Identifying Potential BO2 Oxide Polymorphs for Epitaxial Growth Candidates
Prateek Mehta, Paul A. Salvador and John R. Kitchin, “Identifying Potential BO2 Oxide Polymorphs for Epitaxial Growth Candidates”, ACS Applied Materials & Interfaces 6(5), 3630-3639 (2014). https://doi.org/10.1021/am4059149
This summary was written by a language model from the paper and has not been checked line by line. Rely on the published paper for details.
Question
Metastable oxide polymorphs can have properties superior to the ground state (anatase TiO2 for photocatalysis, for example), and growing them as thin films on a suitable substrate can reorder polymorph stabilities through interfacial energies. Choosing which composition, polymorph and substrate to try is still done largely by trial and error, even with high-throughput approaches such as combinatorial substrate epitaxy (CSE). The paper asks which polymorphs of common dioxides are energetically close enough to the ground state to be realistic targets for epitaxial stabilization, and how much the answer depends on the exchange-correlation functional (Mehta et al. 2014).
Approach
The authors computed equations of state and relative energies for six polymorphs (anatase, brookite, rutile, columbite, pyrite and fluorite) of five dioxides (TiO2, VO2, RuO2, IrO2 and SnO2) with VASP and PAW potentials, using four functionals in a consistent way: LDA, PBE, PBEsol and AM05. A 520 eV cutoff and per-structure k-point grids converged to 10 meV per formula unit were used. Energies at a series of fixed volumes were fit to the Birch-Murnaghan equation of state to get equilibrium volumes and bulk moduli, followed by a final full relaxation, and stabilities were compared at each polymorph’s own equilibrium volume. To decide what energy difference epitaxy could plausibly overcome, they used a simple thin-film free energy model in which the bulk volumetric energy difference is balanced by the film/substrate interfacial energy difference. With a formula-unit volume of about 30 Å3, a nucleus about 1 nm thick, and interfacial energies of about 1 J/m2 (incoherent, stable phase) versus about 0 (coherent, metastable phase), this gives roughly 18 kJ/mol, so they adopt 10 to 20 kJ/mol of 0 K DFT energy as a target window, consistent with the roughly 9 kJ/mol difference overcome for hexagonal SmMnO3 on an isostructural substrate. All structures, energies and parameters were released as JSON data in the Supporting Information (Mehta et al. 2014).
Key findings
Rutile was the lowest energy polymorph for every oxide except TiO2. Volume per formula unit generally followed fluorite < pyrite < columbite < rutile < brookite < anatase, and bulk modulus fell roughly linearly with volume, with anatase and brookite VO2 as outliers. LDA underestimated and PBE overestimated volumes, while PBEsol and AM05 were similar to each other and closest to experiment (for example 31.22 Å3 for rutile TiO2 with PBEsol versus about 31.2 measured). The functional affected relative energies most when the polymorphs differed strongly in volume, so the spread across functionals can be used to bracket the energy differences.
For TiO2, anatase, brookite, rutile and columbite all lay within 8 kJ/mol of each other, with GGA functionals placing anatase below rutile, so columbite and brookite were predicted to be epitaxially accessible. For VO2, columbite and brookite lay about 6 to 8 kJ/mol above rutile and anatase about 12 to 16 kJ/mol (with poorly fit equations of state for some functionals); the computed anatase b lattice constant (3.72 Å LDA, 3.75 Å PBEsol) matched pseudomorphic VO2 layers reported on anatase TiO2. For RuO2, pyrite was about 30 kJ/mol more stable than fluorite, columbite sat about 13 kJ/mol above rutile for all functionals, pyrite 5 to 20 kJ/mol depending on functional, and brookite and anatase above 30 and 50 kJ/mol. Since pyrite is below columbite for most functionals, this explains why rutile RuO2 transforms to pyrite under pressure without passing through columbite, so columbite RuO2 is metastable at all pressures and can only be reached by epitaxy. IrO2 followed RuO2 but with rutile more strongly favored: columbite was more than 20 kJ/mol and pyrite more than 18 kJ/mol above rutile. For SnO2, columbite was only about 3 to 5 kJ/mol above rutile, consistent with its known epitaxial films, while brookite, pyrite and anatase lay about 20 to 30 kJ/mol above. The resulting prime epitaxial candidates were columbite and brookite VO2 and pyrite and columbite RuO2, with pyrite and anatase IrO2 and brookite and anatase SnO2 near the edge of the window. Some high-volume targets (anatase SnO2, brookite VO2 and SnO2) cannot be made by compression at all, while pyrite RuO2 and IrO2, normally high-pressure phases, may also be reachable by epitaxy (Mehta et al. 2014).
Significance
The work turns polymorph screening into a guide for thin-film synthesis: a consistent multi-functional DFT library plus a back-of-the-envelope interfacial energy argument yields a ranked list of experimentally testable targets and suggests isostructural substrates (for example columbite under-layers for RuO2). The authors note the limits of using 0 K internal energies in place of synthesis-temperature free energies, and that strain, interfacial and surface energies must be computed for a complete picture (Mehta et al. 2014).