from – 8/19/2007 11:00:00 PM – Semiconductor International

IBM Corp. (Armonk, N.Y.) and TDK Corp. (Tokyo) announced a four-year joint program to develop next-generation MRAM memories based on the spin momentum transfer effect.

The effort brings together IBM’s MRAM effort and TDK’s expertise at developing magnetic tunnel junction-based devices. TDK makes MTJ-based read/write heads in large volumes for the hard disk drive industry, manufacturing at 150 mm wafer fabs in Japan and California.

For IBM, the effort is aimed at adding a high-performance non-volatile memory to the ASICs it produces for foundry customers, said Bill Gallagher, senior manager of exploratory non-volatile memories at IBM’s T.J. Watson Research Center in Yorktown Heights, N.Y.

“We don’t have a non-volatile offering, and our customers would like it. But it doesn’t make sense unless it can be high-performance and high-endurance, to do some of what DRAM does and also be non-volatile. Without some sort of non-volatile memory, we don’t have a complete system,” Gallagher said. MRAM probably would not be used in IBM’s computer-use microprocessors, which do not require a non-volatile memory, he added.

Craig Cocchi, vice president of business development at TDK’s U.S. subsidiary, said TDK has a research group based in Milpitas, Calif., that will work with the IBM researchers. “TDK has been developing magneto tunnel junction technologies for many years. The tunnel junction devices TDK makes for the HDD industry are very similar to the MTJ elements in MRAMs,” he said, adding that TDK has “quite a large patent portfolio” in the field.

While IBM’s goal is to integrate MRAM into ASICs, TDK’s target is less specific. The joint press release refers to applications in the automotive, cell phone, handheld computing, and industrial controls markets. Asked if TDK plans to make standalone MRAMs, Cocchi declined to comment, but added, “That would be a new area for us.”

TDK, with sales exceeding $8B in its latest fiscal year, earlier worked with IBM to develop giant magneto resistive (GMR) read/write heads. “These are research areas that both IBM and TDK have been working on. We both see ways to use spin-polarized current to scale down the size, and increase the density, of MRAMs,” Cocchi said.

Gallagher said, “We worked with TDK on magnetic sensor devices years ago, and the good feeling between the two companies helped us in the early discussions for this MRAM relationship.”

IBM earlier worked with Infineon Technologies to co-develop MRAM technology, ending that joint-development relationship in 2004 after the companies developed a 16 Mb demonstration vehicle.

Spin torque MRAMs use the transfer of spin angular momentum on a ferromagnetic conductor. Gallagher said next-generation spin-torque MRAMs are likely to use tunnel junctions based on magnesium oxide (MgO), which provides a larger signal margin than today’s material stack. IBM researchers are working on other magnetic materials that are likely to come into use after MgO, further improving power and performance, he said.

“Spin torque will help us as we go to smaller dimensions. We do need to work on the materials in order to lower the write power. That is why we have materials development as a fundamental component of the collaboration with TDK,” Gallagher said.

The magnetics research program at the Almaden Research Laboratory (San Jose), headed up by IBM researcher Stuart Parkin, is using a magnetic materials evaluation system, developed over the past few years by Parkin and his colleagues.

Powered by ScribeFire.