Collaborative Research: FuSe: Indium selenides based back end of line neuromorphic accelerators

合作研究：FuSe：基于硒化铟的后端神经形态加速器

• 批准号: 2328742
• 负责人: Priyadarshini Panda
• 金额: $ 40万
• 依托单位: Yale University
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-10-01 至 2026-09-30
• 项目状态: 未结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=2328742&HistoricalAwards=false
• 关键词: Collaborative; Research; FuSe; Indium; selenides

This project aims to use innovative materials called “2D materials” to enhance the capabilities of modern integrated circuits. These materials have unique electronic properties that make them very promising for compute, storage, and sensing technologies. However, integrating them with existing silicon-based technology has been a challenge due to temperature restrictions. Luckily, a new group of materials called “indium-based chalcogenides” offers a solution, as they can be synthesized at low temperatures compatible with current technology. The project team plans to create a range of devices using these materials to accelerate the performance of energy-efficient spiking neural networks (SNNs). These brain-inspired microchips will revolutionize how audio, visual, tactile, and olfactory information is processed, making devices smarter and more responsive. Moreover, these microchips could be used in autonomous vehicles, drones, and robots, helping them navigate and avoid obstacles. The project also focuses on training the next generation of scientists and engineers and promoting diversity and inclusivity in the field.This project aims to address the challenge of integrating novel 2D materials with the state-of-the-art silicon-based complementary metal oxide semiconductor (CMOS) technology at the back end of line (BEOL). The key innovation lies in leveraging indium-based chalcogenides, such as InSe and In2Se3, which can be synthesized at low temperatures, making them compatible with BEOL processes. The team plans to synthesize and characterize these materials to fabricate an array of sensing, encoding, computing, and memory devices for hardware acceleration of energy-efficient spiking neural networks (SNNs). The project will involve a cross-layer co-optimization approach that encompasses material discovery, synthesis and deposition techniques, process flow development, and device-circuit-architecture co-design. The goal is to develop brain-inspired SNN microchips through 2D/CMOS heterogeneous and monolithic integration, which will lead to substantial reductions in energy consumption and pave the way for sustainable computing paradigms. The broader impact of this work extends to applications on the Internet of Things (IoT) domain, where the brain-mimetic SNN microchips will enable advanced audio, visual, tactile, and olfactory information processing. Additionally, the project emphasizes education and training, promoting diversity and inclusiveness in the workforce.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.

该项目旨在利用被称为“2D材料”的创新材料来增强现代集成电路的能力。这些材料具有独特的电子性能，使它们在计算、存储和传感技术方面非常有前途。然而，由于温度的限制，将它们与现有的基于硅的技术相结合一直是一个挑战。幸运的是，一组名为“铟基硫化物”的新材料提供了一种解决方案，因为它们可以在与当前技术兼容的低温下合成。该项目团队计划创造一系列使用这些材料的设备，以加快节能尖峰神经网络(SNN)的性能。这些受大脑启发的微芯片将彻底改变音频、视觉、触觉和嗅觉信息的处理方式，使设备更智能、更灵敏。此外，这些微芯片可以用于自动驾驶车辆、无人机和机器人，帮助它们导航和避开障碍物。该项目还专注于培训下一代科学家和工程师，促进该领域的多样性和包容性。该项目旨在解决在生产线后端(BEOL)将新型2D材料与最先进的硅基互补金属氧化物半导体(CMOS)技术相结合的挑战。关键的创新在于利用了铟基硫化物，如InSe和In2Se3，它们可以在低温下合成，使它们与BEOL工艺兼容。该团队计划合成和表征这些材料，以制造一系列传感、编码、计算和存储设备，用于节能尖峰神经网络(SNN)的硬件加速。该项目将涉及一种跨层联合优化方法，包括材料发现、合成和沉积技术、工艺流程开发以及设备-电路-架构联合设计。其目标是通过2D/CMOS异质和单片集成来开发受大脑启发的SNN微芯片，这将导致能源消耗的大幅降低，并为可持续计算范例铺平道路。这项工作的更广泛影响延伸到物联网(IoT)领域的应用，在物联网领域，模拟大脑的SNN微芯片将实现高级音频、视觉、触觉和嗅觉信息处理。此外，该项目强调教育和培训，促进劳动力的多样性和包容性。该奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。