Advancing Orbital Free Density Functional Theory: Physics, Algorithms, and Applications
Abstract
Abstract The orbital-free density functional theory (OFDFT) method for predicting electron distributions in matter and their attendant material properties has entered an exciting stage of development. In the last four years, algorithmic advances and physics breakthroughs have positioned OFDFT to make important, reliable predictions of material properties as a function of composition and microstructure not just for main group metals, but also for semiconductors, transition metals, molecules, and liquids in the years ahead. Recent major numerical advances include upgrades to a fully linear scaling and parallelized code that can treat unprecedented numbers of atoms quantum mechanically and partial implementation of a new algorithm, called the small box Fast Fourier transform (SBFFT), which takes advantage of the massive parallelism of modern supercomputers. Recent physics breakthroughs produced three improved descriptions of the kinetic energy of electrons solely from the electron density via kinetic energy density functionals (KEDFs), for covalently bonded materials and coinage metals. Also a new framework was developed and validated: angular-momentum-dependent OFDFT (AMD-OFDFT) for accurately treating transition metals. Adding these new models to our arsenal greatly increased the range of elements that can be accurately treated with OFDFT. Before these improvements during the last grant, only nearly-free-electron-like metals and some properties of semiconductors could be reliably simulated respectively with two nonlocal KEDFs we derived from linear response theory 15 years ago (the WGC99 KEDF) and four years ago (the HC10 KEDF). Ample evidence of OFDFT’s ability to accurately simulate large scale features and phenomena in, e.g., aluminum (Al) and magnesium (Mg), as well as alloys and liquids was also demonstrated during the last grant period. Our current objective is to build on these recent successes to work toward full generality and even greater computational efficiency of OFDFT within our open-source code PROFESS. In particular, we plan to extend OFDFT by continued expansion of our arsenal of KEDFs based on our density decomposition theme. This includes combining our newly derived single-point KEDFs to describe localized densities coupled with the WGC99 KEDF to describe interstitial densities. By so doing, we aim to treat heterogeneous systems such as atomic and molecular interactions with solid and liquid metals. We also intend to automate optimization of the nonlocal energy parameters in AMD-OFDFT, so as to extend its reach first to all transition metals and eventually to the entire periodic table. We propose to introduce spin-dependent local (electron-ion) pseudopotentials (LPSs), as well as a more complete atomic basis set in AMD-OFDFT to further improve its accuracy. We will also develop an AMD-OFDFT+U theory for accurate treatment of transition and rare earth metal cations in ionic materials. We will continue implementing SBFFT throughout the heavily-FFT-dependent PROFESS code, with the aim to do OFDFT molecular dynamics (OFDFT-MD) simulations with OFDFT forces on samples containing up to a million atoms or to simulate smaller samples over much longer time scales. We envision applying OFDFT to many phenomena in the years ahead but only detail two examples here: one that can proceed in conjunction with the research proposed above and one that will be undertaken as soon as the ideas above are implemented and validated. The former example seeks to understand the mechanism of retarded natural aging in Al-Mg-Si alloys, while the latter aims to simulate the full stress corrosion cracking phenomenon of Al alloys immersed in seawater subjected to tensile loads.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 12, 2016
- Source ID
- N000141512218
Entities
People
- Emily A. Carter
Organizations
- Office of Naval Research
- Trustees of Princeton University
- United States Navy