<p>This paper presents a novel non-volatile plasmonic memory based on a Metal–Insulator-Metal (MIM) waveguide configuration integrated with a phase-change material (PCM). In this architecture, silver (Ag) is utilized as the plasmonic metal, while air and <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\({\text{G}\text{e}}_{2}{\text{S}\text{b}}_{2}{\text{T}\text{e}}_{5}\)</EquationSource> </InlineEquation> (GST) serve as the dielectric media. The Drude model is employed to characterize the complex dielectric function of the metal, and the overall performance is verified using two-dimensional Finite-Difference Time-Domain (2D-FDTD) simulations. To achieve optimal optical performance alongside highly compact dimensions, a single-mode plasmonic filter was initially designed. In the amorphous state of the GST material, this filter exhibits a well-defined Gaussian resonance mode with a high transmittance of 84%. By integrating Programming, Select, and Probe optical signals into the filter geometry, a multi-functional non-volatile plasmonic memory is realized. This specialized architecture enables efficient in-plane (in-circuit) programming and precise cell addressing within a large-scale memory array. The proposed device successfully executes all-optical Write, Read (logic ‘0’ and ‘1’), and Erase operations. A comprehensive signal isolation analysis has been conducted for the input ports to ensure minimal cross-talk. The device demonstrates a remarkable contrast ratio of 84% between the two material phases and an outstanding Extinction Ratio (ER) of 43.66 dB, guaranteeing high discriminability between logic states. Additionally, the difference in Insertion Loss (IL) between reading bit ‘1’ and bit ‘0’ is 19.69 dB, facilitating a high storage density of 11 bits/<InlineEquation ID="IEq2"> <EquationSource Format="TEX">\({\upmu \text{m}}^{2}\)</EquationSource> </InlineEquation>. Finally, temporal analysis was performed to validate the dynamics during the logical-state writing operation. The proposed memory architecture holds significant potential for application in next-generation Photonic Integrated Circuits (PICs) and all-optical neuromorphic computing architectures.</p>

错误:搜索内容不能为空,请输入英文关键词
错误:关键词超出字数限制,请精简
高级检索

Design of non-volatile plasmonic memory using phase-change material as a neuromorphic synapse

  • Melina Kehtarmanesh,
  • Parviz Keshavarzi,
  • Mohammad Danaie

摘要

This paper presents a novel non-volatile plasmonic memory based on a Metal–Insulator-Metal (MIM) waveguide configuration integrated with a phase-change material (PCM). In this architecture, silver (Ag) is utilized as the plasmonic metal, while air and \({\text{G}\text{e}}_{2}{\text{S}\text{b}}_{2}{\text{T}\text{e}}_{5}\) (GST) serve as the dielectric media. The Drude model is employed to characterize the complex dielectric function of the metal, and the overall performance is verified using two-dimensional Finite-Difference Time-Domain (2D-FDTD) simulations. To achieve optimal optical performance alongside highly compact dimensions, a single-mode plasmonic filter was initially designed. In the amorphous state of the GST material, this filter exhibits a well-defined Gaussian resonance mode with a high transmittance of 84%. By integrating Programming, Select, and Probe optical signals into the filter geometry, a multi-functional non-volatile plasmonic memory is realized. This specialized architecture enables efficient in-plane (in-circuit) programming and precise cell addressing within a large-scale memory array. The proposed device successfully executes all-optical Write, Read (logic ‘0’ and ‘1’), and Erase operations. A comprehensive signal isolation analysis has been conducted for the input ports to ensure minimal cross-talk. The device demonstrates a remarkable contrast ratio of 84% between the two material phases and an outstanding Extinction Ratio (ER) of 43.66 dB, guaranteeing high discriminability between logic states. Additionally, the difference in Insertion Loss (IL) between reading bit ‘1’ and bit ‘0’ is 19.69 dB, facilitating a high storage density of 11 bits/ \({\upmu \text{m}}^{2}\) . Finally, temporal analysis was performed to validate the dynamics during the logical-state writing operation. The proposed memory architecture holds significant potential for application in next-generation Photonic Integrated Circuits (PICs) and all-optical neuromorphic computing architectures.