Phase Stability and Intercalant Ordering in V2O5 Cathodes

Summary

We will use cluster expansions, constructed based on density functional theory (DFT) calculations, to investigate the phase diagram upon intercalation of V2O5, a potential cathode material for non-Li-ion batteries, i.e., batteries that use ions such as Na, Mg, Ca, etc. Cluster expansions are required to be able to run large-scale Monte Carlo simulations. These will be used to introduce temperature to the DFT results, so that phase diagrams for different polymorphs and intercalant concentrations can be obtained. This will also allow for voltage profiles to be calculated.

Job Description

The student will create python scripts using the Atomic Simulation Environment (ASE) framework and the CLuster Expansion in Atomic Simulation Environment (CLEASE) python package to construct cluster expansions for different polymorphs of V2O5 and for different intercalated ions. These expansions will be fitted using machine-learning approaches, and will be based on density functional theory (DFT) calculations, as implemented in the VASP package. These calculations will be parallelized using MPI (both the VASP calculations, but also the Monte Carlo simulations). Results will be stored in databases. The student will interact with high-performance computing resources, and will write code to perform all steps: underlying DFT calculations, fitting of cluster expansions using machine-learning approaches, such as compressed sensing, creating Monte Carlo simulations based on the cluster expansions, and scripts to plot and analyze results.

Computational Resources

The DFT calculations will be performed on XSEDE (in particular on PSC Bridges-2), and will be controlled by scripts written in python (via ASE). Such calculations will typically require 1 node (128 cores on Bridges-2) for 4-6 hours. Fitting of the cluster expansions will be done on the studentâ€™s computer and/or the local cluster, and subsequent Monte Carlo simulations will be done on XSEDE machines/local cluster.

Contribution to Community

This project will train an undergraduate student in the use of HPC resources to perform computational research on materials properties using first-principles codes. This will increase the studentâ€™s knowledge and proficiency in HPC computing and the underlying physics and material science, which are valuable skills. At the end of the semester the student will present her results during a poster session organized by the office for undergraduate research at the University of Kansas. The project scientific goals might lead to better understanding of, and potential improved, battery cathodes for novel non-Li-ion batteries.

Position Type

Intern

Training Plan

The student will continue to refine (and apply) knowledge of python coding, further increase background knowledge of solid state physics, thermodynamics, and quantum mechanics, basic linux shell (cd, mkdir, cp, ls, etc.), and interacting with HPC resources (ssh, scp, SLURM, etc.). This learning will occur through 1-on-1 mentoring with the PI. The student will study the intercalation of V2O5 cathodes based on guiding questions, and prior knowledge of both DFT codes and the cluster expansion methodology. The more experienced the student becomes, the more open questions and tasks will become. This will introduce the student gradually to more aspects of the scientific process (making hypothesis, designing calculations to test these, analyze results, refine hypothesis, and so on). The student will also take part in regular group meetings to learn about other group members' research and discussions about that research. At the end of the semester the student will present her research during a poster session organized by the office for undergraduate research at the University of Kansas.

Student Prerequisites/Conditions/Qualifications

This is a continuation project, and as such the student already possesses all required background (basic solid state physics, thermodynamics, and quantum mechanics) and proficiency with Python and HPC resources, as well as basic understanding of the underlying DFT code. Currently the student is applying knowledge obtained from a cluster expansion fitted using a simple classical pair potential (Summer 2021) to extend this research to use DFT as underlying calculator. This also requires basic understanding of running DFT calculations, as oftentimes convergence is not as straight-forward. The student is developing code to take into account common DFT errors and crashed calculations. A simple binary metal is used as model system, so that the calculations themselves are relatively quick. By end of the fall semester the student will have acquired all knowledge to be able to tackle the more complicated V2O5 battery cathode system.