APS March Meeting 2014
Volume 59, Number 1
Monday–Friday, March 3–7, 2014;
Denver, Colorado
Session D12: Invited Session: Materials Genome: Theory-Led Accelerated Materials Discovery
2:30 PM–5:30 PM,
Monday, March 3, 2014
Room: 205
Sponsoring
Unit:
DMP
Chair: Darrell Schlom, Cornell University
Abstract ID: BAPS.2014.MAR.D12.5
Abstract: D12.00005 : The ``Missing Compounds'' affair in functionality-driven material discovery*
4:54 PM–5:30 PM
Preview Abstract
Abstract
Author:
Alex Zunger
(University of Colorado, Boulder, Colorado 80309)
In the paradigm of ``data-driven discovery,'' underlying one of the leading
streams of the Material Genome Initiative (MGI), one attempts to compute
high-throughput style as many of the properties of as many of the N (about
10**5- 10**6) compounds listed in databases of previously known compounds.
One then inspects the ensuing Big Data, searching for useful trends. The
alternative and complimentary paradigm of ``functionality-directed search
and optimization'' used here, searches instead for the n much smaller than N
configurations and compositions that have the desired value of the target
functionality. Examples include the use of genetic and other search methods
that optimize the structure or identity of atoms on lattice sites, using
atomistic electronic structure (such as first-principles) approaches in
search of a given electronic property. This addresses a few of the
bottlenecks that have faced the alternative, data-driven/high throughput/Big
Data philosophy: (i) When the configuration space is theoretically of
infinite size, building a complete data base as in data-driven discovery is
impossible, yet searching for the optimum functionality, is still a
well-posed problem. (ii) The configuration space that we explore might
include artificially grown, kinetically stabilized systems (such as 2D layer
stacks; superlattices; colloidal nanostructures; Fullerenes) that are not
listed in compound databases (used by data-driven approaches), (iii) a large
fraction of chemically plausible compounds have not been experimentally
synthesized, so in the data-driven approach these are often skipped. In our
approach we search explicitly for such ``Missing Compounds''. It is likely
that many interesting material properties will be found in cases (i)-(iii)
that elude high throughput searches based on databases encapsulating
existing knowledge. I will illustrate (a) Functionality-driven discovery of
topological insulators and valley-split quantum-computer semiconductors, as
well as (b) Use of ``first principles thermodynamics'' to discern which of the
previously ``missing compounds'' should, in fact exist and in which structure.
Synthesis efforts by Poeppelmeier group at NU realized 20 never-before-made
half-Heusler compounds out of the 20 predicted ones, in our predicted space
groups. This type of theory-led experimental search of designed materials
with target functionalities may shorten the current process of discovery of
interesting functional materials.
*Supported by DOE ,Office of Science, Energy Frontier Research Center for Inverse Design
To cite this abstract, use the following reference: http://meetings.aps.org/link/BAPS.2014.MAR.D12.5