Publications

Lunar spherules are small glass beads that are formed mainly as a result of small impacts on the lunar surface; the ages of these impacts can be determined by the 40 Ar/ 39 Ar isochron technique. Here, 155 spherules separated from 1 gram of Apollo 14 soil were analyzed using this technique. The data show that over the last ∼3.5 billion years, the cratering rate decreased by a factor of 2 to 3 to a low about 500 to 600 million years ago, then increased by a factor of 3.7 ± 1.2 in the last 400 million years. This latter period coincided with rapid biotic evolutionary radiation on Earth.

An entry from the Cambridge Structural Database, the world’s repository for small molecule crystal structures. The entry contains experimental data from a crystal diffraction study. The deposited dataset for this entry is freely available from the CCDC and typically includes 3D coordinates, cell parameters, space group, experimental conditions and quality measures.

A procedure has been developed for the selective etching of Ag from Pt nanoparticles of well-defined shape, resulting in the formation of elementally-pure Pt cubes, cuboctahedra, or octahedra, with a largest vertex-to-vertex distance of {approx}9.5 nm from Ag-modified Pt nanoparticles. A nitric acid etching process was applied Pt nanoparticles supported on mesoporous silica, as well as nanoparticles dispersed in aqueous solution. The characterization of the silica-supported particles by XRD, TEM, and N{sub 2} adsorption measurements demonstrated that the structure of the nanoparticles and the mesoporous support remained conserved during etching in concentrated nitric acid. Both elemental analysis and ethylene hydrogenation indicated etching of Ag is only effective when [HNO{sub 3}] {ge} 7 M; below this concentration, the removal of Ag is only {approx}10%. Ethylene hydrogenation activity increased by four orders of magnitude after the etching of Pt octahedra that contained the highest fraction of silver. High-resolution transmission electron microscopy of the unsupported particles after etching demonstrated that etching does not alter the surface structure of the Pt nanoparticles. High [HNO{sub 3}] led to the decomposition of the capping agent, polyvinylpyrollidone (PVP); infrared spectroscopy confirmed that many decomposition products were present on the surface during etching, including carbon monoxide.

This poster describes results associated with a project for the National Science Foundation, grant # 1148011. It was prepared for a PIs meeting on 2018-04-30. The primary results are •The Materials Simulation Toolkit (MAST) for high-throughput defect and diffusion modeling •A Machine Learning extension (MAST-ML) to rapidly generate machine learning models from materials data. •Online defect and diffusion analysis apps on MaterialsHub. •The world’s largest computed and machine learning enhanced diffusion database with easy online search. •Valuable research results using these tools and data, e.g. new fuel cell materials. •Workforce training through the Informatics Skunkworks, an undergraduate materials informatics group.

Vision-based robotics often separates the control loop into one module for perception and a separate module for control. It is possible to train the whole system end-to-end (e.g. with deep RL), but doing it "from scratch" comes with a high sample complexity cost and the final result is often brittle, failing unexpectedly if the test environment differs from that of training. We study the effects of using mid-level visual representations (features learned asynchronously for traditional computer vision objectives), as a generic and easy-to-decode perceptual state in an end-to-end RL framework. Mid-level representations encode invariances about the world, and we show that they aid generalization, improve sample complexity, and lead to a higher final performance. Compared to other approaches for incorporating invariances, such as domain randomization, asynchronously trained mid-level representations scale better: both to harder problems and to larger domain shifts. In practice, this means that mid-level representations could be used to successfully train policies for tasks where domain randomization and learning-from-scratch failed. We report results on both manipulation and navigation tasks, and for navigation include zero-shot sim-to-real experiments on real robots.

Abstract The rapid development of foundation potentials (FPs) in machine learning interatomic potentials demonstrates the possibility for generalizable learning of the universal potential energy surface. The accuracy of FPs can be further improved by bridging the model from lower-fidelity datasets to high-fidelity ones. In this work, we analyze the challenge of this transfer learning (TL) problem within the CHGNet framework. We show that significant energy scale shifts and poor correlations between GGA and r 2 SCAN hinder cross-functional transferability. By benchmarking different TL approaches on the MP-r 2 SCAN dataset, we demonstrate the importance of elemental energy referencing in the TL of FPs. By comparing the scaling law with and without the pre-training on a low-fidelity dataset, we show that significant data efficiency can still be achieved through TL, even with a target dataset of sub-million structures. We highlight the importance of proper TL and multi-fidelity learning in creating next-generation FPs on high-fidelity data.

The ability to create surface and near-surface properties distinct from the bulk properties of a material by means of coating or surface modification techniques is an important new field of development for engineering materials, encompassing ion implantation, plasma processing, reactive deposition methods, ion beam deposition and modification, molecular beam epitaxy, and fast quench thermal processing. Attention is given to specific areas of process control which illustrate (1) the use of process development to control material properties, (2) the importance of surface modification and coating to the tribology and ion-surface interactions of semiconductor film growth, and (3) recent findings in surface science which exemplify potential applications of ion-surface interactions that generate novel material qualities.

High-valent oxocobalt(IV) species have been invoked as key intermediates in oxidative catalysis, but investigations into the chemistry of proton-coupled redox reactions of such species have been limited. Herein, the reactivity of an established water oxidation catalyst, [Co₄O₄(OAc)₄(py)₄][PF₆], toward H-atom abstraction reactions is described. Mechanistic analyses and density functional theory (DFT) calculations support a concerted proton-electron transfer (CPET) pathway in which the high energy intermediates formed in stepwise pathways are bypassed. Natural bond orbital (NBO) calculations point to cooperative donor-acceptor σ interactions at the transition state, whereby the H-atom of the substrate is transferred to an orbital delocalized over a Co₃(μ₃-O) fragment. The mechanistic insights provide design principles for the development of catalytic C-H activation processes mediated by a multimetallic oxo metal cluster.