Magnetoresistance is a property of certain materials in which the electrical resistance changes in response to an applied magnetic field. Giant magnetoresistance, in which there is a marked drop in resistance, was first observed in thin-film materials two decades ago. TMR is a quantum mechanical effect, discovered in 1995, in a sandwich of two ferromagnetic films separated by a thin insulated tunnel barrier. Electrons cross from one electrode to the other by quantum mechanical tunnelling through the barrier, the effective resistance of which changes with an applied field.

The first TMR demonstrations were observed at low temperatures in ferromagnetic sandwiches that required complex fabrication. Research effort has therefore focused on creating a TMR junction that functions at room temperature and has the potential for industrial fabrication. Papers #2 and #3 describe crucial technical advances that will move the devices out of the lab and into computers. Both papers report very large changes in resistance at room temperature: 220% in #2 and 180% in #3.

Stuart Parkin (IBM Almaden Research Center, California) and his co-workers scored a major technical improvement by replacing an amorphous alumina tunnel barrier with a pure MgO (100) crystalline barrier, the latter formed by sputtering. The Japanese group directed by Shinji Yuasa (NanoElectronics Research Institute, Tsukuba, Japan) used molecular beam technology to fabricate a similar crystalline interface. Both groups used Fe/MgO/Fe junctions in a follow-up to theoretical work that had already suggested that high values of TMR could be realized with a crystalline rather than an amorphous barrier.

The high citations for both papers reflect the importance of the discoveries for the achievement of a new generation of memory. The dynamic RAM in a computer relies on the storage of electric charge that must be topped up continuously to combat leakage. Not only is this is a drain on the battery in a laptop, but the computer operating system itself has to be reloaded from disk every time the computer is shut down. So, the IT industry would dearly like a non-volatile memory that does not require endless rewriting. Although flash memory works fine for small devices, it is far too slow for serious computers. Hot Papers #2 and #3 point the way to a new contender for non-volatile memory: magnetic random-access memory (MRAM).

MRAM is realized by having an array of magnetic cells just like the ferromagnetic sandwiches described in #2 and #3. The electron spins in the two ferromagnetic layers can be parallel or antiparallel, which allows one bit of information to be stored. The direction of the spins is controlled by magnetic fields. Physicists envisage that TMR can be used for readout of MRAM. There is so much promise here of nanosecond read and write speeds, and low power consumption, that industry analysts consider MRAM will be a universal memory technology. Parkin’s team points out that their junction, being sputter deposited, is readily manufacturable and highly stable thermally. They speculate that these features will accelerate the development of new families of spintronic devices.

In astronomy and cosmology, meanwhile, the Sloan Digital Sky Survey (SDSS) continues to supply highly cited papers, such as #4, #6, #8, and #9. In newcomer #4, Uros Seljak’s Princeton-led collaboration examine data from SDSS, from the Wilkinson Microwave Anisotropy Probe, and from the latest supernova searches. These complimentary data sets are used to set important constraints on cosmological models. The analysis is performed in the context of current models for structure formation in the universe, without any exotic physics. The result is a significant reduction in the errors for several cosmological parameters. While there are no surprises, the resulting tightness of constraint rules out many alternative cosmological models.