Ab initiocalculations and photoelectron spectra of cyclic oxamides

摘要

1980 1815 Ab initio Calculations and Photoelectron Spectra of Cyclic Oxamides By Roland lsaksson and Tommy Liljefors," Department of Organic Chemistry 3, Chemical Center, University of Lund, P.O.Box 740, S-220 07 Lund, Sweden Photoelectron spectra of five-, six-, and seven-membered cyclic oxamides in which the oxamide unit is forced to adopt an s-cis or twisted s-cis conformation have been recorded. The spectra have been interpreted using ab initio calculations. In comparisons with acyclic s-trans oxamides it is found that the influence of the s-trans-s-cis conformational change on the ionization energies is masked by the effects of the alkyl parts of the ring systems. Ionization energies of cyclic oxamides with different ring sizes show some effects due to twisting around the inter- carbonyl bond, but also in this case the alkyl parts of the ring systems strongly influence the spectra.OXAMIDESand other a-dicarbonyl compounds have s-cis oxamide derivatives, but the p.c. spectra were morc recently been intensively investigated by various conveniently interpreted in terms of a partition into urea spectroscopic methods. Photoelectron (p.e.) spectro- and s-cis glyoxal. ~copy,l-~ In this paper we report the p.e. spectra of some cyclic U.V. absorption and emission ~pectroscopy,~~~ oxamides, in which the oxamide unit is forced to adopt and n.m.r. spectroscopy 9 have been employed in studies on the electronic and steric interactions in this class of compounds. Previous papers on p.e. spectra of oxamides have mainly been concerned with assignments of ionization events,l with the effects of N-alkylation on the ioni- zation energie~,~.~ and with the energy separation of the n +/n ionization process~s.~ Simple additivity rules for N-alkyl shifts ha1-e been established and have been very useful in the interpretation of pee.For oxamide and its N-substituted derivatives, two planar conformations, s-tram and s-cis, are possible. R bsol; R RR s -trans S -cis Steric interactions may force the oxamide unit to adopt twisted conformations. Very little is known about the conformational properties of this class of compounds, but oxamide itself is s-trans and planar in the crystalline state.1deg; N-Alkyl-substituted oxamides are generally also s-trans,* but may be more or less twisted around the carbonyl-carbonyl bond, depending on the degree of substitution.Tetra-alkyl-substituted oxamides are most severely twisted, with a dihedral angle close to go", according to ~.III.I-.~J~J~and dipole moment studies.13 Strong dipole-dipole repulsions between the carbonyl groups and repulsive steric interactions between nitrogen substituents prevent the s-cis conformation from being significantly populated. The planar s-cis conformation may not even be a local minimum on the potential energy surface .14 All oxamide derivatives studied so far by p.e. spectro- scopy have been s-trnns or s-trautsoid (twisted s-trans). The only exception is a study by McGlynn et al. on imidazolidine-2,4,5-trione and some of its hr-alkyl tlerivative~.~ These compounds may be considered as an S-cis-or twisted s-cis-conformation by the constraints of a ring system.Comparisons with the corresponding s-trans-compounds are made. Compounds (I)-(VI) have been studied. In earlier work, semiempirical CNDO/S calculations have been employed in the interpretation of p.e. spectra of o~amides.l-~ This type of calculation is parametrized to fit electronic spectra and it has shown to be well suited to the prediction of p.e. spectra using the negative of the calculated molecular orbital (MO) energies as an approximation to the ionization potentials (Koopmans' theorem). The success of the CNDO/S method in re- producing p.e. spectra implies that the deficiencies in Koopmans' theorem, i.e.its neglect of electron re-bsol;R' (Y) (?a) a;R =H a;R = H b; R = CH3 b;R = CH, organization energy in the molecular ion and of dif- ferences in correlation energies between the ground state molecule and the ion, to some extent are taken care of by the parametrization. Furthermore, this implies that the MO energics and the order of the MOs, as calculated by CNDO/S and similar methods, do not necessarily refer to the ground state of the molecule. It is thus to be expected that in some cases the order of the MOs in the ground state may be different from the order of the ionization events observed in the p.e. spectrum and assigned to these 0rbita1s.l~ In this work we have employed ah iizitio calculations to obtain MO energies, In order to calculate ionization energies for comparisons with experimental valucs from p.e.spectra, we have made empirical corrections for reorganization and correlation effects. This approach has been successful in earlier studies.l65l7 In this way we hope to be able to predict both the order and the energies of the MOs in the ground state nizd the order and energies of ionization events, as they show up in the p.e. spectra. The purposes of the present work are as follows: (i)to investigate the ability of the computational approach used to support the interpretation of p.e. spectra of oxamides, (ii) to study the influence of conformations of oxamides on the ionization energies in p.e. spectra, and (iii) to evaluate the dependence of the ionization energies on the ring size of cyclic s-cisoid oxamides.EXPERIMENTAL AND COMPUTATIONS Photoelectron Spectra.-P. e. spectra were recorded 011 a, Perkin-Elmer model PS-18 photoelectron spectrometer. The ionization energy was provided by the He1 (21.22 eV) resonance line. The range of temperatures used for solid samples was 53-150 "C. Spectra were calibrated with regard to energy and resolution using the 2P3l2line (12.13 eV) of Xe and the 2P3/2line (15.76 eV) of Ar. No decom- position of any molecule studied in this work was observed under the experimental conditions described above. Synthesis.-Tetramethyloxainide (111) was prepared according to Persson and Sandstrom.l* The synthesis of compounds (IVb), (Va), (Vb), and (VIb) will be described elsewhere.l4 Calculations.-The ab initio calculations were carried out using the computer program MOLECULE.l9 For carbon, nitrogen, and oxygen, a Gaussian basis set of seven s-type and three p-type functions contracted to two and one, respectively, was used (a minimal basis). The basis set for hydrogen was made up from four Gaussian s-type functions contracted to one. Calculations on oxamide (I) were also performed with a contraction to a ' split shell basis with four s-type and two ?-type orbitals for carbon, nitrogen, and oxygen and two s-type orbitals for hydrogen. Orbital exponents and contraction coefficients were those given by Clementi et al. (C, N, 0; 7/3 basis set) 2O and Huzinaga (H).21 In the latter case a scaling factor of 1.34 was used.The MOs have been classified accordiiig to the notation introduced by McGlynn et aZ.l The MOs discussed in this work and denoted n+, n-,n@,and xe are shown in the Scheme. MO energies were converted into ionization energies as described in detail below. Geometries.-The oxamide molecule has been studied by X-ray diffraction,lO but for the remaining compounds in-vestigated in this work no experimental geometries are available. Geometries used in the calculations were con- structed from those of analogous molecules. As a starting J.C.S. Perkin I1 point, the X-ray structure of oxaiiiide was modified in order to make the geometry more appropriate for the gas phase.Strong hydrogen bonding in the crystalline state signi-ficantly influences the geometric parameters of this mol-ecule. The niodifications made were guided by observed s -trans s -cis 0.".?:-N N 00Rn4N N 0 0 0 9 N N 0 a SCHEME differences in structural parameters between X-ray ?? and electron diffraction z3 geometries of ainides. The inter-carboiiyl bond length was given a value close to that of the corresponding bond length of glyoxal 24 and biacetyl 26 in TABLE 1 Geometries used in the ab initio calculations ~oiidIcngtlis (A) Bond angles (") C8P'-CSP' 1.52 CCO 123,126 c=o 1.22 C, r-C,pN 116, 106 C8P-N 1.RG C,,-N-H 120 N-H 1.08 CaP*--N-Capa 120, 112.3,b 126.5 N-Clp~ 1.46, H-C,,*N 109.5 1.44 C-H 1.09 C,p2-C,pa-N 110 Cst3-CsIP 1.54 a Thc syrnmctries uscd for thc cyclic compounds werc: C,,, for (IVa),C, for (Va) and (VIa), with CO-CO dihedral angles of 0, 20, and 60deg;,respectively.All amide units were kept planar. Used for (IVa). c Used for (Va). the gas phase, and bond angles were modified to values appropriate for the s-cis-conforniation.2G The structural parameters for the oxamide unit thus constructed were kept as unchanged as possible in the evaluation of the geometries for the cyclic compounds. The remaining parameters for the ring systems were estimated from standard values and from analogous ring compounds. 'The final geometries are given in Table 1. As shown below, moderate changes of the geometric parameters have an insignificant effect on the calculated orbital energies.RESULTS For s-trans-oxamide (I),using a minimal basis, the highest occupied MO is calculated to be xB, but the three MOs xB, n+, and ng are almost degenerate (Table 2). The n- orbital is significantly more stable. The MO order in s-cis-oxamide is calculated to the re, n+, x@, and n- (Table 2), with a larger separation between nB and arid a smaller separ- ation between n+ and n-than was found for the s-trans-conformation. The dependence of the MO energies on twisting arountl the intercarbonyl bond is shown in Figure 1. 132 bsol; CJ b 151 0 30 60 90 120 150 180 w(") 1FIGURE Dependence of orbital energies on dihedral angle for oxamide (I) Using the more extensive ' split-shell ' basis set, all four MOs increase in energy by approximately the same amount, ca.1 eV (Table 2), leaving the separations between the orbitals essentially unchanged and the orbital order un-affected. The same result is observed for s-cis and YOo TABLE2 Molecular orbital energies from ab inztio calculations (eV) Compound ni-=0 =B n-(I) s-trans 12.06 12.14 12.03 14.52 (I) s-trans 11.18 11.13 11.06 13.52 (I) 90" 11.87 12.31 12.06 14.26 (I) goo 10.95 11.42 11.15 13.15 (I) s-cis 12.23 12.10 12.55 14.09 (I) s-cis 11.36 11.20 11.57 12.85 (11) s-trans 12.02 11.69 11.52 14.48 12.02 12.49 11.73 14.69 (Val 12.20 11.26 12.13 14.11 (VW 11.78 11.40 11.55 13.89 ' Split-shell ' basis. Otherwise a minimal basis set was used.twisted oxamide (Table 2). These results justify our use of the minimal basis set for all other calculations described in this work to economize on computer time. The calcul- ated MO energies are given in Table 2. To obtain an estimate of how possible deficiencies in the constructed geo-metries of the cyclic compounds (1V)-(VI) would affect the computed energies, the sensitivity of these energies to moderate changes in geometrical parameters was investi- gated. A 0.02 A change in the C-C or C-N bond lengths resulted in MO energy changes of 0.1eV. The same result was obtained when the CCO bond angle was increased by 5" or the CCN angle decreased by the same amount. Thus, the effects of moderate geometrical changes on the computed MO energies are apparently small.Conversion of MO energies to ionization energies requires corrections for re- organization energies in the molecular ion and for differ- ences in correlation energies between the ground and ionized states. We have used an approach based on ab initio cal- culations,l69l7 which has been successfully employed in studies on the assignments of ionization events in p.e. spectra. The differences between calculated energies for the four highest occupied orbitals in s-trans-oxamide (I) (Table 2) and the corresponding ionization energies as observed .~and assigned by ILlcGlynn et ~ 1(Table 3), were formed and used as empirical corrections to (the negative of) the cnl- TABLE3 Calculated and experimental ionization energies for compounds (1)-(VI) Calculated Compound MO (eV) Experimcntsl (eV) (1) n, 9.80 a 7 10.50 =@ 11.04 n-11.72 (11) n+ 9.33 o 9.627 =a3 10.07 n, 11.20 (ITI) n+ 8.70 9.02 =@3 8.87 }8.9-9.2 9.34 =0 9.24 9.08 n-10.38 10.5 10.49 (IVa) n, 9.76 *B 10.04 XFI 11.60 v n-11.89 (IVb) =a3 9.299.16 } 9.3-9.5 n, re 10.53 10.7 n-11.37 11.5 (V4 ni-9.94 9.70 10.27 10.20Xe =@ 10.44 10.55 n-11.31 11.70 (Vb) =9 9.30 9.20 n+ 9.47 9.30 w3 9.56 9.60 n-10.79 10.36 (VTa) n+ 9.52 =e 9.86 =euro;I 10.41 n-11.09 (VW re3 }8.8-9.2n+ 98::;re 9.44 n-10.57 10.5 o Ref.2. Assignments based on CNDO/S calculations. The carbonyl-carbonyl bond 60" twisted. culated MO energies for other oxamide derivatives. The corrections thus obtained are n+ -2.26, n--2.80, 7re -0.99, and xe -1.69 eV.As discussed by Roos el aZ.,l7 the cor- rections should be constant for each orbital type through- out a series of similar ~nolecules. Since the corrections are largely determined by the gross structure of the orbitals the corrections for X~ and xe should be inverted for s-cis- 1818 J.C.S. Perkin I1 /i ._.J 132 I (VIb) e FIGURE Photoelectron spectra of compounds (III), (TVb), (Va and b), and (Llsquo;Ia). Vertical bars intlicatc calculatcd ionization 2 energies. The numbers above the bars indicate orbital type: 1 = n,, 2 = XB, 3 :Xe, and 4 = n-oxamides. Inspection of the Scheme shows that the gross structure of the reorbital of the s-trans-conformer is similar to that of the re orbital of the s-czs-conformer and vice versa. The approach described above is based on the assumption that variations in ionization energies in a series of similar molecules are reflected in the variations of MO energies.This justifies the use of the same notation for orbital and ionization energies. It is interesting to note that the differences between the MO energies for s-trans-oxamide (Table 2), calculated by the lsquo;split-shell rsquo; basis set, and the corresponding experimental ionization energies (Table 3) are very similar to those obtained by Roos et al. in their studies on azabenzenes l7 and anhydrides,le using a similar basis set. This implies, as discussed by Roos et a2.,l6 that empirical corrections ob- tained froni studies on one class of molecules may be used for other classes as well, provided that the calculations are being done with basis sets of approximately the same quality .All calculations in this work, except for compound (11), were done for tlie N-unsubstituted parent compounds. The MO energies were converted to ionization energies BS described above. N-Methylation effects were taken into account using the N-methyl shifts determined by McGlynn et al. (n+ -0.47, n --0.52, xg, -0.97. and 7cQ -0.88 eV), with nB/xe inversion euro;or s-cis-conipounds.3 rsquo;These shifts have proved to be reinarkably constant and additive for aniides and oxaniides.l Experimental and calculated ionization energies are summarized in Table 3.Experi-mental p.e. spectra in the 8--12 eV region are shown in Figure 2 with the calculated ionization energies and their assignments indicated by vertical bars. The p.e. spectra of most of tlie compounds studied show extensively overlapped bands. The spectrum of (Vs) is however quite well resolved and inay be used as a test case for the approach described above. The calculated ioniz- ation energies agree very well with the experimental ones, especially for the three lowest ionization events (see Figure 2d and Table 3). N-Methylation shifts all ionizations towards lower energies, but the ne and re bands are more sensitive to the alkylation effect than the n+ and n-bands.3 This results in a strong overlap of the n f,re,and re bands in the p.e.spectra of (Vb) (Figure 2b). It is clearly a great advantage to be able to study the N-unsubstituted com-pounds of this class of molecules, since the p.e. spectra are inore resolved with less N-substitution. However, all attempts to synthesize compounds (IVa) and (VIa) have euro;ailed so far.14 The calculatetl ionization energies for these compounds, as given in Table 3, thus constitute a pre-diction of their p.e. spectra. DISCUSSION As mentioned in the introduction, the order of MO energies does not need to be the same as the order of ionization energies. For s-tram-oxamide (I) the MO order is re,n+, x8, and n-(Table 2), while the experi- mental order of ionization energies is nt, ne, xe, and 12-(Table 3).In this case, however, the three highest occupied MOs are nearly degenerate. The case of (Va) is more clear cut. For this molecule, the calculated MO order is xe, n,, n@,and n-, with a significant separation between re and n,/x@. Due to differences in correction terms for orbitals of x-and 0-type (see above), the order of the re and n+ processes is reversed in the experimental p.e. spectra. Great caution should thus be exercised if conclusions about the order of MO le~amp; are to be drawn from p.e. spectra. Since CNDO/S calculations repro- duce p.e. spectra very well, the same caution applies to conclusions drawn from such calculations. N-MethyZ Shijts.-The observed N-methyl shifts for oxamides are close to 0.5 eV for n, and .iz-and 0.9-1.0 eV for x@ and This difference is expected, since the 12 orbitals are largely localized on oxygen, with small amplitudes on nitrogen. In contrast, the xe/xO orbitals have large amplitudes on nitrogen.CNDO/S calcul- ations on oxamides give a fair account of the N-methyl effe~t.~However, the nb irtitio calculations employed in the present work show almost no shift of the n orbitals when calculations on oxamide and NN'-dimethyloxamide are compared. The energies of the namp;ce levels is, on the other hand, increased by ca. 0.5 eV (Table 2). These calculated N-methyl shifts are thus ca. 0.5 eV smaller than the effects observed on ionization energies in p.e. spectra. Although these differences between calculated and experimental values may to some extent be due to the limited basis set used in the calculations, they more probably imply that the effects of N-methyl substitution on p.e.spectra of oxamides (and other classes of com-pounds as well) are a composite of changes in MO energies and changes in reorganization and correlation energies. Conformation of Oxamides and its In.uence on Y.e. Spectra .-The p.e. spectra of s-trarts-NAr'-dimet hyl- oxamide (11) and its cyclic s-cis-analogue (Va) show significant differences (Table 3). Although the order of ionization events (n+,re,x@,and n-) is the same for the two molecules, the ionization energies in the 9-12 eV region are increased by 0.4-0.6 eV on going from (11)to (Va). These differences are not those expected for a conformational effect on going from s-trans to s-cis.CNDO/S calculations indicate small conformational effects, 0.1eV, on the nt and x8 levels. The xa and n-levels are somewhat more affected, showing shifts of +0.2 and -0.3 eV respectively.8 CND0/2 Calculations give exactly the same results.37 Ab irtitio calculations give somewhat larger differences (see Figure 1 and Table 2), but again the n, and ne MOs are less influenced by the s-trans-s-cis conversion than are the re and n-levels. Also in this case the energy changes of the two latter MOs have different signs. Calculations of ionization energies of s-trans (11) and s-cisoid (11) 20" twisted as in (Va), including reorganiz- ation and correlation terms, result in the following shifts on the ionization energies on going from s-trans to s-cisoid (20'): ?zt +O.Ol, riff, -0.27, xe +0.70, n--0.4 eV.It is clear that the observed differences in the spectra of (11)and (Va) cannot solely be interpreted in terms of effects due to conformational changes. The observed shifts must be a composite of a conformational effect and an effect due to differences between two N-methyl groups and an ethylene bridge. This effect, however, is not to be found in the energies of the PI0 levels. The st+/n-and the xcf,levels in (11) and (Va) have very similar energies (Table 2). The re 1110 lies cn. 0.5 eV higher in (Va) than in (11) depending on the CH,-CH, antibonding character of this orbital in (Va). Instead, differences in the reorganization and correlation term iieccssary to con- vert MO energies to ionization energies must be invoked.It should be noted that the procedure used in the present work very satisfactorily reproduces the p.e. spectrum of (Va). This implies that the correction terms evaluated from p.e. spectra and ab initio calculations of oxamide (1) are also valid for the cyclic s-cisoid-oxaniides, that is no extra terms due to the alkyl bridges need be included. As discussed above this is not the case for MO levels if methyl groups are included in the MO calculations. This analysis indicates a serious problem in studies of un-stable or low population conformers by p.e. spectro-scopy. A common technique for such studies is the pre- paration of model compounds in which the desired geo- metrical or conforniational properties are introduced by rigidly fixing part of the molecule by alkyl bridges.This may lead to a complete masking of the desired effect, by the influence of the bridging alkyl part, as in the case described above. In such cases conclusions drawn from p.e. spectra may be totally misleading. Ring Size and Ionization Energies.-Compounds (IVb) -(VIb) form a series of molecules with increasing twist around the intercarbonyl bond. Compound (IVb) is probably planar or close to planar. Models and ana- logous ring compounds suggest that the twisting in (Vb) and (VIb) is ca. 20 and GO", respectively. As indicated by the assignments in Figure 2, the five-membered ring com- pound (IVb) is quite different from the other two and will therefore be dealt with separately.The p.e. spectra of (Vb) and (VIb) are quite similar (Figures 2b and c), the main difference being a small but sigi~ $.ant shift of the low energy band towards lower a ionization energies as the ring size increases. This may be due to a corresponding increase in the intercarbonyl dihedral angle and/or to differences in the lengths of the alkyl parts of the ring systems. Ab iizitio calculations including the correction terms described above nicely reproduce the difference in p.e. spectra between (Vb) and (Vlb) (Table 3) but the picture that emerges is not a simple one. The three ionization events making up the low energy band all qualitatively behave as expected for an increasing twist of tlte oxamide unit.According to calculations summarized in Figure 1, the energies of the n+ and re levels increase on going from a 20 to a 60deg; twist. A small clecrease in the energy of the re level is also expected. These changes show up somewhat en-larged in the calculations of ionization energies of (Vb) and (VIb). However, according to calculations on oxamide (Figure l), the n-level should exhibit a de-crease in energy with increasing twist. Calculations on (Vb) and (VIb) show an increase in the energy of the n-level. It is thus clear that differences in the inductive effects of the alkyl parts of the ring system influence the p.e. spectrum to some extent. This may also be seen in the calculated MO energies of (Va) and (VIa) (Table 2).The differences in energies, especially for the n+ and rB levels, are about twice as large as tlie corresponding differences between 20 and 60" twisted s-cis-oxaniide. The conclusion is that tlie shifts observed in the p.e. spectra of (Vb) and (VIb) are about equally due to conformational changes and to differences in alkyl inductive effects. The p.e. spectrum of tetra- methyloxaniicle (111) shows a further shift of the low- energy band towards lower ionization energies than does (VIb). This shift is also reproduced by the calculations (Table 3), using the MO energies for 60"twisted oxamide anel adding corrections for methyl groups. Since identical geometries and dihedral angles were employed in the calculations of ionization energies for (111) and (VIb), the observed shifts must be interpreted as a dif- ferential influence of the N-alkyl substituents.NN'-Imidazolidine-4,5-dione (IVb) .-This cornpound may be viewed as a model for a planar s-cis-oxaniide unit. The geometry is quite different from that of oxamide itself (Table l),but the main differences are to be found in the bond angles, and these differences, as shown above, are not expected to influence the p.e. spectrum to a significant degree. In contrast to the other oxaniicle derivatives studied in this work, the low-energy ioniz- ation band is calculated to consist of only two ionization energies, 7te and gz+, with the re event of considerably higher energy.This conclusion is supported by tlie analysis of the p.e. spectrum of NA"-dimethylimidazo- .~lidine-2,4,5-trione (VII) by McGlynn ct ~ 1 For this Owo Me-NKN-Me0 (YIT) related compound, the first band in the p.e. spectrum is also due to the re and l.t-ionization processes, witli thc reprocess ca. 0.7 eV higher in energy. Although the p.e. spectra of (1Vb)-(VIb) are quite similar, the assignments of the two bands in the 8-12 eV region sliow important differences. The reason for the unexpectedly high ionization energy of tlie re process lies in the nodal pro- perties of the corresponding hfO. This orbital possesses a node at the methylene group making it quite insrn- J.C.S. Perkin I1 sitive to changes at this position.The ionization energy, 10.5 eV, is thus close to that observed for (VII) (10.97 eV).3 Co.lzclasio?zs.-Ionization energies in p.e. spectra of cyclic s-cisoid-oxamides are remarkably well accounted for by MO energies from ab initio calculations, empiric- ally corrected for reorganization and correlation effects. The influence of the alkyl parts of the cyclic ring systems necessary to force the oxamide unit into s-cisoid-con- formations seriously masks the effects of conformational change. A qualitative picture of the degree of twist of the oxamide unit in different ring compounds may be obtained, but differential effects due to different lengths of cyclic alkyl chains prohibits a more quantitative analysis. 'L'liis work was supported by the Swedish Natural Science Research Council and by the Royal Physiographic Society of Luncl.bsol;bsol;:e are grateful to Lektor L. Henriksen, The H. C. Oersted Institute, Copenhagen, for help with the re-cording of the photoelectron spectra, to Ur. G. liarlstrijiri for computational assistance, and to I)r. 13. E. Carter for correcting the Englisli. 0/49S Rrcrizied, 1st April, l980j REFERENCES J. .L. Meeks, H. J. Maria, P. Brint, and S. P. McGlynn, Chem. Rev., 1975, 75, 603. J. L. Mecks, J. F. Amett, D. B. Larson, and S. P. McGlynn,mer. Chem. Soc., 1975, 97, 3905.J' J. L. Meeks and S. P. McGlynn, J. Amer. Chem. Soc., 1975,97, 5079. z). Dougherty, P. Rrint, and S. P. McGlynn, J. Amel.. Cliewz. SOC., 1978, 100,5597. D. 0.Cowan, K. Gleiter, J. A. Hashmall, E. Heilbronner, and V. Hornung, Angew. Chem., 1971, 83, 405. P. Schang, R. Gleiter, and A. Kieker, Ber. Bunsengesellschaft Pltys. Chem., 1978, 82, 629. J. F. Arnett, G. Newkome, W. L. Mattice, and S. P. McGlynn,J. AWW. Chem. Soc., 1974, 96, 4385. c, D. B. Larson and S. P. McGlynn, J. Mol. Spectroscopy, 1973. 47, 469. K.E. Carter and J. Sandstrom, J. Phys. Chem., 1972,76, 642. lo E. M. Ayerst and J. R. C. Duke, Acla Cryst., 1954, 7, 588. l1 T. H. Siddall and M. L. Good, J. Inoq. Nuclear Chenz., 1979,86, 139. l3 W. E. Stewart and T. H. Siddall, Chem. Rev., 1970, 70, 517. l3 H. Lumbroso, cited in ref. 9. l4 K. Isaksson and T. Liljefors, to be published. l5 R. Gleiter and J. Spanget-Larsen, Topics Current Cli.em., 1979, 86, 139. lo M. lllmemark, J. E. Backvall, C. Moberg, B. kkermark, L. iisbrink, and B. Roos, Tetrahedron, 1974, 30, 2503. l7 J. Almlof, B. Roos, U. Wahlgren, and H. Johansen, J. Elrctron Spectroscopy Related Phenomena, 1973, 2, 51. 1Persson and J. Sandstrom, Acta Chem. Scand., 1964, 18, 1059. 19 J. Rlmldf, USIP Report 72-09, University of Stockholm, 1972. 20 L. Gianolio, R.Pavani, and E. Clementi, Gazzetta, 1978, 108, 183. 21 S. Huzinaga, .I. Phys. Cliem., 1965, 42, 1293. 25 T. Ottersen, ,4cta Chrm. Scnnd., 1975, A29, 939. z3 $1. Kitano and K. Kuchitsu, Bull. Cliem. Soc. Japan, 1973, 46, 3048. z4 K. Kuchitsu, T. Fukuyamn, and Y. Morino, J. Mol. Strzu-turr, 1967---1968,1,463. 2 D. 1.). Danielsson and I. Hedberg, J. Anzer. Cheni. Soc., 1979, 101,3730. 26 A. R. H. Cole, I'. S. Li, and J. 13. Durig. J. Alol. Specirosropy,1!)76, 61,346. p7 R.Isaksson, unpublished results.