Constant-stress Quantum Monte Carlo for High-pressure Hydrogen

Brenda Rubenstein, Columbia University

Photo of Brenda Rubenstein

Despite being the first element of the periodic table and therefore, in theory, the simplest, hydrogen and its high-pressure phase diagram remain shrouded in mystery. In 1935, Wigner and Huntington predicted that normally insulating hydrogen should become metallic at sufficiently high pressures. More recent predictions suggest that metallic hydrogen also may be a room-temperature superconductor with the highest Tc of any known material. These suppositions have galvanized generations of scientists to search for these elusive phases. Nevertheless, experiments seeking metallic hydrogen have been overwhelmingly unsuccessful. Indeed, hydrogen's quantum character and highly reactive nature have posed severe challenges to experiments, leaving the majority of hydrogen's high-pressure phase diagram undetermined. Given this state of affairs, simulation is uniquely positioned to resolve many of the fundamental questions about high-pressure hydrogen. In this talk, I will discuss our recent work at LLNL developing a constant-stress quantum Monte Carlo (QMC) algorithm well-suited for simulations of the high-pressure hydrogen phase diagram. A number of previous studies have used density functional theory (DFT) to explore high-pressure hydrogen, particularly at high temperatures. However, because DFT is fundamentally a mean-field theory incapable of accurately accounting for strong electron correlation and zero-point motions, its answers differ dramatically from parameterization to parameterization and correction to correction. In contrast, our technique can, in principle, treat electron correlation exactly and is capable of predicting lattice structures without any previous information, a feature crucial for identifying previously unknown phases. I will outline the details of our new algorithm and illustrate how it can be used to identify the lattice structures of unknown phases. I will conclude with a discussion of our algorithm's predictions for previously uncharacterized Phase-III hydrogen. It is our hope that our constant-stress algorithm will become a vehicle for answering many of the ages-old questions about hydrogen in the future.

Abstract Author(s): Brenda M. Rubenstein,1 Jonathan DuBois,2 and Berni J. Alder21Department of Chemistry, Columbia University2Quantum Simulations Group, Lawrence Livermore National Laboratory