光子集成电路
Search documents
美军事研究人员与工业界接洽寻求用于信号处理的超大规模光子电路
Xin Lang Cai Jing· 2026-01-27 03:21
(来源:中国航空报) 美国国防高级研究计划局(DARPA)于1月14日发布了"可扩展系统目标光子集成电路架构"(PICASSO)项目的招标 书。PICASSO旨在革新光子电路架构,其应用范围涵盖从计算到光探测和测距等各个领域,通过从单个组件扩展到集 成微电子系统,并在系统层面衡量性能。 尽管光子集成系统有助于提升计算、模拟信号处理和传感领域的带宽、延迟和能效,但如今的光子电路仍存在局限 性,难以超越传统电子器件的性能。电路尺寸和功能扩展主要受两大限制:一是保持光信号完整性并最大限度降低噪 声;二是杂散波相互作用限制了电路的可预测性。 (逸文) 美国军方研究人员正与业界接洽,寻求开发用于计算、模拟信号处理和传感等领域的新型超大规模光子电路。 目前,系统设计人员通过电子方式转换和重构光信号来应对这些挑战,但大量使用电子器件阻碍了系统级延迟、效率 和带宽的提升,而光子器件本身就能解决这些问题。寻找性能理想的组件并不能解决扩展性受限的问题。 PICASSO项目将着重强调四大支柱:通过测试不同的应用场景并排除单一解决方案来提高通用性;通过定义和强制执 行接口边界条件并将其记录在接口控制文档中来提高互操作性;通过要求交付 ...
硅光子技术与激光器集成进展(下)
势银芯链· 2025-10-16 08:11
Core Insights - The article discusses the advancements and commercialization efforts of micro-transfer printing (μTP) technology, which is crucial for the integration of III-V materials on silicon substrates, enhancing photonic integrated circuits [2][5][6]. Group 1: μTP Technology Overview - μTP technology was developed by researchers at the University of Illinois in 2004 and has seen significant advancements, including the ability to transfer individual devices onto target photonic chips [2]. - The process involves using a PDMS stamp to pick and transfer devices, ensuring high alignment accuracy and allowing for large-scale parallel integration [3]. - μTP enables the integration of different materials on a common substrate without modifying the backend processes of silicon photonics [3]. Group 2: Commercialization Efforts - The INSPIRE project, running from 2021 to March 2025, focuses on commercializing wafer-level μTP technology, aiming to combine InP and SiN photonics on a single platform [5]. - Key partners in the INSPIRE project include Eindhoven University of Technology, imec, Smart Photonics, and Cambridge University, among others [5]. Group 3: Integration Challenges - The integration of III-V materials on silicon faces challenges due to lattice mismatch, which can lead to defects affecting the reliability of semiconductor lasers [6]. - Successful applications of single-chip methods in quantum dot devices have been reported, indicating potential for improved laser reliability [6]. Group 4: Upcoming Conference - TrendBank plans to host a conference from November 17-19, 2025, focusing on heterogeneous integration technologies, aiming to foster collaboration between industry and academia [9]. - The conference will cover advanced packaging technologies, including multi-material integration and photonic-electronic co-packaging [9].
AMD收购硅光芯片公司
半导体行业观察· 2025-05-29 01:15
Core Viewpoint - AMD's acquisition of Enosemi aims to accelerate the development of co-packaged optical devices for AI systems, enhancing its position as a full-stack AI provider [1][2][3]. Group 1: Acquisition Details - AMD announced the acquisition of Enosemi, a startup focused on photonic integrated circuits, to boost its capabilities in optical device development for AI [1]. - The financial terms of the acquisition have not been disclosed, and Enosemi was founded in 2023 with 16 employees and 11 investors [1]. - Enosemi's products include integrated photodetectors and custom silicon chips, indicating a strong product portfolio that complements AMD's existing technologies [1]. Group 2: Technological Implications - Photonic integrated circuits utilize photons for data transmission, offering high speed and low power consumption, which is crucial for applications in healthcare, automotive, and communications [1]. - The demand for faster data transmission is increasing with the complexity of AI models, making co-packaged optical devices essential for next-generation AI systems [2][4]. - AMD's strategy includes integrating its leading CPU, GPU, and SoC technologies with enhanced networking, software, and system integration expertise to support advanced AI workloads [2][4]. Group 3: Future Outlook - The acquisition is expected to enhance AMD's capabilities in supporting various photonic and co-packaged optical solutions for AI systems [3][4]. - AMD aims to provide a comprehensive innovation stack that encompasses computing, networking, system architecture, and software, positioning itself uniquely in the AI market [4].