J . CHEM. SOC. DALTON TRANS. 1986 1663Cobalt(i1) Halide Complexes in Aqueous Ammonium Nitrate-Calcium NitrateMelts: Electronic Spectra and Solvent Extraction tDavid H. KerridgeDepartment of Chemistry, The University, Southampton SO9 5NHRurica NikoliC and Dragica StojicThe Boris Kidri6 Institute of Nuclear Sciences, Belgrade, YugoslaviaAddition of chloride or bromide to cobalt(ii) dissolved in a low-temperature aqueous nitrate melt[xN H,NO,*yCa( NO,),~zH,O] changed the electronic absorption spectrum from near-octahedralco-ordination by both nitrate and water to a tetrahedral (or severely distorted octahedral) symmetrywith between two and three halide ligands, the bromide species containing less water. The effectsof varying the concentrations and of increasing temperature are discussed.Solvent-extractionmeasurements were made and the spectra of the organic and salt phases recorded.Spectroscopic investigations of the co-ordination of 'probecations' have been frequently used for insight into meltstructures; the changes occurring with temperature and withvarying concentrations of a second component have both beenused, when variable co-ordination is found, to show whether thechange is one of gradual distortion or due to two coexistinggeometrically distinct species.'q2Cobalt(1r) has often been chosen for such work since it gives astriking red shift to the principal visible absorption band onchanging from octahedral to tetrahedral co-ordination (thelatter exhibits well defined fine structure).This change is clearlyseen in aqueous solutions, when for example concentratedhydrochloric acid is added, and is commonly illustrated in text-b o o k ~ . ~ The analogous change also occurs in molten nitratesolution upon addition of chloride, as shown by the classic studyof Gruen rt d4Similar changes in co-ordination occur when bromide oriodide is added to cobalt(r1) solutions in the lithium nitrate-potassium nitrate eutectic,' though the initial co-ordination inpure nitrate was later claimed to be dodecahedra16 becauseof the spectral similarity to tetranitratocobaltate(I1) whosestructure had been determined by X-ray diffra~tion,~ and onthe basis of resolution of the spectrum into five bands.*Volkov and Buryak showed that complete conversion intothe tetrachloro complex required an extremely large excess( x 5&1 OOO) of added chloride and that alkali-metalchlorides give increased absorption coefficients withincreasing cation radius.applied a computationaltreatment to their spectral curves and were able to calculate thestability constants of the mono-, di-, tri-, and tetra-halogenospecies, using added chloride and also bromide, while showingadded fluoride caused no spectral change, but could notdetermine the number of co-ordinated nitrate anions nor didthey consider their results accurate enough to specify wherethe change from dodecahedral to tetrahedral co-ordinationoccurred; however, Griffiths and Potts l 1 have attempted thiswith the analogous nickel(1r) species.Similar spectral changeswere observed when cobalt(i1) was dissolved in moltenammonium nitrate with and without added chloride.'Analogous changes in co-ordination have been reported forHemmingsson and Holmberg *t Supplementary data available (No. SUP 56549.15 pp.): additionalspectra of cobalt(i1). See Instructions for Authors, J . Chem. Soc., DaltonTrans.. 1986, Issue 1, pp. xvii-xx.other oxyanion melts, particularly the change from octahedralto pseudo-tetrahedral to octahedral and finally, to tetrahedralin potassium sulphate-zinc sulphate glasses with steadilyincreasing ratios of zinc chloride.' Subsequently, in puresulphate melts, cobalt(I1) was shown to be dodecahedral l 4 oroctahedral, changing to dodecahedral at higher temperature.' 'In acetate, the co-ordination of cobalt(I1) changes from octa-hedral to tetrahedral as the proportion of potassium increasesin lead acetate-potassium acetate glasses," though in alkali-metal acetate melts the co-ordination has been reported both astetrahedral l 7 and as dodecahedral.'* The results from purelychloride melts are rather less relevant but may be summarised asoctahedrally co-ordinated cobalt(I1) in alkali-metal chlorides,'distorted tetrahedral in magnesium chloride,20 and tetrahedralin lead(II), tin(rI), and bismuth(II1) chlorides; " in aluminiumtrichloride the co-ordination is octahedral, changing todistorted octahedral and then to tetrahedral as potassiumchloride is added.20However, in all the above systems there have been no morethan two potential ligands to consider.It thus seemed timely toconsider the potential competition for co-ordination to the'probe cation', cobalt(Ir), when there are three potential ligands.The only such system reported so far is cobalt(I1) in a glass [ofcomposition '6KNO3-4Ca(NO3), + 24 mol% H 2 0 doped with1 mol% KSCN'] where the composition ratio was not changedthough the spectrum was determined at two temperatures (20and 50 oC).22A series of spectral measurements of cobalt(r1) in low-meltinghydrated nitrate melts [ x N H ~ N O , ~ ~ C ~ ( N O , ) ~ - ~ H ~ O ] withadded halide were therefore undertaken as a follow-up to thestudies of complex formation using e.m.f.23--25 and solventextraction 26 measurements. The high solubility of these nitratesin water enables their study at low water concentrations andsuch 'aqueous melts' thus have incomplete hydration spheresaround the ions and are therefore intermediate betweenaqueous electrolyte solutions and anhydrous ionic moltensalts.The water concentration in the two water-rich mixturesused corresponds to the same hydration number (4.35) forthe cobalt(r1) ion, according to the simplified calculationsof Gal el the other to a hydration number of3.23.Some metal-ligand association equilibria in these solutionsinvestigated earlier showed both regular and reversed ordersof halide complexation with highly polarised bivalent metalions,23-26 but no spectroscopy has so far been reported inaqueous melts as opposed to glasses1664 J.CHEM. SOC. DALTON TRANS. 1986Table. Slopes derived by limiting-logarithm method (ref. 31)Mole ratio Concentration (mol dm-') Limiting slopeAf \ f A Temperature A \NH,NO, Ca(NO,), H 2 0 co2 + X- ("C) co2 + CI - Br-0.60.60.60.60.70.70.60.70.70.70.40.40.40.40.30.30.40.30.30.3I .251.251.251.251.0750.801.25I .0750.800.800 . 0 0 2 4 . 0 100.00700.00700.00700.00280.00700.00750.00270.00700.00250.570 . 0 7 4 570 . 0 7 4 . 5 70 . 0 7 4 . 5 80.24-0. 5 80.07-0.750.46- 1.06O . M . 8 50.24-0.890.43-1.01929280619292929292921.152.52.72.81.51.92.52.32.93. IExperimentalSpectrophotometric Measurements.-Solvents of the desiredcomposition were prepared from known weights of carefullydried salts (reagent grade Merck products) and distilled water.Calcium nitrate was dried for 24 h at 220 "C, while ammoniumnitrate was dried under vacuum at d80 "C, in order to avoidpossible decomposition above its melting point.Cobalt(l1)nitrate hexahydrate (reagent grade), anhydrous ammoniumchloride or ammonium bromide were added to the melts toobtain the desired concentrations. The mixtures were premeltedat 80 "C in stoppered glass vessels and rectangular quartzspectrophotometric cells of 10-mm pathlength with ground-glass stoppers were filled with the melts, using preheated glasspipettes.Spectra were recorded on a Beckman DK-1A recordingspectrophotometer, with a thermostatted cell compartment.The temperature in the cells was maintained within k0.7 "C.The reference cell was always filled either with pure solvent ofthe corresponding composition, or where stated, with a solutioncontaining the corresponding concentration of cobalt(i1) nitratebut without ligand addition.Spectra were recorded atwavelengths between 400 and 750 nm, the baseline beingchecked immediately before spectra were recorded.All concentrations in spectrophotometric measurements wereexpressed in molarities. The densities of the melts were deter-mined picnometrically, so that molar absorption coefficientscould be calculated.Distribution Measurements.-The melt phase in the dis-tribution experiments was prepared in the same way as forspectrophotometric measurements, except that cobalt(i1) nitratesolutions were labelled with 6oCo.The organic phase was asolution of tri-n-butyl phosphate (0.2 mol dm-j) in an eutecticmixture of biphenyl and naphthalene (55 mol% biphenyl, m.p.The technique of phase equilibration and sampling isdescribed el~ewhere.'~ The radioactive tracer concentration wascounted in both phases. The molal distribution coefficient wascalculated as an average of at least two independent sampleequilibrations, which were in agreement within & 2%.I n the preliminary experiments it was shown that the puresolvent (eutectic mixture of biphenyl and naphthalene) did notextract cobalt(1i) either from nitrate melts or from melts withvarious ligand concentrations added.The molal distributioncoefficient of cobalt(1r) was determined as a function of thechloride and bromide ligand concentrations in the melt phase.The distribution of cobalt(i1) species between the two phaseswas also followed spectrophotometrically. A few distributionexperiments were performed without radioactive traceraddition and the spectra of both the organic and inorganicphases were recorded. Pure organic and inorganic solvents were39.5 "C Z8).used as corresponding reference solutions. The spectra of bothphases were recorded with and without ligand addition in themelt phase before equilibration.ResultsCobalt(i1) spectra were determined at temperatures from 6 1to 92 "C in melts with three initial ammoniumsalcium-waterratios, and though chloride and bromide were added asammonium halides the additional ammonium had an insignifi-cant effect on the spectra.Representative spectra at increasingchloride ratios are given in Figures 1 and 2 and at increasingbromide ratios in Figures 3 and 4. The effect of temperature isillustrated in Figure 5 and of water concentration in Figures6 and 7. In each case a series of cobalt concentrations werestudied (the complete results are too extensive to list here butcopies of all spectra were available as SUP 56549). Plots of logabsorbance at constant chloride concentrations versus logcobalt concentration at the wavelengths of the absorbancemaxima gave straight lines indicating that Beers law wasobeyed.Attempts were made to calculate the stability constants ofindividual halogeno-substituted monomeric cobalt complexesand of their absorption coefficients by using the LETAGROP-SPEFO program employed so successfully by Hemmingssonand Holmberg l o for cobalt halide complexes in the anhydrouslithium nitrate-potassium nitrate eutectic, though in the presentcase without success.Likewise a newly developed modificationof the STEPIT program, which was kindly made a~ailable,~'also gave no consistent results. The older graphical limiting-logarithm method of Bent and French j1 was therefore used,with the results shown in the Table. Plots of log absorbanceversus log halide molality gave consistent results at wavelengthswhere there was no absorption due to regular 'octahedral'species.Solvent extraction showed cobalt(I1) complexes to be presentin both phases, the spectra of the organic and of the salt phase(after equilibration with 0.5 mol dm-j chloride or bromide)together with that of cobalt(1r) nitrate without added halidebeing given in Figures 8 and 9.The dependence of the distri-bution of cobalt(~r) on halide concentration is given in Figure 10.DiscussionIt is clear that in these hydrate melts chloride and bromideanions cause a change in co-ordination of cobalt(i1) which isconventionally referred to as from octahedral to tetrahedral(Figures 1 4 ) and that the values of A,,,, show a blue shift fromthe values in anhydrous melts, thus indicating co-ordination bywater as well as by nitrate, at least in the halide-free systems.TheOctahedral maximum (Figures 1 and 2) appeared at 525 nmJ. CHEM. soc. DALTON TRANS. 1986 1665450 550 A,nm 650 750Figure 1. Variation of the spectra of cobalt(ii) C(2.75 2 0.05) x 10F3rnol dm Co(NO,),] in NH,NO,-Ca(NO,),-H,O (mole ratio0.7:0.3: 1.075) with added NH,CI: zero (0), 0.437 ( l ) , 0.555 (2), 0.717(3). 0.809 (4), and 1.010 mol d m 3 (5)Figure 3. Variation of the spectra of cobalt(rr) C(7.0 k 0.1) x lo-, moldm Co(NO,),] in NH,NO,Xa(NO,),-H,O (mole ratio0.7:0.3:0.8) at 92 C with added NH,Br: 0.399 (I), 0.477 (2), 0.558 (3),0.607 (4), 0.665 (5). 0.71 1 (6), 0.726 (7), and 0.851 mol dm (8)L50 550 A / n m 650 750Figure 2. Variation as in Figure 1, except (7.0 k 0.1) x 10 rnoldm-3 Co(NO,), in NH,NO,-Ca(NO,),-H,O (mole ratio0.6:0.4: 1.25) at 92 “C.Added NH,Cl: zero ( O ) , 0.070 ( I ) , 0.136 (2),0.208 (3), 0.272 (4), 0.355 (5). 0.445 (6), and 0.569 rnol drn (7)AlnmFigure 4. Variation as in Figure 3, except (7.7 f 1.0) x rnol dm-,Co(NO,), in NH,NO,Xa(NO,),-H,O (mole ratio 0.6:0.4: 1.25).Added NH,Br: zero (01, 0.070 (11, 0.136 (2), 0.206 (3), 0.275 (4). 0.358( 5 ) - 0.449 (6). 0.475 (7), 0.510 (81, 0.547 (9), 0.575 (lo), 0.610 ( 1 1). 0.677(1 2), 0.729 ( 1 3), and 0.750 rnol dm-3 (1416660 . 6 -S 0 . Ld nLJ. CHEM. SOC. DALTON TRANS. 1986-L 50 550 6 50 750AlnmFigure 5. Variation of spectra with temperature. Curves: (1-3), 2.7 xlo-, rnol dm-, Co(NO,), in NH,N0,-Ca(N0,),-H20 (mole ratio0.7:0.3: 1.075) with 1.0 rnol dm-, NH,Cl; (4-6), 2.6 x lo-, mol dm-,Co(NO,), in NH,NO,-Ca(NO,),-H,O (0.6:0.4: 1.25) with 0.59 moldm-, NH,Br.Temperature: 61 (1,4), 80 (2,5), and 92 "C (3,6)A 0YI na0 . 2L 50 550 6 5 0 750AlnmFigure 7. Variation as in Figure 6, except 7.0 xCo(NO,), with 0.75 mol dm-, NH,Brmol dm-,L0 2 0 .a0 .L50 550 650 750XlnmFigure 9. Spectra of the organic phase after equilibration. Details as inFigure 8650 750550 Alnm 4 5 0Figure 6. Variation of the spectra of cobalt(i1) C6.9 x l e 3 mol dm-,Co(NO,),] in NH,NO,-Ca(NO,),-H,O (mole ratio 0.6:0.4: Z) with0.56 rnol dm-, NH,CI at 92 "C. Z = 1.25 (I), 1.50 (2), 1.65 (3), 1.73 (4),and 2.00 (5)4 5 0 550 650 750hlnmFigure 8. Spectra of the salt phase after equilibration at 92 'C. Initialconcentrations: 1.33 x lo-' mol dm Co(NO,),.Curves: 1, no C1- orBr-; 2, 0.5 rnol dm-3 NH,Br; 3, 0.5 mol dm-, NH,Cl0 . 2 5 1 . 0- 0 . 00 . 600 0- 0 . 4- 0.210-3 10-2 l o - ' 1~ ~ - l / m o l d m - 3Figure 10. Dependence of the distribution coefficient ( D ) of cobalt(r~) o nNH,X concentration; X = Cl (a), Br (A). Initial composition of saltphase: NH,N0,-Ca(N0,)2-H,0 (mole ratio 0.6:0.4: 1.25),1.33 x 10 mol dm-3 Co(NO,)J. CHEM. soc. DALTON TRANS. 1986 1667midway between the 510 nm in aqueous solutions and 550 nmin anhydrous LiN0,-KNO, at 180 0C,4 though co-ordinationby as many as three water molecules is not necessarily impliedsince some shift due a change in cation and in temperature couldalso have occurred.The tetrahedral maximum is also at a slightly shorterwavelength than in the corresponding pure halide melt (660 nmwhenchlorideis present, Figures 1 and 2,asagainst 710,4695,19 or695-71 5 nm 2o in anhydrous chloride), again indicating someco-ordination by nitrate and/or by water (Amaz.at 660 nm inLiN03-KNO, with 0.48 mol dm-, C1-, where on averageone nitrate is co-ordinated lo). Similarly with bromide thetetrahedral maximum is at 695 nm (Figures 3 and 4), ascompared to 720 nm in the pure bromide melt." Bromide-containing solutions show three maxima plus two shoulders asopposed to two maxima plus one shoulder with chloride-containing melts. The greater number of peaks may well indicatethe simultaneous presence of two complexes at significant con-centrations and thus less different values of successive stabilityconstants in the case of the bromide spectra, as found earlier withanhydrous nitrate melts.'' Water would also appear to beinvolved in the co-ordination process since increasing the waterconcentrations causes a reduction in the absorption coefficientof the tetrahedral maximum (Figures 6 and 7), though there wasno significant change in the wavelength of this maximum.The multiple absorptions with maxima at 660 nm for chlorideand 695 nm for bromide referred to above as 'tetrahedral, couldalso be attributed to severely distorted octahedral species [cJ:Co" in KCl-AlCl, (49.9: 50.1 m01%)].~' Certainly some inter-mediate octahedrally co-ordinated species might be expected ashalide is substituted, before the eventual production of tetra-hedral species, as suggested by many author^.'*^.^*^*^^ As iscustomarily found, 'tetrahedral' co-ordination (as indicated bythe size of the absorption coefficient) increases with halideconcentration (Figures 14).An increase in temperature alsocauses an increase in this absorption coefficient (Figure 5),indicating either that the tetrahedral complex is relatively morethermally stable, or that the absorbing species is actuallyseverely distorted octahedral, because octahedral species, incontrast to tetrahedral, have increased absorption coefficientswith increasing temperat~re.~~Families of curves (e.g. Figures 1 4 ) are ideally treated bycomputation to obtain a series of stepwise stability constants forincreasing halide substitution, exemplified by the studies ofHemmingsson and Holmberg." However with the present databoth the programs used in the latter studies and one recentlydeveloped by Holmberg,' gave values for the stability con-stants and for the absorption coefficients which could not beoptimised satisfactorily, primarily because the range of halideconcentration was insufficient (the maximum halide concentr-ation used in each case is close to the solubility limit forammonium halide in the solvent).Nevertheless it must be bornein mind that the programs assume only one ligand activity to bechanging, in this case halide, so that the failure to obtainconsistent results may thus also indicate changing water and/ornitrate activities within the sets of spectra.However the limitingslope plots (Table) indicated that the complexes were mono-nuclear in cobalt(r1) and contained between two and threehalides at highest ligand concentration (ie. at the solubility limitof the halides in the particular solvent).The number of halide ligands in the limiting complex can beseen to decrease, as might have been anticipated, with increasingtemperature, cobalt concentration, water concentration, andammonium-calcium ratio at constant hydration number. Thelimiting slope for nitrate ligands indicated virtually no nitrateco-ordination and thus co-ordination by one water molecule,if the absorbing species is four-co-ordinate tetrahedral, inreasonable agreement with the absorption maxima found.Under comparable conditions the number of bromide ligandswas greater than that of chloride, though it should be noted thatthe solubility of ammonium bromide was approximately 50%higher than that of ammonium chloride, and thus that bromidecomplexes tended to contain less water than chloride com-plexes, as expected from electrostatic and steric considerations.The finding that bromide co-ordinates to a greater extent is infact the reverse of the trend found with anhydrous melts." Inthis connection it is relevant that the order of the stabilityconstants for chloride and bromide complexes of zinc(rr), a smallhighly polarising transition-metal cation like cobalt(Ir), was alsoreversed on going from aqueous to anhydrous melts.25The results obtained by solvent extraction broadly supportthe conclusions from the spectroscopic measurements. Thehigher distribution coefficients for cobalt(I1) bromides than forthe chlorides indicate the higher solubility of the former in theorganic phase, largely due to lower water co-ordination.Thedistribution coefficient for the chlorides begins to decrease atthe same chloride concentration (0.2 mol dm-,) at which thesignificant change from octahedral absorption is observed(Figure 2), i.e. a concentration where negatively charged speciesare formed, which are not extractable by tri-n-butyl phos-However, with bromide, no decrease in distributioncoefficient was observed even at the higher possible ammoniumbromide concentration, probably because of the presence of oneor more complexes with a high solubility in the organic phase,whose concentration even increases as the bromide concentr-ation reaches a maximum.The phases after equilibration gave spectra which showedcobalt(I1) to be tetrahedrally co-ordinated in both the meltand in the organic phase.With chloride the absorption wasprobably due to tetrahedral species with negative charge in thesalt phase, e.g. [CoCl,(N03),]("'"- ')- while in the organicphase the spectrum (Figure 8) suggested the simultaneouspresence of octahedral as well as tetrahedral species probablycontaining chloride, nitrate, water, and tri-n-butyl phosphate(tbp) as possible ligands. Conversely with bromides the organicphase showed absorption typical of blue tetrahedral neutralspecies such as CoBr,*2tbp which is known to be soluble in theorganic solvent;34*36 the salt phase showed absorption due toboth octahedral and tetrahedral species similar to that obtainedat 0 .4 5 4 . 5 mol dm-, bromide (Figure 3).In conclusion, although nitrate is always present as apotential ligand, it competes effectively only when no halide ispresent, the octahedral complexes then formed having approxi-mately equal numbers of water and nitrate ligands. However inthe presence of both chloride and bromide, nitrate ligands areincreasingly displaced and eventually also some water ligands,finally giving tetrahedral species with approximately threehalides and one water molecule as ligands.AcknowledgementsThe authors are grateful to Mrs.R. KeneSki and Mrs. D.Simovic for their technical assistance. Thanks are also extendedto the Scientific Council of Serbia for financial support, tothe British Council for assistance through the Academic LinksScheme, and particularly to Dr. Bertil Holmberg for so freelygiving of his expertise in the computer programs and for com-puting the present data. Mr. J. Comor*s efforts in computationsand program testing are also acknowledged.References1 C. A. Angel1 and D. M. Gruen, J. Phys. Chcm., 1966,70, 1601.2 W. E. Smith, J. Brynestad, and G. P. Smith, J. Chem. Phys., 1970,52,3 F. A. Cotton and G. Wilkinson, 'Advanced Inorganic Chemistry,'3890.3rd edn., Interscience, New York, 1972J . CHEM. SOC. DALTON TRANS. 19864 D. M. Gruen, S.Fried, P. Graf, and R. L. McBeth, Proc. 2nd. U.N.5 I. V. Tananaev and B. F. Dzhurinskii, Dokl. Akad. Nuuk SSSk, 1960,6 J. A. Duffy and M. D. Ingram, J. Am. Ceram. Soc., 1968, 51, 544.7 F. A. Cotton and J. G. Bergman, J. Am. Chem. Soc., 1964,86, 2941.8 K. W. Fung and K. E. Johnson, Can. J. Chem., 1970,47,4699.9 S. N. Volkov and N. I. Buryak, Ukr. Khim. Zh. (Russ. Ed.), 1972,38,Int. Conf. Peaceful Uses Atomic Energy, 1958, vol. 28, p. 112.135, 94.663.10 S. Hemmingsson and B. Holmberg, Inorg. Chem., 1980, 19, 2242.1 1 T. R. Griffiths and P. J. Potts, Inorg. Chem., 1975, 14, 1039.12 A. G. Keenan and I. J. Ferrer, J. Phys. Chem., 1972, 76, 2844.13 M. D. Ingram and J. A. Duffy, J. Chem. Soc. A , 1968, 2575.14 J. R. Dickinson and K. E. Johnson, J. Mol. Spectrosc., 1970, 36, 1.15 J. R. Dickinson and M. E. Stone, Can. J. Chem., 1972, 50, 2946.16 M. D. Ingram, G. G. Lewis, and J. A. Duffy, J. Phys. Chem., 1972,76,17 R. A. Bailey, M. El Guindy, and J. A. Walden, fnorg. Chem., 1969,8,18 J. A. Duffy and M. D. Ingram, J. Chem. Soc. A, 1969, 2398.19 B. R. Sundheim and M. Kukk, Discuss. Faraduy SOC., 1961, 32, 49.20 H. A. 0 y e and D. M. Gruen, fnorg. Chem., 1965, 4, 1173.21 K. W. Fung and K. E. Johnson, Can. J. Chem., 1970,48, 3635.1035.2526.22 A. Barkatt and C. A. Angell, J. Phys. Chem., 1978, 82, 1972.23 I. J. Zsigrai, R. M. NikoliC, and I. J. Gal. J. Chem. Soc., Dalton Trans.,24 I. J. Zsigrai, I. J. Gal, and R. M. Nikolic, J. Chem. Soc., Dalton Trans.,25 R. Nikolik and D. Stojic, Electrochim. Actu, 1980, 25, 165.26 R. M. Nikolik and I. J. Gal. J. Inorg. Nucl. Chem., 1974, 36, 3827.27 I. J. Gal, R. M. Nikolic, and G. Herakovic, J. Chem. Soc., Dalton28 V. H. Troutner, U.S. Atomic Energy Commission, Report No. HW-29 R. M. Nikolic and I. J. Gal, J. Inorg. Nucl. Chem., 1968, 30, 1963.30 B. Holmberg, personal communication.31 H. E. Bent and C. L. French, J. Am. Chem. Soc., 1941,63, 568.32 K. E. Johnson and T. S. Piper, Discuss. Faraday SOC., 1962, 32, 32.33 G. Pedro Smith, in 'Molten Salts Chemistry,' ed. M. Blander,Interscience, New York, 1964, p. 445.34 L. I. Katzin, J. Inorg. Nucl. Chem., 1957, 4, 187.35 M. Zangen, fnorg. Chem., 1968, 7, 1202.36 H. Irving and D. N. Edgington, J. Inorg. Nucl. Chem., 1959, 10, 306.1976, 879.1978, 549.Trans., 1976, 104.57431, 1958.Received 25th July 1985; Paper 51127