Page 26

EETE JAN 2015

MEMORY & DATA STORAGE Universal memory for instant-on computing By R. Colin Johnson Combo magnetoelectric memory technology - based on a multiferroic called bismuth ferrite - uses 10 times less energy to switch bits than its closest next-generation contender (spin-torque memory). The nonvolatile magnetic memory can be switched by a voltage potential alone - with no current flowing - making it a strong contender for replacing the entire memory hierarchy for instant-on operation in computers, according to its inventors at Cornell University in New York City. Its biggest caveat right now is fatigue - it can only be switched a few times before failing - but the team is gung-ho that they can lick that problem. Darrell Schlom, a professor in the Department of Materials Science and Engineering, told EE Times: “This multiferroic memory could substitute for all sorts of volatile and nonvolatile memory, because of its lower energy consumption. It isn’t simply replacing the active material in another memory technology with bismuth ferrite, but rather replacing the entire memory technology with a bismuth ferritebased memory technology. This is because the readout scheme makes use of the coupling between the bismuth ferrite multiferroic with the overlying spin valve structure (Pt/Co0.9Fe0.1/Cu/ Co0.9Fe0.1).” Schlom’s collaborators at Cornell include physics professsor Dan Ralph, postdoctoral associate John Heron and his doctoral advisor Ramamoorthy Ramesh, at the University of California, Berkeley who first demonstrated in 2003 that bismuth ferrite can store bits in extremely thin films with enhanced properties compared with its bulk counterparts. The fact that no current flows when switching a bit’s state and its low-voltage operation, also makes multiferroic the coolest running of the next-generation memory technologies. If only the fatigue problem can be solved. “The biggest challenge to commercialization is fatigue - that is, extending the number of times you can switch it and it still works - right now it keeps working just a few times like in the early days of FRAMs. In fact, scientists had given up on FRAMS in the 1960s, and only came back to them when Ramesh came up with a solution to the fatigue problem”, Schlom told us. The team is counting on Ramesh’s experience with solving the fatigue problem with FRAMs, to solve their fatigue problem with bismuth ferrite. Multiferroics are different from FRAMs or MRAMs, which depend on a single mechanism to switch bits. But when you reverse the voltage potential across a multiferroic film, the magnetic pole gets dragged along with the reversing electrical dipole - giving you the best of both worlds. Easy readouts - since the magnetic pole changes the resistance of the circuit - plus easy writing, because no current has to flow to make the switch. History Multiferroics have been the subject of experimentation before, but the fatigue problems turned most engineers to other materials that were harder to switch, but at least didn’t fatigue for many thousands of switches, such as flash (which endures about 10,000 switches without fatigue). The Cornell team, however, has built a single-bit thin-film switch that works at room temperature with lower power than any other next-generation memory contender, making it worth the trouble of solving the fatigue problem. Once solved, the team aims to measure how scalable the memory cells will be. If they can reach the multi-Gbit range, then the technology will probably be licensed out to foundries, or purchased by one of the big memory houses. Other collaborators worked on this project at the University of Connecticut, University of California at Berkeley, Tsinghua University (Beijing) and the Swiss Federal Institute of Technology in Zurich. Funding was provided by the National Science Foundation and the Kavli Institute at Cornell. Artist’s rendering of magnetization reversal with an electric field (blue) applied across the gold capacitors. The compass needles under the electric field are rotated 180 degrees from those not under the field. (Image: Cornell) 26 Electronic Engineering Times Europe January 2015 www.electronics-eetimes.com


EETE JAN 2015
To see the actual publication please follow the link above