Narrow Heater Bottom Electrode‐Based Phase Change Memory as a Bidirectional Artificial Synapse

Abstract

Phase change memory can provide a remarkable artificial synapse for neuromorphic systems, as it features excellent reliability and can be used as an analog memory. However, this approach is complicated by the fact that crystallization and amorphization differ radically: crystallization can be realized in a very gradual manner, very similarly to synaptic potentiation, while the amorphization process tends to be abrupt, unlike synaptic depression. Addressing this non‐biorealism of amorphization requires system‐level solutions that have considerable energy cost or limit the generality of the approach. This work demonstrates experimentally that an adaptation of the memory structure associated with an initialization electrical pulse followed by a sequence of identical fast pulses can overcome this challenge. A single device can then naturally implement gradual long‐term potentiation and depression, much like synapses in biology. This study evidences through statistical measurements the reproducibility of the approach, discusses its physical origin, as well as the importance of the device architecture and of the initial electrical pulse. Through the use of system‐level simulation, it is shown that this device is especially adapted to a neuroscience‐inspired learning. These results highlight how nanodevices can be suitable for bioinspired applications while retaining the qualities of industrial technology.

Effect of Electrode Material on the Crystallization of GeTe Grown by Atomic Layer Deposition for Phase Change Random Access Memory

“…The JMA kinetics are applied to examine the behavior of the crystalline growth of each device shown in Figure 2c [9]. For this analysis, the crystallization fraction, x ( t ), of the phase-changing material, is assumed to be related to the resistance of the memory cell at a certain time t ( R ( t )) according to Equation (1).…”

“…PCRAM has been seriously considered as a new storage-class memory and can be used for 3D Xpoint memory owing to its high read/write speed, high density, and nonvolatile characteristics [4,5]. In addition, PCRAM can provide a feasible artificial synapse for neuromorphic hardware systems due to its multilevel memory functionality and high reliability [6,7,8,9]. Despite the significant improvement made in the device fabrication technique and material innovation during the past decades, a too high operation current has been the major hurdle for high-density low-power devices.…”

Bridging Biological and Artificial Neural Networks with Emerging Neuromorphic Devices: Fundamentals, Progress, and Challenges

“…[78,79] In order to achieve gradual conductance switching, several approaches have been proposed, e.g., using two PCM per synapse to realize potentiation and depression respectively, one PCM per synapse with carefully tuned programming pulses. [81] Another issue with PCM is the conductance drift, or relaxation, which is frequently observed especially for the amorphous state (i.e., off-state). [81] Another issue with PCM is the conductance drift, or relaxation, which is frequently observed especially for the amorphous state (i.e., off-state).…”

Emerging Memristive Artificial Synapses and Neurons for Energy‐Efficient Neuromorphic Computing

“…The transition between the amorphous and crystalline phase is generally considered as the underlying switching principle for phase change memristive synapses ( Figure 2c). [40,[77][78][79][116][117][118] This synapse consists of a phase change material (e.g., Ge 2 Sb 2 Te 5 (GST)) between two conducting electrodes and can be programmed by the degree of the phase change that is modified by electrical current-driven heat. [118,119] Notably, several structures contain a heater surrounded by an insulator on the bottom electrode to concentrate the current and confine the phase change region.…”