UNSW develops AI workflow designed to speed discovery of hybrid perovskite semiconductors

An artificial intelligence-assisted workflow developed at the University of New South Wales is designed to sharply cut the time needed to discover next-generation semiconductor materials, replacing traditional trial-and-error with a goal-driven approach, according to the research team.

The system focuses on hybrid perovskites, a class of semiconductors used in applications such as solar cells and light-emitting diodes. These materials combine inorganic layers with organic molecules, and small changes to those molecules can dramatically affect performance.

That complexity has made development slow, with researchers historically testing large numbers of combinations experimentally. UNSW’s approach reverses the process. It starts with a desired performance outcome—such as how efficiently a material should handle electrical charge—and works backwards to identify suitable molecular candidates.

The workflow then screens millions of possible combinations and filters out those that are impractical to synthesise, narrowing the field to a more manageable set of viable options. The remaining candidates are assessed using detailed computational simulations.

The materials identified through this process have not yet been tested in a laboratory. Even so, the researchers say the workflow could significantly reduce the time required to develop new materials for electronics and energy applications by steering efforts toward the most promising targets.

The study addresses a long-standing challenge in materials science, where progress has often relied on incremental adjustments to known compounds rather than systematic exploration of large chemical spaces. By enabling more targeted discovery, the team believes the approach could accelerate innovation in semiconductors, solar energy and advanced electronic devices.