Search for: All records

Real-time, on-chip learning has become increasingly important as artificial intelligence edge-computing systems are deployed in dynamic environments ranging from autonomous vehicles to robotics, where static pretrained models are insufficient. In this context, analog in-memory computing hardware has been explored to alleviate energy and latency bottlenecks, but its limited reconfigurability has hindered the implementation of reinforcement learning in deployed artificial intelligence agents. Here, we demonstrate split-gate MoS2 memtransistors where local field-effect gating of the Schottky contacts and semiconducting channel enables improved control of memristive switching ratios and conductance states, respectively. These characteristics enable efficient implementation of reinforcement learning due to the combination of nonvolatile synaptic weight updates and rapid parameter adjustments. Adaptive reinforcement learning performance is benchmarked with a cartpole balancing task that serves as an elementary robotic example for embodied decisions and actions, ultimately showing a 6-fold improvement in total reward and a 5-fold reduction in the number of programming steps. 

The rapid deployment of AI agents in IoT, healthcare, robotics, and consumer electronics, combined with the stalled scaling and limited energy efficiency of silicon CMOS computing architectures, have placed unsustainable demands on electricity and water use in data centers. These problems are exacerbated for edge computing applications that require on-chip learning in evolving environments. Recent research in 1D and 2D nanomaterials exemplifies co-design approaches to discover and optimize new materials, devices, circuit architectures, and algorithms. These low-dimensional nanomaterials have shown promise for the realization of high-performance digital electronics and bioinspired approaches for energy-efficient edge AI. This work is fueled by a growing library of low-dimensional nanoelectronic materials, 3D integration, and vdW heterojunctions, resulting in simplified and reconfigurable neuromorphic circuits. 

Memristive devices based on two-dimensional (2D) materials have emerged as potential synaptic candidates for next-generation neuromorphic computing hardware. Here, we introduce a numerical modeling framework that facilitates efficient exploration of the large parameter space for 2D memristive synaptic devices. High-throughput charge-transport simulations are performed to investigate the voltage pulse characteristics for lateral 2D memristors and synaptic device metrics are studied for different weight-update schemes. We show that the same switching mechanism can lead to fundamentally different pulse characteristics influencing not only the device metrics but also the weight-update direction. A thorough analysis of the parameter space allows simultaneous optimization of the linearity, symmetry, and drift in the synaptic behavior that are related through tradeoffs. The presented modeling framework can serve as a tool for designing 2D memristive devices in practical neuromorphic circuits by providing guidelines for materials properties, device functionality, and system performance for target applications. 

Artificial intelligence (AI) algorithms are currently executed using silicon-based hardware, resulting in excessively high energy demand for data centers. Edge computing AI for healthcare, robotics, and autonomous vehicles presents even stricter power and latency constraints, which are unmet by incumbent computing architectures. With vastly superior energy efficiency, biological neuronal networks provide hints towards alternative computational approaches including memory-logic colocation, asynchronous parallelism, and spike-triggered computation. Here, we draw inspiration from the cerebellum to demonstrate asymmetric-contact-gated MoS2 memtransistors that exhibit bias-polarity-dependent excitatory/inhibitory short-term plasticity. Arrays of these cerebellum-inspired memtransistors exploit the evolving interplay between excitatory and inhibitory responses to emulate the emergent synaptic differentiation of the cerebellum, enabling rapid identification of novel events. When applied to electrocardiogram data, arrhythmias are detected within a single heartbeat with 10,000-fold fewer operations than existing silicon-based approaches. In this manner, cerebellum-inspired neuromorphic hardware provides a pathway to computationally efficient, high-speed novelty detection for edge intelligence. 

Artificial neurons that reproduce the rich dynamical behaviour of biological spiking are essential for neuromorphic hardware and biohybrid interfaces, yet scalable solution-processed devices with physiologically relevant spiking characteristics remain elusive. Here we demonstrate aerosol-jet-printed memristive networks of MoS2 nanosheets that exhibit thermally activated filamentary switching and snap-back negative differential resistance, enabling volatile threshold switching in fully printed graphene/MoS2/graphene devices on flexible substrates. In situ thermal imaging and circuit modelling reveal that current-constricted filaments formed through Joule heating govern the nonlinear switching dynamics. These printed memristors enable oscillatory and spiking neuron circuits with tunable frequencies up to 20 kHz and stable operation over more than 10^6 cycles. Simple neuristor circuits realize first-, second- and third-order spiking complexity, including integrate-and-fire behaviour, spike latency, tonic firing, class 1 excitability, tonic bursting and phasic dynamics. The generated spike waveforms match physiological timescales and stimulate Purkinje neurons in mouse cerebellar slices. Our results establish printed nanosheet memristive networks as a scalable platform for bio-realistic neuromorphic hardware and flexible brain–machine interfaces. 

Recent breakthroughs in brain-inspired computing promise to address a wide range of problems from security to healthcare. However, the current strategy of implementing artificial intelligence algorithms using conventional silicon hardware is leading to unsustainable energy consumption. Neuromorphic hardware based on electronic devices mimicking biological systems is emerging as a low-energy alternative, although further progress requires materials that can mimic biological function while maintaining scalability and speed. As a result of their diverse unique properties, atomically thin two-dimensional (2D) materials are promising building blocks for next-generation electronics including nonvolatile memory, in-memory and neuromorphic computing, and flexible edge-computing systems. Furthermore, 2D materials achieve biorealistic synaptic and neuronal responses that extend beyond conventional logic and memory systems. Here, we provide a comprehensive review of the growth, fabrication, and integration of 2D materials and van der Waals heterojunctions for neuromorphic electronic and optoelectronic devices, circuits, and systems. For each case, the relationship between physical properties and device responses is emphasized followed by a critical comparison of technologies for different applications. We conclude with a forward-looking perspective on the key remaining challenges and opportunities for neuromorphic applications that leverage the fundamental properties of 2D materials and heterojunctions. 

Soft materials with reversible electrical and mechanical properties are critical for the development of advanced bioelectronics that can distinguish between different rates of applied strain and eliminate performance degradation over many cycles. However, the current paradigm in mechano‐electronic devices involves measuring changes in electrical current based on the accumulation of strain within a conductive material that alters the geometry through which electrons flow. Attempts have been made to incorporate soft materials like liquid metals and concentrated solutions of conjugated polymers and salts to overcome materials degradation but are limited in their ability to detect changes in the rate of the applied strain. Herein, the anisotropic electrical performance of a soft semiconducting composite prepared with silver‐coated microspheres dispersed within a swollen copolymer gel is demonstrated. This composite exhibits an electrical response proportional to the magnitude of the applied shear force to enable a rate‐of‐strain dependent conductivity. Furthermore, a 100‐fold increase in the conductivity of the composite is observed when the electric field is oriented parallel to the flow direction. This improvement in the electrical response can be attributed to the enhanced alignment of microspheres in viscoelastic media and can be leveraged in the development of mechanically responsive electronic devices.