THEELECTRONIC SPECTRUM OF THE CYCLOOCTATETRAENYL RADICAL ANION The second assumption to be checked is the range of the structure-stabilizing effect of the tetraalkylammonium ions. We have tacitly assumed that the water lattice is most strongly stabilized in the vicinity of the great cations and that this effect decays fairly rapidly with increasing distance from the cations. Some preliminary information regarding this decay could be obtained from a recent investigation carried out by Maijgren and Odberg,47who measured the line width of the 85Rb resonance in 0.5 M aqueous solutions of RbCl and tetraalkylammonium bromides in the concentration range 0.2-1 M . They found that the width of the 85Rb signal stayed within the range of change attributable to the viscosity changes of the solution, indicating that a t tetraalkylammonium concentrations 2 0 . 2 M the stabilization effects on the water structure remaining outside the zone of nearest approach between the R b + ion and the complex cation are unobservable by the measuring techniques so far employed.
2813
Acknowledgments. Professor H. G. Hertz is heartily thanked for providing information regarding his experimental results, especially those for the relaxation of 170in aqueous solutions, prior to publication and for stimulating suggestions and comments to a preprint of this paper. Thanks are due to Professor Henry Frank and to Dr. Felix Franks for valuable comments at discussions and in writing and Dr. Gunilla Gillberg is thanked for helpful discussions in the course of the work. Financial support given by the Swedish Natural Research Council and by the Swedish Council for Applied Research is gratefully acknowledged. The nmr spectrometer was made available to the Nmr Research Group, the Royal Institute of Technology, by a generous grant from Knut and Alice Wallenberg’s Foundation.
(47) B. Maijgren and L. Gdberg, personal communication.
The Electronic Spectrum of the Cyclooctatetraenyl Radical Anion1 by Paul I. Kimme12aand Herbert L. Department of Chemistry, University of California at Berkeley, Berkeley, California Q.Jr20 (Received J a n u a r y 6, 1968)
The cyclooctatetraenyl radical anion has been made by electrolysis in liquid ammonia. The electronic spectrum of the anion is compared with the results of a series of calculationsfor different possible geometries, We conclude that the anion differs little, if at all, from D 8 h symmetry. The conformations and the electronic structure of cyclooctatetraene and its anions have been of interest for many years. The cyclooctatetraene molecule is, of course, nonplanar and nonaromatic. However the dianion is almost surely planar and shows such aromatic features as a ring ~ u r r e n t . ~The , ~ conformation of the anion radical is more difficult to establish.4 Simple theoretical arguments suggest that the resonance energy to be gained by the anion on assuming the planar position should be about half of that gained by the dianion. Analysis of the results of polarography and of electron spin resonance experiments has shown that the anion differs significantly from the neutral molecule in configuration. The hypothesis that the anion is planar is consistent with all the available evidence, but the possibility that the anion is somewhat distorted from a regular octagon cannot be excluded.6 This article describes t,he electronic spectrum of the cyclooctatetraenyl anion and compares the spectrum
with calculations performed assuming several possible geometries. We conclude that the geometry is indeed close to that of a regular octagon.
Experimental Section The cyclooctatetraenyl anion was generated in liquid ammonia both by electrolytic reduction and by the use of potassium metal in ammonia and by electrolytic reduction in methylamine. Both the solvents and the (1) Supported in part by a grant from the Petroleum Research Fund of the American Chemical Society. (2)(a) Alfred P. Sloan Fellow, 1966-1968. (b) Address correspondance to this author at Department of Chemistry, Douglass College, New Brunswick, N. J. 08903. (3) T. J. Kats, J . Amer. Chem. SOC.,82, 3784, 3786 (1960). (4) H. L. Strauss, T. J. Kats, and G. K. Fraenkel, ibid., 85, 2360 (1963). (6) See A. Carrington, H. C. Longuet-Higgins, R. E. Moss, and P. F. Todd, Mol. Phya., 9, 187 (1966), for the ear spectrum of the monodeuteriocyclooctatetraenyl anion, and R. E. Moss, ibid., 10, 601 (less), for the esr spectrum of alkylcyclooctatetraenyl anions.
Volume 78, Number 8 August 1968
2814
salts used as supporting electrolytes were purified by standard methods. The cyclooctatetraene was obtained from the Aldrich Chemical Co., was distilled, and was stored over nitrogen at 0". The electron spin resonance spectra were taken in an apparatus described in detail by Levy.6 Optical spectra were taken in a similar cell (Figure l), which rested in a Styrofoam-insulated box. The box fit into the sample compartment of a Cary Model 14 spectrophotometer and a stream of cold nitrogen was used to keep the cell and sample compartment sufficiently cold. The cyclooctatetraenyl radical was first made by electrolysis in liquid ammonia using tetramethylammonium iodide as a supporting electrolyte. The anion was identified by its electron spin resonance spectrum, which consisted of the expected nine lines, equally spaced, with a splitting of 3.278 f 0.005 G.' For comparison, the spacing of the hyperfine lines in the spectrum of the cyclooctatetraenyl anion made by lithium reduction in tetrahydrofuran (THF) is 3.209 f 0.007 G4 and the spacing of the lines in the spectrum of the anion made by electrolytic reduction in dimethylformaniide (DRIF) is 3.23 rt 0.03 G.8 The radical anion is quite stable in our system. The signal obtained was extremely strong and lasted for a t least 0.5 hr after the electrolysis current was turned off. The optical spectra were taken using a variety of different salts as supporting electrolytes. All the salts used were perchlorates instead of iodides, since the iodide ion absorbs strongly in the 2500-8 region. Tetramethylammonium, lithium, and sodium perchlorates were all used with ammonia as a solvent. The tetran-butylammonium salt was used in methylamine, since the tetramethylammonium salt was not sufficiently soluble in this solvent. The spectrum of the anion. in ammonia using tetramethylammonium perchlorate (TNIAP) as a supporting electrolyte is shown in Figure 2. There is a possibility that a much stronger peak exists a t about 2400 8, but we are uncertain, owing to the strong absorption of the solvent. There is certainly strong absorption by the anion a t this and lower wavelengths. On greatly increasing the electrolysis current, the anion peaks disappeared and the rather nebulous spectrum of the dianion became visible. The dianion spectrum agreed closely with that of Farrell and Mason.9 On turning off the current, this spectrum disappeared and the anion reappeared, gradually, showing that the anion is indeed stable under these conditions. The anion spectra taken with the alkali metal salts were virtually identical with the spectrum of Figure 2. The spectrum taken in methylamine using the TMAP was very similar except for a small red shift. Attempts to observe the spectrum of the anion in methylamine with alkali metal salts as supporting electrolytes were unsuccessful. The Journal of Physical Chemistry
PAULI, KIMMELAND HERBERTL. STRAUSS f
To
] T u n g s t e n Wire T u n g s t e n a t joint
(?
Pt Glass Seal (made vacuumtight w i t h epoxy)
F o i l Anode
-Pt
- - - G l a s s Frit
'M
Pt Cathode Covered
-Beckmann I cm Quartz C e l l Figure 1. Diagram of the cell used for taking the optical spectra. 2,o- 1
I
I
I
-
1.8
.-E a" -
1.4 -
-
c
.-" 0
-
-
1.6-
1.21.0-
0.s0.6-
-
-
0.4
0.2I
2000
2500
3000
3500
4000
x in 8. Figure 2. Electronic spectrum of the cyclooctatetraenyl radical anion a t a concentration of approximately 2 X loT8 M . The ammonia solvent absorbs strongly below about 2300 d. Another possible peak exists in the anion spectrum at about 2400 2%.
Discussion I n relatively nonpolar solvents such as T H F and
DMF, the cyclooctatetraenyl anion (COT-) is known to disproportionate into the neutral molecule and the dianion (6) D.H.Levy a r d R. J. Myers, J. Chem. Phys., 41, 1062 (1964). (7) This spectrum was first obtained with the help of Levy; see D. H. Levy, Ph.D. Dissertation, University of California, Berkeley, Calif., 1965. (8) R. D. Allendoerfer and P. H. Rieger, J. Amer. Chem. rSoc., 87, 2336 (1965). (9) P. G. Farrell and 8.F. Mason, 2.Naturforsch., 16b, 848 (1961).
THEELECTRONIC SPECTRUM OF 2COT- +COT
THE
2815
CYCLOOCTATETRAENYL RADICAL ANION
+ COT2-
(1)
Allendoerfer and Reiger8 have shown that this disproportionation in the more polar solvent dimethylformamide is dependent on the presence of alkali metal cations, equilibrium shifting to the left in the absence of these anions and to the right in their presence. It has also been shown that the anion is relatively stable in dimethyl sulfoxide,* acetonitrile,'O and tetrahydrofuran" in the absence of alkali metal cations. These results indicate that the strong ion pair preferentially formed between the alkali metal cation and the cyclooctatetraenyl dianion is capable of markedly shifting equilibrium 1 l 2 In ammonia, 'however, the anion is stable even in the presence of sodium or lithium ions, and we take this to be evidence of the existence of relatively little ion pairing in this solvent. The anion spectrum was easily obtained in methylamine in the absence of alkali metal cations but could not be obtained in the presence of these anions; this suggests that some ion pairing with the resulting shift in the disproportionation equilibrium takes place in methylamine as the solvent. We may assume then that the electron spin resonance spectra and the optical spectra taken in 1iquid.ammonia represent the spectra of the relatively uncomplexed cyclooctatetraenyl anion. The difference between the behavior of the anion in ammonia and in methylamine and other solvents can be rationalized by noting that the ion pairing is a very steep function of the dielectric constant of the solvent over a critical range of parameters.l3 The hyperfine splitting of the anion in ammonia is slightly greater than that in the other solvents. Much larger shifts than these have been observed for the benzene anion as a function of ternperat~re.'~ The variation in the benzene case was attributed to the variation of the ion pairing with dielectric constant and thus with temperature, but the variation could also be due to a variation of the interaction of the anion with the solvent molecules. The optical spectrum of the cyclooctatetraenyl anion appears to arise from the three distinct transitions listed in Table I. A number of other small features are visible in the spectrum (Figure 2), but these are presumably due to vibrational structure in the electronic bands. We have carried out an extensive series of calculations as a function of the molecular geometry in an attempt to deduce the geometry of the anion from the information of Table I.
MOs -
Energy
I
Table I: Experimental Transitions of COT-
Wavelength,
Energy,
A
eV
Oscillator strength
5 3.83 3.14
Large 0.028 0.0024
+5
$4 #21#3
dl
-
-
--
-
02
a+PI-P2
01
~ + P ~ - P I
e
a
bl
.+PI+P2
-
m
Figure 3. Hiickel energy levels and representations for cyclooctatetraene with different possible symmetries. The symmetry designations follow those of G. Herzberg, "Molecular Spectra and Molecular Structure]" Vol. 111, D. Van Nostrand Co., Inc., Princeton, N. J., 1966.
The calculations were performed assuming three different geometries for the radical anion: (1) planar with all C-C bonds of equal length, D8h symmetry; (2) planar, with alternate C-C bonds of different length, D 4 h symmetry; and (3) tub shaped with either equal or alternating C-C bond lengths, D2d symmetry. The Huckel energy levels for these three cases are shown in Figure 3. The anion has nine *IT electrons, and the regular D8h geometry would, therefore, lead to a degenerate ground state. Distortions from such a geometry are then to be expected by the Jahn-Teller theorem.'6 Various distortions from the regular structure are possible, but the most likely one is a change of bond lengths to provide a D4h structure.6J6 Although distortion to a D2d structure is not predicted by the Jahn-Teller theorem, it is not excluded either, and it is considered here since the neutral cyclooctatetraenyl molecule has this tub-shaped structure. The electronic structure was calculated for two different bond lengths for the D8h structure; the first is 1.397 A, the bond length for benzene;lB the second is 1.409 8, the bond length calculated by Coulson for the (10) A. H.Maki, unpublished work. (11) G. K. Fraenkel, unpublished work. (12) See also F.J. Smentowski and G . R. Stevenson, J . Amer. Chem. Soc., 89, 6120 (1967). (13) R. M. Fuoss and C. A. Kraus, ibid., 55, 1019 (1933). (14) R. W. Fessenden and 8. Ogawa, ibid., 86, 3591 (1964); G. Vincow, J . Chem. Phys., 47,2774 (1967),presents some temperaturedependent factors of the isolated radicals. (15) A. D. McLachlan and L. C. Snyder, J . Chem. Phys., 36, 1159 (1962). (16) A. Langseth and B. P. Stoicheff, Can. J. Phys., 34, 350 (1956). Volume 79,Number 8 August 1068
PAULI. KIMMELAND HERBERT L. STRAUSS
2816 Table I1 : Predicted Transition Energies and Oscillator Strengths of COT-
0-
= 1 3 5 O (planar)--Energy, Oscillator eV strength
---e
= 133’ (D$d)-Energy, Oscillator eV strength
-
0 126O 46‘ (angle of neutral molecule) (D2d) Oscillator Energy, eV strength
r1 = r2
= 1.397
ila
6.27 6.22 3.93 3.62 3.41
1.79 1.81 0.000189 0.0209 0.00005
8.57 6.6 6.10 4.15 3.03
0 00442 3.31 0.206 0.0163 0.0790
13.23 11.24 7.58 5.98 1.80
0.000463 0.0116 2.90 0.0565 0.197
=
= 1.409
k
6.15 6.12 3.83 3.53 3.28
1.78 1.81 0,000184 0.0209 0.00005
8.36 6.44 5.88 4.05 2.95
0.00482 3.29 0.186 0.0163 0.0765
12 06 11.09 7.48 5.88 1.79
0.000454 0.0123 2.92 0.0571 0.197
= 1.432 h;,
6.93 6.33 4.82 3.83 3.32
0.00326 3.41 0,00858 0.0151 0.0734
9.34 6.74 5.22 4.58 2.76
0.0000024 3.42 0.244 0.0192 0.145
13.88 11.82 7.83 6.36 1.60
0.000969 0.00849 2.71 0.0644 0.212
rl
= 1.462 A, r2 = I ,334 h; (same as neutral COT)a
8.96 7.14 6.91 4.95 2.95
0.000568 2.04 1.02 0.0138 0.225
11.49 9.34 7.62 5.79 2.37
0.000692 0.0220 2.93 0.0253 0,242
16.30 13.54 8.90 7.64 1.56
0.00157 0.00320 2.42 0.0782 0.267
r1
=
r2
=
I .397 Ab
6.06 5.81 4.23 3.46 3.15
1.50 1.59 0.0008 0.080 0.0002
r1 =
r2
= 1.409
Ab
5.95 5.70 4.12 3.35 3.02
1.53 1.59 0.00085 0.086 0.0002
r1
TI
r2
r2 = 1.387 k
Pariser and Parr.21
I
Mataga and Nishimoto.21
dianion.I7 The D4h structures were taken with the bond lengths calculated by Snyder1*and another set of bond lengths that correspond t o the neutral molecule.1g For the D2dstructure various angles between the 135” angle of the hypothetical planar molecule and the 126” of the neutral molecule were taken. For the calculations in which the molecule was not assumed to be planar, the various many-center integrals were taken proportional to the overlap integrals between the relevant p orbitals centered on the different atoms.20 In the calculations, the configuration interaction was taken into account between all five sets of doublet states that arise from filling the lowest Huckel Dsh orbitals with nine electrons and then exciting one electron one level. The corresponding configurations were considered in the less symmetric cases. The calculations were checked by comparing the D8h and D4h calculations in the limit that the D~I,bond lengths became equal. The check is a particularly good one, since the determinantal wave functions look quite The Journal of Physical Chemistry
different when written out for these two different symmetries. Further details are to be found in ref 21. The results of these calculations are to be found in Table 11. The various sets of results for the Dsh structure agree a t least qualitatively with the experimental results. The oscillator strength of the lowest energy transitions is appreciably lower than that deter(17) C. A. Coulson, Tetrahedron, 12, 193 (1964). (18)L. C. Snyder, J . Phys. Chem., 6 6 , 2299 (1964). (19) 0.Bastiansen, L. Hedberg, and K. Hedberg, ibid., 27, 1311 (1958). (20)A. Streitwieser, Jr., “Molecular Orbital Theory of Organia Chemists,” John Wiley and Sons, Inc., New York, N. Y., 1961. (21) The computer programs used and other details are described at length in P. I. Kimmel, Ph.D. Dissertation, University of California, Berkeley, Calif., 1967. The integrals yere calculated by the methods of Pariser and Parr (R. G. Parr, Quantum Theory of Molecular Electronic Structure,” W. A. Benjamin, Inc., New York, N. Y., 1963) and by those of N. Mataga and K. Nishimoto, 2.Phys. Chem. (Frankfurt), 13, 140 (1957). The method of calculating intensities is discussed in C. Sandorfy,“ Electronic Spectra and Quantum Chemistry,” Prentice-Hall, Inc., Englewood Cliffs, N. J., 1964.
DETERMINING THE ABSORPTION SPECTRA OF MIXEDPHOTOCHROMIC ISOMERS mined experimentally, but the excess oscillator strength may well be contributed by a vibronic mechanism.22 I n a vibronically allowed band, one might expect to see vibrational fine structure, and indeed the lowest energy band does show such structure. The calculated intensity of the second transition agrees well with experimental results. The third transition is expected to be very weak and would not be seen under the nearby strong bands. Finally, the calculations predict much stronger bands at higher energy, and this agrees with the in-
2817
crease of absorption observed at high frequency, though not with our possible maximum at about 5 eV. As the geometry is changed away from D8hsymmetry, calculations predict an enormous increase in the intensity of the lowest energy transition. This is a clear contradiction to the experimental results and would seem to rule out all but a very small distortion of the cyclooctatetraenyl anion from D g h symmetry. (22) W. Moffitt, J . Chem. Phys., 2 2 , 320 (1954).
A Procedure for Determining the Absorption Spectra of Mixed Photochromic
Isomers Not Requiring Their Separation' by Joseph Blanc and Daniel L. Ross RCA Laboratories, Radio Corporation of America, Princeton, New Jersey 08640
(Received January 11, 1968)
By correlating changes in absorption spectra on irradiation with the changes in the intensity of fluorescence emitted by solutions containing a pair of reversible photochemical isomers, one of which is fluorescent, it is possible to calculate the extinction coefficients of the nonfluorescent isomer as a function of wavelength, even where the absorption spectra of the two isomers show considerable overlap. By appropriate calculations, close approximations to the extinction coefficients of the fluorescent isomer can also be obtained over much of the range of its absorption. The procedure has been applied to thioindigo and N,N'-diacetyl indigo. The results indicate that the extinction coefficients of the nonfluorescent cis isomers can be determined to an accuracy of about 375, while those of the trans isomers a t their absorption maxima can be estimated to within approximately 4%.
Certain indigoid dyes are known to show the property sists of essentially 100% of the trans isomer. Recently, Fischere has proposed a method for determining the of light-induced cis-trans isomerization. 2-6 We have been studying such systems and wished to evaluate the absorption spectrum of one of a pair of isomers if the extinction coefficients of the other are known and if it quantum yields of the photochromic process. To obcan be assumed that the ratio of the quantum yields tain this information requires an accurate knowledge of the extinction coefficients of each of the two isomers at for the forward and reverse photochemical reactions is the wavelengths of interest or, in other words, the independent of wavelength . absorption spectra a t known concentrations. Other workers have previously attempted to obtain absorp(1) Presented at the Informal Symposium on Reversible Photochemition spectra of pure cis- and trans-indigoids by chromacal Processes, Dayton, Ohio, May 1967. tographic separation of mixtures of the isomers,216but (2) G.M. Wyman and W. R. Brode, J. Amer. Chem. SOC.,7 3 , 1487 (1961). the difficulties inherent in this technique (necessity to (3) W. R. Brode and G. M. Wyman, J . Res. Nat. Bur. Stand., 47, exclude light, low solubility of the compounds, and 170 (1961). the inevitable thermal equilibration of the isomers) have (4) W. R. Brode, E. G. Pearson, and G. M. Wyman, J. Amer. Chem. Soc., 76,1034 (1954). led to the development of methodsbased on mathe(6) matical analysis of absorption spectra of m i ~ t u r e s . ~ * ~(1966). J G. M. Wyman and A. F. Zenhllusern, J . Org. Chem., 30, 2348 These earlier methods have in common an assumption (6) D. A. Rogers, J. D. Margerum, and G. W. Wyman, J. Amer. which is not independently testable, namely that there Chem. SOC.,79,2464 (1957). (7) W. R. Brode, J. H. Gould, and G. M. Wyman, ibid., 74, 4641 are one or more regions in the spectrum where only one (1962). of the two isomers has any appreciable absorption. (8) I. Y. Bershtein and Y. L. Kaminskii, Opt. Spectrosc., 15, 381 Another assumption that has been made7 is that a (1963). thermally equilibrated system, kept in the dark, con(9) E.Fischer, J. Phy8. Chem., 71,3704 (1967). Volume 72,Number 8 Augu8t 1868
