Physics · Northeastern University

AI-enabled experimentation for quantum materials.

We aim to understand and control quantum materials through experiments that learn from their results.

Our research connects automated atomic fabrication, intelligent microscopy, and closed-loop materials discovery.

Explore our research Join the lab

Conceptual illustration of atomic fabrication, adaptive characterization, and learning-guided experimental decisions. — Atomic control, adaptive measurements, and learning-guided decisions. AI-generated conceptual illustration.

Help build the lab. We welcome inquiries from prospective students and postdoctoral researchers.

Research opportunities

Research directions

From atoms to discovery.

Develop experimental control at the atomic scale, learn from measurements, and use that knowledge to guide materials research.

Automated atomic fabrication

Develop ways to create and modify atomic-scale structures, then test how those changes affect their properties.

Atomic-scale control

Intelligent microscopy

Build scanning probe and other characterization workflows that tune instruments and adapt measurements to the sample.

Adaptive measurements

Closed-loop discovery

Connect fabrication, characterization, and machine learning so that experimental results determine the next step.

Learning through experiments

Selected research

Measurements that guide
the next experiment.

Selected publications

Autonomous materials characterization

Exploring a composition library

Measurements of roughness, piezoresponse, and hysteresis-loop area guide where to measure next across an (Al, Sc, B)N library.

Multi-objective characterization of an Al-Sc-B composition library: measurements inform Gaussian-process models, acquisition maps, and the next measurement location. — Measured responses, model predictions, and experimental decisions in a closed loop.View full-size figure

Multi-objective active learning

Connecting local structure
with material response

Image patches and measured objectives inform deep kernel learning models. The resulting predictions help select the next location for an experiment.

Piezoresponse images and local changes feed multi-objective deep kernel learning models, which generate Pareto and acquisition maps for the next measurement. — Multi-objective deep kernel learning for adaptive scanning probe experiments.View full-size figure

Principal investigator

Richard Liu

Assistant Professor of Physics at Northeastern University, joining the Quantum Materials and Sensing Institute in Burlington. His research brings together quantum materials, scanning probe microscopy, and machine learning for experimental science.

Meet the principal investigator