J. CHEM. SOC. FARADAY TRANS., 1994, 90(11), 1495-1499 Electroanalytical/X-Ray Photoelectron Spectroscopy Investigation on Glucose Oxidase adsorbed on Platinum G. E. De Benedetto, C. Malitesta" and C. G. Zambonin Laboratorio di Chimica Analitica, Dipartimento di Chimica, Via Orabona, 4, 70726 Bari, Italy Glucose oxidase has been adsorbed on Pt from a solution of the enzyme (0.5-15 pmol dm-3) in phosphate or acetate buffer (pH 3.9-8.0). The amperometric response to glucose of the modified electrodes so prepared has been evaluated. A spectroscopic (XPS) investigation of the system surface indicates that the enzyme molecules are stacked on the Pt surface, some directly interacting with the metal and the remaining part adsorbed on them. At the surface, the enzyme has a conformation in which its carbohydrate part tends to cover the proteic portion.Biosensors represent a major field of investigation, owing to their widespread potential applications. Among them, immo- bilized enzyme biosensors represent an important class. The immobilization of enzymes at electrode surfaces, however, is still a critical step in amperometric biosensor assembling. Such an operation has classically been performed by methods which produce unstable devices (e.g. adsorption) or which consist of complex procedures (e.g. chemical binding). Elec- trochemical immobilization in an electrosynthesized polymer matrix has recently emerged as a strategy to overcome both of these drawbacks in a one-step operation. In addition, a proper choice of entrapping polymers1-6 can produce, at the same time, a significant improvement in interferent rejection, which is classically obtained by use of additional membranes.Adsorption of the enzyme at the electrode surface before the electrochemical polymerization has an important role' in determining the enzyme loading and so the final response of the sensor. So, a study of the adsorbed biomolecule at the surface could be of paramount importance in designing these kinds of biosensors. In this respect, X-ray photoelectron spec- troscopy (XPS), which is a surface technique able to give chemical information, appears to offer interesting pos-sibilities. In fact, amino acids and proteins have already been investigated'.' by this approach, as well as the adsorption"-12 of proteins on a solid surface.In spite of this, only two13*14 attempts, to our knowledge, have been made in using XPS for investigating immobilized enzyme biosensors. In particular, both works investigated only the electrode sur- faces at which the enzyme had to be bound, giving informa- tion on chemical surface modifications following some pretreatments, a base from which to attempt the interpreta- tion of some properties of the electrode+xzyme system sub- sequently assembled. A glucose electrode based on glucose oxidase (GOx) bound at a Pt electrode surface is, by far, the most studied biosensor. Nonetheless, the adsorption of GOx on Pt is only classified as irreversible and no spectroscopic investigation has been undertaken.Consequently, an XPS study has been carried out in this laboratory and relevant results are reported. An electrochemical characterization of active enzyme on the surface was also performed and the results critically com- pared with the spectroscopic data. Experimental Chemicals GOx (VII type from Aspergillus niger) and 8-D-glucose were obtained from Sigma. 0.5 mol dm-3 glucose solution (freshly prepared in phosphate buffer, pH 7.0, every week) was allowed to mutarotate at 4°C overnight before use. Triply distilled water was used in all experiments. All of the other chemicals were of analytical grade and were used without further purification. GOx solutions were prepared by dilution of 15 pmol dm -stock solutions in the selected buffers.Apparatus All electrochemical experiments were carried out using a PAR 174A polarographic analyser (EG&G Princeton Applied Research) coupled to a Hewlett Packard model 1070 XYt recorder (HP, Palo Alto, CA). The conventional three- electrode cell contained an Ag I AgCl reference electrode and a Pt foil counter electrode. When necessary the solutions were magnetically stirred. XPS was performed using a Leybold LSHlO spectrometer. The spectrometer energy scale was calibrated for Cu 2p,,, = 932.6 eV and Au 4f,,, = 84.0 eV. Preparation of Samples with Adsorbed Enzyme Unless otherwise specified, GOx was adsorbed on a Pt(Au) disc (1 mm diameter) sealed in glass. The electrode surface was polished with emery paper and alumina powder (0.3 pm), then sonicated.Finally, the electrode (Pt) was etched for a few minutes by hot HNO, or (for Au) cycled15 in phosphate buffer (pH 6.0) between -0.9 and 1.0 V us. Ag IAgCl. Enzyme adsorption was performed, immediately before use, in a selec-ted buffer containing GOx (0.5-15.0 pmol dm-3, typical con- centrations in biosensor preparation). The enzyme electrode was then thoroughly washed with triply distilled water for removing the loosely bound enzyme. XPS samples were simi- larly prepared by employing Pt sheets (20 mm x 10 mm) which were fully immersed in the relevant solution. The excess water was eliminated by a flow of nitrogen. Drying was then completed in the spectrometer vacuum. Response to Glucose of the Adsorbed Enzyme The enzymatic activity present on each enzyme electrode was estimated amperometrically in selected buffers : acetate (I = 0.1, pH = 3.9, 5.2) and phosphate (I = 0.1, pH = 6.0, 6.5, 7.3, 8.0).For this purpose, the Pt I GOx electrode was main- tained at +0.7 V us. Ag I AgCl in a stirred buffer up to a low, nearly constant current value. A known amount of glucose solution was then added and the i us. t curve for the oxida- tion, at a given potential, of the H20,produced according to the known enzymatic process glucose + 0, -+ gluconolactone + H202 (1) was recorded. As each experiment was carried out on a new (because of the cleaning procedure) electrode surface, a nor- malization procedure for the glucose response was designed, consisting of the addition of a known amount of H,O, after glucose injection.The resulting anodic current increment was recorded. Results are presented as 'normalized glucose response' [i,(glucose)/i,(H20,)]. In this manner the standard deviation of the response improves by about 40%.The nature of the buffer does not influence the response, as verified by performing measurements at a selected pH (5.63),realized by either acetate or phosphate buffer. XPS Experiments The first few experiments were performed by cooling samples to 100 K in order to avoid degradation. In the following we observed that some degradation occurs (see Appendix 1) only if traces of the buffers are left on the samples. So, samples described in the paper have been thoroughly washed with water before XPS analysis at room temperature.For all samples wide-scan (FRR mode) and high-resolution 0 Is, Pt 4f, N Is, C 1s (FAT mode) spectra were acquired. Analysis of the spectra were performed as reported.16 The spectrum energy scale was corrected by setting the hydrocarbon con- taminant, C 1s = 284.8 eV. The occurrence of a local charg- ing effect different from site to site was excluded on the basis of the shape of the N 1s signal which was always nearly sym- metric and had the proper full width at half maximum (FWHM). Results and Discussion Following immersion of Pt in GOx solution a rapid adsorption l7 occurs. Preliminary results from EQCM (electrochemical quartz crystal microbalance) measure-ments" have been obtained recently. They show qualitatively that adsorption consists of a first rapid step followed by a further slower one.The system produced in this laboratory by enzyme adsorp- tion has been electrochemically and spectroscopically charac- terized. All of the reported results are relevant to a 15 pmol dm-enzyme solution for enzyme electrode preparation. The data obtained by employing different enzyme concentrations, in the range 0.5-15 pmol dm-3, were qualitatively similar. Electrochemical Results A typical response to glucose of the enzyme adsorbed on platinum is shown in Fig. 1. The response time to glucose is shorter than that measured for other devices with the enzyme differently immobilised (see e.g.ref. 19). This is probably due to the absence of any kind of membrane or film that could slow down the response. The effect of the pH of the buffers in which the measurements were carried out on the activity of the adsorbed enzyme was studied, measuring the response of the sensor to a 5 mmol dm-3 glucose solution. The plot of normalized response us. pH, shown in Fig. 2 for two different times of adsorption, appears interesting: it seems to exhibit two maxima, which are not very marked, particularly at short times of adsorption, but which are reproducible. In contrast, only one peak is shown by the activity of GOx in solution2' or when immobilized2' on a solid such as a porous glass. Tentatively, since the enzyme in solution has its maximum of activity at around pH 5.5, it seems reasonable to attribute the low pH curve maximum to enzyme molecules slightly modi- fied by adsorption. The second component, with a maximum around pH 7, could be due to the enzyme directly and J.CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 j*O"A 10 s c c g 30 it glucose time Fig. 1 Typical current-time response to glucose and to H,O, for a Pt I GOx electrode obtained by 10 min adsorption (15 pmol dm-3 GOx solution in phosphate buffer, pH = 7.3, I = 0.1). E = +0.7 V vs. Ag I AgC1. open-circuit potential ( + 0.8 V us. Ag I AgCl), the Pt electrode is likely to be covered by negative species (electrode anions, or enzyme molecules if pH > 4.2 which is the enzyme isoelectric" point) and a shift of the maximum activity region to higher pH is expected under these conditions.22 Two find- ings seem to be in agreement with the above view: (1) an increase of the low-pH peak (larger than that of the high-pH peak) at longer adsorption times, and (2) a shift of the high-pH peak towards higher pH at longer adsorption times (larger amount of negatively charged molecules on the electrode).A similar model was proposed by Wilson and co- worker~,~~who postulated for the adsorption of GOx on a different electrode material, that first a layer of directly bound molecules builds up, on which other enzyme layers can later be adsorbed. For the sake of comparison with the immobilized enzyme biosensors, the stability of Pt I GOx was studied by measuring the response to glucose at different times from its prep- aration.The electrodes when not in use were stored in acetate buffer. The response, nearly constant in the first 2 h, shows a decrease of up to 50% (Fig. 3) after 5 h. Later, the anodic n~ I I 0 3 4 5 6 7 8 1 I PH Fig. 2 pH-activity profile of Pt I GOx electrodes prepared for differ- ent adsorption times: 10 s(.)and 10 min (@). Each point represents strongly adsorbed onto the electrode surface. In effect, at the mean value on at least seven replicates performed over 15 days. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 1401 0 0 0:'i 0 0 20- current measured upon adding the glucose was stable up to about 100 h after preparation. This is consistent with the relatively rapid loss of the weakly bound enzyme species which are responsible for the low-pH peak in Fig.2. A comparison between the active enzyme loadings on Pt and Au was also performed by means of normalized anodic response to glucose. The results for Au were 40% higher than those for Pt. This suggests that the enzyme adsorbed on the gold electrode is either in greater abundance or simply more active than that adsorbed on platinum. On the other hand, the absolute response to glucose (and to H202) for an Au electrode was definitely lower than for a Pt electrode of the same area. This fact indicates that many fewer centres active for H202oxidation are present on the gold surface. Finally, a comparison can be made between the present sensor and the one-step amperometric sensor based' on GOx immobilized in an electropolymerized PPD film.The anodic current measured with the former was 65-70% of that obtained with the latter device. The importance of adsorption in the building of sensors such as the Pt I PPD IGOx sensor is confirmed. The remaining 30-35% of the current is likely to be due to entrapping of additional enzyme in the membrane, even if it is only 100 A thick. Spectroscopic Investigations Fig. 4 shows a widescan spectrum for a sample of GOx adsorbed on Pt. The presence of an N Is signal, coupled to intense 0 1s and C 1s peaks, suggests that, even after the 1497 Table 1 Binding energies of XPS signals recorded on GOx adsorbed on Pt from a 15 pmol dm-3 solution of GOx in phosphate buffer (pH = 6.0, I = 0.1);possible errors up to 0.3 eV signal E,IeV Pt 4f,,, 70.7 0 1s 531.4 N 1s 399.7 284.8 286.2 288.4 washing procedure (see Experimental), a certain amount of enzyme remains strongly adsorbed on the Pt surface.At the same time, the presence of the Pt 4f signal in the spectrum indicates that the enzyme layer is either discontinuous or, if continuous, is formed according to the smallest dimension of the enzyme molecule (vide infra). As far as the high-resolution spectra are concerned, only the C 1s signal showed an evident multicomponent profile and was studied in detail. The binding energies (EJ of all photoelectronic signals are reported in Table 1.N and 0 signals showed E, values in good agreement with those already reported8Pi2 for amino acids, peptides and proteins as powders and adsorbates. Fig. 5 reports a typical C 1s spectrum for a sample pre- pared by 10 s adsorption. Three components are present: C-H, C-0 and C=O (see Table 1). C-H represents the hydrocarbon component. The second contribution due to the C-0 component belongs to the carbohydrate chain and to the protein present in a pr~bable~~-~~ ratio of 78 :22. The C=O component is due to contributions of the peptidic (ca. 90%)and the carbohydrate (ca. 10%)chains. On the basis of the relevant signal areas obtained with 10 s samples, atomic ratios of C(C-0) : N = 0.93(k0.08):1 and C(C-0) : C(C-0) = 1.3q40.13): 1 were calculated.While the former agrees with its theoretical value24-26 (1.0 :1) the latter is very24-26 different (0.6 : 1). Theoretical values were calculated by considering a homogeneous distribution of the functional groups in the GOx molecule. Actually, a large part of the carbohydrate moiety is deeply buried in the polypep- tide chain,25 so that the theoretical value for the C(C-0) : C(C-0) ratio is likely to be lower. The discrep- ancy relevant to C(C-0) :C(C-0) could indicate that in the conformation of the adsorbed GOx the exterior part (which gives the largest contribution to the XPS signal) con- sists of the carbohydrate chain. In other words, this chain should cover the protein portion which is obviously the nearest part to the Pt surface and is likely to be involved in the bonding responsible for adsorption.The agreement rele- vant to C(C-0) :N is not surprising considering that the C-0 present in GOx belongs to amidic groups for about 500 1000 1500 kinetic energy/eV 1 I I 1 I Fig. 4 Typical overall spectrum for GOx adsorbed on a Pt foil (10 s 290 288 286 284 adsorption in 15 pmol dmP3 GOx solution in acetate buffer kinetic energy/eV pH = 5.2, I = 0.1). The sample was washed with water after the adsorption step. Fig. 5 High-resolution C 1s XP spectrum for the sample of Fig. 4 Table 2 N : Pt ratios for the Pt IGOx,,, sample adsorption time N :Pt coverage 10 s 0.39 :1 0.82 10 min 0.67 :1 0.88 The 10 s sample was obtained by 10 s adsorption from a 15 pmol dm-3 GOx solution in phosphate buffer (pH = 6.0, I = 0.1).After XPS measurements it was reimmersed in the enzyme solution for a total of 10 min, thus obtaining the 10 rnin sample. 83%; in any case, the proximity of the two species guarantees that in the most part C-0 and N are sampled simulta- neously, producing equivalent signals. Qualitatively similar results were obtained with samples prepared by a 10 rnin immersion in the enzyme solution. Information on the coverage and arrangement of ellipsoidal' enzyme molecules on the platinum surface was inferred from N :Pt ratios: some of them are collected in Table 2. These results are the product of an experiment designed to investigate accurately the correlation between enzyme loading and adsorption time.A sample was prepared by 10 s adsorption, analysed and then again exposed to enzyme solution for a total period of 10 min. The sample was then reanalysed. Similar results were obtained for 10 s and 10 min samples prepared in different experiments. N: Pt ratio can be correlated to the coverage, 0 (see Appendix 2), employing the approach suggested by Dilks.,' To this end, the enzyme was first considered to be distributed in one monolayer, consisting of molecules oriented perpen- dicular or parallel to the surfa~e.'~ More complex arrange- ments were not considered from a quantitative point of view. The N :Pt ratio for a full coverage of the electrode surface by horizontally oriented molecules is 0.26 :1 (the maximum theoretical value), well below any measured ratio (see e.g. Table 2).On this basis, it was discarded. In contrast, when molecules are in a standing position N :Pt can span from 0 to co.Table 2 also reports the coverages resulting from the latter model and calculated from the relevant N :Pt ratios. The devised simple picture cannot explain the higher current response (+ 30%) found (vide ante) for 10min in com- parison to 10 s adsorption: in fact, the coverage correspond- ing to the relevant N :Pt values increases by only 6% (Table 2). An increase of the activity of the same amount of enzyme for longer times of contact with Pt has to be excluded because the opposite effect is often rep~rted.'~,~~ A compari-son between electrochemical and XPS data is not always pos- sible since, in general, modifications could occur in the sample under XPS measurements (e.g.drying). Nonetheless, in this case the comparison is meaningful. Since, N is homogeneously distributed in the enzyme molecule, the thickness (t) of the monolayer is the main factor influencing the N :Pt ratio. t could, at most, be reduced (for example in the drying step). This would produce a limiting value for R, (see Appendix 2) less than 0.26, and cause the 8, values in Table 2 to become even closer, confirming the above observa- tions. It was therefore necessary to adjust the model since the enzyme molecules were partly vertically and partly horizon- tally oriented for both adsorption times. A possible explana- tion could be that the first molecules are adsorbed in a horizontal position (maximising the contact surface area) and then some of them stand up, leaving surface available as further molecules approach the electrode.In this manner, a much greater number of molecules can be accommodated on the surface even if the coverage changes slightly. An alterna- tive picture is possible, i.e. enzyme molecules arranged as multilayer islands over the electrode surface. These islands could perhaps consist of a layer of horizontally oriented mol- J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 ecules directly adsorbed onto the electrode over which other enzyme layers may be adsorbed. The fact that GOx prefers to grow in multilayers instead of exhaustively covering the Pt surface, could simply be the result of a pretreatment method that is unable to distribute uniformly active sites on the surface.In this case, the uncovered Pt should simply form the part free of active sites. The island model appears to be more consistent with the electrochemical data and agreed7 with some literature reports. This view could also explain the apparent lack of any change in the measured Ebvalues fol- lowing interactions (adsorption) between Pt and GOx. A definitive choice between these models requires further inves- tigation and a more suitable method of sample preparation able to produce monolayer fraction coverage, In particular, a study by atomic force microscopy (AFM) is planned for the near future.Appendix 1 The reader interested in performing XPS experiments of the kind described in the present work must be advised that it is mandatory to wash out thoroughly any trace of the buffer employed in the adsorption step before executing the spectro- scopic analysis. We found that in the presence of acetate or phosphate, C 1s spectra exhibited a fourth component at higher binding energy (5.2-6.2 eV higher than the hydrocar- bon peak, often detectable from the beginning of the experiment), probably attributable to degradation products. While this effect is particularly evident (Fig. Al) when the acetate buffer was used, the component was present even with phosphate. This constant presence seems to indicate that the above-mentioned component at high is also related to an enzyme degradation.Nonetheless, the absence of the effect in thoroughly washed samples suggests that the buffer has a role in the degradation phenomena. On the basis of the E, value, the possibility that the component is due to either carbon monoxide or carbon dioxide adsorbed over Pt seems to be excluded.28 These C species, at present unknown, seem to be organic carbonates. 29*3 Appendix 2 In obtaining the relationship between coverage and N :Pt ratio for a GOx monolayer two different arrangement^'^ of I I I I II 293 291 289 287 285 Fig. A1 C 1s spectrum for a sample of GOx adsorbed on a Pt foil (10 s adsorption in 15 pmol dm-3 GOx solution in acetate buffer, pH = 5.2, I = 0.1). The sample was washed with the buffer after the adsorption step.The carbonyl component contains in this case the signal due to acetate as well. The selected spectrum is relative to an experiment in which the degradation component (X) was particularly evident. J. CHEM. SOC. FARADAY TRANS., 1994, VOL. 90 Fig. A2 Arrangements of the GOx molecule on the Pt surface: (a) parallel, (b) perpendicular to the surface the ellipsoidal enzyme molecule on the Pt surface were con- sidered: (a) parallel and (b) perpendicular to the Pt surface (Fig. A2). Calculations were performed employing the approach sug- gested by Dilks.,’ For this purpose, the shape of the mol- ecule was approximated to a cylinder (base, ellipse with axes 140 and 50 A; height, 50 A for the horizontal molecule; base, circle of radius 25 A; height, 140 8, for the vertical molecule).The intensity of the N 1s signal for a homogeneous N dis- tribution and the direction for electron collection normal to the surface can be expressedz7 as: 1, = FkN UN AN 8S[i -eXp(-t/&)]cN (All where F is the X-ray flux, k, is an instrumental parameter also depending on the kinetic energy of the photoelectrons, 0, is the cross-section for the photoemission, iN is the inelas- tic mean free path for N 1s photoelectrons, S is the sampled surface, 8 is the coverage (i.e. the fraction of S covered by the enzyme), t is the thickness of the ‘N’ layer and C, is its N concentration. The term in parentheses spans from 0 (t = 0) to 1 (t = a).The last condition is verified in practice when t 2 31.1 values in the enzyme for N 1s (32.1 A) and Pt 4f (37.8 A)electrons (excited by Mg-Kcr) were calculated by the equation reported3 for organic materials taking the density of the enzyme as ca. 1 g cmd3 [estimated from molar volume (cylindrical shape) and weight (186 OOO g mol-‘)I.The intensity of the Pt 4f signal contains two contribu- tions: Ipt= Fk, optAbt 6Scptexp(-t/Apt) + Fkpt gpt Ah(1 -6)SCpt (A2) The first term of eqn. (A2) represents the Pt lying under the enzyme and whose signal is revealed after attenuation through the enzyme layer. The second represents the contri- bution of uncovered Pt. The apex in Abt means that the value is relevant to the metallic platinum medium. From2’ kPt Opt &dkN gN AN = SPt/SN (A31 where si is the relative sensitivity factor of the element ‘i’, it is possible to define the quantity, R : = (lN/SN)/(lPt/SPt) (A41 which can be easily correlated to 8.The relevant equations for the two arrangements of the enzyme molecules [(a) and (b)in Fig. A2, respectively] are: 6, = RJ0.069 + 0.73Rl) (A5) 6,+O*R,-+O (A5’) 6, + 1 R, + 0.26 (A57 6, = Rz/(0.087 + R,) (A61 8, + O= R, + 0 (A67 6, + 1 * R, -+ 00 (A6) The reason for a finite limit in the case of horizontal mol- ecules can be easily rationalised if one considers that even at full coverage Pt 4f electrons can be detected after travelling through the enzyme layer since 3RPt > 50 A. Francesco Palmisano (University of Bari) and Robert Hill- mann (University of Leicester) are gratefully acknowledged for helpful discussions.The work was carried out with the financial support of Minister0 dell’universita e della Ricerca Scientifica (MURST) and Consiglio Nazionale delle Ricerche (CNR). References 1 C. Malitesta, F. Palmisano, L. Torsi and P. G. Zambonin, Anal. Chem., 1990,62,2735. 2 D. Centonze, A. Guerrieri, C. Malitesta, F. Palmisano and P. G. Zambonin, Fresenius J. Anal. Chem., 1992,342,729. 3 D. Centonze, A. Guerrieri, C. Malitesta, F. Palmisano and P. G. Zambonin, Ann. Chim. (Rome), 1992,82,219. 4 D. Centonze, C. Malitesta, F. Palmisano and P. G. Zambonin, Electroanalysis, 1994, in the press. 5 F. Palmisano and P. G. Zambonin, Anal.Chem., 1993,65,36. 6 F. Palmisano, D. Centonze, A. Guerrieri and P. G. Zambonin, Biosensors, 1993,8, 393. 7 P. N. Bartlett and R. G. Whitaker, J. Electroanal. Chem., 1987, 224,37. 8 D. T. Clark, J. Peeling and L. Colling, Biochem. Biophys. Acta, 1976,453,533. 9 K. D. Bomben and S. B. Dev, Anal. Chem., 1988,60, 1393. 10 B. D. Ratner, T. A. Horbett, H. R. Thomas and D. Shuttleworth, J. Colloid Interface Sci., 1981,83, 630. 11 B. D. Ratner, T. A. Horbett, H. R. Thomas and R. W. Paynter, J. Colloid Interface Sci., 1984, 101, 233. 12 H. Fitzpatrick, P. F. Luckham, S. Eriksen and K. Hammond, J. Colloid Interface Sci., 1992, 149, 1. 13 A. Proctor, J. F. Castner, L. B. Wingard Jr. and D. M. Hercules, Anal. Chem., 1985,57, 1644. 14 F. A. Armstrong, P.A. Cox, H. A. 0. Hill and V. J. Lowe, J. Electroanal. Chem., 1987,217, 331. 15 A. Szucs, G. D. Hitchens, and J. O’M. Bockris, J. Electrochem. SOC.,1989,136,3748. 16 C. Malitesta, L. Sabbatini, P. G. Zambonin, L. Peraldo Bicelli and S. Mafi, J. Chem. SOC.,Faruday Trans. I, 1989,85, 1685. 17 Chem. Abstr., 1990,112, no. 51261. 18 A. R. Hillman and M. J. Swann, personal communication. 19 H. Gunasingham and C. B. Tan, Analyst (London), 1989, 114, 695. 20 D. Keilin and E. F. Hartree, Biochemistry, 1948,42, 221. 21 K. B. Ramachandran and D. D. Perlmutter, Biotechnol. Bioeng., 1976, 18,669. 22 L. Goldstein, in Methods in Enzymology, ed. K. Mosbach, Aca- demic Press, London, 1976, vol. XLIV, ch. 29. 23 B. S. Hill, C. A. Scolari and G. S. Wilson, J. Electroanal. Chem., 1988,252,125. 24 J. H. Pazur, K. Kleppe and A. Cepure, Arch. Biochem. Biophys., 1965,111,351. 25 S. Nakamura and S. Hayashi, FEBS Lett., 1974,41,327. 26 H. J. Hecht, H. M. Kalisz, J. Hendle, R. D. Schmid and D. Scomburg, J. Mol. Biol., 1993,229, 153. 27 A. Dilks, in Electron Spectroscopy, ed. C. R. Brundle and A. D. Baker, Academic Press, London, 1981, vol. 4, ch. 5. 28 P. R. Norton, Surf: Sci., 1474,44, 624. 29 U. Gelius, P. F. Heden, J. Hedman, B. J. Lindberg, R. Manne, R. Nordberg, C. Nordling and K. Siegbahn, Phys. Scr., 1970,2,70. 30 G. Beamson and D. Briggs, High Resolution XPS of Organic Polymers, Wiley, Chichester, 1992, App. 1. 31 M. P. Seah, in Practical Surface Analysis, ed. D. Briggs and M. P. Seah, Wiley, Chichester, 2nd edn., 1990, vol. 1, p. 209. Paper 4/00555D; Received 28th January, 1994