Vibrational Analysis of trans-Azobenzene - The Journal of Physical

Florian O. Koller, Rossana Reho, Tobias E. Schrader, Luis Moroder, Josef Wachtveitl, ..... Jörg Henzl , Michael Mehlhorn , Heiko Gawronski , Karl-Hei...
2 downloads 0 Views 2MB Size
J. Phys. Chem. 1995,99,17825-17831

17825

Vibrational Analysis of trans-Azobenzene D. R. Armstrong, J. Clarkson; and W. E. Smith* Department of Pure and Applied Chemistry, University of Strathclyde, 125 Cathedral Street, Glasgow GI XL,Scotland, U.K. Received: April 12, 1995;In Final Form: September 11, 1999

Molecular orbital calculations were performed to determine the normal modes and vibrational energies of azobenzene. A semiempirical calculation using the PM3 Hamiltonian and an ab initio calculation carried out at the SCF level using the 6-31G basis set gave unsatisfactory predictions especially for vibrations dominated by azo atom displacements. High-level electron correlation ab inifio calculations carried out at the MP2 level improved the fit with experiment but the choice of basis set was found to be critical. When the basis set for the nitrogens of the azo group was changed to the 6-31+G(d) basis set, the calculation gave a satisfactory fit. Normal-mode diagrams and energies are presented, and assignments to experimentally observed vibrational energies of azobenzene are made. The main azo stretch, Y I O ,observed at 1440 cm-I, is theoretically predicted at 1450 cm-I. The calculation correctly predicts an increase in frequency in the azo stretch mode upon deuteration of the phenyl rings. Coupling of several phenyl modes with azo vibrations are revealed by the calculation, in agreement with previous assignments of the vibrational spectra of azobenzene and azobenzene derivatives. The calculation indicates why certain in-plane stretching frequencies give rise to relatively intense Raman and resonance Raman scattering. In Raman scattering, the modes giving rise to the strongest scattering involve displacements along the N-N and C-N bonds. The same modes give intense resonance Raman scattering with the stretches along the azo bond providing the greatest intensity.

Introduction Azobenzene is the simplest example of a wide range of theoretically and practically important molecules containing the azo chromophore. Vibrational spectroscopy is often used to elucidate the in situ properties of these molecules. For example, there are recent studies on cis-trans isomerization and the acid base equilibria of azo comp~undsl-~ and in addition, resonance Raman spectroscopy has been used to study azobenzene derivatives which show tautomerism between the azo and hydrazo structures and therefore are of value as indicator^.^,^ Thus, a normal-mode calculation would be of value in providing an improved understanding of the nature of the vibrations in azoben~ene.~-l Raman scattering, resonance Raman scattering, and infrared spectroscopy of trans-azobenzene have been studied by several groups.8-'' There have also been a few surface-enhanced Raman scattering ~ t u d i e s . ' ~The - ' ~ assignment of the azo bands has been aided by the use of isotopically substituted derivat i v e ~ ~ . ' ~and - ' ~by studying the depolarization ratios of the Raman lines.'* In particular, Gruger et al. have published a series of papers detailing the vibrational spectra of trans-azob e n ~ e n e ,cis-azobenzene, ~ and cis-az~xybenzene'~and also trans-azoxybenzene,20 making use of isotopically substituted derivatives. The azo stretch was assigned to the 1470 cm-' Raman band and there was evidence that it is coupled to phenyl mode 19alb. They also found evidence that other phenyl modes, 12 or 1 and 6a, couple to azo vibrations. However, recent resonance Raman studies support a previous assignment of the mode at 1440 cm-' to the azo stretch.21-23 The other main Raman active azo band, the N=N-Ph symmetric bend, has been assigned to the 1143 cm-' band in these studies. The coupling of the azo vibration with the phenyl modes, especially the C-H Current address: Department of Biochemistry, Case Western Reserve University, 10900 Euclid Ave., Cleveland, OH 44106-4935. * To whom correspondence should be addressed. Abstract published in Advance ACS Abstracts, November 15, 1995. @

0022-365419512099-17825$09.00/0

bending modes such as 9a, is believed to contribute to the difficulty in assigning the azo trans-Azobenzene has two absorption bands in the visible and near-W regions, which are assigned to n* n and n* n transitions, respectively. The lower energy band occurs near 450 nm and is symmetry forbidden and less intense than the intense peak near 320 nm which is symmetry allowed.',2 The 320 nm band is shifted towards the visible region by the addition of electron withdrawing or electron-donating groups on the phenyl rings.5 Initial observations of the intensity enhancement of the high-energy Raman bands of azobenzene when the excitation wavelength is changed from the red to the blue were interpreted as a resonance effect involving the n* n tran~iti0n.I~However, resonance Raman profiles (REPS) of azobenzene2' obtained from carbon tetrachloride in the wavelength range 361 -494 indicate that the intensities are due to a preresonance effect involving the n* n transition. Further studies by the same group concluded that the highenergy symmetric vibrations derive intensity from the n* n electronic transiti~n.",~~ However, detailed studies of the REPS of azobenzene on the low-energy side of the n* n electronic transition revealed evidence for a contribution to the Raman intensity from this absorption f e a t ~ r e . Several ~ ~ . ~ ~groups have commented on the absence of any complete normal-mode calculations giving a precise description of the vibrations of azobenzene. The conformation of trans-azobenzene in solution is not known precisely. It is thought to be either planar and of C2h symmetry or slightly twisted from planarity and of Ci symmetry. The crystal structure shows a slight twist from planarity, but this could be due to crystal packing forces.27 A twist of approximately 30" around the N-Ph bonds has been detected by gas electron diffraction.** Kellerer et al. found that they could not detect any depolarized Raman peaks in the Raman spectra of azobenzene and its isotopically substituted derivatives." They concluded that azobenzene has Ci molecular

-

-

-

-

-

0 1995 American Chemical Society

-

Armstrong et al.

17826 J. Phys. Chem., Vol. 99, No. 51, I995

TABLE 1: (A) Predicted Azo Bond Lengths (A) and Azo Stretch Frequencies (cm-l) for Different Basis Set9 and (B) Results of SCF Calculation for Different Basis Sets Used To Arrive at the Final 6-31Ga Basis See (A) Predicted Azo Bond Lengths and Stretch Frequencies

SCF bond N=N N-C

X-ray25 1.247 (1.240) 1.428 (1.429, 1.427)

PM3 1.232 1.441

EDz8

1.259 (1.268) 1.420 (1.427)

6-31G 1.233 1.420

MP2 6-31G* 1.220 1.447

6-31G 1.310 1.447 MP2

SCF

azo band N=N phenyl mode3’

experimental9 1440 19a/b

PM3 1713

6-31G 1624 8ah

8ah

6-31Ga 1.275 1.431

6-31G* 1296 9a

6-31G 1296 9a

6-31Ga 1450 19ah

(B) Results of SCF Calculation

bond N=N C-N

6-31G** 1.220 1.421

6-31 1G 1.234 1.423

6-31Gfl on N 1.231 1.422

6-3lG+ 1,d on N 1.217 1.425

6-31+G(d) on N 1.218 1.422

a The equivalent phenyl modes that couple to the azo stretch varies from calculation to calculation. The modes concerned are given using the numbering system for benzene. The basis sets are described in the standard way for Gaussian 92,34J8The composition of 6-31Ga is described in the text. ED = gas phase electron diffraction. The basis sets are described in the standard way for Gaussian 92.34.381 is equivalent to an s and p function on the nitrogens. The numbers in parentheses after the X-ray distance refer to different forms of azobenzene in the crystal. The numbers following the electron diffraction refer to C, rather than C2 symmetry.

symmetry in solution and this has been supported by other IR and Raman investigations.14,17,22 In a short report on a normal coordinate analysis of azobenzene, Klima et al. assumed that azobenzene is planar and has C2h symmetry.” Barker et al. report that the resonance Raman and UV-visible spectra of azobenzene are unaffected by both polar and nonpolar solvents. They conclude that the conformation of azobenzene is mainly determined by intramolecular interactions, in particular the n-n and n-n interactions of the azo group with the phenyl rings, which tend to cause ring twisting about the axis of the N=N bond.22 There is also evidence of repulsive interactions of the ortho hydrogens of the phenyl ring with the lone electron pair of the distal nitrogen.* Substituents in the para positions, which are strong electron donors, confer greater planarity to the azobenzene derivative structure.29 To provide the theoretical background for a fuller interpretation of the vibrational spectra of azobenzene and related dyes, a normal-mode analysis of azobenzene utilizing molecular orbital calculations was undertaken. Semiempirical quantum mechanical methods (modified neglect of diatomic overlap (MNDO), using the parametric method 3 (PM3) Hamiltonian) have proved useful in assigning the vibrational spectra of the hydrazone form for the azo dye Solvent Yellow 14 (C112055).30 The change in energy and form of some of the modes of Solvent Yellow 14 upon exchange of the hydrazone hydrogen with deuterium was well predicted by the PM3 calculation giving confidence in their use in assigning the vibrational spectra of Solvent Yellow 14.3’ This method was used to predict the normal modes of azobenzene. However, the results proved to be unsatisfactory, and more rigorous ab initio calculations were performed. An initial geometry of azobenzene was inputted into the geometry optimization procedure with the phenyl rings titled in accordance with Ci symmetry. However, the resulting optimized structure was planar and of C2h symmetry. This symmetry was used in the normal-mode analysis. Methods

The semiempirical calculation for azobenzene used the MOPAC network of programs (version 6), with the PM3 H a m i l t ~ n i a n . ~An * . ~ab ~ initio calculation was performed at the SCF level with the 6-31G basis set using the Gaussian 92

program.34 However, both these calculations had unsatisfactory features and in particular the predicted azo bond length and the azo stretch frequency are unsatisfactory. An ab initio calculation using the larger basis set (6-31G*) with polarization functions (d orbitals) on the carbon and nitrogen atoms also predicted the main azo stretch to be at a higher energy than that found experimentally and the N=N bond length to be smaller than the experimental X-ray and electron diffraction value (Table 1A). Further geometry optimizations with various larger basis sets produced similar or smaller N=N bond lengths to the 6-31G basis set (Table 1B). To obtain better agreement with experimental values, account was taken of electron correlation effects by carrying out calculations at the Moller-Plesset level, MP2 (Table 1). Electron correlation ab initio techniques do not require the scaling factor of 0.89 which is usually required for standard SCF calculations to be applied to the frequencies. The MP2 ab initio calculation using the 6-31G basis set predicted the main azo stretch at too low an energy compared to that found experimentally and moreover coupled this stretch to the phenyl ring mode 9a for which there is no experimental evidence. In addition, the N=N bond length increased due to the introduction of electron correlation. To improve the calculation, a previous basis set which at the SCF level decreased the N=N bond length in comparison with the 6-31G set was used at the MP2 level. The set chosen was based on the 6-31G basis set but with the orbitals on nitrogen replaced by orbitals from the 6-31+G(d) basis set. This adds diffuse functions by incorporating an extra d and 1 (s, p) orbitals and is ideal for atoms with lone pairs of electrons such as the azo nitrogen atoms. This basis set was denoted as 6-31Ga. Results and Discussion The vibrational modes with large displacements on the carbon atoms are relatively easily fitted, as for the hydrazo system studied previously. The difference between the present azo calculation and previous hydrazo calculations is that the nitrogens are less involved with bonding to carbons of the phenyl rings. Consequently, if displacements on the nitrogens are not correctly predicted, large discrepancies can occur in vibrations with large nitrogen displacements. However, the inclusion of larger basis sets increases the time required for the calculation. Thus, a basis set with an extended set on the nitrogens alone

J. Phys. Chem., Vol. 99, No. 51, 1995 17827

Vibrational Analysis of trans-Azobenzene

P

r-

V m 1623 @ab)

v6 1622

(m) r

V7 1614 (SJb)

V8 1519 (1%)

v 9 I s 0 4 (1%)

V52 1524 (1%)

V53 I494 (1%)

@ V y 1270 (C-N.13h)

VH 1392 (14)

$

J

V&J 1047 (I&)

V59 1114 (I&)

Vi8 1024 (12)

Vi9 933 (C-N,IZ/I)

V m 691 (GN.6)

v21 634

(a)

$8 V u 307 (C-N,%)

V u 555 ( C - N , )

V23 226 (C-N,%)

c

Figure 1. Calculated in-plane modes of azobenzene with energies (cm-l) and assignments for the A, vibrations. The differing lengths of arrow show differences in amplitude of the vibrating atoms. They are multiplied by 5 times their true value.

was optimum and sufficient to give reasonable fits. The Mp2/ 6-31Gacalculation proved successful in fitting both the bond lengths and the main azo stretch. The experimental value is 1440 cm-I and the predicted value is 1450 cm-'. As a consequence of the predicted C 2 h symmetry, of the 66 normal modes, 33 are IR active (1 1 A,, out-of-plane and 22 B,, in-plane modes), and 33 Raman active (10B,, out-of-plane and 23 Ag, in-plane modes). The calculated frequencies of azobenzene are displayed in Figures 1-4 and listed in Table 2 for azobenzene, azobenzene substituted with 15N, and azobenzene with deuterated phenyl rings. Comparison between the experimental and theoretical values for azobenzene is given in Table 3. The two phenyl rings can be regarded as monosubstitued benzenes. As such, the displacements of the phenyl ring modes are related to those of benzene. The relationship of particular vibrations to the parent phenyl modes is also given in Table 3.

V66 86 (GN,%)

Figure 2. Calculated in-plane modes of azobenzene with energies (cm-l) and assignments for the B, vibrations. The differing lengths of arrow show differences in amplitude of the vibrating atoms. They are multiplied by 5 times their true value.

V25 851 (l7a)

8

V x 825 (1%)

vn ne (loa)

V28 702 (11)

v29 444

In-Plane Vibrations The in-plane modes of azobenzene are the A, and B, modes shown on Figures 1 and 2. The modes dominated by C-H stretching vibrations, 211-215 and 2145-2149, have been omitted. The highest frequency C-C modes for azobenzene, 2150, 2151 and 216, 217, are related to the 8a/b mode of benzene. 2150 and Y6 are similar in character with regard to each ring, but 2150 is an infrared active mode, where the phenyl rings displacements are out of phase with each other, whereas 216 is the Raman active mode, and the displacements are in phase. Of the in-plane phenyl ring modes there is usually one infrared active mode for each Raman active mode. Mode 2110 (1450cm-l) is the main Raman active azo stretch, but the A, modes 218 and 219 also have a considerable azo

v33

4s (C-N)

Figure 3. Calculated out-of-plane modes of azobenzene energies (cm-l) and assignments for the B, vibrations. The diameters of the circles represent qualitativedifferences in the'amplitude of the vibrating atoms; open circles are positive displacements and closed circles negative.

displacement. Comparison of the calculated frequencies of these modes upon isotopic substitution of the nitrogens and hydrogens, with the results of experimental isotopic substitution studies, confirmed the azo stretch as 2110. This assignment is made to

17828 J. Phys. Chem., Vol. 99, No. 5I, I995

w

8 VM 828 (1%)

Vjs 851 (178)

@ V38 708 (I I)

v42 283 (C-N.16b)

v43 59

(C-W

v44 -20 (C-W)

Figure 4. Calculated out-of-plane modes of azobenzene energies (cm-I) and assignments for the A, vibrations. The diameters of the circles represent qualitativedifferences in the amplitudeof the vibrating atoms; open circles are positive displacements, and closed circles negative.

the experimental bands at 1410 cm-I for (C&I6l5N)2 and at 1470 cm-I for (C6D6N)2.9 Thus, a decrease of 29 cm-I for this band upon I5N substitution of the azo group is predicted for Y I Oin reasonable agreement with experiment. An increase in energy upon deuteration of the phenyl rings is also predicted by the calculations, though it is not as great as is found experimentally. The azo stretch, Y I O ,is the only high-energy mode to increase in energy upon phenyl ring deuteration; the others decrease. This increase may at first appear unusual. However, comparison of Y I O displacements for (C&jN)2 and ( c a d 9 2 reveals differences in the phenyl ring displacements. The azo contribution is greater and the C-H/D contribution less for (C6Da)2 than for (C6H6N)z.

I"

Armstrong et al. tion to the mode Y I Ofrom the azo group and the decreased contribution from the phenyl ring. The in-plane azobenzene vibrations not involving the azo group are relatively easy to assign to phenyl ring displacements. It appears that these vibrations are behaving as if the molecule consisted of two uncoupled monosubstituted phenyl moieties. This simplistic approach also holds true for the C-N modes. Monosubstituted benzenes, where X is the substituent, exhibit modes involving C-X in-plane stretching and bending, and outof-plane bending, assigned as monosubstituted benzene modes 13, 9b, and 16b, re~pectively.~~ These features can also be observed in the modes of azobenzene. Mode Y56 and ~ 1 are 5 described as C-N stretching modes, assigned to phenyl ring modes 13 and 9a. The carbon rings appear to vibrate in a manner similar to mode 13. However, the C-H bending contribution is very similar to 9a. Both these modes show a large shift down in energy upon ring deuteration, showing a large dependence on the C-H bending contributions. This mixing of the azo group vibration with C-H bending modes has been suspected from previous vibrational studies of azo dyes and has led to difficulty in assigning the azo vibration^.^-^ Previous studies have assigned the Raman active mode at 1143 cm-I to the C-N symmetrical stretch; the present study supports this and assigns this band to ~ 1 5 . Modes of 192, ~ 2 3 ,Y65, and Y M are C-N in-plane bending modes assigned to the phenyl ring mode 9b. Only Y65 of this group shows a high degree of C-N involvement revealed by the large 10 cm-' shift of this mode upon substitution of the nitrogens with I5N. All of the in-plane modes of azobenzene show some contribution from recognizable modes, to which the azo vibrations can couple. In addition to the modes discussed above, ~ 1 and 9 Y62 exhibit appreciable azo contributions, as seen from the large shift upon 15N substitution. These modes appear to be coupled to phenyl modes 12 or 1. Modes Y61 and ~32,which have been assigned to phenyl mode 12, show many similarities to modes ~ 1 and 9 'v62. Azo vibrations are also coupled to phenyl mode 6a in modes ~ 1 and 9 vu. The coupling of the azo vibrations to 12 or 1 and 6a phenyl modes was suggested by Gruger et al? Our data are in keeping with the experimental finding and also the suggestions of vibrational coupling of the A, modes of the phenyl rings and the azo Out-of-Plane Vibrations

L (c6H6N)2 v10

1450 cm", A,, (N=N, 19&)

(C6D6N)2 v8

1458 cm", A,, (N=N, 19a/b)

The azo stretch mode of azomethane appears at 1576 cm-I and can be considered a pure azo stretch, with little contribution from the methyl groups.18 Diazine, H-N-N-H, the simplest azo compound has an N=N stretch assigned to 1529 cm-1,35 which has been predicted accurately by a6 initio calculation^.^^ Experimentally, the diazine azo stretch is found to increase to 1539 cm-I upon deuterium substitution to D-N-N-D. This increase is in part accounted for by the mixing of this mode with the symmetric NNH bend which is observed at 1583 cm-' for H-N=N-H and 1215 cm-I for D-N=N-D.35 Thus, the behavior of the azo stretch is in agreement with previous studies, with the value being lower than the simple systems due to electron delocalization. The increase in the azo stretch upon ring deuteration can be accounted for by the increased contribu-

All six out-of-plane vibrations of benzene can be observed in the calculated modes of azobenzene. Only phenyl mode 16b, a C-X out-of-plane bending mode, couples to the azo group. The out-of-plane mode ~ 4 is 2 the C-N out-of-plane bending vibration, assigned to the phenyl ring mode 16b. Mode ~ 3 is9 also assigned to this out-of-plane displacement but has only a small C-N contribution. The low-energy vibrations ~ 3 2 , 1 9 3 , ~43,and YM do not relate to recognizable phenyl modes. The earlier simplistic approach of treating azobenzene as two monosubstituted benzene moieties breaks down for these modes and the whole molecule must to be considered. 'vucan be described as a phenyl torsion in which the phenyl rings are twisted in opposite directions. ~32,~33, and v43 are described as phenyl flaps. The calculation gave two modes with negative values (~33and 214).This is attributed to the combined basis set, 6-31Ga, in which the basis set for the nitrogens was changed to 6-31+G(d) and the carbons and hydrogens were treated with the 6-31G basis set. This problem did not appear in any of the many vibrational analysis we performed on azobenzene using other basis sets.38 Both of these vibrations, ~ 3 and 3 YM,involve a twisting motion. The twist

Vibrational Analysis of trans-Azobenzene

J. Phys. Chem., Vol. 99, No. 51, 1995 17829

TABLE 2: Modes of Azobenzene and Isotopic Derivatives from MP2/6-31Ga Calculation (CsH615N)z 322 1 3210 3197 3187 3178 1621 1613 1517 1501 1411 1394 1363 1234 1226 1189 1110 1047 1023 915 689 634 307 225 857 85 1 825 778 702 444 400 381 21 1 -83

mode

(CsHd\Jh

(C6D6N)Z

3221 3210 3197 3187 3178 1622 1614 1519 1504 1450 1395 1368 1235 1226 1195 1110 1047 1024 933 69 1 634 307 226 857 85 1 825 778 702 444 400 381 213 -85

2387 2378 2365 2354 2346 1586 1577 1458 1391 1371 1357 1165 1080 984 92 1 898 886 848 826 662 609 289 215 674 670 619 608 534 407 384 343 20 1 -83

mode Au

v34 v35 v36

v37 v38 v39 v40

V41 v42

v43 V44

B"

v45 v46 v47 v48 v49

v50

v5 I v52 v53 v55 v54 v56 v57 v58 v59 V60

v6 I v62 v63 VfA v65 v66

(Cd-1615N)2 860 85 1 828 775 708 542 418 387 280 59 -20 3221 3210 3197 3187 3178 1623 1613 1523 1494 1391 1362 1264 1226 1210 1113 1047 1024 827 640 548 534 86

(C6H6N)Z 860 85 1 828 775 708 542 418 387 280 59 -20 3221 3210 3197 3187 3178 1623 1609 1624 1494 1392 1363 1270 1226 1211 1114 1047 1025 834 640 555 524 86

(C6Dd\J)2 682 674 635 607 549 483 407 344 267 56 -18 2387 2378 2365 2354 2346 1587 1570 1404 1377 1352 1202 1078 984 902 886 856 849 777 615 541 515 81

TABLE 3: Experimental Results9 Assigned to Calculated Azobenzene Modes (in em-') experimental

'

IR

R

1591 1586 1484 1454

1591

1307 1295 1218 1181 1155 1070 1021 1@)cl 983 967 926

1491 1470 1440 1312

theoretical, MP2/6-3 lGa

IR 1623(Bu) 1609(Bu) 1524(Bu) 5 1494(B,) 3

v50 v51 v52 ~

~ 5 1392(Bu) 4

~ 5 1363(BJ 5

1185 1155 1143 1068 1021 967 935 914

v56 1270(Bu) ~ 5 1211(Bu) 7 ~ 5 1226(B,) 8 ~ 5 1114(Bu) 9

~~1047(B,) V ~ 1025(Bu) I ~ 3 860(Au) 4 ~ 3 85 5 ~(Au) V36 828(Au) v62

834(BJ

R

experimental description

IR

R

8ah 8a/b 19a 19b N=N, 19ah 14 3 C-N, 13/9a 9a 9b C-N, 13/9a 18b 18a 12 17a 5 17b C-N, 12/1 C-N, 12/1

834 774 690

834 773 689 667 61 1 541

shown in the gas phase electron diffraction structure is greater than that found in the crystal structure. However, the relationship between the vibrational data and the twist angle is not sufficiently detailed to c o n f i i that these two modes are particularly sensitive to the angle. V M , a phenyl torsion mode, suggests a slight twist of the phenyl rings to lower the symmetry to C;.~ 3 3 a, phenyl flap mode, shows a large contribution from the azo bond and the ortho hydrogens, suggesting a slight twist of the phenyl rings to increase the separation between the ortho hydrogens and the distal nitrogen. However, comparison of the displacements for these modes calculated with the 6-31G basis set on the whole molecule at the MP2 level show that the forms of the vibrations and the positive values of the energies are virtually identical.

617 544 544 523 409

406

theoretical, MP2/6-31Ga

IR

R

description 10a 11

C-N, 6b C-N, C-N, C-N, 16b 16a 4

6a 6a 16b 9b

362

C-N, 16b C-N, 9b

315 253 250 218 86

C-N, 9b C-N C-N, 9b

To change the basis set to 6-31+G(d) for the whole azobenzene molecule would be prohibitively expensive in computing time. The low-energy modes of azobenzene have some similarities to those of t r a n s - ~ t i l b e n e . ~They ~ , ~ ~both show a phenyl torsion mode for the lowest energy Au mode: VM in azobenzene. Stilbene also has modes that are described as phenyl flaps and show similarities to ~ 3 2 1, 9 3 , and ~ 4 in 3 azobenzene. The normal-coordinate analysis of trans-stilbene shows many similar features to the normal-mode analysis of azobenzene presented trans-Stilbene was treated as possessing CU, symmetry and displays a progression of phenyl modes, some of which are coupled to the ethylenic b1idge.4~3~~ The frequencies of stilbene and azobenzene modes are close for those modes dominated by the phenyl vibrations. Stilbene, in a similar manner to azo-

Armstrong et al.

17830 J. Phys. Chem., Vol. 99, No. 51, 1995 TABLE 4: FT Raman Scattering from Azobenzend freauencv (cm-') assignment 1590 m 1491 m 1472 m 1440 st 1313 w 1182 m 1146 st 1001 m i s t

st, strong: m, medium; w, weak.

benzene,28 has been shown to twist approximately 30" about the C-Ph bonds in the gas phase.43 Both molecules are likely to have C, symmetry in solution. Comparison between Theoretical and Experimental Data

Table 3 shows a new assignment to the experimental data published by Gruger et ~ l .to, the ~ normal modes of azobenzene, (C&N)2. There is a good match for the two sets of data in the high-energy region, above 900 cm-I, where the vibrations are all in-plane modes. Although the absolute values of the predicted energies are not the same as the experimentally observed energies (there can be differences from a few to over 100 cm-I), there are strong similarities. For example, the difference in the energies between the infrared and Raman active azobenzene modes assigned to phenyl modes 18a ( v 7 , v d and 18b (139, vl6) reflect the experimental results. The difference in frequency between modes v59 and is appreciable and is observed experimentally, whereas for vi7 and v60 it is not appreciable and is not observed experimentally. In addition, I5N and 2H substitution produces comparable theoretical and experimental changes as discussed. The theoretical model predicted C2h symmetry for azobenzene, while the experimental evidence suggests that azobenzene adopts C; symmetry in solution. This difference is thought to be the reason for some of the bad fit between experimental and theoretical energies. However the modes presented here are thought to be a reasonable representation of the molecular displacements. Raman scattering has been studied with excitation from the near-infrared to the near-ultraviolet. The Fourier transform (FT) Raman spectrum of the solid (Table 4) differs only very subtly from that obtained with red excitation.2* All intense modes are due to in-plane vibrations with C-N and N=N displacements. The assignments are given in Table 4. All other vibrations are appreciably weaker than those listed in Table 4 including lowfrequency modes and C-H stretches. The reason for the intensities can be explained qualitatively on the basis of the displacements shown in Figure 1. The two modes at about 1600 cm-I, ?.6' and v7, appear as one feature. It is not clear whether this is due to lack of resolution, a difference in intensity, or exact redundancy. However, v6 involves a displacement in a direction that elongates the molecule and would be expected to produce a significant change in polarization and hence to be more intense in Raman scattering. ~ 1 has 3 some analogous features. All other high-frequency in-plane modes involve significant C-N or N=N displacements and some C-C displacements, but the displacements are small and less likely to produce a significant polarization. The complete absence of V I 6 and ~ 1 for 7 which there is no N=N or C-N displacement indicates the importance of displacements along the linking bonds between the rings in producing polarization changes. In addition, the lower frequency modes ~ 1 and 9 v20 have large nitrogen displacements perpendicular to the azo bond.. These do not seem to produce effective scattering. vl5 and V I 8 also have this feature and a

characteristic displacement pattem on the phenyl rings. It would appear that this pattem gives rise to strong Raman scattering. Thus, three features give rise to intense Raman scattering, namely in-plane elongation of the molecule, N=N and C-N stretches, and symmetrical in-plane displacements of alternative carbon atoms in the rings. In a study of azobenzene using different red excitation frequencies, the same bands were observed as in the FT Raman spectrum. However, the relative intensity of v10 varied compared to vg and v9 depending on the state of the sample (solid or liquid) or the solvent.** The clear difference between Y I O and the other modes is that v10 is dominated by the large displacement on one bond, the azo bond, and the direction is not along the axis of the molecule. It would appear that environmental changes affect the relative Raman scattering efficiency from v10 differently from vibrations with a greater contribution from the other atoms in the molecule, presumably due to the differences in intramolecular forces round the phenyl rings and round the azo bond. With blue excitation, the dominance of the x* n transition in providing resonance intensity enhancement is evident. Substituted azos give resonance excitation profiles with a separate peak in the visible region between 450 and 500 nm in most cases. With azobenzene, the major change observed is a regular increase in scattering towards the UV,24where the n* n transition is situated. The most enhanced bands are vg, v9,and v10with v6 decreasing in relative intensity compared to ~ 1 5 .The common feature of vg, v9, and Y I Ois the displacement along the azo bond, suggesting that one difference between the x and n* states which is important for resonance intensity enhancement is the displacement of electron density along that bond. In addition, resonance excitation profiles generated in studies of the trans to cis conversion of azobenzene25x26indicate subsidary maxima in the region of the x* n transition. v6, v8, and v10all have strong underlying increases in intensity in this region but lower frequency modes have more definite peaks. This effect is explained in terms of constructive and destructive interference. It is now clear that those vibrations with large displacements along the azo bond have a larger constructive than destructive interference effect. Lower frequency bands give weaker resonance Raman scattering, but selected bands are enhanced. The calculation can make assignments to suitable bands, but with the lower energies the assignments are tentative. The situation is made more complex by the possibility that some bands (e.g., those at 939 cm-' and 910 cm-I) may be due to overtone^.^^ A specific study has been made of the mode at 220 cm-'.** The likely assignment is to ~ 3 2 , an out-of-plane mode with a large displacement on the nitrogen atoms and involving a twist of the N=N bond out of the plane. This displacement suggests a good n* x overlap during the vibration. Thus, similar criterion to those required for efficient Raman scattering apply in resonance, indicating that the n* n Franck-Condon overlap on the phenyl rings is important. Additional Raman intensity is obtained when there is good overlap during displacement along the azo bond either in-plane or out-of-plane.

-

-

-

-

-

Conclusions The ab initio calculation carried out at the MP2 level using a combined basis set gave a reasonable fit for bands assigned experimentally as azo vibrations and as phenyl ring modes in previous assignments. Lower level calculations failed to predict the frequencies of the azobenzene modes with reasonable accuracy. This failure is thought to lie in the neglect of electron correlation and in the ability to correctly treat the properties of the azo bond.

Vibrational Analysis of trans-Azobenzene Using the extended basis set, the high-energy vibrational modes were predicted well and could be assigned to experimentally observed bands with confidence. The good fit for the main azo vibrations especially, YIO, was only possible using the combined 6-31Gabasis set. There is agreement with the observed increase in energy of the azo stretch mode upon deuteration of the phenyl rings. The modes show vibrational coupling between the phenyl modes and the azo vibrations as previously suspected from experiment. The low-energy modes were more difficult to fit with experimental results. The form of the displacements indicated by the calculation gave a pattem from which it was possible to explain the intensities of the bands in the Raman spectrum. Raman scattering is strongest for modes with C-N and N=N vibrations and from modes which elongate the molecule causing a change in polarization of the molecule along its length. All vibrations which are intense involve carbon and nitrogen atoms; C-H displacements do not give strong Raman scattering experimentally. Resonance Raman scattering is most pronounced for the same modes. However, an additional increase in intensity is obtained for modes with N-N and C-N displacements in the direction of the azo bond. In addition to using the calculation to interpret resonance Raman data, they form a basis for a better understanding of many Raman studies of substituted azo compounds. Acknowledgment. We thank Ciba Pigments PLC for a grant in support of one of us (J.C.), Professor I. Macpherson and Dr. 1. Fraser of Ciba for discussions, and Dr. G. Dent of Zeneca and Ms. C. Rodger of Strathclyde University for the use of €T Raman data. References and Notes (1) Rau, H. Angew. Chem., lnr. Ed. Engl. 1973, 12, 224. (2) Foster, C. L.; Kelusky, E. C.; Bunce, N. J.; Zemer, M. C. J . Am. Chem. Soc. 1985, 107, 5884-5890. (3) Cimiraglia, R.; Hofmann, H.-J. Chem. Phys. Lett. 1994, 217, 430. (4) Machida, K. In Raman Spectroscopy: Sixty Years On Vibrational Spectra and Structure; Bist, H. D., During, J. R., Sullivan, J. F., Eds.; Elsevier Science Publishers B.V.: Amsterdam, 1989; p 421. (5) Cataliotti, R. S.; Morresi, A,; Paliana, G.; Zgierski, M. Z. J . Raman Spectrosc. 1989, 20, 601. (6) Kumar, K.; Carey, P. R. Can. J . Chem. 1977, 55, 1444. (71 Cataliotti. R. S.: Mureia. S. M.: Paliana, G.: Poletti. A.; Zgierski, M. Z. J. Raman Spectrosc. 19’85, 16, 251. 18) Lin-Vien, D.: Colthun N. B.; Fateley, W. G.; Grasselli, J. G. The Handbook of Characteristic Frequencies of Organic Molecules; Academic Press: New York, 1991. (9) Gruger, A.; Le Calve, N.; Dizabo, P. J . Chim. Phys. 1972,69,291. (10) Klima, M. I.; Kotov, A. V.; Gribov, L. A. J . Struct. Chem. 1972, 13. 987.

J. Phys. Chem., Vol. 99,No. 51, 1995 17831 (11) Biancalana, A.; Campani, E.; Di Domenico, G.; Masetti, G. J . Raman Spectrosc. 1993, 24, 43. (12) Nishihara, C.; Shindo, H.; Hiraishi, J. J . Electroanal. Chem. 1985,

191, 425. (13) Gao, P.; Gostola, D.; Weaver, M. J. J. Phys. Chem. 1997,92,7122. (14) Hacker, H. Spectrochim. Acta 1%5, 21, 1989. (15) Hacker, H. H. Inaugural Dissertation, University of Munich, Germany, 1968. (16) Schrotter, H. W. In Raman Spectroscopy; Szymanski, H. A., Ed.; Plenum: New York, 1970; p 69. (17) Kellerer, B.; Hacker, H.; Brandmuller, J. Indian J. Pure Appl. Phys. 1971, 9, 903. (18) Hertberg, G. lnji-a-red and Raman Spectra of Polyatomic Molecules; Van Nostrand: New York, 1945. (19) Gruger, A.; Le Calve, N.; Dizabo, P. J . Chim. Phys. 1972,69,743. (20) Gruger, A.; Le Calve, N. Spectrochim. Acta 1972, 28A, 1253. (21) Koide, S.;Udagawa, Y.; Mikami, N.; Kaya, K.; Ito, M. Bull. Chem. Soc. Jpn. 1972, 45, 3542. (22) Barker, I. K.; Fawcett, V.; Long, D. A. J . Raman Spectrosc. 1987, 18, 71. (23) Biancalana, A.; Campani, E.; Gorini, G.; Masetti, G.; Quagli, M. J . Raman Spectrosc. 1992, 23, 155. (24) Houben, J. L.; Masetti, G.; Campani, E.; Gorini, G. J . Raman Spectrosc. 1982, 13, 15. (25) Biswas, N.; Umapathy, N. Chem. Phys. Lett. 1995, 236, 24. (26) Okamto, H.; Hamaguchi, H. Chem. Phys. Lett. 1986, 130, 185. (27) Bouwstra, J. A.; Schouten, A.; Kroon, J. Acta Crystallogr. 1983, c39, 1121. (28) Traetteberg, M.; Hilmo, I.; Hagen, K. J . Mol. Struct. 1977.39, 23 1. (29) Lunak Jr., S.;Nepras, M.; Hrdina, R.; Mustroph, H. Chem. Phys. 1994, 184, 255. (30) Munro, C . H.; Smith, W. E.; Armstrong, D. R.; White, P. C. J . Phys. Chem. 1995, 99, 879. (31) Armstrong, D. R.; Clarkson, J.; Munro, C. H.; Smith, W. E., Unpublished work. (32) Dewar, M. J. S.; Thiel, W. J . Am. Chem. Soc. 1977, 99, 4907. (33) Stewart, J. J. P. J . Comput. Chem. 1989, IO, 209. (34) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Gill, P. M. W.; Johnson, B. G.; Wong, M. W.; Foreman, J. B.; Andres, J. L.; Raghavachari, K.; Binkley, J. S.; Stewart, J. J. P.; Pople, J. A. Gaussian 92/DFT, Revision G.1; Gaussian Inc., Pittsburgh, PA, 1993. (35) Bondybey, V. E.; Nibler, J. W. J . Chem. Phys. 1973, 58, 2125. (36) Harrison, R. J.; Fitzgerald, G. B.; Laidig, W. D.; Bartlett, R. J. Chem. Phys. Lett. 1986, 124, 291. (37) Varsanyi, G. Vibrational Spectra of Benzene Derivatives; Academic Press: New York, 1969. (38) Foresman, J. B.; Frisch, E . Exploring Chemistry With Electronic Structure Methods: A guide to Using Gaussian; Gaussian, Inc.: Pittsburgh, PA, 1993. (39) Haller, K.; Chiang, W.-Y.; Laane, J. Proc. XlVth lnt. Con$ Raman Spectrosc. 934, 1994, 934. (40) Chiang, W.-Y.; Laane, J. J . Chem. Phys. 1994, 100, 8755. (41) Negri, F.; Orlandi, G.; Zerbetto, F. J . Phys. Chem. 1989, 93,5124. (42) Palmo, K. Spectrochim. Acta 1988, 44A, 341. (43) Traetteberg, M.; Frantsen, E. B.; Mijlhoff, F. C.; Hoekstra, A. J . Mol. Struct. 1975, 26, 57.

JP95 10498