ANALYST, AUGUST 1992, VOL. 117 1235 Am perometric Biosensors Based on an Apparent Direct Electron Transfer Between Electrodes and Immobilized Peroxidases* Plenary Lecture Lo Gorton, Gunilla Jonsson-Pettersson, Elisabeth Csoregi, Kristina Johansson, Elena Dominguezt and Gyorgy Marko-Varga Department of Analytical Chemistry, University of Lund, P.O. Box 124, S-22 7 00 Lund, Sweden An apparent direct electron transfer between various electrode materials and peroxidases immobilized on the surface of the electrode has been reported in the last few years. An electrocatalytic reduction of hydrogen peroxide stars at about +600 mV versus a saturated calomel (reference) electrode (SCE) at neutral pH. The efficiency of the electrocatalytic current increases as the applied potential is made more negative and starts t o level off at about -200 mV versus SCE. Amperometric biosensors for hydrogen peroxide can be constructed with these types of peroxidase modified electrodes.By co-immobilizing a hydrogen peroxide-producing oxidase with the peroxidase, amperometric biosensors can be made that respond t o the substrate of the oxidase within a potential range essentially free of interfering electrochemical reactions. Examples of glucose, alcohol and amino acid sensors are shown. Keywords: Biosensor; direct electron transfer; amperometry; electrode; immobilized enzyme Amperometric biosensors have been at the focus of electro- analytical research since the first 'enzyme electrode' for the detection of glucose was reported by Updike and Hicks in 1967.' More than 1000 papers have been published since then on amperometric biosensors for glucose and for a series of other analytes.The field of enzyme based amperometric biosensors was recently reviewed by Bartlett et a1.2 One of the major obstacles to be solved in the construction of enzyme based amperometric biosensors is how to optimize the electron transfer reaction between the cofactor of the redox enzyme used and the electrode. All redox enzymes rely on a cofactor as the redox active compound for activity. In all classes of redox enzymes, except the nicotinamide dependent dehydrogenases, the cofactor is strongly bound within the enzyme structure causing steric hindrances for a direct electron transfer between the active site of the enzyme and the electrode.3 Only in a few cases has a direct electron transfer reaction been claimed to occur and in still fewer has the electron transfer reaction been efficient enough to allow the construction of a sensor.2.4 The electron transfer has mostly been brought about using a soluble species that can diffuse from the active site to the electrode and vice ~ersa.2~4 Sometimes redox mediator modi- fied electrodes have been used to facilitate the electron transfer.2.4 A special case is the use of conducting salt electrodes in combination with redox enzymes .2,5 Peroxidases have been intensively studied for the construc- tion of amperometric biosensors not only for sensing hydrogen peroxide and small organic peroxides but also in the combina- tion with hydrogen peroxide producing oxidases for serving the substrate of the oxidase, e.g.glucose, alcohol, amino acids and xanthine.2 In most cases soluble mediators610 or mediator modified electrodes have been used in conjunction with these sensors.2.I 1 However, in 1977, Yaropolov et al.12 reported on what appeared to be a direct and a very efficient electron transfer from carbon black to adsorbed horse radish peroxidase (HRP, E.C. 1.1.11.7) in the presence of hydrogen peroxide. An electrocatalytic reduction of hydrogen peroxide started to occur at about +600 mV versus a saturated calomel (refer- ence) electrode (SCE) at pH 7. Making the applied potential more negative resulted in a much increased current. Not until about -200 mV did the current level off to a more constant value. Similar effects were reported by Paddock and Bow- den13 on cytochrome c peroxidase adsorbed on edge oriented pyrolytic graphite, by Jonsson and Gorton on HRP adsorbed on spectrographic graphite,14 by Kulys and Schmid on fungal peroxidase on graphite,'5 by Wollenberger et al. on HRP immobilized on graphite and Pt,16 by Gorton et al.on HRP, lactoperoxidase and microperoxidase adsorbed on various graphites, glassy carbon and coal electrodes,17-19 by Wata- nabe et al. on immobilized microperoxidase on Sn02 elec- trodes,20 by Gorton et al. on HRP immobilized in carbon paste,4 and by Wollenberger et al. also on HRP and fungal peroxidase immobilized in carbon paste and in graphite epoxy resins.21 The peroxidase modified electrodes can thus work as amperometric biosensors for hydrogen peroxide detection within the optimum potential range, -200 and 0 mV versus SCE, where the risk for interfering reactions is minimized and also where the background current and noise levels take their lowest values.Sensors for oxidase substrates have also been studied using this effect by co-immobilizing a hydrogen producing oxidase with the peroxidase. Kulys and Schmidls studied sensors for alcohol, choline, and glucose by co-immobilizing alcohol, choline and glucose oxidase together with fungal peroxidase on solid graphite electrodes. Gorton et a1.17318 and Jonsson- PetterssonZ2 reported on glucose sensors based on co-immobi- lizing glucose oxidase with HRP on heat-treated spectro- graphic graphite. Here we report on the latest achievement in this direction from our laboratory.We have been involved in the investiga- tion on the background mechanism, construction of co- immobilized oxidase-peroxidase based carbon-paste sensors, and construction of microsensors based on carbon fibres. * Presented at the meeting on Analytical Applications of Chemi- cally Modified Electrodes, Bristol, UK, January 7-8, 1992. 1- Permanent address; Department of Nutrition and Food Analysis, University of Alcala de Henares, 28871 Alcala de Hcnares, Madrid, Spain. Background Mechanism The reaction between hydrogen peroxide and peroxidases occurs according to the following. In a first 2e- transfer step hydrogen peroxide is reduced to water and the bound cofactor1236 2 H H ~ ~ H202 ANALYST, AUGUST 1992, VOL. 117 hemine IX,, ( HRP )r e- E-proto- Carbon paste hemine IXred electrode (in most cases ferroprotoporphyrin IX) is oxidized.This oxidized form of peroxidase is usually denoted compound-1 ; H202 + HRP + H20 + compound-1 (1) Compound-1 is then reduced in a first 1 e- electron step to form compound-11; Compound-I + e- + compound-I1 Compound-I1 + e- + HRP (2) (3) followed by a second le- step back to HRP; Virtually any reducing agent is capable of donating electrons to compound-I and compound-11. The necessary protons are donated either by the reducing agent or taken from the surrounding media. Examples of commonly used reducing agents in this context are; phenols, o- and p-hydroquinones, pyrogallol, resorcinol, N, N- alkylanilins, o- and p-phenylene- diamines, iodide, hexacyanoferrate(u), ferrocenes, o-toli- dine, o-toluidine , o-dianisidine, 4-aminoantipyrine, etc, used both in spectrophotometric and electrochemical detection systems to follow peroxidase catalysed reactions.It is important to elucidate the mechanism behind the electron transfer between electrode and attached enzyme, because if known it can hopefully be further optimized. One problem encountered with these peroxidase sensors is the restricted linearity of the calibration curves for hydrogen peroxide. Many peroxidases are rather small redox enzymes with relative molecular masses ranging between 40000 and 100 OOO u.23 The amino acid sequence of one of the isoenzymes of HRP is known.24 The enzyme consists of a hemin prosthetic group, 2Ca2+, a single polypeptide chain, and also 8 neutral carbohydrate chains.From a possible structure of this isoenzyme given by Welinder,24 it is reasonable to believe that the bound cofactor is not situated in the centre but rather at a distance more close to the outer surface of the enzyme molecule. The distance between the bound cofactor in the active site of an immobilized HRP molecule and the electrode could thus be small enough to allow a reaction between the cofactor and the electrode surface. Strong evidence for a direct electron transfer from the electrode to adsorbed peroxidases have been claimed by Yaropolov et al. ,l2 by Paddock and Bowden,13 by Kulys and Schmid15 and by Wollenberger et al. 16 One indication in this direction is the appearance of an electrocatalytic reduction current of hydrogen peroxide close to the I?" value of the reactions in eqns.(2) and (3).25 However, cyclic voltammo- grams of peroxidase modified electrodes do not show any significant voltammetric waves in the absence of hydrogen peroxide that can be correlated to oxidation-reduction reactions with the bound ferriprotoporphyrin IX cofactor. Kulys and Schmid15 found a strong correlation between the open circuit steady-state potential of an electrode modified with fungal peroxidase in a solution containing hydrogen peroxide and the E"' values of eqns. (2) and (3). However, the steady-state potentials obtained in similar experiments repor- ted from this laboratory with adsorbed HRP on various graphites and on glassy carbon could not be directly correlated with the I?" values.19 Our findings were that with thermally pre-treated electrodes (700 "C, 1.5 min) higher catalytic efficiencies could be obtained especially for glassy carbon and some high density ultra-pure graphite.Oxidative electrochem- ical pre-treatments of carbon electrodes at high potentials (== +1.5-2.0 V versus SCE) on the other hand strongly decrease the efficiency of the catalytic effect and the amount of HRP that could be adsorbed on the electrode surface. Electrocat- alytic activity of a deactivated electrode could be regained by the deliberate immobilization of strongly adsorbing mediators containing o-quinone or other quinoid functionalities. 19 A possible explanation might be that the thermal pre- treatment procedure, most efficient for obtaining the catalytic effect, introduces oxygen containing functionalities on the For covalent coupling of enzyme(s) I Fig.1 Possible structure of graphite with functionalities responsible for the electrocatalytic effect in combination with immobilized peroxidases and of functionalities that can be used in conjunction with carbodiimide coupling of the enzyme 1 E-oroto- Fig. 2 Reaction scheme for hydrogen peroxide reduction at a carbon paste electrode chemically modified with HRP electrode surface capable of mediating the electron transfer from the electrode to compounds I and I1 [eqns. (2) and (3)]. By analysing the hydrodynamic voltammograms of electrodes modified with HRP, lactoperoxidase and microperoxidase obtained in the presence of hydrogen peroxide it was revealed that at pH 7 the catalytic current increases abruptly at about +200 and -150 mV.19This could be in accordance with having different o-quinone functionalities created on the electrode surface,26 e.g.as depicted in Fig. 1, expected to have values close to +200 and -150 mV, respectively. The electrodes that revealed the most efficient electrocatalytic current for hydrogen peroxide reduction when modified with HRP were also shown to be the most efficient for electrochem- ical oxidation of ascorbate and NADH.19 (The experiments with ascorbate and NADH were performed in the absence of HRP.) It is well known that o-quinones and electrodes modified with o-quinones catalyse the electrochemical oxida- tion of these compounds.26~27 A similar explanation was given by Staskeviciene et a1.28 for the electron transfer reaction between immobilized lactate dehydrogenase (cytochrome b2) and carbon black.However, no such explanation can be given for the peroxidase-modified electrodes based on Pt or Sn02. 16,20 All initial work on reagentless amperometric sensors with peroxidases were reported for solid electrodes. Peroxidases can also be immobilized in carbon paste electrodes while retaining their electrocatalytic effect for hydrogen peroxide reduction.4Jl Fig. 2 shows the reaction scheme for catalytic reduction of hydrogen peroxide at a carbon paste electrode chemically modified with HRP. We found that a short (15 s) thermal pre-treatment of the graphite powder resulted in a slightly higher catalytic current than when taking the powder as it is delivered by the manufacturer (Fluka, Cat. No.50870). The importance of a close coupling between the graphite and the added peroxidase was shown when the enzyme was added to a pre-made carbon paste material. No catalytic current could be traced for this preparation. If, however, peroxidaseANALYST, AUGUST 1992, VOL. 117 1237 0.5 1 1.5 2 2.5 a 0 0.02 0.04 0.06 [HzOzl/mmol dm-3 Fig. 3 Calibration graphs for hydrogen peroxide of two carbon paste electrodes chemically modified with covalently bound HRP. A, using only carbodiimide; and B, using both carbodiimide and glutaral- dehyde. (a) Shows the entire and (b) the lower part of the concentration range investigated 0.25 - 0.15 - f a 0.05 - -0.05 - 400 -200 0 200 400 600 UlmV ! 800 Fig. 4 Hydrodynamic voltammograms of two equivalent HRP modified carbon paste electrodes obtained for 0.1 mmol dm-3 hydrogen peroxide.A, Indicates a starting potential of +600 mV; and B of -200 mV. The electrode surface is also covered with six layers of an ion exchange membrane (Eastman AQ 29D30) was first allowed to adsorb onto the graphite powder [1400 U (1 U = 16.67 nkat) of HRP taken per 100 mg of graphite powder] prior to the addition of the pasting liquid the electrocatalytic properties were retained.4 It was shown too, that the choice of the pasting liquid affected the response to hydrogen peroxide. Comparing the results obtained for paraffin oil, two silicon oils (Dexsil 400 GC and silicon oil GE SF-96), and silicone DC 710 (50% phenylmethylsilicon oil), the last three being common stationary phases in gas chromatography, the best one was found to be the phenyl- methylsilicon oil (40 mm3 per 100 mg of graphite powder) judged on the basis of a high response to hydrogen peroxide and a low background current.All electrodes simply based on adsorbing HRP lacked long-term stability. By first activating the graphite powder with l-cyclohexyl-3-(2-morpholinoethyl)cardodiimide metho- p-toluenesulfonate [4.2 mg of carbodiimide dissolved in 300 mm3 of 0.05 mol dm-3 acetate buffer (at pH 4.8) per 100 mg of graphite, left to react for 2 h, then rinsed with pure water seven times and dried in a vacuum before the addition of the enzyme dissolved in 200 mm3 of 0.1 mol dm-3 phosphate buffer at pH 7.0, left to react for 16 h, and dried before addition of the pasting liquid] both the electrocatalytic response (>2 times) and the long-term stability were much improved.The functionalities on the surface of the graphite 0.24 0.20 4 A a 0.16 0.12 3 4 5 6 7 8 9 PH Fig. 5 Variation of the response with pH to 0.1 mmol dm-3 hydrogen peroxide at -50 mV versus Ag-AgCI for four equivalent carbon paste electrodes chemically modified with covalently bound HRP. The electrode surface is also covered with six layers of an ion exchange membrane (Eastman AQ 29DW) 9 0.08 3 4 5 6 7 8 PH Fig. 6 Variation of the response to 0.7 mmol dm-3 glucose with the pH of the 0.1 mol dm-3 phosphate buffer of four equivalent electrodes modified with both HRP and GOD. The electrode surface is also covered with six layers of an ion exchange membrane (Eastman AQ 29DW) powder used for coupling to carbodiimide are depicted in Fig.1. Further addition of covalent coupling agents, e.g. glutaral- dehyde, to the reaction mixture (0.2% glutaraldehyde) was also beneficial for the long-term stability of the paste electrodes and also had some additional influence on the magnitude of the response current. Fig. 3 shows the calibration characteristics for hydrogen peroxide of HRP-modified carbon paste electrodes with an applied potential of -50 mV versus Ag-AgCI (0.1 mol dm-3 KCl) when using carbodiimide activated graphite and also when using glutaraldehyde. The electrodes (surface area 0.25 cm2) were mounted in a flow-through amperometric cell of the wall-jet type29 connected to a single line flow injection (FI) system with a flow rate of 0.6 ml min-1 and with an injection volume of 50 mm3.The calibration graphs reveal the restricted linear response characteristics. An apparent Michaelis- Menten constant, KMapp, could be evaluated from electro- chemical Eadie-Hofstee plots and gave values of about 0.8-1.0 mmol dm3, which is in accordance with the linear part of the calibration curve seen in Fig. 3(b). When concentrations of hydrogen peroxide of more than about 2 mmol dm3 were used an irreversible decrease in the sensor response was noticed. The electrodes obtained using both carbodiimide and glutaraldehyde lost about 40% in response to hydrogen peroxide after 24 h. The stability could be much increased by covering the electrode surface with an anion-exchange mem- brane of the Eastman AQ type.4.30 Even though the initial response decreased by about 75% on covering with six layers30 of the membrane, the response of the electrodes was virtually constant after 1 d up to about 3 weeks with analyses carried out every 5 d and with storage at 4 "C between experiments.Fig. 4 shows the hydrodynamic voltammograms obtained between -200 and +600 mV for electrodes covered with the anionic membrane. The hydrodynamic voltammograms show clear increases in the catalytic current at about +200 and -150 mV as was referred to above, irrespective of the startingANALYST, AUGUST 1992, VOL. 117 1238 0.6 0.4 0.2 B I 1 I I I 1 2 3 4 5 6 I I I I I [Substratel/mmol dm-3 0 0.02 0.04 0.06 0.08 0.1 Fig. 7 Calibration curves for A, hydrogen peroxide and B, glucose at pH 5 and at -50 mV versus Ag-AgC1 of a carbon paste electrode chemically modified with covalently bound HRP and GOD.The electrode surface is also covered with six layers of an ion exchange membrane (Eastman AQ 29D30). (a) Shows the entire and (b) the lower part of the concentration range investigated R-CHZOH R-CHO Carbon paste electrode I Fig. 8 lized AOD and HRP in a carbon paste electrode Reaction scheme for an alcohol sensor based on co-immobi- potential (-200 or +600 mV) of the experiment. The current versus pH profile is depicted in Fig. 5 showing the highest response to hydrogen peroxide between pH 4 and 5. Four different sensor types have so far been studied in our laboratory based on co-immobilizing HRP in the paste along with hydrogen peroxide producing oxidases, viz. glucose oxidase (GOD), alcohol oxidase (AOD), D-amino acid oxidase (D-AAOD), and L-amino acid oxidase ( L-AAOD).The ratio between the taken amounts of HRP and GOD was studied for the glucose sensor to find the amounts giving the maximum glucose response. A similar immobilization proce- dure was used as when HRP was bound alone. Carbodiimide activation was followed by addition of either 4.5 mg of HRP (1400 U) and 4.5 mg of GOD (680 U) per 100 mg of graphite or 4.5 mg of HRP (1400 U) and 2.3 mg of GOD (347 U) per 100 mg of graphite. Glutaraldehyde was also added to the reaction mixture just at the moment of mixing with the graphite powder. The maximum response for glucose was obtained with an HRP:GOD ratio of 2 : 1. All glucose electrodes were also covered with the anionic membrane, as above.Fig. 6 shows the response versus pH of four equi- valently prepared electrodes. The pH profile reveals clearly the pH effect of GOD, cf. Fig. 5, showing the maximum activity for glucose at pH 5. Fig. 6 also shows the variation in the reproducibility between electrodes prepared from the same batch of chemically modified carbon paste and after membrane formation. Fig. 7 shows the calibration characteris- tics for glucose and hydrogen peroxide obtained at pH 5. As is 4 5 6 7 8 9 1 0 PH Fig. 9 Variation of the response to 1.0 mmol dm-3 ethanol with pH at -50 mV versus Ag-AgC1 obtained for three equivalent carbon paste electrodes chemically modified with covalently bound HRP and AOD. The electrode surface is also covered with six layers of an ion exchange membrane (Eastman AQ 29D30) clear, the response for glucose is much lower than that for hydrogen peroxide and that the response to both substrates is almost completely non-linear over the whole investigated concentration range. Analysing the KMaw value for HRP when co-immobilized with GOD resulted in lower values than when HRP was immobilized alone.A similar effect was noticed in a previous investigation when HRP and GOD were co-immobilized on a solid graphite electrode22 even though an increase in the value might have been expected owing to a higher diffusion hindrance. No explanation to this behaviour can be given at this time. The non-linear calibration charac- teristics for glucose is expected to be partly due to the low KMapp of HRP. Usually linear response curves for glucose are obtained over at least two orders of magnitude when GOD is immobilized directly onto the surface of carbon electrodes or in carbon paste electrodes.30-32 Alcohol oxidase was also co-immobilized with HRP in a carbon paste to produce an alcohol sensor.Although it is a rather unselective enzyme showing activity not only for methanol and ethanol but also for propanol, isopropanol and butanol with a decreasing turnover rate with an increased length of the carbon chain. It belongs to the group of oxidaseb that have low reported reaction rates with alternative electron acceptors, e.g. ferrocinium derivatives, to molecular oxy- gen,ls which are commonly used as mediators to facilitate the electron transfer between various redox enzymes and elec- trodes.2 It is thus of special importance if the problems encountered with sensor construction, using oxidases that have low reaction rates with alternative electron acceptors, can be solved using co-immobilization of these oxidases with a peroxide and hence use can be made of the apparent direct electron transfer mechanism between the electrode and the immobilized peroxidase.The ratio of AOD and HRP taken was guided by the experiments with co-immobilizing GOD and HRP and was 3 : 1 [4.5 mg (137 U) of AOD and 1.5 mg (394 U) of HRP per 100 mg of graphite]. The same immobilization procedure was followed using carbodiimide activation and addition of gluta- raldehyde to the immobilization reaction mixture. Fig. 8 shows the reaction scheme for the alcohol sensor.Fig. 9 shows the variation of the response to ethanol with the pH of the contacting buffer of three equivalently prepared electrodes. Compared with the responses to hydrogen peroxide or glucose a clear shift into the alkaline region is seen, which reflects the preference of AOD to work optimally in an alkaline environ- ment and a pH of 7.5 was chosen for further experiments. Contrary to the GOD electrode the alcohol sensor responded linearly to ethanol between 0.1 and 1 mmol dm-3 (not shown). However, the co-immobilization had a similar effect on the & ~ P P of HRP as when co-immobilizing it with GOD. Fig. lO(a) line A shows FI peaks for ethanol and hydrogen peroxide and shows much smaller tailing peaks for ethanol than for hydrogen peroxide, cf.below.ANALYST, AUGUST 1992, VOL. 117 1239 "i t ' 10min ' (bl t 10min' Time - Fi 10 FI responses to (H) 50 pmol dm-3 hydrogen peroxide and to yi70.5 mmol dm- e thanol at H 7.5 and at -50 mV versus Ag-AgC1; a ) with carbon paste e1ectrod)es chemical1 modified with covalently bound HRP and AOD (b) and in (a) but wiere PEI was also added to the immobilization reaction (0.12%). A denotes electrodes with no electropolymerized layer on the surface whereas B, C and D denote electrodes subjected to 2,5, and 20 tential scans in a 5 mmol dm-3 o-phenylenediamine containing b u g r , res ctively, between 0 and +650 mV versus SCE (scan rate 50 mV s-lp" a s 0 0.04 0.08 0.12 [Substratel/mmol dm-3 Fig. 11 Calibration curves for: A, hydrogen peroxide; B, methanol; and C, ethanol, at pH 7.5 and at -50 mV versus Ag-AgC1 of a carbon paste electrodes chemically modified with covalently bound HRP and AOD, PEI, also covered with a layer of electropolymerized mixture of 1.5 mmol dm-3 rn-phenylenediamine and 1.5 mmol dm-3 resorcinol obtained after 3 scans (50 mV s-l) between 0 and +650 mV versus SCE and with six layers of an ion exchange membrane (Eastman AQ 29Dm).(a) shows the entire and (b) the lower part of the concentration range investigated In the last few years several papers have reported improved biosensor characteristics when covering the electrode surface with electropolymerized layers.33-38 The surface can be protected against interfering and fouling agents, an increased diffusional barrier will be obtained, which increases the value of an unfavourable KMapp of an immobilized enzyme thereby increasing the linear response range of the substrate, and an increased long-term stability can be obtained for the enzyme immobilized either within the layer or directly onto the electrode surface.The electropolymerization reaction with these reagents starts at relatively low potentials (- +650-700 mV versus SCE) and was reported not to destroy the activity of the enzymes investigated. Different monomers were therefore tried in order to form the electropolymerized layers to protect the electrode surface, increase the long-term stability and improve the sensor characteristics. The mono- mers tried were: aniline, 0.1 mol dm-3 in 0.1 mol dm-3 phosphate buffer at pH 7.0;33 pyrrole, 0.1 mol dm-3 in 1.0 mol dm-3 KC1;N phenol, 0.05 mol dm-3 in 0.1 mol dm-3 phosphate buffer at pH 7.0;35 o-phenylenediamine, 5 mmol dm-3 in 0.1 rnol dm-3 acetate buffer at pH 5.2;36*3' and a mixture of rn-phenylenediamine and resorcinol, 1.5 mmol dm-3 of each in 0.1 rnol dm-3 phosphate buffer at pH 6.5.38 Electrodes were dipped into these solutions and a series of cycled potential scans was allowed to proceed between 0 mV and a higher potential registered to allow the monomer to be oxidized and the polymer to be formed.The membrane formation for all of these reagents resulted in an additional increase in the background current and a decreased response (as expected) but also a much more pronounced tailing of the ethanol peaks was noticed when the membrane covered electrodes were operated at -50 mV, i.e.as sensors in a 50 n A I B I I 10 5 tlmin Fig. 12 FI recordin s at H 7.0 and at -50 mV versus Ag-AgC1 of: A, 0.1 mmol dm-f hy&ogen peroxide; and B, 1 mmol dm-3 L-phenylalanine, with a carbon paste electrode chemically modified by deposition of HRP, L-AAOD and PEI 25 r I 0 500 1000 1500 2000 2500 3000 HzOz/pmol dm-3 Fig. 13 Calibration curve for hydrogen peroxide at pH 6.0 of a bundle of carbon fibres chemically modified by adsorbed HRP1240 ANALYST, AUGUST 1992, VOL. 117 the FI system. Positive results regarding increased stability and linear calibration characteristics were only obtained with o-phenylenediamine and with the mixture of rn-phenylene- diamine and resorcinol. Figs. lO(a) B-D show the effect on the response to 50 pmol dm-3 hydrogen peroxide and 0.5 mmol dm-3 ethanol after covering the alcohol sensor with an electropolymerized layer obtained after 2, 5 and 20 scans between 0 and +650 mV vesus SCE in a solution containing 5 mmol dm-3 o-phenylenediamine. A further stabilization of the electrode could be found when the surface was also covered with layers of an anionic membrane, Eastman AQ 29 D4730 after the electropolymerization reaction.When aniline, phenol or pyrrole was used for making the electropolymerized membrane even lower response currents for ethanol and hydrogen peroxide were registered and also much increased background currents. These reagents might either be detri- mental to the enzymes or destroy the electrode material. In a previous study of immobilizing alcohol dehydrogenase and its necessary cofactor, NAD+, in a mediator modified carbon paste,39 we found that the addition of poly- ethyleneimine (PEI)40 had a very beneficial effect not only on the response to ethanol, but also on the stability and gave a much reduced peak tailing.Fig. lO(6) line A shows the effect on the response when adding PEI (Sigma Cat. No. P3143) to the carbon paste electrode compared with Fig. 10(a) line A. A final concentration of 0.12% PEI was taken for the enzyme immobilization reaction mixture and added to the carbodi- imide activated graphite powder. A much higher response current to ethanol was obtained, along with less tailing of the FI peaks, and a significantly less noticeable decrease in current after forming the electropolymerized layer of poly-o-phenyl- enediamine, Figs.10(b) lines b-d. As in the previous case with immobilized alcohol dehydrogenase, no certain explanation to this behaviour can be given. However, at pH 7.5 many enzymes are expected to have a net negative charge. The PEI is positively charged and the electrostatic interaction between PEI and the enzyme is obviously beneficial for the charge transfer reaction and also for the stability. Fig. 11 shows the calibration curves for hydrogen peroxide, methanol and ethanol for the electrodes with a membrane formed from electropolymerization of rn-phenylenediamine with resorcinol and also covered with six layers of the Eastman AQ membrane. As is clear from this figure, much improved sensor characteristics can be obtained by the introduction of proper reagents into the paste.When D- or L-amino acid oxidases were first co-immobilized with HRP using carbodiimide activated graphite powder and with the addition of glutaraldehyde to the reaction mixture no responses to any amino acid could be traced. The response to hydrogen peroxide was, however, retained, indicating that the reagents covalently binding to the amino acid oxidases caused an inactivation of these enzymes. Experiments with only one covalent binding reagent was also negative. Excluding both reagents and just adding PEI to the enzyme solution before addition to the graphite powder resulted in retained activities for these enzymes in the past electrodes. Previous experiments when immobilizing L- and D-AAOD on controlled-pore glass show that the choice of immobilization reagent can be critical.41 Fig.12 shows the FI response to 0.1 mmol dm-3 hydrogen peroxide and to 1.0 mmol dm-3 L-phenylalanine for a carbon paste electrode containing L-AAOD and HRP co-immobilized with the addition of PEI to the paste. L-Phenylalanine was previously shown to be one of the most active substrates for immobilized L-AAOD.~~ Much of the current work in the analytical field is devoted to the miniaturization of the analytical equipment. One par- ticular area is for microelectrodes with their possible use as sensors in microflow systems.42 A series of different carbon fibres was therefore tested to see whether the catalytic effect for hydrogen peroxide reduction could be obtained with carbon fibre electrodes modified with immobilized peroxi- dases. A small effect was traceable on all different types studied except for graphitized carbon fibres where high catalytic currents were obtained.Fig. 13 shows a calibration curve obtained in a beaker for hydrogen peroxide registered using a bundle of about 2&30 fibres (Polycarbon LGR 10-ply Z-twist) 4 mm in length and modified with HRP (6200 U cm-3), by allowing the enzyme to adsorb for 10 min. These electrodes were as expeeted not stable long-term. Some preliminary experiments show that higher response currents to hydrogen peroxide can be obtained with thermally and/or electrochemically pre-treated electrodes and that the stability can be substantially increased with covalent attachment of the enzyme to the electrode surface. This work was financially supported by the Swedish Natural Research Council (NFR), the Swedish Board for Technical Development (STU) and the Swedish National Energy Administration (STEV).The authors thank Ms. Valentina Kacaniklic for the work with the amino acid oxidases and Mr 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25, 26 27 28 Alexander Rappensberger for drawing the figures. References Updike, S. J., and Hicks, G. P., Nature (London), 1967, 214, 986. Bartlett, P. N., Tebbutt, P., and Whitaker, R. G., Progr. React. Kinet., 1991, 16, 55. Armstrong, F. A,, Hill, H. A. O., and Walton, N. J., Acc. Chem. Res., 1988, 21, 407. Gorton, L., Csoregi, E., Dominguez, E., EmnCus, J., Jonsson- Pettersson, G., Marko-Varga, G., and Persson, B., Anal. Chim.Acta, 1991, 250,203. Kulys, J. J., Biosensors, 1986, 3, 3. Kulys, J. J., Pesliakiene, M. V., and Samalius, A. S., Bioelectrochem. Bioenerg., 1981, 8, 81. Kulys, J. J., Laurinavicius, V.-S. A., Pesliakiene, M. V., and Gureviciene, V. V., Anal. Chim. Acta, 1983, 148, 13. Yao, T., Sato, M., Kobayashi, Y., and Wasa, T., Anal. Chim. Acta, 1984, 165,291. SBnchez, P. D., Blanco, P. T., Alvarez, J. M. F., Smyth, M. R., and O’Kennedy, R., Electroanalysis, 1990, 2,303. Wang, J., Wu, L.-H., and Angnes, L., Anal. Chem., 1991,63, 2993. Frew, J. E., Harmer, M. A., Hill, H. A. O., and Libor, S. I., J. Electroanal. Chem., 1986, 201, 1. Yaropolov, A. I., Malovik, V., Varfolomeev, S. D., and Berezin, I. V., Dokl. Akad. Nauk SSSR, 1979,249, 1399. Paddock, R. M., and Bowden, E. F., J.Electroanal. Chem., 1989,260, 487. Jonsson, G., and Gorton, L., Electroanalysis, 1989, 1,465. Kulys, J., and Schmid, R. D., Bioelectrochem. Bioenerg.? 1990, 24, 305. Wollenberger, U., Bogdanovskaya, V., Bobrin, S., Scheller, F., and Tarasevich, M., Anal. Lett., 1990, 23, 1795. Gorton, L., Bardheim, M., Bremle, G., Csoregi, E., Persson, B., and Pettersson, G., in Flow Injection Analysis (HA)-Based on Enzymes or Antibodies, ed. Schmid, R. D., Gesellschaft fur Biotechnologische Forschung (GBF), Monographs, VCH, Weinheim, vol. 14, 1991, p. 305. Gorton, L., Bremle, G., Csoregi, E., Persson, B., and Jonsson-Pettersson, G., Anal. Chim. Acta, 1991, 249, 43. Csoregi, E., Jonsson-Pettersson, G., and L. Gorton, J. Biotech- nol., submitted for publication. Tatsuma, T., and Watanabe, T., Anal. Chem., 1991,63, 1580. Wollenberger, U., Wang, J., Ozsoz, M., Gonzalez-Romero, E., and Scheller, F., Bioelectrochem. Bioenerg., 1991,26,287. Jonsson-Pettersson, G., Electroanalysis, 1991, 3, 741. Dundford, H. B., and Stillman, J. S., Coord. Chem. Rev., 1976, 19, 187. Welinder, K. G., Eur. J. Biochem., 1979, 96, 483. Hayashi, Y., and Yamazaki, I., J. Biol. Chem., 1979,254,9101. Scheurs, J., van den Berg, J., Wonders, A., and Barendrecht, E., Recl. Trav. Chim. Pays-Bas, 1984, 103,251. Tse, D. C.-S., and Kuwana, T., Anal. Chem., 1978, 50, 1315. Staskeviciene, S. L., Cenas, N. K., and Kulys, J. J., Anal. Chim. Acta, 1991,243, 167.ANALYST, AUGUST 1992, VOL. 117 1241 29 30 31 32 33 34 35 36 37 Appelqvist, R., Marko-Varga, G., Gorton, L., Torstensson, A., and Johansson, G., Anal. Chim. Acta, 1985, 169,237. Gorton, L., Karan, H. I., Hale, P. D., Inagaki, T., Okamoto, Y., and Skotheim, T. A., Anal. Chim. Acta, 1990,228,23. Gorton, L., Scheller, F., and Johansson, G., Studia Biophys., 1985, 109, 199. 40 Jonsson, G., and Gorton, L., Anal. Lett., 1987, 20,839. Shinohara, H., Chiba, T., and Aizawa, M., Sens. Actuators, 41 1988, 13, 79. Aizawa, M., Anal. Chim. Acta, 1991, 250, 249. 42 Bartlett, P. N., and Whitaker, R. G., Biosensors, 1987/88, 3, 359. Malitesta, C., Palmisano, F., Torsi, L., and Zambonin, P. G., Anal. Chem., 1990,62,2735. Sasso, S . V., Pierce, R. J., Walla, R., and Yacynych, A. M., Anal. Chem., 1990, 62, 1111. 38 39 Geise, R. J., Adams, J. M., Barone, N. J., and Yacynych, A. M., Biosens. Bioelectron., 1991, 6, 151. Dominguez, E., Lan, H. L., Okamoto, Y., Hale, P. D., Skotheim, T. A., Hahn-Hagerdal, B., and Gorton, L., Bio- probe, in the press. Bahulekar, R., Ayyangar, N. R., and Ponrathnam, S., Enzyme Microb. Technol., 1991, 13, 858. Dominguez, E., Marko-Varga, G., Carlsson, M., and Gorton, L., J. Pharm. Biomed. Anal., 1990,8,825. Electroanalysis, 1990, 2, Special Issue (Microelectrodes). Paper 21013270 Received March 12, 1992 Accepted April 24, 1992