Optical sensors

Fresenius Zeitschrift fiir

Fresenius Z Anal Chem (1988) 332:255-257

9 Springer-Verlag1988

Optical sensors Part 20. A fibre optic ethanol biosensor* Otto S. Wolfbeis and Hermann E. Posch Analytical Division, Institute of Organic Chemistry, Karl-Franzens-University, A-8010 Graz, Austria Optische Sensoreu. Teil 20. Ein faseroptischer Biosensor fiir Ethanol Summary. A fibre optic biosensor for ethanol was developed, which is based on the enzymatic oxidation of ethanol. The sensor layer contains an oxygen-sensitive fluorescing indicator which reports the decrease in the local oxygen partial pressure as the result of the enzymatic oxidation. The sensor measures in the 50 - 500 retool/1 ethanol range, with an accuracy of + 4 mmol/1 at 100 retool/1. The detection limit is 10 retool/1 ethanol.

Introduction Continuous determination of ethanol plays an important role in proper management of various biotechnological processes including beer brewing and wine fermentation. Ethanol assay in blood is another important field of application. Consequently, a number of methods for the quantitation of ethanol has been developed [1, 2]. However, only a few of them lend themselves to continuous and reversible recording. Mention should be made of ethanol sensors based on field effect transistors [3], some of which have been coupled to microbial reactions [4], and of polarographic (amperometric) oxygen electrodes coupled to enzymatic reactions. Alcohol oxidases (AOD) and alcohol dehydrogenases have been employed most frequently in these biosensors [ 5 - 7]. Alternative methods for ethanol assay, particularly in alcoholic beverages, include headspace isolation procedures and detection with a fuel cell sensor [8] and continuous-flow bioluminescence assay [9]. Guilbault et al. [10] compared enzyme electrode assays based on polarographic oxygen electrodes with enzyme thermistor alcohol assays and found both methods to give adequate sensitivities. The Provesta Corp. (Bartlesville, OK 74004) and Universal Sensors (New Orleans, LO 70148) supply oxygen electrodes for coupling with immobilized alcohol oxidase in gel form [11]. Other methods employed for alcohol biosensors have been reviewed [12], and an excellent monograph on the state of art in biosensing has appeared [13]. This paper describes a fibre optic approach to sense ethanol, using an oxygen-sensitive optosensor as a trans* Presented in part at the Biosensor International Workshop 1987 at GBF, Braunschweig, June 1987 Offprint requests to: O. S. Wolfbeis

ducer, coupled to an enzymatic reaction. The device is likely to have the typical advantages of fibre optic sensors over other kinds of sensing devices [14]. Recently, Walters et al. [15] have described an entirely different method to sense ethanol via fibre optics: the enzyme alcohol-dehydrogenase (ADH) was retained in position at the distal end of an fibre bundle. In the presence of ethanol, non-fluorescent NAD + (which is the coenzyme of ADH) is reduced to give fluorescent NADH. The rate of NADH production is monitored and can be related to the ethanol concentration. An optical ethanol sensor has also been described in a brief note, but no experimental details were given in this work [16]. Moreover, since all measurements have to be made in the UV, this system cannot be applied to conventional glass or plastic fibres with their poor transmission below 420 nm.

Experimental Reagents Alcohol oxidase (EC 1.1.3.13), from Pichia pastoris, dissolved in a 60% solution of sucrose in water, was provided by Sigma (Munich, FRG). Catalase (EC 1.11.1.6), from bovine liver, purity ca. 50% as assayed by photometry, was provided by the Institute of Biochemistry of the KFU. Both were used as received. Their specific activities were 5 units/ mg and 6000 units/mg, respectively. It is essential to use an alcohol oxidase of high specific activity. Unbuffered and air-saturated aqueous stock solutions of ethanol were used throughout. Sensor layer preparation The method for the preparation of ethanol-sensitive layers is very similar to the one reported [17] for a new type of oxygen sensor, having the indicator embedded in an aqueous environment. A cross-section of the sensing layer is shown in Fig. 1. Basically, the preparation comprises the steps of (a) preparing an aqueous solution of alcohol oxidase (12.8 mg/ml), catalase (0.1 mg/ml), and tris-(2,2'-dipyridyl)ruthenium(II) dichloride (0.5 mmol/1) in phosphate buffer (0.015) tool/1 o f p H 7.0, (b) soaking silica gel (KieseIgel, type Si300, from Merck, FRG) with this solution, and (c) mixing silica gel with silicone prepolymer (E43, Wacker, FRG) in a ratio of 1:2. A 100 ~tm thick film of this mixture is spread onto a 1 x I cm glass slide and left for curing in moist atmosphere at room temperature for several hours. The layers were stored at 4~ under water.

256 sample

EtOH

14

0 2 HP

I q qaA 10

7///5 (, 'r

7//////,

H

6

2 Exr

Fig. 1. Left Cross-section through the ethanol-sensitive membrane used in this work. The silica gel beads (B) contain indicator, catalase and alcohol oxidase and are embedded in a 100 gm thick silicone rubber membrane that is glued onto a glass support. The directions of exciting light and fluorescence are also given. Right Chemical species and the direction of their diffusion in the ethanol biosensor. EtOH ethanol; HP hydrogen peroxide; A A acetaldehyde

Experimental arrangement

460 nm light from a 200 W tungsten halogen lamp is filtrated out from the total lamp spectrum, using the Aminco SPF 500 fluorimeter double monochromator system, and focussed into one end of a 1.5 m bifurcated fibre optic light guide, the common end of which is directed onto the slide containing the sensing layer described above. The fluorescence of the layer is collected by the bundle of light guides of the second arm and guided to the emission monochromator of the fluorimeter set to 610 rim. A PMT detects fluorescence intensities which are graphically recorded on a HP 7225 A X/Y plotter. All measurements were performed in a room thermostatted to 22~ The slide with the sensor layer was used to form one wall of a flow-through cell, with the sensing membrane exposed to the sample solution and the glass support attached to the fibre end. The flowthrough cell was optically isolated from ambient light by covering it with black teflon spray. Airsaturated stock solutions of ethanol in water were pumped through the cell and the changes in fluorescence recorded versus time. Results

The detection principle is based on the following two reactions that are catalyzed by the two enzymes: C2HsOH + 0 2 - + CH3CHO + H202 (alcohol-oxidasecatalyzed) 2 I-I202 -+ 2 H 2 0 + O2 (catalase-catalyzed) Catalase is added in order to destroy the hydrogen peroxide which decomposes AOD. Overall, one equivalent of oxygen is consumed when ethanol is oxidized to acetaldehyde according to 2 C2HsOH + 02 = 2 CH3CHO + 2 HaO. In other words, the consumption of 1 equivalent of molecular oxygen indicated the oxidation of 2 equivalents of ethanol so that AO2 = k x [EtOH]/2. The resulting decrease in oxygen is seen by the oxygen sensor, whose working principle is based on dynamic quenching of the fluorescence of the ruthenium dye by molecular (triplet) oxygen. The various chemical species and diffusion processes involved in this sensing scheme are shown in Fig. 1. Both

0.1 0,2 0.3 0.l. ethanol concentration [mot/tl

0.5

Fig. 2. Change in optical signal (delta I) in % of the fibre optic ethanol sensor versus ethanol concentration at 22 ~ in air saturated solutions. The bars give the ranges as obtained with 4 independently prepared sensors

oxygen and ethanol diffuse through the silicone rubber into the aqueous silica gel compartments where ethanol is oxidized, resulting in consumption of oxygen. A steady state will be established after some time which is governed by the rates of diffusion of oxygen and ethanol into the aqueous phase, by the activities of the enzymes, and by the rate of diffusion of acetaldehyde out of the sensing membrane. While alcohol oxidase is a fairly "slow" enzyme, catalase has an extremly high turnover number, so that it is unlikely to slow down the equilibration process and that large amounts of hydrogen peroxidase will leave the beads and the sensing membranes, respectively. The response time of the sensor is rather quick (ca. 2 rain) with the best sensitivity being between 50 and 500 mmol/1 ethanol. This is the interesting concentration range in most types of fermentation processes including beer brewing. Detection limits appear at present to be at about 10 mmol/1. Although not specifically tested yet, it is likely that alcohols other than ethanol, which were shown to interfere in the electrochemical ethanol sensor [12], will interfere in this case as well. Typical interferents are methanol and the higher alcohols. The lifetime of the sensor is in the order of 2 weeks, after which its response is reduced to about 10% of its initial value. This is mainly due to the poor stability of the enzyme. We also noticed a varying response of the sensor when different lots of the enzyme were used, a fact that required recalibration of the system whenever a new sensor was used. Assuming a 0.5% precision in the determination of fluorescence intensity, a precision of + 10 mmol/1 in the 0.5 tool/1 ethanol concentration range is to be expected.

Discussion

The sensor principle introduced here is an adaption of the electrode-based sensing scheme to the fibre optic technology. Thus, it is likely to have all the advantages of optical sensors such as lack of reference electrodes, ease of miniaturization, and the possibility of remote sensing. The required mass transfer of analyte into the reaction volume, which occasionally was considered to be a disadvantage of fibre optic sensors [18], turns out to be an advantage: First, it

257 contributes to the selectivity of the sensor in that only volatile substances can diffuse through the membrane. Secondly, it protects the enzymes from being adversely affected by the sample with its varying p H and which may contain enzyme inhibitors such as heavy metal ions or irreversibly binding substrates. The relative change in oxygen partial pressure would be even larger in the absence of the catalase, but hydrogen peroxide is an extremely reactive species that causes a rapid deactivation of ethanol oxidase, so that it has to be removed immediately after formation [10]. Even though, the system has a rather short lifetime which we think is due to the formation of acetaldehyde, a species that can react with the amino group of enzymes in various ways, thereby leading to inactivation. A further improvement in lifetime may therefore be achieved by addition of a third enzyme that oxidizes the aldehyde to the acid, or by adding another species that reacts with acetaldehyde. The dye used in this work does not inhibit the enzymes. Almost identical results were obtained when dye and enzymes were contained in the same silica gel phase, or when dye and enzymes were incorporated into different silica gel beads which then were mixed with silicone. The sensor has a Stern-Volmer type of response provided that sufficient oxygen is available (Fig. 2). The graph is linear, showing that up to 500 retool/1 ethanol there is sufficient oxygen supply not to become rate limiting. This is due to the low amount of alcohol oxidase in the sensor beads. However, if diffusion of oxygen becomes rate-limiting, or when high concentrations of alcohol oxidase are present, a downward curvature of the respective plot is observed (not

shown). Since the sensing principle is based on the detection of consumed oxygen, a change in the supply of oxygen can cause changes in the signal, even at constant ethanol concentrations. In such a situation (which is a realistic assumption in case of bioreactors and related fermenters) a two-sensor approach has to be made that accounts for changes in oxygen supply. In this approach, two sensors are to be used, their only difference being that one contains enzymes, and the other not. The Stern-Volmer equation predicts the following relations between optical signal and analyte concentrations: Sensor A (containing no enzymes) responds as in Eq. (1) Co/l" - I = G "

[02]

(1)

with I a and I a being the fluorescence intensities measured in the absence and presence of oxygen, respectively, ksv is the Stern-Volmer quenching constant, and [02] is the concentration of oxygen. The response of sensor B, i.e., the one containing the enzymes, can be described by a modified Stern-Volmer equation [Eq. (2)]: Po//b - 1 = ksv [O2] - ksv-[02]

(2)

with Po being the fluorescence intensity of sensor B in the absence of both oxygen and ethanol, and/b being the intensity in the presence of certain levels of these species. [O2] is the respective oxygen partial pressure [same as in Eq. (1)], and [02] is the change in oxygen partial pressure that is caused by the oxidation of ethanol. The relation between [O2]' and [EtOH] can be described by Eq. (3): [O2]' = f " [EtOH]/2

(3)

where factor f accounts for the unknown rate of conversion of ethanol into acetaldehyde and hydrogen-peroxide, and factor 2 has to be introduced because i mole of oxygen is consumed when 2 moles of ethanol are oxidized. Equation (2) can therefore be written as Ibo/lb -- 1 = ksv" [02] -- ksv' f " [EtOH]/2 .

(4)

Setting ~ for ( I ~ / I - 1) (which is the signal of sensor A) and fi for ( I ~ o / [ b - - 1) (the signal of sensor B), and subtracting Eq. (4) from Eq. (1) we obtain -//=

ksv - f - [EtOH]/2 .

(5)

Thus, the difference between signals c~ and ~ is linearly related to the ethanol concentration via a constant c which is equal to ksv" f / 2 . A plot of c~-fi versus [EtOH] should result in a straight line that goes through zero and has a slope K. Using this calibration graph, the concentration of ethanol may be calculated from the two signals even under varying levels of oxygen. In fact, the two-sensor technique may be used for continuous recording of both oxygen and ethanol in fermenters. Acknowledgement. This work was supported by the "Fonds zur

F6rderung der wissenschaftlichen Forschung", project P5977C, which is gratefully acknowledged.

References

1. Mansoom M, Townshend A (1986) Anal Chim Acta 185:49 and references cited 2. Prencipe L, Iaccheri E, Manzati C (1987) Clin Chem 33:486 3. Josowicz M, Janata J (1986) Anal Chem 58:514 4. Kitagawa Y, Ameyama M, Nakashima K, Tamiya E, Karube I (1987) Analyst 112:1747 5. Birch SW, Turner APF, Ashby RE (1987) Process Biochem 22:37 5, Takei H, Nakashima T, Adachi O, Shinagawa E, Ameyama M (1985) Clin Chem 31:1985 7. Verduyn C, Van Dijken JP, Schaeffers WA (1983) Biotechnol Bioeng 25 : 1049 8. Criddle WJ, Parry KW, Jones TP (1986) Analyst 111:507 9. Girotti S, Roda A, Ghini S, Piacentini AL (1986) Anal Chim Acta 183:187 10. Guilbault GG, Danielson B, Mandenius CF, Mosbach K (1983) Anal Chem 55 : 1582 11. Mascini M, Guilbault GG (1986) Biosensors 2:147, 169 12. Scheller FW, Pfeiffer D, Schubert F, Renneberg R, Kirsten D (1987) In: Turner APF, Karube I, Wilson G (eds) Biosensors. Fundamentals and applications, Chap 18. Oxford University Press, Oxford 13. Schmid RD (1987) Biosensors Workshop GBF Monograph, vol 10. Verlag Chemic, Weinheim 14. For a review, see: Wolfbeis OS (1988) Optical and fiber optical sensors in analytical and clinical chemistry. In: Schulman SG (ed) Molecular luminescence spectroscopy, part 2, chap 3. Wiley, New York 15. Walters BS, Nielsen T J, Arnold MA ( 1988) Talanta 35 : 151 16. V61klKB, Opitz N, Liibbers DW (1980) Fresenius Z Anal Chem 301:162 17. Wolfbeis OS, Leiner MJP, Posch HE (1986) Mikrochim Acta 1986III: 359 18. Seitz WR (1984) Anal Chem 56:16A Received May 3, 1988