PORTLAND, Ore. -- A Japanese team says it has extended a high-performance perpendicular tunneling magneto-resistance (TMR) process to non-volatile logic devices, claiming that device dimensions of 40 nanometers correspond to logic chips with 8 Gbits of nonvolatile memory.
"When measured in terms of commodity RAM, the size of our device makes it possible to realize an 8-Gbit chip with the use of vertical transistors," said professor Hideo Ohno at Tokohu University. "When these devices are made on top of vertical transistors, the cell dimension can be made 4F2--this means that you can accommodate 8 Gbits in a 100-mm2 chip assuming 50 percent cell area."
The tunneling magneto-resistance effect is an extension of spin-valve giant magnetoresistance effect used to record on hard disks. Used as a spin valve, the electron spins are oriented vertically to a thin insulating tunnel barrier, allowing "perpendicular recording" of high-area densities. Since electrons must tunnel--that is, pop out on one side of an insulator and pop back on the other side--TMR is a quantum effect unexplained by Newtonian physics. Similar magnetic tunneling junctions also enable MRAM memories and various reprogrammable magnetic logic devices.
The new perpendicular logic device process was created by Ohno and other researchers at Tohoku University in cooperation with Hitachi Ltd. Their technique makes use of a large perpendicular magnetic anisotropy at the interface between the insulator and its magnetic electrode. As a result, the compact vertical architecture enables bit cells to measure just four times the square of device dimensions--rather than 64 times the square of device dimensions. The results was a 16-fold reduction in the size of logic circuitry using TMR.
According to the researchers, their approach to TMR simultaneously achieves high density, low write current (49 microAmps), a tunneling-magneto-resistance ratio of 124 percent and a thermal resistance of 350 ļ C. That makes the process compatible with the standard fabrication techniques used for logic ICs.
The researchers also report that unlike competing architectures, their devices will be cheap to manufacture (since they use inexpensive noble metals) while promising ultra-low power consumption for system-on-chip applications.
Funding was provided by the Japan Society for the Promotion of Science and by the "High-Performance Low-Power Consumption Spin Devices and Storage Systems" program at Tohoku University.
It should be interesting to see which electronic device based on any quantum effect of one or more electrons or maybe atoms makes it into the main stream. After the proof of concept of course comes the integration, or, maybe the start of a new industry. There are not many quantum engineers yet but in time they will spin out from the universities.