A glucose biosensor with enzyme-entrapped sol–gel and an oxygen-sensitive optode membrane

Abstract

An optical biosensor for the continuous determination of glucose in beverages based on the canalisation of glucose oxidase into a sol–gel is presented. The enzyme was entrapped within a glass matrix by the sol–gel method. The matrix was ground to a powder form and packed into a laboratory-made flow cell. This minireactor was positioned in a spectrofluorimeter connected to a continuous sample flow system. An oxygen-sensitive optode membrane was fabricated from tris(4,7-diphenyl-1,10-phenanthroline)ruthenium(II) didodecyl sulfate adsorbed on silica gel particles and entrapped in a silicone-rubber film. The membrane was situated against the wall of the flow cell to sense the depletion of oxygen content upon exposure to glucose. The change of the luminescence intensity of the optode membrane can be related to glucose concentration. The effects of temperature and pH on the response of the biosensor were investigated. Storage, stability and repeatability of the biosensor were also studied in detail. The analytical range of the biosensor was from 0.06 to 30 mmol dm⁻³ glucose and the time taken to reach a steady signal in a flowing solution was 5–8 min. The detection limit was found to be 6 μmol dm⁻³. Common matrix interferents such as fructose, galactose, lactose, raffinose, rhamnose, stachyose, sucrose and other components in beverage samples showed no interference. The glucose biosensor has been successfully applied to the determination of glucose contents of beverage samples.

---

Introduction

In the past two decades, huge efforts in academe and in industrial laboratories have been devoted to developing biosensors for an array of analytes.¹ However, after the expenditure of an enormous amount of effort, only a few biosensor systems has been successfully commercialised. The growing demand for a more practical and reliable test has been spurring on the development of biosensor devices. In the clinical diagnosis and food industries, a glucose sensor is considered to be very desirable for analysis and process control. Clark reported the first successful electrochemical biosensor using immobilised glucose oxidase in conjunction with an oxygen electrode over 40 years ago.² Since then, the literature reporting glucose biosensors has been accumulating at a very high rate. The majority of them are electrochemical glucose biosensors. Among them, the most famous practical device for the determination of blood glucose content was developed by Yellow Springs Instruments in the early 1970s.³ Most of the electrochemical biosensors are based on measuring H₂O₂ (which is generated from the oxidation of glucose by oxygen) or the response to other electroactive species (which are previously added to the system and react with the product of the reaction between oxygen and glucose). These biosensors are very sensitive to the presence of glucose. However, a very complicated biosensor system is often specifically designed to overcome the interferents in different biological samples. The components of the sample that would be tested have to be known initially in order to correct any error in measurement. The serious drawbacks of these types of devices are that the sensing layer is so delicate that it has to be regenerated frequently; and the weak electrical signal of the device cannot withstand electric and magnetic interferences, especially in a harsh working environment.

A sol–gel matrix has been proved to be a very useful solid support for the immobilisation of enzymes as it can retain the enzyme activity and is considered to be the best way of immobilising glucose oxidase to date.⁴ Unfortunately, electrochemical-based biosensors are not easy to adapt to the sol–gel techniques for the fabrication of their glucose sensing layer because the sol–gel matrix does not provide good electrical conductivity.

Optical biosensors have been developing very rapidly since the mid-1970s.⁵ The promising features of these devices are often related to simple sensor design, easy operation, freedom from electric and magnetic interference and suitability for in situ or on-line remote monitoring. However, most optical biosensors developed so far are not as sensitive as the electrochemical biosensors. In addition, they also suffer from interference from some species in biological samples. This drawback makes the optical biosensor device very complicated in design in order to reduce the effects of interferences. Another major problem is that these biosensors are not robust. As a result, the real potential of optical biosensors is seldom realised.

For biosensor application, a sol–gel matrix encapsulated with glucose oxidase has been well studied in recent years.^4,6,7 The encapsulated glucose oxidase exhibits excellent characteristics in terms of activity, lifetime and optical transparency. In this paper, we report an improved technique for the fabrication of a sol–gel glucose biosensor. The enzyme was initially entrapped within a glass matrix by a sol–gel method.⁴ The matrix was then ground to a powder form and was packed into a flow cell together with an oxygen-sensitive optode membrane previously positioned in the flow cell. In this way, a flow-through system was set up for the determination of glucose. The optode membrane was fabricated from tris(4,7-diphenyl-1,10-phenanthroline)ruthenium(II) didodecyl sulfate [Ru(dpp)₃(DS)₂] adsorbed on silica gel particles and entrapped in a thick silicone-rubber film. The change in the luminescence intensity of the optode membrane can be related to glucose concentration.⁸ In this design, the biosensor exhibited an extremely long lifetime and showed high sensitivity to glucose. Other favourable attributes of our work are (1) the development of a very sensitive and extremely stable optical oxygen transducer, (2) the fabrication of highly active glucose oxidase entrapped in sol–gel matrix and (3) the assembly of a more practicable flow-through bed system consisting of sol–gel powder with an almost interference-free biosensor system. The effects of temperature and pH on the response of the glucose biosensor were investigated. The properties of storage, stability, repeatability, response time and interference of the biosensor were also studied in detail. We successfully applied the proposed method to determine the glucose content of some beverage samples.

Experimental

Materials

Glucose oxidase (EC 1.1.3.4. from Aspergillus niger) with a specific activity of 25 000 units per gram of solid, glucose standard solution (0.10 g cm⁻³) and β-D-glucose were obtained from Sigma (St. Louis, MO, USA), 4,7-diphenyl-1,10-phenanthroline, ruthenium(II) chloride pentahydrate and tetraethyl orthosilicate (TEOS) from Aldrich (Milwaukee, WI, USA) and sodium dodecyl sulfate (SDS) from Riedel-de Häen (Seelze, Germany). Tris(4,7-diphenyl-1,10-phenanthroline)- ruthenium(II) didodecyl sulfate dye ion pair [Ru(dpp)₃(DS)₂] was synthesised and purified as described in the literature.⁹ Silica gel particles (60 Å, 50 μm) were obtained from Matrex (Merck, Darmstadt, Germany). The one part silicone sealant SELLEYS (Selleys Chemical, Padstow, NSW, Australia), was purchased from a local supermarket. All other reagents were of analytical-reagent grade and used without further purification. The buffer solution for preparing glucose standards was 0.05 mol dm⁻³ sodium phosphate solution (pH 7.0). All solutions were prepared with de-ionised (DI) water.

Preparation of oxygen-sensitive optode membrane

A 50 mg amount of Ru(dpp)₃(DS)₂ was dissolved in 10 cm³ of acetone and 50 cm³ of ethanol. This solution was mixed with 2.0 g of silica gel particles, stirred for 2 h at 40 °C, cooled and filtered. The silica gel particles were washed with 60 cm³ of DI water three times, then dried for 6 h at 110 °C. A 0.1 cm³ portion of this oxygen indicator adsorbed on silica gel was mixed thoroughly with about 0.3 g of silicone sealant. By the spreading method the mixture was stuck tenaciously to the surface of a glass plate or a transparent film to form a silicone-based oxygen-sensitive film. It was left at 55 °C for 24 h to cure. The thickness of the oxygen sensing layer was estimated to be approximately 100 μm.

Preparation of enzyme-doped silica gel powder

A 10.4 g amount of TEOS, 1.8 g of water, 4.6 g of ethanol and 30 mm³ of 0.1 mol dm⁻³ HCl were mixed and then stirred using a magnetic stirrer at room temperature for about 8 h to prepare a clear stock sol–gel solution. A 3.0 cm³ aliquot of the stock sol–gel solution was placed in a small vial and stirred under vacuum for 20 min in order to evaporate most of the ethanol. The pH of the solution was adjusted to about pH 4.5 by adding 20 mm³ of 20 mmol dm⁻³ sodium phosphate buffer (pH 7.4) (solution A). In a separate small vial, 18 mg of glucose oxidase and 0.250 cm³ of 20 mmol dm⁻³ sodium 4-(2-hydroxyethyl)-1-piperazineethanesulfonate buffer solution (pH 7.5) were mixed, then solution A was added. A vacuum was applied to the stirred mixture until a gel was formed. The gel was rinsed with 2 cm³ of water three times. The gel was allowed to dry at 4 °C for 6 d. The dried gel was collected and ground to a powder form. Unless stated otherwise, this gel was used for most studies.

Assembly of sensing system

The laboratory-made flow-through cell used in this work was machined from stainless steel and had a chamber volume of approximately 0.45 cm³ (Fig. 1). An oxygen sensing film plus a blank glass plate were positioned as the window of the flow cell. The enzyme-doped silica gel powder was subsequently packed into the flow cell to form a small packed flow bed resulting in a minireactor or biosensor ready for glucose sensing. This minireactor was situated in a spectrofluorimeter in conjunction with a continuous sample flow system. When the glucose biosensor was not in use, it was stored at 4 °C.

Fig. 1 Schematic diagram of the flow-through cell packed with sol–gel powder and an oxygen optode membrane. (1) Stainless steel cell body; (2) sol–gel powder; (3) oxygen optode membrane; (4) transparent glass plate; (5) sample inlet; (6) sample outlet; (7) excitation light beam; and (8) emission light beam.

Instrumentation

Fluorescence intensity was measured on a Perkin-Elmer (Beaconsfield, Bucks., UK) LS-50B spectrofluorimeter which was controlled by FL WinLab software. The fluorescence emission intensity at 602 nm was collected at an excitation wavelength of 460 nm. All measurements were made with 3 nm bandwidths for both the emission and excitation monochromators. For gas-phase measurements, oxygen and nitrogen were mixed and flowed via mass flow controllers [Read Out & Control Electronics 0154 (Brookes Instrument BV, Veenendaal, The Netherlands)] directly to the sealed oxygen sensing flow-through cell. All measurements were performed in air-saturated buffer solutions. Using a MasterFlex C/L Model 77120-62 (Cole-Parmer Instrument Co., Chicago, IL, USA) peristaltic pump, the air-saturated buffer or the air-saturated glucose solutions were pumped through the flow cell at a typical flow rate of 1.0 cm³ min⁻¹. Unless stated otherwise, all fluorescence measurements were made under batch conditions at 20 ± 2 °C and at a pressure of 101.3 kPa.

Results and discussion

Response behaviour of oxygen transducer

An oxygen sensing film acting as a transducer was employed to measure the rate of oxygen consumption in the enzymatic oxidation of glucose. The optical sensing is based on collision quenching of the fluorescence of Ru(dpp)₃(DS)₂ molecules by oxygen molecules.^10,11 Hence the biosensor response composed of a dynamic balance in the diffusion of glucose into the silica gel powder and oxygen into the silicone-rubber film, and consumption of oxygen in the enzymatic reaction, resulting in a steady-state decreased oxygen level and, consequently, an increase in fluorescence intensity. Quenching can be quantified by intensity quenching measurements. The oxygen quenching process is described by the well-known Stern–Volmer equation:^9,11–13

I₀/I = 1 + K·pO₂ (1)

where I is the fluorescence intensity, the subscript 0 denotes the absence of oxygen, K is the Stern–Volmer constant and pO₂ is the partial pressure of oxygen. A plot of I₀/I versus the partial pressure of oxygen should give a straight line with a slope K and an intercept of unity on the ordinate. Fig. 2 shows the curved Stern–Volmer plot of an oxygen sensing film on exposure to various oxygen concentrations. The curvature was attributed to the distribution of slightly different quenching environments for the Ru(dpp)₃(DS)₂ complex, particularly when quenching occurs in a solid matrix.⁹ The emission spectrum of Ru(dpp)₃(DS)₂ of the oxygen sensing layer was very similar to that in the literature.¹⁴ The fluorescence intensity showed a very broad dynamic range for both gaseous and dissolved oxygen measurements, which was more than 20- and 6-fold, as shown in Fig. 2 and 3, respectively. Certainly, the sensitivity of response to dissolved oxygen is far better than that in the literature;^11–16 95% of the steady forward and reverse responses can be reached within 25 s. The concentration of Ru(dpp)₃(DS)₂ had a great effect on the fluorescence intensity. The fluorescence intensity reached a maximum when the amount of Ru(dpp)₃(DS)₂ was at 5.2 mg per gram of silica gel particles. At higher concentrations of Ru(dpp)₃(DS)₂ the fluorescence intensity decreased significantly, which strongly suggests that a high content of Ru(dpp)₃(DS)₂ may cause self-quenching.⁹ The compound in the silicone-rubber film was extremely stable and could be stored for a long period (>1 year) without any degradation.

Fig. 2 Stern–Volmer curve of the oxygen sensing film at excitation and emission wavelengths of 460 and 602 nm when subjected to various oxygen concentrations.

Fig. 3 Response curves for the oxygen sensing film cycled between (1) oxygenated water and (2) deoxygenated water.

Sol–gel enzyme bed

Proteins entrapped in nanometre-scale cages of sol–gel formed by the cross-linking of silicone and oxygen units in a sol–gel process represent a convenient, flexible and efficient immobilisation technique for enzymes which can retain their biological function in both aged gels and xerogels. It provides an efficient design that restricts the movement of the encapsulated recognition molecule and inhibits their intermolecular interaction but allows free permeation of small analyte molecules.^17,18 This study was performed on xerogels. The glucose oxidase to be encapsulated is added to the sol solution, which is first subjected to vacuum to remove most of the ethanol because solvents such as methanol and ethanol have been found to denature enzymes, resulting in decreased enzymatic activity.

The sol–gel procedures are apparently not detrimental to protein stability and resulted in optically transparent solids that are chemically, thermally and dimensionally stable immobilised proteins. The enzyme activation level is easily adjusted simply by changing the amount of enzyme. One of the main advantages of our biosensor is that the sol–gel takes the form of powder and it increases the contact surface area of the biosensor with the solution, which can assist more analytes and products to diffuse into or out of the sol–gel cages. The powder packed in a flow-through bed seems to be more suitable for real application in a flow-through analysis. The packed bed was very stable. It can be stored for over 10 months at 4 °C without apparently losing the enzyme activity. Even when kept under ambient conditions for 5 months, the enzyme activity does not decrease by more than a few per cent.

Dynamic range

The sensing scheme includes the use of glucose oxidase, an enzyme catalysing the oxidation of glucose by the dissolved oxygen in the analytical solution. Oxygen has a relatively low solubility in water, the concentration of oxygen in water in equilibrium with air being only 9.2 ppm at 20 °C and standard atmospheric pressure.¹¹ The changes in pO₂ are detected via quenching of the fluorescence intensity with the oxygen sensing film. The decrease measured in the oxygen partial pressure when glucose is oxidised by the enzyme gives an indirect indication of the glucose concentration. The fluorescence intensity, due to oxygen consumption, hence increases and reaches a plateau, the variation being proportional to the glucose concentration over a wide range. The magnitude of the analytical signal of the glucose biosensor is therefore determined by the oxygen quenching constant, the oxygen concentration and the glucose concentration inside the oxygen sensing membrane. The response behaviour of the oxygen sensor, the concentration of oxygen in the analytical solution, the amount or activity of glucose oxidase in the sol–gel powder, the temperature of the biosensor system and the flow rate of analytical solution are factors which can strongly affect the working range of the glucose biosensor. A typical calibration curve for this biosensor in a flow analysis system is displayed in Fig. 4. The relative signal change is defined as:¹³

R_s = (I_test)/(I_baseline) (2)

where I_test and I_baseline represent the detected fluorescence signals from the biosensor exposed to glucose solution and buffer solution, respectively. The maximum R_s measured with this device increases over fivefold when the biosensor was changed from an air-saturated buffer solution to an air-saturated 30 mmol dm⁻³ glucose solution. In terms of both relative signal change, R_s and detection limits, our glucose biosensor shows significant improvements over other optical biosensors of this kind.^5,6,8 The most sensitive and linear working range is 0.06–2.0 mmol dm⁻³ glucose (I_test/I_baseline = 1.0296[glucose] + 0.9968; r² = 0.995). Even though the sensitivity of the biosensor decreases above 2.0 mmol dm⁻³ glucose, the whole working range of the biosensor can still function between 0.06 and 30 mmol dm⁻³ glucose (Fig. 4) when a calibration curve has initially been prepared that covers this range of glucose concentration. The biosensor performance is not very sensitive to variations in flow rates from 0.8 to 1.2 cm³ min⁻¹. Increasing the flow rate resulted in an increased glucose saturation concentration. On the other hand, increasing the activity of the glucose oxidase in the sol–gel powder can lead to a lower detection limit and a lower glucose concentration saturation signal. It is noteworthy that the dynamic working range and detection limit of the biosensor can be modulated by controlling the relative amount of glucose oxidase entrapped in the sol–gel powder or the amount of sol–gel packed in the flow cell. It can fit the multiple requirements of an analytical method. In this work, a highly sensitive method for determination of glucose has been realised with a detection limit of 6 μmol dm⁻³ glucose. The lowest detection limit obtained was 1 μmol dm⁻³ when the flow system was in the stop flow mode.

Fig. 4 Calibration curve for the glucose biosensor at various glucose concentrations. The inset displays the linear regression curve for lower glucose concentrations from 0.060 to 2.0 mmol dm⁻³.

Effect of pH

The effect of pH was studied over the range 4.5–8.0. The biosensor was subjected to a 4.0 mmol dm⁻³ glucose standard using phosphate buffer solutions of various pH. The results showed that the optimum pH value is about 6.0. It was found that pH changes apparently do not affect the dynamic working range of the biosensor. It is anticipated that the biosensor will work satisfactorily in the pH range 5.0–8.0.

Effect of temperature

It is well known that the analytical performance of both enzyme-immobilised silica gel and oxygen transducers is highly sensitive to variations of temperature. The unquenched excited-state lifetimes of ruthenium(II)–diimine complexes at different temperatures were 5.8 (0 °C), 5.9 (25 °C), 4.8 (38 °C) and 3.3 μs (60 °C).¹² Higher temperatures would result in a significant decrease in lifetime and fluorescence intensity yield and also decrease the Stern–Volmer quenching constant of the ruthenium(II) complex, resulting in a decrease in the sensitivity of the glucose biosensor. However, raising the working temperature has a counteracting effect on the biosensor. The activity of an encaptured enzyme is governed by the kinetics of the enzymatic reaction. The reaction rate will be increased by raising the working temperature. Therefore, a study of the effect of temperature on the minireactor was carried out over the range 15–40 °C. The signal sensitivity depended strongly on the temperature of the biosensor analytical system, as expected. The fluorescence intensity changed from 986 to 780 units when in contact with 100% nitrogen as the temperature increased from 18 to 40 °C whereas the intensity shifted from 48 to 31 units when the temperature increased from 18 to 40 °C on exposure to 100% oxygen. The dynamic working range of the biosensor was reduced and the signal saturation point was reached earlier when the working temperature increased. However, the response rate of the biosensor increased sharply at higher working temperatures. The possible reasons are that the enzyme can acquire higher activity at higher temperatures and subsequently give a more pronounced signal change with oxygen consumption at a faster rate in the enzymatic oxidation of glucose. Although the analytical sensitivity was higher at around 40 °C, for practical purposes temperatures lower than 40 °C are recommended so as to prolong the lifetime of the biosensor since enzyme can easily be denatured at high temperature.

Response time, repeatability and stability of glucose biosensor

In this study, the response time is defined as the time taken to obtain a full steady state signal when the biosensor is in contact with an air-saturated buffer solution and then switches to a known concentration of air-saturated glucose standard solution. The response time depends on the enzyme activity, the working temperature, the flow rate of the analytical solution, the particle size of the sol–gel powder and the thickness of the oxygen sensing indicator layer. When the oxygen sensor material is not dispersed into a silicone-rubber film, it can be used to detect gaseous oxygen with a very fast response time (<1 s). When it is dispersed in a silicone-rubber film a few micrometres thick, it can be employed to determine dissolved oxygen in solution with a response time of ≡1 min. As a result, the response time of our glucose biosensor will be mainly determined by the enzymatic reaction rate of the enzyme and the analyte. A thicker oxygen indicator layer such as about 100 μm can also be employed as an oxygen transducer for the fabrication of our glucose biosensor. The typical time response curve depicted in Fig. 5 exhibits exponential-like behaviour. The response times of the biosensor were from 5 to 8 min. The relatively short response times are attributed to the increase in the surface area of the sol–gel powder, which enhances the exposure of the glucose oxidase to the analyte with a concomitant effect on the biosensor for reaching a fast steady-state signal. Moreover, if an enzyme entrapped in the sol–gel matrix has a higher activity, the response time should also be further shortened.

Fig. 5 Response time and reversibility of the glucose biosensor at excitation and emission wavelengths of 460 and 602 nm when subjected to various concentrations of glucose using pH 7.0 phosphate buffers (0.05 mol dm⁻³). (1) 0.010; (2) 0.10; (3) 0.50; (4) 1.0; (5) 5.0; and (6) 10 mmol dm⁻³.

Fig. 5 displays the signal changes of the biosensor when it is exposed to concentration step changes from 0.010 to 10 mmol dm⁻³ of glucose in 0.05 mol dm⁻³ phosphate buffer at pH 7.0. It demonstrates that the biosensor exhibits a very desirable analytical feature of excellent repeatability although only two repeats have been performed. The long-term stability of the sensor was tested over a 280 d period. When the biosensor was stored in a refrigerator at 4 °C and measured intermittently, the relative signal change of the biosensor on exposure to 5.0 mmol dm⁻³ glucose was found to be above 73% of its initial value over this period. The good stability of the glucose biosensor can be explained by two possible reasons. First, most of the positive charges at the channel entrance are balanced by the negative charges of the nearby acidic residues. Hence there is no detrimental electrostatic interaction with anionic sites of the immobilising matrix with the enzyme; as a result, the enzyme is stabilised upon immobilisation in hydrated silica.¹⁹ Second, the cavity in the sol–gel is tailored to the size and shape of the glucose oxidase. The bottleneck effect of the silica sol–gel prevents the enzyme from leaking.^18–20 The sol–gel powder is so strong that it can withstand any slightly striking, pressing and liquid flow pressure without causing cracking and leaching of the enzyme.

Interference test

The interference test was mainly performed in two parts. The purpose of the first part was to investigate the effects of some common substances on the quenching response of the oxygen sensing film. It was carried out with two groups of solutions. The first group of air-saturated solutions consisted of some potential interferents. The second group of solutions consisted of some potential interferents together with 2% Na₂SO₃. The results showed that there were no significant signal changes of the oxygen sensing film on exposure to the interferents. However, caffeine caused a slight interference (Table 1).

Table 1 Effect of potential interferents on the oxygen transducer and glucose biosensor

		Signal change for oxygen transducer		Signal change for glucose sensor
Interferent	Concentration/ mmol dm^−3	Air-saturated DI water	2% Na2SO3 solution	No glucose	1.25 mmol dm^−3 glucose
a Cetyltrimethylammonium bromide.
Saccharose	20	No	No	No	No
Fructose	20	No	No	No	No
Lactose	20	No	No	No	No
Galactose	20	No	No	No	No
Raffinose	20	No	No	No	No
Rhamnose	20	No	No	No	No
Stachyose	20	No	No	No	No
NaCl	100	No	No	No	No
KCl	20	No	No	No	No
MgSO4	5	No	No	No	No
CaCl2	Saturated	No	No	No	No
CuCl2	Saturated	No	No	No	≡9% decrease
FeCl3	2	No	No	No	No
Sodium citrate	10	No	No	No	No
Lauryl sulfate	1.5	No	No	No	No
Ascorbic acid	10	No	No	No	≡3% increase
CTMAB^a	2	No	No	No	No
Vitamin E (in 2  mM CTMAB^a)	3	No	No	No	No
Sodium benzoate	50	No	No	No	No
Caffeine	2	No	≡1% decrease	≡1% decrease	≡3% decrease

---

The purpose of the second part was to examine the effect of these interferents on the response of the glucose biosensor. It was evaluated with two groups of solutions. The first group of solutions contained only interferents in 0.05 mol dm⁻³ phosphate buffers. The second group of solutions consisted of interferents and 1.25 mmol dm⁻³ glucose in 0.05 mol dm⁻³ phosphate buffers. The results are given in Table 1. It is found that most potential interferents did not give any significant interference on the response of the glucose biosensor. However, in the presence of glucose, CuCl₂ can give some interference on the response of the glucose biosensor. It was interesting to find that the interference from CuCl₂ could be diminished from 9 to 4% if the test solution also contained 10 mmol dm⁻³ ascorbic acid. In the presence of glucose, ascorbic acid and caffeine had a slight interference on the response of the glucose biosensor. However this interference was reduced to nearly zero if glucose was absent from the tested solutions. In brief, we were confident that most substances often found in soft drink samples did not exhibit significant interference on the determination of glucose using our proposed biosensor method.

Glucose determination in beverage samples

Eight beverage samples of seven brands produced in various places were bought from local supermarkets and used as our test samples. The pH of a 10.0 dm³ aliquot of each sample solution was adjusted to about 7.0 by addition of a small volume of 0.2 mol dm⁻³ Na₂HPO₄ solution. It was then diluted with suitable phosphate buffer to yield a test sample solution of pH 7.0. The glucose contents in the samples were determined by the glucose biosensor (Table 2). The relative signal change of each sample solution was measured and compared with that of a set of glucose standard solutions. The recovery tests for glucose were performed by adding various amounts of glucose to the sample solutions. The amounts of added glucose were then evaluated by using our glucose biosensor. All sample solutions were air-saturated before testing. The results of the recovery of the samples are summarised in Table 2. The recovery tests demonstrate that the glucose biosensor offers an excellent, accurate and precise method for the determination of glucose in beverages with almost no effect of interferences from common matrix substances or components in the beverage sample solution. The glucose biosensor is characterised by good long-term stability, selectivity, very high sensitivity, a broad dynamic working range and a relatively fast response.

Table 2 Results of the glucose assay and the recovery test on beverage samples

Beverage (source)	Glucose content^a/ mmol dm^−3	RSD (%)	Glucose added/ mmol dm^−3	Glucose found^b/ mmol dm^−3	Recovery (%)	RSD (%)
a The glucose content is an average of seven tests, determined with the glucose biosensor.b An average of five tests.
Sprite	61.0	3.22	50.0	48.34	96.7	2.31
(Hong Kong)	100.0	98.02	98.0	3.74
150.0	149.2	99.3	2.98
Sprite	50.1	3.12	50.0	46.86	93.7	3.17
(China)	100.0	100.5	101	2.88
150.0	151.4	101	4.45
Cola	185	1.32	50.0	48.86	97.7	6.83
(Hong Kong)	100.0	101.4	101	4.74
200.0	201.4	101	3.31
Pocari Sweat	72.5	1.74	50.0	51.7	103	2.06
(Japan, 1998)	100.0	99.1	99.1	3.21
200.0	199.2	99.6	2.23
Pocari Sweat	96.7	0.59	50.0	50.26	101	2.63
(Japan, 1999)	100.0	100.14	100	2.59
200.0	198.8	99.4	3.33
Striker	153	2.21	50.0	52.44	105	1.98
(Japan)	100.0	103.7	104	3.04
200.0	201.4	101	1.74
Gatorade	105	1.39	50.0	45.52	91.0	5.01
(USA)	100.0	97.08	97.1	1.72
200.0	198.8	99.4	3.97
Lucozade	299	1.76	50.0	45.56	91.1	4.78
(UK)	100.0	97.36	97.4	3.58
200.0	197.4	98.7	4.65

---

Acknowledgements

The work described in this paper was partially supported by a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (project No. HKBU 2058/98P).

References

1. J. Janata, M. Josowicz, P. Vanysek and D. M. DeVaney, Anal. Chem., 1998, 70, 179R.
2. L. C. Clark Jr., Trans. Am. Soc. Artif. Int. Org., 1956, 2, 41.
3. E. Magner, Analyst, 1998, 123, 1967.
4. Q. Chen, G. L. Kenausis and A. Heller, J. Am. Chem. Soc., 1998, 120, 4582.
5. M. C. Moreno-Bondi, O. S. Wolfbeis, M. J. P. Leiner and B. P. H. Schaffar, Anal. Chem., 1990, 62, 2377.
6. U. Narang, P. N. Prasad, F. V. Bright, K. Ramanathan, N. D. Kumar, B. D. Malhotra, M. N. Kamalasanan and S. Chandra, Anal. Chem., 1994, 66, 3139.
7. S. A. Yamanaka, F. Nishida, L. M. Ellerby, C. R. Nishida, B. Dunn, J. S. Valentine and J. I. Zink, Chem. Mater., 1992, 4, 495.
8. W. Trettnak, M. J. P. Leiner and O. S. Wolfbeis, Analyst, 1988, 113, 1519.
9. I. Klimant and O. S. Wolfbeis, Anal. Chem., 1995, 67, 3160.
10. J. N. Demas and B. A. DeGraff, Anal. Chem., 1991, 63, 829A.
11. C. McDonagh, B. D. MacCraith and A. K. McEvoy, Anal. Chem., 1998, 70, 45.
12. J. R. Bacon and J. N. Demas, Anal. Chem., 1987, 59, 2780.
13. B. D. MacCraith, C. M. McDonagh, G. O’Keeffe, E. T. Keyes, J. G. Vos, B. O’Kelly and J. F. McGilp, Analyst, 1993, 118, 385.
14. R. C. W. Lau, M. M. F. Choi and J. Lu, Talanta, 1999, 48, 321.
15. B. Meier, T. Werner, I. Klimant and O. S. Wolfbeis, Sens. Actuators B, 1995, 29, 240.
16. E. R. Carraway, J. N. Demas, B. A. DeGraff and J. R. Bacon, Anal. Chem., 1991, 63, 337.
17. A. K. Williams and J. T. Hupp, J. Am. Chem. Soc., 1998, 120, 4366.
18. B. C. Dave, B. Dunn, J. S. Valentine and J. I. Zink, Anal. Chem., 1994, 66, 1120A.
19. J. Heller and A. Heller, J. Am. Chem. Soc., 1998, 120, 4586.
20. D. Avnir, S. Braun, O. Lev and M. Ottolenghi, Chem. Mater., 1994, 6, 1605.

---

Footnotes

† Presented at the Fifth Asian Conference on Analytical Sciences, Xiamen University, Xiamen, China, 4–7 May 1999.

‡ Visiting scholar on leave from the College of Chemistry and Chemical Engineering, Hunan University, Changsha, China.

---