You are on page 1of 9

Journal of Molecular Structure 993 (2011) 43–51

Contents lists available at ScienceDirect

Journal of Molecular Structure


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

Theoretical and experimental investigation on the spectroscopic properties


of indigo dye
Anna Amat a,⇑, Francesca Rosi a, Costanza Miliani b, Antonio Sgamellotti a,b, Simona Fantacci b,c
a
Dipartimento di Chimica sdi Perugia, Università degli Studi di Perugia, via Elce di Sotto 8, 06123 Perugia, Italy
b
Istituto di Scienze e Tecnologie Molecolari del CNR (CNR-ISTM), C/o Dipartimento di Chimica, Università di Perugia, Via Elce di Sotto 8, I-06123, Perugia, Italy
c
Italian Institute of Technology (IIT), Center for Biomolecular Nanotechnologies, Via Barsanti 73010, Arnesano, Lecce, Italy

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

Article history: The spectroscopic properties of indigo, one of the most important natural dyes present in nature, have
Available online 30 November 2010 been investigated by means of DFT and TD-DFT calculations and Raman and IR spectroscopies. The
absorption spectra of this dye, in vacuo and in different solvents, have been computed. The formation
Keywords: of aggregates in solvent have been investigated by computing the electronic absorption spectra of the
Indigo dimer and the trimer, thus evaluating the effects of the aggregation on the optical properties of indigo.
DFT The IR and Raman spectra have been measured and computed. The comparison between the experi-
TD-DFT
mental and theoretical spectra and the potential energy distribution (PED) of the computed normal
Raman
IR
modes have been used to perform the assignment of the experimental features in terms of functional
Hydrogen bond group displacements. Finally, the effects of the intermolecular hydrogen bond present in the solid state
have been evaluated by computing the vibrational spectra of the dimer.
The intention of the present work is to give an insight into both the vibrational and optical properties of
indigo as well as to evaluate DFT and TD-DFT potentialities in the study of organic dyes’ spectroscopic
properties of interest in the cultural heritage field.
Ó 2010 Elsevier B.V. All rights reserved.

1. Introduction The absorption spectrum of indigo has been shown to be depen-


dent on the environment ranging from red (540 nm) in the gas
The historical use of indigo as a blue dyestuff dates to the pre- phase to violet (588 nm) in tetrachloromethane or to blue
Vedic period in India where Indigofera tinctoria L. was its more (606 nm) in polar solvents such as ethanol [8,9] (see Fig. 1).
common source. India was a primary supplier of indigo to Europe Several theoretical studies can be found on the electronic
in the Greco-Roman era. India’s association with indigo is reflected absorption spectra of indigoids. For indigo a good agreement is
in the word for the dyestuff, which was indicum (by Pliny and found in polar solvents with CASPT2 although gas-phase calcula-
Vitruvius) or indicon (by Dioscorides) which passed into Italian tions underestimates the first transition by ca. 0.30 eV [8]. The first
dialect and eventually into English as the word indigo. In Europe work employing TD-DFT uses two different xc-functionals (B3LYP
it was also obtained from dyer’s woad but prior to the import of and BPW91) reporting non negligible differences between the
larger amounts of indigo from India in the 16th century they were two functionals and with the experiment [10]. Quite large errors
not recognized as containing the same colouring matters [1–3]. are retrieved by Matsuura et al. [11] using semi-empirical and
Indigo has always attracted great interest and many studies can TD-DFT calculations in a set of organic dyes as the solvation effects
be found on this dye and its related derivatives [4]. Indigo is also where neglected. The best agreements with the experiment are
known as the chromophore of Maya Blue, a pre-colombian pig- found in a series of works on indigoids by Jaquemin et al. Hybrid
ment prepared by precipitating indigo on a white clay, usually functionals and large basis sets together with a polarizable contin-
palygorskite. Maya Blue was used in wall paintings and ceramics uum model (PCM) are used by these authors for indigo [12], thio-
and it has been widely studied due to its unusual stability and indigo derivatives [13] and for indirubin, isoindigo and related
hue of colour. To gain an insight into the great stability of Blue derivatives [14] obtaining very good results.
Maya, interactions between indigo and palygorskite have been Experimental evidence of indigo aggregation in solution has
studied, by means of quantum mechanical methods and molecular been detected with the appearance of a new band around
dynamics [5–7]. 700 nm at large indigo concentrations that has been assigned to
the formation of a dimer [15]. On the other hand, dimer and trimer
⇑ Corresponding author. Tel.: +39 075 585 5638; fax: +39 075 585 5606. attached to the textile have been proposed as feasible aggregates
E-mail address: anna@thch.unipg.it (A. Amat). when dyeing wool with solutions of high indigo concentration

0022-2860/$ - see front matter Ó 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.molstruc.2010.11.046
44 A. Amat et al. / Journal of Molecular Structure 993 (2011) 43–51

2. Experimental and computational methodology

Micro-Raman experimental spectra have been obtained using a


JASCO Ventuno double-grating spectrophotometer equipped with
a charge-coupled device (CCD) detector cooled to 50 °C and cou-
pled to an optical microscope. Raman spectra have been excited
using green radiation from an Nd:YAG laser (532 nm). The laser
power at the sample was kept between 0.8 and 4 mW, the expo-
sure time varied between 10 and 20 s with 5–10 accumulations
Fig. 1. Indigo molecular structure.
and the resolution was about 2 cm 1. Calibration of the spectrom-
eter was accomplished using the Raman lines of a polystyrene
[16,17]. Semiempirical simulations of the absorption spectrum standard. Transmission IR spectra were collected with a Jasco FTIR
using 2 and 3 molecule clusters based on the crystalline structure 470-plus spectrophotometer equipped with a ceramic light source,
have been performed [16,17] to simulate the aggregation effects on a Michelson interferometer, a DLATGS detector. The spectra were
the optical properties. Finally, a limited analysis has been reported carried out with 100 scans at a resolution of 2 cm 1 and an effec-
by us in a general paper on the role of computational chemistry in tive range of 4000–400 cm 1 in KBr pressed disks.
the study of art materials [18]. DFT and TD-DFT calculations have been carried using the
Investigations of the indigoids vibrational spectra by means of Gaussian03 code [23]. The Becke’s hybrid exchange functional B3
ab initio and DFT methods have been reported [19–22]. First [24] with the Lee–Yang–Parr correlation functional LYP [25,26]
assignment of indigo vibrational bands with ab initio calculations (B3LYP) have been used in our calculations. The solvation effects
was performed by Tatsch and Schraeder in 1995 [19] using HF/3- have been included by means of the conductor-like polarizable
21G level of theory as implemented in Gaussian 92. In this work continuum model CPCM [27].
the experimental bands are compared to the theoretical spectra Geometry optimizations, with no symmetry constrains, have
for which wavenumbers but no intensities are reported and the been performed using 6-31G** and 6-311++G** basis set and fre-
normal mode assignment is performed. A few works have been quencies have been computed, at the same level of theory, to check
found revising this information. Sánchez del Río et al. [20] using the nature of stationary points.
the same level of theory (HF/3-21G) assign the Raman spectrum Vibrational spectra have been computed using 6-31G** and a
and report the computed Raman activities. Tomkinson et al. [21] 0.9614 scaling factor has been applied to the computed frequencies
use DFT calculations with B3LYP/6-31G** and Inelastic Neutron according to the values suggested by Scott et al. [28] and a Lorentz-
Scattering (INS) spectroscopy to study the vibrational spectra of in- ian convolution has been performed on the scaled frequencies to
digo. The results of the IR and Raman computed spectra are re- obtain the computed spectra. Normal coordinate analysis calcula-
ported with no intensities and the band assignment is performed tions in terms of potential energy distributions (PEDs) have been
on the basis of a previous assignment on isatin and thioindigo performed using VEDA4 program [29] allowing us to assign unam-
spectra. Finally, in a very recent paper, Baran et al. [22] assign biguously the computed frequencies to the functional group
the main experimental features of the IR and Raman spectra below displacements.
1700 cm 1 using B3LYP/6-311++g** calculations and a potential The absorption spectra of indigo in chloroform and in vacuo
energy distribution (PED) analysis. have been simulated by computing the lowest singlet-singlet exci-
Due to the great experimental and theoretical interest in indigo tations following three different procedures:
we report a comprehensive investigation rationalizing some
incompletely addressed aspects such as the aggregation in solution (a) 6-31G** // 6-31G** for the vertical excitations // geometry
or the effects of crystal packing in the vibrational spectra. Regard- optimizations.
ing the vibrational properties, the Raman and IR spectra have been (b) 6-311++G** // 6-311++G** for the vertical excitations //
measured and computed. geometry optimizations.
The comparison between the measured and the computed (ab) 6-311++G**// 6-31G for the vertical excitations // geometry
vibrational spectra, together with the PED analysis, has allowed optimizations.
us to assign the experimental modes in terms of functional groups
displacements. Finally, the dimer optimized structure in vacuo has The comparison between the theoretical results and the exper-
been used as a simple model to evaluate the hydrogen bond effects, imental data has allowed the best methodology to be individual-
present in the solid state. The vibrational spectra of the dimer have ized. This methodology has then been used in the analysis of the
been computed and compared to those of the monomer and to the absorption spectra in different solvents and in the study of indigo
experiment. aggregation in solution.
A computational methodology to reproduce the electronic
absorption spectra of indigo has been calibrated and used to com-
pute the spectra in vacuo and in different solvents. Results have 3. Results and discussion
been compared to experimental data, evaluating the accuracy of
DFT methods in reproducing the solvation effects and rationalizing 3.1. Absorption spectra
the observed bathocromic effects.
Successively, the aggregation in solution has been studied by Indigo optimizations in vacuo and in chloroform with both
hypothesizing the presence of dimer and trimer aggregates and 6-31G** and 6-311++G** basis sets have been performed. No signif-
analyzing the changes of the indigo geometry, electronic distribu- icant differences have been retrieved between the geometry opti-
tion and absorption spectrum. mized in vacuo and in chloroform. Neither the basis set seem to
We present therefore a complete theoretical and experimental affect the results obtaining differences that are kept within
study on the absorption and vibrational spectroscopic properties 0.03 Å and 0.5° for the bond distances and angles, respectively.
of indigo. In a more general framework we intend to use indigo Results are reported in Supporting Info (SI).
as a prototype molecule to setup and evaluate a computational To rationalize the effect of the basis set and the solvent on the
strategy suitable for the study of cultural heritage materials. optical properties, the absorption spectra of indigo has been
A. Amat et al. / Journal of Molecular Structure 993 (2011) 43–51 45

computed in vacuo and in chloroform with the three computa- with respect to the vacuo case, in which differences are within
tional strategies described in Section 2. In all the cases, the main 0.05 eV and can be considered negligible, with differences of
absorption band, responsible for the colour, is originated by the 0.15 eV that are, however, in the limits of the method accuracy. Ba-
lowest singlet–singlet transition (S0–S1). The computed transitions sis set effects retrieved in solvent are ascribable to the response
are reported in Table 1 compared to the maxima of the main exper- part of the calculations given that no differences have been de-
imental absorption band. The charge distribution of the highest tected in the ground state. Previous article reported the need of
occupied molecular orbital, HOMO (hereafter H), and the lowest using large basis set to quantitatively reproduce the absorption
unoccupied molecular orbital, LUMO (hereafter L), computed using experimental spectra of indigo [12] and 6-311G(d,p) and 6-
6-311++G**, reported in Fig. 2, are equivalent to the results ob- 311+G(2d,p) for the geometry optimization and the vertical excita-
tained using 6-31G**, reported in SI. tions, respectively, was identified as the best computational strat-
The H and L involved in the first transitions are completely delo- egy. Our calculations point out that the use of larger basis set on
calized over the whole molecule and have a p and p* character, the geometry optimizations has no influence on the structural
respectively. The H presents a high degree of conjugation that and electronic properties and does not improve the absorption
can be visualized with the charge delocalization on all the mole- spectra calculation accuracy.
cule through the C@C ring. Having established that (ab) procedure give results in excellent
In vacuo, the increase of the basis set does not affect the molec- agreement with the experiment both in vacuo and in solution at a
ular geometry, the charge distribution, the H–L gap or the first sin- lower computational cost than (b) procedure, this strategy has
glet-singlet excitation energy. On the contrary, TD-DFT been used to explore the spectral changes in different solvents
calculations in solution seem to be more sensitive to the basis sets and the aggregation in solution.
A bathochromic shift of indigo absorption spectra with the sol-
vent polarity has been experimentally observed [32–49]. To get in-
Table 1 sight into these effects the absorption spectra, and electron density
Energy [nm(eV)] and composition of the indigo computed lowest singlet–singlet
distribution change between S0 and S1 in vacuo and in different sol-
excitations together with the experimental absorption maxima.
vents have been computed. Results of the computed transitions, re-
a b ab Exp. ported in Table 2 (vacuo, chloroform, and DMSO) and in SI (CCl4,
Vac. 536 (2.31) 548 (2.26) 553 (2.24) 539 (2.30)a benzene, and ethanol) are in very good agreement with the exper-
H ? L (70%) H ? L (72%) H ? L (72%) iment and consistent with previous data [12] and are able to repro-
H-5 ? L (3%) H-5 ? L (2%) H-5 ? L (3%)
duce even the small bathochromic shifts observed in the
Chlor. 565 (2.20) 600 (2.07) 605 (2.05) 604 (2.05)b experimental absorption spectra. This shift has been interpreted
H ? L (70%) H ? L (74%) H ? L (74%)
as resulting from the increased stability of the charge-separated
H-5 ? L (2%)
structure in the excited state rather than in the ground state,
(a)Computed using 6-31G** // 6-31G** for the excitations // geometry optimizations. resulting in a decrease of the excitation energy [12,32].
(b) Computed using 6-311++G** // 6-311++G** for the excitations // geometry
Going from vacuo to chloroform and DMSO, a estabilization/sta-
optimizations (ab) Computed using 6-31++G**// 6-31G** for the excitations //
geometry optimizations.
bilization on the H/L is retrieved, resulting in a decrease of the H–L
a
From Ref. [30]. gap. The computed absorption maxima have similar composition,
b
From Ref. [31]. except for a minimal contribution of the H-5 ? L in vacuo, and
are in good agreement with the experimental values.
The composition of the H and L in terms of atomic orbitals show
a major contribution of the 2pz and 3pz of the N and the O, respec-
L L tively. For the first transition, that has manly a H to L character, this
is translated in a charge transfer upon excitation from the electron
2.40 eV donor ortho NH to the electron acceptor C@O group, as it can be
2.49 eV
observed from the electron density change computed in vacuo be-
H H tween S0 and S1 reported in Fig. 3. Upon excitation, other than the
decrease of the charge in the nitrogen atoms and the increase in
the oxygen atoms, it can be observed a loss of electron density in
Fig. 2. H and L isodensity plots and H–L gap energy of indigo computed in vacuo the center of inversion of the molecule, reflecting a decrease of
(left) and in chloroform (right). the conjugation in the excited state. The electron density change

Table 2
Computed H and L stabilities and composition in terms of atomic orbitals, H–L gap, and computed and experimental absorption spectra maxima [nm(eV)] in vacuo, chloroform
and DMSO.

H (eV) H comp. L (eV) L comp. H–L (eV) Comp. S1 Exp.


Vac. 5.67 41% N (2pz, 3pz) 3.18 28 %O (2pz, 3pz) 2.49 553 (2.24) 539 (2.39)a
15% C(C@C) (2pz, 3pz) 21% C(C@C) (2pz, 3pz) H ? L (74%)
15% C(CO) (2pz, 3pz) 18% C(CO) (2pz, 3pz) H-5 ? L (3%)
Chlor. 5.64 41% N (2pz, 3pz) 3.25 28 %O (2pz, 3pz) 2.38 605 (2.05) 604 (2.05)b
15% C(C@C) (2pz, 3pz) 21% C(C@C) (2pz, 3pz) H ? L (74%)
14% C(CO) (2pz, 3pz) 20% C(CO) (2pz, 3pz)
DMSO 5.63 41% N (2pz, 3pz) 3.27 28 %O (2pz, 3pz) 2.36 614 (2.02) 619 (2.00)c.
14% C(C@C) (2pz, 3pz) 21% C(C@C) (2pz, 3pz) H ? L (74%)
14% C(CO) (2pz, 3pz) 20% C(CO) (2pz, 3pz)
a
From Ref. [30].
b
From Ref. [31].
c
From Ref. [33]
46 A. Amat et al. / Journal of Molecular Structure 993 (2011) 43–51

Within each indigo unit no relevant differences are retrieved


between the dimer and the monomer optimized structures and
are also equivalent between vacuo and solvent calculations. Con-
trarily, some structural differences are retrieved in the unit
arrangements, being the angle between the two indigo units
29.2° (25.8°) in chloroform(vacuo) and the average values of the
hydrogen bonds between the two molecules 2.027(1.999) Å.
The geometry optimization in solution has been also performed
for the indigo trimer providing the structure reported in Fig. 4(b)
which shows inappreciable differences with respect to the dimer,
with average intermolecular hydrogen bond distances and dihedral
angles of 2.035 Å and 28.9°, respectively. Energetically, the dimer
Fig. 3. Isodensity plot of the electron density difference between S0 and S1 in vacuo.
A blue (white) colour indicates a decrease (increase) of the electron density upon and the trimer are computed 16.67 and 26.74 kcal/mol more sta-
excitation. ble, respectively, than the corresponding isolated monomer,
though these results are probably overestimated due to the basis
set superposition error (BSSE) that has not been corrected.
On the optimized structures in solution, the lowest singlet–sin-
glet transitions have been computed. The results, reported in Table
3, point out the appearance of new transitions at lower energies
2.027
(1.996) that, even though they are characterized by small oscillator
strengths for both aggregates, they qualitatively reproduce the
2.035 new band detected experimentally upon increase of the indigo
concentration. The same results were obtained with calculations
28.9°
performed in dichloroethane. Experimental spectra from Ref. [15]
29.2°
(25.8°) at different concentrations together with the positions of the com-
puted vertical excitations for the monomer, the dimer and the tri-
(a) (b) mer are reported in Fig. 5.
The new experimental band at ca. 700 nm (1.77 eV) is com-
Fig. 4. Indigo dimer(a) and trimer(b) optimized structures. Average values of the
hydrogen bonds (Å) and dihedral angles (°) between the molecules are reported for puted sligthly blue-shifted by 0.10 eV, a deviation that is within
the chloroform(vacuo) calculations. the accuracy of the methodology. The molecular orbitals are
grouped in almost isoenergetic pairs(trios) for the dimer(trimer).
Only the molecular orbitals belonging to the first pair(trio) are in-
computed in chloroform and DMSO are very similar to those com- volved in the lowest transitions responsible for the colour band.
puted in vacuo and are reported in SI. Their electronic distribution has been analyzed for the dimer. H,
H-1, L and L+1 molecular orbitals energies and electronic distribu-
tions are reported in Fig. 6.
3.2. Aggregation in solution
The electronic delocalization in each indigo unit is very similar
to that computed for the isolated molecule, except for a slight de-
Experimental evidence of indigo aggregation in solution with
crease of the charge delocalization in the HOMOs and an increase
the appearance of a broad shoulder at 700 nm (1.77 eV) has been
of the charge in the N in the LUMOs. In the H-1 and the L, the indigo
reported. The spectral position of the aggregate indicates the for-
unit orbitals have the same symmetry while the contrary is true for
mation of more polar species than the monomer and excludes
the H and L+1. Transitions are computed with a very low oscillator
the formation of an H-type aggregate [15]. More likely an hydro-
strength that might be affected by an incomplete treatment of the
gen-bonded structure is formed retrieving similar effects on the
weak intermolecular hydrogen bond forces and to a not accurate
absorption spectra as the ones detected for the solid state where
description of the long-range charge-transfer excitations [51–53].
the formation of an hydrogen-bonded structure has been proved
[50]. We have therefore investigated the geometry and absorption
spectra of two different aggregates (dimer and trimer) in chloro- 3.3. Raman and IR spectra
form following the strategy defined for the monomer.
The indigo dimer optimizations in vacuo and in solvent leaded, Vibrational spectroscopy is one of the most powerful tools in
in both cases, to stable hydrogen-bonded structures with all posi- modern materials research. Quantum chemical calculations pre-
tive frequencies. Structures are reported in Fig. 4(a) together with dicting harmonic wavenumbers and spectral intensities are essen-
the main distances and angles. tial when interpreting experimental spectra, particularly for large

Table 3
Computed TD-DFT transitions of indigo dimer and trimer in chloroform.

Dimer Trimer
Th. nm/eV f MO Th. nm/eV f MO
660/1.88(S1) 0.0071 (85%)H ? L 670/1.85(S1) 0.0113 (83%)H ? L
(7%) H-1 ? L+1 (4%)H ? L+2
647/1.92(S2) 0.0217 (63%)H ? L+1 652/1.90 (S3) 0.0392 (67%)H ? L+2
(37%)H-1 ? L (30%)H-2 ? L
612/2.02(S3) 0.0220 (80%)H-1 ? L+1 618/2.01(S5) 0.0298 (47%)H-2 ? L+2
(2%)H ? L (17%)H-1 ? L+1
603/2.06(S4) 0.5863 (50%)H-1 ? L 602/2.06(S7) 0.8436 (33%)H-2? L
(23%)H ? L+1 (26%)H-1 ? L+1
(26%)H? L+2
A. Amat et al. / Journal of Molecular Structure 993 (2011) 43–51 47

Our simulated and experimental vibrational spectra of indigo


are compared and reported in Fig. 7. Experimentally, the Raman
spectra was partially covered by its own fluorescence complicating
the analysis of the Raman spectra above 3000 nm, therefore exper-
imental data from Ref. [19] has been used for the assignment in
this region.
Even though it has been observed that a better agreement with
experiment is found using a different scaling factor at long wave-
numbers (>1800 cm 1) [55], we have chosen to perform a uniform
scaling in all the spectrum and therefore bands computed in the
high wavenumber region are quite overestimated.
Indigo observes a planar molecule that belongs to the C2h group
and has 84 fundamental modes divided into C = 29Ag + 13B-
g + 14Au + 28Bu. The rule of mutual exclusion holds and gerade
vibrations are Raman active only while ungerade are IR active only.
Ag and Bu types are planar while Au and Bg ones are non-planar.
Fig. 5. Experimental absorption spectra in solvent at different indigo concentra-
tions from Ref. [15] and positions of the computed vertical excitations for the Although our calculations have been performed without symmetry
monomer (red), the dimer (blue) and the trimer (green). (For interpretation of the constrains, the optimized structure neglectably deviates from the
references in colour in this figure legend, the reader is referred to the web version of C2h symmetry, therefore the computed modes are in perfect agree-
this article.) ment with the expected mode character distribution of the C2h
group and its 84 modes numbers are Ag(1–29), Au(30–43), Bg(44–
56) and Bu(57–84). The most intense modes of our experimental
and computed spectra are reported in Table 4 alongside with Tats-
ch’s assignment while the complete table with all the experimental
and computed bands is reported in SI. The experimental signals
that have not been related to the computed normal modes are
highlighted in yellow.
Overall, our experimental spectra is very similar to those mea-
sured in precedence with only slight differences mainly in the sig-
nals intensities, probably due to the use of different incident laser
wavelengths. However, a substantial difference is found on the
strong band that we measure at 1075 cm 1 in the IR spectrum that
is reported as a shoulder at 1074 cm 1 by Tatsch et al. No strong
bands are reported in the 1000–1100 cm 1 region by these authors
while a strong band is reported at 1065 cm 1 by Baran et al. [22]
We compute this band at 1052 cm 1 with an intensity of 54% with
respect to the most intense one, in line with the experimental find-
ings and we assign it to in-plane bending modes on the pyrrole
rings.
The computed frequencies show that most of the IR normal
Fig. 6. MO electronic distribution of indigo dimer in solvent.
modes have its correspondent mode in the Raman spectrum, with
shifted wavenumbers, differing only from their symmetry.
In the region above 2000 cm 1 10 normal modes are retrieved,
molecules in which the high density of states results in spectral divided in ungerade/gerade pairs with almost the same frequen-
complexity. Accurate vibrational assignments for such systems cies, that can be ascribed to NH and CH stretching. While medium
are necessary for the material characterization and they can be intensity mCN and mCH bands are retrieved in the Raman spectrum,
proposed on the basis of wavenumber agreement between the no intense mCH band is computed in the IR spectra on this region.
computed harmonics and the observed fundamentals [54]. C@C and C@O stretching modes are computed in the 1700–
The IR and Raman spectra have been experimentally measured 1500 cm 1 range, where we compute the most intense bands.
and computed and their comparison has allowed an assignment of These modes are followed by a series of mCC (inter-ring stretching)
the experimental features in terms of functional groups displace- and in-plane dCH bendings. All the 27 out of plane modes, Au and
ments. Our results have been also compared with previous data Bg are computed below the 1000 cm 1.
available. Among the theoretical works [19–22] present in litera- Computationally, the main differences with previous article
ture we report, alongside our results and assignment, Tatsch’s [19] are retrieved in the assignment of the normal modes to the
assignment [19] which is, the reference paper most recalled by specific functional groups probably due to the different methodol-
experimental works on indigo. Moreover is the only work that cov- ogies used to perform the assignments and not to the different le-
ers the whole spectral range and reports IR and Raman spectra vel of theory. Similar wavenumbers are computed and all the
with the experimental intensities and activities. optical character coincide except for the low intensity bands of
In Tatsch’s work, normal modes assignment for indigo were the Raman spectrum at 1097 cm 1 that we compute Ag, see SI, in
performed on the basis of visual inspection. Usually, many atoms contrast with the Bg reported by Tatsch, in line with the discrepan-
are involved in a vibrational mode and the assignment to a func- cies retrieved by Tomkinson [21].
tional group in terms of stretching (m), in-plane bending (d), and In general a good agreement is found with the experiment. The
out-of-plane deformation (c) is not straightforward. We have bigger discrepancies with the measured wavenumbers are re-
therefore performed a normal coordinate analysis in terms of po- trieved in the high wavenumber region, as discussed above, and
tential energy distributions (PEDs) which provides a strong sup- in the IR band at 1627 cm 1 that we compute at 1660 cm 1. Low
port to the frequency assignment. intensity bands are sometimes difficult to assign but the good
48 A. Amat et al. / Journal of Molecular Structure 993 (2011) 43–51

Fig. 7. Comparison between the computed (red) and the experimental (black) Raman (a) and IR (b) spectra. (For interpretation of the references in colour in this figure legend,
the reader is referred to the web version of this article.)

agreement with the experiment and an accurate analysis taking with harmonic DFT calculations. Firstly, anharmonic effects are
into account the computed mode character and relative intensities not considered in this computational methodology although they
has allowed us to determine almost all the experimental signals. are partially corrected by the scaling factor used. Secondly, exper-
imental spectra can present resonance effects that are not taken
3.4. Inter-molecular hydrogen bond effects on the vibrational spectra into account in our calculations and that may enhance particular
frequencies. Finally, calculations regard the isolated molecule,
In general, three main sources of divergences can be individu- therefore the possible effects of the intermolecular interactions
ated when comparing experimental spectra and those computed are ignored.
A. Amat et al. / Journal of Molecular Structure 993 (2011) 43–51 49

Table 4
Vibrational spectra of Indigo. Experimental and computed wavenumbers, computed relative intensities and raman activities, normal modes numbers, computed modes, this work
assignment, and previous work assignment.
1 1
Exp. IR (cm ) Exp. Ra. (cm ) Th. W. IR Int. Ra. Act. m Sym. Mode (prev. work) Modea
a
3270(vw) 3480 2% 23% 1 Ag 100% mNH mNH
3268(w)a 3479 14% 3% 57 Bu 80% mNH mNH
a
3060(w) 3091 <1% 32% 2 Ag 85% mCH mCH
3056(vw)a 3091 7% 3% 58 Bu 78% mCH mCH
1692(s) 1710 0% 8% 6 Ag 26% mC@C mCO
mC@C
1627(vs) 1660 54% 0% 62 Bu 40% mOC mCO
1624(m) 1631 0% 47% 7 Ag 24% mOC mCC
624% mC@C dCH
13% mCC
1614(s) 1602 100% 0% 63 Bu 26% mCC mCC
dCH
1581(s) 1596 0% 100% 8 Ag 13% mOC mC@C
16% mC@C mCO
18% mCC
1585(m) 1575 3% <1% 64 Bu 25% mCC mCC, dCH
1483(s) 1469 48% 0% 65 Bu 20% mCC mCC
29% dHCC
1462(s) 1441 30% 0% 66 Bu 35% dHCC mCC
dCH
1395(m) 1400 58% 0% 67 Bu 9% mNC dNH
16% dHNC dCH
10% dHCC
1362(s) 1361 0% 2% 12 Ag 18% mNC dNH
7% dHCC dCH
1347 0% 13% 13 Ag 32% mNC dNH
1317(m) 1319 14% 0% 68 Bu 33% mCC mCC
12% dHCC
1308(w) 1303 0% 15% 14 Ag 19% mCC mCC
26% dHCC
1299(m) 1274 20% 0% 69 Bu 18%dHCC dCH
1247(m) 1227 0% 15% 15 Ag 16% mNC dCH
22% dHCC dCO
1199(m) 1169 26% 0% 71 Bu 11% mCC mCC
5% dCCC mCN
10% dHCC
1173(m) 1152 63% 0% 72 Bu 13% mNC dCC
7% dHNC
5% dHCC
1127(s) 1122 53% 0% 73 Bu 9% mNC dCC
9% dHNC dC-NH-C
20% dHCC
1075(s) 1052 54% 0% 75 Bu 12% dCCC mC-NH-C
10% dOCC
10% dCNC
754(m) 736 12% 0% 34 Au 32% cHCCC dCH
8% cNCCC dNCC
713(m) 690 4% 0% 79 Bu 27% dCCC c5-Ring
566(m) 550 10% 0% 36 Au 19% cCCCC c5-Ring
6% cNCCN
548 11% 0% 81 Bu 24% dCCC
547(m) 533 0% 3% 26 Ag 12% mCC dC@C–CO–C
39% dCCC
508(w) 475 18% 0% 37 Au 58% cHNCC dC@C–C@O
5% cNCCN
a
Form Ref. [19].

Anharmonicity and resonance effects can be taken into account


a=9.24 Å ; b=5.77 Å with the more advanced quantum mechanical codes and, even
c=12.22 Å ; ß = 117° though it would be interesting to perform a detailed analysis of
their influence on the computed spectra, such a study is beyond
the aim of this work, where we have focused our attention on
the intermolecular interactions and their effects on the optical
properties of indigo.
In the solid, molecules are disposed in parallel planes with two
different orientations (see Fig. 8).
Each indigo molecule form four hydrogen bonds of ca 2.113 Å
with the four neighboring molecules belonging to the other orien-
tation planes. Frequency calculations on a 5 unit indigo structure
Fig. 8. Crystal structure of indigo from Ref. [50] were too time demanding and attempts to optimize a 3 unit cluster
50 A. Amat et al. / Journal of Molecular Structure 993 (2011) 43–51

from the crystal structure, with indigo bonding the two neighbor- 1621 cm 1, respectively, improving the agreement with the exper-
ing molecules in parallel planes, were vain. However, it can be rea- imental bands measured at 1581 and 1624 cm 1. A new band ap-
sonably expected that most of the solid effects are due to the pears at 1395 cm 1 associated to dCNH that matches the
formation of hydrogen bonds rather than to weaker forces as Van unassigned experimental shoulder at 1411 cm 1 (see SI). In the
der Waals forces between planes. Therefore, the optimized dimer 1100–1250 cm 1 region wavenumbers computed for the dimer
structure in vacuo, where intermolecular hydrogen bonds are pres- are almost unvaried with respect to the corresponding frequencies
ent, has been used to evaluate the effects of the hydrogen bonding in the monomer though the appearance of several new bands
in the solid. The use of the dimer is clearly a simplification of the changes the spectrum shape improving the agreement with the
real case but we expect the hydrogen bonding effects to be at least experimental spectrum.
qualitatively reproduced by the model. In the IR spectrum the mNH modes are also shifted to smaller
Comparison between the computed Raman and IR spectra of the wavenumbers improving the agreement with the experiment. In
monomer and the dimer is reported in Fig. 9. this case a very intense band at 3566 cm 1 is computed corre-
In the Raman spectrum of the dimer, the band conputed in the sponding to an asymmetric mNH in the hydrogen bonded NH. Con-
3400 cm 1 region is divided into not hydrogen-bonded mNH trarily to what retrieved for the Raman activities in the 1600 cm 1
(3459 cm 1) and hydrogen-bonded mNH (3423 cm 1). This duplica- region, mCO and mC@C IR relative intensities differences diminish
tion is due to the simplified model used, where only one side of the becoming equivalent manly due to an increase of intensity of the
molecule is bonded to another unit. In the solid, where all the NH asymmetric mCO band in line with the experiment where the mCO
are bonded to the neighboring molecules, band at 3459 cm 1 band at 1627 cm 1 is retrieved more intense than mCC at
should disappear. The hydrogen-bonded mNH is computed at short- 1614 cm 1. This band is also shifted to 1651 cm 1 in the dimer
er wavenumbers with respect to the monomer improving the improving the agreement with the experiment (1627 cm 1), with
agreement with the experiment. Also mC@C band at 1710 cm 1 is respect to the monomer (1660 cm 1) and to wavenumber com-
shifted to 1702 cm 1 in the dimer, being comparable with the puted in previous article [19] (1692 cm 1) although it is still quite
1692 cm 1 measured value. Other differences can be found in the overestimated. In the 1100–1250 cm 1 region the IR spectrum of
1600 cm 1 region where a new low intensity band appears at the dimer is almost unchanged with respect to the monomer, since
1653 cm 1 corresponding to mCO. On the other hand, frequencies bands in this region do not involve NH and CO modes.
at 1596 and 1631 cm 1 related to mCO and mC@C, increment their As it could be expected, all the perceptible changes in the vibra-
relative intensity differences and are shifted to 1585 and tional spectra of the hydrogen-bonded aggregates with respect to

Fig. 9. Comparison between the monomer (black) and the dimer (red) Raman (a) and IR (b) computed spectra. (For interpretation of the references in colour in this figure
legend, the reader is referred to the web version of this article.)
A. Amat et al. / Journal of Molecular Structure 993 (2011) 43–51 51

the monomer can be traced back to modes involving the CO and References
NH. Other modes affected involve the C@C bond that is tightened
with the aggregation. Obviously, our simple two-molecule model [1] J.H. de Graaff, The Colourful Past, Archetype Publications, 2004.
[2] K. Hunger (Ed.), Industrial Dyes, Wiley-Vch, 2003.
can not completely reproduce all the solid state interactions but [3] E.W. Fitzhugh (Ed.), Artists’ Pigments, vol. 3, Oxford University Press, 1997.
can give hints regarding the effects of the hydrogen bonds between [4] C.J. Cooksey, Biotech. Histochem. 82 (2007) 105–125.
molecules. [5] E. Fois, A. Gamba, A. Tilocca, Micropor. Mesopor. Mater. 57 (2003) 263–272.
[6] M.E. Fuentes, B. Pena, C. Contreras, A.L. Montero, R. Chianelli, M. Alvarado, R.
Olivas, L.M. Rodriguez, H. Camacho, L.A. Montero-Cabrera, Int. J. Quant. Chem.
4. Conclusions 108 (2008) 1664–1673.
[7] A. Tilocca, E. Fois, J. Phys. Chem. C 113 (2009) 8683–8687.
[8] L. Serrano-Andrés, B. Roos, Chem-Eur. J. 3 (1997) 717–725.
A combined experimental–theoretical strategy has been used in [9] M. Klessinger, W. Lüttke, Tetrahedron 19 (1963) 315–335.
this work to further understand the vibrational and electronic [10] D. Guillaumont, S. Nakamura, Dyes Pigments 46 (2000) 85–92.
properties of indigo, paying particular attention to the role of [11] A. Matsuura, H. Sato, W. Sotoyama, A. Takahashi, M. Sakurai, J. Mol. Struc.
(Theochem) 860 (2008) 119–127.
aggregation on the absorption and vibrational spectra. [12] D. Jacquemin, J. Preat, V. Wathelet, E. Perpète, J. Chem. Phys. 124 (2006)
The absorption spectra in vacuo and in different solvents have 074104.
been computed and compared to the experiment, evidencing that [13] D. Jacquemin, J. Preat, V. Wathelet, M. Fontaine, E.A. Perpète, J. Am. Chem. Soc.
128 (2006) 2072–2083.
TD-DFT calculations provide excellent results, able to reproduce [14] E. Perpète, J. Preat, J. André, D. Jacquemin, J. Phys. Chem. A 110 (2006) 5629–
even small bathocrhomic shifts. These shifts are related to a desta- 5635.
bilization/stabilization of the H/L and a decrease of the H–L gap. [15] C. Miliani, A. Romani, G. Favaro, Spectrochim. Acta A 54 (1998) 581–588.
[16] R.J.H. Clark, C.J. Cooksey, New J. Chem. 23 (1999) 323–328.
The appearance of a new band at ca. 700 nm in the absorption
[17] C.J. Cooksey, Dyes Hist. Archaeol. 16 (1998) 97–104.
spectrum of indigo solution at high concentrations has been inves- [18] S. Fantacci, A. Amat, A. Sgamellotti, Acc. Chem. Res. 43 (2010) 802–813.
tigated by simulating the absorption spectra of different size clus- [19] E. Tatsch, B. Schrader, J. Raman Spectrosc. 26 (1995) 467–473.
[20] M. Sánchez del Río, M. Picquart, E. Haro-Poniatowski, E.V. Elslande, V.H. Uc, J.
ter aggregates. Stable dimer and trimer aggregates in vacuo and in
Raman Spectrosc. 37 (2006) 1046–1053.
solution have been computed, revealing that aggregation takes [21] J. Tomkinson, M. Bacci, M. Picollo, D. Colognesi, Vib. Spectrosc. 50 (2009) 268–276.
place with the formation of hydrogen bonded structures. For the [22] A. Baran, A. Fielder, H. Schulz, M. Baranska, Anal. Methods 2 (2010) 1372–
studied indigo aggregates a series of new electronic transitions 1376.
[23] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman,
are computed. These transitions, retrieved at lower energies with J. Montgomery, T. Vreven, K.N. Kudin, J.C. Burant, J.M. Millam, S.S. Iyengar, J.
respect to the monomer ones, can be directly related to the exper- Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G.A. Petersson,
imental band at 700 nm, thus confirming this band as the signature H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T.
Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J.E. Knox, H.P. Hratchian,
of aggregation. Moreover, the optimized aggregates recall the indi- J.B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R.E. Stratmann, O.
go crystal structure, their intermolecular hydrogen bonds being Yazyev, A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, P.Y. Ayala, K.
similar the interactions present in the solid state. Morokuma, G.A. Voth, P. Salvador, J.J. Dannenberg, V.G. Zakrzewski, S.
Dapprich, A.D. Daniels, M.C. Strain, O. Farkas, D.K. Malick, A.D. Rabuck, K.
The IR and Raman spectra have been measured and computed, Raghavachari, J.B. Foresman, J.V. Ortiz, Q. Cui, A.G. Baboul, S. Clifford, J.
finding a good agreement between theory and experiment. Normal Cioslowski, B.B. Stefanov, G. Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.L.
modes have been analyzed by means of potential energy distribu- Martin, D.J. Fox, T. Keith, M.A. Al-Laham, C.Y. Peng, A. Nanayakkara, M.
Challacombe, P.M.W. Gill, B. Johnson, W. Chen, M.W. Wong, C. Gonzalez, J.A.
tion allowing to perform an unambiguous assignment to the func-
Pople, Gaussian 03, 2003.
tional groups. Only a few experimental bands remain not assigned [24] A.D. Becke, J. Chem. Phys. 98 (1993) 5648.
or present important differences with the computed values. Due to [25] C. Lee, W. Yang, R.G. Parr, Phys. Rev. B 37 (1988) 785.
[26] B. Miehlich, A. Savin, H. Stoll, H. Preuss, Chem. Phys. Lett. 157 (1989) 200.
the similarity between the optimized aggregates and the crystal in-
[27] M. Cossi, N. Rega, G. Scalmani, V. Barone, J. Comp. Chem. 24 (2003) 669–681.
digo structure, the effects of the hydrogen bonds present in the so- [28] A.P. Scott, L. Radom, J. Phys. Chem. 100 (1996) 16502–16513.
lid on the vibrational properties have been evaluated by computing [29] M.H. Jamróz, Vibrational Energy Distribution Analysis: VEDA 4.0 program,
the Raman and IR spectra on the optimized dimer aggregate. Re- Warsaw, 2004.
[30] E. Wille, W. Luttke, Angew. Chem., Int. Ed. Engl. 10 (1971) 803.
sults point out the main differences regarding the NH and CO [31] G. Miehe, P. Süsse, V. Kupcik, E. Egert, M. Nieger, G. Kunz, R. Gerke, B. Knieriem,
stretching, that are shifted to shorter wavelengths, and modes M. Niemeyer, W. Lüttke, Angew. Chem. Int. Edit. 30 (1991) 964–967.
involving C@C bond, tightened with aggregation. Some of the [32] S.E. Sheppard, P.T. Newsome, J. Am. Chem. Soc. 64 (1942) 2937–2946.
[33] G. Haucke, G. Graness, Angew. Chem. Int. Ed. 34 (1995) 67.
uncertainties and inaccuracies in the assignment were improved [34] E. Wille, W. Lüttke, Angew. Chem. 10 (1971) 803.
by computing the vibrational spectra of the dimer. [35] M. Klessinger, W. Lüttke, Chem. Ber. 99 (1966) 2136.
This work gathers important information for the comprehen- [36] P.W. Saddler, J. Org. Chem. 21 (1956) 316.
[37] G. Haucke, R. Paetzold, Nova Acta Leopold 11 (1978) 1.
sion of the spectroscopic properties of indigo, one of the most stud- [38] M. Klessinger, W. Lüttke, Tetrahedron 19 (1963) 315.
ied natural dyes present in nature. The computational strategy [39] A.R. Monahan, J.E. Kuder, J. Org. Chem. 37 (1972) 4182.
described and successfully used in this study, can be extended to [40] M.I.G. Travasso, P.C.S. Santos, A.M.F. Oliveira-Campos, M.M.M. Raposo, N.
Prasitpan, Adv. Col. Sci. Tech. 6 (2003) 95.
the investigation of the optical and vibrational properties of differ-
[41] W.R. Brode, G.M. Wyman, J. Am. Chem. Soc. 73 (1951) 4267.
ent dyes of interest in the art heritage field, thus complementing [42] G. Miehe, P. Süsse, V. Kupcik, E. Egert, M. Nieger, G. Kunz, R. Gerke, B. Knieriem,
the experimental results. M. Niemeyer, W. Lüttke, Angew. Chem. 30 (1991) 964.
[43] J. Weinstein, G.M. Wyman, J. Am. Chem. Soc. 78 (1956) 2387.
[44] W.R. Brode, E.G. Pearson, G.M. Wyman, J. Am. Chem. Soc. 76 (1954) 1034.
Acknowledgements [45] G. Pfeiffer, W. Otting, H. Bauer, Angew. Chem. 15 (1976) 52.
[46] R. Pummerer, G. Marondel, Chem. Ber. 99 (1960) 2834.
This work supported by the project Eu-ARTECH (CT-2004- [47] W. Lüttke, D. Hunsdiecker, Chem. Ber. 99 (1966) 2146.
[48] W. Lüttke, M. Klessinger, Chem. Ber. 97 (1964) 2342.
506171) within the 6th FP of the EC and the European Union [49] R. Gerken, L. Fitjer, P. Müller, I. Usón, K. Kowski, P. Rademacher, Tetrahedron
FP7-Research Infrastructure programme under the CHARISMA pro- 55 (1999) 14429.
ject (GA228330). [50] P. Süsse, M. Steins, V. Kupcik, Z. Kristallogr. 184 (1988) 269–273.
[51] M. Wanko, M. Garavelli, F. Bernardi, T. Niehaus, T. Frauenheim, M. Elstner, J.
Chem. Phys. 120 (2004) 1674.
Appendix A. Supplementary data [52] D.J. Tozer, R.D. Amos, N.C. Handy, B.O. Roos, L. Serrano-Andrés, Mol. Phys. 97
(1999) 859.
[53] A. Dreuw, J.L. Weisman, M. Head-Gordon, J. Chem. Phys. 119 (2003) 2943.
Supplementary data associated with this article can be found, in [54] N. Peica, W. Kiefer, J. Raman Spectrosc. 39 (2008) 47–60.
the online version, at doi:10.1016/j.molstruc.2010.11.046. [55] M. Halls, J. Velkovski, H. Schlegel, Theor. Chem. Acc. 105 (2001) 413–421.

You might also like