email

fosterd@
   physics.oregonstate.edu

links

Oregon State University
Physics Department
Guenter Schneider Group


Computational Materials Science Research

I switched fields when I became a postdoc at Oregon State University. I am performing various types of electronic structure calculations using variants of density functional theory. I primarily use the commercial PAW code VASP which I occasionally have altered and recompiled (for example, removing the second effective smearing step of the imaginary part of ε helps remove an unphysical tail in absorption). I have also used the all electron (FLAPW) code flair which has been coauthored by my supervisor Guenter Schneider. Electronic structure, both many-body theory and density functional theory, are fascinating branches of mathematical physics.

Hybrid (HSE) and GGA+U calculations on potential solar absorber Cu3PSe4 and other materials

The potential solar absorber material Cu3PSe4 has an experimental and calculated direct band gap of 1.4 eV, an "excellent start" for a candidate photovoltaic (solar cell) light absorbing material. From a computational standpoint, the material is seems particularly suited for treatment with a hybrid density functional known as HSE06, which gives unexpectedly accurate results for both bond lengths and band gap. Optical HSE calculations show absorption similar to GaAs.
charge density showing P-Se antibonding
The figure above shows the charge density of the single particle state at the CBM. The P-Se antibonding character of this state means its energy level is sensitive to the P-Se bond length. Relaxation of the structure and lattice in LDA, GGA, or GGA+U results in the P-Se bond lengths being to long; when a more accurate method such as HSE or GW is used to determine the band gap without altering the structure, the result can differ by as much as -0.5 eV from the result obtained with the experimental structure. Relaxing atomic positions with HSE, however, results in almost no change in bond length or band gap.

Calculations of Point Defects

Point defects calculations not only estimate the energy and concentration of atomic vacancies, interstitial atoms, and atomic site substitutions, but they also all estimate the intrinsic electronic and/or hole doping. Plane wave based calculations of the energy of an isolated, possibly charged defect are made more accurate by the use of several electrostatic and band-filling corrections. Use of the GGA+U method statistically increases accuracy of both total energies (leading to heat of formation of defects) and the valence band position. The latter also affects heat of formation, and possibly whether a defect state lies in the bandgap and is very localized in space, or whether it lies inside the valence band, forming a hydrogen-like state that is only localized on the scale of tens of nanometers. A GGA+U+G0W0 calculation gives an even more accurate determination of the valence band maximum.

Interface Calculations

The picture at the top of my site is the charge distribution of the highest occupied (valence) band (averaged over k-points) of a Si/ZnS (111) supercell with two particular substitution-compensated interfaces. The calculation is performed with periodic boundary conditions, so the modeled structure repeats in all directions, forming a superlattice. A small amount of charge transfer is visible from the center interface, which is a locally compensated donor interace to the other interface, which is an acceptor. The substituted atom at each interface lowers the interface energy, reduces interface charge, and modifies the valence (and conduction) band offsets.

Semiconductor modeling with Domain Decomposition