The relentless shrinking of silicon components has led to exponential improvements in chip performance, but we’re starting to hit physical limits. Now researchers have developed a way to integrate materials just 10 atoms thick into conventional chips.
For decades, rapid advances in miniaturization meant the number of transistors on a microchip doubled approximately every two years, a phenomenon dubbed Moore’s law. But as these components started reaching dimensions of just a few nanometers, progress started to stall.
This left researchers and chip companies casting about for new ways to squeeze computing power into ever smaller spaces. So-called “2D materials” are a promising way forward. These crystalline structures are just a few atoms thick and exhibit exceptional electronic capabilities.
So far, it’s been challenging to integrate such exotic materials into conventional electronics. But now researchers at Fudan University in China have created a chip that combines a memory core made of the 2D material molybdenum disulfide (MoS₂) with CMOS circuits.
“This work provides a promising technical pathway to bring promising 2D electronics concepts to real-world applications,” the authors write in a paper about the new process published in Nature.
The main reason it’s been hard to combine 2D materials and standard chips is that the rough surface of conventional silicon circuits prevents them from adhering evenly and can damage their atomically thin layers.
To get around this, the researchers developed a fabrication method they call ATOM2CHIP, which introduces an ultra-smooth glass layer between the 2D material and the silicon. This provides both a mechanical buffer and a way to electrically isolate the MoS₂ layer from the electronics.
The team used the new method to create a flash memory array composed of a 10-atom-thick MoS₂ layer stacked on a 0.13-micrometer CMOS platform responsible for transmitting instructions to program, read, and erase the memory.
The chip could program bits in 20 nanoseconds and consumed just 0.644 picojoules per bit—significantly less energy than conventional flash memory. An accelerated aging test showed it could also retain data for more than 10 years at 55 degrees Celsius. Programming accuracy was only 93 percent, which is well below what you’d expect from a commercial chip but still promising for an early prototype.
Kai Xu at King’s College London, told New Scientist the ultrathin design may also help solve a long-standing problem in silicon electronics—signal leakage. Transistors work by using a “gate” to control when current flows through a channel, but as they get smaller it’s easier for current to slip through that barrier.
This means they are never truly off, which leads to extra power consumption and noise that can interfere with nearby signals. But the physics of 2D materials mean transistors made with them have much more effective gates, providing an almost perfect on/off switch.
“Silicon has already hit obstacles,” said Xu. “The 2D material might be able to overcome those effects. If it’s very thin, the control at the gate can be more even, can be more perfect, so there’s less leakage.”
One significant challenge for the approach is that the glass layer central to the technique is not yet compatible with standard fabrication lines. “This is a very interesting technology with huge potential, but still a long way to go before it is commercially viable,” Steve Furber at the University of Manchester told New Scientist.
Nonetheless, the work suggests that if we want to kickstart Moore’s law, we may be better off abandoning the search for ever smaller transistors and instead focus on ever thinner chips.
