Engineers develop new magnetoelectric computer memory

(—By using electric voltage instead of a flowing electric current, researchers from UCLA's Henry Samueli School of Engineering and Applied Science have made major improvements to an ultra-fast, high-capacity class of computer memory known as magnetoresistive random access memory, or MRAM.

The UCLA team's improved , which they call MeRAM for magnetoelectric , has great potential to be used in future for almost all electronic applications, including smart-phones, tablets, computers and microprocessors, as well as for , like the solid-state disks used in computers and large data centers.

MeRAM's key advantage over existing technologies is that it combines extraordinary low energy with very high density, high-speed reading and writing times, and non-volatility—the ability to retain data when no power is applied, similar to hard disk drives and flash memory sticks, but MeRAM is much faster.

Currently, magnetic memory is based on a technology called spin-transfer torque (STT), which uses the of electrons—referred to as spin—in addition to their charge. STT utilizes an electric current to move electrons to write data into the memory.

Yet while STT is superior in many respects to competing memory technologies, its electric current–based write mechanism still requires a certain amount of power, which means that it generates heat when data is written into it. In addition, its is limited by how close to each other bits of data can be physically placed, a process which itself is limited by the currents required to write information. The low bit capacity, in turn, translates into a relatively large cost per bit, limiting STT's range of applications.

With MeRAM, the UCLA team has replaced STT's electric current with voltage to write data into the memory. This eliminates the need to move large numbers of electrons through wires and instead uses voltage—the difference in electrical potential—to switch the magnetic bits and write information into the memory. This has resulted in that generates much less heat, making it 10 to 1,000 times more energy-efficient. And the memory can be more than five-times as dense, with more bits of information stored in the same physical area, which also brings down the cost per bit.

The research team was led by principal investigator Kang L. Wang, UCLA's Raytheon Professor of Electrical Engineering, and included lead author Juan G. Alzate, an electrical engineering graduate student, and Pedram Khalili, a research associate in electrical engineering and project manager for the UCLA–DARPA research programs in non-volatile logic.

"The ability to switch nanoscale magnets using voltages is an exciting and fast-growing area of research in magnetism," Khalili said. "This work presents new insights into questions such as how to control the switching direction using voltage pulses, how to ensure that devices will work without needing external magnetic fields, and how to integrate them into high-density memory arrays.

"Once developed into a product," he added, "MeRAM's advantage over competing technologies will not be limited to its lower power dissipation, but equally importantly, it may allow for extremely dense MRAM. This can open up new application areas where low cost and high capacity are the main constraints."

Said Alzate: "The recent announcement of the first commercial chips for STT-RAM also opens the door for MeRAM, since our devices share a very similar set of materials and fabrication processes, maintaining compatibility with the current logic circuit technology of STT-RAM while alleviating the constrains on power and density."

The research was presented Dec. 12 in a paper called "Voltage-Induced Switching of Nanoscale Magnetic Tunnel Junctions" at the 2012 IEEE International Electron Devices Meeting in San Francisco, the semiconductor industry's "pre-eminent forum for reporting technological breakthroughs in the areas of semiconductor and electronic device technology."

MeRAM uses nanoscale structures called voltage-controlled magnet-insulator junctions, which have several layers stacked on top of each other, including two composed of magnetic materials. However, while one layer's magnetic direction is fixed, the other can be manipulated via an electric field. The devices are specially designed to be sensitive to electric fields. When the electric field is applied, it results in voltage—a difference in electric potential between the two magnetic layers. This voltage accumulates or depletes the electrons at the surface of these layers, writing bits of information into the memory.

"Ultra-low–power spintronic devices such as this one have potential implications beyond the memory industry," Wang said. They can enable new instant-on electronic systems, where memory is integrated with logic and computing, thereby completely eliminating standby power and greatly enhancing their functionality."

Explore further

Magnetic random-access memory based on new spin transfer technology achieves higher storage density

Citation: Engineers develop new magnetoelectric computer memory (2012, December 14) retrieved 18 August 2019 from
This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.

Feedback to editors

User comments

Dec 14, 2012
nice to see that those technologies we talked about in 2000 and were supposed to come about in 5 years, finally getting close to the market, micron also brings phase change memory, its about dram time!

Dec 14, 2012
"Ultra-low–power spintronic devices such as this one have potential implications beyond the memory industry," Wang said. They can enable new instant-on electronic systems, where memory is integrated with logic and computing, thereby completely eliminating standby power and greatly enhancing their functionality."

Will be interesting to see what robot programmers and video game developers can do with that, if and when they ever understand it.

Aside from that, I'm wondering about the timing too. This has been in development for 11 or 12 years now. Heck, I remember talking about this on the old OpenTechSupport site back when it was first mentioned. I think 40GB HD and a few hundred megs of RAM was a big deal back then...

Dec 14, 2012
I won't say no to a terabyte of MeRAM in my next computer. Hopefully it will be cheap to produce. Although I wonder how long it retains its memory before a refresh is required.

Dec 15, 2012
Just recently they invented a better way to have light waves directly detectable at the 90 nanometer scale on the chip. So I guess they will use the photovoltaic effect to emit an electron (causing a voltage change) so that the change in voltage to interface to MeRAM, and then somehow have the emitted electron get reabsorbed, once per clock cycle. Then you could have a computer with NO CURRENT FLOW ? Nothing but changing energy levels in the computer. That would be somehow zero energy consumption, other than the energy required to load the registers at the beginning of a clock cycle, and the energy to detect the registers at the end of a clock cycle. But I'm a computer programmer who knows nothing about electronics! So I mayt be wrong.

Dec 15, 2012
. Although I wonder how long it retains its memory before a refresh is required.

As noted in teh article: it's non-volatile (i.e. it will retain memory indefinitely). Which is good from a power perspective, but very bad from a security perspective.
That would be somehow zero energy consumption,

No, because you're still inducing phonons and radiating light. The power consumption is much lower, but you won't ever get a zero power computer.

Dec 16, 2012
@antialias, what I was thinking was emitting and reabsorbing photons, but I accidentally said 'electron'. So you're right, it didn't make sense. My main "thinking" at the time was that this is cool because voltage requires no actual flow of electrons. :( I shouldn't comment when I'm half asleep! lol.

Dec 18, 2012
The old magnetic core memory used current and retained the data when power was removed and then this was abandoned for other higher density methods and now we go back to improved magnetic memory decades later. Who would have guessed?
And by the way when you change the potential voltage on anything there is indeed an unavoidable flow of current until the voltage is stable. (capacitive reactance)

Dec 18, 2012
@unknownorigin, your last sentence is true only in a closed circuit where current can flow. It is possible to have a voltage potential between two objects that have never even touched each other before. Only once the two objects with relative potential make contact will you see any current flow. If the new type of circuit can sense voltage differences, then it is possible for that sensing to happen with no current flow at all (theoretically). Just think of a resistor with almost total (infinite resistance), as an example. If this circuit is constantly sensing voltage differentials for sending digital 0s and 1s for example that could mean that the 1 is flowing one way and the 0 is flowing the other way, and you would see a net flow of zero electrons over even short periods of time, yet information would be getting transmitted. Kinda reminds me of AC versus DC in a way, but that's different I know. DC has a constant flow, but AC uses less current because of a similar push/pull.

Dec 18, 2012
BTW, no flow of electrons has to be happening as the negative side of a capacitor collects a surplus of electrons (or charge), while the other has a net positive charge (lack of electrons). I'm sure you know how capacitors work. The key thing, related to the article, is how much current MUST flow for their Voltage sensing gate to do its switching.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more