A problem for widespread adoption of MRAM, or Magnetic RAM, has been capacity at a reasonable cost, but the partnership between IBM and TDK aims to change that with the new spin momentum transfer technology, which reduces the cell sizes in MRAM chips, thereby increasing MRAM capacities while maintaining MRAM's nonvolatility benefits.
A smaller, high-capacity, nonvolatile memory cell is the intended offspring of a joint research and development endeavor announced Sunday by IBM and TDK. The companies said they will use "spin momentum transfer" technology to develop high-capacity Magnetic RAM.
MRAM technology, which uses magnetic fields instead of electrical charges for memory storage, offers several advantages over other memory technologies. Like flash memory, nonvolatile MRAM memory does not need continual electrical power to keep its data stored. But MRAM is faster than flash, has lower power requirements, and provides unlimited read-write cycles.
A key problem for widespread adoption of MRAM, however, has been capacity at a reasonable cost.
'Best of All Worlds'
According to the companies, the new spin momentum transfer technology will enable MRAM chips to have smaller cell sizes, thereby increasing memory capacity while maintaining the low power, endurance, and nonvolatility advantages. Such fast, nonvolatile memory could find uses in cell phones and other mobile devices, cars, industrial controls, and other markets.
One type of MRAM is spin-transfer torque RAM, or STT-RAM. This type of memory could be competitive against flash memory within a few years, according to Santa Clara, California-based Grandis, a company that has been a pioneer in this field. President and CEO Farhad Tabrizi has written that the STT-RAM technology offers "the best of all worlds," in that it combines nonvolatility, scalability, and endurance with low power requirements and fast read-write times.
In STT-RAM, an electrified magnet is used to change the direction of a magnetic field in a storage layer, such as up becoming down or left becoming right. This change results in a change in resistance, and the resistance levels can be read as the ones and zeroes needed to store data.
New Approach Needed
With all of the advantages that a higher-capacity, reasonable-cost MRAM might have, some observers have suggested it might lead to the holy grail of a single memory technology to replace the many kinds available now. But Gartner analyst Martin Reynolds predicted that, while there might eventually be fewer memory technologies, possibly replacing flash or even DRAM, there will never be only one kind of memory.
Both IBM and TDK are well-positioned to launch a new approach to MRAM. TDK officials noted that magnetic materials have been its core technology since 1935, and that it has been a leader in applying magnetic tunnel junction, or MTJ, technology for hard drive recording heads. IBM has been a pioneer in the R&D of magnetic tunnel junction technology and of the spin momentum transfer effect.
The research work will be conducted at IBM's TJ Watson Research Center in Yorktown Heights, New York, IBM's Almaden Research Center in San Jose, California, IBM's ASIC Design Center in Burlington, Vermont, and TDK's subsidiary R&D Center in Milpitas, California.