UC Berkeley

This dissertation broadly describes efforts related to materials discovery, design and optimization

using a first-principles approach based on density functional theory (DFT).

The two main application areas comprise transition-metal alloys and functional materials.

The research involving transition-metal alloys aims at finding cost-effective replacement

strategies for rhenium (Re). Elemental Re exhibits profuse deformation twinning under

mechanical loads and has a high ductility. It is shown that the twinning characteristics of

Re and its ductility are correlated with its anomalously low (11-21) twin-boundary energy.

The origin of this twin-energy anomaly is related to the presence of icosahedral structural

units on the (11-21) twin boundary. These structural units are stabilized near d-band fillings

corresponding to Re. The (11-21) twin-boundary energy can be lowered further by

decreasing the d-band filling with respect to elemental Re. This increases the intrinsic

ductility according to three independent ductility parameters employed in this work: i)

Pugh's ratio of bulk to shear modulus, ii) Yoo's ratio of surface to twin energy and iii) an

ideal-strength criterion. Based on new insights in the relation between d-band filling, defect

energies and intrinsic ductility, several candidate alloys for Re replacement are proposed.

Ru-based alloys with additions of Ta, W and Re are shown to be potential replacement

candidates. The last part of this dissertation describes high-throughput calculations that

have culminated in the two largest databases of elastic and piezoelectric tensor properties

available to-date. The workflow for doing such calculations is described, along with various

checks to ensure the accuracy of the calculated physical properties. The database with

elastic-tensor properties is expected to be of use in a number of fields where discovery of

new materials with desired values of elastic stiffness or (lattice) thermal conductivity are of

interest. Further, data mining and machine learning can be applied to better understand

elastic properties and their physical descriptors. Several novel piezoelectric materials are

discovered as part of this work. Some of these exhibit a high intrinsic piezoelectric response

and may serve as a starting point for a process in which the piezoelectric response is optimized

by alloying and microstructure engineering. As such, this database can support a search for future

replacement candidates of the widely used piezoelectric materials such as Pb(ZrxTi1?-x)O3 (PZT) that contain lead.