A disk revolution

Albert Fert and Peter Grnberg jointly win this years Nobel Prize in Physics for the discovery of giant magnetoresistance.

THE first computer hard disk drive (HDD) was introduced by IBM in 1956. Called the 305 RAMAC (Random Access Method of Accounting and Control), this first data storage system comprised 50 disks, each about 60 centimetres in diameter, and stored about five megabytes of information. This corresponded to a data density of a mere 200 bits per square centimetre. The density in todays hard disks is touching a whopping 10 billion (or giga) bits per sq cm. That is a 50-million-fold increase in hard disk storage capacity in just 50 years, a rate much faster than the rate at which the number of components crammed into a silicon integrated circuit (IC) chip has grown in the same period (see graph).

While the latter has determined the rapid growth in the speed of the computer processor, the former is an equally, if not more, important component of the ongoing information technology revolution. Ever-shrinking hard disks have not only led to slenderer and sleeker laptops and more powerful servers and search engines, they have also changed the very concept of personal entertainment, with the development of pocket-size devices such as the MP3 player and the iPod, which can store hours of music and images.

The empirical Moores Law, which states that the number of components on a chip doubles every 18 months and which the growth in processor technology has followed until now, is essentially governed by advances in the technology of packaging transistors into an IC. There have been attempts to describe the exponential growth in data storage capacity by a similar law, called Kryders Law, but it has not conformed to any simple relationship like Moores Law. This is because advances in HDD technology have been due to the constant interplay between progress in basic science which is difficult to predict and its quick application. This years Nobel Prize in Physics has been awarded for one such unexpected but fundamentally new discovery called giant magnetoresistance (GMR) that has revolutionised hard disk technology.

During 1988-89, Albert Fert of Paris-South University, Orsay, France, and Peter Grnberg of the Jlich Research Centre, Germany, who share the Nobel award, independently discovered the new physical effect of GMR, which refers to very weak magnetic changes giving rise to unexpectedly large changes in electric resistance in certain systems. Such a GMR system has proved to be the perfect device for retrieving densely packed digital data from hard disks and has provided the much-needed technological leap for read-out heads in the present generation of HDDs.

A better read-out capability implies that a greater amount of information can be packed in a hard disk. This immediately led to the development of smaller and thinner hard disks with increasing storage capacities, which have now found their way into the new pocket-size electronic wonders. That a new scientific discovery can get translated into industrial-scale technology in less than a decade is unique for any finding. IBM introduced the first 16.8-gigabyte GMR-based HDD in 1997. Describing this development, the IBM website says: To some people, 10 years = a decade; to IBM research, 10 years = a revolution.

After bubble memory devices and flash memory cards, the development of data storage devices seemed to have hit a technology wall. It was as if the rapidly advancing information-driven world was waiting for a breakthrough like GMR. It is the concurrent advances in related disciplines that made the immediate exploitation of the discovery possible. Indeed, HDD is only the first of the possible applications of GMR, and its impact is already enormous. But the effect holds immense promise for future technology developments in the field.

How does an HDD work? Understanding this will help one appreciate the significance of the discovery. Information in a hard disk is stored digitally in the form of tiny magnetised regions. Magnetisation in one direction could be taken to represent the binary digit 0 and that in another to the binary digit 1. In modern HDDs, different elements of a single device write and read data as it flies over the spinning disk. The read-out head picks up the different fields of magnetisation as it scans the disk. As hard disks become smaller, each magnetised area, the bit, also correspondingly shrinks. In a tightly packed hard disk, the magnetic fields of individual bits, therefore, become weaker and require more and more sensitive read-out devices.

Induction coils, which work on the principle that a varying magnetic field induces a current through an electric coil, formed the basis for early read-out heads. But these were not sensitive enough to pick up minute field changes across shrinking hard disks. While induction coils are still used in writing heads, the phenomenon of magnetoresistance (MR) has been found to be better for the read-out function.

What is MR? It was discovered 150 years ago. In 1857, the British physicist William Thomson (Lord Kelvin, as he came to be known) found that the presence of a magnetic field influenced the electric resistance of materials such as iron. He wrote: I found that iron, when subjected to a magnetic force, acquires an increase of resistance to the conduction of electricity along, and a diminution of resistance to the conduction of electricity across, the lines of magnetisation. This difference in the resistance of a magnetic material along different directions is called magnetoresistance. Today, we understand this effect in terms of the quantum mechanical property of electrons called spin, which makes them behave like tiny magnets. Electron spin can assume two states: the state called up representing one polarity and the state called down representing the opposite polarity.

In general, MR effects are quite small, at best about a few per cent. But as sensor technology improved, the MR effect became important technologically, particularly for read-out heads in HDDs, and this was the direct predecessor of GMR. The most useful material in this context was the iron-nickel alloy called permalloy (Fe20 Ni20). This is how an MR-based HDD works. A pattern of electric pulses in the coil of the writing element produces a corresponding pattern of magnetic fields in the gap between the disk and the writing head. This alters the magnetic orientations in the bit regions of the disk so that they represent the data. During the read-out, the reverse process occurs: the bit magnetic fields induce resistance changes in the MR materials in the reading element, which in turn causes corresponding variations in the electric current output. The electric signals are decoded to reveal the data.

In 1991, dual-element MR heads began replacing single-element induction coils in commercial HDDs. IBM made the first public demonstration of a 150 million bits/sq cm MR head in late 1988. Efforts to achieve higher storage densities with smaller MR element heads could not, however, go beyond 1 gigabit/sq cm. This was because, since the time of Kelvin, there had not been any significant improvement in the performance of conventional MR materials, such as permalloy. To keep pace with the growing demands of higher and higher data densities in storage devices, the capability to detect feebler and feebler magnetic fields was required. Indeed, the belief that it was not possible to improve the performance of MR-based sensors further gained ground among physicists in the 1980s.

Enter GMR. It came as a big surprise to the scientific community when the French and the German research groups, led by Fert and Grnberg respectively, discovered materials that displayed very large MR, now called GMR. The materials they used were magnetic multilayers, in which layers of magnetic and non-magnetic metals are alternately stacked. The thickness of individual layers in the materials that they used was of nanometre (nm) size. A nanometre is a billionth of a metre (10{+ndash}{+9} m), and the typical atomic size is about a tenth of that. The techniques for producing materials with a thickness of a few atomic layers, a few nanometres, which had evolved during the 1970s and the 1980s, were thus crucial for the discovery.

At the atomic level, matter behaves differently from bulk material, and nano-scale structures, therefore, often exhibit entirely new and unexpected behaviour with regard to properties such as magnetism and electric conductivity. Nanotechnology, thus, concerns the manipulation of material assemblies comprising only a few strata of atoms, and GMR may be regarded as one of the first major applications of nanotechnology.

In the original experiments, while Grnbergs group used a three-layer system of iron-chromium-iron (Fe/Cr/Fe), Ferts group used a large number of Fe-Cr layers, as high as 60, stacked on one another (see picture). Both used the method called epitaxy, in which gases of iron and chromium are deposited in near-vacuum conditions as alternating layers. In such a set-up, atoms gradually attach themselves to the surface, stratum by stratum, to make the composite multilayer structure. In the presence of high magnetic fields, both experiments displayed significant magnetisation- dependent changes in resistance, the MR effect.

While Ferts system showed a decrease in resistance by as much as 50 per cent, the effect seen by Grnberg was much smaller (6 to10 per cent). This was essentially for two reasons: one, Fert was working with many more layers than Grnberg, and two, the French experiment was performed at a very low temperature (2680 C) as against the room temperature measurements used by Grnberg. The basic effect and the underlying physics, however, were the same in both cases. And both groups realised that what they had observed was an entirely new phenomenon. With classical MR, more than 1 per cent effect had not been seen. It was Fert who coined the term giant magnetoresistance and also foresaw in his first publication that the discovery could lead to important applications. Grnberg, too, realised the possible implications of the effect and filed for a patent along with his research paper.

However, for GMR to become the basis of an industrial technology, it was necessary to find a process to make stacks of nanolayers on a mass scale. The method of epitaxial growth, employed by Fert and Grnberg, is expensive and laborious and more suited to research laboratories. At IBM, Stuart Parkin and his colleague Kevin Roche, therefore, tried the faster and less precise process of sputtering, which was common in the manufacture of HDDs. They found that they could achieve the same effect in materials produced this way too. This demonstration helped Bruce Gurney and associates of IBM to search and identify multilayers that displayed GMR at room temperatures and at low magnetic fields. This paved the way for the development of a super-sensitive sensor for HDDs, called spin-valve, which IBM brought out in 1997. The same year, Parkin, along with Grnberg and Fert, won the European Physical Societys prestigious Hewlett-Packard Europhysics Prize for his pioneering work.

So what is the underlying basic physics that gives rise to GMR? In a metal, electric conduction takes place because of the transportation of electrons in a specific direction. Electric resistance is because of the deviation from the straight path owing to the scattering of electrons off defects and impurities; the more they scatter, the greater the resistance. In magnetic materials, the direction of magnetisation further influences the scattering of electrons as a result of the quantum spin that they possess.

Most of the spins in magnetic metals, such as iron, cobalt or nickel, are parallel to each other. There are, however, a smaller number of spins that point in the opposite direction (anti-parallel). It is this imbalance that contributes to the differing magnetisation of magnetic materials. This also causes conduction electrons to scatter differently depending on their spins relative to the spins in the bulk magnetic material, including at the interfaces between materials. On the other hand, in non-magnetic metals, both up and down electrons contribute equally to the electric resistance. Thus, the spin up carrier electrons and the spin down carrier electrons or the spin up current and the spin down current experience rather different resistances, and it is the material properties that determine which type of electrons scatters most and, hence, has greater resistance.

Consider the simplest three-layer structure, consisting of a layer of non-magnetic metal sandwiched between two layers of magnetic metal, in which GMR can arise. The current-carrying electrons with different spins experience different resistances within the first magnetic material and at the first interface (between the magnetic and the non-magnetic metals), with larger resistance for electrons that are not aligned in the direction of magnetisation of the metal. As the current enters the non-magnetic material, the resistance is the same for both types, which is generally negligibly lower than that in the magnetic layer. At the second interface and in the second magnetic material too, electrons that are not aligned will experience more resistance than those that are aligned.

In the case where both magnetic metals are magnetised in the same direction (as would be the case in the presence of an applied external magnetic field), the spins of most electrons will be aligned with the direction of magnetisation and the electrons will, therefore, pass through the entire structure without facing much resistance. However, if the magnetisation of the two magnetic layers is opposed (as can be the case in the absence of an external magnetic field), all the electrons will be oppositely aligned in one of the two layers. This means that no electrons will be able to pass through easily and the electric resistance will be at a maximum. An analogy with polaroids may be helpful in understanding this effect. A pair of crossed polaroids shuts off light completely. Similarly, a pair of magnetic layers with crossed magnetic polarities (or magnetisation) offers high resistance to the flow of electric current.

A structure as described above works as follows in a read-out device of an HDD. The magnetisation of the first layer is held fixed, or pinned, and the magnetisation of the third layer is free to move. When a weak magnetic field, such as that from a bit on a hard disk, comes under the structure, the magnetisation of the unpinned layer rotates relative to the pinned layer and because of the GMR effect causes a significant change in the electrical resistance and hence in the current signal leaving the read-out head. A high current may represent a binary 1 and a low current a binary 0.

An important reason why this discovery would not have been possible before techniques to grow nanoscale layers were known is the following. In order to exhibit the GMR effect, the mean free path length of the conduction electrons the average distance that an electron traverses before it is scattered has to exceed greatly the interlayer separation so that the electrons can travel through the magnetic layers and pick up the GMR effect. Without the new techniques, it would not have been possible to meet this requirement, and GMR would not have revealed itself. In this context, it may be pointed out that before the work of Fert and Grnberg, there were experimental observations of enhanced MR (of about a few per cent) but none was recognised as a new effect. Nanometre separation between magnetic layers is also important for an effective mutual magnetic coupling between them via the electrons of the non-magnetic layer so that their relative magnetisation is maintained in the absence of an external field.

A variation in the GMR system is the use of an insulating material instead of a non-magnetic material in the sandwich structure so that no electric current can ordinarily pass. But if the insulator is only a few atomic layers thick, electrons can quantum-mechanically tunnel through with a significant probability and generate a current through the multilayer. Such a system is called tunnelling magnetoresistance (TMR). The use of TMR can result in an even larger change in resistance with weak magnetic fields. In 2004, a Japanese group and Parkins group independently showed, using the insulator magnesium oxide sandwiched between iron layers, that TMR values as high as 200 per cent could be obtained. The latest generation of read-out heads uses TMR technology.

After the discovery of high-temperature superconductivity in 1986 in a certain class of oxide materials called perovskites, intense investigations into their conductivity and magnetic properties began. Following the discovery of GMR, the search for similar effects in perovskites led to the discovery in 1993 of an MR effect much larger than GMR in certain manganese perovskites. This effect is known as colossal magnetoresistance (CMR), CMR, however, is not yet useful in the technology of data storage systems or magnetic sensors because the required magnetic fields are high and the stability of the effect is also an issue.

GMR is also one of the first major applications of the emerging field of spintronics, or magnetoelectronics, which aims to exploit simultaneously in the same system the electric charge of the electron as in conventional electronics as well as its spin. Since the spin direction of an electron can only be maintained over very short distances which is why GMR itself had to wait until nanomaterials could be fabricated nanotechnology will form the basis of spintronics as well. That is, emerging nanotechnology was a prerequisite for the discovery of GMR; now spintronics has become the driving force for new applications of nanotechnology.

Besides its potential applications in spintronics, such as magnetic random access memory (MRAM) where memory devices can be switched on or off by the action of electron spins and spin-driven miniature microwave oscillators (which Fert is working on now), GMR could soon find applications in biology.

Detection of genetic material is one, which Grnberg is greatly excited about. You can attach antigenes to antibodies, you can attach the antibodies to so-called magnetic beads and you can then detect genetic material via magnetoresistive sensors. This is a topic which is very broad and, if it works, it has many, many applications, he says. The Nobel Prize-winning discovery of GMR and the developments that it has led to demonstrate clearly how fundamental science and technology go hand in hand and reinforce each other.