Research Projects

The two most pressing problems for large-scale artificial intelligence implementation are space-efficient handling of big data and energy-efficient computing. Current data processing accounts for around 10% of global energy use, with data centres consuming 205 TWh in 2018. By 2040, server stations using current technologies would demand more power than terrestrially produced. Microelectronics approaches face challenges due to fundamental constraints of existing computing paradigms. To address these challenges, new materials, exceptional device functionalities, and efficient architectures are needed. Memristive devices are a promising solution for energy- and space-efficient computing platforms, such as in-memory, analog, and neuromorphic computing. However, designing nanoscale memristors with as low switching energy per bit is challenging due to their reliance on stochastic mechanisms. The proposed approach involves designing molecular memristors where molecules can be reliably switched for billions of cycles with as low as attojoule switching energy, potentially offering six orders of magnitude improvement over the state-of-the-art. From spectroscopic techniques, strong indications suggest this number is attainable, potentially leading to the lowest switching energy device ever made. This approach could potentially lead to the lowest switching energy device ever made.