Traditional machine learning (ML) metrics overestimate model performance for materials discovery. We introduce (1) leave-onecluster-out cross-validation (LOCO CV) and (2) a simple nearestneighbor benchmark to show that model performance in discovery applications strongly depends on the problem, data sampling, and extrapolation. Our results suggest that ML-guided iterative experimentation may outperform standard high-throughput screening for discovering breakthrough materials like high-T c superconductors with ML.Materials informatics (MI), or the application of data-driven algorithms to materials problems, has grown quickly as a field in recent years. 9 Across all of these applications, a training database of simulated or experimentally-measured materials properties serves as input to a ML algorithm that predictively maps features (i.e., materials descriptors) to target materials properties. Ideally, the result of training such models would be the experimental realization of new materials with promising properties. The MI community has produced several such success stories, including thermoelectric compounds, 10,11 shapememory alloys, 12 superalloys, 13 and 3d-printable high-strength aluminum alloys. 14 However, in many cases, a model is itself the output of a study, and the question becomes: to what extent could the model be used to drive materials discovery? Typically, the performance of ML models of materials properties is quantified via cross-validation (CV). CV can be performed either in a single division of the available data into a training set (to build the model) and a test set (to evaluate its performance), or as an ensemble process known as k-fold CV wherein the data are partitioned into k nonoverlapping subsets of nearly equal size (folds) and model performance is averaged across each combination of k-1 training folds and one test fold. Leave-one-out crossvalidation (LOOCV) is the limit where k is the number of total examples in the dataset. Table 1 summarizes some examples of model performance statistics as reported in the aforementioned studies (some studies involved testing multiple algorithms across multiple properties).In Table 1, the reported model performance is uniformly excellent across all studies. A tempting conclusion is that any of these models could be used for one-shot high-throughput screening of large numbers of materials for desired properties. However, as we discuss below, traditional CV has critical shortcomings in terms of quantifying ML model performance for materials discovery. Issues with traditional crossvalidation for materials discoveryMany ML benchmark problems consist of data classification into discrete bins, i.e., pattern matching. For example, the Design, System, ApplicationMachine learning (ML) has become a widely-adopted predictive tool for materials design and discovery. Random k-fold cross-validation (CV), the traditional gold-standard approach for evaluating the quality of ML models, is fundamentally mismatched to the nature of materials discovery, and leads to ...
Increasing the activity of Ag-based catalysts for the oxygen reduction reaction (ORR) is important for improving the performance and economic outlook of alkaline-based fuel cell and metal–air battery technologies. In this work, we prepare CuAg thin films with controllable compositions using electron beam physical vapor deposition. X-ray diffraction analysis indicates that this fabrication route yields metastable miscibility between these two thermodynamically immiscible metals, with the thin films consisting of a Ag-rich and a Cu-rich phase. Electrochemical testing in 0.1 M potassium hydroxide showed significant ORR activity improvements for the CuAg films. On a geometric basis, the most active thin film (Cu70Ag30) demonstrated a 4-fold activity improvement vs pure Ag at 0.8 V vs the reversible hydrogen electrode. Furthermore, enhanced ORR kinetics for Cu-rich (>50 at. % Cu) thin films was demonstrated by a decrease in Tafel slope from 90 mV/dec, a commonly observed value for Ag catalysts, to 45 mV/dec. Surface enrichment of the Ag-rich phase after ORR testing was indicated by X-ray photoelectron spectroscopy and grazing incidence synchrotron X-ray diffraction measurements. By correlating density functional theory with experimental measurements, we postulate that the activity enhancement of the Cu-rich CuAg thin films arises due to the non-equilibrium miscibility of Cu atoms in the Ag-rich phase, which favorably tunes the surface electronic structure and binding energies of reaction species.
The chloroplast contains densely stacked arrays of light‐harvesting proteins that harness solar energy with theoretical maximum glucose conversion efficiencies approaching 12%. Few studies have explored isolated chloroplasts as a renewable, abundant, and low cost source for solar energy harvesting. One impediment is that photoactive proteins within the chloroplast become photodamaged due to reactive oxygen species (ROS) generation. In vivo, chloroplasts reduce photodegradation by applying a self‐repair cycle that dynamically replaces photodamaged components; outside the cell, ROS‐induced photodegradation contributes to limited chloroplast stability. The incorporation of chloroplasts into synthetic, light‐harvesting devices will require regenerative ROS scavenging mechanisms to prolong photoactivity. Herein, we study ROS generation within isolated chloroplasts extracted from Spinacia oleracea directly interfaced with nanoparticle antioxidants, including dextran‐wrapped nanoceria (dNC) previously demonstrated as a potent ROS scavenger. We quantitatively examine the effect of dNC, along with cerium ions, fullerenol, and DNA‐wrapped single‐walled carbon nanotubes (SWCNTs), on the ROS generation of isolated chloroplasts using the oxidative dyes, 2’,7’‐ dichlorodihydrofluorescein diacetate (H2DCF‐DA) and 2,3‐bis(2‐methoxy‐4‐nitro‐5‐sulfophenyl)‐2H‐tetrazolium‐5‐carboxanilide sodium salt (XTT). Electrochemical measurements confirm that chloroplasts processed from free solution can generate power under illumination. We find dNC to be the most effective of these agents for decreasing oxidizing species and superoxide concentrations whilst preserving chloroplast photoactivity at concentrations below 5 μM, offering a promising mechanism for maintaining regenerative chloroplast photoactivity for light‐harvesting applications.
Thomas F (2020) Nitride or Oxynitride? Elucidating the Composition-Activity Relationships in Molybdenum Nitride Electrocatalysts for the Oxygen Reduction Reaction. Chemistry of Materials.
Silver-based bimetallic catalysts for the oxygen reduction reaction (ORR) are promising for a wide variety of renewable energy technologies, including alkaline fuel cells and metal-air batteries. The activity of bimetallic catalysts can sometimes surpass that of either constituent element, but the origin of the enhanced performance is still debated. At a given active site, two complementary mechanisms are proposed to explain the performance improvements: the binding energy of intermediate adsorbates can be tuned by direct electronic contributions from the alloying element or by changes in the bond lengths from lattice distortion. To distinguish between these effects and elucidate the respective roles of each element in the bimetallic, it is critical to study catalysts at the molecular scale under reaction conditions. In this work, we use in situ X-ray absorption spectroscopy (XAS) alongside density functional theory (DFT) to show that direct electronic rather than geometric effects are the primary cause of improved ORR activity in a bimetallic CuAg catalyst. Our results indicate that the local bonding as well as the electronic structure of Ag are virtually unchanged by the presence of Cu, whereas the electronic states of Cu in CuAg are significantly altered. DFT calculations support these experimental findings. We show strong evidence that the activity of the bimetallic CuAg catalyst exceeds the sum of the activities of Cu and Ag, not by incremental improvement of the active Ag sites, but by creating highly active Cu-centered catalytic sites. The insight that the main role of Ag in bimetallic catalysts may be to promote its fellow element through local electronic interactions provides a new design principle for engineering the next generation of bimetallic catalysts for the ORR and beyond.
Rigorous in situ studies of electrocatalysts are required to enable the design of higher performing catalysts. Non-platinum group metals for oxygen reduction (ORR) catalysis containing light elements such as oxygen, nitrogen, and carbon are known to be susceptible to both ex situ and in situ oxidation leading to challenges associated with ex situ characterization methods. We have previously shown that bulk O content plays an important role in the activity and selectivity of Mo-N catalysts, but further understanding the role of composition and morphological changes at the surface is needed. Here, we report the measurement of in situ surface changes to a molybdenum nitride (MoN) thin film under ORR conditions using grazing incidence x-ray absorption and 2 reflectivity. We show that the halfwave potential of MoN can be improved by ~ 90 mV by potential conditioning up to 0.8 V vs RHE. Utilizing electrochemical analysis, dissolution monitoring, and surface sensitive x-ray techniques, we show that under moderate polarization (0.3 -0.7 V vs RHE) there is local ligand distortion, O incorporation, and amorphization of the MoN surface, without changes in roughness. Furthermore, with a controlled potential hold procedure, we show that the surface changes concurrent with potential conditioning are stable under ORR relevant potentials.Conversely, at higher potentials (≥ 0.8 V vs RHE) the film incorporates O, dissolves, and roughens, suggesting that in this higher potential regime the performance enhancements are due to increased access to active sites. Density functional theory calculations and Pourbaix analysis provide insight into film stability and oxygen incorporation as a function of potential. These findings coupled with in situ electrochemical-surface sensitive x-ray techniques demonstrate an approach to studying non-traditional surfaces in which we can leverage our understanding of surface dynamics to improve performance with the rational, in situ tuning of active sites.
The drive for greater efficiency in turbomachinery has led to increasingly stringent specifications for the materials used. Current methods for optimizing alloy composition and processing to meet these requirements typically rely on a combination of expert judgment and trial and error. Machine learning offers an alternative approach that leverages data resources to significantly accelerate the optimization timeline through systematic data-informed decision making. In this paper, we demonstrate the effectiveness of machine learning methods for three different alloys classes: aluminum alloys, nickel-based superalloys, and shape memory alloys. In the first two alloy classes, models are built for the alloy mechanical properties based on the composition and processing information. In the case of shape memory alloys, a model is trained to predict the austenite to marten-site transformation temperatures. In addition to achieving high baseline performance, we leverage recent methodological developments to provide well-calibrated, heteroscedastic uncertainty estimates with each prediction. By wrapping these models in an inverse design routine that takes full advantage of uncertainty information, we are able to demonstrate the feasibility of designing new alloys to meet prescribed specifications. The results indicate that this approach has the potential to fundamentally change how new structural and functional alloys are developed.
The electrochemical conversion between oxygen and water is a key process gating the wider deployment of a variety of renewable energy technologies. In particular, the slow kinetics and limited selection of suitable catalysts for the oxygen reduction reaction (ORR) are main impediments to the expansion of critical clean-energy technologies such as hydrogen fuel-cell vehicles or grid-scale energy storage. While Pt-based catalysts are the conventional choice for acidic electrolytes, favorable oxygen reduction kinetics and stability considerations that occur in alkaline electrolytes provide possibility for using non-platinum group metal catalysts, including silver. In particular, bimetallic or alloyed systems of silver and copper have been predicted by theory to be active for the ORR [1]. While these two metals have a large miscibility gap in the bulk, it has been shown that small nanoparticles can support intermixing of the silver and copper in a stable or metastable phase [2]. In addition, we have shown that co-sputtering silver and copper as both thin films and nanoparticles results in ORR catalytic activity that surpasses that of either metal on its own. In this work we investigate these bimetallic catalysts using in situ x-ray absorption spectroscopy (XAS) in order to help understand the activity enhancements seen when combining copper and silver for the ORR. [1] K. Shin, D.H. Kim, and H.M. Lee. ChemSusChem (2013), 6: 1044-1049. [2] P. Grammatikopoulos, J. Kioseoglou, A. Galea, J. Vernieres, M. Benelmekki, R.E. Diaz, and M. Sowwan. Nanoscale (2016), 8: 9780-9790.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.