J . Phys. Chem. 1987, 91, 1357-1360 bent. If the electronic transition is forbidden but vibronically induced, several of the bands would have to be assigned as magnetic dipole allowed transitions. In the absence of further information, the exact nature of the electronic transition cannot be determined. A careful search was made, out to 8000 A, for the corresponding triplet-singlet transition but no further bands were observed. Initially, in the analysis of the H2CCS spectrum, some of-the weak bands in the red end were thought to belong to the 5 X system, because they appeared double-headed. However, they_ are s,o readily incorporated into the vibronic pattern of the A X transition that this hypothesis was rejected. In sharp contrast to the spectra of formaldehyde and thioformaldehyde, the tripletsinglet system in thioketene is too weak to be observed by conventional absorption spectroscopy using long-path multiple reflection techniques. The lack of rotationally discrete structur_e and the absence of observable fluorescence indicate that the A state of thioketene is predissociated. In gas-phase flash photolysis experiments, a strong absorption spectrum of the thioketyl radical, HCCS, has been observed immediately after the photolysis of thi~ketene.~' Whether the radical is a primary photochemical product or results from a secondary abstraction reaction such as CH2 H2CCS CH3 + H C C S is unknown.
-
-
+
-
(27) Clouthier, D. J.; Ramsay, D. A., unpublished.
1357
Assuming that comparisons of the excited states and electronic transitions of ketene and thioketene are valid, an in-plane bent excited-state structure is suggested for ketene, in agreement with the conclusion of Dixon and Kirby2 and the results of extensive calculations.5 The absence of any evidence for thioketene triplet-singlet bands agrees with the reassessment of the ketene vibronic spectrum by Laufer and Keller3 as a single band system. However, the thioketene bands s an a rather narrow region of about 8000 cm-' centered at 5000 whereas the ketene spectrum covers some 17 000 cm-I centered a t about 3700 A. This large difference probably reflects to some degree the limited pressure-path attainable with thioketene due to its instability, but also suggests more extensive Franck-Condon activity and a more distorted excited state in the ketene case. Such a conclusion is consistent with the a b initio results.
1,
Acknowledgment. The author thanks Dr. D. A. Ramsay for generously providing time on the laboratory facilities at the Herzberg Institute of Astrophysics, Mr. M. Barnett for his invaluable aid in photographing the spectra, and Dr. M. Torres for advice on the synthesis of the thiadiazoles. This work was supported in part by the Director, Office of Energy Research, Office of Basic Energy Sciences, Chemical Sciences Division of the U S . Department of Energy under Contract No. DE-FG05-86ER13544 and in part by a grant from Research Corporation. Registry No. Thioketene, 18282-77-4; thioketene-& 4789-21-3; ketene, 463-51-4.
Gas-Phase Molecular Structure and Conformation of Benzil As Determined by Electron Diffraction Quang Shen*+and Kolbjarn Hagen** Department of Chemistry, Colgate University, Hamilton, New York 13346, and Department of Chemistry, AVH, University of Trondheim, N-7000 Trondheim, Norway (Received: September 4, 1986)
The molecular structure of benzil has been investigated by gas-phase electron diffraction at 175 ' C . One conformer with a O=C-C=O torsion angle of I$ = 117 (3)' (4 = 0' for a syn conformer) and with the phenyl rings nearly coplanar with the carbonyl groups (dB= 10 (1)') was identified. This gas-phase conformation is close to the conformation earlier observed in the solid phase. Values of the bond distances (r,) and valence angles (La), with estimated 3a uncertainties, are r(C-H) = 1.096 (11) A, r(C=O) = 1.220 (4) A, r(C-C)r,ng= 1.399 ( 2 ) A, r ( C l - C l l )= 1.488 (8) A, r(C1-C2)= 1.546 (16) A, LC-C=O = 119.9 (14), and LC-C-C = 118.7 (9)'.
Introduction The conformation of benzil has interested chemists for many years. It is an a-dicarbonyl system with the general formula cox-cox (x = c6H5). The simplest molecule of this type, glyoxal, with X = H, has been reported to exist as two different forms with the carbonyl groups anti or syn to each other.' When the two hydrogen atoms are replaced by methyl groups, as in 2,3-butanedione (biacetyl), the anti form is the only stable conformer. Hedberg and Danielson found that even at 525 OC there was no evidence for the presence of any conformer but the anti.* The existence of the planar forms is probably due to the presence of conjugation in the a-dicarbonyl system. Nonplanar forms have also been observed in systems of this type. In oxalyl chloride (X = Cl)3 and oxalyl bromide (X = Br)4 the molecules exist in the vapor phase as a mixture of anti and gauche conformers. The nonplanar form has been rationalized in terms of repulsive forces (such as nonbonded repulsion, electron-pair repulsion, and dipole-dipole repulsion) being larger than the conjugation stabi'Colgate University. f University of Trondheim.
0022-3654/87/209 1- 1357$01 SO10
lization in the syn configuration. But in all of these molecules the most stable form has been found to be the planar anti. In benzil (X = C6H5) the phenyl rings introduce two important factors into the a-dicarbonyl system: (a) they are bulky and should be sterically important, and (b) the K orbitals in the phenyl ring can interact with the carbonyl K system and influence the conjugation in the O=C-C=O system. This may affect the conformation of benzil and make it different from other COX-COX systems. The conformation of benzil in the solid state has been studied several time^.^,^ Only one form has been observed with the carbonyl groups rotated about 110' from one another. Mea(I,) Kuchitsu, K.; Fukuyama, T.; Morino, Y . J . Mol. Struct. 1968, l , 463. Curne, G.N.; Ramsay, D. A. Can. J . Phys. 1971, 31 7, 49. Durig, J. R.; Tong, C. C.; Li, Y. S. J . Chem. Phys. 1972, 57, 4425. (2) Danielson, D. D.; Hedberg, K. J . Am. Chem. SOC.1979, 101, 3730. (3) Hagen, K.; Hedberg, K. J . Am. Chem. SOC.1973, 95, 1003. (4) Hagen, K.; Hedberg, K. J . Am. Chem. SOC.1973, 95, 4796. (5) Gabe, E. J.; Le Page, Y.;Lee, F. L.; Barclay, L. R. C. Acta Crystallogr., Sect. B 1981, B37, 197. (6) Brown, C. J.; Sadanaga, R. Acra Crysfallogr.1965, 18, 158.
0 1987 American Chemical Society
Shen and Hagen
1358 The Journal of Physical Chemistry, Vol. 91, No. 6, I987
9
P
TABLE I: Final Parameter Values for Benzil
parameter" r(C-H) r(C=O) r(C-C)n,g r(CI-Cl1) r(C1-Cz) LCl--C,=O LC2-Cl-CII
db &Be
1.096 (1 1) 1.220 (4) 1.399 (2) 1.488 (8) 1.546 (16) 119.9 (14) 118.7 (9) 116.9 (34) 9.9 (10)
calcd
refined
0.076 0.039 0.045 0.047 0.049
0.044 (1)
Selected Dependent Distances 2.420 2.496 2.604 2.793 2.834 2.943 3.052 3.127 3.289 3.600 3.769 3.869 4.215 4.264 4.767 5.010
Figure 1. Molecular model of the anti (6 = 180') conformer of benzil, CI4Hl0O2, showing the atomic numbering.
surements of Kerr constants and dipole moments7v8for benzil have shown an O=C-C=O torsion angle of 90-100' (4 = 0' for the syn conformer). Evans and Leermakersg were the first to conclude that the ground- and excited-state structures of benzil were different by comparing the absorption and emission spectra of benzil and biacetyl. The excited state of benzil is near-planar, while the ground state is nonplanar. These results were later confirmed by Morantz et al., Bera et al.," and Chan et a1.12 In the triplet excited state of benzil, a nearly planar molecule with a torsion angle of 4 = 160' was found. It therefore appears that the ground-state gas-phase structure of benzil is nonplanar, and this would be the first small a-dicarbonyl system where the most stable conformer is nonplanar. No gas-phase structure has been reported for benzil, and we therefore initiated an electron diffraction investigation to determine the structure and conformation of benzil in the gas phase.
Experimental and Data Analysis A commercial sample of benzil was obtained from Aldrich Chemical Co. (>99%) and was used without further purification. Diffraction patterns were recorded with the Oslo electron diffraction unitI3 on Kodak Electron Image plates with a nozzle tip temperature of 448 K. The voltage/distance calibration was done with benzene as reference. The nozzle-to-plate distances were 480.70 and 200.71 mm for the long and the short camera experiments. Five plates from the long and four plates from the short camera distances were selected for analysis. Optical density of the scattering intensity was measured with a single-beam microdensitometer constructed in Oslo. The data, after the usual data reducing procedures, were converted to integral units of q ( q = 40/X sin 0 = s l O / r ) and averaged together to form a long and a short camera experimental intensity curve. Refinements of the structure were carried out by the method of least squares14 based on intensity curves using a unit weight matrix. Atomic scattering and phase factors used in the analysis were obtained from the tables of Schafer et al.Is (7) Cureton, P. H.; Le Fevre, C. G.; Le Fevre, R. J. J . Chem. Sac. 1961, 4447. (8) Le Fevre, C. G.; Le Fevre, R. J. W. Reu. Pure Appl. Chem. 1955, 5, 261.
(9) Evans, T. R.; Leermakers, P. A. J . Am. Chem. SOC.1967,89, 4380. (10) Morantz, D. J.; Wright, A. J. C. J . Chem. Phys. 1971, 54, 692. (1 1) Bera, S. C.; Mukherjee, S. C.; Chowdhury, M. J. Chem. Phys. 1969, 51, 754. (12) Chan, I. Y.; Heath, B. A. J . Chem. Phys. 1979, 71, 1070. ( 1 3) Bastiansen, 0.;Hassel, 0.;Risberg, E. Acra. Chem. Scand. 1955, 9, 232. (14) Hedberg, K.; Iwasaki, M. Acfa Crystallogr. 1964, 1 7 , 529
0.055 0.063 0.066 0.059 0.088 0.102 0.155 0.121 0.092 0.070 0.065 0.071 0.092 0.086 0.070 0.081
0.060 (3) 0.064 (7)
Distances (To) and vibrational amplitudes are in angstroms, angles are in degrees. Parenthesized values are 3u from least-squares torsion angle. = 0 for the syn conrefinements. bO=C-C=O former. eC2-Cl-Cll-C12 torsion angle. (La)
The geometrical parameters chosen to define the structure of benzil are as follows: r(C-H), r(C=O), r(CI-C2), r(Cl-Cll), r(Cll-C12), LC,-C,=O, L C ~ - C ~ - Cand ~ ~ the , two C-C torsion angles +B and 4 (see Figure 1 for atom numbering). In the phenyl ring all carbon-carbon bonds are assumed to be equal and all valence angles are 120'. 4B,the C12-C11-Cl-C2 torsional angle, defines the deviation from planarity of the phenyl ring with respect to the -CCO group. When 4B> 0 both C12and C22are rotated clockwise with respect to the Cl-C2 bond at the back. 4 is the O=C-C=O torsion angle, and a positive value of 4 corresponds to a clockwise rotation of the O2 atom relative to the C 1 4bond at the back. Vibrational amplitudes (I) and perpendicular amplitude corrections ( K ) were calculated from a force field developed from force fields of related molecules. Radical distribution (RD) curves (Figure 2) were calculated by Fourier transformation of the experimental intensity curves (Figure 3) and of theoretical curves calculated for different models. The planar anti form (4 = 180°, 4B = 0') was first tested, and a poor agreement between experimental and theoretical curves was obtained. Rotation of the phenyl rings out of the C 1 0 1 C 2plane (dB > 0) did not improve this agreement. When 4 was changed, a significant improvement of the fit between experimental and theoretical data was obtained. In the least-squares refinement the converged value of 4 depended on the starting value. With a starting value of 4 = 80', the least-squares refinement converged to a model with 6 = 56'. With a starting value of 4 > 90°, the converged value for the torsion angle was 4 = 117'. The least-squares minimum with 4 = 1 17' clearly was a better minimum than 4 = 56' with a lower R factor (R= 7.8% vs. 9.8%) and a better fit between experimental and theoretical RD curves. In all of the refinements 4B was also allowed to refine simultaneously with 4, and for all models a small value of 4Bwas observed (4B< 20'). For a model with only one conformer the best agreement was obtained with 4 = 117' and 4B= 10'. Small deviations from planarity are, however, not very well determined by the electron diffraction methode. The value (15) Schafer, L.; Yates, A. C.; Bonham, R. A. J . Chem. Phys. 1971, 55, 3055.
The Journal of Physical Chemistry, Vol. 91, No. 6,1987 1359
Electron-Diffraction Study of Benzil
TABLE II: Parameter Values Obtained for Benzil and Some Related Molecules O=CX-XC=O
parameter"
(COW2
(COW2
(COCHJ2
(COCh
(COBr)2
(COOW2
L61b
1.546 (16) 1.488 (8) 1.220 (4) 119.9 (14) 118.7 (9) 116.9 (34)
1.530 (14) 1.515 (7) 1.210 (2) 119.5 (6) 116.6 (2) 180
this work
1.528 (3) 1.745 (2) 1.181 (2) 124.3 (2) 111.8 (2) 180 56 (4) 3
1.540 (5) 1.927 (3) 1.178 (2) 124.8 (3) 111.6 (3) 180 62 (10) 4
1.548 (8) 1.339 (4) 1.208 (2) 123.1 (18) 111.9 (18) 180
ref
1.526 (3) 1.132 (8) 1.212 (2) 121.2 (2) 112.2 (17) 180 0 1
r(C-C) r(C-X) r(C=O) LC-c=o LC-c-x
&be
2
18
'Distances are in angstroms, angles in degrees. b O C C O torsion angle for conformer with lowest energy. c O C C O torsion angles for a second conformer. TABLE III: Bond Lengths (A) and Angles (deg) Obtained for B e n d and 4,4'-Dinitrobenzil from X-ray and Electron Diffraction Exwriments
parameter" r(C=O) r(CI-C2)
r(C-CPhO) r(C-C),i,g LC-C=O
Lzl--Iz Diff.
a
2
a
6
4
perimental intensity curve.
\ I V
v
6BC method ref
" Parenthesized values are
1.219 (21) 1.533 (21j 1.486 (21) 1.398 (21) 118.1 (12) 119.3 (12) 111.2 6.9
1.214 (9) 1.542(9) 1.460 (9) 1.375 (9) 116.0 (6) 120.0(6) 108.4 6.5
1.212 (15) 1.535 (isj 1.489 (1 5) 1.382(15) 117.0 (12) 119.6 (12) 111.5 (12)
X-ray
X-ray
X-ray
6
5
30..
O=C-C=O
16
torsion angle.
C 2 - C l - C l l - C , 2 torsion angle.
results from this refinement are shown in Table I. Only a few of the vibrational amplitudes were refined; the rest were kept constant at the calculated values.
Discussion The ED experiment has shown that the conformation of benzil
" -
@)b
1.220 (4) 1.546 (16) 1.488 (8) 1.399 (2) 119.9 (14) 118.7 (9) 116.9 (34) 9.9 (10) ED this work
nitrobenzil
18A
Figure 2. Radial distributioncurves calculated from the intensity curves of Figure 3 after multiplication by Zc2/Ac2and with damping coefficient B = 0.0009 A2. Theoretical data were used for q C 7 A-' in the ex-
1
LC-C-Ph
4.4'-di., . _.
benzil
--
Diff.
-
in the vapor phase is nonplanar, and there is no evidence for the presence of any significant amount of a second conformer. These results are consistent with previous spectroscopic studies. The observed O==C-C=O torsion angle in the vapor phase is found to be about 117'. There does not therefore appear to be a great change in the conformation between the solid and gas phase. Table I1 summarizes the important geometrical parameter values for benzil and some related molecules. The central C-C bond in benzil (1S46 (16)A) is long, as in all the cases presented. The nonplanar conformation and the long C-C bond suggest the presence of little, if any, conjugation in the bicarbonyl moiety. In a system like benzil, the overlaps between the ?r systems of the phenyl and the carbonyl and between the two carbonyl groups should favor a planar anti conformation, but the bulky phenyl groups make this form unfavorable. The steric hindrance can be reduced by (a) rotating the phenyl ring out of the C-C+ plane or by (b) rotating the bicarbonyl system out of plane. The values ~ suggest that alternative b is favored. of 4 (1 17') and $ J (10') This could be the result of a compromise between maximum reduction of steric hindrance and minimum loss of delocalization. The C1-Cll bond is observed to be short (1.488 (8) A) and the C=O bond is long (1.220 (4)A), and this is as expected since the phenyl ring and the carbonyl group are nearly coplanar, making electron delocalization possible. The C1-C1, bond distance has almost the same value as that observed in terephthalaldehyde, O==€H-Ph-HC=0,17 and it is also close to the value observed (1.466 (21) A) and in p-ethylbenzin 2-chloroben~aldehyde'~
i
0
20
40
6a
80
180
120
148
imq
Figure 3. Average experimental molecular intensity curve for benzil shown together with the theoretical curve calculated from the parameter values of Table I. Difference curve is experimental minus theoretical.
of 4Bmay also be affected by the calculated perpendicular amplitudes, and the uncertainty in &, may therefore be somewhat larger than the estimated value of 1.0'. In view of the conformational composition of the oxalyl halides (anti and gauche), a two-conformer model was also tested. The anti and gauche forms were assumed to have the same geometrical parameters except for the O=C-C=O torsion angle. Refinements failed to converge with the presence of the anti form, and when its composition was allowed to vary, a negative value was obtained for the amount of anti. From this electron diffraction study there is therefore no evidence for the presence of an anti form. Other two conformer models were also tested, but none of them provided an improved fit with the experimental data. In the final refinement a one-conformer model was used, and the
(16) Kimura, M.; McCluney, R. E.; Watson, W. H. Acta Crystallogr., Sect. B 1979, B35, 483. (17) Domenicano, A,; Hargittai, I.; Portalone, G.; Schultz, G.In Proceedings of the X V Meeting of the Italian Crystallographic Association: Monte Porzio Catone: 1984, as referenced in J . Mol. Struct. 1985, 129, 81. (1 8) Nahlovska, 2.;Nahlovsky, B.; Strand, T. G. Acta Chem. Scand. 1970, 24,
2617.
(19) Schafer, L.; Samdal, S.; Hedberg, K. J . Mol. Srruct. 1976, 31, 29.
J . Phys. Chem. 1987, 91, 1360-1365
1360
aldehydez0(1.471 (10) A). The carbon-carbon bond distances in the phenyl ring have the same value as in other similar molecules. Table I11 compares the molecular structure obtained for benzil in the gas and solid phase. Also included are the results determined for 4,4’-dinitroben~il.’~The agreement between the different investigations is very good. The old X-ray6 and our ED investigation give almost identical results for benzil, and both structure and conformation therefore seem to depend very little on what phase the molecules are in. In the gas phase the molecules are a little closer to the planar anti form, and if any effect was to be (20) Brunvoll, J.; Kolonits, M.; Bohn; R. K.; Hargittai, I. J . Mol. Struct. 1985, 131, 177.
expected, it would be the opposite. But the difference is small.
Acknowledgment. We are grateful to cand.rea1. Arne Almenningen and sivhg. Ragnhild Seip for help with the electron diffraction experiment and to Ms. Snefrid Gundersen for technical assistance. Financial support from the Norwegian Marshall Fund and from the Norwegian Research Council for Science and the Humanities is acknowledged. Registry No. Benzil, 134-8 1-6. Supplementary Material Available: Tables listing correlation matrix for the refined parameters, total scattered intensity, sL‘I,(s), for each plate, and average molecular intensity after background subtraction (6 pages). Ordering information is given on any current masthead page.
Effect of Added Salt on the Optlcal Absorption Spectra of Solvated Electrons in Liquid Ammonia Sidney Golden,+Thomas R. Tuttle, Jr.,* and Salia M. Lwenje Department of Chemistry, Brandeis University, Waltham, Massachusetts 02254 (Received: September 25, 1986)
Optical absorption spectra of solvated electrons in ammonia solutions of sodium iodide at three different temperatures and at three different salt concentrations all exhibited essentially the same identical shape as that observed in pure ammonia. At each salt concentration, the spectra shifted to longer wavelengths as temperature was increased. At each temperature, the spectra shifted to shorter wavelengths as salt concentration was increased. The salt-effected shifts are shown to arise from the interaction of solvated electrons with the added ionic components, principally through ion pairing with sodium cations. Intrinsic shifts of the optical absorption spectra of the pertinent solvated-electronspecies are shown to be related to the changes produced in their standard free energies by the added salt. Theoretical estimates of the intrinsic shifts compare quite well with those obtained from the analysis of the experimental data.
Introduction To date, several publications have dealt with the changes that occur in the optical absorption spectra of solvated electrons in ammonia when salts are added to their solutions. The results published so far, however, do not enable a full assessment to be made of the effects of the added salts. For example, the solvated electron spectra in ammonia have been reported to be blue-shifted slightly when sodium iodide is added to the solutions.l,2 A similar effect has been indicated for dilute alkali metal solutions in amm ~ n i a and ~ - ~in perdeuterioammonia.6 Nevertheless, the optical absorption spectra of the blue solutions which were generated electrolytically in solutions of alkali, alkaline earth, and quaternary ammonium iodides in ammonia have been described as unaffected by changes in the salt concentration.’ In addition, it has been ~ l a i m e dthat ~ , ~a new absorption shoulder at 12 500 cm-I appears in the optical spectra of dilute sodium-ammonia solutions to which sodium iodide has been added. This result has not been confirmed by other^,'-^,^,' however. It is evident that a precise determination of the salt-effected changes which are produced in the optical absorption spectra of solvated electrons in ammonia, as well as their characterization in molecular and structural terms, remains yet to be obtained. This is the motivation for the present investigation. Accordingly, we have determined the optical absorption spectra of increasingly dilute solutions of metallic sodium in liquid ammonia which contained four different fixed concentrations of sodium iodide at three different fixed temperatures. The solvated electron spectra corresponding to each of these fixed conditions were obtained by extrapolating the relevant spectra to infinite Emeritus Professor of Chemistry
dilution in metal. The experimental procedures and results are described briefly in the Experimental Results section. Within experimental precision, the effect of increasing concentrations of sodium iodide is to produce a slight, monotonically increasing blue shift of the solvated-electron spectrum without any significant change in its shape. In the Theoretical Results section, the theory used in analyzing the shift data quantitatively and the results of its application are presented. They lead to the conclusion that the shifts are primarily due to the ion pair of the (solvated) solvent-anion complex8 comprising the solvated electron with the (solvated) sodium cation. A discussion of the results is given in the Discussion section.
Experimental Results Essentially the same procedures as those described in earlier investigations carried out in this laboratory1,2~1s’2 were employed (1) Rubinstein, G. Ph.D. Dissertation, Brandeis University, 1973. (2) Rubinstein, G.; Tuttle, T. R., Jr.; Golden, S. J . Phys. Chem. 1973, 77, 282. (3) Gold, M.; Jolly, W. L. Inorg. Chem. 1962, I , 818. (4) Clark, H. C.; Horsfield, A.; Symons, M. C. R. J . Chem. SOC.1959, 2478. (5) Catterall, R.; Symons, M. C. R. J . Chem. SOC.1964, 4342. (6) Burrow, D. F.; Lagowski, J. J. In Soluared Electron, Gould, R. F., Ed.; American Chemical Society: Washington, DC, 1965; Advances in Chemistry Series No. 50, p 125. (7) Quinn, R. K.; Lagowski, J. J. J . Phys. Chem. 1968, 72, 1374. 1969, 73, 2326. (8) Golden, S.; Tuttle, T. R., Jr. J . Chem. Soc., Faraday Trans. 2 1982, 78, 1581. (9) Hurley, I. Ph.D. Dissertation, Brandeis University, 1970.
0 1987 American Chemical Society 0022-3654/87/2091-1360$01.50/0
