ANALYST, JUNE 1986, VOL. 111 605 Amperometric Enzyme Electrode System for the Flow Injection Analysis of Glucose* G. J. Moody, G. S. Sanghera and J. D. R. Thomas Department of Applied Chemistry, Redwood Building, UWIST, P.O. Box 13, Cardiff CFI 3XF, UK A flow injection system incorporating an amperometric enzyme electrode is described. Glucose oxidase immobilised on nylon mesh and held over a platinum electrode formed the basis of the electrode, which was incorporated in a three-electrode amperometric Stelte cell, modified for flow injection analysis. The flow and enzymatic reaction conditions have been optimised for maximum glucose response. The system exhibited good linearity (0.01 to 3.0 mM glucose), short response times (<45 s), good capacity (24 h with a continuous flow of 2.5 mM glucose) and long lifetimes (up to 4 months with storage at 4 "C).Data for the determination of glucose in foodstuffs with the enzyme electrode were similar t o those obtained by the Yellow Springs Instrument glucose analyser and a soluble enzyme test kit (Boehringer Mannheim method). Keywords: Amperometric glucose sensor; enzyme electrode; flow injection analysis; food analysis; glucose analysis Ever since the pioneering work of Clark and Lyons1 there has been increasing interest in the enzyme electrode field. Most immobilised enzyme probes have arisen for organic and biological substrates for which simple analyses were not available. Continuing interest in this field has largely been due to the availability of a wider range of enzymes, advances made in immobilisation technology and improvement of the asso- ciated sensor.The range of probes available today covers numerous substrates, and excellent reviews have been given by Carr and Bowers2 and by Guilbault.' Various suppport materials have been used for enzyme immobilisation,3-5 including alumina, charcoal, glass, silica, polyacrylamide gel, PVC matrix, nylon and proteins, such as bovine albumin. Attachment of enzyme directly on to the base sensor itself rather than on an adjacent support matrix has recently been reported.6 The use of enzyme probes has branched into several areas of analytical chemistry. Direct probes may employ poten- tiometry , amperometry, enthalpimetry or chemiluminescence as the base sensor and the amperometric approach is the most popular. Glucose electrodes, owing to their great scope in clinical chemistry, have been the subject of considerable interest, as can be deduced from reviews and literature s ~ r v e y s .~ ~ ~ ~ g Glucose flow injection analysis systems based on enzymes normally have a glucose oxidase reactor in a separate chamber from the signal detector.9 Present trends in clinical appli- cations are directed towards the miniaturisation and implan- tation of a glucose sensor as part of a regulated insulin infusion device.10 Active research in the area of chemically modified electrodes and direct electron transfer electrodes is of great interest. Thus, amperometric-type chemically modified elec- trodes for glucose have been reported6 and recent workll well illustrates the electron transfer electrode based on a ferrocene mediator.The determination of glucose in foodstuffs has received relatively little attention when compared with clinical glucose analysis. Among the current enzyme-based methods employed are a soluble enzyme test kit (Boehringer Mann- heim) and the Yellow Springs Instrument (YSI) glucose analyser. Foodstuffs requiring analysis may range from raw glucose products, such as simple syrups or powders, to already processed foods, such as biscuits. * Presented at the 30th IUPAC Congress, Manchester, UK, September 8-13th, 1985. This paper describes an amperometric enzyme electrode system in a modified electrochemical cell design for flow injection analysis (FIA) (Fig. 1). The excellent mechanical strength of nylon mesh with immobilised glucose 0xidase12.~~ gave a robust membrane for covering a standard platinum electrode forming the basis of the enzyme electrode.This electrode was incorporated in a three-electrode Stelte micro- cell modified for FIA, permitting the rapid determination of glucose in foods. Experimental Reagents Glucose oxidase (E.C. 1.1.3.4, 100 IU mg-1, purified from Aspergillus niger) , lysine monohydrochloride, dimethyl sul- phate, 25% glutaraldehyde solution and P-D( +)-glucose were obtained from Sigma Chemical Co. The enzyme was stored in a refrigerator. Glucose standards were prepared from a stock solution of P-D( +)-glucose in sodium dihydrogen orthophosphate (100 mM) buffer obtained from BDH Chemicals (Poole). The range of food samples was from the Laboratory of the Government Chemist (London). Nylon (6,6) mesh was obtained from Henry Simon Ltd.(Cheshire) with the following character- istics: open surfaces 38%, thread thickness 80 pm, mesh count 47.15 cm-1 and mesh aperture 132 pm. Immobilisation of Enzyme Glucose oxidase was immobilised on nylon mesh by a modification of the method devised by Hornby and Morris.12 Thus, enzyme was attached to single (1 cm2) and master (6 cm2) nylon mesh membranes, the procedure being the same in each instance. Nylon mesh was treated with dimethyl sulphate in a water-bath (75 "C) for exactly 5 min, followed by immersion in ice to stop the reaction. After cooling, the membrane was washed twice with anhydrous methanol, the first wash yielding a white precipitate. Lysine was attached to the methylated nylon mesh by immersion of the membrane in 50 cm3 of lysine (50 mM, pH 9.0) for 2 h at room temperature. After thorough washing with sodium chloride solution (100 mM), the mem- brane was placed in a 12.5% VWsolution of glutaraldehyde in saturated borate buffer (100 mM, pH 8.5) for 45 min.The enzyme was attached to the mesh by dipping the membrane in glucose oxidase (50 mg) in 5 cm3 of phosphate buffer (100 mM, pH 7.0) for 2 h at room temperature and then overnight at606 ANALYST, JUNE 1986, VOL. 111 4 “C. The enzyme attachment to nylon mesh procedure is achieved by the agency of the bifunctional glutaraldehyde with the lysine acting as a spacer between the nylon and the enzyme structure: NYLON MESH I +NH=CNHCH( COO-)CH2CH2CH2CH2N I1 CH ENZYME-N=CH The enzyme electrode was then fabricated by stretching the immobilised enzyme membrane (of specific immobilised activity 22.1 nmol cm-2 min-1) over a smooth platinum electrode of the Stelte cell. The excellent mechanical strength of the mesh permitted a taut fit over the electrode with the aid of an O-ring.Apparatus Fig. 1 shows the apparatus assembly used. Electrode poten- tials were controlled and currents monitored by means of a potentiostat (Metrohm VA-detector E611), capable of measuring current in the nanoamp range. A linear ylt chart recorder (Model 500) was used to record the FIA signals. Sample propulsion was made with a four-channel peristaltic pump (Ismatec Model IP-4) with sample injections being made with a manual (PTFE) valve (Tecator).All connecting tubing was of PTFE (nominal i.d. 1.27 mm). Pump pulsation was reduced with a suppressor situated immediately after the pump. Static noise was greatly reduced by earthing the flowing stream immediately after the injection valve. The detector was based on a three-electrode assembly incorporating a modified (Metrohm EA1102) Stelte micro-cell with a platinum - enzyme working electrode, a glassy carbon auxiliary electrode and a silver - silver chloride reference electrode [Fig. l(b)]. Electrical contact between the two chambers was achieved with a cellulose acetate membrane. The modification of the Stelte cell consisted of a Perspex block in the cell chamber in order to reduce the dead volume and to I- Linear Model 500 chart recorder C B Inject ion I d valve €@-& Waste Model IP/4 Modified Stelte cell l--tH peristaltic pump with enzyme electrode Fig.1. (a) Amperometric flow injection analysis apparatus, with ( b ) details of the modified three-electrode Stelte micro-cell and (c) a section through D to E of the micro cell. A, Reference electrode, silver - silver chloride; B, auxiliary electrode, glassy carbon; C, enzyme electrode chamber; D, sample inlet; E, sample outlet; F, reference and auxiliary electrode chamber; G, V notch on back of Perspex block; H, Perspex block; I, etched channel; and J, enzyme electrode produce a “wall jet” type of working electrode chamber [Fig. l(c)]. A channel was etched along the Perspex block in order to connect the inlet and outlet ports of the cell, enhancing the diffusion of sample to the electrode.The apparatus was used in the normal mode for flow injection analysis with standards and sample solutions of appropriate volume being introduced through the injection valve (Fig. 1). Results Optimisation for Glucose Analysis Enzymatic and flow parameters were optimised to obtain the best response to glucose. The potential of the working electrode and the pH of the carrier were optimised with a flow-rate of 2.3 cm3 min-1 and a sample volume of 850 mm3. For the optimisation of the working electrode potential, glucose (2.5 mM) was injected for 100 mV applied potential intervals over the range +300 to +1100 mV. Peak heights equivalent to the change in current (AZ, A) were plotted against potential [Fig. 2(a)]. In a similar fashion, the effect of pH on glucose response was investigated.The pH of the carrier and glucose standard (1.0 mM) was adjusted over the range 5.0-8.5 with sodium hydroxide [Fig. 2(b)]. With the potential and the pH optimised, the effect of the flow-rate of the carrier stream over the range 0-5 cm3 min-1 was determined for sample volumes of 500 and 850 mm3 (Fig. All further work used a potential of +600 mV relative to the Ag - AgCl reference electrode, a carrier stream and standard pH of 7.0, a flow-rate of 2.3 cm3 min-1 and a sample volume of 5UO mm3. 3). Electrode Calibration Prior to calibration of the enzyme electrode for glucose, the associated platinum electrode was calibrated for hydrogen peroxide. The hydrogen peroxide standardised with potas- sium permanganate solution (20 mM) was serially diluted with phosphate buffer (100 mM, pH 7.0) from an aqueous 100-volume solution.Standards were injected in duplicate and Fig. 4(a) shows a typical chart recorder output and Fig. 4(b) the corresponding calibration graph; response times to the peak and wash times are shown in parentheses. The calibration procedure was similar for the enzyme electrode. Fig. 5(a), showing the recorder output for the duplicate injection of glucose standards, and Fig. 5(b), the calibration graph, illustrate the response similarities between the peroxide - platinum and the glucose - enzyme electrodes. Repeated sampling of two glucose standards (2.5 and 5.0 mM) yielded mean currents of 116 and 192 nA, respectively, the corresponding standard deviations being 1.5 and 5.5 nA (n = 10).601 ,/ ‘I I I 1 5 6 7 8 9 I $ , 300 500 700 900 1100 Electrode potentiah\/ vs. Ag - AgCl Phosphate buffer (100 mM) pH Fig. 2. Optimisation of (a) working electrode otential (2.5 mM glucose) and (b) pH of the enzymatic reaction f1.0 mM glucose). Sample size, 850 mm3; and flow-rate, 2.3 cm3 min-IANALYST, JUNE 1986, VOL. 111 607 5 mM I20 nA 2 mM 1.5 2.5 3.5 Flow-rate/cm3 min-' Fig. 3. mm3; and B, 500 mm3. 2.5 mM glucose Optimisation of flow-rate for two sample volumes: A, 850 A 2.5 I T 300nA 200nA - Scan 1 4 3 1 I 1 -2 -1 0 Log([hydrogen peroxidelim~) Fig. 4. ( a ) Typical recorder out ut and (b) calibration gra h for hydrogen peroxide detection by tge platinum electrode. In (8) the numbers above the line indicate the response times and those below the line indicate wash times I ' I 0.5 mM 0.1 mM I 0.05 mM mM w 5 min , - Scan -2 -1 0 Log([glucosel/rn~) Fig.5. (a) Typical recorder output and (6) glucose calibration with the enzyme electrode. In (b) the numbers above the line indicate response times and those below the line indicate wash times. Determination of Glucose in Foodstuffs The selectivity of the enzyme electrode relative to seven common sugars was investigated. Each of the sugars was injected at the same concentration as the glucose standard (10 mM) and at ten times this ccrncentration (i. e., 100 mM for the Table 1. Effect of individual sugars on the glucose electrode Response normalisation value relative to glucose for the various sugars Interfering sugar lop2 M lo-' M Galactose . . .. . . 0.05 0.08 Maltose . . . . . . 0.10 0.20 Arabinose . . . . . . 0.04 0.06 Fructose . . . . . . 0.01 0.04 Sucrose . . . . . . 0.01 0.05 Lactose . . . . . . 0.05 0.12 Saccharin . . . . . . 0.02* 0. 15a * Commercially available sweetener (0.1 and 1% mlV) was used.608 ANALYST, JUNE 1986, VOL. 111 3.65 , 9.25 9.15 22.12 21.90 21.5 21.1 Table 2. Determination of glucose in foodstuffs by three different methods 9.2 22.0 21.3 Analytical data proposed by the Laboratory of the Government Chemist Sample Strawberry ice cream . . . . Vanillaicecream . . . . . . GlucosesyrupI . . . . . . Glucosepowder . . . . . . Horlicks . . . . . . . . Molasses . . . . . , . . Unrefinedglucosesyrup . . . . Glucoseinflour . . . . . . Sample pre-treatment Treat with Carrez solutions I and I1 Treat with Carrez solutions I and I1 Dissolve in phosphate buffer, gentle warming (35 "C) Dissolve in phosphate buffer, gentle warming (35 "C) Treat with Carrez solutions I and I1 Treat with Carrez solutions I and I1 Dissolve in phosphate buffer, gentle warming (35 "C) Treat with Carrez solutions I and I1 Method of analysis Soluble enzyme kit Soluble enzyme kit YSI glucose analyser Soluble enzyme kit Soluble enzyme kit Soluble enzyme kit YSI glucose analyser Soluble enzyme kit Glucose, YO by mass 4.4 4.7 14.8 88.0 4.8 9.2 21.4 22.0 interfering sugar).All standards and sugars investigated were prepared in phosphate buffer (100 mM, pH 7.0). Each peak height was normalised relative to the glucose signal (Table 1). Glucose in foodstuffs was determined with the enzyme electrode and the results obtained were compared with data obtained for other methods (Table 2).For the simplest foodstuffs, sample pre-treatment involved the leaching of glucose into phosphate buffer by gentle warming (35 "C). Protein-containing samples were treated with 5 cm3 of potassium hexacyanoferrate(II1) solution (80 mM) and 5 cm3 of zinc sulphate solution (250 mM) (Carrez solutions I and 11, respectively) in phosphate buffer and diluted to 100 cm3. After thorough shaking, the filtered solution was analysed for glucose content (Table 2). Discussion Optimisation of Glucose Analysis When optimising the potential of a platinum electrode for the anodic decomposition of hydrogen peroxide, the effect of pH must be considered. As the pH of the solution is decreased, the current - voltage oxidation wave becomes more anodicI4; this behaviour can be considered from the reaction scheme for the oxidation of peroxide: H202+02+2H+ +2e- .. . . (1) When applied to enzyme systems, this pH dependence is important for the optimum pH [Fig. 2(b)]. The potential producing the maximum glucose response was determined and the greatest response to hydrogen peroxide was attained with the electrode poised between +600 and +700 mV (vs. Ag - AgCl) [Fig. 2(a)]. Hence the working electrode potential chosen was +600 mV. With regard to the effect of pH on the electrode responses studied here from pH 5.0 to 9.0 [Fig. 2(b)], it can be seen that the optimum pH found lies between 6.8 and 7.2. Other workerslSJ6 who studied the pH dependence of solubilised glucose oxidase reactions found a broad range of pH 4-7 with a maximum around pH 5.5 The extent of such pH shifts are very much dependent on the immobilisation method.As mentioned earlier, pH 7.0 was adopted for the carrier and sample streams in this work. With regard to flow-rate (Fig. 3 ) , the plateau region is unexpected because, as the pumping rate is increased, the residence time of the substrate in contact with the electrode is reduced and, hence, a smaller peak height might be expected. This behaviour is observed between 0 and ca. 1.5 cm3 min-1, but between 1.5 and ca. 3.0 cm3 min-1 there is little change in the peak height (Fig. 3). At flow-rates greater than ca. 3.0 cm3 min-l, the peak height-determining factor again seems to be the residence time of the substrate.Initial problems with reproducibility were overcome by choosing a flow-rate in the plateau region (Fig. 3). At the adopted flow-rate of 2.3 cm3 min-1, fluctuations in the pump speed of up to k0.2 cm3 min-1 had a negligible effect on the response to glucose. Electrode Lifetime and Selectivity Important parameters when considering immobilised enzymes are the lifetime, durability and storage stability of the system. The excellent mechanical strength of nylon mesh membranes with immobilised enzyme eases handling during electrode fabrication and membranes can be assembled or detached without a decrease in enzyme performance. Membranes stored at 4 "C in buffer respond to glucose for up to 4 months. In order to determine membrane lifetime with respect to substrate, glucose (2.5 mM) was continuously pumped (2.3 cm3 min-1) over the immobilised glucose oxidase membrane electrode.At 2-h intervals, the electrode was washed and calibrated in order to measure the electrode capacity and robustness. The studies showed that membranes may with- stand 24 glucose hours before the onset of loss of enzyme activity, further demonstrating the ruggedness of this immobi- lised enzyme system. Selectivity studies of the glucose electrode relative to seven common sugars investigated (Table 1) produced a relatively small interfering signal, even when the interferent was ten times the glucose standard concentration. Determination of Glucose in Foodstuffs For the determination of glucose in foodstuffs, the electrode was pre-calibrated for glucose.Linearity was obtained over the range 0.01-3 mM [Fig. 5(b)], which was slightly greater than the range for the bare platinum electrode for peroxide standards. However, the peroxide [Fig. 4(a)] and glucose [Fig. 5(a)] response patterns are otherwise similar, although the current generated at the bare platinum electrode [Fig. 4(b)] is ten times greater than that for the enzyme electrode [Fig. 5(b)] for corresponding concentrations. This indicates that only a small fraction (about 10%) of the glucose that can be converted into peroxide is detected at the surface of theANALYST, JUNE 1986, VOL. 111 platinum electrode; hence, the upper limit of linearity for glucose is slightly increased. On this premise, the enzyme electrode might be expected to exhibit linearity up to 10 mM glucose (that is, 1 mM peroxide).However, at concentrations greater than 3 mM glucose the enzyme itself becomes saturated with substrate. Hence, increasing the concentration of glucose produces little change in the peak height at greater than 3 mM glucose. With regard to the analysis of food samples for glucose content with the enzyme electrode, there is good agreement between results obtained by the alternative methods and the glucose enzyme electrode method in this work (Table 2). Each sample was duplicated and the difference in peak heights varied between 0.7 and 2.2% for each sample with a mean variation of 1.75% for the sixteen peaks obtained for the eight samples. Conclusion Glucose oxidase immobilised on nylon mesh and fitted over platinum to form an indicator enzyme electrode in a modified Stelte micro-cell is suitable for the flow injection analysis of glucose.This is facilitated by the thin nylon mesh membrane with immobilised glucose oxidase permitting a fast response by the associated platinum electrode. The robustness and long lifetime of the system have advantages of economy of enzyme and of minimising diffusion effects by having the enzyme- catalysed reaction occurring at the sensing electrode. On-line dilution of samples could be used to increase the linear range in order to analyse clinical samples. The authors thank the Department of Trade and Industry (Laboratory of the Government Chemist) for financial sup- port and for glucose-containing samples and analytical data.Thanks are also extended to Mrs. Geraldine Alliston, Mr. 609 D. G. Porter and Mr. I. Lumley of the Laboratory of the Government Chemist for very helpful and inspiring discus- sions and interest. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. References Clark, L. C., and Lyons, C., Ann. N. Y. Acad. Sci., 1962, 102, 29. Carr, P. W., and Bowers, L. D., “Immobilized Enzymes in Analytical and Clinical Chemistry,” Wiley, New York, 1980. Guilbault, G. G., Ion-Sel. Electrode Rev., 1982, 4, 187. Silman, I. H., and Katchalski, Annu. Rev. Biochem., 1966,35, 873. Weetall, H. H., Methods Enzymol., 1976, 44, 134. Yao, T., Anal. Chim. Acta, 1983, 148, 27. Moody, G. J . , and Thomas, J. D. R., Ion-Sel. Electrode Rev., 1983, 5 , 243. Moody, G. J . , and Thomas, J. D. R., Ion-Sel. Electrode Rev., 1984, 6,209. Masoom, M., and Townshend, A,, Anal. Chim. Acta, 1984, 166, 111. Clark, L. C., and Duggan, C. A,, Diabetes Care, 1982,5,174. Cass, A. E. G., Davis, G., Francis, G . D., Hill, H. A. O., Aston, W. J., Higgins, I. J., Plotkin, E. V., Scott, L. D. L., and Turner, A. P. F., Anal. Chem., 1984, 56, 667. Hornby, W. E., and Morris, D. L., in Weetail, H. H., Editor, “Immobilized Enzymes, Antigens, Antibodies and Peptides,” Marcel Dekker, New York, 1975. Mascini, M., Ianello, M., and Palleschi, G . , Anal. Chim. Acta, 1983, 146, 135. Guilbault, G. G., and Lubrano, G . J., Anal. Chim. Acta, 1973, 64,439. 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. Paper A51352 Received October 2nd, 1985 Accepted December 2nd, 1985