Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

J. Phys. D Appl. Phys. 25 (1992) 1-23.

Printed in the UK

1 1

REVIEW ARTICLE

Magnetic characterization of recording media

R W Chantrellt and K O'Grady#

*
t

Physics Department. University of Keeie, Keele, StaffordshireST5 5BG, UK School of Electronic Engineering Science, UCNW, Bangor, Gwynedd LL57 1U T UK Received 1 May 1991, in final form 2 September 1991

Abstract. Magnetic recording media are important technology materials whose

behaviour remains very poorly understood. However, magnetic measurements form a very important part of the practical characterization of recording media. This review outlines the current techniques o f magnetic characterization and considers the effects of current research developments.

1. Introduction

Magnetic recording has been around in some form since early this century, although most of the growth has occurred during the past 2&30 years. The field now encompasses a diversity of applications such as the consumer market of audio and video recording, in addition to the storage of digital information on rigid and floppy disks. This varied range of materials and techniques is often referred to as magnetic information technology (MINT). Magnetic storage of information, in whatever form, has many innate advantages, particularly in terms of erasability coupled with long-term stability of the stored data, and is likely to remain viable for the foreseeable future. However, far from being a mature and established technology, MINT is presently facing enormous challenges, particularly as regards the increased requirements of information storage density and faster access times. These challenges are presently being met by developments in particulate recording media, although many alternative materials and techniques are under consideration. These include thin film media and magneto-optic recording, and the vertical hloch line technique which has considerable theoretical promise. It is not possible to cover the many aspects of these materials in a brief review therefore this review will concentrate on the properties and physical problems confronting the development of existing media. It is not possible, however, to consider the physical properties and characterization of the medium in isolation from the recording process. For this reason we start with an overview of the analogue and digital recording processes, which provides the necessary background
0022-37271921010001 + 23 $03.50 @ 1992 IOP Publishing Ltd

for the full consideration of the scientific problems presented by the media themselves. It should also he acknowledged that although the physics of recording media is a central problem in the field, magnetic recording spans many scientific and engineering disciplines, and covers areas outside the scope of this review. There are, however, many excellent hooks covering the whole of the field of magnetic recording, particularly those by Jorgensen (1980) and Mee and Daniel (1987).
2. The recording process

A recording system consists of a transport mechanism for the medium and one or several transducers by means of which information is transferred to and from the medium. In addition to this there is also external circuitry to process the information. In this review we shall consider only the physics involved with the storage and replay of the information, which depends crucially on the recording technique. It is useful to consider in detail the most common techniques, which are broadly categorized as saturation recording of digital information and analogue recording of audio and video signals. Each of these techniques makes many different demands on the recording medium and so it is necessary to consider the behaviour of recording media in the specific context of the intended application.
2.1. Digital recording

This is the most easily visualized form of recording. The medium, in the form of a rigid or floppy disk
1

R W Chantrell and K OGrady

LOGIC and TIMING C I R C U I l S

WRITE AMPLIFIERS

Motion O f m e d i m

!
Region 1
M

n,
M 2 Region 2

recorded

signal

replayed signal
(b)

Figure 1. (a) Schematic of the digital recording process using a ring head. (b) Schematic write and read waveforms

for the digital recording process.

carrying a thin magnetizable film is rotated at high speed beneath a magnetizing head, as is shown in figure l(a). Another important configuration is the standard reel-to-reel transport of magnetic tape. This is a relatively inexpensive form of digital information storage, but it does not have the rapid access times offered by the rigid or floppy disk and is only useful for long-term storage and back-up facilities. In all cases the storage of a binary digit is achieved by means of a current pulse in the recording head. This consists of a gapped toroid of soft magnetic material, wound with an energizing coil to which the current pulse is applied. The result is a fringing field in the medium, which can be made larger than the coercive force, thus causing the medium to be magnetized. On readback, the recorded information gives rise to an induced voltage produced by the Hux changes in the head. The write and read
2

waveforms are of the form shown in figure l(b); electronic processing is applied afterwards to the read signal in order to retrieve the original square pulse. The digital recording process has been investigated analytically by Middleton and Davies (1984) and Potter (1970) and many others. Magnetically the process is similar to the acquisition of the remanence following the application of a nonsaturating field, the isothermal remanent magnetization (IRM) to be defined later, although importantly, the process takes place in a very short timescale, often less than a microsecond. A further complication is the spatial variation of the head field throughout the medium, which is taken into account in the vector model of the recording process (Ortenburger and Potter 1979, Pottcr and Beardsley 1980, Beardsley 1982). These models are limited by an imperfect understanding of the recording medium itself, for reasons to be discussed fully later. Similar considerations also apply in the case of analogue recording. Before considering analogue recording it is important to point out some of the purely mechanical problems of magnetic recording which arise because of the nature of the head/medium interface. Consider, for example, a Winchester disk, which may have a rotational speed in excess of 3000 RPM. The Hying height of the head over the medium is of the order of tenths of microns, so there is an obvious requirement of a clean surface for the medium. The motion actually creates an air bearing effect which reduces wear, but on the other hand introduces a spacing loss which is an important limit on the resolution of recorded data. In the case of tapes there is often contact between the head and medium, resulting in considerable problems of wear, which can be minimized by the inclusion of lubricants in the tape formulation. There is also a pressing need to keep the surface roughness as low as possible. Such tribological problems form an important area of study which can make a vital contribution to the further development of recording technology. For the most part, however, this review will concentrate on the magnetic behaviour and characterization of recording media, since in this area considerabie progress is possible on a fundamental level.

2.2.
AC

AC

bias recording

bias recording, although somewhat complex as a magnetic process, is an extremely neat solution to the problem of maximizing the linearity and response of the medium in analogue recording. The problem is illustrated in figure 2. Curve a here shows the isothermal remanent magnetization (IRM) as a function of the applied field. This measurement will be described in more detail later, but is essentially the DC response of the system. It can be seen that the response is very non-linear. The first attempt to linearize the response involved the application of a DC bias field Hd to ensure that the medium was operating in the optimum range.

Magnetic characterization of recording media


IQ

P
8

-x \

Z 6

i
r

I I

3
2
1

Figure 3. Schematic hysteresis loop defining the primary magnetic quantities used for the characterization of recording media.

8 IQ 12 applied field / n e

1I

I6

I8

20

magnetization, showing the extended linear region lor the latter case.

Figure 2. An experimental compar.son 01 the tsothermal (curve a) and anhysteretic (curve b) remanent

The technique of AC bias is a later and much more effective method. Essentially the technique involves the application of a large, high-frequency bias field which is reduced to zero in a timescale over which the signal does not vary appreciably. The closest physical measurement to this process is anhysteresis, in which the signal field remains completely static while the AC field is reduced. Anhysteretic measurements are also shown in figure 2 (curve b). In low fields the response is very linear, and characterized by a large value of the anhysteretic susceptibility xi.. In practice, as the tape leaves the region of the head the signal and AC field strengths reduce simultaneously resulting in a reduction of xi. by perhaps a factor of two. This, however, remains the most successful method of recording analogue information. The techniques of analogue recording utilize two formats. Audio recording uses a stationary head, whereas in video recording the head is rotated at an angle to the tape transport direction (helical scan recording). This makes optimal use of the whole area of the tape for recording, as is necessary in order to store the high information density required for recording video signals. A similar technique is used in the rotary digital audio tape (RDAT) system. AC bias recording is a very important technique, but one whose physical basis is very poorly understood. This is essentially because of the strong dipolar interparticle interactions which are known to dominate the process. A further complication is the lack of understanding of the noise spectrum of a recording medium which is also strongly dependent on interactions. In

fact, the interaction problem is central to the fundamental physical investigations of these materials, as will become clear later. A further difficulty is that in general tapes are characterized in terms of their static magnetic properties, whereas anhysteresis and the AC bias recording process are clearly dyriamic phenomena. The relation between the static measurements and the practical (dynamic) response of the medium during the recording process is still very much an unsolved problem. Real progress in this area will only be possible given a much better fundamental understanding of the physics of recording media. The current high level of activity in the field is very promising in this respect.

3. Basic materials requirements

We shall shortly go on to consider the physics of the most common types of recording media. Before doing this, however, it is worthwhile stating explicitly the basic specifications for a useful recording medium in order to provide a link with the technical aspects of recording and additionally to introduce the primary magnetic quantities by which a medium is characterized. The basic physical attribute of a magnetic material which makes it usable for information storage is non-equilibrium behaviour. This can be considered as introducing a memory which is clearly essential to any information storage technique. In the case of a magnetic material the non-equilibrium behaviour is represented by the hysteresis loop obtained by measuring the magnetization M as the applied field H is cycled. Figure 3 is a schematic hysteresis loop illustrating the primary magnetic quantities. These are the coercivity Hc, the saturation magnetization M , and remanent magnetization M,. From this we can also define the squareness of the loop S, = M,/M,. A useful recording medium requires a large enough value of M , coupled with a large value of squareness, the actual values depending on the details of the application. The requirements relating to the coercivity are somewhat more stringent. Ideally H , should he large, in order to
3

R W Chantrell and K OGrady

resist demagnetizing effects, although clearly increasing H , makes the write process more difficult. The practical value of H , depends upon the application. The smallest values of H , tend to be in low-density applications. Here, coercivities range from -300400Oe and tend to increase with increasing information density, up to values in excess of 1000 Oe as the specification for video recording. No magnetic material has a perfectly well defined field at which magnetization reversal takes place, since this would result in a square hysteresis loop which is never observed in practice. It is perhaps useful to consider a recording medium as comprised of characteristic activation volumes, each of which has its own 'switching field' at which it reverses its state of magnetization. This parameter depends on detailed considerations of the material and the reversal mechanism as will be discussed later. What is important in the present context is the fact of a distribution in these intrinsic coercivities, known as the 'switching field distribution' (SFD). The SFD naturally tends to reduce the squareness of the hysteresis loop, and a convenient measure of the width of the SFD is obtained from the Williams-Comstock (1971) construction which is shown in figure 3. The parameter (1 - S*), which characterizes the width of the SFD, is essentially a measure of the slope of the hysteresis loop at H = H,:

(1) More recently, Bertram (1986) gave a more general relation for a medium with a finite value of (1 - S*):
a = d(l+*)+[(

a = 61r/Hc.

d(l

- S') ) 2 + - ]

Ir6d 'I2

(2)

nQ

nQK

where Q is a parameter related to the head field and d is the head to medium separation. Generally speaking equation (2) is dominated by the second term with only a weak dependence on (1 - S*). It should be noted that if equation (2) predicts a smaller transition width than the demagnetization limited case (equation (1)) the latter should he chosen. Thus in saturation digital recording the relevant static properties are I, and H,. From the previous discussion it is is clear that the primary characterization of recording media should be in terms of the static properties, and this is certainly the case in practice. However, there are many additional important magnetic and electromagnetic properties which characterize a medium. For the remainder of the review we shall concentrate on introducing those properties and relating them to, on the one hand, the record/replay process itself, and on the other the underlying physics of the medium. Before this, however, we describe briefly the preparation of particulate and thin film recording media, which introduces many important factors relevant to their characterization.

This parameter has been recommended (Koester 1984) as the hest single parameter for the characterization of recording media on a practical basis. Other methods for measurements of the SFD, based on remanence curves will be discussed in detail later. The SFD has a considerable effect on the analogue recording process as shown by Koester et a1 (1981) who found that the high-frequency output was drastically reduced by increasing (1 - S*) in a series of tapes prepared by mixing cobalt modified iron oxides. The optimum bias current was found to decrease at a rate of -5.6dB per unit (1 - S*). The origin of this effect is ascribed to the intrinsic spread of coercivities associated with (1 - S*) coupled with the overbiasing phenomenon to be discussed later. Generally a small (1 - S') value is required for good recording properties. In digital recording the information storage density is limited by the minimum transition width as determined by demagnetizing effects. Essentially, the maximum demagnetizing field for a thin medium with an arctangent magnetization transition is Hd = -16/2n where a is the transition width (here, and in the rest of the paper, the symbol I is used to represent the magnetization in EMU cm-) ( M = 4x0). Consequently the important parameters are the coercivity H , film thickness 6 and remanence I,. Potter (1970) has shown that the minimum transition width is (for a square hysteresis loop with 1 - S* = 0)
4

4. Preparation of magnetic recording media


4.1. Particulate media

Elongated particles consisting of a single magnetic domain are the most common magnetic recording media in use today. A wide variety of particles are used and in this section only a brief description of the preparation techniques commonly used is possible. Of course, much of the detail of the particle preparation processes is proprietary to the major manufacturers and :hi ieadir should bc iiiidci iio ilktsion as :o :hc level of care and sophistication that is involved in the bulk preparation of the highly uniform particles which are in current use. To prepare particles which have good magnetic properties and dispersibility, direct precipitation techniques should ideally he used, which will result in particles which have smooth surfaces and are highly uniform in terms of size, anisotropy, etc. Unfortunately many of the particles which have the appropriate properties cannot be prepared directly and more elahorate techniques are required. These techniques are reviewed below.
4.1.1. Preparation of elongated y-Fe,O, particles. The

most widely used material for all recording applications consists of elongated gamma ferric oxide particles. The particles used are between 0.4 and 0.8pm long with

Magnetic characterization of recording media

lFeOlOH

Goethite

i orlhorhombir I

Feso, no 0 i1 , , . " , 4 Na OH I

Figure 4. The principal stages in the production of elongated y-Fe203 particles lor recording media.

aspect ratios between 6:l and 12:l. The basic reaction to form the particles is shown in figure 4. Needle-shaped iron oxyhydroxide FeOOH is grown on precipitated seeds from a solution containing iron salts (typically FeCI2). Usually orthorhombic aFeOOH, synthetic geothite is used. The dehydration and reduction processes usually require temperatures up to 700C which can result in the particles sintering together. Accordingly the FeOOH particles are usually coated with chemical complexes often incorporating other elements such as zinc, nickel or tin which inhibit sintering and are also instrumental in determining the size and the aspect ratio of the particles. It is this step in the preparation which is vital to the morphology, dispersibility and magnetic properties of the final product. Thus the precise nature of the additives and the method of their utilization is proprietary to the manufacturers and is not revealed. The reduction process is usually undertaken in a hydrogen or hydrogen rich atmosphere produced using a variety of reducing oils. The final controlled oxidation of the synthetic magnetite to the defect spinel y-Fe20, is undertaken by heating to between 300 and 400C. This stage of the process much also be carefully controlled to avoid the transformation of y-Fe203 to The final value of the coerweakly magnetic a-Fe203. civity can be enhanced at this stage by failing to complete the oxidation process leaving a final product with a composition (Fe,O,),(Fe,O,), --I (synthetic betholite). The final product is mechanically compressed between rollers to reduce the porosity of the particle although care must be exercised at this final stage to avoid damaging the particles which results in a well known reduction in the coercivity. During this final stage initial dispersants and other chemicals may be added to coat the particles thereby assisting the eventual dispersibility.

In this process it is essential to recall that the properties of the final product are essentially determined by the initial seed formation process. In recent years synthetic lepidocrocite (y-FeOOH) has also been used commercially as the precursor for y-Fe203. The eventual product is a brown (tan) free-running powder with coercivities which range from 260 to 385 Oe. For particles which include a limited amount of Fe30a the coercivity can be as high as 425 Oe. In general the final value of the coercivity for the particles when dispersed and coated onto the substrate can differ from the original powder by up to 20 Oe. For a full review of the preparation of y-Fe203 particles the reader is referred to Bate (1980). Table 1 summarizes the typical properties and applications of y-Fe203particles.
4.1.2. Metal particles. In a purely historical context it

is interesting to note that the first magnetic recording tape was coated with elemental iron particles produced by the thermal decomposition of iron pentacarbonyl. However, small particles of the ferromagnetic elements are highly reactive and in fact may be pyrophoric when exposed to air. They may also be attacked by the dispersants or binders used in the tape coating process. Accordingly they were not used until the recent increase in demand for media with high moment and coercivity for applications such as digital audio tape and 8 mm video. Elongated metal particles can be prepared by two basic routes: either the reduction of oxides or salts of the metals or the decomposition of organometallic compounds or complexes. The former method usually involves the reduction of any of the precursors of yFe,O, particles using hydrogen or other organic reducing agents. Alternatively iron, cobalt or iron-cobalt alloy particles can be prepared by the borohydride
5

R W Chantrell and K OGrady


Table 1. Typical properties and applications of y-Fe203particles. Source: Bayer UK technical data

sheet. Application Audio and computer tape, floppy discs Computer tape, floppy discs Low noise audio and video tape Professional audio tape Particle length (pm)
0.7 0.6

Aspect ratio
6:l 8:l 10:l 7:l

M, (=4nls)

Coercivity
(Oe)

(Gauss)
4000 4000 41 00 4100

0.5 0.4

285 305 385 300-380

reduction of solutions containing the metal salts (Oppegard et a1 1961). The latter process includes carbonyl decomposition, formate decomposition, etc, but these processes are expensive due to the cost of the precursors and have not found wide application. The key to the preparation and use of metal (usually iron) particles lies in the passivation of the particle surface. Due to the reactivity of the metal, organic compounds are in general unsuitable and other metal or oxide coatings are preferred. For example Aonuma (1975) incorporated chromium and potassium sulphate in the borohydride reduction of metal salts to inhibit oxidation and produced a tape with H , = 1000 Oe and a loop squareness of 0.81. The preferred and currently the only utilized technique for the passivation of iron particles for recording is the controlled oxidation of the particle surface, leaving a core of metal which comprises approximately 50% of the particle volume. The oxidation reduces the effective value of the specific magnetization of the bulk material to around 150 EMU g-' but still results in a remanence of double that of a fully oxidized particle. As is the case for y-Fe203the coercivity is entirely controlled by the particle elongation. However, due to the high moment, particle alignment during the coating process is greater resulting in a loop squareness which is generally higher than that achieved for y-Fe203and comparable to the values obtained with C r 0 2 (section 4.1.4).
4.1.3. Cobalt-modified iron oxides. The substitution of

cobalt ions for iron gives rise to a large increase in the magnetocrystalline anisotropy of iron oxides. The cobalt saturated oxide COO Fe203, cobalt ferrite, is unique amongst the inverse spinel ferrites since it has cubic anisotropy with three easy axes along the cube edges. Cobalt ferrite has a large intrinsic anisotropy constant, K , , of 2 x 106ergcm-3 so it is hardly surprising that cobalt inclusions in y-Fe20, can raise the coercivity to the region of 1OOOOe or greater if required. The inclusion of cobalt into iron oxides can be achieved by a variety of techniques. y-Fe203 can be doped uniformly (body doped) with cobalt by adding cobalt salts to the solution before the precipitation of FeOOH or by precipitation of CoOOH onto the FeOOH particles after formation. The remainder of the process is as described in section 4.1.1. The resulting particles exhibit multi-axial or isotropic magnetic behaviour and are in general spherical or have
6

axial ratios <2:1. It has been suggested that such particles may have advantages as recording media since the magnetization can more closely follow the head field, thus using perpendicular components of the field (Lemke 1982). Unfortunately the magnetic properties of bodydoped particles exhibit a strong temperature dependence and all currently available commercial materials are usually surface doped by one of two techniques. Surface doping can give rise to strong surface anisotropies which give the required increase in coercivity for high density (=high frequency) applications, without producing the strong temperature dependence effects of body-doped materials. Surface-doped y-Fe,03 particles are produced by precipitation of cobalt hydroxide onto the surface of already prepared y-Fe203 particles. The cobalt ions are found to remain on the surface unless annealing above 100C occurs when diffusion into the particle occurs resulting in properties analogous to body doped materials. It is also possible to prepare epitaxial doped yFe203in which cobalt ferrite is doped onto the surface by precipitation of Fe2+ and Co2+ onto the particle surface. The doping is undertaken simultaneously in a similar manner to cobalt doping alone, although care ions into must be taken to avoid diffusion of the FeZ+ the body of the particle thereby destroying its uniaxial anisotropy (Witherell 1984). The quantity of cobalt ions doped onto the surface of the particles varies between 1 and 5% depending on the coercivity required, hut it should be noted that the anisotropy remains uniaxial and hence the maximum loop squareness remains as 0.5 for an unoriented sample. The enhancement of the coercivity arises from the coupling of the magnetocrystalline anisotropy of the cobalt surface layers with the shape anisotropy of the original particle. Whilst the increase in coercivity with doping is observed using quasi-static measurements a significantly smaller increase is found using dynamic measurements in a E-H loop tracer. For a full discussion of time- and temperature-dependent effects in cobalt-doped particles see Eiling (1987) and for full details of the preparative techniques see Bate (1980). The potential applications for cobalt surface-doped y-Fe203 particles are as high storage density media such as the IBM 3480 format, video tape, etc. Table 2

Magnetic characterization of recording media Table 2. Typical properties and applications of cobalt modified iron oxides. Source: Bayer UK technical data sheet. Application High bias audio tape, Betamax video tape VHS video tape High-density data tape, video tape Particle length (pm)
0.45 0.4 0.4

Aspect ratio
10:1 1O:l
8:l

M, (=4n/.)

(Gauss)
4400 4400 4300

Coercivity (Oe)
595 630 760-980

Table 3 . Typical properties and applications of CO2. Source: DuPont product information. Particle Application length (pm) Data tape 0.4-0.6 Video tape 0.3-0.4 Video tape 0.4 Aspect M, (=4n/,) Coercivity ratio (Gauss) (Oe)

1O:l
1 O : l 1 O : l

4390 4390 4390

47C-670 470-620 485-540

gives examples of various particles which are available and their typical properties.
4.1.4. Chromium dioxide. At the present time elon-

gives some typical properties and applications of CrO, particles. The properties are controlled by varying the diameter of the needles, which is achieved by careful control of additives and preparation conditions. It is also noteworthy that due to the near perfect nature of the crystallites, the particles are relatively easy to align during the tape production process, giving squareness ratios of up to 0.9 which is higher than can be achieved with iron oxide media and which results in a higher output from the tape.
4.1.5. Barium ferrite particles. Barium ferrite is unique

gated particles of chromium dioxide (CrO,) are second only in importance to y-Fe,O, as particles for recording media. CrO, particles have found application in video recording and high-density data recording media, although Co-doped y-Fe,O, particles are now finding applications in these areas. The common use of CrO, arises because it is possible to grow particles in the form of needles with very parallel sides, excellent morphology and an absence of dendritic growth. There are several methods available for the preparation of CrO, particles, but the main technique used is the hydrothermal reduction of CrO, (Swoboda et al 1961). The trioxide is mixed in equimolar ratio with water and heated to between 400 and 500C at a pressure of around 5 x lo6 Pa. The reaction takes place in a double-walled vessel with an inner surface of platinum or glass in order to prevent reaction with the vessel. The size and axial ratio of the particles is controlled by the addition of small quantities of metals, the most commonly used being antimony and tellurium (typically 0.15%). These additions result in perfectly crystallized, highly uniform particles between 0.3 and 0.4pm i n length with an axial ratio of 1O:l. Iron is also often added to increase the crystalline anisotropy. Finally, since C r 0 , is an oxidizing agent, the particles are coated with ,hromium oxyhydroxide (CrOOH). In hulk production the particles have coercivities which range from 400 to 700Oe, and are used in a range of applications. There is one unusual application of CrO,: since it has a very low Curie point (-125 "C) it can he used to make thermoremanent copies of video tape material by heating the tape, for example by laser heating, whilst in contact with a (high Curie point) master tape carrying the signal to be copied. Table 3

among recording media since the particles exhibit strong crystal habit, reflecting the HCP crystal structure, and exist as hexagonal platelets. Due to their crystal structure the particles exhibit strong (uniaxial) magnetocrystalline anisotropy. For pure barium ferrite this results in coercivities in the range 2-3 kOe. For practical application the coercivity must be reduced. This is achieved, with little effect on the saturation magnetization, by doping with cobalt and titanium to produce particles of composition BaFe,, _,Co,Ti,O,, which, when x = 0.75, have a coercivity of 900 Oe. The particles can be prepared by the glass crystallization method (Kuho et a1 1982) in which a mixture of BaO, Fe,03 and B,03 is melted and rapidly cooled between metal rollers. Dopants are included in the melt in oxide form. The boron provides a glassy matrix in which the barium ferrite is formed. After heating to crystallize the barium ferrite particles, the boron and often excess barium is removed with hot acetic acid leaving hexagonal platelets of BaO. 6Fe,O3 approximately 0.1 pm in diameter and less than 0.05 pm thick. Despite the presence of CO and Ti the easy axis of magnetization remains the hexagonal axis of the particles, giving rise to particles which have opposing crystalline and shape anisotropies. Consequently the particles have a positive temperature coefficient of coercivity of around 4.8 Oe "C-', in contrast with all other recording media particles, for which H , decreases with increasing temperature. Furthermore the platelet aspect of the particles means that potentially they can be incorporated in tapes for perpendicular recording. However, the dispersion and orientation of these particles in difficult, and to date they have found limited application in flexible disc media, although development of high-density tape products for video and data applications is expected.
7

R W Chantrell and K OGrady

4.2. Production of magnetic recording tape

backing 1011

The production of a magnetic recording tape is essentially a two stage process. Firstly a dispersion of the pigment (particles) is produced by grinding the particles usuallv , in a mixture of solvents and disoersants. lubricants, anti-wear agents, anti-static agents and binders. Secondly the dispersion is coated onto (usually) a polyester film and treated in such a way as to produce a very smooth continuous coating with a high gloss and more importantly free of voids. In this second stage a magnetic field is applied, to align the particle easy axes whilst the coating is wet.
~~~

dlrperdon

E +

knife

4.2.1. Preparation of particle dispersion. The particles

are dispersed by ball milling or, more usually, sand milling in a combination of solvents possibly including several from MEK, MIK, cyclohexanone, tetrahydrofuran, dioxane, or several other compounds. Included in this stage of the process are usually a 'cocktail' of dispersants which may be anionic, for example sulphonates; cationic, such as quaternary ammonium salts or imidazolinum salts; or amphoteric, such as betaines or amino oxides. The particles are milled typically for several hours with the dispersion quality monitored by viscosity measurements or other techniques. Once the particles are dispersed, cross linked polymers or other binders are incorporated together with lubricants such as silicone oils or hydrocarbon oils or more recently solid lubricants such as polyfluor hydrocarbons or carbon black are used, the latter having the advantage of reducing the dielectric constant of the medium. In addition other anti-static agents are added to produce a resistivity less than 10 R cm-*. Other wear resisting agents, e.g. alumina or carborundum, are also included. The final dispersion is then diluted with appropriate solvents and resins to produce a 'lacquer' with the correct viscosity for coating onto the polyester film.
4.2.2. The coating process. Coating technologies in

Icoating ro11
metering roll

I b l

barking mli

dispersion

doctor
pressure- fad

I C 1

Figure 5. The tinal coating process by which the particulate dispersions including binders and lubricants is laid down on the medium. (a) Gravure coating. (b) Knife

coating. (c)Reverse roll coating.

magnetic recording and other technologies are highly advanced versions of an essentially very simple process. The polyester film or web over half a inetre wide is passed through rollers and fed into a region where the dispersion is simply poured onto its surface across the full width of the web. The web then passes beneath a knife edge which leaves a very uniform thin layer of the dispersion on the film. If required, the particles are aligned as the web passes by an array of permanent magnets and then dried. The drv or almost drv coatine is then pressed between heaied rollers in- a carefully controlled environment to produce a very smooth, high gloss finish. This pressing or calendering also serves to minimize voids in the tape. The tape is then split by knife blades and wound onto spools, the outer few thicknesses being discarded. Often the tape, particularly if it is to be used in digital applications, is tested by writing and reading data as it is wound.
8

Currently it is at this point that the characterization of the medium begins. Samples of tape are routinely taken for a range of detailed analyses of their magnetic, recording, adhesive, morphological properties and it is a tribute to the formulation chemists and coating engineers that they can relate failures in the final product 10 any one of the many ingredients or processes in the tape production procedure. Of course the precise details of the formulations used are proprietary as indeed are the precise coating techniques used. In figure 5 schematic representations of the most common coating techniques are shown.
4.3. Thin film recording media

The thickness of a magnetic medium plays a crucial role in determining the maximum recording density for longitudinal recording. It is shown in a straightfoward argument by Bate (1966) that the pulse width is some function of I,tlH, where t is the thickness of the medium. Clearly it is not possible to reduce the remanence I, too far since this would reduce the flux available at the replay head. Therefore, there is considerable advantage in recording on films of smaller

Magnetic characterization of recording media

component of magnetization. The magnetizing field is created by


a single-pole head.

Figure 6. A possible recording system utilizing the perpendicular

thickness. Using ultrathin films and a magnetoresistive head, workers at IBM have recently reported experimental CoPtCr thin film media capable of operation at an areal density of 1 Gbit per square inch which represents the current world leading position (Tsang et a 1 1990). I Although thin film recording media have significant advantages in recording in the longitudinal mode, an equally large impetus for their development has been the possibility of perpendicular magnetic recording. This has been a theoretical possibility for some time, and has the advantage of increased recording densities due to the reduction of the demagnetizing effects on the transition region between bits. The major breakthrough in this area was made by Iwasaki and coworkers (1977, 1979) who proposed not only a new (cobalt-chromium alloy thin film) recording medium, but also a novel single pole head which is more efficient than a ring head in the perpendicular mode. The recording system is shown in figure 6. The essential requirement for a perpendicular recording medium is an intrinsic anisotropy large enough to support a perpendicular component of magnetization against the large demagnetizing energy (of the order of 2 ~ 1 : ~We ) . can define a figure of merit as the ratio of the anisotropy and demagnetization energy density, i.e. K/2x& which must be B l for a medium capable of supporting a perpendicular component of demagnetization. Thus the basic material requirement must be for a high anisotropy and a perpendicular texture, in the case of perpendicular recording. We shall now briefly review the major techniques for the production of thin film media for perpendicular and longitudinal recording, with emphasis on the relation between microstructure and magnetic properties which remains an important problem in this field.
4.3.1. Sputtered metallic media. RF sputtered thin films of cobalt-chromium alloys have been extensively investigated with a view to applications as perpendicular recording media. Sputtering is carried out in a vacuum

system evacuated to 1O-6 to W7 Torr and then filled with a working gas (e.g. argon) to IO-' to 10-3Torr. The ionized gas atoms are accelerated by means of a DC or RF field and gain sufficient energy to knock off surface atoms from the target material. This forms a plasma from which deposition onto the substrate occurs. Under the correct preparation conditions sputtered CoCr films have an HCP crystal structure with a large intrinsic magnetic anisotropy and a c-axis orientation perpendicular to the film, resulting in a high figure of merit for the perpendicular recording mode. Optimum conditions are obtained by adjustment of the most important parameters of the sputtering process which are the RF sputter voltage, the inert gas pressure and the substrate temperature. Important factors in the characterization of perpendicular media are those which represent the degree of perpendicular texture. Examples of this are the hysteresis loop, torque magnetometry and the x-ray rocking curve method, all of which will be discussed in detail later. The structural properties and morphology of thin films are of enormous importance. These are investigated by a variety of techniques including x-ray diffraction and electron microscopy. This is particularly revealing about the microstructure of the films which seem to consist of columns oriented perpendicular to the film and spatially distinct. A micrograph of the structure of a Co7,Cr2, film (Grundy er a1 1984) is shown in figure 7. Such observations give rise to a description of thin film media as strongly coupled particulate systems. Magnetic isolation of individual grains is assumed here to be enhanced by migration of Cr to the grain boundaries. This is in marked contrast to models of thin films as continuous media in which domain wall motion is the dominant magnetization process.
4.3.2. Metal evaporated (ME)thin films. The production

technique here consists of evaporating a ferromagnetic alloy in a vacuum chamber and depositing the atoms continuously onto a moving plastic substrate. A review
9

R W Chantrell and K OGrady

Figure 7. Electron micrograph of a cross section of a CoCr

thin film clearly showing the columnar microstructure (courtesy of Dr P J Grundy).

related to the microstructure-Tada et a1 (1986) have shown that small amounts of Pr give rise to finer grains in CoNi with a corresponding increase in corrosion resistance. Corrosion resistance of the films is often characterized by monitoring the decrease in remanence with time at an elevated temperature and humidity. A decrease in M , is often associated (see, for example, Sugaya and Tomago (1983)) with an increase in H,. presumably via a decrease in the average grain size. Clearly corrosion is a problem which needs to be overcome. It is possible to reduce corrosion by overcoating with sputtered protective films of Si, although this has the disadvantage from the recording viewpoint of increasing the head to medium separation and consequently the spacing loss. Currently, metal evaporated tapes are available as commercial products, principally in Japan. Some o f these materials consist of a number of layers in which angle of incidence alternates, producing a rather complicated microstructure. These tapes represent an important development.

Figure 8. Schematic of the apparatus used for the production of metal evaporated films.

of the behaviour of ME film for recording has been given by Sugaya and Tomago (1983). The process is critically dependent o n many factors including the surface roughness which has a bearing on the radiation effects on the film, which-must he minimized by effective shielding of the source. This is shown in figure 8 which illustrates the apparatus used for the production of ME films. Another important factor is the angle of incidence of the atoms, because of a shadowing effect which tends to produce a particulate structure oriented at an angle to the film. The coercivity and remanence are critically dependent on the angle of incidence of the atoms. The presence of reactive gases such as OXYgen also has an effect on the magnetic properties as shown for example by Gau et al(1986) for CoNi films. This was attributed to a shape transition in the columnar structure and the formation of oxides in and around the polycrystalline columns, influencing the nucleation and pinning of domain walls. A problem with thin metallic films is corrosion in air and other reactive atmospheres. This is perhaps
10

4.3.3. Electroless deposition. Films of Co/Ni/P with the inclusion of small amounts of rare earth elements can be prepared by electroless deposition. This is essentially an autocatalytic chemical reaction in which a catalytic substrate is immersed in a solution of metal salts. Small palladium particles adsorbed onto the substrate are used for catalysis. When the initial deposition has occurred the reaction becomes autocatalytic, since CO and Ni themselves act as very good catalysts. Electroless deposition has one major advantage; it is a continuous process and therefore much more convenient than sputtering and evaporation which must be carried out as batch processes in vacuum. It is, however, a complex chemical reaction which is difficult to control, and presents problems with the maintenance of chemical concentrations during the production process. Generally the plating process produces films which are isotropic in the plane and useful as longitudinal recording media. An interesting d e v e k q " t , however is the addition of Mn and 'Re in small quantities to a CoNiP alloy (Goto eta[ 1984, Takano and Matsuda 1986). This had the effect, as demonstrated by magnetic measurements and x-ray rocking curves, of producing a significant Perpendicular anisotropy. As a consequence, electroless Plating could be a viable technology for the Production of Perpendicular recording media. At present, however, the use of electroless plating is restricted to a relatively low volume of outf rigid disks for standard longitudinal digital reput o cording.
5. The magnetic properties of recording media
5.1. Static propertiesthe hysteresis loop

Because of their relative ease of measurement the static properties of recording media are an established means

~-

Magnetic characterization of recording media by which their behaviour is characterized. As discussed previously there is a relation between the primary magnetic quantities (Hc, 1 - S*, etc) and their recording performance, although this is by no means fully understood. In this section we outline the basic static magnetic properties and the principal parameters which govern them, for the major types of recording media. The archetypal demonstration of hysteresis is the M / H loop (with M = 4nf the magnetization in Gauss) which is generally measured by means of a vibrating sample magnetometer, or a E-H looper whose technical details will be discussed later. The hysteresis loop (see, for example, figure 3) is a non-equilibrium phenomenon arising from local energy barriers which can arise via many mechanisms dependent on the type of recording medium. In particulate media the dominant mechanism is the magnetic anisotropy intrinsic to the particles. This itself can arise from a combination of magnetocrystalline, shape and strain anisotropies. The particle size in particulate media is small enough for the particle to be considered as a single domain with a magnetic moment p = f,,V, with V the particle volume. The origin of the high coercive force in particulate dispersions can be explained by the StonerWohlfarth (1948) theory. For a system of particles with uniaxial anisotropy and easy axes oriented at an angle 0 to the field H , the energy has the form

-h
Case

Aligned

1 . 5

Figure 9. Calculated magnetization cuwes using StonerWohlfarththeory for systems with random and aligned easy axes. The beneficial effects of alignment include the increased coercivity and loop squareness.

where K is the anisotropy energy density and q is the angle between the easy axis and the magnetic moment. Numerical solution of equation (3) with variable H gives the theoretical M / H curve for a given orientation. Particularly illuminating is the case 0 = 0 (a perfectly aligned system) which retains all the essential physics while being analytically soluble. It is straightforward to show that under these circumstances there exist two energy minima separated by an energy barrier

A E = KV(1 - H/Hk)*

(4)

where Hk ( = 2 K / l S b ) is the anisotropy field. Clearly, when H = Hk the energy barrier vanishes and the magnetic moment can make a transition between the minima. Thus Hk corresponds to the coercive force at absolute zero. In practice the coercive force is generally lower than this value for several reasons. The first of these is the angular distribution of easy axes or orientational texture. Numerical solution of equation (3) in the original paper by Stoner and Wohlfarth (1948) demonstrates this effect, as is shown in figure 9 which shows the coercivity of a randomly oriented system to be just less than half the aligned value. There are, however, more important and fundamental reasons for the relatively small observed values of coercivity. The first of these is the effect of thermal agitation which gives rise to transitions for a finite value of the energy barrier with a consequent reduction of H , as was shown by Gaunt (1968). The effects of thermal agitation are most apparent in the

observed time dependence and print-through of recording media which will be discussed in detail shortly. The temperature effects are strongly dependent on the particle size. As the particle volume is reduced, a critical value V, is reached at which the energy barriers are n o longer large enough to sustain a remanent magnetization in the presence of thermal agitation. The particles exhibit reversible magnetic behaviour and are termed superparamagnetic by Bean and Livingston (1959). Clearly it is crucial for the particles of recording media to have volumes greater than V,. A second major factor contributing to the reduction of H, is the mode of nucleation of the magnetization reversal. The simplest reversal mechanism is coherent rotation which is dominated by the exchange energy and in which the spins of individual atoms remain parallel during the reversal. This mechanism tends to take place only in very small particles. In larger particles the demagnetizing energy is correspondingly large and incoherent reversal modes in which the surface pole density is reduced, become important. Perhaps the most extensively investigated mode is the curling mode described in terms of its micromagnetic basis by Shtrikman and Treves (1963). This predicts a decrease in coercivity with the square of the particle diameter. Reversal mechanisms in individual particles is very much a current problem. Della Torre (1985,1986) carried out computer simulations which show that nonuniform demagnetizing effects at the ends of the particles can drive the nucleation process. A similar idea lies behind the flipping model of Knowles (1986). Solution of this problem is necessary if realistic models of recording media are to be developed. Generally, the approach to this problem is via the numerical solution of the micromagnetic equations for a particle. Currently, this field of numerical micromagnetics is very important and a number of advances have been made in the understanding of magnetization reversal in both particles and thin films. However, this is beyond the scope of the current review. For a review
11

R W Chantrell and K OGrady

of micromagnetics as applied to particles the reader is referred to the work of Schabes (1991). The final major factor determining the magnetic properties of particulate media is the dipolar interaction between particles which gives rise to a variety of phenomena, chiefly (as regards the static magnetic properties) a decrease in coercivity with packing density. This is well documented experimentally (Martin and Carmora 1968, Umeki et al 1981, Morrish and Yu 1955, Corradi and Wohlfarth 1978) and has been extensively investigated theoretically, most recently by computer simulation (Lyberatos and Wohlfarth 1986, Knowles 1985). Interactions present an important practical problem in the production of recording media, since the degree of dispersion of the particles by the m e t h d s oii:lincd picvvouslj; inus: be very carefully controlled in order to produce a consistent microstructure. The static properties of thin film recording media are also imperfectly understood. The columnar microstructure (see figure 7) often observed in films produced by evaporation and sputtering (Weilinga and Lodder 1986, Grundy and co-workers 1983, 1984) seems to suggest a description of these media in terms of a system of strongly coupled fine particles. Models such as those of Andra et al(1984) and Victora (1987) are based on this premise, the latter giving encouraging agreement with experimental data. However, the observa:iGns of doinain structures in fi:ii~s(Okhoshi ci al 1983, Schmidt e f al 1985) support the models of thin film media which take domain nucleation and the motion of domain walls as the central factors in the magnetization process. Hoffmann (1986) reviewed the experimental situation and concluded that films with low coercivities are continuous, but that there seems to be a transition to particulate behaviour a t the higher coercivities which are required for high-density recording media. Thus, the basic ideas of fine particle magnetism appear likely to be applicable t o thin films usable as recording media. Successful models of longitudinal thin film media viewed as strongly (sometimes exchange) coupled grains have also been developed (Zhu and Bertram 1988, Miles and Middleton 1990).
5.2. Remanence curves

-0.01 -0.015

c
0 500
1000 applied
field

I, IHI

I
2000
2500
3000

1500

IOeI

Fijliie :e. (a) Piiiicipai remanence curves jaiier Sprati ei a/ (1988)). The definition of the remanence coercivity H, and alternative estimate H: are given. In principle, for a non-interacting system H: = H,. (b) Switching field distribution obtained by differentiation of the DC demagnetization remanence curves of figure lO(a).

field distribution, as shown in figure 10(b) which gives an example of the SFD determined by this method from the demagnetization remanence curve. A further characteristic of the remanence curves is their dependence on many-body interaction effects. This was first demonstrated by Henkel(l964) who plotted &(H) against I , ( H ) for a variety of magnetic materials. In the absence =f m-ary-b~dy effcc:s i : has been shown (Wohlfarth 1958) that the reduced remanence curves are related by

Although the primary standards for the characterization of recording media are derived from the M / H loop, further useful information can be derived from the remanence curves. The principal remanence curves are of the form shown in figure lO(a). The isothermal remanence I,(H) is obtained after the application and removal of a field H with the sample initially demagnetized. The DC demagnetization remanence I,(H) is obtained from the saturation remanent state by the application of increasing demagnetizing fields. Generally, these curves are normalized to the saturation remanence I,(m). The reduced remanence is defined as I = I/Ic(m). By differentiation of the remanence loops one obtains a detailed representation of the switching
12

I#)

1- 2 i , ( ~ ) .

(5)

The experimental data showed a pronounced deviation from linearity, which was ascribed to many-body effects. Similar effects have also been observed in NdFeB permanent magnets by Pinkerton (1986) and Gaunt et al (1986). The effects are also present in recording media as demonstrated by the experimental work of Spratt et al (1988). Corradi and Wohlfarth (1978) observed a similar effect but represented it in terms of a single parameter, the interaction field factor IFF = (H, - H:)100/Hc where H : is an estimate, of the remanence coercivity H , obtained from the r,(H) curve as shown in figure lO(n). The IFF is a measure of the deviation from linearity and is potentially a useful par-

Magnetic characterization of recording media Street and Woolley (1949) who used the term magnetic viscosity to describe the effect. The phenomenon originates in thermally activated transitions over local energy barriers, provided in the case of particulate systems by the intrinsic anisotropy. In the presence of thermal agitation a particle bas a characteristic relaxation time for magnetic reversal given by the Arrhenius-N6el law:

c =foexp(-AE/kT)
-0.21 0

(6)

200

400

800 1000 applied field I Oe 1

600

1200

1400

Flgure 11. Modified Henkel plot for a CO-P thin film (after Kelly et a/ (1989)).

ameter characterizing the strength of interactions in a recording medium. Certainly the observation of an increase in IFF with packing fraction by Corradi and Wohlfarth is consistent with this hypothesis. Kelly et ai (1989) used a similar technique to investigate C-P thin films. The approach essentially measures the deviation from linearity in the Henkel plot y - @ , where f yis the using the parameter 6r = j measured value of reduced DC demagnetization remanence and @< is the value predicted from the IRM curve using equation (5). A positive Sf indicates that interactions have a tendency to stabilize the magnetized state. The data for a C e P film are given in figure 11. The low-field region shows an initially positive S f , which distinguishes the behaviour of the film from standard particulate media in which d is generally negative. However, this is followed by a rapid change to a negative S i during the demagnetization transition region. This is indicative of strong cooperative reversal which has been predicted by Hughes (1983). The use of Henkel plots and 6 i i s becoming increasingly common in the investigation of interaction effects f ai (1990) in particulate and thin film media. Mayo e used these techniques to examine the dispersion of barium ferrite particles and found a strong correlation between the form of the interaction effects as measured by the techniques and the degree of dispersion characterized by the milling time. Clearly Henkel plots are a useful device in the characterization of recording media. The theoretical basis has been examined by Fearon et ai (1990) using a Monte Carlo simulation to examine the departure from linearity of the Henkel plot for a system of particles dispersed on a lattice. The work predicts behaviour similar to that observed experimentally and confirms a strong dependence on the magnetic and physical microstructure. This is also in agreement with experimental observations. 5.3. Time-dependent magnetization and print-through Time dependence of magnetization occurs in all magnetic systems and was first observed some time ago by

where AE is the energy barrier and fo 109s- is a characteristic frequency factor. From equation (6) we can firstly estimate the critical volume for superparamagnetic behaviour mentioned previously. In zero field the particle energy barrier is KV and on setting r equal to the time of measurement f we can solve for V to give the criterion:
Vp = (In tfo)kT/K. (7) Taking f 100 s for quasi-static measurements we have VP-25kT/K. This simple criterion enables a minimum volume to he estimated for particles to exhibit ferromagnetic behaviour and hence to be usable as recording media. In fact it is possible to refine this estimate somewhat by consideration of the printthrough phenomenon as will be shown shortly. Consideration of the magnetization dynamics of individual particles leads to the expectation of an exponential time decay of the magnetization with a characteristic time given by equation (6). In fact this behaviour is never observed experimentally; in practice it is found that the decay is more closely logarithmic, characterized by the time dependence coefficient S = dM/d In f. This is true for a wide variety of materials, specifically weakly interacting particulate dispersions (OGrady et ai 1981) and recording media (Sharrock and McKinney 1981, OGrady and Chantrell 1986, Oseroff e f ai 1985). The origin of this behaviour lies in the distribution of energy barriers which exists in all systems and which consequently gives rise to a spread of relaxation times through equation (6). It is the summation of the exponential decays which gives rise to an approximately logarithmic relation. In order to illustrate this, consider a particulate dispersion at the remanent state after previous saturation. The system is assumed to have a distribution of volumes j ( V ) dV representing the fraction of the total magnetic volume having particle volumes between V and V dV. The situation is shown schematically in figure 12. The remanent magnetization arises from those thermally stable particles with V > V,. However, from equation (7) it is clear that the critical volume increases linearly with In t and this-results in a decrease in the remanent magnetization 0~ f(V,) In f. The energy barrier appropriate to the determination of the relaxation time is generally dependent upon the applied field, the simplest example being a fully aligned system for which AE has the form given by equation (4). Using this expression it is possible to

13

R W Chantrell and K OGrady

>

the

Figure 12. The distribution of the particle volumes showing origin of magnetic viscosity in terms of the time dependence of the critical volume V,. generalize the calculation of V , to include an external field and thereby define a critical volume V,(H) given
hv -,

The corresponding expression for the time dependence is (OGrady et al 1981)

where S , is the maximum time dependence and U is the standard deviation of the particle size distribution (assumed lognormal). Discussion of time dependence effects is generally based on the empirical relation (9) which is often used to represent experimental data. Use of equation (9) has been criticized by Aharoni (1985) on the grounds that it must fail in the regions of small and large f. However, equation (9) remains a useful relation over time intervals as large as 2-3 decades. Clearly S = S ( H , T ) . At a constant temperature the form of the variation wiih field is similar to that shown in figure 13, having a maximum value S , , , which occurs at a field close to the coercivity. Recently Oseroff et al (1985) measured the variation of S , , , with temperature, which can be explained in terms of a model (Charap 1988) which takes into account the distribution of anisotropy fields. In the context of the characterization of recording media it is interesting to relate the time dependence phenomenon to two important practical problems, namely print-through and the variation of the coercivity with measurement frequency. The term printthrough refers to the unwanted transfer of a signal from one layer of tape to another due to the stray field. Print-through tends to increase with time and temperature which suggests a relation with the timedependence phenomenon. On this basis print-through is simply the long-term response of the medium to a
14

small applied field. The print-through phenomenon has recently been investigated in detail by Flanders and Sharrock (1987) who give an explanation involving a time-dependent coercive force which is a natural consequence of the time decay of the magnetization. In this work the print-through was found to correlate well with the time dependence of the magnetization, although in this case the behaviour was characterized by a plot of M against (In t). n = 1 and n = 1 both gave closely linear relations, n = 4 being used by Flanders and Sharrock to effect a comparison with the print-through measurements. From the previous explanation it is clear that the particles involved in the print-through process are those with volumes just above the critical volume for super- paramagnetism whose relaxation times will he strongly influenced by small external fields. This suggests that the small-field time-dependence would be a useful parameter characterizing recording media, although such measurements are very rarely made. In order to minimize the print-through it is necessary to ensure that the volume distribution is narrow, consequently ensuring a minimal number of particles in the size range just above V,. Essentially this means keeping the switching field distribution as narrow as possible SFD since the is intimately related to the distribution of particle volume. Measurements of time-dependent effects are technically somewhat difficult, so it is interesting to take the relation with the SFD one stage further. Early studies of time dependence proposed the following empirical relation

= constant -

S In f

S = XirrHf (10) where xicris the irreversible susceptibility and H , = kT/VI$,, is a fluctuation field Characterizing the thermal agitation. For a fully aligned system of uniaxial particles it has been shown by Chantrell el al (1986a) that H f can he calculated analytically, giving S = x s ( l - H/H,)H,/50 (11) where in this case xicris defined as the differential of either of the principal remanence curves. This is particularly useful since xirc is a readily accessible experimental .measurement. Although equation (11) refers to a special case it has been shownto give the correct form of the variation of S with H for a partially aligned tape (Uren er al 1988) although the magnitude of S is not correctly predicted. More recently, a more detailed investigation by el-Hilo et al (1990) has shown that taking into account a distribution of anisotropy fields (which to some extent also represents the angular dispersion) gives a much better agreement with experimental data. This is a fruitful area for further research as regards the characterization of recording media since it would appear that relatively simple measurements of I,(H) can potentially provide information on printthrough via equation (11). Since xir,is formally identical to the switching field distribution this makes the

Magnetic characterization of recording media

-0.5

H I kOe 1

-1.0

-1.5

-2.0

Figure 13. The variation of the time-dependent behaviour with field and temperature. There is a characteristic variation with field having a maximum close to the coercivity H, (data for CrOZtape after Oseroff et a/ (1985)).

SFD even more a central parameter characteristic of the behaviour of recording media. Coverdale ef a1 (1990) examined theoretically the variation of the fluctuation field with applied field for fine particle systems where the anisotropy axes are not necessarily aligned parallel to the applied field and found a complex variation having a minimum in H , close to the coercivity. In a separate study de Witte er al (1990) examined experimentally the variation of Ht and the related activation volume parameter VAC determined by H , = kT/(VAcIsb) for systems of fine cobalt particles. Reasonable agreement is obtained with the calculations of Coverdale et al, confirming the validity of the activation volume approach. Further work on recording media .particles indicates that VAC is sensitive to the reversal mechanism, a fact which is potentially important in the study of all recording media since the magnetization reversal mechanism is a central problem. The value of the activation volume is significantly smaller than the particle volume, consistent with the work of Flanders and Sharrock. This is presumably due to the incoherent reversal modes, which do not involve simultaneous switching o f the whole particle. A further consequence of the relaxational behaviour of recording media is the variation of coercivity with frequency which has been observed by Sharrock f and McKinney (1981), Sharrock (1984) and Corradi e al (1987). This is important since measurements of H, are normally made quasi-statically in vibrating sample magnetometers, yet the magnetization process during digital recording now takes place over timescales of the order of a microsecond. Thus the static characterization of recording media must be used carefully

since their actual dynamic behaviour when used in recording systems can differ widely from these measurements. Much work still needs to be done in order to establish the relationship between static and dynamic properties more firmly. This is possibly a long-term aim since the relationship is likely to be governed by a combination of intrinsic particle properties and manybody effects, which make a rather complex contribution to the dynamic behaviour of recording media. Clearly significant progress has been made recently in the understanding of time-dependent phenomena. However, further study in this area is important since it is the long-term coercivity which will determine the lower limit of particle size usable in recording media (Sharrock 1990). Also it would appear that it is the behaviour of subunits of the particles that must be considered since they determine the value of S and ultimately the archival stability of the stored data. This presents a very important area for further research.
6. Dynamic properties

In this section we concentrate on those properties which are most relevant to the performance of analogue recording. The response of the medium is determined by its anhysteretic behaviour which is an alternating field process to be described in detail shortly. We shall also consider the noise of the medium which is a limiting factor in its performance.
6.1. Anhysteretic behaviour

The anhysteretic remanent magnetization (ARM) of a system is obtained by firstly applying a large AC field
15

R W Chantrell and K OGrady

HAC and small DC field Hoc to the sample. The AC field is then reduced to zero after which the DC field is removed. For relatively small values of Hoc, the ARM is directly proportional to Hoc and the response is ~ characterized by the anhysteretic susceptibility x , = dARM/dHoC. The highly linear response is ideal for recording analogue information in contrast to the DC response (characterized by I , ( H ) ) which is very nonlinear. The technique of AC bias recording is rather more complex than the anhysteretic process described here. However, anhysteresis remains the closest related physical measurement. A useful description of AC and DC bias recording is given by Rossing (1981). Fundamentally anhysteresis is a fascinating process. Essentially the AC field provides the energy to switch the magnetic moments over the energy barriers and must he large enough to melt the system magnetically (in the sense that all the moments follow the AC field excursions) at the beginning of the process. In this respect it is very similar to the thermoremanent magnetization (TRM)which produces an equilibrium magnetic configuration. For a system of non-interacting ideal Stoner-Wohlfarth particles, x. is infinite. Early treatments of anhysteresis concentrated o n the introduction of interactions (Wohlfarth 1957, Nee1 1943) in order to explain the finite measured values. A later series of papers by Jaep (1969, 1971a, b) introduced ~finite the dynamics of the process and showed that x , is at a finite temperature although up to two orders ot magnitude larger th;n observed values. A mean field approach was used to introduce interactions in order to reconcile theory and experiment. Clearly anhysteresis is dominated by the interactions between particles, a problem which is only really amenable to modern computational techniques. The first use of these was by Bertram (1971) who essentially treated anhysteresis as a growth process during which the magnetic configuration and hence the interparticle interactions varied. A more recent model has been developed by Lyberatos et al(1985) and Chantrell et al (1986b) which takes account of the dynamics of the process. The interactions between particles are predicted to have an important bearing on the dynamics of the process via fluctuations in the local interaction field. These tend to act as an effective temperature which helps the system to escape from local energy minima and thereby to approach the equilibrium state which is characteristic of anhysteresis. This was demonstrated by increasing the rate of reduction of the AC field which essentially damped out the fluctuations and led to the production of non-equilibrium states. In an actual recording system the tape moves at a constant speed past the head and responds to a field and a consisting of the signal field (analogous t o IfDc) higher frequency AC bias field. As the tape moves out of the recording region both the AC and DC fields reduce to zero at the same rate, in contrast to the ideal situation so far considered. The practical process is referred to as modified anhysteresis, and can result in a phenomenon known as overbiasing. This arises due
16

to the application of too large an AC field. Under this condition by the time that the AC field is low enough to allow the particles to freeze magnetically the signal field is too small to give rise to a significant magnetic response. This situation is demonstrated by calculations (Lyberatos 1986, Chantrell et al 1986b) the results of which are illustrated in figure 14. Since the anhysteretic susceptibility xi. represents the response of the recording medium there is clearly an optimum bias point as can be seen in figure 14. The optimum bias point is characteristic of the tape material used, being dependent on the coercivity of the medium. Because of the rather complex nature of the experiment anhysteresis is not suitable for the day to day characterization of recording media. This tends to rely on the use of static measurements. Although the link between static and dynamic properties is not fully understood they are undoubtedly coupled. Thus any change in the dynamic properties might be expected to be reflected in variations in the static properties represented by H c , squareness, (1 - S*),etc, which on this basis remain perfectly good production monitoring and control parameters. Because of the wealth of information contained in anhysteretic measurements, however, these remain an excellent tool for research and development purposes.

6.2:Medinm nnise
The noise from a recording medium arises from several major factors, and is usually much greater than noise arising from other parts of the recording channel. Large scale variations in the thickness of the recording medium and head to medium spacing give rise to modulation noise which has been studied by Coutellier and Bertram (1987). A second contribution arises from the discrete amount of flux contributed by an individual particle at the read head. This gives rise to statistical fluctuations in the readback voltage, and consequently noise. Particulate noise depends critically on the magnetic state of the system and on the local magnetic structure which is strongly influenced by dipolar interactions. Essentially thrsr rrsuii in iocai nragueiic currelations, often involving closed loop configurations which tend to reduce the noise. Noise measurements are often used for the characterization of recording media, generally by comparison with the results for standard tapes which exist for all the common types of recording media. For example, bias noise, which refers to the noise of a medium having been exposed to a high-frequency AC field as the tape passes a record head is often measured as a function of the bias current. Often the total noise power is measured, rather than the actual spectrum, which nonetheless carries considerable information about the medium. Here we shall discuss briefly the origin of the noise power flux spectrum which with modern instrumentation and powerful methods of on-line statistical analysis is itself potentially useful for the characterization of recording media. According to Thurlings (1982) the problem of

Magnetic characterization of recording media


o

WEAL ANHYSTERETIC PROCESS MOOlFIEO ANHYSTERETIC


0

PROCESS
0 0

I
1.08

0.5HAC I HK

Chantrell et ai (1986)).

noise in particulate media was first investigated by Mann (1957) and Daniel (1960) who produced expressions for random noise. Noise theories assuming completely uncorrelated particle moments are in poor agreement with experiments. Most successful theories take into account the microscopic magnetic structure induced by interparticle interactions. For example, Arratia and Bertram (1984) investigated theoretically the noise of AC erased media (which are essentially subjected to an ARM process with zero static field) using the earlier model of Bertram (1971) to carry out the erasure. More recently Fearon et a1 (1987) have shown that the dynamics of the AC erase process have a significant effect on the local correlations. This is a possible explanation for the observation (Ragle and Smaller, 1965) that bias noise is a few dB higher than AC noise. Although these measurements are made after the medium is subjected to similar processes, a difference in the dynamics could result in such small differences in the noise. The noise originates in statistical fluctuations in the readback voltage and is strongly dependent on the microscopic magnetic configuration. There is direct evidence for this in that AC erased noise differs vastly from DC erased noise in which the system is subjected to a large applied field to cause magnetic saturation. Using a cross correlation technique Thurlings (1985) demonstrated the differences in noise between the different magnetic states including the demagnetized

virgin state of an as-prepared tape and an AC demagnetized tape. The variation of noise with magnetic state is well illustrated by the calculations of Fearon ef a1 (1987). Here different magnetic microstructures were achieved by AC demagnetization with different rates of reduction of the AC field which are shown in figure 15. Highly correlated states with low noise were observed with small AC decrements, which resulted also in complete AC demagnetization. As the rate of reduction is increased by increasing the AC field decrement the sample achieves a non-equilibrium state characterized by finite remanent magnetization at the end of the process. Thus, essentially the system moves continuously from the AC erased state (small AC decrement) to the DC erased state (large AC decrement). The corresponding increase in the noise is very pronounced and is in accord with experimental data. Essentially the rapid noise increase from the calculations follows the equally rapid change of magnetization in the non-equilibrium region. In order to fully understand noise in recording media further physical measurementsarepossible. Oneofthese is the technique of neutron depolarization which is sensitive to magnetic correlations and also to density variations which can be an important source of noise, Magnetic correlations have been studied using neutron depolarization in CrO, tapes by Rosman et a1 (1988) and in alumite by Kraan and Rekveldt (1990). Tape measurements revealed the presence of oriented superdomains.
17

R W Chantrell and K OGrady


A

mediumitself and are thuscharacteristicof its behaviour. Although the relationship between noise, staticmagnetic properties and eventual recording characteristics is as yet not fully understood, the present sophistication of measurement techniques makes this an interesting possibility for the characterization of recording media. Considerable further work is required in order to gain the fundamental insight required in order to fully realize the practical use of noise in this context.
7. Experimental techniques

i-25-30

t
-2
1 I

7.1. The vibrating sample magnetometer

(VSM)

401 -1

-1 drrrnnmt 1109 10 1

AC

Flgure 15. Calculated noise power flux spectra lor systems demagnetized at varying rates (after Fearon et a/ (1987)).

The data were in qualitative agreement with the fact that the bulk erased noise was about 6dB lower than the DC erased noise, Although interpretation of neutron depolarization data is rather difficult it is an interesting probe of correlations and inhomogeneities in magnetic materials which could provide useful information for studies of the fundamental noise mechanisms. A further important form of noise is transition noise which occurs in digital recording. It is related to the imperfect structure of the transition region between the recorded bits (Belketa11985) and is apotentiallimitation ontheachievable bit density. Experimental work (Arnoldussen and Tong 1986) has shown that the noise can be related to the characteristiczig-zag structure arising from non-uniformities across the track width. Micromagnetic calculations (Hughes 1983, Zhu and Bertam 1988, Miles and Middleton 1990) have demonstrated the existence of this type of structure. Much work remains to be done in the understanding of the transition noise. In this context it is interesting to note the work on remanence curves described earlier which found evidence for cooperative reversal, a possible contribution to transition noise, in the modified Henkel plots in both C O P thin films and barium ferrite media. Clearly such a relationship between the macroscopic magnetic properties and the noise is of importance as regards the fundamental understanding of the materials properties. I n addition it emphasizes the potential importance of remanence curves for the characterization of practical media. This brief introduction to noise in recording media is intended to demonstrate that the noise power flux spectrum contains considerable information about the magnetic and physical microstructure of the recording medium. These are fundamental properties of the
18

The VSM, originally due to Foner (1959) is one of the most commonly used ins!rumen?s for the meas"remen! of the magnetic properties of recording media and other magnetic materials. The basic design of the VSM is shown in figure 16. The sample is caused to vibrate at a frequency between 25 and 100Hz by either a motor and cam arrangement or more commonly by a loudspeaker drive system. The moment induced in the sample is detected as an AC signal in the detection coils whose amplitude is proportional to that moment. The amplitude of the AC signal is then determined either via a lock-in amplifier or via digital analysis to give a DC output proportional to the induced moment. Thus, when the magnetometer is calibrated with a sample of known mnm.en! it is capable nf giving mea.si~rem~n!s over a wide range of moments (10F EMU to IO2 EMU); and over a wide range of applied fields (0 to 150 kOe). The field is usually measured via a Hall probe of known calibration, or via calibration of a superconducting solenoid for high-field systems. The reasons for the common usage of the VSM are principally concerned with its versatility and reliability. Temperature variation is simple, its principle of operation and operating procedures are easy to understand, but the instrument does suffer from some disadvantages. Considerable care is necessary when using a VSM which all too often is not taken. Areas where special care is needed are described below. (1) The measurements are madc ii; ai; opci; magnetic circuit so sample shape demagnetizing effects must be taken into consideration. These effects are small when measuring with the applied field parallel to the plane of a tape or thin film but are of great importance when measuring dispersions of particles or perpendicular to the plane of thin film coatings. ( 2 ) High-purity nickel metal at saturation is often used as the calibration standard but this is rarely adequate. The saturation moment of nickel is not known precisely and, as with other ferromagnetic metals, the field required to saturate the sample will depend on its mechanical history. Nickel calibration samples need to be carefully and frequently annealed and cleaned and the field required to saturate them checked. In reality the calibration sample used should be of a similar moment and have similar dimensions to the sample to be measured. This is most readily

Magnetic characterization of recording media


transducer assembly oscillator moveable plates

A C signal subject to variatiins with changes in vibration amplitude and frequency.

reference signal time constan

OC voltage proportional to moment and independent of changer in vibration amplitude and frequency.

differential amplifier

synchronous detector

/
amplifier

to output display circuits

but subject to variations with changes in vibration amblitude and frequency.


sample

A C difference signal which is independent


of changes in vibration amplitude and

frequency.

magnet pole pieces

Figure 16. The basic design and operating principle of the vibrating sample magnetometer.

achieved by using a paramagnetic calibration sample such as palladium, whose moment is well known and which, by varying the field can be used to give a wide range of moments, if the field calibration is accurately knnwn and the sample temperature controlled. (3) Because of the time-dependent effects described in section 5.3 the sweep rate of the magnetic field must be carefully controlled or monitored so that comparisons between different samples can be meaningfully made. Failure to control the sweep rate can give rise to variations in the measured value of the coercivity of up to 15% in extreme cases and the use of a slow sweep rate can result in a variation of the switching field distribution when the Koester (1984) method is used.
7.2. The B-H loop tracer

The basic arrangement of this instrument is shown in figure 17. The outer coil generates an AC field H which induces an AC moment in the sample which is usually in the form of a tape. The inner search coil then detects B (B = H + 4 n M ) and electronic analysis of the two signals enables the M-H curve for the sample to be displayed on an oscilloscope. The B-H looper was until recent years the most commonly used instrument for the measurement of magnetic properties in the recording industry. It suffers disadvantages relative to the VSM; i.e. the control of temperature and its variation are difficult since the

field coil will cause heating. Also the maximum field attainable is only of the order of 5 kOe which is barely sufficient to saturate high-coercivity media. Similarly to the VSM the B-H looper is subject to calibration error particularly with respect to the measurement of B and it is also an open circuit technique requiring care to be taken with regard to demagnetizing effects. The major advantages of this technique are the removal of sweep rate problems since this is accurately known, although the disparity in the mains frequency between Europe and the US can still cause problems for an international industry, and the speed of measurement, literally a few seconds compared with several minutes for a VSM. This latter point in particular will ensure that B-H loopers continue to be used for quality control in media manufacture. It should of course be noted that the B-H looper can only be readily used to measure the basic hysteresis loop whereas the VSM is capable of measuring a far wider range of parameters.

7 . 3 . Noise measurements
In general the noise from a recording system is a combination of noise from the replay head and associated electronics in addition to the required noise spectrum. As such, noise measurements must be carefully made and interpreted. A widely used experimental configuration is to employ a high-quality recorder with a standard tape transport scheme using a high-quality
19

R W Chantrell and K OGrady


R

AC power source

Figure 17. Diagram of a B-H loop plotter.

head and amplifier. In all cases, the background noise of the head and amplifier must be subtracted from the total signal in order to arrive at the noise produced by the medium itself. The head used should have a low value of magnetostrictive noise to which some attention has been given by Thurlings (1982). Noise of this form is very much a complicating factor since magnetostrictive effects are dependent on the magnetic field. By measuring noise for tapes in contact with the head and at a small distance away, Thurlings concluded that for his purposes a sendust head was not significantly affected by magnetostrictive effects whiist in contact with the tape, allowing noise measurements to be made in contact mode. The experimental configuration used by Thurlings (1982) and Luitjens et al (1985) abandons the use of the standard tape transport system. The experiment consists of a turntable supported in air bearings driven by a constant-speed crystal-controlled motor. Samples in the form of glass plates can be held t o the turntable by vacuum and thin films on a flexible substrate can be held in a 'drumhead' configuration (Luitjens et a1 1985). The recording head is positioned in contact with the magnetic layer which rotates with respect to the head at a constant speed. This seems a particularly suitable technique for investigations o f sinall scale experimental samples. The noise power flux spectrum contains detailed information about the recording medium which is ignored in many standard techniques which merely measure the total noise power. However, by combining the use of a spectrum analyser with the power of present microcomputers it is possible to carry out a very detailed analysis on the whole spectrum. One particular advantage is the capability of measuring cross correlation functions between spectra which can he very revealing about the physical and magnetic microstructure of the medium. In addition, modulation noise, which arises from the variations i n amplitude of a recorded signal has been shown by Coutellier and Bertram (1987) to he principally determined by surface effects. Thus noise measurements have the capability of analysing surface, bulk and microscopic effects, in
20

a way which is only now being realized due to rapid advances in instrumentation. There is no doubt that noise measurements are presently an exciting area, with considerable potential for the future characterization of recording media.
7.4. Measurements of anisotropy and texture

Anisotropy measurements are not commonly made on magnetic recording media despite the fact that they can give valuable information concerning the intrinsic properties of the materiais. 'Many recording particies can be considered as having uniaxial anisotropy with an energy density given by (12) where 8 is the angle between the magnetization and easy axis. If the magnetization is rotated away from the easy axis then the sample exerts a torque on the mechanism which suspends it. If K 2 Q K 1 then the torque L (= -dE/d0) is given by L = -K, sin(20). (13) Fig. 18 shows the variation of torque with angle. K , can simply be determined from the torque curve. Torque curves are measured using aii iiistiiimeii: called a torque magnetometer. These are generally home made, and the principle of operation is quite simple. A sample is suspended over a magnet and either the magnet or the sample is rotated through some angle. The resulting torque is measured using either a calibrated fibre whose torsion constant is known or via an electromechanical measurement device, for example a force microbalance operated vertically. For a review of the various designs of torque magnetometer see Pearson (1979). Some samples of recording media are isotropic in the plane, and hence exhibit no external torque. In such cases an alternative technique, termed rotational hysteresis can be used, which utilizes the following principle. When a sample is rotated in field H less than the anisotropy field Hk there is an energy loss associated with the switching of particles between energy minima. E = K O+ K , sin20 + K , sin40

Magnetic characterization of recording media

Figure 18. Typical torque curve

The energy loss is low in small fields where the field energy is not large enough to pull the moment out of the easy axis direction. It is also small in large fields where, conversely, the anisotropy is not large enough to switch the magnetization from the field direction. At intermediate fields the energy loss goes through a maximum. In principle, the energy loss W, vanishes at Hkwhich should give a sharp cut-off at high fields and , . Particulate media tend a very accurate measure of H to have a significant distribution of Hk, and this tends, for these materials, to produce a tailing-off at high fields rather than a sharp cut-off which makes the determination of Hk somewhat uncertain. It is conventional to calculate the rotational hysteresis integral:

Torque magnetometry can be used to measure two important parameters.

(1) The value of Hk. This is obtained by plotting the value of W, against 1/H and extrapolating to W, = 0 which occurs at the anisotropy field Hk. The tail in this plot, characteristic of the spread of Hkin particulate media, often makes the extrapolation difficult. (2) The value of RH. This has for some time been used to characterize the reversal mechanism of a given medium. A low value indicates that the particles reverse by the coherent reversal mechanism of Stoner and Wohlfarth (1948).
Corradi et a1 (1980) used rotational hysteresis for a detailed analysis of anisotropy in cobalt-doped y-Fe203 particles. The data suggested that epitaxially doped materials were more inclined to coherent rotation than other surface treated particles. It was suggested that this was probably due to the smoother particle surfaces produced by this technique. A technique which can in principle measure anisotropy is reversible transverse susceptibility (RTS). The technique measures the response of the medium

to a small AC field applied perpendicular to a bias field. Theoretical predictions for a randomly oriented particulate dispersion (Aharoni et a1 1957) based on Stoner-Wohlfarth theory have been verified by Pareti and Turilli (1987). In essence, the theory predicts a divergent behaviour of the RTS at a bias field H,, equal to the anisotropy field Hkfor those particles oriented at right angles to HDc. In principle this provides a technique for the determination of Hk. However, the divergence as the orientation approaches n / 2 is relatively weak, and is somewhat smeared out by a distribution of the easy axis directions, resulting in a cusp LeDang et a1 (1987) demonstrated the rapid at Hk. angular dependence by calculations which give good agreement with data obtained for a thin film with a high degree of orientation. The technique has the advantage that the apparatus is simple to build and operate and requires only a calibrated DC field sensor (Hall probe) to perform accurately. It can be used to measure HK as a function of temperature and can also detect the presence of polydomain particles. This is increasingly important in the study of barium ferrite. It is likely that the RTS technique will become more commonly used in materials characterization. Recording media generally have some characteristic form of texture, ranging from field induced in-plane alignment to enhance the performance of longitudinal media, to the very pronounced perpendicular texture necessary for the perpendicular magnetic recording process. One of the most obvious observations of texture is simply to measure hysteresis loops parallel and perpendicular to some direction within the medium. More detailed information is available from the magnetic data, for example using a method due to Shtrikman and Treves (1960) which is based on the angular variation of the remanent magnetization. This method is illustrated by the recent measurements of Kelly and OGrady (1988) which show the existence of a preferred orientation in a COP thin film which was nominally isotropic in-plane. Hoon and Paige (1987) have given an interesting method by means of which torque magnetometry can
21

R W Chantrell and K OGrady

be used to determine the distribution of particle orientation and anisotropy field. The technique is based on the energy loss occurring during small angular cycles of the direction of the applied field, which is determined by a small fraction of the particles having specific orientations and anisotropy fields. A recent development (Chantrell ef a l 1989) has been the study of a modified transverse susceptibility technique in which the change in magnetization AM parallel to the bias field in response to a small transverse AC field is measured. This response occurs at twice the frequency of the applied field. Application of Stoner-Wohlfarth theory to this situation shows that AH a H t . Thus it is possible to define a non-linear transverse susceptibility xq = limHac-,(AM/H;c). The response characterized by xq is strongly divergent so that even in a randomly oriented system the angular dispersion is not sufficient to mask the divergence as is the case in the RTS measurement described earlier. In principle this provides a technique for the determination of the dispersion of Hkand, by measuring the orientational dependence, the orientational texture of the system. Finally, a method of texture determination important in perpendicular thin films is the x-ray rocking curve method. This is a diffraction technique in which the sample is tilted through small angles in the diffraction plane and the intensity recorded. The width at half height is often used as a measure o f the angular distribution of easy axes within the sample.

Acknowledgments
Parts of this work were carried out within the framework of the CAMST (Community Action on Magnetic Storage Technology) project of the E C Science programme. This work is dedicated to the memory of the late Professor Peter Wohlfarth, whose inspiration is a continuing influence.

References
Aharoni A 1985 J . Appl. Phys. 57 4702 Aharoni A. Frei E H. Shtrikman S and Treves D 1957 Bull. Res. Councii isr. 6~ ii5 Andra W, Danan H and Ropke U 1984 I E E E Trans. Magn. 20 102 Aonuma M 1975 JaDanese Patent 75 104164 Arnoldussen T C and Tong H C 1986 I E E E Trans. M a p . 22 889 Arratia R A and Bertram H N 1984 IEEE Trans. Magn. 20
~~ ~

412

Bate G 1966 J . Appl. Phys. 37 1164 -1980 Handbook of Ferromagnetic Materials ed E P Wohlfarth (Amsterdam: North-Holland) Bean C P and Livingston J D 1959 J . Appl. Phys. 30 120s Beardsley I A 1982 J. Appl. Phys. 53 2582 Belk N, George P K and Mowry G S 1985 I E E E Trans. Magn. 21 1350 Bertram H N 1971 J. Phvsique Call. 32 C1 2-3 684 -1986 Proc. IEEE 74 1494 Chantrell R W, Fearon M and Wohlfarth E P 1986a Phys.
Status Solidi a 97 213

8. Conclusion
Due to the nature of recording media the techniques used in their characterization are very diverse. In this review we have attempted to describe the major techniques used in the magnetic characterization of recording media from the point of view of the underlying physics. There is no doubt that recording media are in a stage of remarkable development which is a result of the potential of various competing technologies. For this reason a large amount of research is being carried out with the aim of achieving the understanding of the materials which is necessary for further product development. This work is naturally linked with the improved characterization of the materials. The major areas of research in recording media concern the properties of thin metallic films with potential for high-density recording, magneto-optical media and the optimization of the behaviour of particulate recording media. All these materials have a diversity of problems, for example of stability and tribology in addition to magnetics in thin films. In particulate media much remains to be done in the characterization of the colloidal dispersions which form the precursor to the final tape.
22

Chantrell R W, Hoare A, Melville 0 , Lutke-Stetzkamp H J and Methfessel S 1989 I E E E Trans. Magn. 25 4216 Chantrell R W, Lyheratos A and Wohlfarth E P 1986b 1. M a p . Magn. Mater. 54-57 1693 Charap S H 1988 J. Appl. Phys. 63 2054 Corradi A R, DiNitto C A, Bottoni G , Candolfo D, Cecchetti A and Masoli F 1987 IEEE Trans. Magn. 23 48 Corradi A R, Visigalli P G, Bottoni G, Candolfo D, Cecchetti A and Masoli F 1980 Proc. Inf. Conf. on Ferrires p 526 Corradi A R and Wohlfarth E P 1978 I E E E Trans. Maan. 14 861 Coutellier J M and Bertram H N 1987 I E E E Trans. Magn.
I

23 195

Coverdale G N, Chantrell R W and O'Gradv K 1990 J . Magn. Magn. Mater. 83 442 Daniel E D 1960 Amoex Research Report AEL-1 Della Torre E 1985 iEEE Trans. Magn. 21 1423 -1986 I E E E Trans. Magn. 22 484 Eiling A 1987 IEEE Trans. M a p . 23 16 Fearon M , Chantrell R W, Lyberatos A and Wohlfarth E P 1987 IEEE Trans. M a p . 23 174 Fearon M, Chantrell R Wand Wohlfarth E P 1990 J . Magn. Magn. Mater. 86 197 Flanders P J and Sharrock M P 1987 J . Appl. Phys. 62 2918 Foner S 1959 Rev. Sci. Instrum. 30 584 Gau J S, Spahn A G and Majumdar D 1986 I E E E Trans. Mann. 22 582 GauntP 1968 Phil. Mag. 17 263 Gaunt P, Hadjipanayis G and Ng C 1986 J . Magn. Magn. Mater. 54-57 841 Goto F, Osaka T, Koiwa I, Okabe Y, Matsubara H, Wada A and Shiota N 1984 I E E E Trans. Magn. 20 803 Grundy P J and Ali M 1983 1. Magn. Magn. Mater. 40 154

Magnetic characterization of recording media Grundy P J, Ali M and Faunce C A 1984 IEEE Trans.
Magn. 20 794 Henkel 0 1964 Phys. Starus Solidi 7 919

Oppegard A L, Damell F J and Miller H C 19611. Appl. Phvs. 33 S 184s ~. . OrtenLurger I B and Potter R I 1979 J. Appl. Phys. 50
~~ ~

el-Hi10 M, Uren S H, OGrady K, Popplewell J and Chantrell R W 1990 I E E E Trans. Magn. 26 244 Hoffmann H 1986 IEEE Trans. Magn. 22 472 Hoon S R and Paige D M 1987 IEEE Trans. Magn. 23 183 Hughes G F 1983 J . Appl. Phvs. 54 5306 Iwasaki S and Nakamura Y 1977 IEEE Trans. Magn. 13
<*?* I','

?7Q7

Oseroff S B, Clark D, Schultz S and Shtrikman S 1985


IEEE Trans. Maan. 21 1495

Iwasaki S, Nakamura Y and Ouchi K 1979 IEEE Trans,


M a p . 15 1456

-1971b A I P Conf

Jaep W F 1969 1. Appl. Phys. 40 97 -1971a J . Appl. Phys. 42 2790


Proc. 5 786

Pareti L and Turilli G 1987 J. Appl. Phys. 61 5098 Pearson R F 1979 Magnetic anisotropy in Experimental Magnetism vol VI ed G M Kalvius and R S Tebhle (New York: Wilev) D 137 Pinkerton F E 1986 i E E E Trans. Magn. 22 922 Potter R I 1970 J . ADP/.Phvs. 41 1647 Potter R I and Bearddey I A 1980 IEEE Trans. Magn. 16
1967

Jorgensen F 1980 The Complete Handbook of Magnetic Recording (Tab Books) Kelly P E and OGrady K 1988 1. Physique Coll. 49 1833 Kelly P E, OGrady K, Mayo P I and Chantrell R W 1989
IEEE Trans. Magn. 25 3881 Knowles J E 1986 1. Magn. Magn. Mater. 61 121 -1985 IEEE Trans. Magn. 21 2576 Koester E 1984 IEEE Trans. Maan. 20 81 Koester E, Jakusch H and Kullm~nn U 1981 IEEE Trans. M a w 17 2550 KraanW H and Rekveldt M Th 1990 IEEE Trans. Magn. 26 219 LeDang K, Veillet P, Suran G and Ounadjela K 1987 J. Appl. Phvs. 62 3328 Kubo 0, Id0 T and Yokoyama H 1982 IEEE Trans. Magn. 18 1122 Lemke J U 1982 1. Appl. Phys. 53 2561 Luitjens S B, Schrauwen C P G, Bernards J P C and Zieren V 1985 IEEE Trans. Magn. 21 1438 Lyberatos A 1986 Thesis University of London Lyberatos A and Wohlfarth E P 1986 J. Magn. Magn. Mater. 59 L l i l Lyberatos A, Wohlfarth E P and Chantrell R W 1985 IEEE Trans. Magn. 21 1277 Mann P A 1957 Arch Elektris Uberiragung 11 97 Martin A and Carmora F 1968 IEEE Trans. Magn. 4 259

Ragle-H U and Smaller P 1965 IEEE Trans. Magn. 1 105 Rosman R, Rekveldt M Th and Cramer H A J 1988 J
Physique Coll. 49 1983

Rossing D T 1981 Am. J . Phys. 49 655 Schmidt F, Rave W and Hubert A 1985 IEEE Trans.
Magn. 21 1596

-1984 IEEE
~

Schahes M 19911. Magn. Magn. Mater. 95 249 Sharrock M P 1990 IEEE Trans. Magn. 26 193
Trans. Magn. U) 754

Sharrock M P and McKinney R A 1981 I E E E Trans.


Maen. 1730211 . . .

Shtrikman S and Treves D 1963 Magnetkm ed G T Rado and H Suhl (New York: Academic) Shtrikman S and Treves D 1960 1. Appl. Phys. 31 585 Spratt G W D, Bissell P R, Chantrell R W a n d Wohlfarth E P 1988 J. Magn. Magn. Mater. 75 309 Stoner E C and Wohlfarth E P 1948 Phil. Trans. R. Soc. A
240 599

Street R and Woolley J C 1949 Proc. Phys. Soc. A 62 562 Sugaya H and Tomago A 1983 Proc. Symp. Magnetic Media and Manufacruring Methods paper C-2-1 Swoboda T J, Arthur P, Cox N L, Ingraham J N, Oppegard A L and Saddler M S 1961J. Appl. Phys. 32
274'2 _I,Ty

Tada J, Akahiro M, Takei K, Satoh T and Suzuki T 1986


IEEE Trans. Magn. 22 343

Takano 0 and Matsuda H 1986 Metal Finishing June 1986


DD

a-4

Mayo P I, Erkkila R M, Bradbury A and Chantrell R W


1990 IEEE Trans. Magn. 26 1894

Thiriings L F G 1982 Thesk Technical University of Eindhoven


-1985 IEEE Trans. Magn. 21 36

Mee C D and Daniel E D (eds) 1987 Magnetic Recording (New York: McGraw-Hill) Middleton B K and Davies A V 1984 IERE Conf. Proc. 59
27

Tsana C, Chen M M, Yogi T and Ju K 1990 IEEE Tram


I

Miles J J and Middleton B K 1990 IEEE Trans. Magn. 26


204

Morrish A H and Yu S P 1955 1. Appl. Phys. 26 1049 Nee1 L 1943 Cahiers Phys. 17 47 Okhoshi M, Toba H, Honda S and Kusuda T 1983 J.
Mann. Maen. Marer. 35 266

Magn. 26 1689 Umeki S, Sugihara H, Taketomi Y and Imaoka Y 1981 IEEE Trans. Magn. 17 3014 Uren S, Walker M, OGrady G and Chantrell R W 1988 I E E E Trans. Magn. 24 1808 Victora R A 1987 Phys. Reu. Lett. 58 1788 Weilinga T and Ladder J C 1986 Phys. Status Solidi 96 255 Williams L and Comstock R L 1971 A I P Conf. Proc. 5 738 Witherell F E 1984 IEEE Trans. M a p . 20 739

O G r a i y K a l d Chantrell R W 1986 1. Magn. Magn.


Mater. %57

755

OGrady K, Chantrell R W, Popplewell I and Charles S W


1981 IEEE Trans. Magn. 17 2943

de Witte A M, O'Grady K, Coverdale G N and Chantrell R W 1990 1. M a p . M a p . Mater. 88 183 Wohlfarth E P 1957 Phil. Mag. 2 719 -1958 J . Appl. Phys. 29 595 Zhu J G,and Bertram H N 1988 1. Appl. Phys. 63 3248

23

You might also like