1975 73?Electron Spin Resonance and Raman Spectra of [ReOF,]- and RelatedSpecies in Aqueous Hydrofluoric AcidBy John H. Holloway and J. Barrie Raynor,' Department of Chemistry, The University, Leicester LE1 7RHRaman spectra and electron spin resonance measurements on the products of partial hydrolysis of rhenium hexa-fluoride in aqueous hydrofluoric acid show that the main species is green [ReOF,]-. Assignments of the observedfundamental frequencies from the Raman spectra are as follows: 1008 (vl), 736 (va), 700 (va), 590 (VJ, 575 (VJ.387 (vl,,), 370 (v,), 330 (v4), 298 (v,,), and 233 (v6) cm-l which confirms the expected CI1, symmetry. Thee.s.r. parameters are: gll = 1.72, gl = 1 -74,All (Re) = 0.0960cm-l.Al (Re) = 0.0500 cm-l, and lQ'l = 0.0045cm-l, together with fluorine hyperfine coupling.The bonding parameters are p1 = 0.78. pz = 0.85, and E = 0.92showing considerable covalency. A transient blue species is considered to be an oxygen- or fluorine-bridgedpolymer.RHENIUM HEXAFLUORIDE dissolves quite readily in an-hydrous hydrogen fluoride to give a colourless solution.The introduction of traces of water result in the solutionbecoming pale blue. When slightly larger amounts ofwater are introduced the blue colour is only transient andafter a few minutes the solution turns green.In an investigation by Ippolitov in 1962, the partialhydrolysis of oct afluororhenates gave blue solids whichwere separated and characterized as salts of generalformula M[ReOF,] (11 = K, Rb, or Cs) on the basis ofchemical analysis, magnetic measurements, and Noddakvalency determinations.Solutions in organic solventswere said to remain blue for several hours but those inwater turned green after a few minutes.We have been unable to reproduce part of the Russianwork. In particular, the colour changes described in thevarious modes of hydrolysis of the potassium salt do notoccur in our samples, nor do any of the resulting solutionsyield e.s.r. signals. Whilst our results do not rule out thepossibility that [ReOF,]- salts are blue, the reportedready solubility 1 of the salts in ethyl methyl ketone andthe fact that ReOF, itself is blue suggest the possibilitythat the hydrolysis of M2ReF8 may in fact give rise toReOF, and KF.Our Raman and e.s.r. measurements confirm earlierRaman work which suggested that ReF, in anhydrousHF is very little distorted from the octahedral configur-ation found in the gas phase,3 there being little interactionwith the solvent, and Raman, e.s.r., and optical spectrahave shown that the green solutions contain monomeric[ReOF,]-.We propose that the transient blue colour isdue to the formation of polymeric intermediates such asF F F F \ I \ I-0 - Re-F F I\ F ' 'ForF F F FEXPERIMENTALPure rhenium hexafluoride was obtained by direct fluorin-The reac- ation of rhenium metal with high purity fluorine.tion was carried out in a closed nickel reactor a t 6 atm and300 "C. To avoid contamination of the sample with ReF, asmall excess of rhenium metal was used.4 After purificationby trap-to-trap sublimation, the absence of ReF, was verifiedby the complete absence of bands a t 300 and 355 cm-l in thei.r.spectrum.6 Potassium octafluororhenate was preparedby direct reaction of ReF, with the stoicheiometric amountof potassium fluoride under pressure.Samples were prepared for the various spectroscopicmeasurements by condensing ReF, onto purified anhydrousH F (specific conductivity x 1 x L2-l cm-1) in carefullydried, preseasoned Kel-F (polytrifluorochloroethylene) tubesand cells, and then allowing the ReF, to dissolve slowly as themixture was allowed to warm to room temperature. Theblue and green oxygen-containing solutions were obtainedby condensing varying small amounts of water on to the re-frozen ReF6-HF solutions and allowing these to warm onceagain to room temperature.Raman spectra were measured on a Coderg PH1 instru-ment with 488.0 nm excitation.U.V. measurements weremade using a Unicam SP 700 spectrophotometer. E.s.r.spectra were obtained on a Varian E-3 instrument withfacilities for variable-temperature control.RESULTSRaman Spectra.-The Raman shifts for the colourlessReF6-HF solution a t room temperature are compared withpreviously reported values 2 and those of ReF, vapour inTable 1.We were unable to obtain unambiguous Raman spectra ofthe blue solution because of rapid conversion into the greensolution in the laser beam. The spectrum of the greensolution showed numerous shifts which are listed with theirassignments in Table 1.Bands a t 330, 575, 736, and 1008cm-l are polarised.Optical Spectra.-The spectra of the blue solution ex-hibited three main features: a peak a t 14 300 cm-l, ashoulder a t ca. 16 400 cm-1, and a shoulder a t ca. 29 000 cm-1.It was not possible to measure extinction coefficientsaccurately but the relative intensities of the absorptionswere about 1.25 : 1.0 : 2.5.The spectra of the green solution was significantly differentand consisted of a peak a t 14 000 cm-l, a shoulder a t ca.19 700 cm-l, and a peak a t 25 500 cm-l. Beyond this therewere intense charge-transfer absorptions. The relativeE. G. Ippolitov, Russ. J . Inorg. Cham., 1962,7,486.B. Frlec and H. H. Hyman, Inorg. Chem., 1967, 6, 1696.H. H. Claassen, G. L. Goodman, J.H. Holloway, and H.J. G. Malm and H. Selig, J . Imrg. Nuclear Chem., 1961, 20,B. Weinstock and J. G. Malm, J . Inorg. Nuclear Chem., 1966,Selig, J . Chem. Phys., 1970, 58, 341.189.2, 380738 J.C.S. Daltonintensities of the main absorptions were similar to those ofthe blue solution.E.s.Y. Spectm-The colourless KeF, solution in anhydrousHF exhibited no e.s.r. spectrum at 77 K as expected for amolecule of such high symmetry.The blue solution gave no signal at room temperature anda broad line around 1500 G at 77 K. Upon addition of atrace of water, the spectrum of the green solution wassuperimposed upon the broad line.perpendicular features between 2135 and 5060 G. Theanalysis of the spectrum was aided by a computer simulationwhich yielded the following parameters: Ail (Re) = 0-0960cm-l, A 1 (Re) = 0.0500 cm-l, gll = 1-72, gl = 1.74, IQ'I =0.0045 cm-l.Although rhenium has two nuclei with nuclearspin (ls5Re, 37.07%, px = 3.143 nm, I = 5/2; 1*7Re,62.93%, px = 3.1760 nm, I = 5 / 2 ) , signals from the twoisotopes were never resolved.Under higher resolution numerous additional lines andTABLE 1Raman bands (in cm-l) and their assignments for ReF,, ReOF,, and [ReOFJAssignments[ReOFsI - r---------h---7ReOF, e (This work) c4v o h125vw234vw260 IR309vw P334s367s640m P652m715 IR989.8s P7 3 7 . 6 ~ ~ P233s298s330m, sh P370w387s575w P590w- 7 0 0 ~ , sh1008s P736vs PReF,Ref. a This work Gas147 (calc)257 (calc)247s588m715 (calc)753.7vs249vw578vw757vsa Ref.2. * Ref. 3. c Ref. 7. P Polarised band. IR Observed only in the i.r. spectrum.No spectrum for the green solution was observed at roomtemperature, but as the temperature was lowered six broadlines appeared, which were narrowest (ca. 350 G wide) atGU. 176 K. These were attributed to hyperfine coupling toI I 1 1 I IFIGURE 1 E.s.r. spectrum of the green solution at 77 Kthe rhenium. The field positions of the lines were asfollows: ca. 1500, 2060, 2730, 3475, 4475, and 5670 G andthe hyperfine coupling, after correction to second order,6 wascalculated to be 734 G with gav = 1.800. The line at lowestfield was obscured by an absorption at ca. 1500 G whichcould possibly be attributed to a Am* = 2 transition ofresidual blue species.At 77 K, a complex spectrum was obtained (Figure 1)which was analysed in terms of an axially symmetricalsystem, Six well-resolved parallel features were observedat field positions between 738 and 6665 G and corresponding246m587vw756vsshoulders appeared and were attributed to superhyperfinecoupling to fluorine atoms.On the lowest-field rheniumparallel feature, the second-derivative spectrum clearlyshowed a multiplet of five equally spaced lines with a separ-ation of 33 G and relative intensities ca. 1 : 4 : 6 : 4 : 1. Weassign these features to interaction with four equivalentequatoral fluorine atoms. There was no evidence of furtherinteraction with an axial fluorine atom, thus indicating thatsuch coupling, if any, was small.The lowest-field rheniumperpendicular feature also exhibited fairly well resolvedsuperhypefine structure showing five features with anapproximately equal spacing of 45 G. The relative inten-sities and spacings were slightly distorted because of thegeneral shape of perpendicular features of a powder spec-trum.I)ISCUSSIONThe Colourless ReF, in HF.-The three vibrations(Table 1) are readily assigned to the vl, v2, and v5 funda-mentals in Oh symmetry since these are the only Raman-active modes. The close similarity between the gas-phase and solution spectra is strong evidence that themolecular structure in solution is similar to that in thegas phase, viz. discrete molecules. Our spectra arebetter resolved than those previously published.2The assignmentof the Raman shifts was aided by the published assign-ments for ReOF, 7 and ReF,.3 The main problem wasto decide whether the species in solution was ReOF, or[ReOFJ-.Table 1 correlates the vibrational modes forReF, and ReOF,. Our observation that the bands at330, 575, 736, and 1008 cm-1 are polarised allows us toassign them to the v4, v3, %, and v1 fundamentals by com-parison with the assignments for ReOF,. Other bandsB. Rleaney, Phil. Mag., 1951, 42, (7), 441.J . H. Holloway, H. Selig, and H. Claassen, J. Chem. Phys.,The Green Solution.-Raman spectra.1971, 54, 43051975 739are assigned by analogy with ReOF, and KeF,. Had thespecies in solution been ReOF,, then the v3 and vl0fundamentals of [ReOF,]- would be absent.The un-ambiguous detection of four polarised bands confirmsthat the species is [ReOF,]-.By analogy with the correspondingchromium and molybdenum oxide pentafl uoride anions,we assign the unpaired electron to a b2(d,) orbital inC4u symmetry. If we take the relative order of themolecular-orbital energy levels as that for other dl metaloxide halide complexess79 then b2(d,) < e(d,,,) <bl(d5*-yy4) < al(dZa). The co-ordinate scheme is givenin Figure 2, where X , Y , and 2 represent the molecular-axis scheme and x, y, and x the local-axis scheme at eachfluoride, where the x axis is along each metal-ligandbond. 8 And 4 are spherical polar angles which relatethe external magnetic-field vector H to the 2 and Y axesrespectively. The orbitals necessary for the followingdiscussion are given below, where Pl, P2, and E are orbitalcoefficients.The e.s.r.g-tensor.$P2 = P2$(d,) - P2%Y$PI = P,$(dZ8-Yd - PlYLZ$E = E$(dzz,,z) - E‘$Lz(1)(2)(3)#LY Represents the ligand p , orbitals which can overlapwith the dzy orbital by in-plane x-bonding, $L, representsthe ligand 9, orbitals associated with out-of-plane x-bonding and $ L ~ represents the ligand 9, orbitals asso-ciated with the o-bonding.First-order perturbation theory predicts that gll <gl < 2, and this is true for many dl oxide pentahalideions. However, for some, notably those with chlorideor bromide as the halide, gl and gll are reversed. Mano-haran and RogerslO have explained this in terms of acontribution to the g-tensor by a spin-orbit interactionwith ligands having a large spin-orbit parameter.Ourresults follow the first-order predictions and are in linewith those of other oxide pentafluorides (Table 2).Optical spectra. The three weak absorptions at 14 000,19 700, and 25 500 cm-l are assigned to the three orbitallyforbidden transitions which are expected to be weaklyobserved, namely b2+e, b2+b1, and b2+al respectively.The magnitude of the energy of these transitions isreasonable by comparison with the careful assignmentsof Wentworth and Piper on single crystals of [VOC1,]3-using polarised light.The rhenium hyperfine tensor. The signs of the experi-mental hyperfine couplings to rhenium are all taken to benegative since this is the only combination of signs whichgives a negative isotropic hyperfine coupling, as almostalways found for transition-metal ions, and a negativeprincipal value of the anisotropic tensor, as expected fora b, ground state.ll In atoms where suitable wave-functions are available, values can be calculated for theR.A. D. Wentworth and T. S. Piper, J. Chem. Phys., 1964,K. de Armond, B. R. Garrett, and H. S. Gutowsky, J. Chem.lo P. T. Manoharan and M. T. Rogers, J. Chewz. Phys., 1968,49,41, 3884.Phys., 1965, 42, 1019.5510.expected hyperfine coupling assuming 100% occupancyof a 5d orbital. Such values are available for the 3d and4d ions in many electronic configurations,12 but for rhen-ium it is only known for the 5ds configuration, (ls5Re+,-154.8 G).Since the value increases considerably aselectrons are removed, then our value of -307 xcm-l is entirely reasonable. Unfortunately, without areliable value for rhenium in a dl configuration, we cannotestimate the electron population in the Sd,, orbital bythis m e t h ~ d . ~In order to calculate thebonding coefficients of the molecular orbitals 1-3 usingthe measured e.s.r. parameters, we have followed thebasic theory of the g- and metal hyperfine-tensorsdeveloped by Abragam and Pryce,13 and used the expres-sions given by Manoharan and Rogers and followed theirnotation :The bonding parameters.Sb,, etc. are overlap integrals between the metal orbitaland ligand orbitals of the same symmetry. We wereunable to evaluate these since no proper radial wave-functions for the higher ionic species of rhenium areavailable. We have used the values estimated byManoharan and Rogers for [MoOF,]2-, namely Sb, =0.12, S, = 0.16, sb, = 0-20.It is unlikely they will begreatly different from this, and in any case Manoharanand Rogers have shown that changes of up to &0.04 makelittle significant difference to calculated bonding co-efficients. The only other equations required are of thetypePI2 + - 2P1P,’Sbl = 1 (8)which allows calculation of the orbital coefficient of theligands (dashed). The spin-orbit coupling constant forfluorine, h~ = 272 cm-l,l* whilst that for rhenium isl1 B. A. Goodman and J. B. Raynor, Adv. Inorg. Radiochem.,l2 B. A. Goodman and J.B. Raynor, J. Inorg. Nuclear Chem.,l3 A. AbragamandM. H. L. Pryce, Proc. Roy. Soc., 1951, A205,l4 D. S. McClure, J. Chem. Phys., 1949, 17, 905.1970, 13, 135.1970,32, 3406.135; 1951, A M , 164740 J.C.S. Daltonsolutions and our interpretation of tlie fluorine super-hyperfine couplings is less certain than might otherwisehave been the case.The axial symmetry of the g-tensor shows that all fourequatorial fluorine atoms are equivalent. The hyperfinetensor for each fluorine has to be considered in terms ofthe local symmetry axes as defined in Figure 2. Thesplitting of the rhenium parallel features in the spectruminto five equally spaced lines and appropriate intensityratios confirms the equivalence of the four fluorines.The hyperfine coupling of 33 G is the z component of tlietaken as 4200 cm-l based upon the known value forRe6+.We now have sufficient information to solve equations(4)-(7) for the bonding coefficients. However, thefunctional forms are complicated by the second-orderterms in equations (4)-(7).Fortunately, these arerelatively small and they can be approximated by aniterative procedure in which the second-order terms areinitially ignored , the bonding coefficients calculated tofirst order, and these coefficients then used to approxi-mate the second-order terms for the final calculation.P Was taken as 0.046 cm-l.llTABLE 2g-Tensor ant1 fluorine hyperfine couplings in various d1 complexes (units >: 1O-O cm-l cxccpt where G = gauss)[Mo01;,]*- [MoOF,]*- iTiF.J+iCrOF,]- [CrO I?,] 2- sol.crystal [wOl;,J*- in CH,OH [\-0FJ2- [ReOF,j -&I 1.959 1.959 1.874 1.874 1.566 1.944 1.764gl 1.968 1.968 1.91 1 1.91 1 1.685 1.988 1.816A d e 9 ) 5.87 6.68 13.1 7 GA , - 22.97 - 22-97 - 24.98 - 20.1A , + 36.75 + 36.76 + 50.6 1 f63.1A* - 6.8 - 4.4 + 3.7A(ax) 1-74 0} 46Gav } 10G- 16.3 -33 Ga C I1 This work a, d , f , g a, e a, g g, b Ref.(I Ref. 10. 13. S. Xbradakhmanov, N. S. Garif’yanov, and E. I. Scmcnova, Zlruv. Strukt. Khinr., 1968,9,463. c E. L. WctersandA. H. lMaki, Phys. Rev., 1962, 125, 233. c N. S. Garif’yanov, V. S .Fedotov, and N. S. Kucheryavenko, Bull. Acad. Sci. U.S.S.R. Chem. Div., 1964,4, 689; D. I . Ryabchikov, I. N. IIarov, Y. N. Dub-rov, V. K. Belyaeva, and A. N.Ermakov, Doklady Akad. Nauk. S.S.S.R., 1966,169, 1107; J. T. C. Van Kemenade, J. L. Verbeek,and P. R. Cornaz, Rec. Trav. chim., 1966,85,629. J. T. C. VanKemenade, Rec. Trav. chim., 1970,89, 1100. * G. M. Larin, Y. U. Buslaev, M. E. Dyatkina, and 1. V. Miroshnichenko, Zhur. Slvrrkl.Khim., 1969, 10, 993.d N. S. Garif’yanov, Doklady Pliys. Client., 1964, 155, 24?.S. L. Verbeek and P. I;. Cornaz, Rec. Trav. chim., 1967,86, 209.Repetition of the procedure resulted in the followingbonding coefficients :p1 = 0.78 pl’ = 0.80p2 = 0*85 p2’ = 0.64E = 0.92 8’ = 0.57As a consequence of the various assumptions andapproximations, these values will not be highly accuratebut they should be of the correct order of magnitude.The most significant thing here is the virtually purecovalent x-bonding in the molecule.We will comparethese values with related complexes later in this paper.ItReFIGURE 2 Axis notation for [ReOl:,,-Superhyperjne coupling to @mine. Because of thedifficulty of manipulating the solutions, we did notattempt to grow magnetically dilute single crystals.As a consequence, measurements were made on frozenexperimental tensor. Determination of the separate xand y components is not possible directly since the super-hyperfine coupling on the rhenium perpendicular featuresrepresents the average value ( x + y)/2 as has been shownin certain square planar cupric complexes.ls However,the average ( A , + A , + A,)/3 is A*,, and can be com-puted since A , = -+33 G, ( A , + Ay)/2 = &45 G ; thenthe four possible combinations are :(a) ,4, = +33 G, A , + A , = +90 G, Az90 = +41 G(b) A , = -33 G, A , + A , = +9O G, Ado = +19 G(c) A , = -33 G, A , + A , = -90 G, A i s o = -41 G(d) A , = +33 G, A , + 14, = -90 G, Ri.ro = -19 GIt is difficult to decide which is more likely, but sinceknown values of Aho for other transition-metai d1fluorides do not exceed 11 G (Table 2), and that a positivevalue is expected,*J6 then the value (+19) is the mostprobable and will be assumed in subsequent discussion.Since there will be positive spin density in the fluorine p ,orbital because of the direct overlap with the d , orbital,then the amount of this can be calculated directly fromthe bonding parameters.The total unpaired-electrondensity in all four p , orbitals of the equatoral fluorines isgiven by the Mulliken population analysis to be equal to(p12)2 - p2p2Sb,.lS For our case, it equals 0.34 or 8.5%Is C.M. Guzy, J . B. Raynor, and 31. C. It. Symons, J. Chew.SOC. ( A ) , 1969, 2329; J . B. Raynor, Z. Naturfovsch., 1969, 24b,776.l8 l<. S. Mullikcn, J. Chem. Phys., 1955, 25, 18331975 741in each eY orbital, from which we can estimate themagnitude o f the anisotropic tensor in the y direction.Since lWy0 occupancy of a p orbital on fluorine yields asplitting of -+lo85 G, then by proportion the tensorcomponent A,, n-ould be +92 G.e.s.r. spectrum at 77 K and showing three opticalabsorptions at different energies. We have strong e.s.r.evidence that the green solution consists of monomeric[ReOF,]- ions and propose that the blue solution con-tains polymeric O-ReF,-O-FeR,-, O-ReF,-F-ReF,-,TABLE 3Bonding paraiiieters for some conipleses related to [lieOl;,]-[Re0F5j - [\'OC1,]3- [\'O(HZO)~~~' [CrOC1,I2- [MoOF,]*- [MoOC~,]~- [MoOBr,12-0-78 <1 0.981 0.692 0-89 1 0.764 0.6961 0.886 0-966 0.906 0.932 P IL 0.90 1 0.844P 2 0.86 < l0.92 0.992 0.962 0-968 0.960Ref.This work b b c U a cL. A. Dalton. H. D. Bereman, and C. H. Brubaker, Inovg. Clreni., 1969, 8, 2477. * Ref. 10. Ref. 9.Since A , = -33 G and A i s o = +19 G, then the tensorcomponent A , = A , - Aho = -52 G. Also, A , =+92 G, therefore A , = -40 G. This tensor needs tobe corrected for the dipolar interaction of the electron inthe dq orbital by using the point-dipole approximationAdip = 2 g ~ p ~ / @ where r is taken as 1.85 x astypical of a metal-fluorine bond length.17 The principalvalue of the dipolar tensor is then +S G and directedalong the x axis.The corrected hyperfine tensor is nowA , = -48 G, A , = +96 G, A , = -48 G. Unpairedelectron density reaches the 9, orbital by direct in-planez-bonding with the metal dzv orbital. However, adifferent mechanism is needed to account for theanisotropic dipolar interaction in the p, and 9, orbitalsand also for the apparent spin density in the s orbitalsof fluorine. The most likely explanation is via spinpolarisation of the f l e d b, and e bonding molecularorbitals for the 9, and 9, orbitals (to yield negative spindensity) and either via polarisation of the filled b, or bypolarisation of the 2s and 29, fluorine orbitals by spindensity in fir which will lead to positive spin density inthe 2s orbital.8.16No interaction was observed withthe axial fluorine.This is not surprising since there isno mechanism for direct delocalisation of electron densitynor much chance of the spin-polarisation mechanismyielding hyperfine couplings large enough to be seen.Comparison with other complexes. Table 3 lists thebonding parameters for some related complexes. Thebonding in [ReOF,]- is more covalent than in [MoOFJ2-but rather similar to that in [MoOC1,]2-. This reflectsthe larger radial distribution of the dq orbital in Re andC1 compared with Mo and F. The most striking featureis the covalency of the a-bond in [ReOF,]-.Very few e.s.r. studies have been made on other d1rhenium complexes and none in sufficient detail to makemeaningful comparisons.18The Blue SoZzction.-The blue solution differs signifi-cantly from the green solution in exhibiting no resolvedComm., 1973, 321.The axidjuorine.l7 D. Bruce, J . H. Holloway, and D. R. Russell, J.C.S. Chem.or [O-ReF4-O-ReF4-0I2- units. The close proximityof the rhenium atoms would permit an antiferromagneticinteraction between the electrons and prevent observ-ation of resolved e.s.r. spectrum at 77 K. Our observ-ation of a possible Ams = 2 transition supports this.The optical spectra can now be explained if we say that theblue polymeric species has a larger axial distortion thanthe monomeric [ReOFJ-. The effect upon the energylevels is seen in Figure 3. Since the e and b, levels inReF6 IReOF;] -0-7,e-0- \ /FIGURE 3 Order of the energ levels for ReF,, [ReOFJ-,and -0-Re-O- T// \[ReOF,]- are relatively close (ca. 5000 cm-l), then anincreased axial distortion would increase the separationof the b, and e orbitals and also that of the b, and a,orbitals. The effect could be a reversal of the b and eenergy levels in the blue species.We are indebted to Professors I. N. Marov and R. D.Peacock for carrying out some preliminary measurementsand we thank the Royal Society for an Equipment Grant.[3/1043 Received, 18th May, 19733l8 N. S. Garif'yanov, Imest. Akad. Nauk. S.S.S.R. Ser khim.,1968, 8, 1902; N. S. Garif'yanov, Soviet Phys. JETP., 1963, 18,1246; E. I. Steifel and H. B. Gray, J. Amer. Chem. Soc., 1966,87,4012