Analytical Communications, January 1996, Vol33 (23-25) 23 Direct Non-enzymic Amperometric Glucose Sensor Based on a Novel Glucose Selective Membrane ~ Yazid Benmakroha", Ian Christie and Pankaj Vadgama Department of Clinical Biochemistry, University of Manchester, Hope Hospital, Eccles Old Road, Salford, UK M6 8HD A selective amperometric glucose sensor is described, based on direct electrochemical oxidation without recourse to voltage or time resolution techniques. In the presence of ionic and also uncharged interferents, selectivity is provided by a novel membrane composite of sulfonated polyether-ether sulfonepolyether sulfone copolymer and PVC plasticized with isopropyl myristate.Ratios for equimolar glucose : interferent are 220 : 1 for ascorbate, urate and ethanol and 210 : 1 for acetaminophen.Linear response extends to at least 600 mmol 1-1, encompassing clinical use and agrofood applications. Many of the strategies explored for the electrochemical measurement of glucose have been enzyme based but the enzymic devices are inherently subject to constraints of enzyme kinetics: notably, there is a finite linear range. Most direct glucose sensors have been based on oxidation of the aldehyde group.Work on the fuel cell principle' and also voltammetry has been moderately successful. Marincic et aL2 found complex, potential dependent mechanisms of parallel or sequential reactions but inhibition was n0ted;3.~ in further work, pulsed voltammetric techniquess improved the selectivity. The use of membrane barriers has been discussed6 and examples of M , 500 cut off are in use; however, the effects of H+, hydrogencarbonate and amino acids are unlikely to be elimi- nated.Measurements from 2-20 mmol 1-l were achieved by the compensated net charge (cnc) and the cnc ratio technique^.^.^ A further technique9 was based on potential jumps from 1600 mV, oxidizing the Pt surface, to 100 mV to discharge the electrode capacitance and then to the working potential of 400 mV, current integration giving a measure of glucose concentration.Impedance experiments have also been undertaken to improve the performance of the potential jump method.9 Bockris et a1.1" have postulated a mechanism for alkaline oxidation of glucose whereby the carbohydrate repeatedly cleaves one carbon atom and is simultaneously oxidized to a terminal-COOH, giving for total completion C6H12O6 + 6H20 + 6C02 + 24H+ 24e- Over-all charge transfer numbers ( n ) of 2-20 have been noted for RDE sweeps of -0.2 to +0.2 V, confirming oxidation beyond gluconic acid where n = 2.Over time n decreases, indicating passivation and a change in electrode processes taking place. Recent work has highlighted the role of the electrode material in the oxidation of glucose; ruthenium dioxide-carbon paste electrodes have exhibited catalytic be- haviour.11 * Present address: Department of Biomedical Engineering and Medical Physics, North Staffordshire Hospital Centre, Thornburrows Drive, Hartshill, Stoke on Trent, UK, ST4 7QB.Membranes used to exclude interferents from amperometric sensors have commonly exploited the anionic nature of clinical interferents.Plasticized PVC excluded charged species by a differential partitioning mechanism, * whereas sulfonated poly- ether-ether sulfonated polyether-ether sulfone/polyether sul- fone (SPEES-PES) carries a charged sulfonate group. l3314 This work describes the use of SPEES-PES and plasticized PVC mixtures for the fabrication of membranes permselective to glucose.l 5 Materials and Methods All assays were conducted in phosphate-chloride buffer with sample solutions at pH 7.4 unless indicated. Glucose, urate, ascorbate and acetaminophen were reagent grade, and were made up in buffered stock solution and diluted in the electrochemical cell to the required concentration. Stock glucose solutions were left overnight prior to use.The 'ethanol' was 95% ethyl alcohol and 5% isopropyl alcohol by volume. Undiluted human blood and serum (collected in fluoride- oxalate tubes) was obtained from the Department of Clinical Biochemistry Laboratory, Hope Hospital. Cuprophan dialysis membranes were removed from a hemodialysis cartridge (Gambro, Lund, Sweden). Other mem- branes were solvent cast in a covered Petri dish.The PVC solution contained 0.06 g of PVC (M, 200 000, Merck, Poole, UK) and 0.15 ml isopropyl myristate (IPM) plasticizer dissolved in 5 ml of tetrahydrofuran. The SPEES-PES solution contained 0.4 g of SPEES-PES (a gift from ICI Chemicals and Polymers, Runcorn, UK), dissolved in 7.5 ml of dimethylforma- mide and 2.5 ml of methoxyethanol. SPEES-PES-PVC mem- branes were cast from a mixture of 3 ml SPEES-PES solution and 3 ml of PVC solution.15 The electrochemical cell and potentiostat have been de- scribed earlier.16 In the case of stability studies in serum and blood, hydrogen peroxide measurements were made following periods of specified exposure of the sensor to undiluted human serum and blood.Results Glucose is oxidized electrochemically as pH is raised, particu- larly above 10 (Fig.1). Under these conditions on a Pt anode (+0.65 V versus Ag/AgCl) no plateau region of pH indepen- dence was observed. When the applied polarizing voltage was varied (Fig. 2) using pH 13.5 for optimal response, a sigmoidal potential dependence was seen, and though there was no definite plateau, +0.65 V versus Ag/AgCI was adopted as a standard.Responses to glucose in the range of interest in clinical monitoring (0-30 mmol 1-1) are shown in Fig. 3 (+0.65 V versus Ag/AgCl, pH 13.4). Glucose did not permeate PVC, but there are, however, responses when SPEES-PES is employed, which are amplified when the mixed membrane of SPEES-24 Analytical Communications, January 1996, Vol33 200 160 p 120 .u, PES-PVC is used, such that the response with the mixed membrane is not intermediate between that of either polymer alone but significantly enhanced ( X 3 at 20 mmol 1-1 of glucose) over SPEES-PES alone. Response with the mixed membrane is also linear over a greater glucose range than with SPEES-PES alone, up to at least 600 mmol 1- l , i.e., - 10% m/v glucose (y = 0.16~ - 1.2, I^ = 0.99).For all levels of glucose down to 5 mmol 1-1, current intensity was sufficient for assay purposes, but thinner membranes, or more precise tailoring of membrane composition, offer a means of signal enhancement at low concentrations. Partitioning of the buffer was maintained for prolonged periods with no significant signal drift during 8 h periods of use on repeated occasions.The selectivity of the membranes was investigated (Table 1) for commonly encountered interferents in clinical analyses. Ascorbate and urate signals are eliminated using the SPEES- . 40 / 7 8 9 10 1 1 12 13 14 PH Responses of a Cuprophan covered platinum electrode to 10 Fig. 1 mmol 1-1 glucose at varying pH (+0.65 V versus Ag/AgCl). 300 p 200 g . v) d 100 0 0 0.2 0.4 0.6 0.8 1 .o Polarising voltageN Fig.2 Responses of a Cuprophan covered platinum electrode to 10 mmol 1-1 glucose in pH 13.5 buffer at varying polarizing voltages (versus Ag/AgC1) h Al 0 5 10 15 20 25 30 [GI ucose]/mmol I-' Fig. 3 Responses to glucose at an electrode covered by PVC (a), SPEES- PES (A) and SPEES-PES-PVC (0) membranes, overlying a Cuprophan membrane bathed in pH 13.4 buffer (+0.65 V versus Ag/AgCl).PES-PVC membrane, as is the response to acetaminophen, a common problem even with other selective membrane materials including PVC12 and celluose acetate.16 Ethanol only shows any interference at higher concentrations. The biocompatibility of the membranes was studied by comparing responses to 2 mmol 1-1 H202 at membrane- protected non-enzyme electrodes used at +0.65 V versus Ag/ AgCl during exposure to serum and blood, with an underlying Cuprophan membrane included with the test membrane (inter- nal electrolyte pH 7.4).When a selective membrane was not used, over 60% signal attentuation occurred over a 200 min interval. Unplasticized PVC gave a final 85% remaining signal, SPEES-PES 90% and plasticized PVC 92%. The membrane with the least reduction in signal, however, was the mixed SPEES-PES-PVC membrane with a 95% retention of response (90% when blood was used).A general pattern of early rapid ( 10 min) decrease in signal was followed by an extended period of relative stability. Discussion Glucose signals at high pH are attributed to deprotonation of the slightly ionic 0-H bond on the C-1 of glucose, causing ring rupture.17 It appears likely that the involvement of an enolate ion,18 formed under alkaline conditions, is implicated in the increase in glucose oxidation at high pH.Glucose exclusion by PVC is possibly related to the polar OH groups of glucose. SPEES-PES permeability may be due to its polar SO3- groups interacting with the OH groups of glucose or enabling a degree of hydration14 such that partitioning of solute into the membrane is favoured.The likely explanation for the high responses with the composite membrane is a fine balance between the non-polar character of PVC-IPM constituents and polar SPEES-PES.15 The extended linearity with this sensor is attributed to the avoidance of enzyme saturation and the restriction of glucose flux through the membrane.The elimina- tion of both aromatic (acetaminophen) and charged (urate, ascorbate) interference is remarkable, and it appears fine tuning of permeability towards glucose is accompanied by exclusion of species of both charged and hydrophobic character. Also, ethanol interference is low, in part due to the intrinsically low electrochemical activity of the molecule, but also reduced by a factor of approximately 10 compared to a plasticized PVC covered electrode (Table 1).The discrimination between polar, uncharged species merits future study. Electrochemical oxidation of certain interferents produces a passivating film on the electrode surface: small molecule passivation of a polarized electrode surface by diffusible solutes has been reported by us.19 The selectivity shown by the highly perm-selective membranes results in the exclusion of such passivating species, not achieved by Cuprophan.At the outer Table 1 Responses (nA) to pH 7.4 solutions with electrode covered by Cuprophan dialysis membrane and overlaid polymer membranes (+0.65 V versus Ag/AgCl). Concen- tratiord SPEES- SPEES- mmol 1-1 PVC PES-PVC PES pH 7.4 electrolyte Ascorbate 10 <0.1 <0.1 0.30 Urate 10 <0.1 <0.1 0.25 Acetaminophen 5 335 <0.1 <0.1 pH 13.4 electrolyte - 2.0 Glucose 10 Ascorbate 1 <0.1 Ethanol 20 - - - Acetaminophen 1 - <0.1 - - < 0.2 -Analytical Communications, January 1996, Vol33 25 sensor surface the plasticizer component of the membrane could possibly account for non-adhesion of protein and cells during exposure to blood, by presenting a quasi-fluid surface, similar to that of a plasticized PVC membrane.The slight superiority in non-fouling shown for the mixed polymer membrane requires further elucidation. The surface heterogeneous polarity with hydrophobic (PVC) and charged (SPEES) domains could be a further factor in reducing surface fouling. Potential amphiphilic protein adherents may therefore encounter an interface that diminishes surface denaturation, reducing the tendency for further adhesion of proteins.Conclusions A sensor for glucose has been developed based on direct electrochemical oxidation of glucose at a platinum anode. The selectivity required has been imparted by the use of a novel membrane composite, in marked contrast to conventional enzyme glucose sensors in which the enzyme reaction product is selectively permeable through an inner membrane.The perm- selectivity properties shown are not predictable from those of either constituent membrane, and further study is being undertaken to establish the parameters involved, particularly the balance between polar and non-polar constituents.The work has opened up a route to more selective amperometric detection and sophisticated membrane materials that can be tailored to offer new avenues of sensor design. The authors thank SERC for financial support for this work. The SPEES-PES polymer was a gift from ICI Chemicals and Polymers (Runcorn, UK). References 1 Yao, S. J., ‘Chemistry and Potential Methods for In Vivo Glucose Sensing’ in Bioinstrumentation and Biosensors, ed. Wise, D.L., Marcel Dekker, NY, USA, 1991, p. 229. Marincic, L., Soeldner, J. S., Colton, C. K., Giner, J., and Morris, S., J . Electrochem. Soc., 1979, 126, 43. 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 Marincic, L., Soeldner, J. S., Giner, J., and Colton, C. K., J . Electrochem. Soc., 1979, 126, 1687. Gough, D. A., Anderson, F.C., Giner, J., Colton, C. K., and Soeldner, J. S., Anal. Chem., 1978, 50, 941. Lewandowski, J. J., Malchesky, P. S., Moorman, M. J., Nalecz, M., and Nose, Y., IEEE Trans. Biomed. Eng., 1986, 33, 147. Yao, S. J., Chan, L.-T., Wolfson, S. K., Krupper, M. A., and Zhou, H. F., IEEE Trans. Biomed. Eng., 1986, 33, 139. Lemer, H., Giner, J., Soeldner, J. S., and Colton, C. K., An.N.Y. Acad. Sci., 1984, 428, 263. Sarangapani, S., Giner, J., Soeldner, J. S., Colton, C. K., Picha, G., Mayhan, K. G., and Drake, R. F., ‘Electrocatalytic Glucose Sensor’ in Implantable Glucose Sensors-The State of the Art (Hormone and Metabolic Research Supplement Series, Vol. 20), ed. Pfeiffer, E. P., and Kerner, W., Georg Thieme Verlag, Stuttgart, Germany, 1988, von Lucadou, I., Luft, G., Preidel, W., and Richter, G.J., ‘The Electrocatalytic Glucose Sensor’ in Implantable Glucose Sensors- The State of the Art (Hormone and Metabolic Research Supplement Series, Vol. 20), eds. Pfeiffer, E. P., and Kemer, W., Georg Thieme Verlag, Stuttgart, Germany, 1988, p. 41. Bockris, J. O’M., Piersma, B. J., and Gileadi, E., Electrochim. Actu, 1964, 9, 1329. Lyons, M. E. G., Fitzgerald, C. A., and Smyth, M. R., Analyst, 1994, 119, 855. Christie, I. M., Treloar, P. H., and Vadgama, P., Anal. Chim. Actu, 1992, 269, 65. Vadgama, P., Spoors, J., Tang, L. X., and Battersby, C., Biomed. Biochim. Acta, 1989, 48, 935. Bunn, A,, and Rose, J. B., Polymer, 1993, 34, 1 1 14. Benmakroha, Y., Christie, I., Desai, M., and Vadgama, P., Analyst, in the press. Koochaki, Z., Christie, I., and Vadgama, P., J . Memb. Sci., 199 1, 57, 83. Delahey, P., and Strassner, J. E., J . Am. Chem. Soc., 1952, 74, 893. Solomons, T. W. G., Fundamentals of Organic chemistry, Wiley, New York, USA, 1994, vol. 4, p. 904. Higson, S. P. J., Desai, M. A., Ghosh, S., Christie, I., and Vadgama, P., J . Chem. Soc., Faruday Trans., 1993, 89, 2847. p. 43. Paper 5107604H Received July 17, I995 Accepted November 21,1995