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

Подождите немного. Документ загружается.

a ﬁrst coating is generally prepared by chemically

plating the aluminum alloy disk with about 10 mmof

a structurally amorphous and mechanically hard NiP

alloy. This layer is then often mechanically textured

on the micrometer scale and the resulting uneven

surface of the ﬁnished disk reduces problems with

stiction in the start and stop phases of the head slider

ﬂight in systems in which the head is ‘‘parked’’ on the

disk at rest. However, the move to higher and higher

storage densities requires smaller and smaller

(B10 nm) ﬂying heights.

This is leading to smoother ﬁnishes on the disk and

the adoption of procedures such as locally focused

ion beam and laser etching to produce relatively

rough areas on which to land and park the head.

Faster rotation rates, essential to reduce access times

and increase data rates, require stronger, harder,

stiffer, and lighter substrate materials. These are usu-

ally nonmetallic to avoid mechanically induced dis-

tortion and glass, glass ceramic, aluminum, boron,

and silicon carbides, and silicon are being used or are

being explored. However, the substrate material must

be thermally stable and also process compatible. Al-

so, the nucleation and growth of the several layers

on the disk may very well depend on the substrate

material.

The various layers, apart from the NiP layer, are

now without exception put down by physical vapor

deposition (PVD) methods, nearly always sputter

deposition, at fairly fast rates of B10 nms

1

onto

variable temperature substrates. The use of PVD

techniques offers the necessary control and repro-

ducibility in the deposition process. The different

forms of magnetron sputtering that may be used are

variations on a technique that offers excellent control

of deposition rate, ﬁlm thickness, alloy composition,

substrate conditions and environment, and efﬁcient

use of the target material.

2. The Underlayer

The magnetic layer is deposited on one or more chosen

underlayers or ‘‘seed’’ layers (Figs. 1 and 2). The usual

role of the underlayer(s) is to promote the growth of a

particular crystallographic texture and grain size and

morphology in the magnetic layer. The magnetic layer

is, without exception, a polycrystalline, cobalt-based

alloy ﬁlm, typically CoCrTa,Pt that crystallizes with

a hexagonal crystal structure. The magnetically pre-

ferred orientation of the grains is with their [001] c-axis

in the plane of the ﬁlm. As discussed below, there is an

exception to this in media designed for the perpendic-

ular recording mode in which the c-axis should be

perpendicular to the ﬁlm plane. However, commercial

drives use ‘‘longitudinal’’ recording and in-plane mag-

netization is required.

Figure 2

High-resolution transmission electron micrograph of a

cross-sectional sample prepared from an experimental

thin-ﬁlm disk showing the silicon substrate, the

chromium underlayer, and the CoCrPt magnetic layer.

Figure 1

Schematic diagram showing a cross-section of one side

of a typical thin-ﬁlm hard disk.

660

Magnetic Reco rding: Rigid Media, Preparation

The heteroepitaxy between the underlayer(s) and

the magnetic layer, which is so apparently beneﬁcial

to recording performance, clearly must depend on the

lattice spacings of the layers as well as their crystal

structures. Important additions to the basic CoCr

alloy in the magnetic layer are tantalum and platinum.

The alloy is still hexagonally packed but there are

subtle changes in the c/a ratio on alloying. Chromium

has been the usual underlayer but b.c.c. alloys based

on chromium, such as CrV and CrTi, can be used

(Shiroishi et al. 1993) to improve the crystallographic

ﬁt and perhaps also to produce elastic distortions in

the magnetic layer that in turn may increase aniso-

tropy and coercivity. Growth orientations of the

cobalt alloy layer on the immediate underlayer that

have been identiﬁed (Lee et al. 1995, Grundy 1998)

are typically (10.0)Co//(112)Cr, (11.0)Co//(100)Cr.

The crystallographic texture, grain size, and mor-

phology developed in the underlayer depend on the

substrate type and process conditions. Important

process conditions are base vacuum, substrate tem-

perature, deposition rate, sputter gas pressure, and

ﬁlm thickness. Various strategies are used to promote

oriented chromium or chromium alloy layers on glass

and other substrates, including the initial deposition

of special seed layers such as NiAl (Lee et al. 1995)

under the chromium.

The requirements for the magnetic layer are that it

has optimized magnetic properties and also that it

provides the recording and replay performance re-

quired from the ﬁnished disk. The microstructure and

many magnetic properties, such as magnetocrystalline

and shape anisotropies and intergranular exchange, of

the overlying magnetic layer depend on the micro-

structure of the underlayer(s) (Grundy 1998). Their

crystallographic orientation, grain structure, crystal

lattice parameters, and thermal and chemical proper-

ties control the growth of the magnetic layer. It is

found that conditions restricting adatom mobility,

such as low substrate temperature and high sputtering

pressures (X1 Pa), encourage unoriented ﬁlms with a

rough surface. Low deposition pressures and higher

substrate temperatures (B200 1C) favor increased

mobility and oriented and smoother chromium and

chromium alloy deposits.

3. The Magnetic Layer

The improved recording and replay performance re-

quired of magnetic disks has led to the development

of CoCr-based ternary alloys, such as CoCrTa,Pt,

quaternary, and, indeed, quinternary alloys and

the deposition of polycrystalline ﬁlms with small

(20–50 nm), oriented but segregated grains (Fig. 2).

The size distribution of the grains should be narrow,

with a mean size as small as possible, but one that is

greater than the so-called superparamagnetic limit so

that the recorded signal is stable.

It is known that media noise, the dominant noise

limiting recoverable signals in replay, is reduced when

the magnetic medium is subdivided into isolated,

noninteracting particles. However, smaller particles

require a larger value of magnetic anisotropy for sta-

bility (Lu and Charap 1994). To satisfy this need,

media with increased magnetocrystalline anisotropy

are required.

Chromium was originally used to improve the cor-

rosion properties of cobalt ﬁlms but it was also found

to help in increasing coercivity and reducing replay

noise. This is achieved by segregation of chromium at

the grain boundaries producing a nonmagnetic layer

that interrupts exchange interactions between grains

or clusters of grains (Doerner et al. 1993). There is

also evidence that compositional gradients exist with-

in grains (Hono et al. 1995). The tantalum and plat-

inum additions substitute into the majority cobalt

lattice and increase the magnetocrystalline anisotropy

with a concomitant increase in coercivity. However,

cobalt and its alloys can exist in other crystalline

forms, notably a f.c.c. structure. The magnetization

of the f.c.c. form of cobalt is not very different from

that of the hexagonal phase but it has four easy axes

of magnetization along the /111S directions and the

magnetocrystalline anisotropy is much less than that

of the h.c.p. phase. Cubic grains would have a del-

eterious effect on the recording behavior of the ﬁlm

(Dova et al. 1999).

In efforts to increase coercivity even further, what

might be called the confusion principle is being ap-

plied in the investigation of even more complex qua-

ternary and quinternary alloys such as CoCrPtXY

(Akimoto and Shinohara 1998), where X and Y are

tantalum, niobium, silicon, or boron. Boron enters

the lattice interstitially and changes the c/a ratio, as

do the substitutional solutes. It is therefore probable

that anisotropy will be changed and segregation

effects made more complex.

4. Overlayers

The integrity of the thin-ﬁlm disk in the drive is pre-

served by the application of a mechanically hard

overlayer. This prevents wear of the magnetic layer in

any stop–start action. Such wear would involve loss

of the storage medium and the information it con-

tains and it would also introduce debris into the drive

mechanisms. Low friction in the system is ensured by

a ﬁnal coating of a high molecular weight polymer.

The most common hard overlayer is a thin (5–10 nm)

ﬁlm of modiﬁed carbon. Sputter or plasma beam

deposition of carbon in atmospheres containing

hydrogen or hydrocarbons can result in so-called

diamond-like carbon (DLC) ﬁlms (Cord and Scherer

2000). The presence of hydrogen in the ﬁlms favors

the formation of sp

bond orbitals at the expense

of sp

orbitals that are characteristic of graphitic

661

Magnetic Reco rding: Rigid Media, Pr eparation

bonding. The resultant density and hardness of the

ﬁlms is a function of the hydrogen content of the

ﬁlms and the relative proportion of sp

/sp

bonding.

Values in the literature are in the range 5–20 GPa.

This compares with the hardness of, for example,

tungsten carbide, WC, at 3 GPa.

Typical lubricants are the perﬂuoropolyethers, such

as Fomblin, with molecular weights of 2000 and

more. These lubricants have the necessary properties

of thermal and chemical stability, low vapor pressure,

and low surface tension (Bushan 1990). The coating is

at a monolayer thickness and binds to the carbon

overcoat by interactions between end groups in the

linear polymer and the carbon surface. The strong

binding ensures that lubricant remains between the

head slider and the disk where contact occurs and also

that it does not leave the disk when it spins at high

speed. However, some mobility is needed so that areas

from where lubricant is lost can be recovered.

5. Perpendicular Recording Media

The thin-ﬁlm media discussed so far have in-plane

magnetization and record information in the longi-

tudinal mode. A second mode of recording has been

proposed (Iwasaki and Nakamura 1977) in which the

transitions separate regions magnetized perpendicu-

larly to the ﬁlm surface. This situation requires a

magnetically uniaxial ﬁlm in which a strong, perpen-

dicular anisotropy, K

, outweighs the demagnetizing

energy, i.e., K

/2, and supports magnetization

normal to the ﬁlm. Several types of material have

been investigated, including particulate barium ferrite

ﬁlms and sputter-deposited crystalline ﬁlms of cobalt

alloyed with transition metals (Grundy and Ali 1983).

The media investigated mostly so far are cobalt

alloyed with chromium at compositions round

(Wright et al. 1985) A well oriented, co-

lumnar or equiaxed microstructure can be obtained

by growth on a variety of substrates; including rigid

substrates, such as glass, and glass coated with f.c.c.

crystalline underlayers and NiFe (the NiFe acts as a

magnetic closure layer) and amorphous underlayers

such as germanium, and ﬂexible substrates such as

polyimide and melinex. The metallic underlayers are

preferentially oriented with a /111S ﬁber texture,

which encourages heteroepitaxy in the CoCr layer,

whereas the amorphous plastics and germanium pre-

sumably favor the nucleation of the high surface en-

ergy basal planes in the cobalt alloy. The main

advantage of the perpendicular scheme is that, in

principle, the recording transition width is narrower

than for the longitudinal mode and hence higher bit

densities are possible. At the same time, the magnetic

structure is resistant to demagnetization, even at high

packing densities.

There is as yet no perpendicular recording disk

drive on the market. There is no doubt of the potential

of the scheme for very high storage densities. Calcu-

lations (Bertram and Williams 2000) have sugges-

ted that, although acceptable limits for longitudinal

recording may be upgraded to B100 Gbits in

2

(15 Gbits cm

2

), perpendicular recording may support

B500 Gbits in

2

(75 Gbits cm

2

). There have been

problems in developing a fully integrated system,

and the very rapid improvements in longitudinal me-

dia and associated GMR heads are providing strong

competition. However, the great potential of the per-

pendicular mode in resisting demagnetization and

thermal instability is rekindling a great deal of activity

in the development of this technology. It may yet

provide the transitional mode of magnetic storage

between the longitudinal scheme and patterned media.

6. Fabrication

Commercial deposition equipment attempts to re-

produce the precision of laboratory systems and yet

operate with the cost effectiveness and reproducibility

of mass production methods. They usually take one

of two forms. The ﬁrst is the in-line coater in which

disks are unloaded from a cassette in a load-lock

chamber and progress through a series of in-line

process chambers in which the various layers are

sequentially deposited onto the disk.

In this type of coater, such as the Leybold A400

system (Leybold AG, Hanau, Germany), the indi-

vidual deposition chambers for, say, chromium,

CoCrTa,Pt, and carbon depositions are separated

by individually pumped buffer chambers with nar-

row-aperture slit load locks that allow continuous

passage of the palates loaded with disks between

large rectangular magnetron cathodes. The second

system is the single-station system where most of the

layers are deposited onto the disk from a cluster set of

circular magnetrons. The Balzers CIRCULUS M12

or M14 (Balzers Process Systems, Balzers, Liechten-

stein), have 12 or 14 process stations in a circular

layout (Fig. 3). Each disk progresses round the circle

of magnetrons from the input station to a holding

cassette at the output. Each station is isolated and has

its own pumping and cold trap system, gas supply,

and measurement devices.

In both systems rapid processing demands sputter

deposition at rates greater than 10 nm s

1

, and sub-

strate heating and cooling for the various deposition

regimes. Layers are deposited with little interrup-

tion to avoid contamination and to achieve repro-

ducibility. Uniformity across the disk can in principle

be enhanced by using magnetrons with on-line ad-

justable magnet ﬁelds. The base vacuum standard

in both systems is that of conventional high vacuum,

i.e., p10

5

Pa, with partial pressures of o10

4

Pa,

of H

O and o5 10

5

Pa of N

and O

, respec-

tively. Disk production demands, of course, a clean

room environment up to semiconductor fabrication

662

Magnetic Reco rding: Rigid Media, Preparation

standards in the initial presentation of the disks and

in any subsequent testing and in packaging after fab-

rication. Commercial deposition systems can now

produce fully coated disks at rates approaching 1000

per hour.

There are continuing efforts to improve processing

procedures and to introduce new technologies in disk

fabrication (Cord and Scherer 2000). For instance,

research (Takahashi et al. 1997) has shown that

improved properties, including larger magnetic coer-

civity, can be obtained in samples deposited in systems

which have ultrahigh vacuum base pressures (o5 

7

Pa) and use very pure target materials and sput-

tering gas. This is thought to produce uncontaminated

thin ﬁlms with grains that are less faulted, which for a

given composition should have an increased anisotro-

py ﬁeld and greater coercivity. Any moves to adopt

such preparation standards commercially will need

to take account of increased costs of fabrication,

although investment in plant is essentially a one-off

expenditure. However, the required increases in stor-

age densities discussed above will probably lead to

more demanding standards in disk fabrication.

See also: Magnetic Recording: Rigid Media, Re-

cording Properties; Magnetic Recording: Rigid Me-

dia, Tribology; Magnetic Recording Devices: Head/

Medium Interface; Magnetic Recording Media:

Advanced; Magnetic Recording Technologies: Over-

view; Micromagnetics: Basic Principles

Bibliography

Akimoto H, Shinohara M 1998 Magnetic interaction in Co-

CrPtTaNb media—utilization of micromagnetic simulation.

IEEE Trans. Magn. 34, 1597–9

Bertram H N, Williams M 2000 SNR and density limit esti-

mates: a comparison of longitudinal and perpendicular re-

cording. IEEE Trans. Magn. 36, 4–9

Bushan B 1990 Tribology and Mechanics of Magnetic Storage

Devices. Springer, New York

Cord B, Scherer J 2000 Future media manufacturing—more

than just conventional sputtering. IEEE Trans. Magn. 36,

67–72

Doerner M F, Yogi T, Parker D S, Lambert S, Hermsmeier B,

Allegranza O C, Nguyen T 1993 Composition effects in high

density CoPtCr media. IEEE Trans. Magn. 29, 3667–9

Dova P, Laidler H, O’Grady K, Toney M F, Doerner M F 1999

Effects of stacking faults on magnetic viscosity in thin ﬁlm

magnetic recording. J. Appl. Phys. 85, 2775–81

Grundy P J 1998 Thin ﬁlm magnetic recording media. J. Phys.

D: Appl. Phys. 31, 2975–90

Grundy P J, Ali M 1983 The magnetic and microstructural

properties of CoCr thin ﬁlms with perpendicular anisotropy.

J. Magn. Magn. Mater. 40, 154–62

Hono K, Yeh K, Maeda Y, Sakurai T 1995 Three dimensional

atom probe analysis of a sputter-deposited CoCr thin ﬁlm.

Appl. Phys. Lett. 66, 1686–8

Iwasaki S, Nakamura Y 1977 An analysis for the magnetization

mode for high density magnetic recording. IEEE Trans.

Magn. 13, 1272–7

Lee L L, Laughlin D E, Fang L, Lambeth D N 1995 Effects of

Cr intermediate layers on CoCrPt thin ﬁlm media on NiAl

underlayers. IEEE Trans. Magn. 31, 2728–30

Lu P, Charap S H 1994 Magnetic viscosity in high density

recording. J. Appl. Phys. 75, 5768–70

Shiroishi Y, Hosoe Y, Ishikawa A, Sugita Y, Ohno T, Ohura M

1993 Magnetic properties and read/write characteristics of

CoCr(Pt,Ta)/(CrTi,Cr) thin ﬁlm media. J. Appl. Phys. 73,

5569–71

Takahashi M, Kikuchi A, Kawakita S 1997 The ultra-clean

sputtering process and high density magnetic recording

media. IEEE Trans. Magn. 33, 2938–43

Wright C D, Lacey E T M, Grundy P J 1985 Optimisation

of the magnetic properties of CoCr ﬁlms for perpendicular

recording applications. J. de Phys. 46, C6-101–4

P. J. Grundy

University of Salford, Manchester, UK

Magnetic Recording: Rigid Media,

Recording Properties

Rigid disk drive performances are being pushed

ahead at such an enormous rate that disk recording

has become the ﬂagship for state of the art develop-

ments in digital recording. Other applications of re-

cording beneﬁt from and follow this lead but at a

lesser pace. This article will describe briefly the

theoretical background to digital recording and iden-

tify which properties of the recording media are

important.

Recording media used in rigid disk drives are with-

out exception of high coercivity and very thin. These

requirements have been known for many years and

have grown to be accepted as a result of early exper-

imental and theoretical work completed largely in the

Figure 3

Schematic plan of the Balzers CIRCULUS M12

deposition system showing the disk input and output

stage and the 12 magnetron positions.

663

Magnetic Reco rding: Rigid Media, Recording Properties

1960s and 1970s (e.g., Bertram 1994, Comstock 1999,

Hoagland 1963, Hoagland and Monson 1998, Mid-

dleton 1987, Middleton 1989, Richter 1999). Since

then, technology has marched ahead and so has the

understanding of the nature of the record and replay

processes. The values of the parameters involved have

also changed dramatically but the basic requirements

remain for thin media of high coercivity.

Figure 1 shows a simple diagram illustrating the

complete digital record–replay cycle. A current into

the record head is magnetizing a section of the me-

dium to the right, whereas previously a current of the

opposite polarity caused a magnetization to be writ-

ten to the left. The magnetization transition between

these states is used to represent a bit of stored infor-

mation and a sequence of recorded transitions, or

absences of them at particular locations along the

track, is used to represent the recorded data. The

width of a transition is indicative of the space needed

to store a bit of information and so this is an impor-

tant parameter ultimately determining the density at

which bits are stored. When a transition moves under

the replay head the stray ﬁeld arising from its diver-

gent magnetization couples into the head structure

and generates an output voltage pulse. This article

considers the nature of the recorded magnetization

transitions and the corresponding output waveforms,

and draws conclusions about the requirements for

high-density recording.

1. The Recording of a Transition

In this theory (Middleton 1987, Middleton 1989), only

components of magnetization in the plane of the re-

cording medium, which have been assumed to arise

from the inﬂuence of the planar components of the

head ﬁeld, will be considered. Figure 2 shows detail of

how the head ﬁeld magnetizes a medium to the right

which had previously been magnetized to the left. The

magnetization variation in the transition is determined

by the hysteresis loop M(H

) of the medium and the

total ﬁeld H

it experiences. The latter is made up of

head ﬁeld H

plus any demagnetising ﬁeld H

pro-

duced by the medium. The head ﬁeld is given by:

7tan

1

g=2 þ x



þ7tan

1

g=2  x



ð1Þ

where H

¼ niZ=g is the ﬁeld along the surface of the

head (y ¼0) where n is the number of turns, Z the head

efﬁciency, i the record current, g the gap length, and x

and y are deﬁned in Fig. 2.

The recorded magnetization in the transition is as-

sumed, for simplicity, to take the form:

M ¼

tan

1

x  x



ð2Þ

where the center of the transition is at x ¼x

, near to

the trailing edge of the gap, and the magnetization

varies smoothly from a value of þM

on one side to

Figure 1

Schematic of the record and replay processes.

Figure 2

Close up of the head gap, the ﬁeld distribution, and the

recorded transition to the right of the gap.

664

Magnetic Reco rding: Rigid Media, Recording Properties

M

on the other. M

is the remanent magnetization

of the medium and a is proportional to the transition

width and is known as the transition width para-

meter. Using Eqn. (2) the demagnetizing ﬁeld arising

from a thin medium of thickness d is:

dðx  x

pððx  x

þ a

ð3Þ

The gradient of the magnetization in the transition

is given by:

dMðH

dðH

þ H

ð4Þ

where dM(H)/dH is the slope of the hysteresis loop

and dH

/dx is the total ﬁeld gradient. The head ﬁeld

gradient in a medium spaced y ¼d from the head is

given by dH

/dx ¼aH

/d where the quantity a has

values, in the literature, ranging from (2/p) to 1 de-

pending on the size of the head gap and the magni-

tude of the head current. Here a ¼ O3=2 corresponds

to maximization of head ﬁeld gradient for a narrow

gap head. At the center of the transition, where

x ¼x

, M ¼0, and H

¼H

, the slope of the hysteresis

loop is w and so, using Eqns. (3) and (4), it is easily

shown that:

¼ w

O3H





ð5Þ

Solving this equation leads to a transition width of:

a ¼

O3pwH

ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

O3pwH

O3pH

ð6aÞ

When the hysteresis loop is steep sided and w is large:

a ¼

ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

O3pH

ð6bÞ

This is the transition width in the presence of the

head ﬁeld but as the medium moves away from the

record head, and the inﬂuence of its ﬁeld wanes,

the transition demagnetizes to a greater width. When

the transition then moves to the vicinity of the replay

head it induces within it magnetic charges which im-

age those within the medium. The net reduced pole

strength reduces the demagnetizing ﬁelds and nar-

rows the transition widths to values approaching

those given by Eqn. (6). Thus Eqn. (6) is a reasonable

ﬁrst approximation of the ﬁnal transition width.

When the head ﬁeld gradient is large, it can, ac-

cording to Eqns. (6) and (3), lead to a transition suf-

ﬁciently narrow as to give rise to a demagnetizing

ﬁeld which is larger than the coercivity of the record-

ing medium. Such a ﬁeld cannot be consistent with a

hysteresis loop and so the transition widens to reduce

the ﬁeld to less than or equal to the coercivity. The

minimum possible transition width is when the de-

magnetizing ﬁeld is equal to the coercivity. Thus

equating the maximum of the demagnetizing ﬁeld,

given by Eqn. (3) at (xx

) ¼a , to the coercivity, the

minimum value for the transition width is:

a ¼

2pH

ð7Þ

This is sometimes known as the self-demagnetizing

limited transition width.

The transition width occurring in a recording me-

dium is the larger of the write- or self-demagnetizing

limited values given, respectively, by Eqns. (6) or (7),

and these are plotted in Fig. 3 as a function of (M

). Experimental results conﬁrm these relationships

and in each case the important ratio in determining

recording performance is (M

d/H

) which should,

obviously, be kept small.

2. The Replay Process

During the replay process (Middleton 1987, Middle-

ton 1989) transitions pass under the replay head

which is sensitive to their presence. The variation of

sensitivity with position under the head can be rep-

resented by a sensitivity function which has the same

mathematical form as the ﬁeld distribution per unit

current, H

, obtained from Eqn. (1). The replay head

voltage generated as the magnetization gradient

moves through the sensitive region of the head is:

eð

xÞ¼m

vwd

þN

N

ðx 

xÞ

ðx; dÞdx ð8Þ

Figure 3

The broken line shows the write limited transition width,

from Eqn. (6b) with d ¼40 nm and w ¼N, while the

solid line shows the demagnetizing limited case from

Eqn. (7).

665

Magnetic Reco rding: Rigid Media, Recording Properties

where m

is the permeability of free space, v the me-

dium velocity, w the track width,

x ¼vt and t is time.

Substitution of Eqn. (1) for the head sensitivity func-

tion along with the magnetization from Eqn. (2), into

Eqn. (8) and evaluation leads to the output pulse

shape:

eð

xÞ¼

2m

vwnZM

7tan

1

g=2 

d þ a



þ 7tan

1

g=2 

d þ a



ð9Þ

This pulse shape is identical to that of the sensi-

tivity function of the head except that the spacing d is

replayed by (d þa). Three pulse shapes are shown in

the inset of Fig. 4, where the amplitude e(0) and

width p

are shown on one of them. From Eqn. (9)

these quantities are given by:

eð0Þ¼

4m

vwnZM

7tan

1

g=2

d þ a



ð10aÞ

and

¼ 2

ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

ða þ dÞ

þðg=2Þ

ð10bÞ

It is easily seen from Eqns. (10a) and (10b) that

maintaining a narrow transition width keeps the

pulse amplitudes high and the pulse widths narrow, in

line with the requirement for high density recording.

At high densities when the transitions are close to-

gether, the total voltage can be predicted using linear

superposition. This says that the total waveform

(

x)arising from a sequence of transitions is given by

the sum of the isolated pulse shapes as:

xÞ¼

n¼þN

n¼N

ð1Þ

eð

x  nl=2Þð11Þ

where l/2 is the spacing between the transitions.

Figure 4, inset, shows three overlapping pulses,

their sum, and the consequent reduction of output.

The variation of normalized output voltage with

packing density, for a long sequence of pulses, is

shown in Fig. 5 for a typical thin-ﬁlm recording me-

dium. This type of graph is often referred to as a

roll-off curve and is one of the standard measure-

ments of recording performance. The full line repre-

sents theoretical predictions and the points represent

experimental values. Their coincidence shows the

validity of linear superposition but such agreement is

not found in all circumstances (Richter 1999) because

intertransition interactions can cause their widths

and/or shapes to change with a consequent loss of

agreement.

The summation, Eqn. (11), of pulse shapes, Eqn. (9),

can be represented in closed form and at short wave-

lengths approximates to:

ð0Þpe

2pðaþdÞ=l

ð12Þ

Now the transition width appears in an exponential

term and its role in causing a fall of output becomes

very strong. This re-emphasizes the need to keep a,

and therefore (M

d/H

), small so as to keep the output

high at high recording densities.

The simpliﬁed recording theory given here has, in

its full exposition, had much success in explaining

Figure 4

Inset: overlapping pulses and their sum showing a

reduction of output and the occurrence of peak shift.

Main graph: peak shift as a function of packing density

for pulses calculated using Eqn. (9) with a ¼50 nm,

d ¼40 nm, and g ¼220 nm.

Figure 5

A roll-off curve showing normalized output voltage as a

function of packing density. The solid theoretical curve

is by linear superposition and the points are for a thin

ﬁlm medium with M

d ¼20 mA, H

¼159 kA m

1

a ¼50 nm, and with d ¼40 nm and g ¼220 nm.

666

Magnetic Reco rding: Rigid Media, Recording Properties

practical experience and guiding the way forward

with prediction of future performances. While this

theory has shortcomings with the simplicity of its

basic assumptions and analysis, more sophisticated

numerical modeling has not altered the basic media

requirements.

The requirement for narrow transitions for high

density operation has led to the widespread adoption

of thin ﬁlm media with high coercivity. Other re-

quirements include media with steep sided hysteresis

loops, small to medium head spacings and heads with

narrow gaps. Practical media satisfying these require-

ments are now around 20 nm or less in thickness, and

are reaching over 250 kA m

1

in coercivity. They are

invariably of alloys of cobalt or cobalt–nickel with

small additions of tungsten, platinum, or other met-

als. Future media suitable for the predicted very high

storage densities are expected to rise to around

400 kA m

1

3. Recorded Magnetization Transitions

Direct observation of recorded magnetization distri-

butions in thin media obtained by either bitter colloid

techniques or electron microscopy reveal, in plan

view, a transition structure which varies in the cross

track direction as shown in Fig. 6. This is known as a

‘‘saw-tooth’’ pattern whose nature is not indicated by

Eqn. (2). Early interpretations of these patterns were

that they came about to spread the magnetic poles

over longer transitions, thereby reducing demagnet-

izing ﬁelds and self-demagnetizing energies. This may

be so in wider transitions where bitter colloids have

shown the presence of demagnetizing ﬁelds around

the saw-teeth, and some record theories have fol-

lowed this interpretation.

However, micromagnetic modeling of narrow tran-

sitions in higher coercivity ﬁlms has shown that the

saw-tooth like structures arise as a result of part of

the magnetization distributions forming ﬂux closure

patterns. These do not contribute ﬁelds which inﬂu-

ence the recording process, and subtraction of them

from the total magnetization shows the resulting

charge carrying transition to have little cross track

structure, not unreasonably represented by Eqn. (2).

Indeed, micromagnetic calculations carried out on

self-demagnetization limited transitions found that

cross track average magnetizations, where ﬂux clo-

sure components sum to zero, have distributions not

unlike Eqn. (2) and transition widths which are close

to Eqn. (7). This goes some way to explaining why

record theory based on such simple assumptions, as

used here, continues to apply even when the detailed

structures of transitions appear so complicated.

4. Practical High Density Recording

It has been shown above that at high densities the

output voltage is significantly reduced and therefore

so is the signal to noise ratio. At high densities there

is also a loss of timing accuracy in the replayed

waveforms because the superposition of pulses leads

to a total waveform where the peaks are shifted in

position. This phenomenon, known as peak shift, in-

creases rapidly as densities increase and this is illus-

trated in Fig. 4. These timing shifts, in concert with

the reduced signal to noise ratios, lead to increased

error rates in the recovered data.

5. Conclusions

This chapter has concerned itself with basic record

theory to identify those properties of recording me-

dia, thinness and high coercivity, which are needed to

allow high-density recording to take place. It has

discussed these ﬁndings in view of the rather compli-

cated magnetization distributions which are found in

practice, and which appear at ﬁrst sight not to be

consistent with the theory quoted.

See also: Magnetic Recording: Rigid Media, Tribo-

logy; Magnetic Recording Devices: Head/Medium

Interface; Magnetic Recording Measurements; Mag-

netic Recording Technologies: Overview; Micro-

magnetics: Basic Principles; Micromagnetics: Finite

Element Approach

Bibliography

Bertram H N 1994 Theory of Magnetic Recording. Cambridge

University Press, Cambridge

Comstock R L 1999 Introduction to Magnetism and Magnetic

Recording. Wiley, New York

Hoagland A S 1963 Digital Magnetic Recording. Wiley, New

York

Hoagland A S, Monson J E 1998 Digital Magnetic Recording,

2nd edn. Krieger

Middleton B K 1987 The recording and reproducing pro-

cesses. In: Mee C D, Daniel E D (eds.) Magnetic Recording:

Technology. McGraw-Hill, New York, Vol. 1, Chap. 2,

pp. 22–97

Figure 6

Diagrammatic representation of the plan view of the

appearance of a ‘‘saw-tooth’’ transition.

667

Magnetic Reco rding: Rigid Media, Recording Properties

Middleton B K 1989 The recording and reproducing processes.

In: Mee C D, Daniel E D (eds.) The Magnetic Recording

Handbook: Technology and Applications. McGraw-Hill, New

York, Chap. 2, pp. 24–100

Richter H J 1999 Recent advances in the recording physics of

thin media. J. Phys. D: Appl. Phys. 32, 147–68

B. K. Middleton and M. M. Aziz

University of Manchester, Manchester, UK

Magnetic Recording: Rigid Media,

Tribology

Magnetic recording disk drives are generally designed

so that the slider ﬂies over the disk during steady-

state operation of the drive. The ﬂying of the slider is

based on the formation of a thin air bearing which

separates the slider and the disk. If the disk drive is

shut down, the slider is in contact with the surface of

the disk. During start/stop of the drive, sliding mo-

tion between the slider and the disk occurs until the

air bearing is formed and the slider is supported

completely by hydrodynamic pressure forces.

The design of a magnetic recording slider during

steady-state operation is a problem of hydrodynamic

lubrication, requiring the solution of the compressible

Reynolds equation. However, during start/stop of a

drive, contacts between the slider and the disk occur,

requiring the understanding of boundary lubrication

and wear. Since a magnetic disk drive may be turned

on and off more than 50 000 times during its life,

great care has to be taken in the design and imple-

mentation of the head/disk interface in order to

guarantee reliable operation of the disk drive.

In order to obtain maximum signal resolution and

highest recording density, it is important that the

spacing between the slider and the disk be as small as

possible. In fact, from a magnetics point of view, the

spacing between disk and slider (or ‘‘head’’) should

be zero. This situation, generally described as ‘‘con-

tact recording’’, causes problems with friction and

wear. Thus, a compromise must be made so that the

spacing between the slider and the disk is small

enough to give adequate signal strength, but is large

enough to establish a strong air bearing to enable

ﬂying of the slider over the disk. Flying heights of

typical magnetic recording sliders are of the order of

20 nm.

1. Reynolds Equation

The equation governing the hydrodynamic ﬂying of a

magnetic recording slider is the compressible Reynolds

equation, given by

QðKnÞph

@ph



QðKnÞph

@ph



¼ 6mU

@ph

þ 6mV

@ph

þ 12m

@ph

ð1Þ

where p is the pressure, h is the spacing between slider

and disk, m is the viscosity of the lubricant ﬁlm, and U

and V are the disk velocity components in the x-and

y-direction, respectively. The factor Q is a function of

the local Knudsen number, Kn (ratio of mean free

path and local spacing), and is an indication of the

deviation of the air ﬂow between slider and disk from

continuum ﬂow theory.

The Reynolds equation is a nonlinear partial dif-

ferential equation which can be solved numerically

using ﬁnite difference and ﬁnite element methods

(Hu and Bogy 1996, Wahl et al. 1996). Two of the

most important considerations in the design of mag-

netic recording sliders are the requirements for a

slider to ﬂy at constant ﬂying height from the inner to

the outer disk radius, and the need of a slider to be

independent of altitude changes.

2. Slider Desi gn

A typical air bearing slider is shown in Fig. 1(a).

The slider consists of two side rails, a leading edge

bridge, and a subambient cavity. Also shown is the

positive pressure buildup over the slider rails and the

subambient pressure region over the cavity portion

(Fig. 1(b)).

3. Disk Characteristics

For ﬂying heights of the order of 20 nm, the rough-

ness of a disk has to be well controlled. The peak-to-

valley roughness of a typical disk is of the order of

Figure 1

(a) Typical magnetic recording subambient pressure

slider and (b) the pressure distribution over the slider.

668

Magnetic Reco rding: Rigid Media, Tribology

10–20 nm. A roughness plot for a hard disk measured

by atomic force microscopy is shown in Fig. 2. The

substrate of choice for 65 mm laptop disk drives is

glass, while the substrate for 95 mm disk drives

is aluminum. In the case of aluminum substrates, a

10–15 mm thick layer of nickel phosphorous is ﬁrst

deposited on the disk, followed by a 10–50 nm thick

layer of magnetic alloy. On top of the alloy is a thin

carbon ﬁlm that, in turn, is coated with a 1–2 nm

thick lubricant ﬁlm.

4. Tribology of the Head/Disk Interface

4.1 Friction and Stiction

During shutdown and startup of a disk drive the

slider is in contact with the disk, resulting in fric-

tion and wear at the interface. The variation of the

coefﬁcient of friction is shown in Fig. 3 as a function

of the number of start/stop cycles (CSS) between

slider and disk. In general, the coefﬁcient of friction is

found to increase with the number of start/stop cycles

and is inﬂuenced by environmental conditions such as

temperature and humidity (Zhao and Talke 1999a).

The coefﬁcient of friction shows, in general, a

sharp increase during startup with a well-deﬁned

peak approximately 100–200 ms after motion starts.

This peak in the friction force is generally described

as ‘‘stiction’’.

The stiction force, F

, encountered at the startup of

a disk can be expressed as follows:

¼ mðW

þ F

ÞþF

ð2Þ

where m is the coefﬁcient of friction, W

is the sus-

pension load of the slider, F

is the meniscus force,

and F

is the viscous force in the lubricant ﬁlm.

The meniscus force, F

, is related to the presence

of lubricant ﬁlms or adsorbed water ﬁlms on the disk

surface. A number of models have been developed to

predict stiction as a function of design parameters

and environmental conditions (Gui and Marchon

1995, Zhao and Talke 1999b).

4.2 Laser Zone Texturing of Disks

A large stiction force can prevent the motor from

starting the disk drive after shutdown of the drive. In

order to prevent failure of the head/disk interface

owing to excessive stiction, texturing of the disk sur-

face in the start/stop region of a magnetic disk has

been implemented using laser texturing. Laser tex-

turing is performed by focusing a laser beam onto the

disk surface and creating a number of small and

evenly spaced ‘‘bumps’’ in the ‘‘landing’’ zone of the

disk by localized melting of the disk surface (Fig. 4).

The height and size of the individual laser ‘‘bumps’’

Figure 3

Coefﬁcient of friction as a function of contact start/stop

cycles.

Figure 2

Surface roughness of hard disk measured with atomic

force microscopy.

Figure 4

Crater-shaped laser bump.

669

Magnetic Recording: Rigid Media, Tribology