Beyond Water Splitting: Efficiencies of Photo‐Electrochemical Devices Producing Hydrogen and Valuable Oxidation Products

Coproduction of valuable chemicals such as halides or hydrogen peroxide concomitant with hydrogen production is presented as a strategy to boost the industrial applicability of photo-electrochemical and photocatalytic systems. Highly efficient two-photon devices can be fabricated (solar-to-chemical efficiencies of >27%), whereas single absorber devices will be appropriate for, for example, HBr splitting and efficiencies of almost 19% can be obtained. Seasonal fluctuations in renewable energy production have stimulated investigation of approaches to photosynthetically produce fuels, such as hydrogen.1-3 These so-called solar fuels need to be scalable to the terawatt level, and therefore the considered coproduct of fuel is usually oxygen.1-5 Since oxygen has negligible economic value, an attractive alternative approach would be to produce value-added oxidation products.6, 7 Production of economically favorable products (see Table 1) or purification of waste streams,8 might considerably reduce time to market of photo-electrochemistry (PEC)-based devices. Oxidation of halides is an interesting alternative to oxidation of water. Anodic conversion of halide-containing salts (NaCl, KBr), or acids (HCl, HBr) leads to gaseous products (i.e., Cl2, Br2). Cl2 is typically used to make PVC, or is used in the production of pharmaceuticals. Cl2 is already produced commercially via electrolysis, thus the fundamental kn owledge for future PEC chlorine electrolysis is already established. The annual production of chlorine is the equivalent of 0.2 TJ worth of H2 storage. While this amount is insufficient for worldwide energy storage requirements, it is still significant. Bromine is frequently used for production of chemical intermediates and its production has already been considered by Texas Instruments using PEC-based devices.15 Various researchers have evaluated this path recently.16, 17 Hydrogen peroxide is another interesting product, due to its high economic value. It is used as disinfectant, bleach, or soft oxidant. While industrially H2O2 is produced using the anthraquinone oxidation process,18 it is also well known that H2O2 can be produced via electrochemical oxygen reduction.19, 20 Finally, H2O2 can be produced via electrochemical oxidation of water, albeit at relatively low faradaic efficiency, due to competitive O2 production. For carbon-based electrodes, efficiencies of 30%–50% have been reported.21 Additionally, oxides such as MnOx appear promising catalysts.22, 23 Table 1 shows that the current market price of the mentioned alternatives is significantly more attractive than oxygen. The material requirements (e.g., to be corrosion resistant) and the techno-economic viability of these devices should share many of the same traits as reported for PEC water splitting.9, 24, 25 Since all of the aforementioned reactions are simple 2e− oxidation reactions with minimal overpotential, losses should be less important than encountered in water oxidation.26, 27 PEC systems for water splitting17, 28-33 have been extensively modeled, and to less extent PEC systems for CO2 reduction have been as well.34, 35 To the best of our knowledge, in-depth modeling of halide oxidation, or oxidation of water to H2O2, has yet to be published. In this perspective, we therefore model single and dual absorber devices where hydrogen (as a fuel) and valuable oxidation products are simultaneously produced. We discuss the effect of the thermodynamic potential and kinetics of the different oxidation reactions on the solar-to-hydrogen (STH) efficiencies of PEC devices and point out how new catalysts for partial oxidation of water to H2O2 will increase PEC efficiencies. Utilizing state-of-the-art modeling approaches provided by a JavaScript, web based model (WBM) developed by Seger et al.,32 we will demonstrate that bromine, chlorine, and hydrogen peroxide based STH conversion efficiencies are high, and PEC technology may be economically viable. Nevertheless, the 2e− oxidation process of Cl2 is favored if well-designed electrodes are used inducing a low overpotential for the Cl2 production reaction, while reducing the high overpotential for the thermodynamically favored OER (in acidic solutions) is challenging.36-42 BrER with only one reaction intermediate can easily be kinetically favored as well.16, 43, 44 Using the WBM we have calculated solar-to-chemical/solar-to-hydrogen (STC/STH) efficiencies of different devices (for further information see the Supporting Information). The baseline parameters used to model these devices were chosen in agreement with previous work by Seger et al.32 (see Table 2 and Table S1 and Figure S1, Supporting Information). For BrER and ClER the parameters were modified to account for changes in overpotential and ionic conductivity. For further information about the parameters, see the Supporting Information. Figure 1a,b shows the obtainable STC/STH efficiencies for a PEC device containing a stacked, two-photon absorber, with the oxidation reactions being BrER (Figure 1a and Figure S2, Supporting Information) and ClER. The corresponding STH efficiency considering OER is shown in Figure S1 (Supporting Information). With increasing thermodynamic potential of the aimed product, the STH efficiency gradually increases, that is, for BrER a maximum STH of 19.7% can be obtained, whereas for OER and ClER, theoretical STH efficiencies of 22.0% and 24.5% can be achieved. The minimal necessary operating voltage for the BrER, OER, and ClER devices at their maximum operating point is 1.48, 1.73, and 1.73 V, respectively (see the Supporting Information for further information). This analysis shows that even though the BrER reaction has a lower overpotential than the OER reaction, it actually is predicted to result in less efficient PEC devices. This peculiarity is rationalized by the definition of efficiency that relates to the chemical potential difference between reactants and products. When water is oxidized this difference is 1.23 V, whereas bromide ions are oxidized at 1.09 V. As noted by Vesborg and Seger, PEC reactions with operating potentials near 2 V are optimally suited to match the efficiency of two-photon tandem PEC devices.35 Given that the BrER reaction operates at 1.48 V, whereas both the OER and ClER operate at 1.73 V, it is clear that deviation from the optimal tandem device voltage significantly affects solar-to-hydrogen fuel production and the BrER efficiency is mitigated by the misalignment of the reaction operating potential with the optimal tandem device potential. On the other hand, the choice of the oxidation reaction only slightly affects the most suitable band gap combination of top and bottom cell absorber. For all cases the optimal top cell absorber should have a band gap of Etop = 1.78 eV. To construct a PEC device with the highest STC efficiency a matching bottom cell absorber with a band gap of Ebot = 0.68–0.94 eV (with the narrowest range of (Ebot = 0.92–0.94 eV) for the OER) should be used. The lower voltage needed for the BrER also means that for most band gap combinations the device will be limited by photocurrent, rather than photovoltage. While at top cell band gaps above ≈1.75 eV, the top cell is providing the limiting photocurrent, at top cells below ≈1.75 eV the bottom cell is providing the limiting photocurrent. Figure 1c allows for the top cell photoabsorber to be thinned to optimize photon matching between both photoabsorbers. The corresponding percentage of thinning is shown in Figure 1d. This greatly expands the region of highly efficient devices, and allows the use of Si bottom cell absorbers with various top cell absorbers. Si is highly efficient and a well-studied material in PEC water splitting. The same top cell thinning approach was also calculated for the ClER (Figure S3, Supporting Information), resulting in a slightly smaller positive effect. Simulated STC efficiencies for PEC devices calculated with more dilute electrolytes are shown in Figure S4 (Supporting Information). For HBr, decreasing the concentration to 0.1 m does not affect the maximum efficiency while decreasing the concentration to 0.01 m HBr results in a drop in efficiency of 7%. Hence, electrolyte-induced losses are negligible compared to other losses for electrolyte concentrations above 0.1 m. Below 0.1 m HBr ionic losses start to dominate.46 PEC devices that produce hydrogen peroxide (Figure 2) are characterized by a thermodynamic redox potential significantly more oxidative than the OER (1.78 V compared to 1.23 V for OER). This makes selective H2O2 production experimentally challenging. Nevertheless, selective production of hydrogen peroxide by oxidation of water was recently reported to occur on carbon materials and MnOx,21, 22 with these catalysts typically achieving a faradaic efficiency of 40%. Since these oxidation catalysts need basic conditions, the PEC modeling was done using either 0.1 or 1 m KOH with HER overpotentials slightly worse than our previous reactions done in acid (ɳHER = 35 mV45 @ 1 mA cm−2 with a Tafel slope of 40 mV dec−1). The hydrogen peroxide overpotential (ɳH2O2) losses were fixed (15022 and 350 mV). The other conditions were chosen similar to BrER and ClER. Figure 2a displays the STC efficiencies for three different conditions assuming that catalytic/engineering improvements can increase the faradaic efficiency for oxidative hydrogen peroxide production to 100%. In reality, for example, in cases where hydrogen peroxide is produced with lower FE, mixtures of oxygen and hydrogen peroxide are produced. Hence, the resulting STC will be based on two different thermodynamic potentials, that is, for OER and H2O2. To achieve a good comparison between the different processes, 100% FE for H2O2 was used, similar to the FE that can be achieved for BrER, ClER, and OER, respectively. In general, all conditions allow high STC, with Figure 2a showing a maximum of up to 27.55% for a 1.87 eV/1.24 eV band gap combination assuming low overpotentials and a highly basic environment. This STC efficiency is quite high, due to the optimal operating potential near 2 V (see the Supporting Information for further information). Given that to date no catalyst providing faradaic efficiency of 100% for the selective two-electron oxidation of water to hydrogen peroxide is known, it should be emphasized that finding a good catalyst driving this reaction with high selectivity is one of the challenges that should be addressed. Still, even with suboptimized STC efficiencies, this might be quite an economically promising approach given the value of H2O2, as well as realizing that H2 produced in this device would still be running at 100% FE. Figure 2b shows that a 200 mV increase in catalytic losses only decreases the efficiency slightly to 25.32%. However, switching to 0.1 m KOH decreases the efficiency more drastically to 21.8%. Si (Eg = 1.12 eV) would be a well-suited bottom cell. For the conditions used in Figure 2b,c, that is, higher overpotential and lower electrolyte conductivity, the optimal bottom material requires a high band gap; a property of, for example, (expensive) GaAs. In Figure 2d, the contour plot obtained for top cell “thinning” is shown (Figure S5, Supporting Information shows the “thinning %” of the top cell absorber). From these data it is obvious that for a tandem device consisting of two stacked materials with similar band gap of 1.55 eV, “thinning” results in an increased (as well as total) STC efficiency of 19.7%. Considering two stacked GaAs cells (band gap of Eg ≈ 1.42 eV), STC efficiencies greater than 10% can be achieved by “thinning” down the top cell absorber by ≈52%. Given that GaAs produces much better photovoltages than used in the model, the actual efficiency could likely be even higher.47 For materials prepared by epitaxial growth this approach might reduce the cost significantly. An interesting point discovered with PEC water splitting is that while the highest potential efficiency can be obtained with a two-photon tandem PEC cell,17, 28-31 recent techno-economic analysis suggests that a single absorber approach utilizing a semiconductor with a large band gap is more feasible in terms of price per kg of hydrogen.24, 25 The authors clearly show that the capital costs for the implementation of hydrogen production facilities using a PEC tandem device approach are circumventing economical hydrogen production from water. Instead, a slurry-based particle approach with one-photon absorber might be more suitable as capital costs can be significantly reduced compared to tandem approaches.24, 25 Figure 3 investigates the approach of a single photoabsorber for BrER, ClER, and H2O2 by modeling the STC efficiencies using the same parameters as used in Figures 1 and 2, that is, faradaic efficiencies of 100%. In comparison to the tandem approach, the order in STC efficiencies is different, with HBr splitting being significantly more efficient than HCl or water-splitting. For HBr splitting, and applying an optimal single absorber with a band gap of 1.78 eV, STC efficiencies of 18.9% can be achieved almost exceeding the maximum efficiency of 19.0% for a dually stacked absorber device. This can be attributed both to the lower thermodynamic potential, as well as to the low band gap bottom cell. In tandem devices, the efficiency of the bottom cell is hampered by low energy photon absorption of the electrolyte.35 For single-cell absorbers, reactions with operating potentials near 1 V are optimally suited to achieve high efficiencies, as is the case for bromine oxidation. The high efficiencies calculated here for single band gap absorbers in HBr conversion are in good agreement with results obtained by McFarland and co-workers16, 48 proving the suitability of efficient single cell absorber devices. As can also be seen from Figure 3, higher STH efficiencies can be obtained for water and HCl conversion in tandem devices, as compared to single-cell absorbers, due to the higher operation potential required to drive these reactions. Finally, the data suggest that hydrogen peroxide production with a single absorber is relatively inefficient with an STC of 6.2%, and is less suited for large-scale solar hydrogen production. Photo-electrochemical splitting of water is considered a promising route to a hydrogen-based economy, however, high economic barriers are impeding its implementation. While costs will decrease with improving performance,24 in this work we show the feasibility of coproduction of valuable chemicals such as halides or H2O2 concomitant with H2 production. By doubling the amount of valuable products, the economic barrier is decreased, allowing solar H2 to enter the market. While the market for halides and/or H2O2 is small in comparison to H2, it is still large enough to allow for the development of a desperately needed photo-electrochemistry industry. It is anticipated that the high value of the oxidation products considered in this study might also allow building devices with lower STC/STH efficiencies. Still a detailed techno-economic feasibility study is required to support this assumption. Based on the modeling results presented in this perspective, solar-to-hydrogen efficiencies producing Br2, Cl2, or H2O2 are as high, or even higher than the efficiency of producing O2. The low thermodynamic potential for HBr activation allows for efficiencies for a single photoabsorber on par with that of a tandem device. Here, only Pt-based HER catalysts were considered, but materials with higher stability, that is, with higher corrosion resistance or lower tendency for halide adsorption,16 should be used in an actual device. For the devices discussed in this paper still several challenges need to be addressed before a successful implementation can be realized. As for PEC water splitting the lifetime of the device is a crucial factor and the harsh conditions used in these processes (but also in PEC water splitting) suggest that protection layers need to be used to limit the corrosion of the light-absorbing semiconductor. For devices producing hydrogen peroxide, catalyst development and testing is required, as to date selective electrocatalysts for oxidative hydrogen peroxide generation are not known. Besides the presented approaches, there might be other reaction combinations of interest for construction of economically viable solar-to-chemical production devices. In particular, devices producing hydrogen peroxide by two-electron reduction of oxygen, coupled to two-electron water oxidation to H2O2 should be considered. F. Diongi and J. Vos are acknowledged for fruitful discussions. As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.