128 0

Enhanced Analog Switching Behavior of a-Si / Dielectric Bilayer Memristor Device by Ion Implantation

Enhanced Analog Switching Behavior of a-Si / Dielectric Bilayer Memristor Device by Ion Implantation
Alternative Author(s)
Issue Date
2021. 2
Artificial intelligence (AI), the core of the fourth industrial revolution, is developing at the unprecedented rate. In the near future, the current Von Neumann based computing system, however, will face challenges in terms of processing speed and energy consumption as the amount of data increase. Neuromorphic computing is being widely studied as an alternative to tackle these issues. In particular, emerging devices for artificial neuron and synapse as basic building blocks are being actively studied for the development of energy-efficient neuromorphic computing systems. Among various emerging devices, memristor is being considered as a promising artificial synaptic device for information transmission. Conductive bridging random access memory (CBRAM) has been considered to be a promising emerging device for artificial synapses in neuromorphic computing systems. Good analog synaptic behavior, such as linear and symmetric synapse updates, are desirable to provide high learning accuracy. Although numerous efforts have been made to develop analog CBRAM for years, the stochastic and abrupt formation of conductive filaments hinders its adoption. In this study, we propose a novel approach to enhance the synaptic behavior of a SiNx/a-Si bilayer memristor through Ge implantation for the purpose of suppressing a strong filaments and inducing multiple weak filaments. The SiNx and a-Si layers serve as switching and internal current limiting layers, respectively. Ge implantation induces structural defects in the bulk and surface regions of the a-Si layer, enabling spatially uniform Ag migration and nanocluster formation in the upper SiNx layer and increasing the conductance of the a-Si layer. As a result, the analog synaptic behavior of the SiNx/a-Si bilayer memristor, such as the nonlinearity, on/off ratio, and retention time, is remarkably improved. In order to improve the synaptic performances of the bilayer memristor, a further study on optimizing material combinations of an active metal (Ag, Cu, Ni) and a switching layer (SiOx, SiNx) was conducted. Depending on the combination of materials, characteristic behaviors in terms of synapse parameters were observed. The Ag/SiOx memristor exhibited only a short-term memory behavior compared with the Ag/SiNx device. The Cu/SiOx memristor showed a long-term memory comparable to the Ag/SiNx device. The Cu/SiNx and Ni/SiOx devices did not show reliable switching behaviors revealed by quasi-static DC sweep experiments. All the synapse parameters of Ag/SiOx and Cu/SiOx were extracted and compared with the Ag/SiNx device. The cross-sectional TEM images showed clearly that the metal nanoclusters were formed in the switching layers and play a main role in the gradual switching. We discussed the morphology and distributions of the metal nanoclusters in the switching layers in correlation with the synaptic behaviors. Lastly, an MNIST pattern recognition simulation was performed based on a memristor-based neural network with consideration of the synapse update model, which was experimentally determined in potentiation and depression under repetitive identical pulses for the devices of Ag/SiNx/(unimplanted or implanted) a-Si, Cu/SiOx/implanted a-i. The neuromorphic system with the Ag/SiNx/implanted a-Si showed a highest learning accuracy of 88.3 %, whereas the Ag/SiNx/unimplanted a-Si system only showed an accuracy of 62.8 %. The enhanced learning accuracy of the implanted a-Si memristor based system is mainly attributed to the improved synapse linearity in potentiation and depression.
Appears in Collections:
Files in This Item:
There are no files associated with this item.
RIS (EndNote)
XLS (Excel)


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.