(Invited) Oxide Neuron Transistor for High-Performance Biochemical Sensors

Abstract

Synergic integration of presynaptic inputs from the dendrites plays an important role for sensory information process and cognitive computation, and the idea of building bio-inspired solid-state devices has been around for decades. In 1992, Shibata et al proposed the Si-based neuron transistors with multiple input gates that are capacitively coupled to a floating gate . The “on” or “off” state of the neuron transistors depends on the integrated effect of the multiple input gates. One of the unique features of the neuron transistors is the ultralow power dissipation during the calculation due to the gate-level sum operation in a voltage mode. From then on, Si-based neuron transistors have attracted much attention for chemical and biological detection due to the easy adjustment of threshold voltage. But, up to now, electrolyte-gated neuron transistors with amorphous oxide channel layers for biochemical sensing applications are not reported. For portable applications, low-voltage operation is preferred. Electrolyte gated electric-double-layer (EDL) transistors can act as potential candidates with a low operation voltage due to the strong EDL modulation at the electrolyte/channel interface. Recently, oxide-based EDL transistors gated by solid-state inorganic electrolytes were proposed by our group. At the same time, artificial synapses and neuron transistors with low power consumption and fundamental biological functions were mimicked in individual device. The sensitivity of a standard ion-sensitive field-effect transistor is limited to be 59.2 mV/pH (Nernst limit) at room temperature. Here, inspired by the dendritic integration and spiking operation of a biological neuron, oxide-based neuron transistors with multiple input gates are fabricated on flexible plastic substrates for pH sensor applications. When the neuron transistor sensor is operated in a quasi-static dual-gate synergic sensing mode, it shows a high pH sensitivity of ~105 mV/pH, which is higher than the Nernst limit. Our results also demonstrate that single-spike dynamic mode can remarkably improve pH sensitivity and reduce response/recover time and power consumption. At last, we find that appropriate depression applied on the sensing gate electrode can further enhance the pH sensitivity and reduce the power consumption. Our flexible neuron transistors provide a new-concept sensory platform for biochemical detection with high sensitivity, rapid response and ultralow power consumption.