730 Analyst, August, 1979, VoL 104, PP. 730-738 Ion-selective Polymeric-membrane Electrodes with Immobilised Ion-exchange Sites Part I. Development of a Calcium Electrode L. Ebdon, A. T. Ellis and G. C. Corfield Department of Chemistry, She$eld City Polytechnic, Pond Street, Shefield, S1 1 WB A new type of ion-selective electrode is described in which ion-exchange sites are immobilised in a polymeric membrane by covalent bonding. Membranes were prepared by cross-linking a styrene - butadiene - styrene triblock copolymer with triallyl phosphate. After subsequent hydrolysis these membranes were evaluated as Ca2+ sensors. The electrodes formed exhibited an extended Nernstian response to Ca2+ (10-'-10-.6 M) and good selectivity over other alkaline earth and alkali metals. Such electrodes offer very fast response times, extended lifetimes and a wide y H working range.The possibilities of this new class of ion-selective electrode are also discussed. Keywords : Ion-selective electrode ; calcium analysis ; electroanalytical chemistry ; polymeric membrane ; immobilised ion-exchange groups Ion-selective electrodes offer considerable advantages in analytical chemistry. They are simple and inexpensive to operate, sensitive and suited to on-line measurement. Thus, in addition to their application to laboratory analyses, applications to pollution monitoring, in vivo biological measurements and process control have been proposed. The major problems with ion-selective electrodes seem to be that certain of them lack selectivity and mechanical strength. The range of ion-selective electrodes until about 10 years ago was limited to classical glass electrodes or electrodes using certain crystalline membranes. The introduction of electrodes based on liquid ion exchangers, e.g., the calcium electrode first reported by Ross1 in 1967, greatly extended the range of available electrodes.Ross used rather impure calcium didecyl phosphate (DDP) as sensor in conjunction with dioctylphenyl phosphonate (DOPP) solvent mediator supported on a Millipore filter. This type of electrode has been marketed com- mercially (e.g., Orion 92-20). Such electrodes were recognised as a significant advance and were the first of a range of liquid ion exchanger ion-selective electrodes. In their design it was necessary to keep three liquids, the internal filling solution, the ion exchanger and the test solution, in electrolytic contact but to prevent mixing.This posed some constraints as did the limited lifetime of the electrodes, caused by leaking of the DDP and DOPP. This leaking also tends to raise the obtainable limit of detection. Severe interference froin H+ and some heavy metals and the expense of construction have also been cited as drawbacks of the Ross-type calcium ion-selective electrode. In 1970, Moody et aZ.2 reported a modification of the Orion system in which they used pure DDP and DOPP solvent mediator incorporated into a poly(viny1 chloride) (PVC) matrix. This convenient construction had the benefits of cheapness and ease of fabrication over the Ross system, but such electrodes still had a limited lifetime as the exchanger was entangled in the PVC matrix and so could still be leached out.The advantages of the PVC system were, however, obvious and it can be simply constructed3 or purchased from commercial suppliers, finding widespread use for many applications. Since this early work, the sensor and solvent-mediator systems have been the subject of much research, resulting in further improvements in the Ca2+ ion-selective electrode. In particular, it is now recognised that the solvent mediator performs an essential function in controlling the selectivity of the liquid ion exchanger electrodes. As an example, the use of calcium bis [di(2-ethylhexyl) phosphate] as sensor results in a Ca2+ ion-selective electrode if DOPP is used as solvent and as a divalent (water hardness) Ca2+/Mg2+ test electrode if decan-1-01 is used.4 From work carried out by RdiiCka et aL5 and later by Moody et aZ.,6 a better sensor for Ca2+ ion-selective electrodes seems to be calcium bis[di-4-(1,1,3,3-tetra- methylbutylphenyl) phosphate] with DOPP.Work has also been carried out on the effectsEBDON, ELLIS AND CORFIELD 73 1 of nitrating7ss the sensor and/or the mediator, but although giving useful information on the synergistic sensor - mediator system, nitration gave no real improvement in electrode perf onnance. In addition to the use of the organophosphoric acid type of sensor, good results have been reported for a Ca2+ ion-selective electrode using the non-cyclic neutral carrier NN'-di [ (1 1- ethoxycarbonyl)undecyl]-NN'-4,5-tetramethyl-3,6-dioxaoctanediamide as sensor and 2- nitrophenoxyoctane as solvent rnediat~r.~ Although this system also shows high Ca2+ selectivity, it is also of the PVC matrix type and so the system will be leachable and subject to limited lifetimes, particularly when used in flowing systems.A neutral carrier system has also been used in which the tetraphenylborate salts of a calcium adduct of poly(propy1ene glycol),l0 in conjunction with DOPP, is entangled in a PVC matrix. The resulting electrodes showed, in general, poorer Ca2+ selectivity than the phosphate systems and had shorter lifetimes. I t can be seen that there has been much work directed towards understanding and therefore improving the sensor exchanger - mediator system, but very little work has been carried out with different matrices.Schultz et aZ.ll made an effort in 1968 to incorporate a dialkyl phosphate sensor into collodion, with little success, and in 1972 Griffiths et aZ.12 reported poor electrode quality for some cellulose, collodion and pyroxylin entangled membranes. They suggested that the hydrophilicity of these materials is undesirable as leaching of the exchanger takes place very quickly. Schafer13 used poly(viny1 isobutyl ether) as a matrix material to prepare a divalent ion electrode, which showed no advantage over the PVC system and was again a polymer-entangled system from which the exchanger could be leached. The advantages of PVC for electrode fabrication arise from its ready availability and lack of hydrophilicity. PVC is, however, stabilised by organometallic compounds that can block the ion-exchange sites and when these are removed its stability, e.g., to ultraviolet light, is poor.More significantly, the ion exchanger in the membranes described above is not covalently bound. Thus, eventually the ion exchanger will migrate from the membrane and the resultant deterioration in electrode response as this leaching proceeds limits the useful lifetime of the membrane. We are therefore studying different polymer systems and the possibility of immobilising ion-exchange sites into polymer matrices by covalent bonding. The immobilisation of the active ion-exchange groups should overcome the problems associated with leaching referred to above and lead to the development of ion-selective electrodes with extended lifetimes and sensitivity.Additionally, the careful selection of polymer systems should perrnit optimisa- tion for enhanced mechanical and electrochemical properties. This new approach to the fabrication of ion-selective electrode membranes could lead to greater control over steric properties and hence to a significant improvement in selectivity. Two other groups of workers have also directed attention towards covalently bonding ion-exchange groups to a polymer, although their work became known to us only after we had started the study described here. Keil et aZ.14 phosphorylated a vinyl chloride - vinyl alcohol copolymer with decyl dihydrogen orthophosphate. The copolymer was then mixed with PVC in which DOPP was entangled, to produce a Ca2+ ion-selective electrode.Extended electrode lifetimes were not obtained in their study. bound ion-exchange groups to the ends of PVC chains both by the use of an amine as chain transfer agent during polymerisation and by the use of the 'SO3- radical anion as polymerisation initiator. The electrodes produced exhibited selectivity towards anionic or cationic surfactants and, although the electrode lifetimes were extended in comparison with the liquid ion exchanger surfactant ion-selective electrodes, the plasticiser used (tricresyl phosphate) was leached from the polymer and this limited the lifetime of the electrodes. In the study reported here, membranes were prepared by cross-linking a styrene- butadiene - styrene (SBS) triblock copolymer with triallyl phosphate using free-radical initiation.After subsequent hydrolysis to give covalently bound pendant dialkyl phosphate exchanger groups, these membranes were evaluated as sensors for Ca2+ ion-selective electrodes. Cutler et Experimental Chemicals Tetrahydrofuran (THF) was freshly distilled from aluminium lithium hydride to dry it732 EBDON et al. : ION-SELECTIVE POLYMERIC-MEMBRANE Anallyst, VoZ. 104 and to remove stabilisers. Triallyl phosphate (TAP) was used as received (Aldrich Chemical Co.) but the initiator a,a-azobisisobutyronitrile (ABIN) was recrystallised from methanol. Analytical-reagent grade chemicals were used during electrode evaluation. The polymer used was Cariflex SBS 1101 (Shell Chemicals, London), which was purified by dissolving it in THF and re-precipitating it in cold, well stirred methanol.Gel permeation chromato- graphy showed the number-average relative molecular mass (Mn) as 9.8 x lo4 g mol-l with 70% SBS triblock poly(styrene-h-butadiene) , 26% SB diblock and 4% homopolystyrene content. The polydispersity (Mw/Mn) was 1.51 and 100-MHz nuclear magnetic resonance spectroscopy indicated 27% m/m of polystyrene in the polymer, and hence 73% m/m as polybutadiene ; 300-MHz nuclear magnetic resonance spectroscopy showed the butadiene units to be 90% 1,4- and 10% 1,Zconfiguration. Preparation of Polymeric Membranes Re-precipitated SBS (4 g) was dispersed in the solvent (40 cm3) and allowed to dissolve overnight, after which the required amounts of TAP and ABIN were added with mixing. The polymer solution was then poured into a glass ring (100 mm i.d.) resting on a Cellophane sheet on a glass plate.A heavy weight and a front-silvered mirror were arranged such that a seal was maintained with the plate and ultraviolet radiation could be reflected normally on to the curing membrane. Membranes were usually cured after 6 h, depending on solvent volatility, forming strong, clear membranes about 1 mm in thickness. The membranes were removed from the Cellophane and portions hydrolysed to give the master membrane. Evaluation of Electrodes Discs, 10- in diameter, were cut from the master membranes and stuck to the end of clear, plasticised PVC tubing by using a cyanoacrylate adhesive (IS 12, Loctite, Dublin). Calcium chloride solution (10-1 M) saturated with silver chloride was used as the internal reference solution, the electrodes being mounted on modified pH electrodes in a similar fashion to PVC membrane electrodes3 The mounted electrodes were soaked overnight in M Ca2+ solution in order to replace the Na+ or K+ form of the exchanger with the Ca2+ form.The time required for soaking was again dependent on the degree of cross-linking, but 12 h was the maximum required by any electrode. EMF measurements were made using the cell - - Hg.HgC1, I KCl (sat.) 11 test solution I membrane I CaC1, (0.1 M) I AgC1.Ag using varying test solutions. A single-junction reference electrode (Orion 90-01, Orion Research Inc., Cambridge, Mass., USA) and a digital voltmeter (Orion 701) reading to h O . 1 mV were used. The test solutions were stirred magnetically and maintained at 25 & 0.5 "C.Calcium-ion activities were calculated from ac,2+ = cy where y, the activity coefficient, was calculated using an extension of the Debye - Huckel equation: -log y(Q+ = z2 (;yd7 -- - 0.H) where I , the ionic strength, is given by I = &2cizi2. Potentiometric selectivity coefficients were determined by a mixed solution method using an interferent-ion level of 10-3~and changing the calcium-ion concentration. Coefficients were evaluated using the IUPAC recommended rnethodl6 : where z = 2 and n = charge of interference ion M, a,*+ = M and a,,2+ is taken at the point when the mixed solution calibration differs from the extrapolated linear portion by l8/2 mV (Le., 9 mV for Ca2+). Hydrogen-ion interference is expressed as the pH range over which there is no measurable change in electrode potential at a constant calcium-ion activity.The pH was changed by using 0.1 M hydrochloric acid and 0.1 M sodium hydroxide solution in conjunction with a combination glass pH electrode (Activion Ltd., Halstead, Essex) and a pH meter (Pye, Model 290, Pye Unicam, Cambridge).August, 1979 ELECTRODES WITH IMMOBILISED ION-EXCHANGE SITES. PART I Results and Discussion 733 Incorporation of Ion-exchange Groups The ion-exchange group chosen for this study was the classical dialkyl phosphate group, as this offered a well tried system that could be incorporated into a polymeric structure by a simple copolymerisation mechanism. A poly(styrene-b-butadiene) triblock elastomer (SBS) was chosen as the polymer because it contains the necessary C=C unsaturation for cross- linking, has mechanical behaviour similar to natural rubber vulcanisates without requiring cross-linking, can be dissolved in solvents and is easy to process.In order to incorporate the phosphate exchanger groups, the unsaturated C=C bonds are cross-linked by a free- radical initiated addition mechanism using triallyl phosphate (TAP). This mechanism can be represented by the scheme shown in Fig. 1. uv (CH,),CN=NC(CH,), - 2(CH,),C' + N, I CN 1 1 CN CN ABlN R' R'+ - C H 2 ~ H - C H 2 C H = C H C H , - - ~ ~ ~ c ~ - ~ ~ , ~ ~ I I - S B S -S-CH,CH=CHCH,-CH,CH-S- I ),CH R-CH, (CH2=CHCH20)3P=0 I I -S-CH,CH=CHCH,-CH,CH-S - CH / \ R-CH, CH,-'CHCH o 2 \ ,P=O (3) (C H 2=C H CH , 0) , S-€34 I 0 II I ,CH2 I I -S-CH, CH=CHCH2-CH2CH- S CH A\ R-CH, CH,-CHCH20-P(OCH2CH=CH,), ( 4 ) 'CH -S-CH,CH=CHCH,--CH,CH-S - Fig.1. Immobilisation of phosphate groups. This simplified mechanism shows that initiation of SBS should occur [(l) and (2)], leading to incorporation of phosphate groups in the structure via covalent bonding (3) and cross- linking (4). In order to obtain the dialkyl phosphate sensor unit, it is necessary to hydrolyse the trialkyl phosphate system with alkali (Fig. 2). The stability of the resonance-stabilised dialkyl phosphate salt is such that further hydrolysis to remove other alkyl groups is very difficult and will not proceed under the conditions used. This leaves the dialkyl phosphate734 EBDON et aZ. : ION-SELECTIVE POLYMERIC-MEMBRANE Analyst, VoZ.104 group covalently bound to the polymeric membrane. Strong alkaline hydrolysis results in attack on the polymer itself, presumably at the residual unsaturation, and at a sodium hydroxide concentration of approximately 25% m/m the polymeric material breaks down completely. One disadvantage of this hydrolysis procedure is that the K+ or Na+ salt of the acid is formed and so the electrodes must be conditioned overnight in 1 0 - 2 ~ Ca2+ to create the Ca2+ form of the exchanger. OH - + +-OK Fig. 2. Hydrolysis of trialkyl phosphate grouping. The initial stage of this work was concerned with optimising the polymerisation and hydrolysis steps. The first consideration was the amount of cross-linking in the membrane, which is controlled primarily by the amount of TAP and ABIN present, with factors such as temperature and light intensity of lesser importance.Table I shows the effect of changing the amount of TAP and ABIN on the physical and electrochemical properties of the cast membrane. From these results, it can be seen that the best membrane (B) resulted from 5% m/m of TAP with 2.5% m/m of ABIN. High levels of TAP resulted in oily, non- functional membranes, and larger amounts of ABIN reduced the response and the clarity of the membrane. TABLE I EFFECT OF MEMBRANE COMPOSITION ON PHYSICAL AND ELECTROCHEMICAL PROPERTIES TAP, ABIN, THF Physical properties of Membrane % m/m yo m/m solubility* cast membranes A .. . . 2.5 0 Soluble Elastic, clear B . . .. 6 2.5 Insoluble Elastic, clear, slightly c . . . . 2.5 5 Insoluble Elastic, opaque D .. . . 5 10 Insoluble Clear with opaque E . . . . 5 15 Insoluble Bubbled, opaque F .. . . 15 2.5 Insoluble Elastic, clear, yellow G .. . . 50 5 Insoluble Oily, yellow, rigid H .. .. 100 6 Insoluble Thick, rigid with oily yellow inclusions inclusions surface * Before hydrolysis. t After hydrolysis for 6 h in 1% sodium hydroxide solution under reflux. Electrochemical propertiest Short linear range, low Functional membrane slope No response No response No response Very short, linear range, low slope Poor response Poor response Table I1 shows the effects of both the hydrolysis conditions and the degree of cross-linking on the properties of the membrane. The use of mild aqueous alkaline conditions for hydrolysis gave membranes that showed near-Nerns tian slopes but with short lifetimes.Those with longer lifetimes had significantly sub-Nernstian calibration slopes. Aqueous hydrolysis produced membranes that were more soluble in THF than the unhydrolysed membranes, presumably owing to further attack on the polymer structure in addition to hydrolysis of phosphate groups. Hydrolysis with 5% m/m methanolic potassium hydroxide solution proceeded smoothly, resulting in an insoluble membrane with a Nernstian response and an extended lifetime (membrane 5 ) . Reductions in the amounts of TAP and ABIN eventually gave membranesAugust, 1979 ELECTRODES WITH IMMOBILISED ION-EXCHANGE SITES. PART I TABLE I1 EFFECT OF MEMBRANE COMPOSITION AND HYDROLYSIS CONDITIONS ON THE PHYSICAL AND ELECTROCHEMICAL PROPERTIES OF POLYMERIC ION-EXCHANGE ELECTRODES TAP, ABIN, Membrane yo m/m % m/m Hydrolysis* 1 5.0 2.5 5 h, 1% aq.NaOH 2 4.5 2.5 3h,1% aq.NaOH 3 4.86 2.6 1h,1% aq.NaOH 4 4.54 2.5 20 h, H,O 5 4.4 2.5 5h,5y0 KOH-MeOH 6 4.4 2.5 6h,10% KOH-MeOH 7 2.2 1.26 6h,6% KOH-MeOH 8 2.2 1.26 5h, 10% KOH-MeOH 9 1.0 0.55 6h,5% KOH-MeOH 10 1.0 0.55 5 h, 10% KOH - MeOH 11 0 0 5h,6%KOH-MeOH Solubility Slope/ in THF mV decade-' Soluble + 16 Soluble (12 h) + 24.2 Slightly soluble + 29.5 Slightly soluble + 29 Insoluble + 30 Slightly soluble + 35, decreasing Insoluble + 24. decreasing to +17 after 20 d to +is after 3 ;i Soluble - 19 Soluble + 25 Soluble - 12 Soluble - 17 735 Lifetime >6 months >6 months 10 days -3 days >6 months 20 days 5 days 3 days 3 days 3 days > 3 months Appearance Milky, opaque Clear, becoming yellow - brown on hydrolysis Mainly clear, yellow on hydrolysis Dark brown and brittle Clear, pale yellow, tough and elastic Clear, yellow - brown Clear, pale yellow Brown, becoming brittle Clear and colourless Brown Clear and colourless.Pure SBS membrane * Conditions refer to reflux. After hydrolysis. that, after hydrolysis under similar conditions to those used with membrane 5, did not appear to be strongly cross-linked and showed a calibration slope below Nernstian with a short life- time. When the hydrolysis was more vigorous, a similar increase in solubility and a marked tendency to age became apparent (membrane 6). Membranes 8 and 10 showed negative calibration slopes, which may have been due to the passage of small anions through these membranes, which were not cross-linked.Porous membranes would be expected to respond to C1- ions because of the internal reference electrode. The properties of membrane 5 provide clear evidence for the immobilisation of the phosphate sensor in the polymer structure by covalent bonding. The insolubility of the polymeric material indicates that cross-linking has occurred and the presence of C-0-P and P=O absorption bands in the infrared region, using both transmission and attenuated total reflectance methods, shows the incorporation of phosphorus into the membrane. Quantitative phosphorus analysis was carried out by using acid digestion followed by a molybdenum blue spectrophotometric method. Tests on membrane 5 showed no leaching of phosphorus even after 10 d in THF, confirming the covalent binding of the phosphate group to the polymer structure.Evaluation of Electrode Response In all instances, the range of linear response is wide and in some instances extends below 1 0 - 6 ~ , although the use of diluted standards rather than a calcium-buffered system at this level is unlikely to give reliable results. This significantly extended performance on the low concentration side TABLE I11 RESPONSE AND SELECTIVITY DATA FOR POLYMERIC ION-EXCHANGE ELECTRODES Table I11 shows the electrode responses using the membranes prepared. Effective linear Membrane range/M Slope a t 25 OC/mV decade-' 1 10-6-10-2 + 16 2 10-6-10-1 +24.2 3 10-6-1 0-1 + 29.5 4 10-6-10-1 + 29 5 lo-a-lo-' 4- 30 6 10-6-10-1 + 35, decreasing to 7 10-0-10-1 + 24, decreasing to $17 after 20 d +15 after 3 d 8 10-'-10-' -19 9 10-0-10-' + 25 10 10-~-10-' - 12 11 10-~-10-' - 17 For decade change of 10-6-10-4 M Caa+ concentration.t Concentration of interferent ion 10-8 M. Static response time*/s 5 5 -120 5 2 2 5 5 6 5 6 Selectivity coefficient,t $gtM Ba Mg K Na pH range 0.3 2 00 30 5-10 0.80 0.82 0.35 0.17 4.5-10 0.4 -lo2 -10' Major interference from all cations 0.8 0.3 2.5 6 4-10 0.65 0.2 12 6 4-10 0.8 Initially negligible but 20 mV pH-I increases with ageing Responding to anions 0.17 7 20 4.5-10 Responding to anions Responding to anions736 EBDON et al. : ION-SELECTIVE POLYMERIC-MEMBRANE Analyst, Vol. 104 compared with the Orion 92-20 type is thought to be due to the non-leachable ion-exchange groups and a calcium-buffered system might be expected to extend the range below M.The range above 10-1 M Ca2+ has not been investigated, the range of concern in this study being 10-5-10-2 M Ca2+. Some initial comments on the calibration slopes of some electrodes have been made above. The results obtained confirmed that optimum electrochemical properties were obtained with electrodes that were cross-linked and contained a high density of bonded phosphate groups in the membrane. It was not possible to increase the phosphate content indefinitely without producing membranes with poor physical characteristics. Early results showed that it was necessary to use a sufficient intensity of ultraviolet irradiation during polymerisation to ensure cross-linking. Membranes 1 and 2 gave electrode slopes of +16 and +24.2 mV decade-l, and this was attributed to insufficient radiation, which resulted in poor initiator efficiency and hence lack of cross-linking.Membrane 11 was a pure SBS membrane and when treated under conditions identical with those used with membrane 5 gave a THF- soluble membrane with a negative calibration slope. Again, we would conclude that this pure SBS membrane is porous to small anions. Response times were generally very short, with static times being in the region of 5 s for the range 10-1-10-4 M Ca2+, increasing to 15 s for the and lo4 M solutions. Dynamic response times were measured by making 10-fold changes in Ca2+ concentration by addition of concentrated calcium chloride solution from a piston syringe to 100 cm3 of well stirred solution.The whole range from 10-6 to 10-1 M Ca2+ was covered in this manner and gave response times of 1 s for most solutions, only increasing to about 5 s for 10-6 M solutions. The reason for these fast response times compared with other organophosphorus-based electrodes is thought to lie in the mechanism of the electrode. In the liquid ion exchanger and PVC electrodes, the response relies on an ion-transport mechanism, whereas in this type of exchanger the mechanism might in some ways be analogous to the ion-exchange mechanism found in glass electrodes. Selectivity studies were initially performed on a limited number of cations showing the general order of selectivity to be Ca2+ > Ba2+ > Mg2+ > M+. Potentiometric selectivity coefficients for the divalent ions were kzFBa = 0.8 and kglt,, = 0.3, whilst interference from monovalent cations was less, thus making it possible to generate the K+ or Na+ form of the exchanger on hydrolysis and then to convert this into the Ca2+ form by soaking over- night in 10-1 M calcium chloride solution. These selectivity data are also presented, perhaps more meaningfully, in calibration graphs as in Fig.3, where the interferent level is maintained at The influence of increasing sodium concentration on kzitN, and electrode response is presented in Fig. 4, showing this type of electrode to be functional in the presence of 10-2 M Na+, but with loss of Ca2+ response at the 10-1 M Na+ level. The H+ interference was evaluated by varying the pH at a constant Ca2+ concentration M) (see Fig.5). Table I11 includes the pH range over which there is no measurable change in the electrode potential for different electrodes. With the electrodes reported M. > E ‘3 .- *.’ 0) Q a3 0 *.’ - - - 10-6 10-5 10-4 lo-’ Activity, aCa2+ Fig. 3. Calibration graphs for a calcium ion- selective polymeric-membrane electrode with immobilised ion-exchange sites. The inter- ferent effect of Ba2+, Mg2+, K+ and Na+ solutions ( l o - 3 ~ as chloride in each instance) on this calibration is also shown.August, 1979 ELECTRODES WITH IMMOBILISED ION-EXCHANGE SITES. PART I 737 50 > E 1. .- *I’ C 8 0 n - s -50 ’ 1O-’M Na Activity, aCa2+ Fig. 4. Effect of Na+ concentration on the calibration graph for a calcium ion-selective membrane electrode with immobilised ion-exchange sites.here, for well cross-linked electrodes the pH range was wide (pH 4-10) and showed none of the “dips” seen with earlier organophosphate-based systems. This is perhaps to be expected as these “dips” were attributed to the solvent mediator and the system here has no such mediator. The useful alkaline end is limited analytically by the formation of calcium hydroxide, although the electrode may still be capable of giving the true Ca2+ activity. Poorly cross-linked membranes showed narrower pH working ranges and were unstable at low pH values; there was, however, no poisoning of the electrodes by Hf. 2 3 4 5 6 7 8 9 10 11 PH Fig. 5. Graph of electrode response against pH in 10-3 M calcium chloride solution. In terms of general analytical behaviour, the electrodes yielded a reproducibility of & 1 mV, but the poorly cross-linked membranes resulted in poorer reproducibility, especially when they also had a limited lifetime.Drifts with the better electrodes were of the order of 1 mV d-l, but again the poorer cross-linked electrodes gave higher drifts (about 5 mV d-l) and increased noise levels. The lifetime of this type of electrode was found, as predicted, to be significantly longer than those of polymer entrapped liquid ion exchanger electrodes. Poorly cross-linked membranes lasted only a matter of days, but the well cross-linked membranes gave electrodes with lifetimes in excess of 6 months. These latter membranes showed no physical deteriora- tion and the electrode response remained Nernstian with good selectivity. In the less cross-738 EBDON, ELLIS AND CORFIELD linked membranes, shorter lifetimes of less than 1 month are accompanied by a noisy response, decreasing calibration slope and increasing monovalent cation interference.The lifetimes quoted are for electrodes stored in 1 0 - 2 ~ Ca2+ solution and calibrated at least once per week. Conclusion It is possible to form immobilised ion-exchange sites in an unsaturated polymer matrix by cross-linking SBS with triallyl phosphate and hydrolysing the resulting membrane to give pendant dialkylphosphoric acid exchange groups. The best membrane was prepared by cross-linking 4 g of SBS with 4.5% m/m TAP and 2.5% m/m ABIN initiator and hydrolysing the polymer with methanolic potassium hydroxide solution ; the resulting membrane gave an ion-selective electrode of high physical strength and stability.This ion-selective electrode gave a Nernstian response to Ca2f ions with selectivity over Ba2+, Mg2+ and alkali metal cations. It showed a wide pH working range (pH 4-10) and a lifetime in excess of 6 months. Further work on the polymer system and ion-exchange sites is expected to result in a new family of ion-selective electrodes extending to other ions. Early indications are that this type could combine improved mechanical properties, such as robustness and lifetime, with advantageous electrochemical properties, including fast response speeds and lower limits of detection. We are grateful to the Trustees of the Analytical Chemistry Trust Fund for the award of an SAC studentship to one of us (A. T. E.), which has made this work possible. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. References Ross, J. W., Science, N.Y., 1967, 156, 1378. Moody, G. J., Oke, R. B., and Thomas, J. D. R., Analyst, 1970, 95, 910. Craggs, A., Moody, G. J., and Thomas, J. D. R., J . Chew Educ., 1974, 51, 541. Craggs, A., Keil, L., Moody, G. J., and Thomas, J. D. R., Talanta, 1975, 22, 907. RbiiCka, J., Hansen, E. H., and Tjell, J. C., Analytica Chim. Acta, 1973, 67, 155. Moody, G. J., Nassory, N. S., and Thomas, J . D. R., Analyst, 1978, 103, 68. Jagner, D., and 0stergaard-Jensen, J. P., Analytica Cham. Acta, 1975, 80, 9. Keil, L., Moody, G. J., and Thomas, J. D. R., Analytica Chim. Acta, 1978, 96, 171. Amrnann, D., Giiggi, M., Pretsch, E., and Simon, W., Analyt. Lett., 1975, 8, 709. Jaber, A. M. Y., Moody, G. J., and Thomas, J. D. R., Analyst, 1977, 102, 943. Schultz, F. A., Petersen, A. J., Mask, C. A., and Buck, R. P., Science, N.Y., 1968, 162, 267 Griffiths, G. H., Moody, G. J., and Thomas, J. D. R., Analyst, 1972, 97, 420. Schafer, 0. F., Analytica Chim. Acta, 1976, 87, 495. Keil, L., Moody, G. J., and Thomas, J. D. R., Analyst, 1977, 102, 274. Cutler, S. G., Meares, P., and Hall, D. G., J . Electroanalyt. Chem., 1977, 85, 145. IUPAC, Pure Appl. Chem., 1976, 48, 129. Received October 31st, 1978 Accepted Muvch 13th, 1979