Novel Materials for AI Chip - Two-Dimensional Materials and the Third-Generation Semiconductors

AI chips have a rising demand for high computing power and low power consumption.Traditional silicon-based semiconductors are constrained by inherent properties and process limits, making it difficult to meet the highperformance requirements of AI chips.Therefore the research and exploration of novel materials are particularly significant.This paper focuses on the currently popular two-dimensional (2D) materials and the third-generation semiconductors, deeply analyzing the material properties and comparing four key parameters: bandgap width, electron mobility, thermal conductivity, and breakdown voltage.It is found that graphene has ultra-high electron mobility and thermal conductivity, which can help solve the challenges of high-speed computing and thermal management in AI chips.Molybdenum disulfide with adjustable bandgap width and variable breakdown voltage has excellent switching characteristics, making it suitable for chip memory modules.GaN and SiC can adapt to high-power and high-frequency scenarios by their wide bandgap, high electron mobility and high breakdown voltage characteristics.The four new materials each have their own advantages in physical and electrical properties, demonstrating great application potential in the field of AI chips.If novel materials can overcome the obstacles in manufacturing processes and production costs, they are expected to promote the development of AI chips in the future.