摘要:

J . G e m . Soc. Faraday Trans. 1 1986,82 1161-1 167 New Host-Guest Carriers - feedback + molecular design J. C. Lockhart Department of Inorganic Chemistry The University Newcastle upon Tyne NEl 7RU The design of ligands suitable as ionophores for ion-selective electrodes (i.s.e.) is discussed in terms of the carrier hypothesis emphasising the balance needed between maximum selectivity and a fast response time. A new type of crown ether is described which incorporates a triarylmethyl (propeller) substituent which can be fluxional or rigid. Theimproved kinetic performance of crowns with rigid propellers suggests further possible exploitation in future designs of ionophore. Our approach to the production of ionophores (L) for ion-selective electrode work with alkali-metal cations (M+) in particular is based on the carrier hypothesis and is shown in scheme 1 .1 complexation feedback 4 kinetic - - and mechanistic studies ligand synthesis feedback feedback i.s.e. passive transport studies Scheme 1. We design a ligand (L) on paper or using molecular graphics synthesise and evaluate getting feedback at each stage for a better design. We have paid particular attention to the complexation process. It is important to optimise the kinetics of the process to provide selectivity together with a fast response time. We are trying to find what features of the macrocycle affect the kinetics so that we can improve them - a glib term for this might be ‘kinetic engineering’. Selectivity and the Complexation Process It is convenient to look at the complexation process in terms of eqn (1) for our arguments The equilibrium constant ( K ) for the formation reaction is given by the ratio k,/kd.It has been established for many ionophore/cation combinations that the rate of complexa- tion shows very little variation and is within a factor of ca. 100 of the diffusion- controlled rate,l although there are exceptions particularly for rigid ligands where the formation rate is s10wer.~ Nevertheless for one ligand and a range of alkali-metal cations the formation rate constant varies little between metals. If this is general eqn (2) may 39-2 1161 1162 New Host-Guest Carriers solvents. Fig. cryptands; Data 0 at 25"C taken with permission from ref.(2). A MeOH cryptands; cations 1. Plot of log DMF k us. -5.0 cryptands; log 0 I K for 2.0 I a . range H,O 4.0 I of crowns; crowns log 6.0 I K and V 8.0 I DME cryptands 10.0 I crowns; 12.0 with I 0 alkali EtOH cryptands. in a several PC A plot of log k versus log K is found to be a straight line of unit slope for many macrocycles as first shown by Cox and Schneider,l with the intercept on the log k axis giving the (common) formation rate constant (see fig. 1). Incorporating some of Cram's data on spherands3 (which complex more slowly) a similar straight line of different intercept results. The discrimination of any one ligand between one cation and another in a single phase thus depends on the relative rates of dissociation of their respective complexes.1.s.e. containing macrocycles can show discrimination between alkali-metal cations so this is unlikely to arise from the complexation process but from the reverse dissociative step. Although particular attention has been paid by others3 to maximising the equilibrium constant for complexation it is apparent that the dissociation rate constant is a critical factor. There will certainly be kinetic limitations for i.s.e. and other applications which require a fast response time. There must be a lower limit to the viable dissociation rate constant k, below which dissociation will not be fast enough to permit i.s.e. response. Suppose for argument that the half-life for a bound cation prior to dissociation should be 0.1-0.01 s this would correspond to a log k of ca.1-2 and log K of 6-7 at most. In three-phase passive transport measurements it has actually been shown4. that the transport rate has a maximum for systems with log K of ca. 5-8. This maximum probably depends on the dissociation rate rather than K since a number of exceptions are known where systems with the same log K transport at very different rates. Returning to i.s.e. (since the performance may be limited by the diffusion rate for complexation on the one hand and electrode response time on the other) the maximum A log K between two ions will be ca. 6f 1. This is probably the maximum selectivity which can be expected for a carrier-type of i.s.e.depending solely on the macrocycle for discrimination. Electrochemists working with the complete system may be able to get round this problem. Following this interpretation we needed to know how to manipulate the kinetics of the macrocycle complexation to achieve maximum selectivity. It has been noted by J. C. Lockhart 1163 Sutherlands that rigidity is an important feature for a macrocycle which can improve selectivity by a CaVity-SiZe effect. Simple CrOwnS are fluxional and not selective while the very rigid spherands are very selective but would provide very slow response times We sought a macrocycle type where we could add rigidity and also one to which we could attach groups suitable for n.m.r. studies SO that we could determine rates and mechanisms and how these were affected by the rigidity which would help us in the design of better electroactive materials for i.s.e.We accordingly designed the propeller ethers7 of general formula 1. 2 X = o - O M e Y = H n = l 3 X = m m' p-(OMeh Y = H n = 2 5 X = m rn' p-(OMe) Y = But n = 3 1 6 X = 0 0'-C12 Y = But n = 3 4 A = D H Background of the Three-bladed Propeller Structural determinations on several thousand molecules with three aryl groups attached to a central atom have not disclosed any which are not propeller in form.8 The name derives from the aryl rings regarded as blades all having the same pitch radiating from a central hub and rendering the molecule ~ h i r a l . ~ Many of these propeller molecules are fluxional in solution and switch pitch.Potential intramolecular mechanisms are defined with respect to a reference plane joining the three apex carbons of the aryl rings and the process of lowest activation energy is believed to be a two-ring flip in which two aryl rings pass in one direction perpendicular to the reference plane while the third passes in the opposite direction through the reference plane (fig. 2). Our objectives were (1) to synthesise a range of propeller crowns; (2) to establish the propeller structure in solid and in solution; (3) to look at the way in which the flip affects the conformation of the ether ring; (4) to look at the complexation processes and how the flip affects these; ( 5 ) to evaluate the ligands in passive transport experiments and i.s.e.; and (6) to feed useful features of the macrocycle back to the design step. We have achieved a number of these objectives. Some forty crowns have been synthesised with substitution mostly at position 6 of the naphthalene rings as well as at all of the phenyl positions and at the carbons closest to the naphthalene rings in segment i (see 1). Crystallographic Results for Propeller Systems 1164 New Host-Guest Carriers Fig. 2. The two-ring flip mechanism found for many of the propeller crowns in this paper. The labelling indicates the two rings (2 and 3) which pass perpendicular to the reference plane. Fig. 3. Crystal structure showing the normal isomeric form of the propeller for the crown 2 projected on the reference plane taken from ref. (6).studies of crowns e.g. 2’ and salt complexes indicated only one of the 16 or 32 possible isomers of the propeller was present in the solid exceptlo for the free ligand 3 which formed a second isomer. Structures are shown of the common isomer in fig. 3 and the unusual one in fig. 4. Although the propeller was constant in form the ether conformation varied considerably in the crystal. N.m.r. studies in solution indicated two types of spectrum corresponding to a fluxional propeller which was apparently symmetric about the pseudo-axis shown in 1 and indicating only one set of naphthyl signals and a static or ‘ locked’ propeller which showed separate signals for each naphthyl group and differentiable sides to the phenyl ring.’l The minimum AGS for the symmetrisation process for the propellers fluxional at room temperature was 43.0 kJ mol-l.The propeller could be ‘locked’ on the n.m.r. timescale at room temperature using sterically demanding substituents in the ortho positions of the phenyl ring. To show how the movement of the ether strand would be affected by that of the propeller we used the ‘locked’ crown 4. The collapse of the But signals on the naphthyl rings measured the rate of symmetrisation of the propeller while the pairwise collapse of the four distinct deuterium signals of the two segments on opposite sides of the molecule measured the movement of segment i. The two processes are clearly correlated (see table 1). The rest of the ether chain is still mobile swivelling around the 1165 process 350 Y = But 358 D on segment i 2 3-flip segment i (T G) segment iT -9.18 (0.02) 6.36 (0.02) 6.88 (0.03) 4.76 (0.02) 6.26 (0.02) - 10.22 (0.04) 1-2 1-3 1-4 2-3 2-4 3-4 J.C . Lockhart Fig. 4. Crystal structure showing the unusual isomeric form of the propeller for the crown 3 projected on the reference plane taken from ref. (9). This propeller is related to that in fig. 3 by a 1 2 flip. Table 1. Free energies of activation for the symmetrisation of ligand 4 AGt/kJ mol-1 T/K signals used 76.6 & 0.5 76.2f 1.0 Table 2. Coupling constants for crown 4a* segment iG - 10.84 (0.01) 3.53 (0.01) 9.25 (0.01) 2.92 (0.01) 3.56 (0.02) - 10.86 (0.01) a Standard deviation in brackets. The rotamer populations for segment iT (iG) are Ng = 0.41 (0.81) Ng2 = 0.21 (0.19) and N t = 0.38 (0.0).deuterium-bearing carbon since the coupling constants (shown in table 2) are an average over several rotamers. This swivelling process was slowed down in the ' locked ' propellers on cooling but rate parameters have not yet been obtained; however it is much faster than the symme trisa ti on. 1166 We then proceeded to examine processes in which excess ligand is exchanged with complex (which could model the situation inside a membrane) or excess cation is exchanged with complex (which could model the aqueous side of the membrane interface).12 From the kinetics we hoped to determine the relative importance of (3) (4) The ligand had one proton which was very considerably shifted in the complex and this signal was used to determine the rate of ligand exchange [eqn (411.The rate of sodium exchange [eqn (3)] was determined using 23Na n.m.r.13 The analysis of the kinetics (see scheme 2) shows that the dissociative mechanism is predominant for New Host-Guest Carriers bimolecular steps such as those of eqn (3) and (4) and the dissociative step of eqn (1) Na+ + NaL+ k2 NaL+ + Na+ (for excess Na+ ) L + NaL+ e k2 NaL+ + L (for excess ligand). For dissociative and bimolecular steps operating together kobsd = k2[B1 + kd[BI/[Al where B = [NaL] A = “a+] (excess Na+) or [L] (excess L). A plot of kobsd/[B] versus 1/[A] gives kd and k,.13 Scheme 2. sodium exchange and for ligand exchange. The key result was that locked propellers were found to exchange sodium more slowly than the fluxional ones.This was the type of effect we were looking for and we proceeded to see how the slower rate of dissociation affected selectivity in multiple phases. Three-phase Passive Transport and I.S.E. Transport measurements were made with a simple chloroform membrane4 containing the crown and a source phase containing alkali-metal picrate and followed by the appearance of the picrate counter-ion in the receiving phase using the u.v.-visible absorption band of the picrate. As expected the results showed for propeller crowns with seven oxygen donors the selectivity sequence Rb > K > Na for those with six oxygen donors the sequence K > Rb > Na and for those with five oxygen donors the sequence Na 4 K.However the most interesting feature was that locked propellers within any one series showed higher transport rates than fluxional ones.14 A comparison of Rb uersus Na selectivity was also made with i.s.e. (PVC membranes) containing one fluxional (5) or one locked (6) seven-donor propeller crown. The locked propeller was again more selective. l5 Conclusions Summary of Kinetic Results The kinetic results for propeller complexes indicate that the symmetrisation process is much slower in the complex than in the free crown. The dissociation of the complex is faster than the symmetrisation process for either free or complexed propeller crown. The 1167 J . C. Lockhart dissociation process was found to be slower for complexes of locked propellers which were also found to promote more rapid three-phase transport.Feedback for Future Designs The propeller units have been found to provide suitable n.m.r. labels for kinetic studies of the crown-cation interaction. The hydrophobic nature and the bulk of the triaryl moiety as well as its complex chirality are exploitable features. A critical finding is that the introduction of a locked propeller provides a control over the kinetic performance of the ligand in several areas. While these current propeller crowns are not outstanding in relation to previously known crowns the key features noted can now be incorporated in more complex structures to control kinetic properties and impart other desirable steric features.I thank my coworkers Drs M. B. McDonnell P. D. Tyson Mr M. Todd and Mrs L. Cook who performed many of the experiments referred to Drs C. Spencer (Sheffield) I. Sadler (Edinburgh) and S. Hill (Newcastle) for the high field n.m.r. spectra and Dr W. Clegg who analysed the X-ray crystal structures. Dr A. K. Covington and the Sensors Group in Newcastle provided facilities for i.s.e. work carried out by H. Grey. Provision of high field n.m.r. and X-ray diffraction facilities by the S.E.R.C. are also gratefully acknowledged. References 1 B. G. Cox J. Garcia-Rosas and H. Schneider J. Am. Chem. SOC. 1981 103 1054; 1384. 2 J. C. Lockhart in Advances in Inorganic and Bioinorganic Mechanisms ed. A. G. Sykes (Academic Press New York 1982) vol. 1 pp. 217-268. 3 D. J. Cram and G. M. Rein J. Am. Chem. SOC. 1985 107 3657. 4 J. D. Lamb J. J. Christensen J. L. Oscarson B. L. Nielsen B. W. Asay and R. M. Izatt J. Am. Chem. SOC. 1980 102 6820. 5 J-P. Behr M. Kirch and J-M. Lehn J. Am. Chem. SOC. 1985 107 241. 6 I. 0. Sutherland J. Chem. SOC. Faraday Trans. I 1986 82 1145. 7 J. C. Lockhart M. B. McDonnell and W. Clegg J. Chem. SOC. Chem. Commun. 1984 365. 8 E. Bye W. B. Scweizer and J. Dunitz J . Am. Chem. SOC. 1982 104 5893. 9 K. Mislow Ace. Chem. Res. 1976 9 26. 10 W. Clegg J. C. Lockhart and M. B. McDonnell J. Chem. SOC. Perkin Trans. 1 1985 1019. 11 M. B. McDonnell Ph.D. Thesis (Newcastle upon Tyne 1984). 12 M. Todd and M. B. McDonnell personal communication. 13 E. Shchori J. Jagur-Grodzinski and M. Shporer J. Am. Chem. SOC. 1973 95 3842. 14 L. Cook personal communication. 15 H. Grey personal communication. Paper 511888; Received 21st October 1985

ISSN:0300-9599    

DOI:10.1039/F19868201161

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. SOC. Faraday Trans. I 1986,82 1169-1178 Biosensors based on Reversible Reactions at Blocked and Unblocked Electrodes Richard P. Buck Kenan Laboratories of Chemistry University of North Carolina Chapel Hill North Carolina 27514 U.S.A. The establishment of the potential difference at both blocked and unblocked interfaces is described and contrasted. Various applications of these interfaces to biosensors are discussed. Enzyme amplification or amplification by vesicle release can achieve the necessary sensitivity for immunoassays. Other approaches are to use carrier-antigen linked sensors or to rely on the kinetics of a Donnan membrane. Fully blocked interfaces have also been used but the potential differences are small and the experiments are not easy.Electrochemical sensors based on potentiometry amperometry ax. conductivity and capacitance measurements are already well established in analytical and clinical chemistry laboratories. These sensors are indifferent to sample colour or turbidity require little complex equipment and are relatively easily calibrated in aqueous solutions. In most cases they sense and measure concentrations and activities of ions and neutral species typically from 1 mol dm-3 to ca. low5 mol dm-3 although there are examples of wider sensitivity ranges. In every case the sensor is designed to possess an interface at which the generated potential the mass-transport-controlled current or the a.c. responses can be directly related quantitatively to desired species concentrations or activities.How well the actual device functions depends on a long list of factors from the categories [thermodynamic kinetic (heterogeneous and homogeneous) and mass transport] that determine observables [response range response stability (noise drift) response selectivity response time temperature coefficient lifetime pretreatment requirements and others]. As sensor systems become more elaborate by virtue of multiple interfaces of passive membranes and thin surface coatings with sandwiched reaction zones for interposed chemical reactions there is a great need for recognizing how the factors can be optimized. In particular what types of behaviour should be sought when choosing the potential- generating or potential-controlled interface. The choices include wide ranges of behaviour arising from the coupling processes at the principal interfaces.The latter may be unblocked blocked or kinetically controlled (e.g. lying between blocked and unblocked). Manipulation of processes at interfaces has been an important achievement of electrochemistry since 1947. Therefore it is not a foregone conclusion that unblocked reversible interfaces always provide the best sensors. However the majority of chemical sensors for application in biology and medicine described in the literat~rel-~ and available commercially are resistively coupled. The blocked or capacitively coupled interfaces are less well known and applied as sensors but have been intensively studied for almost a century by electrochemists.1° The aim of this paper is to summarize the basic ideas of blocked and unblocked interfaces and the literature covering the details as applied to sensors.Recent advances on some of the applications to immuno-biosensors are presented. 1169 Reactions at Blocked and Unblocked Electrodes 1170 Charge Redistribution at Interfaces Potential differences (pds) across interfaces are accompanied by a number of dis- tinguishable regions or layers of charge altogether called ‘the double layer’ consisting of one or more superficial layers in each of two phases.ll The charge originates by redistribution from electroneutral phases when placed in contact and from the external circuit. There are four principal sources. lo (1) Charge transfer across the interface; this charge may be electronic for redox- containing electrolytes in contact with metals or electronic semiconductors; it may be ionic for electrolytes contacting ion-conductors such as ionic defect solid-state materials liquid and solid ion exchangers and for electrolytes contacting extracting phases such as inert polymers containing ionophores (siderophores).An extensive list of examples has been published.12 For mixed conductors interfacial-ion and electron exchange can occur simultaneously. (2) Unequal specific (contact) adsorption of ions of opposite charge; at metal interfaces aqueous polarizable anions often absorb more strongly than cations and create a space charge in the phase containing the salt. However if ‘adsorption’ is understood more generally a single salt with a hydrophilic cation and hydrophobic anion partitioned at equilibrium also generates positive aqueous-phase space charge and residual negative organic-phase charge regardless of the phase into which the salt is originally introduced.This process is implied under category (I) although specific adsorption may also be involved. (3) Adsorption and orientation of dipolar molecules; penetration of electron clouds from electronic conductor surfaces into electrolytes and superimposed electrostatic fields account in part for orientation of solvent molecules at interfaces. The resulting dipolar layer changes the detailed profiles of field and potential. (4) Deformation of polarizable atoms or molecules in the unsymmetric force field at an interface. The total reorganization of chemical species and charge in the interfacial regions is established by minimization of electrostatic energy.Some factors especially (1) and (2) dominate responses depending on the degree of reversibility e.g. kinetics of the system. Blocked and unblocked interfaces are idealizations of extreme kinetic behaviour of charge transport across the interfaces. Classical reversible non-polarizable interfaces that pass electrons or ions at high exchange current densities are nearly ideally ‘unblocked’. The current can be mainly faradaic and an interface can be nearly ohmic at high mass-transport rates to and from an interface. Of course there is space charge and some non-Faradaic (charging) current flowing simultaneously when an applied potential difference is varied.The d.c. current-voltage curve can be represented as a straight line crossing the zero-current value at a constant potential difference. The a.c. circuit is a parallel RC with R -+ 0 (depending on electrode placement) and C typically 1-100 pF cm-2. Unblocked interfaces are susceptible to current control by mass transport because the interfacial kinetics are relatively very fast. In fact the i against Y curve for reversible electrodes are never quite vertical lines even at the zero-crossing for two reasons. There is always an iR drop between the electrodes used to apply or sense the potential differences and at high current flow mass transport rate in one phase or the other causes current limitations. Diffusion or diffusion-migration limited currents are typical of concentration-polarized interfaces.The well known relation between amperometric and potentiometric operation of a given electrode is apparent ;13 the zero-crossing potential is logarithmically dependent on species activity while the limiting current is directly proportional to species concentration. Blocked electrodes possess polarizable interfaces and pass no Faradaic current because there is no exchange process or all charge exchanges are so completely irreversible 1171 R. P. Buck (kinetically slow) as to be virtually absent. The interface seen as a contact is completely capacitively coupled just the opposite of ohmic-reversible. Current flowing to and away from the electrode interface is capacitive charging/discharging current.The d.c. current- voltage curve is ‘flat’ e.g. no d.c. current flows for any applied voltage. The a.c. circuit is again a parallel RC with R approaching infinity. Real electrodes of classical types and the more modern membrane sensors show intermediate behaviour lying between the two ideal extremes. Practical potentiometric sensors are based on nearly unblocked reversible-electron and ion-exchanging interfaces. Practical amperometric sensors can make use of reversible or kinetically slow interfacial transfers. The latter application arises because slow rates can be increased by application of an overpotential that appears across the inner layers decreases the electrostatic barrier and increases the surface reaction rate. Of course this strategy does not apply to processes controlled by very slow coupled homogeneous reactions but can be useful for many membrane-coated sensor devices.Insulator-covered semiconductor electrodes and other chemically sensitive semiconductor devices and simple coated wire electrodes using purely ionic conductors coated on purely electronic conductors are examples of nearly blocked interfaces. Two illustrative examples are provided by fluoride and by glass pH electrodes. The LaF,/Pt interface is nearly blocked and potentiometrically unstable because an ionic conductor contacts an electronic conductor. Stability is achieved by a charge-exchange unblocked interface sequence LaF,/AgF/Ag. Similarly glass/metal shows conductor-type mismatch unless mercury sodium or silver metals are used together with a source of Hgl Na+ or Ag+ to provide ion exchange between the two phases.Both ideally blocked and unblocked interfaces can be treated mathematically by the methods of electrochemical thermodynarnicsl4* l5 to establish useful potential/charge/ activity relationships. The unblocked electrodes are treated most readily to derive relations between potential differences and ion activities using energy considerations equality of the electrochemical potential of species at equilibrium in two or more phases. The resulting expressions for many practical electrode examples have been catalog~ed.~~? 16! l7 For electrodes large compared with the double-layer dimension (the Debye length) the potential-difference-activity relations are independent of electrode size shape and charge ! At the opposite extreme blocked electrodes are treated using Gibbs’ general thermo- dynamic adsorption analysis combined with electrostatic analysis according to Gouy- Chapman-Stern theory.ls The principal result of the theory is that the potential-activity relationships depend on charge at an interface (equal and opposite on the two sides).Acase studied by Esin and MarkovlQ is the dependence of the potential-difference-activity relation on charge for high levels of + or - charge e.g. electrodes far from the potential of zero charge (P.z.c.). They demonstrated that blocked electrodes could show Nernst-like response for simple single-salt solutions. However for complex solutions such as blood the potential difference depends on all of the ionic species that contribute to the space charge.The Esin-Markov effect per se will probably not be a principle for designing sensors. However its analysis is instructive for interpreting responses of coated wire electrode and simple hydrated insulator pH ISFET response^.^^ Several examples of blocked-interface responses have been described and a few responses ana1y~ed.l~ When charge exchange is rapid and reversible specific adsorption of charge does not figure in the thermodynamic response. Unless a sufficiently impenetrable film is formed on a reversible sensor so that charge exchange is markedly slowed then reversible electrode responses in principle do not depend on intermediate adsorption or on the exact charge profiles.Presumably adsorption and charge distribution adjust to satisfy the equality of electrochemical potentials from one phase to the other. On the other hand blocked electrodes depend on the charge and specific adsorption is dominant. Attempts to use field effect devices at constant charge have not been altogether 21 It 1172 Reactions at Blocked and Unblocked Electrodes is not easy to hold constant localized charge because of imperfect dimensional stability and imperfect insulators. Also there are numerous parasitic processes ofcharge cancellation such as surface group ionization and redox discharges. Platinum electrodes with surface oxides are proton exchangers. Platinum electrodes in the double-layer potential difference region are prone to electron exchange by ambient oxygen.14 These factors may be crucial in establishment of stable pds at some ion-exchange/wire interfaces.Experimentally unblocked interfaces can be fouled or otherwise altered to slow charge-exchange rates and blocked interfaces can be partially unblocked by parasitic Faradaic processes. Thus practical devices depending on processes at blocked and unblocked interfaces should not be expected to behave ideally. Designs and expectations should account for uses in hostile environments. Progress on Biosensors using Unblocked Interfaces In the past 15 years composite systems involving basic electrodes sensitive to ions and interposed chemical reactions have been prepared. The purpose is to extend the response of the basic electrodes to new often neutral charge species by converting the new species into one sensed by the basic electrode.The electrode can be simply overlaid with membrane layers that contain the reactants needed and may exclude undesired reactants. The simplest of these is a gas sensor. When enzymes are immobilized on basic sensor surfaces or trapped between a permeable membrane and an ion or gas sensor electrodes are formed that are sensitive to many otherwise electrochemically inactive substances. So-called enzyme electrodes have enjoyed spectacular popularity as research topics. 1-9 Despite a large catalogue of examples enzyme-based sensors and reverse-operating sensors that measure enzyme activities have not found commercialization except in a few cases. Evidently the problems of stability and shelf-life have not been solved.Living bacterial-based electrodes provide another route to highly sensitive membrane- based sensors. Instead of isolating enzymes from bacteria the whole culture is immobilized on an electrode surface in a buffered medium isolated by a permeable membrane. Although response times tend to be longer than conventional base electrodes and the shelf-life is short a number of examples have been demonstrated and some are given in a recent review.22 One can expect great selectivity from these sensors although interferences from intermediate metabolic products are frequent. Shelf-lives of 20-30 days can be achieved with cold-storage. Tissue-slice electrodes and membranes sectioned from various plants are similar to bacterial electrodes in construction details.24 Charged drug-sensing electrodes have been made for drugs of both sign using hydrophobic ion pairs as ion exchangers. The general theory which discloses factors determining sensitivity and selectivity will soon be published.23 Numerous examples are catalogued in recent Immunochemical Related Sensors using Unblocked Interfaces Indirect Measurements Protein determinations using silver ion-sensing electrodes have been 26 The sulphydryl groups on protein bind silver ions after denaturation in basic electrolytes. Human serum albumin could be determined at the 0.5-30 pg ~ m - ~ level after addition and incubation with antibody. The precipitation reaction product was removed by centrifugation and the supernatant or precipitate was denatured and mixed with a known low level of silver ion.Comparison of potential shift between a blank and a sample led to a quantitative analysis. Similarly antigens or haptens could be immobilized on a solid An unknown antibody sample was applied and retained on the support. Other non-specifically held proteins were washed out. The specific antibody could then be 1173 R. P. Buck removed with an eluting buffer and delivered to a denaturing silver ion-containing solution for measurement. When a hapten-selective electrode can be made such as a membrane electrode for trimethylphenylammonium ion antibodies for the hapten can be determined by changes in the hapten concentration upon application of antibody.28 36 Amplification of Responses There have been reported two types of amplified responses.The first is an electrode-based enzyme immunoassay analogous to the well known radioimmunoassay . Enzymes have been shown to be adequate substitutes for radioisotopes in many cases. The enzyme label must have a high turnover rate and produce or consume an ion that can be measured by an electrochemical sensor. One examplez9 is the use of an iodide sensor that detects a decrease in iodide concentration resulting from oxidation by peroxide. In the sequence of events to determine hepatitis B surface antigen corresponding antibodies were first immobilized on an artificial protein membrane anchored on the iodide electrode. When immobilized antibody was reacted with an unknown small amount of hepatitis B surface antigen an excess of antibody remained.The latter was determined by addition of peroxidase-labelled immunoglobin IgG. When the electrode was removed washed and inserted into an optimized solution of peroxide the labelled IgG catalysed removal of peroxide and the residual peroxide oxidized iodide. Linear calibrations were observed in the range 0.5-50 pg dm-3. The analysis time was ca. 6-7 h by virtue of the long incubation required. Another example is determination of adenosine-3’,5’-cyclic phos- phate c-AMP and bovine serum albumin BSA using urease as the enzyme label with an ammonia gas-sensing e l e c t r ~ d e . ~ ~ The second type of amplification used direct chemical release rather than indirect enzyme reactions. Vesicles such as red blood-cell ghosts were loaded with an easily detected non-physiologic ion trimethylphenylammonium.The cells naturally have specific antigenic determinants on their surfaces. An unknown concentration of an antibody that binds to the surface will do so and the complement enzymes will recognize the surface complex and lyse the cell to release the marker ions. The antibody haemolysin has been determined by this method.31 A more subtle example was determination of BSA antiwody that was first mixed with antiserum and complement. Some of the complement was consumed by the BSA-antiserum so that a lesser rate of lysis occurred upon addition of the vesicles. BSA antibody could be determined in the low pg ~ m - ~ range and the results correlated with the manufacturers’ stated values.32 Phospholipid vesicles can be made from phosphatidylcholines cholesterol and other liposome formers.33 Liposomes can trap marker-ions inside and the excess can be removed to produce a suspension of marker-loaded liposomes.In one example ganglioside was incorporated in the vesicle wall. Treatment with a specific antibody released the trapped tetraphenylammonium ions. When liposomes were made with dinitrophenol in the walls dinitrophenol antiserum and complement lysed the vesicle to release the detectable marker ions. Free dinitrophenol was used to titrate DNP antibody and the latter was sensed by the liposome. Levels of 10-7-10-8 mol dm-3 of hapten could be determined.34 Another more complicated system has been devised to monitor avidin or biotin using vesicles loaded with trimethylphenylammonium The practitioners in this field consider the use of loaded vesicles as a general assay method.Vesicles can contain large numbers of detectable ions and the red-cell ghosts can carry most haptens and antigens. The potential drawbacks include lack of specificity of some antibody-hapten reactions especially sensitivity to metabolic products of haptens. Also binding constant variability can lead to degradation of detection limits. 1174 Reactions at Blocked and Unblocked Electrodes 38 Carrier-Antigen Linked Sensors An active antigen or hapten can be covalently attached to an oil-soluble neutral carrier of the types used for preparation of ion-selective electrodes. In each case the neutral carrier while chemically labelled retains its function.For example a modified crown ether that produces a potassium or a sodium sensor retains its carrier property and yields a well behaving ion-selective electrode with reversible slope and some minor change in standard potential. When an antibody is admitted to the electrolyte phase bonding with the antigen or hapten changes the standard potential of the ion-selective electrode response possibly by changing the carrier concentration at the surface in the space-charge region in the membrane interface or on the side exposed to antibody. The first immuno-sensor to be demonstrated involved the use of the hapten dinitro- phenol covalently linked to dibenzo- 18-crown-6-ether. This conjugate was dispersed in a plasticised poly(viny1 chloride) membrane and mounted on a conventional electrode body.At constant sodium or potassium activities the response was constant. However when dinitrophenol antibodies were admitted the potential changed in 6-16 min to a new steady-state value. Surprisingly the response to whole serum containing many serum proteins was still selective. There was a lack of non-specific protein adsorption on the membrane. Although the mechanism of the response is presently unclear the selectivity of the electrode to various cations is modified by the presence of antibody. Perhaps the carrier-hapten mobility is changed as a result of bonding to the antibody. Very recently another example of this type of system was reported by Re~hnitz.~~ Digoxin-modified crown ethers retain their response to potassium ion.However upon addition of antibody at constant potassium activity the response to antibody was again observed. By loading the membrane at about the same concentration as the antibody-containing bathing electrolyte a near-Nernstian response was found even at the extraordinarily low concentrations typically encountered for antibody preparations. Immuno-sensors have also been based on unblocked membrane interfaces including cases in which the antibody or antigen is placed inside the electrode interface while the antibody is placed in the external bathing electrolyte solution on one side. The electrode is so far restricted to membrane type and the reference electrode is conventional e.g.silver/silver chloride in junction configuration. Aizawa et aZ.*O have reported blood-typing membrane electrodes containing membrane-bound blood-group substances. In the presence of added antibody (agglutin) in serum on one side of the membrane the membrane potential is changed significantly by the agglutination reaction that seems to occur mainly at the surface. The interesting mode of operation (use of a fixed non-unity ratio of NaCl bathing electrolytes) suggests that the measured membrane potential change before and after agglutination is a diffusion-potential change due to NaCl transport through the membrane. The classical membrane electrochemical interpretation of the experiment is that the membrane is probably a low site density ion exchanger that is undergoing some degree of Donnan exclusion failure.The result is that the ‘pilot’ salt NaCl is not freely diffusing as a salt through an inert membrane but shows a lesser flux because one ion is partly excluded. The steady-state diffusion potential is larger than the typical free-flowing junction-potential analogue. Addition of the antibody alters the fixed site concentrations the salt flux and the resulting diffusion potential. Membranes studied originally included triacetyl cellulose and poly(viny1 chloride). To demonstrate the presence of charged sites membrane potentials were measured for triacetyl cellulose membranes modified by addition of surface groups bromine amine and epoxy residues. When the modification was done on both sides of the membrane and the unsymmetric bathing solutions were employed membrane potentials of positive or negative sign were 1175 R.P. Buck expected depending on pH of the bathing electrolytes. Results were consistent with this expectation. When proteins egg albumin and lysozyme were immobilized on one side membrane potentials were indeed changed.41 An immuno-responsive membrane was then demonstrated by blending cariolipin antigen complexes in a cellulose acetate The membrane was responsive to the Wasserman antibody but non-specific adsorption of proteins interfered. The assay system was further investigated as an immuno-sensor for syphilis.43 The most recent theoretical analysis of potentials generated by Donnan exclusion failure was applied to liquid membrane^,^^-^^ and earlier references to fixed-site membranes are contained in these papers.Both the diffusion potential and reversible interfacial potential differences are affected by site density and the total effect is directly measureable by potentiometry during salt encroachment and transfer. Thus Donnan failure or the reverse perm- selectivity generation is a bulk process in membranes and can occur when interfacial processes are reversible. However the experiments show that the measured potentials are generated by very low activities of charged species such as antibodies. It seems probable that some other effects involving surface rate control are operative. Another group of effects that leads to potential differences at electrolyte/membrane interfaces are associated with Frumkin.The basic notion is that leaky membranes admit salts from bathing solutions but in contrast to the Donnan failure interpretation above the interfacial potential-dependent extraction process for kinetically slow ion transfers establishes the potential difference. When the membrane is subsequently modified by a surface reaction with a large molecule such as a protein the rates of pilot transfers are changed and the new potential difference becomes a monitor of the protein surface reaction. Theoretically this surface effect also requires imperfectly permselective materials and it arises when there is a failure of co-ion exclusion accompanied by slow interfacial kinetics of ion crossing^.^' Whenever the site density of ion exchanging materials is low in comparison with the external electrolyte concentration multiplied by the salt-extraction coefficient it is probable that ions of both signs will enter a membrane.Thus for organic liquids and polymers used as film materials it is probable that salts from the bathing electrolyte will dissolve in the organic phase. This effect has been observed and verified by Collins and Janata,4s and it is likely that electrodes made from site-free materials for binding haptens enzymes and antibodies may be plagued by salt encroachment from bathing electrolytes. An important assumption of the electrochemical kinetics-based analysis is that away from extraction equilibrium rates of extraction and rates of removal of pilot salt ions must produce equal net fluxes of ionic charge in each phase.Otherwise electroneutrality will be violated. Interfacial shifts in potential difference are required in response to surface charge to provide equality of fluxes. Nearer to equilibrium and equilibrium interfacial potential differences become independent of rates. At equilibrium the individual forward and backward fluxes of pilot ions can be quite different. For example an extracting interface at equilibrium can receive cations and anions at different rates because their simultaneous removal rates balance the extraction rates. Immunochemical Related Sensors using Blocked Interfaces The first example of a blocked-interface immuno-responsive electrode consisted of a surface-oxidized titanium wire onto which human chorionic gonadotropin (hCG) antibody was anchored.49 The measured electrode potential us.a reference showed a small response to added hCG at pH 8.5. Since the anchored surface charge was negative at this pH the reaction with hCG should have involved cancellation or redistribution of charge density with a positive directional change in potential. Similarly an electrode with anchored hCG exposed to antibody showed a small negative change in potential. The Reactions at Blocked and Unblocked Electrodes 1176 change in potential was a few to tens of millivolts and the potential stability was poor. Some improvement was possible by limiting the surface area and making sure that the surface was well overed.^^^ 51 Since oxides themselves are sensitive to pH as well as to the redox state of the solution it is remarkable that the charge-cancellation/redistribution effect could be observed.Another example was an oxidized titanium wire with chemisorbed dye Cibacron Blue F3G-A.52 The reference electrode was an identical oxidized wire but without the dye. The difference potential was sensitive to addition of HSA in physiological buffer. The response was linear to HSA between 0 and 15 pg ~ m - ~ . The response was reversible; the electrode could be used for several months but was sensitive to interferences by myoglobin and rabbit gamma globulin. Lysozyme and ovalbumin did not interfere. Charge cancellation at a blocked interface with redistribution of charge in the oxide semiconductor is likely to be a more specific event than the Esin-Markov effect.When a metal or semiconductor with a bound charged dye a surfactant or an antigen is equilibrated with an electrolyte there will be an interfacial potential difference with inner layer and diffuse potential components as determined by the ionic concentration and dipole effects. Upon addition of a specific counter-charge reagent the bound layer may be discharged partially completely or even charged with opposite sign. The measured potential should then reflect this charge cancellation reaction. The probable requirements are (1) a fully blocked interface with respect to the charged biospecies (2) establishment of charged interface using a closed circuit (3) measurement of potential difference at nearly open circuit using an electrometer (4) measurement of potential change upon charge cancellation and (5) a large favourable equilibrium constant for the charge cancellation reaction.The sensitivity may be quite large for a change in surface concentration Ac/cm2 where the charge change is Aq = Ac/cm2 x n(charge/species) x F. (1) For pmol cm-2 changes and a blood electrolyte with the Gouy-Chapman capacitance value of 86.6 pC V-l cm-2 the expected voltage change is dd2/dc = 1.1 In cm2 mV pmol-l. (2) By using integral capacitance data determined experimentally in terms of actual total rational electrode potential a lower capacitance value is found. Using this more realistic integral capacitance of 10-20 pF cm-2 the result 5-10 cm2 mV pmol-l is found. These experiments are not easy to perform.There are several severe problems that are sources of error or failure (1) charge leakage to the exterior bathing solution by loss of surface coating or spreading of charge along the surface by imperfect insulations (2) charge leakage to the interior underlying electrode via ion penetration or electron exchange at the incompletely blocked interface and (3) an unfavourable chargecancellation equilibrium constant. Summary The ideas of ideally blocked and unblocked interfaces are explained in the context of electrochemical biosensors. Progress on the design of sensor systems using imperfect examples of both types of interface is illustrated with various proposed immunochemical sensors. Electrochemical sensors are not restricted to reversible charge-exchanging inter- faces and the probability that some forms of nearly blocked interfacial sensors can be used in successful devices is very good.In addition the wide region of intermediate ' kinetic' controlled interfaces are promising and have already shown success under controlled-potential amperometric conditions. I thank the National Research Council (U.S.A.) and the National Science Foundation (grant CHE-8406976) for partial support. 1177 R. P. Buck References 1 P. L. Bailey Analysis with Ion-selective Electrodes (Heyden London 1976). 2 Ion-selective Electrodes ed. E. Pungor and I. Buzas (Akademia Kiado Budapest 1973,1977,1978,1981 and 1985) vol. 1-5. 3 Ion-selective Electrodes in Analytical Chemistry ed. H. Freiser (Plenum Press New York 1978 and 1980) vol.1 and 2. 4 Ion-selective Electrode Methodology ed. A. K. Covington (CRC Press Boca Raton Florida 1979) vol. 1 and 2. 5 Medical and Biological Applications of Electrochemical Devices ed. J. Koryta (John Wiley and Sons New York 1980). 8 G. G. Guilbault in Biomedical Applications of Immobilized Enzymes and Proteins ed. T. M. S. Chang 6 Ion and Enzyme Electrodes in Biology and Medicine ed. M. Kessler L. C. Clark D. W. Lubbers I. A. Silver and W. Simon (Urban and Schwarzenberg Munich 1976). 7 Ion Measurements in Physiology and Medicine ed. M. Kessler D. K. Harrison and J. Hoper (Springer- Verlag Berlin 1985). (Plenum Press New York 1977) p. 163. 9 A. S. Barker and P. J. Somers in Topics in Enzyme and Fermentation Biotechnology ed. A. Wiseman (Ellis Horwood Ltd.Chichester 1973) vol. 2 chap. 3. 10 R. Parsons in Modern Aspects of Electrochemistry ed. J. O’M. Bockris and B. E. Conway (Butterworths Scientific Publications London 1954 chap. 3 pp. 103-1 79. 11 R. delevie in Chemically Sensitive Electronic Devices Principles and Applications ed. J. Zemel and P. Bergveld (Elsevier Sequoia S.A. Lausanne 1981) pp. 97-1 10. P. Bergveld (Elsevier Sequoia S.A. Lausanne 1981) pp. 137-196. 12 R. P. Buck in Chemically Sensitive Electronic Devices Principles and Applications ed. J. Zemel and and Sons New York 1980) chap. 1 pp. 1-12. 14 R. P. Buck in Theory Design and Biomedical Applications of Solid State Chemical Sensors ed. 13 J. Koryta in Medical and Biological Applications of Electrochemical Devices ed.J. Koryta (John Wiley P. W. Cheung D. G. Fleming W. H. KO and M. R. Neuman (CRC Press W. Palm Beach Florida 1978) pp. 3-36. P. W. Cheung D. G. Fleming W. H. KO and M. R. Neuman (CRC Press W. Palm Beach Florida 15 J. Janata in Theory Design and Biomedical Applications of Solid State Chemical Sensors ed. P. Bergveld (Elsevier Sequoia S.A. Lausanne 1981 pp. 197-260. 1978) pp. 41-48. 16 R. P. Buck in Chemically Sensitive Electronic Devices Principles and Applications ed. J. Zamel and 17 R. P. Buck in Comprehensive Treatise of Electrochemistry vol. 8 of Experimental Methocis in Electrochemistry ed. R. E. White J. OM. Bockris B. E. Conway and E. Yeager (Plenum Press New York 1984 chap. 3 pp. 137-248. 18 I. Lauks in Chemically Sensitive Electronic Devices Principles and Applications ed.J. Zemel and P. Bergveld (Elsevier Sequoia S.A. Lausanne 1981) part I pp. 261-288; part 11 pp. 393-402. 19 0. A. Esin and B. F. Markov Acta Physicochim. URSS 1939 10 353. 20 R. M. Cohen and J. Janata J. Electroanal. Chern. 1983 151 33; 41. 21 R. M. Cohen and J. Janata Thin Solid Films 1983 109 329. 22 R. L. Solsky Cr. Rev. Anal. Chem. 1982 14 1. 23 R. P. Buck and V. V. Cosofret in Chemical Sensors-Principles and Applications ed. D. Schuetzle (A.C.S. Symp. Ser. Am. Chem. SOC. Washington D.C. 1985). 24 V. V. Cosofret and R. P. Buck Ion-Selective Electrode Rev. 1984 6 59. 25 P. W. Alexander and G. A. Rechnitz Anal. Chem. 1974,46,250. 26 P. W. Alexander and G. A. Rechnitz Anal. Chem. 1974,46 1253. 27 R. L. Solsky and G.A. Rechnitz Anal. Chim. Acta 1978,99 241. 28 M. E. Meyerhoff and G. A. Rechnitz Science 1977 195 494. 29 J. L. Boitieux G. Desmet and D. Thomas Clin. Chem. 1979 25 318. 30 M. E. Meyerhoff and G. A. Rechnitz Anal. Biochem. 1979,95,483. 31 P. D’Orazio and G. A. Rechnitz Anal. Chem. 1977,49,2083. 32 P. D’Orazio and G. A. Rechnitz Anal. Chim. Acta 1979 109 25. 33 K. Shiba T. Watanabe Y. Umezawa S. Fujiwara and H. Momoi Chem. Lett. 1980,1980 155. 34 K. Shiba Y. Umezawa T. Watanabe S. Ogawa and S. Fujiwara Anal. Chem. 1980,52 1610. 35 E. A. Bayer and M. Wilchek Metho& Biochem. Anal. 1979 26(1) 19. 36 C. R. Gebauer and G. A. Rechnitz Anal. Biochem. 1980 103 280. 37 R. L. Solsky and G. A. Rechnitz Science 1979 204 1308. 38 R. L. Solsky and G. A. Rechnitz Anal. Chim. Acta 1981 123 135. 39 G. A. Rechnitz Report at FACSS Meeting Philadelphia PA. September 27 1983. 40 M. Aizawa S. Kato and S. Suzuki J. Membr. Sci. 1980 7 1. 41 S. Kato M. Aizawa and S. Suzuki J. Membr. Sci. 1978 3 29. 1178 42 M. Aizawa S. Kato and S. Suzuki J. Membr. Sci. 1977 2 125. 43 M. Aizawa S. Suzuki Y. Nagamura R. Shinohara and I. Ishiguro J. Solid-Phase Biochem. 1979 4 25. 44 R. P. Buck F. S. Stover and D. E. Mathis J. Electroanal. Chem. 1977,82 345. 45 F. S. Stover and R. P. Buck J. Electroanal. Chem. 1978,94 59. 46 R. P. Buck F. S. Stover and D. E. Mathis J. Electroanal. Chem. 1979 100 63. 47 R. P. Buck IEEE Trans. Elec. Div. 1982 ED-29 108. 48 S. G. Collins and J. Janata Anal. Chim. Acta 1982 136 93. 49 N. Yamainoto Y. Nagasawa S. Shuto M. Sawai T. Sudo and H. Tsubomura Chem. Lett. 1978,245. 50 N. Yamamoto Y. Nagasawa M. Sawai T. Sudo and H. Tsubomura J. Immunol. Meth. 1978,22,309. 51 N. Yamamoto Y. Nagasawa S. Shuto H. Tsubomura M. Sawai and H. Okumura Clin. Chem. 1980 26 1569. 52 C. R. Lowe FEBS Lett. 1979 106,405. Reactions at Blocked and Unblocked Electrodes Paper 511889; Received 21st October 1985

ISSN:0300-9599    

DOI:10.1039/F19868201169

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. SOC. Faraday Trans. I 1986,82 1179-1 186 Design of Anion-selective Membranes for Clinically Relevant Sensors Urs Oesch Daniel Ammann Hung V. Pham Urs Wuthier Richard Zund and Wilhelm Simon* Department of Organic Chemistry Swiss Federal Institute of Technology Uniuersitatstrasse 16 CH-8092 Zurich Switzerland Requirements for the applicability of anion-selective electrodes in clinical analysis are discussed in terms of selectivity response time lifetime and stability. Basically classical anion-exchangers electrically charged carriers and electrically neutral carriers may be used as anion-selective components in solvent polymeric membranes. While classical ion-exchanger-based mem- brane electrodes are only of limited practical relevance sensors containing ion carriers seem to bear a broad potential with regard to the feasibility of realizing the required properties.The control of the selectivity determining properties of electrically neutral tin organic anion carriers probably allows the development of sufficiently selective membranes for the assay of clinically relevant anions. Furthermore the incorporation of membrane technology into the design of such carrier membranes may lead to fully optimized sensors. Solvent polymeric membranes have proved to be especially suited for clinical analysis with ion-selective sensors since they can easily be manufactured in different sizes and shapes and are less affected by the presence of biological substrates (e.g. proteins).l. For clinical ion measurements in blood and urine a well balanced optimization of all the electrode characteristics relevant for the application should be discussed rather than the superiority of a single pr~perty.~ Such relevant properties are selectivity e.m.f.stability response time and lifetime of the sensor. In previous contributions we focussed on the design of clinically relevant sensors for cation^.^-^ Here we report on design features for solvent polymeric membranes exhibiting selectivities for an assay of anions in whole blood blood serum and urine. Theoretical Using the Nicolsky-Eisenman formalisme and assuming representative physiological concentration ranges required selectivities (log Kpt) can be calculated for a maximally tolerable error of 1% due to interfering anions (table 1).In the case of serum plasma or whole blood as the physiological sample one can claim required selectivities for a worst case i.e. highest activity of the interfering ion and lowest activity of the primary ion within their normal physiological ranges (see Experimental). This is reasonable since the ranges are rather well defined for blood electrolytes. This does not hold at all in the case of urine as a sample. On one hand the urine electrolyte concentrations depend on the daily fluid excretion volume which may vary between 0.5 and 2 dm3.' On the other hand the range of the daily electrolyte excretion is usually reported as extreme values (table 2). Therefore a required selectivity estimation for urine is more reasonably done on the basis of a typical concentration.Such obtained required selectivities represent a guideline rather than a guarantee for the 1 % interference of another anion. Calculations for other than typical concentrations can easily be performed using the equations given in the Experimental section. 1179 1180 Table 1. Required potentiometric selectivity factors K Y for a maximal interference of 1 % by other anions Clinically Relevant Sensors OH- C1- primary ion 1% -0.8 -6.2 -3.4 -5.6 -0.1 -5.6 -2.8 -4.9 1.4 -2.9 0.1 (a) for serum plasma and whole blood applications (worst case) 0 0 -7.6 -4.8 -7.0 0.7 - 1.4 0 3.4 -0.6 -6.1 2.7 -2.7 1.7 (0.8) 0 -5.4 -3.8 (-4.8) -4.7 -6.8 -6.8 0.9 -1.2 -1.3 -2.3 (-3.2) -3.2 -5.2 -5.3 0 0 -1.8 (-2.7) -3.4tl.5) 3.5 (1.6) 1.1 (0.1) 0 3.0 (1.1) 0.2 -1.9 -1.9 -3.3 -5.4 -5.4 -4.7 -4.7 1.6 -2.6 OH- c1- Br- HCO; SCN- H,PO; HP0;- so:- COZ- 6.6 -4.2 6.7 -4.1 6.2 -4.6 0 0 -8.9 -5.7 1.2 4.9 1.7 -5.2 1.7 -5.3 -2.1 2.8 -4.1 -0.9 10.8 -3.1 3.4 3.9 (b) for urine applications (typical case) 0 - 5.7 1.3 - 1.9 0 0 0 -7.9 -6.9 -7.2 -0.9 -0.3 -4.1 -4.7 -5.0 -4.2 -4.7 -5.0 -2.5 -2.8 -2.6 -0.9 -0.4 -1.4 OH- Cl- Br- SCN- H,PO HPOi- SO:- 11.4 -2.5 a Values for non-smokers values for smokers given in parentheses.In view of clinically meaningful results the performance of the cell assembly should allow an adequate subdivision of the e.m.f.ranges corresponding to the physiological concentration ranges (table 2). For a five-fold subdivision (95% confidence limit) of the physiological normal range in serum plasma or whole blood the required standard deviations of single assays are as given in table 2. For urine applications where the e.m.f. ranges for the various ions are rather large (see table 2) demands on the required e.m.f. stabilities are therefore comparatively modest. The acceptable drift or offset of the signal e.g. due to the calibration routine or the response time of the system has to be compatible with these e.m.f. stabilities. In commercial clinical analysers one has to cope with contact times of the sensor with the sample of only ca. 30 s. Among the many different processes that contribute to the dynamic response behaviour of the measuring system only the intrinsic response time of the ion-selective membrane may be influenced significantly by adequate membrane technology.8 For a high speed of response of neutral carrier-based electrodes plasticizers of low dielectric constant are usually advantageous.8 The e.m.f.stability and the lifetime of neutral carrier-based solvent polymeric membranes seem to be related.5 Since the lifetime is to a large extent dictated by the loss of membrane components (e.g. ion carrier plasticizer) into the sample solution membrane technological measures may be taken to ensure a desired lifetime. To guarantee that neutral carrier-based solvent polymeric membrane sensors in a typical commercial flow analyserg exhibit a continuous use lifetime of at least one month in contact with undiluted serum or whole blood a lipophilicity log P (partition coefficient P between water and octan-1-01) of 12.8 and 8.4 for plasticizer and ion carrier respectively is necessary.For the less lipophilic urine samples values of 4.1 and 2.3 for plasticizer and ionophore respectively are ~ufficient.~ log Kg:t interfering ion Jzj SCN-a Br- HCO; 1.5 -2.8 1.0 -3.4 - 0.8 3.5 4.0 H,PO; HPOi- SO:- C0:- - 4.9 0.6 - 3.4 0.0 - 3.5 - 2.9 - 1.8 - 1.7 0 0 1.7 -2.5 1.2 -3.0 -3.0 0 0 anion OH- c1- Br- HCO; SCN-f H,PO;Q HPO:-g so;- cop a Concentration values given on the basis of an average daily fluid excretion volume of 1 dm3 although it may vary individually between 0.5 and 2 dm3.7 Calculated according to the corresponding activity range assuming constant ionic strength of I = 0.160 mol dmU3 and ideal (Nernstian) electrode slope at 25 "C.Required standard deviation of the e.m.f. for a five-fold subdivision of the given physiological range with a 95% confidence limit. Calculated according to the corresponding activity range assuming a constant ionic strength of I = 0.35 mol dm-3 and ideal (Nernstian) electrode slope at 25 "C. Indefinite owing to the lower limit. f Values for non-smokers for smokers given in parentheses. Q Values calculated on the basis of total inorganic phosphate pH range and the H,PO;/HPO;- equilibrium constant (pK = 7.23). Values calculated on the basis of bicarbonate concentration range pH range and the HCO;/COi- equilibrium constant (pK = 10.33).6.9 25 0.0 E.M.F. Measurements Measurements were performed on cells of the type at 20-21 "C with the equipment described earlier.1° The data were corrected for changes in the liquid-junction potentials and single-ion activities were evaluated as shown previously.* Hg Hg2Cl,; KCl(sat.) I 1 mol dmA3 LiOAc I sample (1 membrarie 11 Ion-selective Membranes Membranes had the compositions indicated in fig. 1-3. For preparation see ref. (1 1). 0.3-1 .O 0.10-0.17 Materials Origin preparation and purification of the ion-selective components have been mentioned previously.l09 12-14 The other membrane materials were obtained from Fluka AG Switzerland.All other chemicals were of the highest purity avai!able. 1181 urinea extreme concentration e.m.f. ranged U. Oesch et al. Table 2. Representative physiological concentration ranges of anions serum plasma whole blood required e.m.f. e.m.f. typical rangeb stabilityC concentration7 /mV 95% normal concentration range4/mmol dmP3 /mV range7 /mmol dm-3 /mmol dm-3 /mV ~~ 0.58 1.6 x 10-5 11.7 217 2 . 6 ~ 10-4 to 4.1 x 10-4 31 28 e e 4.5 x 10-7 to 2.2 x 10-3 80-270 0.037-0.107 0.1-160 0.0-0.1 135 0.082 0.0 0.07 13.1 0.19 3.8 0.28 1.1 1.8 0.80 3.8 75.3 5.6 22.7(78) 36.9 15.9 95-1 10 0.009-0.17 2 1.3-26.5 0.007-O.O 17 (0.15) 0.034-0.14 0.26-0.89 0.03-38 0 .O 1 7-3 8 2.5-45 183 198 74 e 0.77 0.34 15.4 6.7 0.0-8.4 Experimental 0.01 mol dm-3 NaCl; AgC1 Ag 1182 PVC -ao,- -SCN- -11- - --NO; 7 1 - 5 4 3 2 - - x 1 Q n m u - 1 - 2 - 3 - L - 5 - 6 CLASSICAL ANION- EXCHANGER CHARGED CARRIER MTDDACl ( 1 '10) HOCO(m)ACCP(l%) TBTCI ( 1 'lo) TOTCI (1.5 'lo) TBTCI ( 20.0%) TOT BBPA (66%) BBPA ( 6 6 %) BBPA ( 66 X ) BBPA (65.7 %) BBPA (52 8 %) BBPA (66.1 %) BBPA ( 66 5%) ( 33 %I PVC ( 33 %) PVC ( 33%) PVC ( 32.8 XI PVC (27.2 %I PVC ( 32.2 %) PVC ( 33.5 %) JSCN- -SCN- Fig.1. Selectivity factors KEY; for membranes based on a classical ion-exchanger (column a charged carrier (column 2) various neutral carriers (columns 3-6) and for a blank membrane (column 7) as determined by the separate solution method.MTDDACl methyl tridodecyl- ammonium chloride; HOCo(III)ACCP a,b,d,ef,g-hexamethyl c-octadecyl Co-aquo-Co-cyano- cobyrinate perchlorate; TBTC1 tributyltin chloride TOTC1 trioctyltin chloride; TOT tetra- octyltin ; BBPA bis( 1 -butylpentyl) adipate ; PVC poly(viny1 chloride). -so;- Clinically Relevant Sensors ELECTRICALLY NEUTRAL ANION CARRIERS -JNO; -NO; -HPO;- Selectivity Factors These were evaluated using the separate solution technique6 with 0.1 mol dm-3 buffered solutions (0.1 mol dm-3 tris/H,SO, pH 7.5) of the sodium salts [for membranes with MTDDACl (fig. 1) see ref. (1 5) and with the C o I I I cobyrinate (fig.1) see ref. (lo)]. Required potentiometric selectivity factors16 for blood and urine have been calculated using eqn (1) and (2) respectively where at low is the lowest value of the normal physiological activity range of the ion 1% to be measured aj low is the highest value of the normal physiological activity range of the interfering ion Jzj and p i j is the maximal tolerated error (%) caused by the ion Jzj and zi and zj are the charges of the ion I and J respectively where ai typical and aj typical are the typical physiological activities of ions I and J respectively. -HCO; -so,'- - 8 3 HP0:- BLANK ( 1.7 %) I -7 1% aa Fig. 2. Electrode function of a cell assembly with a membrane based on tributyltin chloride measured in aqueous NaCl solutions.1 wt % TBTCl 66 wt % BBPA 33 wt % PVC. I -6 \ -58.4 mV I I l l I I U. Oesch et al. I I I -4 -3 - 2 -5 - 3 -4 log a HCO Fig. 3. Electrode function of a cell assembly with a membrane based on tributyltin chloride measured in tris/HCl-buffered (0.0 1 mol dm-3 tris) NaHCO solutions containing a constant background of 0.1 mol dmP3 C1- ions. 33 wt % TBTC1 32 wt % DOS 35 wt % PVC. - 2 1183 1 1 - 1 0 mV I I - 1 Clinically Relevant Sensors 1184 ~cog.15~20 Results and Discussion There are three inherently different types of material that induce anion selectivity in solvent polymeric membranes (a) classical anion-exchangers (b) positively charged anion carriers and (c) electrically neutral anion carriers.The most widely used classical anion-exchangers (e.g. salts of quaternary ammonium ions which are predominantly active in the membrane phase in their dissociated form) induce a selectivity sequence which is given by the partition coefficient of the anions between aqueous sample phase and membrane phase (column 1 in fig. 1). For a more detailed discussion see ref. (15). Deviations C10 > SCN- > from I- > NO- the > Br- so-called > C1- > HCO; Hofmeister = OAc- x selectivity SO:- - HP0:- sequence can be expected by stabilizing a given primary ion via a selective interaction with a membrane component. Some success in this respect has been reported for C0:-l7 and p-toluene sulphonate18 as the primary ions.For C1- and HCO; however this type of sensor material will barely lead to a competitive clinical sensor in the long run since classical ion-exchanger-based sensors are likely to suffer from interference by lipophilic anions (see column 1 in fig. 1). This interference may be somewhat reduced by using more hydrophilic plasticizer~l~ or by eliminating them completely.20 The reduced selectivity for lipophilic anions unfortunately parallels some loss in the selectivity of C1- over Positively charged anion carriers (e.g. a,b,d,ef,g-hexamethyl c-octadecyl Co-aquo- Co-cyanocobyrinate perch1orate)'O which complex a given primary anion may induce selectivities in membranes which drastically deviate from the Hofmeister series (see column 2 in fig. 1). The selectivity pattern presented in column 2 of fig.1 is not attractive for a clinical application of the corresponding sensors. Recently however it was shown that lipophilic MnIII complexes of porphyrins21 may be interesting ionophores for Cl- sensors but they still suffer from interference by SCN-.21 On the other hand it is likely that a vigorous application of membrane technological design features will make it possible to meet the requirements in respect to speed of response e.m.f. stability and sensor lifetime. Organotin compounds are attractive candidates for electrically neutral carriers in anion-selective electrodesl29 l3 [see also ref. (22)-(24)]. Rather surprising differences in the induced selectivities are observed if trioctyltin chloride (TOTCl) tributyltin chloride (TBTCl) or tetraoctyltin (TOT) are incorporated into a membrane phase (see fig.1). Fig. 4 offers one mode of action of organotin compounds as neutral carriers for anions Y-. The behaviour of TBTCl and TOTCl is in line with the configurational unstability of most triorganotin halides in the presence of anions or nucleophiles such as pyridine (see fig. 4).25 The configurational stability of tetraorganotin compounds even in the presence of large quantities of nucleophiles underscores the fact that TOT does not induce anion selectivity in membranes (column 6 in fig. 1). Indeed studies by l19Sn n.m.r. corroborate an interaction of TOTC112 and TBTC113 with Cl-. When C1- is added to these trialkyltin chlorides large l19Sn chemical shift changes are induced whereas with TOT no such effect is observed.l21 1 3 9 26 The scheme in fig.4 further suggests that compounds of the type R,SnX where X is an electronegative substituent (e.g. F C1 Br I OH OCOR CN) may undergo exchange reactions leading to an equilibrium mixture of ion carriers in the membrane phase.13 The selectivity of a membrane may therefore depend both on the substituents R at the tin centre (see TBTCl and TOTCl in columns 3 and 4 of fig. 1) and on the sort and distribution of the electronegative substituents at the tin centres which in turn may depend on the composition of the sample s o l ~ t i o n ) . ~ ~ ~ 25 The selectivity may further depend on the total concentration of carrier in the membrane phase. In view of a clinical application of C1- sensors based on tinorganic compounds TBTCl perfectly meets the selectivity requirements when buffered sample solutions are used (see 1185 R' + Y - ~....l! R" U. Oesch et al. - y - - x - S Y-Sn' + x - Sn-X / R' \ R L,,. . I \ R R R ' R' %.. I_ + Y-Sn-X R Fig. 4. Scheme of the interaction of anions Y- with trialkyltin halides. table 2 and column 3 in fig. 1). The electrode response in unbuffered sample solutions exhibits an almost theoretical slope (fig. 3). The requirements in respect to lipophilicity clearly seem not to be met. Indeed sensors based on TBTCl show a comparatively poor e.m.f. stability. Furthermore a membrane [33 wt % TBTCl 32 wt % bis(2-ethylhexyl) sebacate (DOS) 35 wt % poly(viny1 chloride)] exposed for 3 days to a large volume of water loses 11% of its weight within one week.14 An increase in the alkyl chain length in order to improve the lipophilicity of the ionophore seems to be paralleled by a reduction in clinically relevant ion selectivities (see columns 3 and 4 in fig.1). Interestingly a considerable change in the ionophore concentration induces quite a different selectivity pattern (see columns 3 and 5 in fig. 1). The change in the selectivity for bicarbonate is so significant that it gets close to the clinical requirement for an assay of HCO; (see fig. 3). The antagonistic behaviour of the required properties makes membrane design an ongoing challenge. This work was partly supported by E.I. du Pont de Nemours & Co.(Inc.) and by the Schweizerischer Nationalfonds zur Forderung der wissenschaftlichen Forschung. References 1 H. B. Jenny D. Ammann R. Dorig B. Magyar R. Asper and W. Simon Mikrochim. Acta 1980,II 125. 2 J. D. Czaban Anal. Chem. 1985,57 345A. 3 U. Oesch P. Anker D. Ammann and W. Simon in Ion-Selective Electrodes ed. E. Pungor and 1. Buzas (Akaddmiai Kiad6 Budapest 1985) p. 81. 4 P. C. Meier D. Ammann W. E. Morf and W. Simon in Medical and Biological Applications of Electrochemical Devices ed. J. Koryta (John Wiley Chichester 1980) p. 13. 5 U. Oesch D. Ammann U. Schefer E. Metzger and W. Simon Proceedings of the International Symposium on Ion-Selective Electrodes June 10-14 1985 Shanghai People's Republic of China in press. 6 G. G. Guilbault R.A. Durst M. S. Frant H. Freiser E. H. Hansen T. S. Light E. Pungor G. Rechnitz N. M. Rice T. J. Rohm W. Simon and J. D. R. Thomas Pure Appl. Chem. 1976,48 127. 7 Wissenschaftliche Tabellen Geigy Yol. 2 (Ciba-Geigy AG Basel 1979). 8 W. E. Morf E. Lindner and W. Simon Anal. Chem. 1975,47 1596. 9 U. Oesch 0. Dinten D. Ammann and W. Simon in Ion Measurements in Physiology and Medicine ed. M. Kessler D. K. Harrison and J. Hoper (Springer Verlag Berlin 1985) p. 42. 10 P. Schulthess D. Ammann B. Krautler Ch. Caderas R. Stipanek and W. Simon Anal. Chem. 1985 57 1397. 11 P. Anker E. Wieland D. Ammann R. E. Dohner R. Asper and W. Simon Anal. Chem. 1981 53 1970. 12 U. Wuthier H. V. Pham R. Zund D. Welti R. J. J. Funck A. Bezegh D. Ammann E. Pretsch and W. Simon Anal.Chem. 1984 56 535. 13 U. Wuthier H. V. Pham E. Pretsch D. Ammann A. K. Beck D. Seebach und W. Simon Helv. Chim. Acta 1985 68 1822. 14 R. Zund Thesis ETHZ No. 7144 (Zurich 1982). 15 D. Wegmann H. Weiss D. Ammann W. E. Morf E. Pretsch K. Sugahara and W. Simon Mikrochim. Acta 1984 III 1. 16 D. Ammann P. Anker E. Metzger U. Oesch and W. Simon in Ion Measurements in Physiology and Medicine ed. M. Kessler D. K. Harrison and J. Hoper (Springer Verlag Berlin 1985) p. 102. Clinically Relevant Sensors 1186 17 J. A. Greenberg and M. E. Meyerhoff Anal. Chim. Acta 1982 141 57. Chem. 1980 35 1381. 24 W. Simon CH Patent 636451 1983. 25 M. Gielen in Top. Curr. Chem. 1982 104 57. Chemistry (formerly Org. Magn. Reson.) in preparation. 18 T. Fujinaga Philos. Trans. R. Soc. London Ser. A 1982 305 631. 19 K. Sugahara and J. Mori German Patent DE 3040269 C2 1984. 20 H. J. Marsoner Chr. Ritter and M. Ghahramani International Symposium on Ion-Selective Electrodes Shanghai People’s Republic of China summary of papers 1985 p. 36. 21 B. Krautler M. Huser P. Schulthess D. Ammann and W. Simon in preparation. 22 M. S. Frant and J. W. Ross Jr US Patent 3406102 1968. 23 V. A. Zarinskii L. K. Shpigun V. M. Shkinev B. Y. Spivakov and Y. A. Zolotov Russ. J. Anal. 26 H. V. Pham U. Wuthier E. Pretsch D. Ammann D. Welti and W. Simon Magnetic Resonance in Paper 5/ 1890; Received 16th September 1985

ISSN:0300-9599    

DOI:10.1039/F19868201179

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. SOC. Faraday Trans. I 1986,82 1187-1 193 Ion-selective Electrodes based on Siderophores Derek Midgley C. E.G. B. Central Electricity Research Laboratories Kelvin Avenue Leatherhead Surrey KT22 7SE Siderophores are natural products used by plants and bacteria to transport iron across cell membranes. One class of these the mycobactins are hydrophobic and were judged to be suitable for inclusion in the organic membrane phase of a liquid ion-exchange electrode responsive to iron(m) ions. When tested the electrodes had no response to iron but did respond to salicylate with a sensitivity of ca. - 29 mV per decade. The selectivity for salicylate over other anions was unusually large for a liquid membrane electrode and the effect of anion structure on the extent of interference was investigated with implications for the mechanism of the electrode.Siderophores (from the Greek sideros iron and phorein to carry) are substances produced by plants and bacteria to enable them to extract iron from their surroundings and transport it across their cell wa1ls.l Most siderophores are soluble in water but the mycobacteria have waxy coatings on their cell walls and produce hydrophobic siderophores known as myc0bactins.l Fig. 1 shows the structure of mycobactin S extracted from Mycobacterium smegmatis. Eight other mycobactins have been identi- fied,2 differing mainly in the lengths of side-chain (methyl and C, in mycobactin S) pro- truding from the left of the structure in fig. 1 the three chelating groups being identical in all cases.The biological function and hydrophobic properties of the mycobactins made them obvious candidates for inclusion in liquid-membrane ion-selective electrodes for iron(II1). The outcome of the attempts to make an iron-selective electrode however was an electrode with an unusually high selectivity for the salicylate ion. Response in Solutions of Iron(m) Membranes were made by dissolving mycobactin S in decan- 1-01 or dinonyl phthalate until it was almost saturated (ca. 1 % w/w). The solutions were placed in the ion-exchanger compartment of an Orion series 92 liquid ion-exchange electrode. Further experimental details are given el~ewhere.~ These electrodes gave no response to iron in solutions of iron(m) nitrate or iron(II1) chloride buffered at pH 4 (acetate-acetic acid) or pH 7 (Tris-HC1).Salicylate was then added to fill the role of exochelin2 (a class of water-soluble substances produced by the mycobacteria to facilitate extraction of iron from the environment). Although ineffective in natural media where it is not a strong enough chelating agent salicylate enables mycobacteria to take up iron from synthetic media in which the iron is not otherwise strongly complexed. With iron in excess over salicylate no satisfactory response was obtained from the electrode but in the presence of an excess of salicylate the e.m.f. decreased as the excess of salicylate increased indicating a response to a negatively charged species such as a salicylatoiron complex or salicylate itself. Further tests showed that the electrode was responding only to the free salicylate and not to the complex and work was concentrated on the response to salicylate in the absence of iron.1187 1188 I C H 2 5 d E Ion-selective Electrodes based on Siderophores NCH 2 CH I } 10 mV "+& sX1O3 [ Sal]/mol dm-3 Fig. 2. Calibration curves in background media with different pH values 0 pH 3 nitric acid; x pH 4 acetic acid-acetate; 0 pH 7; + pH 8 Tris-HC1; A pH 9 borax. l o 2 Response to Salicylate Ion Nernstian responses were obtained to total salicylate concentration as shown in fig. 2 for the membranes using decan- 1-01 as the solvent. The electrodes were the same as those used in iron solutions. The best sensitivity of - 29 mV per decade at 25 "C was obtained at pH 7 (Tris buffer) whereas with the dinonyl phthalate as solvent the best sensitivity of -30 mV per decade was obtained at pH 4 (acetate buffer).The electrodes responded quickly steady e.m.f. values being obtained in <2 min for twofold changes in concentration and the standard potential drifted at a rate of only PH [H Sal-] [H Sal] 50 50 3 (HNO3) 4 (acetate) 5.35 (acetate) 7 (Tris) 9 (Tris) 9.2 (borax) 7.3 0.35 10-2 10-3 10-4 93 99.6 100 100 1 00 a With respect to pH 7 (Tris). At 9.1 x mol dm-3. 30 mV per decade. Fig. 2. - 1 mV per day (after an initial rate of ca. - 5 mV per day over the first 24-48 h of use). Removal of an electrode’s inner aqueous reference solution (0.05 mol dm-3 sodium salicylate plus 0.15 mol dm-3 sodium chloride buffered at either pH 4 or pH 7) after several weeks revealed a faint pink colour indicative of a slight degree of extraction of iron from the membrane by formation of coloured salicylatoiron complexes.The consequent decrease in free salicylate inside the electrode explains the downward drift in standard potential. - 15 - 35 The Nernst equation D. Midgley Table 1. Effect of pH on salicylate speciation and e.m.f. proportion (% ) of total salicylate [Sa12-] 5 x 10-9 10-7 2 x 10-6 10-4 10-3 10-2 ZF RT In (10) log WI - 30 - 65 90 50 0 E = E ” + where z is the charge (with sign) on the ion X implies that the electrodes’ responses of ca. - 30 mV per decade were caused by doubly charged salicylate species.Table 1 shows however that the change in e.m.f. with pH is much less than predicted from protonation of the doubly charged salicylate ion even after allowance for the diminished slope at the more extreme pH values. Moreover the concentrations of the doubly charged ion at the lower pH values are much lower than the levels normally considered capable of detection by liquid-membrane electrodes. Changing from the pH 7 Tris buffer to a pH 7 buffer containing N-2-hydroxyethylpiperazine-N’-2-ethanesulphonic acid (HEPES) produced a shift of -20 mV which is more than could be explained by differences in activity coefficients in the two media. The questions of the slope factor pH response and medium effect will be discussed after a consideration of the electrodes’ selectivity.Selectivity No response was observed up to concentrations of 0.01 3 mol dm-3 of common inorganic anions such as sulphate phosphate nitrate and chloride nor from strong complexing agents such as EDTA citrate tartrate and oxalate. Other complexing agents interfered however in very specific ways according to their structure. Table 2 shows the interferences from substituted salicylic acids and the isomers of salicylic acid. The non-chelating isomer 4-hydroxybenzoate does not interfere at all suggesting that ability to chelate iron is essential if a substance is to interfere. The attachment of charged groups to the aryl ring of salicylate also results in non-interference although these compounds still form strong complexes with iron in aqueous solution.It is inferred that these additional charges 1189 e.m.f. changea predicted observedb with observed sloped with ideal slopeC 80 95 120 ,.. 17 14 0 58 23 0 - 22 - 52 1190 responseb substance - 0.18 0 Ion-selective Electrodes based on Siderophores Table 2. Response to salicylate isomers and substituted salicylates at pH 7 stability constantsa P I P3 mVper decade mV (mmol dm-3)-1 30 36.1 17.4 - - - ca. 10 14.4 82 28.1 - - ca. 23 25.2 0 0 salicylate 3-hydroxybenzoate 4-hydroxybenzoate 4-aminosalicylate 5-sulphosalicylate - 3-5 0 0 0 a Quoted as log 8 value (overall constant) from ref.(4H6). True response in Roman mol dm-3. approximate response in italic. All tested in the range 1 x 10-3-1.3 x - ca. 31 32.2 10-2 5 x l o 3 concentration/mol dm-3 Fig. 3. Response to interferences for mycobactin4ecanol membrane electrode in pH 7 Tris medium 0 SCN-; + resorcinol; A catechol; 0 SO;- and x 3-hydroxybenzoate. prevent the substituted salicylates from entering the organic phase and that for a substance to interfere it must be able both to enter the organic phase and then to chelate the iron in the mycobactin complex. 3-Hydroxybenzoate should have no more difficulty than salicylate in entering the organic phase but its relative inability to complex the iron makes it avery weak interferent. Fig. 3 shows that the interference from 3-hydroxybenzoate gives a linear response rather than the Nernstian log-linear form at least over the range of concentrations tested.The pseudo-log-linear response is included in table 2 for comparison. Interference tests with a wider range of substances confirm these trends but also 1191 PH4 PH 7 - 0.3 0.13 0 4 - 27c 10 6-9 3 0 0 0 0 D. Midgley Table 3. Response to interfering complexing agents responsea at substance mV per decade mV (mmol dmP3)-l mV per decade mV (mmol dm-3)-1 - 0.35 - - 0.76 0. 7c 0.25 0.64 0 I .3 2OC 5 I0 catecholb resorcinol h ydroquinone phthalate maleate thiocy anated 10-15 a Tested in the range 1 x 10-3-l.3 x lo- mol dm-3. True responses in Roman approximate responses in italic.Stepwise stability constants (logarithmic) from ref. (4)-(6) K = 15 K3 = 9.5. Overall stability These results are for electrodes with dinonyl phthalate as the organic phase. constants (logarithmic) from ref. (4)-(6) p1 = 3.1 a = 5.3 p3 = 6.2. indicate a still further degree of specificity. Fig. 3 shows that these interferents generally gave a directly linear response and the results in table 3 are shown in italics unless a definite linear or log-linear response was established. The weak interference from catechol is surprising since it complexes iron strongly in aqueous solution and seems no less likely than salicylate to enter the organic phase. The smallness of this interference indicates a very specific recognition by the membrane of salicylate-type groupings.The 1,3- and 1,4-isomers of catechol have respectively even weaker and zero interferences showing a parallel trend to the three hydroxybenzoic acids in table 2. Phthalate gave the most marked interference among the substances tested being the only one to have a Nernstian response (with the dinonyl phthalate membrane) but the pH dependence indicates that it is hydrogen phthalate (predominant at pH4) that interferes while the dianion (predominant at pH 7) does not. Maleate which has the same conformation of chelating groups as phthalate shows exactly similar trends but the interference is weaker. The absence of the aryl ring presumably makes extraction into the organic phase more difficult. Note that the complexing (but not chelating) and relatively lipophilic thiocyanate ion also interferes.Discussion The response to salicylate and the various interferences shows that the electrode responds only to complexing and lipophilic anions. The presence of additional charged or polar groups inhibits the response completely even though the iron is still complexed in aqueous solution. This makes the electrode much more selective for salicylate than the familiar quaternary ammonium liquid ion-exchange electrodes. An electrode with a membrane consisting of trioctylmethylammonium salicylate in decanol' had selectivity coefficients of 0.13-0.22 for 3-hydroxybenzoate and 0.4 for nitrate. An otherwise similar electrode using tetraheptylammonium salicylate8 was equally selective to other aromatic anions such as benzoate and 4-hydroxybenzoate.The mycobactin membrane electrode is therefore much more selective than conventional liquid anion-exchange electrodes. This selectivity may be compared with the behaviour of mycobacteria which accumulate salicylate but not other phenolic acids or phenols9 and which assimilate 4-aminosalicylate only slightly.1° FAR 1 40 1192 (2) k k (3) Ion-selective Electrodes based on Siderophores The anion sensitivity of the electrode requires that the membrane contains a relatively immobile cationic species.ll Candidates for this species include a protonated ferrimyco- bactin complex or a cation (other than a proton) extracted from the aqueous test and filling solutions The extracted-cation hypothesis gives some thermodynamic grounds for the differences in potential in buffers of the same pH although kinetic (diffusion) factors could also be involved.In the tests done so far the only cations available for extraction were those in the buffer used for pH control (Na+ or Tris H+) or the sodium from the salicylate standard solutions. With the inner surface of the membrane considered to be in a constant state the boundary potential may be written as where [Sal],, is the concentration of salicylate in the membrane immediately adjacent to the surface and [Sal] is the bulk aqueous concentration. Electroneutrality requires that [Sal],, equals the concentration of extracted cation at the surface [C+],,,. The extraction equilibrium has a constant K = [SalIaq [C+Iaq a [Sallorg [C'lorg Hence [Sal]Xrg = K[Sal], [C+],,.On substitution in eqn (1) = E" - log [Sal], +- log K+- k log [PI, 2 2 2 where k = RT In (lO)/F. In eqn (3) the prelogarithmic term of k/2 accords with the observed slope factor of ca. -30 mV per decade for salicylate and the log K term provides a qualitative explanation for differences between buffers (in the absence of extractability data). It is expected that K for the large organic Tris H+ ions in Tris buffer would be greater than that for the sodium ions in HEPES buffer and that a change from the former to the latter would produce a decrease in e.m.f. The observed shift was -20 mV. The log[C+], term in eqn (3) predicts that halving the cation concentration (at constant pH and [Sail,,) should produce a shift of - 8.7 mV.When tested with Tris buffer at pH 7 halving the concentration produced a shift of only - 2 mV but the assumption of relative immobility for the extracted cation implies that equilibration times should be much longer than the measurement time of a few minutes. Changing from pH 7 to pH 8 at constant total Tris and total salicylate should produce a shift of -6 mV compared with the - 15 mV observed. The protonated mycobactin hypothesis leads to equations similar to eqn (2) and (3) except that [C'], = [H+], and [C+],, E [FeMH+],, where M represents mycobactin. The same dependence on log[SalIaq is predicted as before but also with a 30 mV per decade dependence on pH which is greater than was observed (table 1).The log K term in this case should be independent of the buffer and this hypothesis does not account even qualitatively for this aspect of the observations. That the observed medium effects are intermediate between the predictions of the two hypotheses may indicate a hybrid mechanism. Note that in a valinomycin-based potassium-selective electrode the 'fixed ' anions in the membrane consisted of chloride ion extracted from the test solution and hydroxide ions produced by hydrolysis.12 Both 1193 D. Midgley mechanisms however are successful in predicting the unusual -30 mV per decade response to the singly charged salicylate anion. Some light may be thrown on the failure of the electrode to respond to iron by consideration of iron uptake by mycobacteria.It has been shown13 that uptake can be followed on a timescale of tens of seconds and that transfer of iron from the surrounding aqueous medium to the cell wall does not involve enzymic processes neither of which suggests problems in making an ion-selective electrode. The difficulty seems more likely to arise during the removal of iron from the mycobactin complex. In the living cell the release of iron involves an NADH-linked reductase that reduces iron(Ir1) to iron@) in the complex.13 The weakly bound iron(@ can then be removed from the complex without difficulty and iron transport is facilitated. Such a mechanism is not readily available in an ion-selective electrode. Although a chemical or enzymic reductant could be included in the inner filling solution problems of exhaustion interference with the inner reference electrode and accumulation of iron could be expected.I thank Prof. C. Ratledge of the University of Hull for providing the mycobactin. This work was carried out at the Central Electricity Research Laboratories and is published by permission of the Central Electricity Generating Board. References 1 J. B. Neilands Annu. Rev. Biochem. 1981 50 715. 2 C. Ratledge in The Biology of Mycobacteria ed. C. Ratledge and J. Stanford (Academic Press London 1982) vol. 1 p. 186. 3 D. Midgley Anal. Chim. Acta in press. 4 A. E. Martell and L. G. Sillin Stability Constants (Special Publ. no. 17 The Chemical Society London 1964). 5 A. E. Martell and L. G. Sillin Stability Constants Supplement No. I (Special Publ. no. 25 The Chemical Society London 1971). 6 D. D. Pemn Stability Constants of Metal-Ion Complexes Part B. Organic Ligands (Pergamon Oxford 1979). 7 H. James G. Carmack and H. Freiser Anal. Chem. 1972 44 856. 8 W. M. Haynes and J. H. Wagenknecht Anal. Lett. 1971,4491. 9 C. Ratledge and F. G. Winder Biochem. J. 1962 84 501. 10 K. A. Brown and C. Ratledge Biochim. Biophys. Acta 1975 385 207. 11 W. E. Morf The Principles of Ion-selective Electrodes andof Membrane Transport (Elsevier Amsterdam 1981) p. 274. 12 A. P. Thoma A. Viviani-Nauer S. Arvanitis W. E. Morf and W. Simon Anal. Chem. 1977,49 1567. 13 C. Ratledge and B. J. Marshall Biochim. Biophys. Acta 1972 279 58. Paper 5/1891; Received 3rd October 1985 40-2

ISSN:0300-9599    

DOI:10.1039/F19868201187

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. SOC. Faraday Trans. I 1986,82 1195-1208 Solid-state Ion Sensors Theoretical and Practical Issues Robert G. Kelly* and Alan E. Owen Department of Electrical Engineering University of Edinburgh Edinburgh EH8 9YL Integrated ion-sensor devices may be constructed using the methods of microcircuit fabrication. The suitabilities of the silicon integrated circuit and thick-film hybrid circuit processes for this purpose are compared and it is argued that the hybrid approach will offer advantages in many cases. Theoretical questions about the operation of both the hybrid and ISFET (ion-sensitive field-effect transistor) types of device are discussed and possible origins are suggested for the output drift that is commonly observed in ISFET. Practical issues which might affect the commercial development of integrated ion-sensor devices are noted.The remarkable progress in the technology of integrated circuits over the past 10-1 5 years has led to the availability of highly sophisticated measuring instruments at relatively low cost. Improvements to the signal conditioning and data processing electronics have not generally been paralleled in the ‘sensors’ or ‘transducers’ which are used to convert the measured variable into an electrical signal. As a consequence the sensor is now seen as the ‘weak link’ in many measurement systems.l We may identify two broad ways in which the methods of microelectronics might be used directly or indirectly to improve sensor performance. First the availability of sophisticated data processing capability at low cost can be used to compensate for the inherent deficiencies of sensors e.g.by storing a suitable calibration curve as a ‘look-up table’ in an electronic memory and using it to compensate automatically for sensor non-linearity . Zero-drift problems can also be reduced by using auto-calibration methods under microcomputer control but it should not be forgotten that this requires a suitable stable standard against which the instrument can calibrate itself; data processing methods alone cannot compensate for random unpredictable drift in the sensor. Microelectronics can be used in this manner in conjunction with a sensor of conventional type. The sensor need not be physically incorporated into the microelectronic device although further benefits might be gained if it is.Secondly the manufacturing processes that have been developed for the fabrication of microelectronic devices might be used directly to produce new kinds of sensor which incorporate the improvements in performance and cost that are already evident in electronics. It is an obvious corollary of this approach that the sensor and its associated electronics will be fabricated as a single integrated structure. It is with the latter approach to the sensing of ions that this paper is concerned. Microcircuit Fabrication Methods The technologies of both silicon integrated circuits (SIC) and thick- or thin-film ‘hybrid’ microcircuits are applicable to sens0rs.l Silicon circuits are fabricated by the diffusion (or implantation) of donor and acceptor impurities into a monocrystalline semiconductor wafer using photomasking to define the required p- and n-type regions.Insulating films usually of silicon oxide or nitride 1195 1196 Solid-state Ion Sensors may be grown by reaction of the silicon substrate with appropriate gases or may be deposited e.g. by chemical vapour deposition (c.v.d.). Conductors comprise deposited films of metal or polycrystalline silicon. SIC processes produce very thin films typically hundreds of nanometres in thickness. Extreme miniaturisation has always been an essential feature of SIC technology. The requirement for small devices was originally imposed by the existence of defects in silicon wafers; the larger the individual device the greater the probability of it incorporating a defect leading to a lower ‘yield’ of working circuits (or ‘chips’) from the wafer.Nowadays the main driving force for further reductions in feature sizes arises from the demand for microcomputers and similar devices of greater and greater sophistication and complexity. Considerations of cost reliability circuit operating speed and power consumption determine that the greatest possible functional capability must be imple- mented on each chip. In order to achieve micron or sub-micron feature sizes SIC processes have employed increasingly sophisticated and expensive manufacturing meth- ods. The high capital cost of the equipment can only be justified by optimising its use through the production of ‘ standard ’ devices in very large volumes.Modifications to a standard silicon process are expensive e.g. the introduction of an additional photo- lithographic step or even more so the incorporation of non-standard materials like chemically sensitive films. There is a marked contrast between what is technically possible on a research scale and what will be economically practicable in terms of a commercial silicon fabrication plant. The low unit cost of SIC devices is achieved by fabricating many circuits simultaneously on each silicon wafer. The final stages of the production process involve the division of the wafer into individual circuit chips which are then encapsulated in a ‘package’ which provides environmental protection and a means for making electrical connection to the device.The packaging processes contribute heavily to the total cost of the device because they involve the handling of individual chips rather than wafers. Also because packaging involves mechanical and manual processes the dimensions of the packaged device are large compared with the on-chip feature sizes which are determined by optical processes. The hybrid microcircuit technologies are less well known than silicon but they are widely employed as a method of achieving reliable interconnection between SIC chips and for the implementation of precise and stable resistor networks. A hybrid circuit is fabricated on an inert substrate of ceramic or glass. Conductor and resistor patterns are deposited by vacuum evaporation of thin metallic films or in the case of thick-film circuits by using a screen printing process followed by treatment at elevated temperatures to ‘dry’ and subsequently ‘fire’ the deposited films which are typically 20 pm thick.Active devices (transistors or SIC chips) may then be bonded to the conductor areas of the device. The hybrid circuit technologies may be regarded as being intermediate between conventional (printed circuit board) electronic circuits and SIC. Less expensive equipment is required than for SIC fabrication and the processes are economic for medium-volume production. Conductor and resistor line widths are much larger than for SIC (typically 250 pm for thick-film circuits) so less precise photolithographic equipment is acceptable although component densities are much smaller.The require- ments for chemical purity are less stringent than for semiconductor circuits and the processes can be more readily modified. Properties and Performance of Ion Sensors The characteristics of ion-sensing devices can be considered under two broad headings. The first covers those aspects of performance that are due to the chemical properties of the ion-sensitive material itself e.g. sensitivity and selectivity to a specific ion resistance to poisoning by materials which might be present in the analyte and problems resulting from chemical processes such as the leaching out of the electroactive material or the R. G. Kelly and A . E. Owen 1197 requirement for hydration or ‘ conditioning’ before use (by exposure to solutions containing the ion to be measured).These matters are familiar from experience with conventional membrane ion-selective electrodes (ISE). The second heading encompasses properties that are mainly related to the physical structure of the sensor device e.g. its size and mechanical robustness or to its electrical characteristics especially its output impedance which can introduce problems of noise and poor response time. The direct benefits of applying microelectronic methods to sensor fabrication will be mainly in this second category. More speculatively improvements to the sensor materials themselves might follow from improved control of film properties and dimensions or by allowing the use of ultra-thin films of electroactive material which could not be fabricated by conventional means.From the user’s point of view an ‘ideal’ ion sensor should produce an easily handled electrical output (i.e. an analogue signal from a low output impedance source or better a digital word suitable for directly interfacing to a display or computer). The output should be uniquely dependent on the activity of a single species in the analyte and be unaffected by other species total ionic strength or temperature. The device should be small requiring the minimum possible volume of analyte material and its output should respond rapidly to changes in the activity of the measured ion. It should be highly reliable physically robust usable over a wide range of temperature and pressure and require only occasional calibration. Finally it should be inexpensive.The relative importance of these properties depends on the particular application. We may contrast the in vivo measurement of blood pH during surgery with the monitoring and control of pH in an industrial process plant. In the former case the device must be small enough to be housed in a catheter for insertion into the blood vessel and it must not induce adverse biological reactions. It must be able to withstand sterilisation before use. On the other hand it need only have a limited lifetime being discarded after use and hence a short recalibration interval would be acceptable but it must be inexpensive. In many cases the clinically significant activity variations are large so that high sensitivity is not essential but the immediate availability of data ‘on-line’ is of immense assistance to the surgeon and cannot be offered by conventional laboratory analysis.In the industrial application however the size of the sensor might not be critical and it might well represent only a small proportion of the cost of the whole measurement system. Reliability and ruggedness would be essential however and the necessary frequency of recalibration would significantly influence maintenance costs. The adoption of microelectronics fabrication methods can contribute to improvements in many aspects of sensor performance. However it will not of itself eliminate those problems that originate from the chemical nature of the electroactive material or that are associated with the chemistry of the analyte solution. These aspects can be especially troublesome when measurements are to be made outside the laboratory by unskilled operatives.Close liaison between research workers and users is required to identify how the particular strengths of the microelectronic approach can best be exploited in appropriate applications. On the other hand we must avoid being too constrained by the perceived requirements of current users of conventional ISE. The cost reduction which can be achieved by microelectronics in volume production can open up hitherto unidentified markets. It is not many years since the microprocessor was being described as ‘a solution in search of a problem’; today it is central to the rapidly expanding field of information technology. The Application of Microcircuit Technology to Ion Sensors In order to retain the cost and performance benefits of microelectronics technology the processes required to fabricate the sensor should follow as closely as possible the stages used in the production of ordinary circuits.The sensor device (whether it is of the SIC 1198 Solid-state Ion Sensors or hybrid type) will therefore incorporate a film of electroactive material deposited on a planar substrate in close proximity to standard electronic circuitry for signal conditioning. The use of an entirely solid-state structure in place of the conventional filling solution and inner reference electrode has obvious practical advantages in respect of physical ruggedness and ability to withstand wider ranges of temperature (e.g. for sterilisation) and pressure.(Theoretical issues concerning the electrochemical nature of the solid-state devices are discussed later.) The most obvious characteristic of microcircuit devices is their small size. This is clearly essential for in vivo applications and can also be advantageous in conventional laboratory situations if it is desired to use small volumes of the analyte. We must be careful though to distinguish the size of the active area from that of the entire device. Using SIC technology for example it is possible to define feature sizes of the order of 1 pm whereas 250 pm is typical for a thick-film hybrid device. The size of a completed device however is determined by the requirement to make electrical connections to it and to encapsulate it and there is little difference between the sizes of typical SIC and hybrid devices in their final packaged form.Furthermore the feature sizes noted above refer to dimensions on the surface of a two-dimensional planar device. The fabrication of three-dimensional (probe type) structures with minimum tip size involves substantial modifications to standard circuit-processing procedures. For many sensor applications extremely small active areas might even be a disadvantage e.g. where fouling by solid matter is a problem. For sensor purposes in general therefore the minute size of microelectronic devices is not in itself their main advantage. It is the indirect benefits of small size such as improved reliability and reduced cost which will be more significant. From these points of view the hybrid processes can offer a viable alternative to SIC methods and their thicker films might well be advantageous from the point of view of durability.The incorporation of pre-amplifier circuitry giving reduced noise improved response time and greater user convenience and refinements such as in situ temperature sensing and compensation and the use of multi-sensor arrays are possible using either approach. Commercially the relative advantages of the silicon and hybrid technologies will be heavily influenced by the prospective volume of production. The introduction of variations into standard silicon processes in particular will only be justified for very large markets. It has been shown that the silicon and hybrid technologies offer alternative and complementary approaches to solid-state ion sensors but relatively little work has been done on the hybrid method (although it has been extensively used for gas sensors).2 The most widely researched type of microelectronic ion sensor has been the ISFET (ion- sensitive field effect transistor) which uses SIC te~hnology.~ ISFET development has been directed mainly towards biomedical applications.ISFET Devices A cross-section through a typical ISFET is shown in fig. 1. It is essentially an insulated gate field-effect transistor (IGFET) in which the metal gate electrode has been replaced by a film of an ion-sensitive material (ISM). The device is encapsulated to protect it from the analyte solution. A window in the encapsulation allows the solution to contact the electroactive material adjacent to the channel region of the device which lies between the diffused source and drain contacts.(The practical problems of encapsulation will be discussed later.) A reference-electrode system is used to define the electrical potential of the silicon substrate with respect to that of the analyte solution. As in a conventional ISE electrochemical processes take place at the ISM-solution interface establishing across it an electrical potential difference which is a function of the activity of the measured ion. The sum of this potential and the reference electrode half-cell potential together with other potentials which are substantially independent of t- ‘d R . G. Kelly and A . E. Owen analyte solution encapsulant p-type silicon substrate Fig.1. ISFET cross-section. ion activities (e.g. liquid junction potential metal-silicon contact potential) is developed across the insulator-silicon interface and hence modulates the concentration of mobile charge carriers in the ‘channel’ region close to the silicon surface. Variations in channel conductivity modify the drain-source current-voltage characteristics of the transistor as in a conventional IGFETS4 Hence the drain current (Id) serves as a low impedance output signal related to solution ion activity. Two main schools of thought have evolved in ISFET development in recent years. One approach has been to employ ISM of the kinds previously used in conventional ISE; a macrocyclic ion carrier like valinomycin dissolved in PVC is a typical example.ISFET sensitive to a range of ions have been fabricated in this but some difficulties have been experienced owing to poor adhesion of organic films to the underlying substrate.* The second approach has investigated the use of inorganic films e.g. of silicon nitride or aluminium oxide as sensors principally for P H . ~ ~ lo These materials can be deposited on to the ISFET substrate by processes like chemical vapour deposition which appear to be more compatible with the conventional silicon wafer processing environment. Devices of this kind are innovative in both of the broad aspects of sensor design identified above i.e. they employ novel ion-sensitive materials in a new type of device structure. Furthermore as will be described later it has been necessary to invoke different mechanisms to explain their ion-sensitivity.It is not always clear from the published work whether a particular aspect of the performance of these devices is a consequence of the ISM used or of the device structure; the commonly observed phenomenon of long-term drift is an important example. 1199 Solid-state Ion Sensors 1200 Hybrid Devices The possibility of implementing a high-input-impedance amplifier as a hybrid circuit sufficiently small to incorporate into a conventional ISE probe has been recognised by several l2 The difficulties associated with the transmission of high impedance level signals through a metre or more of cable to a remote voltmeter can thus be minimised.In those cases where the ISE itself is of the ‘all solid-state’ variety (e.g. employing direct metallic connection to LaF or Ag,S/AgI crystals) the device enjoys the further advantages of small size and entirely solid-state construction.13 None of these devices can strictly be described as ‘ integrated ’ sensors however because the ion-sensitive membrane itself is fabricated separately by conventional methods not as an integral part of the circuit fabrication process. A short connecting lead is used between the membrane itself and the thick-film amplifier circuit which may therefore be readily produced using any standard hybrid circuit process. Other workers have addressed themselves to the fabrication of a more fully integrated type of hybrid sensor in which the sensor material itself is deposited directly on to the substrate using thick-film methods.l** l5 This approach involves some modification of the standard microcircuit process as is the case with the ISFET.For the reasons suggested previously however it is felt that such modifications are more likely to prove commercially acceptable with the hybrid technology than with an SIC process. A thick-film process dedicated to sensor fabrication could be a practicable proposition. In our own laboratory we have focussed our attention on the use of pH-sensitive glass as the ion-sensitive material. The measurement of pH still accounts for a major part of the ISE market especially for industrial applications. The thick-film process appears to be an appropriate method for depositing glass films because conventional thick-film conductor and resistor materials are themselves based on glass matrices.Theoretical questions about the nature of a solid-state contact to an ion-sensitive glass will be considered later. A basic experimental form of hybrid sensor is shown in fig. 2. A conducting contact is first deposited on to the inert substrate by printing and firing a standard thick-film conductor composition. An ‘ink’ or ‘paste’ containing the ion-sensitive glass is then prepared by grinding the glass to a particle size of ca. 10pm and dispersing it in an organic binder material which has flow properties appropriate to the screen printing process. The glass film is then printed and dried to remove the organic solvents.(Several prints might be used to give a film of adequate thickness and free of pinholes.) The glass is then fired at a temperature at which appreciable flow occurs and then cooled slowly through its annealing range. Electrical contact is made to the conductor film and the contact area is protected from the analyte solution by some form of encapsulation. Structures of this kind are being used to develop suitable techniques for film deposition and to evaluate the ion-sensitivity of the deposited films and the effect of different contact materials. For practical devices signal conditioning electronics will be implemented on the same substrate using standard circuit processes. The first requirement of a practical thick-film process is that the substrate and the various printed films must have compatible coefficients of thermal expansion over the wide ranges of temperature used in the firing process.The development of cracks in the glass film owing to thermal stress destroys the sensor by allowing direct contact between the analyte solution and the conductor film. Good adhesion and chemical compatibility between the constituent films is also necessary; difficulties can arise from the diffusion of materials notably silver from the conductor into the ion-sensitive glass during firing. The composition of the ‘ink’ containing the ion-sensitive glass has to be designed (in conjunction with the mesh size and thickness of the printing screen) to give a printed pattern of well defined shape and appropriate uniform thickness free of occlusions.Adverse effects on the ion-sensitivity of the resulting film must be prevented. These can 1201 encapsulant analyte solution output * contact - encapsulated R. G. Kelly and A . E. Owen I pH-glass / contact Fig. 2. Experimental hybrid ion sensor (a) cross-section (b) plan view. arise for example from the dissolution of sodium out of the finely ground glass into the medium or by the incomplete removal of organic residues after firing. The peak firing temperature and the heating and cooling rates are also important. At the time of writing we have developed a process to produce visually satisfactory and pin-hole free films of a simple soda-lime pH glass having the composition of ‘ Corning 01 5 ’. The pH-sensitivity of these sensors varies greatly however even between devices fabricated under nominally similar conditions.Nernstian responses have sometimes been observed but the ‘yield’ of satisfactory devices is poor. An investigation into these problems is now in hand. We note that these difficulties are principally a consequence of the choice of glass as the ion-sensitive material but continuing importance of glass electrodes in the industrial sphere noted previously suggests that the development of a glass based thick-film sensor will be of value. Practical Issues In the previous sections we have set out to contrast the SIC and hybrid circuit approaches as alternative routes to the commercial production of integrated ion-sensor devices. There are certain practical matters that are relevant to both approaches however specifically concerning (i) encapsulation and (ii) the requirement for an appropriate reference electrode sys tem.An integrated ion sensor must be encapsulated in a way which allows the active sensor area to be exposed to the analyte solution while at the same time preventing contact between the solution and the ‘electronic’ parts of the sensor. Apart from the obvious possibility of damage to the device owing to corrosion there is the requirement to maintain extremely high levels of electrical insulation at the input stages of the electronics consistent with the generally high electrical resistances of sensor membranes. The problems involved are much more severe than those of packaging conventional microcircuits which can be totally ‘potted’ and are seldom exposed directly to such an unfavourable environment.Most of the experimental sensor devices reported in the literature have used organic encapsulants e.g. epoxy resin silicone rubber or polyimide. Their application (including the definition of the window exposing the active area) has 1202 Solid-state Ion Sensors analyte solution sensor body -. substrate - Fig. 3. Self-encapsulating hybrid sensor. been by hand in most cases. Degradation of the encapsulation is commonly believed to be the reason for the relatively poor lifetime of many devices. The problems are especially severe with ISFET because of the need to make electrical connection to the source and drain which are on the same face of the chip as the exposed sensing area and in close proximity to it.The delicate bonding wires are very vulnerable to damage when applying epoxy or similar material to the chip. Methods have been proposed to circumvent this difficulty by providing a conducting path through the silicon substrate so that external connection can be made on the reverse side of the chip.161 l7 These methods require extensive modifications to standard SIC processes however and will incur corresponding cost penalties. It has previously been pointed out that packaging contributes heavily to the cost of standard SIC devices because it involves operations on individual chips rather than complete wafers. Any additional costs at this stage resulting from the special requirements of packaging a sensor device will weigh heavily against the economic viability of the ISFET.Some workers in the ISFET field consider that methods of encapsulation have now been developed into a form suitable for quantity production but they have not yet been employed commercially to our knowledge. The encapsulation of hybrid sensors presents similar problems but the thick-film process will be more adaptable to the special requirements of the sensor device. One approach currently under investigation is the incorporation of a metal filament into the ceramic substrate to provide electrical connection through it. The pre-amplifier circuitry and external electrical connections will then be implemented on the opposite side of the substrate from the sensor material as shown in fig.3. The sensor material will then be self-encapsulating completely covering the contact metal. Both ISFET and hybrid sensor devices like conventional ISE must be used in conjunction with a reference electrode to complete the electrical measurement circuitry. (Confusion about this matter which persisted for several years after the introduction of the ISFET has now been resolved.) For applications in which the analyte solution contains an appropriate ion of known and constant activity it is sometimes possible to use a second sensor selective to this ion as a reference. Both sensors could be integrated in a single solid-state device. A ‘differential’ hybrid sensor based on this principle has been rep0~ted.l~ In many cases however a reference system of constant potential is required and is traditionally provided by a single or double liquid junction system.At the present time there is no proven method of implementing a satisfactory reference 1203 R. G. Kelly and A . E. Owen system of constant potential in a solid-state form compatible with a solid-state ion sensor. (The problem has been recognised however and some preliminary work has been done.)187 l9 The requirement to use a conventional reference system will greatly restrict the benefits to be gained from the solid-state sensor for many applications. 22 26 Theoretical Issues The relationship between the various kinds of ‘ microelectronic’ ion-sensors and conventional ‘membrane’ and ‘coated-wire’ ISE has been examined in a recent review by the present authors.20 The electrochemical processes that take place at the interface between the ISM and the metal (or semiconductor) contact in the various kinds of solid-state sensor have excited controversy for many years.In particular the possible role of ‘blocked’ or ‘polarised’ interfaces (which do not support a charge transfer or ‘Faradaic’ process) has been widely debated. The matter has practical as well as theoretical significance because it relates closely to the question of how reproducible and stable these devices can be expected to be In practice it is widely (if not universally) accepted that the standard potentials of coated wire electrodes (CWE) and the corresponding parameter for ISFET are not closely reproducible even for devices fabricated under nominally similar Drift of these parameters2l9 23 during the device’s operating lifetime also appears to be greater than for membrane electrodes.The practical utility of the devices has been strongly defended however and it has been pointed out that even membrane electrodes require regular calibration in a single standard solution at least and that the magnitude of the zero-correction required is then of little consequence. The hybrid sensor of fig. 2 and 3 is electrochemically identical to the ‘coated-wire’ type of ISE in which direct metal connection is made to one side of the ion-sensitive membrane. The precise nature of the charge exchange processes which take place at the metal contact in a CWE are not well established but they are usually assumed to be non-blocking and several possible mechanisms have recently been reviewed by Cattrall and Hamilton.21 The drift and poor reproducibility of the standard potentials of CWE would be an expected consequence of an inadequately defined contact process.CWE usually employ polymeric sensor membranes but we have observed generally similar behaviour with glass pH-electrodes that have a direct metallic contact to the inner surface of the Their pH-sensitivities have quite reproducible and nearly Nernstian values but their standard potentials vary widely. The standard potential of an individual device has been observed to vary by < 11 mV over several weeks under well controlled conditions of use but variations in the temperature of the device or the passage of small electrical currents through it can initiate prolonged drift of the standard potential.Various attempts have been made to improve the stability of the solid-state contact to glass membranes e.g. by using intermediate films of the oxide or halide of the contact It is the view of the present authors that much more research will be required to elucidate the operating mechanisms of both CWE and hybrid sensors but we believe that the best approach to solid-state sensors lies in the implementation of a suitably well defined half-cell in solid-state form to create a stable non-blocked contact to the ISM. In the case of the ISFET (fig. I) the presence of the gate insulator material implies that the internal interface to the ISM is blocked and hence the troublesome question of its standard potential does not arise.We shall show however that a detailed examination of the electronic and electrochemical structure of the device raises fundamental questions about ISFET operation which suggest other possible sources of instability. Fig. 4 (a) represents a conventional metal gate IGFET.4 The gate-substrate bias voltage (V,,) is supplied by a battery and is essentially developed across the gate insulator. The structure is equivalent to a capacitor in which equal and opposite charges are induced 1204 "P insulator (Si3N4) \ I / - I t metal gate (electronic charge) electrical potential I reference electrode / analyte solution liquid junction \ substrate contact (electronic charge) substrate contact (electronic \ charge) Solid-state Ion Sensors enwasulant I- I insulator (Si3N4) / I asforlGFET I Fig.4. Operation of (a) IGFET and (b) ISFET devices. on the metal gate and at the silicon surface. In the case of the semiconductor the charge is diffuse and its surface conductivity is a sensitive function of the concentration of mobile charge carriers. The conductivity measured between the source and drain contacts therefore depends on GS. The electrical potential varies linearly with distance through the charge-free insulator (constant electric field) but varies non-linearly within the semiconductor because the concentration of electrons and holes at any point depends on the local electrical potential.R. G. Kelly and A . E. Owen 1205 Fig. 4(b) represents a typical ISFET in which an ion-sensitive membrane is deposited on top of a silicon nitride insulator layer which is assumed to represent a perfect barrier to both electronic and ionic conduction between the ISM and the silicon surface. By analogy with the metal gate device we can see that the electrical field in the insulator and the silicon surface potential (and conductivity) will depend on the electrical potential difference that is established by electrochemical effects between the substrate contact and the surface of the ISM adjacent to the nitride insulator. Note that conditions within the insulator and at the silicon surface are then identical to those in the IGFET; electrochemical processes are excluded due to the ion-blocking function of the nitride and the electric field varies linearly with distance through the insulator.We note here that an alternative model for ISFET operation has been proposed in which ionic species originating in the analyte solution diffuse through the insulator to the silicon surface where they directly modify the surface conductivity e.g. by creating local surface states or traps for electrons or (A sensor functioning in this way would not comprise a ‘ field-effect ’ device of course.) However the observed response times of most ISFET are too fast to be compatible with the diffusion process implied by this although it might be present as a second-order effect and could contribute to output drift.The electrical potential at the ISM-insulator interface in fig. 4(b) may be calculated by considering the electrochemical process at each interface between the silicon and the ISM going via the ‘external’ path through the reference electrode. Potentials at the semiconductor-contact interface at the reference electrode and across the liquid junction will be substantially independent of ion activity in the analyte. The essential sensing process arises at the solution-ISM interface through ion-exchange or absorption mechanisms selective to the measured ion. In addition however a diffusion potential can occur inside the ISM which is usually permeable to the ions involved in the sensing process and will contribute to the potential at the ISM-insulator interface.There is an important difference in this respect between the ISFET and a conventional membrane electrode with an internal filling solution. A glass electrode for example can be regarded as an ion-exchange membrane with fixed sites. From the theory developed by Eisenman and coworkers in the 1960s it can be shown that the total potential across a glass membrane is the sum of phase-boundary potentials at each hydrated glass-solution interface together with a diffusion potential in the bulk membrane.29 Now the potential of a glass electrode attains a new steady value quite quickly after the solution pH is changed and yet the diffusion processes would not be expected to reach a steady-state condition within the timescale of normal measurements.It has been shown however that the diffusion potential is determined entirely by the boundary conditions of ion concentrations in the bulk glass just inside the hydrated surfaces. These boundary conditions are in turn determined by the phase-boundary ion-exchange processes which are assumed to reach equilibrium rapidly following a change in solution activity. The diffusion potential can therefore attain a steady value even though the ionic concentrations and electrical potential profile within the membrane continue to change.30 In the case of an ISFET with a glass sensing layer however the internal surface of the glass is in contact with an ion-impermeable film of silicon nitride. Ion concentrations inside this surface of the glass film are not stabilised by a rapid exchange process.Hence if these concentrations differ between devices (e.8. owing to varying degrees of sodium loss during device fabrication) or if they change in use as a consequence of diffusion within the bulk membrane or accidental polarisation of the device for example the total membrane potential would be subject to irreproducibility or drift as a result. A similar conclusion can be inferred from the theoretical analysis of polarisable electrodes by Lauks31 who has shown that the electrical potential at the inner surface of the ISM is a function of the electrochemical potentials of ions inside the ISM phase. We also note that a device of this kind is not entirely analogous to a conventional 1206 Solid-state Ion Sensors IGFET as has sometimes been suggested.In the metal gate device [fig. 4(a)] the external gate bias battery incorporates two half-cells which determine the equilibrium potential between electrons in the silicon and electrons in the gate metal. In the ISFET case [fig. 4(b)] however the potential at the inner surface of the ISM is associated with an overall state of equilibrium between electrons in the silicon and ions in the ISM. As Lauks has observed ‘the ISM in contact with the solution cannot be considered as an electrode’.31 It was noted previously that an interesting aspect of ISFET work has been the discovery that films of materials like silicon nitride or aluminium oxide can function as pH sensors. Because these materials are known to be effective barriers to penetration by ions it has been necessary to postulate a mechanism for the potential determining process that depends entirely on selective surface adsorption effects at the solution interface which is assumed to be ‘blocking’ or ‘ polarisable’.The mechanism is based on the ‘site binding model’ of colloid chemistry.32 The model assumes that the interior of the ISM is free of net electrical charge i.e. that the electric field is constant through the ISM as is the case in the gate insulator of a conventional IGFET. The assumption that ions are excluded from the interior of the ISM appears to circumvent the problem discussed above concerning the definition of the electrochemical potentials of ionic species in the bulk membrane. The electrical potential within the ISM is purely a matter of electrostatics and the problem of instability as a consequence of a varying diffusion potential does not arise.However the use of a blocked interface as a model for the sensing mechanism raises conceptual problems with regard to the electrical operation of the device. Can a voltage source which has a theoretically infinite output resistance serve as a satisfactory basis for a practical measurement system? We recognise that present-day instruments enable measurements to be made from much higher resistance sources than was once the case. However we should distinguish between say a glass electrode which has a high resistance but nevertheless represents a complete cell capable of supplying a continuous current however small and a sensor in which the potential determining process essentially excludes any current flow.In any practical device parasitic electrical conductances will inevitably be present because insulation levels must be finite however effective the device’s encapsulation may be. The blocked sensor would be highly sensitive to these effects and the likelihood of instability again arises. These questions have been explored by Janata.33 It has been pointed out that the difficulties associated with parasitic electrical effects will increase with the length of the connection between the sensing interface and the input of the amplifier or voltmeter which measures the potential. It has been argued that the ISFET will be superior to the CWE or hybrid sensor in this respect because it represents the limiting case of minimum connection length.33 In the case of a hybrid sensor however the connection need comprise only a few millimetres of conductor track and this does not appear likely seriously to prejudice it as a practicable alternative approach.Conclusion The SIC and hybrid microcircuit processes offer complementary routes to the fabrication of integrated solid-state ion sensors. Most of the research effort in the field so far has focussed on ISFET however and the thick-film devices are less well developed at present. An integrated sensor will not necessarily be superior to a conventional ISE with respect to its sensitivity selectivity etc. but it will offer advantages related to its physical structure and electronic performance.The significance of these improvements will depend on the application of the sensor. Much of the work on ISFET in particular has been motivated by the special requirements of in vivo measurement which are very different from those of the industrial market for example. For many applications the hybrid structure will offer similar performance to the ISFET and the relative advantages of the two approaches will be heavily influenced by matters of cost and hence market size. SIC 1207 R. G . Kelly and A . E. Owen manufacturing processes have been highly refined in response to economic conditions in the integrated circuit market. The very low unit costs which have been achieved in this field will not necessarily extend to the modified processes which will be required for sensor fabrication.The methods used for the encapsulation of the device will have to be appropriate to quantity production and must not introduce excessive cost penalties at this expensive stage of the fabrication process. The encapsulation must be sufficiently durable to give acceptable device lifetime under practical conditions of use ; exactly what is acceptable will depend greatly on the application. The reproducibility and stability of ISFET and some hybrid devices appears to be poor compared with membrane ISE. Theoretical considerations suggest that these problems might have fundamental origins but it has alternatively been argued that they are due only to technological deficiencies in current methods of fabri~ation.~~ Like CWE however ISFET have found practical application in spite of theoretical doubts about their operating mechanism.Electronic data-processing techniques have been used to reduce the effects of drift to an acceptable level for certain applications.22 It is the opinion of the present authors that many of the applications for integrated sensors will require a degree of reproducibility between devices and long-term stability that are at least comparable with conventional ISE. We believe that this will only be achieved by devices which incorporate well defined non-blocking contacts and research is continuing into ways of implementing those contacts in solid-state form. Finally the full benefits of an integrated ion-sensor system will not be realised until a compatible solid-state reference system is devised.R. E. Belford F. Daley R. A. Parr and J. Wilson have contributed to the development of a process for the thick-film deposition of pH glass. We are grateful to Elf (U.K.) Ltd and to the S.E.R.C. for financial support for this work. References 1 S. Middelhoek D. J. W. Noorlag and G. K. Steenvoorden Electrocompon. Sci. Tech. 1983 10 217. 2 G. Velasco J. Ph. Schnell and M. Croset Sensors and Actuators 1982 2 371. 3 A. Sibbald IEE Proc I Solid-State and Electron Dev. 1983 130 233. 4 S. M. Sze Physics of Semiconductor Devices (Wiley New York 1969). 5 P. T. McBride J. Janata P. A. Comte S. D. Moss and C. C. Johnson Anal. Chim. Acta 1978 101 239. 6 B. Shiramizu J. Janata and S. D. Moss Anal. Chim. Acta 1979 108 161.7 U. Oesch S. Caras and J. Janata Anal. Chem. 1981 53 1983. 8 G. Blackburn and J. Janata J. Electrochem. SOC. 1982 129 2580. 9 H. Abe M. Esashi and T. Matsuo IEEE Trans. 1979 ED-26 1939. 10 T. Akiyama Y. Ujihira Y. Okabe T. Sugano and E. Niki IEEE Trans. 1982 ED-29 1936. 11 J. Langmaier K. Stulik and R. Kalvoda Anal. Chim. Acta 1983 148 19. 12 T. J. Mellor M. Haskard and D. E. Mulcahy Anal. Lett. 1982 15 1549. 13 T. A. Fjeldly K. Nagy and B. Stark Sensors and Actuators 1983 3 11 1. 14 M. A. Afromowitz and S. S. Yee J. Bioeng. 1977 1 55. 15 S. I. Leppavuori and P. S. Romppainen Electrocompon. Sci. Tech. 1983 10 129. 16 A. Neidig G. Popp and G. Gilbers U.S. Patent 4232326. 17 C.-C. Wen T. C. Chen and J. N. Zemel IEEE Trans. 1979 ED-26 1945. 18 P. A. Comte and J. Janata Anal. Chim. Acta 1978 101 247. 19 T. Matsuo and H. Nakajima Sensors and Actuators 1984 5 293. 20 R. G. Kelly and A. E. Owen IEE Proc I Solid-state and Electron Dev. 1985 132 227. 21 R. W. Cattrall and I. C. Hamilton Ion-Selective Electrode Rev. 1984 6 125. 22 P. Bergveld IEE Colloquium Digest 1985154 p . 211. 23 L. Bousse and P. Bergveld Sensors and Actuators 1984 6 65. 24 R. G. Kelly Ph.D. Thesis (University of Edinburgh 1979). 25 Beckman Instruments Ltd U.K. Patent 1260065. 26 M. F. Nichols U.S. Patent 4312734. 27 P. Bergveld N. F. De Rooij and J. N. Zemel Nature (London) 1978 273 438. 1208 Solid-state Ion Sensors 28 Y. G. Vlasov A. V. Bratov and V. P. Letavin Anal. Chem. Symp. Ser. 1981 8 387. 29 G. Eisenman The Origin of the Glass-electrode Potential in Glass Electrodes for Hydrogen and other Cations ed. G . Eisenman (Marcel Dekker New York 1967) p. 133. 30 C. Conti and G. Eisenman Biophys. J. 1965 5 247. 31 I. Lauks Sensors and Actuators 1981 1 261. 32 W. M. Siu and R. S. C. Cobbold IEEE Trans. 1979 ED-26 1805. 33 J. Janata Sensors and Actuators 1983 4 255. Paper 511892; Received 21st October 1985

ISSN:0300-9599    

DOI:10.1039/F19868201195

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

Devices (ISFETs) J. Chem. SOC. Faraday Trans. I 1986,82 1209-1215 Recent Advances in Microelectronic Ion-sensitive The Operational Transducer Arthur K. Covington* and Peter D. Whalley Electrochemistry Research Laboratories Department of Physical Chemistry University of Newcastle upon Tyne Newcastle upon Tyne NEl 7RU The development of ion-selective field-effect transistors (ISFETs) their advantages over ion-selective electrodes (ISEs) and their intrinsic dis- advantages are reviewed. Some of these disadvantages can be overcome by utilising the ISFET as the integral element in an analogue circuit to provide an operational transducer on a 2 x 2.5 mm silicon chip. The resultant device has small thermal sensitivity and linear response. The performance of a nitrate ion-sensitive device appraised by a continuous dilution computer- controlled technique is described.Successful and adequate encapsulation of the semiconductor excluding the chemosensitive gate region is mandatory and recent improvements to the polyimide-photopolymer process are described. Since the original pioneering work of Bergveldlq on ion-selective field-effect transistors (ISFETs) interest has grown in spite of very real practical difficulties in this interdis- ciplinary area linking solid-state integrated-circuit technology to analytical potentiometry with ion-selective electrodes (ISES).~~ Although there are a number of types of chemically sensitive semiconductor devices most are based on the field-effect transistor (FET),5 or insulated-gate field-effect transistor (IGFET).Structurally the ISFET is very similar to the IGFET and a typical construction of an n-channel IGFET is shown in fig. 1. It consists of a p-type silicon substrate with source and drain diffusions separated by a channel which is overlain by SiO as insulator and a metal gate. The polarity and magnitude of the gate voltage ( VG) applied between the substrate and the gate are chosen so that an n-type inversion layer forms in the channel between the source and drain regions. The magnitude of the drain current (i,) will be determined by the effective electrical resistance of the surface inversion layer and the voltage difference ( V,) between the source and the drain. It is an advantage to operate at low VD in the unsaturated mode.The ISFET (fig. 2) differs from the IGFET in several respects; first the analyte solution is in direct contact with the gate insulator layer(s) and a reference electrode in the solution replaces the metal gate. Previously it had been thought1* that the reference electrode was unnecessary but this is not SO.^ Secondly the introduction of Si,N4 overlying the SO provides a charge blocking interface and furthermore other films such as PVC containing valinomycin which is the ion-selective electroactive material used in potassium ion-selective electrodes can be added to confer other ion selectivities to the ISFET.3 Lastly the successful encapsulation of all regions of the device other than the gate region to be exposed to the analyte solutions is mandatory.' If the applied gate bias potential ( VB) is fixed then changes at the solution-electroactive material interface are reflected in changes in i,.However if the drain current is maintained at a constant value by means of an operational amplifier which directly controls the applied gate bias potential using a negative-feedback loop (fig. 3) then the 1209 1210 Ion-sensitive Devices p-Si F I Fig. 1. Schematic diagram of an IGFET. A Metal gate B SiO insulator C source and drain metal contacts D drain E source and F substrate. I I ' Fig. 2. Schematic diagram of an ISFET. REF Fig. 3. Constant-current operating-mode circuit. output is a voltage which varies with change in activity of the sensed ion in accordance with the Nernst equation. The output is therefore effectively the same as that from an ISE.The method involving constant drain current has disadvantages if the circuit is broken when the analyte solution is changed. This can cause the feedback potential to assume large positive values and lead when it is re-established to voltage transients which polarise the reference electrode and/or the electroactive gate causing spurious offset potentials and even encapsulation breakdown. A . K. Couington and P. D. Whalley 121 1 An equivalent gate voltage circuit may be used7 in which the drain current is inverted undergoes i/V conversion and the resultant voltage is used to drive a further circuit incorporating a matching IGFET so that the IGFET drain current mirrors that of the ISFET. This requires several operational amplifiers voltage reference sources and a means of gain adjustment to compensate for slight ISFET/IGFET differences.Four-function ISF'ETs for hydrogen potassium sodium and calcium ion have been developed in Newcastle and very successfully applied to on-line analysis of these ions in whole blood during surgery.8 ISFETs have also been made sensitive to nitrate carbonateg and phenobarbital anions.l0 Since all these devices with the exception of that for hydrogen ions which is based on silicon nitride have utilised the electroactive materials developed for ISEs they suffer from the same disadvantages with respect to lack of selectivity as do ISEs. The advantages of ISFETs include their small size low-impedance output and lower production costs (if one excludes the high capital cost of the required semiconductor fabrication facility).Since the problem of achieving satisfactory selective encapsulation of the semiconductor chip and associated bonding wires using a polyimide-photopolymer process has been ~olved,~ their remaining disadvantages are (a) short lifetimes if the electroactive material leaches out from the thin coating covering the gate region (b) inherent thermal sensitivity (c) non-linear response and ( d ) non-buffered device output which is unsuitable for driving multiplexing circuitry . Recently Sibbaldll in Newcastle has overcome the last three disadvantages with the operational transducer which incorporates a matched ISFET/IGFET pair connected in source-coupled dual-differential configuration with the output voltages coupled to a differential/single-ended converter and then to several d.c.amplification stages so that the ISFET is the non-inverting input device and the IGFET the inverting input device. The operational transducer (Optran),ll with its simpler power requirements linear output and low thermal sensitivity is therefore an important development in micro- electronic chemical sensors. Improvements have been made to the polyimide (PI) - photosensitive polymer encapsulation process7 by using a PI coating applied selectively at the wafer stage of chip fabrication. An advantage of this technique is mechanical keying of polymeric electroactive membranes to the surface of the ISFET. Results for a nitrate-sensitive Optran device are given.Method and Materials The Chip The Ep358A Optran device has been described in detail elsewhere.ll A schematic representation of the Optran with a reference electrode is shown in fig. 4. To facilitate encapsulation a 600 nm polyimide coating (PI-2555 Du-Pont Stevenage) was spun on at a final stage of wafer fabrication exposing the chemically responsive gate and bond pads only. An adhesion promotor (a-aminopropyltriethoxysilane VM-65 1 Du-Pont) was used prior to the application of the PI. Patterning of the PI was achieved using positive photoresist as a mask. The alkaline developer for the photoresist also removed the partially imidised PI. A high temperature (350 "C) was used to imidise fully this layer. Electroactive Material Tetradodecylammonium nitrate (TDDA-N),l29 l3 for use as the nitrate-responsive electroactive material was prepared from bromide salt (Fluka) by repeated exchange with aqueous KNO, and recrystallised from ethanol-water mixtures.Composition was confirmed by elemental analysis. The membrane was composed of 4% TDDA-N 8% dibutylphthalate (Aldrich) and 28 % PVC (Fluka). A tetrahydrofuran-cyclohexanone solvent mixture was used for membrane casting. 1212 N reference Ion-sensitive Devices m &-& I+” Fig. 4. Operational transducer (Optran). Sensor Construction The chip was mounted on a 16 pin dual in-line ceramic header. After cleaning the chip was wire-bonded to the gold-plated header (cleaning is not as critical for successful encapsulation as with previously used ISFETs with bare silicon nitride surfaces’).PI (PI-2555 Du-Pont) was applied to the chip and bond wires in a thin coat and partially imidised. The negative photoresist (KPR-4 Eastman-Kodak Liverpool) was pre- thickened under partial vacuum (400 Torr)? for several days at 40 “C. This reduced the solvent concentration and removed dissolved gas allowing a single coat of photoresist to be used and decreased bubble formation in the final high-temperature cure cycle. Optical masking of the single gate of the Optran was achieved using a 1 mm drafting pen. Cross-linking of the negative photoresist was initiated using a medium-pressure Hg vapour lamp (125 W) with a measured incident radiation of 20 mW cm-l(205-uv power meter Optical Associates Inc.).The composite encapsulation had a combined thickness of ca. 50pm after developing the photoresist etching the PI and final hard baking at high temperature. Polymeric membranes were applied to the well formed by the encapsulation above the gate of the chip using a disposable pipette tip. Several castings of the membrane were necessary to obtain an approximate thickness of 30 pm. Devices were stored under vacuum for 48 h to ensure the removal of the less volatile cyclohexanone solvent. After membrane application V-shaped flow caps (plate 1) were cemented into position using silicone rubber (Dow-Corning RTV-3 145).’ Sensor Operation All solutions were prepared from distilled water using analytical-grade dried salts. Before use sensors were conditioned in 0.1 mol dm- KNO,.A & 5 V regulated power supply for the transducer and a suitable variable reference electrode voltage source was used. Sensor flow cell assemblies were characterised by filling with solution connecting a reference electrode (saturated calomel Russell pH Ltd type CR) and sweeping the reference voltage from - 1 to +4 V to establish the linear operating region. This also permitted an appropriate gate bias potential to be selected. The ability of the encapsulation to preclude solution contact with electrical connections was verified for each device by applying varied gate bias voltages to the reference electrode and monitoring any resulting current flowing between the reference electrode and the encapsulated sensor. The importance of applying a varied bias voltage has been t 1 Torr = 101 325/760 Pa.J . Chem. Soc. Faraday Trans. I Vol. 82 part 4 Plate 2. SEM of membrane cross-section. (Facing p . 12 12) A. K. COVINGTON AND P. D. WHALLEY Plates 1 and 2 J. Chem. SOC. Faraday Trans. 1 Vol. 02 part 4 Plate 4. SEM of membrane physically removed from the gate region showing impression of FET gate. A. K. COVINGTON AND P. D. WHALLEY Plates 3 and4 1213 A . K . Covington and P . D. Whalley Fig. 5. Computer-controlled constant-dilution apparatus. A initial solution; B diluent ; C magnetic stirrer; D dilution vessel; E flow cell; F reference electrode; G peristaltic pump; - I c I H to waste; I ISFET interface; J chart recorder. emphasised elsewhere.' Leakage currents have previously7 been detected with a high- precision digital electrometer.An alternative system is now used. Small currents (< 0.1 nA) are measurable using an operational amplifier (RCA CA 3140) in a current-to-voltage configuration with a 100 M a feedback resister in parallel with a 470 pF capacitor the output being fed to a digital multimeter so that 1 V corresponds to 10 nA; leakage currents of < 10 pA are measurable. A constant-dilution techniquel3* l4 was used to obtain nitrate calibration curves and estimate selectivity coefficients of chloride and nitrite interferences on the nitrate response of the device. The design and configuration of the constant-dilution apparatus is given schematically in fig. 5. The electrical output from the Optran and reference electrode was monitored with a digital multimeter (Hewlett-Packard type HP-3468A) and a chart recorder (Linseis).Data acquisition and estimation of activities (using a Debye-Huckel expression for activity coefficients) was by microcomputer (Hewlett-Packard HP86B). Output voltages were recorded every 30 s during the dilution of the primary ion from 10-1 to mol dm-3. A constant flow rate of between 4.5 and 5.5 ~ m - ~ min-l resulted in a complete calibration being obtained in ca. 50 min. The relationship between flow rate initial concentration of the primary ion and concentration after a given time is given by14 log ct = log c - t - log e F V where c and ct are the initial concentration and concentration of primary species at time t respectively F is the flow rate and V is the volume of the dilution vessel.The concentration of interfering ion (Cl- or NO;) was kept constant throughout the dilution at 10-1 mol dm-3 by including the ion in both the initial solution ( 10-1 mol dm-3 NO;) and the diluent (corresponding to the mixed-solution method for estimating selectivity data15). All measurements were made at room temperature (typically 20 "C). Results and Discussion Testing of the chips prior to encapsulation was not possible. The yield of functional devices after mounting and encapsulation was ca. 50%. (This compares with ca. 80% success rate for conventional ISFETP .) The application of PI under clean conditions and before the growth of surface oxides is an important enhancement of the previously developed PI-photopolymer composite encapsulation technique.' The second PI coat applied during encapsulation adheres well 1214 Itoresist -+ mern brane FET channel PI 2-* PI 1-b Si N4 surface Fig.6. Schematic representation of cross-section of encapsulation and polymeric membrane. (PI 1 polyimide applied at chip fabrication stage PI 2 polyimide applied during encapsulation.) 4 5 Ion-sensitive Devices 3 bias potentiallv Fig. 7. Electrical characteristics of NO;-responsive Op tran. to the first layer. No device failures attributable to the penetration of solution between the silicon nitride of the chip and the PI layer have been seen as revealed by the presence of interference fringes at this interface. A further advantage of this encapsulation technique is the formation of an undercut by the etching of the PI under the photoresist.This undercut acts as a mechanical key for the polymeric electroactive membrane. The undercut is visible in the scanning electron microscope (SEM) image of a cross-section of a membrane (plate 2; a certain amount of distortion is apparent as a consequence of physically removing the membrane from the chip) and shown diagrammatically in fig. 6. In the process of encapsulation the chemosensitive region is optically masked so the photoresist is protected from the u.v.; hence a well with vertical sides is formed (plate 3) after the removal of non-crosslinked material by solvent. The PI etchant (a 1 1 mixture of diaminoethane and hydrazine hydrate’) etches in both the horizontal as well as the vertical direction thus undercutting the photoresist and keying the electroactive polymeric membrane.The ability of the encapsulation to cause the membrane to be in intimate contact with the FET gate is clearly seen in the SEM image of the impression of one end of the gate in a PVC membrane (plate 4). This method of ensuring membrane contact with the FET gate is believed to be simpler than the suspended mesh approach of Blackburn and Janata. l7 is shown in fig. 7 and showed a linear operating region between ca. + 1 and +4.5 V. An electrical characteristic plot of the Optran with a nitrate-sensitive membrane used 1215 300' A . K. Covington and P. D. Whalley I 1 1 I I 4 -3 -4 -5 -2 - log aN0; Fig. 8.NO calibration curve (a) and the influence of mol dmW3 C1- (b) and NO (c) on the response of a NO;-sensitive Optran. A typical calibration curve and the effect of chloride and nitrite on the response of the nitrate-sensitive device is shown in fig. 8 (typical selectivity coefficients are - log = 2.2; -log K N 0 3 N O ; = 1.5). These results compare well with those obtained previously using nitrate electroactive membranes on ISFETs.13 Device lifetimes (typically several weeks) appear to be limited by leaching of the electroactive from the PVC membrane and not by encapsulation. Financial support from the S.E.R.C. is gratefully acknowledged. Thanks are due to Dr A. Sibbald for his role in the Newcastle work on ISFETs to the staff especially Alan Gundlach of Edinburgh Microfabrication Facility for device processing to the following coworkers Mr J.R. Dodgson Mr T. R. Harbinson Mrs R. Kataky Mr S. Papacostas and Mr S. Reveley and to Mr D. A. Jeffrey for his valuable technical assistance. References 1 P. Bergveld IEEE Trans. 1970 BME-17 70. 2 P . Bergveld IEEE Trans. 1972 BME-19 342. 3 J. Janata and R. Huber in Ion-selective Electrodes in Analytical Chemistry ed. H. Freiser (Plenum New York 1981) vol. 2. 4 A. Sibbald IEE Proc. 1983 I 130 233. 5 J. N. Zemel Anal. Chem. 1977,47 255A. 6 R. G. Kelly Electrochim. Acta 1977 22 1. 7 A. Sibbald P. D. Whalley and A. K. Covington Anal. Chim. Acta 1984 159 41. 8 A. Sibbald A. K. Covington and R. F. Carter Med. Biol. Eng. Comput. 1985 23 329. 9 A. K. Covington T. R. Harbinson and A. Sibbald in preparation. 10 A. K. Covington and T. R. Harbinson Anal. LRtt. 1982 15 (A17) 1423. 11 A. Sibbald Sensors and Actuators 1985 7 23. 12 H. J. Nielsen and E. H. Hansen Anal. Chim. Acta 1976 85 1. 13 A. K. Covington and P. D. Whalley in preparation. 14 G. Horvai K. Toth and E. Pungor Anal. Chim. Acta 1976,82 45. 15 A. K. Covington in Ion-selective Electrode Methodology ed. A. K. Covington (CRC Press Florida 1979) vol. I. 16 P. D. Whalley unpublished work 1985. 17 G. Blackburn and J. Janata J. Electrochem. Soc. 1982,129 2580. Paper 511893; Received 28th October 1985

ISSN:0300-9599    

DOI:10.1039/F19868201209

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

Coated Wire Ion-Selective Electrodes J. Chem. SOC. Faraday Trans. I 1986,82 1217-1221 Principles and Practice H. Freiser Strategic Metals Recovery Research Facility Department of Chemistry University of Arizona Tucson Arizona 85721 U.S.A. Ion-selective electrodes prepared by coating polymer films containing electroactive species have been incorporated on a metallic substrate and have proven to be effective for a wide variety of inorganic and organic anions and cations. Their characteristics are reviewed here. In addition current scan polarography and chronopotentiometry across an aqueous-immiscible organic solvent interface have been applied to the study of ion-transport processes such as that involved in the potassium-crown ether ion-selective electrode systems.A mechanism for such systems is described. The ion-selective electrode (i.s.e.) approach to trace analysis is advantageous because of the speed and ease of i.s.e. procedures in which little is required. Further i.s.e. possess wide dynamic ranges and are relatively low in cost. These characteristics have inevitably led to sensors for several ionic species and the list of available electrodes has grown substantially over the past two decades. In many cases the traditional barrel configuration has been used. However the large size of this type of i.s.e. along with the requirement that it be used in a nearly upright position renders it somewhat cumbersome to use and unnecessarily expensive. In our laboratory these disadvantages have been overcome with the development of the coated wire electrode (c.w.e.).This sensor having response characteristics equal to and occasionally better than conventional types is only 1-2 mm in diameter (further size reduction can be easily achieved) can be used at any angle and costs only a few pennies to make. Indeed they can be considered ‘disposable’ though with proper handling lifetimes over six months have been realised. During the course of c.w.e. investigations here the list of analyte species has been lengthened to include not only most common inorganic ions of interest but also organic species which are anionic or cationic under appropriate solution conditions (table 1). The first electrode of this type was based on the Ca2+-didecylphosphate/dioctylphenyl phosphonate system.l An effective Ca2+ selective c.w.e.resulted when a 6 1 mixture of 5% PVC in cyclohexanone and 0.1 mol dm-3 Ca didecylphosphate in dioctylphenyl- phosphonate was dried on the end of a platinum wire. Favourable comparison of this electrode’s response characteristics against those of the commercial counterpart encour- aged further studies with other membrane components. The Ca2+ electrode response relied upon the complexation of aqueous Ca2+ by didecylphosphate dispersed in the organic or membrane phase. In a similar manner incorporation of methyltricaprylam- monium (Aliquat 336s) salts in polymer membranes produced c.w.e. for their respective anions.2 A 60% (v/v) solution of Aliquat 3363 in decanol was first converted to the desired anionic form by shaking with a 1 mol dm-3 aqueous solution of the appropriate Na+ salt.A 10 1 mixture of PVC in cyclohexanone and this decanol solution was then used to coat copper wires by repeated dipping and drying until a small bead completely encapsulated their ends. Applications ranged from critical micelle determination using laurylsulphonate sensor^,^ analyses of atmospheric NO pollutants with nitrate electrodes4 and assay of phenobarbital tablets using a phenobarbital anion c . w . ~ . ~ 1217 1218 Coated Wire Ion-Selective Electrodes lipophilic cation-based halide C1- Br- I- CNS- oxyanion NO, ClO; organic anion RCOO- RSO; Table 1. Coated wire electrodes //O ‘OH amino acid H,N-C(-R)-C neutral carrier-based K+-valinom y cin dinon ylnapthalenesulphonate-based quaternary ammonium ions drugs of abuse e.g.phencyclidine methyl- amphetamine methadone /I-adrenergic drugs e.g. acebutalol Ca-blockers e.g. verapamil phenothiazines e.g. chloropromazine In many cases the selectivities of the c.w.e. were significantly better than the conventional ‘ barrel-type ’ counterparts. This along with absence of the traditional internal reference electrode raised fundamental questions surrounding the charge conduction mechanism occurring in the membrane and at the polymer-substrate interface. Calculation of activation energies from the temperature dependence of conduction suggested that an electronic mechanism was operative such as that observed in organic semiconductors.6 Later studies of the pressure dependence of conduction gave strong evidence for ionic conduction because much larger activation volumes than could be expected from an electronic mechanism were obtained.’.* As such the existence of a redox couple at the substrate-polymer interface probably functions as an ‘internal reference’. This hypothesis is further reinforced when one considers that conditional standard potentials shift by significant and reproducible amounts from one type of metal substrate to another. Our attention next turned to the development of cation-selective electrodes in order to develop methods for protonated alkylammonium ions. Initial studies in this area were aimed at improving selectivity among similarly charged cations by using a mobile exchange site facilitating membrane response to changing counter ions.9 These membranes were comprised of dinonylnaphthalene sulphonic acid (DNNS) a lipophilic anionic extractant dissolved in a PVC membrane plasticized with dioctylphthalate.Extremely high selectivity was observed for alkylammonium ions over common inorganic ions. Among organic species selectivity increased regularly with the number of carbon atoms of the analytes tested. This indicated great promise for sensors of pharmaceutical interest since many such compounds are high-molecular-weight protonated amines in the physiologic pH range. Currently DNNS based c.w.e. are made by dissolving the amine of interest in 5% PVC in tetrahydrofuran (THF) which is also 0.5% in DNNS and 4.5% in plasticizer (usually dioctylphthalate).This solution is then used to coat the end of a copper wire which is elsewhere insulated with non-plasticized PVC. Following conditioning in a 10-4-10-3 mol dm-3 solution of the analyte the electrodes are ready for use. C.w.e. are available9-12 for drugs of abuse such as phencyclidine (PCP) cocaine methylamphetamine and methadone as well as P-adrenergic and calcium blockers for cardiac treatment such as propranalol protriptyline acebutalol lidoflazine verapamil diltiazem and nicardipine and psychotherapeutic agents such as the phenothiazines (e.g. chlorpromazine). 1219 59.2 H . Freiser Electrode selectivity is described by the ‘ selectivity coefficient ktyt which is calculated from e.m.f. responses of the electrodes sampled in the presence and in the absence of an interfering ion using the following relationship E = Ee+- ( a i + k ~ t a ~ l Z ~ ) n where n and zi are charges of the primary and interfering ions respectively and ai and aj are their activities.Experimentally computer-generated solutions of specific ratios of a,/aj are made followed by correction of the ‘new’ activity of ion i due to addition of ion j . Systematic selectivity studies of both cation- and anion-responsive c.w.e. reveal the importance of solvent extraction parameters. Using tributylammonium as a primary ion log kF,:t values were determined for various substituted alkylammonium ions.lo9 l1 As was the case for Aliquat-based electrodes for organic species,2 selectivity improved considerably with increasing molecular weight.These observations are closely related to analyte extractability. From solvent extraction studies and from calculated extraction constants,13 close one-to-one correspondence is observed between Kex and electrode selectivity coefficients kf’,;t. Hence one can see the use in considering solvent extraction parameters in predicting electrode selectivity behaviour. An interesting example of a ‘neutral carrier’ i.s.e. is the potassium-valinomycin electrode developed by Simon,l* which we later found to function very well as a c.w.e. prepared by using a PVC matrix and di-n-decyl-phthalate as ~1asticizer.l~ More recently we have begun to study ion-transport mechanisms involved in electrochemical and other processes. In recent years the study of the Faradaic ion transfer across an aqueous/immiscible organic solvent interface has been developed and recognized as a powerful tool for the investigation of liquid-liquid (L-L) interfacial phenomena.16 This method possesses basic features which are quite similar to the classical electrochemical system in which a solid or liquid-metallic electrode-solution (S-L) interface is employed.When these interfaces are externally polarized ion transfer takes place predominantly at the L-L interface because both liquid phases are ion-permeable while electron transfer predominates at the S-L interface owing to the high electron conductivity of the solid electrode phase. Electrochemical study of the aqueous-organic interface can provide a model for biological membranes as previously pointed out by Koryta et all6 The advantage of using such systems over an artificial membrane system is that one can isolate the information on interfacial phenomena because mass transfer in the liquid phase is simple diffusion which is not the case in the membrane system.Employing this technique Koryta and Samec’s group investigated the facilitated transfer of ions (alkali-metal and alkaline-earth ions and protons) by such neutral carriers as valinomycin crown ethers and nonactin present in the organic phase.17* l8 Through these studies they arrived at a transport mechanism of potassium-valinomycin complex ion concluding that complex formation (and dissociation) occurs in the organic phase following phase transfer of the hydrated potassium ion and a surface reaction between aqueous potassium ion and the carrier molecule in the organic phase.On the other hand in our worklg on the analogous ion transfers involved in metal chelate solvent extraction systems i.e. where metal ions are transported (extracted) into the organic phase in the presence of neutral carrier extractants complex formation was demonstrated to occur in the aqueous phase following the transport of the carrier from the organic to the aqueous phase even with carriers having low aqueous solubility. The reaction mechanism of the potassium-ion transfer from aqueous to nitrobenzene phases which is facilitated by valinomycin present in the organic phase was elucidated by polarography at the ascending water electrode and chronopotentiometry at the stationary water electrode.The complex ion is formed in the aqueous-phase reaction between potassium ion and valinomycin that migrated from the organic phase. The Coated Wire Ion-Selective Electrodes 1220 -201 Fig. 1. Current Scan Polarogram at a.w.e. Aqueous phase .MgSO, 1 mol dm-3. KCl (1) 0.01 (2) 0.1 (3) 1 rnol dm-3. DCE phase THA-TPB 0.01 mol dm-3 DBC 0.4 mmol dm-3. accumulation of the complex ion in the aqueous phase is controlled by the diffusion of valinomycin from the bulk organic phase to the water-nitrobenzene interface. In further pursuit of greater understanding of these highly interesting transfer processes we extended our study to the system of K+-dibenzo-18-crown-6 (DBC) in dichloroethane (DCE) using the ascending water electrode (a.w.e.) in a current scan polarographic system described earlier.,O The range of K+ concentrations in the aqueous phase which was also 1 mol dm-3 in MgSO, was from 1-00 x 10-4-l.00 mol dm-3.In DCE which was 0.01 mol dm-3 in tetraheptylammonium tetraphenylborate the DBC concentration ranged from (2.00-6.00) x lo- mol dm-3. At the concentrations of KC1 in the aqueous phase at or below 0.02 mol dm-3 a single wave is obtained whose limiting current is directly proportional to the concentration of DBC in the DCE phase and to the square root of the height of the hydrostatic head of the aqueous electrolyte. The logarithmic analysis of this wave shows a straight line with a slope of 52 mV. The product of the applied current and the square root of the transition time was found to be independent of the applied current in constant current chronopotentiometry at the stationary water electrode (s.w.e.).The square root of the transition time z increases linearly with the square root of the standing time t i.e. the time elapsed before the s.w.e. system undergoes constant current electrolysis. It is clear from these results that the transfer process is diffusion-controlled and the transferring species from the aqueous phase into the organic phase is the complex ion of potassium with crown ether in accord with.19 If the potassium ion itself were transferring one would not expect any accumulation as represented by an increase of z with t at the s.w.e. The half-wave potential El,, shifts to less positive values with CK+ increasing.When the K+ concentration reaches 0.01 mol dm-3 or higher however a new series of phenomena are observed. At CK+ = 0.01 mol dm-3 Ellz reaches a value which it maintains until the wave disappears entirely at CK+ = 0.2 mol dm-3. Furthermore the limiting current of the first wave in the region 0.04 mol dm-3 < CK+ < 0.2 mol dm-3 exhibits a first-order dependence on electrostatic head as well as an independence of DBC concentration. Starting at CK+ = 0.04 mol dm-3 an additional wave appears at less-positive potential which increases in height at the expense of the original one which finally disappears at CK+ = 0.2 mol dmU3. The half-wave potential of this second wave moves to less positive values with further increases in CK+ increasing the difference of the half-wave potential between the two waves.The limiting current of this wave exhibits a square-root dependence on the hydrostatic head. Chronopotentiometry conducted at 1 mol dm-3 KC1 where only the second wave is observed results in z values that are independent 1221 H. Freiser of t . Finally it should be noted that the sum of the heights i.e. limiting currents of the two waves remains constant throughout the entire range of CK+ and remains proportional to the crown ether concentration in DCE.21 different from the first. The first wave at CK+ < 0.01 mol dm-3 displays all the charac- These findings demonstrate unequivocally that the second wave is fundamentally teristics of a process which involves diffusion of DBC from DCE to the aqueous phase where it reacts with K+ to form K(DBC)+ the complex ion which transfers electrochemically across the interface.At CK+ > 0.01 mol dm-3 the diffusing DBC reacts with K+ in the interfacial region explaining the adsorption characteristic (iah1a0) observed. At still higher CK+ the second wave which may represent the electrochemical transfer of the simple K+ appears. Even this process depends on the rate of diffusion of DBC to the interface from the bulk organic phase. This study was supported by grants from the Office of Naval Research. References 1 R. W. Cattrall and H. Freiser Anal. Chem. 1971 43 1905. 2 H. James G. D. Carmack and H. Freiser Anal. Chem. 1972,44,853; 856. 3 T. Fujinaga S. Okazaki and H. Freiser Anal. Chem.1974 46 1842. 4 B. M. Kneebone and H. Freiser Anal. Chem. 1973,45449. 5 G. D. Carmack and H. Freiser Anal. Chem. 1977 49 1577. 6 G. D. Carmack and H. Freiser Anal. Chem. 1973 45 1976. 7 G. D. Carmack and H. Freiser Anal. Chem. 1975,47 2249. 8 G. D. Carmack and H. Freiser Anal. Chem. 1977 49 767. 9 C. R. Martin and H. Freiser Anal. Chem. 1980,52 562. 10 C. R. Martin and H. Freiser Anal. Chem. 1979 51 803. 11 L. Cunningham and H. Freiser Anal. Chim. Acta 1982 139 97. 12 L. Cunningham and H. Freiser Anal. Chim. Acta 1981 132 43. 13 C. Hansch and J. L. Leo Substituent Constants for Correlation Analysis in Chemistry and Biology (Wiley New York 1979). 14 Z. Stefenac and W. Simon Microchem. J. 1967 12 125. 15 R. W. Cattrall S. Tribuzio and H. Freiser Anal. Chem. 1974 46 2223. ed. H. Gerischer and C. W. Tobias (Wiley-Interscience New York 1981) vol. 12 p. 113. 16 J. Koryta and P. Vanysek in Advances in Electroanalytical Chemistry and Electrochemical Engineering 17 A. Hofmanova L. Q. Hung and W. Khalil J. Electroanal. Chem. 1982,135 257. 18 Z. Samec D. Homolka and V. Marecek J. Electroanal. Chem. 1982 135 265. 19 Z. Yoshida and H. Freiser J. Electroanal. Chem. 1984 162 307. 20 Z. Yoshida and H. Freiser J. Electroanal. Chem. 1984 179 31. 21 S. Lin and H. Freiser J. Electroanal. Chem. in press. Paper 511894; Received 16th September 1985

ISSN:0300-9599    

DOI:10.1039/F19868201217

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. SOC. Faraday Trans. I 1986,82 1223-1236 Potentiometric Monitoring of Proteins Part 3.-Direct Potentiometry with a Copper Electrode U.S.A. Michael L. Hitchman"? and Frazier W. M. Nyasuluj Department of Chemistry and Applied Chemistry University of Salford Salford M.5 4 WT Results are presented for the direct potentiometric determination of L- cysteine L-histidine and a number of proteins with copper electrodes which are cathodically treated before measurement. Comparisons are made with untreated and chemically treated electrodes. It is shown that provided the cathodic pretreatment technique is used the electrode response is fast and reproducible enough for the method to form the basis of a simple reliable and inexpensive procedure for the monitoring of proteins.The Nernstian responses and equilibration times together with the improved behaviour of the electrochemically pretreated electrodes are discussed in terns of various models for the complexation of copper with amino acids and proteins. It is well known that CuII will complex with amino acids,l and this fact has been used as the basis for the potentiometric determination of amino acids with metallic copper electrode^^-^ and copper-selective membrane electrode~.~-~ However in using copper- based electrodes for monitoring organic complexing agents there are problems owing to the sluggish response to changes in ligand concentration the susceptibility of the electrodes to poisoning (leading to a lack of reproducibility) and the marked dependence of the electrode signal on the history and pretreatment of the electrode.8~9 So for example in some recently reported work7 on the indirect potentiometric determination of a-amino acids with a copper-selective electrode Athanasiou-Malaki and Koupparis chose instead of the direct reaction of amino acids with CuII ions in solution to use the more complicated technique of reacting the amino acids with excess of copper(I1) phosphate suspension and then measuring the copper ions produced on acidifying the filtrate.This technique allows the determination of a number of a-amino acids in the concentration range 5 x lo-* to 1 x lod2 mol dm-3 but it does depend on the a-amino acid forming a soluble copper complex which was found to exclude a considerable number of amino acids.Alexander and Maitra have found2 that bright copper wire and tubular electrodes have higher sensitivities and more reproducible responses to amino acids than copper-selective membrane electrodes and they have applied3 copper metal electrodes in flow-through configurations to the detection of amino acids eluted from a reversed-phase h.p.1.c. column. However the work of Alexander et al. shows that the potential dependence of metallic copper electrodes on amino acid concentration is generally non-Nernstian and that the electrode response deteriorates with time especially in the presence of a thiol group. Therefore while copper or copper-selective electrodes offer in principle the basis of a simple and rapid method for monitoring amino acids clearly unless steps can be taken to improve electrode response then the use of copper- based electrodes will be very limited for amino acid determinations and there will be little Present address Department of Pure and Applied Chemistry University of Strathclyde 295 Cathedral Street Glasgow G1 1XL.1 Present address Department of Chemistry Oklahoma State University Stillwater Oklahoma 74078 41 FAR 1 1223 1224 Direct Potentiometry with a Cu Electrode possibility of extending their use to the determination of more complex compounds such as proteins. In two recent publicationslO7 l1 we have discussed ways of improving the reproducibility and response of a metallic silver electrode for the monitoring of thiol-containing proteins. For a silver electrode it has been observed12 that chemical cleaning with ammonia solution before monitoring the end-point in potentiometric titrations of thiols leads to improved electrode response.Cleaning with ammonia solution indicates the removal of a surface layer by formation of a diammine complex Ag(s)/AgX(ads)+2NH,OH(aq)~ Ag(s)+Ag(NH3)~(aq)+2H20+X-(aq). (I) This idea suggested the use of cathodic electrochemical cleaning to achieve the same end (11) Ag(s)/AgX(ads) + e- * Ag(s)/Ag(ads) + X-(aq). We have shownllv l3 that indeed electrochemical cleaning of silver electrodes does allow the direct potentiometric monitoring of proteins. Compared with chemical cleaning the technique produces more rapid stabilisation of- electrode potentials and linear Nernst plots with significantly reduced scatter of data.Also the degree of reproducibility of electrode potentials with electrochemical pretreatment is good enough for quantitative protein assay over a concentration range from ca. mol dm-3 to the solubility limit of the protein (typically three decades of concentration). The success of electrochemical cleaning of silver electrodes coupled with the observation of Alexander and Maitra2 that direct potentiometry with metallic copper is ' reproducible for quantification of amino acids providing bright copper wire is used' suggested that cathodic cleaning of copper electrodes could also lead to significantly improved behaviour of such electrodes for monitoring amino acids and proteins. We have earlier briefly reported13 that the cathodic treatment of a copper electrode does give rise to similar improvements to those found with silver electrodes.In this paper we present more detailed results for the direct potentiometry of a number of proteins and also for two amino acids. The results for the two amino acids L-histidine and L-cysteine are particularly pertinent to the protein study since they represent two extremes of the type of response expected with a copper electrode. Of the ten a-amino acids investigated by Alexander and Maitra,2 L-histidine was found to give the most successful results with the bright copper electrodes. This therefore suggests it could be a model compound to be used in preliminary studies for optimising monitoring conditions for proteins. L-Cysteine on the other hand was found by the same workers to give the least successful results and this was attributed to be due in some way to the presence of the thiol group.This suggests that this amino acid could provide an indication of the problems that one might encounter with thiol-containing proteins. Experimental All electrode-potential measurements were made with conventional instrumentation. Variations of potential with time were monitored on a recorder with the aid of a high-impedance voltage follower to ensure minimum electronic disturbance of the electrodes. Indicator electrodes used were high-purity (99.99 % )copper wires (Goodfellow Metals) of 1 mm diameter. Reference electrodes used were a saturated calomel electrode with a salt bridge and a silver/silver chloride double-junction electrode.Two types of electrode arrangement were used. Batch measurements were made with a small beaker as the cell and the solution stirred with a magnetic follower. Continuous-flow measure- ments were made with the configuration shown in fig. 1. Solutions were pumped with a peristaltic pump (Pharmacia AB Type P3) at flow rates in the range 30-60 cm3 h-l. The performance of potentiometric detectors is known to depend to some extent on 1225 M. L. Hitchman and F. W. M. Nyasulu cw t i Fig. 1. The flow-through electrode design. cw copper wire; i inlet; pc plastic connector; pe platinum electrode; re reference electrode; wo waste outlet. electrode configuration,14 and this will be discussed subsequently in the context of flow-injection analysis and protein monitoring.The following amino acids and proteins (Sigma Chemical Company) were used as received L-cysteine (free base) L-histidine (free base) ribonuclease A (from bovine pancreas) lysozyme (from egg white) bovine albumin (Cohn fraction V) human albumin (Cohn fraction V) bovine y-globulin (Cohn fraction 11) and human y-globulin (Cohn fraction 11). Standard solutions were prepared immediately before use by direct weighing and dissolution in an appropriate' buffer solution with gentle stirring. Solutions in the concentration range 0.1-10 mg ~ m - ~ were generally used the upper limit being established by the solubility of the protein and the lower limit by the detection limit. Buffer solutions were prepared from analytical-grade reagents with compositions given in standard tables15 and a constant ionic strength of 0.1 mol dm-3 was maintained by the addition of sodium nitrate.For studying the effect of pH on the electrode signal either nitric acid or sodium hydroxide solution was added to the test solution and the ionic strength in this case was maintained at 2 mol dm-3 again with the addition of sodium nitrate. Electrode-cleaning procedures investigated included mechanical abrasion with fine emery paper followed by rinsing with distilled water chemical cleaning of the electrode with nitric acid (1 volume concentrated acid 2 volumes distilled water) again followed by a rinse and electrochemical cleaning. The electrochemical cleaning procedure was done by placing the copper electrode and a platinum foil electrode in the buffer solution alone or more usually in the buffered test solution and making the copper cathodic with respect to the platinum.The simple d.c. circuit consisted of a 1.5 V battery a switch and a resistor in series. Various combinations of circuit resistances and cleaning times were was adequate but that a much lower resistance (ca. 10-250 a) compared with that used tried. It was found that as with a silver electrode a cathodic treatment time of ca. 10 s 41-2 1226 tlmin Direct Potentiometry with a Cu Electrode 6 2 1 4 3 5 0 7 1.0- 1 (d 1 0.9 - 0.8 - 0.7 - Fig. 2. Stabilisation curves for electrode potentials with L-cysteine (0.1 mg ~ m - ~ ; pH 8). (a) No cleaning; (b) chemical cleaning; (c) electrochemical cleaning with 1 MR resistance; ( d ) electro- chemical cleaning with 50 R resistance.for the case of silver (ca. 1 MR) gave faster stabilisation times; this point is discussed further below. After an electrode had been cleaned it was briefly rinsed with distilled water and then introduced into the test solution. The potential variation with time was monitored until there was <0.5 mV drift in a five minute period. The time dependences of the potential and the steady-state values were used for comparison with theoretical equations. When cathodic treatment was done in the test solution switching in the external circuit allowed direct changeover to the measurement of electrode potential. In order to minimise washing and rinsing procedures solutions were monitored in order of increasing concentrations but where reproducibility was being investigated a random order was used; for studies of precision seven potential measurements were made for each amino acid and protein concentration.In some cases potentials are reported as the difference (AE) between the value in the test solution and that in the blank. All measurements were made at 23 1 "C. Results and Discussion Amino Acids L- Cys teine Fig. 2 compares the times needed to reach steady-state potentials (stabilisation times) for differently pretreated copper electrodes in L-cysteine (0.1 mg ~ m - ~ ) solution at pH 8. Similar plots were obtained with other cysteine concentrations at various pH values in the range pH 4-8. With a 1 MR resistor in the d.c.cleaning circuit as used in the case of pretreatment of silver electrodes,ll there is no significant improvement in the stabilisation time as compared with chemical cleaning with nitric acid. This contrasts strongly with the case of silver where electrochemical cleaning led to stabilisation times ca. three times faster than those found with chemical cleaning. There is however a marked improvement in the stabilisation time for the copper electrode if the resistance in the external circuit is reduced to a much lower value in the range 10-250 R. The reason for the need for a much lower resistance in the case of copper is at this stage not completely proven but we can propose a model which agrees with the empirical observations. We have shownll that the charge required to activate a silver electrode corresponds approximately to a monolayer coverage so with similar cleaning times the lower resistance or higher current for copper would indicate a greater amount of the 1227 M.L. Hitchman and I;. W. M . Nyasulu surface species which is hindering the metal-amino acid interaction. Just such a species could be the metal oxide. In the case of copper the formation of oxides is thermo- dynamically favourable and it has been shown thatl6 the metal in contact with water and oxygen readily forms oxides which grow up to a limiting thickness of 6.5-9 nm. Silver on the other hand will only adsorb half to one monolayer of oxygen either as adsorbed oxygen or in the form of oxides of ~i1ver.l~ Furthermore the cathodic reduction of silver oxides is thermodynamically more favourable than the reduction of copper oxides and is very efficient.18 These considerations are in accord with what has been found experimentally.It is also interesting to note that if oxides are the cause of surface inhibition then for oxides of copper one would expect dilute nitric acid to be an effective chemical surface-cleaner and for silver oxide an ammonia solution; nitric acid might also be expected to work for silver and this has been observed,19 but it is less effective than ammonia. The oxide model has in addition another interesting consequence. In the case of silver we have previously suggestedlO that there is direct adsorption of a thiol-containing ligand onto the electrode surface to form an absorbed silver-thiol complex.The steady-state potential attained would then be that associated with an electrode of the second kind and there is some circumstantial evidence to support this suggestion from SERS studies of chemisorbed thiourea complexes on copper and silver electrodes.20 This work showed that there is direct bonding of the thiourea via the sulphur atom to the electrode surface and that for both copper and silver this occurs through metal adatoms. The electrochemical reduction of surface oxides would give metal adatoms [cf. reaction (II)] but since we expect more oxide on the copper than on the silver the reduced copper surface will be likely to be a rougher surface with more active sites for metal-amino acid interaction than the reduced silver.Now of course there may well be different kinetic considerations for the interaction of copper and silver with the same ligand and indeed on the basis of the Nernst plots for L-cysteine we shall argue below that this is the case. However all other things being equal a reduced copper surface will be catalytically more active than the reduced silver and so equilibrium will be more rapidly achieved. Interestingly the stabilisation time of ca. 0.1 min for the copper- L-cysteine system with electrochemical cleaning is significantly shorter than the stabilisation time ca. 6 min which we have reported for the silver-I.-cysteine systems. Recently it has been suggested4* 21y 22 that the potential response of a copper indicator electrode to various complexing ions and ligands may be due to the disturbance of the local copper ion concentration at the electrode surface by the complexing agent.This model is not incompatible with the idea of the bonding of the ligand to the surface through an adatom for in this case adsorption could occur after complexation with the coordination configuration of the metal atom being completed with a bond to the surface. The function of the cathodic pretreatment would still be to produce a highly active surface either to facilitate the adsorption of the metal-ligand complex onto the surface or to facilitate the dissolution of metal atoms and hence the formation of metal ions. Further electrochemical studies are in progress to try and distinguish between these various possibilities.Table 1 summarises the least-squares Nernstian slopes with the associated standard errors for the various pretreatment procedures with L-cysteine over the concentration range 10-2-10 mg (0.083-83 mmol dm-3) and at pH 8. The improved behaviour of the cathodically pretreated electrodes is again apparent. Results at other pH values with different buffer solutions showed similar improvements for cathodic cleaning. Except for the untreated electrode the Nernst slopes are all ca. -30 mV per decade which suggests a two-electron process. In the absence of a sulphur atom complexation of transition metals with simple a-amino acids occurs with the a-amino nitrogen and the carboxylic oxygen.23 Stability constants increase though in the ligand atom series 0 < N < S.24 So in the presence of a thiol 1228 Direct Potentiometry with a Cu Electrode Table 1.Slopes of Nernst plots for L-cysteine pretreatment procedure 28.1 f0.8 23.2f 1.1 28.1 f 0.7 none emery paper chemical elect roc hemical batch solution flowing solution group complexation will be likely to involve the thiol sulphur and the a-amino nitrogen and we can write the potential determining equilibrium at a copper adatom as SH NH + Cu-Cu(ads) + 2 CH2-CHCOO-(aq) " - slope/mV per decade I 31.8 k0.3 30.5 & 0.6 ) (ads) + 6H+(aq) + 4e-. (111) I I cu-cu ( 1 S-CH NH-CHCOO- 2 Assuming the copper adatoms and the chemisorbed chelate have unit activity the corresponding Nernst equation is RT [~-cysteine] 2F In [H+I3 * E = @ - - - The copper complex of reaction (111) is different from that with e.g.glycine in that the a-amino nitrogen does not act as a donor atom but loses a proton to form a normal covalent bond with the copper. Certainly a donor bond formed initially from the nitrogen to the copper will bestow some acid character on the remaining hydrogens and facilitate proton release. Without this type of complexation it is difficult to obtain an expression which not only gives a Nernst slope of -30 mV per decade but which also has a pH dependence in accord with experiment. Fig. 3 shows this dependence. The slope in the range pH 4-8 is - (77.3 1.8) mV per pH unit which is in reasonable agreement with that predicted by eqn (1). Furthermore the lower dependence of potential on pH above pH 9 is what one would expect for the involvement in complexation of the thiol group (pK 8.5)25 and the amino group (pK 10.8).26 Also plotted in fig.3 is the pH dependence of the copper electrode in the absence of L-cysteine. The lack of any systematic dependence of potential on pH up to pH 10 is a welcome feature; the slope of -(59.1 k5.0) mV per pH unit above pH 10 is close to that expected for a simple Cu I Cu(OH) couple. A precision study showed that at the 95% confidence level for seven measurements potentials could be measured with a precision of k0.3 mV. For a concentration of lo- m g ~ m - ~ the difference in potential (AE) between the test solution and the background electrolyte was 20mV and so the precision is clearly good enough for quantitative assay above 0.1 mmol dm-3.Using the suggestion of Midgley,' of taking the detection limit as the concentration at which there is only 5% chance of failing to detect the determinand the detection limit for L-cysteine was in fact shown to be ca. 16 pmol dm-3. M. L. Hitchman and F. W. M . Nyasulu 800 I / / 600 - \ I I 1 I / T > E -1000 2 10 8 I 4 6 PH Fig. 3. The effect of pH on the potential of a copper electrode in (a) L-cysteine (10 mg ~ m - ~ ) plus sodium nitrate (2 mol dm-3) and (6) sodium nitrate (2 mol dm-3). tlmin t/min 1 .o 0 . 0.9 2 4" 08 Fig. 4. . Stabilisation curves for electrode potentials with L-histidine (0.02 mg ~ m - ~ ; pH 7).(a) Chemical cleaning; (6) electrochemical cleaning. 1229 12 14 1230 -2 - E n -. 1.4" Qi -3 d v - 4 0 I v c s - - - n ri; -3.0 - -5.0 Direct Potentiometry with a Cu Electrode -1.0. - 2.0 - -4.0 - 2 t/min ( b ) I I I 1 1 3 I L-Histidine Fig. 4 shows that cathodic electrochemical pretreatment again leads to faster potential stabilisation than does chemical cleaning. Applying the same theoretical analysis given previouslylO for the particular case of a thiol interaction with a silver electrode to the more general case of the formation of a metal-amino acid complex on the metal-electrode surface the variation of the electrode potential with time is expected to have the form where E,, Ee and Et are the potentials initially finally and at time t respectively OM, and 8, are the initial surface fractions of electrode covered with metal adatoms and the complex a is a system parameter [see eqn (16) ref.(lo)] and k; and k& are the exchange frequencies on metal adatoms surrounded by other adatoms and metal adatoms adjacent to sites with the impurity X. The impurity X we have argued above could well be the oxide of the metal. The two different exchange frequencies will then simply reflect the fact that adsorption of an amino acid on a surface area composed largely of reduced metal atoms is likely to be kinetically different from that on an area of mixed metal and metal oxide sites. Fig. 5 shows data from fig. 4 plotted in the form of In (1 - Et/Ee) us.t. For chemical cleaning there are two sections to the plot which on the basis of eqn (2) would correspond to reactions on two types of sites. Using the metal-oxide model this would suggest that chemical pretreatment does not remove all the oxide. The slope of - 1.2 min-l for the region marked M and of - 0.9 min-l for that marked X indicates 1231 H M. L. Hitchman and F. W. M. Nyasulu a ratio of activities for the two types of sites of ca. 1.3 1. With electrochemical cleaning only one type of site is indicated which suggests that cathodic pretreatment is much more efficient at generating clean copper sites. Also the slope of - 11.5 min-l means that the copper sites produced by electrochemical reduction are very much more active than either type of copper site for the chemically cleaned electrode.If the oxide model is correct this is not too surprising for as we have pointed out above cathodic reduction of the oxide will be likely to produce a very active surface. In this instance the two-site model provides a reasonable explanation of the observed potential transient but it is interesting to note that an alternative general model for heterogeneous kinetics has recently been suggested28 in which there is a Gaussian distribution of sites with a corresponding distribution of exchange frequencies. When the dispersion of the Gaussian distribution is large such as one may have on an untreated surface a semi-logarithmic plot of the type given in fig. 5 will be expected to have several linear regions or in some instances to have no part of the plot linear over any significant time interval.On the other hand when activation of the electrode surface is carried out e.g. by cathodic pretreatment the dispersion will tend to zero and a classical first-order decay plot will be obtained. In the light of these comments it would be interesting for future work to make a more detailed examination of the variation of potential with time especially for the earlier and final parts of the transient. With electrochemical pretreatment the Nernst plot of L-histidine has a slope of - (5 1.6 +_ 0.7) mV per decade over the concentration range 10-3-10 mg (6.5 x 1OV3-65 mmol dmP3). This slope is sufficiently close to 59 mV per decade to suggest a one-electron equilibrium.The variation of potential with pH at constant L-histidine concentration is linear with a slope of - (37.1 & 0.6) mV per pH unit over the pH range 3-9. These results suggest the release of ca. one proton for every two electrons exchanged. On the basis of the increase of stability constants normally found in the ligand atom series 0 < N < S one might expect complexation of the copper solely with the a-amino and imidazole nitrogens. However detailed studies of the copper-histidine system in solution have that in fact there are probably at least six different structures for the complexes with both 1 1 and 1 2 ratios for metal ligand being possible and with chelation occurring with the carboxylate oxygen and imidazole nitrogen as well as just with the two types of nitrogen.An equilibrium which is in accord with a suggested structure for a copper-histidine complex and with the experimental observations given above is as follows Cu-Cu(ads) + 2 N - - f C W O O - y H 3 i H H (as) (ads) + H+(aq) + 2e-. The corresponding Nernst equation predicts a slope of -59 mV per decade change of L-histidine concentration and a slope of ca. -30 mV per pH unit which are in reasonable accord with the observed values of - 52 mV per decade and - 37 mV per pH unit. Above pH 9 there is little potential dependence on pH and this is in accord 1232 Direct Potentiometry with a Cu Electrode tlmin tlmin Fig. 6. Stabilisation curves for electrode potentials with bovine albumin (10 mg (a) Chemical cleaning and (b) electrochemical cleaning.with the involvement of the a-amino group with pK = 9.2.26 A mixed chelate of the type given in reaction (IV) but with two water molecules to complete the octahedral symmetry is one of the major complexes suggested for Cu2+ and L-histidine interaction; the formation constant for this complex has a value of Some of the other proposed complexes have comparable values for their formation constants and contributions from these complexes could account for the deviations from simple Nernstian behaviour of the potential with both the concentration and pH. + 0.2 mV at the 95 % confidence level which would readily allow quantitative assay for A precision study similar to that made for L-cysteine monitoring gave a precision of solutions showing a potential ca.20 mV different from that given by the background electrolyte. For L-histidine such a value of AE corresponds to ca. 0.01 mmol dm-3. The detection limit was found to be 2 pmol dm-3. Both the precision and detection limits for L-histidine are ca. an order of magnitude better than the corresponding values for L-cysteine suggesting a greater stability of the copper-L-histidine complex compared to the copper-L-cy s teine complex. Proteins Album ins Fig. 6 shows potential stabilisation curves for bovine albumin with chemical and electrochemical cleaning of the electrode. More rapid stabilisation is found with electrochemical cleaning as in the case of the amino acids. A resistance of 250 R in the external circuit was used to obtain the electrochemical result in fig.6. Reducing the resistance to 10-50 R again led to a faster stabilisation typically 0.5-1 min. Analysis of the potential stabilisation curves according to eqn (2) gives for chemical cleaning a plot with two linear segments of slopes - 1.2 min-l and -0.9 min-l. These slopes are identical to those obtained for L-histidine under comparable conditions. This is in contrast to the case of a silver electrode in contact with an amino acid or an albumin pH 7). Cu (ads) + M. L. Hitchman and F. W. M . Nyasulu (ads) + 3H+ (aq) + 3e-. 1233 where we foundll distinctly lower exchange fluxes for bovine albumin ( - 0.1 1 1 min-1 and - 0.053 min-l) than for L-cysteine ( - 0.282 min-l and - 0.132 min-l).This is what one might expect for interaction of an adatom with a larger molecule where steric factors could retard the rate of complexation. The results for bovine albumin with chemical cleaning of a copper electrode are therefore a little surprising. However with electro- chemical cleaning the behaviour is more in accord with expectations while an improved potential response is obtained with the plot according to eqn (2) showing simple first- order behaviour and with a slope (-3.1 min-l) higher than either segment of the chemical cleaning case the improvement is not as marked as for L-histidine (- 11.5 min-l). Similar results were obtained with human albumin again with the significant feature being the faster potential stabilisation when electrochemical pretreat- ment of the copper electrode was carried out.Under conditions where the test solution (PH 7) was flowing past the copper indicator electrode bovine albumin and human albumin gave Nernstian slopes of -(19.5 20.3) and - (20.2 0.2) mV per decade respectively over the concentration range 0.2-10 mg ~ m - ~ (ca. 1-60 pmol dm-3) for both albumins. These slopes suggest a three- electron transfer and at least in the case of bovine albumin are in accord with a model suggested307 31 for the binding of one equivalent of copper with the protein. This model postulates the involvement of the terminal a-amino group of the aspartyl residue the deprotonated amide nitrogen atoms of the threonyl residue and the histidyl residue and the nitrogen of the imidazole group of the same histidyl residue.The following equilibrium can therefore be proposed Other albumins although not having exactly the same amino acid terminal sequence nevertheless do have a histidyl residue in position three of the peptide chain. So for example human albumin has the sequence aspartyl-alanyl-histidyl. It appears that the bulk of the albumin molecule has relatively little influence on the properties of the terminal sequence. Accordingly the Nernst slope for human albumin would suggest a similar equilibrium to that proposed in reaction (V) for bovine albumin. A study of the binding of copper by the peptide comprising the first twenty-four residues of bovine albumin has produced evidence31 that in addition to the preferred Direct Potentiometry with a Cu Electrode 1234 0 0 6 a 10 12 PH Fig.7. The effect of pH on the potential of a copper electrode in bovine albumin (10 mg cm-3) plus sodium nitrate (2 mol drn-9. site involving the three terminal residues there is interaction with other residues and a second equivalent of copper releases at least one further proton. Interestingly if the Nernstian behaviour of copper electrodes with albumin is examined in batch solutions with prolonged contact the slopes obtained with cathodic cleaning of the electrode are -(16.1 k0.3) mV per decade for bovine albumin and -(14.9_+0.5) mV per decade for human albumin; the corresponding slopes with chemical cleaning are similar but the statistical errors are larger. These slopes indicate a four-electron equilibrium and it could be that in the case where there is a longer contact time between the protein and the electrode there is the possibility of interaction between a second copper adatom and another part of the protein terminal peptide chain.Fig. 7 shows the potential variation with pH for a solution of bovine albumin; human albumin showed the same variation. No measurements were made below pH 5 because precipitation of the protein tended to occur. There is clearly a marked difference between the pH dependence for L-cysteine (fig. 3) and L-histidine discussed earlier and the albumins. For the albumins in the range of pH 6-8 the potential remains practically constant and even above pH 8 there is only a shallow slope (ca. -20 mV per pH unit) for a further two pH units.Above pH 10 the slope is significantly higher but this is not unexpected [cf. fig. 3 curve (b)]. The weak dependence of potential on pH is a welcome feature since it means a copper electrode can usefully be employed for quantitatively monitoring in gradient gel-chromatography of proteins without having to compensate for the changing pH of the eluate. Nevertheless since the proposed equilibrium (V) does involve proton release a more marked dependence of potential on pH might have been expected. The reason for the observed weak dependence is not fully understood at this stage. It could be that the pH of the solution is not exactly that expected for the buffer used since it is well known that proteins themselves have an excellent buffer action because of the large numbers of groups that interact reversibly with protons over a wide pH range; much of the buffer capacity of body tissues for example is due to protein buffers.This effect could be accentuated close to the electrode surface where such a large entity as a protein molecule could radically change the ionic structure of the double layer and annul any normal buffer action of the added buffers. M. L. Hitchman and F. W. M . Nyasulu 1235 For bovine albumin a precision of k0.4 mV at the 95% confidence level was found for AE x 20 mV corresponding to a concentration of ca. 7 pmol dm-3. A detection limit of ca. 4 pmol dm-3 was determined. Similar results were obtained with human albumin. The results show that quantitative determinations down to the micromolar level are feasible.Globulins Studies of bovine and human y-globulins produced results very similar in all respects to those found for the albumins. For example electrochemical cleaning of copper electrodes with a low resistance (10-50 n) in the external circuit gave stabilisation times of 0.5-1 min for both globulins. The slopes of the Nernst plots were also the same as those obtained with the albumins. In flowing solution bovine y-globulin gave a slope of - (1 8.9 f 0.2) mV per decade and human y-globulin a slope of -(19.3 k0.3) mV per decade. The linear portions of the Nernst plots corresponding to these slopes were however over a narrower concentration range than for the case of the albumins 1-10 mg ~ m - ~ and 0.2-1 0 mg ~ m - ~ respectively.The Nernst slopes again indicate a three-electron transfer being associated with complexation. No information has been found in the literature about the complexing of copper with globulins but the similarity to the albumins in the Nernstian response and the known tendency of copper to form square-planar chelate complexes does suggest the possibility of an amino amido imidazole interaction. In batch solutions the Nernst slopes again indicate the complexing of a second equivalent of copper with another part of the molecule since the values fall to - (1 5.9 k 0.2) and - (1 5.0 f 0.4) mV per decade for bovine y-globulin and human y-globulin respectively. The distinctive lack of pH dependence of the electrode potential in the range of pH 6-10 commented on above for the albumins was also found for the globulins.Ribonuclease A and Lysosyme These two proteins gave practically no response with silver electrodes owing to the absence of a thiol group.ll Since copper is able to bind effectively with nitrogen and oxygen atoms a Nernstian response is expected. This is only found however if cathodic cleaning is carried out beforehand. With this pretreatment slopes in flowing solution of -(17.4f0.4) and -(16.6+0.2) mV per decade were obtained for lysozyme and ribonuclease respectively but without it the Nernst plots have slopes close to zero. As reported earlier,13 this improved Nernstian behaviour together with the improved stabilisation time has been exploited for monitoring proteins eluted from a chroma- tography column.More recently we have extended the application to monitoring proteins separated on electrophoresis gels and we shall be reporting on this subsequently. We thank the British Council for a training award for F.W.M.N. and Pharmacia AB Uppsala for the provision of materials and equipment. We also thank Prof. C. J. Suckling and Drs R. Bucher M. Dufton W. E. Smith and M. Trojanowicz for helpful discussions. References 1 A. E. Martell and R. M. Smith Critical Stability Constants VoL 1 Amino Acidr (Plenum Press New York 1974). 2 P. W. Alexander and C . Maitra Anal. Chem. 1981,53 1590. 3 P. W. Alexander P. R. Haddad G. K. C. Low and C. Maitra J. Chromatogr. 1981 209 29. 4 P. W. Alexander P. R. Haddad and M. Trojanowicz Anal. Chim. Acta 1985 171 151.5 A. Libicky and L. Wunsch Collect. Czech. Chem. Commun. 1961 26 2663. 6 Y. A. Gawargious A. Besada and M. E. M. Hassouna Microchim. Acta 1974 1003. 1236 Direct Potentiometry with a Cu Electrode 7 E. M. Athanasiou-Malaki and M. A. Koupparis Anal. Chim. Acta 1984 161 349. 8 T. Sekerka and J. F. Lechner Anal. Lett. 1978 A l l 415. 9 J. C. Westall F. M. Morel and D. N. Hume Anal. Chem. 1979 51 1792. 10 M. L. Hitchman Anal. Chim. Acta 1985 171 131. 11 M. L. Hitchman A. Aziz D. D. K. Chingakule and F. W..M. Nyasulu Anal. Chim. Acta 1985 171 141. 12 T. Y. Toribara and L. Koval Talanta 1970 17 1003. 13 M. L. Hitchman F. W. M. Nyasulu A. Aziz and D. D. K. Chingakule Anal. Chim. Acta 1983 155 219. 14 B. Karlberg and S. Thelander Analyst (London) 1978 103 1154.15 Handbook of Chemistry and Physics ed. R. C. Weast (The Chemical Rubber Co. Boca-Raton Cleveland 1970) p. D102. (Marcel1 Dekker New York 1974) vol. 11 p. 384. 17 J. P. Hoare in Encyclopaedia of Electrochemistry of the Elements ed. A. J. Bard (Marcel Dekker 16 U. Bertocci and D. R. Turner in Encyclopaedia of Electrochemistry of the Elements ed. A. J. Bard New York 1974) vol. 11 p. 191. 20 B. H. Loo Chem. Phys. Lett. 1982,89 346. 21 P. W. Alexander P. R. Haddad and M. Trojanowicz Anal. Chem. 1984 56 2417. 22 P. W. Alexander P. R. Haddad and M. Trojanowicz Chromatographia 1985 20 179. 23 T. Moeller Inorganic Chemistry (John Wiley New York 1952) p. 238. 24 C. S. G. Phillips and R. J. P. Williams Inorganic Chemidry (Clarendon Press Oxford 1966) vol. 11 p. 268. 25 T. Y. Lim in The Proteins ed. H. Newrath (Academic Press New York 1977) vol. 111 p. 245. 26 H. D. Jakubke and H. Jeschkeit Amino Acids Peptides and Proteins (Macmillan Press London 1977) 27 D. Midgley Analyst (London) 1979 104 248. p. 25. 28 W. J. Albery P. N. Bartlett C. P. Wilde and J. R. Darwent J. Am. Chem. SOC. 1985 107 1854. 29 D. R. Williams J. Chem. Soc. Dalton Trans. 1972 790. 30 T. Peters and F. A. Blumenstock J. Biol. Chem. 1967 242 1574. 31 R. A. Bradshaw W. T. Shearer and F. R. N. Gurd J. Biol. Chem. 1968,243 3817. 18 N. A. Shumilova and G. V. Zhutaeva in Encyclopaedia of Electrochemistry of the Elements ed. A. J. Bard (Marcel Dekker New York 1974) vol. 111 p. 1. 19 A. Aziz MSc. Thesis (University of Salford 1981). Paper 511895; Received 28th October 1985

ISSN:0300-9599    

DOI:10.1039/F19868201223

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J. Chem. SOC. Faraday Trans. I 1986,82 1237-1243 Amperometric Enzyme Electrodes Monika J. Green* Genetics International Inc. 11 NuBeld Way Abingdon OX14 1RL H. Allen 0. Hill Inorganic Chemistry Laboratory South Parks Road Oxford OX1 3QR The design of a new type of amperometric enzyme electrode exploiting ferrocene as mediator between enzyme and electrode is described. Its use is illustrated by application to the determination of glucose through the use of glucose oxidase. Other flavoproteins accept ferrocene as a mediator as do peroxidases. The synthesis of ferrocene analogues of drugs e.g. lidocaine and theophylline permits the development of analytical methods employing components of the immune system. Enzyme electrodes are mostly associated with systems or devices in which the true analyte is a substrate or product of the enzyme reaction detected potentiometrically or amperometrically.A typical example of the former method of detection is one in which the enzyme reaction causes a change in the pH of the solution in close proximity to the sensing electrode. There seems little limit to the ingenuity that can be brought to the combination of traditional methods with modern physical methods of detection. We have been concerned to exploit enzyme electrodes in which an artificial mediator traps electrons at the active site of the enzyme and transports them to or from the electrode. Glucose Electrode The archetypal amperometric enzyme electrode is that' used to analyse for glucose in which the disappearance of dioxygen or the appearance of hydrogen peroxide is associated with the glucose oxidase-catalysed reaction glucose + 0 -+ gluconolactone + H,O,.The flavin within glucose oxidase is reduced and it in turn reduces oxygen to hydrogen peroxide. The reaction can be followed electrochemically either through the loss of oxygen or by oxidation of the hydrogen peroxide formed. Sample dilution is necessary owing to the low oxygen tension within samples. Many configurations based on this reaction are available in which the different methods of immobilisation of the enzyme or different membranes are employed to enhance selectivity. The latter need arises because to a certain extent the selectivity inherent in the properties of the enzyme is lost since its redox properties are not monitored directly.It would seem attractive to couple these enzymes directly to electron transfer to the electrode. Unfortunately to date only very slow electron-transfer reactions of such complex enzymes have been observed and devices which exploit this direct electrochemistry of the enzyme tend to have long response times. The application of lessons learnt from the study of simpler proteins may remove these problems. Several attempts have been made to design an enzyme-based glucose electrode that obviated the necessity of sample dilution. In many of these devices attempts have been made to replace the electron acceptor (oxygen) by a spectrophotometric dye. Unfortu- nately these dyes although possessing good spectrophotometric properties (i.e.high 1237 1238 A mper ome t r ic Enzyme Electrodes Fe 0.0 0.1 Fig. 1. Bis(cyclopentadieny1)iron. EIV vs. SCE 0.2 0.3 0.4 0.5 Fig. 2. D.c. cyclic voltammogram of ferrocene monocarboxylic acid (a) in the presence of D-glucose and (b) (a) plus glucose oxidase. extinction coefficients) have pH-sensitive Ea values are autoxidisable or do not show good reversible electrochemistry. A mediator is required that is not pH-sensitive is not autoxidisable shows good one-electron electrochemistry similar to the ferricyanide/ ferrocyanide couple but can be derivatised to tune the potential and yet maintain good electrochemistry. For some applications the requirement for dioxygen is disadvan- tageous. In order to overcome this problem a new generation of oxygen-insensitive mediators of electron transfer has been introduced.These are based on the organometal- lic compound ferrocene bis(cyclopentadienyl)iron fig. 1. This has a well behaved redox couple exhibiting fast electron-transfer reactions with properties easily modified by sub- stitution. It actszp4 as a most effective mediator to a wide range of oxidoreductases including glucose oxidase. A wide range of derivatives exist which spans a potential window over several hundred mV. 30 25 2 20 \ CI 0 15 10 5 M . J . Green and H . A . 0. Hill 10 I I I I I Fig 3. Dependence on the current produced by the glucose oxidase electrode on glucose. 20 30 40 50 60 70 [glucose]/mmol dm-3 Table 1. Substituted ferrocenes as mediators to glucose oxidase Ei/mV us.SCE k/10-5 dm3 mol-l s-l ferrocene derivative 1,l '-dicarboxylic monocarboxylic hydroxymethyl 2-aminoethyl Ferrocene monocarboxylic acid shows good one-electron quasireversible electro- chemistry AEp = 60 mV (see fig. 2) in the presence of D-glucose. On addition of enzyme a striking change is observed and a large catalytic current flows at oxidising potentials which is particularly evident at low scan rates (< 10 mV s-l). This observation can be interpreted as resulting from the reduction of the flavin within glucose oxidase. The reduced enzyme-bound flavin can then be reoxidised by the ferricinium ion and the ferrocene formed is oxidised at the electrode. This mediated enzyme electrode gives a linear response to glucose up to 30 mmol dmd3 (fig.3). Using a range of ferrocenes the second-order rate constants obtained under conditions of substrate excess are such that a linear relationship does not exist between Ei and the rate constant. From these results it is apparent that the charge on the mediator is important29 (table 1). Positively charged ferrocenes (2-aminoethyl ferrocene pK = 9.5) are far better mediators to glucose oxi- dase than negatively charged ferrocenes (dicarboxylic acid). This may be interpreted as showing that the active site of glucose oxidase is negatively charged and therefore would favour positively charged species. These observations are consistent with the observa- tions of Kulyss who found on sulphonating naphthoquinone derivatives that they were less efficient as mediators to glucose oxidase than the unsulphonated naphthoquinones even though the value of Ei of the mediator increased on sulphonation.The immobilised ferrocene enzyme electrode and the soluble system are both insensitive to pH over a range 6-9. Furthermore the commonly found electroactive 1239 I 0.26 1.80 9.00 44.00 395 275 185 200 1240 Amperometric Enzyme Electrodes Table 2. The use of ferrocene monocarboxylic acid as a mediator to a number of flavoprotein oxido-reductases substrate / c / ~ O - ~ dm3 mol-l s-l enzyme 4.0 0.2 0.2 0.1 2.0 pyruvate xanthine sarcosine NADH NADPH co 4.0 67.0 2.0 L-lactate methanol pyruvate oxidase xanthine oxidase sarcosine oxidase diaphorase glut at hi one reductase carbon monoxide oxidase flavocytochrome b alcohol dehydrogenase species in serum do not interfere with the activity of the electrode to any appreciable extent.Oxygen interference becomes a problem if the enzyme electrode is expected to operate in a neat atmosphere of oxygen. Thus the resulting electrode exhibits excellent linearity of response to glucose over the physiologically relevant range shows essentially no difference in response whether the analysis is carried out under aerobic or anaerobic conditions and provided an anion exclusion membrane is placed over the electrode exhibits no interference from other metabolites commonly found in blood. It works equally well in blood (the latter having been treated with heparin) or plasma.Until such times as devices based on direct electron transfer become available it is likely to provide the basis for a wide range of clinically useful electrodes. The results obtained with the ferrocene enzyme electrode compare favourably with the results obtained using a Yellow Springs glucose analyser. An interesting extension of the use of the glucose electrode is to couple it to other analytes using enzymes that compete with glucose oxidase for its substrate. In the presence of hexokinase glucose is diverted to glucose-6-phosphate provided that ATP is present in solution. It is not surprising therefore that the glucose electrode in this configuration can be used to detect ATP. Through the detection of ATP NAD and NADP an even wider range of analytes become accessible including clinically relevant enzymes.For e~ample,~ the association of the glucose electrode with hexokinase ADP and creatine phosphate allows the detection of the enzyme creatine kinase used in the diagnosis of myocardial infarction. Other Flavoprotein Oxidases The ferricinium ion acts4 as an electron acceptor to a number of flavoproteins and some of these are given in table 2. In these cases the mediator has not been optimised towards the enzymes. Two enzymes tried oxalate oxidase and choline oxidase did not couple with ferrocene which is consistent with the observation that the only published mediator to these enzymes is oxygen. A limited study on the effect of charge on the mediator on the second-order rate constant of ferrocenes with diaphorase has been undertaken.In this case ferrocenes without any charge seem to be better mediators. This assay can be used to detect NADH generated by over 200 different dehydrogenases. It is normally unsatisfactory to detect NADH directly by oxidation at an electrode as a large overpotential is required. This also results in radical formation and tends to complicate its use in coupled reactions. Peroxidases In the previous examples given the ferricinium ion acts as an electron acceptor. However ferrocene and its derivatives can also act as electron donors. Horseradish peroxidase will reduce hydrogen peroxide to water and ferrocene can then act8 as an electron donor to d - 0 6 a M. J. Green and H.A . 0. Hill [H2021/pmol dm-3 4 2 Fig. 4. Calibration graph for the horseradish peroxidase-ferrocene monocarboxylic acid system. [cholesterol linoleate]/mmol dm-3) L n 2d Fig. 5. Current response to cholesterol linoleate. the oxidised iron in the haem protein. The ferricinium ion can be reduced at the electrode. This enzyme coupled system can be used to detect hydrogen peroxide formed by a number of oxidases. The sensitivity of the assay is such that 0.5 nmol dm-3 hydrogen peroxide may be detected (fig. 4). Cholesterol Current assays for cholesterol are enzymatically based spectrophotometric assays. Unlike the glucose assay they require several enzymes. A number of electrochemical assays can be envisaged for detection of cholesterol.In all cases the cholesterol ester must first be dissociated from a lipoprotein complex usually by the use of a surfactant. The ester is then hydrolysed by an esterase to form the free cholesterol. It can then be (a) oxidised by a dehydrogenase the NADH formed being subsequently detected by a coupled reaction based on diaphorase ; (b) oxidised by a specific flavoprotein oxidase coupled to 1241 l o 1242 Amperometric Enzyme Electrodes 0 6 5 u c 4 -. 9 g 3 2 .3 u 0 3 2 * v 1 0 COCH,N(C 100 ‘C H 80 120 40 0 60 conjugate concentration/pmol dmh3 Fe Fig. 6. Ferrocene derivative of lidocaine. 20 Fig. 7. Relationship between the current produced and the concentration of the ferrocene derivative of lidocaine.ferricinium ion; (c) oxidised by a non-specific oxidase that is reoxidised by oxygen; the hydrogen peroxide generated can then be detected by the peroxidase coupled assay. All three assays are in principle sound. However in practice the peroxidase based assay proved to be the easiest to use and a linear response to cholesterol linoleate was obtained9 over the physiologically relevant range (fig. 5). Enzyme-linked Homogeneous Immunoassays Drug concentrations often have to be monitored carefully too low an effective concentration in the plasma and the drug is not effective; too high and the drug becomes extremely toxic. These drugs which are usually small-molecular-weight compounds tend to be therapeutically effective in the micromolar to nanomolar range.In some cases the drugs may be electroactive but direct electrochemical detection leads to non-specific analyses unless it is associated with e.g. a column separation. In some cases if the drug is charged it may be detected by an ion-selective electrode although in many cases this will require sample pretreatment. It would be extremely valuable to link together immunochemistry with electrochemistry. Thus one would have the specificity obtained 1243 M. J . Green and H . A . 0. Hill by using the former coupled with the ease of operation of the latter. It was with this in mind that an amperometric immunoassay was designed.l0 In this case a modified ferrocene was employed to act as a mediator to certain flavoprotein oxidases; a ferrocene modified by a drug should also act as mediator to e.g.glucose oxidase. However in the presence of the antibody to the drug which also binds the drug-ferrocene complex the diffusion properties of the modified ferrocene are greatly affected and it is no longer able to act as a mediator to the flavoproteins. The immunoassay can then be configured in which a fixed amount of antibody ferrocene-labelled drug enzyme and substrate are used; the only variable is the unknown concentration of the drug. In the absence of any drug the ferrocene-drug conjugate will bind to the antibody and no ferrocene-drug complex will be available to mediate to the flavoprotein. In the presence of the drug however a competition will occur between ferrocene-drug complex and drug for the antibody.The more drug is present in the sample the less ferrocene-drug complex will be bound and the greater the catalytic current will become. Ferrocene derivates of lidocaine (fig. 6) a drug used in the treatment of arrythmia and theophylline used in asthma were prepared. Both derivates acted as electron acceptors from reduced glucose oxidase. Dose response curves were obtained as illustrated in fig. 7 for the lidocaine assay. These ferrocene derivates can then be used in rapid homogeneous enzyme immunoassay s. Conclusions As judged by the development of either direct amperometric enzyme electrodes or mediated amperometric enzyme electrodes there seems no basic impediment to their exploitation. They should find ready acceptance since they can be cheap the measurements are easily made and the output readily understandable.To increase the degree of specificity exhibited they may incorporate for example components of the immune system or features inherent in the so-called DNA probes. They might even be exploited in vivo in cases where that is desirable. The deficiences of current devices are those common to all electrodes achievement of reproducibility in manufacture and of stability in storage and use. We thank Professor I. J. Higgins and Drs M. Ball A. E. G. Cass G. Davis J. E. Frew S. I. Libor C. J. McNeil and D. Scott for their help. References 1 L. B. Wingard Fed. Proc. Fed. Am. SOC. Exp. Biol. 1983,42 288. 2 A. E. G. Cass G. Davis G. D. Francis H. A. 0. Hill W. J. Aston I. J. Higgins E. V. Plotkin L. D. L. Scott and A. P. F. Turner Anal. Chem. 1984,56 667. 3 G. Davis I. J. Higgins and H. A. 0. Hill J. Biomed. Eng. 1984 6 174. 4 A. E. G. Cass G. Davis M. J. Green and H. A. 0. Hill J. Electroanal. Chem. 1985 190 117. 5 D. Scott personal communication. 6 N. K. Conos J. J. Kanapieniene and J. J. Kulys Biochim. Biophys. Acta 1984 767 108. 7 M. J. Green G. Davis and H. A. 0. Hill J. Biomed. Eng. 1984 6 176. 8 J. Frew M. A. Harmer H. A. 0. Hill and S. I. Libor J. Electroanal. Chem. in press. 9 M. Ball J. E. Frew M. J. Green and H. A. 0. Hill to be published. 10 K. di Gleria M. J. Green H. A. 0. Hill and C. J. McNeil Anal. Chem. in press. Paper 5 / 1896; Received 23rd October 1985

ISSN:0300-9599    

DOI:10.1039/F19868201237

出版商:RSC    

年代:1986

数据来源: RSC

摘要:

J . Chem. Soc. Faraday Trans. I 1986,82 1245-1258 Chemically Modified Electrodes for the Electrocatalytic Oxidation of Nicotinamide Coenzymes Lo Gorton Analytical Chemistry University of Lund P.O. Box 124 S-221 00 Lund Sweden Methods for the oxidation of the cofactor NADH are discussed and the criteria for a successful mediator are set out. The kinetics of the system are analysed and linear free-energy relationships found for a number of different mediators. The variation of the reaction rate with pH is discussed. The successful oxidation of NADH using a modified electrode with Meldola’s blue is described. A biosensor is a combination of biologically derived material (enzymes antibodies organelles or whole cells) with a transducer having an electrical output ~ignal.l-~ Rapid progress in biochemistry as well as in electrochemistry has improved the possibilities of making novel types of biosensors for a variety of substrates.This development could make substantial impact in a number of areas e.g. biotechnology food and agricultural product processing health care medicine and pollution monitoring. Enzymatic redox reactions are particularly closely related to electrochemical trans- ducers since an electron-transfer step is involved in the natural cycle of the enzyme. The coupling is however far from straightforward. All redox enzymes depend on a cofactor as the natural partner in the exchange process. Charge transfer between the redox enzyme and an electrode can be either through a direct transfer between the active site and the electrode or through a soluble charge-carrier.Direct electron exchange with electrodes has been reported for a number of redox enzymes with bound c~factor~.*-~ In most cases however either large reaction barriers or steric hindrance prevent the electron transfer between these cofactors and electrodes or other redox couples. The inaccessibility of the charge is probably the result of a natural selection process acting to preserve the energy within the electron-transfer chain of the cells.9 The largest group of redox enzymes known today is the one where the enzymatic process depends on a soluble nicotinamide coenzyme. These enzymes are usually termed dehydrogenases. Some 250 depend on the NAD+/NADH couple and some 150 on NADP+/NADPH.7 A successful marriage between these enzyme-catalysed reactions and a transducer might therefore be expected to be of great utility. The active sites of the nicotinamide-dependent dehydrogenases seem to be well embedded into the enzyme however and there is a cleft for a very specific binding of the soluble coenzyme. The prospect of finding a direct reaction at an electrode material is therefore bleak. It has however been reported that an alcohol dehydrogenase-NAD+-carbon black complex could oxidize ethanol with a half-wave potential of -40 mV us. SCE.1° An inhibitor blocked the wave and the reduced coenzyme-carbon black alone required + 180 mV. The enzyme seems therefore to be involved in the reaction but the mechanism is unknown at present. All work on biosensors for NAD+-dependent dehydrogenases have with the possible exception mentioned above utilized the reaction scheme where S is the oxidized and SH the reduced (e.g.alcohol glucose amino acid) form of the substrate. The soluble t NAD+ = nicotinamide adenine dinucleotide ; NADP+ = nicotinamide adenine dinucleotide phosphate; NADH NADPH = the reduced forms of NAD+ and NADP+ respectively. 1245 1246 nicotinamide coenzymes discussed in this paper are excellent examples where biological redox species exhibit irreversible redox behaviour at electrode surfaces and also exhibit a strict fastidious choice of electron transfer with other redox species (charge carriers). In this paper only the non-enzymatic oxidation of NADH is relevant to the discussion as the non-enzymatic reduction of NAD+ faces severe problems and has not yet been utilized in biosensor applications.(2) Elec troca taly t ic Oxidat ion of Nico t inam ide Coenzymes -e- NADH -+ NADH'+ m N A D * e NAD+. fast +e- This scheme was tentative and did not explain all the experimental observations; the conclusion by Jensen and E l ~ i n g ~ ~ that there is still considerable uncertainty about the detailed nature of the electrochemical oxidation of NADH therefore still holds true. The recent suggestion by Samec and E l ~ i n g ~ ~ that the oxidation at solid Pt and Au electrodes might occur through a mediated mechanism is particularly interesting because it gives some hope that studies of chemically modified electrodes (CME) might help in the interpretation of reactions on metal electrodes.One of the steps would involve a bond between the hydrogen atom to be transferred from NADH and a surface atom by the formation of the intermediates NAD' and OH;,,(H,O). The radical side reactions are of less importance at low NADH concentrations and a quantitative yield of enzymatically active NAD+ may be of less importance for a transducer. The success of the direct oxidation concept has been demonstrated by a number of biosensor application~.~~-~~ Oxidation of NADH -H+ +Hf intermediate Direct Electrochemical Oxidation Early experiments by Burnett and Underwoodll indicated a quantitative recovery of NAD+ upon electrochemical oxidation of NADH and Blaedel and Jenkins12 used the reaction analytically.Coughlin et al. l3 and Jaegfeldt et all4 showed that enzymatically active NAD+ was produced almost with 100% efficiency under specified conditions. Work done by Elving and c o ~ o r k e r s ~ ~ - ~ * showed that the adsorption of NAD+ played an important role on carbon electrodes and that it was essential to control such experimental parameters as electrode material pretreatment and conditioning of the electrode. The quantitative recovery of NAD+ at platinum was also dependent on a pretreatment and possibly on the fast rotation of the electrode so that intermediates were removed from the electrode surface. Electrode fouling is a major reason for the less than 100% yield of NAD+ and for the shortened lifetime of the electrode especially at NADH concentrations > 0.5 mmol dm-3.Mechanistic s t ~ d i e s l ~ - ~ ~ have been made on NADH or on model compounds and it has been discussed whether the reaction is a two-electron transfer or involves two single-electron transfer steps. Diagnostic tests over a wider concentration range than that used earlier22 showed that the evidence for consecutive one-electron transfer steps became stronger the more the concentration was increased. There has also been some discussion whether an acid-base reaction is involved in the rate-limiting step.22* 23 Tests at higher NADH concentrations indicate22 a second order pH-dependent chemical reaction and the following scheme was proposed -e- slow L. Gorton 1247 The high applied potential necessary for the direct oxidation of NADH due to the high overvoltage at solid electrode~l~-~l may however open up the analytical system not only to high background currents but also to many interfering reactions that can contribute to the response.Mediated NADH Oxidation r------ electrode (3) where M, and Mred denote the oxidized and the reduced forms of the mediator respectively. In a number of p a p e ~ s ~ ~ ~ 35-41 homogeneous electron-transfer mediators have success- Soluble Mediator One way to increase the electron-transfer rate (and decrease the large overvoltage) is the introduction to the analyte solution of an electron-transfer mediator that (i) in a first step reacts very quickly with NADH and (ii) in another step is reoxidized electrochemically at a potential substantially lower than that at which NADH is directly oxidized.Only a restricted number of mediators have been found to provide redox coupling between electrodes and biological redox 34 The most important ones for NADH are the ortho- and para-quinones quinone imines and phenylene diimines derivatives thereof and those where these crucial structural elements have been incor- porated into larger molecules such as in indophenols phenazines and phenoxazine~.~~ The reaction scheme including the mediator is fully been used for the detection of NADH. Mediator Immobilized on Electrode Surface Chemical modification of electrode surfaces provides a powerful tool for giving the electrode electrocatalytic proper tie^.^^^ 43 It was shown in many instances than an electrode with a surface-immobilized redox mediator will display similar or the same properties as those of the soluble mediator.43 A more compact design of the NADH mediating system mentioned above would thus be to attach the mediator directly to the electrode surface; the mediator could then be reused for many cycles.The introduction of the mediator functionality on an electrode can be achieved in either two ways. (i) A drastic oxidation of a carbon electrode will introduce in a rather unselective way oxygen-containing functionalities on its surface.l29 329 44-49 In a recent paper it was clearly demonstrated that ortho-quinone groups were (ii) A mediator that is known to react very fast with NADH homogeneously in solution can be immobilized on the electrode surface.The attachment of the compound can be done in different ways covalent binding or adsorption on the electrode surface inclusion into the electrode material or the introduction of the mediator into a polymeric backbone that is deposited on the electrode surface. As given in table 1 the initial reports involved attachments of molecules containing ortho-quinone group^,^^^^ followed by alkylphena~inium,~~ p a r a q ~ i n o n e s ~ ~ ~ ~ phenylenediimines62 and p h e n o x a z i n e ~ . ~ ~ ~ ~ A successful transducer has to meet a number of demands (1) a large decrease in overvoltage during electrocatalysis; (2) good long-life stability (weeks to months) (a) irreversible immobilization of mediator function (b) chemical stability of the functional group (hydrolysis light decomposition chemical oxidation etc.) (c) electrochemical stability ( d ) stability in the presence of NADH (no radical side reactions); (3) fast Table 1.Chemically modified electrodes for electrocatalytic oxidation of NADH mediator group immobilization stability k,/dm3 mol-l s-l approximate decrease in ,/V as. SCE overvoltage/mV ref. 200-250 50 51 52 53 200-250 200-250 200-250 200-250 200-250 covalent covalent polymer polymer polymer adsorbed +0.170 +0.170 +0.160 (PH 6.35) +0.170 +0.170 +0.170 54 55 56 +0.100 -0.150 -0.160 adsorbed mixed with carbon adsorbed adsorbed mixed with carbon 300 500 500 550 350 300 550 orthoquinone ort hoquinone orthoquinone ort hoquinone orthoquinone ort hoquinone orthoquinone 172-naphthoquinone N-methyl phenazinium (NMP+) N-ethyl phenazinium chloranil mercaptoquinone phen ylenediimines Meldola Blue (phenoxazine) 3-B-naphthoyl-Nile Blue -0.210 + 0.050 57 58 58 59 61 62 63 65 64 elect r o pol ymer mixed with carbon adsorbed adsorbed +0.100 -0.175 - 0.220 (phenoxazine) adsorbed 66 + + + 1x103 600 550 300 0 - + 0.200 -0.210 400450 - - - + + 3~ 1 0 4 + + + + - - + + 8 ~ 1 0 ~ ~ 4547,49 + + + + 1 .3 ~ 104-1.5x 67,68 69 300 1,2-benzophenoxazine-7-0ne oxidized carbon conducting radical salt of NMP+TCNQ-C immobilized diaphorase and ferrocene + + 105 a From ref.(49). Calculated values from ref. (68). TCNQ = tetracyanoquinodimethane. L. Gorton 1249 reaction rates (a) fast electron-transfer rate between the electrode and the immobilized mediator (b) fast charge transfer within the film (the latter applying to polymeric and multilayer coatings) (c) fast reaction rate between NADH and the immobilized mediator. These will now be discussed. Decrease in Overvoltage The decrease in overvoltage is closely related to the formal potential (E"') of the mediating group on the electrode surface. In principle this means that the lower the E"' of the mediator is the more is the overvoltage decreased see table 1. A measure of the decrease is either (1) the difference between the peak potentials (E,) of cyclic voltammograms or (2) the difference between the of rotating electrode scans for the catalysed and uncatalysed waves.For the simplest case where only the mass transport of NADH and/or the rate constant [k (in dm3 mol-l s-l)] between NADH and the mediator can limit the reaction (see below) Ep and El12 for the catalysed waves are given by eqn (4)'O and (5):'l Ep = Eo'+- [0.78 +In (D#iDH/kl r) +In ( ~ F u / R T ) ~ ' ~ ] RT nF where and id(L) = 0.620 nFAD#i,H v - ' / ~ CNA, w1i2. (4) (7) DNA, is the diffusion coefficient of NADH (in cm2 s - l ) C,,, is the bulk concentration of NADH (in mol ~ m - ~ ) I' is the coverage of the mediator (in mol cm-2) m is the rotational speed (in rad s-l) v is the hydrodynamic viscosity (in cm2 s-l) and v is the sweep rate (in mV s-l) and all other parameters have their usual meanings.As can be seen factors other than E"' affect the effective decrease in overvoltage e.g. sweep rate (u) coverage (r) rotational speed (m) etc. The picture becomes even more complex when all other steps that may contribute to the kinetics of the charge transport between NADH in solution and the electrode are influential (see When used as a sensing device the optimal applied voltage of the chemically modified electrode for NADH determinations should be in the range where the background current switches sign and where interferents contribute as little as possible. As many of these interferents are also subjected to high overvoltages e.g. ascorbic acid uric acid bilirubin neurotransmitters oxygen etc.the increased electron-transfer rate obtained by the immobilized mediator should be selective towards NADH. The optimal voltage at pH 7.0 is generally ca. - 100 to + 100 mV us. SCE i.e. the decrease should thus be ca. 450-600 mV cf. table 1. Lifetime The increased understanding of the operating mechanism has resulted in an increase in lifetime from a single cycle50 to several weeks,72 as indicated in table 1. Chemically modified electrodes made with orthoquinones could be made fairly stable by themselves but their lifetime was limited in the presence of NADH. An intermediate ~emiquinone~~ was assumed to react with a radical intermediate of the coenzyme NADH+ to give an inactive compound which poisons and blocks off the surface (see scheme 1).This radical side reaction seems to be of little or much less importance for chemically modified electrodes made with alkylphena~inium~~ or p h e n ~ x a z i n e ~ ~ - ~ ~ mediators. The 1250 Electrocatalytic Oxidation of Nicotinamide Coenzymes $ONHz C-ONH + - o-tl - Scheme 1. Deactivation mechanism proposed for an adsorbed mediator of the orthoquinone type and NADH.56 former are light-sensitive and can be dealkylated electrochemically and are thus less stable by themselves than the phenoxazines. The latter may however hydrolyse in alkaline solutions with different rates depending on 66 The work up to now has demonstrated that an adsorbed phenoxazine can be used for electrodes with lifetimes of one day to several weeks in neutral or acid s o l u t i o n ~ .~ ~ ~ 73 The half-life of adsorbed 3-#I-naphthoyl-Nile Blue (6 aromatic rings positively charged in the oxidized state) was 8 days in flowing solutions and 15 days in static The covalent immobilized compounds and the polymers remain on the surface for extended periods of time whereas the adsorbed mediators desorb spontane~usly.~~ The rate of desorption decreases with an increased number of conjugated aromatic rings in the mediator and will also vary with the heteroatom or functional groups in the molecule. A positive or negative charge on the mediator seems to give less stable electrodes than those prepared from neutral but these are otherwise similar in structure. The conducting salts of N-methylphenazinium (NMP+ ) and the tetracyanoquinodi- methane radical (TCNQ'-)67 68 forms a special group in which desorption of a surface molecule has little effect since an identical molecule is exposed instead.Electrodes made from pellets or deposition68 were both stable for practical applications and the potential for further improvement should be good. Reaction Rates Electron Transfer between Electrode and Immobilized Mediator The apparent electron-transfer rate [k (in s-l)] of a surface-immobilized redox com- pound can be evaluated in cyclic voltammetry from the variation of the peak potentials of the oxidation and reduction waves with the sweep 75 The peak separation of the oxidation and reduction waves will be zero if the rate is sufficiently fast compared with the sweep rate.Variation of peak potentials with sweep rate can also be caused by a number of other factor~,'~ e.g. uncompensated resistances local pH changes and movements of counter-ions. The sweep method will therefore only give the lowest value of the rate constant. Typical values were 6-17 s-l depending on pH for adsorbed pheno~azines,~~~ 66 3-6 s-l for adsorbed quin0nes~~9 56 and 1 s-l for oxidized carbon Electron transfer therefore generally seems to be fast compared with the chemical reaction (see below). When the mediator is adsorbed in multilayers or when it is incorporated into a polymeric backbone the charge propagation in the film may contribute to the rate-limiting kinetics in the catalytic cycle of a chemically modified 77-79 The combined effects of electron and counter-ion transport as well as conformational changes and solvent molecule movements in the film during an electrochemical cycle are collectively denoted charge transport.It was shown in the films of polymer-coated electrodes that the charge transport obeys Fick's laws of diffusion and hence could be defined by a diffusion constant. The partition coefficient of the substrate between the contacting solution and the polymer film the diffusion coefficient of the substrate in the polymer as well as the reaction rate between the substrate and the mediator in the film may also contribute to the rate-limiting kinetics. 5 6 9 1251 The rate constant of the chemical reaction between the NADH and the surface-attached mediator [k (in dm? mo1-1 s-l)] can be evaluated by cyclic ~oltammetry,~~t 70 but is more appropriately done when the electrode is mounted in a rotated-disc assembly.42 The situation is straight-forward at chemically modified electrodes bearing monolayers or less of immobilized mediator.The general equation for the catalytic current (icat ) is if r < monolayer and the stationary applied potential 2 120 mV more positive than E"' of the mediator (10) L. Gorton k r by the heterogeneous rate constant k (in cm s-l). Eqn (8) reverts to the Levich equation for an irreversible reaction when we replace Inversion of eqn (8) gives the Koutecky-Levich equation Eqn (9) shows that a plot of l/icat us. 1/c0l/~ extrapolating to 1 / d 2 - 0 gives the electrocatalytic rate from the intercept expressed as the product k r (in cm s-l).The coverage is evaluated from the integration of the area of the anodic or cathodic wave of a cyclic voltammogram. For a number of chemically modified electrodes k was evaluated; see table 1. For adsorbed phenoxazine 66 k was found to decrease in a distinct pattern with an increase in the bulk concentration of NADH. Similar experimental results were obtained for the NMP+TCNQ'- electrode.68 A charge-transfer complex between the phenoxazine mediators and NADH was postulated :657 66 k-1 charge-transfer (1 1) (13) 1 (14) l/icat = 0.620 nFADgkD H 'v - 'I6 CNADH o1l2 nFA( or 6 6 v 68 NADH + Mo k+l [NADH MI + k+2 NAD+ + Mred. complex Combination of the rate constants k+, k- and k+ yields (cf.Michaelis-Menten kinetics) = (k-l + k+2 )/k+l* Now the overall rate constant k, can be expressed as ki = k + 2 / ( K M -k CNADH 1. Inversion of eqn (12) gives k+2 llkl = K M / k + 2 + CNADH/k+2 and substitution of k, eqn (12) into eqn (9) gives 1 1 + ) r c N A H KM + CNADH If this assumption of a charge-transfer complex holds true plots of Ilk against CNADH as well as l/icat against I/CNADH would result in straight lines. Moreover the intercept of eqn (1 5) should be independent of CNADH and cu. Experimental results confim"5' 66 eqn (1 3)-( 15). The rate constants k+ and KM could be evaluated either by eqn (1 5) alone or by combining eqn (9) (12) and (1 3). The close agreement between the results obtained either way further supports the mechanism given in eqn 1252 Electrocatalytic Oxidation of Nicotinamide Coenzymes Table 2.Rate constants at pH 7.0 for the catalysed NADH oxidation ref. k+2/~-1 KM/mmol dm-3 mediator Meldola Blue (adsorbed) 65 66 68 1.1 0.21 3-la 30 0.22 13-1.5a pH 3 x 104 7.0 6.0 1,2-benzophenoxazine-7-0ne (adsorbed) NMP+TCNQ'- (radical salt) Depending on the method of deposition of the radical salt. Table 3. Variation of k,(CNA,H = 0) with pH for two adsorbed phenoxazine derivatives mediator ki(CNA, = O)/dm3 mol-l s-l Meldola Blue 1,2-benzophenoxazine-7- one 8.0 7.0 6.0 5.0 4.0 8 x lo4 0.9 x 103 1.1 x 103 1.6 x 103 2.9 x 103 5.9 x 103 Table 4. Variation of log k+ and log KM with pH for adsorbed 1,2-benzophenoxazine-7-0ne pH log (k+2/s-1) log (KM/dm3 mol-l s-l) -4.1 - 3.7 - 3.4 8 7 4 6 5 - 1.1 - 0.66 - 0.22 0.52 1.1 CNADH + 0 [cf.eqn. (1211- - 3.0 - 2.6 Slope of log k+ US. pH = -0.55; slope of log K M US. pH = -0.40. This equation tells us that for a constant CNADH (up to ca. KM) icat will vary linearly with r for small coverages and will be virtually independent of r for high coverages. Experiments confirm this? 73 For a comparison of the data the overall rate constant k was calculated for zero concentrations [eqn (1 3)] of NADH i.e. k + k,,/K when The kinetics of the reaction between adsorbed phenoxazine derivatives and NADH were found to be strongly influenced by the pH in the contacting phosphate buffers (see tables 3 and 4).At a constant pH however eqn (13x16) will still hold. The overall rate constant k increased with a decrease in pH (table 3). A plot of the logarithm of the rate constants k, and KIM against pH for adsorbed 1,2-benzophenoxazine-7-0ne results in straight lines with slopes of -0.55 and -0.40 respectively. and L. Gorton The redox reactions of the two phenoxazines are MB+ + 2e- + H+ t MBH (2e- 1 H+) BPO+2e-+2HS =S BPOH (2e- 2H+) (1) I where MB+ and MBH denote the oxidized and the reduced forms of 7-dimethylamino- 1,2-benzophenoxazinium (Meldola Blue) (1) and BPO and BPOH denote the oxidized The overall mediated reaction with NADH and MB+ 1253 (19) (20) (21) If the reactions were as simple as postulated in these equations the rate with MB+ should be independent of pH and log k+ should vary linearly with pH for BPO with a slope Tanaka and coworkers have in a series of recent publications,80-82 shown that the reaction between an NADH analogue with both one-8o and two-electron acceptors8Q-82 proceed via a charge-transfer intermediate and according to a multistep mechanism.The stability of the charge-transfer complex varied with the polarity of the solvent and the mechanism was greatly influenced by the presence and concentration of protons. With these results in mind a more complex reaction mechanism between NADH and the phenoxazine derivatives MB+ and BPO could explain the puzzling pH dependence of the rate constants.66 Unlike the paraquinones MB+ contains two nitrogen and BPO one nitrogen and one oxygen atom in para positions.These can however like the oxygens in paraquinones form quinoid resonance structures and may behave in a similar manner as the quinone. In an aprotic media the NADH analogue (AH) reacts with the quinone (Q)80-82 fast and reduced forms of 1,2-benzophenoxazine-7-0ne (2). NADH + MB+ k+l e [NADHMB+] 2 NAD+ + MBH k-1 does not show any proton influence whereas the reaction with BPO does NADH + BPO 2 [NADHBPO] [NADHBPO] + H+ - k-i k+2 NAD+ + BPOH,. of -1. AH +Q KCT f [AHQ] 2 [AH'+ Q'-] kH [AH'+ Q'-] + [A'QH' ] -+ fast A+ + QH- QH- +Q -- QH' +Q'- 2QH' -+ fast QH + Q 1254 Electroca t a b t ic Oxidat ion of Nicotinam ide Coenzymes I I I I I -400 -300 -200 -100 E",L,lmV vs.SCE Fig. 1. Variation of log k with E"' for some adsorbed phenoxazine derivatives in 0.25 mol dm-3 phosphate buffer pH 7.0. (a) Meldola Blue 7-dimethylamino-l,2-benzophenoxazinium (1); (b) 1,2-benzophenoxazine-7-0ne ; (c) Resorufin 3-hydroxyphenoxazine-7-one ; ( d ) Nile Blue 3-amino- 7-diethylamino- 1,2-benzophenoxazinium ; (e) gallocyanine 7-dimethylamino-4-hydroxy-3-oxo- 3H-phenoxazine- 1 carboxylic acid ; cf) Methyl Capri Blue 3,7-bis(dimethylamino)phenoxazinium; (26) whereas in the presence of a base (B) a reaction that will compete with reaction (23) will occur (g) = Ethyl Capri Blue 3,7-bis(diethylamino)phenoxazinium. in the presence of acid protons will influence reaction (24) Q'-+ H+ -+ QH' kl3 [AH'+Q'-]+B+A'+Q'-+BH+.(27) If now AH and Q are exchanged to NADH and BPO respectively we can see that in protic media like aqueous phosphate buffers eqn (27) may compete with eqn (23) so that a mixed reaction mechanism may be found thereby explaining the slope (ca. -0.5) of the plot of log k+ us. pH. The influence of p?I is found both in eqns (26) and (27) where B in eqn (27) may correspond to hydroxide ions or the base character of H,O. In two recent investigations of the homogeneous reaction between NADH and ortho- and para-quinones in aqueous ethanola3 (also one-electron acceptors of quinoid struc- ture) and in aqueous ~ ~ l ~ t i o n ~ ~ ~ it was found contrary to the results mentioned above that k for the quinones (two-electon acceptors) did not vary with pH but rather with the percentage of ethanoLa3 Logk was found in the aqueous ethanol buffers to vary linearly with Eo' of the quinone-hydroquinone redox couple,a3 while in the aqueous buffers a linear correlation was found with EP;- of the quinonedeprotonated hydroquinone couple (i.e.Q+H++2e- E& + QH-).84 With one of the one-electron acceptors investigated (TCNQ'-) however a slope of - 0.5 for the plot of log k us. pH was A linear dependence of the plot of log k us. pH was also found for a 2,6-dichlorophenol-indophenol mixture with NADH analogues in phosphate buffer- methanol ~ ~ l u t i o n ~ . ~ ~ L. Gorton 1255 R 88 Scheme 2. Proposed dependence of reaction rate with NADH on structure of ??? R R Scheme 3. Questioned dependence of reaction rate with NADH on structure for 3-dialkylamino- (left) and 3,7-bis(dialkylamino)-(right) substituted phenoxazine derivatives.At this laboratory the values for k at pH 7.0 for a series of adsorbed phenoxazine derivatives were evaluated according to eqn (1 I)-( 15). The values of k were found to vary with EO’ of the mediator in such a manner that a plot of log k against E”’ resulted in two straight lines (see fig. l) with slopes of 1.2 x and 0.9 x mV-l. These lines will cross very close to -560 mV us. SCE which is Eo’ of the NAD+/NADH couple. When calculating k from its logarithm at this potential the result was an effective rate constant of zero. Thus there seems to be an indication that both E”’ and the structural elements of the phenoxazine derivatives are important for the overall rate constant with NADH.As indicated in ref. (83) and (84) there exist linear correlations between E”’ (or Eg-) and logk of some classes of electron acceptors with NADH orthoq~inones,~~ para- quinones and semiquinones of phenothiazinesEe and some other compounds. Contrary to this no such correlation was found when a series of phenylenediamines and aminophenols was i n v e ~ t i g a t e d . ~ ~ ~ ~ ~ The value of k depended on the structure of the mediator rather than on E”’ indicating that charged imines were superior to neutral imines which in turn were superior to a keto group (see scheme 2). The series of phenoxazine derivatives investigated so far (fig. 1) is far too small to prescribe a strategy for the optimal configuration of the mediator with respect to the rate constant k,.Other commercially available phenoxazine derivatives have yet to be examined. Synthetic organic work where crucial changes are made to existing compounds will also certainly give more evidence of the detailed mechanism and will point the way to the best mediator. So far adsorbed Meldola Blue (E,”& = - 175 mV us. SCE) seems to be the fastest mediator (see tables 1 and 3); 93 and 100% current efficiencies were obtained at pH 7.0 and 6.0 respectively for 0.1 mmol dm-3 NADH The phenoxazine derivative 3-B-naphthoyl-Nile Bluee4 (E,”; = - 220 mV us. SCE) gave some 70% current efficiency for a 1 mmol dm-3 NADH solution at pH 7.0. Even if k for this mediator was not evaluated it seems as if it would fit the upper line in fig.1. Thus some guidelines can be drawn. A phenoxazine derivative based chemically modified electrode with E,”& x - 200 to - 150 mV vs. SCE should give sufficient current efficiency to be used as an NADH sensor. Information is still too scarce as to whether a charged imino group in position 7 should be preferred over a neutral one or an oxygen atom [cf. (l) (2) and scheme 21 and also whether the phenoxazine should be mono- or di-substituted in positions 3 and 7 (see scheme 3). The ingenious work by Hucke4 to add two more rings in position 3 of Nile Blue to FAR 1 42 1256 Elec trocataly t ic Oxidation of Nico t inamide Coenzymes produce 3-a-naphthoyl-Nile Blue not only increased the stability of the adsorption but Ei& was raised from - 430 to - 220 mV us.SCE. By reacting Nile Blue with 1 -pyrene aldehyde,89 four conjugated rings can be added forming 7-diethylamino-3-( 1 -pyrenyl)- imino- 1,2-benzophenoxazinium (3). This mediator should have even more stable adsorption and as E,"h = - 160 mV vs. SCE the rate constant k should be more favourable than that of 3-P-naphthoyl-Nile Blue. The phenoxazines are subjected to decomposition in alkaline ~ o l u t i o n . ~ ~ ~ 91 This is a crucial drawback for these mediators as many dehydrogenase-catalysed reactions have a more favoured substrate (SH,)-product (S) equilibrium in the alkaline region. The sensitivity to hydroxide ions seems however to differ with the substituent~,~~~ 65 and this problem might be solved by molecular engineering.Other techniques for preparing chemically modified electrodes such as covalent binding of or polymer contaning phenoxazines have not been tested yet but should have a favourable future. The conducting salt of NMP+TCNQ'- seems to operate well as a NADH sensor. The mechanism of charge transport from NADH to the electrode body seems still to be questioned. Both NMP+ and TCNQ'- alone catalyse the oxidation of NADH as well as many other organic substances that are normally subjected to high overvoltages as does the conducting salt. Some experiments in this department on making conductive salts of MB+ and TCNQ'-92 seemed promising for selective NADH oxidation but unfortunately successful syntheses of the conductive salt could not be repeated.I thank Prof. G. Johansson for valuable discussions concerning this paper. Financial support from the Swedish Natural Research Council and the Swedish Board for Technical Development is gratefully acknowledged. References 1 C. R. Lowe Trends Biotechnol. 1984 2 59. 2 H. Y. Neujahr Biotechnol. Gen. Eng. Rev. 1984 1 7. 3 B. N. Zaba and S. Bone Lab. Pract. 1985,34 1 1. 4 F. Scheller G. Stmad B. Neumann M. Kiihn and W. Ostrowski J. Electroanal. Chem. 1979,104,117. 5 H. A. 0. Hill N. J. Walton and I. J. Higgins FEBS Lett. 1981 126 282. 6 M. J. Eddowes and H. A. 0. Hill J. Am. Chem. SOC. 1979 101,4461. 7 R. M. Ianniello T. J. Lindsay and A. M. Yacynych Anal. Cnem. 1982,54 1098. 8 H. Durliat and M. Comtat Anal. Chem. 1982 54 856. 9 C.M. Visser Origins Life 1982 12 165. 10 E. A. Yastrebova I. V. Osipov S. D. Varfolomeevand P. K. Agasyan Zh. Anal. Khim. 1982,37 1278. 11 J. N. Bumett and A. L. Underwood Biochemistry 1965,4 2060. 12 W. J. Blaedel and R. A. Jenkins Anal. Chem. 1975,47 1337. 13 R. W. Coughlin M. Aizawa B. F. Alexander and M. Charles Biotechnol. Bioeng. 1975 17 515. 14 H. Jaegfeldt A. Torstensson and G. Johansson Anal. Chim. Acta 1978,97 221. 15 J. Moiroux and P. J. Elving J. Electrounal. Chem. 1979 102 93. 16 J. Moiroux and P. J. Elving Anal. Chem. 1979,51 346. 17 J. Moiroux and P. J. Elving J. Am. Chem. Soc. 1980,102,6533. 1257 (pub. 821028). L. Gorton 18 W. T. Bresnahan J. Moiroux Z. Samec and P. J. Elving Bioelectrochem. Bioenerg. 1980 7 125. 19 W. J. Blaedel and R.G. Haas Anal. Chem. 1979,42,918. 20 P. Leduc and D. Thevenot Bioelectrochem. Bioenerg. 1974 1 96. 21 R. D. Braun K. S. V. Santhanam and P. J. Elving J. Am. Chem. Soc. 1975,97 2591. 22 H. Jaegfeldt J. Electroanal. Chem. 1980 110 295. 23 Z. Samec and P. J. Elving J. Electroanal. Chem. 1983 144 217. 24 R. L. Blankespoor and L. L. Miller J. Electroanal. Chem. 1984 171 231. 25 M. A. Jensen and P. J. Elving Bioelectrochem. Bioenerg. 1978 5 526. 26 G. G. Guilbault and T. Cserfalvi Anal. Lett. 1976 9 277. 27 A. Malinauskas and J. Kulys Anal. Chim. Acta 1978 98 3 1. 28 T. Yao and S. Musha Anal. Chim. Acta 1979 110 203. 29 W. J. Blaedel and R. C. Engstrom Anal. Chem. 1980 52 1691. 30 H. M. Eggers H. B. Halsall and W. R. Heineman Clin. Chim. 1982 28 1848.31 J. M. Laval and C. Bourdillon J. Electroanal. Chem. 1983 152 125. 32 J. M. Laval C. Bourdillon and J. Moiroux J. Am. Chem. Soc. 1984 106,4701. 33 R. Szentnmay P. Yeh and T. Kuwana in Electrochemical Studies of Biological Systems ed. D. Sawyer (ACS Washington D.C. 1977) p. 143. 34 M. L. Fultz and R. A. Durst Anal. Chim. Acta 1982 140 1. 35 J. H. Ottaway Biochem. J. 1966 99 253. 36 M. D. Smith and C. L. Olsson Anal. Chem. 1975,47 1074. 37 H. Winartasaputra S. S . Kuan and G. G. Guilbault Anal. Chem. 1982 54 1987. 38 M. D. Smith and C. L. Olsson Anal. Chem. 1974 46 1544. 39 S. T. Sulaiman and M. M. N. M. Saleem Fresenius’ 2. Anal. Chem. 1984 317 751. 40 S. T. Sulaiman and M. M. Najeeb Michrochem. J. 1985 31 37. 41 K. Matsumoto H. Ukeda and Y. Osajima Agric.Biol. Chem. 1984,48 1879. 42 R. W. Murray in Electroanalytical Chemistry ed. A. J. Bard (Dekker New York 1984) vol. 13 p. 191. 43 J. Zak and T. Kuwana J. Electroanal. Chem. 1983 150 645. 44 J. F. Evans T. Kuwana M. T. Henne and G. P. Royer J. Electroanal. Chem. 1977,80,409. 45 N. Cenas J. Rozgaite A. Pocius and J. Kulys J. Electroanal. Chem. 1983 154 121. 46 L. Falat and H-Y. Cheng J. Electroanal. Chem. 1983 157 393. 47 K. Ravichandran and R. P. Baldwin Anal. Chem. 1984 56 1744. 48 J. Schreurs J. van den Berg A. Wonders and E. Barendrecht Recl. Trav. Chim. Pays-Bas 1984 103 251. 49 N. K. Cenas J. J. Kanapieniene and J. J. Kulys J. Electroanal. Chem. 1985 189 163. 50 D. C-S. Tse and T. Kuwana Anal. Chem. 1978 50 1315. 51 C. Ueda D. C-S.Tse and T. Kuwana Anal. Chem. 1982,54 850. 52 C. Degrand and L. L. Miller J. Am. Chem. SOC. 1980 102 5728. 53 M. Fukui A. Kitani C. Degrand and L. L. Miller J. Am. Chem. Soc. 1982 104 28. 54 N. K. Lau and L. L. Miller J. Am. Chem. Soc. 1983 105 5271. 55 H. Jaegfeldt A. B. C. Torstensson L. G. 0. Gorton and G. Johansson Anal. Chem. 1981,53 1979. 56 H. Jaegfeldt T. Kuwana and G. Johansson J. Am. Chem. Soc. 1983 105 1805. 57 K. Ravichandran and R. P. Baldwin J. Electroanal. Chem. 1981 126 293. 58 A. Torstensson and L. Gorton J. Electroanal. Chem. 1981 130 199. 59 H. Huck and H-L. Schmidt Angew. Chem. 1981 93,421. 60 N. Cenas J. Rozgaite and J. Kulys Liet. TSR Mokslu Akad. Darb. Ser. B 1983 3 13. 61 G. Arai M. Matsushita and 1. Yasumori Nippon Kagaku Kaishi 1985 5 894.62 K. Ravichandran and R. P. Baldwin Anal. Chem. 1983 55 1586. 63 L. Gorton H. Jaegfeldt A. Torstensson and G. Johansson Patent No. 82/SE 59 Al(8203729) 64 H. Huck Fresenius’ 2. Anal. Chem. 1982 313 548. 65 L. Gorton A. Torstensson H. Jaegfeldt and G. Johansson J. Electroanal. Chem. 1984 161 103. 66 L. Gorton G. Johansson and A. Torstensson J. Electroanal. Chem. 1985 1% 81. 67 J. J. Kulys Enzyme Microbiol. Technol. 1981 3 344. 68 W. J. Albery and P. N. Bartlett J. Chem. SOC. Chem. Commun. 1984 234. 69 A. P. F. Turner Biotech’85 (Europe) (Online Publications Plenar 1985) p. 181. 70 C. P. Andrieux and J. M. Savkant J. Electroanal. Chem. 1978,93 163. 71 N. Oyama N. Oki H. Ohno Y. Ohnuki H. Matsuda and E. Tsuchida J. Phys. Chem. 1983,87,3642.72 H. Huck A. Schelter-Graf and H-L. Schmidt Bioelectrochem. Bioenerg. 1984 13 199. 73 R. Appelqvist G. Marko-Varga L. Gorton A. Torstensson and G. Johansson Anal. Chim. Acta 1985 169 237. 74 E. Laviron J. Electroanal. Chem. 1979 101 19. 75 M. Sharp M. Petterson and K. Edstrom J. Electroanal. Chem. 1979 95 123. 76 E. Laviron in Electroanalytical Chemistry ed. A. J. Bard (Dekker New York 1982) vol. 12 p. 53. 77 W. J. Albery and A. R. Hillman J. Electroanal. Chem. 1984 170 27. 42-2 1258 78 C. P. Andrieux J. M. Dumas-Bouchiat and J. M. Saveant J. Electroanal. Chem. 1984 169 9. 79 C. P. Andrieux and J. M. Saveant J. Electroanal. Chem. 1984 171,65. 80 S. Fukuzumi Y. Kondo and T. Tanaka J. Chem. SOC. Perkin Trans. 2 1984 673. 81 S. Fukuzumi N. Nishizawa and T. Tanaka J. Org. Chem. 1984,49 3571. 82 S. Fukuzumi N. Nishizawa and T. Tanaka J. Chem. SOC. Perkin Trans. 2 1985 371. 83 N. K. h a s J. J. Kanapieniene and J. J. Kulys Biochim. Biophys. Acta 1984 767 108. 84 B. W. Carlson and L. L. Miller J. Am. Chem. Sac. 1985,107,479. 85 K. Wallenfels and M. Gellrich Ann. Chem. (Leipzig) 1959 621 149. 86 J. Grodkowski F. Neta B. W. Carlson and L. Miller J. Phys. Chem. 1983 87 3135. 87 A. Kitani and L. L. Miller J. Am. Chem. SOC. 1981 103 3595. 88 L. Gorton unpublished results. 89 A. Kitani Y-H. So and L. L. Miller J. Am. Chem. SOC. 1981 103,7636. 90 M. KotouEek J. Tomagova and S . DurEGkova Collect. Czech. Chem. Commun. 1969 34 212. 91 M. KotouEek and J. Zavadilova Collect. Czech. Chem. Commun. 1972,37 3212. 92 L. Gorton and T. Svensson unpublished results 1985. Electroca taly t ic Oxidat ion of Nico t inamide Coenzymes Paper 5/1897; Received 21st October 1985

ISSN:0300-9599    

DOI:10.1039/F19868201245

出版商:RSC    

年代:1986

数据来源: RSC