Beyond-DFT Simulation of Energetic Barriers and Photoexcited Dynamics
LaboratoryLawrence Livermore National Laboratory (LLNL)
Capability ExpertMiguel Morales-Silva, Alfredo Correa Tedesco, Tadashi Ogitsu
ClassComputational Tools and Modeling
Node Readiness Category2: Photoelectrochemical (PEC)
This capability describes a suite of computational tools that offer unprecedented accuracy for describing energetics of reactions and photoexcited electron dynamics in photoabsorber materials. The techniques are intended to address issues associated with the level of approximation in describing ground and excited states with more conventional density functional theory (DFT) simulations. For highly accurate calculations of adsorption, reaction, and transport barriers, a simulation capability based on Quantum Monte Carlo (QMC) will be provided. QMC is a powerful stochastic quantum-mechanical technique for determining electronic ground-state energies, and is currently the most accurate atomistic simulation method that can be applied to extended systems such as solids and liquids. The LLNL team co-develops the QMC code QMCPACK under DOE BES and NSF funding. The extreme computational expense of QMC has traditionally limited its application; however, recent advances in QMCPACK algorithms have optimized performance for excellent scalability on high-performance supercomputers. As a result, QMC can now be applied to more complex systems such as those relevant to PEC, as tested on LLNL supercomputers. Examples of cases where QMC could be especially useful include more accurate catalyst screening based on intermediate absorption energetics (Sabatier's principle), diffusion barrier estimates, and reaction barriers for corrosion chemistry. For modeling of photoexcited electron dynamics in photoelectrodes, LLNL also maintains a simulation capability based on Time-dependent DFT Molecular Dynamics (TDDFT-MD) within the Ehrenfest formalism, as implemented by our team in the Qbox/qb@ll code. This is a state-of-the-art method that can be directly applied to investigate recombination rates and help understanding about how photoactive defect states affect the photoabsorber chemistry and stability. Such information could be used to determine effective strategies for optimizing materials synthesis processes.
On currently available DOE supercomputer facilities, QMCPACK can perform energetics simulations of a system containing up to several hundred unique atoms. Qbox/qb@ll can perform TDDFT-MD simulations of a system containing up to a few thousand atoms, limited to ultrafast processes that occur within a few hundred femtoseconds. The TDDFT-MD implementation has mostly been applied to other application spaces such as radiation damage assessment2 and may require some modification for full application to PEC systems. Both techniques are very resource intensive and require extreme-scale computing.
The experts have extensive collective experience in extreme-scale computing, optimization, and application of both QMC and TDDFT-MD to problems of similar complexity. The TDDFT-MD code is maintained at LLNL. QMCPACK is a focus of a newly minted DOE BES Computational Materials Science Software Center, with LLNL as the key contributor to code development and functionality. The experts have more than 10 years of experience using and developing these methods, including specific experience with FCTO projects for PEC hydrogen production and hydrogen storage. Team members have also collaborated on similar large-scale theory-experiment collaborations on excited electron dynamics as part of LLNL investments in ultrafast science. The codes are optimized and have been tested extensively on the high-performance computing facilities at LLNL.
QMCPACK is a public, open-source code that can be applied immediately. The TDDFT-MD implementation is currently not in the public release version of Qbox/qb@ll, but has been tested extensively at LLNL. Some modifications may be required to adapt the TDDFT-MD implementation to PEC H2 research. Both techniques require expert assistance and high-performance computing allocations. Computing facilities at LLNL can be made available to HydroGen projects. However, larger projects necessitating more extensive CPU time will likely require an additional separate allocation through DOE/ASCR, which can be granted via a secondary review process.
Accurate description of ground state and excited state will enhance the level of confidence of the other types of computational capabilities, and also assist in improving robustness of interpretation of experimental results. This capability could play a critical role in mitigating uncertainties in other simulation capabilities. It would also be useful for assessing the role of photoactive defects in dissipative processes detrimental to photoabsorber performance, such as carrier recombination and photocorrosion.
Snapshots of electronic density change (electron wake) produced by H+ moving in Cu with a kinetic energy of 81 keV along a  channel. The brown balls represents the Cu atoms and the single gray ball with light-blue iso-density contours represents the host time-dependent electron density perturbation affected by the presence of H+. Snapshots show the evolution from initial conditions to a representative steady condition after t = 91.64 attoseconds.
1. M.A. Morales, C. Pierleoni, E. Schwegler, D.M. Ceperley, "Evidence for First-Order Liquid-Liquid Phase transition in High Pressure Hydrogen from Ab-Initio Simulations", PNAS 107, 12799 (2012).
2. A.A. Correa, J. Kohanoff, E. Artacho, D. Sanchez-Portal, A. Caro, "Nonadiabatic Forces in Ion-Solid Interactions: The initial Stages of Radiation Damage", Phys. Rev. Lett. 108, 213201 (2012).
3. B.I. Cho, K. Engelhorn, A.A. Correa, T. Ogitsu, C.P. Weber, H.J. Lee, J. Feng. P.A. Ni, Y. Ping, A.J. Nelson, D. Prendergast, R.W. Lee, R.W. Falcone, P.A. Heimann, "Electronic Structure of Warm Dense Copper Studied by Ultrafast X-ray Absorption Spectroscopy", Phys. Rev. Lett. 106, 167601 (2011).