J . Chem. Soc. Faraday Trans. I 1986,82 1051-1070 Sensors from Polymer Modified Electrodes Mark W. Espenscheid Amiya R. Ghatak-Roy Robert B. Moore 111 Reginald M. Penner Marilyn N. Szentirmay and Charles R. Martin" Department of Chemistry Texas A and A4 University College Station Texas 77843 U.S.A. The construction and operation of ionomer-film modified electrodes are described including the fundamentals of the ion-exchange processes. Ion- exchange selectivity coefficients for a series of alkylsubstituted pyridinium cations in Nafion have been measured and the large values of these coefficients are attributed to the hydrophobic effect; a free-energy relation- ship relating the coefficient to the size of the ion has been established. The use of Nafioncoated electrodes for the in uiuo measurement of neurotrans- mitters and for the preconcentration of analytes present in low concentra- tion is discussed Experiments on the dynamics of the response of ionomer modified electrddes are reported and very high ionic transport rates can be achieved for novel ionically conductive composite polymer membranes.The covalent attachment of redox groups such as ferrocene allows one to control the release of counterions by switching the redox state of the ferr ocene . This paper will deal primarily with ionomer-film modified electrodes. An ionomer is a linear or branched organic polymer which contains covalently attached ionizable groups. Ionomers generally contain rather low quantities of ionizable groups (i.e. less than ca.10 mol% ionizable monomers) and are not covaIently cross-linked. Thus ionomers are chemically and morphologically distinct from conventional organic ion-exchange materials (e.g. styrenesulphonate-based resins). The structures for the two ionomers to be discussed in this paper are shown in fig. 1. Ionomer-film modified electrodes were prepared by coating ca. 1 pm thick ionomer films onto vitreous carbon or Pt surfaces using conventional solution-casting techniques2 Ionomers are suitable for a variety of electrochemical processes and devices including energy conversion and storage systems3 and industrial electrolytic pro~esses.~ Nearly all of these applications rely on the ability of the ionomer to exchange and transport counterions and reject co-ions (ions of the same charge as the fixed site).It seems likely that these properties could be of use in an electroanalytical system and we are investigating the possibility of using ionomer-film modified electrodes as electrochemical sensors. This paper will focus on fundamental aspects of ionomer-modified electrode sensor response. Topics to be discussed include (1) the thermodynamics of ionomer- counterion interactions (2) the dynamics of ionomer modified electrode response and ( 3 ) in situ film regeneration in ionomer modified electrodes. Thermodynamics of Ionomer-Counterion Interactions Fundamentals of the Ion-exchange Process in Perfluorosulphonate Ionomers Our interest in this subject stems from observations made in Prof. Freiser's laboratory concerning the selectivity of Nafion (see fig.1) membrane-based ion-selective electrodes.6 These electrodes showed remarkable affinities for large hydrophobic cations (e.g. alkylammonium ions) relative to simple monovalent or divalent inorganic cations.6 This unusual ion-exchange selectivity was later echoed in voltammetric studies on Nafion-film 1051 1052 0 I 0 I SO3H Polymer Modified Electrode Sensors C I FZ CF-CF3 0 I I C F2- C F2 - SO H ( a ) C I F2 c I F ( b ) Fig. 1. Structures of perfluorosulphonate ionomers. (a) DuPont Nafion (b) Dow PFSA. modified electrodes conducted in Prof. Bard's laboratory.' From a qualitative point of view these studies suggest that if Ogn+ represents a hypothetical hydrophobic organic cation the position of equilibrium in the ion-exchange reaction shown in eqn (1) would lie very far to the right Ogn+(aq) + nNa+(film) -+ Ogn+(film) + nNa+(aq).Ogn+(ag) + Ogn+(film). (1) If this is the case the ionomer film could be used to extract and preconcentrate hydrophobic organic cations from a contacting aqueous phase. By coating the film on an electrode surface a subsequent electrochemical experiment could be used to quantitate the film entrapped cation (vide infra). The extent of the preconcentration effect depends on the magnitude of the ion-exchange selectivity coefficient. Since no ion-exchange data for hydrophobic cations in Nafion had been reported in the literature we set out to determine selectivity coefficients for a variety of organic cations in Nafion.An electrochemical ion-exchange isotherm methods and a chromatographic method9 were used to obtain these ion-exchange data. In the isotherm method a Nafion film-modified electrode (initially in the Na+ form) was exposed to a very dilute (ca. lo+ mol dm-3) solution of the electroactive cation whose selectivity coefficient was to be measured (the solution was also 0.1 mol dm-3 in supporting electrolyte NaClO,). The film was then equilibrated with the solution by rapid rotation of the electrode. Depending on the ion and its concentration this process could take as long as 25 h.8 When film-solution equilibrium was achieved the quantity of electroactive ion exchanged into the film was determined coulometrically.8 This procedure was then repeated for increasingly higher concentrations of the electroactive ion.The extent of the exchange of the electroactive cation (Ogn+) for Na+ in the Nafion film can be expressed in terms of both a distribution coefficient k, and an ion-exchange selectivity coefficient at. kD is the equilibrium coefficient for the hypothetical (vide infra) partition reaction ( 2 ) More importantly kD is the slope of the partition isotherm for the electroactive counterion (see fig. 2) and as such gives a measure of the extent of the ion-exchange reaction. is the equilibrium coefficient for the ion-exchange reaction shown in eqn (1) and is given bylo where the a are activities in the solution phase and the x are the equivalent ionic fractions of -SO; sites occupied by each ion in the film phase.Details of this method and of the calculations involved are presented in ref. (8). Fig. 3 shows voltammograms for Nafion-coated electrodes after equilibration with 1 M. W. Espenscheid et al. I . . . . I . . . . 1 . . . 1 . 4 2.5 0 solution concentration/ 1 0-8 mol dm-3 10 5 solution concentration/ 1 0-8 mol dm-3 2 3 solution concentration/ 1 O-' rnol dm-3 1.5 0.5 1 2 solution concentration/ lo-' mol dm-3 Fig. 2. Ion-exchange isotherms for (a) Ru(bpy)g+ (bpy = 2,2-bipyridine) (b) Ru(NH,)i+ (c) methyl viol ogen and (d) ferrocen ylme thy1 trime t h ylammoni um . very dilute solutions of the various electroactive counterions. These voltammograms are the first indicators of the magnitudes of the ion-exchange selectivity coefficients for these electroactive counterions.No signals above background currents would be detected at uncoated electrodes in these extremely dilute (< 5 x mol dm-3) solutions. Clearly as suggested above the Nafion film preconcentrates these cations and the equilibrium film concentrations are orders of magnitudes higher than the solution concentrations. Ion-exchange reactions should be described by an equilibrium-coefficient expression like that shown in eqn (3) which takes into account the solution and film concentrations of both the exchanging and exchanged ions. However when the concentration of the exchanged ion (Na+) in the aqueous phase is high and the concentration of the exchanging ion (Ogn+) in the film is very low the aNa and zNa terms in eqn (3) are constant and the reaction may be regarded as a partition reaction [eqn (2)].119 Simplifying the ion-exchange reaction to a partition reaction is attractive because the extent of the reaction may be conveniently visualized by plotting a partition isotherm ([Ogn+],il us.[Ogn+Iw)3 l4 Fig. 2 shows partition isotherms for the various electroactive cations studied here. The linear portions of these isotherms show the (low) concentration regions over which it is appropriate to use eqn (2) to describe the ion-exchange process. The slopes of these 0 0 1053 l2 1054 I I I I 1.2 0.8 -0.2 I Polymer Modified Electrode Sensors u I I 1 I 0.2 I I -0.4 0.0 0.8 I I I -0.6 I -0.8 (c) 4.36 x 0.4 EIV us.SCE Fig. 3. Cyclic voltammograms recorded at a scan rate of 0.1 V s-l for Nafion-coated electrodes in equilibrium with (a) 1.51 x mol dm-3 Ru(bpy)$+ (6) 2.78 x mol dm-3 MV2+ ( d ) 3.40 x 0.100 mol dmP3 NaClO,. mol dm-3 Ru(NH3):+ mol dm-3 FA+. Supporting electrolyte I I 1 0 . 4 0.0 0.0 1055 cation Ni3+ kD M. W. Espenscheid et al. Table 1. Ion-exchange distribution coefficients kD and selectivity coefficients Kg:? K M ~ + 7.9 x 105 2.5 x lo6 3.7 (k0.5) x lo4 2.1 x lo7 5.7 (+ 1.1) x lo6 1.1 x 106 2.6 x 104 1.5 (k0.2) x 104 7.3 (k0.9) x 104 740d MV2+ a FA+ Ru(NH3):+ Ru(bPY):+ R~(NH,);+ c a Methyl viologen. Ferrocenylmethyltri- methylammonium. Calculated from Ee. shift. Assuming zNa+ = 0.90.isotherms k, are shown in column 2 of table 1. The immenseness of these slopes clearly shows that Nafion greatly prefers these electroactive counterions over Na+. Furthermore these k are essentially Nafion-water preconcentration factors and their immense sizes suggest that Nafion would indeed be a useful preconcentration material for these or similar ions. While the partition isotherm allows for a convenient visualization of the extent of the ion-exchange reaction if ion-exchange data obtained here are to be compared with existing data for this or other ion-exchange systems and if the data outside of the linear isotherm region are to be treated ion-exchange selectivity coefficients [eqn (3)] should be calculated. Ion-exchange selectivity coefficients for the variouselectroactive counterions are shown in column 3 of table 1.These coefficients were calculated from points on both the linear and non-linear portions of the isotherms. While as indicated by the standard deviations there is some scatter in these coefficients no trends in PNi with xos were observed. The magnitudes of these ion-exchange selectivity coefficients are truly remarkable. Yeager and Steck studied exchange reactions of alkali-metal and alkaline-earth ions in Nafion.loV l5 (While these authors used H+ as the exchanged ion and we used Na+ they have shown that K p z 1 ;lo this allows for rough comparisons of their coefficients with ours.) The largest selectivity coefficient for a monovalent ion observed by Yeager and Steck was SHS which has a value of 9.1 ;lo the largest for a divalent ion was K p which had a value of ca.30.15 Selectivity coefficients for the ions studied here are from 3 to 6 orders of magnitude larger than these values; clearly Nafion shows tremendous preference for the ions studied here over the simple inorganic ions studied by Yeager and Steck. Conventional cation-exchange materials (i.e. sulphonated styrenedivinylbenzene resins) show ion-exchange selectivity coefficients for the alkali-metal and alkaline-earth ions of the same order of magnitude as those for Nafion (i.e. l-1O).l6 However since most of the ions studied here are hydrophobic organic cations it is of interest to compare the ion-exchange data obtained here with selectivity coefficients for organic ions on conventional ion-exchange resins.Gregor and Bregman studied ion-exchange reactions of a variety of alkyl- and phenylalkyl-ammonium ions on resins containing ca. 1-35% di~iny1benzene.l~ While selectivity coefficients as large as ca. 15 were observed,17 none of the resins showed the remarkably large ion-exchange selectivity coefficients exhibited by Nafion. Because most of the ions in table 1 are hydrophobic these data corroborate the conclusion reached by Martin and Freiser (from ion-selective electrode experiments) that Nafion shows tremendous affinity for hydrophobic cations6 This conclusion is also corroborated by our luminescence studies of Nafion.l89 l9 However because the ions in table 1 have different structures and charges it is impossible to assess quantitatively the 1056 Polymer Modified Electrode Sensors Table 2.Ion-exchange selectivity coefficients on ODs-Nafion columna pyridinium ion 9.2 x lo2 2.3 x 103 6.8 x lo3 2.2 x 104 ethylp yridinium prop ylpyridinium butylpyridinium pentylpyridinium a Mobile phase = 0.95 mol dmP3 aqueous TMABr flow rate = 1 cm3 min-l low3 cm3 sam- ples of mol dm-3 aqueous pyridinium solu- tion injected. contribution of the hydrophobic interaction to the magnitude of the selectivity coefficient. The contribution of the hydrophobic effect could be assessed by evaluating selectivity coefficients for an homologous series of organic counterions.20 Because the electrochemical isotherm approach is tedious and time consuming we decided to study selectivity for such an homologous series using a high-performance liquid chromatographic method (HPLC).21 Commercially available octyldecyl-derived silica (ODS) particles (5 pm average diameter) were coated with Nafion by exposing the particles to a solution of Nafion and then evaporating the solvent.21 After sizing these Nafion-coated particles were packed into stainless-steel HPLC columns using conventional slurry packing techniques.21 An homologous series of N-alkylsubstituted pyridinium cations was prepared by reacting pyridine with the appropriate alkylbromide.21 These ions were injected onto the Nafion-based columns and eluted with aqueous tetramethylammonium bromide. The ion-exchange selectivity coefficients for the alkylpyridinium ions were calculated from the measured retention volumes using standard HPLC procedures.21 Ion-exchange selectivity coefficients for the various alkylpyridinium ions are shown in table 2.Note again that the selectivity coefficients for these hydrophobic organic cations are orders of magnitude greater than selectivity coefficients obtained by Yeager and Steck for inorganic cations.1° Furthermore the selectivity coefficients shown in table 1 are measured us. tetramethylammonium (TMA+) whereas Yeager’slO are measured us. H+. Because inorganic ions will not elute the larger pyridinium ions from the Nafion values of KgY if they could be measured directly would be even larger than the values (m. TMA+) reported here. The role of hydrophobic interactions in the binding of counterions to Nafion is more dramatically demonstrated and placed on more quantitative terms through a plot of log KgLA us.number of carbon atoms in the pyridinium ion. This ‘free-energy plot’20 (fig. 4) is linear (slope = 0.380 correlation coefficient = 0.996) demonstrating that each increment of molecular bulk (i.e. hydrophobicity) adds an inwement of thermodynamic driving force to the ion-exchange reaction. From the slope of this line it can be calculated that each additional carbon contributes an additional -0.518 kcal mol-l to the standard free-energy change for the ion-exchange reaction of an alkylpyridinium ion. Fig. 4 shows quantitatively that the hydrophobic effect is responsible for the enormity of ion-exchange selectivity coefficients for organic cations in Nafion.As noted above conventional ion-exchange materials do not show this marked preference by hydrophobic counterions.17 We believe that the difference between the exchange characteristics of Nafion and conventional resins can be explained by considering the structural differences between these polymers. First conventional ion-exchange resins are covalently cross- linked while Nafion is not. The Gibbs-Donnan equation (the fundamental thermo- 1057 M . W. Espenscheid et al. 51 1 6 11 8 7 9 no. of C atoms 10 Fig. 4. Plots of log K;GA vs. number of carbon atoms in the pyridinium cation for (a) Nafion (1 100 E.W.) and (b) Dow PFS/(1065 E.W.) (based on HPLC columns). dynamic equatiotl for ion-exchange reactions)22 accounts for the effect of cross-linking through inclusion of a PV term.22 This term discriminates against ions of large size (large V).179 22 Since Nafion is not cross-linked there should be no PV-based discrimina- tion against large ions and therefore large ions may be partitioned into Nafion without thermodynamic penalty.The second important structural difference between Nafion and conventional ion- exchange resins is that in Nafion only ca. 1 in every 8 monomer units is sulphonated while conventional resins are close to 100% sulphonated. We believe that the large segments of uncharged chain material allow for a greater extent of interaction with the non-polar parts of the exchanging ion. Sensor Development Because a Nafion film can preconcentrate a hydrophobic organic cation at a substrate electrode surface Nafion-coated electrodes show promise for development as sensitive and selective sensors for electroactive organic cations.Indeed we propose that Nafion or other ionomer-coated electrodes could be used in a trace analytical technique which we have called ion-exchange voltammetry. Ion-exchange voltammetry (i.e.v.) is qualitatively similar to anodic stripping voltammetry (a.s.v.) ; a comparison of these two techniques is shown in fig. 5. In a.s.v. a metal ion is preconcentrated at a small-volume Hg electrode by reduction of the ion to the amalgam. After this cathodic preconcentration the amalgamated metal is oxidatively stripped from the Hg electrode and the resulting anodic current is recorded and related to the concentration of the metal ion in the solution.1.e.v. also uses a preconcentration step to achieve high sensitivity; in this case however the inherent thermodynamic driving force of the ion-exchange reaction is used to preconcentrate an organic cation into a small-volume ionomer film on the electrode surface. The film-entrapped ion is then either oxidized or reduced and the resulting current is recorded and related to the concentration of ion in the aqueous solution (fig. 5). We have recently shown that ion-exchange voltammetric determination of the 1058 Polymer Modijied Electrode Sensors (a) a.s.v. (1 ) preconcentrate electrochemically M2+ + Hg + 2e- + Hg(M) (2) reoxidize-record current Hg(M) + Hg + Mz+ + 2e- Fig. 5.Schematic representations of anodic-stripping voltammetry and ion-exchange voltammetry. DA DHB NE 5-HT og2+ + e- - o’g. Fig. 6. Structures of neurotransmitters dopamine (DA) norepinephrine (NE) and 5-hydroxytriptamine (5-HT) and of dihydroxybenzylamine (DHB). dication methyl viologen at a Nafion-film-coated electrode yields an improvement of three orders of magnitude in detection limit relative to an uncoated electrode.* In a collaborative effort with Prof. Adams of the University of Kansas we are currently assessing the feasibility of using ion-exchange voltammetry at Nafion-coated microelectrodes for in vivo determinations of dopamine and other neurotransmitters (see fig. 6 for structures) in rat brain.23 Because these transmitters are amines they are protonated at physiological pH values and thus can be ion exchanged into Nafion.On the other hand the biogenic amine metabolites and the important brain substance ascorbic acid are anions at physiological pH values and thus should be rejected by the Nafion film. A recent report from Professor Adams’ laboratory shows that Nafion-coated carbon microelectrodes do indeed show very little anion interferen~e.~~ This differen- I Y ( b ) i.e.v. ( 1 ) preconcentrate via ion exchange Og” + 2(P-S03H) + (P-S03)20g+2HS (2) reduce film-entrapped ion- record current OH 1059 M. W. Espenscheid et al. Table 3. Ion-exchange selectivity coefficients for protonated forms of the neurotransmitters* no. ofOH KNT+ TMA+ no. of molar mass/ OH DOUPS groups molar mass/g ion 170 140 154 56.7 70.0 77.0 K ; s 2.5 x lo2 1.5 x 7.5 x 102 4.5 x 103 8.3 x lo2 5.0 x 1.6 x 104b norepinephrine 3,4-dihydroxybenzylamine dopamine serotonin 2 3 2 1 177 2.7 x 103 177 a 3,4-Dihydroxybenzylamine included for comparison.Calculated from KgZ and conversion factor obtained by eluting 3,4-dihydroxybenzylamine with both TMABr and NaBr. tiation of the neurotransmitters (NT) from metabolites and ascorbate has been sorely needed for in viuo electro~hemistry.~~9 24 In addition to the selectivity advantage since NT are organic cations they should be preferentially partitioned into a Nafion film at an electrode surface; thus the coated electrode should show better detection limits than an analogous naked electrode.Since hydroxy substitution renders a molecule more hydrophilic this preferential partitioning should be partially mollified by the hydroxy groups present on these ions and we anticipated seeing a mass (hydrophobicity increases ion-exchange selectivity coefficient) us. hydroxy substitution (hydrophilicity decreases selectivity coefficient) tradeoff. This hydrophobic-hydrophilic tradeoff is evident in the aqueous-Nafion ion-exchange data for NT (table 3). The largest ion-exchange selectivity coefficient is observed for 5-HT+ (see fig. 6) which has the highest molecular weight and the fewest hydroxy groups; the smallest coefficient is observed for NE+ which while not having the lowest molecular weight has the largest number of hydroxy groups.In fact a plot of KzE us. the empirical parameter obtained by dividing the molar mass by the number of hydroxy groups (table 3) is quite linear (slope = 114 correlation coefficient = 0.995). While this empiri- cal parameter seems to be devoid of thermodynamic significance it is apparently useful for predicting ion-exchange selectivity coefficients at least within the limited class of compounds studied here. The adherence of the ion-exchange data for NT+ to the predicted trends reinforces our earlier conclusions regarding the importance of hydro- phobic interactions to counterion binding in N a f i ~ n . ~ - ~ Since all of the selectivity coefficients in table 3 are quite large a Nafion film should preconcentrate these ions and ion-exchange voltammetry at a Nafion-coated electrode should result in improved detection limits.This is demonstrated by the data shown in fig. 7. A Nafion-coated electrode was equilibrated for 3 min (with stirring) in a solution lob7 mol dm-3 in DA+ and 0.01 mol dm-3 in Na+ (phosphate buffer pH 7.0). After the equilibration period a cyclic voltammogram was obtained and the process was repeated for successively higher concentrations of DA+. Fig. 7 [curve (a)] shows a plot of anodic peak current (background corrected) us. concentration of DA+ at a Nafion-modified electrode. Note that because of the preconcentration effect the detection limit (d.1.) for DA+ is ca. mol dm-3. This d.1. is two orders of magnitude lower than the corresponding d.1. at a naked electrode.While curve (a) in fig. 7 clearly shows that significantly improved d.1. can be obtained at Nafion-coated electrodes such electrodes might suffer from interference not observed at naked electrodes. Organic cations present in the analyte solution which are not electrochemically active could (if their concentrations or selectivity coefficients were large enough) partition into the film and occupy a fraction of the Nafion -SO; sites. Since this would decrease the number of sulphonate sites available for the analyte cation the sensitivity for the analyte cation could be decreased. This possibility was explored by 1060 9 - 8 - 7 - c g a .i 5 - 6 - -. 4 - 3 - 2 - 0 1 2 3 4 5 6 7 8 DA' concentration/ 1 0-6 mol dm-3 Fig.7. Peak current for oxidation of DA+ (background corrected scan rate = 100 mV s-l) at a Nafion-modified electrode (8.9 x mol SO; sites) 0s. [DA+]. Supporting electrolyte was 0.004 mol dm-3 Na,HP0,/0.002 mol dm-3 NaH,PO, pH = 7.0. Concentration of phenylethyl- amine (a) 0 (b) 2 x and (c) 2 x obtaining DA+ calibration curves in the presence of phenylethylamine which is protonated at pH 7.0 but not electrochemically active. As indicated in curves (b) and (c) of fig. 7 reduced sensitivities to DA+ are seen when phenylethylamine is present. From a practical point of view this means that analyses with Nafion-modified electrodes will require some knowledge of the composition of the analyte solution. An identical situation is encountered in analyses with ion-selective electrode~.~~ Nafion-coated microelectrodes (100-300 pm) were checked for such inter- ferences by calibrating in brain homogenates at pH 7.4.23 The calibration curves obtained showed no significant decrease in slope when compared to buffer-only calibration curves.23 Dynamics of Ionomer Modified Electrode Response and New Fast Ion-transporting Membranes The maximal preconcentration advantage for an ionomer film-modified electrode will be obtained when ionomer film-analyte solution equilibrium is achieved.The time required for establishment of equilibrium depends on the size charge structure and concentration of the analyte counterion and on the ionic strength of the analyte solution. We have shown that relatively small counterions (e.g.DA+) at the lO-'mol dm-3 concentration level reach equilibrium with a 1 pm thick Nafion film in less than 3 mhZ3 In contrast larger counterions [e.g. Ru(bpy):+] at the mol dm-3 concentration level may require as long as one day to reach equilibrium,8 which from an analytical viewpoint is an unacceptably long time. One can of course decide to wait only a small fraction of the total equilibration time before assaying the quantity of analyte partitioned into the film but the maximal preconcentration advantage will not be realized.8 Polymer Modijied Electrode Sensors mol dm-3. 1061 M . W. Espenscheid et al. ionomer impregnated pores t ineit PTFE porous membrane Fig. 8. The ionically conductive composite polymer membrane concept. In general the rate of film-solution equilibration is limited by the rate of ionic diffusion in the film.8 Thus membranes with higher ion-transport rates would produce modified electrode sensors with faster response times; we are currently attempting to develop such fast ion-transporting membranes.We have recently described a new series of ionically conductive composite polymer membranes.26 The concept behind these membranes is illustrated schematically in fig. 8. A porous inert host material which is initially neither wetted by water nor ionically conductive is impregnated with an ionomer which renders the membrane water swollen and counterion conductive. We reasoned that because the counterion moves through a water-filled pore ionic diffusion in these membranes should be fast.Prototype membranes of this type were prepared by impregnating Gore-tex (a commercially available porous polytetrafl~oroethylene)~~ with Nafion. A Gore-tex membrane (0.2 pm mean pore diameter) was stretched over the face of a wax-impregnated graphite (WIG) electrode and held into place with a collar of heat-shrinkable Teflon tubing (fig. 9). Nafion was loaded into the Gore-tex by immersion of the membrane into an ethanolic Nafion solution (0.7-2.5 w/v %).26 The electrode was then removed from the impregnating solution and the solvent was allowed to evaporate (room tempera- ture 5 h). Fig. 10 curve (a) shows a ‘cyclic voltammogram’ for a native Gore-tex membrane- modified WIG electrode in aqueous supporting electrolyte. Because the Gore-tex membrane is not wetted by water the electrode surface is completely insulated from the solution phase and no background currents are observed.However Nafion-impregnated Gore-tex (NIGT) is wetted and ionically conductive and background voltammograms at NIGT-modified WIG [fig. 10 curve (b)] and naked WIG electrodes are essentially identical. The ionic conductivity of NIGT is even more dramatically illustrated by the voltammograms shown in fig. 11 ; a NIGT-modified WIG electrode was immersed into a solution which was 3 mmol dm- in Ru(NH,):+ and the potential was scanned over the Rurrl/I1 redox wave. An Ru(NH,)i+ ‘loading voltammogram’ very similar to that observed at a Nafion-film modified electrode is obtained.2 Ru(NH,)g+ is entering the NIGT membrane diffusing through the membrane to the electrode surface and undergoing reduction and reoxidation at the electrode surface.When Ru(NH,)i+ loaded NIGT-modified WIG electrodes are transferred to supporting electrolyte solutions a fraction of the complex is retained by the membrane (fig. 12); this is again analogous to the results obtained with Nafion film-coated electrodes.2 As shown in fig. 12 the differences in potential between the anodic and cathodic peaks for 1062 Polymer Modijied Electrode Sensors wire 6 m m glass tube heat shrinkable Teflon sleeve 9 mm glass tube mercury - wax impregnated graphite h e a t shrinkable' Tef Ion sleeve 10 p A I t I I Nafion impregnated Gortex membrane Fig. 9. The Nafion-impregnated Gore-tex film-modified wax-impregnated graphite electrode.( b ) I I I T I I I U.b (a) U n u .4 U .L -0.2 -0.4 -0.6 -0.8 EIV vs. SCE Fig. 10. Background voltammograms (200 mV s-l) in 0.1 mol dm-3 NaClO,. (a) Native Gore-tex membrane-modified WIG electrode (b) Nafion-impregnated (14.6 w/w % ) Gore-tex membrane- modified WIG electrode. the Ru(NH3):+I2+ waves decrease as the quantity of Nafion incorporated increases. This suggests as might be expected that the ionic conductivity of NIGT increases with the quantity of Nafion incorporated. As noted above we initiated this research in the hope of developing fast ion- transporting membranes. To obtain a quantitative measure of the facility of ion transport chronocoulometric experiments2 were used to determine diffusion coefficients for Ru(NH,)g+ and Ru(NH,)f+ in NIGT membranes.As shown in table 4 these diffusion coefficients are ca. three orders of magnitude larger than the analogous diffusion coefficients in Nafion. These data clearly show that ionically conductive composite membranes can produce very high ion-transport rates. A multitude of questions remain to be answered before the practical utility of these membranes can be assessed; these include how do pore size of the host and quantity of the ionomer incorporated affect the rate of ion transport how do pore-size distribution and chemical properties of the host affect the rate of ion transport how chemically and thermally stable are these composites and can permselective membranes be prepared using this approach? We are currently addressing these questions.1063 n M. W. Espenscheid et al. SCE Fig. 1 1. Loading voltammograms for incorporation of Ru(NH,)i+ into a NIGT membrane at a WIG electrode surface. Solution was 3 mmol dm- in Ru(NH,);+ and 0. 1 mol dm- in NaClO (every fifth scan recorded). I n Situ Film Regeneration Ionomer-modified electrodes are examples of chemical sensors in which a membrane is placed between the analyte solution and the detection system; the purpose of the membrane in such a sensor is to provide selective chemistry such that the detection system ‘sees’ or has opportunity to respond to only the desired analyte species. We have discussed chemical selectivity in ionomeric systems in this paper. In fact ionomers provide a very rudimentary form of chemical selectivity (see Conclusions section) yet ionomer-based electrochemical sensors can be quite useful for certain types of analyses (vide supra).In addition to ‘building’ chemical selectivity into the membrane one must also ‘ build in ’ facile mass-transport ; if not the sensor’s response time will be prohibitively long. We have presented a general scheme for providing facile mass-transport in polymer membranes. There is another factor which relates both to selectivity and to mass transport which must be considered when designing a sensing membrane. Consider a well stirred solution containing the analyte species which is in contact with a sensor membrane. Furthermore assume that the membrane has been appropriately designed so that (1) there is a large FAR 1 36 1064 Polymer Mod$ed Electrode Sensors 0.2 0.0 A/ Fig.12. Ru(NH3),3+-loaded NIGT-WIG electrodes. Voltammetry in 0.1 mol dm-3 NaClO solution. Nafion content of membranes (a) 7.33 (b) 10.9 and (c) 14.7%. Table 4. Comparison of diffusion coefficients in Nafion (1 100 EW) and Nailon-impregnated Gore-tex (0.2 pm mean pore diameter Gore-tex; 9 w/w % Nafion) diffusion coefficients / c m 2 s-1 NIGT Nafion diffusing ion 2.3 x 10-9 3.4 x 10-9 1.9 x 10-6 4.1 x 10-6 Ru(NH,);+ Ru(NH&+ thermodynamic driving force for the partitioning of the analyte species into the membrane and (2) mass transport within the membrane is facile. Because the solution is well stirred solution mass-transport is also facile and the combination of these three factors will insure that a large quantity of the analyte will be partitioned very rapidly into the membrane phase.This is clearly the ideal situation for detection of the analyte but it presents a problem after the detection process is complete. Unless this (electrochemical) detection step drastically alters the thermodynamics of the partition reaction the analyte will be preferentially retained by the film after analysis rendering the film essentially useless for further analyses. The bottom line is we have designed a membrane which from a thermodynamic point of view loves the analyte and from a dynamic point of view can rapidly satisfy that love by quickly incorporating this analyte. The question then becomes since one has 1065 X i ...TO s polymer film M . W. Espenscheid et al. I(cH~;HHcH~- electrochemical ox idat io n 1 2 support1 ng electrolyte s u P P ~ rti ng electrolyte Fig. 13. Proposed model for electrorelease of counterions from hypothetical electroactive ion exchange polymer film-modified electrode. stacked the deck in the opposite direction how does one then remove this analyte when the analysis is complete? One possible route for purging the membrane is to remove the sensor from the analyte solution and treat it chemically (e.g. with a very high ionic strength solution) to remove the analyte. This chemical treatment could however introduce additional problems and would in any case be unacceptable if continuous monitoring (e.g.in vivo or process-stream analysis) was required. We have recently described a new series of electroactive ion-exchange polymers or electroactive ionomers which offer an elegant alternative to the film purging problem.28 Electroactive ionomers are polymers which contain both electroactive and ion-exchange functionalities. We prepared a prototype series based on styrene vinylferrocene (the electroactive functionality) and styrenesulphonate (the ion-exchange monomer). These terpolymers were prepared from the commercially available monomers using conventional free-radical initiator techniques2* When coated onto an electrode surface and exposed to an electrolyte solution the ferrocene group in these terpolymers can be reversibly oxidized and rereduced ; when the ferrocene group is in the reduced form the styrene sulphonate group needs a counterion (cation) and in analogy to ion-exchange voltammetry at a Nafion-modified electrode this cation could be an analyte species (e.g.DA+). When oxidized however the ferricinium group created might function as the counterion for the styrenesulphonate (see fig. 13) resulting in the expulsion of the analyte counterion from the film. Thus we reasoned that an electroactive ionomer film should be able to electrorelease an incorporated analyte counterion. To test this electrorelease hypothesis terpolymers having the following approximate mole percentages of styrene (STY) styrenesulphonate (SS-) and vinylferrocene (VFc) were prepared :2a high-SO; = (STY),,-(VFc),,-(SS-) ; mid-SO; = (STY),,-(VFc),,- (SS-) ; low-SO; = (STY),,-(VFc),,-(SS-),.The electrochemical characteristics of these polymers were studied using polymer film-modified glassy carbon electrodes in conjunc- tion with aqueous supporting electrolyte solutions. Cyclic voltammetric experiments at these modified electrodes produced the rather unusual voltammograms shown in fig. 14. Note that polymers containing high mole percentages of SS- show two distinct Fc+/O redox waves one with an E" similar to E" for Fc+/O in water and one with a significantly more positive E". These data suggest that there are two chemically distinct classes of Fc in these polymer films. We believe that the low E" form corresponds to Fc in an aqueous-like microenvironment (note that the quantity of this form scales with the quantity of SS- and therefore the quantity of water in the film) and that the high E" form results from Fc in a drier chemical microenvironment.We are further investigating the genesis of this interesting electrochemical behaviour. 36-2 1066 Polymer Modijied Electrode Sensors E/V vs. SCE (bl E/V us. SCE Fig. 14. Steady-state cyclic voltammograms for (STYHVFcHSS-) terpolymer film-modified electrodes. (a) (STY),,-(VFc),,-(SS-), high -SO; ; (b) (STY),,-(VFc),,-(SS-), mid -SO; ; (c) (STY),,-(VFc),,-(SS-), low -SO;. Supporting electrolyte = 0.1 mol dm-3 Na,SO,. Scan rate = 200 mV s-l. The modified electrodes in fig. 14 were exposed to 1 mmol dm- methyl viologen solution rinsed and placed back in a solution containing only supporting electrolyte.By analogy with Nafion these polymers incorporate methyl viologen via ion exchange as evidenced by the new voltammetric wave at ca. -0.7 V (fig. 15). Note that as would be expected the quantity of methyl viologen incorporated increases with the mole percentages of SS- in the polymer. Fig. 16 shows that these polymers can electrorelease counterions. Voltammogram (a) in fig. 16 was obtained after exposing a terpolymer modified electrode to a solution containing Ru(NH,)i+. A prominent voltammetric wave for electrostatically incorporated RulI1 is seen. The electrode was then held at +0.8 V for 1 min. According to the scheme shown in fig. 13 this should cause Ru(NH,)i+ to be expelled from the film; the diminution in the RulI1/I1 voltammetric wave [curve (b) fig.161 shows that Ru (NH,)i+ has indeed been expelled. When the potential is held for another minute at +0.8 V the RuIII’II wave is barely preceptable [curve (c) fig. 161 indicating that nearly all of the Ru(NH,)i+ has been discharged. This process is perfectly reversible; Ru(NH,)i+ can be reincorporated 1067 I 1 - I / E/Vvs. SCE M . W. Espenscheid et al. I 1 I I T \ 1 f"" o.L o ' 2 / '/-=.a' T EIVus. SC CE I Fig. 15. Steady state voltammograms for terpolymer film-modified electrodes after exposure to 1 .O mmol dm-3 methyl viologen solution. Terpolymer compositions and scan conditions are given in the caption to fig. 14. via re-exposure to the Ru(NH,)i+ solution and discharged again by holding the potential positive of the Fc+/O wave.It is in principle possible that the loss of Ru(NH,)i+ from the film (fig. 16) resulted from ion exchange of Ru(NH,)i+ with Na+ from the supporting electrolyte. To test this possibility the experiment described above was repeated but the potential of the electrode was held at 0 V (where Fc remains in the reduced form) rather than at +0.8 V. While some loss of Ru(NH,)Z+ is observed (fig. 17) note that even after 9 min at 0 V most of the complex has remained in the film. Clearly leaching can not account for the dramatic loss of Ru(NH,)i+ shown in fig. 16. The analytical use of these electroreleasing polymers is demonstrated in fig. 18. Curve (a) in fig. 18 is an ion-exchange voltammetry calibration curve (vide supra) for methyl viologen at a terpolymer film-modified electrode.Note that the detection limit at this electrode is < lo-' mol drn-, over two orders of magnitude better than at a naked electrode. These data show that the ion-exchange voltammetry preconcentration advan- tage is observed at these terpolymer-modified electrodes. After exposure to the most concentrated calibration solution a large quantity of methyl viologen has been partitioned into the film and a prominent redox wave is seen [curve (a) fig. 191. The potential of 1068 Polymer Modijied Electrode Sensors 5bA .I Fig. 16. Cyclic voltammograms for a (STY),,-(VFc),,-(SS-), film-modified electrode immersed in 0.1 mol dm-3 Na,SO, 200 mV s-l. (a) After exposure to 1 .O mmol dm-3 Ru(NH)i+; (b) as in (a) but after electrode had been potentiostatted for 1 min at +0.8 V and then potentiostatted for 1 min at 0.0 V; (c) after repeating sequence described in (b) (see text).Fig. 17. Cyclic voltammograms for polymer film-modified electrode described in fig. 5 immersed in 0.1 mol dm-3 Na,SO, 200 mV 0. (a) After reloading in Ru(NH3);+ solution; (b) (c) and (d) after potentiostatting at 0 V for 3 6 and 9 min respectively. the electrode was then held at +0.8 V for 30 min which [as indicated by the loss of the methyl viologen redox wave curve (b) in fig. 191 resulted in the expulsion of the methyl viologen from the film. The calibration process can then be repeated [curve (b) fig. 181 producing a calibration curve essentially identical to the first. Conclusions We have shown that ionomer-modified electrodes are potentially useful electroanalytical sensors.Because of their unusual ion-exchange selectivity perfluorosulphonate ionomer- modified electrodes can preconcentrate electroactive organic counterions resulting in dramatically improved detection limits for these ions relative to the naked electrode. We M. W. Espenscheid et al. [ MVz']/ 1 0-5 mol dm-3 Fig. 18. Ion-exchange voltammetry calibration curves for methyl Gologen at a (STYXSS-WFc) film-modified electrode. (a) First calibration curve ; (b) after electrorelease of methyl viologen introduced during first calibration. Supporting electrolyte = 0.1 mol dm-3 Na,SO,. Scan rate = 200 mV 0. EIV us. Ag I AgCl Fig. 19. Voltammograms for electrode described in fig.18. (a) Before electrorelease of methyl viologen ; (b) after electrorelease of methyl viologen. call this preconcentration-based electroanalytical scheme ion-exchange voltammetry. Ion-exchange voltammetry has been shown to be quite useful for in vivo analyses of neurotransmitters. Further studies of this technique particularly of the effect of potential-scan waveform on detection limits are in progress. We have also briefly examined the dynamics of ionomer-modified electrode response. At very low concentrations of exchanging (analyte) ion the rate of film-solution equilibration can be prohibitively slow. Since film mass-transport is in general the rate- determining step there is clearly a need for ionomer membranes with higher ionic diffusivities.We have shown that the ionically conductive composite polymer membranes developed in our laboratory can have very high ionic diffusivities. Thus these composite 1069 Polymer Modified Electrode Sensors membranes show promise for use in electrochemical sensors and in a variety of other electrochemical processes and devices. Further fundamental research will be required before the practical utility of these membranes can be assessed. Finally we have discussed film regeneration in electroanalysis with ionomer-modified electrodes. We have shown that a new class of electroactive ionomers can accomplish film regeneration by electroreleasing the incorporated analyte counterion. Films of these electroactive ionomers also preconcentrate the analyte counterion ; thus ion-exchange voltammetry at electroactive ionomer-modified electrodes produces detection limits superior to those obtained at a naked electrode.In closing it is important to point out that while ionomer-modified electrodes have greater selectivity than a naked electrode ionomers provide only a very rudimentary form of selectivity (charge type and mass or hydrophobicity selectivity). The challenge to the electrochemist (or better still to the electrochemist in collaboration with a synthetic- polymer chemist a biochemist etc.) is to ' build' better and more selective chemistry into membranes so that sensors with higher chemical specificity can be obtained. One obvious approach is to incorporate chemically selective biochemical species into membranes; indeed potentiometric sensors based on this concept are now commercially a ~ a i l a b l e .~ ~ There are also other approaches to building chemically selective membranes; we hope to be able to report on some of our attempts at developing such membranes soon. This work was supported by the Office of Naval Research and the Robert A. Welch Foundation. 1070 References 1 Ions in Polymers Advances in Chemistry Series 187 ed. A. Eisenberg (American Chemical Society Washington 1980). 2 C. R. Martin and K. A. Dollard J. Electroanal. Chem. 1983 159 127. 3 F. G. Will J . Electrochem. Soc. 1979 126 36. 4 R. S. Yeo and D-T. Chin J. Electrochem. Soc. 1980 127 549. 5 W. Grot Chem.-Ing.-Tech. 1978 50 299. 6 C. R. Martin and H. Freiser Anal. Chem. 1981 53 902.7 C. R. Martin I. Rubinstein and A. J. Bard J. Am. Chem. Soc. 1982 104 4817. 8 M. N. Szentirmay and C. R. Martin Anal. Chem. 1984 56 1898. 9 R. B. Moore 111 J. E. Wilkerson and C. R. Martin Anal. Chem. 1984 56 2572. 10 H. L. Yeager and A. Steck Anal. Chem. 1979 51 862. 11 0. Samuelson Ion Exchange in Analytical Chemistry (John Wiley New York 1953) p. 37. 12 E. R. Tompkins and S. W. Mayer J. Am. Chem. Soc. 1947 69 2859. 13 H. S. White J. Leddy and A. J. Bard J. Am. Chem. Soc. 1982 104 481 1. 14 J. R. Schneider and R. W. Murray Anal. Chem. 1982 54 1508. 15 A. Steck and H. L. Yeager Anal. Chem. 1980 52 121 5. 16 J. A. Marinsky J. Phys. Chem. 1967 71 1572. 17 H. P. Gregor and J. I. Bregman J. Colloid Sci. 1951 6 323. 18 N. E. Prieto and C. R. Martin J. Electrochem. Soc. 1984 131 751. 19 M. N. Szentirmay N. E. Prieto and C. R. Martin J. Phys. Chem. 1985,89 3017. 20 J. A. Marinsky and Y. Marcus Ion Exchange and Solvent Extraction (Marcel Dekker New York 1974) vol. 6 p. 5. 21 Details of this method are described in R. B. Moore 111 J. E. Wilkerson and C. R. Martin Anal. Chem. 1984 56 2572. 22 J. Fietelson in Zon Exchange ed. J. A. Marinsky (Marcel Dekker New York 1969) vol. 2 chap. 4. 23 G. Nagy G. A. Gerhardt A. F. Oke M. E. Rice R. N. Adams R. B. Moore 111 M. N. Szentirmay and C. R. Martin J. Electroanal. Chem. 1985 188 85. 24 G. A. Gerhardt A. F. Oke G. Nagy B. Mughaddam and R. N. Adams Brain Res. 1983 290 390. 25 C. R. Martin Trends Anal. Chem. 1982 1 175. 26 R. M. Penner and C. R. Martin J. Electrochem. Soc. 1985 132 514. 27 Gore-tex is a registered trademark of W. L. Gore and Associates. 28 M. W. Espenscheid and C. R. Martin J. Electroanal. Chem. 1985 188 73. 29 R. P. Buck J. Chem. Soc. Faraday Trans. I 1986,82 1169. Paper 5/ 1956; Received 6th November 1985