Artificial inorganic synapses achieved using atomic switch technology
T. Tsuruoka , K. Terabe , T. Hasegawa , M. Aono
We demonstrated that atomic switches can emulate the synaptic plasticity underlying memory functions in the human brain. The change in the conductance of the atomic switch is considered analogous to the change in the strength of a biological synapse which varies according to stimulating input pulses. The atomic switch therefore has potential for use as an essential building block for neural computing systems.
In biological systems, there are two types of synaptic plasticity: short-term plasticity (STP), in which changes in synaptic strength last for only a very short time and then it quickly returns to the original state; and long-term potentiation (LTP), in which the enhancement of synaptic strength can last from a few hours to the life of the living organism. The appearance of STP and LTP corresponds to the formation of short-term memory (STM) and long-term memory (LTM), which depends on the strength, frequency, and number of stimuli.
Atomic switches can mimic this synaptic behavior when device conductance varies depending on the repetition rate, amplitude, and number of input voltage pulses (Fig. 1(a)). The STM and LTM behaviors were first discovered in a gap-type Ag2S atomic switch[1]. For practical applications, the realization of synaptic properties by a gapless-type atomic switch is more important, because it can be easily integrated into complementary metal–oxide–semiconductor (CMOS) circuits. We demonstrated synaptic behavior in an Ag/Ta2O5/Pt device based on the transport of metal ions[2]. A temporary increase in conductance and its spontaneous decay over time is observed with lower repetitions of input stimuli, but persistent enhancement is achieved by higher input repetitions, as shown in Fig. 1(b). The transition from STM to LTM over a wide time scale can also be achieved using the transport of oxygen vacancies in a Pt/WO3-x/Pt device (Fig. 1(c))[3].
Our results show that individual atomic switches enable a new functional element suitable for the design of neural systems that can work without the poorly scalable software and preprograming employed in current CMOS-based neural networks. These artificial synapses will contribute to the achievement of next-generation neural computing systems.
Main Papers
- “Short-term plasticity and long-term potentiation mimicked in single inorganic synapses”, T. Ohno, T. Hasegawa, T. Tsuruoka, K. Terabe, J.K. Gimzewski, M. Aono, Nat. Mater. 10 (2011) 591.
- “Conductance quantization and synaptic behavior of a Ta2O5-based atomic switch”, T. Tsuruoka, T. Hasegawa, M. Aono, Nanotechnology 23 (2012) 435705.
- “Synaptic plasticity and memory functions achieved in a WO3-x-based nanoionics device by using the principle of atomic switch operation”, R. Yang, K. Terabe, Y. Yao, T. Tsuruoka, T. Hasegawa, J.K. Gimzewski, M. Aono, Nanotechnology 24 (2013) 384003.