Exploiting the excellent electronic properties of two-dimensional (2D) materials to fabricate advanced electronic circuits is a major goal for the semiconductor industry1,2. However, most studies in this field have been limited to the fabrication and characterization of isolated large (more than 1 µm2) devices on unfunctional SiO2–Si substrates. Some studies have integrated monolayer graphene on silicon microchips as a large-area (more than 500 µm2) interconnection3 and as a channel of large transistors (roughly 16.5 µm2) (refs. 4,5), but in all cases the integration density was low, no computation was demonstrated and manipulating monolayer 2D materials was challenging because native pinholes and cracks during transfer increase variability and reduce yield. Here, we present the fabrication of high-integration-density 2D–CMOS hybrid microchips for memristive applications—CMOS stands for complementary metal–oxide–semiconductor. We transfer a sheet of multilayer hexagonal boron nitride onto the back-end-of-line interconnections of silicon microchips containing CMOS transistors of the 180 nm node, and finalize the circuits by patterning the top electrodes and interconnections. The CMOS transistors provide outstanding control over the currents across the hexagonal boron nitride memristors, which allows us to achieve endurances of roughly 5 million cycles in memristors as small as 0.053 µm2. We demonstrate in-memory computation by constructing logic gates, and measure spike-timing dependent plasticity signals that are suitable for the implementation of spiking neural networks. The high performance and the relatively-high technology readiness level achieved represent a notable advance towards the integration of 2D materials in microelectronic products and memristive applications.
23 December 2021 in IEEE Journal on Exploratory Solid-State Computational Devices and Circuits
This work discusses the proposal of a spintronic neuromorphic system with spin orbit torque-driven domain wall motion (DWM)-based neurons and synapses. We propose a voltage-controlled magnetic anisotropy DWM-based magnetic tunnel junction (MTJ) neuron. We investigate how the electric field at the gate (pinning site), generated by the voltage signals from pre-neurons, modulates the DWM, which reflects in the nonlinear switching behavior of neuron magnetization. For the implementation of synaptic weights, we propose a 3-terminal MTJ with stochastic DWM in the free layer. We incorporate intrinsic pinning effects by creating triangular notches on the sides of the free layer. The pinning of the domain wall and intrinsic thermal noise of the device lead to the stochastic behavior of DWM. The control of this stochasticity by the spin orbit torque is shown to realize the potentiation and depression of the synaptic weight. The micromagnetics and spin transport studies in synapses and neurons are carried out by developing a coupled micromagnetic non-equilibrium Green’s function ( MuMag-NEGF ) model. The minimization of the writing current pulsewidth by leveraging the thermal noise and demagnetization energy is also presented. Finally, we discuss the implementation of digit recognition by the proposed system using a spike time-dependent algorithm.
21 November 2022 in IEEE Transactions on Electron Devices
Magnetic skyrmion-based data storage and unconventional computing devices have gained increasing attention due to their topological protection, small size, and low driving current. However, skyrmion creation, deletion, and motion are still being studied. In this study, we propose a skyrmion-based neuromorphic magnetic tunnel junction (MTJ) device with both long-and short-term plasticity (LTP and STP) (mixed synaptic plasticity). We showed that plasticity could be controlled by the magnetic field, spin-orbit torque (SOT), and the voltage-controlled magnetic anisotropy (VCMA) switching mechanism. LTP depends on the skyrmion density and is manipulated by the SOT and magnetic field while STP is controlled by the VCMA. The LTP property of the device was utilized for static image recognition. By incorporating the STP feature, the device gained additional temporal filtering ability and could adapt to a dynamic environment. The skyrmions were conserved and confined to a nano track to minimize the skyrmion nucleation energy. The synapse device was trained and tested for emulating a deep neural network. We observed that when the skyrmion density was increased, the inference accuracy improved: 90% accuracy was achieved by the system at the highest density. We further demonstrated the dynamic environment learning and inference capabilities of the proposed device.
Pre-shaped microbeams, curved or inclined, are widely used in MEMS for their interesting stiffness properties. These mechanisms allow a wide range of positive and negative stiffness tuning in their direction of motion. A mechanism of pre-shaped beams with opposite curvature, connected in a parallel configuration, can be electrothermally tuned to reach a near-zero or negative stiffness behavior at the as-fabricated position. The simple structure helps incorporate the tunable spring mechanism in different designs for accelerometers, even with different transduction technologies. The sensitivity of the accelerometer can be considerably increased or tuned for different applications by electrothermally changing the stiffness of the spring mechanism. Opposite inclined beams are implemented in a capacitive micromachined accelerometer. The measurements on fabricated prototypes showed more than 55 times gain in sensitivity compared to their initial sensitivity. The experiments showed promising results in enhancing the resolution of acceleration sensing and the potential to reach unprecedent performance in micromachined accelerometers.
30 October 2020 in Journal of Microelectromechanical Systems
Beam electrothermal actuators amplify the thermal expansion of pre-shaped beams and use the symmetrical structure to create a linear motion. These actuators, including V and Z shapes, are widely used in microsystems. Explicit analytical expressions are derived in this paper governing the structural deformation of these actuators due to electrothermal expansion and interaction with external lateral load. The analytical expressions are developed for an arbitrary initial shape of the actuator beams with symmetry and uniform cross-section. The modeling is based on the elastic beam theory, and all the modes of buckling are considered in the analytical solution. The solution and analytical expressions account for axial forces in tension and compression. The modeling considers the buckling with the third mode, which occurs at a certain limit of the axial compression and leads to significant variation in the stiffness of the actuator. This phenomenon is well studied for bistable pre-shaped beams, but related studies are limited for beam electrothermal actuators. As a case study, the V shape actuator is specifically investigated. The modeling shows very good agreement with finite element simulations and experimental data based on micro-machined in-plane silicon actuators. [2020-0262]
25 March 2022 in Neuromorphic Computing and Engineering
We present a magnetic tunnel junction (MTJ) based implementation of the spike time-dependent (STDP) learning for pattern recognition applications. The proposed hybrid scheme utilizes the spin–orbit torque (SOT) driven neuromorphic device-circuit co-design to demonstrate the Hebbian learning algorithm. The circuit implementation involves the (MTJ) device structure, with the domain wall motion in the free layer, acting as an artificial synapse. The post-spiking neuron behaviour is implemented using a low barrier MTJ. In both synapse and neuron, the switching is driven by the SOTs generated by the spin Hall effect in the heavy metal. A coupled model for the spin transport and switching characteristics in both devices is developed by adopting a modular approach to spintronics. The thermal effects in the synapse and neuron result in a stochastic but tuneable domain wall motion in the synapse and a superparamagnetic behaviour of in neuron MTJ. Using the device model, we study the dimensional parameter dependence of the switching delay and current to optimize the device dimensions. The optimized parameters corresponding to synapse and neuron are considered for the implementation of the Hebbian learning algorithm. Furthermore, cross-point architecture and STDP-based weight modulation scheme is used to demonstrate the pattern recognition capabilities by the proposed neuromorphic circuit.