Improving Upon the Design of Microelectronic Devices

Andrew Corselli

A new study led by researchers at the University of Minnesota Twin Cities is providing new insights into how next-generation electronics, including memory components in computers, breakdown or degrade over time. Understanding the reasons for degradation could help improve efficiency of data storage solutions.

Advances in computing technology continue to increase the demand for efficient data storage solutions. Spintronic magnetic tunnel junctions (MTJs) — nanostructured devices that use the spin of the electrons to improve hard drives, sensors, and other microelectronics systems, including Magnetic Random Access Memory (MRAM) — create promising alternatives for the next generation of memory devices.

MTJs have been the building blocks for the non-volatile memory in products like smart watches and in-memory computing with a promise for applications to improve energy efficiency in AI.

Using a sophisticated electron microscope, researchers looked at the nanopillars within these systems, which are extremely small, transparent layers within the device. The researchers ran a current through the device to see how it operates. As they increased the current, they were able to observe how the device degrades and eventually dies in real time.

“Real-time transmission electron microscopy (TEM) experiments can be challenging, even for experienced researchers,” said First Author Hwanhui Yun, Ph.D., Postdoctoral Research Associate in the University of Minnesota’s Department of Chemical Engineering and Material Sciences. “But after dozens of failures and optimizations, working samples were consistently produced.”

By doing this, they discovered that over time with a continuous current, the layers of the device get pinched and cause the device to malfunction. Previous research theorized this, but this is the first time researchers have been able to observe this phenomenon. Once the device forms a “pinhole” (the pinch), it is in the early stages of degradation. As the researchers continued to add more and more current to the device, it melts down and completely burns out.

“What was unusual with this discovery is that we observed this burn out at a much lower temperature than what previous research thought was possible,” said Senior Author Andre Mkhoyan, Professor and Ray D. and Mary T. Johnson Chair in the University of Minnesota Department of Chemical Engineering and Material Sciences. “The temperature was almost half of the temperature that had been expected before.”

Looking more closely at the device at the atomic scale, researchers realized materials that small have very different properties, including melting temperature. This means that the device will completely fail at a very different time frame than anyone has known before.

“There has been a high demand to understand the interfaces between layers in real time under real working conditions, such as applying current and voltage, but no one has achieved this level of understanding before,” said Senior Author Jian-Ping Wang, Distinguished McKnight Professor and Robert F. Hartmann Chair in the Department of Electrical and Computer Engineering at the University of Minnesota.

The researchers hope this knowledge can be used in the future to improve design of computer memory units to increase longevity and efficiency.

Here is an exclusive Tech Briefs interview with Mkhoyan, edited for length and clarity.

Tech Briefs: What was the biggest technical challenge you faced while trying to observe this pinhole within a device and how it degrades in real time?

Mkhoyan: There were two big issues. One is to prepare the device so it will be possible to put in a microscope and have it be electron transparent so that you can see things happening there on an atomic level. This means it needs to be around 50 to 70 nanometers thick, no more than that. Then, you need to have that device fully functioning inside the microscope. You need to have contacts and all of that to be sure that device operates as it would on a regular wafer.

Then, once all of that is done, you’ll operate your microscope so that you'll see the atomic-level details and you will not miss anything.

Tech Briefs: How did this project come about? What was the catalyst for your work?

Mkhoyan: We have these devices that sometimes function and then they don't. They just don't work, or die really quickly. That brought up the question of, ‘It’s not only these devices having this issue, almost every nanoscale device you make, they're going to die. Can we study how these things die?’

People know before and after; before it's functional, after it's completely dead. But, they never had a good understanding on the specifics — what starts the degradation, how that degrades, and how it eventually burns out. And you need to do that on an atomic level because some of these devices are small, so there's no other way around. That made us get on this path and try to help the entire field of nanotechnology with better diagnostics and how you can understand and directly visualize high degrades and quantify it. Then, once you know how to do that, you can design new devices accordingly. You will minimize these issues.

Tech Briefs: Can you explain in simple terms how you went about the process and the obstacles you faced?

Mkhoyan: Challenge number one was to prepare a device so it will be electron transparent. We spent quite a bit of effort to try to figure it out. Keep in mind these devices are about 100 nanometers in size and sometimes it's 50 nanometers. That's 50-100 nanometers sitting on a piece of wafer. So, identifying a sample and cutting and preparing a TM-friendly sample was a big challenge. My postdoc spent a year or so just getting that one sorted out.

You need to do it in a way that you won’t harm it. So, the way we did it was by using gallium-iron beams — digging from the sides and cutting a piece and then take it out.

Then, you need to have it connected. So, you need to prep the sample in a way where you cut the piece and they still have functional electrodes coming out that you could couple it. Then these electrodes need to be connected to the power supply, which is outside of the microscope. So, we need these things to somehow be connected — these tiny electrodes that are also electron transparent to physically connect to our supply outside of the microscope. We had a specialized holder that provides that linking holder for a TM sample. But we designed the entire structure.

A third challenge was doing actual measurements. You connected it, let's say, and tested some of them, and it is all working now. Now, you want to test it in a way that will be closest to the natural working conditions and then do it in a way while you are still running a microscope on an atomic level, so you can see the atoms moving around.

Tech Briefs: What are your next steps on this?

Mkhoyan: Since we figured out how to develop this method and how to make our samples, we're already working on a different projects with this setup. So, we plan to use the different devices and different type of contacts. We also filed a patent for how to prepare this setup.

---