Stanford researchers have established a new phase-change memory that could help computers process large amounts of data faster and more efficiently.
We are tasking our computers with processing ever-increasing amounts of data to increase drug discovery, improve train artificial intelligence, weather and climate predictions, and much more. We need faster and more energy-efficient computer memory, to keep up with this demand.
Enhancing Computing Efficiency
Presently computers store and process data in separate locations. Volatile memory – handles the processing and it’s fast but disappears when your computer turns off. Nonvolatile memory – which isn’t that fast but can hold information and takes care of the long-term data storage, without constant power input.Shifting information between these two locations can cause bottlenecks while the processor waits for huge amounts of data to be retrieved.
It takes a lot of energy to shuttle data back and forth, mostly with the current computing workloads, says Xiangjin Wu, co-lead author on the paper and a doctoral candidate co-advised by Pop and Philip Wong, the Willard R. and Inez Kerr Bell Professor in the School of Engineering. With this type of memory, we’re hoping to bring the memory and processing closer together, eventually into one device, so that it uses less energy and time.
There are various technical hurdles to achieving an effective, commercially viable universal memory that is capable of both long-term storage and fast, low-power processing without sacrificing other metrics, but the new phase change memory that developed in Pop’s lab is as close as anyone has come so far with this technology. The scientists hope that it will encourage further development and adoption as a universal memory.
The Promise of GST467 Alloy
The memory relies on GST467, an alloy of four parts germanium, six parts antimony, and seven parts tellurium, which collaborators at the University of Maryland developed. Pop and his colleagues found ways to sandwich the alloy between many other nanometer-thin materials in a superlattice, a layered structure they’ve previously used to achieve good nonvolatile memory results.
The unique composition of GST467 gives it a particularly fast switching speed, says Asir Intisar Khan, who earned his doctorate in Pop’s lab and is co-lead author of the paper. Integrating it within the superlattice structure in nanoscale devices allows low switching energy, gives us good endurance, and very good stability, and makes it nonvolatile – it can retain its state for 10 years or longer.
Setting a New Bar
The GST467 superlattice clears many important benchmarks. Phase change memory can sometimes drift over time – essentially the value of the ones and zeros can slowly shift – but according to their tests, it shows that this memory is very stable. It also functions at below 1 volt, which is the goal for low-power technology and is much faster than a typical solid-state drive.
A few other types of nonvolatile memory can be a bit faster, but they operate at higher voltage or higher power, says Pop. With all these computing technologies, there are tradeoffs between energy and speed. The fact that we’re switching at a few tens of nanoseconds while functioning below one volt is a big deal.
Into a small space, the superlattice also packs a good amount of memory cells. The scientists have shrunk the memory cells down to 40 nanometers in diameter – which is less than half the size of a coronavirus. That’s not quite as dense as it could be, but scholars are exploring ways to compensate by stacking the memory in vertical layers, which is possible thanks to the superlattice’s low fabrication temperature and the techniques used to produce it.
The fabrication temperature is well below what you need, says Pop. To increase density, people are thinking about stacking memory in thousands of layers. This type of memory can enable such future 3D layering.
