With this in mind, researchers from Nagoya University designed graphene-diamond junctions capable of mimicking the characteristics of biological synapses and key memory functions. Researchers led by Kenji Ueda demonstrated optoelectronically controlled synaptic functions using junctions between vertically aligned graphene (VG) and diamond. The fabricated junctions mimic biological synaptic functions, such as the production of “excitatory post-synaptic current” (EPSC) — the charge induced by neurotransmitters at the synaptic membrane — when stimulated with optical pulses, and exhibit other basic brain functions, such as the transition from short-term memory (STM) to long-term memory (LTM).
“Our brains are well equipped to sieve through the information available and store what’s important. We tried something similar with our VG-diamond arrays, which emulate the human brain when exposed to optical stimuli,” Ueda said. “This study was triggered due to a discovery in 2016 when we found a large optically induced conductivity change in graphene-diamond junctions.”
Apart from EPSC, STM, and LTM, the junctions also show a paired pulse facilitation of 300% — an increase in postsynaptic current when closely preceded by a prior synapse.
The VG-diamond arrays underwent redox reactions induced by fluorescent light and blue LEDs under a bias voltage. The team attributed this to the presence of differently hybridized carbons of graphene and diamond at the junction interface, which led to the migration of ions in response to the light and in turn allowed the junctions to perform photo-sensing and photo-controllable functions similar to those performed by the brain and retina. Additionally, the VG-diamond arrays surpassed the performance of conventional rare-metal-based photosensitive materials in terms of photosensitivity and structural simplicity.
“Our study provides a better understanding of the working mechanism behind the artificial optoelectronic synaptic behaviors, paving the way for optically controllable brain-mimicking computers with better information-processing capabilities than existing computers,” Ueda said.
The research was published in Carbon (www.doi.org/10.1016/j.carbon.2021.06.060).