Quantum AI Designs Semiconductors

Researchers have successfully employed quantum machine learning (QML) to streamline semiconductor design and fabrication. This innovative approach uses a hybrid of artificial intelligence and quantum computing. Scientists at Australia's CSIRO have demonstrated the world's first use of QML in fabricating semiconductors, specifically testing their QML model on 159 samples of GaN HEMT semiconductors, which offer enhanced performance over silicon-based transistors.

The quantum kernel extracts key features, and a classical algorithm then uses this information to guide fabrication, identifying crucial parameters for optimisation. The QKAR technique outperformed seven classical machine learning algorithms. This method addresses the increasing constraints of data scarcity and process complexity in the semiconductor industry. Quantum models can capture patterns that classical models might miss, especially in high-dimensional, small-data regimes.

This research marks a significant advancement, showcasing the potential of quantum methods to improve real-world semiconductor fabrication processes. The team focused on modelling Ohmic contact resistance, a challenging but critical property in semiconductor design, demonstrating that QML can outperform classical AI in modelling this process.