operation per second) with ultralow power consumption. [1,2] The biological synapses are fast conductive connections between typical cortical neurons which is capable to transmit and receive electrochemical signals. [3] The parallel data processing is ensured since the neurons have one to ten thousand synapse links to connect with each other. Inspired by the high efficient brain, emerging nanoelectronic devices and complementary metal-oxide-semiconductor technology have been recently demonstrated to emulate the synaptic functions including short-term potentiation (STP), long-term potentiation (LTP), long-term depression (LTD), paired-pulse facilitation (PPF), paired-pulse depression (PPD), and post-tetanic potentiation, which are expected to settle the von Neumann architecture bottleneck. [4][5][6][7] In biological synapse, the plasticity defined as signal-modulated conductivity originated from ion movements is less than 15% in term of asymmetric behavior which is related to signal direction. [8] The LTP and LTD observed in biological synapses correspond to the continuously increased and decreased conductivity in the artificial counterpart in which the desirable variation should be both symmetric (direction of change) and linear (differential magnitude). [2] One of the most vigorously studied synaptic devices is memristor which can be operated between the high resistance state and the low resistance state with continuous internal resistance states with applied bias history. [9][10][11][12][13] Nevertheless, it is hard for memristors to exhibit either linear or symmetric conductivity variation in function with successive SET and RESET operations. Alternatively, integrating augmentative functionalities into a single synapse to enhance the plasticity and parallelism allows the development of robust neuromorphic circuit. [14] Field-effect transistor (FET)-based flash memory with an extra gate-control is promising to achieve hysteresis for mimicking dynamic, linear plasticity. In this structure, the charge carriers can be trapped and released from the charge trapping sites by reversed polarity of bias conducted programming and erasing operations in which the type of trapped charge carriers can be verified by judging the shifting direction of the threshold voltage (V th ). [15][16][17][18][19][20][21][22][23][24] For Employing flash memories that function as the analogue synapse cleft for implementation of neural systems is a popular approach. For practical application in synapse mimicking, ambipolar trapping of both holes and electrons enables the weight updated linearity, symmetric conductivity variation, and desired variation margin, which are consistent with biological synapse. Nevertheless, most of the flash memory exhibits unipolar charge trapping behavior. Here, the first polyoxometalates (POMs)-modulated reduced graphene oxide (RGO)-based flash memory with ambipolar charge trapping characteristics is reported. Keggin type [PW 12 O 40 ] 3− (PW) is selected to function as both catalyst for in situ photoreduction ...

No abstract

No abstract