Combinatorial synthesis and screening for discovery of electrocatalysts has received increasing attention, particularly for energy-related technologies. High-throughput discovery strategies typically employ a fast, reliable initial screening technique that is able to identify active catalyst composition regions. Traditional electrochemical characterization via current−voltage measurements is inherently throughput-limited, as such measurements are most readily performed by serial screening. Parallel screening methods can yield much higher throughput and generally require the use of an indirect measurement of catalytic activity. In a water-splitting reaction, the change of local pH or the presence of oxygen and hydrogen in the solution can be utilized for parallel screening of active electrocatalysts. Previously reported techniques for measuring these signals typically function in a narrow pH range and are not suitable for both strong acidic and basic environments. A simple approach to screen the electrocatalytic activities by imaging the oxygen and hydrogen bubbles produced by the oxygen evolution reaction (OER) and hydrogen evolution reaction (HER) is reported here. A custom built electrochemical cell was employed to record the bubble evolution during the screening, where the testing materials were subject to desired electrochemical potentials. The transient of the bubble intensity obtained from the screening was quantitatively analyzed to yield a bubble figure of merit (FOM) that represents the reaction rate. Active catalysts in a pseudoternary material library, (Ni−Fe−Co)O x , which contains 231 unique compositions, were identified in less than one minute using the bubble screening method. An independent, serial screening method on the same material library exhibited excellent agreement with the parallel bubble screening. This general approach is highly parallel and is independent of solution pH.

As the materials science community seeks to capitalize on recent advancements in computer science, the sparsity of well-labelled experimental data and limited throughput by which it can be generated have inhibited deployment of machine learning algorithms to date. Several successful examples in computational chemistry have inspired further adoption of machine learning algorithms, and in the present work we present autoencoding algorithms for measured optical properties of metal oxides, which can serve as an exemplar for the breadth and depth of data required for modern algorithms to learn the underlying structure of experimental materials science data. Our set of 178 994 distinct materials samples spans 78 distinct composition spaces, includes 45 elements, and contains more than 80 000 unique quinary oxide and 67 000 unique quaternary oxide compositions, making it the largest and most diverse experimental materials set utilized in machine learning studies. The extensive dataset enabled training and validation of 3 distinct models for mapping between sample images and absorption spectra, including a conditional variational autoencoder that generates images of hypothetical materials with tailored absorption properties. The absorption patterns auto-generated from sample images capture the salient features of ground truth spectra, and band gap energies extracted from these autogenerated patterns are quite accurate with a mean absolute error of 180 meV, which is the approximate uncertainty from traditional extraction of the band gap energy from measurements of the full transmission and reflection spectra. Optical properties of materials are not only ubiquitous in materials applications but also emblematic of the confluence of underlying physical phenomena yielding the type of complex data relationships that merit and benefit from neural network-type modelling. † Electronic supplementary information (ESI) available: Details of model structure, additional model results, and le containing all model weights. See