bottleneck and improve the system efficiency since the direct integration of ultrahigh memory layer on the processor chip is feasible. In another aspect, photonic memories are expected to speed up the von Neumann bottleneck and supercharge the performance of serial computers since the light signal can be regarded as the additional terminal of the underlying basic devices to ensure low power consumption. [7][8][9] The growing pursuit of practical photonic memories drives the rapid development of photonic technologies, especially in the region of nanofabricationcompatible optical signaling. In photonic memory, optical signals have to be converted into electrical signals and vice versa. However, it is difficult to make use of nearinfrared (NIR) light in photonic memory due to the inferior NIR sensitivity of most semiconductor materials originating from their broadband absorption. [10,11] In addition, although the decryption technology for visible light is mature in photonic memories, NIR photonic memristors are less progressed. [12][13][14] Upconversion materials are an anti-Stokes-shift type of photoluminescent compound that absorb several photons with long wavelength and emit one photon with shorter wavelength. [15,16] The upconversion nanoparticles (UCNPs) have been proposed as vital materials in various kinds of photonic applications including in vivo therapeutics, [17] biomedical imaging probes, [18] and optoelectronic [7,19] and optogenetic devices [20] due to their sharp emission bandwidth, high photochemical stability, and large anti-Stokes shift (up to several hundred nanometers). The UCNPs based on lanthanide ion (Yb 3+ , Er 3+ )-doped NaYF 4 have a narrow absorption band at 980 nm due to electronic transition between the energy levels in the lanthanide ion. [15] Thus, UCNPs exhibit extension of the applications in high-performance optoelectronics and multi-modal imaging in NIR band. However, the weak photo-absorption and insulating property of UCNPs limits their photon-electron conversion efficiency. The family of 2D materials including insulating hexagonal boron nitride (h-BN), semiconducting molybdenum disulfide (MoS 2 ) to semimetallic graphene and infrared-gapped black phosphorus has been demonstrated with distinct optical, electronic, and mechanical properties from conventional bulk materials. [21][22][23][24][25] Integration of UCNPs with 2D semiconducting materials results in heterostructures featuring increased sensitizing centers and energy transfer, which ensures the formation of more excitons and subsequently high sensitivity Photonic memories as an emerging optoelectronic technology have attracted tremendous attention in the past few years due to their great potential to overcome the von Neumann bottleneck and to improve the performance of serial computers. Nowadays, the decryption technology for visible light is mature in photonic memories. Nevertheless, near-infrared (NIR) photonic memristors are less progressed. Herein, an NIR photonic memristor based on MoS 2 -NaYF 4 :Yb 3+ , Er 3+ upconve...

This review presents the development of photonic memory, with a view towards inspiring more intriguing ideas on the elegant selection of materials and design of novel device structures that may finally induce major progress in the fabrication and application of photonic memory.

A multi-state information storage state could be achievedviaa configurable SET process with non-volatile devices based on Ti3C2nanosheets.

The inherent limitations of traditional silicon technology in the implementation of machine learning could be addressed by neuromorphic computing since it is able to mimic the resilience, versatility, and efficiency of the human brain. [1][2][3][4] Synaptic It is desirable to imitate synaptic functionality to break through the memory wall in traditional von Neumann architecture. Modulating heterosynaptic plasticity between pre-and postneurons by another modulatory interneuron ensures the computing system to display more complicated functions. Optoelectronic devices facilitate the inspiration for high-performance artificial heterosynaptic systems. Nevertheless, the utilization of near-infrared (NIR) irradiation to act as a modulatory terminal for heterosynaptic plasticity emulation has not yet been realized. Here, an NIR resistive random access memory (RRAM) is reported, based on quasiplane MoSe 2 /Bi 2 Se 3 heterostructure in which the anomalous NIR threshold switching and NIR reset operation are realized. Furthermore, it is shown that such an NIR irradiation can be employed as a modulatory terminal to emulate heterosynaptic plasticity. The reconfigurable 2D image recognition is also demonstrated by an RRAM crossbar array. NIR annihilation effect in quasiplane MoSe 2 /Bi 2 Se 3

The mimicking of both homosynaptic and heterosynaptic plasticity using a high‐performance synaptic device is important for developing human‐brain–like neuromorphic computing systems to overcome the ever‐increasing challenges caused by the conventional von Neumann architecture. However, the commonly used synaptic devices (e.g., memristors and transistors) require an extra modulate terminal to mimic heterosynaptic plasticity, and their capability of synaptic plasticity simulation is limited by the low weight adjustability. In this study, a WSe2‐based memtransistor for mimicking both homosynaptic and heterosynaptic plasticity is fabricated. By applying spikes on either the drain or gate terminal, the memtransistor can mimic common homosynaptic plasticity, including spiking rate dependent plasticity, paired pulse facilitation/depression, synaptic potentiation/depression, and filtering. Benefitting from the multi‐terminal input and high adjustability, the resistance state number and linearity of the memtransistor can be improved by optimizing the conditions of the two inputs. Moreover, the device can successfully mimic heterosynaptic plasticity without introducing an extra terminal and can simultaneously offer versatile reconfigurability of excitatory and inhibitory plasticity. These highly adjustable and reconfigurable characteristics offer memtransistors more freedom of choice for tuning synaptic weight, optimizing circuit design, and building artificial neuromorphic computing systems.