Schematic diagram of the quantum machine learning-based modeling process of OHMIC contact formation in GAN HEMTS. credit: Advanced Science (2025). doi:10.1002/advs.202506213
Semiconductor processing is notoriously difficult. This is one of the most complicated feats of modern engineering, as the extreme precision required and hundreds of steps, such as etching and layering, create a single chip.
However, researchers at the Federal Institute of Scientific and Industrial Research (CSIRO), the world's number one research institute in Australia, use Quantum Machine Learning to manufacture semiconductors. Their research could revolutionize how chips are made.
Research into teams published in the journal Advanced Sciencefor the first time, we show that semiconductor manufacturing can be improved by applying quantum methodologies to actual experimental data.
They focused on key steps in the semiconductor design process. The ohmic contact resistance of semiconductor materials was modeled. This is a measure of the electrical resistance generated when a semiconductor comes into contact with a metal, and it affects how easily the current flows.
Modeling problems
One of the points of attachment to date is the extremely difficulties in modeling the contact resistance of ohms. The current approach uses classic machine learning (CML) algorithms, but requires large data sets and results in poor performance in nonlinear settings for small samples.
The Australian researchers, led by Muhammad Usman, CSIRO's Quantum Systems professor and principal, went in a different way.
They adopted a quantum machine learning (QML) approach for data from 159 experimental samples of Gan Hemt (Gallium High-Electron-Mobility Transistor) semiconductors. This clever method blends classical techniques with quantum techniques.
Quantum ablation studies to optimize QKAR performance. credit: Advanced Science (2025). doi:10.1002/advs.202506213
First, they narrowed down many manufacturing variables to those that have a significant impact on performance.
They then developed a quantum kernel-adapted regressor (QKAR) architecture to convert classical data into quantum states to begin the machine learning process. Once all features were extracted from the data, classical algorithms were trained to obtain the information and then direct the manufacturing process.
The QKAR technology outperformed seven different CML algorithms developed for the same problem.
“These findings illustrate the potential of QML to effectively handle high-dimensional small-sample regression tasks in the semiconductor domain, and point to promising means of deployment in future real-world applications as quantum hardware continues to mature,” the researchers write.
In addition to potentially reducing manufacturing costs and improving device performance in the semiconductor industry, this study could have other widespread results. As quantum technology continues to evolve, it may help solve complex problems beyond the capabilities of classical computers.
Written by author Paul Arnold, edited by Gaby Clark, and fact-checked and reviewed by Robert Egan. This article is the result of the work of a careful human being. We will rely on readers like you to keep independent scientific journalism alive. If this report is important, consider giving (especially every month). You'll get No ads Account as a thank you.
detail:
Zeheng Wang et al., Quantum kernel learning for small dataset modeling in semiconductor manufacturing: application to OHMIC contacts, Advanced Science (2025). doi:10.1002/advs.202506213
©2025 Science X Network
Quote: Quantum Machine Learning will improve semiconductor manufacturing for the first time on July 6, 2025 from https://techxplore.com/news/2025-07-Quantum-machine-machine-ductor.html (July 3, 2025)
This document is subject to copyright. Apart from fair transactions for private research or research purposes, there is no part that is reproduced without written permission. Content is provided with information only.
