Prof. Zlatan Aksamija
Institution: University of Massachusetts Amherst
Department: Electrical and Computer Engineering
Proposed research ideas:
I am primarily interested in ab initio calculations of materials properties based on density functional theory (DFT). DFT has been successfully used to obtain electronic and vibrational structure of many semiconductors and 2-dimensional materials (ex. graphene, MoS2, etc.) A secondary interest is in classical simulation of heat propagation via the Molecular Dynamics method. Both DFT and MD are computationally expensive, especially for complex materials with large unit cells, or those with ample disorder such as alloys. One of our present limitations in studying alloys from first principles is the large computational cost of DFT and DFPT calculations on large supercells–in order to accurately capture the electronic structure and vibrational spectra of realistic semiconductor and 2D alloys, a large simulation supercell containing many unit cells and hundreds of atoms is required. Same is true when studying interfaces where a large expanded supercell is needed to avoid artificial periodic effects. The resulting computational cost of many-hundred-atom first principles calculations could be reduced through parallelization and deployment on an HPC platform such as NERSC. This is an opportunity to collaborate on a study of vibrational spectra and thermal conductivity in alloys, especially 2D alloys of TMDs (ex. Mo_x W_1-x S2) and thermal resistivity of their interfaces. Alloy composition can be used to “tune” or alter the vibrational modes of a material (generally, introducing a heavier element into an alloy lowers the vibrational frequencies, scatters phonons, and reduces thermal conductivity). At the same time, alloy composition can be used to create materials with intermediate vibrational properties to “bridge” an interface between otherwise mismatched materials with incompatible vibrational spectra. Thus alloys have potential as both thermal barriers or thermoelectric converters (requiring low thermal conductivity) and interface materials (requiring tuneable vibrational spectra).
I would like to forge stronger links to DOE labs in the future. My research is predominantly computational materials for energy applications; consequently, the best collaboration opportunities, and the best HPC hardware, are all within DOE national labs. While a graduate student at the University of Illinois Urbana/Champaign, I was funded by the DOE through the Computational Science Graduate Fellowship (CSGF). The CSGF program exposed me to the national lab system through summer practicums (2007 at Los Alamos National Lab with Bobby Philip in T-7 see http://math.lanl.gov/~aksamija/ and 2008 at Argonne National Lab with Paul Fischer in MSC). Since moving to academia, I’ve been looking for opportunities such as this one to rekindle that collaborative spark and introduce my own graduate students to the broad palette of world-class research opportunities at the National Labs.