Complete Protection of O2-Sensitive Catalysts in Thin Films.

Energy conversion schemes involving dihydrogen hold great potential for meeting sustainable energy needs, but widespread implementation cannot proceed without solutions that mitigate the cost of rare metal catalysts and the O2 instability of biological and bioinspired replacements. Recently, thick films (>100 μm) of redox polymers were shown to prevent O2 catalyst damage but also resulted in unnecessary catalyst load and mass transport limitations. Here we apply novel homogeneous thin films (down to 3 μm) that provide protection from O2 while achieving highly efficient catalyst utilization. Our empirical data are explained by modeling, demonstrating that resistance to O2 inactivation can be obtained for nonlimiting periods of time when the optimal thickness for catalyst utilization and current generation is achieved, even when using highly fragile catalysts such as the enzyme hydrogenase. We show that different protection mechanisms operate depending on the matrix dimensions and the intrinsic catalyst properties and can be integrated together synergistically to achieve stable H2 oxidation currents in the presence of O2, potentially enabling a plethora of practical applications for bioinspired catalysts under harsh oxidative conditions.

Introduction

The need for a global implementation of sustainable energy schemes drives the search for cost-effective clean energy sources and storage technologies. In energy conversion processes, catalysts based on earth-abundant elements are needed to replace rare-metal-based anodes such as platinum for the efficient generation and conversion of key energy carriers such as dihydrogen. Recent advances in molecular catalysis,[1−5] inspired by the understanding of enzymes,[6] have raised expectations for the development of cheap, highly active catalysts. However, in contrast with materials based on noble metals, molecular and biological systems easily deactivate when used under operating conditions that are relevant for real-world applications.[7] One particular challenge is an oxidative degradation, as observed for molecular catalysts for H2 oxidation[8] and enzymes capable of H2/H+ interconversion such as hydrogenase (H2ase, Figure A).[9] Because the strict exclusion of O2 over a long-term operation in water splitting cells or fuel cells is impossible, due to membrane crossovers or contamination by O2 from air, the current paradigm in chemical energy conversion is to increase the resistance of the catalyst to O2-induced damage.

Figure 1

Redox films as protection matrices for hydrogenases. (A) Structure of the [NiFe(CN)2CN] active site of D. fructosovorans NiFe hydrogenase, used herein as a model of O2-sensitive catalyst, showing the nearby valine 74 and its deactivation/reactivation in the presence of O2. The whole enzyme is globular, with a diameter of ∼10 nm, Mw = 90 kDa. (B) Schematic representation of viologen-modified dendrimers (detailed structure is given in ref (20), diameter around 5 nm, Mw = 20 kDa) used as building blocks for the assembly of the redox matrix. The viologen moieties act as electron relays and O2-reducing catalysts. The thioacetate group is used as a cross-linking functionality for gelation, leading to film formation. (C) Scheme of the redox matrix homogeneously constituted of hydrogenase molecules and viologen-modified dendrimers. Under turnover for H2 oxidation in the presence of O2, several reaction layers form within the matrix. A first layer of H2 oxidation reaction (H2 ox.) in the vicinity of the electrode produces electrons used for current generation (electrocatalytic layer). A second H2 ox. layer associated with a layer of O2 reduction reaction (O2 red.) near the film/electrolyte interface reduces the incoming O2 (protection layer). (D) Kinetic scheme of the reaction–diffusion processes considered for modeling of the two proposed protection mechanisms: (1) reduction of the incoming O2 and (2) reactivation of the inhibited catalyst. (Definitions of all parameters are given in Table S1.) (E) Fraction of the catalyst contributing to current (top) and catalytic current (bottom) under anaerobic conditions as a function of the film thickness. Regimes V, VII, and III are defined by Bartlett and Pratt[22] as cases in which the current is primarily limited by catalysis, electron transfer, and mass transport, respectively. VIIa and VIIb are the subregimes specific to protection from O2 described in this Article. (F) Depiction of the fluxes of H2 (gray), O2 (red), and electrons (blue) associated with electrocatalysis and protection from O2 in each of the different operating regimes (III, VIIb, and VIIa).

However, recent developments in protective matrices have also allowed the possibility of using O2-sensitive catalysts in an oxidative environment. Such matrices consist of a film of redox polymer deposited onto an electrode and embedding a catalyst such as a hydrogenase. Matrix-bound viologens (Figure B) can serve as redox moieties, owing to their ability to shuttle electrons within the film, their fast charge exchange with hydrogenases, their matching redox potential for H2 oxidation, and, most importantly, their catalytic activity for O2 reduction. These properties allow for the diversion of electrons generated from H2 oxidation toward the reduction of the O2 molecules that penetrate the film at the matrix/solution interface,[10−12] so that the hydrogenase molecules immobilized within the film are shielded from O2 (labels (1) in Figures D and 1F). Hence the enzymes do not experience the oxidative conditions of an operating fuel cell and thus maintain their activity under conditions that are relevant for practical applications. This breakthrough in catalyst protection, initially developed for electrocatalytic H2 oxidation with a NiFe hydrogenase,[10] has since been applied to various other catalysts.[12−15]

However, the actual use of such protective matrices has been limited because this protection mechanism was demonstrated only for thick films (>100 μm), in which the H2 oxidation reaction layer that sends electrons toward the electrode (electrocatalytic layer) is very thin. Hence, the vast majority of the catalyst does not contribute to current generation (Figure E, top). Moreover, the catalytic performances of the film are suboptimal because the H2 oxidation current is limited by the H2 transport within the film (Figure E, bottom). Therefore, thick films provide protection at the cost of mediocre properties in terms of catalyst usage within the matrix and absolute current output.

Until now, thin films have not been explored in this context because thickness was assumed to be an inherent requirement for protection. In addition, the preparation of thin films embedding hydrogenase was impossible until recently because of the poor control of matrix morphology.

Here we apply a new method based on redox dendrimers that enables the formation of catalytic films of controlled and homogeneous thickness. Our results demonstrate not only that the robust O2 protection of a catalyst in a film 6 μm thick is feasible but also that the observed catalytic protection is actually superior to that provided by analogous thick films. This protection coincides with the maximum performance in terms of the catalytic current (Figure F) while maintaining the stability and effective catalyst utilization required for technological applicability. We also show that the minimal thickness that affords O2 immunity can be further decreased by using modified hydrogenases whose rate of reductive reactivation is increased. Under these conditions, the fraction of sacrificial catalyst is negligible. Theoretical models of the protection mechanisms that predict currents and concentration gradients within the redox matrix explain our empirical observation of the effectiveness of thin redox films against catalyst inactivation by O2 and indicate that quasi-infinite protection can be achieved for the hydrogenase even when it is used in minimized loading. These are the key requirements for cost-effective and practical implementations of cheap molecular and bioinspired catalysts under the harsh oxidative conditions of energy-converting devices.

Results

Bioanodes utilizing Desulfovibrio fructosovorans NiFe hydrogenase as a catalyst in viologen-modified dendrimer films[20] of distinct thicknesses (66, 6, and 3.5 μm) were generated (see SI Section S2) to determine the enzyme stability and performance (Figure ). For comparison, wild-type (WT, i.e., unmodified) NiFe hydrogenase and a previously described O2-resistant V74C mutant[16] were cast in thin films. Whereas an O2-sensitive NiFe hydrogenase directly connected to an electrode quickly becomes completely inactivated in the presence of O2,[21] the same enzyme incorporated in a redox matrix is protected from O2 damage,[10] as shown in Figure A. However, Figure shows that the variation in H2 oxidation catalytic current upon exposure to O2 strongly depends on the film thickness and on the nature of the catalyst. Whereas the thickest film (66 μm) shows a loss in current that is fully reversed by the removal of O2 (Figure A), films that are one order of magnitude thinner (6 μm) are apparently completely unaffected by O2 (Figure B). With a film that is only slightly thinner (3.5 μm), the catalytic current rapidly collapses and is only partially recovered upon returning to anaerobic conditions (Figure C). In contrast, when the V74C H2ase mutant is used, the aerobic current of this thinnest film drops to a steady-state value and fully recovers after O2 is removed (Figure D).

Figure 2

Experimental current responses of viologen-modified films embedding Desulfovibrio fructosovorans NiFe hydrogenase, catalyzing H2 oxidation, and being exposed to O2. Wild-type (WT, i.e., unmodified) NiFe hydrogenase embedded in films with decreasing thicknesses from left to right: (A) 66 μm, regime III; (B) 6 μm, regime VIIb; and (C) 3.5 μm, regime VIIa. (D) O2-resistant site-directed mutant (V74C[16]) embedded in a film with a thickness corresponding to regime VIIa for comparison with panel C. The behavior of films in regimes III and VIIb using the V74C mutant is shown in Figure S1. The top panels show the O2 concentration (the shaded areas represent O2 exposure), and the bottom panels show the change in H2 oxidation current divided by the value under anaerobic conditions, J/J0, against time, t, counted from the beginning of exposure to O2. The anaerobic conditions correspond to 100% H2. Aerobic conditions correspond to 5% O2 in 95% H2. Note that 5% O2 is just below the explosion limit for H2/O2 gas mixtures[17] and well above what would be expected under operational conditions in a fuel cell. In all cases, iodide was present in the electrolyte (phosphate buffer, 0.1 M, pH 7 with KI 0.1 M) to minimize H2O2 accumulation;[18,19] this is more convenient than including large enzymes such as catalase. The electrodes were held at a constant potential of 0.21 V versus SHE and rotated at 2000 rpm. The current behavior of films with thicknesses near the regime transitions (4.5 and 33 μm) is shown in Figure S2.

We used theoretical modeling (Figures –6) to explain these unexpected experimental observations and elucidate the underlying kinetics in the reaction–diffusion systems (Figure D). We have considered the diffusion and reaction of the various species (reduced and oxidized viologen; reduced, oxidized, and inactive enzyme; O2 and H2) within the depth of the film, the thickness of which is . We also considered the diffusion of substrate H2 and O2 (diffusion coefficients DS and DO, respectively) and electron hopping between the immobilized viologen moieties. (This is also a diffusion process, with an apparent diffusion coefficient of the electron DA; AΣ is the viologen concentration.) The enzyme (total concentration EΣ) is immobilized and reacts with the oxidized viologen (rate constant ka) and with the substrate H2 (Michaelis parameters kcat, KM); it can also be inactivated by O2 (bimolecular inactivation rate constant ki) and reactivated upon the reaction with reduced viologen (bimolecular rate constant kr). The reduced viologen also reduces O2 with a bimolecular rate constant ko.[11] The reaction–diffusion system depends on 16 independent parameters (three diffusion coefficient, four concentrations, seven kinetic parameters, film thickness, and time). Figures –6 and Figure S1 show numerical solutions of the model obtained by varying only two parameters (the thickness of the film and the rate, kr, at which the inactive enzyme is reactivated upon reduction); all other parameters were fixed to their experimentally determined values. (See the details in SI Sections S3 and S4).

Figure 3

Simulated concentration profiles within the thick film (66 μm, regime III) explaining the current response shown in Figure A and predicting the protection lifetime. (A) Anaerobic and (B) aerobic dimensionless calculated concentration profiles within the viologen-modified films embedding WT hydrogenase, catalyzing H2 oxidation (bottom panels). The black arrow illustrates the rate of progress of the front of the inactive enzyme. A discontinuous x axis is used to zoom in on the thin reaction layers near the inner and outer edges of the film. The local rates of consumption (top panels) of H2 (gray) and O2 (red) are obtained from the second derivative of the concentration profile of each species; the horizontal dotted lines mark Y = 0. The dimensionless distance to the electrode ξ is x/. (C) Change in simulated current divided by the value under anaerobic conditions, j/j0, against dimensionless time, τ (τ = tkaEΣ, counted from the beginning of exposure to O2). (D) Simulated position of the front of the inactive enzyme, ξf, against dimensionless time; the dashed red line is the model. (See SI Section S6.4 and eqs S6.62 and S6.70).

Figure 6

Effect of the reactivation of the enzyme by reduced viologen on the behavior of the thin film in regime VIIa, as observed in Figure D. (A) Anaerobic and (B) aerobic dimensionless simulated concentration profiles within the viologen-modified films embedding O2-resistant site-directed mutant (V74C) hydrogenase, catalyzing H2 oxidation (bottom panels). The black arrow illustrates the rate of progress of the front of the inactive enzyme. The local rate of consumption of H2 (gray) and O2 (red) was obtained from the second derivative of the concentration profile of each species (top panels); the horizontal dotted lines mark Y = 0. The dimensionless distance to the electrode ξ is x/. (C) Change in simulated current divided by the value under anaerobic conditions, j/j0, against τ. (D) Simulated position of the front of the inactive enzyme ξf against dimensionless time τ. (E) Dependence of the normalized position of the front of the inactive enzyme, ξf, and (F) the aerobic current, j/j0, as a function of the dimensionless rate of reactivation kr/ki for various values of the dimensionless thickness, κ; see SI Section S6.5. The black dot indicates the situation in Figure D. Calculations were for the values of the parameters in Table S2.

To frame the various rates of diffusion, kinetic reactions, and protective states within the films, we discuss our results in terms of different “regimes”, each of which defines a region of the parameter space for which the system behaves in a certain manner. Bartlett and Pratt[22] defined seven regimes for the case of a uniform film containing a mediator and an enzyme under anaerobic conditions. For example, the so-called “regime III” (described for the anaerobic conditions in ref (22) and for the aerobic conditions in ref (11)), is observed when the dimensionless thickness of the film is large; it corresponds to the situation of very thick films, where the current response is limited by the long-range diffusion of the substrate and electrons, as described before.[11] In SI Section S5, we demonstrate the kinetic regime assignments before O2 exposure.

Thick Films (Regime III, Figures A and 3)

For the sake of comparison with our new results, we start by recalling the experimental behavior[10] and theory[11] for a thick viologen-modified film containing hydrogenase.[22] In thick films, the oxidation of H2 is confined to a region close to the electrode where the concentration profiles of oxidized viologen and hydrogen overlap. (The peak in gray in the top panel of Figure A shows where H2 is consumed.) The resulting steady-state catalytic anaerobic current density J0 is limited by the mass transport of both H2 and electrons; the thicker the film, the stronger the limitations due to mass transport, and the smaller the currentThe anaerobic current is essentially proportional to the bulk concentration of H2 (= S∞), as indeed experimentally observed for the 66 μm film (Figure S6A). According to the parameter values in SI Section S3, films thicker than 29.5 μm are in regime III. This is confirmed by the experimental current response to O2 of a film with a thickness just above this regime transition (33 μm, Figure S2B), which displays similar behavior to the 66 μm thick films. Moreover, for films with > 29.5 μm, the experimental anaerobic current density decreases with increasing film thicknesses (0.62 ± 0.06 mA/cm2 for ≈ 30 μm compared with 0.38 ± 0.1 mA/cm2 for ≈ 60 μm), which is in agreement with eq and thus further supports the regime III assignment.

Introducing O2 into the solution slightly decreases the steady-state catalytic current density, J, as observed in Figure A (66 μm film) and previously predicted[11]The numerical application of eq based on the parameter values listed in Table S1 gives J/J0 = 0.96, in agreement with the experimental results in Figure A and Figure S2 and the simulation in Figure D. Regarding the transient response and the results from the model given in Figure (and also Figures –6 and Figure S1), the only significant difference between the experimental and simulated currents is that the former shows a gradual current change, whereas the latter show a sharp step upon the addition of O2. This is because we simulate the current for an instantaneous change in O2 concentration, whereas the experimental current reacts to a gradual change in O2 concentration (controlled by purging the electrolyte with a gas mixture containing 5% O2). The transient in the current follows the experimental change in O2 concentration given in the top panels of Figure .

Figure 4

Simulated concentration profiles within the thin film (6 μm, regime VIIb) explaining the current response shown in Figure B and predicting the protection lifetime. (A) Anaerobic and (B) aerobic dimensionless calculated concentration profiles within the viologen-modified films embedding WT hydrogenase, catalyzing H2 oxidation (bottom panels). The black arrow illustrates the rate of progress of the front of the inactive enzyme. The local rate of consumption of H2 (gray) and O2 (red) was obtained from the second derivative of the concentration profile of each species (top panels). (C) Change in simulated current divided by the value under anaerobic conditions, j/j0, against τ. (D) Simulated position of the front of the inactive enzyme ξf = xf/ against dimensionless time, τ (O2 is introduced at τ = 0); the dashed red line is the model. (See SI Section S6.4 and eqs S6.62 and S6.70). See SI Section S7 and Figure S11 for the result of a simulation of constant exposure to O2 until the film collapses.

The protection mechanism can be understood by examining the corresponding concentration profiles (Figure B).[11] Dioxygen (whose concentration profile is shown in red) penetrates the film, inactivates the enzyme, and oxidizes the viologen near the film/solution interface. The oxidized viologen (pink) diffuses inward, allowing catalytic hydrogen oxidation in an outer catalytic layer (peak at ξ = 0.93 in Figure B, top panel). The resulting reduced viologen (blue) diffuses out and consumes the incoming O2 in the outermost reaction layer (red peak at ξ = 0.97 in Figure B). Exposing the film to O2 decreases the hydrogen oxidation current because O2 reduction uses a fraction of the incoming flux of H2. The enzyme is inactivated by O2 in an outer layer comprised between x = xf and x = , where is the thickness of the film. The size – xf of this layer of inactive enzyme (black in Figure B) keeps increasing, very slowly, in proportion to the logarithm of time[11] (eq )The dashed red line in Figure D shows this prediction. The only difference from Fourmond et al.[11] is that we better estimated the duration of the transient (that is, the time delay between the exposure to O2 and the moment the front of the inactive enzyme starts moving in at a steady-state rate, log10(τ) = 1.2 in Figure D; see SI Section S6.4 and eqs S6.62 and S6.70). This transient (log10(τ) < 1) corresponds to the initial penetration of O2, followed by the inactivation of the enzyme in this outer region of the film; the latter step is the slowest (with the values of the parameters in Table S1) and fully determines the duration of the transient, which is therefore proportional to the reciprocal of ki. The rate of progression of the front of inactive enzyme (eq and red line in Figure D) does not depend on ki, but the duration of the transient (eq S6.70), and therefore the lifetime of the film (eq S6.73), depends on ki. However, eq S6.73 shows that the parameters present in eq (which are all independent of the properties of the catalyst) have a much bigger impact on the lifetime than ki. We also note that despite the movement of the front of inactive enzyme, the aerobic current is constant (eq ).[11]

Most importantly, the protection of thick films in regime III results from the separation of reaction zones (Figure B, top panel) and from the very slow movement of the invading layer of inactive enzyme, which together provide very long lifetimes. This protection comes at a cost: Most of the catalyst is unused or used for protection rather than catalysis (Figure F), and the current is limited by long-range diffusion.

Thin Films in Regime VII under Anaerobic Conditions (Figures B,C, 4, and 5)

The 6 and 3.5 μm films are in what Bartlett and Pratt referred to as regime VII.[22] Under anaerobic conditions, the concentration profiles are as shown in Figures A and 5A (bottom panels): There is little H2 depletion in the film, and the current is limited by the rate of electron transfer. H2 oxidation occurs near the surface of electrode in a steady-state reaction layer whose size is determined by the mutual compensation of H2 oxidation and electron diffusion, and the anaerobic current is therefore independent of (22)Figure S6B,C confirms for the 6 and 3.5 μm films the nonlinear dependence of the value of the current on the H2 concentration under anaerobic conditions, in quantitative agreement with eq . The experimental value for the anaerobic current density in regime VII is 0.55 ± 0.1 mA/cm2 (for the four electrodes in regime VII shown in Figure and Figure S2).

Figure 5

Simulated concentration profiles within the thinnest film (3.5 μm, regime VIIa) explaining the current response shown in Figure C and predicting the protection lifetime. (A) Anaerobic and (B) aerobic dimensionless calculated concentration profiles within the viologen-modified films embedding WT hydrogenase, catalyzing H2 oxidation (bottom panels). The black arrow illustrates the rate of progress of the front of the inactive enzyme. The local rate of consumption of H2 (gray) and O2 (red) was obtained from the second derivative of the concentration profile of each species (top panels); the horizontal dotted lines mark Y = 0. The dimensionless distance to the electrode ξ is x/. (C) Change in the simulated current divided by the value under anaerobic conditions, j/j0, against τ. (D) Simulated position of the front of the inactive enzyme, ξf, against dimensionless time, τ.

In considering the additional effect of O2, we shall distinguish two subcases of regime VII, VIIb (Figure , 6 μm) and VIIa (Figure , 3.5 μm).

Thin Films under Aerobic Conditions (Regime VIIb, Figures B and 4)

Unexpectedly, we found that decreasing the film thickness does not come at the expense of resistance to O2. Exposure to O2 actually has no effect on the current of the 6 μm thick film (Figure B), consistent with the corresponding simulation in Figure C. The aerobic concentration profiles in Figure B (bottom panel) explain why the current is nearly independent of the presence of O2: As in regime III, the reduction of O2 decreases the incoming flux of H2 and the concentration of H2 in the inner reaction layer, but because the concentration of H2 remains greater than the Michaelis constant, the enzyme remains saturated. Therefore, the decrease in H2 concentration has little impact on the catalytic rates and thus on the current.

In SI Section S6.3.1, we demonstrate that the relative change in current upon exposure to O2 in regime VIIb isThe observed decrease in current is actually tiny because the KM of NiFe hydrogenase is small (∼5 μM, ref (21)). The above estimation assumes that under anaerobic conditions, there is no depletion of H2; numerical application gives 1 – J/J0 = 7 × 10–5 compared with 7 × 10–4 in Figure C. A more accurate estimation, taking the depletion of H2 into account and giving 1 – J/J0 = 1.1 × 10–4, is described in SI Section S6.3.1. Note that eq is independent of the bulk concentration of O2.

In this regime, although the current is not significantly affected by the presence of O2, the layer of inactive enzyme keeps moving in, as shown in Figure D, where the dashed line shows the theoretical prediction: The rate of progression is the same as in regime III (eq ). Supplementary Figure S12 shows a plot of simulated and calculated lifetimes versus for the conditions in Figure . Under the conditions of Figure B, we calculate that the film becomes totally invaded with an inactive enzyme at τ = 108.2, corresponding to 1 year (with τ/t = kaEΣ = 4.8; see Figure S11). More generally, increasing the film thickness by 1 μm increases the lifetime by over two orders of magnitude (Figure S12). From a practical perspective, this means that the hydrogenase in the electrocatalytic reaction layer can be protected quasi-infinitely from O2 in this regime.

Even Thinner Films under Aerobic Conditions (Regime VIIa, Figures C and 5)

The independence of the catalytic hydrogen oxidation current on O2 observed in Figure B results from the separation of the two H2 oxidation reaction layers (Figure B, top panel). A distinct subclass of regime VII (that we call VIIa) is observed when the film is so small that the two H2 oxidation catalytic layers merge, as shown in Figure C with a film of 3.5 μm. This film is still in regime VII under anaerobic conditions, so the equation for the anaerobic current is as described above (eq ), but the relative decrease in current just after the film is exposed to O2 is significant, as explained in SI Section S6.2. The formation of a single H2 oxidation catalytic layer implies that current generation and protection from O2 compete for electrons, which significantly decreases the reduced viologen concentration gradients (Figure B). Hence the flux of electrons toward the film–electrolyte interface decreases to the extent that it does not counterbalance the flux of O2. The imbalance is amplified by the formation of an inactive enzyme, which leads to fast O2 penetration in the very thin film (Figure B, bottom panel, and Figure D), and, indeed, a simulation such as that in Figure D but with continuous exposure to O2 predicts a short lifetime of ∼10 min.

The two H2 oxidation reaction layers in Figure B (top panel) have the same size, equal to the size of the unique reaction layer under anaerobic conditions (Figure A, top panel). Therefore, the transition between the thicker and thinner regime VII occurs for a film thickness that is twice that of the anaerobic reaction layer (eq and eq S6.44)Numerical application using the parameter values in Table S1 gives = 3.25 μm for the transition between regimes VIIa and VIIb. This value is just below the thickness of the film of Figure C, suggesting that the latter is actually in regime VIIb; however, the progression of O2 inside the film rapidly makes the two catalytic regions collapse so that the profiles in Figure B are compatible with the “thinnest films” behavior in which protection is not sustained (VIIa), as indeed demonstrated by the fast decay of the catalytic current observed upon exposure to O2 (Figure C). For comparison, a slightly thicker film (4.5 μm, Figure S2A), shows a clear regime VIIb behavior (experimental catalytic current unaffected by O2), which supports the value of 3.25 μm as the transition between regimes VIIa and VIIb.

Additional Protection Mechanism Operational in the Thinnest Films (VIIa, Figures D and 6)

The inactivation of NiFe hydrogenase by O2 is mostly reversible: It leads to a mixture of inactive states that can be reactivated by reduction.[23] However, the reaction with O2 is strongly dependent on the side chain of the position 74 residue (Figure A),[24,25] and, in particular, replacing valine 74 with cysteine accelerates the reactivation of the oxidized enzyme.[16,24,26] Here we observe that the V74C mutation abolishes the O2 sensitivity in the very thin hydrogenase film: Figure D shows that the current obtained with a very thin film (3.2 μm) incorporating the V74C hydrogenase mutant reaches a steady-state value under 5% O2 instead of quickly decreasing, as observed for the WT enzyme in Figure C.

The experimental behavior is simply accounted for by the model by a single change: setting a nonzero value of the rate of reactivation of the enzyme by the reduced viologen (whereas kr = 0 in the simulations in Figures –5, all other parameters being the same). This reaction is indicated by the label (2) in Figures D and 1F. The concentration profiles in Figure B reach a true steady state, implying that the protection can theoretically be effective forever (compared with a stability of minutes for the corresponding film incorporating the WT enzyme, Figure C). Stabilization is possible because as the front of the inactive enzyme moves in, the concentration of O2 at the front decreases, whereas the concentration of the reduced mediator remains the same. Thus there is an O2 penetration depth for which the inactivation of active enzymes by O2 is exactly balanced by the reactivation of inactive enzymes.

Figure E shows the position of the stationary front of an inactive enzyme, ξf, as a function of the ratio of the rate constants of reactivation over inactivation for various values of the film thickness (all in dimensionless forms; see SI Section S6.5). The greater the kr/ki, the smaller the O2 penetration depth (ξf is closer to 1). Increasing the film thickness (κ) decreases the value of kr/ki that is necessary to stabilize a certain value of ξf. The exact value of κ matters very much: Increasing κ from 5.5 to 7.5, that is, <30%, decreases the value of kr/ki that stabilizes a given ξf (here ξf = 0.6, horizontal gray dashed line in Figure E) by more than two orders of magnitude. The curves in Figure E all end vertically below a certain value of kr/ki, which strongly depends on the film thickness: This is the minimum value of kr/ki that affords long-lasting protection. Figure F shows the steady-state value of j/j0 corresponding to the penetration depth of Figure E. It increases with both κ and kr/ki; if kr/ki, is large enough that the film is protected, then the steady-state current is nearly independent of kr/ki. The black circle in Figure E marks the value predicted for the experiment shown in Figure D. Note the close agreement between the exact value of j/j0 (0.86 in Figures C) and the value approximated here (0.965, marked by a dot in Figure F).

Control experiments where we embedded the V74C mutant in thicker films are shown in Figure S1. In regimes VIIb and III, the current responses for the wild-type hydrogenase and V74C mutant are identical (Figures A,B and Figure S1, respectively). This confirms that the protection is primarily defined by the O2 reduction reaction catalyzed by the viologen in these regimes. Nevertheless, the faster reactivation rate of V74C leads to a steady-state position of the front of the inactive enzyme (Figure S1, row 7) because a balance between reactivation and deactivation is reached, like for the thin films of V74C hydrogenase in regime VIIa (Figure D).

Discussion

The experimental and theoretical data presented here describe a novel protection regime for O2-sensitive catalysts in thin films (<6 μm) consisting of cross-linked viologen-modified dendrimers. It was previously reported that protection could occur only in thick films (>100 μm) in which only a tiny fraction of the catalyst (e.g., ∼0.3% for a thickness of 100 μm, Figure E) contributed to the current generation. Here we demonstrate that not only are O2-sensitive catalysts protected from damage in thin redox films with a thickness of just 6 μm, but also, remarkably, the protection is even more robust than that reported for thick films. Indeed, in thick films (regime III), the current drops significantly after exposure to O2 (Figure A). In thin films (VIIb), the catalytic current is virtually unaffected by the exposure to O2, and the protection remains long-lasting (Figures B and 4). Thinner films (3.5 μm, regime VIIa) do not resist O2 (Figures C and 5) unless an additional protection mechanism is operational to provide long-term protection (Figures D and 6).

The change in catalytic current over time (or the absence thereof) when the hydrogenase/polymer film is exposed to O2 (Figure ) reveals the extent of the protection in each case considered, but the underlying protection mechanism cannot be deduced from the electrochemical response alone. Therefore, we used a reaction diffusion model to predict the time-resolved concentration gradients and positions of the catalytic layers inside the film; the behavior of the system depends on the values of parameters that could all be independently determined. The model succeeds in predicting the dependence of current on the O2 and H2 concentration as well as the transition between the two newly identified regimes, namely, VIIa and VIIb.

Despite different current responses between regimes III and VIIb, the protection can be explained by the exact same mechanism involving the viologen-catalyzed O2 reduction close to the outer film boundary, which balances the inward flux of O2. The condition for efficient protection is that the two catalytic layers for H2 oxidation remain spatially separated to avoid any competition for electrons between the process for current generation and the process for protection. The reason for the peculiar behavior in regime VIIb (the absence of the effect of O2 on the current) is that in thin films the concentration of H2 remains well above the Michaelis constant, even if it is slightly depleted as a result of O2 consumption. The lifetime of the film is defined by two contributions, namely, the initial rate of enzyme deactivation at the film/solution interface just after the exposure to O2 and the rate of the subsequent progression of the front of the inactive enzyme within the film. Whereas the duration of the former depends on the kinetic constant for catalyst inactivation, ki (eq S6.70), the latter depends only on the parameters that define the film properties (eq ). In particular, the lifetime of the film strongly depends on the viologen concentration and the apparent diffusion coefficient of the electron (AΣ and DA). With the system made of viologen dendrimers described herein, the value of the product DA, which defines the progress of the front of the inactive enzyme (eq ), is about five times larger than that for the previously reported redox matrices.[11]Equation and Equation S6.73 also show that the lifetime of the film scales exponentially with its thickness. The model predicts that the lifetime of a 6 μm film made of these viologen-modified dendrimers is 1 year (Figures S11 and S12); a further increase in film thickness by just 2 μm theoretically leads to a lifetime of 22 000 years. These values are well beyond the intrinsic anaerobic stability of hydrogenase, with a half life of 6 weeks,[10] or of the matrix itself, whose lifetime is limited by the H2O2 generation that degrades the viologen moieties.[27]

Importantly, although the film lifetime is proportional to the reciprocal of the rate constant of the inactivation of the catalyst by O2 (ki), this effect can easily be compensated by adjusting the film parameters because the latter contribute to the exponential component defining the lifetime in eq S6.73. For instance, an eight order of magnitude increase in inactivation rate can be compensated by just a 4 μm increase in film thickness.

Overall, the very thin films are particularly advantageous because the fraction of catalyst contributing to current is enhanced by two orders of magnitude (up to half of the catalyst load in regime VIIb), the current output is maximal (owing to fast mass transport of H2) and immune to O2, and the lifetime remains well above that required for real-world applications.

The minimal thickness that affords long-term protection according to the mechanism described above defines the boundary between regimes VIIa and VIIb, where the two catalytic layers for H2 oxidation merge (eq ), but we also demonstrated that the redox matrix can be used to induce reactivation, which leads to the prediction of everlasting protection, even in the thinnest films (3.2 μm thick, regimes VIIa). The reductive reactivation of hydrogenase occurs for hydrogenases either undergoing direct electron transfer with electrodes[24,28,29] or interfaced by redox films,[15,30] but these previous reports suggested that the electrons needed for reactivation originated from the electrode. Such protection was effective only at a low electrode potential[15] because the rate of reactivation decreases exponentially as the electrode potential increases.[29] High electrode potentials, which do occur in fuel cells under anode-limiting conditions, resulted in complete and irreversible fuel cell shutdown even when using the most resistant hydrogenases.[31] In contrast, we demonstrate here that in the redox-active films the electrons used for reactivation are exclusively generated from the H2 oxidation catalyzed by an active hydrogenase in the vicinity of the layer of the inactive enzyme. This matrix-triggered reactivation is analogous to biological pathways that enable the reactivation of hydrogenase through their neighbors via electron transfer.[32,33] With a mechanism that is independent of the electrode as an electron source, reactivation becomes possible regardless of the operating conditions of the fuel cell and thus serves as a built-in safeguard against uncontrolled oxidative shutdown. Films operating in regime VIIb and embedding enzymes that can reductively reactivate are particularly advantageous because the protection from O2 theoretically lasts forever and the anaerobic currents are fully immune to O2.

Conclusions

We experimentally demonstrated and theoretically elucidated the mechanism of protection of catalysts in redox films in the complete range of catalytic regimes related to the film thicknesses relevant for applications. Two major and unexpected findings could make the research community rethink catalyst use and design. First, micrometer-thick films are sufficient to efficiently protect fragile catalysts such as hydrogenase, enabling their use in energy-converting devices without compromising the current generation efficiency or immobilizing large amounts of unused catalyst. The only requirement is that the matrix efficiently diverts electrons from the catalysts for the reduction of O2. Because even a large inactivation kinetic constant of the catalysts can be easily counterbalanced by minor changes in parameters defining the film properties (in particular, its thickness), highly O2-sensitive catalysts can now be considered for applications without the need to decrease their inactivation rate. Second, the matrix can further enhance protection from O2 for an infinite time period by using a fraction of the electrons diverted from H2 oxidation (rather than from the electrode) to reactivate the catalyst and thus stop the progression of the layer of the inactive catalyst. This creates novel opportunities to utilize reactivation kinetics, even under the uncontrolled and oxidative conditions encountered in operating fuel cells. The quantitative description of the requirements in terms of the catalyst reactivation kinetics and the film properties presented here will guide the future development of biological or bioinspired catalysts that reductively reactivate, as indeed recently observed in synthetic mimics of the NiFe hydrogenase active site,[34] to achieve robust operation, even under the harsh oxidative conditions of H2/O2 fuel cells and other energy-converting devices.