Electrochemical Biosensors: Recommended Definitions and Classification Daniel Thevenot, Klara Toth, Richard Durst, George Wilson

Home , Bioelectronics

Electrochemical biosensors: recommended definitions and classification Daniel Thevenot, Klara Toth, Richard Durst, George Wilson

To cite this version:

Daniel Thevenot, Klara Toth, Richard Durst, George Wilson. Electrochemical biosensors: recommended definitions and classification. Biosensors and Bioelectronics, Elsevier, 2001, 16, pp.121 -131. 10.1081/AL-100103209. hal-01084678

HAL Id: hal-01084678 https://hal-enpc.archives-ouvertes.fr/hal-01084678 Submitted on 19 Nov 2014

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés.

Distributed under a Creative Commons Attribution| 4.0 International License Biosensors & Bioelectronics 16 (2001) 121–131 www.elsevier.com/locate/bios

Technical report Electrochemical biosensors: recommended deﬁnitions and classiﬁcation

Daniel R. The´venot a,*,1, Klara Toth b, Richard A. Durst c, George S. Wilson d

a Centre d’Enseignement et de Recherche sur l’Eau, la Ville et l’En6ironnement (Cere6e), Faculte´ de Sciences et de Technologie, Uni6ersite´ Paris XII-Val de Marne, 61 A6enue du Ge´ne´ral de Gaulle, 94010 Cre´teil Cedex, Paris, France b Institute of General and Analytical Chemistry, Technical Uni6ersity of Budapest, Gellert ter 4, H1111, Budapest, Hungary c Department of Food Science and Technology, Cornell Uni6ersity, Gene6a, NY 14456-0462, USA d Chemistry Department, Kansas Uni6ersity, Lawrence, KS 66045, USA

Abstract

Two Divisions of the International Union of Pure and Applied Chemistry (IUPAC), namely Physical Chemistry (Commission I.7 on Biophysical Chemistry formerly Steering Committee on Biophysical Chemistry) and Analytical Chemistry (Commission V.5 on Electroanalytical Chemistry) have prepared recommendations on the definition, classification and nomenclature related to electrochemical biosensors; these recommendations could, in the future, be extended to other types of biosensors. An electrochemical biosensor is a self-contained integrated device, which is capable of providing specific quantitative or semi-quantitative analytical information using a biological recognition element (biochemical receptor) which is retained in direct spatial contact with an electrochemical transduction element. Because of their ability to be repeatedly calibrated, we recommend that a biosensor should be clearly distinguished from a bioanalytical system, which requires additional processing steps, such as reagent addition. A device that is both disposable after one measurement, i.e. single use, and unable to monitor the analyte concentration continuously or after rapid and reproducible regeneration, should be designated a single use biosensor. Biosensors may be classified according to the biological specificity-conferring mechanism or, alternatively, to the mode of physico-chemical signal transduction. The biological recognition element may be based on a chemical reaction catalysed by, or on an equilibrium reaction with macromolecules that have been isolated, engineered or present in their original biological environment. In the latter cases, equilibrium is generally reached and there is no further, if any, net consumption of analyte(s) by the immobilized biocomplexing agent incorporated into the sensor. Biosensors may be further classified according to the analytes or reactions that they monitor: direct monitoring of analyte concentration or of reactions producing or consuming such analytes; alternatively, an indirect monitoring of inhibitor or activator of the biological recognition element (biochemical receptor) may be achieved. A rapid proliferation of biosensors and their diversity has led to a lack of rigour in defining their performance criteria. Although each biosensor can only truly be evaluated for a particular application, it is still useful to examine how standard protocols for performance criteria may be defined in accordance with standard IUPAC protocols or definitions. These criteria are recommended for authors, referees and educators and include calibration characteristics (sensitivity, operational and linear concentration range, detection and quantitative determination limits), selectivity, steady-state and transient response times, sample throughput, reproducibility, stability and lifetime. © 2001 Elsevier Science B.V. All rights reserved.

Keywords: Electrochemical biosensor; deﬁnition; single use biosensor; calibration characteristics; performance criteria; analytes

Abbre6iations: Ab, antibody; Ag, antigen; BLM, bilayer lipid membrane; BSA, bovine serum albumin; CME, chemically modified electrode; ENFET, enzyme field-effect transistor; FET, field-effect transistor; FIA, flow injection analysis; HPLC, high performance liquid chromatography; IMFET, immunological field-effect transistor; ISE, ion selective electrode; ISFET, ion sensitive field-effect transistor; LP, lactose permease; tL, life time; LOD, limit of detection; LOQ, limit of quantification; NAD, nicotinamide adenine dinucleotide; PU, polyurethane; PVAL, poly(vinyl alcohol); SAM, self assembled monolayer; SPR, surface plasmon resonance; TCNQ-, tetracyanoquinodimethane; TTF+, tetrathiafulvalene. Membership of the working party for the present project during the period 1993–1999 was as follows: D.R. The´venot, R.P. Buck, K. Cammann, R.A. Durst, K. Toth and G.S. Wilson * Corresponding author. Tel.: +33-145-171625; fax: +33-145-171627. E-mail address: [email protected] (D.R. The´venot). 1 International Union of Pure and Applied Chemistry: Physical Chemistry Division, Commission I.7 (Biophysical Chemistry); Analytical Chemistry Division, Commission V.5 (Electroanalytical Chemistry).

1. Definition and limitations While all biosensors are more or less selective (non-specific) for a particular analyte, some are, by design and 1.1. Biosensor construction, only class-specific, since they use class enzymes, e.g. phenolic compound biosensors, or whole A chemical sensor is a device that transforms chemi- cells, e.g. used to measure biological oxygen demand. cal information, ranging from the concentration of a Because in sensing systems present in living organisms/ specific sample component to total composition analy- systems, such as olfaction, and taste, as well as neuro- sis, into an analytically useful signal. Chemical sensors transmission pathways, the actual recognition is contain usually two basic components connected in performed by cell receptor, the word receptor or biore- series: a chemical (molecular) recognition system (receptor is also often used for the recognition system of a ceptor) and a physico-chemical transducer. Biosensors chemical biosensor. Examples of single and multiple are chemical sensors in which the recognition system signal transfer are listed in Table 1. These examples are utilises a biochemical mechanism Cammann, 1977; limited to the most common sensor principles, exclud- Turner et al., 1987. ing existing laboratory instrumentation systems. The biological recognition system translates informa- The transducer part of the sensor serves to transfer tion from the biochemical domain, usually an analyte the signal from the output domain of the recognition concentration, into a chemical or physical output signal system, mostly to the electrical domain. Because of the with a defined sensitivity. The main purpose of the general significance of the word, a transducer provides recognition system is to provide the sensor with a high bi-directional signal transfer (non-electrical to electrical degree of selectivity for the analyte to be measured. and vice versa); the transducer part of a sensor is also

Table 1 Types of receptors used in biosensors and the electrochemical measurement techniques, linked to them that recognize speciﬁc speciesa.

Analytes Receptor/Chemical recognition Measurement technique/ Transduction mode system

1. Ions mixed valence metal oxides potentiometric, voltammetric permselective, ion-conductive inorganic crystals trapped mobile synthetic or biological ionophores ion exchange glasses enzyme(s) 2. Dissol6ed gases, 6apours, odours bilayer lipid or hydrophobic in series with 1. membrane inert metal electrode amperometric enzyme(s) amperometric or potentiometric antibody, receptor amperometric, potentiometric or impedance, piezoelectric, optical 3. Substrates enzyme(s) amperometric or potentiometric in series with 1. or 2. or metal or carbon electrode, conductometric, piezoelectric, optical, calorimetric whole cells as above membrane receptors as above plant or animal tissue as above 4. Antibody/antigen antigen/antibody oligonucleotide amperometric, potentiometric or impedimetric, duplex, aptamer piezoelectric, optical, surface plasmon resonance enzyme labelled in series with 3. chemiluminescent or fluorescent optical labelled 5. Various proteins and low molecular weight specific ligands as 4. substrates, ions protein receptors and channels enzyme labelled fluorescent labelled

a Biological receptors, which are part of electrochemical biosensors, are indicated in bold characters Bergveld and The´venot, 1993. Besides quantiﬁcation of the above mentioned analytes, biosensors are also used for detection and quantiﬁcation of micro-organisms: receptors are bacteria, yeast or oligonucleotide probes coupled to electrochemical, piezoelectric, optical or calorimetric transducers. D.R. The´6enot et al. / Biosensors & Bioelectronics 16 (2001) 121–131 123

Table 2 Type of electrochemical transducers for classiﬁed type of measurements, with corresponding analytes to be measured Bergveld and The´venot, 1993.a

Measurement type Transducer Transducer analyte

1. Potentiometric ion-selective electrode (ISE) K+,Cl−,Ca2+,F− glass electrode H+,Na+...

gas electrode CO2,NH3 metal electrode redox species

2. Amperometric metal or carbon electrode O2, sugars, alcohols... chemically modified electrodes (CME) sugars, alcohols, phenols, oligonucleotides... 3. Conductometric, interdigitated electrodes, metal electrode urea, charged species, oligonucleotides... impedimetric 4. Ion charge or field ion-sensitive field effect transistor (ISFET), enzyme FET (ENFET) H+,K+... effect

a Non electrochemical transducers are also used within biosensors: (a) piezoelectric (shear and surface acoustic wave); (b) calorimetric (thermistor); (c) optical (planar wave guide, fibre optic, surface plasmon resonance...) called a detector, sensor or electrode, but the term Although biosensors with different transducer types, transducer is preferred to avoid confusion. Examples of e.g. electrochemical, optical, piezoelectric or thermal electrochemical transducers, which are often used for types, show common features, this report is restricted to the listed types of measurement in Table 1, are given in electrochemical biosensors (indicated in bold characters Table 2, together with examples of analytes which have in Table 1). Optical, mass and thermal sensors will be been measured. Transducers are classified by recogni- described in future IUPAC reports. For example opti- tion element type (Table 1) or by electrochemical trans- cal biosensors will be described by IUPAC commission ducer mode (Table 2). V.4 in Spectrochemical and other optical procedures for Finally, chemical sensors, as well as biosensors de- analysis (project number 540 / 19 / 95). scribed below, are self-contained, all parts being pack- aged together in the same unit, usually small, the 1.3. Limitations in the use of the term ‘biosensor’ biological recognition element being in direct spatial contact with the transducing element. Since a biosensor is a self contained integrated device, we recommend that it should be clearly distin- 1.2. Electrochemical biosensor guished from an analytical system which incorporates additional separation steps such as high performance An electrochemical biosensor is a biosensor with an liquid chromatography (HPLC), or additional hard- electrochemical transducer (Table 2). It is considered a ware and/or sample processing such as specific reagent chemically modified electrode (CME) Durst et al., 1997; introduction, as flow injection analysis (FIA). Thus, a Kutner et al., 1998 since electronic conducting, semi- biosensor should be a reagentless analytical device, conducting or ionic conducting material is coated with although the presence of ambient co-substrates, such as a biochemical film. water for hydrolases or oxygen for oxidoreductases, A biosensor is an integrated receptor-transducer may be sufficient for the analyte determination. On the device, which is capable of providing selective quantita- other hand, it may provide, as part of an integrated tive or semi-quantitative analytical information using a system, some separation or amplification steps achieved biological recognition element. Thus, biological exam- by inner or outer membranes or reacting layers. In ples given in Table 1 are shown in bold characters. conclusion, an HPLC or FIA system may incorporate a A biosensor can be used to monitor either biological biosensor as a detecting device, and FIA is often conve- or non-biological matrixes. Chemical sensors, which nient to evaluate the biosensor analytical performance incorporate a non-biological specificity-conferring part (see Section 5.) On the contrary, an FIA system con- or receptor, although used for monitoring biological taining a reagent reservoir, an enzymatic or immuno- processes, as the in vivo pH or oxygen sensors, are not logical reactor and, downstream, an electrochemical biosensors. These sensors are beyond the scope of the sensor, is not a biosensor. present report. Similarly, physical sensors used in bio- Because of the importance of their ability to be logical environment, even when electrically based, such repeatedly calibrated, we recommend that the term as in vivo pressure or blood flow sensors, are also multiple-use biosensor be limited to devices suitable for excluded from this report. monitoring both the increase and decrease of the ana- 124 D.R. The´6enot et al. / Biosensors & Bioelectronics 16 (2001) 121–131 lyte concentrations in batch reactors or flow-through viously or have been manufactured. Thus, a continuous cells. Thus, single-use devices that cannot rapidly and consumption of substrate(s) is achieved by the immobi- reproducibly be regenerated should be named single-use lized biocatalyst incorporated into the sensor: transient biosensors. Various terms have been used for such or steady-state responses are monitored by the inte- disposable and non-regenerable devices, e.g. bioprobes, grated detector. Three types of biocatalyst are com- bioindicators. At present, none of these names have monly used: been generally accepted by the scientific community and 1. Enzyme (mono- or multi-enzyme), the most com- we recommend designating them as single-use mon and well developed recognition system, biosensors. 2. Whole cells (micro-organisms, such as bacteria, Finally, as is seen in the various sections of this fungi, eukaryotic cells or yeast) or cell organelles or report, the diversity of the molecular recognition sys- particles (mitochondria, cell walls), tems and of the electrochemical transducers incorpo- 3. Tissue (plant or animal tissue slice). rated in each biosensor appears to be very wide. The biocatalytic-based biosensors are the best known Nevertheless, common features, related to their operat- and studied and have been the most frequently applied ing principles, are significant. They mainly depend upon to biological matrices since the pioneering work of the type of transducer and molecular receptor used: Clark Clark et al., 1962. One or more analytes, usually because of the nature of their operational principle, named substrates S and S%, react in the presence of amperometric sensors, including biocatalytic amper- enzyme(s), whole cells or tissue culture and yield one or ometric sensors, alter the concentration of the ana- several products, P and P%, according to the general lyte in their vicinity; these sensors may reach a reaction scheme:

steady-state but they never reach equilibrium. biocatalyst Knowledge of the rate-limiting step of their re- S+S% P+P% sponse, i.e. mass transport rate versus analyte consumption reaction rate, is very important for There are four strategies that use adjacent transduc- understanding their operational characteristics; ers for monitoring the analyte S consumption by this potentiometric as well as biocomplexing based sen- biocatalysed reaction: sors usually operate at or near equilibrium and are detection of the co-substrate S’ consumption, e.g., not subject to such transport limitations; on the oxygen depleted by oxidase, bacteria or yeast react- other hand, the magnitude of their apparent equi- ing layers, and the corresponding signal decrease librium constant and kinetics, under experimental from its initial value; conditions, will define the continuity of the sensor recycling of P, one of the reaction products, e.g., + response and the necessity for reagent introduction. hydrogen peroxide, H ,CO2,NH3, etc. production If these sensors operate without requiring reagent by oxidoreductase, hydrolase, lyase, etc., and corre- addition and are capable of rapid and reproducible sponding signal increase; regeneration, then they are referred to as multiple- detection of the state of the biocatalyst redox active use biosensors. centre, cofactor, prosthetic group evolution in the presence of substrate S, using an immobilized mediator which reacts sufficiently rapidly with the biocata- 2. Classification lyst and is easily detected by the transducer; various ferrocene derivatives as well as tetrathiafulvalene-te- Biosensors may be classified according to the biolog- tracyanoquinodimethane (TTF+ TCNQ−) organic ical specificity conferring mechanism, or to the mode of salt, quinones, quinoid dyes, Ru or Os complexes in signal transduction or, alternatively, a combination of a polymer matrix, have been used Bartlett et al., the two. These might also be described as amperomet- 1991; ric, potentiometric, field-effect or conductivity sensors. direct electron transfer between the active site of a Alternatively, they could be termed, for example, as redox enzyme and the electrochemical transducer. amperometric enzyme sensors Inczedy et al., 1998. As The third strategy attempts to eliminate sensor re- an example, the former biosensors may be considered sponse dependence on the co-substrate, S%, concentra- as enzyme- or immuno-sensors. tion and to decrease the influence of possible interfering species. The first goal is only reached when reaction 2.1. Receptor: biological recognition element rates are much higher for immobilized mediator with biocatalyst than those for co-substrate with biocatalyst. 2.1.1. Biocatalytic recognition element An alternative approach to the use of such mediators In this case, the biosensor is based on a reaction consists in restricting the analyte (substrate) concentra- catalysed by macromolecules, which are present in their tion within the reaction layer through an appropriate original biological environment, have been isolated pre- outer membrane, whose permeability strongly favours D.R. The´6enot et al. / Biosensors & Bioelectronics 16 (2001) 121–131 125 co-substrate transport Scheller and Pfeiffer, 1978; sensitivity of immuno-sensors, enzyme labels are Bindra et al., 1991. frequently coupled to Ab or Ag, thus requiring When several enzymes are immobilized within the additional chemical synthesis steps. Even in the case same reaction layer, several strategies for improving of the enzyme-labelled Ab, these biosensors will biosensor performance can be developed. Three follow- essentially operate at equilibrium, the enzymatic ing possibilities have been most frequently proposed: activity being there only to quantify the amount of several enzymes facilitate the biological recognition complex produced. As the binding or affinity con- by sequentially converting the product of a series of stant is usually very large, such systems are either enzymatic reactions into a final electroactive form: irreversible (single-use biosensors) or placed within this set-up allows a much wider range of possible an FIA environment where Ab may be regenerated biosensor analytes Wollenberger et al., 1993; by dissociation of complexes by chaotropic agents, multiple enzymes, applied in series, may regenerate such as glycine-HCl buffer at pH 2.5. the first enzyme co-substrate and a real amplification 2. Receptor/antagonist/agonist. More recently, at- of the biosensor output signal may be achieved by tempts have been made to use ion channels, mem- efficient regeneration of another co-substrate of the brane receptors or binding proteins as molecular first enzyme; recognition systems in conductometric, ISFET or multiple enzymes, applied in parallel, may improve optical sensors Sugawara et al., 1997a. For example, the biosensor selectivity by decreasing the local con- the transport, protein lactose permease (LP), may be centration of electrochemical interfering substance: incorporated into liposome bilayers thus allowing this set-up is an alternative to the use of either a coupling of sugar proton transport with a stoichio- permselective membrane (see Section 4.2) or a differ- metric ratio of 1:1, as demonstrated with the fluores- ential set-up, i.e., subtraction of the output signal cent pH-probe pyranine entrapped in these generated by the biosensor and by a reference sensor liposomes Kiefer et al., 1991. These LP-containing having no biological recognition element The´venot et liposomes have been incorporated within planar al., 1979. lipid bilayer coatings of an ISFET gate sensitive to A recent development of enzyme based biosensors pH. Preliminary results have shown that these involves their operation in an organic solvent matrix: a modified ISFETs enable rapid and reversible detec- hydrophilic microenvironment is often maintained tion of lactose in an FIA system. Protein receptor within the enzyme and the substrate partitions between based biosensors have been recently developed Sug- the matrix and the enzyme active site. awara et al., 1997b. The result of the binding of the analyte, here named agonist, to immobilized channel 2.1.2. Biocomplexing or bioaffinity recognition element receptor proteins, is monitored by changes in ion The biosensor operation is based on interaction of fluxes through these channels. For example gluta- the analyte with macromolecules or organized molecu- mate, as target agonist, may be determined in the lar assemblies that have either been isolated from their presence of various interfering agonists, by detecting original biological environment or engineered Aizawa, Na+ or Ca2+ fluxes, using conductivity or ion 1991. Thus, equilibrium is usually reached and there is selective electrodes. Due to the dependence of ion no further net consumption of the analyte by the channel switching on agonist binding, there is usu- immobilized biocomplexing agent. These equilibrium ally no need for enzyme labelling of the receptor to responses are monitored by the integrated detector. In achieve the desired sensitivity. some cases, this biocomplexing reaction is itself moni- A developing field in electrochemical biosensors is tored using a complementary biocatalytic reaction. the use of chips and electrochemical methods to detect Steady-state or transient signals are then monitored by binding of oligonucleotides (gene probes) (Table 1). the integrated detector. There are two approaches currently developed. The fist 1. Antibody-antigen interaction. The most developed one intercalates into the oligonucleotide duplex, during examples of biosensors using biocomplexing recep- the formation of a double stranded DNA on the probe tors are based on immunochemical reactions, i.e. surface, a molecule that is electroactive. The second binding of an antigen (Ag) to a specific antibody approach directly detects guanine that is electroactive. (Ab). Formation of such Ab-Ag complexes has to In conclusion, biocomplex-based biosensors although be detected under conditions where non-specific in- showing promising behaviour, have not yet reached the teractions are minimized. Each Ag determination advanced development stage of the biocatalyst-based requires the production of a particular Ab, its isola- systems. Being based on equilibrium reactions, they tion and, usually, its purification. Several studies generally present a very narrow linear operating range have been described involving direct monitoring of of concentration and are often unable to monitor con- the Ab-Ag complex formation on ion-sensitive-field- tinuously the analyte concentration. Furthermore, some effect transistors (ISFETs). In order to increase the of these biosensors may be difficult to operate in a 126 D.R. The´6enot et al. / Biosensors & Bioelectronics 16 (2001) 121–131 biological matrix because their sensing layer has to be steady-state or transient, but it is never an equilibrium in direct contact with the sample and because it may response. The situation is more complex for enzyme-la- not be possible to incorporate an outer membrane to belled immuno-sensors: although the Ab-Ag complex is separate the sensing element from the sample matrix. expected to reach an equilibrium and reactions to be either reversible or irreversible, the labelled enzyme 2.2. Detection or measurement mode: electrochemical activity is measured under steady-state analyte con- transduction or detection sumption conditions. Another important feature of the ISE based biosen- 2.2.1. Amperometry sors, such as pH electrodes, is the large dependence of Amperometry is based on the measurement of the their response on the buffer capacity of the sample (see current resulting from the electrochemical oxidation or Section 4.2) and on its ionic strength. reduction of an electroactive species. It is usually performed by maintaining a constant potential at a Pt, Au 2.2.3. Surface charge using field-effect transistors or C based working electrode or on array of electrodes (FETs) with respect to a reference electrode, which may also An important variation of the systems used to deter- serve as the auxiliary electrode, if currents are low mine ion concentrations are the ion-sensitive field-effect −9 −6 (from 10 to 10 A). The resulting current is directly transistors (ISFETs). An ISFET is composed of an correlated to the bulk concentration of the electroactive ion-selective membrane applied directly to the insulated species or its production or consumption rate within the gate of the FET Covington, 1994. When such ISFETs adjacent biocatalytic layer. As biocatalytic reaction are coupled with a biocatalytic or biocomplexing layer, rates are often chosen to be first order dependent on the they become biosensors, and are usually called either bulk analyte concentration, such steady-state currents enzyme (ENFETs) or immunological (IMFETs) field- are usually proportional to the bulk analyte effect transistors. Operating properties of ENFET and concentration. IMFET-based devices are strongly related to those of the ISE based biosensors. 2.2.2. Potentiometry Potentiometric measurements involve determination 2.2.4. Conductometry of the potential difference between either an indicator Many enzyme reactions, such as that of urease, and and a reference electrode, or two reference electrodes many biological membrane receptors may be monitored separated by a permselective membrane, when there is by ion conductometric or impedimetric devices, using no significant current flowing between them. The trans- interdigitated microelectrodes Cullen et al., 1990. Be- ducer may be an ion-selective electrode (ISE) which is cause the sensitivity of the measurement is hindered by an electrochemical sensor based on thin films or selec- the parallel conductance of the sample solution, usually tive membranes as recognition elements Buck and a differential measurement is performed between a sen- Lindner, 1994. The most common potentiometric sor with enzyme and an identical one without enzyme. devices are pH electrodes; several other ion (F−,I−, − + + 2+ + CN ,Na,K,Ca ,NH4 ) or gas (CO2,NH3) selective electrodes are available. The potential differences between these indicator and reference electrodes 3. Analytes or Reactions monitored are proportional to the logarithm of the ion activity or gas fugacity (or concentration), as described by the Biosensors may be further classified according to the Nernst-Donnan equation. This is only the case when (i) analytes or reactions that they monitor. One should the membrane or layer selectivity is infinite or if there is clearly differentiate between the direct monitoring of a constant or low enough concentration of interfering analytes, or of biological activity, and the indirect ions; and (ii) potential differences at various phase monitoring of inhibitors. boundaries are either negligible or constant, except at the membrane/sample-solution boundary. 3.1. Direct monitoring of analyte, or, alternati6ely, of When a biocatalyst layer is placed adjacent to the biological acti6ity producing or consuming analytes potentiometric detector, one has to take into account of, as for any biocatalyst sensor: (1) transport of the Direct monitoring of analytes has clearly been the substrate to be analysed to the biosensor surface; (2) major application of biosensors. Nevertheless, one analyte diffusion to the reacting layer; (3) analyte reac- should be aware that the same biosensor can be a useful tion in the presence of biocatalyst and (4) diffusion of tool also for the direct monitoring of enzyme or living reaction product towards both the detector and the cell activities, by measuring, continuously or sequen- bulk solution. The response of potentiometric biocata- tially, the production or consumption of a given lytic sensors is, as for amperometric biosensors, either compound. D.R. The´6enot et al. / Biosensors & Bioelectronics 16 (2001) 121–131 127

3.2. Indirect monitoring of inhibitor or acti6ator of the SAMs or multilayers, avidin-biotin silanization, biochemical receptor some of such activated membranes being commer- cially available; Alternatively, biosensors have been developed for 5. Bulk modification of entire electrode material, e.g. indirect monitoring of organic pesticides, or inorganic enzyme modified carbon paste or graphite epoxy (heavy metals, fluoride, cyanide, etc.) substances which resin Gorton, 1995. inhibit biocatalytic properties of the biosensor. How- Receptors are immobilized either alone or they are ever such devices are often irreversible. As for immuno- mixed with other proteins, such as bovine serum albu- sensors, their original biological activity can be usually min (BSA), either directly on the transducer surface, or restored only after chemical treatment and such sensors on a polymer membrane covering it. In the latter case, are not classified as reagentless devices. Their potential preactivated membranes can be used directly for the use, especially for environmental monitoring, is thus enzyme or antibody immobilization without further often more as a warning system, not requiring exact chemical modification of the membrane or measurement of the analyte concentration. We recom- macromolecule. mend that they be referred to as single-use biosensors, Apart from the last example, reticulation and cova- except when they can be rapidly and reproducibly lent attachment procedures are more complicated than regenerated, such as the cyanide biosensor using the entrapment ones, but are especially useful in cases inhibition of a cytochrome oxidase which is regenerated where the sensor is so small that the appropriate mem- by washing with phosphate buffer at pH 6.3 Amine et brane must be fabricated directly on the transducer. al., 1995. Under such conditions more stable and reproducible activities can be obtained with covalent attachment.

4. Biosensor construction 4.2. Inner and outer membranes

4.1. Immobilization of biological receptors Besides the reacting layer or membrane, many biosensors, especially those designed for biological or Since the development of the enzyme-based sensor clinical applications, incorporate one or several inner or for glucose, first described by Clark in 1962, in which outer layers. These membranes serve three important glucose oxidase was entrapped between two membranes functions: Clark et al., 1962, an impressive literature on methods 1. Protecti6e barrier. The outer membrane prevents of immobilization and related biosensor development large molecules, such as proteins or cells of biologi- has appeared. These methods have been extensively cal samples, from entering and interfering with the reviewed elsewhere Turner et al., 1987; Guilbault, 1984; reaction layer. It also reduces leakage of the reacting Mosbach and (Ed.), 1988; Cass and (Ed.), 1990; Go¨pel layer components into the sample solution. This et al., 1991; Blum et al., 1991; Kas et al., 1996. Biolog- function of the outer membrane is important, for ical receptors, i.e. enzymes, antibodies, cells or tissues, example, for implanted glucose sensors, since its with high biological activity, can be immobilized in a glucose oxidase is of non-human origin and may thin layer at the transducer surface by using different cause immunological reactions. Furthermore, a procedures. The following procedures are the most properly chosen membrane exhibits permselective generally employed: properties, which may be additionally beneficial to 1. Entrapment behind a membrane: a solution of en- the biosensor function. It may decrease the influence zyme, a suspension of cells or a slice of tissue is, of possible interfering species detected by the trans- simply, confined by an analyte permeable membrane ducer. For example, most in vivo or ex vivo glucose as a thin film covering the electrochemical detector; biosensors present a negatively charged inner cellu- 2. Entrapment of biological receptors within a poly- lose acetate membrane in order to decrease the meric matrix, such as polyacrylonitrile, agar gel, interfering effect of ascorbate or urate, electrochem- polyurethane (PU) or poly(vinyl-alcohol) (PVAL) ically detected together with enzymatically gener- membranes, sol gels or redox hydrogels with redox ated hydrogen peroxide. +/2+ centers such as [Os(bpy)2Cl] Rajagopalan et 2. Diffusional outer barrier for the substrate. As most al., 1996; enzymes follow some form of Michaelis-Menten 3. Entrapment of biological receptors within self as- kinetics, enzymatic reaction rates are largely non- sembled monolayers (SAMs) or bilayer lipid mem- linear with concentration. Nevertheless, linear dy- branes (BLMs); namic ranges may be large if the sensor response is 4. Covalent bonding of receptors on membranes or controlled by the substrate diffusion through the surfaces activated by means of bifunctional groups membrane and not by the enzyme kinetics. This or spacers, such as glutaraldehyde, carbodiimide, control is achieved by placing a thin outer mem- 128 D.R. The´6enot et al. / Biosensors & Bioelectronics 16 (2001) 121–131

brane over a highly active enzyme layer Scheller and methods, e.g. precision, accuracy, interlaboratory and Pfeiffer, 1978; Bindra et al., 1991: the thinner is this interpersonal reproducibility, it is recommended that membrane, the shorter is the biosensor response standard IUPAC definitions be followed Inczedy et al., time. Furthermore, such diffusional barrier also 1998; Buck and Lindner, 1994. makes the sensor response independent of the Most of the discussion below relates to enzyme-based amount of active enzyme present and improves the biosensors. In the case of immunosensors, a key issue is sensor response stability. the capture capacity of the surface, i.e. the number of 3. Biocompatible and biostable surfaces. Biosensors are molecules on the surface which are actually biologically subject to two sets of modifications when they are in active. One of the methods for assessing this parameter direct contact with biological tissues or fluids, i.e. consists in measuring the specific activity, i.e. the ratio implanted in vivo or, more generally, in biologically of the number of active molecules/the total number of active matrices, such as cell cultures: immobilized molecules. This figure is very dependent on modification of the host biological sample by the mode of immobilization (molecular orientation, various reactions caused by biosensor introduc- number of points of attachment) and can range from tion and toxicity, mutagenicity, carcinogenicity, about 0.15 to 0.3, rarely reaching 1. This capture thrombogenicity or immunogenicity of its capacity becomes especially important when the surface elements, is decreased, as in microfluidic applications. Another modification of the biosensor operating proper- important issue for immunosensors is the question of ties by sample components or structure: external whether the surface can be regenerated without signifi- layer or inner detector fouling, inhibition of the cant loss of activity (see Section 2.1.2). biorecognition reaction, substrate and/or co-sub- The rapid proliferation of biosensors and their diver- strate transport rate towards the biorecognition sity has led to a lack of rigour in defining performance area. criteria. Although each sensor can only truly be evalu- Apart from molecular recognition systems or trans- ated for a particular application, it is still useful to ducers which require direct contact between sample and establish standard protocols for evaluation of perfor- biological receptor, the choice of an outer layer is mance criteria, in accordance with standard IUPAC generally essential for the stability of the response after protocols or definitions Inczedy et al., 1998. These implantation. Depending upon sensor diameter, i.e. protocols are recommended for general use and include centimeter or sub-millimeter range, pre-cast mem- four sets of parameters, described below. branes, such as those made of collagen, polycarbonate or cellulose acetate, or, alternatively, polymeric materi- 5.1. Calibration characteristics: sensiti6ity, working and als deposited by dip- or spin-coating (cellulose acetate, linear concentration range, detection and quantitati6e Nafion or polyurethane) may be used. Microsize determination limits biosensors are often prepared by entrapping the enzyme by an electropolymerization step. Sensor calibration is performed, in general, by adding If the implantation of the biosensor does not materi- standard solutions of the analyte and by plotting ally affect the normal functioning of the host medium steady-state responses Rss, possibly corrected for a and if the medium does not materially affect the normal blank (often called background) signal Rbl, versus the operation of the biosensor, then the biosensor is consid- analyte concentration, c, or its logarithm, log c/c°, ered to be biocompatible. where c° refers to a reference concentration, usually 1 mol l−1, although such high concentration value is never used, the highest values reaching usually 1–10 5. Performance criteria: guidelines for reporting mmol l−1. Transient responses are important for se- characteristics of the biosensor response quential samples but are less significant for continuous monitoring: within several possibilities, they are gener- As for any sensor based on molecular recognition ally defined as the maximum rates of variation of the

Buck and Lindner, 1994, it is important to characterize sensor response (dR/dt) max, after addition of analyte a biosensor response: it is even more important here into the measurement cell. A convenient way to per- since operating parameters may indicate the nature of form such calibrations, under well-defined hydrody- the rate-limiting steps (transport or reaction) and facili- namic conditions, is to place the biosensor in a FIA tate biosensor optimization in a given matrix. This system for sequential sample analysis. section will briefly list main performance criteria and The sensitivity and linear concentration range of discuss their relation to properties of the receptor and steady-state calibration curves are determined by plot- transducer parts of electrochemical biosensors. When ting the ratio (Rss −Rbl)/c or (Rss −Rbl)/log c/c° versus performance criteria are not specific to biosensors but log c/c°. This method is much more concise than common to most types of chemical sensors or analytical plotting the usual calibration curves (Rss −Rbl) versus c D.R. The´6enot et al. / Biosensors & Bioelectronics 16 (2001) 121–131 129 or log c/c° since it gives the same weight to low and Naught and Wilkinson, 1997; Umezawa et al., 1995. It high analyte concentration results. Likewise, sensitivity depends both upon the choice of biological receptor and linear range of transient calibration curves are and transducer. Many enzymes are specific. Neverthe- determined by plotting the ratio (dR/dt)max/c or (dR/ less, class (non-selective) enzymes, such as alcohol, dt)max/log c/c° versus log c/c°. In both cases sensitivity group sugar or amino-acid oxidases, peroxidases, lac- is to be determined within the linear concentration case, tyrosinase, ceruloplasmin, alcohol or glucose range of the biosensor calibration curve. NAD-dehydrogenase, etc, have been used for the devel- Electrochemical biosensors always have an upper opment of class biosensors, such as those for determi- limit of the linear concentration range. This limit is nation of phenols, used in environmental monitoring or directly related to the biocatalytic or biocomplexing food analysis. Bacteria, yeast or tissue cultures are properties of the biochemical or biological receptor, naturally non-specific. Whereas oxygen electrodes, pH although in the case of enzyme-based biosensors, it may electrodes and ISFETs show appropriate selectivity, be significantly extended by using an outer layer diffu- metal electrodes are often sensitive to numerous inter- sion barrier to substrate S (see Section 4.2.). The com- fering substances. This direct selectivity can be modified promise for such an extension in the linear when these transducers are associated with receptors. concentration range is, obviously, the decrease of sen- For example, when pH-sensitive ENFETs are used as sor sensitivity. The local substrate concentration, within transducers, their responses are influenced by the buffer the reaction layer, can be at least two orders of magni- capacity of the sample, since part of the released pro- tude lower than in the bulk solution. In relation to the tons react with the buffer components and only the usual parameters for Michaelis-Menten kinetics, i.e. KM remainder is sensed by the transducer. In this case, it is, and Vmax, enzyme based biosensors are often character- in fact, the sensitivity of the biosensor, which is ized by their apparent KM and (Rss −Rbl)max: the first modified, and not its selectivity. parameter represents the analyte concentration yielding When transducer interfering substances are well iden- a response equal to half of its maximum value, (Rss − tified, such as ascorbate or urate in glucose sensors

Rbl)max for infinite analyte concentration. When the based on hydrogen peroxide detection, their influence apparent KM is much larger than its value for soluble may be restricted by the application of appropriate enzyme, it means either that a significant substrate inner or outer membranes (see Section 4.2.). Alterna- diffusion barrier is present between the sample and the tively, a compensating sensor may be introduced in the reaction layer, or that the rate of reaction of the set-up, without biological receptor on its surface The´v- co-substrate, S%, with the enzyme is increased. As for enot et al., 1979. Such a differential design is frequently enzyme solution kinetics, the apparent KM is usually used for ISFET- or ENFET-based sensors. Within determined by Lineweaver-Burk reciprocal plots, i.e. various methods for biosensor selectivity determination,

1/(Rss −Rbl) versus 1/c. As for any electrochemical two are recommended depending upon the aim of its sensor, one should state the composition and the num- measurement. The first one consists in measuring the ber of standards used and how the sample matrix is biosensor response to interfering substance addition: a simulated or duplicated. It may be necessary to specify calibration curve for each interfering substance is plot- procedures for each biosensor type and application. ted and compared to the analyte calibration curve, This is especially important for single-use biosensors under identical operating conditions. Selectivity is ex- based on immuno affinity (see Section 2.1.2) or on pressed as the ratio of the signal output with the inhibition reactions (see Section 3.2). analyte alone and with the interfering substance alone, The sensitivity is the slope of the calibration curve, at the same concentration as that of the analyte. In the i.e. (Rss −Rbl) versus c or log c/c°. One should always second procedure interfering substances are added, at avoid confusion between sensitivity and detection lim- their expected concentration, into the measuring cell, its. The limit of detection (LOD) and of quantification already containing usual analyte concentration, at the (LOQ) take into account the blank and the signal mid-range of its expected value. Selectivity is then fluctuation (noise). Their definition is not specific to expressed as the percentage of variation of the biosen- biosensors and IUPAC recommendations should be sor response: although more easily quantified than the used. The working concentration range, which may calibration curve comparison performed in the first considerably extend the linear concentration range, is procedure, the second method is characteristic of each determined by the lower and upper limits of application and presents a more restricted significance. quantification. Such selectivity may depend on the analyte concentration range that is determined. 5.2. Selecti6ity and reliability The reliability of biosensors for given samples depends both on their selectivity and their reproducibility. Biosensor selectivity is determined and expressed as It has to be determined under actual operating condi- for other amperometric or potentiometric sensors Mc- tions, i.e. in the presence of possible interfering sub- 130 D.R. The´6enot et al. / Biosensors & Bioelectronics 16 (2001) 121–131 stances. In order to be reliable for an analyst, the factors mentioned above on response time Eddowes, biosensor response should be directly related to the 1990. Modeling is somewhat limited by the necessary analyte concentration and should not vary with fluctua- knowledge of a large number of sensor parameters tions of concentrations of interfering substances within (thickness, partition and diffusion coefficients of each the sample matrix. Thus, for each type of biosensor and membrane or layer for each species, distribution of sample matrix, one should clearly state the reasonable biocatalytic or biocomplexing activity within the sensor interference that should be considered and how its layers, transducer operating properties, etc.). Often, influence should be quantified. This reliability determi- such modelling is restricted to steady-state operation nation is necessary for accuracy assessment for each and is not sufficiently advanced for the evaluation of application. transient responses and response in general Albery et al., 1987. 5.3. Steady-state and transient response times, sample throughput 5.4. Reproducibility, stability and lifetime

Steady-state response time is easily determined for Definition of reproducibility is the same for electro- each analyte addition into the measurement cell. It is chemical biosensors as for any other analytical device: the time necessary to reach 90% of the steady-state reproducibility is a measure of the scatter or the drift in response Lindner et al., 1986. Transient response time a series of observations or results performed over a corresponds to the time necessary for the first derivative period of time. It is generally determined for the ana- of the output signal to reach its maximum value (dR/ lyte concentrations within the usable range. The operational stability of a biosensor response may dt)max following the analyte addition. Both response times depend upon the analyte, co-substrate and vary considerably depending on the sensor geometry, product transport rates through different layers or method of preparation, as well as on the applied recep- membranes. Therefore, the thickness and permeability tor and transducer. Furthermore it is strongly depen- of these layers are essential parameters. Both response dent upon the response rate limiting factor, i.e. a substrate external or inner diffusion or biological recog- times also depend upon the activity of the molecular nition reaction. Finally, it may vary considerably de- recognition system. The higher this activity, the shorter pending on the operational conditions. For operational is this response time. Finally, they also depend upon stability determination, we recommend consideration of the mixing conditions of the sample into the batch the analyte concentration, the continuous or sequential measurement cell: such mixing time may not be negligi- contact of the biosensor with the analyte solution, ble. A simple way to better define such hydrodynamic temperature, pH, buffer composition, presence of or- conditions in the biosensors vicinity is to use a FIA ganic solvents, and sample matrix composition. Al- system for sample introduction. When biosensors are though some biosensors have been reported usable part of FIA systems, their response time is defined as under laboratory conditions for more than one year, for any other FIA detector: if the analyte concentration their practical lifetime is either unknown or limited to is varied stepwise, steady-state and transient response days or weeks when they are incorporated into indus- times are defined as in batch; alternatively, if analyte trial processes or to biological tissue, such as glucose pulses are introduced into the circulating fluid, only biosensors implanted in vivo Pickup and The´venot, transient responses are available. Finally, when sensors 1993. For storage stability assessment, significant are implanted in vivo or placed in or in the vicinity of parameters are the state of storage, i.e. dry or wet, the industrial reactors, their operational response time also atmosphere composition, i.e. air or nitrogen, pH, buffer incorporates the analyte and co-substrate transport composition and presence of additives. rates towards the sensor site. While it is relatively easy to determine the laboratory When biosensors are used for sequential measure- bench stability of biosensors, both during storage and ments, either in batch or flow-through set-ups, the operation in the presence of analyte, procedures for sample throughput is a measure of the number of assessing their behaviour during several days of intro- individual samples per unit of time. This parameter duction into industrial reactors is much more complex takes into account the steady-state or transient response and difficult to handle. In both cases, i.e. bench or times but also includes the recovery time, i.e. the time industrial set-ups, it is necessary to specify whether needed for the signal to return to its base line. lifetime is a storage (shelf) or operational (use) lifetime Both types of response times, as well as sample and what the storage and operating conditions were, throughput, may depend on sample composition, ana- and specify substrate(s) concentration(s), as compared lyte concentration, or the sensor history: such depen- to the apparent Michaelis-Menten constant KM (see dencies should be tested and quantified. Section 5.1). Knowledge of the biosensor rate limiting Theoretical modelling of biosensor operation enables step or factor is especially important for the under- a better understanding of the relative importance of the standing of stability properties. D.R. The´6enot et al. / Biosensors & Bioelectronics 16 (2001) 121–131 131

Finally, the mode of assessment of lifetime should be Amine, A., Alafandy, M., Kauffmann, J.M., 1995. Anal. Chem. 67, specified, i.e. by reference to initial sensitivity, upper 2822. Bartlett, P.N., Tebbutt, P., Whitaker, R.G., 1991. Prog. React. Kinet. limit of the linear concentration range for the calibra- 16, 55. tion curve, accuracy or reproducibility. We recommend Bergveld, P., The´venot, D.R., 1993. In: Turner, A.P.F. (Ed.), Ad- the definition of lifetime, noted tL, as the storage or vances in biosensors, Supplement 1. JAI Press, p. 31. operational time necessary for the sensitivity, within the Bindra, D.S., Zhang, Y., Wilson, G.S., Sternberg, R., The´venot, D.R., linear concentration range, to decrease by a factor of 10 Moatti, D., Reach, G., 1991. Anal. Chem. 63, 1692. Blum, L.J., Coulet, P.R. (Eds.), 1991. Biosensor principles and appli- (tL10) or 50% (tL50). For the determination of the stor- cations. Marcel Dekker, New York. age lifetime, we suggest comparison of sensitivities of Buck, R.P., Lindner, E., 1994. Pure Appl. Chem. 66, 2527. different biosensors, derived from the same production Cammann, K., 1977. Fresenius Z. Anal. Chem. 287, 1. batch, after different storage time under identical condi- Cass, A.E.G. (Ed.), 1990. Biosensors: a practical approach. I.R.L. tions. Biosensor stability may also be quantified as the Press, Oxford. Clark, L.C., Lyons, J.R., Lyons, C., 1962. Ann. N.Y. Acad. Sci. 102, drift, when the sensitivity evolution is monitored during 29. either storage or operational conditions. The drift de- Covington, A.K., 1994. Pure Appl. Chem. 66 (3), 565. termination is especially useful for biosensors which Cullen, D.C., Sethi, R.S., Lowe, C.R., 1990. Anal. Chim. Acta 231, 33. evolution is either very slow or studied during rather Durst, R.A., Baümner, A.J., Murray, R.W., Buck, R.P., Andrieux, short period of time. C.P., 1997. Pure Appl. Chem. 69, 1317. Eddowes, M.J., 1990. In: Cass, A.E.G. (Ed.), Biosensors: a practical approach. I.R.L. Press, Oxford, pp. 211. Go¨pel, W., Jones, T.A., Kleitz, M., Lundstro¨m, I., Seiyama, T., 1991. 6. Conclusion Chemical and biochemical sensors. In: Go¨pel, W., Hesse, H., Zemel, J.N. (Eds.), Sensors — a comprehensive survey, vols. 2–3. VCH, New York. Some characteristics of biosensors are common to Gorton, L., 1995. Electroanal. 7, 23. different types of electrochemical sensors. Others are Guilbault, G.G., 1984. Handbook of immobilized enzymes. Marcel more specific to biosensor principles but may be com- Dekker, New York. mon to different types of transducers. Responses of Inczedy, J., Lengyel, T., Ure, A.M. (Eds.), 1998. JUPAC compendium biosensors will be controlled by kinetics of recognition of analytical nomenclature. 3rd edition, Blackwell Science, Oxford. Kas, J., Marek, M., Stastny, M., Voll, R., 1996. Chapter 6. In: Brabec, and transduction reactions, or by mass transfer rates. V., Valz, D., Milazzo, G. (Eds.), Experimental techniques in Determination of the rate-limiting step is clearly essen- bioelectrochemistiy. Birkaüser Verlag, Basel, pp. 361–447. tial for the understanding, optimization and control of Kiefer, H., Klee, B., John, E., Stierhof, Y.D., Ja¨hnig, F., 1991. such biosensor performance criteria. Biosens. Bioelectron. 6, 233. As with most nomenclature documents on complex Kutner, W., Wang, J., L’Her, M., Buck, R.P., 1998. Pure Appl. Chem. 70 (6), 1301. technological developments, the definitions, terminol- Lindner, E., Toth, K., Pungor, E., 1986. Pure Appl. Chem. 58 (3), 469. ogy, and classification of electrochemical biosensors McNaught, A.D., Wilkinson, A., 1997. Compendium of chemical cannot unambiguously address every detail, nuance and terminology, Second edition. Blackwell Science, Oxford. contingency of this diverse subject. There will invari- Mosbach, K. (Ed.), 1988. Methods in enzymology. Vol.137, Academic ably be exceptions to some of the nomenclature and Press, New York. Pickup, J., The´venot, D.R., 1993. In: Turner, A.P.F. (Ed.), Chemical classification recommendations. However, this is a liv- sensors for in vivo monitoring: advances in biosensors. Supplement ing document and, as such, will be revised periodically 1, JAI Press, pp. 201. as needed to address ambiguities and new technological Rajagopalan, R., Aoki, A., Heller, A., 1996. J. Phys. Chem. 100, 3719. developments as they arise in the evolution of electro- Scheller, F.W., Pfeiffer, D., 1978. Z. Chem. 18, 50. chemical biosensors. Comments on this document are Sugawara, M., Sato, H., Ozawa, T., Umezawa, Y., 1997a. In: Scheller, F.W., Scubert, F., Fedrowitz, J. (Eds.), Frontiers in biosensorics actively solicited from scientists working in this, and — fundamental aspects. Birkhaüser Verlag, Basel, pp. 121–131. related, fields of research. Sugawara, M., Hirano, A., Rehak, M., Nakanishi, J., Kawai, K., Sato, H., Umezawa, Y., 1997b. Biosens. Bioelectron. 12 (5), 425. The´venot, D.R., Coulet, P.R., Sternberg, R., Laurent, J., Gautheron, D.C., 1979. Anal. Chem. 51, 96. References Turner, A.P.F., Karube, I., Wilson, G.S. (Eds.), 1987. Biosensors, fundamentals and applications. Oxford University Press, Oxford. Aizawa, M., 1991. Anal. Chim. Acta 250, 249. Umezawa, Y., Umezawa, K., Sato, H., 1995. Pure Appl. Chem. 67, Albery, W.J., Craston, D.H., 1987. In: Turner, A.P.F., Karube, I., 507. Wilson, O.S. (Eds.), Biosensors: fundamentals and applications. Wollenberger, U., Schubert, F., Scheller, F.W., 1993. Trends Biotech- Oxford University Press, Oxford, p. 180. nol. 11, 255.