Chemcial Physics Seminar - CMSP

[Condensed Matter Section]

CMSP Chemical Physics Seminar
=============================

Monday 14 March at 11:00 a.m.
Euler Lecture Hall, Terrace Level, Leonardo Building


Giancarlo TRIMARCHI
Dept. of Physics, Northwestern University, Evanston, U.S.A.


"Designing Material by 'ab initio' Methods and Optimization
Algorithms: The Search for New Transparent Conductors as a
Case Study"

_Abstract:_
Computational materials design is an emerging research paradigm
in which state-of-the-art electronic structure methods are applied
to predict materials with target properties within families of  candidate
compounds. In this seminar, as an example of materials design driven
by ab initio methods, I will describe the application of density functional
theory (DFT), and its corrections and extensions, to predict new p-type
transparent conducting materials (TCMs). While TCMs are crucial in
optoelectronics, transparency and p-type conductivity rarely coexist in the
same material. To identify new candidate p-type TCMs, we screened the family of
known Ag and Cu oxides and identified Ag3VO4 and KAg11(VO4)4 as p-type
conductors transparent to the red light. In these oxides, the Ag vacancies act
as intrinsic hole-generating defects, and the hole effective mass is lighter
than in CuAlO2, the prototypical p-type transparent conducting oxide. However,
ab initio methods require the crystal structure as input information to predict
materials properties. Therefore, the ability to predict the crystal structure
of hypothetical compounds without constraints or assumptions on the lattice
symmetry is essential in order to discover new functional materials beyond the
repertoires of known solids. Here, I will describe a crystal structure
prediction method based on an evolutionary algorithm to search for the global
total-energy minimum of a solid calculated by DFT as a function of the lattice
vectors and atom positions. I will illustrate the application of this global
space-group optimization (GSGO) algorithm to selected binary and ternary
solids. Finally, I will discuss an extension of the GSGO algorithm in which the
stoichiometry is also optimized so as to predict at the same time the composition
and crystal structure of the thermodynamically stable phases of a solid system.**