KAIST researchers unlock secrets to next-gen ReRAM memory

Every time you tap, click, or scroll, your device stores and processes data with memory. Memory technology today, however, is getting to its limits. It's becoming tougher to make it faster, more efficient, and reliable to accommodate the growing demand for speed and data.

That is where ReRAM—Resistive Random Access Memory—comes into action. The novel memory is created from oxide material. It provides great speed, long retention time of data, and far less complex structure than the traditional memory. Because of that, researchers all over the world regard it as one of the strongest candidate types to upstage old technologies like flash memory.

KAIST researchers have taken a giant leap towards realizing how this new technology works in the first place. For the first time ever, they have actually seen visually what happens inside the ReRAM when writing or erasing data. This breakthrough could change the memory technology and neuromorphic computing future.

ReRAM differs from regular memory in that it stores data not just temporarily. Like flash memory, it also holds information when the power is shut down. But ReRAM does it faster and with less equipment, which could lead to devices that are smaller and consume less power in the future.

As the central element of ReRAM's functionality is a thin oxide material layer—titanium dioxide (TiO₂), to be precise. When an electric signal is applied, something miraculous happens: the oxide defects—oxygen vacancies—start to move. These defects are regions where there is a preferential lack of one oxygen atom, and they play a significant role in whether or not current can pass.

When many oxygen vacancies come together, they create a path in which electricity can travel freely. But when the vacancies are scattered, the flow of current is severed. This difference—between conducting and obstructing current—is what gives ReRAM its ability to retain the "on" and "off" states of memory.

Before this, scientists had primarily learned about this process from indirect speculation or computer models. But the KAIST researchers figured out how to see these changes.

Using an instrument called a multi-modal scanning probe microscope (multi-modal SPM), they watched the ReRAM device function at the nano-scale. This advanced device is made up of three types of microscopes with one body. Each has a different purpose:

All these combined gave the researchers a clear idea of what happens when ReRAM writes and deletes data.

By applying an electric signal to a TiO₂ film, they were able to see directly the way oxygen vacancies started and moved. This demonstrated that minute changes were behind the device changing its conducting and non-conducting states.

Study leader Professor Seungbum Hong called it, "This is an example that proves we can directly observe the spatial correlation of oxygen defects, ions, and electrons through a multi-modal microscope."

This ability to see the process in real-time is all it takes. It turns a theory into a fact.

The researchers also discovered something important about the dependability of memory. During their experiments, they found that if oxygen ions were dynamically inserted into the oxide layer during the "reset" or erasure process, the device stayed off longer periods of time. In technical terms, this means staying in a "high resistance state."

This is important news because one of the largest problems with memory design is keeping data stable over a period of time. If a memory toggles states when it shouldn't, then it will result in data loss or system error. But if ReRAM can reliably stay off—or on—when it needs to, then it's a much better prospective choice for future technology.

Even better still, the researchers did not just test a tiny piece of material. They saw how these alterations occurred on a larger surface area—several micrometers in breadth. Having that wide view helped support the fact that it's not merely the oxygen defects that matter, but how electrons behave in their presence. It's the interaction between the two that shapes the performance of the memory.

Professor Hong reported that the findings offer "a new chapter in the research and development of various metal oxide-based next-generation semiconductor devices."

So, what's the overall picture? Memory technologies like ReRAM potentially have a gigantic effect on everything from smartphones and PCs to artificial intelligence. In fact, neuromorphic computing—a discipline that tries to build computer systems based on the human brain—relies strongly on advanced types of memory.

That's because neuromorphic processors need memory that can both compute and store data extremely quickly with barely a whisper of power. ReRAM is perfect for that because it operates a lot like the manner in which neurons spike within your brain.

The work carried out at KAIST brings us one step closer to being able to create that kind of technology. To developers, it offers suggestions about how they could design memory devices not just fast, but also stable and power-efficient. To manufacturers, it tells them what material properties or design techniques could lead to better performance.

While progress toward totally replacing older memory technology isn't complete, this finding provides one of the most straightforward blueprints to achieve that.

Research findings are available online in the journal ACS Applied Materials and Interfaces.

Like these kind of feel good stories? Get The Brighter Side of News' newsletter.