arXiv subjects

Kristin A. Persson

Publications and source records attributed to Kristin A. Persson.

60 records · Page 4Linked to original sources

arXiv2020

Incorporating electronic information into Machine Learning potential energy surfaces via approaching the ground-state electronic energy as a function of atom-based electronic populations

Machine Learning (ML) approximations to Density Functional Theory (DFT) potential energy surfaces (PESs) are showing great promise for reducing the computational cost of accurate molecular simulations, but at present they are not applicable to varying electronic states, and in particular, they are not well suited for molecular systems in which the local electronic structure is sensitive to the medium to long-range electronic environment. With this issue as the focal point, we present a new Machine Learning approach called bpopNN for obtaining efficient approximations to DFT PESs. The methodology is based on approaching the true DFT energy as a function of electron populations on atoms, which may be realized in practice with constrained DFT (CDFT). The new approach creates approximations to this function with deep neural networks. These approximations thereby incorporate electronic information naturally into a ML approach, and optimizing the model energy with respect to populations allows the electronic terms to self-consistently adapt to the environment, as in DFT. We confirm the effectiveness of this approach with a variety of calculations on Li$_n$H$_n$ clusters.

physics.chem-ph↗

arXiv2019

Two-Dimensional Forms of Robust CO$_2$ Reduction Photocatalysts

Novel photoelectrocatalysts that use sunlight to power the CO$_2$ reduction reaction will be crucial for carbon-neutral power and energy-efficient industrial processes. Scalable photoelectrocatalysts must satisfy a stringent set of criteria, such as stability under operating conditions, product selectivity, and efficient light absorption. Two-dimensional materials can offer high specific surface area, tunability, and potential for heterostructuring, providing a fresh landscape of candidate catalysts. From a set of promising bulk CO$_2$ reduction photoelectrocatalysts, we screen for candidate mono-layers of these materials, then study their catalytic feasibility and suitability. For stable monolayer candidates, we verify the presence of visible-light band gaps, check that band edges can support CO$_2$ reduction, determine exciton binding energies, and compute surface reactivity. We find for SiAs, ZnTe, and ZnSe monolayers, visible light absorption is possible, and reaction selectivity biases towards CO production. We thus identify SiAs, ZnTe, and ZnSe monolayers as targets for further investigation, expanding the chemical space for CO$_2$ photoreduction.

cond-mat.mtrl-sci↗

arXiv2019

Efficient Pourbaix diagrams of many-element compounds

Pourbaix diagrams have long been an essential tool for determining the phase stability of solids and their associated ionic species under electrochemical conditions. In recent years, Pourbaix diagrams have been used for applications ranging from corrosion-resistance alloy design to electrocatalysis, and the data from which they are generated has been enhanced by the availability of materials data in various online databases. However, generation of multi-element Pourbaix diagrams has a critical bottleneck which makes 3-element systems difficult to analyze quickly, and 4 and 5 element systems intractable. In this work, we present a method for constructing Pourbaix diagrams which uses a pre-processing step to circumvent the most egregious computational bottleneck and make many-element Pourbaix diagrams computationally efficient.

cond-mat.mtrl-sci↗

arXiv2019

Pawpyseed: Perturbation-extrapolation band shifting corrections for point defect calculations

Significant progress has been made recently in the automation and standardization of ab initio point defect calculations. However, the task of developing, implementing, and benchmarking charge corrections for density functional theory (DFT) point defect calculations is still an open challenge. Here we present a high-performance Python package called pawpyseed, which can read PAW DFT wave functions and calculate the overlap between wavefunctions from different structures. Using pawpyseed, we implement a new band shifting correction derived from first order perturbation theory. We benchmark this method by calculating the transition levels of several point defects in silicon and comparing to experimental and hybrid functional results. The new band shifting method can shift single-particle energies to improve transition level predictions and can be automated and parallelized using pawpyseed, suggesting it could be a useful method for high-throughput point defect calculations.

cond-mat.mtrl-sci↗

arXiv2019

2DMatPedia: An open computational database of two-dimensional materials from top-down and bottom-up approaches

Two-dimensional (2D) materials have been a hot research topic in the last decade, due to novel fundamental physics in the reduced dimension and appealing applications. Systematic discovery of functional 2D materials has been the focus of many studies. Here, we present a large dataset of 2D materials, with more than 6,000 monolayer structures, obtained from both top-down and bottom-up discovery procedures. First, we screened all bulk materials in the database of Materials Project for layered structures by a topology-based algorithm, and theoretically exfoliate them into monolayers. Then, we generated new 2D materials by chemical substitution of elements in known 2D materials by others from the same group in the periodic table. The structural, electronic and energetic properties of these 2D materials are consistently calculated, to provide a starting point for further material screening, data mining, data analysis and artificial intelligence applications. We present the details of computational methodology, data record and technical validation of our publicly available data (http://www.2dmatpedia.org/).

cond-mat.mtrl-sci↗

arXiv2015

Elucidating the Structure of the Magnesium Aluminum Chloride Complex electrolyte for Magnesium-ion batteries

We present a rigorous analysis of the Magnesium Aluminum Chloro Complex (MACC) in tetrahydrofuran (THF), one of the few electrolytes that can reversibly plate and strip Mg. We use \emph{ab initio} calculations and classical molecular dynamics simulations to interrogate the MACC electrolyte composition with the goal of addressing two urgent questions that have puzzled battery researchers: \emph{i}) the functional species of the electrolyte, and \emph{ii}) the complex equilibria regulating the MACC speciation after prolonged electrochemical cycling, a process termed as conditioning, and after prolonged inactivity, a process called aging. A general computational strategy to untangle the complex structure of electrolytes, ionic liquids and other liquid media is presented. The analysis of formation energies and grand-potential phase diagrams of Mg-Al-Cl-THF suggests that the MACC electrolyte bears a simple chemical structure with few simple constituents, namely the electro-active species MgCl$^+$ and AlCl$_4^-$ in equilibrium with MgCl$_2$ and AlCl$_3$. Knowledge of the stable species of the MACC electrolyte allows us to determine the most important equilibria occurring during electrochemical cycling. We observe that Al deposition is always preferred to Mg deposition, explaining why freshly synthesized MACC cannot operate and needs to undergo preparatory conditioning. Similarly, we suggest that aluminum displacement and depletion from the solution upon electrolyte resting (along with continuous MgCl$_2$ regeneration) represents one of the causes of electrolyte aging. Finally, we compute the NMR shifts from shielding tensors of selected molecules and ions providing fingerprints to guide future experimental investigations.

cond-mat.mtrl-sci↗