Interpretable Machine Learning Models for Practical Antimonate Electrocatalyst Performance.

Computationally predicting the performance of catalysts under reaction conditions is a challenging task due to the complexity of catalytic surfaces and their evolution in situ, different reaction paths, and the presence of solid-liquid interfaces in the case of electrochemistry. We demonstrate here how relatively simple machine learning models can be found that enable prediction of experimentally observed onset potentials. Inputs to our model are comprised of data from the oxygen reduction reaction on non-precious transition-metal antimony oxide nanoparticulate catalysts with a combination of experimental conditions and computationally affordable bulk atomic and electronic structural descriptors from density functional theory simulations. From human-interpretable genetic programming models, we identify key experimental descriptors and key supplemental bulk electronic and atomic structural descriptors that govern trends in onset potentials for these oxides and deduce how these descriptors should be tuned to increase onset potentials. We finally validate these machine learning predictions by experimentally confirming that scandium as a dopant in nickel antimony oxide leads to a desired onset potential increase. Macroscopic experimental factors are found to be crucially important descriptors to be considered for models of catalytic performance, highlighting the important role machine learning can play here even in the presence of small datasets.

Protocol for assembling and operating bipolar membrane water electrolyzers.

Renewable energy-driven bipolar membrane water electrolyzers (BPMWEs) are a promising technology for sustainable production of hydrogen from seawater and other impure water sources. Here, we present a protocol for assembling BPMWEs and operating them in a range of water feedstocks, including ultra-pure deionized water and seawater. We describe steps for membrane electrode assembly preparation, electrolyzer assembly, and electrochemical evaluation. For complete details on the use and execution of this protocol, please refer to Marin et al. (2023).1.

Designing a Zn-Ag Catalyst Matrix and Electrolyzer System for CO2 Conversion to CO and Beyond.

CO2 emissions can be transformed into high-added-value commodities through CO2 electrocatalysis; however, efficient low-cost electrocatalysts are needed for global scale-up. Inspired by other emerging technologies, the authors report the development of a gas diffusion electrode containing highly dispersed Ag sites in a low-cost Zn matrix. This catalyst shows unprecedented Ag mass activity for CO production: -614 mA cm-2 at 0.17 mg of Ag. Subsequent electrolyte engineering demonstrates that halide anions can further improve stability and activity of the Zn-Ag catalyst, outperforming pure Ag and Au. Membrane electrode assemblies are constructed and coupled to a microbial process that converts the CO to acetate and ethanol. Combined, these concepts present pathways to design catalysts and systems for CO2 conversion toward sought-after products.

A non-precious metal hydrogen catalyst in a commercial polymer electrolyte membrane electrolyser.

We demonstrate the translation of a low-cost, non-precious metal cobalt phosphide (CoP) catalyst from 1 cm2 lab-scale experiments to a commercial-scale 86 cm2 polymer electrolyte membrane (PEM) electrolyser. A two-step bulk synthesis was adopted to produce CoP on a high-surface-area carbon support that was readily integrated into an industrial PEM electrolyser fabrication process. The performance of the CoP was compared head to head with a platinum-based PEM under the same operating conditions (400 psi, 50 °C). CoP was found to be active and stable, operating at 1.86 A cm-2 for >1,700 h of continuous hydrogen production while providing substantial material cost savings relative to platinum. This work illustrates a potential pathway for non-precious hydrogen evolution catalysts developed in past decades to translate to commercial applications.

Charging and Transport Dynamics of a Flow-Through Electrode Capacitive Deionization System.

We present a study of the interplay among electric charging rate, capacitance, salt removal, and mass transport in "flow-through electrode" capacitive deionization (CDI) systems. We develop two models describing coupled transport and electro-adsorption/desorption which capture salt removal dynamics. The first model is a simplified, unsteady zero-dimensional volume-averaged model which identifies dimensionless parameters and figures of merits associated with cell performance. The second model is a higher fidelity area-averaged model which captures both spatial and temporal responses of charging. We further conducted an experimental study of these dynamics and considered two salt transport regimes: (1) advection-limited regime and (2) dispersion-limited regime. We use these data to validate models. The study shows that, in the advection-limited regime, differential charge efficiency determines the salt adsorption at the early stage of the deionization process. Subsequently, charging transitions to a quasi-steady state where salt removal rate is proportional to applied current scaled by the inlet flow rate. In the dispersion-dominated regime, differential charge efficiency, cell volume, and diffusion rates govern adsorption dynamics and flow rate has little effect. In both regimes, the interplay among mass transport rate, differential charge efficiency, cell capacitance, and (electric) charging current governs salt removal in flow-through electrode CDI.

Quantifying the flow efficiency in constant-current capacitive deionization.

Here we detail a previously unappreciated loss mechanism inherent to capacitive deionization (CDI) cycling operation that has a substantial role determining performance. This mechanism reflects the fact that desalinated water inside a cell is partially lost to re-salination if desorption is carried out immediately after adsorption. We describe such effects by a parameter called the flow efficiency, and show that this efficiency is distinct from and yet multiplicative with other highly-studied adsorption efficiencies. Flow losses can be minimized by flowing more feed solution through the cell during desalination; however, this also results in less effluent concentration reduction. While the rationale outlined here is applicable to all CDI cell architectures that rely on cycling, we validate our model with a flow-through electrode CDI device operated in constant-current mode. We find excellent agreement between flow efficiency model predictions and experimental results, thus giving researchers simple equations by which they can estimate this distinct loss process for their operation.