Read/Write Heads

The great increases in disk storage capacity over the last few decades have been made by finding ways to maximize the areal density of the disk, defined as the product of the linear density and track density. Linear density is the number of bits per unit length of track; track density is the radial density of tracks upon the disk. In order to increase the total areal density of a disk, it is necessary to minimize the track width and bit size, both of which depend on the heads used to write to and read from the recording media.

Inductive Heads

There are several desirable properties for a good inductive head, of which the most important are:

• The magnetization in its material should easily and linearly increase with the application of a magnetic field (low coercitivy).
• To write on a high-coercivity medium, high fields from the head must be generated at reasonable currents (high saturation).
• Occasional contacts between head and disk can wear away at the materials; thus, good resistance against wear and corrosion are necessary.
• Magnetic properties of heads change under mechanical stress (a phenomenon known as magnetostriction). Low-magnetostriction materials can endure the stresses they are put under without affecting the head performance.

Initially, a single inductive head was used as both the read and write head in a disk drive. By the 1970s, ferrite heads or their close counterparts, the metal-in-gap (MIG) heads, were the most commonly used; close-up views of these heads are shown in Figure 1.

The gap in a ferrite head is formed by putting down a layer of glass between two ferrite slabs. For high linear densities, submicron gaps are desirable, a feat not achievable with a ferrite head. Ferrite heads also have saturation magnetizations that are too low for them to be effective in writing to new media with higher coercivities. These and other considerations eventually caused a move to thin-film heads, reasearched since the 60's and making their first appearance in commercial disk drives in 1980.

Thin-film inductive heads have several advantages over their predecessors. As they are made with a combination of electroplating and lithographic techniques, the width of the head gap can be carefully controlled and minimized. Thin poles (the tips of the films on either side of the gap) result in narrow read-back pulses.

Figure 2 shows a schematic representation of one of the first thin-film inductive heads, the IBM 3370. The head is deposited on top of an alumina substrate, one layer at a time. The first is a magnetic layer of Permalloy (a nickel/iron alloy), which actually consists of two individual layers: the second does not reach all the way to the pole tip, leaving a thin layer at the tip for high resolution, but a thicker layer elsewhere to carry more flux from the inside of the head to the gap. Next, the gap layer -- usually alumina -- is deposited. (A hole is left in this insulation layer near the coil center to allow the later, second layer of Permalloy to make contact with the first and form a complete circuit.) After the gap layer, another insulating layer, usually photoresist, is deposited to create a level surface. The copper coil layer is lithographically patterned on top of the insulator, and electroplated on. Another layer of insulating photoresist is deposted on top of the copper, filling in the gaps between copper "wires", and followed the second magnetic layer of Permalloy (again, in two layers).

MR Heads

Magnetoresistive (MR) heads can detect signals at high densities and can produce a high voltage output from reasonably-sized magnetization changes, making them ideal magnetic recording sensors. Permalloy, an alloy already highly-studied and well-understood because of all its applications to magnetic recording, experiences a change in resistivity of 2-3% when exposed to a magnetic field; it has become the material of choice in MR sensors.

One advantage of MR heads is that the signal they generate is independant of the velocity of the disk medium. The change in resistivity depends on the flux from the medium, unlike the inductive head, which reacts to the change in flux with time. This means that MR heads are equally sensitive in disk drives of varying speeds, making them useful for slowly-rotating laptop drives.

Unlike inductive heads, which utilize a copper coil of many turns, the MR head has a very low internal inductance. Inductive head noise increases with the number of turns, whereas MR noise depends only on the sensor dimensions. MR heads are also active devices; that is, their voltage output depends on the amount of current being run through them, a factor that can be changed to help boost the signal.

The MR sensor is a read-only element, whereas the inductive head can both read and write to the medium. While this may at first appear to be a point in the inductive head's favor, the separation of write and read elements allows each head to be optimized to do its part. The MR read head can be made narrower than the track (and the write head), making it less likely to pick up noise from adjacent tracks.

A schematic of a typical MR sensor is shown below in Figure 3. While the MR element is being formed by the deposition of a thin Permalloy film on a substrate, a magnetic field is applied parallel to the plane of the film, creating an easy axis as shown. The sensor's magnetization M would usually align with the easy axis, but in the applied field H from the magnetic medium, it rotates toward H. The resisitivy of the MR strip changes as a function of the angle between the magnetization M and the current direction, reaching its maximum when the magnetization is perpendicular to the current.

Typical MR curves are shown in Figure 4; it should be noted that the change in resistivity is nonlinear in reasonably-sized applied fields. Since a linear response is the most desirable, MR sensors are usually biased by being placed next to a permanent magnet. This has the effect of shifting the curve as shown in Figure 5, so that it can be operated in the nearly-linear region of the curve.