Читать книгу Biofuel Cells - Группа авторов - Страница 24

For the enzymatic redox reactions to be useful in systems containing bioelectrodes (such as fuel cells), the electrode must replace one of the half reactions of the enzyme in its natural environment. This is, the electrode must function as the final electron acceptor or donor, depending on whether it is working as a bioanode or biocathode, respectively. For this to occur, the electrons must travel between the enzyme active sites and the electrode surface in a process termed “electron transfer”. In the context of enzymatic electrodes, this term is employed more loosely than in pure electrochemical sciences, in that the process can involve non-electrochemical charge transport steps, for instance, the diffusion of a mediator (see below) to/from the electrode surface.

In some enzymes, like glucose oxidase, the presence of the enzyme and its substrate at the electrode is, in general, not enough to produce an electrochemical response [3, 55]. As pointed out in Section 1.2.1, the active site of glucose oxidase is buried deep inside the protein. The distance between the redox cofactor at the active site and the electrode surface is too large for efficient electron tunneling. In its initial experiment, Davies noted that, when methylene blue was added to the GOx/glucose solution, an electrochemical response was observed [3]. This charge transport mechanism, termed mediation, opened the door to a whole new class of electrodes.

Mediated electrodes incorporate a molecule that acts as an electron carrier between the enzyme’s active site and the electrode surface. After reacting with their substrate, enzymes change their oxidation state. In their native environment, the enzyme then reacts with another molecule (e.g. O₂ for GOx) and returns to their original oxidation state. The mediators must be able to substitute these natural electron donors/acceptors and readily exchange electrons with the enzyme cofactor. To this end, they must possess the adequate size and charge to be able to access the active site, which can be insulated by oligopeptide and saccharide shells. Furthermore, mediators must also be capable of undergoing a reversible reaction on the electrode surface.

A variety of transition metal (e.g. Os, Fe, Ru, Co) coordination compounds have been employed to this end [16, 56]. Among the ligants used are derivatives of cyclopentadiene [7, 44, 51, 63–65], bipyridine (bpy) [50, 56, 66] and phenanthroline [130]. Careful modification of these ligands with electron donor or withdrawing groups allow for a fine tuning of the redox potential of the mediator. In general, electron donating substituents will increase the electron density of the mediator, therefore making it easier to oxidize. Experimentally, this can be observed as a shift in its formal redox potential in the negative direction. The opposite effect is observed for electron withdrawing functional groups.

Organic molecules have also been used as mediators. As pointed out earlier, methylene blue (a phenothiazine) was one of the first mediators employed, although it is no longer common in enzymatic biofuel cells. Other phenothiazines have been used as well, including methylene green [42], toluidine blue [131] and thionine [6]). Pyrroloquinoline quinone (PQQ), a cofactor of several enzymes, including glucose dehydrogenase has also been used as a mediator for enzymes that do not naturally employ it. Glucose oxidase and lactate dehydrogenase, for example, have been shown to couple with PQQ-containing self-assembled monolayers to shuttle electrons to Au surfaces [132]. It is believed that the fact that the mediator reaction is a 2-electron one, similarly to the enzymatic reaction, can help simplify the electron transfer process [133].

Ramanavicius’ group has studied the use of phenanthroline derivatives (not coordinated to a metal) as mediators for glucose oxidase. They reported that amine electron-donating substituents performed more favorably than nitro electron-withdrawing groups [55]. They note this is in contrast with the above-described results for ligands in metal complexes and cannot be rationalized purely in terms of the effects of electron density on the redox potential. As well, they successfully used a dione phenanthroline derivative to mediate the anode reaction in a GOx-based biofuel cell [13].

In order for a molecule to work as a mediator, its formal redox potential must have the appropriate value relative to the one for the enzyme cofactor . At the anode, must be higher than so that the mediator is capable of reoxidizing the reduced enzyme back to its active state. Conversely, must be lower than at the cathode (Figure 1.10). It is important to realize that these requirements necessary introduce thermodynamic potential losses, therefore decreasing the open circuit voltage of the cell. Therefore, a balance must be sought to have a large enough separation between and to drive the mediation reaction thermodynamically but to minimize the open circuit potential losses. In any case, the potential losses are compensated for with the increased kinetics caused by the presence of the mediator.

Depending on the enzyme employed, the reaction between the enzyme and the mediator might be in competition with the one between the enzyme and its natural acceptor. Such is the case of GOx, in which oxygen competes for the electrons with the mediator. In these cases, the concentration of the mediator must be high enough to favor a predominant enzymemediator reaction. On the other hand, when using glucose dehydrogenase for example, the enzyme activity is not dependent on oxygen and therefore no competition exists [5].

Figure 1.10 Schematic of the i − E curves for mediated and non-mediated electrodes in a fuel cell. The shaded areas represent twice the power of a fuel cell working at a given current value.

Mediator molecules are usually immobilized on the electrode surface along with the enzyme and significant theory has been developed for these systems. Particularly, Barlett and Pratt and have classified the behavior of mediated electrodes in a series of “cases” according to the relative amounts of enzyme, substrate and mediator, as well as the relative rates of charge transport, charge transfer and reagent and product diffusion [134]. This is important because, since the observed current is the result of many sequential steps, different experimental conditions can result in a variety of limiting steps. The understanding of the effect of these parameters on the rates can help make sense of the observed experimental dependencies and design a system accordingly.

Early strategies of mediator immobilization included the use of thiol self-assembled monolayers (SAMs) on Au substrates. The initial SAM would be modified with a mediator molecule that contained the enzyme cofactor at its distal end. Addition of the apoenzymes would result in reconstitution and enzymatic activity [132, 133]. Monolayer-based electrode modification is limited by the available surface area of the electrode. Therefore, surface roughening is employed to maximize the immobilization area.

A different alternative to increase the current per unit of electrode area is the incorporation of the mediator in the chains of a polymer, thus creating the so-called redox polymers [135]. Poly(vinylimidazoles)s, poly (vinylpyridine)s, poly(allylamine)s and poly(ethyleneimine)s have been covalently modified with mediators and cross-linked along with enzymes to form hydrogels. Poly(vinylimidazoles)s, poly(allylamine)s and poly(ethyleneimine)s have primary and/or secondary amine groups that, as discussed in Section 1.3.1, can react with some of the cross-linkers. Poly(-vinylpyridine)s, on the other hand, possess only tertiary amines that are less reactive with the cross-linkers. Redox polymer hydrogels concentrate a large number of mediator molecules close to the surface of the electrode, thus producing high currents. When analyzed via cyclic voltammetry in the absence of enzyme and/or substrate, these systems typically show a diffusion-like behavior. This can be explained though the theory of electron hopping developed in the 60s and 70s by Dahms [136] and Ruff and Friedrich [137]. According to this model, neighboring redox moieties undergo a self-exchange reaction. The flow of electrons in the three-dimensional network of mediator molecules can be described by the same Fick laws that describe diffusion of dissolved substances. Therefore, the rate at which electrons are exchanged in the redox polymer hydrogel is represented by an apparent electron diffusion coefficient (D_e). Consequently, a diffusion-like layer of thickness (D_et)^1/2—where t is time—can be calculated.

If the thickness of this diffusion-like layer is much smaller than the hydrogel thickness, the voltammograms represent a process limited by semi-infinite diffusion [47]. However, if enough time is given so that (D_et)^1/2 grows larger than the film thickness (using slow potential scan rates), thin layer behavior is expected. Of course, this approach is applicable not only to redox polymers, but also to electroactive molecules immobilized by any other method, like ion exchange for example [138].

In early works, poly(allylamine) was modified with [Fe(CN)₅]^3−/2−, [Ru(HN₃)₅]^2+/3+ and [Os(bpy)₂Cl(PyCH₂)] and proven to communicate with immobilized GOx [47, 48]. Poly(vinylpyridine) has as well been modified with [Os(bpy)₂Cl] and cross-linked with PEGDGE in a multienzyme mixture that includes glucose oxidase in an interesting study that compares multiple immobilization approaches [39]. Poly(vinylimidazole) has been modified with [Os(bpy)₂] and compared the performance when cross-linking the hydrogel with GA and PEGDGE. While GA seems to produce higher currents, it presents lower stability than the PEGDGE counterparts [56].

Recent years have seen an increase in the use of poly(ethyleneimine)-based redox polymers as components of enzyme/mediator hydrogels. In particular, linear poly(ethyleneimine) (LPEI) has been attracting considerable attention because it presents less toxicity than branched PEIs [139], which is desirable for implantable electrodes. LPEI has been commonly modified with ferrocene (Fc-C₃-LPEI) [51, 140–145] and its dimethylated (FcMe₂-C₃-LPEI) [51, 54, 63, 65, 146–148] and tetramethylated (FcMe₄-C₃-LPEI) derivatives [63, 64]. Direct comparisons between Fc-C₃-LPEI and FcMe₂-C₃-LPEI [51], and between FcMe₂-C₃-LPEI and FcMe₄-C₃-LPEI [63], showed an 85–90 mV decrease in the peak potentials for each pair of methyl groups added.

Mediator-less charge transport between the enzyme and the electrode surface is an attractive idea. These systems would allow the use of the whole potential difference between the cofactors on each enzyme. As well, the absence of the mediator means there is one less component that can get lost or damaged during operation. In fact, nowadays stability problems are associated more to the mediator than to the enzyme itself [94]. Direct electron transfer (DET) is, however, quite more challenging than MET in general. Let us consider the case of glucose oxidase. As mentioned earlier, it is difficult to achieve DET between GOx and the electrode because of its deeply buried active site. Attempts to covalently immobilize the FAD cofactor on the flat electrode surface and then add the apoenzyme to reconstitute the holoenzyme were futile [149]. This is most likely due to the presence of the surface that sterically hinders the proper fitting of the cofactor therefore altering the conformation of the enzyme.

A number of reports detail the preparation of electrodes which incorporate carbon nanotubes and GOx. Some of them claim that direct electron transfer takes place on the basis of the observation of the oxidation and reduction peaks of the cofactor (FAD) through cyclic voltammetry [150]. Upon addition of oxygen to the solution, an increase of the cathodic current is interpreted as proof of biocatalysis by the FAD in GOx. Furthermore, addition of glucose results in a decrease of the reductive current, which is taken as evidence of the retention of the activity of the immobilized GOx. It has been pointed out by Milton and Minteer, however, that these responses cannot be unequivocally ascribed to GOx DET [5]. A mixture of dissociated cofactor at a proper distance to transfer electrons to the electrode, and remaining active enzyme not in the right distance/conformation for DET would result in the same response. They suggest that, in order to ascertain the presence of DET, a few strategies can be employed, including the analysis of the reaction product and the evaluation of the reaction using denatured, inhibited or mutated enzymes.

Enzymes with more exposed active sites are more suitable for their use in DET. Such is the case, for example, of laccase (see Section 1.2.1). In order to have efficient DET, however, the active site must be within tunneling distance of the electrode [52]. This means that it is not enough to have an active site close to the enzyme surface. Also, the orientation of the enzymes must be such that the active site is pointing towards the electrode. Of course, most enzyme immobilization methods do not control the enzyme orientation; instead, it is random and only a few molecules are in the proper orientation. However, it is possible to use the enzyme tertiary structure itself to direct the immobilization. Certain molecules resembling the substrate (typically polyaromatic compounds) can be immobilized in the electrode surface. When laccase approaches this surface during enzyme immobilization, the substrate-binding pocket where the T1 Cu is located tends to interact (“dock”) with these groups and acquire a proper orientation for DET. These molecules have been termed DET promoters and have also been shown to work with other metalloenzymes like bilirubin oxidase [5].

Not many direct comparisons have been performed between enzymatic fuel cells in which MET or DET take place. Ishida et al. compared the performance of bioanodes containing glucose dehydrogenase, either immobilized with CNT for DET or with ferricyanide for MET [151]. Figures 1.11 c and d show that the onset for oxidation in DET is, as expected occurs at less negative potentials compared to MET (see Figure 1.10). The difference is, however, only of about 100 mV, small value compared to the difference in redox potentials between the ferricyanide and the FAD cofactor of the GDH (~300 mV). The authors attribute the high potential required for MET to the overpotential for FAD oxidation produced by the distance between the CNT and the enzyme active site. Polarization curves of complete cells sharing the same cathode (Figures 1.11e,f) are consistent with the cyclic voltammograms. A slightly higher OCV is obtained for the DET case, at the expense of current and, therefore, power. This example shows that each electron transfer mode has its advantages and disadvantages, and the most suitable strategy must be chosen in a case-by-case basis.

Figure 1.11 Comparison of direct and mediated electron transfer in electrodes containing glucose dehydrogenase. DET is achieved through the addition of carbon nanotubes while ferricyanide is used for MET. (a–d) Cyclic voltammetry of GDH anodes along with the respective negative controls. Black and red curves are in the absence and presence of glucose, respectively. (e–f) Polarization curves of fuel cells produced with the DET (e) and MET (f) anodes. Cathodes used Pt as catalyst. Reprinted with permission from (Ishida, K., Orihara, K., Muguruma, H., Iwasa, H., et al., Comparison of Direct and Mediated Electron Transfer in Electrodes with Novel Fungal Flavin Adenin Dinucleotide Glucose Dehydrogenase. Anal. Sci., 34, 783-787, 2018.). Copyright (2018) Japan Society for Analytical Chemistry.