New synthetic route to WSFb and its solution-phase structure as determinedby tungsten L o d g e extended X-ray absorption fine structure studiesKulbinder K. Banger, Christopher S. Blackman and Alan K. Brisdon *Department of Chemistry, University of Manchester Institute of Science and Technology ( UMIST),PO Box 88, Manchester M60 lQD, UKThe compound WSF, was synthesized in one step by the reaction of WF, with (Me,Si),S in solution atambient temperatures. Solution-phase structural data have been obtained for the first time for it and relatedtungsten chalcogenide tetrahalides via a tungsten L(II1)-edge EXAFS (extended X-ray absorption finestructure) study. This demonstrates that WOF,, WOCl, and WSF, in acetonitrile or dichloromethane solutionare monomeric. For WSF, analysis of the EXAFS data yielded bond lengths of 2.026(8) (W=S) and 1.863(3) A(W-F).The W=S bond length is shorter than previously determined, by gas-phase electron diffraction, but ismore consistent with that found in related structures.A large number of fluoride oxides of transition metals areknown,’ they are generally well characterised and a wide rangeof experimental methods are available for their synthesis. In thesolid state most adopt fluoride-bridged structures2 and it hasbeen suggested 3*4 that the sulfur and selenium analogues arebased on similar fluoride-bridged polymeric structures. Bycomparison the structure of the analogous chlorides appear tobe based on distorted chloride-bridged dimeric units. 536Probably the most studied of these compounds are thechalcogenide tetrahalides of tungsten, i.e.WEF, (E = S, Se orTe), with the principal synthetic routes being based on high-temperature autoclave methods.’ However the amount ofphysical data available is very limited, which may be ascribedto a number of factors including their extreme reactivity tomoisture. This lack of data is unfortunate since the combinationof traditionally ‘hard’ and ‘soft’ ligands on the same high-oxidation-state metal centre makes such species interestingfrom structural and bonding aspects.Structural data for the gas-phase tetrahalide sulfide com-pounds WSX, (X = F, CI or Br) have been obtained fromhigh-temperature electron diffraction studies which concludedthat under such experimental conditions they all adoptmonomeric square-based pyramidal structures.Howeverexamination of the bond lengths obtained reveals an unusualtrend for the tungsten-sulfur distance of the tungstentetrahalide sulfides. The reported W=S bond length decreasedfrom 2.109 8, in WSBr, to 2.086 A in WSCl,, as anticipated, butthen increased to 2.104 8, in WSF,. This is clearly not what mightbe expected on the basis of simple electronegativity arguments,’nor is the trend mirrored in the data obtained for the W=O bondlengths in the oxide tetrahalides obtained by electrondiffraction (1.684, WOBr,; 1.685, WOCl,; 1.666 A, WOF,).However the authors noted that problems were encounteredstudying WSF, primarily due to sample decomposition.The extreme reactivity of these species is amply demonstratedby the fact that just a single communication of thedetermination of the crystal structure of WSF, exists., Thisreveals a fluoride-bridged polymeric chain [d(W=S) = 2.07,d(W-F),, = 1.87 A].It also reports the structure of anacetonitrile adduct (WSF,-MeCN) as a monomer with a W=Sbond length of 2.034 A. In both cases the W=S bond distance issignificantly shorter than the value derived from the electrondiffraction experiments and indeed shorter than the distancein WSCl,. In view of previous successes in the use of solid-and solution-phase EXAFS (extended X-ray absorption finestructure) studies to determine structural parameters for someof the least-stable fluorine-containing species, a re-evaluation ofthe structure of monomeric WSF, was undertaken.We reportan extension of a previously published method” for theconvenient one-step solution-phase synthesis of WSF, and ourfindings regarding the W=S bond length as determined bysolution-phase tungsten L(II1)-EXAFS studies.Experiment a1Hexamethyldisilathiane and deuteriated acetonitrile anddichloromethane (Aldrich) were used as supplied afterdegassing by a number of freeze-thaw cycles. Diethyl ether(BDH) was dried by standing over sodium wire for ca. 1 d andsubsequently refluxed over CaH, under a dinitrogenatmosphere. The compound WF, (Fluorochem) was used assupplied after verification of its purity by spectroscopicmethods. Fluorine NMR spectra were recorded of solutions inheat-sealed prepassivated FEP tubes (outside diameter 4 mm,inside diameter 3 mm) held concentrically in a 5 mm glass NMRtube on a Bruker AC200 spectrometer operating at 188.296MHz and referenced against CFCl,.All reagents were handledunder a nitrogen atmosphere in a dry-box (Belle Technology,UK) or on a glass and PTFE vacuum line.In a typical reaction a heat-sealed FEP tube fitted with PTFEvalve (STD/VC-4 Production Techniques Ltd., Fleet, UK) wasevacuated and passivated with WF, on a vacuum h e . Afterremoving the WF, the tube was taken into a dry-box where(Me,Si),S (0.06 g, 0.34 mmol) was dissolved in solvent(acetonitrile or dichloromethane, 0.35 cm3). The tube wasresealed and reconnected to the vacuum line followed bypassivation with WF, of those parts of the line that had beenexposed to the atmosphere.The compound WF, (0.1 g, 0.34mmol) was condensed into the reactor held at 77 K followed bya further 0.15 cm3 of solvent. On warming to room temperaturean immediate reaction ensued which resulted in the formationof a small amount of a dark brown precipitate. After allowingthe solid to settle the pale yellow solution was decanted off intoa second prefluorinated 4 mm FEP tube and the tube heatsealed. Fluorine NMR spectra of this solution exhibited anintense singlet at 6 86.2 showing tungsten-1 83 satellite couplingof 33.4 Hz. A weak peak at 6 66.9, also showing tungstensatellites, due to WOF, was observed on occasions. A ten-linemultiplet centred at 6 - 157.1 was also observed. This signalis assigned to SiMe3F on the basis of its chemical shift andthe magnitude of the observed coupling constants, 7.4 HzC3J(’H-’’F)] and satellites ‘J(”Si-”F) = 272.7 Hz (4.7%J.Chem. SOC., Dalton Trans., 1996, Pages 2975-2978 297natural abundance). ' ' The compounds WOF, and WOCl,were prepared by standard routes 2 3 1 3 and characterised byelemental analysis and spectroscopic methods.The EXAFS studies were carried out either on solutions ofthe tungsten chalcogenide tetrahalide dissolved in deuteriatedsolvents used for recording NMR data or on fresh samplesprepared by dissolving a known mass of solid in acetonitrile,diethyl ether or dichloromethane in a dry-box (concentrationca. 0.05 mol drn-,). In both cases solutions were transferred toheat-flattened 4 mm FEP tubes (path length ca.1 mm) whichwere then sealed. The tungsten L(II1)-edge EXAFS transmissionspectra were recorded at the Daresbury Synchrotron RadiationSource operating at 2 GeV (ca. 3.2 x lo-'' J) with an averageoperating current of 180 mA on station 7.1 using an order-sorting Si( 1 1 1) monochromator offset to 50% of the rockingcurve for harmonic rejection. Several data sets were collected atroom temperature in k space and averaged to improve thesignal-to-noise ratio. The monochromator was calibrated usinga thin tungsten foil. The EXAFS data treatment utilised theprograms EX l 4 and EXCURV92. ' The background wasremoved by fitting the pre-edge by a straight line andsubtracting this from the spectrum. The atomic contributionto the oscillatory part of the absorption spectrum wasapproximated using a polynomial and the optimum fit judgedby minimising the intensity of chemically insignificant shells atlow r in the Fourier transform.Curve fitting used single-scattering curve-wave theory with phase shifts and back-scattering factors calculated using the X, option, based onnormal ab initio methods,' within EXCURV92.Results and DiscussionWhen hexamethyldisilathiane dissolved in acetonitrile ordichloromethane is added to a solution of tungstenhexafluoride dissolved in the same solvent at low temperaturefollowed by warming to room temperature an immediatereaction ensues. According to fluorine NMR measurements,tungsten tetrafluoride sulfide and trimethylsilyl fluoride aregenerated as the principal fluorine-containing solution-phasespecies.In most reactions a variable amount of a dark browninsoluble compound was also formed. A number of separateexperiments were performed with slight variations in the solventand conditions used, however they were generally unsuccessfulin completely eliminating this side reaction although it may beminimised by working at low temperature (- 30 "C) and withmoderately dilute (ca. 0.1 mol drn-,) solutions. Solid WSF, wasisolated by allowing the dark solid to collect at the bottom ofthe reactor followed by decantation of the solution andsubsequent removal of the solvent and volatile trimethylsilylfluoride resulting in a light yellow solid which is extremelymoisture and oxygen sensitive.When exposed to the atmos-phere it darkens immediately and H2S is liberated.The 19F NMR studies confirm the identity of the tungsten-fluorine product as WSF, by comparison with the previouslyrecorded chemical shift of 6 86.9 with respect to CFCI, and thetungsten-1 83 coupling constant., An additional weak singlet,also exhibiting tungsten- 183 satellites, could occasionally beobserved at 6 66.9 corresponding to a small amount of WOF,,presumably arising from limited hydrolysis. Integration of thesignals due to WSF, and WOF, suggests that typically WOF, ispresent at less than 1%.The reaction also produced a variable, but small, amount ofbrown material. The small scale meant it was not possibleunambiguously to identify the solid; elemental analysis showsthat it contains tungsten and sulfur and it is in all likelihoodWS,.However its apparent complete insolubility in thesesolvents resulted in facile separation from the desired productand does not pose a problem. This route therefore provides asimple one-step synthesis of WSF, in accordance with equation(1).WF, + (Me,Si),S - WSF, + 2Me,SiF (I)Attempts to generate WS,F2 by increasing the ratio ofhexamethyldisilathiane to WF, to 2: 1 and beyond wereunsuccessful, resulting instead in a greater mixture of WSF,and the brown compound.The EXAFS studies were carried out either on fresh samplesdissolved in acetonitrile, diethyl ether or dichloromethane or onsamples dissolved in deuteriated solvents which were used inNMR measurements.Spectra were recorded in transmissionmode out to k = 15 k' (k = photoelectron wave vector), butdue to a poor signal-to-noise ratio the data sets were truncatedat k = 13.5 A-'. Several data sets for each compound werecollected and averaged and the data multiplied by k3 tocompensate for the fall-off in intensity at higher k. Nosmoothing or Fourier filtering was applied to the data.The Fourier transforms of the background-subtractedEXAFS data for WOF, and WOCl, dissolved in dichlorometh-ane exhibit two major peaks. The more intense feature wassuccessfully modelled to the appropriate four halide ligands andthe less intense peak to the oxide ligand around the tungstencentre as expected for a monomeric species.In both cases theFourier transforms also exhibit more-distant weaker features ataround 4 A. These could not be successfully modelled as a secondtungsten atom arising from a polymeric structure, and thedistances are too great to be due to bridging ligands. Howeverwe note that they occur at a similar distance to the solvationspheres that have been observed for metal hexafluoridesdissolved in anhydrous HF. l 7 Modelling of these more distantfeatures was performed by stepwise addition of further shellscontaining one and two chlorine atoms which were iterated inthe usual way, and the best fits tested for statistical significance.For both compounds the addition of the solvation spheresresulted in a decrease in the R factor whilst the distances of thebonded shells remained essentially invariant.No attempts weremade to determine accurately the occupancy numbers andDebye-Waller factors for these non-bonded shells. Checks onthe co-ordination numbers of the first two shells wereperformed by mapping the occupation number of a shell againstits Debye-Waller parameter. The distances, and otherparameters, determined from these EXAFS analyses arepresented in Table 1 where they are compared with the datapreviously obtained by other techniques.The Fourier transform of the EXAFS spectrum of WSF, indichloromethane exhibits a single, broad peak centred around2.0 8, with weaker features at ca. 2.9 and 3.6 8, [Fig. I(b)]. Initialmodelling of these data was undertaken based on a two-shellmodel (4 F, 1 S).Both of these contributions are containedwithin the single broad peak envelope seen in the Fouriertransform. However modelling of the data in terms of these twoshells alone resulted in a R factor of just under 20%. Aconsiderable contribution to R appears to be due to thepresence of the additional features at longer distance. Indeedthis is amply demonstrated by removing the contribution tothe EXAFS from the more distant shells completely byFourier filtering (Fourier window 0.2-2.5 A) followed byiteration which results in a lowering of R to 13% withoutsignificant changes to the other parameters. There are a numberof possible interpretations of these more distant shells: theycould be due to the presence of bridging atoms, sampleimpurities or solvation shells.The non-Fourier-filtered data setwas analysed for each of these possibilities. The distances (ca.2.97 and 3.60 A) are too great to be due to bridging fluorineatoms and modelling of these features to tungsten atoms atthese distances resulted in unacceptably large Debye-Wallerfactors, thus WSF, appears to be monomeric in solution. Thesecond possibility is that these features could arise from animpurity, for example due to sample decomposition. This wasinvestigated by modelling the additional two shells to the mostlikely non-fluorine-containing impurity WS,, which is reported2976 J. Chem. Soc., Dalton Trans., 1996, Pages 2975-297Table 1 Bond distances and EXAFS parameters for the compounds WEX, (E = 0 or S, X = F or C1)EXAFSBonded shells Non-bonded shellsElectron CrystalCompound diffraction a structure DistanceIA 2 0 2 ‘/A2 R DistanceIA 202 ‘/A2 VPI/eV AFAC R e (%)WOF, d(W=O)/A 1.666(7)d(W-F)/A 1.847(2)WSF, d(W=S)/A 2.104(7)d(W-F)/A 1.847(3)WOCI, d(W=O)/A 1.685( 15)d(W-Cl)/A 2.280(3)WSCl, d(W=S)/A 2.086(6)d(W-Cl)/A 2.277(3)2.1 1(4)1 1.686(3) 0.002(1)1.84(4)1 1.852(2) 0.004(1)2.07 2.026(8) 0.005(2)1.87g 1.863(3) 0.007( 1 )1.81 1.691(3) 0.004(1)2.28 2.303( 1) 0.006( 1)2.098 j2.299( 1 1)’ j2.73.028(7) 0.015(3)3.659(12) 0.014(2) -4.71 0.869.72.969(7) 0.012(1)3.603(7) O.OlO(2) -5.86 0.937.43.887(15) 0.014(4)4.488(8) 0.0 10( 1) - 6.43 0.934.66.74.5a Ref.8. This work, CD,C12 solution; standard deviation in parentheses, systematic errors in bond distances arising from data collection andanalysis procedures are ca.k 0.02 A. ‘ Debye-Waller factor. R = [J(xlheor - ~ ~ ~ p ~ ’ ) k ~ d k / ~ ~ ~ ~ p ‘ ’ k ” d k ] x 100% obtained after modelling for shells 1and 2 only. Final R factor obtained after inclusion of non-bonded shells in modelling. Fluoride-bridged tetramer.3 Fluoride-bridged p01ymer.~Chloride-bridged dimer.6 Chloride-bridged dimer. Analysis not possible because of low-quality data due to poor sample solubility.1 2 t ( a ) -t -!0 2 4 6 8 10rfAFig. 1 ( a ) Background-subtracted EXAFS (-, experimental x k3;- - - -, curved-wave theory x k”) for WSF, dissolved in CD,CI, and(b) the corresponding Fourier transforms (-, experimental; - - - -,theoretical)as a product of WSF, in extended contact with a~etonitrile,~by comparison with a previous EXAFS study of tungstensulfides [WS,, d(W=S) = 2.41 A].However after iteration themodel was not able to reproduce the correct distances or co-ordination numbers from the EXAFS data for any knownbinary tungsten sulfide and thus this possibility was alsorejected. This leaves the third hypothesis, that these moredistant shells are due to the co-ordination of solvent molecules.For the data recorded in dichloromethane solution we were ableto reproduce satisfactorily the additional features observed inthe Fourier transform by incorporating shells corresponding toco-ordination spherescontaining 1 C1 at 2.97 A and 2 C1 at 3.60A.After iteration this resulted in a decrease in R to 16.7% and theparameters given in Table 1.The EXAFS data recorded foracetonitrile solutions of WSF, resulted in W=S and W-Fdistances which are essentially invariant within the precision ofthe EXAFS technique. Well defined, although slightly lessintense, non-bonded shells were also obvious from the Fouriertransform of the EXAFS data at 2.86 and 3.48 A. These areslightly shorter non-bonded distances than those obtained fromdichloromethane solution and modelling suggests that theycorrespond to two and five nitrogen atoms respectively. Againno attempts were made to determine accurately the occupancynumbers and Debye-Waller factors for these non-bondedshells. However we note that the relatively low occupationnumbers may be an indication of some form of orderedinteraction.Interestingly, incorporation of a single acetonitrilemolecule co-ordinated trans to the sulfur atom at a similardistance to that observed in the crystal structure of the WSF,.MeCN adduct (2.25 A)’ did not result in a significantimprovement to the fit although on the basis of our modellingthe possibility of such a species existing in solution cannot becompletely discounted.Two further checks on the modelling of the data wereperformed. Sequential addition of the shells was tested forstatistical significance using the tests of Joyner et a1.,I9 all theshells passed at the 1% level or better. A check on the primary-shell co-ordination numbers was also undertaken by includingin the iteration process the occupancy of the first and secondshells (4 F and 1 S) and by mapping the occupation number foreach shell against its Debye-Waller factor.A minimum wasobserved for co-ordination numbers of 4.2 fluorine atoms and0.86 sulfur atoms. Although the error associated with thesevalues is believed to be k 10% it has been shown that by carefulanalysis of EXAFS dataz0 it is possible to obtain reasonableconfidence in occupancy numbers. We are, therefore, confidentthat our EXAFS studies show that WOF,, WOCI, and WSF,all adopt monomeric structures in acetonitrile and dichloro-methane solution.It is instructive to compare the bond lengths derived from thevapour-phase electron diffraction study and those obtained inthis work; Table 1 summarises the final parameters obtainedfrom our EXAFS studies.It is obvious that there is goodagreement between the two sets of derived W=O and W-X bondJ. Chem. SOC., Dalton Trans., 1996, Pages 2975-2978 297lengths for both WOF, and WOCl,. For WSF, the W-F bonddistances obtained from our data [1.863(3) A] and from electrondiffraction [1.847(3) A] are similar and consistent with valuesin related tungsten(v1) compounds.21 However we obtain asignificantly shorter W=S bond length [2.026(8) 8, comparedwith 2.104(7) 8, by electron diffraction]. In fact the W=S bondlength derived from our EXAFS study is more in line with thatwhich may be expected from the observed distances in WSBr,and WSCl, and simple electronegativity arguments.’ The trendto decreasing W-S distances for the lighter, more electronegativehalides is also consistent with that observed for the W=O bonddistances of the tetrahalide oxides.This is contrary to thefindings of the electron diffraction study. We also note that thisdistance (2.026 A) is, as expected, now shorter than the W=Sdistance reported for the crystal structure of WSF,-MeCN(2.034 A) and thus mirrors the behaviour of the W=O distance inWOF, (1.66 A) and its adducts (1.77 A).”There is therefore a considerable difference between the W=Sdistance determined in this study and the only other availabledistance reported for monomeric WSF, obtained by the electrondiffraction study. As previously noted the latter experimentsuffered from a moderately high degree of sample decomp-osition; the results were interpreted in terms of a samplecomposition of 23% WOF, and 77% WSF,.This howevershould have had minimum effect on the derived W=S bonddistance since the W=S and W=O distances are quite distinct.There were also reported problems with the vibrationalamplitudes obtained during modelling which resulted in theseparameters having to be constrained to remain withinreasonable limits. * However the most important differencebetween these two sets of experiments was the sample state.The electron diffraction study was carried out on a high-temperature gas-phase species which would be undergoingconsiderable thermal and vibrational motion.ConclusionThis work demonstrates that WSF, may be prepared by thereaction of (Me,Si),S and WF, at or below room temperaturein acetonitrile or dichloromethane solution. The identity of theproducts in solution is confirmed by fluorine NMR studies.Tungsten L(II1)-edge EXAFS studies of solutions of WSF,,WOF, and WOCl, have been made for the first time anddemonstrate that these compounds are monomeric inacetonitrile, dichloromethane and diethyl ether solutions.Fromthe EXAFS data a W=S distance of 2.026(8) 8, and a W-Fdistance of 1.863(3) 8, is derived for WSF, adopting amonomeric structure. This new value for the W=S distance is inline with that expected by comparison with bond distances ofother tungsten chalcogen tetrahalides and their adducts.AcknowledgementsWe thank the Directory of the Daresbury Laboratory for theprovision of facilities and UMIST for financial support.References1 J.H. Holloway and D. Laycock, Adv. Inorg. Chem. Radiochem.,1984,28, 73.2 A. J. Edwards and G. R. Jones, J. Chem. Soc. A, 1968, 2074;I. R. Beattie, K. M. S. Livingstone, D. J. Reynolds and G. A. Ozin,J. Chem. Soc. A, 1970, 1210.3 M. J. Atherton and J. H. Holloway, J. Chem. Soc., Chem. Commun.,1977,424.4 J. H. Holloway, V. Kaucic and D. R. Russell, J. Chem. Soc.. Chem.Commun., 1983, 1079.5 H. Hess and H. Hartung, Z . Anorg. Chem., 1966,344, 157.6 M. G. B. Drew and R. Mandyczewsky, J. Chem. Soc. A , 1970,2815.7 See, for example, M. J. Atherton and J. H. Holloway, Adv. Inorg.Chem. Radiochem., 1979, 22, 171; D. A. Rice, Coord. Chem. Rev.,1978, 25, 199.8 D. A. Rice, K. Hagan, L. Hedberg, K. Hedberg, G. M. Staunton,and J. H. Holloway, Inorg. Chem., 1984,23, 1826.9 T. Moritani, K. Kuchitsu and Y. Morino, Inorg. Chem., 1971, 10,344.10 V. C. Gibson, T. P. Kee and A. Shaw, Polyhedron, 1990, 9,2293.11 G. R. Holzman, P. C. Lauterbur, J. H. Anderson and W. Koth,J. Chem. Phys., 1956,25, 172.12 H. Selig, W. A. Sunder, F. C. Schilling and W. E. Falconer,J. Fluorine Chem., 1978, 11, 629.13 P. C. Crouch, J. Inorg. Nucl. Chem., 1970,32, 333.14 A. K. Brisdon, EX, A program for EXAFS Data Reduction,University of Leicester, 1992.15 N. Binsted, S. J. Gurman and J. W. Cambell, EXCURVE, SERCDaresbury Laboratory Program, 1992.16 S. J. Gurman, N. Binsted and I. Ross, EXCURVE, J. Phys. C, 1984,17, 143; 1986, 19, 1845.17 S. A. Brewer, A. K. Brisdon, J. H. Holloway and E. G. Hope,Polyhedron, 1994, 13, 749.18 S. P. Cramer, K. S. Liang, A. J. Jacobson, C. H. Chang andR. R. Chianelli, Inorg. Chem., 1984, 23, 1215.19 R. W. Joyner, K. J. Martin and P. Meehan, J. Phys. C, 1987, 20,4005.20 I. R. Beattie, P. J. Jones and N. A. Young, J. Am. Chem. Soc., 1992,114,6146.21 A. K. Brisdon, J. H. Holloway, E. G. Hope, W. Levason,J. S. Ogden and A. K. Saad, J. Chem. Soc., Dalton Trans., 1992,139.22 L. Arnaudet, R. Bougon, B. Ban, P. Charpin, J. Isabey, M. Lance,M. Nierlich and J. Vigner, Inorg. Chem., 1989, 28, 257.Received 22nd February 1996; Paper 6/0 1279E2978 J. Chem. Soc., Dalton Trans., 1996, Pages 2975-297