Charge-Sensitive Vibrations in p-Chloranil: The Strange Case of the

We have combined DFT calculations with single-crystal polarized infrared spectra to reinvestigate the assignment of the C C antisymmetric stretching m...
0 downloads 0 Views 107KB Size
12844

J. Phys. Chem. B 2007, 111, 12844-12848

Charge-Sensitive Vibrations in p-Chloranil: The Strange Case of the CdC Antisymmetric Stretching Paolo Ranzieri, Matteo Masino, and Alberto Girlando* Dip. Chimica GIAF and INSTM-UdR Parma, UniVersita` di Parma, Parco Area delle Scienze, 43100-I Parma, Italy ReceiVed: July 14, 2007; In Final Form: August 23, 2007

We have combined DFT calculations with single-crystal polarized infrared spectra to reinvestigate the assignment of the CdC antisymmetric stretching mode b2uν18 of p-chloranil (CA). The frequency of this mode indeed seems to display a nonlinear dependence on the average charge on the CA molecule (ρ), at variance with the behavior of the antisymmetric CdO stretching frequency. The DFT calculations show that the origin of the problem is a drastic, 2 orders of magnitude decrease of the infrared intensity of the CdC antisymmetric stretching upon electron addition. Therefore, no infrared band can be easily associated to this mode in charge-transfer (CT) solids with F J 0.5. On the other hand, a linear relationship between ρ and the b2uν18 frequency is found in quasi-neutral CT complexes of CA.

1. Introduction The easy electron exchange between π-donor and -acceptor molecules is the origin of the rich phase diagram and of the diverse physical properties of low-dimensional charge-transfer (CT) solids.1 The degree of ionicity, or average charge per molecules (ρ), is one of the fundamental parameters characterizing the physical properties of these systems. A determination of ionicity based on its definition is not an easy task.2 However, one can use indirect methods, for instance, the change of bond length3 or of vibrational frequencies4 upon variation of the molecular charge. Both of the above methods require a calibration, namely, (i) structural or vibrational studies of a set of compounds whose ionicity is known from stoichiometry and (ii) assumption about the functional dependence of ρ from bond lengths or vibrational frequencies. Vibrational spectroscopy has been the most used method to determine ρ for its easy implementation, in particular, when the ionicity has to be followed as a function of temperature or pressure.5 In addition, there is theoretical background to support the idea that ρ has a practically linear dependence on vibrational frequencies.4,6 A number of studies on the most important π-electron-donor and -acceptor molecules has shown that the expected linear relationship between frequencies and ionicity is indeed found in most cases.4,7-11 Most of the CT solids formed by tetrachloro-p-benzoquinone (chloranil, hereafter CA) with electron donors like tetrathiafulvalene (TTF) or N,N′-tetramethyl-p-phenylendiamine (TMPD) crystallize with a mixed stack motif, with ρ considerably different from the extreme values of 0 and 1.5 Temperature or pressure may change ρ, up to induce a true phase transition, the so-called neutral-ionic phase transition, implying an abrupt change of ionicity.5 Complete vibrational assignments of CA and of its radical anion have shown that the vibrations most sensitive to average molecular charge are the symmetric and antisymmetric carbonyl stretching, agν1 and b1uν10.12,13 These modes indeed lower by 175 and 160 cm-1, respectively, when ρ changes from 0 to 1. Since ag modes are perturbed by electron-molecular vibration coupling in mixed stack crystals,6

the ionicity of CT solids involving CA has been always estimated through the b1uν10 mode. When estimating the ionicity through just one vibrational frequency, one should make sure about its linear dependence. However, no direct experimental proof, of the type devised for the TTF or tetracyanoquinodimethane (TCNQ) modes,14 is available for CA of the b1uν10. This lack of experimental proof has been among the motivations of a Car-Parrinello molecular dynamics investigation of CA in various oxidation states.15 This study confirms the linear dependence of the b1uν10 mode from ρ. On the other hand, in contradiction with the above calculations, the frequency of another CA fundamental mode, the b2uν18 CdC antisymmetric stretching, apparently shows a highly nonlinear dependence on ρ. The strange behavior of this mode may be the cause of misinterpretation of the spectra,16 and indeed, our recent work on TTF-CA17,18 and on dimethyltetrathiafulvalene-chloranil (DMTTF-CA)19 suggests that its assignment has to be carefully reconsidered. In this paper, we present a reanalysis of the vibrational assignment of fully ionic CA, based on polarized infrared spectra of the K salt, assisted by a DFT calculation of neutral and ionic CA. We revise the assignment of the CdC antisymmetric stretching, explain the reason for the previous misassignment,13 and show how the frequency of this mode can be reliably used to estimate the ionicity of CA-based CT crystals. 2. Experimental and Computational Methods CA was a commercial product (Fluka), used without further purification. K+CA- crystals were grown from acetone according to the method of Torrey and Hunter.20 The infrared (IR) spectra were obtained with a Bruker IFS-66 FTIR spectrometer, equipped with an A590 microscope. The control Raman spectra, not reported in this paper for the sake of brevity, were recorded with a Renishaw System 1000 microspectrometer with 647.1 nm excitation from a Kr ion laser (Lexel 3500). The spectral resolution of the IR and Raman spectra was 2 cm-1. Quantum chemical calculation were performed with the GAMESS package,21 using the DFT (B3LYP) method and the

10.1021/jp075510v CCC: $37.00 © 2007 American Chemical Society Published on Web 10/17/2007

Charge-Sensitive Vibrations in p-Chloranil

J. Phys. Chem. B, Vol. 111, No. 44, 2007 12845

TABLE 1: In-Plane Fundamental Modes of Neutral CA calculated νj/cm-1

mode agν1 ν2 ν3 ν4 ν5 ν6 b3gν23 ν24 ν25 ν26 ν27 b1uν10 ν11 ν12 ν13 ν14 b2uν18 ν19 ν20 ν21 ν22

experimental νj/cm-1

rel. int.b

1757 1625 991 484 322 200 1240 851 727 338 264 1760 1109 906 464 205 1590 1227 736 378 214

100.0 91.9 2.0 13.6 3.9 1.5 34.7 5.7 0.1 1.1 0.2 100.0 99.5 4.6 1.0 ∼0.0 66.6 27.1 51.8 1.3 ∼0.0

a

rel. int.b

1693 1609 1007 496 330 200 1247 852

100.0 55.5 2.2 42.0 12.8 11.8 ∼15.0 2.7

307 267 1685 1110 908 473 186? 1572 1260 755 376 206

0.8 1.4 100.0 65.0 4.5 0.4 0.4 38.7 7.1 26.3 0.8 0.4

a

From ref 12. b Raman and IR intensities are relative to a value of 100 assumed for the highest intensity band in both types of spectra. Raman intensities are from the present work; IR intensities are from ref 12.

TABLE 2: Calculated and Observed Totally Symmetric (ag) Fundamentals of the CA Anion calculated mode

νj/cm-1

ν1 ν2 ν3 ν4 ν5 ν6

1541 1607 1000 493 326 202

rel.

int.b

18.6 100.0 1.7 4.4 3.2 0.3

experimentala νj/cm-1 1518 1594 1028 517 338

rel.

int.b

8.9 100.0 4.4 7.7 5.2

ionization freq. shift ∆νjcalc/cm-1 ∆νjexpt/cm-1 -217 -16 +9 +9 +4 +2

TABLE 3: Proportionality Factors for the IR Intensities of K+CA-

-175 -15 +21 +21 +8

a From ref 13. b Intensities are relative to a value of 100 assumed for the ν2 mode.

6-31G(d) basis set. This combination is know to satisfactorily reproduce the molecular vibrations of relatively large organic molecules, provided tight geometry optimization and fine grids are used in the DFT analysis.22 D2h symmetry was imposed for both neutral and ionic CA. The results were analyzed and visualized with the help of Jmol23 and Molden24 packages. 3. Neutral Chloranil, CA The assignment of neutral CA vibrational modes, based on single-crystal polarized Raman and IR spectra, is well established and essentially complete.12 We then use this assignment to assess the reliability of DFT calculations. Table 1 compares calculated and experimental frequencies for the ag and b3g (Raman active) and the b1u and b2u (IR active) in-plane modes.12 The agreement is very good, the average difference between experimental and calculated frequencies being 17 cm-1. The only exception is the CdO symmetric and antisymmetric stretching mode, with deviations of 64 and 75 cm-1, respectively, with an error of less than 5%. Notice that the reported calculated frequencies are not scaled. Scaling is often used to quickly correct for anharmonic effects22 but is clearly not needed in this case. The error in the calculated CdO stretching frequencies is more likely attributable to inadequacy of the basis set.22

a b c

b1u

b2u

b3u

0.13 0.87 0.00

0.65 0.09 0.26

0.22 0.04 0.74

In Table 1, we also compare calculated and experimental relative intensities of Raman and IR bands. Again, the agreement is quite satisfactory, the correct trend being always properly reproduced. We mention that the same agreement between calculated and observed frequencies and intensities is found also for the out-of-plane modes, not reported here for the sake of brevity. Therefore, the present results confirm that the DFTB3LYP method with the 6-31G(d) basis set offers a reliable description of the normal modes of relatively large organic molecules22 and gives us confidence in the calculations for the radical anion. 4. Chloranil Potassium Salt, K+CAThe assignment of CA- vibrations has been based on the spectra of powders and solutions of the potassium salt.13,25,26 For the Raman, we shall use the available assignment, which is limited to the totally symmetric ag modes due to resonance Raman effects.13 Table 2 compares calculated and experimental ag frequencies of CA-. As already observed,13,15 upon the addition of one electron, the description of the two highest frequency modes mixes. In the neutral molecule, the highest frequency mode is the symmetric CdO stretching, and the second one is the CdC stretching, whereas in the anion, there is strong mixing due to the larger ionization shift of the CdO stretch. We have kept the symbols ν1 and ν2 to label the CdO and CdC stretches, respectively, in both CA and CA-. The agreement between calculated and experimental frequencies in Table 2 is again quite satisfactory. The average deviation is slightly larger (20 cm-1) than that in the case of neutral CA, but the frequency of the CdO stretching has an error of 33 versus 64 cm-1 in neutral CA. As a consequence, the calculated ionization frequency shift of the CdO ν1 mode is somewhat higher than that of the experiment (columns 6 and 7 in Table 2). On the other hand, all of the other ionization frequency shifts have the correct (positive or negative) trend, and the shift of the CdC stretching is quantitatively reproduced. The relative intensities (columns 3 and 5) are also well reproduced in the calculation To clear up the assignments of the ungerade, IR active modes of CA-, we shall use polarized spectra of K+CA- microcrystals. The appropriate spectral predictions are in order. K+CA- is reported to have more than one polymorph, but the R phase is most often obtained by the method used in our preparation.20,27 The R-K+CA- crystal is orthorhombic, space group P212121 (D2), with four molecules per unit cell, residing on the general position (C1). The factor group analysis then predicts that each molecular mode is split into four components in the crystal, of symmetry A, B1, B2, and B3. All of the four components are Raman active, and the B1, B2, and B3 components are also IR active, with polarization along the c, b, and a axes, respectively. According to the above factor group analysis, the IR spectra of K+CA- should be rather complicated, as all of the molecular modes, including the gerade ones, should become IR active and split into two components in the absorption spectra of a principal crystal face. However, the previously published data13 show that the factor group splittings are not visible in the powders spectra

12846 J. Phys. Chem. B, Vol. 111, No. 44, 2007

Ranzieri et al.

Figure 2. Eigenvectors of the b2uν18 mode of CA and CA-. The corresponding HOMOs are also shown. Figure 1. Polarized IR spectra of K+CA-. Continuous and dashed lines indicate spectra polarized parallel and perpendicular to the c crystal axis, respectively. The assignment of the b3uν28 mode is noted.

TABLE 4: Calculated and Observed Ungerade In-Plane Fundamentals of the CA Anion calculated mode

νj/cm-1

b1uν10 ν11 ν12 ν13 ν14 b2uν18 ν19 ν20 ν21 ν22

1562 1136 901 443 208 1464 1133 701 358 215

experimental

ionization freq. shift

rel. int.

νj/cm-1

∆νjcalc/cm-1

∆νjexpt/cm-1

100.0 44.9 36.6 ∼0 1.0 0.6 28.8 42.1 0.3 ∼0

1525 1149 918 468 200 ?? 1174 722 374 212

-198 +27 -5 -21 +3 -126 -94 -35 -20 +1

-160 +39 +18 -5 - 86 -33 -2 +6

and are probably small, like in the case of neutral CA.12 In addition, the gerade molecular modes are not detected in the crystal IR spectra. Therefore, the intermolecular interactions are small, and under these circumstances, we can use the oriented gas model to predict the relative intensities of the b1u, b2u, and b3u molecular modes, with the oscillating dipole moment directed along the z, y, and x molecular axes, respectively. The relative intensities are then proportional to the squared direction cosines between the molecular and crystal axes.12 The K+CAsquared direction cosines are given in Table 3. From the Table 3 predictions, one expects that the b1u modes should be completely polarized along the b axis and the b3u (out-of-plane) modes along the c axis. The b2u modes should have a less pronounced polarization. Figure 1 reports the IRpolarized absorption spectra of K+CA- from 650 to 1650 cm-1. The crystal plane certainly contains the c axis, by far the shortest crystal axis,27 and is most likely the bc plane since the b3u mode at 694 cm-1 is completely polarized. The same polarization is displayed by the vibronically activated bands around 1010 and 1490 cm-1, corresponding to the ag modes of the isolated molecule.26 The b1u modes, which display an opposite polarization, are easily identified. Further assistance in the assignment comes from DFT calculations of CA- vibrations reported in the second and third columns of Table 4. The previous assignment13 of the b1u modes is entirely confirmed. On the other hand, the calculation helps to assign the b2uν19 mode to the relatively weak, but correctly polarized, band at 1174 cm-1. Curiously, this band was properly assigned in the first K+CA- data collection,28 but subsequently, the assignment was rejected on the basis of empirical normal coordinate analysis.13

We finally discuss the assignment of the CA- CdC antisymmetric stretch, b2uν18, which is the central issue of this work. This mode was previously assigned to the band at 1545 cm-1, a shoulder of the band at 1525 cm-1, corresponding to the Cd O antisymmetric stretching.13 This assignment was motivated by the lack of plausible alternatives at lower frequency and by considering that the corresponding symmetric CdC stretch, agν2, has a rather small ionization frequency shift (ref 13 and Table 2). However, the present experiment and calculation are against the assignment of the CdC antisymmetric stretch to the 1545 cm-1 band. Figure 1 shows that both the 1525 cm-1 band and its shoulder at 1545 cm-1 are strongly polarized along the b axis and, accordingly, belong to the b1u species (in Figure 1, the strong CdO stretching band saturates the spectrum; therefore, the doublet structure is not visible). Therefore, the 1545 cm-1 band is likely a combination band, possibly enhanced by Fermi resonance, as it happens, for instance, in p-benzoquinone.29 In addition, both the present and previous15 calculations show that the ionization frequency shift of the b2uν18 mode is on the order of 100 cm-1. Notice that the order of magnitude of all of the other sizable ionization frequency shifts is properly accounted for (Tables 2 and 4). Then, the band corresponding to the CdC antisymmetric modes should be located around 1460 cm-1, but only very weak bands, barely visible in Figure 1, are present around this frequency. The present DFT analysis explains why no reasonable candidate for the antisymmetric CdC stretching mode can be found in the spectra. Upon ionization, the IR intensity of the b2uν18 mode is completely washed out, from 66 to 0.6% in relative terms (Tables 1 and 4) or from 6.13957 to 0.05566 D2/ amu Å2 in absolute terms. We believe it is a useless endeavor to try to find some candidate; at this intensity level, any combination or even sample impurity bands may be confused with the fundamental mode. For this reason, in Table 4, we leave the b2uν18 unassigned. 5. Discussion and Conclusions It is not easy to understand the origin of the drastic intensity lowering of the CdC antisymmetric stretching upon electron addition, as the oscillating dipole moment depends on the overall electronic distribution. Figure 2 compares the eigenvectors of the b2uν18 fundamental in neutral and ionized CA, together with the corresponding HOMO orbitals. The eigenvectors change a little: In the anion, the vibration involves also a small ring deformation. However, the motion can be essentially described as an antisymmetric CdC stretching in both the neutral molecule and in the anion. Therefore, the intensity washing out is more likely attributable to a change in the electron distribution.

Charge-Sensitive Vibrations in p-Chloranil

Figure 3. Ionicity dependence of the frequency of CdO (squares) and CdC (circles) antisymmetric stretching in CA. Empty symbols are from refs 17 and 18 (TTF-CA), and filled symbols are from ref 19 (DMTTF-CA). Triangles label the frequencies of fully neutral and fully ionic molecules.

Incidentally, we note that the effect does not depend on the basis set since it is obtained also by using the 6-31G and the minimal STO-3G basis sets. It is well-known, and schematically illustrated by the HOMOs in Figure 2, that the additional electron goes on an antibonding orbital for the CdO and on a bonding orbital for the ring C-C bonds. Both the CO and CC double bonds are weakened, the electronic distribution shifting from quinoid to benzenoid. As a consequence, the CdO and CdC stretching frequencies are lowered. On the other hand, the IR intensity of the antisymmetric CdO stretching changes very little, whereas that of the CdC is drastically weakened. An intuitive justification of the above difference can be advanced as follows. The change in the electron density distribution upon electron addition does not affect the intensity of the b1u CdO antisymmetric stretching because the distribution change is along the z axis, the direction of the oscillating dipole moment. On the other hand, the oscillating dipole moment generated by the b2u antisymmetric CdC stretching is along the y axis, perpendicular to the bonds, where the nodes of the CA- HOMO occur (Figure 2). The depletion of the electron density precisely all along the direction of the oscillating dipole moment then gives reason for the IR intensity weakening, although such a large effect would not have been predicted on the basis of these simple considerations. Apart from the anomalous intensity behavior, a more important question to address is whether or not the frequency of the chloranil b2uν18 fundamental gives reliable information about the average charge residing on the molecule. To answer this question, we have plotted in Figure 3 the frequencies of the antisymmetric CdO and CdC stretchings as a function of ρ. The upper dashed line in Figure 3 gives the usual empirical linear dependence of the CdO stretching frequency from ρ, νjCO ) 1685 - 160ρ. Superimposed on it, we show the ionicities estimated from just this frequency in TTF-CA and DMTTFCA CT solids, as a function of temperature and pressure for the former17,18 and as a function of temperature for the latter.19 We then used the ionicities estimated from the CdO stretching frequency to obtain the corresponding plot for the frequencies of the antisymmetric CdC stretching (circles in Figure 3). The reported frequencies are limited to the neutral (ρ < 0.5) side of the diagram since we cannot identify, with certainty, bands corresponding to this mode due to the above-discussed drastic intensity lowering. The continuous line in Figure 3 is a linear regression fit of the experimental points, including the frequency of the neutral molecule (triangle). The fitting is very good, the fitting line being expressed by νjCdC ) 1566 - 116ρ. Notice also that by extrapolating the line to ρ ) 1.0, one obtains 1450

J. Phys. Chem. B, Vol. 111, No. 44, 2007 12847 cm-1, closely matching the DFT calculated frequency (1464 cm-1, Table 4). In conclusion, we have shown that the apparent nonlinear dependence of the CA CdC antisymmetric stretching mode on ρ was due to a misassignment in the IR spectra of K+CA-.13 Actually, we cannot identify the b2uν18 band even in spectra where CA is only partially ionic (ρ > 0.5) due to the washing out (2 orders of magnitude) of the IR intensity of this mode upon electron addition. On the other hand, we have also shown that when ρ < 0.5, the frequency of the CA b2uν18 is an ionicity indicator as good as the frequency of the antisymmetric CdO stretching, b1uν10.13,15 The fact that the frequency of the b2uν18 mode cannot be identified for ionic or quasi-ionic CA CT solids is, of course, a drawback. However, the disappearance of the CdC antisymmetric stretching band when the ionicity of a CT solid increases upon temperature or pressure change can be used as an indication of the crossing of the neutral-ionic borderline.18,19 The same considerations apply also to other haloquinones, like bromanil13 or fluoranil. Preliminary investigations suggest that the ring CdC stretching IR intensity weakening occurs also in TCNQ/TCNQ-; although in this case, the mode mixing makes the analysis more difficult. Acknowledgment. Work supported by Parma University, FIL projects. P.R. thanks the Italian Interuniversity Consortium on Materials Science and Technology (INSTM) for the contribution to his Ph.D. fellowship. References and Notes (1) Soos, Z. G.; Klein, D. J. In Treatise on Solid-State Chemistry; Hannay, N. B., Ed.; Plenum: New York, 1976; Vol. III, p 689. (2) Garcı´a, P.; Dahaoui, S.; Katan, C.; Souhassou, M.; Lecomte, C. Faraday Discuss. 2007, 135, 217. (3) Umland, T. C.; Allie, S.; Kuhlmann, T.; Coppens, P. J. Phys. Chem. 1988, 92, 6456. (4) Pecile, C.; Painelli, A.; Girlando, A. Mol. Cryst. Liq. Cryst. 1989, 171, 69, and references therein. (5) Girlando, A.; Painelli, A.; Bewick, S. A.; Soos, Z. G. Synth. Metals 2004, 141, 129, and references therein. (6) Painelli, A.; Girlando, A. J. Chem. Phys. 1986, 84, 5655. (7) Lunardi, G.; Pecile, C. J. Chem. Phys. 1991, 95, 6911. (8) Wang, H. H.; Ferraro, J. R.; Williams, J. M.; Geiser, U.; Schlueter, J. A. J. Chem. Soc., Chem. Commun. 1994, 1893. (9) Katan, C. J. Phys. Chem. A 1999, 103, 1407. (10) Drozdova, O.; Yamochi, H.; Yakushi, K.; Uruichi, M.; Horiuchi, S.; Saito, G. J. Am. Chem. Soc. 2000, 122, 4436. (11) Yamamoto, T.; Uruichi, M.; Yamamoto, K.; Yakushi, K.; Kawamoto, A.; Taniguchi, H. J. Phys. Chem. B 2005, 109, 15226. (12) Girlando, A.; Pecile, C. J. Chem. Soc., Faraday Trans. 2 1973, 69, 1291. (13) Girlando, A.; Zanon, I.; Bozio, R.; Pecile, C. J. Chem. Phys. 1978, 68, 22. (14) Bozio, R.; Pecile, C. In The Physics and Chemistry of Low Dimensional Solids; Alca´cer, L., Ed.; Reidel: Dordrecht, The Netherlands, 1980; p 165. (15) Katan, C.; Blochl, P. E.; Margl, P.; Koenig, C. Phys. ReV. B 1996, 53, 12112. (16) Matsuzaki, H.; Takamatsu, H.; Kishida, H.; Okamoto, H. J. Phys. Soc. Jpn. 2005, 74, 2925. (17) Masino, M.; Girlando, A.; Brillante, A.; Della Valle, R. G.; Venuti, E. Mater. Sci. 2004, 22, 333. (18) Masino, M.; Girlando, A.; Brillante, A. Phys. ReV. B 2007, 76, 064114. (19) Ranzieri, P.; Masino, M.; Girlando, A.; Leme´e-Cailleau, M. H. LANL.arXiV.org 2007, arXiv:cond-mat/0703716. (20) Torrey, H. A.; Hunter, W. H. J. Am. Chem. Soc. 1912, 34, 702. (21) (a) Schmidt, M. W.; Baldridge, K. K.; Boatz, J. A.; Elbert, S. T.; Gordon, M. S.; Jensen, J. H.; Koseki, S.; Matsunaga, N.; Nguyen, K. A.; Su, S.; Windus, T. L.; Dupuis, M.; Montgomery, J. A. J. Comput. Chem. 1993, 14, 1347. (b) Gordon, M. S.; Schmidt, M. W. In Theory and Applications of Computational Chemistry: The First Forty Years; Dykstra, C. E., Frenking, G., Kim, K. S., Scuseria, G. E., Eds.; Elsevier: Amsterdam, The Netherlands, 2005; p 1167.

12848 J. Phys. Chem. B, Vol. 111, No. 44, 2007 (22) Barone, V.; Festa, G.; Grandi, A.; Rega, N.; Sanna, N. Chem. Phys. Lett. 2004, 388, 279. (23) Jmol: An Open-Source Java Viewer for Chemical Structures in 3D. http://www.jmol.org/. (24) Schaftenaar, G.; Noordik, J. H. J. Comput.-Aided Mol. Des. 2000, 14, 123. (25) Girlando, A.; Bozio, R.; Pecile, C. J. Chem. Soc., Chem. Commun. 1974, 87.

Ranzieri et al. (26) Bozio, R.; Girlando, A.; Pecile, C. Chem. Phys. 1977, 21, 257. (27) Konno, M.; Kobayashi, A.; Marumo, F.; Saito, Y. Bull. Chem. Soc. Jpn. 1973, 46, 1987. (28) Girlando, A.; Morelli, L.; Pecile, C. Chem. Phys. Lett. 1973, 22, 553. (29) Becker, E. D.; Ziffer, H.; Charney, E. Spectrochim. Acta, Part A 1961, 19, 1871.