For the first time, researchers have successfully used quantum machine learning to create semiconductors, paving the way for a more efficient chip-making process. This breakthrough could transform how modern devices are manufactured.
In a study published in June 2023, scientists in Australia developed a quantum machine learning technique that combines artificial intelligence and quantum computing principles to improve the challenging task of modeling electrical resistance inside chips. This is a key factor affecting how efficiently a chip performs.
Quantum machine learning blends classical data with quantum computing methods, allowing it to uncover complex patterns in data that would be difficult for classical systems to detect. The process involves encoding classical data into quantum states, which enables the quantum computer to identify relationships in the data.
The researchers focused on modeling Ohmic contact resistance, a critical challenge in chipmaking. This is a measure of how easily electricity flows between metal and semiconductor layers; lower values indicate faster performance. Engineers typically rely on classical machine learning algorithms for this task, but these often struggle with small, noisy datasets containing nonlinear patterns.
To address this limitation, the researchers developed a new algorithm called the Quantum Kernel-Aligned Regressor (QKAR). This method converts classical data into quantum states, allowing the quantum system to identify complex relationships. A classical algorithm then learns from these insights, creating a predictive model to guide chip fabrication.
The QKAR was tested on five new samples not included in the training data and outperformed seven leading classical models, achieving significantly better results than traditional methods. The researchers believe this technology can be deployed on quantum machines as they become more reliable, paving the way for real-world applications in chip production.
This breakthrough demonstrates the potential of quantum machine learning to effectively handle high-dimensional, small-sample regression tasks in semiconductor domains. As quantum hardware continues to evolve, this method could soon be applied to real-world chip production, transforming the manufacturing process.
Source: https://www.livescience.com/technology/computing/scientists-use-quantum-machine-learning-to-create-semiconductors-for-the-first-time-and-it-could-transform-how-chips-are-made