Interfaces create properties that neither material possesses alone. They decide whether a quantum state survives integration, and whether a contact in a low-energy transistor — the hardware that sets the energy cost of computing, AI included — can both switch and conduct. I understand and design interfaces with first-principles and localised quantum models, alongside experiment and industry. At Sydney I would build First Contact: a program that turns interface physics into a physical language that AI can learn from, generate with and be tested against.
Interfaces, and why AI needs physics
Interfaces can reverse semiconductor polarity or improve carrier injection, and machine-learning models can now propose favourable interface configurations. But a successful configuration does not establish which interactions caused its response, whether they can be recreated elsewhere, or how much variation they tolerate — so AI-designed interfaces can be neither trusted nor reproduced. My program asks which interactions are necessary, when they are sufficient, and where the description transfers, and expresses the answers in a form that learning and generative methods can use and physics can test (right).
Track record
After a PhD (Monash, 2016) and seven years in the ARC Centre of Excellence FLEET, I have built an independent program on surfaces, interfaces and disorder across 46 refereed articles.
Original interface methods. I developed approaches that connect first-principles calculations with efficient localised models, and showed how surrounding layers and interface symmetry control the states of two-dimensional bismuth — weak substrate coupling and symmetric encapsulation each preserve its edge states (New J. Phys. 2021; Mater. Today Phys. 2023, 2025).
Interfaces in real devices. Our pick-and-place transfer of arbitrary-metal electrodes connects atomic interactions with van der Waals device fabrication (ACS Nano 2025, co-first). As Lead CI of the A$650K AEA Ignite project and lead inventor on its US patent application, I am taking an engineered transistor channel towards manufacturing with TQ Transistors. With experimental partners I showed that giant Berry-curvature responses survive complete structural disorder (Matter 2025, co-first; Nature Communications 2026, co-first).
Preliminary foundation. My TQ Transistors programme adds interface design, agent-assisted coding and high-throughput screening. Its completed calculations (Figure 1) reveal the opportunity: assembly narrows the zero-field gap from 1.78 eV to about 0.4 eV, yet the complex stack retains a far stronger field response than the bare bilayer.
