Chinese scientists have achieved a significant breakthrough by developing the world`s first two-dimensional flash memory chip. This innovation utilizes complementary metal-oxide-semiconductor (CMOS) technology, featuring silicon material of atomic thickness. This development promises a substantial increase in data processing speed and energy efficiency, particularly for artificial intelligence systems, and could establish a new standard in memory technology.
The chip was created by a research team from Fudan University, and their findings were published in the esteemed scientific journal Nature.
This groundbreaking invention represents a hybrid 2D silicon-based flash chip that successfully integrates ultra-fast flash memory with a CMOS structure. The chip exhibits an impressive memory cell performance of 94.3%, supports 8-bit command operations, 32-bit high-speed parallel operations, and random access. Its operational speed significantly surpasses that of current flash memory technologies, marking it as the first engineering realization of such a hybrid 2D silicon flash chip.
The escalating demand for faster and more energy-efficient data access has been driven by the rapid advancements in artificial intelligence. Existing technologies often face limitations in data transfer speeds and incur high power consumption.
In contrast to conventional silicon chips, which typically have wafer thicknesses ranging from hundreds of microns down to tens of nanometers, these two-dimensional semiconductor materials boast an atomic thickness—less than 1 nanometer.
Professor Zhou Peng, head of the research group, highlighted that 2D semiconductors, as a novel material class, are currently not employed in any integrated circuit manufacturing facilities globally.
The researchers intend to collaborate with technology companies to further develop and implement this project. They are optimistic that their innovative technology will fundamentally transform traditional memory architecture, enabling quicker and more power-efficient processing of vast amounts of data for AI and big data applications.

