Albert Fert and Peter Gruenberg, for the discovery of giant magnetoresistance. This is one of those prizes that’s pretty well-deserved, because giant magnetoresistance is now applied for data readout on pretty much every magnetic hard drive being produced, such as those in one’s iPod. It’s not really my field of study, but let me take a stab at explaining what giant magnetoresistance is. I’ll do that by first discussing the application: reading and writing to computer/iPod hard drives. Odds are I’m going to screw up the field directions in all the pictures, but hopefully the arguments will be sound…
A computer hard drive essentially consists of a disc of magnetic material, and data is stored on that disc by modifying the magnetic properties to create ’0′ and ’1′ bits. It is perhaps most helpful to imagine the disc as a large collection of microscopic bar magnets, with north and south poles. If we define a ’0′ as the north pole pointing to the left, and a ’1′ as the north pole pointing to the right, we have a technique for encoding data, as illustrated below:
The arrows in the picture indicate the directions of the magnetic fields in each of the bar magnets. If you’re wondering why these magnets don’t push each other around (i.e. two adjacent north poles pushing away from one another), the answer is: that can be a problem, but is a fixable one.
Changing a bit on the disc is simply a matter of applying a magnetic field to the ‘bar magnet’ to force it to flip to the opposite direction. This is typically done via Ampere’s law: a changing electric current is used to produce a magnetic field in the write-head of the disc:
Some fields from the write-head ‘leak’ into the data layer, forcing the bar magnet in that layer to realign itself.
Once the data is written to the disc, how do we read it again? This problem is more challenging. We could essentially reverse the write process, by letting the magnetic field of the data induce electric currents in the write-head, i.e. Faraday’s law. Our data bits are extremely small, however, and produce a very small magnetic field, which means very little current is produced and the readout is subject to errors.
Instead of using Faraday’s law, we can use one of several phenomena collectively referred to as magnetoresistance to detect the magnetic field of our data. Electrical resistance may be loosely considered to be a measure of how ‘easily’ electrons travel through a given material. As electrons travel through a material, they interact with the atoms of the material. The more the electrons ‘bounce off’ the atoms of the material, the harder it is for them to travel from point A to point B, i.e. there is more ‘resistance’ in the material to electrical motion. Electrical resistance is quantified in Ohm’s law: V = IR, where V is the voltage applied across the resistor, I is the current passing through it, and R is the resistance. If we apply a fixed voltage across a circuit, and the resistance changes, we will see a change in current. Magnetoresistance is the observation that when a magnetic field is applied to a material, the resistance of that material can change. The application of magnetoresistance to hard drives is straightforward: we make a read-head which consists of a current traveling through a magnetoresistant material: by measuring the changes of current as the read-head travels over the bar magnets, we can (hopefully) determine what position the bar magnet is in.
‘Ordinary’ magnetoresistance is observed in common non-ferromagnetic metals. An applied magnetic field makes the electrons in the metal travel along helical paths, instead of straight lines. This ‘meandering’ of the electrons makes them more likely to hit an atom of the metal, and the resistance is therefore increased:
This effect is perhaps analogous to a drunk person wandering through a crowded party: because of his meandering, the drunk person will bump into people more often than a sober person will!
Ordinary magnetoresistance is a small effect, too small for application in hard drives. If we make our read-head out of a ferromagnetic material instead of an ordinary metal, we find a stronger effect known as anisotropic magnetoresistance (AMR). In essence, when current propagates in a material with built-in magnetization (such as a bar magnet), current flows easier when it travels in the direction of magnetization than when it travels perpendicular to the direction of magnetization. This effect is due to what is known as spin-orbit coupling between the current electrons and the atomic electrons in the magnetic read-head. Roughly speaking, an electron spins like a top, i.e. it possesses spin angular momentum. Electrons moving about atoms in a magnetic material have a preferential way of circling the atom,i.e. they possess orbital angular momentum. The two types of angular momentum can interact, and that interaction turns out to be strongest when the electrons are moving along the direction of magnetization.
Descriptions of how an AMR read-head works are rather hard to find, unless one is skilled in reading patent jargon. From what I can figure, they typically work as follows: A strip of magnetic material to be used as a read-head is pre-magnetized along a direction diagonal from the direction the current will flow, as roughly illustrated below:
Note that this means that, if we treat our read-head as a collection of bar magnets, the poles are more at the diagonal ends of the strip, rather than at the longitudinal ends.
What happens when we put this read-head near one of our data bits? The results of both directions is illustrated below:
The magnetic field of our data bits force the magnetization of our read-head to change. For the right-going bit, the magnetization rotates towards the perpendicular, while for the left-going bit, the magnetization rotates towards the parallel of the current. This means that we will see more current for our ’1′ bit, and less current for our ’0′ bit.
We finally come to giant magnetoresistance, the Nobel-prize winning discovery! As the size of bits on a disc is reduced, the corresponding field produced by that bit gets proportionally smaller. At some critical threshold, the field produced by the bit is too small to produce any appreciable change in resistance using AMR. In other words, there’s only so much data one can cram onto a disc using AMR.
Fert and Gruenberg looked at current traveling through a multilayer stack of magnetized materials separated by nonmagnetic spacers. If the spacer size is chosen appropriately, the magnetic layers will have opposite magnetizations:
As mentioned, the electron itself has spin (its own bar magnet), however, and every time it passes into a region which has a parallel magnetization, it must ‘flip’ itself over to the opposite direction. This change of energy state results in a net increase in resistance. Every electron must be pointing ‘up’ or ‘down’: when passing through the multilayer as pictured, then, every electron experiences at least one resistance-increasing flip.
Let us imagine putting this device in a magnetic field. A strong enough magnetic field will flip all the layers to a single direction, which means that some of the electrons will pass through without flipping! In the end, one ends up with a significant change in resistance, up to 10 times as large a change as found in AMR. A larger resistance change means we can use a smaller magnetic field and a smaller bit size.
How does one use this in a read-head? Let us assume we have only two oppositely magnetized layers, but one of them is ‘hard’ (doesn’t change magnetization easily) while the other is ‘soft’ (changes magnetization readily when exposed to an external field). Our read-head will have the following two configurations:
The data structure’s magnetization will effect only the soft layer, and we get a high current for a ’0′, and a low current for a ’1′. This configuration is known as a ‘spin valve’, as the upper magnetization layer acts very much like a valve that allows current to flow.
And that’s it! The discovery of GMR allowed the development of ultra-compact hard disc drives, which find important application in computers and iPods.
There’s also an even stronger form of magnetoresistance called colossal magnetoresistance, which apparently does not currently find use in disc drives due to the extremely large magnetic fields required.
Whew! Writing this took more time and effort than I thought it would. I’m still not sure I’ve gotten everything completely right (arrows for fields and magnetization always throw me), so feel free to yell at me if I’ve screwed it up.
For those interested in reading about magnetization in more technical detail, a very nice undergraduate level book is Magnetic Materials, by Nicola Spaldin. I used this book, along with ample internet searches, to understand magnetoresistance.
To conclude, let me add that this Nobel Prize also demonstrates how silly it is to criticize ‘Old Europe’ as stagnant and decadent. The prize winners are from France and Germany, respectively, and their discovery is arguably a cornerstone of modern technology.