Fig. 4: Characteristics of volatile DG MPBTFT-based LIF neuron.
From: Analog reservoir computing via ferroelectric mixed phase boundary transistors
a Schematic of the volatile DG MPBTFT-based LIF neuron (left panel) and excitatory/inhibitory pulse schemes for the BG/TG (right panel). The DG MPBTFT-based LIF neuron exhibits high area efficiency and functional versatility as it can process both excitatory and inhibitory pulses within a single device. Excitatory pulses applied to the BG increase the ID of the DG MPBTFT, whereas inhibitory pulses applied to the TG decrease the ID. The voltage of the excitatory pulses ranges from Vbase to Vhigh, while the voltage of the inhibitory pulses ranges from –3.0 V to Vinh. Every neuronal behavior was investigated under a VDS of 0.1 V. b Demonstration of the neuronal behavior with self-reset characteristics of the DG MPBTFT when excitatory pulses are applied. The neuron can still fire in response to the excitatory pulses after the self-reset. Modulation of neuronal behavior through the c Vbase and d Vhigh. As the Vbase increases, the leaky effect of the neuron is decelerated, and the integration function is strengthened. An increase in Vhigh leads to more significant polarization switching in the HZO film and increases the ID change. On the contrary, insufficient Vhigh degrades the integration function of the neuron. Neuronal behaviors in response to the application of both excitatory and inhibitory pulses with various e Vinh, f inhibitory pulse width (twidth,inh), and g the number of inhibitory pulses (N). The inhibitory efficiency of the neuron is enhanced as Vinh, twidth,inh, and N increase. h Neuronal behavior of DG MPBTFT-based LIF neuron. The inset represents the input pulse train. The neuron demonstrates three primary functions: integration, firing, and resetting functions.
