Buschow K.H.J. (Ed.) Concise Encyclopedia of Magnetic and Superconducting Materials

Подождите немного. Документ загружается.

stripe height point of the reader to the zero throat

height point of the writer is extremely critical, as only

one of the layers can be used as the reference layer for

lapping.

Control of the lapping process is done with some

lithographically deﬁned feature that changes in a

systematic way with lap depth. In early generations of

thin ﬁlm head technology, optical lapping guides

(OLGs) were used. These were patterned structures

thick enough to be seen in cross-section from the

airbearing surface during lapping. They were typi-

cally triangular in shape such that their visible width

changed in a systematic way with lapping depth.

Lapping was stopped when a particular dimension

was reached in the visible structure.

Since this process did not lend itself easily to

automation, electronic lapping guides (ELGs) be-

came the standard. These structures were made from

the same layer as the MR stripe, and were similar in

shape. Their resistance increased as their stripe height

was reduced in a predictable fashion and could be

used as an endpoint detection for lapping. Once this

process was automated, it was possible to sense

variations across the row by monitoring a number of

lapping sensors and to actually ﬂex the row slightly,

increasing the pressure in slow lapping points to

obtain maximum uniformity during the lapping

process. Absolute resistance is typically not used as

a stopping criterion, as it is too susceptible to process

variation. Typically a differential or ratio technique is

used to make the sensors as insensitive as possible to

process variations in ﬁlm thickness, stripe height, etc.

Once the rows have lapped to stripe height, an

airbearing contour is formed on the airbearing

surface of each slider in the row. Older-style air

bearings were simply machined rails, which allowed

the head to be near the disk while allowing a lot of air

to rush through a channel between the head and

medium. Air bearings have become patterns that

cannot be formed with rectilinear batch machining,

and are formed through one or more steps of ion

milling. This requires photolithography on the air-

bearing surface, allowing air bearings of arbitrary

shape. Some of these are quite complicated with

negative and positive pressure regions, ﬂy heights

that are insensitive to speed, and excellent stability.

State-of-the-art commercial air bearings (ca. 2001)

allow head to medium spacings of o25 nm, with

further reductions of at least a factor of three under

development. A good review of ABS technology is

given in Hoyt (1997).

Once the air bearings are complete, the rows are

diced or parted into sliders, again with precision

sawing. Individual sliders are then mounted on

stainless steel suspensions, using adhesive between

the gimbal mounting point and the slider. In some

advanced designs, these suspensions are active

devices, incorporating secondary ﬁne actuators that

use piezoelectric, electrostatic (Kim et al. 1999), or

electromagnetic (Lin and Ong 1999) forces to provide

ﬁne tracking. It is widely assumed that secondary

actuation is required for very high track densities

(Fan et al. 1995).

3. The Future of Recording Head Fabrication

The future of magnetic recording likely holds at least

four major paradigm shifts in the coming years. Some

of these will have a profound impact on head

fabrication. Not all of these will become a reality,

but it is almost certain that at least one that will have

major effect on head architecture and fabrication will

reach commercialization. These four trends are (in

order of increasingly radical departure from today’s

head manufacturing paradigm);

patterned media

perpendicular or vertical recording

thermally assisted (or hybrid) recording

micro-array based storage

Patterned media do not in themselves require any

changes in the head, although it will certainly change

the system architecture, especially the channel (Nair

and New 1998, Hughes 1999). It may change the

head, if the required ﬁeld magnitudes or rise times

become radically different than they are today. If

(and/or when) patterned media do become commer-

cialized, the head changes to support it will be mainly

evolutionary.

Perpendicular recording will almost certainly

change the write head design, although similar steps

will be used in fabrication. It is possible to record on

and read from perpendicular media with a standard

head for longitudinal recording, but more optimum

designs are possible (Khizroev et al. 1997). Some kind

of probe head with a relatively large gap is likely for

writing. Soft underlayers in the media will be used to

increase the head ﬁeld. Long magnetic circuits with

distant return paths must be avoided to avoid stray

ﬁeld sensitivity (Cain et al. 1996). Reading will be

magnetoresistive, but whether the sensor will be at

the airbearing or recessed, and the exact type of

sensor, is still to be determined.

Thermally assisted recording will precipitate a

radical change in the head geometry. Some source

of heat, most likely light will need to be conducted

to the medium by the head. At the densities where

this will become necessary due to thermal stability

considerations (Ruigrok et al. 2000), near-ﬁeld optics

will be necessary. This will require optical coupling

and focusing elements on the head, and many new

fabrication techniques.

Several research groups are currently working

on microarray-based storage, including IBM and

Hewlett Packard (described in Toigo and Hayashi

2000), Kionix and Carnegie Mellon (Carley et al.

2000). If these devices become widely commercia-

lized, they will represent radical departures in head

570

Magnetic Reco rding Devices: Inductive Heads, Properties

fabrication technology. Heads will be tightly inte-

grated with Si electronics, on Si substrates. They will

be massively parallel, and they may employ any

number of physical mechanisms to read and write

including magnetic. Whether such radical architec-

tural changes will completely replace standard disk

drives is an open question. It almost certain, though,

that they will at least ﬁnd some niche market, and

introduce some radically different designs to the data

storage head.

See also: Magnetic Recording Technologies: Over-

view; Magnetic Recording Devices: Future Techno-

logies; Magnetoresistive Heads: Physical Phenomena;

Magnetic Recording Devices: Head/Medium Inter-

face

Bibliography

Ashar K G 1997 Magnetic Disk Drive Technology: Heads,

Media, Channel, Interfaces and Integration. IEEE Press, New

York

Bajorek C, Krongelb S, Romankiw L T, Thompson D 1974 An

integrated magnetoresistive read, inductive write high density

magnetic recording head. In: Proc. Magnetism and Magnetic

Materials Conf.

Brug J A, Anthony T C, Nickel J H 1996 Magnetic recording

head materials. MRS Bull. 21, 23–6

Cain W C, Payne A P, Baldwinson M, Hempstead R 1996

Challenges in the practical implementation of perpendicular

magnetic recording. IEEE Trans. 32, 97–102

Carley L R, Bain J A, Fedder G K, Greve D W, Guillou D F,

Lu M S C, Mukherjee T, Santhanam S, Abelmann L, Min S

2000 Single chip computers with MEMS-based magnetic

memory. J. Appl. Phys. 87, 6680–5

Church M A, Jones R E Jr 1980 Thin ﬁlm magnetic head

assembly. US Patent 4 219 854

Coffey K R, Gurney B A, Heim B E, Lefakis H, Mauri D,

Speriosu V S, Wilhoit D R 1994 Spin valve magnetoresistive

sensor with self-pinned laminated layer and magnetic

recording system using the sensor. US Patent

Daniel E D, Mee C D, Clark M H 1999 Magnetic Recording:

The First 100 Years. IEEE Press, New York

Fan L-S, Ottesen H H, Reiley T C, Wood R W 1995 Magnetic

recording-head positioning at very high track densities using

a microactuator-based, two-stage servo system. IEEE Trans.

Indust. Electron. 42, 222–33

Guo W, Wang Z, Yao X, Huang T, Bi C 1998 A high

bandwidth piezoelectric suspension for high track density

magnetic data storage devices. IEEE Trans. 34, 1907–9

Hannon D, Krounbi M T, Christner J 1994 Allicat magnetor-

esistive head design and performance. IEEE Trans. 30,

298–302

Hoyt R F 1997 Head-disk interface. In: Ashar K G (ed.)

Magnetic Disk Drive Technology: Heads, Media, Channel,

Interfaces and Integration. IEEE Press, New York, Chap. 10,

pp. 268–305

Hughes G F 1999 Read channels for patterned media. IEEE

Trans. 35, 2310–2

Jones R E Jr 1980 Thin ﬁlm magnetic head. US Patent 4 219

855

Jones R E Jr, Mee C D, Tsang C 1995 Recording heads. In:

Mee C D, Daniel E D (eds.) Magnetic Recording Technology.

(2nd edn). McGraw-Hill, New York, Chap. 6, pp. 1–91

Khizroev S K, Bain J A, Kryder M H 1997 Considerations in

the design of probe heads for 100 Gbit/in

recording density.

IEEE Trans. 33, 2893–5

Kim B-H, Park S, Kim H-C, Chun K, Cho D-I D, Lee J-W, Lee

H-J, Kim S-H, Seong W-K, An Y-J 1999 MEMS fabrication

of high aspect ratio track-following microactuator for hard

disk drive using silicon on insulator. In: Proc. 12th IEEE Int.

Conf. on Micro Electro Mechanical Systems. Orlando, FL,

USA

Krounbi M T, Voegeli O, Wang P-K 1991 Magnetoresistive

read transducer having hard magnetic bias. US Patent 5 018

037

Lazzari J P 1974 Integrated magnetic recording heads. Review

and outlook. Proc. AIP. 18, 990–1004

Lazzari J P 1996 Planar silicon head: the modern solution for

the disk drive industry. Proc. Electrochem. Soc. 95, 30–2

Lazzari J P 1997 Planar silicon heads for disk drive industry.

In: Transducers 97, Int. Conf. Solid-state Sens. Actuators 2,

pp. 1077–80

Lederman M 1999 Performance of metallic antiferromagnets

for use in spin-valve read sensors. IEEE Trans. 35, 794–9

Lin H, Ong E H 1999 Design and optimization of a dual-stage

high bandwidth piggy-back electromagnetic milli-actuator.

IEEE Trans. 35, 3673–5

Mallary M L, Dipalma L, Gyasi K, Sidman A L, Wu A 1994

Advanced multi-via heads. IEEE Trans. 30, 287–90

Mallary M L, Ramaswamy S 1993 A new thin ﬁlm head which

doubles the ﬂux through the coil. IEEE Trans. 29, 3832–6

Mallinson J C 1996 Magneto-resistive Heads: Fundamentals and

Applications. Academic Press, San Diego, CA

McGuire T R, Potter R I 1975 AMR effect review. IEEE Trans.

11, 1018–38

Nair S K, New R M H 1998 Patterned media recording: noise

and channel equalization. IEEE Trans. 34, 1916–8

Romankiw L T, Croll I M, Hatzakis M 1970 Batch-fabricated

thin-ﬁlm magnetic recording heads. IEEE Trans. 6, 597–601

Ruigrok J J M, Coehoorn, Cumpson S R, Kesteren H W 2000

Disk recording beyond 100 Gb/in

: Hybrid recording?

J. Appl. Phys. 87, 5398–403

Shen Y, Chen W, Jensen W, Ravipati D, Retort V, Rottmayer

R, Rudy S, Tan S, Yuan S 1996 Comparative study of MR

head designs with abutted vs overlaid permanent magnet

bias. IEEE Trans. 32, 19–24

Tamura K, Nagao E, Ohdoi Y 1990 Development of a 48-turn

thin-ﬁlm head. J. Appl. Phys. 67, 4857–9

Tang D D, Lee R E, Su J L, Chu F, Lo J, Santini H A A, Lane

L, Robertson N, Ponce M, Cisneroz P, Guthmiller G

1994 Omega head—an experimental 120-turn inductive head.

J. Appl. Phys. 75, 6397–9

Toigo J W, Hayashi A M 2000 Avoiding a data crunch. Sci.

Am. 58–71

Tsang C, Chen M-M, Yogi T, Ju K 1990a Performance study

and analysis of dual element head on thin-ﬁlm disk for

gigabit-density recording. IEEE Trans. 26, 2949–53

Tsang C, Chen M -M, Yogi T, Ju K 1990b Gigabit density

recording using dual element MR/inductive heads on thin-

ﬁlm disks. IEEE Trans. 26, 1689–93

Tsang C, Fontana R E Jr 1982 Fabrication and wafer testing of

barber-pole and exchange-biased narrow-track MR sensors.

IEEE Trans. 18, 1149–51

White R L 1992 Giant magnetoresistance: a primer. IEEE

Trans. 28, 2482–7

571

Magnetic Reco rding Devices: Inductive Heads, Properties

Zak B, Curland N, Giusti J H, Ash K P, Cameron G P 1996

Design and performance of a recessed thin ﬁlm inductive

transducer. IEEE Trans. 32, 74–9

Zhu J-G, O’Connor D J 1995 A new design for an ultra-high

density magnetic recording head using a GMR sensor in the

CPP mode. IEEE Trans. 31, 2597–9

Zhu J-G, O’Connor D J 1996 Impact of microstructure on

stability of permanent magnet biased magnetoresistive heads.

IEEE Trans. 32, 54–60

Zieren V, Somers G, Ruigrok J J M, de Jongh M, van Straalen

E, Folkerts W, Draaisma E, Pronk F 1993 Design and

fabrication of thin ﬁlm heads for the digital compact cassette

audio system. IEEE Trans. 29, 3064–8

J. A. Bain

Carnegie Mellon University, Pittsburgh

Pennsylvania, USA

Magnetic Recording Heads: Historical

Perspective and Background

Magnetic recording heads are a key enabling tech-

nology for the recording of information on magnetic

media since they both write and read the magnetic

transitions. The ﬁrst recording heads were inductive,

composed of a bulk ferromagnetic material such as

iron and a coil to both magnetize the iron and to

provide a means to detect the magnetic ﬂux from a

magnetic transition.

The inductive head served to both read and write in

the same structure. Figure 1 shows a simpliﬁed dia-

gram of an inductive recording head. As current

ﬂows through the coil a magnetic ﬁeld ﬂows in a

circular path through the core and gap of the device.

Some of the ﬂux in the gap fringes outside of the core.

It is this ﬂux that writes the transition on the media.

By a principle known as reciprocity the ﬂux from a

transition beneath the head will be picked up by the

coil and sensed as voltage proportional to the change

in ﬂux under the gap; i.e., the signal is proportional

to the product of the velocity of the head over the

media and strength of the ﬂux emanating from the

transition.

Oberlin Smith, an American, suggested the ﬁrst

magnetic recording head in 1888. It featured a wire

recording medium with an electromagnet consisting

of a coil wound around a rod of iron as shown in

Fig. 2. Valdemar Poulsen of Denmark built the ﬁrst

functioning recorder using this head in 1898.

The wire was replaced by a steel tape in the 1930s.

In Germany, the Stabltone-Bandmachine was ﬁrst

offered for sale in 1935. In England, the Blattner-

phone was ﬁrst offered for lease in 1931 and in

1933 the rights to the Blattnerphone were sold to

Marconi’s Wireless Telegraph Company, Ltd.

Marconi, in cooperation with Stille, one of the

developers of Stabltone-Bandmachine, developed a

second-generation machine. The head on this ma-

chine consisted of a dual pole head, one on each side

of the tape. The ﬂux was concentrated by tapering the

tips of each pole. In 1935, Marconi modiﬁed the head

design to single-pole design. It is interesting to note

the pole head is favored for perpendicular recording,

which has attracted a lot of attention lately in the

push to higher areal densities.

In 1933, Eduard Schuller of AEG in Germany took

out a patent on the ring head. At AEG, tape record-

ers were built using ring heads for recording, play-

back, and erasure. They are still used today for

virtually all writing on magnetic recording media.

For ﬁve decades ring heads were used for reading

and writing in most recording applications. The basic

Figure 1

Inductive ring head.

Figure 2

(a) The apparatus used by Poulsen for his ﬁrst

experiments. A steel wire is stretched between A and B.

The electromagnet E was moved along the wire by hand

to record or replay sound. In (b), the wire is shown

partially surrounded by the pole piece of the

electromagnet.

572

Magnetic Reco rding Heads: Historical Perspective and Background

design remained the same but improvements in ma-

terials and the manufacture of smaller heads with

tighter tolerances extended them to more advanced

application. The original steel heads were replaced by

mu metal and laminated mu metal. This was followed

by NiZn and then MnZn ferrite. Manufacturing

technology improved to allow for batch fabrication

of many cores together which were then sliced up like

loaves of bread to yield individual cores.

Core winding moved overseas. Epoxy and then glass

was used to bond together the cores and to bond the

cores into the sliders. The last evolution in this batch-

fabricated technology was the MIG (metal in gap)

head, which featured a ferrite core with high-moment

material such as Sendust deposited in the gap.

The ﬁrst commercial thin-ﬁlm single-turn heads

were made by Burroughs in 1975 on the B9470 head/

track disk ﬁle. In 1979, IBM introduced thin ﬁlm

heads. These heads were made by a semiconductor-

type process with hundreds and then thousands of

heads fabricated on a single wafer. The wafers were

then sliced up, made into sliders, and mounted on

head gimbal assemblies (HGA) as shown in Fig. 3.

This batch process eventually reduced the cost and

improved the performance of the magnetic recording

heads until MIG ferrite heads were driven from the

market place for disk drives around 1995. Ferrite

heads continue to ﬁnd application in ﬂoppy disks,

video, and other tape recorders.

As the areal density increased in the 1980s the

smaller tracks required more and more turns be put

on the heads to make up for the signal lost as the

track width was decreased. This was not entirely

without cost since the inductance increased with the

square of the number of turns and the increasing

turns reduced the frequency response of the heads by

increasing the resonant frequency of the coil. The

thin-ﬁlm heads had an advantage in this respect over

the ferrite heads since their smaller size resulted in

lower inductance than the larger ferrite heads.

The ﬁrst IBM thin-ﬁlm heads had a single-layer

eight-turn coil. Two-layer coils followed these. In the

early 1990s, three-layer and four-layer coils were

developed. Inductive thin-ﬁlm heads reached their

limit with heads manufactured by Dastek Corpora-

tion consisting of four layers and 54 coil turns. The

last read/write inductive head drive was shipped by

Western Digital Corporation in 1995.

Figure 3

Thin-ﬁlm head process—wafer to slider (courtesy of Seagate Technology, LLC).

573

Magnetic Reco rding Heads: Historical Perspective and Background

In 1991 IBM introduced the MR (magneto-resis-

tive) read head using the AMR (anisotropic magne-

toresistive) effect in which the resistance of the element

was determined by the angle between the sense cur-

rent and the magnetization. This AMR device was

coupled with an inductive write head (see Fig. 4).

From this point MR and GMR heads proceeded

to dominate the disk storage business within the next

ﬁve years.

1. Air Bearings

Hand-in-hand with the improvements in transducers,

the increases in areal density of the previous four

decades were enabled by the decrease in head-to-

media spacing (HMS), usual referred to as ‘‘ﬂy height’’

(FH). In 1953, Bill Goddard of IBM designed and ﬂew

a hydrostatic air bearing as part of the development

of the IBM RAMAC disk drive. This was an exter-

nally pressurized air bearing where the air cushion

was supplied by air piped into the head assembly as

shown in Fig. 5.

The heads were mounted on a gimbal spring that

allowed the head to pitch and roll so that it could

align with the surface of the rotating disk underneath.

All subsequent ﬂying heads used a gimbal assembly

to allow alignment of the head and disk surfaces. This

is referred to as a head gimbal assembly (HGA).

The ﬁrst self-acting air bearing for magnetic re-

cording was developed at IBM beginning in 1955.

Jake Hagopian developed the ﬁrst concept and Bill

Gross, Ken Haughton, and Russ Brunner reﬁned it.

The self-acting air bearing was introduced in the IBM

1301 disk drive in 1962. This air bearing and the

following ones had a cylindrical or spherical crown

and featured a large mass and load pressure (several

hundred grams force). The use of these cylindrical

self-acting air bearings reduced the ﬂy height from

800 mm on the RAMAC to 50 mm on the IBM 3330.

They were of the load/unload type, which means they

were lifted from the disk surface before the disk

stopped rotating and loaded when the disk was up to

speed.

In the early 1970s, IBM developed and shipped the

3340 ‘‘Winchester’’ disk drive which featured a two-

rail taper ﬂat air bearing with a low-low mass slider

ﬂying over a lubricated ferric oxide disk in a contact

start–stop mode. The head ﬂew at about 20 micro-

inch when the disk was rotating but was in contact

when the disk was stopped. At about the same time

Digital Equipment Corporation designed and

shipped the RS04 disk drive featuring a taper-ﬂat,

contact/start–stop, air bearing ﬂying on a lubricated

thin-ﬁlm plated disk.

These taper-ﬂat, contact/start–stop sliders were

used on many drives until the introduction of the

negative-pressure air bearings in 1991 by IBM in the

Corsair Drive. In 1995, Read-Rite Corporation in-

troduced the Tri-Pad air bearing featuring a pseudo-

contact air bearing where the head and disk were

almost in contact. This approach only worked with

inductive heads due to the sensitivity of the MR and

Figure 4

Merged MR head (courtesy of Seagate Technology,

LLC).

Figure 5

RAMAC head (courtesy of IBM Corporation).

574

Magnetic Reco rding Heads: Historical Perspective and Background

GMR heads to contact problems such as electrostatic

discharge (ESD) and thermal noise due to the contact

with asperities on the disk surface.

Current air bearings are of the negative-pressure

type and ﬂy below 1 microinch (25 nm). A negative-

pressure air bearing has regions of both above- and

below-ambient pressure as shown in Fig. 6. The re-

sultant pressure is positive and is balanced by the

head load; however the negative-pressure air bearing

has the advantage of being stiffer for the same load

and being amenable to designs that are insensitive to

changes in pressure (altitude) and velocity (radius).

The increasingly smooth surfaces needed to facilitate

these low ﬂying heights have led to a resurgence of

the load/unload concept due to the wear and stiction

problems associated with two very ﬂat surfaces in

contact.

The complex shapes required for negative-pressure

designs are created by photopatterning after the bar

is cut from the slider. The cavities are formed by ion

etching of the air-bearing surface.

To be useful in a disk drive, the slider must be

mounted on a HGA and assembled into a positioner

in the drive as shown in Fig. 7. Basic ﬂexure design

has not fundamentally changed since the early days

of disk storage. They have become less stiff and apply

smaller loads as the sliders have become smaller. The

discrete wires of the early heads have given way to

miniature ﬂex circuits for connecting the head to the

electronics.

The higher performance of tomorrow’s drives has

given rise to consideration of moving the critical front

end electronics closer to the head itself and even in-

tegrating the electronics with the head on the slider.

The ever-increasing number of tracks per inch (tpi)

has resulted in much development on microactuator

designs. These are structures which move the head a

very small distance very quickly and significantly

enhance the servo performance of the heads.

2. The Longitudinal Write Head

The simplest approximation to the write ﬁeld from a

ring head is the Karlqvist head equation published in

1954 (Karlqvist 1954) as shown in Fig. 8. It ignores

the bulk of the head and concentrates on the gap area

where a very high (inﬁnite) permeability is assumed.

The head is assumed to be inﬁnitely wide, reducing

the problem to two dimensions and the magnetic ﬁeld

in the gap region is assumed to be uniform.

In the limit of a very small gap, the equation

reduces to

¼ðH

g=pÞðy=x

þ y

¼ðH

g=pÞðx=x

þ y

The ﬁeld in the gap can be approximated by

¼ðNI=gÞE

where E is the head efﬁciency and I is the current in

the coil. The head efﬁciency is a measure of how dif-

ﬁcult it is to pass ﬂux through the core of the head.

The efﬁciency is dependent on the head material,

head geometry and, if there is saturation, the right

Figure 6

Negative-pressure air bearing (courtesy of Seagate

Technology, LLC).

Figure 7

HGA head assembly (courtesy of Seagate Technology,

LLC).

575

Magnetic Reco rding Heads: Historical Perspective and Background

current. In a good modern design, the write efﬁciency

can vary from 40% to 80%.

The Karlqvist equation predicts that when the

wavelength of the written signal, l ¼g/n, where n ¼1,

2, 3,y, then the readback will be zero. This is called

the gap null. It must be considered when using the

inductive head to readback but no null is created

when writing at gap null wavelengths. The Karlqvist

equation is an approximation and more exact calcu-

lations show that the ﬁrst gap null is at 1.136 g.

In a real inductive head the magnetic ﬁelds pro-

duced by the head are complex and difﬁcult to model

because of several factors including the complicated

head geometry, the nonlinear magnetic materials, and

the size of the calculation to do a full micromagnetic

calculation.

The important parameters that a write head de-

signer will consider are:

The strength of the ﬁeld at the media,

The write-ﬁeld gradient at the point where the

transition is written,

The off-track writing or fringing,

The frequency response of the head.

The head designer using a standard thin ﬁlm ring

head has the following parameters to work with:

(i) The gap length can be adjusted and write ﬁeld

and gradient traded off against fringing.

(ii) The number of turns can be adjusted and in-

ductance traded off against write current.

(iii) The yoke can be made shorter to gain better

NLTS (nonlinear transition shift) and frequency re-

sponse.

(iv) The frequency response can be increased by

laminations, thinner poles, higher resistivity pole

material, and materials with high-frequency perme-

ability.

(v) The ﬁeld and ﬁeld gradient can be increased in

the head by increasing the deep gap ﬁeld using higher

moment materials.

The design improvements in the head show a trend

to heads moving to fewer turns, shorter yokes, and a

tendency to run closer to the saturation of the pole

tips. Actual saturation of the pole tips can lead to

very undesirable nonlinear effects.

High-resistivity materials such as CoZrTa and Co-

ZrNb have been used in poles or in the part of the

pole close to the gap. Although these materials reduce

eddy currents, they have a B

of only about 1.3 T so

they can not produce the high ﬁelds of alternative

materials. B

is the saturation magniﬁcation and is

the highest magnetization that can be induced in a

material. Laminated materials have been used to sup-

press eddy currents in the yokes of heads; however,

they have not been widely used.

The materials used in head poles are moving to

higher saturation moments with a common compo-

sition now being Ni

¼1.6 T). These alloys

are in common use and are deposited by electroplat-

ing. Sputtered alloys with higher moments such as

FeNX (X ¼tantalum, aluminum, zirconium, etc.)

¼2.0 T) and other alloys such as FeCoX (X ¼zir-

conium, niobium, boron) (B

¼2.2 T), and FeCoNi

¼2.0 T) are also known. High-moment materials

are often used with other lower-moment materials in

the pole tips as shown in Fig. 9. The high-moment

material is placed closed to the gap where the high

ﬂux is needed. Lower-moment material guides the

ﬂux to the gap. Generally, the lower-moment mate-

rials such as Permalloy have superior soft properties

to the higher-moment materials.

The maximum B

obtained in a bulk alloy is 2.4 T

in Fe

. Some materials have been reported with

higher B

such as Fe

but never in stable phases of

Figure 9

Dual-layer write head (courtesy of Seagate Technology,

LLC).

Figure 8

Karlqvist head.

576

Magnetic Reco rding Heads: Historical Perspective and Background

enough material to make a pole. This lack of a soft

material of high B

places a fundamental limit on how

high a coercivity can be written. The coercivity, in

turn, is directly related to the stability of the grains in

the media and thus the B

of the pole material is one

of the key factors in limiting the size of the grains in

magnetic media.

Since smaller grains are required for higher signal-

to-noise, the B

of the pole material limits the den-

sity of magnetic recording. Researchers continue to

search for higher-moment materials that have ade-

quate soft properties, are stable over the operating

temperatures of the disk drive and the processing

temperatures of head manufacture, and can be made

in large enough pieces for pole fabrication.

3. Perpendicular Write Heads

The longitudinal heads discussed above are limited

to a ﬁeld of less than 2p M

(o1.2 T for known

materials) at the media. Perpendicular recording, on

the other hand, can produce ﬁelds of almost 4p M

(2.4 T) due to the more favorable geometry utilizing

a soft underlayer (SUL) beneath the recording layer

in the media. This is one of the main attractions of

perpendicular magnetic recording.

Perpendicular recording utilizes a so-called pole

head to write on the media as shown in Fig. 10. In

this process, the vertical ﬁeld from the head is used

to write the media. In perpendicular recording, the

SUL in the media is really part of the head ﬂux path

providing the return path for the ﬂux from the head.

As we approach the ‘‘superparamagnetic limit,’’ tak-

ing into account thermal stability and SNR in a

model with a number of simpliﬁcations, Bertram

(1994) has shown that perpendicular recording can

have a 4–5 times advantage in areal density over

longitudinal.

While there are factors other than SNR and ther-

mal stability that may affect the relative densities

of perpendicular and longitudinal recording, there

has been a reawakening of interest in perpendicular

recording for high densities.

The perpendicular recording head can be modeled

by a Karlqvist head turned on its side as shown in

Fig. 11. In this case, the ﬁeld is given by

head

tan

1

d þ t=2



þ tan

1

d þ 3t=2



where H

is the deep gap ﬁeld.

The considerations for designing a perpendicular

write head are similar to those for a longitudinal.

Frequency response can be improved by using lami-

nation or using high-resistivity materials and by

making the structure smaller. The write efﬁciency is

generally higher in perpendicular heads so lower

currents can be used for the same number of turns.

Off-track writing is not affected by the gap and can

be less than longitudinal in the ideal case. The fring-

ing will, however, increase with increasing skew angle

and this may be a problem for heads with ﬁnite-

length pole tips. The SUL causes some unique pro-

blems since it provides the return path for the head

ﬂux and is part of the media.

Its orientation, permeability, and saturation mag-

netization are important. The SUL also extends far

beyond the normal extent of a normal ring head and

this requires careful treatment in models. The shape

of the pole is important in preventing the pole from

becoming magnetized when not being driven by the

coil. Work in the past has also shown perpendicular

heads to be more susceptible to the pick up of stray

ﬂux. Whether perpendicular recording will replace

longitudinal at densities above 100 Gb/in

depends

on overcoming these and other problems to realize

the potential density increases promised by thermal

stability and SNR considerations.

Figure 10

Perpendicular write head.

Figure 11

Karlqvist model for perpendicular.

577

Magnetic Reco rding Heads: Historical Perspective and Background

4. The Read Head

In 1988, the giant magnetoresistive (GMR) effect was

discovered by Baibich et al. (1988) in Fe/Cr multi-

layers (see Giant Magnetoresistance). In this effect

layers, of ferromagnetic (FM) material alternate with

layers of nonmagnetic metals. These structures ex-

hibit a large change in resistance when switched from

the low-resistance state, where the FM layers are

parallel to each other, to a high-resistance state,

where the layers are antiparallel.

There are two types of architecture of potential use

in read-head geometry, the CIP (current in plane) and

CPP (current perpendicular to the plane) as shown in

Fig. 12. The vast majority of today’s heads are in a

simple two-layer CIP conﬁguration referred to as a

spin valve.

Spin valve heads (see Magnetic Recording Systems:

Spin Valves) have separate read and write elements as

shown in Fig. 4, which allows separate optimization

of the read and write function as compared to the

inductive only head. These heads are also much more

sensitive to ﬂux than inductive or the MR readers. An

inductive-style writer is usually built on top of a

magnetoresistive reader. As shown in Figs. 13 and 14

the GMR sensor consists of several layers.

In the bottom spin valve shown here there is a seed

layer, an antiferromagnetic (AF) pinning layer that

holds the pinned layer magnetization in a constant

direction through exchange coupling with the pinning

layer which has no net moment. Between the pinned

layer and the free layer is a thin layer of copper. The

free layer is free to rotate under an applied ﬁeld as

shown in Fig. 15. The structure is protected with a

capping layer.

The change in voltage of the spin valve is given by

DV ¼

IDRsinY

where I is the bias current ﬂowing through the sensor,

Y is the angle of rotation of the free layer wrt in its

quiescent state and DR is the maximum change in

resistance when the two layers change from parallel

to antiparallel. The response of a spin valve to a

transition on the disk is shown in Fig. 16.

In a real spin valve the magnetization is not uni-

form and the sinY must be averaged over the active

area of the spin valve.

Figure 12

CPP and CIP geometries.

Figure 13

GMR spin valve sensor stack.

578

Magnetic Reco rding Heads: Historical Perspective and Background

DR/R values of over 20% have been observed in

spin valves but values of 7–12% are typical of today’s

production. Typical sensor resistances are 40–50 O.

Bias currents are typically about 5 mA. Output volt-

ages of over 1 mV/mm have been achieved.

This spin valve structure is sandwiched between

two insulating layers so that the device will not short.

This structure is then placed between two soft mag-

netic layers called shields as shown in Fig. 14. These

shields prevent the sensor from picking up ﬂux from

bits both down-track and up-track from the sensor

and limit the pickup to the ﬂux immediately under the

device.

In order to remain stable under the operating

conditions of disk drives, these heads require longi-

tudinal stabilization, which keeps the magnetiza-

tion in these devices more coherent and prevents the

breakup into smaller domains. This function is served

by attaching a permanent magnet structure (PM) to

the sides of the GMR sensor and magnetizing it in

a direction across the track. The PM is combined

with a lead structure to bring the sense current to

the device. The change in resistance is then typically

read out as a change in voltage with a constant sense

current.

In addition to the gap and track width, the height

of the sensor is important. This is referred to as the

stripe height and is accomplished by precision lap-

ping of the sensor after it is cut from the wafer in the

form of a bar. In order to precisely determine the

stripe height, electronic lapping guides are formed

simultaneously with the sensor and provide an exact

reference to the back of the stripe.

The modern spin valve head is made up of several

layers of very thin ﬁlms (10–200 A

commonly)

and small feature sizes (submicron track widths)

(see Fig. 13). These sensors require ultraclean vacuum

processing and thickness control better than 1 atomic

Figure 14

Cross section of a GMR reader (courtesy of Seagate Technology, LLC).

Figure 15

GMR sensor basics (courtesy of IBM Corporation).

579

Magnetic Reco rding Heads: Historical Perspective and Background