Research Uncovers Innovative Computer Memory Design to Cut Down on Energy Usage

A new computer memory design has been developed by researchers that has the potential to enhance performance and reduce energy consumption in internet and communication technologies. These technologies are predicted to consume almost 30% of the world’s electricity in the next decade.

The study was published in the journal ‘Science Advances’.

A team led by the University of Cambridge created a device that processes information in the same way as synapses in the human brain. The devices are made of hafnium oxide, a material already used in the semiconductor industry, and small self-assembled barriers that can be raised and lowered to allow electrons to pass through.

This method of changing the electrical resistance of computer memory devices and allowing computing and memory to coexist could lead to the development of computer memory devices with significantly higher density, higher performance and lower energy consumption. The results were published in the journal Science Advances.

Our data-hungry world has led to an increase in energy demand, which makes reducing carbon emissions even more difficult. In the next few years, artificial intelligence, internet usage, algorithms and other data-driven technologies are expected to consume more than 30 percent of the world’s electricity.

“This explosion in energy demand is largely due to shortcomings in current computer memory technologies,” said first author Dr Markus Hellenbrand from Cambridge’s Department of Materials Science and Metallurgy. “Traditional computing has memory on one side and processing on the other, and data is shuffled back and forth between the two, which takes both energy and time.”

One possible solution to the problem of inefficient computer memory is a new type of technology known as resistive switching memory. Traditional memory devices are capable of two states: one or zero. However, a functioning resistive switching memory device would be capable of a continuous state region – computer memory devices based on this principle would be capable of much higher density and speed.

“A typical continuous-range USB stick could hold, say, 10 to 100 times more data,” Hellenbrand said.

Hellenbrand and his colleagues developed a prototype device based on hafnium oxide, an insulator material already used in the semiconductor industry. A problem associated with the use of this material in resistive switching memory applications is known as the uniformity problem. At the atomic level, hafnium oxide has no structure, and hafnium and oxygen atoms mix randomly, making it challenging to use in memory applications.

However, the researchers found that by adding barium to the thin hafnium oxide films, unusual structures began to form in the composite material perpendicular to the hafnium oxide plane.

These vertical “bridges” containing barium are highly structured and allow electrons to pass through, while the surrounding hafnium oxide remains unstructured. At the point where these bridges meet the contacts of the device, an energy barrier was created that the electrons could cross. The researchers were able to control the height of this barrier, which in turn changes the electrical resistance of the composite material.

“This allows multiple states to exist in the material, as opposed to traditional memory, which only has two states,” Hellenbrand said.

Unlike other composite materials that require expensive high-temperature fabrication methods, these hafnium oxide composites self-assemble at low temperatures. The composite material showed high performance and uniformity, making them very promising for next-generation memory applications.

Cambridge Enterprise, the university’s commercialization arm, has applied for a patent for the technology.

“What’s really exciting about these materials is that they can act like a synapse in the brain: they can store and process information in the same place as our brains can, which makes them very promising for the rapidly growing fields of artificial intelligence and machine learning,” said Hellenbrand. (LETTER I)