Transition metal oxides are challenging to model from first principles, due to the substantial presence of electron correlations. Most conventional first-principles methods treat electron correlation in an approximate manner. This approximate treatment sometimes results in qualitatively incorrect predictions about the physics of a material. Manganese oxide is one such example: although the rock salt structure is energetically stable, most conventional electronic structure methods predict the zinc blende phase to be lower in energy. For challenging problems such as these, we are interested in developing the framework for applying highly-accurate quantum Monte Carlo methods to the description of real material properties. Quantum Monte Carlo methods are well-established in physics research, but their application to real engineering materials is a new, emerging area. Additionally, due to its direct treatement of electron correlation within a many-body framework, QMC offers the possibility of parameter-free and systematically improvable analysis. For manganese oxide we find that, when carried out properly, QMC can obtain accurate descriptions of the two phases, including their relative energies and band gaps.
Even at small concentrations, the presence of point defects can greatly modify the properties of a host semiconductor, especially in photovoltaics. Because point defects dominate optical and electronic properties, accurate modeling is crucial to understanding and designing materials. Due to the simultaneous presence of localized defect states together with semiconductor host states, point defects present a challenge for first-principles modeling (most techniques are unable to treat both on equal footing). To add another tool to the computational toolbox, we are establishing the use of quantum Monte Carlo methods for quantitative descriptions of the properties of point defects in semiconductors. The QMC approach secures high accuracy through the use of stochastic approaches to solve (to a large extent) the many-body interacting Schrödinger equation. Historically, large computational costs have hampered the use of QMC for predicting material properties. With on-going enhancements of computational power and numerical algorithms, it is now becoming viable to use QMC approach for real materials analysis. We are interested in the calculation of fundamental properties such as defect formation energies, thermal transition levels, and optical ionization energies. Additionally, we are interested in developing diagnostic assessment methods that help to reveal the physics that may be missing from other modeling approaches. Some current topics of interest include doping in zinc oxide, DX center formation in zinc selenide, and point defects in thin-film photovoltaic materials such as chalcopyrites and CIGS.
Magnetocaloric shape memory alloys present an intriguing solid-state alternative to conventional approaches to cooling and refrigeration. Magnetic cooling is potentially an environmentally green, energy-efficient technology capable of outperforming conventional gas-compression refrigeration. The discovery of the near-room-temperature giant magnetocaloric effect over fifteen years ago expanded the scope of magnetic cooling from cryogenic to standard room temperature refrigeration. In the magnetocaloric effect, the application or removal of a magnetic field induces a structural phase transformation, and the latent heat of the phase transition induces a temperature change in the material (adiabatic demagnetization). Physically, phonons are emitted or absorbed as the atoms rearrange during the phase transformation. To optimize the effect, the entropy change of the phase transition must be as large as possible. In collaboration with Prof. Huseyin Sehitoglu's group, we are focusing on the identification and optimization of ferromagnetic shape memory alloys exhibiting high magnetocaloric cooling capacity. We have developed a framework that couples first-principles calculations magnetic exchange interactions together with statistical mechanics models to predict the properties of the magnetic Heusler shape memory alloys. We have validated our approach on two known materials, and now are extending it to as-of-yet unknown material compositions.
In collaboration with Prof. Lane Martin's group at UC Berkeley, and Prof. Ed Seebauer and Angus Rockett, we are interested in the design of heterojunctions between TiO2 and correlated metal oxides for photocatalysis. One such example is SrRuO3/TiO2, in which the correlated metal SrRuO can help to absorb parts of the solar spectrum that are not absorbed by TiO2. Correlated metal oxides such as SRO often exhibit interesting characteristics such as metal-like conductivity, and semiconductor-like optical properties. We are modeling the electronic structure of these heterojunctions and helping guide the experimental work to design optimized systems.
MoS2 is an exciting material with many potential applications. Much like the material graphene, MoS2 can be made in a single layer. A very thin amount of MoS2 can absorb relatively large amount of incident light. The next generation of solar devices are going to be very thin, and MoS2 is a perfect candidate material for these devices. MoS2 may also be used in transistors of the future. Field effect transistors (FET) control the flow of electrons using an applied electric field. Compared to transistors used today, MoS2 FETs would be more efficient in terms of power consumption.