J. Phys. Chem. 1993,97, 1121611220
11216
The NSCl/AgNCO Reaction: A He I Photoelectron, Photoionization Mass Spectroscopy, Mid-Infrared, and ab Initio Study of the Gaseous SzNzCO Molecule Richard H. d e h t , Louise Durham, Edward G. Livingstone, and Nicholas P. C. Westwood' Guelph- Waterloo Centre for Graduate Work in Chemistry, Department of Chemistry and Biochemistry, University of Guelph, Guelph, Ontario, Canada NIG 2Wl Received: July 26, 1993; In Final Form: August 20, 1993"
Attempted generation of the NSNCO molecule via an NSCl(g)/AgNCO(s) reaction leads to the formation of a gaseous product identified as SzNzCO. This molecule is characterized by H e I photoelectron spectroscopy employing a position sensitive detector, photoionization (He I, HL,) mass spectroscopy, and mid-infrared spectroscopy, as a planar five-membered ring with alternating S and N atoms, and an exocyclic carbonyl group. Previous synthetic work has identified this species as a yellow crystalline solid; the present study is aimed at elucidating gas-phase information. The electronic and geometric structures obtained from semiempirical (MNDO, PM3), and ab initio (6-3 1G*) calculations establish the cyclic structure, with the observed gas-phase ionization potentials and vibrational frequencies in close correspondence to theory.
Introduction The hitherto unidentified molecule NSNCO, thionitrosyl isocyanate,is a possible target species for study by electronic and geometric structure methods and is a natural extension of our interest in the photoelectron and infrared spectroscopyof unstable isocyanates and other pseudohalides.1-3 The reaction of gaseous NSCl with solid AgNCO is an anticipated source of this species, whereby a labile C1 group could be replaced with an NCO group. The species NSNCO may be considered a form of pseudohalide, with an NS moiety replacing classic halogens, e.g.,X in XNCO,M or pseudohalogens, e.g.,NC in NCNC07-90r NCNCS.'O Clearly of interest would be the structure of NSNCO where the ambidentate nature of NCO could lead to either cyanate or isocyanate formulations, or the substituentNS could bond through either Nor S. Anadditionalcomplicatingfeatureisthe feasibility of further rearrangement to give molecules with CNO, SNO, or NSO groups." Despite the above comments regarding the intended target molecule NSNCO, design and experiment contradict in this particular case and lead to the observationthat reaction of NSCl(g) with AgNCO(s) leads not to NSNCO (or indeed any of its incarnations) but to the species S2N2C0, a previously known crystalline material never before characterized in the gas phase. This paper describes the formation of this five-membered heterocycle via the NSCI/AgNCO reaction, the gas-phase characterization of the synthesized molecule by ultraviolet photoelectron (PE), photoionization mass (PIMS), and infrared (IR) spectroscopies and an investigation of its electronic and geometric structure by semiempirical and ab initio methods.
Experimental Section PE spectra were recorded on a second generation spectrometer which employs a position sensitive detector.12 Kinetic energy (KE) selected photoelectrons are imaged onto a chevronned pair of 50- X 8-mm microchannel plates, which are then read out via a resistive anode, matched preamps, and timing circuitry involving zero-crossing detectors.13 This permits collection of windows of data in much less time than would be needed if each KE were Abstract published in Advance ACS Absrructs, October 1, 1993.
0022-365419312097-11216$04.00/0
scanned individually. In addition, improved pumping capabilities provide an operating pressure ((2-6) X 1 W Torr) an order of magnitude better than our previous spectrometers, giving less contamination, and the ability to work with lower vapor pressure molecules. The electron energy analyzer, mean radius 10.16 cm, is preceded by a four-element lens system14J5 which permits focusing of the electrons prior to entering the analyzer and adjustment of their KE to the desired transmission energy (TE). From the size of the analyzer and detector, windows of data are approximately 19%of the TE so that a 20-eV TE gives a window witha KEspreadof 3.8eV. Thus,spectracan bequicklyaquired in a few large windows. Nevertheless, since resolution tracks as TE, high resolution requires the use of many windows. Calibration was achieved using Ar and MeI. The operating resolution for these experiments was 45 meV. Overlapping windows of data were collected on a 286 PC and linearized, normalized, and assembled using LOTUS 123. Capability also exists for the collection of the concomitant ion with a quadrupole mass analyzer (Hiden Analytical, 320 amu), which has its conventional E1 source removed, ionization being provided by He1 (21.22 eV) or unfiltered HL,B,7 (10.2-12.7 eV) radiation. At present, PE and PIMS spectra cannot be collected simultaneouslybecause of competing polarity requirements for extraction of the photoejected electrons or ions. They can, however, be collected sequentially within seconds,and it is assumed that for a given PE spectrum the subsequent PIMS is of the same compound. Mid-infrared spectra were collected on a Nicolet 2OSXC interferometer equipped with either a variable path length White cell or a 20-cm single-pass cell. Both cells had KBr windows, giving a spectral range to 400 cm-1. The path length of the White cell was kept at 16 m. In either case the effluent from thereaction tube or sample (below) was pumped slowly through the cell using a rotary pump while maintaining the pressure between 50 and 500 mTorr. Spectra were collected at 2- or 0.5-cm-l resolution. Initial experiments carried out on a single-channel detector PE spectrometerindicated that S2NzC0was the primary product obtained by passing gaseous NSCl (generated by heating the solid cyclic trimer N3S3Cl3toca. 70 "C) over heated solid AgNCO. The optimum temperature for reaction was 125 OC. AgNCO was prepared by precipitation from aqueous KNCO and AgN0316 and dried under vacuum. It was loosely packed into a 9-mm-i.d. 0 1993 American Chemical Society
The NSCl/AgNCO Reaction
The Journal of Physical Chemistry, Vol. 97, No. 43, 1993 11217
X 200-mm glass tube attached directly to the inlet of either a PE/PIMS spectrometer or an IR cell. Following the identification of the molecule (below), a bona-fide sample was synthesized and investigated using the multidetector system with its enhanced sensitivity at low pressure. PE, PIMS, and IR of the synthesized SZNZCOwere obtained by analyzing the vapor over the solid. This material (5-oxo1,3X4,2,4-dithiadiazole)was prepared from N,N'-bis(trimethy1sily1)sulfur diimide and chlorocarbonylsulfenyl chloride,17using as solvent for the reaction Et20 instead of CHzC12. The crude product was recrystallized from ether and purified by sublimation at 30 OC/O.l Torr. The final product, large, translucent yellow, cubiccrystals, had a melting point of 40-41.5 OC (literature 40.5 OC).l*J9 The chlorocarbonylsulfenylchloride used in the reaction was prepared by hydrolyzing perchloromethyl mercaptan, CCl3SCl (Aldrich), with concentratedsulfuric acid and stoichiometric amounts of water and then collecting the distillate fraction at 98 OC.20
I
1
SN+
30
a
40 50 60 70
80 90
100 110 120 130 140 150
b
Computational Methods Ab initio calculations were performed at the 3-21G* and 6-3 1G* level using Gaussian 86.21 Frequency calculations were carried out on the stationary points to confirm that all structures were true minima (zero negative eigenvalues). Semiempirical geometry optimizationswere carried out for the NSNCO target molecule, several of its isomers, and SzNzCO and its isomers using MNDO and PM3 as implemented by MOPAC version 5.00.2zAll optimizationswere run with the PRECISE option as suggested by Stewart et al.23 Frequency calculations were run on these structures to characterizethe stationary points. Ab inirio calculations were performed on an IBM 308 1 at the University of Guelph; MOPAC was implemented on a 33-MHz 386 PC.
30
40 50 60 70
-
80 90 100 110 120 130 140 150
Figure 1. (a) He I and (b) HL.J,, photoionization mass spectra (amu) of SzNzCO.
ResultS In theinterestsofexpediency wesummarizehere thePE, PIMS, and IR results for the NSCl/AgNCO reaction and the evidence leading to the determination of the formation of a cyclic S ~ N Z C O species. We then go on to discuss in more detail the gas-phase spectroscopyand computationalstudy of the characterized S2N2CO molecule. PE spectra of the NSCl/AgNCO reaction lead to a gaseous product with distinct low-energy IPS at 10.0, 10.5, 11.8, 12.0, 13.3, and 13.8 eV (in some cases excess NSCl is 0bserved2~~2~); the form of the spectrum indicates an absence of NCO or C1 groups. Calculated (6-3 1G*,MNDO) IPSfor NSNCO, bearing in mind the limitations of Koopmans' theorem but taking into account our experience with other SN- and NCO-containing molecules,2.3.26 do not fit the observed IPS. Neither do the calculated vibrational frequencies concur with those obtained experimentally. The IR spectrum shows a strong band at 1742 cm-l, indicative not of an NCO group but of a > C = O group. We note in some cases, with AgNCO temperatures in excess of 150 OC, that some gaseous NCNCO is also formed, a known The product obtained by overheating AgNCO in uacu0.2~-~~ PIMS data are perhaps the most illuminating;with an IP of 9.95 eV, we can make use of both He I and unfiltered HL, radiation to determine the parent ion. In both cases there are peaks at m / z = 46 (s), 78 (s), 92 (vw), and 120 (s), and additionally, m / z = 32 occurs with He I; the high mass end is favored with the lower energy light source as we have demonstrated previously for the six-membered SN heterocycles, S3N3. and S4Nz.30.31 The strong peaks show a weak isotope (ca. 5%) 2 amu higher, suggestingthe presence of sulfur. If the parent ion is taken to be m / z = 120, then the peak at m / z = 78 (SzN+) represents a loss of 42 amu (NCO). A possible formula for a parent ion of m / z = 120 that could produce this loss fragment is SZNZCO,with the IR result
T 2000
1600
1200 800 WAVENUMBER
4 D
Figure 2. FTIR spectrum of S2N2CO (2200-400 cm-'); scaled ob inffio values (6-31G') are shown with the calculated intensities. Inset shows detail of band at 1742 em-'.
pointing to a > C 4 group. A molecule of this formula, 5-OXO1,3X4,2,4-dithiadiazole,is known in the solid state and has been characteri~edl~-~~-~~-~~ as a cyclic five-membered ring containing an exocycliccarbonyloxygen. It has a similar mass spectrum'7J* and a very strong solid-state IR band at 1727 ~ m - 1 ~ ~ ~ 9 ~ ~ Figure la,b shows the He I and HL, photoionization mass spectra, Figure 2 shows the FTIR spectrum from 2200 to 400 cm-l, and Figure 3 presents the He I PE spectrum of pure SIN*CO, the subject of the following discussion.
11218 The Journal of Physical Chemistry, Vol. 97, No. 43, 1993
deLaat et al.
TABLE I: Observed and Calculated Vibrational Frequencies
9
13
11
15
19
17
IONIZATION ENERGY (eV) F l p e 3. He I photoelcctron spectrum of SzN2CO.
0 II
91.21 9
Crystal Structure
Valence Structure
H
H
1.197 121.9 -129.8
1.181 121*:--129.0 1.825/
-
1.852) 9 . 3 8 2 b s -&118.7N8 1.65h,,, b
b
-9S.2114.6. a
1A51
YN=l1
Gl 1 . % 2
3-21G* 6-31G’ Figure 4. Crystal structure, valence structure, and calculated semiempirical (MNDO, PM3)and ab initio structun (3-21G*,6-31G’) for
SzNzCO.
Discussion Calculated Structure of Caseous SNzCO. The PIMS data (Figure 1) argue against open carbonyl compounds of the type 0
/c\ S I N
0
II
I
I
/c\N N b S
N f s
b
S
N f s
s--N
on the basis that such structures cannot lead to a pronounced S2N+ fragment. MNDO calculations on various optimized conformations (all nonplanar) of these structures show much inferior energies to the proposed five-membered ring structure. Figure 4 shows the crystal structureg2of cyclic S2N2CO compared to gas-phase C, structures obtained from MNDO, PM3, and ab initio (3-21G* and 6-3 1G*) calculations. The semiempirical
(cm-1) and Assignments for W 2 C O calculated& observed 6-31G* intensitv assinnt and approx descriptiond 1742 1787(A’) 774 V I CO str 1068 ll06(A’) 173 v2 ring str 95W l006(A’) 43 v3 ring bend 926e 993 (A’) 63 v4 C-N str 780 772(A’) 78 VI ring str 34 YIO NCO oop bend/CO wag 675? 662 (A”) NO 616(A‘) 13 V6 ring bend 607(A’) 2 v7 ring str NO 525(A’) 4 vg NC(0)S bend NO 424(A”) 20 v11 ring oop bend NO 386(A‘) 6 v9 CO in-plane bend NO 191 (A”) 1 v12 ring oop bend NO “The values shown are for the best energy C, structure (ET = -1016.549 159 hartrccs) and are 0.9 X 6-31G*calculated values. Symmetries are in parentheses. b Within C, symmetry all in-plane vibrations are A’and all out-of-planevibrations are A”. The u principal axis is approximately through the molecule following the CO bond. Intensity in km mol-’. doop = out-of-plane; NO = not observed. e Overlapping bands. methods give inferior geometries to those obtained by ab inirio methods. MNDO furnishes ring bond lengths (except for C-N) that are too short by 0.01-0.04 A,whereas PM3 gives ring bond lengths that are all much too long (from 0.03 to 0.09 A). The principal problem with the semiempirical methods is, however, their inability to represent accurately SI1and SIv in the same molecule, thereby making the distinction between the N,S.Nb diimide formulation and the divalent sulfur (Sb) less obvious. The 6-31G* structure, on the other hand, is in reasonable agreement with experiment, clearly showing almost equivalent S,N,andSaNbbonds(1.549and 1.535A)forthediimidefragment and a much longer length (1.65 1 A) for the single N a b bond. Nevertheless, the bonds in thediimide fragment are still calculated too short, whereas the nominally single N a b and SC bonds are calculated too long. The net result of this is that the angles about carbon differ by a few degrees from experiment. The final structure is thus lopsided, a result of the unsymmetrical ordering of the heteroatoms with respect to the carbonyl group and the SI1 and SIvformulation; this valence structure is also shown in Figure 4. The 3-21G* structure is not discussed here as it closely follows that calculated at the 6-31G* level with only slight increases in calculated bond lengths. Infrared Spectrum. The gas-phase FTIR spectrum of S2N2CO (Figure 2) shows a predominant band with PQR structure at 1742cm-l, corresponding to the CO stretching frequency. The 6-3 1G* calculated vibrational frequencies (scaled by 0.9 to account partially for harmonic effects) and intensities are also shown in the figure and are included in Table I. They illustrate the strong correspondence between experiment and theory. In addition to the very strong band at 1742 cm-l, distinct lower energy bands at 1068,926, and 780 cm-l are observed, together with a weak shoulder at around 950 cm-l. These correspond to various ring modes (Table I) with an average calculated deviation from experiment of 43 cm-1. Other bands are expected below 700 cm-l, but the calculated intensities indicate that they are too weak to observe with the possible exception of a tentative band at 675 cm-l. In general, this gas-phase data tracks those strong bands above 700 cm-1 observed for the crystalline material.19 PhotoelectrooSpectroscopy of W K O . The He I photoelectron spectrum of SIN~CO(Figure 3) has five strong bands below 14 eV which can be assigned to seven of the valence molecular orbitals (MOs) by comparison with calculation. The observed IPSare given in Table I1 together with the calculated MNDO, PM3, and 6-31G* (the latter scaled by 0.92 as a Koopmans’ correction) orbital energies and assignments. In general, the experimental IPS agree with the calculated orbital energies at the 6-31G* level and with both MNDO and
The Journal of Physical Chemistry, Vol. 97, No. 43, 1993 11219
The NSCl/AgNCO Reaction
TABLE II: Observed and Calculated IPS(eV) and Orbital Assignments for SZNZCO ~
~~~~
observed (9.85)’ 9.95 10.53
LUMO
7a”
0
I
~
6-3lG* a
calculated MNDob
PM3b
9.26 (6a”)
10.03 9.86 10.81 (24a’) 11.10 10.88 11.75 11.91 (Sa”) 12.38 11.96 12.00 12.26 (23a’) 12.48 11.64 13.77 (22a’) 14.25 13.40 13.26 14.19 (4a”) 14.51 14.36 13.83 13.9 14.82 (2la’) 15.05 13.78 15.3 16.25 (20a’) 16.40 15.75 15.9 16.67 (3a”) 16.59 16.29 17.17 16.7 17.22 (19a’) 17.24 17.8 20.38 (18a’) 21.93 20.16 NO 22.01 (17a’) 22.10 20.95 a Scaled, 0.92 X 6-31GSvalues for minimum C, symmetry structure. Unscaled values for minimum C, symmetrystructureusing the MNDO and PM3 Hamiltonians. AHf (MNDO) = 40.49 kcal mol-’; AHf (PM3) = 16.19 kcal mol-’. Adiabatic IP in parentheses; vibrational structure, 880 A 60 cm-I.
0 0
I
HOMO 6a”
0
I
O S 5a”
PM3. Experimentally, there is a pair of Ips at the lowest binding energies followed by two pairs of closely spaced peaks. The principal disagreement with ab inirio theory comes in the relative separations of the first three experimental IPS; the second IP is closer to the first than the third. Interestingly, both semiempirical methods do better (Table 11) despite the fact that the calculated structures are inferior. MNDO has the improved parameters for sulfur,35and PM3 is considered better for hypervalent molecules,36 although as noted before, this species is difficult with SI1and S I v in the samemolecule. On the basisof a comparison of experiment and theory, PM3 does perform better for the first seven IPS(albeit deficient on the structure) than MNDO: average deviation (MNDO) = 0.65 eV and average deviation (PM3) = 0.26 eV. Schematic MOs are shown in Figure 5 for the lowest unoccupied MO (LUMO) and the six highest occupied MOs; the subscripts a and b refer to the labeling of atoms. The highest occupied molecular orbital (HOMO), 6a”, is a nonbonding A orbital (A*b A * ~ with ) a substantial component on SI,,and the first IP (9.95 eV) which is well predicted by both MNDO and PM3 shows some weak vibrational structure (880 f 60 cm-I), probably corresponding to heavy atom movement. Although located on the SNSN fragment, it is higher by 2 eV than the corresponding IP in 1,3,2,4-benzodithiadiazine,C ~ H & N Z due , ~ ~to the added 0 PA contribution and no benzene destabilization. The second band at 10.53 eV is assigned to an in-plane nonbonding orbital (24a’) with a major contribution from 0 and some admixture from Na, Nb, and s b . Calculation (6-31G*) places it at 10.81 eV; MNDO and PM3 also do a reasonable job here. The next band, with a maximum at 12.00 eV and a lowenergy shoulder (1 1.75 eV), corresponds to the Sa’’ and 23a’ MOs. These are calculated within 0.35 eV by all three methods, although they disagree as to the relative ordering (Table 11). These orbitals are of mixed ring character, principally USN~--USN, (23a’) and Ab-A, (sa”) (Figure 5) with the large S b component of Sa” implying it will have the largest cross section with He I, thereby corresponding to the stronger band at 12.00 eV. The comparable A band in CsH4S2N2 is located at 11.65 eV.37 Above 13 eV two discrete bands are seen at 13.26 and 13.83 eV with a weak shoulder at 13.9 eV; all three computational methods agree that these must comprise three MOs (22a’, 4a”, and 21a’) separated by less than 1.05 eV. Again these MOs are delocalized over the whole ring, e.g., 22a‘ is principally usaband UNG,, whereas 4a” shows ASNS together with substantial CO PA bonding. Above 14.5 eV, broad, weaker bands are also observed (15.3, 15.9, 16.7, and 17.8 eV) corresponding to the remaining four valence orbitals (20a’, 3a”, 19a’, and 18a’. all bonding in nature) predicted (6-31G*, 16.25, 16.67, 17.22, and 20.38 eV)
+
23d
UI
\
I
4a”
Figure 5. Schematic orbital diagrams for the LUMO, HOMO, and the remaining five highest-occupied MOs for SzN2CO.
within the He I range. The feature with the most intensity (15.9 eV) is probably the ANSNS bonding MO. Formation of SzNzCO in the NSCl(g)/AgNCO(s) Reaction. We have established that S ~ N Z C is Othe product of the reaction of NSCl(g) and AgNCO(s), and so something quite different from what was originally envisaged must be occurring. It is wellknown that in the solid phase NSCl exists as the trimer N3S3C13, a six-membered ring of alternating nitrogen and sulfur atoms. One possibilityis that NS3Cl3 reacts as the trimer with the packed AgNCO(s) allowing for numerous secondary reactions, possibly from a tris(isocyanat0) derivative. Given the temperature of the AgNCO, however (125 “C), this is unlikely, and therefore one must consider reactions based on monomeric NSC1. One possibility, built on the knowledge that thiazyl chloride can behave as a sulfur analogue of a cyanide, is for a reaction of the type ClSN
-
+ AgNCO + */gSs
OCS,N,
+ AgCl
RCN (instead of AgNCO) is known to react in this way.38 This fits with the observation that free sulfur is formed early in the reaction, manifesting itself on the walls of the reaction tube.
Conclusions Generation of NSNCO was sought by reacting gaseous NSCl with solid AgNCO and monitoring the effluent gas. The product was analyzed using PE spectroscopy, FTIR, and PIMS using both He I and HL, light sources. Comparison of the PIMS data from the two light sources affords a parent ion of m / z = 120 and major fragment ions of m / z = 78 and 46. On the basis of this, the strong carbonyl stretching frequency in the FTIR, and the form of the PE spectrum, the product was deduced to be 5-oxo1,314,2,4-dithiadiazole,S Z N ~ C Oa ,molecule characterized previously in the solid state. To confirm this, S2NzCO was synthesized, with its gas-phase PE, PIMS, and FTIR spectra concurring with the results of the NSCl/AgNCO reaction.
11220 The Journal of Physical Chemistry, Vol. 97, No. 43, I993
Geometry optimizationsfor the molecule were carried out via MOPAC (MNDO and PM3) and at the 3-21G* and 6-31G* SCF level using ab initio methods. All geometries were of planar C, symmetry and consisted of five-membered ring with an exocyclic oxygen. The optimized structure at the 6-3 lG* level compares favorably to the crystal structure, the semiempirical structures less so due to their inability to represent accurately SI1 and SIv in the same molecule. Frequency calculations at the 6-31G' level provide a good correspondence to the experimental FTIR. Orbital assignments for the observed bands in the PE spectrum of S2N2CO were made by comparison with MNDO, PM3,and the scaled (6-31G*) orbital energies.
Acknowledgment. N.P.C.W. thanks the Natural Sciences and Engineering Research Council of Canada (NSERC) for grants in support of this work. E.G.L. thanks NSERC for the award of a graduate studentship. K. T. Bestari and R. T. Oakley are thanked for assistance with the synthesis. References and Notes (1) Westwood, N. P. C. Chem. SOC.Rev. 1989,18,317. (2) Alaee, M.; Hofstra, P.; Kirby, C.; Westwood, N. P. C. J. Chem. Soc., Dalton Trans. 1990. 1569.
( 5 ) Frost, D. C.; MacDonald, C. B.; McDowell, C. A.; Westwood, N. P. C. Chem. Phys. 1980,47,111. (6) Langhoff, S. R.; Jaffe, R. L.; Chong, D. P. Int. J. Quantum. Chem. i983,23,875. (7) Mayer, E. Monarsh. Chem. 1970,101, 834. (8) Hocking, W. H.; Gerry, M. C. L. J. Mol. Spectrosc. 1976,59,338. (9) Frost, D. C.; Kroto, H. W.; McDowell, C. A.; Westwood, N. P. C. J . Electron. Spectrosc. Relar. Phenom. 1977,I I , 147. (10) King, M. A.; Kroto, H. W. J. Am. Chem. SOC.1984,106,7347. (1 1) Vondrak, T.; Westwood, N. P. C. Unpublished work. (12) Alaee, M.; DeLaat, R. H.; Westwood, N. P. C. To be published. (1 3) Comstock Inc., Oak Ridge, TN.
deLaat et al. (14) Wannberg, B. W.; Skollermo, A. J. Electron Specrrosc. Relat. Phenom. 1977,10, 45. (15) Gelius, U.; Asplund, L.; Basilier, E.; Hedman, S.;Helenelund, K.; Siegbahn, K. Nucl. Instrum. Merhods Phys. Res. 1984,B1, 8 5 . (16) Neville, R. G.; McGee, J. J. Inorg. Synrh. 1966,8,23. (17) Neidlein, R.; Leinberger, P.; Gieren, A,; Dederer, B. Chem. Ber.
1978,111,698. (18) Roesky, H. W.; Wehner, E. Angew. Chem., Int. Ed. Engl. 197$,14, 498. (19) Roesky, H. W.; Witt, M. Inorg. Synth. 1989,25,49. (20) Zumach, G.; Kiihle, E. Angew. Chem., Inr. Ed. Engl. 1970,9,54. (21) Frisch,M. J.;Binkley, J. S.;Schlegel,H. B.;Raghavachari,K.;Melius, C. F.; Martin, R. L.; Stewart, J. J. P.; Bobrowicz, F. W.; Rohlfing, C. M.; Kahn, L. R.; Defrees, D. J.; Seeger, R.; Whiteside, R. A.; Fox, D. J.; Fleuder, E. M.;Pople, J. A. Gaussian 86; Carnegie-Mellon Quantum Chemistry Publishing Unit: Pittsburgh, PA, 1984. (22) Stewart, J. J. P. QCPE 455,MOPAC A General Molecular Orbital Package, version 5.00; Frank, J. Seiler Research Laboratory, US.Air Force Academy: Colorado Springs, CO, 1990. (23) Boyd,D. B.;Smith,D. W.;Stewart, J. J.P.; Wimmer,E. J. J.Compur. Chem. 1988,9, 387. (24) Cowan, D. 0.;Gleiter, R.; Glemser, 0.;Heilbronner, E. Helv. Chim. Acra 1972,55, 2419. (25) DeKock, R. L.; Shehfeh, M. A.; Lloyd, D. R.; Roberts, P. J. Trans. Faraday Soc. 1976,807. (26) French, C. L.; Oakley, R. T.; Westwood,N. P. C. Unpublishedresults. (27) Eysel, H. H.; Nachbaur, E. Z . Anorg. Allg. Chem. 1971,381,71. (28) Gottardi, W. Monarsh. Chem. 1972,103,1150. (29) Devore. T. C. J. Mol. Struct. 1987. 162. 287. (30) Lau,W. M; Westwood, N. P. C.; Palmer; M . H. J. Am. Chem. SOC. 1986,108,3229. (31) Palmer, M. H.; Lau, W. M.; Westwood, N. P. C. Z . Narurforsch. 1982.37A. 1061. (32) R&ky, H. W.; Wehner, E.;Zehnder, E.-J.; Deiseroth, H.-J.;Simon. A. Chem. Ber. 1978,1 1 1 , 1670. (33) Roesky, H. W.; Miiller, T.; Wehner, E.; Rodek, E. Chem. Ber. 1980, I I 3.2802. (34) Heitmann, A.; Edelmann, F. Z . Naturforsch. 1983;388, 521. (35) Dewar, M.J. S.; Reynolds, C.H. J. Compur. Chem. 1986,7, 140. (36) Stewart, J. J. P. J. Comput.-Aided Mol. Design 1990,4, 1. (37) Hitchcock, A. P.; DeWitte, R. S.; van Esbroeck, J. M.; Aebi, P.; French, C. L.; Oakley, R. T.; Westwood, N. P. C. J. Electron. Spectrosc. Relat. Phenom. 1991, 57, 165. (38) Banister, A. J. Phosphorus Sulfur Relai. Elem. 1978,5, 147.
