The brain-inspired codes of the AI revolution primarily run on conventional silicon computer architectures that were not designed for this purpose. This will soon lead to unrealistic energy consumption as AI continues to grow. Neuromorphic architectures offer the promise of lower energy consumption by mimicking the brain’s basic components—neurons and synapses—ideally using a single material. Unfortunately, among the few quantum materials which naturally act as spiking "neuristors" (artificial neurons), non-volatile "synaptor" memory (artificial synapses) has been hard to implement. One promising candidate functioning at room temperature is vanadium dioxide (VO2). Interestingly, this material exhibits multiscale fractal electronic patterns during its insulator-metal transition that must be understood and controlled before hoping to fully use it in a neuromorphic device.
To achieve this, we have developed a new optical microscopy method that allows for the precise sub-micron recording of these patterns. This cluster image series first enabled us to generate Tc maps [1] and reveal the underlying interactions using newly developed machine learning techniques [2]. More recently, by utilizing various temperature sweeps [3], a focused laser beam, and an AFM scanning tip [4], we have been able to modify and control these insulator-metal patterns. This breakthrough offers, for the first time, the possibility of creating rewritable synaptic connections between neuristors in a single VO2 material neuromorphic chip.
[1] arXiv:2301.04220 (2023)
[2] Phys. Rev. B 107, 205121 (2023)
[3] Adv. Electron. Mater. 2023, 9, 2300085 (2023)
[4] To be published in Adv. Electron. Mater. (2024)