Chinese researchers have developed a method that extends the writing endurance of an emerging memory material to more than 10 billion cycles.
The result addresses a major reliability problem that has limited the use of wurtzite ferroelectrics in advanced memory devices. The research was carried out by scientists from Xidian University, City University of Hong Kong and Fudan University.
The team focused on wurtzite ferroelectrics, a group of materials that can hold one of two electrical states. These states can be used to represent data in a memory device. One material attracting particular attention is aluminium scandium nitride (AlScN).
AlScN has several properties that make it interesting for future semiconductor applications. It can switch states quickly and has the potential to operate with low energy use. The material can also work with established semiconductor manufacturing processes, which may make integration into memory hardware easier.
READ ALSO: https://modernmechanics24.com/post/gosun-solar-tractor-ditches-diesel/
The main problem has been durability during repeated writing. Conventional AlScN-based devices have generally reached failure after about 100 million writing cycles. That level of endurance is well below the billions of cycles expected from practical high-performance memory applications.
The researchers identified nitrogen vacancies as a major factor behind this degradation. A nitrogen vacancy is a point in the material where a nitrogen atom is missing. These defects can move through the material as electrical switching is repeated.
Controlling Atomic Defects
The team found that the movement of nitrogen vacancies was more important than simply the number of defects present. Over repeated switching, the vacancies can migrate and gather in particular areas of the material. Their accumulation can eventually create routes through which electrical current leaks.
Wang Ruiqing, a doctoral researcher at Xidian University and one of the study’s authors, compared the material to a planted cornfield.
In the comparison, nitrogen vacancies are like spaces where seedlings are missing. The researchers found that the problem develops when these empty spaces move and collect together.
The study examined this process at the atomic level. Researchers had previously observed that ferroelectric devices deteriorated after repeated operation. The new work identified how nitrogen vacancies move and how their movement contributes to electrical leakage and eventual failure.
The researchers then designed a layered structure to restrict the movement of the vacancies. The structure limits their ability to travel through the material and prevents them from gathering into damaging pathways. This approach slowed the deterioration of the ferroelectric material.
More Than 10 Billion Cycles
The modified material demonstrated more than 10 billion writing cycles during the researchers’ tests. That is roughly 100 times the endurance previously reported for the same type of material. The result represents a significant increase in the number of times the memory can be repeatedly written before its performance deteriorates.
WATCH ALSO: https://modernmechanics24.com/post/ai-robots-physical-ai-real-world/
The research was published in the journal Science on Thursday. It was led by Xidian University in the northwestern Chinese city of Xi’an, with researchers from City University of Hong Kong and Fudan University also taking part.
The findings remain at the laboratory research stage. The researchers have demonstrated improved endurance in the material, but further work is needed to determine how the approach performs in complete memory systems and under practical operating conditions.
Implications For Future Chips
The work comes as semiconductor researchers seek memory technologies that can handle demanding computing workloads. Artificial intelligence systems require large amounts of data to be moved and stored, placing greater pressure on memory performance, power consumption and reliability.
Ferroelectric memory is being studied as one possible option for future computing hardware. Its ability to switch between electrical states at high speed gives it potential for applications where fast and repeated data operations are required. Improving its endurance addresses one of the main technical obstacles to wider adoption.
The researchers’ method also offers a way to approach the reliability problem through material design. Instead of simply accepting nitrogen vacancies as an unavoidable source of degradation, the team focused on controlling their movement inside the material.
READ ALSO: https://modernmechanics24.com/post/largest-electric-ferry-south-america/
The more than 10 billion-cycle result does not by itself establish commercial readiness. Manufacturing at scale, long-term stability, device performance and compatibility with complete semiconductor systems will still need to be evaluated.
If the approach continues to perform well in larger and more complex devices, controlling atomic defects may help move wurtzite ferroelectric memory closer to practical use in future high-performance and AI-focused hardware.
