18.3 Cooperative Supramolecular Systems


2.18.3.1 Introduction
Cooperativity and multivalency are closely related concepts.26,52,53 In particular, allosteric cooperativity in recognition arises
when an effector binding affects the affinity of another guest-binding site (Fig. 1). Meanwhile, multivalent interactions mean
simultaneous interactions of multiple binding sites and often also result in enhancement of the binding between components
(Fig. 17).5,54,55 It is noteworthy that multivalency does not necessarily mean cooperativity, since noncooperative, additive
accumulation of binding forces can be a case of multivalent interactions.
In “Macrocyclic and Macropolycyclic Receptors Exhibiting Allosteric Cooperativity” section, allosteric receptors whose
structures are regulated by an effector are introduced. Fixation of the host structures realizes organization of the guest-binding sites
and results in specific binding. Preorganization is an important concept also seen in templated syntheses, where complex and
elaborate (supra)molecular structures are built up by assembling the precursor units in the presence of the template molecules.56
Thus, the cooperative systems and the template syntheses share a common feature, where an effector and a template play a role in
regulating host structure and in organizing the precursors, respectively. Furthermore, molecules built up by templated synthesis
are often proven to be excellent cooperative systems, for their organized structures with many functionalities. In this article,
some examples of templated syntheses are explained, although they might not be a cooperative system in a strict sense.
Cooperativity is expressed not only in closed systems but also in open, polymeric systems. Fig. 18 shows the nucleation and
elongation of self-assembled oligomers of bifunctional monomer M. Investigation of cooperativity in such systems is essential
to understand biological events like the formation of amyloid fibers,57 but also for the formation of artificial supramolecular
polymers.58 The binding constant Ki of each elongation step is simply expressed as Eq. (15).
½Miþ1 
Ki ¼ (15)
½Mi ½M
If all the binding constants have the same value (K1 ¼ K2 ¼ / ¼ Kn), the system is noncooperative and called isodesmic polymer-
ization. If K1 is smaller than Ki (i > 1) and/or Ki became larger as the increase of i, the system has positive cooperativity and
elongation of supramolecular polymers is favored. On the other hand, if K1 is larger than Ki (i > 1) and/or Ki became smaller as
the increase of i, the system exhibits negative cooperativity and the small oligomers such as dimer M2 become favored species.
This article focuses on closed, discrete systems but does not deal with the open polymeric cooperative systems, because the detailed
theories and many excellent examples of the analysis of cooperative polymerization are discussed elsewhere.7,26,59–67

Figure 17 Multivalent interaction.

Figure 18 Cooperativity in supramolecular polymers.


Cooperativity 393

2.18.3.2 Templated Synthesis and Cooperativity


2.18.3.2.1 Ion as a preorganization effector
A pioneering work of the use of template would be the synthesis of crown ethers in the presence of Naþ.68 An oligoethylene glycol
chain is wrapped and preorganized around the metal ion template, which enables the intramolecular reaction between the terminal
functional groups of linear precursor. This case utilizes the template inside the (pre)macrocyclic framework. Tucker, Desvergne, and
coworkers reported the synthesis of anthracene-based crown ether that responds to two kinds of effectors (Fig. 19).69 Intramolecular
photodimerization reaction of two terminal anthracene units of a cyclization precursor 12 was investigated. In the absence of
effectors, the reaction quantum yield FR was 0.19. 12 has a central bipyridyl group and ethylene glycol linkers as the structure-
regulating unit. In the presence of Naþ as an effector, ethylene glycol units work as a chelating units and 12 takes a conformation
where anthracene units are brought in proximity. As a result, FR increases to 0.29 in the presence of Naþ. 12 is also responsive to
Hg2 þ, which is complexed to the bpy unit and fixes its conformation, in a similar manner to the bpy-crown ether 3 in Fig. 10. FR is
0.26 in the presence of Hg2 þ, which is also improved compared to the absence of the effector. Furthermore, in the presence of both
Naþ and Hg2 þ, the yield FR increases up to 0.32. To summarize, in this case, the photoreaction quantum yield FR can be seen as the
factor to quantify the degree of cooperativity of the system.
Multimetal supramolecular systems can realize precise tuning of their functions utilizing various characteristics of metal
elements and cooperative interactions between them.2,70–80 An oligooxime ligand H613 with chiral end groups shows interesting
property to switch its helicity in response to multiple kinds of effectors (Fig. 20).77 That is, successive addition of metal ions as

Figure 19 Intramolecular photocycloaddition of anthracene units of 12 in the presence of metal cations as effectors.69

Figure 20 Successive helicity switching realized by different metal ion effectors.77


394 Cooperativity

effectors in the order of Zn2 þ, Ba2 þ, and La3 þ results in the stepwise helicity inversion by replacing the effector inside the helix,
which was monitored by circular dichroism signals. Here, lability of coordination bonds is effectively applied to the dynamic
regulation of the molecular system.
Templating methods can be used for the synthesis of more complex structures such as interlocked molecules. Sauvage and
coworkers reported the synthesis of catenanes by creating an intersection point utilizing coordination of Cuþ to phenanthrolines,
following the cyclization reaction to attach oligoethylene glycol chains.81 Here, the template Cuþ ion can be removed after the
synthesis of interlocked molecules. In general, functional groups that had been used for the preorganization can be utilized as
guest-binding sites after the removal of templates. Clever and coworkers reported a series of interlocked double cages constructed
from banana-shaped bis(pyridyl) ligands and PdII ion.82 The quadruply interlocked cage 14 is firstly formed by coordination-
driven self-assembly of an acridone-based ligand and Pd(BF4)2 (Fig. 21).83 Three BF4  anions act as templates by exerting
electrostatic interactions with four Pd2 þ centers. The halide ions (Cl or Br) as effectors replace the two BF4  anions at the upper
and bottom cavities, which are contracted upon anion exchange. This structural change results in the widening of the central
cavity, and enables encapsulation of a small organic guest such as benzene, cyclohexane, and norbornadiene. The binding
constants and uptake kinetics were investigated by 1H NMR spectroscopy, in combination with the characterization by
ESI-MS and X-ray crystallography.
Significant contribution of an effector to preorganization of the structure can also be seen in the hierarchically self-assembled
structure shown in Fig. 22.84 The preorganization of a tripodal ligand 15 by a transition metal ion into a helical building block
is requisite to form a triple-stranded interlocked helicate, which is formed by the Schiff-base formation of each component by
diamine 16. Here, a PF6  anion is encapsulated at the center of the interlocked helicate, which is presumably considered to stabilize
the structure as a template.

2.18.3.2.2 Cooperativity in Multivalent Receptors


Receptors with multiple binding sites can exhibit chelate cooperativity and/or interannular cooperativity as explained in “Models of
Cooperativity” section. Shinkai and coworkers reported multivalent receptors based on porphyrin double-decker complexes,
whose porphyrin ligands can rotate around the C4 symmetry axis of the porphyrin core.23,85,86 The LaIII double-decker porphyrin
17 possessing benzo-15-crown-5 moieties exhibits interesting difference in cooperativity depending on the alkali metal guest
(Fig. 23).86 In the recognition of Naþ, which coordinates to the crown-ether unit in the 1:1 ratio, strong cooperativity is not
observed. Meanwhile, Kþ is sandwiched between crown-ether units of the upper and lower porphyrin ligands. Consequently,

Figure 21 A quadruply interlocked receptor 14 whose guest recognition properties are triggered by halide effectors.83
Cooperativity 395

Figure 22 Formation of interlocked helicate via self-assembly of building blocks preorganized by transition metals.84

Figure 23 A LaIII double-decker porphyrin receptor 17 showing positive cooperativity in the binding of Kþ ion.86

the first binding of Kþ restricts the rotation of the porphyrin ligands, and increases the affinity of the other pairs of crown ethers.
Thus, positive cooperativity is observed in the recognition of Kþ. The overall binding constant Kf is 1.0  1014 M 4 in CHCl3/
CH3CN ¼ 1:1, and the Hill coefficient of the system is as high as 4.0.
Anderson and coworkers reported a series of multiporphyrin macrocycles connected via diyne linkers.87–92 Various polypyridyl
templates that offer multivalent coordination bonds to Zn-porphyrin units are utilized to synthesize macrocycles with different
numbers of repeating units. A hexapyridyl template 18 with a hexaphenylbenzene core is utilized for the synthesis of a cyclic
porphyrin hexamer 19.89 The chelate cooperative effect and multivalent interactions between the cyclic or linear porphyrin
oligomers and polypyridyl guests were investigated in detail.90 Stepwise binding constants in the case of porphyrin hexamers
and a hexapyridyl guest are expressed as in Fig. 24, using the reference binding constant K between a pyridyl group and a
Zn-porphyrin from one side and stepwise effective molarities EMi. The binding constants are too high to be determined by normal
titration experiments (an overall binding constant as 1:1 binding between 18 and 19 was 1036 M 1), thus evaluated by
396 Cooperativity

Figure 24 Stepwise binding constants between a cyclic porphyrin hexamer 19 or a linear porphyrin hexamer 20 and a hexapyridyl guest 18. A repre-
sentative isomer is shown for each binding step. (Ar ¼ 3,5-di-tert-butylphenyl (for 19), 3,5-bis(octyloxy)phenyl (for 20). THS ¼ tri(n-hexyl)silyl.).90

denaturation experiments7 using incremental addition of quinuclidine as a monodentate competitive ligand. As a result, for the
cyclic hexamer, the stepwise effective molarities EMi (i ¼ 3–6) are as high as 102–103 M 1. On the other hand, the values for the
linear hexamer 20 are about 0.05 M 1. Thus, this is an excellent example of the strong chelate cooperative effect of the multivalent
macrocyclic receptor.
The hexapyridyl template 18 is also utilized for the synthesis of a porphyrin macrocycle larger than the hexamer. The Vernier
templating method, which utilizes a preorganized intermediate (Vernier complex) of which interaction sites the number n is the
common multiple of that of the receptor and that of the template, is applied to the synthesis of a cyclic porphyrin dodecamer 21
(Fig. 25).91 Specifically, a linear porphyrin tetramer 22 and the hexapyridyl template 18 result in the formation of the Vernier
complex 223$182 with 12 coordination bonds, and following diyne coupling achieves the formation of a cyclic dodecamer
21a. Dimerization behavior of the cyclic dodecamer 21b linked by 1,4-diazabicyclo[2,2,2]octane (DABCO) was investigated
Cooperativity 397

Figure 25 (Upper) Synthesis of a cyclic porphyrin dodecamer 21a utilizing a preorganized Vernier complex (Ar ¼ 3,5-di-tert-butylphenyl).91 (Lower)
Highly cooperative dimerization of the cyclic porphyrin dodecamer 21b by DABCO (Ar ¼ 3,5-bis(octyloxy)phenyl).92

by UV–vis–NIR titration (Fig. 25).92 The titration experiment revealed that the self-assembly can be regarded as an all-or-nothing,
two-state process. The self-assembly is highly cooperative, with the Hill coefficient nH ¼ 9.5, and the overall binding constant Kf
(M 13) defined as Eq. (16) is as high as logKf ¼ 131  1.
½21b2 $DABCO12 
Kf ¼ (16)
½21b2 ½DABCO12
Cooperativity in molecular motion was discussed in a molecular elevator system that consists of threefold bistable rotaxanes
(Fig. 26).93 The “platform” 23 is a trivalent receptor with dibenzo[24]crown-8 rings, and the “rig” 24 is a tripodal unit with
ammonium and bipyridinium station on each leg. In an acidic condition, the ammonium moieties are protonated and the
crown-ether moieties strongly interact with the ammoniums, thus the platform is in the “upper” level. The addition of the
phosphazene base results in deprotonation of the ammoniums and the ring is moved to the bipyridinium units, which transfers
the platform to the “lower” level. The initial and final states together with the intermediates in the transportation process were
investigated by 1H NMR, UV–vis titration, cyclic voltammetry, and molecular mechanics calculations. It was concluded that not
simultaneous but stepwise deprotonation occurs upon the incremental addition of base, thus the regulation of the transportation
system is perfectly stepwise.
Evaluation of multivalency and chelate cooperativity was performed with a triply threaded pseudorotaxane 25$26 (Fig. 27),94
which possesses the same topology as the molecular elevator 23$24. The pseudorotaxane 25$26 consists of a trivalent tetralactam
receptor 25 and a tripodal guest 26 with three diamide units. Monovalent compounds 27,28 (Fig. 27) and divalent compounds
(not shown) were also synthesized, and utilized for the double-mutant cycle analysis.95 The cycle can be considered as the relation-
ship between four situations A–D. The individual binding constants KA–KD are expressed as shown in Fig. 27, using the appropriate
statistical factors, microscopic binding constant K, and stepwise effective molarities EMi (i ¼ 2,3). As a result, the overall equilibrium
constant K for the cycle is calculated according to Eq. (17).
398 Cooperativity

Figure 26 A molecular elevator system 23$24 that regulates transportation by acid–base stimuli.93

 
KA KD 3
K¼ ¼ EM2 EM3 (17)
KB KC 4
KA–KD were experimentally determined by ITC measurements, and EM2 can be estimated from the double-mutant cycle analysis
of the corresponding divalent receptor. Thus, independent determination of EMi for the second and third binding is possible for the
formation of the triply threaded pseudorotaxane. The second binding is found to have slight positive cooperativity, while the third
binding is nearly noncooperative. Thus, compared with the case of Fig. 24, it can be said that not all multivalent systems lead to
strong positive cooperativity. Conformational freedom in both receptors and guests, total electrostatic valances, and many other
factors have to be considered to design cooperative supramolecular systems with desired functions.

2.18.4 Supramolecular Cooperative Catalytic Systems

In the biological systems, various enzymes function to produce various physiological chemical species and degrade unnecessary
substances. To keep the whole system constant and robust (i.e., homeostatic), elaborate multistep allosteric regulations are in
action, which are responsive to fluctuations of the amounts of substances. In the previous sections, artificial molecular receptors
are introduced that achieve regulation of molecular recognition ability by changing the shape and properties of their guest-
binding sites. Here, if the effector binding causes a certain structural change of a catalytic site in the same molecule/supramolecule,
the molecular system can work as an allosteric catalyst. In the allosteric receptors, the change in binding constants K (M 1 unit in 1:1
bindings) is discussed to evaluate the cooperativity. By contrast, in the case of allosteric catalyst, the reaction rates k (s 1, M 1 s 1,
etc.) or simply reaction yields in the presence and absence of effectors are discussed. Michaelis–Menten kinetic analyses are also used
to discuss the consequences of the effectors.96 Detail explanations of the analyses are not given here, but some excellent examples of
artificial cooperative catalysts are presented in this section. For reference, see other excellent reviews that summarize examples of
such “artificial enzymes.”24–26,97–101
CuII or ZnII complexes exhibit Lewis acidity, thus some complexes show catalytic activity toward hydrolysis. Various hydrolases
are known to have CuII or ZnII in their reaction centers.102 Inspired by such enzymes, artificial model mononuclear and polynuclear
complexes were synthesized and widely investigated.103 In particular, for the polynuclear complexes with two or more catalytic
centers, the catalytic activity can be actively controlled by changing the relative geometry of the catalytic units in response to suitable
effectors.104–109 An artificial phosphodiesterase 29$Mn (M ¼ Zn or Cu) is the case, which bears two dpa (2,20 -dipycolylamine) metal
complexes as catalytic units and bpy (2,20 -bipydiryl) linker as a regulating unit (Fig. 28).109 Dpa has a higher affinity than bpy
toward these metal ions. In the presence of 2 equiv. of metals, a dinuclear complex 29$M2 whose dpa units are selectively
complexed is formed. 29$M2 has low catalytic activity toward the hydrolysis of HPNP (2-hydroxypropyl-p-nitrophenyl phosphate)
because the two dpa metal complex units are apart from each other. Further addition of the metals leads to the complexation at the
Cooperativity 399

Figure 27 Double-mutant cycle analysis for the evaluation of chelate cooperativity in a triply threaded pseudorotaxane 25$26.94
400 Cooperativity

Figure 28 Allosteric catalysts for the hydrolysis of phosphodiester regulated by the presence of metal ions.109

bpy part to form 29$M3 (together with the bis(bpy) complex 292$M5). The stepwise complexation event was carefully examined by
UV–vis titration experiments. As the result of the fixation of the bpy linker, two dpa moieties are brought in proximity, and the
catalytic activity of 29$M3 allosterically increases. The pseudo-first-order rate constant kobs for the hydrolysis of HPNP is monitored
under different concentrations of metal ions. The enhancement of kobs by converting 29$M2 to 29$M3 are 3.3 and 19.4 for the cases
of ZnII and CuII, respectively. It is remarkable that a clear switching in catalytic activity is achieved at a certain concentration
threshold.
Mirkin and coworkers reported a series of allosteric catalysts whose molecular frameworks are converted by the weak-link
approach, where the formation and dissociation of coordination bonds in the supramolecular metal complexes are reversibly
controlled.110,111 A macrocyclic bis CrIII-salen-based catalyst 30 in Fig. 29 has RhI complexes as structure-regulating units.112 In
its closed form, the hemilabile terminal coordinating group works as a P,S-bidentate chelating ligand to the RhI center. Meanwhile,
after the addition of CO and Cl as strongly coordinating ligands, structural switching at the RhI complexes occurs to result in the
open macrocyclic complex. Compared to the closed form, the open form shows a doubling in the reaction rate of the ring opening
of cyclohexene oxide. The rate-determining step of the same reaction catalyzed by a mononuclear complex was found out to involve
a bimetallic intermediate in the preceding study,113 so the reason for the high activity of 30 can also be ascribed to the cooperative
function by the two CrIII-salen complexes in the cavity of the macrocycle.
Mirkin and coworkers developed this approach to a triple-layer catalyst 31 (Fig. 30).114 In its closed form, the bulky
bis(biphenyl)amine-derived substituents cover both the faces of the AlIII-salen unit, thus its catalytic ability is inactivated. Upon
the addition of the effector, Cl, or CH3CN, the RhI complex units are converted to semiopen form by ligand exchange. This alters
the molecular framework and allows access of the substrate to the catalytic center. The semiopen AlIII-salen complex works as a good
catalyst for ring-opening polymerization of 3-caprolactone. The in situ allosteric control of catalytic ability realizes the regulation of
molecular weights of the resulting polymer.
As a bis-salen complex whose catalytic activity is regulated by structural transformation induced by the effector, a podand-based
dimeric CrIII-salen complex was reported.115 To both ends of an oligoethylene glycol chain are appended chiral CrIII-salen units. In
the absence of an effector, the catalyst takes a linear form with smaller interaction between the CrIII-salen units. Meanwhile, a Kþ
cation as an effector complexes to the oligoethylene glycol chain and fold the ligand framework around it. As a result, the

Figure 29 A macrocyclic allosteric catalyst 30 whose activity is controlled by structural change induced by strongly coordinating ligands.112
Cooperativity 401

Figure 30 A triple-layer allosteric catalyst 31 that can regulate ring-opening polymerization of 3-caprolactone.114

conformational freedom of two terminal chiral CrIII-salen units is restricted, and improvements in yield and enantioselectivity are
observed in the Henry reaction. Here, enantiomeric excess (ee) is used to discuss the effect of cooperativity, which is also considered
to involve the interplay of the catalytic centers during the reaction process.
A strong guest-binding ability induced by an effector is also used for the regulation of the catalytic system. Pentafoil knot 32
shown in Fig. 31 is synthesized via self-assembly of a linear tris(bpy) ligand and a FeII ion, covalent linking of terminal olefin groups

Figure 31 A pentafoil knot 32 whose halide trapping ability is allosterically regulated by ZnII.116
402 Cooperativity

by metathesis, and removal of the FeII ion by EDTA.116 32 reversibly binds the ZnII to form the pentanuclear complex, which
strongly binds halide ions such as Cl and Br in its cavity. In other words, this halide trapping ability is allosterically regulated
by the ZnII ion. The pentanuclear complex of 32 was employed for the catalytic hydrolysis of Ph2CHBr and for the initiation of Lewis
acid-catalyzed Michael addition and Diels–Alder reactions by generating trityl carbocation from trityl bromide.
Many allosteric catalysts developed so far were designed by combining and attaching regulating units to an established catalyst.
ON/OFF switching of their catalytic activities has been achieved, but the function in the ON state is more or less the same as the
reference catalyst. A development of cooperative catalysts realizing synergistic reactions that cannot be achieved by noncooperative
catalysts would greatly advance this field. In another aspect, as seen in the example of Fig. 31, the artificial allosteric catalyst can be
employed to multistep cascade reactions, which is reminiscent of signal transduction in the biological system. A few pioneering
works in this direction have been reported.117–119 Systems combining multiple artificial allosteric catalysts have a huge potential
for the realization of functioning complex systems.

2.18.5 Summary and Outlook

In this article, we have covered the concept of cooperativity, models of cooperativity in guest binding, and many examples of
cooperative supramolecular systems. In an allosteric sense, cooperativity means the change in affinity of a guest-binding site by
means of a certain effector. However, cooperativity can be interpreted as the change in a structure or a physical property induced
by the effector. As the physical properties, not only the binding constant K and/or the effective molarity (EM), but also the ratio
of the isomers, chiroptical outputs, degree of self-assembly, rate of reaction, and many other factors can be employed as a measure
to evaluate the cooperativity. Understanding and controlling of cooperativity, that is, how a certain substance influences the state of
a target molecule, have been essential to the supramolecular chemistry. Its importance would increase more and more and open
a new way to create the next generation of functional supramolecular systems including self-learning systems, structural and
functional evolution systems, supramolecular complex systems, etc., where numerous chemical species as the building units of
the system mutually interact with each other and precisely communicate on demand.

References

1. Perutz, M. F. Q. Rev. Biophys. 1989, 22, 139–237.


2. Nabeshima, T. Bull. Chem. Soc. Jpn. 2010, 83, 969–991.
3. Baldwin, J.; Chothia, C. J. Mol. Biol. 1979, 129, 175–220.
4. Nabeshima, T., Ed. Synergy in Supramolecular Chemistry; CRC Press: Boca Raton, FL, 2014.
5. Mammen, M.; Choi, S.-K.; Whitesides, G. M. Angew. Chem. Int. Ed. 1998, 37, 2754–2794.
6. Mitchell, M. Complexity: A Guided Tour; Oxford University Press: New York, NY, 2009.
7. Hunter, C. A.; Anderson, H. L. Angew. Chem. Int. Ed. 2009, 48, 7488–7499.
8. Ercolani, G.; Schiaffino, L. Angew. Chem. Int. Ed. 2011, 50, 1762–1768.
9. Schneider, H.-J. Org. Biomol. Chem. 2016, 14, 7994–8001.
10. Connors, K. A. Binding Constants: The Measurement of Molecular Complex Stability; John Wiley & Sons: Chichester, 1987; pp 78–86.
11. Ercolani, G.; Piguet, C.; Borkovec, M.; Hamacek, J. J. Phys. Chem. B 2007, 111, 12195–12203.
12. Schally, C. A., Ed. Analytical Methods in Supramolecular Chemistry, 2nd, Completely Revised and Enlarged Edition;; Wiley-VCH: Weinheim, 2012.
13. Valeur, G.; Berberan-Santos, M. N.; Martin, M. M.; Plaza, P. In Analytical Methods in Supramolecular Chemistry; Schally, C. A., Ed., 2nd, Completely Revised and Enlarged
Edition;; Wiley-VCH: Weinheim, 2012; pp 287–336.
14. Hirose, K. In Analytical Methods in Supramolecular Chemistry; Schally, C. A., Ed., 2nd, Completely Revised and Enlarged Edition;; Wiley-VCH: Weinheim, 2012; pp 27–66.
15. Schmidtchen, F. P. In Analytical Methods in Supramolecular Chemistry; Schally, C. A., Ed.; 2nd, Completely Revised and Enlarged Edition; Wiley-VCH: Weinheim, 2012; pp
67–103.
16. Ceroni, P.; Credi, A.; Venturi, M. In Analytical Methods in Supramolecular Chemistry; Schally, C. A., Ed.; 2nd, Completely Revised and Enlarged Edition; Wiley-VCH: Weinheim,
2012; pp 371–457.
17. Stephan, H.; Kubeil, M.; Gloe, K.; Gloe, K. In Analytical Methods in Supramolecular Chemistry; Schally, C. A., Ed.; 2nd, Completely Revised and Enlarged Edition; Wiley-VCH:
Weinheim, 2012; pp 105–127.
18. Nabeshima, T.; Saiki, T.; Sumitomo, K.; Akine, S. Tetrahedron Lett. 2004, 45, 6761–6763.
19. Yamaki, Y.; Nakamura, T.; Suzuki, S.; Yamamura, M.; Minoura, M.; Nabeshima, T. Eur. J. Org. Chem. 2016, 2016, 1678–1683.
20. Akine, S. TitrationFit, version 1.1.0., 2013. http://chem.s.kanazawa-u.ac.jp/coord/index_e.html (accessed Aug 08, 2016).
21. Gampp, H.; Maeder, M.; Meyer, C. J.; Zuberbühler, A. D. Talanta 1985, 32, 257–264.
22. Hirose, K. In Synergy in Supramolecular Chemistry; Nabeshima, T., Ed.; CRC Press: Boca Raton, FL, 2014; pp 261–300.
23. Takeuchi, M.; Ikeda, M.; Sugasaki, A.; Shinkai, S. Acc. Chem. Res. 2001, 34, 865–873.
24. Kovbasyuk, L.; Krämer, R. Chem. Rev. 2004, 104, 3161–3187.
25. Kremer, C.; Lützen, A. Chem. Eur. J. 2013, 19, 6162–6196.
26. Biedermann, F.; Schneider, H.-J. Chem. Rev. 2016, 116, 5216–5300.
27. Bradshaw, S.; Izatt, R. M.; Bordunov, A. V.; Zhu, C. Y.; Hathaway, J. K. In Comprehensive Supramolecular Chemistry, vol. 1, Gokel, G. W., Ed.; Pergamon Press: Oxford,
1996; pp 35–95.
28. Dietrich, B. In Comprehensive Supramolecular Chemistry, vol. 1, Gokel, G. W., Ed.; Pergamon Press: Oxford, 1999; pp 153–211.
29. Lehn, J.-M. Angew. Chem. Int. Ed. Engl. 1988, 27, 89–112
30. McConnell, A. J.; Beer, P. D. Angew. Chem. Int. Ed. 2012, 51, 5052–5061.
31. Kim, S. K.; Sessler, J. L. Chem. Soc. Rev. 2010, 39, 3784–3809.
32. Rebek, J., Jr.; Trend, J. E.; Wattley, R. V.; Chakravorti, S. J. Am. Chem. Soc. 1979, 101, 4333–4337.
Cooperativity 403

33. Nabeshima, T.; Inaba, T.; Furukawa, N. Tetrahedron Lett. 1987, 28, 6211–6214.
34. Nabeshima, T.; Hashiguchi, A.; Saiki, T.; Akine, S. Angew. Chem. Int. Ed. 2002, 41, 481–484.
35. Van Veggel, F. C. J. M.; Harkema, S.; Bos, M.; Verboom, W.; Woolthuis, G. K.; Reinhoudt, D. N. J. Org. Chem. 1989, 54, 2351–2359.
36. Schneider, H.-J.; Ruf, D. Angew. Chem. Int. Ed. 1990, 29, 1159–1160.
37. Cole, K. L.; Farran, M. A.; Deshayes, K. Tetrahedron Lett. 1992, 33, 599–602.
38. Benesi, H. A.; Hildebrand, J. H. J. Am. Chem. Soc. 1949, 71, 2703–2707.
39. Habata, Y.; Bradshaw, J. S.; Zhang, X. X.; Izatt, R. M. J. Am. Chem. Soc. 1997, 119, 7145–7146.
40. Nabeshima, T.; Hashiguchi, A.; Yazawa, S.; Haruyama, T.; Yano, Y. J. Org. Chem. 1998, 63, 2788–2789.
41. Nabeshima, T.; Inaba, T.; Sagae, T.; Furukawa, N. Tetrahedron Lett. 1990, 31, 3919–3922.
42. Nabeshima, T.; Yoshihira, Y.; Saiki, T.; Akine, S.; Horn, E. J. Am. Chem. Soc. 2003, 125, 28–29.
43. Blanc, S.; Yakirevitch, P.; Leize, E.; Meyer, M.; Libman, J.; Van Dorsselaer, A.; Albrecht-Gary, A.-M.; Shanzer, A. J. Am. Chem. Soc. 1997, 119, 4934–4944.
44. Baldes, R.; Schneider, H.-J. Angew. Chem. Int. Ed. Engl. 1995, 34, 321–323.
45. Elemans, J. A. A. W.; Claase, M. B.; Aarts, P. P. M.; Rowan, A. E.; Schenning, A. P. H. J.; Nolte, R. J. M. J. Org. Chem. 1999, 64, 7009–7016.
46. Thordarson, P.; Bijsterveld, E. J. A.; Elemans, J. A. A. W.; Kasák, P.; Nolte, R. J. M.; Rowan, A. E. J. Am. Chem. Soc. 2003, 125, 1186–1187.
47. Thordarson, P.; Coumans, R. G. E.; Elemans, J. A. A. W.; Thomassen, P. J.; Visser, J.; Rowan, A. E.; Nolte, R. J. M. Angew. Chem. Int. Ed. 2004, 43, 4755–4759.
48. Thomassen, P. J.; Foekema, J.; Jordana i Lluch, R.; Thordarson, P.; Elemans, J. A. A. W.; Nolte, R. J. M.; Rowan, A. E. New J. Chem. 2006, 30, 148–155.
49. Veling, N.; Thomassen, P. J.; Thordarson, P.; Elemans, J. A. A. W.; Nolte, R. J. M.; Rowan, A. E. Tetrahedron 2008, 64, 8535–8542.
50. Deutman, A. B. C.; Monnereau, C.; Moalin, M.; Coumans, R. G. E.; Veling, N.; Coenen, M.; Smits, J. M. M.; De Gelder, R.; Elemans, J. A. A. W.; Ercolani, G.; Nolte, R. J. M.;
Rowan, A. E. Proc. Natl. Acad. Sci. 2009, 106, 10471–10476.
51. Cantekin, S.; Markvoort, A. J.; Elemans, J. A. A. W.; Rowan, A. E.; Nolte, R. J. M. J. Am. Chem. Soc. 2015, 137, 3915–3923.
52. Badjic, J. D.; Nelson, A.; Cantrill, S. J.; Turnbull, W. B.; Stoddart, J. F. Acc. Chem. Res. 2005, 38, 723–732.
53. Fasting, C.; Schalley, C. A.; Weber, M.; Seitz, O.; Hecht, S.; Koksch, B.; Dernedde, J.; Graf, C.; Knapp, E.-W.; Haag, R. Angew. Chem. Int. Ed. 2012, 51, 10472–10498.
54. Mulder, A.; Huskens, J.; Reinhoudt, D. N. Org. Biomol. Chem. 2004, 2, 3409–3424.
55. Baldini, L.; Casnati, A.; Sansone, F.; Ungaro, R. Chem. Soc. Rev. 2007, 36, 254–266.
56. Arico, F.; Badjic, J. D.; Cantrill, S. J.; Flood, A. H.; Leung, K. C.-F.; Liu, Y.; Stoddart, J. F. Top. Curr. Chem. 2005, 249, 203–259.
57. Terzi, E.; Hölzemann, G.; Seelig, J. J. Mol. Biol. 1995, 252, 633–642.
58. Harada, A., Ed. Supramolecular Polymer Chemistry; Wiley-VCH: Weinheim, 2012.
59. De Greef, T. F. A.; Smulders, M. M. J.; Wolffs, M.; Schenning, A. P. H. J.; Sijbesma, R. P.; Meijer, E. W. Chem. Rev. 2009, 109, 5687–5754.
60. Zhao, D.; Moore, J. S. Org. Biomol. Chem. 2003, 1, 3471–3491.
61. Chen, Z.; Lohr, A.; Saha-Möller, C. R.; Würthner, F. Chem. Soc. Rev. 2009, 38, 564–584.
62. Smulders, M. M. J.; Schenning, A. P. H. J.; Meijer, E. W. J. Am. Chem. Soc. 2008, 130, 606–611.
63. Smulders, M. M. J.; Nieuwenhuizen, M. M. L.; De Greef, T. F. A.; Van der Schoot, P.; Schenning, A. P. H. J.; Meijer, E. W. Chem. Eur. J. 2010, 16, 362–367.
64. Markvoort, A. J.; Ten Eikelder, H. M. M.; Hilbers, P. A. J.; De Greef, T. F. A.; Meijer, E. W. Nat. Commun. 2011, 2, 509.
65. Korevaar, P. A.; George, S. J.; Markvoort, A. J.; Smulders, M. M. J.; Hilbers, P. A. J.; Schenning, A. P. H. J.; De Greef, T. F. A.; Meijer, E. W. Nature 2012, 481, 492–496.
66. Ogi, S.; Sugiyasu, K.; Manna, S.; Samitsu, S.; Takeuchi, M. Nat. Chem. 2014, 6, 188–195.
67. Kang, J.; Miyajima, D.; Mori, T.; Inoue, Y.; Itoh, Y.; Aida, T. Science 2015, 347, 646–651.
68. Pedersen, C. J. J. Am. Chem. Soc. 1967, 89, 2495–2496.
69. McSkimming, G.; Tucker, J. H. R.; Bouas-Laurent, H.; Desvergne, J.-P. Angew. Chem. Int. Ed. 2000, 39, 2167–2169.
70. Akine, S.; Taniguchi, T.; Nabeshima, T. Angew. Chem. Int. Ed. 2002, 41, 4670–4673.
71. Akine, S.; Taniguchi, T.; Saiki, T.; Nabeshima, T. J. Am. Chem. Soc. 2005, 127, 540–541.
72. Nabeshima, T.; Saiki, T.; Iwabuchi, J.; Akine, S. J. Am. Chem. Soc. 2005, 127, 5507–5511.
73. Nabeshima, T.; Miyazaki, H.; Iwasaki, A.; Akine, S.; Saiki, T.; Ikeda, C.; Sato, S. Chem. Lett. 2006, 35, 1070–1071.
74. Akine, S.; Taniguchi, T.; Nabeshima, T. J. Am. Chem. Soc. 2006, 128, 15765–15774.
75. Yamashita, A.; Watanabe, A.; Akine, S.; Nabeshima, T.; Nakano, M.; Yamamura, T.; Kajiwara, T. Angew. Chem. Int. Ed. 2011, 50, 4016–4019.
76. Akine, S.; Hotate, S.; Nabeshima, T. J. Am. Chem. Soc. 2011, 133, 13868–13871.
77. Akine, S.; Sairenji, S.; Taniguchi, T.; Nabeshima, T. J. Am. Chem. Soc. 2013, 135, 12948–12951.
78. Akine, S.; Matsumoto, T.; Nabeshima, T. Angew. Chem. Int. Ed. 2016, 55, 960–964.
79. Nabeshima, T.; Yamamura, M. Pure Appl. Chem. 2013, 85, 763–776.
80. Frischmann, P. D.; MacLachlan, M. J. Chem. Soc. Rev. 2013, 42, 871–890.
81. Dietrich-Buchecker, C. O.; Sauvage, J.-P.; Kern, J.-M. J. Am. Chem. Soc. 1984, 106, 3043–3045.
82. Han, M.; Engelhard, D. M.; Clever, G. H. Chem. Soc. Rev. 2014, 43, 1848–1860.
83. Löffler, S.; Lübben, J.; Krause, L.; Stalke, D.; Dittrich, B.; Clever, G. H. J. Am. Chem. Soc. 2015, 137, 1060–1063.
84. Nakamura, T.; Kimura, H.; Okuhara, T.; Yamamura, M.; Nabeshima, T. J. Am. Chem. Soc. 2016, 138, 794–797.
85. Sugasaki, A.; Sugiyasu, K.; Ikeda, M.; Takeuchi, M.; Shinkai, S. J. Am. Chem. Soc. 2001, 123, 10239–10244.
86. Robertson, A.; Ikeda, M.; Takeuchi, M.; Shinkai, S. Bull. Chem. Soc. Jpn. 2001, 74, 883–888.
87. Liu, P.; Hisamune, Y.; Peeks, M. D.; Odell, B.; Gong, J. Q.; Herz, L. M.; Anderson, H. L. Angew. Chem. Int. Ed. 2016, 55, 8358–8362.
88. Wang, S.-P.; Shen, Y.-F.; Zhu, B.-Y.; Wu, J.; Li, S. Chem. Commun. 2016, 52, 10205–10216.
89. Hoffmann, M.; Kärnbratt, J.; Chang, M.-H.; Herz, L. M.; Albinsson, B.; Anderson, H. L. Angew. Chem. Int. Ed. 2008, 47, 4993–4996.
90. Hogben, H. J.; Sprafke, J. K.; Hoffmann, M.; Pawlicki, M.; Anderson, H. L. J. Am. Chem. Soc. 2011, 133, 20962–20969.
91. O’Sullivan, M. C.; Sprafke, J. K.; Kondratuk, D. V.; Rinfray, C.; Claridge, T. D. W.; Saywell, A.; Blunt, M. O.; O’Shea, J. N.; Beton, P. H.; Malfois, M.; Anderson, H. L. Nature
2011, 469, 72–75.
92. Sprafke, J. K.; Odell, B.; Claridge, T. D. W.; Anderson, H. L. Angew. Chem. Int. Ed. 2011, 50, 5572–5575.
93. Badjic, J. D.; Balzani, V.; Credi, A.; Silvi, S.; Stoddart, J. F. Science 2004, 303, 1845–1849.
94. Kaufmann, L.; Traulsen, N. L.; Springer, A.; Schröder, H. V.; Mäkelä, T.; Rissanen, K.; Schalley, C. A. Org. Chem. Front. 2014, 1, 521–531.
95. Carter, P. J.; Winter, G.; Wilkinson, A. J.; Fersht, A. R. Cell 1984, 38, 835–840.
96. Bisswanger, H. Enzyme Kinetics: Principles and Methods, 2nd ed.; Wiley-VCH: Weinheim, 2008.
97. Dong, Z.; Luo, Q.; Liu, J. Chem. Soc. Rev. 2012, 41, 7890–7908.
98. Kumagai, N.; Shibasaki, M. Catal. Sci. Technol. 2013, 3, 41–57.
99. Raynal, M.; Ballester, P.; Vidal-Ferran, A.; Van Leeuwen, P. W. N. M. Chem. Soc. Rev. 2014, 43, 1734–1787.
100. Imahori, T.; Kurihara, S. Chem. Lett. 2014, 43, 1524–1531.
101. Kuah, E.; Toh, S.; Yee, J.; Ma, Q.; Gao, Z. Chem. Eur. J. 2016, 22, 8404–8430.
102. Holm, R. H.; Kennepohl, P.; Solomon, E. I. Chem. Rev. 1996, 96, 2239–2314.
103. Weston, J. Chem. Rev. 2005, 105, 2151–2174.
404 Cooperativity

104. Fritsky, I. O.; Ott, R.; Pritzkow, H.; Krämer, R. Chem. Eur. J. 2001, 7, 1221–1231.
105. Scarso, A.; Scheffer, U.; Göbel, M.; Broxterman, Q. B.; Kaptein, B.; Formaggio, F.; Toniolo, C.; Scrimin, P. Proc. Natl. Acad. Sci. 2002, 99, 5144–5149.
106. Scrimin, P.; Tecilla, P.; Tonellato, U.; Veronese, A.; Crisma, M.; Formaggio, F.; Toniolo, C. Chem. Eur. J. 2002, 8, 2753–2763.
107. Kovbasyuk, L.; Pritzkow, H.; Krämer, R.; Fritsky, I. O. Chem. Commun. 2004, 880–881.
108. Scarso, A.; Zaupa, G.; Houillon, F. B.; Prins, L. J.; Scrimin, P. J. Org. Chem. 2007, 72, 376–385.
109. Takebayashi, S.; Shinkai, S.; Ikeda, M.; Takeuchi, M. Org. Biomol. Chem. 2008, 6, 493–499.
110. Gianneschi, N. C.; Masar, M. S.; Mirkin, C. A. Acc. Chem. Res. 2005, 38, 825–837.
111. Lifschitz, A. M.; Rosen, M. S.; McGuirk, C. M.; Mirkin, C. A. J. Am. Chem. Soc. 2015, 137, 7252–7261.
112. Gianneschi, N. C.; Bertin, P. A.; Nguyen, S. T.; Mirkin, C. A.; Zakharov, L. N.; Rheingold, A. L. J. Am. Chem. Soc. 2003, 125, 10508–10509.
113. Hansen, K. B.; Leighton, J. L.; Jacobsen, E. N. J. Am. Chem. Soc. 1996, 118, 10924–10925.
114. Yoon, H. J.; Kuwabara, J.; Kim, J.-H.; Mirkin, C. A. Science 2010, 330, 66–69.
115. Ouyang, G.-H.; He, Y.-M.; Fan, Q.-H. Chem. Eur. J. 2014, 20, 16454–16457.
116. Marcos, V.; Stephens, A. J.; Jaramillo-Garcia, J.; Nussbaumer, A. L.; Woltering, S. L.; Valero, A.; Lemonnier, J.-F.; Vitorica-Yrezabal, I. J.; Leigh, D. A. Science 2016, 352,
1555–1559.
117. Zhu, L.; Anslyn, E. V. Angew. Chem. Int. Ed. 2006, 45, 1190–1196.
118. Yoon, H. J.; Mirkin, C. A. J. Am. Chem. Soc. 2008, 130, 11590–11591.
119. Foy, J. T.; Ray, D.; Aprahamian, I. Chem. Sci. 2015, 6, 209–213.

You might also like