JOURNALO FTHE CHEMICAL SOCIETYDALTON TRANSACTIONSInorganic ChemistrySolubility of Potassium Cyanide in Mixed Aqueous and Non-aqueousMedia; Gibbs Free Energies of Transfer of the Cyanide IonBy Michael J. Blandamer, John Burgess,* and Andrew J. Duffield, Chemistry Department, LeicesterUniversity, Leicester LEI 7RHSolubilities of potassium cyanide in a variety of non-aqueous solvents and binary aqueous solvent mixtures arereported. From these results and from published data on Gibbs free energies of transfer of the potassium ion,Gibbs free energies have been estimated for transfer of the cyanide ion from water into a selection of mixed aqueousand non-aqueous media. With the aid of these transfer parameters, reactivity trends for the reaction of the[Fe(bipy),lz+ cation (bipy = 2,2’-bipyridyl) with cyanide ion in mixed aqueous solvents are analysed for initial-state and transition-state contributions.THERE is little information published on the solubilitiesof cyanides in aqueous, mixed aqueous, or non-aqueousmedia.Potassium cyanide is freely soluble in water,soluble in glycerol, and sparingly soluble in methanol andin ethano1.l The solubility of potassium cyanide inwater is known at 298.2 K (and at other temperatures),2but published solubility data for rnethan01,~ ethan01,~and glycerol refer to the somewhat lower temperatures292.7 or 288.7 K. We have therefore measured solubi-lities of potassium cyanide in several non-aqueoussolvents and in several ranges of binary aqueous mixturesat 298.2 K.In this paper we describe the observedsolubility trends for this salt, estimate Gibbs free energiesof transfer for the cyanide ion, and compare trends withthose for other anions, particularly the chloride ion. Wealso consider the relevance of these results to our earlieranalysis of kinetic studies involving cyanide ion inbinary aqueous solvent mixtures.EXPERIMENTALPotassium cyanide was AnalaR material {B.U.H.), andwas used as supplied. Several batches of material were usedin the course of this investigation; all gave solubilityresults consistent with each other. Organic solvents werepurified by published m e t h ~ d s . ~ Binary aqueous solventmixtures were made up, by volume before mixing or byweight as specified below, using deionised water.Saturated solutions were generated by stirring or a g h t i n gthe respective solvents or solvent mixtures with a largeexcess of potassium cyanide in sealed thermostatted veisels.Aliquots were removed a t intervals for analysis, until i twas evident that equilibrium had been attained.Sampleswere appropriately diluted as soon as they had beenobtaincd, and analysed for potassium and/or cjranide ions.Potassium analysis was carried out using an EEL flamephotometer, calibrated over the whole concentration rangeused (the calibration is non-linear) against standardsolutions made up from AnalaR potassium sulphate.Cyanide analysis was performed by titration against stan-dard silver nitrate solution.6RESULTS AND DISCUSSIONSolubilities of potassium cyanide in a range of non-aqueous solvents and binary aqueous mixtures, in allcases at 298.2 K, are reported in Table 1.No directchecks with previously published values are possible,since other data are either in weight units, inconvertiblein the absence of density data, or at somewhat differenttemperatures (see above). However our results comparesatisfactorily, in a qualitative sense, with these values.SoZzcbiZity of Potassiuwz Cyanide.-The solubility ofpotassium cyanide decreases as the proportion of organicco-solvent increases (Table l), which is the normal andexpected behaviour for a salt consisting of simple, pre-dominantly hydrophilic, ions. Some solubility trendsare illustrated in Figure 1. The shapes of the curvesindicate preferential solvation by water, but of course donot indicate whether this may be attributed to favour-able hydration of the K+, the CN-, or both ions.Ace-tone-dimethylformamide (curves 5 and 4 in Figure 1)comparisons suggest stronger preferential solvation bywater in mixtures containing the former co-solvent thanin those containing the latter. The overall solubilitypattern for potassium cyanide bears a close resemblanceto that for potassium chloride, although for the threeco-solvents methanol, ethanol, and dioxan, stronge2 J.C.S. Daltonpreferential hydration for potassium cyanide, i.e. forcyanide in comparison with chloride, is indicated.Cyanide Transfer Parameters.-The main purpose ofof transfer of cyanide ion, F,p*(CN-), from water intonon-aqueous and into mixed aqueous solvents.Thex2 [y*(x)].solubility results in a number of ways.We have approached the analysis of ourIf it is assumedthis paper is to present estimates for the Gibbs free energy &nlle(KCN) = 2RT 1n[Swy,(w)/Sxy*(x)l (1)that y,(w)/y,(x) is approximately unity, then an indic-ation of the sign and magnitude of 8,,pe(KCN) is obtainedTABLE 1Solubilities of potassium cyanide, in g dm-3, in binary aqueous solvent mixtures a at 298.2 Kyo v/v co-solventMethanolEthanolt-Butyl alcoholEthylene glycolAcetoneDioxanFormamideDimethylformamideDimethyl sulphoxidcAcetonitrilePropylene carbonate5 10 20810 760740 640794 700 b7007 20612820 730814 74075030700580b52058058055054040600460b490540310400b50450350b3 50360b3 00bYo w1w co-solvcnt603 40273bb420180210b70 80 90 95230 110 51180 99 24235100 5.8 1.9b b 7.231044110 64 34b b b 13b b100218.40.10.030 .94.07.80.150.2418025010 20 22 30 33.3 40 45 50 60 80Methanol 780 720 610 540Glycerol 680 550 420 340 260Ethylene carbonate 440 320 3 00a Solubility in watcr 845 g a t 298.2 K. Phase separation occurs a t these solvent compositions.first stage in estimating 8,pe(CN-) is the estimation of8,pe(KCN) from our solubility results. The solubilitiesof this 1 : 1 electrolyte in water, S,, and in a mixedaqueous or non-aqueous solvent, Sx, are related by thedirectly from the solubilitics of Table 1.I t is note-worthy that tlie abo\re approximation does not implythat tlie solutions arc ideal, but merely that the ratio ofactivity coefficicnts is unity. However this may not be a3.0oy 2.00” 1.0d0I justifiable assumption in view of the striking changes in‘ Isolubility and in solvent permittivity. Consequently wehave examined the impact on the calculated transfercoefficients of using various theoretical equations tocalculate y* values. One approach was to use theDebye-Hiickel limiting law,’ but this was rejected on thegrounds that the concentrations were well outside therange over whicli this law is valid. Therefore we usedeither the full Debye-Huckel equation setting a, thedistance of closest approach, equal to the sum of theionic radii,* or the Guntelberg equation.’ The threevalues of 8,p0(KCN) calculated in this way are comparedfor some of the systems in Table 2.A similar approachto the analysis of solubility data is described by Parkerand co-workers * who used the Davies e q u a t i ~ n . ~ How-ever we found that this particular equation produced inmany cases smpe values which were strikingly differentand often unrealistic compared to the three methodsdiscerned above. Consequktly we did not explore thisA comparison of the three values of G,ye(KCN) forh 50 100 approach any further.oro V I VFIGURE 1 Variation of the solubility of potassium cyanide with transfer (0 a series of pure solvents (Table 2) shows thatsolvent composition for binary aqueous solvent mixtures. the assumption y* (w)/y, (x) = 1 .O is reasonable.ThusCo-solvents: ( I ) methanol; (2) ethanol; (3) dimethyl sulph- the order of solvents based on the magnitude of &,ve isoxide; (4) dimethylformamide; (5) acetone hardly affected. Moreover comparison of these quanti-* This approach has very recently been used in estimatingtransfer activity cocfficients for ions from solubility data forpotassium and tetraphenylarsonium picrates and potassium tetra-phenylborate (C. Tissier, Comfit. rend., 1978, C286, 35).difference in standard-state chemical potential, 8 , ~ ~ -(KCN), by equation (1). In this equation y* representsthe mean ionic activity coefficient of the salt in eitheraqueous solution [y+ (w)] or in the mixture of mol fractio1980 3ties for transfer between pure solvents represents the' worst ' case.For transfer between solutions in waterand solutions in water-rich mixtures there is less dis-agreement between the three values for Gmpe(I<CN).Finally we note that these differences are much smallerthan the uncertainties arising from the various assump-tions used in establishing single-ion parameters (seebelow).Having obtained values for Gmpe(KCN) it is nownecessary to assume values for Smpe(K+) in order toobtain values for 8mpe(CN-) [equation (2)]. Several setsG,p"(KCN) = G,pe(K+) + G,p"(CN-) (2)of G,,p*(K+) values exist, for non-aqueous and for mixedaqueous media. They are based on a variety of assump-tions and, unhappily, values for a given medium dependmarkedly on the assumptions involved.Two mainapproaches to the obtaining of single-ion 6mp9 values aresulphoxide-water mixtures have also been obtained 2ofrom e.m.f. measurements of the autoprotolysis constantfor water via the electrostatic model calculation methodfor H+ of Feakins and Watson.21 We shall be makinguse of each of the assumptions and derived Gmpe(K+)values; often it has proved necessary to interpolatefrom published values, and sometimes to convert units,for example, from the mol fraction into the molarityscale .sCyanide Transfer to Pure SoZvents.-Table 2 shows ourestimated values of G,pe(KCN) from water into non-aqueous solvents, the various available estimates ofampe (K+) , and our thence-derived values for ampe (CN -) .Agreement between Gmpe(K+) values derived by differentroutes is not good;22 obviously our Gmpe(CN-) valuesreflect this.The negative values for G,pe(K+) for trans-fer to dioxan and to acetone obtained by the treatmentof the de Ligny school are particularly u n a t t r a c t i ~ e , ~ ~ ? ~ ~TABLE 2Gibbs free energies of transfer of potassium cyanide, 6mpe (KCN), from water into non-aqueous solvents at 298.2 K (molarscale), calculated from measured solubilities : for potassium ion, G,pe(K+) was assumed as indicated ; for cyanide ion,(CN-) was calculated from the foregoing6,,,pe/k J mol-'SolventMethanolEthanolAcetoneDimethyl sulphoxideDimethylforniamideFormamideAcetonitrilePropylene carbonateA N # b41.3 18.637.1 22.912.5 51.019.3 23.216.0 26.539.8 6.018.0 42.840.5K[CN] - C20.325.747.221.825.44 .139.2tie Ligny cox b, f7 f-----h- - d I< + CN- K+ CN-21.6 4.9 13.8 10.0 8 . 627.2 13.4 9.5 15.7h 7.247.1 -15.2 66.2 2.9 48.021.9 -12.1 35.325.6 - 9 . 6 36.13 . 8 -6.3 12.339.1 8.0 34.93.4 36.1Abraham b*g' K + CN-'2.7 15.94.2 18.89.6 41.4-6.7 29.9-2.1 28.610.0 32.8a A N = Acceptor number, see ref. 25. y+- Quantities calculated using the Debye-Huckel equation. yri- Quantities calculated using the Giintelberg equation. Rcfs. 11-13. f Ref. 16. Ref. 19. h Derivedvia the assumption of Popovych and Dill.l*Derived assuming y+(w)/yk(x) = 1.0.relevant here. One type of approach is based on aconsideration of the molecular interactions of the ionwith solvent molecules coupled with a Born treatment;the other type depends on assuming that the Gibbs freeenergy of transfer of a given large cation and of a givenlarge anion (often AsPh4+ and BPh4-) are equal.Wells lohas used the first approach to estimate Gmpe(Kf) in avariety of water-rich binary mixtures, of which thosewith co-solvents methanol, t-butyl alcohol, ethyleneglycol, glycerol, acetone, and dioxan are relevant here.A different aspect of the first approach has been employedby de Ligny and colleagues for 6,pe(K+) values in non-aqueous and in mixed solvents, e.g. with the organic(co-)solvents methano1,ll ethanol,12 acetone,13 and di-oxan.14 The assumption that G,pe(AsPh,+) = RmpO-(BPh,-), currently in has been used for anextensive set of single-ion values in non-aqueous sol-vents,16 and for dimethyl sulphoxide-water l7 andacetonitrile-water l7 mixtures.The similar assumptionthat 6,p~[NRu(i-C,Hll),+] = G,pe(BPh4-) has been usedfor ethanol-water mixtures,18 while a related assumpt.ionthat G,pe(NMe4+) = 0 has also been employed l9 toobtain single-ion values for transfer into several non-aqueous solvents. Values for Gmpe(ions) in dimethyland we therefore discount the Gmpe(CN-) values derivedtherefrom. There is still an uncomfortable measure ofdisagreement between the Gmpe(K+), and thus 6,pe(CN-),values obtained by the ' large-ion ' methods. Theassumption of Cox and Parker that 8,pe(AsPh,+) =G,pe(BPh,-),16~23 and the related assumption of Popovychand Dill l8 that G,pe[NBu(i-C5H1,),+] = G,pe(BPh4-),seem slightly more attractive than that of 8,pe(NMe4+) =0.19 In practice the first-named assumption is of widestuse ; values derived using this assumption are availablefor several mixed solvents (see below) as well as for non-aqueous solvents.Values of 8,Lpe(CNd) for transfer of cyanide from waterinto non-aqueous solvents show some degree of cor-relation with certain empirical solvent pararneter~.~~The most satisfactory correlation is that between G,pe-(CN-) values derived by Cox's method l6 and Gutmann'sacceptor numbers, A N ; 25 the standard error of such aplot is 8%.Cyanide Trans fey to Mixed Aqueous Solvents.-Table3 collects together our estimates for 8,pe(CN-) fromwater into binary aqueous mixtures, with the sources ofGmpe(K+) indicated in each case.Again the choice ofassumptions used in the single-ion split has a disturb4 J.C.S. DaltonTABLE 3Gibbs free energies of transfer of potassium cyanide (calculated from measured solubilities), potassium ion (assumed asindicated), and cyanide ion (calculated from the foregoing) from water into binary aqueous mixtures a t 298.2 K, molarscale. The assumptions used are indicated a t the head of each column of single-ion values for each co-solvent ; columnI entries are all based on Wells's Born-derived model, column I1 top four entries are differently Born-derived, andcolumn I11 entries are based on various large-ion splitting assumptions&p/k J mol-lCo-solventMethanolEthanolt-Butyl alcoholEthylene glycolGlycerolAcetoneDioxanDimethyl sulphoxideAcetonitrile% "v/v10low2020w3030w4040w50607080901020305070809051030507020w33.3w50w60w203050708090103040901020304050607080902030x20.0470.0590.1000.1230.1600.1940.2290.2730.310.340.410.490.580.0330.0720.1170.2350.420.550.730.0090.0210.150.240.430.0470.0890.1640.2270.0580.0950.1960.360.490.690.0230.0830.1230.650.0270.0600.0980.1440.2020.2760.370.500.700.0790.129KCN0.210.380.520.790.951.611.702.223.134.56.510.414.00.671.381.904.47.710.717.70.300.862.44.46.41.052.153.504.630.921.904.210.413.319.00.591.862.7223.70.120.802.13.85.17.010.112.815.90.682.28I I1 111& -7 r A - K+ CN- K+ CN- K+ CN-Wells b de Ligny0.50.70.91.11.31.41.61.61.4- 0.4- 1.40.11.51.20.50.71.0 f1.2-3.9-5.2ca.-9 ca.- 1.8-4.1-4.7-0.3-0.3-0.4- 0.3-0.4-0.20.10.61.70.72.32.32.95.20.51.52.53.44.87.1132.46.07.4-0.1-0.2- 0.3- 0.5- 0.7-0.9 - 1.2- 1.8-2.2- 2.4- 2.9-3.5-4.20.30.60.81.31.62.52.94.15.36.99.413.918.2de Ligny d1.3 0.63.6 0.75.5 2.2de Lignyr- 1.0 1.9- 1.8 3.7 - 4.0 8.2- 7.9 18.3- 10.3 23.6- 13.1 32.1de Ligny h-0.4 1 .o- 1.3 3.2- 1.9 4.6Popovych 00.5 0.20.4 1 .o0.4 1.50.5 3.93.4 4.25.6 5.09.5 8.2ca.7 ca. 17Das and Kundu f cox-5.0 5.1 -0.1 0.2- 10.9 11.7 -0.3 1.1-0.8 2.9-30.1 33.9 - 1.5 5.3- 2.0 7.1- 60.1 67.1 -3.3 10.3-5.3 15.4- 7.3 20.1-9.6 25.5-0.1 0.8-0.3 2.60 w = Weight percent. Ref. 10. e Ref. 11. Ref. 12. Ref. 18. IAn alternative value of G,p*(K+) = 0.3 kJ mol-l givesG,p(CN-) = 3.2 kJ mol-1. oRef. 13. "ef. 14. 'Ref. 20. jRef. 171980 5ingly large effect. Cyanide normally becomes destabil-ised as more organic co-solvent is added, but the extentof destabilisation appears markedly different dependingon the provenance of the 6,pe(K+) values used.Suchdifferences may be most dramatically illustrated by thecase of dimethyl sulphoxide-water mixtures, where Dasand Kundu 20 calculate 8,pe(K+) = -60 kJ moP1 forTABLE 4Comparison of G,pe(anion) values for anions for transferfrom water into aqueous methanol. All valuesobtained by the use of Wells's values for 8,pe(K+);all on the molar scale, at 298.2 I<&,+*(anion) /k J mol-'7 Methanol h(% vlv) CN- c1- so,2- s,o,2-10 - 0.3 0.9 4 220 - 0 . 4 2.1 9 430 - 0 . 4 3.6 13 540 0.1 17 9Source a b c d@ This work. From refs. 31 and 32. Estimated fromthe solubilities of potassium sulphate reported by G. Akerlofand H. E. Turck, J . Amev. Chem. Soc., 1935, 57, 1746. Fromsolubilities of potassium peroxodisulphate (M.J . Blandamer,J . Burgess, and R. I. Haines, J . Inorg. Nuclear Chem., 1979,41, 258.transfer from water into 60 dimethyl sulplioxide,whereas Cox ct aZ.17 propose 6,,pe(K+) = -3.3 kJ mol-lfor the same transfer.In order to compare B,pe(CN-) values with analogousparameters for other anions, we have collected togetherin Table 4 sets of values derived using Wells's method inall cases. They can thus be compared fairly. Table 4brings out clearly the similarity between cyanide ionand the similar size uninegative chloride ion, and theTABLE 5Differences, A, between the Gibbs free energies of transfer ofcyanide and of chloride from water into selected binaryaqueous solvent mixtures [A = 6,pe(C1-) - ?jmpe-(CN-)] a Values of A are in kJ mol-l, on the molarscale, a t 298.2 Kyo Co-solvent b'10 20 30 40 50 60 70 80 90'Methanol 1.5 2 .0 3 . 0 4.0 4.5 5.3 6.4 5.0 3.7Ethanol 0 7 1.4 2 . 4 3.0 3.6 4 . 4 5 . 3 6.8 5.2Glycerol - 0 . 2 -1.1 -1.1 - 1 . 1A - 6,,,pe(KC1) - &pe(KCN), with 8,p~(I<Cl) values cal-culated from published solubility data and tS,pe( KCN) valuesderived as in the text. Percentages by volume before mixingfor methanol and ethanol, by weight for glycerol.differences between these and dinegative anions such assulphate and peroxodisulphate. However a moredetailed comparison between the similar but by nomeans identical cyanide and chloride ions is given inTable 5 , in the form of differences between 6,pe(CN-) and8mp*(Cl-) derived from solubilities of potassium cyanideand of potassium chloride [and thus independent ofassumptions made about 8,pe(K+) values].The intim-ations of different preferential solvation characteristics inthe potassium cyanide-potassium chloride solubilitycomparisons above are confirmed and amplified by thesenumbers.Initial-state and Transition-state Solvation.--We andothers have recently attempted to separate solventeffects on reactivity for inorganic reactions into initial-") '10 v / v methanol!than030-10 --20 -LFIGURE 2 Analysis of solvent effects on reactivity separated intoinitial-state (i.s.) and transition-state contributions for thereaction of the [Fe(bipy),12+ cation with cyanide ion;0 ___ initial -- 0 - - individual reactants, --state, -- + -- Gibbs free energy of activation, --transition statestate and transition-statc compon~nts.~~-~~ Such treat-ment of reactions with cyanide ions30 33 lias beenhampered by a lack of knowledge of 8,,pQ(CY) values.Our earlier analysis on the intuitively reasonable basis 34TABLE GDcrivation of estimates for 6,,pe[Fe(bipy),2'] from waterinto aqueous methanol froni measurements of solubili-ties of its tetraphenylborate6,,,p*(species) /k J mol-'Illethanol [Fc(bipy) a:-(%I vlv) [UPh,l2 [BPli,]- [ Fc(bipy)J2+10 - 19.1 -2.7 - 13.720 - 25.9 - 3.8 - 18.330 - 36.0 -6.4 - 23.240 -440.4 - 7.5 - 25.4(I From measured solubilities, estimated via spectrophoto-metric determination of cation concentration.Via solu-bilities of the caesium salt (K.1. Haines, Ph.1). Thesis, Uni-versity of Leicester, 1977).of equivalence of S,,p"(CN-) and 8,pe(Cl-) is now super-seded.Reaction of [Fe(bipy),I2+ (bipy = 2,2'-Bi$ydyZ) withCyanide.-Figure 2 shows an analysis of solvent effectson this reaction for methanol-water mixtures using ouJ.C.S. Daltonpresent 6,$+(CN-) value^.^^.^^ The two parts of Figure 2use two different sets of estimates for G,pe[Fe(bipy)32+] :graph (a) uses van Meter’s and Neumann’s 35 values andgraph ( b ) uses our values for the iron complex derivedfrom solubilities of its tetraphenylborate salt (ref. 36 andTable 6). Despite the obvious differences, the overallconclusion is that solvation changes around the [Fe-(bipy),12+ moiety dominate, both in the initial and in thetransition state, a t least in water-rich mixtures.Thesame conclusion may be reached for acetone-watermixtures, the only other series for which such an analysisis possible.We thank Dr. B. G. Cox for helpful discussion and forproviding data previous to publication, and the S.R.C. forsupport.[8/1749 Received, 6th October, 19781RE I; E R E N C E SCRC Handbook of Chemistry and Physics, 58th edn., ed.R. C. Weast, CRC Press, Cleveland, Ohio, 1977, p. B144.2 H. Bassett and A. S. Corbet, J . Ckem. SOC., 1924, 1660;A. S. Corbet, abad., 1926, 3190.C. A. L. de Bruyn, Z . phys. Chem., 1892, 10, 782; Rec.Trav. chim., 1892, 11, 112.A . M. Ossendowski, Pharfm. J . (London), 1007, 79, 575;J . Phnrm. Chim., 1907, 26, 162.A. J . Gordon and R.A. Ford, ‘ The Chemist’s Companion,’Wiley, New York, 1972.A. I. Vogel, ‘ Quantitative Inorganic Analysis,’ 2nd edn.,Longmans, London, 1951, p. 263.See, for example, R. A. Robinson and R. H. Stokes, ‘ Elec-trolyte Solutions,’ 2nd edn. (revised), Butterworths, London,1965, ch. 9.A. J . Parker and W. E. Waghorne, Austral. J . Chcm., 1978,31, 1181; R . Alexander, A. J . Parker, J . H. Sharp, and W. E.Waghorne, J . Amer. Chem. SOC., 1972, 94, 1148.C. W. Davies, ‘ Ion Association,’ Butterworths, London,1962, p. 39.lo C. F. Wells, J.C.S. Faraday I , 1973, 984; 1974, 604; 1975,1868; 1976, 601; 1978, 1569.l1 C . L. de Ligny and M. Alfenaar, Rec. Trav. chim., 1965, 84,81 ; M. Alfenaar and C. L. de Ligny, ibid., 1967, 86, 929; D. Bax,C.L. de Ligny, and M. Alfenaar, ibid., 1972, 91, 452.l2 D. Bax, C. L. de Ligny, and A. G. Remijnse, Rec. Trav.chim., 1972, 91, 965.l3 C. L. de Ligny, D. Bax, M. Alfenaar, and M. G. L. Elferink,Rev. Trav. chim., 1969, 88, 1183; D. Bax, M. Alfenaar, and C. L.de Ligny, ibid., 1971, 90, 1002; D. Bax, C. L. de Ligny, and A. G.Remijnse, ibid., 1972, 91, 1225.l4 D. Bax, C. L. de Ligny, and A. G. Remijnse, Rec. Trav.chim., 1973, 92, 374,l5 H. J. M. Nedermeijer-Denessen, C. L. de Ligny, and A. G.Remijnse, J . Electroanalyt. Chem., 1977, 77, 153; C. Treiner,Canad. J . Chem., 1977, 55, 682; J. I. Kim, J . Phys. Chem., 1978,82, 191.l6 B. G. Cox, Ann. Reports ( A ) , 1974, p. 249.l7 B. G. Cox, R. Natarajan, and W. E. Waghorne, J.C.S.0. Popovych and A. J. Dill, Analyt. Chem., 1069, 41, 456.lQ M. H. Abraham, J.C.S. Faraday I , 1973, 1375.2o A. K. Das and K. K. Kundu, J.C.S. Faraday I , 1973, 730;K. 13. Khoo, J , Chem. SOC. ( A ) , 1971, 2932; J . Chem. SOC. ( B ) ,1971, 1178.21 D. Feakins and P. Watson, J . Chem. SOC., 1963, 4734.22 J. Burgess, ‘ Metal Ions in Solution,’ Ellis Honvood,23 B. G. Cox, G. W. Hedwig, A. J. Parker, and D. W. Watts,24 C. Reichardt, Angew. Chem. Internat. Edn., 1965, 4, 29.25 W. Mayer, V. Gutmann, and W. Gerger, Monatsh., 1975,106,1235.26 nI. J . Blandamer and J . Burgess, Pure Appl. Chcm., 1079,51, 2037.27 M. H. Abraham, G. F. Johnston, J . F. C. Oliver, and J . A.Richards, Chem. Comm., 1969, 930; M. H. Abraham, J . Chem.Soc. ( A ) , 1071, 1061.28 M. H. Abraham, Chem. Comm., 1969, 1307.29 M. J. Blandamer, J. Burgess, and J . G. Chambers, J.C.S.30 J . Burgess, J.C.S. Dalton, 1972, 1061; Inorg. Chim. A d a ,31 M. J . Blandamer, J. Burgess, and J. G. Chambers, J.C.S.32 M. J . Blandamer, J. Burgess, J . G. Chambers, R. I. Haines,33 M. J. Blandamer, J. Burgess, S . J. Cartwright, and M.34 N. N. Greenwood, ‘ Ionic Crystals, Lattice Defects and Non-35 F. M. van Meter and H. M. Neumann, J . Amcr. Chem. Soc.,36 M. J. Blandamer, J. Burgess, and R. I. Haines, unpublishedFaraday I , 1979, 86; ibid., submitted for publication.Chichester, 1978, ch. 6 .Austral. J . Chem., 1974, 27, 477.Dalton, 1977, 60.1971, 5, 133.Dalton, 1976, 606.and H. E. Marshall, J.C.S. Dalton, 1977, 165.Dupree, J.C.S. Dalton, 1976, 1158.Stoichiometry,’ Butterworths, London, 1968, pp. 27, 35, 41.1976, 98, 1382.work