摘要:
1978 277A Theoreticat Investigation of the Structure of some Small Nitrogen-S u I p h u r M o I ecu lesBy Michael P. S. Collins and Brian J. Duke,**t Department of Chemistry, University of Lancaster, Bailrigg,The structure of the nitrogen-sulphur compounds NSF. HSN. N2S. N,S, and H8NS has been investigated using theL.C.A.O.M.O. ab initio method and a moderately large basis set. The compound NSF is found to be more stablethan ‘SNF in agreement with experiment, HNS is predicted to be more stable than HSN. NNS (linear) is more stablethan NSN, and the order of stability of the isomers of HaNS is predicted to be H,SNH < H,SN < HSNH, < SNH,.The compound N2S2 is shown to be square planar in agreement with recent experimental evidence.Lancaster LA1 4YATHE chemistry of nitrogen-sulphur compounds hasattracted much recent interest but there are stillseveral species whose structures are not fully character-ised. Surprisingly, there have been few ab initiomolecular-orbital studies of such compounds , althoughthis technique is known to accurately predict the struc-tures and geometries for small molecules if sufficient careis taken with the choice of basis set.In this study wehave investigated a number of small molecules containingN-S bonds. The structure of NSF has now been assignedunambigu~usly,~~~ but HSN or HNS is so far unknown.The compound N,S was prepared4 only after our cal-culation had been completed and initial structural datahave now been publi~hed.~ Similarly, little is known ofthe possible isomers of H,NS.The geometry of N,S,in its adduct with SbC1, has been determinedJb butrecently this has been shown to be different from thefree molecule.6 Our objective has been to study thegeometry of these systems using a moderately large basisset of Gaussian orbitals in the usual L.C.A.O.M.O. abinitio technique. The role of d functions in predictingaccurate geometries is found to be particularly impor-tant.METHODThe ab initio molecular-orbital method using a basis setof contracted Gaussian orbitals (g. t.0.s) was employed.The basis set was selected after extensive tests on H2S andNSF and was a compromise between economy and accuracy.It is nevertheless moderately large and gives good optimisedgeometries. Calculations on H2S reported elsewhere 7 andon NSF reported here indicate that the inclusion of dorbitals on sulphur is crucially important in predictingaccurate geometries, even though the effect of these orbitalson the variational energy may be small. For economy weused single-zeta contractions for the core orbitals anddouble-zeta contractions for the valence orbitals.Thisresults in energies somewhat above those reported else-where. Total energy is not necessarily a good measure ofthe accuracy of a wavefunction. Improvements in theprescriptions of the core orbitals, especially the nuclearcusps, leads to dramatic changes in the molecular energywithout necessarily improving the representation of thevalence regions of the molecule. The quality of the valencet Present address : Department of Chemistry, Bayero Univer-sity, Kano, Nigeria.$ Throughout this paper: 1 Bohr w 5.29 x 10-l’ m; 1Hartree w 2.62 x lo3 kJ mol-l; 1 D R 3.33 xH.G. Heal, Adv. Inorg. Chem. Radiochem., 1972, 15, 375.W. H. Kirchoff and E. Bright Wilson, jun., J. Amer. Chem.F. X. Powell, Chem. Phys. Letters, 1975, 88, 393.C m.* T. Barrow and R. N. Dixon, Mol. Phys., 1973, 26, 109.Soc., 1963, 85, 1726.basis seems to be an important factor in describing bondingand obtaining optimal geometries. For NSF the followingbasis sets were contrasted.SP.-This was the foundational set employed. Itpossessed Basch-Snyder double-zeta functions for thevalence orbitals of the light nuclei and Clementi-Reotti @STO-3G double-zeta representations for the sulphurorbitals. Clementi-Raimondi 10 STO-3G single-zeta formul-ations were used for all the core orbitals.SPP’.-The sulphur p space was improved by the additionof an optimal p STO-3G function.SPP’P’’.-The nitrogen p space was further improved(with respect to SPP’) by the incorporation of an optimal3p STO-3G orbital.SPD.-Initially, an optimised STO-1G (exponent 1.2)d orbital was added to the SP set.Further d functionswere subsequently added to produce SPDD’ and SPDD’D”sets.Results of optimising the N-S bond length and NSF bondangle are shown in the Table. Bond lengths and bondangles at the lowest energy are accurate to 0.005 Bohr 3Nitrogen-sulphur bond length and bond angle in NSFas a function of the choice of basis setBasisSPSPP’SPP’P’’SPDSPDD’SPDD’D”ExperimentTotal energy(Hartreej-6644.898 2-545.092 8-646.153 6-646.086 1-546.143 9-545.172 5N-SDistance3.233.133.132.882.832.782.75( B O WN-S-FAngle108108109113116116116.77(“1and 0.5” respectively.It is clear that only basis setscontaining a d function give results close to the experimentalones. The importance of d orbitals, both for predictingaccurate bond lengths and angles as well as gross structuralfeatures, has been recently discussed by several workers(see ref. 7 and refs. therein). For further work the basis setSP was used as above for S, F, and N with the P” improve-ment of the N p space but without the extra S p function.This was augmented by an optimum (for NSF) STO-1C dorbital on S, although the results are very insensitive to theactual value of the d-orbital exponent.For hydrogen aR. L. Patton and K. N. Ratmond, Inorg. Chem., 1969, 8,2426.C. M. Mikulski, P. J. Russo, M. S. Saran, A. G. MacDiarmid,A. F. Garito, and A. J. Heeger, J. Amer. Chem. Soc., 1975, 97,6358.M. P. S. Collins and B. J. Duke, Chem. Phys. Letters, 1976,42, 364.8 L. C. Sny? and H. Basch, ‘Molecular Wave Functionsand Properties, Wiley, New York, 1972.C. Roetti and E. Clementi, J. Chem. Phys., 1974,60, 4726.lo E. Clementi and D. L. Raimondi, J. Chem. Phys., 1963, 88,2686278 J.C.S. DaltonBasch-Snyder double-zeta basis plus a STO- 1G 2p polaris-ation function (exponent 0.75) was employed.For each molecule, different geometries and differentisomers were studied.The results for the geometries arelikely to be more accurate than those for the differences inenergy between structural isomers. In each case accurateresults are dependent on there being no change in cor-relation energy with change in geometry. From manyrecent calculations i t appears that this assumption must bereasonably accurate. The use of the ab initio molecular-orbital method for predicting structural isomers is nowwidespread. At worst, it gives encouragement to experi-mentalists to study a particular system and a t best i t givesvery accurate predictions. It gives better estimates ofenergy differences for systems having the same number ofelectron pairs than i t does for binding energies or bond-breaking processes.RESULTS AND DISCUSSIONThionitrosyl Fluoride, NSF.-Although this moleculeis now well chara~terised,~~~ it is interesting to note thatwhen first investigated it was assigned as SNF ratherthan NSF.11-13 A full geometry search was thereforecarried out on both species with the following results (alldistances are in Bohrs) :SNF NSFCalc.Calc. Expt.S-N-F 120 N-S-F 115 O 116.46 ON-F 3.08 N-S 2.85 2.75N-S 2.90 S-F 3.21 3.11Mulliken chargesF -0.003 6 -0.004 2N 0.011 7 -0,001 8S -0.008 1 0.006 0The compound NSF is more stable than SNF by 0.011 2Hartree. This, of course, is in contrast to the oxygenanalogue, where ONF is more stable than NOF. Ineach case the electronegative fluorine atom is bonded tothe more electropositive atom.In an ab initio calcul-ation on ONF, Peslak et aZ.14 found the energy differenceto be 0.074 Hartree. The CNDOIZ method also predictsNSF as the most stable isomer although with a muchexaggerated energy difference in comparison to theab initio result. The CND0/2 structure also has veryshort bonds (N-S 2.43 Bohr) even though the bond angleis accurate.HNS or HSN.-This system is apparently unknownand it is therefore unclear whether the most favouredstructure would be HSN or HNS. A full geometrysearch gives the following results:HNS HSNN-S 2.93 N-S 3.08N-H 1.90 S-H 2.56H-N-S 100 O H-N-S 104 OHNS is more stable than HSN by 0.040 8 Hartree.Similarly HNO is more stable than HON by 0.03947, 94.1956, 279, 28.1971, 55, 1993; J .Amer. Chem. SOC., 1971, 98, 5001.l1 0. Glemser, H. Richert, and F. Rogowski, Naturwiss., 1960,la F. Rogowski, 2. phys. Chem., 1961, 27, 277.Is 0. Glemser, H. Schroder, and H. Haesler, 2. anorg. Chem.,l4 J. Peslak, D. S. Klett, and C. W. David, J . Chem. Phys.,Hartree according to an ab initio calculation by Peslaket As a ligand NS behaves differently to H than itdoes to F, and the energy difference is fairly substantial.There is no experimental support for this prediction, butthere are some analogies. For example, in N4S4F, thefluorine is co-ordinated to sulphur, but in N4S4H4 thehydrogen is attached to the nitrogen. In the first-rowoxygen congeners the F or H atom is always bound to thenitrogen.Lithium, however, is thought to bond morefavourably with the oxygen in LiON,14 and a similarsituation may exist l5 in the matrix-isolated metaltricarbonyls [M(CO)J (M = K or Cs). CND0/2 dis-agrees with the ab initio prediction and favours HSN.However, the ab initio result seems to be a clear theoreti-cal prediction and the experimental investigation of thissystem is awaited with interest.Dinitrogen Su@hide.-The compound N,S can beconsidered as a nitrogen atom bonded to NS and is thussiniilar to the previous systems. It was discoveredduring the course of our work by Powell4 who haspublished initial structural data and is currently workingon its structural refinement.16 No other properties suchas dipole moment, ionisation potential, etc., are available.The following possible forms were considered; N(NS)linear and bent (analogous to N,O) and N(SN) linear andbent similar to SO,.The most favoured structure isNNS linear with N-N 2.13 Bohr and N-S 3.08 Bohr.The tentative experimental results agree well with thistheoretical assignment giving N-N 2.15 Bohr and N-S2.99 Bohr. The dipole moment is predicted to be 0.18D and the Mulliken charges are:N-N-S-0.0945 0.0280 0.0665The N-S bond length is again overestimated but theexperimental results are not certain.Dinitrogen Disul9hide.-Experimentally, the structureof this free species was not known before.this comput-ation was initiated, although the geometry of the adductwith SbC1, had been determined. Our results, whichhave appeared in a preliminary comm~nication,~~predict the molecule to be square planar with a N-Sdistance of 3.06 Bohr, in close agreement with therecent experimental result 6 of N-S 3.11 Bohr and N-S-N89.58", and in conflict with the results for the SbC1,adduct where the angle N-S-N was found to be 85".CNDOI2 predicts a bond length of 3.23 Bohr andN-S-N 88".A minimum basis-set (STO-3G) calculationhas recently been reportedla for N,S, which gives anoptimum N-S distance of 3.203 Bohr and N-S-N 86.5'.This result is in accordance with our conclusion that,without the inclusion of d functions on sulphur, the N-Sdistance will be overestimated. Neither the bonddistance nor the angle is in good agreement with experi-ment.1s M.P. S. Collins, B.A. Thesis, University of Oxford, 1973.18 F. X. Powell, personal communication, 1975.17 M. P. S. Collins and B. J. Duke, J.C.S. Chem. Comm., 1976,18 M. Kertesz, S. Suhai, A. Azman, D. Kocjan, and A. I.701.Kiss, Chem. Phys. Lettevs, 1976, 44, 531978 279H,SN, H,SNH, HSNH,, and SNH,.-It was antici-pated that the most favourable isomer of H,SN would be2-89 / 0.0047H H /S-N H H0.0705tS t N f H 358HEnergy changes (Hartree) and optimal N-S distances (Bohr) forfor various H,NS isomersH,NSH analogous to PH,(SH).19 It was only aftercompletion of these calculations that we became awarethat Kerouanton et al. 2o had synthesised and character-a* P. W. Schenk and B. Leutner, Angew. Chem., 1966, 78, 942. ** A. Kerouanton, M.Herlem, and A. Thiebault, Analyt. Letters,1973,6, 171.ised H,NS, which thus appears to be the most stableisomer. This indeed appears to be the most favourablesystem as shown in the Figure. The stability sequenceH,SNH < H,SN < HSNH, < SNH, is not quite inaccord with the trend for the siting of the protons, butthe energy difference between the first two systems isvery small. The tendency for the hydrogen atoms to bebonded to the more electronegative nitrogen of NS, notedfor HSN, comes out strongly in this case.Only the bond lengths were fully optimised. Theangles were idealised for H,SN and SNH, and the valuesof Mezey et aL21 were used for sulphimide. For SNH,the bond lengths are predicted to be 3.58 and 1.91 Bohrfor the N-S and N-H bonds respectively. The N-Sdistance is very long and, with the Mulliken densities,might be used asevidence that the bond isdative H,N+S.Mezey et studied sulphimide using a split valence-shell basis which also described the cores very well.Although their energy is lower than ours, -452.948 1comparedwith -449.648 1 Hartree, the inversion barriersare similar, 10.1 and 12.3 kJ mol-l. Again we feel thatour calculations provide a chemically significant struct-ural prediction for a molecule where little experimentalevidence is available.We thank the S.R.C. for the award of a studentship (toM. P. S. C.) and for a grant of computer facilities at theAtlas Computing Laboratory.[6/2219 Received, 3rd December, 1977121 P. Mezey, A. Kucsman, G. Theodorakopoulos, and I. G.Czismadia, Theor. Chim. A d a , 1975, 88, 115
ISSN:1477-9226
DOI:10.1039/DT9780000277
出版商:RSC
年代:1978
数据来源: RSC