Acquistion of Computer Cluster for First-Principles Predictions of Structure and Properties of Cocrystals
Abstract
The proposal requests funds for acquisition of a computer cluster for calculations of intermolecular force fields needed for predicting crystal structures and properties of energetic, ferroelectric, and nonlinear optical materials. Emphasis will be on crystals built from more than one monomer, called cocrystals. This research is the subject of an ongoing ARO MURI project. The main role of our group in this project is to per- form first-principle computational predictions of structure and properties of notional materials prior to their cocrystallization. Predictions of this type are even more impor- tant for cocrystals than for pure crystals since in the former case the main issue is the very existence of a given hypothetical structure. Theory can also determine properties that are difficult to measure and give insights into the microscopic molecular processes involved. The development of first-principles forcefields is necessary for the project since un- certainties of the currently used empirical force filds|obtained by fitting parameters of the fields to reproduce experimental data on model systems|are the major factor limiting the reliability of crystal structure predictions even for pure crystals. The goal of the proposed research is to create much more accurate fields, fitted to results of ab initio quantum mechanical calculations. This novel approach has been made possible by our development of an accurate method called SAPT(DFT) [symmetry-adapted perturbation theory based on density-functional theory (DFT) description of isolated monomers]. The SAPT(DFT) method has been shown to correctly predict structure and properties of crystals such as RDX or FOX-7. Developments of force fields require significant computational resources. Since access to DoD supercomputer centers became virtually impossible for foreign nationals, a reasonably powerful local computer cluster is needed for realizing the goals of the MURI project. The proposed cluster will have 900 Intel Xeon cores in 45 nodes, each node containing 128 Gbytes of memory and about 2 Tbytes of disk space.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 14, 2019
- Source ID
- W911NF1610167
Entities
People
- Krzysztof Szalewicz
Organizations
- Army Contracting Command
- United States Army
- University of Delaware