1384
J . Phys. Chem. 1985,89, 1384-1387
Ionlc States of Nltrosyi Cyanide 0. Walter, L. S. Cederbaum,* Lehrstuhl fur Theoretische Chemie, Institut fur Physikalische Chemie, Universitat Heidelberg, 0-6900 Heidelberg, West Germany
B. Solouki, P. Rosmus, and H.Bock Fachbereich Chemie der Vniuersitdt Frankfurt, 0-6000 FrankfurtlM, West Germany (Received: March 27, 1984)
The valence-shell ionization spectrum of ONCN is studied by He I photoelectron spectroscopy and by many-body Green’s function calculations. The experimental and theoretical ionization potentials are in excellent agreement allowing for a univocal assignment of the cationic doublet states of ONCN’. Previous assignments of the PE spectrum of ONCN are critically analyzed. The many-body calculations predict the radical anion ONCN- to be bound with a vertical electron affinity of 1.3 eV.
Introduction The ionization of nitrosyl cyanide, ONCN, has been documented by photoelectron (PE) spectroscopy using He I radiation as a photon source.’J A characteristic feature of the PE spectrum which is common to molecules containing CN groups is the occurrence of several closelying ionization bands. A first tentative assignment of the corresponding doublet states of ONCN’ based on experimental considerations and qualitative arguments has been given by Jonkers et al.’ This assignment, however, substantially deviates from the results of a Hartree-Fock-Slater (HFS) calculation performed by the same authors. Generally, a discussion of a PE spectrum is restricted when based on the assumption of a oneto-one correspondence between doubly occupied orbitals and bands in the PE spectrum. As is well-known, the inner-valence regime of the ionization spectrum is often affected by strong correlation effects. The origin of these final state correlation effects is the near degeneracy of the inner-valence single-hole configurations with two-hole-one-particle (2hlp) and higher configurations. The interaction between these two kinds of configurations leads to a redistribution of the intensity associated with single-hole configurations over several more or less closely spaced states. In particular, a distinction between ionic main states and satellite states is often no longer possible and one encounters a breakdown of the molecular orbital picture of i ~ n i z a t i o n . Due ~ to the low symmetry of O N C N and the finding that it possesses a positive electron affinity (see below) such many-body phenomena can be expected to occur already a t relatively low binding energies. A first attempt to include correlation effects has been made by Kimura et aL2 These authors performed configuration interaction (CI) calculations on ONCN and ONCN+. The results of these calculations indicate that already above 14-eV binding energy higher excited ionic states come into play. However, due to the very limited CI procedure used by Kimura et al. the calculations account only for a small portion of the electron correlation and do not yield a univocal interpretation of the PE spectrum. In the present communication we present a new PE spectrum of O N C N recorded by using a “spectrum stripping” procedure together with the results of a Green’s function calculation. Based on this calculation an assignment of the cationic states of ONCN+ is given and the assignments of ref 1 and 2 are critically analyzed. In addition, since the one-particle Green’s function automatically (1) G. Jonkers, R. Mooyman, and C. A. de Lange, Chem. Phys., 57, 97 (1981). (2) K. Kimura, S. Katsumata, Y. Achiba, T.Yamazaki, and S. Iwata,
‘Handbook of He I Photoelectron Spectra of Fundamental Organic Molecules“, Halsted Press, New York, 1981. (3) See,e.g., L. S. Cederbaum, W. Domcke, J. Schirmer, and W. von Niessen, Phys. Scri., 21, 481 (1980).
0022-3654/85/2089- 1384$01.50/0
also provides information on the electronic states of the radical anion, the electronic stability of the ONCN- molecule is discussed as well.
Experimental Section Nitrosyl cyanide (ONCN) used in the present investigations was prepared from nitrosyl chloride (ONCl) and dry silver cyanide (AgCN) according to the method described in the l i t e r a t ~ r e . ~ , ~ The deep-blue gaseous ONCN was purified by repeated distillation at temperatures between 140 and 170 K. In several experiments, however, we were not successful in preparing pure ONCN, which easily decomposes into more stable products like (CN),, NO2, NO, N2, etc., even below 170 K. Since the compound exploded in the trap at about this temperature during the measurement, we did not try further to obtain the pure compound. Instead, a PE spectrum contaminated by (CN), was recorded on a Leybold-Heraeus UPG 200 PE spectrometer connected with a PDP 11/40 computer and the PE spectrum of pure (CN)z was then subtracted from the spectrum of the mixture ONCN/(CN),. Both PE spectra are shown in Figure 1. Even though the ONCN spectrum obtained in this way is improved in its high-energy part relative to the PE spectra of Jonker et al.’ and comparable to the one recorded by Kimura et the shapes of the bands between 13 and 19 eV are somewhat uncertain. This is mainly due to small changes in the resolution of the PE apparatus in consecutive experiments caused by contamination of the analyzer. Especially, it is difficult to decide whether the small band at about 15.1 eV is due to compounds other than ONCN or whether it belongs to the PE spectrum of this molecule. Being obscured by contamination this peak has not been observed before. Computational Details The many-body calculations have been performed with the so-called algebraic diagrammatic construction (ADC) scheme’ for the one-particle Green’s function. The ADC(n) scheme constitutes an approximation which represents an infinite partial summation of the perturbation expansion for the self-energy being exact to nth order in the electronic interaction. The resulting mathematical procedures require the solution of Hermitean eigenvalue problems in restricted spaces of physical excitations. In the present case, we have used the ADC(3) approximation (also called the extended two-particle-hole Tamm-Dancoff approxi(4) R. Dickinson, G. W. Kirby, J. G. Sweeny, and J. K. Tyler, J. Chem. SOC.,Faraday Trans. 2 , 1 4 , 1393 (1978). ( 5 ) B. Bak and 0. J. Nielsen, J . Labelled Compd. Radiopharm., 15, 715 (1978). ( 6 ) 0. Walter and J. Schirmer, J. Phys. B, 14, 3805 (1981). (7) J. Schirmer, L. S. Cederbaum, and 0. Walter, Phys. Reu. A, 28, 1237 (1983).
0 1985 American Chemical Society
Ionic States of Nitrosyl Cyanide
The Journal of Physical Chemistry, Vol. 89, No. 8, 1985 1385
TABLE I: Vertical Ionization Energies I,, (eV) and Spectroscopic Factors txD(")lzfor ONCNa ADC(3)b orbital p
HF -eD
In
8a' 7a' 2a"
13.38 14.36 13.94
6a'
16.89
11.35 13.67 13.86 13.98 14.56 14.92
0.75e 0.6gf 0.88 0.001 0.408 0.448
16.00
0.02h
IxD(")lz
expt
HFSc I ,
CId I ,
InC
11.3 13.8 14.6
11.70 14.10 14.24 14.37 15.46 16.52
11.50 13.56 13.79
la"
18.26
16.78 18.84
0.37 0.45
17.92
5a'
20.07
16.88 17.37 18.55
0.08 0.55 0.1 3
17.50
4a'
22.42
19.74 20.06 20.36
0.21 0.20 0.19
leading configuration
In
11.48 13.58 13.75
(8a'j-' ; state (7a')-' ;A, state (2a")-' ; B state (8a')-'3a" (6a')-' (7a')-' (2a")-' 3a" (8a')-'(2a'')-l3a" (7a')-'(2a")-'3a" (8a')-'(2a1')-'3a''
:::: {
14.45
t
(5a')-' (7a')-'(la'')-' 3a" (8a')-'(la")-'3a" (7a')-'(2a")-'3a" (4a')-' (la")"(na')-'3a" (n = 6, 7, 8) (n = 5 , 6) (2a")"(na')-'3a"
a States with lxp(n)12 < 0.05 are omitted except for the first satellite state in each symmetry, Present results. Reference 1. Reference 2. ' This state has an intensity coefficient I X , ~ ~ ( " ) I ~ = 0.12 for the orbital 7a' ( x , ~ ~ ' "=) -0.35). This state has an intensity coefficient lx8af(n)12= 0.14 for the orbital 8a' (xsa'(") = +0.37). 8 The spectral intensities of these two states are strongly basis set dependent (see text). The accumulated spectral intensity of this state is 0.26 (see text).
m a t i ~ n ) . ~Within .~ this approximation the ionic main states, Le., states that are characterized by a single-hole configuration, are treated consistently through third order in the electronic interaction. The discrepancy between ADC(3) ionization potentials and experimental values is typically 50.2 eV, which at present can be considered as the accuracy of good "state of the art" calculations. The treatment of higher excited 2hlp ionic states is exact through first order which allows for an a t least qualitative description of the corresponding ionization phenomena. A detailed description of the ADC(3) approach and the numerical procedures involved is given e l ~ e w h e r e ~and - ~ need not be repeated here. Neutral-molecule ground-state Hartree-Fock (HF) orbitals were used as input data for the Green's function calculation. These HF one-particle data have been obtained from LCAO S C F calculations using the experimental equilibrium geometry4 of ONCN. The basis set is built of Cartesian Gaussian functions. For each atom 0, N, and C the basis set consists of nine s-type functions and four p-type functions contracted to four s-type functions and two p-type functions. The exponential parameters and contraction coefficients have been taken from the work of Dunning.'O This basis set of double P quality has been augmented by one d-type function on each atom (ad(N) = 0.75, q(0)= 0.8, ad(C) = 0.6). The total S C F energy obtained from this basis set is -221.524 au. For the many-body calculations all ten occupied valence orbitals together with the 27 unoccupied orbitals lowest in energy (emax = 46 eV) have been kept. As a first step the self-energy part is computed and subsequently the Dyson equation is setup and solved. The matrices involved in the self-energy part have dimensions 1512 for the ( N - 1)-particle 2hlp configuration space and 3888 for the ( N 1)-particle 2plh space (2A' symmetry). To keep the numerical effort low, a matrix of 400 selected configurations for each space was diagonalized. The remaining configurations were included in the Dyson equation by taking the corresponding first-order approximations. In this way the Dyson equation was solved for the full space of ( N 1)-particle configurations. The static part of the self-energy was determined via the free oneparticle Green's function. The technical details of the computational procedure are described at length in ref 8. The vertical ionization potentials obtained from the ADC( 3) calculations are listed in Table I together with experimental values.
t
b'
+
(8) W. von Niessen, J. Schirmer, and L. S . Cederbaum, Compur. Phys. Rep., 1, 57 (1984). (9) J. Schirmer and 0. Walter, Chem. Phys., 78, 201 (1983). (10) T. H. Dunning, J . Chem. Phys., 53, 2823 (1970).
t
I
I
.
11 13 15 17 IE(eV) Figure 1. He I spectrum of ONCN. Spectrum b is a PE spectrum of ONCN contaiminated by (CN),. The spectrum of pure ONCN (a) is obtained from (b) by subtracting the PE spectrum of (CN),.
For each ionization potential I,, the spectroscopic factor Ix (")I2 is also given. Here, the index p indicates the H F orbital &om which the corresponding line in the spectrum has "borrowed" its intensity (for more details see ref 3 and references therein). For comparison, the results of Jonkers et al.' and Kimura et ala2are included. The ADC(3) "ionization spectrum" of ONCN, Le., spectroscopic factors vs. ionization potentials, is displayed in Figure 2.
1386 The Journal of Physical Chemistry, Vol. 89,No. 8, 19‘85 1
2 3 4
5
x
-
1.0-
6 7
--
ea1
---
7al 2a’ 6al la’ 5a’ 4a’
12
13
ONCN
0.2 0.1
11
14
15
16
17
18
19
20
21
22
B l n d l n g Energy ( a V )
Figure 2. The ionization spectrum of ONCN up to 22-eV binding energy calculated by using the ADC(3) approximation one-particle Green’s function. Only states with spectroscopic factors Ixp(n)l* 2 0.01 have been included. The number above each line indicates the p orbital from which the spectral intensity originates.
Discussion For the following discussion we divide the PE spectrum of O N C N into an outer-valence part below 14-eV binding energy and an inner-valence regime above 14-eV binding energy. Let us first focus on the former region. The characteristic of this part is the one-to-one correspondence between doubly occupied orbitals and bands in the spectrum. Here, one ficds three ionization phenomena which are assigned as X(2A’), A(2A’), and 8(2Arr) states of ONCN+ corresponding to a removal of an electron from the 8a’, 7a’, and 2a” orbital of the neutral molecule, respectively. (The numbering of the orbitals begins with the valence orbitals.) The nature of these three orbitals is N O in-plane a-antibonding, strongly perturbed C N in-plane ?r-bonding, and C N ?r-bonding, respectively, where the nitrogen atom in C N is the terminal one. As can be seen from Table I, the ADC(3) results and the experimental ionization potentials are in excellent agreement. On the other hand, application of Koopmans’ theorem leads to a different sequence of the ionic states, the states corresponding to the 7a’ and 2a” orbitals being interchanged. This reordering with respect to the HF level arises from the fact that many-body effects contribute differently to the a’ (u-type) and a’’ (a-type) orbitals. Koopmans’ defects are relatively large for the 8a’ (2 eV) and 7a’ orbitals (0.7 eV), whereas the corresponding shift for the 2a” orbital is almost zero (see Table I). Such a reordering is often encountered for planar molecules with low-lying unoccupied orbitals which are localized in space as, e.g., for N2 and F211or “0, FNO, and O3.I2 For those molecules a “rule of can be applied for predicting whether a change in the ordering of the ionization potentials due to many-body effects is to be expected. When the (localized) unoccupied orbital lowest in energy is of a” symmetry (a’ symmetry) a large shift to lower energy is to be expected for the a’-type (a”-type) outer-valence orbitals. Hence, if an a‘-type occupied orbital is situated close, but on the lower binding energy side of a a”-type occupied orbital and the low-lying unoccupied orbital is of a’’ symmetry, a reordering of the corresponding ionization potentials is likely to happen. In the case of ONCN the unoccupied orbital lowest in energy is the bound 3a” < 0). In addition, the occupied 7a’ orbital is situated orbital only 0.4 eV above the 2a” orbital (see Table I). If the above rule is applied to this situation the interchange of the 7a’ and 2a” orbitals can easily be anticipated. The calculations icdicate that the Green’s function is strongly nondiagonal for the X(2A‘) ground state and the A(2A‘) state of ONCN’. In both cases spectral intensity is shared by the 8a’ as well as by the 7a’ orbital (see Table I). For the A(2A’) state for example, the spectroscopic factor associated with the 7a’ orbital (11) L. S . Cederbaum, Chem. Phys. Lett., 25, 562 (1974). (12) D. P. Chong, F. G. Herring, and D. McWilliams, J . Electron Spectrosc. Relat. Phenom., I, 445 (1975).
Walter et al. is as small as 0.69 whereas the factor associated with the 8a’ orbital amounts to 0.14. This effect is caused by a rather strong mixing (16%) of the 8a’ single-hole configuration with the 7a’ single-hole configuration. Such a marked hole mixing in ionic main states is a rather surprising result in view of the relatively large energy gap between the two ionic states associated with the removal of electrons from the 7a’ and 8a’ orbitals. Usually, an ionic main state is well characterized by one single-hole configuration only. This is due to the fact that for HF single-hole configurations the direct coupling matrix element vanishes (apart from a tiny contribution arising from nondiagonal contributions (0.06 eV) of the static self-energy part). A hole mixing is thus mediated by 2hlp and higher excited ionic configurations. These configurations induce an effective hole mixing especially for energetically close-lying states (of the same symmetry).13 The two ionic main states originating from the 8a’ and 7a’ orbitals are well separated in energy thus underlining the strong effect of 2hlp configurations. Other rather prominent examples for this effect are the two molecules propynol and propiolic acidI3 where the corresponding single-hole configurations have almost equal weights. It should be noted that the one-particle picture of ionization is still well applicable in these cases. The corresponding quasi-holes, however, no longer derive from the H F orbitals. The assignment of the first three bands in the PE spectrum based on the ADC(3) results is confirmed by two other ab initio calculations. Jonkers et al.’ performed restricted HFS calculations with a triple { basis set (without polarization functions) and obtained the same ordering for the first three ionization potentials of ONCN. Agreement with experiment is g+ for the two states of 2Ar symmetry, whereas the energy for the B(2A”) state is too high by 0.8 eV. Kimura et ale2calculated the energies via a CI technique. However, due to severe technical approximations as, for instance, a restricted basis set of less than double { quality without polarization functions and a very limited configuration space, these CI results can only be used for a qualitative discussion. Here, the deviations from the experimental values amount up to 0.5 eV. Another shortcoming of this CI calculation should be mentioned. Since energies only are calculated, no assertions can be made about the spectral intensities of the corresponding ionic states in a PE spectrum. This fact is particularly important when satellite states come into play. A typical example is the state of 2A‘r symmetry characterized by the configuration 3a”. Kimura et al. find a binding energy of 14.37 eV; the corresponding ADC(3) result is 13.98 eV. The spectroscopic factor calculated via the one-particle Green’s function is, however, less than 0.00 1, indicating that thii ionic state will not show up in the PE spectrum. In the communication of Jonkers et al.’ the authors propose a different assignment for the ionic states in the outer-valence region. They compare the PE spectrum of ONCN to those of the related molecules N3CN and OCNCN and by analogy arrange the orbitals in the order 8a’, 6a’, 7a’. However, apart from the fact that this assignment is contrary to all a b initio calculations which go beyond Koopmans’ theorem, their argument seems doubtful for another reason. The assignment of the cationic states of OCNCNI4 and N3CN’4s15is based on Koopmans’ theorem and for the latter molecule on a CNDO calculation. However, in the case of ONCN, Koopmans’ theorem applied to the results of a CNDO calculation’ completely fails to reproduce the PE spectrum and it thus seems doubtful whether the assignments made for OCNCN and N3CN are reliable. It cannot be excluded that a calculation which accounts for many-body effects may lead to a reinterpretation of the PE spectra of these two molecules as well. Here, a remark concerning the basis set used seems appropriate. Our underlying SCF calculations on ONCN differs from all other available calculations in that polarization functions are included in the basis set. To check on the importance of the latter functions we have repeated the ADC(3) calculation for the same basis set (13) W. von Niessen, G. Bieri, J. Schirmer, and L. S . Cederbaum, Chem. Phys. 65, 157 (1982). (14) D. C. Frost, H. W. Kroto, C. A. McDowell, and N. P. C. Westwood, J . Electron Spectrosc. Relat. Phenom. , 11, 147 (1977). (IS) B. Bak, P. Jansen, and H. Stafast, Chem. Phys. Lett., 35, 247 (1975).
Ionic States of Nitrosyl Cyanide but omitting the d-type functions. This calculation does not alter the assignment of the outer valence ionization potentials. The impact of the polarization functions is largest on the 7a’ orbital, shifting its ionization potential by 0.3 eV to higher energy. Interestingly, the change of basis affects the orbital energies more than the final results. Another effect worth mentioning in the outer valence region is due to the ionization of the 2a” orbital. When d-type functions are omitted the 2hlp satellite line characterized by the (8a’)-23a” configuration interchanges ordering with 3ts” main line characterized by the (2a9-I single-hole configuration (see Table I). The spectroscopic factor corresponding to this satellite line remains very small. In the energy region above 14-eV binding energy a completely different situation is found. Here, the spectral intensity for the ionization of the 6a’, la”, Sa‘, and 4a’ orbital is shared by several ionic states and it is thus no longer possible to discern an ionic ’main” state. This breakdown of the one-particle picture of ionization, which is a common phenomenon for the ionization of molecular inner-valence electron^,^ is caused by higher excited ionic configurations. The situation is best documented by Figure 2 and Table I where, however, only the most intense lines are shown. Of special interest is the ionization of the 6a’ orbital which is the nitrogen lone pair orbital. The 6a’ single-hole configuration = -16.89 eV) strongly interacts with a 2hlp configuration at -14.69 eV being mainly characterized as an excitation of the 2a” orbital into the empty 3a” orbital upon a hole in the 7a’ or 8a‘ orbital. As a consequence, one no longer finds an ionic 6a’ 2A’ main state but two closely spaced states a t 14.56 and 14.92 eV of almost equal spectral intensity. These two states, which are tentatively assigned to the two peaks at 14.4 and 15.1 eV in the experimental spectrum, are a mixture of all configurations involved, i.e., the 6a‘ single-hole configuration and the 2hlp configurations (7a’)-’ (2at’)”3a’’ and (8a’)”(2a”)-’”’’. The 15.1-eV peak in the experimental spectrum, Figure la, is much less intense than the strong peak at 14.4 eV. We assume that the apparent contradiction between theory and experiment concerning the relative intensity of these peaks is due to sensitivity of this intensity to changes in the basis set. To illustrate this point we have repeated the computation omitting the polarization functions in the basis set (of course, it would be more appropriate to enlarge the basis, but this would increase the numerical effort). Without polarization functions the two 2A’ states at 14.56 and 14.92 eV originating from the 6a’ orbital are shifted apart by 0.2 eV to give 14.47 and 15.02 eV, respectively. However, the distribution of spectral intensity of these two states is markedly changed: the state at lower binding energy now exhibits a spectroscopic factor of 0.72 whereas the spectroscopic factor of the state at higher energy is reduced to 0.13. The computed energies change only slightly with variation of the basis set. The ratio of spectroscopic factors of the two closely lying lines, on the other hand, is very sensitive to the size of the energy gap between them. The calculations thus support the assumption that the peak at 15.1 eV in the experimental spectrum is associated with the ionization of O N C N and is not an artifact caused by contamination of the analyzer. The Green’s function calculation yields a further state of ’A’ symmetry at 16.00 eV which is a mixture of the two configurations (7a’)-’( 2a””’”’’ and (8a’)-’( 2a”)-l3a”. The spectroscopic factor of this state with respect to the 6a’ orbital is rather small (0.02). It gains, however, spectral intensity from all valence orbitals of a’ symmetry. Summing up the amplitudes x?) of the spectroscopic factors and squaring the result, one finds an accumulated spectroscopic factor of 0.26, indicating that this state might show up in an experimental PE spectrum. In the energy region above 16 eV the experimental spectrum exhibits many structures superimposed on a broad band. In this region ionization phenomena are to be expected which derive from ionization out of the three orbitals la”, 5a‘, and 4a’. The la’’
-
The Journal of Physical Chemistry, Vol. 89, No. 8, 1985 1387 orbital is N O ?r-bonding whereas the 5a’ and 4a’ orbitals are mixtures of lone pair electrons on the oxygen and the adjacent nitrogen atoms. As Table I shows for all three orbitals no ionic main state can be discerned; the available spectral intensities are shared between the corresponding single-hole configurations and 2h l p configurations which predominantly involve an excitation into the unoccupied 3a” orbital. Here, a detailed comparison with the PE spectrum is not possible at the moment and more extensive experimental data are needed. As has been documented by many calculations, the ADC(3) reproduces the breakdown phenomenon for the ionization of inner-valence electrons in a semiquantitative manner; the detailed structure of the theoretical spectrum is often short of a quantitative agreement with experiment. Essentially, there are two reasons for this situation. The first deficiency arises from the use of a finite discrete basis set of one-particle states and is common to all finite basis set methods. The breakdown regime might be close to or even comprise the first double ionization threshold to which several Rydberg series of the cation converge. With a finite discrete basis set only the very first members of such 2hlp excited Rydberg states are described properly. One then obtains artificial states and thus an artificial spectral structure at higher excitation energies which can be viewed as an inadequate simulation of higher Rydberg and continuum states. This situation is reflected by a strong basis set dependence of the ADC(3) results in the innervalence region. To check on this point we compare the ADC(3) results using the basis sets with and without polarization functions. In the energy region shown in Figure 2 above 14 eV the two theoretical spectra differed by not more than 0.1 eV as far as the energies are concerned. A possible explanation for the good agreement of the ADC(3) energies with experiment in the breakdown regime is due to the electronic structure of ONCN: the dominant 2hlp configurations involved are based on an excitation into the unoccupied 3a” orbital. This orbital is bound and thus well described within a finite discrete basis set. A second shortcoming of our theoretical approach is inherent to the ADC(3) scheme itself. As has been mentioned the ionization potentials of the 2hlp excited ionic states are treated consistently only up to first order in the perturbation which certainly is not satisfactory. Moreover, some of the lowest 3h2p excitations may explicitly come into play in the inner-valence ionization spectrum. The explicit treatment of these higher excited configurations, however, is beyond the scope of the ADC(3) and is the subject of the more advanced ADC(4) scheme.’ Finally, we briefly address the question of the electronic stability of the negative ion ONCN-. Both fragments N O and C N possess a positive electron affinity and, therefore, it is interesting to see whether the radical anion of O N C N also has a bound electronic (ground) state. As has already been mentioned the one-particle HF energy of the 3a” orbital is negative (ejat, = -0.13 eV), indicating a bound ( N + 1)-particle state of ONCN already on the independent-particle level. Since the one-particle Green’s function yields the energies and spectral intensities of both ( N - 1)- and ( N + 1)-particle states, the ADC(3) approach can also be used for obtaining vertical electron affinities. Our present ADC(3) calculation predicts that the many-body effects lead to a stabilization of the ONCN- ground state corresponding to an electron attachment in the 3a” arbital. The vertical electron affinity is obtained as 1.28 eV which amounts to a shift of 1.1 eV with respect to the HF level. Since no experimental electron affinities for O N C N are available, a comparison with experiment has to be postponed.
Acknowledgment. Fruitful discussions with W. von Niessen and financial support by the Deutsche Forschungsgemeinschaft (L.S.C.) and Fond der Chemischen Industrie are gratefully acknowledged. Registry No. ONCN, 4343-68-4; ONCN’, 95274-29-6; ONCN-, 95342-89-5.
