Free Energies of Transfer of Alkali Metal Fluorides from Water to Hydrogen Peroxide+Water and Methanol+Water Mixtures using Ion-selective Electrodes BY ARTHUR K. COVINGTON* AND JENNIFER M. THAIN Department of Physical Chemistry, University of Newcastle, Newcastle upon Tyne NEl 7RU Received 8th April, 1974 Alkali metal ion-responsive glass and fluoride- or chloride-, responsive ion-selective electrodes have been used to obtain free energies of transfer of alkali metal fluorides and chlorides from water to methanol + water mixtures. The precautions and safeguards necessary to obtain reliable results are discussed. The results obtained for the chlorides are in satisfactory agreement with existing data obtained using amalgam double cells. The method was extended to the determination of free energies of transfer for the alkali metal fluorides from water to hydrogen peroxide + water mixtures and the results obtained were compared with spectroscopically-derived values.The Gibbs free energy of transfer, that is the difference between the standard free energy per mole of electrolyte in a pure solvent, usually water, and that in another solvent or mixed solvent, is an important measure of the differences in interaction between the ions of the electrolyte and the solvent molecules in the two media.l Evaluation of this quantity may be made either from e.m.f. measurements or by solubility methods, but the former is the more satisfactory method. Most attention has been directed towards acids using hydrogen gas and silver-silver halide electrodes but in a number of ways values for salts are easier to interpret.For the latter, amalgam electrodes have been employed together with silver-silver halide electrodes. 39 However, Lowe and Smith reported the use of sodium responsive glass electrodes to obtain free energies of transfer of sodium chloride between H20 and H,0+D20 mixtures. No values have been reported for fluorides because of the lack, until recently,6 of a suitable electrode reversible to fluoride ions. The lanthanum fluoride, single crystal, ion-selective electrode developed by Ross and Frant has been used for precise thermodynamic studies in aqueous solutions 7-9 and is suitable for use in solvents other than purely aqueous ones.1o* l 1 In a preliminary study l 2 on the work described in this paper, we used sodium-responsive glass and fluoride ion-selective electrodes to obtain free energies of transfer of sodium fluoride from water to water + hydrogen peroxide mixtures.Neither amalgam electrodes nor silver-silver halide electrodes are usable in solutions containing hydrogen peroxide and this accounts for the paucity of electrochemical work in this solvent medium apart from the study of Mitchell and Wynne-Jones We now describe the precautions necessary for using two ion-selective electrodes to obtain free energies of transfer, and the test of the method for alkali metal chloride solutions in methanol + water mixtures, for which data are available for comparison purposes from amalgam electrode cells.4 Results will be given for three alkali metal fluorides in this solvent mixture which it has not been possible to determine previously, and for four alkali metal fluorides in hydrogen peroxide + water mixtures.Interest in these results arises from an important correlation l4 between the free energy of preferential solvation, a quantity determinable from n.m.r. chemical shift measure- ments and the free energy of transfer. using hydrogen-responsive glass electrodes. 78A . K . COVINGTON AND J . M. THAIN 79 THEORY The cells studied were without liquid junction, namely anion (X) responsive (1) I MX,H20yS I ion selective electrode. cation (M) responsive glass electrode The difference in e.m.f. (AE,) between two cells of type (I) containing solutions of composition MX(mS), S(x), H,O(l-x) with e.m.f. Ex and MX(mw), S(x = 0), H,0(1 -x = 1) with e.m.f.E,, is given by where k = RT(1n lO)/F; m and y k are respectively a suitable concentration variable, and the mean ionic activity coefficient relative to a hypothetical solution of unit concentration in the solvent mixture of mole fraction x of component S . The problem is to choose the most appropriate concentration scale ; for reasons related to our previous work l4 we have chosen the aquamolality scale, defined as number of moles of solute per 55.51 moles of solvent. If we make ms = mw, then A E g , is obtainable from the measured AE,,,, by making an activity coefficient term correction, and where y;, yl can be calculated from a Debye-Huckel-type expression. It can be shown using the method of Robinson and Stokes,15 that the aquamolality activity coefficient and the molality activity coefficient are the same for a given solution irrespective of the choice of aquamolality or molality scales.Thus the aquamolality activity coefficient can be calculated from expression (3) AE, = E, -Ex = AE& + 2k log(msy;/mwy;) (1) A q X ) = AEt(x)-2k h(Y%/Y3 (2) logy* = - -ln(l f0.03603 m) 1 + BaJc where c is the molarity given by 18.015 md WA+0.01S015 mW,' c = (3) (4) The symbols in eqn (3) and (4) have the following significance : A = molar Debye- Huckel limiting law constant ; a = ion-size parameter taken as 4.56 A ; B = molar Debye-Huckel constant in the denominator of eqn (3) ; d = solution density equated to the solvent density ; W, = molecular weight of the solvent calculated from W, = 18.015(1-x)+xWs where W, is the molecular weight of the co-solvent; W, = molecular weight of the solute.The last term in eqn (3) arises from the need to adjust the Debye-Huckel expression for the rational (mole fraction) activity coefficient to give the molal activity coefficient. The quantities A and B, which are functions of the relative permittivity (dielectric constant) of the solvent mixture (E,) and the temperature (T), were calculated from B/cm-l mol-* dm* K* = 50.29 x 108/(~,T)* where values for the dielectric constant were obtained from a recent compilation.16 The activity coefficient correction in eqn (2) amounts at a maximum to 3-4 % in the free energy of transfer when m = 0.01 aquamolal. Some measurements were made at m = 0.005 and 0.02 aquamolal, which confirmed that this method for making activity coefficient corrections obviated the need for an elaborate procedure for extrapolation to infinite dilution.A/mol-* dmt Kt = 1.8246 x 106/(~,T)*80 TRANSFER OF ALKALI METAL FLUORIDES alkali metal ion (M) responsive glass electrode EXPERIMENTAL The potentials of cells (I) were measured using an integrated circuit (MF1-3 electrometer amplifier : Computing Techniques Ltd., Billinghurst, Sussex) interfacing device to a digita 1 voltmeter (DVM) (Advance Electronics Ltd., Bishop’s Stortford, Herts.) with f 0.1 mV discrimination. The cell potentials could also be opposed by a vernier potentiometer, and the DVM used as a null-reading device. The presence of hum pick-up was detectable by a difference between the directly read DVM value and the backed-off value.With strict attention to shielding and grounding, the affects of hum and stray capacitance were rendered negligible. The output from the interfacing device could also be fed to a potentiometric recorder to give a trace of the variation of cell potential with time. The electrodes used and their sources are shown in table 1. As far as possible, several electrodes of the different types available for a given ion were used in the studies. Glass electrodes were conditioned by soaking in 1 rnol dm-3 solutions of the requisite alkali metal ion in 0.1 rnol dm-3 Tris+HCl solution. Conditioning of the halide-responsive ion- selective electrodes is not necessary. Cell vessels were simple H-type with a tap between the compartments.Sets of six cell vessels were placed in an air thermostat (25.0+O.l0C) and pairs of electrodes were transferred between solutions of steadily increased methanol or peroxide content. The Orion lanthanum fluoride electrode was found to be affected by hydrogen peroxide solutions. The trouble was traced to hydrogen peroxide attacking the epoxy cement used to fix the single crystal into the epoxy body. This attack could be prevented by coating the sensitive annular region with paraffin wax. Alternatively a lanthanum fluoride electrode was constructed from a 5 mm diameter, 2 mm thick, undoped, lanthanum fluoride crystal (Metals Research Ltd., Histon, Cambridge) sealed with Araldite into glass fibre-reinforced, epoxy, drilled rod (Bushing Co. Ltd., Hebburn, Co. Durham).The inner filling system was 0.1 mol dm-3 NaFSO.1 rnol dm-3 NaCl I AgCl I Ag. This home-made electrode was quite resistant to attack by hydrogen peroxide and substantially the same results were obtained from it. MCI, HzO, MeOH C1--responsive AgCl (11) (1-x) x 1 ion-selective electrode. TABLE 1 .-ELECTRODE TYPES AND SOURCES anion responsive cation responsive Orion 94-09 fluoride-responsive* Orion 94-17 chloride-responsive* Thermal electrolytic Ag/AgCl Orion Na+-responsive glass, 94-1 1 A* E.I.L. Na+-responsive glass, GEA 331 3 or 33 1048 loo? Corning monovalent cation-responsive glass 476220: E.I.L. K+-responsive glass 33 1057 2007 * Orion Research Inc., Cambridge, Mass., U.S.A. ; f Electronic Instruments Ltd., Chertsey, Surrey ; Corning-E.E.L., Halstead, Essex. Methanol+ water mixtures were prepared from AnalaR grade methanol and doubly distilled water.Hydrogen peroxide 85 % w/w (Laporte Chemicals Ltd., Warrington, Lancs.), free from stabilising additives was diluted with doubly distilled water, and stock solutions were analysed by titration with permanganate. Salts were the highest grade material readily available and were dried before use. All solutions were made 0.01 aquamolal in salt and rnol dm-3 in Tris to buffer the solutions to a pH where the alkali metal ion responsive glass electrodes would not show a mixed response to hydrogen ions.A . K. COVINGTON AND J . M. THAIN 81 The technique employed consisted of allowing the pair of electrodes to reach a steady potential in the purely aqueous solutions and then to transfer them successively to solutions of increasing methanol content.In this way the glass electrode is subjected to only small changes in solvent environment and large changes in solvent medium, which often provoke drifts in potential, avoided. Return transfers down the series to the purely aqueous solution showed that, although the actual steady potential reached in each cell was not necessarily the same as that reached in the up-series transfers, the dzflerences in potential between successive cells i and .j, AElj, differed from AE,, in the downward series by only 1 %. This is illustrated in fig. 1 which refers to an Orion Na+ responsive electrode conditioned and used in caesium chloride solutions. The transfer e.m.f. A,??,,,, in eqn (1) is obtained from AEtj and AEji by X X that is, the transfer e.m.f.from water into a methanol mixture of mole fraction x is obtained by summing the potential differences between successive cells. Values of AE,,,, and, from eqn (2) with eqn (9, values of the free energy of transfer AG&, are collected in table 2. The apparent absence of a negative sign in eqn ( 5 ) arises from the method of defining the transfer e.m.f. and follows the practice of Feakins and Voice.4 The approach by these workers to calculate free energies of transfer from their measurements on amalgam double cells is related to that outlined above except that Feakins employs the molal standard state (AG&$. The relation between the two free energies of transfer with respect to aquamolality or molality standard states can easily be shown to be where the second term arises from the ratio of molality to aquamolality.AG& = AG,",,,+2RT In 10 log(18.015/WA) (6) 640 590 520 450 340 I timelrnin FIG. 1 .-Transfer of an Orion Na+-responsive and an Orion Cl- responsive electrode between solutions 0.01 aquamolal in CsCl and increasing methanol content, cells 1-6. Fig, 2 compares results obtained with two types of cation-responsive glass electrode with the corrected results of Feakins and Voice for CsCf in methanol+ water mixtures. The results incorporated in table 3 are those obtained with the Orion Na+-responsive electrode. It is clear from fig. 2, that in high methanol content (x > 0.8) solutions the results using the E.I.L. K+-responsive glass electrode82 TRANSFER OF ALKALI METAL FLUORIDES are anomalous by comparison with both the Orion glass and the amalgam electrode results.The high potentials observed probably result from abstraction of water from the gel layer on the surface of the glass electrode, the thickness of which depends on the glass composition and may be thicker for the E.I.L. electrode and more sensitive to methanol than that of the Orion glass. A similar effect has been reported of MeOH 0 0.197 0.201 0.204 0.305 0.401 0.403 0.436 0.478 0.590 0.612 0.616 0.651 0.784 0.789 0.800 0.806 0.899 0.900 1 .ooo x 4 = CI 2 a G a 10 C!l TABLE 2.-vALUES OF AEt,xjmV FOR CELL (11) AND AQUAMOLAL FREE ENERGIES OF TRANSFER AGp(xl/kJ mol-1 FROM WATER TO MeOHf H20 mole fraction A@/ A G ~ ) I AGYl AGyl A G ~ I LiCl NaCl KCl RbCl CSCl AEtlmV kJ mol-1 AEtlmV kJ mol-1 AEt/mV kJ mol-1 AEt/mV kJ mol-1 AEtlmV kJ mol-1 0 0 0 0 47.9 4.72 31.1 3.11 89.1 8.82 60.8 6.12 125.0 12.29 83.8 8.48 0 0 0 0 0 43.8 4.33 47.9 4.72 67.3 6.65 62.3 6.17 84.2 8.35 107.8 10.68 101.8 10.10 121.8 12.13 140.2 13.95 130.3 12.99 167.4 16.66 164.0 16.34 155.2 15.28 107.7 10.86 134.6 13.68 178.8 18.00 182.1 18.09 178.5 17.97 197.2 19.78 187.5 18.72 I I I 1 Q2 OA 0.6 0.8 1-0 methanol mole fraction FIG.2.-Free energy of transfer for CsCl from water to methanol + water mixtures. - , amalgam electrode result^.^ + , Orion Na+-responsive electrode ; 0, x , E.I.L. K+-responsive electrodes.A . K . COVINGTON AND J . M. THAIN 83 previously. * Ivanovskaya, Gavrilova and Shults and Eisenman l 8 have studied certain electrode glass compositions in water, methanol +water and methanol solutions all saturated with NaCl or KCl.The Russian workers l7 refer to the specific effect of methanol on lithium aluminosilicate-based glass electrodes in solutions containing alkali metal ions not contained in the bulk electrode glass, and attributed the observed effects of 20-40 mV to diffusion potential effects in the glass. An alternative method of comparing the present results with those of Feakins and Voice is shown in fig. 3 as plots, against the mole fraction of methanol, of which is a deviation function from the straight line joining AG&( =0) and AG&= 11, a procedure which greatly magnifies the experimental differences. It can be seen from fig. 3, that up to x = 0.7, except for RbCl, all cells show agreement to within I - o x o x + + a r .I I 2 - 0 0 -0. - W rl I - 2 I 0.2 0.4 46 as methanol mole fraction LO LiCl NaCl 8 KCI RbCl CsCl FIG. 3.-Alternative method of comparison of results for methanol + water mixtures with those from amalgam cells in terms of a deviation function AG?(,) - X A G ~ ( ~ = l). 0, Amalgam result^.^ LiCl : +, 0, E.I.L. Na+-responsive electrodes ; x , Orion Na+-responsive electrode. NaCl : x , 0, E.I.L. Na+-responsive electrodes without Tris added to solutions ; 0, E.I.L. Na+-responsive electrode. KCl : 0, E.I.L. K+-responsive electrode ; x , Corning monovalent ion-responsive electrode. RbCl : 0, E.I.L. K+-responsive electrode ; x , Orion Na+-responsive electrode. CsCl : x , 0, E.I.L. K+- responsive electrode ; + , Orion Na+-responsive electrode.84 TRANSFER OF ALKALI METAL FLUORIDES + 5 mV.This is true over the whole mole fraction range if the anomalous results for E.I.L. K+-responsive electrodes are excluded (as has been done in table 2). These high values are observed for KCl, as well as for CsCl already referred to, and are greatly enhanced at low mole fractions for RbCl. In making this comparison with Feakins and Voice’s result^,^ it should be borne in mind that amalgam electrodes are reactive in aqueous solutions and that the e.m.f. results presented were obtained from plateau potentials held for only 10-20 min. Hence the agreement is judged satisfactory particularly since the present intention was to obtain results quickly of only moderate accuracy for comparison with spectroscopically-derived data.Attention should be drawn to points for NaCl in fig. 3 which were obtained by sub- stituting a conventional thermal electrolytic Ag I AgCl electrode for the Orion chloride-responsive ion-selective electrode ; these points confirm the correct function- ing of the latter in methanol + water mixtures. Transfer e.m.f. values for lithium, sodium and rubidium fluorides in methanol- water mixtures were obtained from measurements using cell (111) : alkali metal ion (M) F- responsive LaF, (111) responsive glass electrode 1 ion selective electrode. 78.3 87.9 7.82 8.89 TABLE 3.-\’ALUES OF AEt(,)/rnV FOR CELL (111) AND AQUAMOLAL FREE ENERGIES OF TRANSFER AG?(,)/kJ rnoP FROM WATER TO METHANOL+ WATER mole fraction LiF NaF RbF MeOH *E*(x)lmV A@) AEt(x)/mV 4) AEt(x)’mV A G g ) 0 0 0 0 0 0 0 0.201 53.8 5.89 0.212 48.3 4.77 0.305 88.1 8.66 0.403 107.2 10.57 0.410 94.9 9.39 0.456 0.478 134.0 13.21 0.503 121.6 12.02 0.603 157.7 15.59 135.3 13.42 0.612 161.1 15.92 0.651 166.0 16.43 0.789 207.1 20.49 0.825 177.1 17.60 0.837 116.9 11.85 185.2 18.43 0.900 199.0 19.82 1 .ooo 124.8 12.79 184.0 18.50 221.50 22.10 These and values of AG& obtained from eqn (5) are given in table 3.No previous data are available for comparison. For LiF, the E.I.L. and Orion Na+ responsive glass electrodes were used but for NaF only the former was available. For both salts, cells were well-behaved, reaching steady values on transfer quickly. For RbF, however, the results were less satisfactory, and only those obtained with Corning monovalent ion electrode were used in compiling table 3. E.I.L.K+-responsive electrodes again showed high values in the high methanol content solutions, but additionally the values were less reproducible and dependent on the time the electrodes were kept in high methanol content solutions. Table 4 shows tests for the additivity of single ion free energy of transfer values at four methanol mole fractions. Differences were taken between free energy values for pairs of salts and the discrepancies are expressed in terms of A, which should be zero. The new data meet the additivityA. K . COVINGTON AND J . M. THAIN 85 criterion to within - 3 kJ mol-l. Values in brackets in table 4 were derived using the amalgam cell data for chlorides of Feakins and Voice,4 which effects a slight improvement in the additivity test ( w 1 kJ mole-I). TABLE 4.-ADDITIVITY TESTS FOR FREE ENERGY OF TRANSFER DATA (AG/kJmOl-') FOR METHANOL AND WATER MIXTURES mole fraction MeOH 0.5 0.75 0.9 1 .o { NaCl-LiCl 3.7 (3.8) 3.35 (4.1) 4.2 (5.2) 4.1 (5.2) 1.85 (1.75) 4.15 (3.4) 2.1 (1.7) 2.5 (0.8) Na- Li NaF- LiF 5.55 7.5 6.9 6.6 NaCl-RbCl 0.3 (-3.9) -0.8 (-1.05) 1.1 (1.45) 0 (-1.6) Na- Rb NaF- RbF 2.15 1.4 1.9 - 2.4 1.85 (2.25) 2.2 (2.45) 0.8 (0.45) 2.4 (0.8) i l a rLiC1-RbCl -3.4 (-3.9) -5.15 (-5.95) 5.3 (6.65) -4.1 (6.8) 0 (0.5) 0.95 (0.15) 2.1 (0.75) 5.3 (2.6) Li-Rb -(LiFiRbF -3.4 - 6.1 7.4 - 9.4 LiCI- LiF - 1 .O 0.05 0.2 1.1 RbCl- RbF - 1 .O - 0.9 - 1.9 - 4.2 max A 1.75 3.15 2.7 5.3 Cl-F { NaCI- NaF - 2.75 - 3.1 -2.5 - 1.2 The preliminary study carried out in this laboratory using sodium-responsive glass and lanthanum fluoride F--responsive ion selective electrodes in aqueous hydrogen peroxide solutions of sodium fluoride l 2 has been extended to lithium, potassium and caesium fluoride solutions using cell (IV) alkali metal ion (M) I MF, ql_o;H;O, F- responsive LaF, (IV) responsive glass electrode For KF and CsF, ELL.K+-responsive electrodes and for LiF, E.I.L. Na+-responsive electrodes were used. Results are given in table 5 which includes those for sodium fluoride given previously l 2 in graphical form only. Cell e.m.f. values were steady after a few minutes even in the most concentrated hydrogen peroxide solutions. However, because of the attack of peroxide on the epoxy cement used to secure the lanthanum fluoride crystal in place, and the tendency to bubble formation on the electrode which caused variations in potential, the electrodes were left in the cells for a minimum possible time contingent upon reliable measurements being achieved. ion selective electrode.COMPARISON WITH SPECTROSCOPICALLY DERIVED DATA Elsewhere l9 it has been shown that the free energy of transfer AGFx) can be related to a free energy of preferential solvation AGE obtainable from n.m.r. or U.V. studies of the ions of the salt in the mixed solvent systems. For isodielectric solvent systems such as hydrogen peroxide + water mixtures these quantities were shown to be identical, i.e. AG& = AGg. Values of AGg for Li+, Na+, Cs+ and F- have been determined from n.m.r. measu~ements.~~ By combination of these values to obtain free energies of transfer for the appropriate salts, direct comparison is possible with the e.m.f.-derived data presented in table 5.This comparison has already been made elsewhere for NaF and is shown for LiF and CsF in fig. 4. As with NaF, the agreement is remarkably good and within 1 kJ mol-l in the free energy of transfer. For LiF, the agreement is particularly satisfying since Li+ is preferentially solvated by86 TRANSFER OF ALKALI METAL FLUORIDES water and F- by peroxide so there is a subtraction of contributions leading to a very small negative free energy of transfer as shown in fig. 4, and this system provides an exacting test of the theory. No. n.m.r. data are available for 39K chemical shifts in this mixed solvent system so comparison is not possible.TABLE 5.-vALUES OF AEt(x)/mV FOR CELL (IV) AND AQUAMOLAL FREE ENERGIES OF TRANSFER AGTx,/kJ mol-1 FROM H,O TO Hz02+ HzO LiF NaF KF CsF mole fraction - A G a -AGg)/ - 4 1 -A& of H 2 0 2 -AEt(x)/mV kJ mol-1 -AEt(x) /mV kJ mol-1 -AEt(&nV kJ mol-1 -AEt(x)/mV kJ mol-1 0 0.0310 0.0486 0.0909 0.125 0.133 0.140 0.180 0.282 0.322 0.330 0.376 0.404 0.416 0.427 0.484 0.554 0.569 0.584 0.730 0.753 0 0 0 0 7.6 0.73 13.5 1.30 19.7 1.90 19.3 1.86 0 0 0 0 52.7 5.09 34.2 3.3 8.1 0.78 28.4 2.74 35.3 3.41 17.8 1.72 22.7 2.19 43.9 4.24 101.5 9.79 76.7 7.4 113.8 10.99 97.7 9.427 54.8 5.29 28.0 2.19 42.0 4.05 70.0 6.75 139.8 13.494 155.6 15.086 126.2 12.177 149.2 14.396 I I I I 2 0 0>2 0.4 0.6 0.8 1.0 mole fraction peroxide FIG.4.-Free energy of transfer for CsF (0) and LiF (0) from water to hydrogen peroxide + water mixtures. Comparison of the e.m.f. data for methanol+water mixtures with n.m.r. data which requires an estimation of an electrostatic contribution in addition to the free energy of preferential solvation, in order to calculate the free energy of transfer, will be postponed until a later paper.A. K . COVINGTON AND J . M. THAIN a7 CONCLUSIONS It has been shown that ion-selective electrodes can be used to obtain free energies of transfer of alkali metal halides from water to methanol or to hydrogen peroxide+ water solutions, provided certain precautions are observed. It is necessary to use several cation-responsive glass electrodes of different glass compositions in order to ensure that no specific solvent effects are present.Few studies have been made of the response of these glass electrodes in solutions of pure electrolyte, aqueous or mixed aqueous, although selectivity studies have been reported.l* In the absence of a direct comparison with an established electrode responsive to the same ion in the same solvent mixtures, or of activity coefficient data, it is by no means easy to demonstrate that theoretical response is obtained in mixed solvent systems. However, if no hysteresis is shown when the electrodes are subjected to change of solvent composition and several different electrodes yield substantially similar results then confidence in the results is instilled. With these precautions and safeguards the method is useful for obtaining results of moderate accuracy (& 1 mV).We thank S.R.C. for a research studentship (to J. M. T.). The preliminary measurements on NaF+H,O,+H,O were made by Mr. Michael Wood. An account of this work was presented at the IUPAC-sponsored symposium on ion- selective electrodes held in Cardiff in April 1973. 0. Popovych, Crit. Rev. Anal. Chem., 1970,1,73. R. G. Bates, Determination ofpH(Wiley, New York, 2nd edn., 1973), pp. 211-234. G. Akerlof, J. Amer. Chem. Soc., 1930, 52, 5353. D. Feakins and P. J. Voice, J.C.S. Faraday I, 1972, 68, 1390. B. M. Lowe and D. G. Smith, Chem. Comm., 1972,989. M. S. Frant and J. W. Ross, Science, 1966,154,1553. A. K. Covington and J. M. Thain, J. Chem. Educ., 1972,49,554. K. Srinivasan and G. A. Rechnitz, Anal. Chem., 1968,40,509. J. N. Butler and R. Huston, Anal. Chem., 1970,42,1308. W. E. Bazzelle, Anal. Chim. Acta, 1971, 54, 29. lo J. J. Lingane, Anal. Chem., 1968,40,935. l2 A. K. Covington, K. E. Newman and M. Wood, Chem. Comm., 1972,1234. l 3 A. G. Mitchell and W. F. K. Wynne-Jones, Trans. Faraday SOC., 1955,51,1690; 1956,52,824. l4 A. K. Covington, K. E. Newman and T. H. Lilley, J.C.S. Farahy I, 1973,69,973. l5 R. A. Robinson and R. H. Stokes, ElectroZyte Solutions (Butterworth, London, 2nd. rev. edn., l6 A. K. Covington and T. Dickinson in Physical Chemistry of Organic Solvent Sys?ems (Plenum, l7 I. S. Ivanovskaya, V. I. Gavrilova and M. M. Shults, Souiet Electrochem., 1970, 6, 975. l 8 G. Eisenman in Aduances in Analytical Chemistry and Instrumentation (Wiley, New York, 1965), l9 A. K. Covington, T. H. Lilley, K. E. Newman and G. A. Porthouse, J.C.S. Faraday I, 1970), p. 30. London, 1973), p. 18. vol. 4, pp. 295-305 (reprinted in The Glass Electrode, Interscience Reprint, 1966). 1973, 69, 963.