Magnetic storage of digital data is now possible at densities approaching 1 Terabit per square inch at a cost of only about a tenth of a cent per Megabit. To a large extent, the breathtaking progress in this area of technology is sustained by discovery of bits. The invention of “GMR" sensors based on stacks of ultra-thin films of magnetic metals (for which the Nobel Prize in Physics was awarded in 2007) is a perfect example. In these devices the ability to "pin" the magnetic orientation of one film in the stack while allowing another to respond freely to a magnetic field is a key principle. This "magnetic pinning", or "exchange pinning" is achieved by layering a simple magnet with a more complex magnetic material called an antiferromagnet. This pinning effect discovered over 50 years ago, has been used in devices for over a decade, but is poorly understood. One of the major reasons for this is that the pinning effect is due to atomic defects in the antiferromagnet. Experiments are hindered by the fact that these defects are difficult to probe, while theoretical research is hampered by the fact that it is difficult to incorporate such defects in calculations. Recently, IRG-3 students Jyo Saha, Mike Lund, and Mun Chan, working with postdoc Jeff Parker and IRG faculty Chris Leighton, Randy Victora, and Paul Crowell, have performed one of the most detailed studies of this exchange pinning in materials very similar to those used in hard disks. Most importantly, the experimental results were directly compared to realistic micromagnetic simulations. These supercomputer simulations break the sample up into hundreds of thousands of tiny magnetic elements (see figure), and are sufficiently powerful that the all-important defects can be properly accounted for. This approach has enabled the researchers to understand important factors such as the influence of layer thickness and grain size, complex magnetic switching, and even a mysterious effect where the exchange pinning deteriorates over time, a major problem for applications.