Ab Initio Modeling of Electrochemical Interfaces
LaboratoryLawrence Livermore National Laboratory (LLNL)
Capability ExpertTadashi Ogitsu, Brandon Wood, Tuan Anh Pham
ClassComputational Tools and Modeling
Node Readiness Category2: Low-Temperature Electrolysis (LTE)
1: Photoelectrochemical (PEC)
This capability provides expertise, methods, and software tools for realistic and predictive atomistic modeling of interfacial systems relevant for photoelectrochemical H2 production. It combines large-scale ab initio molecular dynamics (AIMD) for quantum-mechanical simulations of solid-liquid interface chemistry, new techniques for simulating applied external voltage or photovoltage, and many-body perturbation theory for predicting band edges and excited-state properties of photoelectrode/electrolyte interfaces. Collectively, these tools can probe interfacial chemistry and dynamics within a fully quantum-mechanical formalism, considering the full electrochemical interface complexity in a way that conventional simulations cannot. The approaches have been developed over the past several years and successfully applied to III-V PEC materials under EERE/FCTO PEC hydrogen production funding in collaboration with the PEC Hydrogen Working Group. Three software codes will be provided: (1) qball is an LLNL branch of the Qbox AIMD code that is co-developed with the parent code maintained at the new BES Midwest Integrated Center for Computational Materials (MICCoM). It is highly optimized for efficient scalability on state-of-art supercomputers, and has recently been extended to simulate electrochemical systems under (photo)bias with a potentiostat using the Effective Screening Method (ESM) method. (2) MGMol is a linear-scaling O(N) AIMD code developed at LLNL, which has recently demonstrated dynamics of over a million atoms across 1.5 million CPU cores. Though less flexible than qball, MGMol permits much larger-scale simulations, such as electrical double layer formation. (3) The WEST code is a highly scalable many-body perturbation (GW method) code developed at MICCoM, which is able to calculate the absolute positions of band edges with the accuracy of about 0.1-0.2 eV (image c), even under simulated electrochemical conditions that are difficult to probe experimentally. Together, these codes have been used to elucidate relationships between structural, (opto)electronic, and chemical properties within an operating PEC device to guide design.
The qball code can run AIMD of up to ~2000 unique atoms, and MGMol up to ~1.2M unique atoms. Both are limited to time scales of tens of picoseconds, and can therefore only be used to simulate fast processes. Slower chemical reactions can be probed with the aid of advanced sampling techniques if pathways can be elucidated using other characterization methods. The dynamics codes use the DFT/PBE-GGA level of approximation, which can lead to quantitative misprediction of chemical reaction barriers (but generally correctly predicts trends). MGMol is currently unable to simulate metallic electrode systems (although qball has no issues in this regard). The WEST code is quantitatively accurate for calculations of band edge positions, including interfacial systems, but is limited to a few hundred atoms. It can be readily combined with AIMD to give thermally averaged band positions in a liquid environment.
The experts have several years of collective experience in ab-initio modeling of interfaces for PEC H2 R&D under EERE/FCTO funding. These projects have involved close collaboration between theory and in situ experimental characterization. The protocols and methods for electrochemical interface simulation provided by this capability represent the current state of the art, including implementations that are unique even among the U.S. national laboratories, such as the ESM potentiostat for (photo)bias simulations. The software packages are maintained and supported by LLNL and by the DOE BES software center MICCoM, and are optimized to take advantage of the high-performance computing resources at LLNL and other national laboratories. LLNL maintains several state-of-the-art computational facilities, including two of the world's top ten computers, and is part of the CORAL tri-Lab alliance (with Oak Ridge and Argonne) for implementation of emerging computing technologies.
The qball and WEST codes have been well tested and are in mature stages of development. Both are being developed as open source codes (MGMol is under testing and will be released as open source in the future). The capabilities discussed here have been extensively demonstrated on III-V PEC systems as part of past and current EERE/FCTO projects, and can therefore be applied immediately without additional modification. The methods are generally resource intensive and require high-performance computing allocations; LLNL high-performance computing resources can be made available for this purpose.
This capability can provide accurate atomistic information about electrochemical interfaces, and reveal precise relationships between interface atomic structure, optoelectronic properties, and chemical properties. This computational capability can be used to inform multiscale kinetic models to predict macroscopic STH efficiency and durability. It can also operate in conjunction with advanced characterization tools such as in operando spectroscopy to develop a more comprehensive understanding about device behavior and underlying microscopic processes such as H2 evolution and corrosion for a given device component material. Such comprehensive information can be used to develop more effective device improvement strategies.
a) Snapshot of ab-initio molecular dynamics simulations of the water-InP(001) interface showing native surface hydroxylation in a PEC environment; b) snapshot of ab-initio MGMol simulations of water containing 1.2 million atoms; c) calculated band-edge positions against the vacuum level for GaP/InP (110)/(001) surfaces using many-body perturbation theory (GW method) compared with experiments, using the WEST code.
B. Wood, E. Schwegler, W.-I. Choi, and T. Ogitsu, "Hydrogen-Bond Dynamics of Water at the Interface with InP/GaP(001) and the Implications for Photoelectrochemistry," J. Am. Chem. Soc. 135, 15774 (2013).
T. A. Pham, D. Lee, E. Schwegler, G. Galli, "Interfacial Effects on the Band Edges of Functionalized Si Surfaces in Liquid Water," J. Am. Chem. Soc. 136, 17071 (2014).
B.C. Wood, T. Ogitsu, and E. Schwegler, "Ab initio modeling of water-semiconductor interfaces for photocatalytic water splitting: The role of surface oxygen and hydroxyl," J. Photon. Energy 1, 016002 (2011).