Electronic-Structure Modeling for Atomistic Understanding of Catalytic Materials with Real-World Distributions of Facets and Defects


National Renewable Energy Laboratory (NREL)

Capability Expert

Mai-Anh Ha, Ross Larsen


Computational Tools and Modeling

Node Readiness Category

2: Low-Temperature Electrolysis (LTE)


We apply ground-state density functional theory methods to understand the role of heterogeneity (e.g., facets or differential alloying) in determining catalytic properties of surfaces, with the goal of using in silico predictions to understand electrolysis with materials produced under different processing conditions. The focus of this approach is not on large-scale simulation of interfaces, but rather to perform detailed simulations and bonding analysis of materials for catalysis, taking into account the role of defects and the different activities of distinct crystal facets. For promising catalysts such as extended surfaces, the actual interface may be composed of a distribution of surfaces with a range of facets, lattice constants and layers of metal, alloys, and/or metal oxides. The catalytic activity of the full distribution must be accounted for to understand the catalyst system’s stability, activity, and selectivity for electrolysis. Hence, analysis of the activity and selectivity of a catalyst requires an investigation of reaction intermediates as an ensemble of surfaces and adsorbates participating in the reaction. Hence, our DFT-based work results in an atomic and electronic-level understanding of how band structure and bonding interact with the morphology and surface structure of nano-catalysts to determine their performance. This allows us to make predictions for materials with real-world complexity by combining the following approaches:

  1. Characterize realistic surfaces by considering varying lattice constants, defects, facets, and phases of these catalysts (through experimental characterization or computational parameter scans). Cohesive energies, chemical bonding analysis, and adsorption of intermediates indicate stability and reactivity of these supports, providing an in vacuo understanding of electrolysis. Where appropriate, amorphous phases can be modeled through simulated annealing of surfaces utilizing classical molecular dynamics.
  2. Approximate the electric double layer with implicit or explicit solvation. Implicit solvation accounts for the liquid-catalyst interface and explicit solvation through the addition of one or more H2O molecules and appropriate ions.

Capability Bounds‎

Our DFT methods consider surfaces of up to 400 atoms, with large unit cells chosen to a surface to minimize possible spurious interactions between periodic images of multiple adsorbates. For a single system, such as Pt on Pt-Ni and Ni-O, we have successfully modeled circa 150 surfaces, considering all facets and relevant lattice constants. For each surface of interest, we do a sampling of 20-50 unique adsorbate sites. The relevant ensemble of adsorbates is then calculated, using appropriate Boltzmann weighting with in-situ considerations (solvation, electric field) to generate a reaction profile of the catalytic system. Additional considerations of defects or crystalline or amorphous phases require even more permutations of layered surfaces and subsequent sampling of adsorbate sites. We tailor this process to each catalytic system with input from experimentalists to ascertain the most relevant and interesting distribution of interfaces and their in-situ conditions during material processing and operation.

Unique Aspects‎

Our group’s consideration of the complex interface and ensemble reaction intermediates has led to a unique perspective of heterogeneous catalysis. Namely, our theoretical work has complemented and informed or prompted the development of novel catalysts. Recent work on catalysts at the sub-nano (clusters) and nano (extended surfaces) regime includes:

  • Coverage effects: In collaboration with experiment, we identified the highly active Pt7 cluster for ethylene dehydrogenation, with each negatively charged cluster able to adsorb and activate a maximum of 3 ethylenes.1 However, successive de(hydrogenation) results in deactivation as catalyst sites are blocked by carbon (coking). We tempered that high activity and predilection towards coke formation by doping Pt7 with the electropositive boron, which resulted in sustained activity during successive reaction cycles by adsorbing and activating circa 1 ethylene.2
  • Ensemble of interfaces (clusters or surfaces): At the sub-nano scale, Pt7’s high activity over that of a similar cluster size, Pt8, resulted from Pt7’s fluxionality, the ensemble is able to access isomer geometries with more exposed Pt sites. At the nano-scale, NREL’s experimental work on extended surface catalysts of Pt-Ni converged on the same lattice constant and facet distribution as our computational work of the layered interface (see figure below).3
  • Coverage effects, co-adsorption of relevant reaction intermediates, and reaction profiles with thermodynamic descriptors of reaction barriers (Gibbs Free energies and enthalpies) provide a gauge for activity with respect to surface morphology and in-situ considerations.
  • Other considerations with respect to defect surfaces: Projected density of states (PDOS) and charge density plots also reveal the localization of electrons at neighboring atoms due to defects such as metal interstitials and oxygen vacancies.4


Utilization of DOE-EERE sponsored high performance computing resources (Peregrine at NREL) with installed software (VASP, LAMMPS).


This work represents a concerted effort to match theoretical research in heterogeneous catalysis to the complex rigors present in manufacturing and real-world devices through high-performance computing. Our theoretical consideration of the complex interface and ensemble adsorbates has been validated with experimental collaborators. Namely, we have predicted physicochemical properties related to stability, activity, and selectivity by considering the distribution of surfaces and the ensemble of adsorbates available for catalysis.


Figure 1. (a) XRD patterns of Pt—Ni nanowires (7.3 ± 0.3 %wt Pt), as-synthesized and annealed in hydrogen. (b) Pt lattice constants (by Rietveld refinement of XRD patterns) and Pt facet data, as determined by germanium and tellurium underpotential deposition. (c) Surface models of Pt skins on the Ni3Pt alloy. The alloying effect of Ni3Pt was more pronounced than that of Pt3Ni as the Ni-enriched alloy particularly stabilizes both (100) ∼ (111) over (110) with a compressed lattice constant ca. 3.77 Å (at the experimental high performer).3


  1. Baxter, E. T.; Ha, M.; Cass, A. C.; Alexandrova, A. N.; Anderson, S. L. Ethylene Dehydrogenation on Pt4,7,8 clusters on Al2O3: Strong Cluster-Size Dependence Linked to Preferred Catalyst Morphologies. ACS Catalysis 2017, 7, 3322-3335.
  2. Ha, M.; Baxter, E. T.; Cass, A.; Anderson, S. L.; Alexandrova, A. N. Boron Switch for Selectivity of Catalytic Dehydrogenation on Size-Selected Pt clusters on Al2O3. J. Am. Chem. Soc.  2017, 133, 11568-11575.
  3. Alia, S. M.; Ngo, C.; Shulda, S.; Ha, M.; Dameron, A. A.; Weker, J. N.; Neyerlin, K. C.; Kocha, S. S.; Pylypenko, S.; Pivovar, B. S. Exceptional Oxygen Reduction Reaction Activity and Durability of Platinum–Nickel Nanowires through Synthesis and Post-Treatment Optimization. ACS Omega 2017, 2, 1408-1418.