You are on page 1of 11

International Journal of Mass Spectrometry 354–355 (2013) 292–302

Contents lists available at ScienceDirect

International Journal of Mass Spectrometry


journal homepage: www.elsevier.com/locate/ijms

Fragmentation, structure, and energetics of small sodium formate


clusters: Evidence for strong influence of entropic effects
Ágnes Révész a , Tibor András Rokob a , Gábor Maász b,c , László Márk b,c , Helga Hevér d ,
László Drahos a , Károly Vékey a,∗
a
Institute of Organic Chemistry, Research Centre for Natural Sciences, Hungarian Academy of Sciences, Budapest 1025, Pusztaszeri út 59-67, Hungary
b
Department of Analytical Biochemistry, Institute of Biochemistry and Medicinal Chemistry, Medical School, University of Pécs, Pécs 7624, Szigeti út 12,
Hungary
c
János Szentágothai Research Center, University of Pécs, Pécs 7624, Ifjúság útja 20, Hungary
d
Gedeon Richter Plc., Budapest 1103, Gyömrői út 19-21, Hungary

a r t i c l e i n f o a b s t r a c t

Article history: The behavior of small cationic sodium formate clusters [NaOOCH]n Na+ was studied using a combination of
Received 3 May 2013 experimental and theoretical approaches. Energy-dependent tandem mass spectra were measured using
Received in revised form 25 June 2013 a triple quadrupole and an ion trap type instrument, while structure and energetics were obtained from
Accepted 26 June 2013
density functional theory calculations. Analysis of the relative abundances and fragmentation patterns
Available online 5 July 2013
indicate the existence of certain unstable cluster sizes, like the tetramer [NaOOCH]4 Na+ . This observation
Dedicated to the memory of Detlef is in contrast to alkali halides, where this species has enhanced stability. Surprisingly, this ‘anti-magic’
Schröder, an excellent scientist, mentor, sodium formate cluster is shown to be compact, energetically low-lying by quantum chemical calcula-
colleague, and friend. tions. The apparent contradiction can be explained by taking into account entropy effects as calculated
dissociation Gibbs free energy values show remarkably improved agreement with experimental trends.
Keywords: Fragmentation of these clusters is a good example of reactions that are predominantly directed by entropy
Sodium formate and not by energy constraints.
Cluster stability © 2013 Elsevier B.V. All rights reserved.
Magic number
Entropy
Tandem mass spectrometry
Quantum chemistry

1. Introduction their clusters, both in neutral [AX]n and ionic [AX]n A+ or [AX]n X−
forms, have been extensively studied by several experimental and
The investigation of clusters has gained special attention in the computational techniques, including mass spectrometry [3–24].
last few decades for several reasons [1,2]. They represent an inter- Important advantages of this technique over other methods are
mediate state between the condensed matter and the gas phase its sensitivity and mass selectivity, and the latter provides a
exhibiting size-dependent chemical and physical behavior. Their straightforward solution to size-dependent investigation. Mass
properties usually differ from those of both the condensed and spectrometric (MS) measurements have been carried out in com-
the gas phase, caused by their individual and specific structures. bination with various ionization modes, e.g., field desorption [10],
Therefore, the study of the clusters’ stability, reactivity and disso- secondary ion mass spectrometry [11–18], and, more recently,
ciation in naked form, i.e. without additional solvent molecules, the soft electrospray ionization (ESI) [19–23]. ESI has enabled the
may provide insight into the inherent features of the nucleation access of clusters with relatively high values of n. Relative MS
and growth mechanism of crystal structures. abundances and MS/MS fragmentation data can typically be linked
Alkali halides serve as examples for the simplest ionic crystal to energetics, i.e., thermodynamic stability of the clusters [25],
structures, most of them having face-centered cubic lattices. These whereas ion-mobility mass spectrometry provides structural infor-
compounds are inexpensive in their pure forms, and the genera- mation [24,26,27]. Early ESI-MS measurements of ionic alkali halide
tion of the gas-phase clusters can be easily achieved. As a result, clusters were aimed at the investigation of the electrospray ion
formation mechanism itself [3,6,7]. Later, the main focus shifted
to the characterization of the cluster ions in the gas phase, and it
∗ Corresponding author. Tel.: +36 1 438 1157. was found for various alkali halides (NaF [20], LiCl [19,22], NaCl
E-mail address: vekey.karoly@ttk.mta.hu (K. Vékey). [9,19,21,22], KCl [19,22], RbCl [19,22], CsCl [9,19], NaI [9,11], KI

1387-3806/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.ijms.2013.06.029
Á. Révész et al. / International Journal of Mass Spectrometry 354–355 (2013) 292–302 293

to the monoclinic crystal system, but in spite of the different sym-


metry properties of the anions and the crystals, the unit cells of
NaOOCH and NaCl show remarkable similarity with respect to the
arrangement of the ions (compare Figs. 1b and d). In light of these
similarities and differences, a closer study of the cluster ions of
NaOOCH seems of interest.
In the present work, we investigate the structure, energetics
and fragmentation properties of small [NaOOCH]n Na+ clusters by
MS and MS/MS techniques combined with computational methods.
Our aim is to unravel the differences between the alkali halide and
sodium formate clusters and to examine the origin for the absence
of magic numbers in the case of the latter.

2. Methods

2.1. Mass spectrometry

Mass spectrometric investigations were carried out by sev-


eral types of instruments. First, collision-induced dissociation
(CID) experiments were performed using a Bruker Esquire HCT
quadrupole ion trap (IT-MS) mass spectrometer (Bruker Dalton-
ics, Bremen, Germany) equipped with a conventional electrospray
(ESI) source. The [NaOOCH]n Na+ clusters were prepared by
spraying a 1:1:8 mixture of 0.1 mol/dm3 NaOH solution/10%
HCOOH/acetonitrile with a flow rate of 10 ␮l/min. The solution
Fig. 1. 3 × 3 × 1, 3 × 3 × 3 and 3 × 3 × 5 fragments cut out from the crystal lattice of
was infused through a heated capillary, set to a voltage of 3500 V;
NaCl (a–c) [42]; crystal structure of NaOOCH [43,44] (d). Cations are black, anions
are white. The lines connecting the ions within planes are only drawn to guide the the drying gas temperature was kept at 280 ◦ C, and the pressure
eye and have no special chemical meaning. Gas phase structures of [NaCl]n Na+ for of the ion trap was 8.8 × 10−6 mbar. Note that this is uncorrected
the magic numbers n = 4, 13 and 22 have very similar geometry to the parts of the gauge, and the real pressure is known to be 2–3 orders of magnitude
crystal shown here in subfigures a–c, respectively.
higher [45]. He was used as collision gas. The CID was performed
by application of an AC voltage to the end caps of the trap to induce
collisions of the isolated [NaOOCH]n Na+ (n = 2–8) cluster ions with
[11], RbI [11] and CsI [9,11,12,23]) that [AX]n A+ clusters with spe-
the helium buffer gas. The excitation period lasted 20 ms using an
cific values of n, e.g., 4, 13, 22 etc., have increased intensity in the
excitation width of 4 m/z units.
mass spectra and higher stability against dissociation compared to
The second set of MS/MS experiments was carried out on a triple
the other cluster sizes. In some cases, also [AX]n X− species were
quadrupole (QqQ) type instrument; namely, a Micromass Quattro
examined, resulting in the same observations. On the other hand,
micro mass spectrometer (Waters, Manchester, UK) bearing an ESI
monotonically decreasing abundance with increasing size was pub-
source was used. The clusters under study were generated by infus-
lished in a study for CsCl [22], and different favored values of n were
ing the solution described above with a flow rate of 10 ␮l/min. The
found and interpreted for CsI [24]. The appearance of magic clusters
capillary voltage was 3 kV, and nitrogen was used as desolvation
was attributed to their compact structure, corresponding to regu-
gas, kept at 250 ◦ C. The cone voltage was set to 25 V, which provided
lar fragments of the rock-salt crystal lattice (depicted in Fig. 1a–c).
high enough intensity and resulted in little or no in-source frag-
Quantum chemical calculations on singly charged NaCl clusters
mentation. Mass-selected sodium formate clusters collided with
[27–33] and other related systems [33–36] confirmed the assumed
argon gas having a pressure of 3.1 × 10−3 mbar in the collision
preferred structures, as did ion-mobility mass spectrometric exper-
cell. The dissociation studies were also probed on a Waters QTOF
iments [26,27]. Interestingly, [CsX]n Cs+ (X = Cl, I) clusters exhibit
Premier mass spectrometer with analogous source and collision
the same magic numbers as other alkali halides although CsCl and
conditions. It was found that practically the same results can be
CsI have different crystal structure from most other alkali halides
obtained on the QTOF equipment as on the triple quadrupole instru-
(body-centered vs. face-centered cubic). Computations suggested
ment, which is explained by the similar construction of the two
that the CsCl-type structure starts to be preferred energetically over
mass spectrometers with respect to collision induced dissociation.
the rock-salt-type only above a fairly large cluster size of several
For this reason, only the QqQ results are presented in detail in this
hundred atoms [37]. On the other hand, for neutral LiF species,
paper.
hexagonal or octogonal nanotube structures were calculated to
compete in stability with cubic forms [38]. Further, [NaCl]n Na2 2+
dications also show size-dependent stability [39,40], and in this 2.2. Quantum chemical calculations
respect, the computations agreed well with the mass spectrometric
results [41]. To obtain global minimum structures of cluster ions, sev-
Although alkali halides have been studied in detail, the mass eral global optimization techniques have been employed in the
spectrometric literature of other ionic clusters is rather scarce. The recent literature, such as simulated annealing [34,46], basin hop-
mass spectra of NaOOCH and NaOOCCH3 solutions, containing sim- ping [36], or genetic algorithm [14,28,47]. For the relatively small
ple but notably anisotropic anions, were recorded up to m/z 4000, species treated in the present work, we adopted the simpler
showing that, for these salts, the normal magic numbered species approach of selecting a large number of random starting geome-
do not dominate the spectra [19]. However, no tandem mass spec- tries and optimizing them first by molecular mechanics (MM) and
trometric investigation has been carried out, and the possible gas then by density functional theory (DFT). Specifically, 1000 (for
phase cluster structures have not been explored yet. Concerning [NaOOCH]n Na+ or [NaOOCH]n with n ≤ 2) or 10,000 (for 3 ≤ n ≤5)
the solid phase, the crystal structure of NaOOCH [43,44] belongs starting structures were generated by placing sodium and formate
294 Á. Révész et al. / International Journal of Mass Spectrometry 354–355 (2013) 292–302

ions at random positions and with random orientation into a cubic


box with an edge length of 3–5 Å, avoiding atom–atom contacts
shorter than 1.5 Å. These structures were then subjected to full
molecular mechanics minimization using the Cornell force field
[48], with charges fitted to reproduce the electrostatic potential on
the formate ion (see the Supplementary Material for details). From
the obtained optimized geometries, distinct structures having an
MM energy within 0.5 eV from that of the lowest lying isomer were
input to geometry optimization at the M05-2X/6-31G(d) level of
DFT [49–53]. For the DFT-optimized structures, all computed har-
monic frequencies were real, confirming that true minima have
been found. Again, for distinct structures with electronic energy
within 0.4 eV from that of the lowest lying isomer, more accurate
single-point electronic energy was computed with the same func-
tional and the 6-311++G(3df,3pd) basis set [54–57]. This value was Fig. 2. Positive ion mode ESI-MS spectrum of sodium formate solution in the m/z
combined with a zero-point energy calculated from the M05-2X/6- range of 20–700 measured on the QTOF instrument. Note the general trend of
decreasing intensity with size, and the deviations from the trend at n = 2–3 and
31G(d) frequencies, scaled by a factor of 0.958 [58], to obtain the
n = 4–5.
total energy at 0 K.
For the isomers considered for the large-basis-set electronic
energy computation, gas-phase Gibbs free energy was also esti- B97-D computations, respectively. The B97-D calculations were
mated by determining the thermal and entropic contributions done with the resolution-of-identity approximation [66–68].
using the standard ideal gas – rigid rotor – harmonic oscillator Molecular graphics were drawn using Molden [69] and CYLview
approximation. No frequency scaling was applied. The activa- [70].
tion Gibbs free energies for the dissociations were considered to
be equal to the dissociation free energies, which corresponds to 3. Results and discussion
approximating the transitional modes as free product translations
and rotations. Part of the dissociation free energy, namely, the 3.1. Behavior of sodium formate clusters in a mass spectrometer
entropy contribution from the free translation, depends on the
pressure. As we are trying to approximate the transition state of a The positive ion ESI-MS spectrum of sodium formate solution
unimolecular dissociation, there is no reason to identify this pres- measured on the QTOF instrument in the m/z range of 20–700 is
sure with the actual pressure in the mass spectrometer. It may be presented in Fig. 2. The appearance of peaks at m/z = 91, 159, 227,
rather considered as an adjustable parameter, connected with the 295, 363, 431, 499, 567, 635 clearly points to the formation of clus-
looseness of the transition states. Changing the pressure alters all ter ions of the type [NaOOCH]n Na+ (n = 1–9). The relative intensity
dissociation free energies by an additive constant of RT ln p2 /p1 ; is essentially monotonously decreasing with cluster size, i.e., with
hence, the dissociation free energy trends are not affected by any increasing n, which is in agreement with the previously published
specific choice. Concerning the absolute magnitudes, it was found results and stems partially from the higher instrumental sensitiv-
that, with dissociation free energies at p = 1 atm, reasonable rate ity toward lower m/z values. However, it is in marked contrast with
constants at reasonable temperatures can be obtained for the ion the ESI-MS of alkali halide clusters, where magic numbered clus-
trap experiments. Specifically, we determined the temperature for ters, including the n = 4 species, are present with very high relative
each cluster at which the dissociation rate as obtained from the abundance in the spectra. Peaks with somewhat increased inten-
computed dissociation Gibbs free energy via transition state the- sity also appear in the case of sodium formate, but these correspond
ory corresponds to 50% fragmentation after 20 ms. The average of to the cluster cations with n = 3 and 5, i.e., having m/z = 227 and
these temperatures was ∼500 K. Therefore, we report Gibbs free 363. However, the overall appearance of the spectra can be better
energy data for T = 500 K and p = 1 atm throughout. described by stating that the [NaOOCH]4 Na+ and [NaOOCH]2 Na+
For the energetically lowest-lying isomer of all species, test cal- cluster ions have somewhat decreased abundance compared to the
culations with the spin-component scaled MP2 (SCS-MP2) [59] neighboring elements of the cluster series. The mass spectra mea-
with the G3Large triple-␨ basis set [60] were also carried out (results sured on the IT and QqQ instruments show identical trends. The
shown in the Supplementary Material). All trends and conclusions observed remarkable difference in the mass spectrum induced by
based on the DFT data are also supported by the SCS-MP2 results, the change of halide anion to the anisotropic formate prompted us
and even the agreement between the numerical values from the to further investigate the stability and fragmentation properties of
two methods is very good. small sodium formate cluster ions by tandem mass spectrometric
In order to find approximate isomerization and dissociation methods.
pathways, the potential energy surface was mapped using relaxed In order to obtain a more detailed picture, the MS/MS exper-
scans at the B97-D/def2-TZVP [61–63] level of DFT. For each iments were carried out on two different instruments: on an ion
point of a scan, harmonic potentials with force constants of trap and on a triple quadrupole. In both cases, parameters influenc-
∼100–500 eV/Å2 were employed to restrain certain atom–atom ing the amount of energy introduced into the ions were varied. In
distances at specific values, and the geometry was relaxed in 10–25 particular, the fragmentation amplitude on the IT and the collision
optimization cycles. The restraint distances were then repeatedly voltage in the collision cell on the QqQ were varied. The motivation
shifted by small amounts, typically ±0.05–0.1 Å, to obtain the sub- for employing two setups stems from the differences [71] between
sequent points. The pathways located in this way do not correspond the excitation processes occurring in the two types of MS. The most
to the minimum energy pathways of the isomerization or disso- relevant one is that, in IT instruments, the parent ion is selectively
ciation processes, but they do provide an upper estimate of the energized, i.e., the fragmentation of daughter ions is uncommon.
involved barriers. In QqQ mass spectrometers, all ions are excited, facilitating con-
The M05-2X and SCS-MP2 calculations were done with Gaussian secutive dissociations. Further, in IT instruments, the energy of the
09 [64], while for B97-D, Turbomole 5.10 was employed [65]. The actually dissociating ions practically does not change with increas-
“ultrafine” and the “m3” grids were chosen for the M05-2X and ing excitation; it is always slightly above the dissociation threshold.
Á. Révész et al. / International Journal of Mass Spectrometry 354–355 (2013) 292–302 295

Increasing the fragmentation amplitude only leads to more and not lead to this fragment ion. These results further support that
more precursor ions fragmenting. This property of IT is the conse- the observed phenomena are not artifacts linked to a specific type
quence of the large number of collisions with small kinetic energy, of instrument, but they reflect the inherent behavior of sodium
i.e., the so-called “slow heating” [72–74]. In QqQ instruments, sig- formate clusters.
nificantly higher excess energy can be deposited in the ions before
fragmentation. Finally, the time frame of the experiment also dif- 3.2. Calculated structural features
fers, being ∼10−2 s in an IT whereas only 10−5 –10−4 s in a QqQ.
In experiments on both IT and QqQ instruments, the The above experimental analysis highlighted the preferred frag-
[NaOOCH]n Na+ (n = 2–8) clusters show losses of various numbers mentation pathways and the notably different behavior of NaOOCH
of sodium formate units (e.g., monomer loss: m = 68, dimer loss: clusters from that of NaCl. For alkali halides, it was usually found in
m = 136, etc.) at elevated collision energies, resulting in smaller the literature that the computed structures and dissociation ener-
clusters. From the tandem mass spectra recorded at various colli- gies or average binding energies can account well for the observed
sion energies, the breakdown diagrams, i.e., the relative abundance magic numbers [29–32,34,35]. Hence, we also turned to compu-
of precursor and fragment ions as a function of the extent of exci- tational chemistry to obtain an atomic-level description of the
tation, were constructed. These data are shown in Fig. 3, and the involved sodium formate species and to correlate the observed
branching ratios at high excitation on the IT are collected in Table 1. stabilities in the spectrometer with the energetics of the clusters.
Upon inspection of the curves obtained on the IT instrument, Density functional theory was used to optimize geometries and
depicted in Fig. 3 in the left column, several conclusions can derive zero-kelvin internal energies, as a measure of the inherent
be drawn. First, the preferred fragmentation process varies from stability, for the lowest-lying isomers of the involved ions and neu-
cluster to cluster. For example, the [NaOOCH]3 Na+ cluster loses tral species. The unbiased choice of initial structures was ensured
mainly monomer and dimer in an approx. 4:1 ratio, resulting in by a random starting geometry generation. The M05-2X exchange-
[NaOOCH]2 Na+ and [NaOOCH]Na+ fragments. For the next larger correlation functional was chosen on the basis of its high accuracy
[NaOOCH]4 Na+ ion, one major fragment appears belonging to for the thermochemistry of main-group compounds. Specifically, a
monomer loss ([NaOOCH]3 Na+ ), whereas the [NaOOCH]5 Na+ clus- mean unsigned error of 0.17 eV was reported in the literature for a
ter does not dissociate by losing a monomer; rather, [NaOOCH]3 Na+ set of diverse chemical transformations [75]. Considering the sim-
and [NaOOCH]2 Na+ fragment ions are observable, which cor- ple electronic structure of the sodium formate clusters and the error
respond to the leaving of neutral dimer and trimer units. By cancelation in the relative energetics of species of similar nature,
considering all the obtained breakdown diagrams, it can be con- we expect the maximum errors in the energetics described below
cluded that the clusters usually tend to lose smaller neutral not to exceed 0.2–0.3 eV, and the errors in differences in stability or
fragments, e.g., monomer, dimer. Thus, the extra sodium ion, and dissociation energy to be even smaller. This assumption is further
therefore the charge, remains on the larger cluster fragment. There supported by test calculations using the SCS-MP2 wavefunction-
is one notable exception, namely, the [NaOOCH]4 Na+ cluster ion is based method (shown in the Supplementary Information), which
not observable as fragment in considerable amount for any of the do not differ more than 0.2 eV from the DFT results and predict the
investigated clusters. [NaOOCH]5 Na+ dissociates mainly via dimer same trends in dissociation energies as DFT.
loss instead, and, in the case of [NaOOCH]6 Na+ , even the com- The computations were carried out for the [NaOOCH]n Na+
plementary [NaOOCH]2 Na+ is somewhat larger in intensity than (n = 1–5) ions and for the neutral fragmentation products
the [NaOOCH]4 Na+ fragment. [NaOOCH]7 Na+ produces the com- [NaOOCH]n (n = 1–4). The most stable structures, computed via
plementary [NaOOCH]3 Na+ cluster in roughly the same intensity as M05-2X to lie within a 0.1 eV energy window, are shown in
the other two major fragments, [NaOOCH]6 Na+ and [NaOOCH]5 Na+ , Fig. 4 and compared below with literature results on [NaCl]n
and no [NaOOCH]4 Na+ is formed. The same holds true for the one [29,30,32,47,76–78] and [NaCl]n Na+ [29–32].
unit bigger cluster, [NaOOCH]8 Na+ , as well. A comparison of the The smallest neutral species, the NaOOCH ion pair, shows the
breakdown curves of [NaOOCH]4 Na+ with the other clusters reveals expected in-plane coordination of the Na+ to both oxygens of the
that this ion is not extraordinarily prone to dissociation; instead, formate anion. The Na–O distances are 2.18 Å; for larger clusters,
its formation seems unfavored. Hence, in contrast to alkali halides, the distances between Na and O directly in contact show some
where n = 4 is a magic number, it has rather an anti-magic character variation (∼2.0–2.7 Å). This ion pair can be considered as the basic
in case of sodium formate clusters, resulting in the [NaOOCH]4 Na+ building block of the crystal structure of solid NaOOCH, consisting
cluster ion not only having slightly lower intensity in the normal of layers of NaOOCH ion pairs, with parallel alignment within lay-
ESI mass spectrum but also being practically absent as fragment in ers and antiparallel between neighboring layers (see Fig. 1d). The
the CID of larger cluster ions. It is to be noted that the employed most stable structure of [NaOOCH]2 is the planar antiparallel dimer,
mass spectrometers are not designed for the sampling of low mass in complete analogy with [NaCl]2 or other alkali halides. Related
ions and might thus significantly discriminate against the bare Na+ arrangement of two NaOOCH units from two neighboring layers
ion. Therefore, we decided not to monitor this fragment at all and can also be identified in the crystal structure although interaction
recorded the spectra only from m/z = 50. On the basis of our com- with surrounding ions leads to marked deviation from planarity. In
putational results, discussed below, preference for the formation of contrast to the monomer and the dimer, the neutral trimer shows
Na+ seems improbable for n ≥ 3. a compact, truly three-dimensional structure, which is not similar
At first sight, the measurements carried out on the triple to any part of the crystal. Apparently, the absence of the crystal
quadrupole mass spectrometer (Fig. 3, right column) notably dif- packing forces allows the system to avoid the less favorable H· · ·Na
fer due to the presence of consecutive dissociation channels. They contacts and to keep the number of O· · ·Na contacts high. Indeed, 6
demonstrate very clearly the significance of consecutive reactions oxygens from 5 formates coordinate a single Na+ in the crystal, but
in a QqQ instrument, in contrast to their unfavored nature in IT even in the smallest clusters, 3–4 oxygens around Na+ are typical.
instruments, which is the expected behavior. In spite of the con- As a quite general consequence, the occurrence of a Na+ coordi-
siderable differences with respect to ion activation, the identity of nated to both oxygens of two or more formates (a “multiple-double
primary fragments, as well as their relative ratio at the 50% survival coordination”) is a typical feature for small clusters, but in the crys-
yield, agrees well in the two sets of experiments. In particular, the tal, each Na+ is in close contact with both oxygens of only a single
unfavored [NaOOCH]4 Na+ cluster is not produced in the QqQ type formate. For [NaCl]3 , the most stable structures are the planar six-
instrument either; even the consecutive dissociation processes do atom ring and the planar 2 × 3 × 1 fragment of the crystal structure.
296 Á. Révész et al. / International Journal of Mass Spectrometry 354–355 (2013) 292–302

Fig. 3. Breakdown curves of [NaOOCH]n Na+ (n = 3–8) clusters constructed from the tandem mass spectra measured on the ion trap instrument (left column) and on the
triple quadrupole instrument (right column). Note that the fragment Na+ (m/z = 23) was not taken into account. Furthermore, the [NaOOCH]2 Na+ cluster produces only the
daughter ion [NaOOCH]Na+ (m/z = 91); thus, its diagram is not shown.
Á. Révész et al. / International Journal of Mass Spectrometry 354–355 (2013) 292–302 297

Table 1
Branching ratios of dissociation processes of [NaOOCH]n Na+ clusters as obtained from the ion trap measurements. The percentages denote the relative abundance of the
different channels at high extent of excitation, i.e., when equilibrium has already been reached between the various losses.

m/z Composition Relative abundance of losses in %

Monomer Dimer Trimer Tetramer Pentamer Hexamer Heptamer


+
159 [NaOOCH]2 Na 100
227 [NaOOCH]3 Na+ 77 23
295 [NaOOCH]4 Na+ 88 11 1
363 [NaOOCH]5 Na+ 0 75 25 0
431 [NaOOCH]6 Na+ 31 4 48 17 0
499 [NaOOCH]7 Na+ 25 36 1 31 7 0
567 [NaOOCH]8 Na+ 6 41 26 2 22 2 0

Their NaOOCH analogs were also found by our calculations, but exhibit a cubic arrangement of the ions, with minor variation due
they lie 0.30 eV and 0.22 eV higher in energy than the most sta- to the versatility of formate coordination mode. However, the gas-
ble isomer, respectively. The similarity to NaCl is partially restored phase [NaCl]4 cube directly corresponds to a 2 × 2 × 2 fragment of
for the tetramer [NaOOCH]4 , where all located low-lying isomers the crystal, while the orientation of formates in all shown isomers

Fig. 4. Computationally identified isomers of [NaOOCH]n and [NaOOCH]n Na+ within an energy window of 0.1 eV from the most stable structure. Cations are black, anions
are white. Relative zero-kelvin energies with respect to the lowest-lying structure are given in eV. Dashed and dotted lines separate smaller fragments that make up certain
larger structures (see text for details).
298 Á. Révész et al. / International Journal of Mass Spectrometry 354–355 (2013) 292–302

of [NaOOCH]4 is markedly different from that in the solid-phase most favored one, or it is very close to being the most favored. The
structure. monomer loss energies seem to converge to a value around 1.95 eV
In the investigated size range, the positively charged clusters with increasing cluster size, but a definite conclusion in this respect
tend to have more low-lying isomers than the neutral species. would necessitate the study of a larger size interval.
Already for [NaOOCH]Na+ , we could identify two essentially isoen- The computational results reveal that the small [NaOOCH]n Na+
ergetic structures, as a result of the competing preferences for Na clusters with even n values are more stable, i.e., they require more
coordination to both oxygens and for increased Na· · ·Na distances. energy for dissociation, than species with odd n. This might be the
[NaOOCH]2 Na+ is exceptional, having a single isomer within the energetic consequence of the well-defined and compact structures
0.1 eV window, containing a central Na+ over the middle of a six- for the even-n species. Interestingly, our MS studies showed that
membered Na–OO–Na–OO ring. The remarkable stability of this the n = 4 (m/z 295) and the n = 2 (m/z 159) cluster ions have low
“dome” arrangement is corroborated by its appearance as the build- abundance, and in MS/MS, the m/z 295 ion is not produced from the
ing block of larger clusters, but it does not appear as element of fragmentation of the mass-selected higher clusters. On the basis of
the crystal. A low-lying related structure for [NaCl]2 Na+ is an equi- the correlation of the computed energetic stability with structural
lateral triangle of three Na+ with chlorides above and below the features and the above-discussed expected accuracy of the com-
center, but the “dome” contains rather an isosceles Na3 triangle putations, we consider it very unlikely that the computed trends
with one markedly longer side. The analogs of other [NaCl]2 Na+ would be in error. It can thus be concluded that the observed behav-
structures, i.e., the linear chain (5 × 1 × 1 fragment of the crystal) ior is not determined solely by the energies, but other factors also
or the [NaCl]2 square with a pendant Na+ at one corner, are more play an important role.
than 0.5 eV higher for sodium formate. For the next larger cluster,
[NaOOCH]3 Na+ , we found two types of isomers. A central Na+ inter- 3.4. Investigation of fragmentation pathways
acting with five or six oxygens of an almost planar nine-membered
ring is the characteristic of the more stable isomers (A and B). These One possible reason for the disagreement between the trends
are conceptually similar to the most stable geometry of [NaCl]3 Na+ in experiments and in the calculated adiabatic dissociation ener-
although the latter is much less planar due to the small size of chlo- gies is that some dissociations may not be able to produce directly
ride. Somewhat higher in energy, variations of the “dome” structure the lowest energy isomer, and the involved rearrangements may
with an attached [NaOOCH] unit appear (C and D, consider the pose significant energetic barriers. In order to provide a detailed
dashed lines). For all these geometries, the presence of “multiple- description of all dissociation processes, a full map of the cor-
double” coordination excludes the analogy with the solid-phase responding potential energy surface would be necessary. Related
structure. studies exist in the literature in which the isomerization PES was
For sodium chloride, the cationic n = 4 cluster was found to be explored in detail for [LiF]4 [46] and [NaCl]35 Cl− [79]. Such calcu-
a “magic” one, with a planar structure corresponding to a 3 × 3 × 1 lations are beyond the scope of the present work; however, we
slab of the rock-salt lattice. Related local minima for [NaOOCH]4 Na+ carried out simple PES scans to examine whether there are sig-
appear to be missing; instead, the most stable structure (A) is nificant potential energy barriers hindering the dissociations of the
remarkably compact, and it can be described either as the com- most stable isomers of the clusters to the most stable isomers of the
bination of two “domes” with a common Na+ (consider the dashed next smaller cluster and a [NaOOCH] fragment. The energy profiles
line), or as a [NaOOCH]4 cube with an extra, “quadruply-doubly” are shown in the Supplementary Information, only a brief summary
coordinated Na+ in the middle (see the dotted line). The size of is given here. As already apparent from its structure, [NaOOCH]2 Na+
the formate ion seems to play key role in allowing an extra cation can directly dissociate into [NaOOCH] and A-[NaOOCH]Na+ , and no
inside the cube; for [NaCl]4 Na+ , only isomers with Na+ attached to a barrier higher than the endothermicity of the reaction was found.
[NaCl]4 cube via one or two anions from outside were found. A “cube For A-[NaOOCH]3 Na+ , a rearrangement via a barrier of ∼0.8 eV
plus cation outside” arrangement is exhibited by B-[NaOOCH]4 Na+ can occur, which leads to a structure consisting of the “dome” of
(consider the dashed line), but it is somewhat less stable than the [NaOOCH]2 Na+ and a [NaOOCH] unit connected via its edge. This
“combined domes” structure. B-[NaOOCH]4 Na+ can also be inter- structure, ∼0.4 eV higher than the most stable isomer, can then
preted as a folded B-[NaOOCH]3 Na+ structure with an attached directly lose [NaOOCH], again without a barrier higher than the
[NaOOCH] ion pair (see the dotted line). endothermicity. The compact A-[NaOOCH]4 Na+ can interconvert
The [NaOOCH]5 Na+ cluster has a less well-defined structure, with B-[NaOOCH]4 Na+ in a process with an activation energy of
with several similar stable isomers. This feature, as well as the ∼0.7 eV. In accordance with expectations based on its structure,
actual geometries, is reminiscent to [NaCl]5 Na+ . Structures A, B, the direct dissociation of B-[NaOOCH]4 Na+ to A-[NaOOCH]3 Na+
and D can be best described as a [NaOOCH]4 cube with an attached and [NaOOCH] is again feasible without extra barrier. Hence, for
A-[NaOOCH]Na+ unit, while C and E better correspond to A- n = 2–4, the monomer loss can always proceed with a critical
[NaOOCH]3 Na+ with an attached [NaOOCH]2 dimer (dashed lines); energy equal to the adiabatic dissociation energy. For n = 5, i.e.,
nevertheless, alternative “assignments” exist, and a clear-cut clas- A-[NaOOCH]5 Na+ , we only examined a reaction pathway toward B-
sification to two families cannot be made. [NaOOCH]4 Na+ , as this route seemed to be easier to follow with the
employed computational technique. Indeed, we found that simul-
3.3. Calculated energetics of the fragmentation processes taneous rearrangement and expulsion of a [NaOOCH] fragment
can connect A-[NaOOCH]5 Na+ with B-[NaOOCH]4 Na+ . Although B-
Having located the most stable isomers for all involved [NaOOCH]4 Na+ is 0.10 eV higher than the most stable n = 4 cluster
species, one can calculate the energies required to dissociate the A-[NaOOCH]4 Na+ , this result is sufficient to prove that for n = 5, the
[NaOOCH]n Na+ clusters to a smaller cationic cluster and a neu- monomer loss is still preferred energetically over the dimer loss
tral [NaOOCH]k fragment. We assume “adiabatic” dissociations, i.e., (1.90 + 0.10 < 2.14), and it occurs at smaller threshold energy than
we suppose that the fragmentation starts from the lowest-lying the monomer loss from n = 4 (1.90 + 0.10 < 2.04).
structure of the parent ion and yields the most stable isomers of In short, we found that rearrangements seem indeed necessary
the daughter ion and the neutral part. The results are shown in for the “adiabatic” dissociations, but the involved barriers are low.
Table 2 (see the non-parenthesized numbers). In agreement with The threshold energies of the dissociations are not influenced to an
the expectations, loss of a smaller neutral fragment is typically pre- extent which would explain, most importantly, the experimentally
ferred; hence, the monomer loss channel is either the energetically observed preference for the dimer loss from n = 5, and thereby, the
Á. Révész et al. / International Journal of Mass Spectrometry 354–355 (2013) 292–302 299

Table 2
0 K dissociation energies and 500 K dissociation Gibbs free energies (the latter in parentheses) of [NaOOCH]n Na+ clusters for the loss of neutral [NaOOCH]k species, computed
at the M05-2X/6-311++G(3df,3pd)//M05-2X/6-31G(d) level as the difference between the energies of the most stable isomers of the fragments (according to the corresponding
energy scale). The lowest energy channels are emphasized with boldface numbers.

m/z Composition Computed 0 K dissociation energy (500 K Gibbs free energy) (eV)

Monomer loss Dimer loss Trimer loss Tetramer loss


+
91 [NaOOCH]Na 1.77 (1.40)
159 [NaOOCH]2 Na+ 2.17 (1.16) 2.14 (1.42)
227 [NaOOCH]3 Na+ 1.85 (1.37) 2.21 (1.40) 2.32 (2.07)
295 [NaOOCH]4 Na+ 2.04 (0.71) 2.09 (0.94) 2.60 (1.38) 2.53 (1.87)
363 [NaOOCH]5 Na+ 1.90 (1.14) 2.14 (0.71) 2.33 (1.36) 2.66 (1.61)

existence of the “anti-magic” cluster n = 4. From the calculations, Using the discussed protocol, we estimated the Gibbs free ener-
it thus seems that neither the thermochemical stability, nor the gies of all considered isomers of the [NaOOCH]n Na+ ions and the
related energetic barriers can explain the observed abundance and [NaOOCH]k neutral fragments. A single temperature value was
fragmentation patterns of sodium formate clusters. employed throughout, chosen such that the resulting predicted dis-
sociation free energies yield rate constants that roughly correspond
to the experimental time scale of the fragmentations in the ion trap
3.5. Computational estimation of entropic effects experiments. The obtained value, 500 K, is somewhat low but not
unrealistic for activated ions in the ion trap.
Deviation from the stability pattern predicted from dissociation In line with the suggestions for [NaF]n Na+ , our computational
energies was found in the literature for the fission of metastable results highlight that the entropy notably alters the stability
Nan 2+ clusters [80], and for [NaF]n Na+ clusters produced by the order of the isomers of a given cluster. As a general trend, more
gas aggregation technique [36]. In these studies, statistical (i.e., extended (“less ordered”) structures become favored, particularly
entropic) effects were proposed to play important role in determin- for larger cluster sizes. For the neutral [NaOOCH] and [NaOOCH]2 ,
ing cluster stabilities and preferred fragmentation pathways, and the structure with the lowest Gibbs free energy is the still same
such effects may also be involved in the case of sodium formate. A as that with the lowest zero-kelvin internal energy. However, for
rigorous analysis of the influence of entropy in a mass spectromet- [NaOOCH]3 , a flatter, looser structure a-[NaOOCH]3 (see Fig. 5)
ric experimental setting would require a master equation modeling has the lowest G, 0.06 eV below [NaOOCH]3 in Fig. 4. The cubic
of the time-dependent internal energy distribution of the dissoci- structure is preferred also in terms of G for [NaOOCH]4 , but the
ating ions. Reasonable models would be required for the density of most stable isomer becomes the more open B-[NaOOCH]4 . For
states arising from the coupled, anharmonic internal motions, for the charged [NaOOCH]Na+ , there is no change in the order of iso-
the energy exchange upon energizing collisions, and for the energy- mers, but G between A and B is 0.11 eV, significantly larger than
dependent unimolecular rate constants (e.g., RRKM theory). As the the zero-kelvin energy difference. The “dome” of [NaOOCH]2 Na+
goal of the present study was only to assess the importance of remains favored in G as well. For n = 3 and n = 4, the symmetri-
entropic effects, we decided to employ several approximations. cal planar B-[NaOOCH]3 Na+ and the “cube plus cation outside”
First, a thermal equilibrium situation was assumed, implying that B-[NaOOCH]4 Na+ become the lowest-lying structures, respectively.
one can use free energy values to characterize the dissociations. Finally, for n = 5, entropic effects lead to the preference for the
This is probably a realistic assumption in the case of ion trap instru- extended a-[NaOOCH]5 Na+ isomer (Fig. 5), surmounting a E0K of
ments [72]. Second, we considered the activation free energy for the 0.29 eV.
fragmentation to be equal to the dissociation free energy, corre- Having located the structures with lowest G, we can calculate
sponding to the completely separated fragments. In contrast, the the appropriate free energy differences to obtain the dissociation
true transition state in terms of free energy is located at some G values of the [NaOOCH]n Na+ ions, again in the “adiabatic”
finite value of the dissociation reaction coordinate. Our assump- sense. The values are reported in Table 2 (numbers in parentheses).
tion is equivalent to treating the transitional degrees of freedom Apparently, the trends differ notably from those of the zero-kelvin
as free translation and rotation of the products, which definitely internal energies and can account for some of the experimental
introduces error. Rearrangements may also represent significant observations. Accordingly, the monomer loss has the lowest bar-
entropic bottlenecks even if they are not associated with energetic rier for all clusters, except for [NaOOCH]5 Na+ , which is correctly
barriers. Third, we employed the rigid-rotor-harmonic-oscillator predicted to strongly prefer dimer loss, avoiding the anti-magic
(RRHO) approach to model the molecular motions. An error of 1–2% n = 4 cluster upon fragmentation. However, the monomer loss still
in total entropies may be typical for the RRHO procedure even for
rigid molecules [81], leading to errors on the order of 0.1 eV in the
free energies at the present temperature and molecule sizes. The
more strongly anharmonic low-frequency modes of larger clusters
may further increase the error bar. Nevertheless, an inspection of
the structures shows that these modes are still probably better
treated as vibrations than as hindered internal rotations, for which
the RRHO approach would yield large inaccuracies [82–84]. Some
error compensation upon calculating the free energy differences
can also be expected. For these reasons, we refrained from the com-
putationally intensive estimation of hindered rotor or anharmonic
corrections [85]. While all these approximations limit the accuracy
of the treatment of entropic effects, we still expect that the free
energies computed in this way can correctly reflect many qualita-
tive trends that could be obtained from a more rigorous treatment
within a microcanonical framework. Fig. 5. Lowest Gibbs free energy isomers of [NaOOCH]3 and [NaOOCH]5 Na+ .
300 Á. Révész et al. / International Journal of Mass Spectrometry 354–355 (2013) 292–302

range of 550–700 K depending on the extent of activation, expec-


tedly higher than the 500 K obtained from the quantum chemical
treatment due to the assumption of tighter transition states. At
650 K, which roughly corresponds to a 50% survival yield of the
parent ion, the absence of the monomer loss in the experiments
can only be reproduced if the activation free energy of this chan-
nel is larger than 1.66 eV, corresponding to an activation entropy
at least ∼60 J/mol K less than that of the dimer loss channel. For the
dimer and trimer loss channels, activation free energies of 1.48 eV
and 1.55 eV were obtained from the fit, respectively. The corre-
sponding quantum chemical dissociation free energies for 650 K
are 0.98 eV, 0.37 eV and 1.11 eV. These values are understandably
Fig. 6. MassKinetics modeling of the breakdown curves of [NaOOCH]5 Na+ . Circles lower than those from the fit; more importantly, the preference
are experimental data, crosses show results from the fitting. Thermal model was for the dimer loss is exaggerated, and the monomer loss is incor-
used, with a reaction time of 20 ms; a single set of activation parameters for the rectly predicted as the secondary channel, most probably due to
whole figure, and an individual temperature value at each fragmentation amplitude
inaccuracy of entropy estimation. Nevertheless, the conclusion is
were fitted.
clear from both the experimental and computational approaches:
The monomer loss channel does not appear due to its entropically
appears as the second most favorable channel, while experimen- unfavored nature.
tally, trimer loss is observed instead. Furthermore, the absolute
values of the dissociation free energies of the lowest-G chan-
nels show significant variation (0.71–1.40 eV). Specifically, the n = 3 4. Conclusions
cluster seems to have a larger free energy barrier than n = 4 and n = 5,
while experimentally, there is no significant difference between the In the present work, cationic sodium formate clusters
amount of excitation necessary to fragment these clusters. In short, ([NaOOCH]n Na+ ) were subjected to a combined experimental (ESI-
there are several indications that our calculations do not model MS, MS/MS) and computational study. As the crystal structure of
accurately the actual conditions in the ion source and during dis- sodium formate is similar to the extensively investigated alkali
sociation, but the consideration of the Gibbs free energies instead halides, we expected to find many analogies, but we found marked
of zero-kelvin energies still significantly improves the qualitative differences instead. Here, we summarize the main findings in a
agreement with experiments. As the computed 500 K enthalpies qualitative manner, focusing on the tetramer, which is probably
(data in Supplementary Information) exhibit essentially the same the most characteristic example.
behavior as the zero-kelvin energies, we can conclude that entropic Accurate quantum chemical computations show that minimum
effects play a key role in shaping the observable stability pattern of energy conformations of small neutral and cationic sodium formate
sodium formate clusters. clusters are often much different from those expected based on
the crystal structure and also from analogous alkali halide clus-
3.6. Experimental estimation of activation free energy differences ters. Specifically, the structure of [NaCl]4 Na+ resembles a planar
3 × 3 × 1 slice of the crystal, with a sodium atom in the center,
Activation parameters of processes occurring in the mass while [NaOOCH]4 Na+ exhibits a compact, cubic arrangement of the
spectrometer can be estimated by fitting rate theories to the exper- ions, with one sodium in the middle. At the same time, their ener-
imental data although errors can easily reach ∼20% [86]. In order to getic properties are similar. [NaCl]4 Na+ is a stable configuration,
provide a preliminary data analysis of the dissociation processes, resistant to mass spectrometric fragmentation, which results in a
we employed the MassKinetics [87] software to model the disso- ‘magic’ cluster. According to quantum chemistry, fragmentation of
ciation of [NaOOCH]5 Na+ (m/z 363) in the ion trap experiments. In [NaOOCH]4 Na+ also requires high critical energy, higher than that
accordance with the slow-heating property, a thermal equilibrium of either the trimer or the pentamer. This would suggest that, in
was assumed in the modeling. The monomer, dimer, and trimer mass spectrometry, this species would also be a particularly stable,
loss processes were considered, producing the fragments with m/z i.e., magic, cluster.
295, 227, 159, respectively. In the quantum chemical investigation To investigate the behavior experimentally, we measured the
of a series of dissociations, we always found rearrangement barri- electrospray ionization mass spectrum of the [NaOOCH]n Na+
ers to be notably lower than the dissociation energies (see above). clusters and also studied their collision-induced dissociation. Frag-
Therefore, we think it is reasonable to suppose the potential energy mentation of the charged clusters takes place by the loss of a
barriers of these three losses to be equal to the computed zero- neutral monomer or oligomer unit, providing useful information
kelvin dissociation energies (i.e., data from Table 2). The activation on the stability of the clusters: Which clusters are stable, and
entropy for the dimer loss was estimated using the computed which fragment easily, and among parallel processes, which ones
harmonic frequencies, but in this case, transitional modes were are predominant. For [NaOOCH]4 Na+ , experiments show the exact
treated as low-frequency vibrations instead of free translations and opposite of what was expected from calculations: (1) in the ESI
rotations, which corresponds to lower activation entropies. Indi- spectrum, the tetramer has smaller intensity than either the trimer
vidual temperature values corresponding to each fragmentation or the pentamer. (2) Fragmentation of higher oligomers avoids the
amplitude, and the activation entropies of the other two processes tetramer product; e.g., the pentamer loses a neutral dimer and
(more precisely, frequencies of the 6 transitional modes in the not a monomer unit. All of these indicate that [NaOOCH]4 Na+ is
two transition states; common at all amplitudes) were adjusted less stable than neighboring clusters, behaving as an ‘anti-magic’
to obtain the best fit of the experimentally observed branching cluster.
ratios. It was found that the relative abundances can be well fitted The seeming contradiction between calculation and experi-
at each considered fragmentation amplitude (see Fig. 6). The result- ments can be resolved by considering entropy effects. Fragmen-
ing temperature monotonously increases and then levels off with tation of sodium formate clusters requires ca. 2 eV of energy. To
increasing fragmentation amplitude (not shown), thus exhibiting drive fragmentation in the timeframe of the mass spectrometric
qualitatively correct behavior. The temperature values were in the experiments, one has to put a significant amount of internal energy
Á. Révész et al. / International Journal of Mass Spectrometry 354–355 (2013) 292–302 301

into the clusters. In ion trap instruments, this can be reasonably [17] K. Yokoyama, N. Haketa, M. Hashimoto, K. Furukawa, H. Tanaka, H. Kudo, Pro-
described by a temperature, which has to be on the order of ∼500 K duction of hyperlithiated Li2 F by a laser ablation of LiF–Li3 N mixture, Chemical
Physics Letters 320 (2000) 645.
according to our calculations. At this temperature, entropic dif- [18] F.A. Fernández-Lima, C.R. Ponciano, H.D.F. Filho, E. Pedrero, M.A.C. Nascimento,
ferences provide a sizable contribution to Gibbs free energy. One E.F. da Silveira, UV laser induced desorption of CsI and RbI ion clusters, Applied
can expect, and calculations did show, that the energetically stable, Surface Science 252 (2006) 8171.
[19] C. Hao, R.E. March, T.R. Croley, J.C. Smith, S.P. Rafferty, Electrospray ionization
compact [NaOOCH]4 Na+ cluster is entropically strongly unfavored, tandem mass spectrometric study of salt cluster ions. Part 1. Investigation of
as compared to, e.g., the flat and extended [NaOOCH]3 Na+ . As a alkali metal chloride and sodium salt cluster ions, Journal of Mass Spectrometry
result, the energetically very stable [NaOOCH]4 Na+ cluster becomes 36 (2001) 79.
[20] M.P. Ince, B.A. Perera, M.J. Van Stipdonk, Production, dissociation, and gas phase
unstable on the Gibbs free energy scale. This explains why this clus-
stability of sodium fluoride cluster ions studied using electrospray ionization
ter is avoided in the fragmentation of larger ones, and why it has a ion trap mass spectrometry, International Journal of Mass Spectrometry 207
small intensity in ESI spectra. (2001) 41.
[21] A.T. Blades, M. Peschke, U.H. Verkerk, P. Kebarle, Hydration energies in the gas
In short, sodium formate cluster fragmentation is dominated not
phase of select (MX)m M+ ions, where M+ = Na+ , K+ , Rb+ , Cs+ , NH4 + and X− = F− ,
by energy, but entropy effects, in contrast to alkali halides. Although Cl− , Br− , I− , NO2 − , NO3 − . Observed magic numbers of (MX)m M+ ions and their
energetically preferred, the tetramer becomes an ‘anti-magic’ clus- possible significance, Journal of the American Chemical Society 126 (2004)
ter in the ESI spectrum and in dissociation processes due to its low 11995.
[22] A. Wakisaka, Nucleation in alkali metal chloride solution observed at the cluster
entropy. level, Faraday Discussions 136 (2007) 299.
[23] A.S. Galhena, C.M. Jones, V.H. Wysocki, Influence of cluster size and ion activa-
tion method on the dissociation of cesium iodide clusters, International Journal
Acknowledgments of Mass Spectrometry 287 (2009) 105.
[24] F.A. Fernandez-Lima, C. Becker, K. Gillig, W.K. Russell, M.A.C. Nascimento, D.H.
Russell, Experimental and theoretical studies of (CsI)n Cs+ cluster ions produced
Useful discussions with Dr. András Stirling are gratefully by 355 nm laser desorption ionization, Journal of Physical Chemistry A 112
acknowledged. T.A.R. was supported by the János Bolyai Research (2008) 11061.
Scholarship of the Hungarian Academy of Sciences. Financial [25] J.A. Alonso, M.J. López, Growth Ability, Stability indices of clusters, Journal of
Cluster Science 14 (2003) 31.
support was received from TIOP 1.3.1-10/1-2010-0008, TIOP 1.3.1- [26] D.E. Clemmer, M.F. Jarrold, Ion mobility measurements and their applications
07/1, TÁMOP-4.2.2.A-11/1/KONV-2012-0053, and PTE AOK KA to clusters and biomolecules, Journal of Mass Spectrometry 32 (1997) 577.
2009, 2011, 2013. [27] P. Dugourd, R.R. Hudgins, M.F. Jarrold, High-resolution ion mobility studies of
sodium chloride nanocrystals, Chemical Physics Letters 267 (1997) 186.
[28] A.N. Alexandrova, A.I. Boldyrev, Y.-J. Fu, X. Yang, X.-B. Wang, L.-S. Wang, Struc-
ture of the Nax Clx+1 − (x = 1–4) clusters via ab initio genetic algorithm and
Appendix A. Supplementary data photoelectron spectroscopy, Journal of Chemical Physics 121 (2004) 5709.
[29] S. Zhang, N. Chen, Energies and stabilities of sodium chloride clusters based on
Supplementary data associated with this article can be found, in inversion pair potentials, Physica B: Condensed Matter 325 (2003) 172.
[30] M. Lintuluoto, Theoretical study on the structure and energetics of alkali halide
the online version, at http://dx.doi.org/10.1016/j.ijms.2013.06.029.
clusters, Journal of Molecular Structure (THEOCHEM) 540 (2001) 177.
[31] A. Ayuela, J.M. Lopez, J.A. Alonso, V. Luaña, Ab initio study of (NaCl)n Na+ clusters,
Canadian Journal of Physics 76 (1998) 311.
References [32] N.G. Phillips, C.W.S. Conover, L.A. Bloomfield, Calculations of the binding ener-
gies and structures of sodium chloride clusters and cluster ions, Journal of
[1] R.L. Johnston, Atomic and Molecular Clusters, Taylor & Francis, London and New Chemical Physics 94 (1991) 4980.
York, 2002. [33] B.I. Dunlap, The role of alternative geometries in alkali–halide clusters, Journal
[2] T.P. Martin, Shells of atoms, Physics Reports 273 (1996) 199. of Chemical Physics 84 (1986) 5611.
[3] T.P. Martin, Alkali halide clusters and microcrystals, Physics Reports 95 (1983) [34] R.Q. Topper, W.V. Feldmann, I.M. Markus, D. Bergin, P.R. Sweeney, Simulated
167. annealing and density functional theory calculations of structural and energetic
[4] J. Lefebure (Ed.), Halides: Chemistry, Physical Properties and Structural Effects, properties of the ammonium chloride clusters (NH4 Cl)n , (NH4 )+ (NH4 Cl)n , and
Nova Science Pub. Inc., Hauppauge NY, 2013. (Cl)− (NH4 Cl)n , n = 1–13, Journal of Physical Chemistry A 115 (2011) 10423.
[5] S. Zhou, M. Hamburger, Formation of Sodium Cluster Ions in Electrospray Mass [35] A. Aguado, A. Ayuela, J.M. López, J.A. Alonso, Ab initio calculations of structures
Spectrometry, Rapid Communications in Mass Spectrometry 10 (1996) 797. and stabilities of (NaI)n Na+ and (CsI)n Cs+ cluster ions, Physical Review B 58
[6] G. Wang, R.B. Cole, Charged residue versus ion evaporation for formation of (1998) 9972.
alkali metal halide cluster ions in ESI, Analytica Chimica Acta 406 (2000) 53. [36] C. Bréchignac, P. Cahuzac, F. Calvo, G. Durand, P. Feiden, J. Leygnier, Evidence
[7] P. Kebarle, L. Tang, From ions in solution to ions in the gas phase – the mecha- for entropic effects in the dissociation of cationic sodium fluoride clusters,
nism of electrospray mass spectrometry, Analytical Chemistry 65 (1993) 972. Chemical Physics Letters 405 (2005) 26.
[8] D. Schröder, Ion clustering in electrospray mass spectrometry of brine and other [37] A. Aguado, Emergence of bulk CsCl structure in (CsCl)n Cs+ cluster ions, Physical
electrolyte solutions, Physical Chemistry Chemical Physics 14 (2012) 6382. Review B 62 (2000) 13687.
[9] Y.J. Twu, C.W.S. Conover, Y.A. Yang, L.A. Bloomfield, Alkali-halide cluster ions [38] F.A. Fernandez-Lima, A.V. Henkes, E.F. da Silveira, M.A.C. Nascimento, Alkali
produced by laser vaporization of solids, Physical Review B 42 (1990) 5306. halide nanotubes: structure and stability, Journal of Physical Chemistry C 116
[10] F.W. Röllgen, K.H. Ott, On the formation of cluster ions and molecular ions in (2012) 4965.
field desorption of salts, International Journal of Mass Spectrometry and Ion [39] D. Zhang, R.G. Cooks, Doubly charged cluster ions [(NaCl)m Na2 ]2+ : magic num-
Physics 32 (1980) 363. bers, dissociation, and structure, International Journal of Mass Spectrometry
[11] T.M. Barlak, J.E. Campana, R.J. Colton, J.J. DeCorpo, J.R. Wyatt, Secondary ion 195/196 (2000) 667.
mass spectrometry of metal halides. 1. Stability of alkali iodide clusters, Journal [40] N. Mirsaleh-Kohan, S. Ard, A.A. Tuinman, R.N. Compton, P. Weis, M.M. Kappes,
of Physical Chemistry 85 (1981) 3844. Collisional dissociation of salt-cluster dianions, Chemical Physics 329 (2006)
[12] T.M. Barlak, J.R. Wyatt, R.J. Colton, J.J. DeCorpo, J.E. Campana, Secondary ion 239.
mass spectrometry of metal halides. 2. Evidence for structure in alkali iodide [41] A. Aguado, An ab initio study of the structures and relative stabilities of dou-
clusters, Journal of the American Chemical Society 104 (1982) 1212. bly charged [(NaCl)m (Na)2 ]2+ cluster ions, Journal of Physical Chemistry B 105
[13] X.B. Cox, R.W. Linton III, M.M. Bursey, Formation of small cluster ions from alkali (2001) 2761.
halides in SIMS, International Journal of Mass Spectrometry and Ion Processes [42] D.R. Lide (Ed.), CRC Handbook of Chemistry and Physics, 84th ed., CRC Press
55 (1983) 281. LLC, Boca Raton FL, 2004.
[14] F.A. Fernandez-Lima, O.P. VilelaNeto, A.S. Pimentel, C.R. Ponciano, M.A.C. [43] W.H. Zachariasen, The crystal structure of sodium formate, NaHCO2 , Journal of
Pacheco, M.A. Chaer Nascimento, E.F. da Silveira, A theoretical and experimen- the American Chemical Society 62 (1940) 1011.
tal study of positive and neutral LiF clusters produced by fast ion impact on a [44] P.L. Markila, S.J. Rettig, J. Trotter, Sodium formate, Acta Crystallographica Sec-
polycrystalline LiF target, Journal of Physical Chemistry A 113 (2009) 1813. tion B 31 (1975) 2927.
[15] C.R. Ponciano, R. Martinez, E.F. da Silveira, Fragmentation of (LiF)n Li+ clusters [45] R.M. Danell, A.S. Danell, G.L. Glish, R.W. Vachet, The use of static pressures of
in the acceleration region of TOF spectrometers, Journal of Mass Spectrometry heavy gases within a quadrupole ion trap, Journal of the American Society for
42 (2007) 1300. Mass Spectrometry 14 (2003) 1099.
[16] H. Hijazi, H. Rothard, P. Boduch, I. Alzaher, F. Ropars, A. Cassimi, J.M. Ramillon, [46] K. Doll, J.C. Schön, M. Jansen, Ab initio energy landscape of LiF clusters, Journal
T. Been, B. Ban d’Etat, H. Lebius, L.S. Farenzena, E.F. da Silveira, Interaction of of Chemical Physics 133 (2010) 024107.
swift ion beams with surfaces: sputtering of secondary ions from LiF studied by [47] H. Kabrede, R. Hentschke, An improved genetic algorithm for global opti-
XY-TOF-SIMS, Nuclear Instruments and Methods in Physics Research Section B mization and its application to sodium chloride clusters, Journal of Physical
269 (2011) 1003. Chemistry B 106 (2002) 10089.
302 Á. Révész et al. / International Journal of Mass Spectrometry 354–355 (2013) 292–302

[48] W.D. Cornell, P. Cieplak, C.I. Bayly, I.R. Gould, K.M. Merz, D.M. Ferguson, D.C. [65] R. Ahlrichs, M. Bär, H. Baron, R. Bauernschmitt, S. Böcker, N. Crawford, P.
Spellmeyer, T. Fox, J.W. Caldwell, P.A. Kollman, A second generation force field Deglmann, M. Ehrig, K. Eichkorn, S. Elliott, F. Furche, F. Haase, M. Häser, C. Hät-
for the simulation of proteins, nucleic acids, and organic molecules, Journal of tig, A. Hellweg, H. Horn, C. Huber, U. Huniar, M. Kattannek, A. Köhn, C. Kölmel,
the American Chemical Society 117 (1995) 5179. M. Kollwitz, K. May, P. Nava, C. Ochsenfeld, H. Öhm, H. Patzelt, D. Rappoport,
[49] Y. Zhao, N.E. Schultz, D.G. Truhlar, Design of density functionals by combining O. Rubner, A. Schäfer, U. Schneider, M. Sierka, O. Treutler, B. Unterreiner, M.v.
the method of constraint satisfaction with parametrization for thermo- Arnim, F. Weigend, P. Weis, H. Weiss, Turbomole 5.10, University of Karlsruhe,
chemistry, thermochemical kinetics, and noncovalent interactions, Journal of Karlsruhe, 2008.
Chemical Theory and Computation 2 (2006) 364. [66] K. Eichkorn, O. Treutler, H. Öhm, M. Häser, R. Ahlrichs, Auxiliary basis sets to
[50] R. Ditchfield, W.J. Hehre, J.A. Pople, Self-consistent molecular-orbital methods. approximate Coulomb potentials, Chemical Physics Letters 242 (1995) 652.
IX. An extended Gaussian-type basis for molecular-orbital studies of organic [67] K. Eichkorn, F. Weigend, O. Treutler, R. Ahlrichs, Auxiliary basis sets for main
molecules, Journal of Chemical Physics 54 (1971) 724. row atoms and transition metals and their use to approximate Coulomb poten-
[51] W.J. Hehre, R. Ditchfield, J.A. Pople, Self-consistent molecular orbital methods. tials, Theoretical Chemistry Accounts 97 (1997) 119.
XII. Further extensions of Gaussian-type basis sets for use in molecular orbital [68] F. Weigend, Accurate Coulomb-fitting basis sets for H to Rn, Physical Chemistry
studies of organic molecules, Journal of Chemical Physics 56 (1972) 2257. Chemical Physics 8 (2006) 1057.
[52] P.C. Hariharan, J.A. Pople, The influence of polarization functions on molecular [69] G. Schaftenaar, J.H. Noordik, Molden: a pre- and post-processing program
orbital hydrogenation energies, Theoretica Chimica Acta 28 (1973) 213. for molecular and electronic structures, Journal of Computer-Aided Molecular
[53] M.M. Francl, W.J. Pietro, W.J. Hehre, J.S. Binkley, M.S. Gordon, D.J. DeFrees, J.A. Design 14 (2000) 123.
Pople, Self-consistent molecular orbital methods. XXIII. A polarization-type [70] Legault C.Y., Université de Sherbrooke, 2009 (http://www.cylview.org).
basis set for second-row elements, Journal of Chemical Physics 77 (1982) 3654. [71] J. Sztáray, A. Memboeuf, L. Drahos, K. Vékey, Leucine enkephalin – a mass
[54] R. Krishnan, J.S. Binkley, R. Seeger, J.A. Pople, Self-consistent molecular orbital spectrometry standard, Mass Spectrometry Reviews 30 (2011) 298.
methods. XX. A basis set for correlated wave functions, Journal of Chemical [72] S.A. McLuckey, D.E. Goeringer, Slow heating methods in tandem mass spec-
Physics 72 (1980) 650. trometry, Journal of Mass Spectrometry 32 (1997) 461.
[55] A.D. McLean, G.S. Chandler, Contracted Gaussian basis sets for molecular calcu- [73] R.E. March, An introduction to quadrupole ion trap mass spectrometry, Journal
lations. I. Second row atoms, Z = 11–18, Journal of Chemical Physics 72 (1980) of Mass Spectrometry 32 (1997) 351.
5639. [74] R.G. Cooks, R.E. Kaiser Jr., Quadrupole ion trap mass spectrometry, Accounts of
[56] T. Clark, J. Chandrasekhar, G.W. Spitznagel, P.v.R. Schleyer, Efficient diffuse Chemical Research 23 (1990) 213.
function-augmented basis sets for anion calculations. III. The 3-21+G basis set [75] Y. Zhao, D.G. Truhlar, Density functional theory for reaction energies: test of
for first-row elements, Li–F, Journal of Computational Chemistry 4 (1983) 294. meta and hybrid meta functionals, range-separated functionals, and other high-
[57] M.J. Frisch, J.A. Pople, J.S. Binkley, Self-consistent molecular orbital methods 25. performance functionals, Journal of Chemical Theory and Computation 7 (2011)
Supplementary functions for Gausssian basis sets, Journal of Chemical Physics 669.
80 (1984) 3265. [76] J. Lai, X. Lu, L. Zheng, Convergence from clusters to the bulk solid: an ab initio
[58] J.P. Merrick, D. Moran, L. Radom, An evaluation of harmonic vibrational fre- investigation of clusters Nan Cln (n = 2–40), PhysChemComm 5 (2002) 82.
quency scale factors, Journal of Physical Chemistry A 111 (2007) 11683. [77] C. Ochsenfeld, R. Ahlrichs, An ab initio investigation of clusters Nan Cln , Journal
[59] S. Grimme, Improved second-order Møller–Plesset perturbation theory by sep- of Chemical Physics 97 (1992) 3487.
arate scaling of parallel- and antiparallel-spin pair correlation energies, Journal [78] F.M. Bickelhaupt, M. Solà, C.F. Guerra, Table salt and other alkali metal chlo-
of Chemical Physics 118 (2003) 9095. ride oligomers: structure, stability, and bonding, Inorganic Chemistry 46 (2007)
[60] L.A. Curtiss, K. Raghavachari, P.C. Redfern, V. Rassolov, J.A. Pople, Gaussian-3, 5411.
(G3) theory for molecules containing first and second-row atoms, Journal of [79] J.P.K. Doye, D.J. Wales, The dynamics of structural transitions in sodium chloride
Physical Chemistry C 109 (1998) 7764. clusters, Journal of Chemical Physics 111 (1999) 11070.
[61] S. Grimme, Semiempirical GGA-type density functional constructed with a [80] O. Obolensky, A. Lyalin, A. Solov’yov, W. Greiner, Geometrical and statistical
long-range dispersion correction, Journal of Computational Chemistry 27 factors in fission of small metal clusters, Physical Review B 72 (2005) 085433.
(2006) 1787. [81] C. Červinka, M. Fulem, K. Růžička, Evaluation of accuracy of ideal-gas heat
[62] F. Weigend, M. Häser, H. Patzelt, R. Ahlrichs, RI-MP2: optimized auxiliary basis capacity and entropy calculations by density functional theory (DFT) for rigid
sets and demonstration of efficiency, Chemical Physics Letters 294 (1998) 143. molecules, Journal of Chemical & Engineering Data 57 (2012) 227.
[63] F. Weigend, R. Ahlrichs, Balanced basis sets of split valence, triple zeta valence [82] V. van Speybroeck, R. Gani, R.J. Meier, The calculation of thermodynamic prop-
and quadruple zeta valence quality for H to Rn: design and assessment of erties of molecules, Chemical Society Reviews 39 (2010) 1764.
accuracy, Physical Chemistry Chemical Physics 7 (2005) 3297. [83] M. Umer, K. Leonhard, Ab initio calculations of thermochemical properties of
[64] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, methanol clusters, Journal of Physical Chemistry A 117 (2013) 1569.
G. Scalmani, V. Barone, B. Mennucci, G.A. Petersson, H. Nakatsuji, M. Caricato, [84] G. Katzer, A.F. Sax, Beyond the harmonic approximation: impact of anharmonic
X. Li, H.P. Hratchian, A.F. Izmaylov, J. Bloino, G. Zheng, J.L. Sonnenberg, M. Hada, molecular vibrations on the thermochemistry of silicon hydrides, Journal of
M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, Physical Chemistry A 106 (2002) 7204.
O. Kitao, H. Nakai, T. Vreven, J.A. Montgomery Jr., J.E. Peralta, F. Ogliaro, M. [85] V. Barone, Vibrational zero-point energies and thermodynamic functions
Bearpark, J.J. Heyd, E. Brothers, K.N. Kudin, V.N. Staroverov, R. Kobayashi, J. Nor- beyond the harmonic approximation, Journal of Chemical Physics 120 (2004)
mand, K. Raghavachari, A. Rendell, J.C. Burant, S.S. Iyengar, J. Tomasi, M. Cossi, N. 3059.
Rega, J.M. Millam, M. Klene, J.E. Knox, J.B. Cross, V. Bakken, C. Adamo, J. Jaramillo, [86] L. Drahos, K. Vékey, Entropy evaluation using the kinetic method: is it feasible?
R. Gomperts, R.E. Stratmann, O. Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Journal of Mass Spectrometry 38 (2003) 1025.
Ochterski, R.L. Martin, K. Morokuma, V.G. Zakrzewski, G.A. Voth, P. Salvador, [87] L. Drahos, K. Vékey, MassKinetics: a theoretical model of mass spectra incor-
J.J. Dannenberg, S. Dapprich, A.D. Daniels, Ö. Farkas, J.B. Foresman, J.V. Ortiz, J. porating physical processes, reaction kinetics, and mathematical descriptions,
Cioslowski, D.J. Fox, Gaussian 09, Revision A.02, Gaussian, Inc., Wallingford, CT, Journal of Mass Spectrometry 36 (2001) 237.
2009.

You might also like