World’s First! Fudan University Team Unveils Breakthrough 2D–Silicon Hybrid Flash Memory Chip

Fudan University’s National Key Laboratory of Integrated Chips and Systems has achieved a groundbreaking milestone in global semiconductor research. Led by Professors Zhou Peng and Liu Chunsen from the School of Integrated Circuits and Micro-Nano Electronics Innovation Institute, the team developed the “Changying (CY-01)” architecture, integrating the 2D ultrafast flash memory device “PoX (Pox)” with mature silicon-based CMOS technology. This marks the world’s first 2D–silicon hybrid architecture flash memory chip, with the findings published in Nature on October 8 (Beijing time). In the era of big data and artificial intelligence, computing systems demand unprecedented data access speed and efficiency. Traditional memory technologies, constrained by speed and power consumption, have become major barriers to computational performance. Earlier this year, in April, Zhou and Liu’s team published the “PoX” 2D flash memory prototype in Nature, demonstrating 400-picosecond ultra-high-speed non-volatile storage — the fastest semiconductor charge storage technology to date — laying the theoretical foundation to overcome computing bottlenecks. Yet the central challenge remained: how to move from “LAB to FAB” — from experimental breakthroughs to industrial production. Determined to accelerate this transition, the Fudan team engaged deeply with the semiconductor industry. “Among all 2D electronic devices, memory is the most promising for industrialization,” said Professor Zhou. “It requires no extreme manufacturing conditions, yet its performance far exceeds today’s commercial standards, opening possibilities for disruptive applications.” Recognizing CMOS (Complementary Metal-Oxide-Semiconductor) as the dominant global manufacturing process for integrated circuits, the team focused on integrating the 2D ultrafast flash memory into existing CMOS workflows — accelerating innovation while enhancing conventional silicon capabilities. Over five years of experimentation, the researchers overcame significant technical challenges. While conventional chips are built from silicon layers hundreds of micrometers thick, 2D semiconductor materials exist at the atomic scale, less than one nanometer in thickness. Integrating such ultrathin, fragile materials with complex CMOS structures — filled with varying surface heights like a miniature city — was extremely difficult. “If you look at Shanghai from space, it appears flat,” Zhou explained. “But inside the city, there are buildings hundreds of meters high. Laying an atomic-thin film over that uneven surface would immediately cause tearing or distortion.” Introducing 2D materials into CMOS manufacturing also risked contaminating production lines, a concern unacceptable to foundries worldwide. To solve this, the team innovated a modular integration method — instead of altering CMOS, they adapted to it. They fabricated the 2D memory circuit and CMOS control circuit separately, then combined them using high-density monolithic interconnects at the micron scale. This approach enabled atomic-level bonding between the 2D material and CMOS substrate, achieving a record-breaking 94.3% chip yield. The integrated chip supports 8-bit instruction operations, 32-bit high-speed parallel processing, and random access, with performance surpassing existing flash memory technologies. For the first time, the world has witnessed the engineering realization of a 2D–silicon hybrid architecture. Building on this, the team introduced a comprehensive cross-platform design framework, including co-design and interface methodologies for 2D–CMOS integration, naming it the “Changying (CY-01)” architecture. “This is a source technology for China’s integrated circuit industry,” Zhou emphasized. “It gives us strategic control in the next generation of memory core technologies.” Looking ahead, the Fudan team envisions this 2D–silicon hybrid flash memory as a transformative foundation for future computing — replacing multi-layered memory hierarchies with a faster, lower-power, universal memory solution. As artificial intelligence and big data applications expand, the team believes 2D flash memory will become the standard storage architecture of the AI era, redefining the future of global semiconductor and computing ecosystems.