Vibrational Spectroscopy 50 (2009) 268–276
Contents lists available at ScienceDirect
Vibrational Spectroscopy
journal homepage: www.elsevier.com/locate/vibspec
The vibrational spectroscopy of indigo: A reassessment
John Tomkinson a,*, Mauro Bacci b, Marcello Picollo b, Daniele Colognesi c
a
Science and Technology Facilities Council, The ISIS Facility, The Rutherford Appleton Laboratory, Chilton, OX11 0OX, UK
Istituto di Fisica Applicata ‘‘Nello Carrara’’ (IFAC-CNR), Via Madonna del Piano 10, 50019 Sesto Fiorentino (Firenze), Italy
c
Istituto dei Sistemi Complessi (ISC-CNR), Via Madonna del Piano 10, 50019 Sesto Fiorentino (Firenze), Italy
b
A R T I C L E I N F O
A B S T R A C T
Article history:
Received 28 July 2008
Received in revised form 5 January 2009
Accepted 19 January 2009
Available online 30 January 2009
We report the neutron vibrational spectra of indigo and its model compounds thioindigo and isatin. The
neutron data extend the low energy range of the vibrational spectra of these molecules. The assignments,
made with the help of ab-initio calculations, give convincing fits between the observed and scaled
calculated results, and correct errors in the published literature. The indigo eigenvectors are described in
terms of those of its model compound isatin. Finally, candidate modes, that could be used to study
indigoids in matices (e.g. ‘Maya Blue’), are selected.
ß 2009 Elsevier B.V. All rights reserved.
Keywords:
Indigo
Thioindgo
Isatin
Inelastic neutron scattering
Maya blue
1. Introduction
Indigo is a dyestuff that is not only of current commercial
interest [1] but also of considerable historical importance. (The
structures of indigo and the related molecules discussed in this
paper are given in Fig. 1.) Since ancient times it has been widely
used as a dye and a pigment [2]. It has also been found in ‘Maya
Blue’, a very stable blue pigment used by the ancient inhabitants
of Mexico for pottery and mural paintings. Here, the indigo
chromophore is inserted in an inorganic matrix (palygorskite)
[3,4,5]. Despite numerous studies of the spectroscopic behaviour of
indigo it is far from being completely understood. Both electronic
and vibrational spectra are very sensitive to the local environment
around the indigo molecule: for instance the main absorption band
in the visible region shifts from 540 nm (gas phase) to 588 nm in
non-polar solvents and to 606 nm in polar solvents or in the solid
state [6]. These shifts in lmax are probably related to the details of
the intermolecular bonding pattern immediately around the
chromophore, as has been demonstrated by the recent success
in obtaining lmax from ab-initio calculations [7].
However, its vibrational spectra are even more puzzling,
especially in the comparison of pure indigo and ‘Maya Blue’
[8,9,10,11]. Although the vibrational features related to the N–H
group are strongly affected when indigo is inserted into clay, it is
* Corresponding author. Tel.: +44 1235 44 6686; fax: +44 1235 44 5383.
E-mail address: j.tomkinson@rl.ac.uk (J. Tomkinson).
0924-2031/$ – see front matter ß 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.vibspec.2009.01.005
still unknown if this is due to a change in the hydrogen bonding
and, or, the formation of new species such as dehydro-indigo [9].
Fortunately, this spectral sensitivity to its environment offers the
opportunity for a better understanding of ancient artefacts at a
molecular level and hence access to the processes that were used in
their creation. Central to extracting this information, and understanding ‘Maya Blue’, is a reliable appreciation of the vibrational
spectroscopy of pure indigo that, until now, has been lacking.
Further, the low energy spectral region of indigo and related
molecules, a region that might be expected to be most sensitive to
changes in the local environment, is unknown.
Only one published study has addressed the entire vibrational
spectrum of indigo [12], it also provides a convenient entry to
earlier literature where the focus was on the vibrations of the
chromophoric centre, especially n(C C) and n(C O). This work
organised infrared, Raman and ab-initio calculations, to produce a
complete assignment scheme. Sadly, it is marred by a chaotic mode
numbering scheme and the assignment of more fundamental
transitions than the theoretical maximum, 84. Some of these errors
were partly addressed by more recent work [13] but it is still
referenced by workers in the field and even ‘‘substantially
confirmed’’ by some [14]. The magnitude of the problem can be
gauged from Fig. 2, which compares this assignment scheme [12]
with our new data, clearly it is inadequate.
Without settling these issues, and generating a good assignment scheme, a thorough understanding of ‘Maya Blue’ as an
organic-inorganic composite is unlikely to appear. As a basis for
our own future work in this field we have undertaken the study of
J. Tomkinson et al. / Vibrational Spectroscopy 50 (2009) 268–276
269
involving hydrogen motions and, as a consequence, direct study of
the chromophore vibrations, which involve almost pure n(C C)
and n(C O), is difficult. However, it has been demonstrated that
the technique is ideal for the study of organic molecular crystals,
especially those isolated in inorganic matrices, it is described in
detail elsewhere [15].
We shall assign the vibrational spectra of two model
compounds of indigo; namely, thioindigo and isatin. These
assignments will then be used to help assign the vibrations of
indigo. Since isatin is, roughly, one half an indigo molecule (see
Fig. 1), its vibrational eigenvectors are well suited to use as
descriptors of the indigo vibrations. This approach addresses the
inadequacy of describing molecular eigenvectors of larger
molecules like indigo by the motions of a few functional groups,
as has already been touched upon [13]. We shall also discuss the
choice of spectral bands that can be used to determine the nature of
indigoids in inorganic matrices.
Fig. 1. The molecular structures of isatin (above) and indigo (below). The structure
of thioindigo is that of indigo with the NH atoms replaced by sulfur atoms at the
nitrogen positions.
indigo, through its model compounds thioindigo and isatin, using
Inelastic Neutron Scattering (INS) spectroscopy. The results we
report here address, mostly, the correct assignment of these
spectra and the identification of those candidate(s) in the
vibrational spectrum of indigo that will allow its recognition in
mixtures like Maya Blue. (Systems, like ‘Maya Blue’ itself and the
Roman violet lakes, will be considered elsewhere.)
Molecular vibrational spectroscopy with neutrons is a well
established technique that makes use of modern spallation
neutron sources to obtain spectral bandwidths similar to those
obtained in the optical spectroscopy of solids, for a comparison of
INS and optical spectra see [15]. INS has the advantage that its
observed intensities can be simply compared to calculated spectral
intensities. These intensities are obtained straightforwardly from
the mean square vibrational atomic displacements that appear as
output from standard ab-initio programs. This advantage is not
readily available to results from optical spectroscopies and the
difficulty of matching calculated and observed Raman intensities
can be seen from the published literature (see especially Table 1 of
Ref. [13]). The INS technique emphasises those vibrational modes
Fig. 2. A comparison of the observed inelastic neutron scattering spectrum of indigo
(black trace) with that calculated for the currently accepted assignment scheme
(red trace), after [12]. This figure should be compared to the results shown in Fig. 7.
(For interpretation of the references to color in this figure legend, the reader is
referred to the web version of the article.)
2. Molecular and crystal structures, mode numbering
Isatin, see Fig. 1, crystallises in the P21/c (C2h5) space group
with four molecules in the unit cell. Pairs of molecules are
hydrogen bonded, head to head, into centrosymmetric dimers
that interact only weakly with their neighbours [16,17]. Each
dimer can be imagined to be produced from a single indigo
molecule; by rupturing it at the central C C bond, translating the
moieties laterally in the plane, about 2.8 Å´, and terminating both
free valence, C , with an oxygen atom. The isatin molecule is
planar and its 42 fundamental modes are distributed over the
representation: Gvib = 29 A0 + 13 A00 . It is appropriate to note here
that we shall follow the conventional approach to spectroscopic
mode numbering [18]. Modes are first grouped in blocks
according to their symmetry species and the mode frequencies
are ordered in decreasing value within each character block. In the
cases discussed here, the ordered frequency values are those given
by the unscaled ab-initio calculations described below. We shall
use the monomer form of isatin as our basis for its mode
numbering scheme, A0 (1–29) and A00 (30–42). The isatin dimer is a
planar, high symmetry (C2h), arrangement of the monomers and
its 90 fundamentals are distributed over the representation:
Gvib = 31 Ag + 15 Au + 14 Bg + 30 Bu.
Thioindigo can be crystallised in the P21/c (C2h5) space group
with two molecules in the unit cell [19] but our commercial
sample is probably P21/n [20]. We shall treat the molecule as
planar, C2h, with its 78 fundamental modes distributed over
the representation: Gvib = 27 Ag + 13 Au + 12 Bg + 26 Bu, and its
numbering scheme is; Ag (1–27), Au (28–40), Bg (41–52) and Bu
(53–78),
Indigo, see Fig. 1, crystallises in the P21/c (C2h5) space group with
two molecules in the unit cell [21,22]. Apart from the obvious
presence of intra-molecular hydrogen bonds (since the sulfur of
thioindigo has been replaced by an NH group), there is also evidence
of close intermolecular contacts in the crystal. These give rise to
multicentred, non-linear, hydrogen bonds [22]. The molecule is
planar, C2h, and its 84 fundamental modes are distributed over the
representation: Gvib = 29 Ag + 14 Au + 13 Bg + 28 Bu, and its mode
numbers are Ag (1–29), Au (30–43), Bg (44–56) and Bu (57–84). This
numbering is in contradistinction to the confusion found in
other published work, Ag (1–29), Au (28–35, 37–40), Bg (41–52)
and Bu (36, 51, 56–84) with the g(NH)(Au) mode left unnumbered,
Table 1 of Ref. [12].
3. Calculational details
Our calculations were performed in GAUSSIAN98 [23] for the
isolated molecules and the isatin dimer (C2h). The molecular
270
J. Tomkinson et al. / Vibrational Spectroscopy 50 (2009) 268–276
geometries were constrained planar and, using Density Functional
Theory (DFT), were optimised at the B3LYP level with the 6-31G**
basis set. From this geometry the harmonic vibrational frequencies
and atomic displacement vectors were produced, to be used in
ACLIMAX, see below. Preparative to the DFT calculation on indigo a
repeat of earlier work [12,13] was undertaken at the Hartree Fock
(H-F) level, with the 3-21G basis set. The results from this
calculation (when used with ViPA, see below) enabled our DFT
results to be accurately correlated to the earlier assignment
scheme of indigo [12] on a mode by mode basis, thus enabling us to
produce Fig. 2.
There are several approaches to handling the results of ab-initio
calculations, some workers prefer to leave them unchanged whilst
others scale the ab-initio eigenvalues according to more or less
complex schemes. Such approaches have, as their objective, the
aim of testing the adequacy of the different calculational schemes
in reproducing the observed eigenvalues. Our approach is different,
we accept the calculational scheme is adequate and stress the
importance of the eigenvectors. Fortunately, eigenvectors are not
very sensitive to the details of a given calculational scheme [15].
Thus, we shall assume that our calculated eigenvectors are
acceptable but recognise that calculational inadequacies produce
somewhat erroneous eigenvalues. If we are to fully exploit the
calculated eigenvectors in understanding our observed INS spectra
we must overcome the problem of poorly calculated eigenvalues.
Since the eigenvectors are acceptable, if only the eigenvalues were
correct the calculated INS spectrum would reproduce that
observed, within the limits of theory, as would be obvious from
simple visual inspection. We shall correct our eigenvalues by
applying individual scaling parameters.
3.1. Spectral comparison
The vibrations of structurally related molecules can be profitably compared through the form of the atomic displacements in
given modes, their eigenvectors. This comparison might be
attempted by inspection using commercial visualisation packages
but is subjective, non-quantitative and, so, unusable. Fortunately,
the comparison can be made quantitative using the ViPA program
[24]. The normal modes of a given molecule are projected onto the
orthonormal vector space defined by the normal modes of the
reference molecule. Any one mode’s projection is then reported in
terms of the percentage of each of the reference molecule’s modes.
Since the vibrational displacements given by GAUSSIAN are not
mass-weighted Cartesian displacement coordinates, and so not
orthonormal, ViPA produces its own normal modes from the
Cartesian force-fields generated by GAUSSIAN [24]. As is intuitively
correct, and given by ViPA, the C–H stretching vibrations of isatin
are the same as those of indigo, to better than 99%. The mode
descriptions given in Tables 2a and 2b are a very condensed version
of the full ViPA output and enable quantitative comparisons to be
made back, from the vibrations of isatin and indigo, to the
vibrations of benzene and pyrrole-2,3-dione.
4. Experimental
4.1. Samples
The samples of isatin (98%) and commercial dyestuffs
thioindigo and indigo (95%) were obtained from Aldrich, they
were used without further treatment.
4.2. Spectrometer and data visualisation
The samples, about 5 g, were wrapped in aluminium foil and
held in flat sample cells. These were maintained at 20K in the
neutron beam of the TOSCA spectrometer [25], at the ISIS Facility,
The Rutherford Appleton Laboratory, Chilton, OX11 0QX, UK.
TOSCA is a pulsed neutron, indirect geometry, low band-pass
spectrometer with good spectral resolution (DEt/Et 2%) [25].
Data were collected for about 8 h and transformed into the
conventional scattering law, S(Q,v) (arbitrary units), vs. energy
transfer, Et (cm1), using standard programs.
The INS spectra were displayed and compared with the
calculated spectra from GAUSSIAN98 using ACLIMAX [26]. This
free-ware program uses the GAUSSIAN output files (.LOG) to
generate the TOSCA spectrum, although the output from other abinitio programs can also be used. In a simplified form; the INS
intensity, S(Q,vn)l, of the mode, n, observed at the energy transfer,
v, and momentum transfer, Q (determined by the neutron
spectrometer), is given by
X
v u2 Þ
(1)
SðQ 1 vv Þl ¼ ðQ 2 v u2l ÞexpðQ 2
l
V
n
where u2l, is the mean square displacement of the atom, l, in the
mode, as given by GAUSSIAN. A full discussion of the details of INS
intensities is given elsewhere [15]. The exponent shown in Eq. (1)
is the same Debye–Waller factor that will be familiar from
diffraction work and the more an atom moves the smaller will be
the D–W value. Isolated molecule calculations omit those atomic
displacements caused by the, external, lattice modes and to correct
for this an extra term is added to the internal displacements
calculated by GAUSSIAN98.
X
X
X
v u2 þ
v u2 ¼ ða Þ þ ða Þ
v u2 ¼
(2)
l int
l ext
l
l
l
v
internal
external
A typical value for the external contribution term, (al)ext, would
be that of benzene, 0.025 Å´2; for larger, or stiff, molecular systems,
this will fall and for smaller, or limp, molecules it will rise [15].
5. Results
5.1. Thioindigo
The initially calculated, or unscaled, and observed spectra of
thioindigo are detailed in Table 1. The two spectra are very similar,
with bands of similar intensity overlapping one another (either
partly or wholly). The n(CH) stretching region is remarkably sharp
and structured. This structure was freely fitted to three Gaussians
(at 3035, 3091 and 3152 cm1), which we identify as the n(CH)
band origin and its first two phonon-wings [15]. The band origin
value agrees reasonably with the average value of the n(CH),
Raman, bands, 3020 cm1 [12]. The external modes are thus
located below about 60 cm1 (=3091–3035) and their limit nicely
corresponds to the sharp edge at 69 cm1 seen in the observed
spectrum. The observed spectrum below 69 cm1 was taken as
representative of the phonon-wing spectrum. A value for (al)ext
(see above) was obtained, 0.008 Å´2, from the ratio of the band
intensities in the n(CH) region [15]. This value is low compared to
typical values for organic molecules, see above, and is a
consequence of the heavy molecular mass and stiffness of
thioindigo.
As discussed above, we shall exploit the results from the
GUASSIAN98 calculations by individually scaling the calculated
eigenvalues. The process of extracting the necessary individual
scaling parameters presupposes a knowledge of the optical
characters of the molecular transitions as well as the calculated
eigenvectors. In the case of thioindigo the Raman data was used to
identify the Ag modes. (These frequencies were read directly from
Fig. 3 of [12], we estimate that the frequency values are accurate to
about 3 cm1.) We begin at the lowest frequencies, which has the
advantage that fixing a fundamental also identifies its overtones and,
271
J. Tomkinson et al. / Vibrational Spectroscopy 50 (2009) 268–276
Table 1
A comparison of the reported infra-red and Raman spectra of thioindigo with the INS spectrum of this work, in cm1, below 2000 cm1. Both the unscaled (‘DFT’) and the
scaled (‘Assigned’) eigenvalues of the GAUSSIAN98 calculations, see text, are also reported.
This work
DFT
29
74
78
90
137
152
203
213
230
230
270
282
305
407
425
426
445
462
487
492
[12]
INS
62
90 br
99 br, sh
121
170
161
205
236
–
–
280
297
392
421
414
442
462
Assigned
Raman
This work
ir
69
88
97
95
121
170
161
205
236
227
230
270
280
297
392
421
414
445
462
480
480
502
536
594
676
691
720
489
502
536
542
605
676
691
691
703
720
750
782
794
877
877
ir
877
117
154
207
226
237
925
961
996
1022
1075
937
961
961
996
996
1022
1022
1066
1075
1075
295
1134
1138
1126
1126
932
968
1016
1026
1054
1074
1115
1126
1092
1120sh
1195
1163
1219
1228
1281
1293
1317
1321
1124
453
1140
1186
1187
1238
1243
1310
1312
1361
1361
484
532
1168
1224
1270
1290
1323
1126
1168
1168
1224
1224
1290
1290
1325
1325
601
1391sh
1402
679
691
721
1485
1486
1490
1490
1453
1467
1463
1463
1463
1463
750
1517
754
787
794
794
Raman
910
937
973
973
1002
1002
1041
1042
1066
1084
1095
803
824
882
883
Assigned
877
753
754
787
797
797
[12]
INS
889
489 br
505
511
539
542
605
686
695
697
703
729
750
DFT
883
Fig. 3. A comparison of the observed inelastic neutron scattering spectrum of
thioindigo (blue trace) with that calculated for the scaled ab-initio assignment
scheme of Table 1 (red trace). The external mode region appears below 70 cm1.
(For interpretation of the references to color in this figure legend, the reader is
referred to the web version of the article.)
1574
1619
1623
1634
1639
1715
1739
1574
1619
1623
1634
1639
1715
1739
1459
1457
1467sh
1503
1523
1579
1590
1594
1656
1674
Fig. 4. A comparison of the observed inelastic neutron scattering spectrum of isatin
(green trace) with that calculated for the isatin-dimer (red trace), as initially given
by GAUSSIAN98 and left unscaled. (Here the contributions from the NH hydrogen
atoms have been omitted from the calculated spectrum.) (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of
the article.)
272
J. Tomkinson et al. / Vibrational Spectroscopy 50 (2009) 268–276
Table 2a
A comparison of the ir, Raman, INS and unscaled calculated eigenvalues, cm1, of isatin (20 K). Results of both the monomer and dimer calculations are given. The order is
given by the mode numbers of the, Cs, monomer and (A0 )g = Ag, (A0 )u = Bu, (A00 )g = Bg, (A00 )u = Au. Dimer mode numbers, therefore, occasionally appear out of sequence. Mode
numbers for both the monomer, v0 , and dimer, n, are given.
Calculated
Monomer
Observed
INS
This work
v0
Mode Descriptions
x
Dimer
n
u
46
45
90
17
33
51
n
20 K
g
31
60
30
59
62
116
ir
ir
[27]
[29]
Raman
[27]
300 K
80 K
300 K
A00
A00
A0
A0
A00
A0
Description, of dimer vibrations external to the monomer
Antiwag of monomers
Antitwist of monomers
Antirock of monomers
Antitranslation of monomers, along dimer
Antitranslation of monomers, out of plane
Antitranslation of monomers, across dimer
Description, of monomer vibrations
00
42
41
40
29
28
99
140
260
270
314
44
43
42
89
88
116
143
271
275
347
59
58
57
29
28
117
143
271
275
334
A
A00
A00
A0
A0
39
38
27
391
429
484
41
40
87
403
473
486
56
55
27
406
481
491
A00
A00
A0
138
167
275
273
315
334
392
459
480
26
36
37
25
24
37
35
23
546
551
514
647
658
514
731
733
86
39
553
552
85
84
36
38
83
650
675
812
734
742
26
54
51
25
24
551
553
789
648
666
53
23
734
739
A0
A00
A00
A0
A0
A00
A00
A0
550
537
627
657
657
686
717
733
34
33
32
22
768
821
879
889
37
35
34
82
767
826
887
896
52
50
49
22
766
822
887
892
A00
A00
A00
A0
768
810
879
(889)
21
31
30
20
19
944
971
998
1041
1116
81
33
32
80
79
955
971
999
1041
1116
21
48
47
20
19
954
971
999
1041
1116
A0
A00
A00
A0
A0
(944)
951
990
1008
1092
18
17
1167
1195
78
77
1178
1235
18
16
1177
1232
A0
A0
1143
1160
16
1211
76
1210
17
1210
A0
1185
15
14
13
1261
1308
1357
75
74
73
1290
1329
1365
15
14
13
1289
1332
1364
A0
A0
A0
1268
1308
1343
12
11
10
1411
1503
1516
72
71
70
1455
1503
1523
12
11
10
1466
1503
1524
A0
A0
A0
1413
1460
1480
9
8
1642
1665
69
68
1645
1663
9
8
1645
1669
A0
A0
(1606)
(1621)
7
6
1816
1835
66
1818
6
7
1818
1774
A0
Ag
Bu
(1736)
(1740)
67
1800
273
273
336
388
453
478
537
554
–
639
659
659
670
720
734
755
766
816
876
884
915
925
951
952
989
1015
1095
1111
1145
1155
1180
1189
1202
1218
1255
1267
1282
1333
1363
1377
1405
1466
1485
1538
1587
1615
1630
1677
1694
1727
1736
336
383
459
481
540
550
–
638
664
664
700
720
738
754
775
1746
140
173
257
276
335
–
–
488
–
–
551
Antitwist, Bz against Py
Antiwag (butterfly), Bz against Py
Bz[33(20), 10(4)]: Py[51(24)]
Antirock, Bz against Py
Bz[–]: Py[92(17)]
Bz[54(20)]: Py[52(23), 14(20)]
Bz[48(20)]: Py[27(22), 17(21), 14(20)]
Antitranslation, Bz against Py
–
736
Bz[13(20)]: Py[26(13), 21(16),14(15)]
Bz[39(20), 22(8)]: Py[37(21)]
g(NH)
Bz[61(18)]: Py[46(16), 11(13)]
Bz[34(18)]: Py[55(15), 12(21), 11(14)]
g(NH)
Bz[60(8), 20(4)]: Py[29(18), 16(21)]
Bz[33(18), 11(14)]: Py[36(14), 16(10), 11(2)]
–
–
–
886
Bz[42(11), 41(4)]: Py[–]
Bz[48(11)]: Py[70(19), 12(20)]
Bz[53(11), 30(19), 12(2)]: Py[–]
Bz[38(7)]: Py[31(11), 18(13)]
–
–
950
–
1016
1100
Bz[19(6), 15(14), 15(17),10(10)]: Py[30(14), 27(12)]
Bz[83(19), 16(11)]: Py[–]
Bz[61(2), 38(19)]: Py[–]
Bz[61(14), 24(6)]: Py[–]
Bz[45(14), 23(17), 18(7)]: Py[18(8), 11(7)]
1152
Bz[20(17), 18(10), 11(6)]: Py[34(12)]
Bz[33(10), 27(17)]: Py[21(12), 14(9)]
1192
1208
1220
–
1266
1302
1338
–
–
1424
1466
Bz[46(17), 18(7)]: Py[12(9)]
1606
1621
1735
1740
Bz[16(17), 14(14), 11(7)]: Py[24(11), 21(10), 18(8)]
Bz[28(3), 26(9)]: Py[31(9)]
Bz[44(9),39(3)]: Py[12(8), 10(9)]
Bz[24(9)]: Py[34(7), 12(8), 11(2)]
Bz[81(13)]: Py[38(6)]
Bz[80(13)]: Py[–]
Bz[94(16)]: Py[31(6), 19(2)]
Bz[74(16)]: Py[37(3), 24(7)]
Bz[–]: Py[95(5)]
Bz[–]: Py[96(4)]
Our mode descriptions of the internal vibrations of the monomer are taken from the GAUSSIAN98 output and given with respect to the modes of the parent cycles benzene
and pyrrole-2,3-dione, after [24]; Bz[x(y)] implies that this isatin mode has x% of the benzene mode number y [30], similarly for pyrrole-2,3-dione, Py, see Table 2b below (only
contributions greater than 10% are shown).
273
J. Tomkinson et al. / Vibrational Spectroscopy 50 (2009) 268–276
so, prevents strong overtones being mistaken for fundamentals. Then,
taking the lowest calculated frequency, we search for nearby
observed transitions, not only of similar INS intensity but also which
possess an appropriate optical character. Thus, for example, if the
optical character of the transition identifies it as an Ag mode it can
only be associated with a calculated Ag eigenvector. The observed
band is thus assigned to a specific ab-initio eigenvector and
eigenvalue, we then scale the calculated eigenvalue to equal the
observed value, thus each transition has its individual scaling
parameter. Above about 1200 cm1 the INS features weaken and
become undifferentiable, consequently, in this region, our assignment
scheme is mostly dependant on the optical data previously published
[12]. The resulting scaled spectrum clearly agrees very well with
observation, see Fig. 3, and adds support to the assignments of
Table 1.
5.2. Isatin
The observed INS spectrum of isatin was compared with the,
unscaled, calculated spectrum of the dimer using ACLIMAX, a
working value for the (al)ext parameter was found, 0.025 Å´2, which
is more typical of organic systems, see above.
As was the case for thioindigo, the overall distribution of the
calculated bands corresponded well with the observed spectrum.
Except, that is, for two strong features calculated at 789 and
812 cm1, these had no immediate counterparts in the observed
spectrum. The calculated eigenvectors of both modes showed
them to be, predominantly, the out-of-phase and in-phase
components of the out-of-plane N–H vibration, g(NH). To
emphasise this in ACLIMAX, the N–H hydrogen’s neutron
scattering cross-section was set to zero and the observed and
calculated spectra again compared. Whilst several features
suffered minor intensity loss the 789 and 812 cm1 bands were
almost completely eliminated. This left most of the remaining
calculated features in convincing register with the observed
spectrum. This is shown in Fig. 4 and detailed in Table 2. Working
as outlined above, for thioindigo, the calculated frequencies were
scaled to those of nearby observed INS features of the same
intensity and appropriate optical characters [27]. The scaled
results reported in Table 2 are unambiguously assigned, at least
below 1200 cm1, and constitute the first reliable assignment
scheme for isatin.
At the end of this process two observed bands remain to be
assigned, at 627 and 686 cm1, these are known to be out-of-plane,
A00 , modes from optical work [27] and the region is typical for
Fig. 5. A comparison of the observed inelastic neutron scattering spectrum of isatin
(green trace) with that calculated for the scaled ab-initio assignment scheme of
Table 2 (red trace), NH contributions included. The external mode region appears
below 120 cm1. (For interpretation of the references to color in this figure legend,
the reader is referred to the web version of the article.)
g(NH) modes. Indeed, the earliest optical work assigned g(NH) at
742 cm1 (misprinted as 724 cm1 in the original table) [28]. This
was challenged by latter work, where it was assigned at 670 cm1
(at 300 K), increasing to 690 cm1 (at 110 K) [27] but more recent
work has these positions as 681 cm1 (at 300 K), and 700 cm1 (at
80 K) [29]. Clearly the positions of these bands, unlike all other
spectral features of isatin, are rather sample dependant and we
assign our calculated ungerade g(NH) 812 cm1 band to the
observed INS feature at 686 cm1. The discrepancy between
the ab-initio g(NH) band position and that observed is due to the
overly short N. . .O length found in the dimer calculation.
Consequently, the g(NH) frequency should be less than that
calculated in the dimer (where it has an average of 801 cm1) but
more than in the, non-H-bonded, monomer (where it was
calculated at 514 cm1). Correspondingly, the calculated dimer
gerade g(NH) mode at 789 cm1, is assigned to the INS band at
627 cm1. The final, scaled, spectrum clearly agrees very well with
observation, see Fig. 5, and adds support to the assignments of
Table 2.
Finally, each vibrational eigenvector of isatin was correlated to
its equivalent mode in thioindigo, using ViPA. This result will be
Table 2b
The mode numbers, n, unscaled calculated eigenvalues, cm1, and brief description of the vibrations of pyrrole-2,3-dione, for use in conjunction with Table 2a.
n
A0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Calculated frequency (cm1)
3659
3280
3242
1846
1806
1621
1428
1353
1248
1127
1084
1040
794
740
616
555
313
Description
n (NH)
n (CH) in-phase
n (CH) anti-phase
n (C(2) O)
n (C(3) O)
n (C C)
d (CH) in-phase, d (NH) anti-phase
d (CH), d (NH) – all in-phase
n (HC–CO), n (OC–CO), n (OC–N), d (C C–C)
d (CH) anti-phase, d (NH)
n (C(5)–N), d (CH)
n (C(2)–N), d (CH)
d (C C–C)
n (OC–CO), d (C C–C)
d (ring)
d (C O) in-phase
d (C O) anti-phase
n: stretching vibrations; d: in-plane angle bending vibrations; g: out-of-plane vibrations.
n
Calculated frequency (cm1)
Description
A00
18
19
20
21
22
23
24
956
827
734
537
454
283
139
g (CH) anti-phase
g (C O) anti-phase, g (CH) in-phase
g (CH) in-phase
g (C O) in-phase
g (NH)
g (ring), g (NH)
g (ring)
274
J. Tomkinson et al. / Vibrational Spectroscopy 50 (2009) 268–276
Table 3
A comparison of the reported infra-red and Raman spectra of indigo with the INS
spectrum of this work, in cm1, below 2000 cm1. The mode numbers, n, and results
of the GAUSSIAN98 calculations (both DFT and H-F, see text) are also reported. The
DFT are unscaled and the H–F values have been scaled by 0.89 [12]. The scaled DFT
values, see text, are given under ‘Assigned’. (Where they differ from our assignment,
the characters of the optical transitions are given immediately following their
wavenumber value.).
v0
n
x
6
8
8
9
9
7
7
10
10
11
11
12
12
13
14
6
62
7
63
8
9
64
65
10
66
11
67
12
13
69
13
14
15
68
14
15
Ag
Bu
Ag
Bu
Ag
Ag
Bu
Bu
Ag
Bu
Ag
Bu
Ag
Ag
Bu
Bu
Bu
Ag
Ag
15
17
16
16
17
18
18
70
16
17
71
72
18
73
Bu
Ag
Ag
Bu
Bu
Ag
Bu
Optical results
This work
Observed
INS
Raman
ir
Raman
[12]
[12]
[13]
1701
1701
1627
1625
1625
1614
1582
1571
1582
1574
1585
1483
1482
1483
1505
1463
1464
1365
1356
1462
1460
1408
1400
1365
1392
1317
1299
1310
1248
1295
1310
1248
1257
1220
1224
1190
1224
1190
1199
1174
1147
74
19
75
Bu
Ag
Bu
20
20
30
30
21
31
31
20
76
44
30
21
31
45
Ag
Bu
Bg
Au
Ag
Au
Bg
22
22
32
32
33
33
23
23
34
34
35
25
35
24
37
37
24
25
36
26
77
22
32
46
47
33
78
23
34
48
49
79
35
24
36
50
80
25
51
81
Bu
Ag
Au
Bg
Bg
Au
Bu
Ag
Au
Bg
Bg
Bu
Au
Ag
Au
Bg
Bu
Ag
Bg
Bu
36
26
27
37
26
82
Au
Ag
Bu
1182
1147
1128
1121
21
19
19
1220
1139
1128
1105
1096
1097 Bg
1097 Bg
1074
1038
1015 Ag
1090
1012 Bg
1011
1005
989
979
964
970
966
947
940
921
940
921
934
879
868
862
858
797
854
797
765
758
760
788
746
747
699
674
675
716
682
646
637
635
598
597
566
633
600
591
559
588
538
544
542
544
508
508
Assigned
DFT
H–F
1701
1627
1625
1614
1582
1574
1585
1505
1505
1464
1464
1400
1400
1356
1356
1748
1668
1675
1636
1635
1619
1617
1534
1532
1508
1507
1465
1432
1398
1352
1746
1586
1592
1559
1559
1677
1694
1471
1470
1448
1452
1396
1364
1315
1297
1295
1295
1384
1372
1288
1234
1243
1197
1284
1272
1234
1237
1221
1203
1198
1234
1225
1196
1161
1140
1122
1073
1108
1138
1136
1014
1080
1082
1049
1045
1032
1031
965
998
998
991
987
1056
1056
918
1022
1023
904
883
904
902
822
820
788
780
789
788
753
730
725
698
655
643
613
624
587
582
867
840
933
912
943
865
805
737
805
818
780
692
752
679
643
618
593
606
586
563
577
564
528
535
549
508
g(NH)0-2
1220
1220
1182
1182
1160
1145
1130
–
–
1108
1092
1092
–
1005
1005
974
974
965
933
933
–
904
883
856
856
788
788
774
772
748
748
718
718
683
684
655
643
601
630
592
555
–
546
539
505
Table 3 (Continued )
v0
n
x
Optical results
This work
Observed
ir
Raman
[12]
[12]
[13]
38
38
52
38
Bg
Au
467
39
39
29
40
29
53
39
27
54
83
28
40
29
41
55
56
42
84
43
Bg
Au
Ag
Bg
Bu
Ag
Au
Ag
Au
Bg
Bg
Au
Bu
Au
401
40
41
41
INS
Raman
466
427
420
310
275Ag
266
464
424
400
387
312
286
252
244
236
175
129
102
87
72
Assigned
463
424
–
397
384
313
289
280
260
247
236
180
174
129
101
87
71
DFT
H–F
484
447
514
449
425
397
312
300
291
255
249
231
170
166
114
88
69
36
428
404
296
294
277
240
250
222
169
163
105
88
71
35
Two mode numbers are given; in the first column, v0 , are the mode numbers of the
isatin monomer eigenvectors (see, Table 2) that most nearly correspond to the
indigo mode, as numbered in the second column, n. N.B. not all indigo modes have
corresponding eigenvectors in the isatin monomer.
used, after the assignment of indigo, to correlate isatin eigenvectors to those of indigo.
5.3. Indigo
The INS spectrum of indigo is detailed in Table 3. As could be
anticipated, the INS spectra of indigo and thioindigo are rather
similar (with the exception that bands involving the indigo NH are
absent from thioindigo). The same strong features occur in both
spectra, typically to within about 30 cm1 of each other. Below
about 600 cm1 the thioindigo features appear to lower frequency
than the corresponding indigo transitions and above 600 cm1
they appear to higher frequency. Above about 1000 cm1, the
influence of the sulfur substitution has lost its impact and the
corresponding transitions from the two systems appear in close
proximity. This comparison of the two spectra exposed extra
intensity in the indigo spectrum at 646 cm1, at 1257, 1352 and
1397 cm1. The first of these corresponds to g(NH)(0–1), which
would give rise to a harmonic overtone at about 1292 cm1
(=646 2) but observed here at 1257 cm1. The other features are
associated with the in-plane deformation, d(NH), clearly in line
with other work [9], where an ir band, about 1394 cm1, was
shown to disappear upon indigo’s incorporation into inorganic
matrices.
The stretching mode region of the indigo spectrum, about
3000 cm1, is quite sharp and strong but is unstructured and a
direct analysis, as was done for thioindigo, was not possible. We
estimated a value for (al)ext about 0.01 Å´2, similar to, but slightly
larger than the heavier thioindigo. The observed spectrum below
50 cm1 was taken as representative of the phonon-wing
spectrum.
Unfortunately, the unscaled ab-initio INS spectrum of indigo
does not compare to the observed spectrum nearly as straightforwardly as did the results for thioindigo and isatin. Attempts to
determine scaling factors directly from the comparison were
difficult and an indirect approach was adopted. Here, the model
compounds’ spectra were used as a guide. Since the INS spectra of
indigo of thioindigo are so very similar, we compared the scaled
ab-initio spectrum of thioindigo (from Fig. 3) to the INS of indigo,
see Fig. 6. This figure enables us to associate an observed band in
J. Tomkinson et al. / Vibrational Spectroscopy 50 (2009) 268–276
Fig. 6. A comparison of the observed inelastic neutron scattering spectrum of indigo
(black trace) with that calculated for the scaled ab-initio assignment scheme of
thioindigo (red trace), Table 1. (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of the article.)
the indigo spectrum with candidate eigenvalues in its model
compound, thioindigo, and, hence, to candidates in the unscaled
ab-initio spectra of indigo. (Candidate eigenvectors were also
checked to be consistent with those of isatin.) The final assignment
is made by comparing the symmetry of the calculated indigo
candidate eigenvector with the optical character of the indigo
transition; as was done, above, in the assignments of thioindigo
and isatin. In this manner, each indigo transition is associated with
a calculated eigenvector (and eigenvalue) and scaling parameters
can then be extracted.
This rather drawn out process was unambiguous up to
1000 cm1, where two calculated u-g pairs are candidates for
two indigo transitions observed at 970 and 1005 cm1.
Unfortunately, here, and only here, the optical studies [12,13]
have conflicting assignments of the character of the transition
about 1014 cm1 (our 1005 cm1 INS band). We have opted to
assign this to an in-plane mode. This is in agreement with the
earlier optical work [12], consistent with the thioindigo
assignments (above) and in best agreement with the distribu-
Fig. 7. A comparison of the observed inelastic neutron scattering spectrum of indigo
(black trace) with that calculated for scaled ab-initio assignment scheme of Table 3
(red trace). Note the over intense g(NH)(0–1) and (0–2) transitions in the calculated
spectrum, which are not reflected in the observed spectrum, see text for details. (For
interpretation of the references to color in this figure legend, the reader is referred
to the web version of the article.)
275
tion of INS intensities. The remaining assignments are given in
Table 3. (We disagree with the assigned optical character of only
two Raman features, observed at 275 and 1097 cm1, these were
assigned Ag [12] and Bg [12,13] respectively but we have them as
Bg and Ag.) Above about 1200 cm1 the observed INS features are
too indistinct to be reasonably assigned according to their
strengths and, here, our assignment scheme relies on the
published optical data [12,13]. The INS spectrum of indigo can
now be compared directly to its scaled ab-initio spectrum,
shown in Fig. 7. The good fit shown in the figure supports the
assignments of Table 3.
So far we have ignored the N–H vibrations and, as can be seen
from Fig. 6, one strong feature of the observed spectrum remains to
be assigned. Quite fortuitously, we believe, the calculated g(NH), ca
650 cm1, fall almost exactly on the strong INS band at 646 cm1.
The relatively low frequency calculated for g(NH) stems from a
hydrogen bonding environment of rather modest strength. The
intra-molecular contacts, the only contributions to our ab-initio
calculations, show a long N. . .O distance, 2.917 Å´ compared with
isatin, 2.876 Å´. This is compounded by a severely non-linear
geometry, the N–H. . .O bond is bent at 120.08 (c.f. 168.28 in isatin).
In the crystal other hydrogen bonding contacts are present [22] but
cannot be as important as the intra-molecular bond. Where the
intermolecular contacts are important, however, is in the way they
subtly change the atomic displacements involved in the g(NH)
mode. As calculated, the eigenvector for this mode is almost purely
a hydrogen motion (i.e. an oscillator mass about 1.0 amu), with
little movement of other atoms, and it is strong in the calculated
spectrum. Its observed strength, about half that of the calculated
value, shows that the intermolecular contacts have coupled
displacements of other atoms into g(NH). This is also mirrored
in the weakness of the observed (0–2) compared with that
calculated. From the reduced intensity of the fundamental and the
ratio of the fundamental to overtone intensities [15], we estimate
the effective oscillator mass to be about 3 amu. The intensities of
the in-plane deformation modes, d(NH) about 1375 cm1, are
reasonably calculated since they are always found mixed with
heavy atom displacements.
6. Discussion
The mode numbering scheme of indigo, n, is given in Table 3.
Also given there are those mode numbers, v0 , of the isatin monomer
whose eigenvectors most nearly correspond to the eigenvectors of
indigo. (Several different modes of isatin correlate, to a greater or
lesser degree, with any given indigo mode but this information has
been severely condensed in Table 3, where only the leading terms
are retained, the v0 .) By referring back to Table 2, where the
eigenvectors of isatin are described in detail, the approximate form
of the indigo eigenvectors can be determined. Those indigo modes
which imply little, or no, deformation of the isatin unit correspond
to low energy vibrations involving the displacement of isatin
monomers against one another. Otherwise, as could be expected,
each u–g pair of indigo modes represents the in- and out-of-phase
vibrations of a common isatin ‘root’ mode. The indigo mode n44
corresponds to the in-phase vibration of the isatin mode v030 , with
indigo n30 corresponding to the out-of-phase vibrations of the
same isatin mode (the isatin v030 is defined in Table 2). Thus n30 and
n44 involve mostly ‘ring-breathing’ displacements of the benzenoid
motifs in indigo (mode ‘2’ is the ring-breathing mode of benzene
[30]).
In searching for candidate vibrations of indigo that can be used
in INS spectroscopy to identify the chemical species present in
inorganic matrices (like Maya Blue), two criteria present themselves: the modes must be easily observable and sensitive to
different environments. Certainly the INS region above 1200 cm1
276
J. Tomkinson et al. / Vibrational Spectroscopy 50 (2009) 268–276
is not a good frequency range for strong transitions, candidates in
this region may be discounted. Transitions at low frequencies can
also be rejected, their eigenvalues are likely to be too much
influenced by external forces. Such that, in a non-homogenous
system like Maya Blue, very broad features are the most likely
result. However, ease of observation is not limited to the domain of
intense transitions. As can be seen in all the figures there are
several strong features about 1100 cm1 but they have strong
bands nearby and frequency changes in any one of these features
may be masked by its becoming accidentally degenerate with its
neighbours. Strong features separated from other strong features
would represent better candidates.
The assigned spectrum of indigo shows two good candidates in
the mid-frequency range, the g(NH) mode and the band at ca.
750 cm1. (The 750 cm1 band is the degenerate composite of the
u and g components of a g(C–H) mode of the benzene motif. This is
itself a 1:1 mixture of the benzene modes ‘4’ and ‘11’ [30], see Table
2). The g(C–H) and g(NH) bands are both strong and yet isolated
from other strong features in the region. However, the g(C–H)
mode involves atoms somewhat distanced from the part of the
molecule that interacts with its environment. Even if the band
were observed it would give little further information than we
have at present. This is definitely not the case with the g(NH)
mode, so much so, that if dehydro-indigo is the chemical species in
Maya Blue [9] then g(NH) will be completely absent and the
presence of dehydro-indigo could only be implied rather than
observed. However, the two bands, g(C–H) and g(NH), could be
used in conjunction to demonstrate and quantify the presence of
an indigoid, through g(C–H), and how it interacts with its
environment, through g(NH).
Any variations in the g(NH) frequency and, or, intensity can
then be exploited to study the interaction of the incorporated
indigo with its environment. In all probability, the intra-molecular
hydrogen bond of indigo will retain some of its character in the
matrix and variations in the intensity of g(NH) will indicate how
the matrix environment differs from that of its crystal. In the event
that the intra-molecular hydrogen bond is ruptured indigo may
bond directly, or through water, to the matrix. The modes of
intermolecular hydrogen bonds are known to be rather sensitive to
the local bonding environment, increasing in frequency as the
hydrogen bond length falls [31]. Under these circumstances it is
even possible that the g(NH) mode intensity could increase, as the
involvement of heavy atoms in the eigenvector is reduced. In
contrast, the g(C–H) band will remain almost unchanged by
inclusion into the matrix, it will act as an internal standard. Its
strength will be a clear indicator of the experimental statistics
allowing, say, a judgement to be made as to the statistical
relevance of the non-observation of a g(NH) band.
These results clearly show that INS spectroscopy will be
singularly advantaged in the study of Maya Blue since it, alone of
all the molecular vibrational techniques, can readily observe the
g(C–H) and g(NH) modes.
7. Conclusions
Using inelastic neutron scattering techniques and ab-initio
methods, we have extended the measured spectra of isatin,
thioindigo and indigo down to the lowest frequencies. We have
revised previous calculations and provided sound spectral assignments, below 1200 cm1. We have described the vibrational
eigenvectors of indigo in terms of those of its model compound
isatin and identified a g(C–H) and the g(NH) modes, used in
conjunction, as the best candidates for vibrational features that
could be used to determine the nature of the organic residue in Maya
Blue and how it might interact with the inorganic matrix.
Acknowledgements
We should like to thank the CNR for partially funding the
present study and the STFC for access to the neutron beam facilities
of ISIS, The Rutherford Appleton Laboratory, UK. We would also
like to thank Prof. P. Naumov, Graduate School of Engineering,
Osaka University, Japan, who kindly sent his unpublished, lowtemperature, infrared data of isatin.
References
[1] E.P. Akcakoca-Kumbasar, H. Aydin, Z. Ondogan, M. Ozcelik, E.N. Ondogan, Fibres
Textiles 14 (2006) 86–90.
[2] H. Schweppe, Artists’ pigments, in: E. West Fitzhugh (Ed.), A Handbook of Their
History and Characteristics, Oxford University Press, New York, 1997, pp. 81–107.
[3] R.J. Gettens, American Antiquity 27 (1962) 557–564.
[4] A.O. Shepard, H.B. Gottlieb, Notes from a Ceramic Laboratory N1, Carnegie
Institution of Washington, Washington, DC, 1962.
[5] H. Van Olphen, Science 154 (1966) 645–646.
[6] L. Serrano-Andres, B.O. Roos, Chem. Eur. J. 3 (1977) 717–725.
[7] D. Jaquemin, J. Preat, V. Wathelet, E.A. Perpete, J. Chem. Phys. 124 (2006) 074104.
[8] M. Leona, F. Casadio, M. Bacci, M. Picollo, J. Am. Inst. Conserv. 43 (2004) 39–54.
[9] A. Doménech, M.T. Doménech-Carbò, M.L. Vàzquez de Agredos Pascual, J. Phys.
Chem. C111 (2007) 4585–4595.
[10] F.S. Mancia, L. Reza, L.A. Polette, B. Torres, R.R. Chianelli, J. Raman Spectrosc., in
press. Published on line in Wiley Interscience doi:10.1002/jrs.1751.
[11] P. Vandenabeele, S. Bode, A. Alonso, L. Moens, Spectrochim. Acta A61 (2005)
2349–2356.
[12] E. Tatsch, B. Schrader, J. Raman Spectrosc. 26 (1995) 467–473.
[13] M. Sanchez del Rio, M. Picquart, E. Haro-Poniatowski, E. van Eslande, V.C. Uc, J.
Raman Spectrosc. 37 (2006) 1046–1053.
[14] R. Giustetto, F.X. Xamena, G. Ricchiardi, S. Bordiga, A. Damin, R. Gobetto, M.R.
Chierotti, J. Phys. Chem. B109 (2005) 19360–19368.
[15] P.C.H. Mitchell, S.F. Parker, A.J. Ramirez-Cuesta, J. Tomkinson, Vibrational Spectroscopy with Neutrons, World Scientific Press, Singapore, 2005.
[16] N.A. Frolova, V.Kh. Kravtsov, V.N. Biyushkin, Y.M. Chumakov, O.N. Bel’kova, T.I.
Malinovski, J. Struct. Chem. 29 (1988) 491–493.
[17] G.J. Palenik, A.E. Koziol, A.R. Katritzky, W.-Q. Fan, J. Chem. Soc., Chem. Commun.
(1990) 715–716.
[18] T. Shimanouchi, Tables of Molecular Vibrational frequencies Consolidated
Volume 1. NSRDS-NBS 39, U.S. Dept. Commerce, Washington, 1972.
[19] H. von Eller, Bull. Soc. Chim. France 5 (1955) 1438–1440.
[20] W. Haasewessel, M. Ohmasa, P. Susse, Naturwissen 64 (1977) 435–1435.
[21] H. von Eller, Bull. Soc. Chim. France 5eme (1955) 1433–1438.
[22] P. Susse, M. Steins, V. Kupcik, Z. Kristall. 184 (1988) 269–273.
[23] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman,
V.G. Zakrzewski, J.A. Montgomery, Jr., R.E. Stratmann, J.C. Burant, S. Dapprich, J.M.
Millam, A.D. Daniels, K.N. Kudin, M.C. Strain, O. Farkas, J. Tomasi, V. Barone, M.
Cossi, R. Cammi, B. Mennucci, C. Pomelli, C. Adamo, S. Clifford, J. Ochterski, G.A.
Petersson, P.Y. Ayala, Q. Cui, K. Morokuma, D.K. Malick, A.D. Rabuck, K. Raghavachari, J.B. Foresman, J. Cioslowski, J.V. Ortiz, B.B. Stefanov, G. Liu, A. Liashenko, P.
Piskorz, I. Komaromi, R. Gomperts, R.L. Martin, D.J. Fox, T. Keith, M.A. Al-Laham,
C.Y. Peng, A. Nanayakkara, C. Gonzalez, M. Challacombe, P.M.W. Gill, B. Johnson,
W. Chen, M.W. Wong, J.L. Andres, C. Gonzalez, M. Head-Gordon, E.S. Replogle, J.A.
Pople, GAUSSIAN98, Revision A.3, Gaussian, Inc., Pittsburgh, 1998.
[24] A.K. Grafton, R.A. Wheeler, Comp. Phys. Commun. 113 (1998) 78–84.
[25] D. Colognesi, M. Celli, F. Cilloco, R.J. Newport, S.F. Parker, V. Rossi-Albertini, F.
Sacchetti, J. Tomkinson, M. Zoppi, Appl. Phys. A 74 (Suppl.) (2002) S64–S66.
http://www.isis.rl.ac.uk/molecularSpectroscopy/tosca/index.htm.
[26] A.J. Ramirez-Cuesta, Comput. Phys. Commun. 157 (2004) 226–238.
[27] A. Bigotto, P. Galasso, Spectrochim. Acta A35 (1979) 725–732.
[28] K. Kurosaki, Nippon Kagaku Zasshi 82 (1961) 1555–1557.
[29] P. Naumov, F. Anastasova, Spectrochim. Acta A57 (2001) 469–481.
[30] G. Herzberg, Molecular Spectra and Structure: II. Infrared and Raman Spectra of
Polyatomic Molecules, Van Nostrand, New York, 1945, pp 118.
[31] J. Howard, J. Tomkinson, J. Eckert, J.A. Goldstone, J. Chem. Phys. 78 (1983)
3150–3155.