The internet and communications technologies, which are anticipated to consume nearly a third of all electricity worldwide within the next ten years, could greatly benefit from a new design for computer memory that has been developed by researchers.
A device that processes data in the same way that synapses in the human brain do was developed by the researchers, who were led by the University of Cambridge. Small self-assembling barriers that can be raised or lowered to permit electrons to pass are the foundation of the devices, which are based on hafnium oxide, a material that is already in use in the semiconductor industry.
Computer memory devices with significantly higher density, higher performance, and lower energy consumption could be made possible by employing this technique to alter the electrical resistance of memory devices and permit information processing and memory to coexist there. The findings are published in Science Advances.
“In traditional computing, there is memory on one side and processing on the other, and data is shuffled back and forth between the two, which consumes both energy and time.”
Dr. Markus Hellenbrand, from Cambridge’s Department of Materials Science and Metallurgy.
Our data-hungry world has increased energy consumption, making it harder and harder to cut down on carbon emissions. It is anticipated that the use of artificial intelligence, the internet, algorithms, and other data-driven technologies will account for more than 30% of global electricity consumption within the next few years.
Dr. Markus Hellenbrand, who is the first author and works in the Department of Materials Science and Metallurgy at the University of Cambridge, stated, “To a large extent, this explosion in energy demands is due to the shortcomings of current computer memory technologies.” Data is shuffled back and forth between processing and memory in conventional computing, which necessitates both energy and time.
A novel type of technology known as resistive switching memory may offer a potential solution to the issue of inefficient computer memory. There are two states that conventional memory devices can reach: zero or one. However, a computer memory device based on the principle of resistive switching would be capable of far greater density and speed than a working resistive switching memory device.
Hellenbrand stated, “For instance, a typical USB stick based on continuous range would be able to hold between 10 and 100 times more information.”
A prototype device was created by Hellenbrand and his colleagues using hafnium oxide, an insulating material that is already utilized in the semiconductor industry. The uniformity problem arises when using this material for resistive switching memory applications. Hafnium oxide is difficult to use for memory applications because it lacks structure at the atomic level and has oxygen and hafnium atoms mixed at random.
However, when barium was added to thin films of hafnium oxide, the researchers discovered that the composite material began to develop unusual structures that were perpendicular to the hafnium oxide plane.
While the surrounding hafnium oxide remains unstructured, electrons can pass through these vertical barium-rich “bridges,” which are highly structured. An energy barrier that electrons can cross was erected where these bridges meet the device contacts. The researchers were able to control this barrier’s height, which affected the composite material’s electrical resistance.
Hellenbrand stated, “This allows multiple states to exist in the material, unlike conventional memory, which only has two states.”
These hafnium oxide composites, in contrast to other composite materials, which require costly high-temperature manufacturing processes, self-assemble at low temperatures. The composite material’s high performance and uniformity make it extremely promising for applications in next-generation memory.
The University’s commercialization arm, Cambridge Enterprise, has applied for a patent on the technology.
“The fact that these materials can function like a synapse in the brain is really exciting: According to Hellenbrand, “they can store and process information in the same place, like our brains can,” making them extremely promising for the rapidly expanding fields of AI and machine learning.
In order to gain a better understanding of how the high-performance structures are formed, the researchers are currently collaborating with industry to carry out larger feasibility studies on the materials. Since hafnium oxide is a material previously utilized in the semiconductor business, the scientists say it wouldn’t be challenging to integrate into existing assembling processes.
More information: Markus Hellenbrand et al, Thin-film design of amorphous hafnium oxide nanocomposites enabling strong interfacial resistive switching uniformity, Science Advances (2023). DOI: 10.1126/sciadv.adg1946
