IBM and Stanford's spintronics revolution
Electron spin is cool
IBM and Stanford University have announced a major research drive in the field of Spintronics, a new field of nanotechnology that researchers hope will turn electronics on its head, in the same way the invention of the transistor did, 50 years ago.
As more of the ever smaller transistors get packed on to every chip, the power consumption and overheating problems increase. Moving current around just generates too much heat, and the faster you want to move it, the more power you need, and the hotter things get.
The joint venture between Stanford and IBM is called "IBM-Stanford Spintronic Science and Applications Center" or SpinAps, for short. The deal is quite simple: IBM provides the seed money, Stanford supplies additional research brains, and they split the intellectual property (IP).
Spintronics provides an alternative to the traditional die method of processor manufacturing, by sidestepping the issue of current altogether. It focuses instead on the spin, or magnetic alignment, of individual electrons. An electron's spin can be in one of two states: either up or down, and it is this binary nature that makes it the subject of a whole lot of computer science research.
It is possible to control the magnetism of a material by aligning the spins of its consituent electrons, and applying magnetic fields to the material affects the movement of electrons differently, depending on their spin. The aim of the joint venture is to fully understand and control these properties. This, the scientists hope, will lead to properties like low power switching, and non-volatile storage.
If the team manages to make it work, the potential is enormous. It could lead to even smaller and more powerful processors, without the need for a heatsink the size of Wales.
In a statement on the company's webiste, IBM's boffins burble excitedly about creating "new materials and devices with entirely new capabilities - such as reconfigurable logic devices, room-temperature superconductors and quantum computers". Now that is a bit of IP worth having, even if commercial products are not expected to appear for another five years. ®