About this Event
3700 O'Hara Street, Pittsburgh, PA 15261
Title: Atomistic Simulations using Quantum Bond Order and Populations Models
Abstract: Simulations of large-scale chemical phenomena currently require accurate atomistic models. Analytical force fields (e.g., reactive force fields like ReaxFF) and semi-empirical quantum mechanics (SQM) methods have widely been used to achieve structure and energy predictions for systems too large to be treated with reliably accurate computational quantum chemistry methods, but the intrinsic empiricism in these methods limits their predictive reliability. For my PhD work, I propose to develop a new class of SQM methods that have the promise to be significant improvements over existing methods. The two methods I propose are based on using well-conditioned Hartree-Fock (HF) bond order and populations, and these are called the 1) BEBOP and 2) ZPEBOP models. BEBOP can calculate total atomization energy that is physically partitioned across chemical bonds for unique insights into structure/property relationships. ZPEBOP is capable of computing zero-point vibrational energies in molecules using a single energy calculation and avoiding a costly Hessian calculation. BEBOP currently brings the expense of an HF single-point calculation and has been shown to produce competitive results with many existing Kohn-Sham Density Functional Theory (KS-DFT) approaches. ZPEBOP currently brings a similar cost but can provide zero-point vibrational energies on par with SQM methods. Our current models have parameters for molecules containing first-row elements only and we will implement parameters for second-row elements in the future. This proposal plans to resolve the following aims:
Aim 1 will focus on developing corrections to our current BEBOP and ZPEBOP models, e.g., those that include many-body bond and van der Waals corrections. These corrections are important for obtaining reliable data for small to large molecules. We hypothesize that the addition of these terms will improve the accuracy and transferability of our model across chemical space.
Aim 2 includes the development of semi-empirical non-rigid rotor anharmonic oscillator (semi-empirical NRRAO) partition functions using the vibrational bond energies from ZPEBOP as inputs. We plan to parametrize this model to path-integral molecular dynamics data to get accurate results for thermal corrections, e.g., entropy and internal energy, at any arbitrary temperature.
Aim 3 will focus on ameliorating our simulations’ computational cost of computing structure geometry and HF bond orders and populations. We plan to generate approximate values using SQM methods and apply machine learning models to correct the differences between SQM and hybrid KS-DFT model geometries and differences in orbital population and bond orders between SQM and HF. The result will be on-the-fly BEBOP, ZPEBOP, and semi-empirical NRRAO partition function calculations for molecules containing thousands of atoms.
Chair:
Dr. John A. Keith
Department of Chemical & Petroleum Engineering, University of Pittsburgh
Committee Members:
Dr. J. Karl Johnson
Department of Chemical & Petroleum Engineering, University of Pittsburgh
Dr. Christopher E. Wilmer
Department of Chemical & Petroleum Engineering, University of Pittsburgh
Dr. Kenneth D Jordan
Department of Chemistry, University of Pittsburgh
Dr. Geoffrey R. Hutchison
Department of Chemistry, University of Pittsburgh
Please let us know if you require an accommodation in order to participate in this event. Accommodations may include live captioning, ASL interpreters, and/or captioned media and accessible documents from recorded events. At least 5 days in advance is recommended.