ANALYST, JANUARY 1989, VOL. 114 29 Effect of Pre-treatment of Platinum for Modified Platinum Wire Glucose Oxidase Amperometric Electrodes S. K. Beh, G. J. Moody and J. D. R. Thomas* School of Chemistry and Applied Chemistry, University of Wales College of Cardiff, PO Box 972, Cardiff CF7 3TB, UK A micro-amperometric enzyme electrode, suitable for flow injection analysis and based on platinised platinum wire (100 pm diameter), is described. The system was tested for glucose oxidase covalently attached to the activated electrode surface with glutaraldehyde. Comparisons were made with micro-enzyme electrodes based on anodised platinum and thermally oxidised platinum wire, each linked to glucose oxidase. The response and wash times of each system were <25 and ca. 30 s, respectively.The platinised platinum enzyme electrode exhibited enhanced signals and a wider range towards glucose (0.005-30 mM) compared with the anodised platinum and thermally oxidised platinum systems (0.05-30 mM). Also, the lifetimes of 15 h obtained for both the platinised platinum wire and the thermally oxidised platinum wire systems considerably exceeded the 9 h obtained for the anodised platinum electrode when these electrodes were subjected to a continuous flow of 10 mM glucose. The lifetimes for all four systems during normal flow injection analysis exceeded 10 d. Keywords: Amperometric glucose sensor; enzyme electrode; flow injection analysis; modified platinum electrode Enzyme electrodes combine the selectivity of an enzyme reaction with a suitable electrochemical detection system.Such an electroanalytical approach offers considerable poten- tial because of its relative simplicity and ease of interfacing to other equipment. The methods based on oxidase enzyme electrodes frequently involve either monitoring the consump- tion of oxygen using a Clark electrode? as demonstrated by Updike and Hicks,' or monitoring amperometrically the hydrogen peroxide formed using a platinum electrode.2.3 Usually, the enzyme electrode consists of a layer of enzyme immobilised on a suitable matrix held over the sensing electrode tip, but more recently, electrodes have been described in which the enzyme is either adsorbed on modified carbon4 or chemically immobilised directly on to silanised anodised platinum .s Such thin membrane electrodes offer the advantages of improved diffusion of substrate to, and the diffusion of product(s) from, the active sensor zones.In a previous report5 describing the immobilisation of glucose oxidase on platinum wire, the platinum surface was first activated by pulsing the electrode between anodic and cathodic potentialsh.7 so that the platinum surface was cleaned and roughened.' The anodic - cathodic treatment essentially results in metal from the electrode surface dissolving during the anodic sweep and a fraction of it being redeposited during the cathodic sweep. This anodisation process is equivalent to surface evaporation and selective condensation and it pro- duces a clean, fresh metal surface. Alternative approaches to cleaning and roughening the platinum surface include polishing it with alumina powder followed by soaking in hot concentrated nitric acid and thermal oxidation, or platinisation.Both approaches were studied here, each being followed by silanisation and, finally, immobilisation of the glucose oxidase. The performance of the resulting electrodes was compared with that obtained by the anodisation of platinum prior to enzyme immobilisation as described by Moody et a1.5 Experimental Reagents Glucose oxidase (E.C. 1.1.3.4., 100 U mg-1, purified from Aspergillus niger) , 3aminopropyltriethoxysilane, hydrogen * To whom correspondence should be addressed. Table 1. Summary of pre-treatment of platinum wires up to the silanisation step Platinum wire Pre-treatmentstep A B C D Aluminacleaning . . V v V V Hotnitricacid .. . . V' V' V' V Anodisation . . . . .\/ v - - v Platinisation . . . . Thermaloxidation . . - - V V Silanisation(20°/o V/V) V - V V Silanisation (10% VW) - V - - _ - - hexachloroplatinate( IV) hydrate, lead acetate, 25% glutaral- dehyde solution and P-D( +)-glucose were obtained from Sigma (Poole, Dorset, UK) and platinum wire from Goodfel- low Metals (Cambridge, UK). The enzyme was stored in a desiccator in a freezer ( - 5 "C). All other reagents used were of the best analytical-reagent grade available and were used without further purification. A pH 4 sodium dihydrogenorthophosphate buffer (100 mM) was prepared; it was adjusted to appropriate higher pH values by spiking with sodium hydroxide solution (4 M). Glucose standards were prepared from a fresh stock solution of P-D(+)-glucose (0.1 M) in 0.1 M sodium di- hydrogenorthophosphate buffer (pH 7).The flow injection analysis (FTA) carrier stream consisted of the same buffer. Pre-treatment of the Platinum Wire Prior to use, four platinum wires (100 pm diameter) (two were used for control experiments) were cleaned by rubbing them with fine alumina powder followed by soaking in hot concentrated nitric acid, rinsing with de-ionised water and drying at 120 "C. Each wire was then treated differently prior to the silanisation step (Table 1). For the two control experiments involving glucose oxidase immobilised on platinum,5 the platinum wires (A and B) were anodised at +2.50 V (relative to silver - silver chloride) in sulphuric acid for 5 min using a potentiostat.This produces a layer of oxide on the electrode surface. In the first approach used here for oxidising the electrode surface (electrode C) the platinum wire was oxidised in an electric furnace under atmospheric pressure at 900 "C for 5 h .30 ANALYST, JANUARY 1989, VOL. 114 For the second approach, i . e . , platinisation of the platinum wire (electrode D), hexachloroplatinate was reduced galvos- tatically at 450 nA for 2 h in the presence of lead acetate with an electrode system consisting of two platinum wires. The electrolyte solution contained 33 mg cm-3 of hexachloroplati- nate and 0.6 mg cm-3 of lead acetate. Prior to silanisation the platinised platinum electrode was also heated in an electric furnace for 5 h at 900°C. The following scheme illustrates the platinum electrode pre-treatment , silanisation and enzyme immobilisation steps.I NH2 Glutaraldehyde Glucose 1 C H=N- E n zy m e - oxidase Immobilisation of the Enzyme Three of the four pre-treated platinum wires, namely A, C and D, were refluxed for 1 h with a solution of anhydrous 3-aminopropyltriethoxysilane in toluene (20% VlV), whereas wire B was refluxed for 1 h with a 10% V/V solution of anhydrous 3-aminopropyltriethoxysilane in toluene in order to observe the effect of different concentrations of silanising agent. After silanisation, each platinum wire was placed in glutaraldehyde solution (5% V/V in 100 mM phosphate buffer) in a stoppered sample tube for 1 h. To attach the enzyme to the treated wire, the electrode was dipped overnight in a solution of glucose oxidase (30 mg) in phosphate buffer (1 cm3 of 100 mM pH 7 buffer) at 4°C.The same batch of enzyme was used throughout this work. Apparatus An FIA system (Fig. 1) for monitoring glucose was used to evaluate the four different electrodes. The electrode potential was controlled with, and the current monitored by, a Metrohm E 61 1 VA-detector potentiostat. A Servoscribe chart recorder was used to record the output signal. The carrier stream and sample were propelled by a four-channel Watson Marlow peristaltic pump, and an Omnifit sample injection valve was used. All connecting tubes were made from either silicone rubber or PTFE and had a nominal internal diameter of 1.27 mm. A pulse suppressor was fitted between the pump and the injection valve. The hydrogen peroxide produced by the enzymatic reaction was monitored with a three-electrode amperometric flow- through cell system designed for FIA (Fig.1) at 600 mV (relative to a silver - silver chloride electrode). The Perspex flow-through cell [Fig. l(b)] was laboratory-built and con- sisted of a platinum wire based enzyme electrode (I) as the working electrode, in addition to a platinum wire auxiliary ( a ) Servoscri be chart recorder Pump Waste Amperometric flow-through cell I I Ill Fig. 1. Schematic representation of (a) the FIA apparatus and ( b ) the amperometric flow-through cell (A). (I) Enzyme working electrode; (11) auxiliary platinum electrode; (111) silver - silver chloride reference electrode; (E) silicone rubber seals; (F) plastic electrode holders; (G) potassium chloride solution; and (H) sample inlet 2 400 22 a 4.3 5.3 6.3 7.3 PH Fig.2. pH optimisation rofiles for the four different enzyme electrodes. ( A ) A; (0) B; 0) C; and (A) D electrode (11) and a silver - silver chloride reference electrode (111). The cell was designed so that the auxiliary and reference electrodes were placed in a stationary solution of saturated potassium chloride and were in contact with a flowing buffer stream by means of a T junction [Fig. l ( b ) ] . Each electrode was held in place with silicone rubber seals. When not in use the enzyme electrode was stored at 4 "C in 100 mM pH 7 phosphate buffer. Results Optimisation of Conditions for Glucose Determination The sample volume and carrier stream flow-rate were optimised for each electrode using a modified Simplex optimisation algorithm for which the control parameters were the flow-rate and sample volume with respect to the response criteria of peak height and the time taken from sample injection to the attainment of the maximum signal.The bias of the optimisation is the maximisation of peak height but for the minimum time taken from injection to signal with a greater emphasis on peak height. Hence the carrier stream flow-rate was fixed at 2.0 cm3 min-1 and the corresponding sample size was optimised at 0.50 cm3 for all further work.ANALYST, JANUARY 1989, VOL. 114 31 The pH was optimised for each of the four enzyme electrodes by varying the pH between 5 and 9 by increments of 0.5 pH. The resulting peak height - pH plots showed plateaux at pH 6-7.5 (Fig.2). Therefore, all further work was carried out at pH 7 for the different enzyme electrodes. Other workers have reported a broad pH range of 4.0-7.0 with a maximum response around pH 5.5 for solubilised glucose oxidase.*.g The optimum pH range is a direct result of the micro-environment of the enzyme and is related to the immobilisation technique employed and to the nature of the supporting material. Electrode Calibration The electrodes were calibrated with glucose standards over the range 0.005-100 mM using the optimised conditions described above. Fig. 3 shows a typical chart recorder output obtained with electrode D for the triplicate injection of glucose standards and illustrates the utility of the electrode under FIA conditions.The lowest detectable concentration of glucose using the platinised electrode (D) was 0.005 mM, but was only 0.1 mM for the two control anodised platinum electrodes (A and B) and for the thermally oxidised platinum wire electrode (C). The calibration graph was linear up to 30 mM glucose [Fig. 4(a)]; the log - log plots are shown in Fig. 4(b). 12.5 pA I 100 r n M I\ 0 r n M 1 Omn 750 nA1 I 5 min H I Scan time --- Fig. 3. electrode D Typical recorder output towards glucose for glucose oxidasc 40 000 30 000 P a 20000 10 000 0 Discussion The various pre-treatments applied to the platinum wire prior to enzyme immobilisation lead to enzyme electrodes with different response characteristics, particularly for the platin- ised platinum system (D) compared with the other electrodes.Hence electrode D shows considerable enhancement of signal response over both electrode C and the control electrodes A and B, and permits improved sensitivity towards low levels of glucose (down to about 0.005 mM) (Fig. 4). No analytically useful signals were obtained for glucose standards below 0.05 mM with electrodes A, B and C. However, the speed of response, under the optimised conditions, is similar for all four electrodes, being ca. 24 s from base line to peak response in each instance. The wash times, i.e., the times required for the signal to revert to the base line, are only slightly longer, being ca. 30 s (see, for example, reference 5 and Fig. 3). For electrodes A, B and C the platinum wire was oxidised either by anodisation or by thermal means, whereas the platinised platinum wire D was oxidised by means of a final thermal stage aimed at preventing losses or redistribution of the newly formed platinised surface.Each approach was adequate for silanisation of the electrode surface as can be seen in systems A and C for the same concentration of silanising agent. The effect of doubling the concentration of the silanising agent for systems A and B produced little change in the signal. The platinisation of platinum wire increases both the surface area of the electrode and also the number of active sites available for silanisation. This in turn enhances the amount of enzyme that can be immobilised on the electrode surface, thus leading to significantly larger current outputs for electrode D, viz., about seven times greater than for the control electrodes A and B and the thermally oxidised electrode C (see Fig. 4).The advantages of electrode D are apparent from a one-way analysis of variance (ANOVA), the results of which are shown in Table 2. The analysis indicates that electrode D gives a highly significant improvement in response over the other electrodes and that this improvement is considerably greater than for any comparison between electrodes A, B and C. With regard to optimisation of the experimental conditions it should be noted that these were similar to the various electrode types. This, however, is to be expected from the A A A I I -2.5 I -1.5 I 0 ' --5.5 -4.5 -3.5 Log ([glucosel~M) Fig. 4. Glucose calibration plots. ( u ) A1 ( y ) versus [glucose] (x).( A ) Electrode A: y = 0.770(+0.066) + 10.5.510(k2001.9)~, r2 = 0.999: (0) electrode B: v = 0.720(+0.016) + 188970(5889.0)~, r2 = 1.000; (0) electrode C: y = 0.860(+0.066) + 117234(+222.5.8)~, r2 = 0.999; and (A) electrode D: y = .5.300(+0.066) + 732713(+13906.8)x, r2 = 1.000. Prediction intervals for single y values. (A) Electrode A: i2001.9; (0) electrode B: f889.0; (0) electrode C: 5222.5.8; and (A) electrode D: k13906.8. ( h ) Log AZ b) versus log [glucose] (x). (A) Electrode A: y = .5.0380(k0.066) + 1.0053(+0.02l)x, r2 = 0.999; (0) electrode R: y = .5.040.5(~0.016) + 1.0053(+0.00.5)x, r2 = 1.000; (0) electrode C: y = .5.0838(50.066) + 1.00.53(50.021)~, r* = 0.999; and (A electrode D: y = 5.8785(+_0.066) + 1.0049(+_0.021)~, rz = 1.000. Prediction intervals for single y values.( A ) Electrode A: k0.066; (01 electrode B: k0.016; (0) electrode C: 50.066; and (A) electrode D: k0.06632 ANALYST, JANUARY 1989, VOL. 114 Table 2. One-way analysis of variance for the four types of platinum wire electrode. Critical values are 4.41 atp = 0.05 and 8.29 atp = 0.01 with numerator = 1 and denominator = 18 Comparison of electrodes AandB , . AandC . . AandD . . BandC . . BandD . . CandD . . . . . . Difference at Difference at highly significant F-factor significant level level 347.75 v v 8.20 v X 1.3 x 106 \’ v 134.84 v v 1.6 x 106 v v 3.1 x 105 v v nature of the electrode under study, which simply consists of an enzyme layer immobilised directly on the electrode surface. Hence, diffusion conditions are more ideal than in instances where an enzyme membrane on a separate support is placed over the electrode tip.Electrode lifetimes and storage stability are significant factors when considering a wider practical role for biosensors. Hence, when a 1 mM glucose solution was pumped continu- ously through the electrode system until failure occurred, it was found that the platinised platinum system (D) and the thermally oxidised system (C) had lifetimes of 15 h, which was an improvement over the lifetimes of 10 h found for the control anodised systems (A and B). Also, when stored in a buffer overnight at <6”C, and with daily use, all the electrodes (A-D) had a lifetime in excess of 10 d. Therefore, there is a general improvement in the performance of electrodes C and D compared with those in which the enzyme is immobilised on anodised platinum as reported previously.5 The critical factor with regard to lifetime may relate to the stability of the Pt-0 bonds, i.e., cleavage of these bonds would lead to a loss of enzyme and hence deactivation of the electrode.However, the thermal pre-treatment stage employed for electrodes C and D seems to be beneficial. Interestingly, Masoom and Townshendl0 reported a glucose oxidase activity of up to 1 year for 3-aminopropyltriethoxy- silane - glutaraldehyde - glucose oxidase on controlled poros- ity glass (CPG). One factor that can reduce electrode lifetimes is the setting of the electrode to an anodic potential which may facilitate desorption of oxygen, resulting in premature loss of enzyme activity by cleavage of the Pt-0 bonds.Conclusion Platinisation of platinum wire significantly enhances the electrode surface features for the immobilisation of glucose oxidase and yields an improved enzyme electrode (type D) with respect to current response, linear range, detection limits, lifetime and storage stability. Electrodes (type C) prepared from thermally oxidised platinum compare favour- ably with the type D sensor except for the lower signal output and shorter linear response range. Hence both types of electrode offer advantages over the control electrodes (types A and B). The authors thank the Trustees of the Analytical Chemistry Trust Fund of the Royal Society of Chemistry for the award of an SAC Research Studentship (to S. K. B.). References 1. 2. 3. 4. 5 . 6. 7. 8. 9. 10. Updike, S. J . , and Hicks, G. P., Narure (London), 1967, 214, 986. Guilbault, G. G . , “Analytical Uses of Immobilized Enzymes,” Marcel Dekker, New York, 1984. Guilbauit, G. G., Ion-Sel. Electrode Rev., 1982, 4, 187. Marko-Varga, G., Appelquist, R., and Gorton. I . , Anal. Chim. Acta, 1986, 179, 371. Moody, G. J . , Sanghera, G. S . , andThomas, J. D. R . , Analyst, 1986, 111, 1235. Woods, R.. Electroanal. Chem., 1976, 9, 9. Gilman, S . , Electroanal. Chem., 1967, 2, 111. Bright, H. J . , and Appleby, M., J. Biol. Chem., 1969, 244, 3625. Weibel, M. K., and Bright, H. J . , J. Biol. Chem., 1971, 246, 2734. Masoom, M., and Townshend, A., Anal. Chim. Actu, 1984, 166, 111. Paper 8l031.501 Received August 2nd, 1988 Accepted September 26th, 1988