Research

What we develop

Research

Topics, methods, and capabilities that connect materials science, data, and physics.

Topics of Interest

Thermal and Radiation Barrier Coating Complex Oxides

Transition metal-substituted yttria-stabilized zirconia (YSZ) and YTaO₄, and multicomponent rare-earth-based oxide and silicate coatings that protect components in extreme thermal and radiation environments.

Materials for Microelectronics

Ferroelectric and dielectric thin films for next-generation memory and microelectronic devices. Current focus is on HfO₂, (Zr,Hf)O₂ and doped variants.

Corrosion Science

Passivation and electrochemical behavior in structural materials.

Nuclear Physics

Understanding the quark-gluon structure of matter as it emerges from experiment, phenomenology, and first-principles theoretical calculations.

Understanding Synthesis and Processing of Advanced Materials

Process-structure-property relationships in additive manufacturing and other synthesis routes.

Multi-Principal Element Alloys

High-entropy and multi-principal element alloys, spanning mechanical, magnetic, and corrosion-resistant behavior.

Methods & Tools

First-Principles Calculations

Tools: Quantum ESPRESSO (DFT), EPW (electron-phonon coupling), Phonopy (lattice dynamics and phonons), Yambo (GW-BSE, optical properties).

Computes electronic, vibrational, magnetic, and optical properties directly from quantum mechanics.

Machine Learning, Uncertainty Quantification & Explainable AI

Tools: Ensemble learning methods and physics-regularized model architectures; Bayesian and ensemble methods for uncertainty quantification; Partial Dependence Plots (PDP), Friedman’s H-statistic, and SHAP (Shapley values) for explainability.

Builds predictive models directly from experimental and computational data, quantifies how much those predictions can be trusted, and opens up black-box behavior to reveal which features drive predictions and how they interact.

Symbolic Machine Learning

Tools: Symbolic regression, physics-constrained search.

Recovers interpretable governing equations directly from data, rather than an opaque fit.

Active Learning for Materials Discovery

Tools: Bayesian optimization and active learning pipelines.

Chooses the next experiment or calculation most likely to improve a model or reveal a better material.

LLM-Guided Discovery & Agentic AI

Tools: LLM-based literature and data extraction, agentic multi-step research workflows.

Extends beyond structured data into unstructured literature and autonomous multi-step research.