Atomistic Simulations at Micro Time and Length
Scales: Implications and Applications

VIKAS TOMAR
Aeronautics and Astronautics
Purdue University

Abstract:
Repeatedly, experiments have shown that nanomaterials with primary length scale of a few micrometers and nanodevices with
component dimensions ranging from few hundreds of nanometers to micrometers can be used to exploit various advantages in
current nano-bio-energy applications. Mechanics modeling has lagged behind in supporting such experiments and predicting
future experimental outcomes because (1) continuum models cannot effectively capture complex material behavior such as
fracture as a function of dimension and morphology under varied boundary conditions below micrometers and (2) molecular
models have been so far only successful in providing a qualitative non-verified insights by mutliscaling upto a few hundreds of
nanometers using approximate models and quantitative insights for very small subset of requirements such as energetics of
vacancies or grain boundaries in metals and similar defect issues in other materials such as biological materials etc. Use of
supercomputers could increase the length scale but not the time scale of molecular simulations. Molecular simulations and
modeling cannot be circumvented because they are a powerful scheme to produce correct solutions (if they could be verified
with experiments) in multiple domains from a single simple framework unlike continuum models that need fitting in order to
determine constants that change with changing situations. Therefore, direct molecular simulations that explicitly resolve
femtosecond to picosecond material behavior at micrometer dimensions and time for complex composite materials are needed.
Only then the real power of molecular modeling will be realized.

This talk will present one such framework that we have developed along with initial results and outcomes. In this method,
dynamic equivalent crystal lattices are generated to represent a classical analogue of the statistical mechanical description of
the underlying material. During this care is taken that the ergodicity of the system is not compromised. By using the method,
we found that the atomistic system size and time step of molecular dynamics (MD) simulations can be increased without
compromising fundamental system dynamics and thermodynamics. With this new technique, we see a significant speedup in
our MD simulations. Atomistic simulations on single crystal and polycrystalline silicon's mechanical deformation at micrometer
length scale reveal features that are in direct quantitative agreement with experiments. We then discuss the use of this method
in different thermal and mechanical behavior analyses projects undergoing in our lab.