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

Подождите немного. Документ загружается.

from a lower MO output. The increase of Curie tem-

perature with increasing magnetic layer thickness is

also in accordance with the random-phase approxi-

mation Green function calculations of Bruno (1991) in

which he estimates the Curie temperature for ultrathin

h.c.p. (0001) cobalt ﬁlms as a function of the number

of atomic planes. T

of the multilayer decreases with

increasing platinum thickness. For a palladium inter-

layer this effect can be entirely attributed to the strong

ferromagnetic coupling through the interlayer, which

has a strong exchange-enhanced susceptibility. This

coupling is an order of magnitude smaller for plati-

num interlayers and therefore the strong decrease in

Curie temperature with platinum layer thickness can-

not be attributed to this coupling (Bruno 1991).

Therefore, this platinum thickness dependence might

be due to interdiffusion between cobalt and platinum,

which are known to form solid solutions. When gold is

chosen as the interlayer (gold is insoluble in cobalt,

and therefore sharp interfaces are expected) there is

almost no dependence of Curie temperature on the

gold thickness (Bloemen et al. 1991). This result sup-

ports the interdiffusion mechanism. Co/Pt layers with

good MO properties and good perpendicular aniso-

tropy generally are composed of about 3.5–4.5 A

cobalt and 10–15 A

platinum. The T

of such a layer is

350–400 1C. The coercivity will vanish at a lower tem-

perature, say 50 1C lower, and therefore the layer has

to be heated to 300–350 1C in the writing process. This

temperature is higher than desirable. Annealing at

200 1C can lower coercivity (Lin and Do 1990), and

therefore the Curie temperature should be lowered by

about 100 1C. This could be done by changing the

composition of the magnetic layers. This change

should not drastically lower the magnetic moment of

the magnetic layers because the Kerr rotation scales

with the magnetic moment. One way to keep the

magnetic moment high is to alloy the cobalt layers

with another magnetic element. It has been found that

for CoNi/Pt multilayers decreases with decreasing

CoNi thickness but increases with decreasing platinum

thickness (Meng et al. 1996a).

3.5 Thermomagnetic Written Bits

Multilayer disks consisting of Co/Pt have not only

been prepared in research laboratories but also com-

mercially (Lin and Do 1990, Sumi et al. 1991, Zeper

et al. 1990, and Hashimoto et al. 1990). In MO re-

cording the ﬁnal bit-size is determined by the prop-

erties of the media as well as the writing laser beam.

The domain (bit written by laser modulation) be-

comes smaller as the laser power decreases. This has

been measured for Co/Pt multilayer disks (e.g.,

Greidanus et al. 1989) as well as for CoNi/Pt (Meng

et al. 1996a).

Figure 16 shows an example of written bits in a

multilayer with a composition of 1.2 nm platinum

seed layer (0.55 nm CoNi/1.2 nm Pt) 20. The writ-

ten domains change in size as a function of the laser

power. In order to achieve regularly written domains,

a large perpendicular coercivity with a square hys-

teresis loop is required. The optimum thickness in

Co(Ni)/Pt multilayers prepared in the author’s lab-

oratory has been determined to be about t

(or

CoNi

) ¼ 5A

, t

¼10 A

, and N ¼10. An argon sput-

tering pressure of 1.6 10

2

mbar is preferred. The

typical magnetic parameters in an optimum CoNi/Pt

multilayer are H

¼113 kAm

1

, Y

K,310

¼0.31, and

¼255 1C.

(a) Thermal stability

RE/TM multilayers are sensitive to oxidation while

Co/Pd and Co(Ni)/Pt multilayers are very corrosion

resistant. It has been shown (Hashimoto and Ochiai

1990, Zeper et al. 1991) that a Co/Pt multilayer does

not show large changes in H

and Kerr rotation after

being stored for 200 hours at 80 1C and 75% relative

humidity. Another important point is the stability of

the multilayer structure during thermomagnetic writ-

ing. During writing the whole multilayer stack is

heated many times locally to much higher tempera-

tures than 80 1C. Various studies (den Broeder et al.

1997, Wright et al. 1992, Shiomi et al. 1993a, Meng

1996) have reported on the structural and chemical

changes (diffusion) and changing of the magnetic

properties as a function of annealing temperature up

to 500 1C. For all multilayers a general result is that

below 300 1C only very small changes are observed.

Above this temperature depending on the thickness

Figure 16

Lorentz microscopy image of written domains in a

0.5

/Pt multilayer with a composition of 12 A

Pt/(5.5 A

CoNi/12 A

Pt) 20 (sample I) deposited at an

argon pressure of 4.0 10

2

mbar. The domains shown

are written with laser powers of 2.5 mW, 5.0 mW,

7.5 mW, and 10 mW. The writing is performed with a

pulse duration of 0.24 ms and a bias ﬁeld of 50 kAm

1

790

Magneto-optic Multilayers

and preparation technologies changes are reported

and at temperature higher than 400 1C a large inter-

diffusion takes place and perpendicular anisotropy is

very much affected. In the case of CoNi/Pt multilay-

ers two remarkable results have been found during

annealing. One is enhancement of coercivity in the

annealed multilayers with thinner CoNi layers. An-

other is enhancement of Kerr rotation in the annealed

multilayers having thicker CoNi layers. Based on the

reports it is advisable to choose multilayers with

switching temperature below 300 1C. In the case of

Co/Pt multilayers this means that the cobalt layer

must be made thinner, which means that the Kerr

rotation becomes smaller. CoNi/Pt multilayers are at

least from these viewpoints a better solution.

4. Conclusions

The multilayer structures of Co/Pt are tested and

used in very much detail for applications in short-

wavelength MO recording. Research on CoNi/Pt

multilayers started much later. Based on the pub-

lished results a remarkable difference between Co/Pt

and CoNi/Pt multilayers is the lower Curie temper-

ature in CoNi/Pt. The lower T

results in lower values

for the magnetic properties of CoNi/Pt multilayers

but thermomagnetic writing can be more easily per-

formed than in Co/Pt multilayers. The optimal thick-

ness for Co/Pt magnetic layers, regarding magnetic

and MO properties, is about 3 A

for cobalt layers,

whereas it is 5–6 A

for Co

layers. In the Co/Pt

case, the thinnest platinum layer that still gives a

squareness of unity is about 5 A

for 3 A

cobalt layers,

and about 7 A

for 5 A

cobalt layers (Hashimoto et al.

1990). In the CoNi/Pt case, thinner platinum layers

could be applied (down to 2.9 A

platinum for 5 A

magnetic layers) where the perpendicular

anisotropy is still high, the magnetic hysteresis loop

has unity squareness and is almost rectangular. The

highest Kerr rotation (Y

) is found for a 100 A

Pt þ17 (5 A

/5.9 A

Pt) multilayer sput-

tered at 40 mbar argon pressure on a Si(111) substrate.

The Kerr rotation is 0.481 at 4.6 eV (270 nm), but

calculations indicate (Van Drent 1995) that the peak

is shifted to higher photon energies because this mul-

tilayer is optically thin. According to the calculations,

the peak would be 0.461 at 3.8 eV for an optical thick

layer (100 instead of 17 repetitions). The calculations

also indicate that increasing the seed layer thick-

ness of this sample from 100 A

to 400 A

will shift the

peak position back to 4.0 eV, while the Kerr rota-

tion decreases by about 0.011. The ﬁgure of merit

(FOM ¼R

1/2

7, where R is the reﬂectivity) is a

measure of the MO performance of a given medium.

The FOM for CoNi/Pt is calculated for a layer with

the highest Kerr rotation.

In Fig. 17 the FOM for the thin CoNi/Pt sample is

denoted by open circles. The corresponding reﬂectivity

is drawn as a dashed line. As this sample is optically

thin, comparison to other media is not straight-

forward. To compare several media, the media should

be optically thick, without enhancement layers. There-

fore the Kerr rotation, reﬂectivity, and FOM for a

hypothetical 100-bilayer medium based on the thin

CoNi/Pt sample are calculated. The MO Voigt con-

stant of the polarized platinum is determined from the

measurement data, and a prediction for the Kerr ro-

tation and reﬂectivity of a 100-bilayer medium is made

using this constant. The FOM for the 100-bilayer me-

dium is denoted by open squares, the corresponding

reﬂectivity by the dotted line (Fig. 17). For both

N ¼17 and N ¼100 the calculated reﬂectivity is lower

by 5–10% than the values measured by Zeper (1991)

(not shown in Fig. 17).

These values are compared to the FOM for N 

(4 A

Co/9 A

Pt) with N440 calculated from meas-

ured data (Zeper 1991) (ﬁlled squares). The FOMs of

the Co/Pt and the 100-bilayer CoNi/Pt multilayer

overlap in the region 2.4–3.3 eV (375–520 nm). At

3.5 eV the FOM of the Co/Pt multilayer is slightly

higher than that for the CoNi/Pt simulation. From

these results it is clear that the MO performance of

CoNi/Pt multilayers can be made comparable to

that of Co/Pt multilayers in the wavelength range

of interest. For comparison the FOM of a 2000 A

layer (Hashimoto et al. 1990) is drawn

(ﬁlled triangles). The advantage of the multilayers at

short wavelengths is evident.

Figure 17

FOM, R

1/2

7, for several MO layers. Open circles

denote the FOM for a 100 A

Pt þ17 (5 A

5.9 A

Pt) multilayer on a silicon substrate. The

corresponding reﬂectivity is denoted by the dashed line.

Open squares denote the FOM for 100 instead of 17

repetitions. The dotted line denotes the corresponding

reﬂectivity. The FOM for a N (4 A

Co/9 A

Pt) with

N440 is denoted by ﬁlled squares (Zeper 1991). The

ﬁlled triangles denote the FOM for a conventional MO

medium: a 2000 A

layer (Hashimoto et al.

1990).

791

Magneto-optic Multilayers

See also: Magnetic Layers: Anisotropy; Magnetic

Recording Technologies: Overview; Magneto-optic

Recording Materials: Chemical Stability and Life

Time; Magneto-optic Recording: Total Film Stack,

Layer Conﬁguration; Magneto-optics: Inter- and

Intraband Transitions, Microscopic Models of

Bibliography

Bijker M D, Donnet D M, Lodder J C 1996 The effects of

sputter pressure on the microstructure of CoNi/Pt multilay-

ers. J. Magn. Soc. Jpn. 20 (Suppl. S1), 407–10

Bloemen P J H, de Jonge W J M, den Broeder F J A 1991 The

temperature dependence of the magnetization of magnetic

multilayers. J. Magn. Magn. Mater. 93, 105

den Broeder F J A, Donkersloot H C, Draaisma H J G,

de Jonge W J M 1987 Magnetic properties and structure of

Pd/Co and Pd/Fe multilayers. J. Appl. Phys. 61, 4317

den Broeder F J A, Kuiper D, Draaisma H J G 1997 Effects of

annealing and ion implantation on the magnetic properties of

Pd/Co multilayers containing ultrathin Co. IEEE Trans.

Magn. 23, 3696

Bruno P 1991 Theory of the Curie temperature of cobalt-based

ferromagnetic ultrathin ﬁlms and multilayers. J. Magn. Soc.

Jpn. 15 (Suppl. S1), 15

Carcia P F 1988 Perpendicular magnetic anisotropy in Pd/Co

and Pt/Co layered structures. J. Appl. Phys. 10, 55066

Carcia P F, Li Z G, Reilly M, Zeper W B 1993 The magnetic

and microstructural properties of Pt/Co multilayers grown

on ZnO. J. Appl. Phys. 73, 6424

Carcia P F, Meinhaldt A D, Suna A 1985a Perpendicular an-

isotropy in Co/Pd thin ﬁlm layered structures. Appl. Phys.

Lett. 47, 178

Carcia P F, Reilly M, Zeper W B, van Kesteren H 1991a Appl.

Phys. Lett. 58, 191

Carcia P F, Shah S I, Zeper W P 1990 Effect of energetic

bombardment on the magnetic coercivity of sputtered Pt/Co

thin ﬁlm multilayers. Appl. Phys. Lett. 56, 2345

Carcia P F, Suna A, Onn D G, van Antwerp R 1985b Struc-

tural, magnetic and electrical properties of thin ﬁlm Pd/Co

layered structures. Superlattices Microstruct. 1, 101

Carcia P F, Zeper W B, van Kesteren H W, Jacobs B A J,

Spruit J H M 1991b Materials challenges for metal multi-

layers as a magneto-optic recording medium. J. Magn. Soc.

Jpn. 15 (Suppl. S1), 151

Chang C H, Kryder M H 1994 Effect of substrate roughness on

microstructure, uniaxial anisotropy and coercivity of Co/Pt

multilayer thin ﬁlms. J. Appl. Phys. 75, 6864

Chappert C, Bruno P 1988 Magnetic anisotropy in metallic

ultrathin ﬁlms and related experiments on Co ﬁlms. J. Appl.

Phys. 64, 5736

Donnet D M, Tsutsumi K, de Haan P, Lodder J C 1996 Effect

of sputter gas on the physical and magnetic microstructure of

Co/Cu multilayers. J. Appl. Phys. 79 (8), 6243–5

Draaisma H J G, de Jonge W J M, den Broeder F J A 1987

Magnetic interface anisotropy in Pd/Co and Pd/Fe multilay-

ers. J. Magn. Magn. Mater. 66, 351

Freeman A J, Wu R 1991 Electronic structure theory of surface,

interface and thin-ﬁlm magnetism. J. Magn. Magn. Mater.

100, 497

Gamble R, Lissberger P H, Parker M R 1985 A simple analysis

for the optimization of the normal polar magneto-optical

Kerr effect in multilayer coatings containing a magnetic ﬁlm.

IEEE Trans. Magn. 5, 1651–63

Greidanus F J A M, Zeper W B, Jacobs B A J, Spruit J H M,

Carcia P F 1989 Magneto-optical recording in Co/Pt mul-

tilayers. Jpn. J. Appl. Phys. 28 (Suppl. 28–3), 37

Hashimoto S 1994 Adding elements to the Co layer in Co/Pt

multilayers. J. Appl. Phys. 75, 438

Hashimoto S, Maesaka A, Fujimoto K, Bessho K 1993

J. Magn. Magn. Mater. 121, 471

Hashimoto S, Ochiai Y 1990 Co/Pt and Co/Pd multilayers as

magneto-optical recording materials. J. Magn. Magn. Mater.

88, 211

Hashimoto S, Ochiai Y, Aso K 1990 Ultrathin Co/Pt and

Co/Pd multilayered ﬁlms as magneto-optical recording

materials. J. Appl. Phys. 67, 2136

Honda S, Tanimoto H, Ago J, Nawate M, Kusuda T 1991

Origin of coercivity in Co/Pt sputtered multilayers. J. Appl.

Phys. 70, 6047

van Kesteren H W, Zeper W B 1993 Controlling the Curie

temperature of Co/Pt magneto-optic recording media.

J. Magn. Magn. Mater. 120, 271

Kirilyuk A, Rasing Th, Haast M A M, Lodder J C 1998 Prob-

ing structure and magnetism of CoNi/Pt interfaces by non-

linear magneto-optics. Appl. Phys. Lett. 72, 2331–3

Kronmu

ller H, Durst K D, Sagawa M 1986 Analysis of the

magnetic hardening mechanism in RE-FeB permanent mag-

nets. J. Magn. Magn. Mater. 54/57, 1055

Lin C J, Do H V 1990 Magneto-optical recording on evapo-

rated Co/Pt multilayer ﬁlms. IEEE Trans. Magn. 26, 1700

Lin C J, Gorman G L, Lee C H, Farrow R F C, Marinero E E,

Do H V, Notarys H, Chien C J 1991 Magnetic structural

properties of Co/Pt multilayers. J. Magn. Magn. Mater. 93, 194

Meng Q 1996 Magneto-optical recording media CoNi/Pt and

Co/Pt multilayers. Ph.D. thesis, University of Twente

Meng Q, van Drent W P, Lodder J C, Popma Th J A 1996a

Curie temperature dependence of magnetic properties of

CoNi/Pt multilayer ﬁlms. J. Magn. Magn. Mater. 156, 296–8

Meng Q, de Haan P, van Drent W P, Lodder J C, Popma Th J

A 1996b Effects of Pt seedlayer and Air pressure on magnetic

and structural properties of sputtered CoNi/Pt multilayers.

IEEE Trans. Magn. 32 (5), 4064–6

Meng Q, Lodder J C, Popma Th J A 1996c Magneto-optic Kerr

spectroscopy of CoNi/Pt and Co/Pt multilayers. J. Magn.

Soc. Jpn. 20 (Suppl. S1), 287–90

Mes M, Lodder J C, Takahata T, Moritani I, Imamura N 1993

(1–x)

/Pt multilayers for magneto-optical recording.

J. Magn. Soc. Jpn. 17 (Suppl. S1), 44

Ochiai Y, Hashimoto S, Aso K 1989 IEEE Trans. Magn. 25,

3755

Shiomi S, Nakabayashi T, Okada M, Kobayashi T, Masuda M

1993a Magnetic properties and structure of Co/Pt multilayer

ﬁlms evaporated on heated substrates. Jpn. J. Appl. Phys. 32,

791

Shiomi S, Nishimura T, Kobayashi T, Masuda M 1993b De-

pendence of magnetic properties of Co/Pt multilayers on

deposition temperature of Pt buffer layers. Jpn. J. Appl. Phys.

32, L495

Somekh R E 1984 The thermalization of energetic atoms during

the sputtering process. J. Vac. Sci. Technol. B 2, 1285–91

Sumi S, Tanase K, Torazawa K, Tsunashima S, Uchiyama S

1991 Multilayer Pt/Co magneto-optic disk. IEEE Trans.

J. Magn. Jap. 6, 141

Suzuki T, Notarys H, Dobbertin D C, Lin C J, Weller D, Miller

D C, Gorman G 1992 Coercivity mechanism and micro-

structure of (Co/Pt) multilayers. IEEE Trans. Magn. 28, 2754

Van Drent W 1995 CoNi/Pt multilayers for magneto-optical

recording. Ph.D. thesis, University of Twente

792

Magneto-optic Multilayers

Weller D, Brandle H, Gorman G, Lin C J, Notarys H 1992a

Appl. Phys. Lett. 61, 2726

Weller D, Farrow R F C, Marks R F, Harp G R, Notarys H,

Gorman G 1993 Mast. Res. Soc. Symp. Proc. 313, 791

Weller D, Notarys H, Suzuki T, Gorman G, Logan T, McFad-

yen I, Chien C J 1992b Thickness dependent coercivity in

sputtered Co/Pt multilayers. IEEE Trans. Magn. 28, 2500

Wright C D, Glegg W W, Boudjemultilayerine A, Heyes N A E,

Grundy P J, Pollard R J, Greaves S J, Petford-Long A K,

Kim Y H, Jakubovics J P 1992 Micro-annealing studies of

Pt/Co multilayers for magneto-optic recording applications.

IEEE Trans. Magn. 28, 2671

Zeper W B 1991 Magneto-optical recording media based on

Co/Pt multilayers. Ph.D thesis, University of Twente

Zeper W B, Greidanus F J A M, Carcia P F 1989a Evaporated

Co/Pt layered structures for magneto-optical recording.

IEEE Trans. Magn. 25, 3764

Zeper W B, Greidanus F J A M, Carcia P F, Fincher C R 1989b

J. Appl. Phys. 65, 4971

Zeper W B, Greidanus F J A M, van Kesteren H W, Jacobs B A

J, Spruit J H M, Carcia P F 1990 Co/Pt and Co/Pd multi-

layers as a new class of Magneto-optical recording materials.

SPIE 1274, 282

Zeper W B, van Kesteren H W, Jacobs B A J, Spruit J H M,

Carcia P F 1991 Hysteresis, microstructure and magneto-

optical recording in Co/Pt and Co/Pd multilayers. J. Appl.

Phys. 70, 2264

J. C. Lodder

University of Twente, Enschede, The Netherlands

Magneto-optic Recording Materials:

Chemical Stability and Life Time

Thin amorphous rare earth transition metal (RETM)

alloys such as (Gd,Tb)–(Fe,Co) and (Gd,Dy)–

(Fe,Co) are commonly used as a storage layer in

the ﬁeld of magneto-optical (MO) data storage.

These alloys exhibit very good recording properties

such as high carrier-to-noise ratio (CNR) and low

required write/erase laser powers and ﬁelds (Raasch

1993, Raasch et al. 1994a, 1994b). Amorphous layers

can be prepared by rapid quenching from a liquid

phase, evaporation, or sputtering and show a 5–15%

lesser density than their crystalline counterparts, de-

pending on composition and preparation conditions

(Klahn et al. 1990). Amorphous layers have no grain

boundaries and composition can be varied continu-

ously. Their advantages compared to polycrystalline

materials led to the technical breakthrough of RETM

layers in the ﬁeld of MO data storage. The storage

capacity of MO disks has been dramatically increased

compared to the ﬁrst products in 1988. This trend will

continue over the next years opening a wide range of

applications. The long-term stability is one essential

requirement for storage layers.

The stability of RETM ﬁlms is endangered by

crystallization of the amorphous phase, by structural

relaxation processes within the amorphous phase,

and by selective reactions of the rare earth (RE)

atoms, where oxygen is the most critical element.

Degradation processes can be monitored by the

change of important magnetic properties such as

magnetic anisotropy K

, required magnetic ﬁeld to

reverse the magnetization H

, Curie temperature T

compensation temperature T

comp

, and saturation

magnetization M

as a function of aging time. The

general dependencies of those parameters on compo-

sition and preparation conditions are listed in Lan-

dolt–Bo

rnstein (Hansen 1988).

1. Crystallization and Structural Relaxation of

Thin Amorphous RETM Layers

The amorphous phase is a metastable state, of which

the free energy E

is larger than that of the crys-

talline state E

. This energy difference is the reason

for a phase transition towards the crystalline phase,

but a reduction of E

may not result in a better

stability. Furthermore, the mobility of atoms, repre-

sented by the activation energy dE, dominates the

stability of the amorphous phase. For instance, a

0.62

0.38

alloy exhibited dE ¼3.3 eV while

E

¼0.4 eV (Buschow 1985). The required en-

ergy for atomic rearrangement follows the sequence

transition metal (TM) ¼Fe, Mn, Co, and Ni as well

as going from RE ¼La to RE ¼Lu. For RE

1x

alloys (RE ¼Gd, Tb; TM ¼Fe, Co) with x ¼0.3–0.8,

ranges between 500 K and 900 K with a maximum

of x ¼0.35 (T

¼600 K) for Gd–Co alloys (Hansen

1988) and of x ¼0.70 (T

¼900 K) for magnetron-

sputtered GdTb–Fe alloys (Klahn et al. 1987). The

higher T

of Fe alloys contradicts the former expec-

tations and shows the importance of the preparation

process.

It must be taken into consideration that layers

sputtered at low argon pressure exhibit an ideal

amorphous structure with a small number of crystal-

lization grains and a density close to crystalline layers

resulting in a reduced mobility of atoms (Klahn et al.

1990). In addition, the heating rate inﬂuences T

which will decrease with decreasing heating rate. The

correlation between both parameters is described by

Kissinger (1957). A contribution of heating rate im-

plies the possibility of structural changes within the

amorphous phase below T

. These processes are

known as structural relaxation processes effecting the

near range ordering due to a limited atomic mobility.

It is very difﬁcult to comment on the importance of

structural relaxation on stability of RETM alloys

because this effect has to be separated from oxida-

tion. Nevertheless, Klahn et al. (1987) found that

structural relaxation mainly affect the coercive energy

(required energy to reverse the magnetization), while

793

Magneto-optic Recording Materials: Chemical Stability and Life Time

other magnetic properties remained nearly constant.

This process performs on a logarithmic time scale.

The required time to half the coercive energy on an

aging temperature of 150 1C was estimated to be

about 17 years though for lower temperatures it

would be much longer.

2. Corrosion Ki netics of Thin Amorphous RETM

Layers

A large number of elements are able to react with a

bare RETM layer. As oxygen is the most critical el-

ement, this article will be mainly restricted to oxygen.

Klahn et al. (1987, 1990) prepared unprotected

GdTb–Fe layers of 100 nm thickness on glass subst-

rates by magnetron sputtering and oxidized the sam-

ples in a hot dry climate chamber. Oxygen depth

proﬁles were determined after 15 hours by secondary-

ion mass spectroscopy (SIMS) as a function of aging

temperature. Directly after preparation the layers ex-

hibited a surface oxide layer of about 10 nm thickness

consisting of Tb

and a-Fe

. Beneath the sur-

face, a layer was detected with an oxygen saturation

level of 25%, enough to oxidize the RE atoms com-

pletely and not effecting the TM atoms. The thick-

ness of this partially oxidized layer increases with

time and temperature. The remaining layer is mag-

netically still unaffected. The binding energies for

RE–O (Gd–O, 716 kJmol

1

; Tb–O, 707 kJmol

1

; and

Dy–O, 611 kJmol

1

) are much larger than for TM–O

(Fe–O, 410 kJmol

1

and Co–O, 368 kJmol

1

) (Weast

1976). This leads to a strong preference of oxidation

of the RE atoms. The energy differences enable a

reduction of oxidized TM atoms by RE atoms. Saito

et al. (1986) established that the oxide of GdTb–Fe

consists of RE oxide and b.c.c. Fe clusters.

If the oxygen exposure at higher temperatures con-

tinues after complete oxidation of the RE atoms the

totally oxidized surface layer will grow into the par-

ticular oxidized bulk. The oxidation process can be

characterized by an Arrhenian temperature depend-

ency with an activation energy E

¼0.370.1 eV, which

shows only a weak dependency on composition and

preparation (Klahn et al. 1990). The oxidation speed

decreases considerably with increasing Co content. The

most affected quantity is the magnetic anisotropy.

Mergel et al. (1993) listed data of different additives

and its inﬂuence on K

, where a distinction between

additives with a metallic bond and those with covalent

bonds has to be made. An addition of only 2 at.%

oxygen or 4–8 at.% nitrogen will destroy the aniso-

tropy totally. Instead of this, up to 20 at.% of Bi, Pb,

Al, Be, or Cr are still acceptable from this point of view

(see Sect. 3). Nevertheless, a dilution of the magnetic

system will effect T

, T

comp

, M

, and Kerr rotation.

The inﬂuence of several common additives regarding

magnetic and recording properties is discussed else-

where in more detail (Raasch et al. 1995, Raasch 1996).

3. Aging Behavior of Magneto-optical

Storage Disks

The weak chemical stability of RETM layers requires

additional measures to enhance the environmental

stability of the MO layer stack, especially when the

MO layers are grown on plastic substrates. Two meth-

ods can be applied separately or simultaneously: (i) the

use of protective layers, such as SiN or AlN or (ii)

doping active elements directly to magnetic layers in

order to protect them from further corrosion by passi-

vation of chemical reactions with air or water. If pro-

tective layers are grown perfectly, a long-term stability

of more than 50 years is obtained. The possibility of

local defects within these layers will cause corrosion of

the MO layer and can be prevented by using a limited

amount of additives though this must be restricted to

elements with a metal bond. The magnetism of RETM

alloys is caused by the localized 4f electrons of the RE

and the delocalized 3d electrons of TM. Therefore, a

differentiation must be made between elements with

partly and fully occupied d shells. The addition of

metals with partly occupied d shell (Ti, Cr, Nb, Zr, Hf,

Ta, and Pt) shows positive effects on the chemical

stability. But the large negative effect on the magnetic

properties, such as T

and Kerr rotation, can not be

compensated. Elements with fully occupied d shells

(Be, Al, and In) show only a minor inﬂuence on mag-

netic properties. In particular, TbFeCo–In layers with

In contents of up to 25 at.% exhibit sufﬁcient magnetic

properties. Therefore, recording performance simulta-

neously to a substantial improved environmental sta-

bility compared to nondoped RETM layers (Raasch

1996), proven for 150 hours at 85 1Cand85%relative

humidity. The later values represent common acceler-

ated aging test conditions.

See also: Amorphous and Nanocrystalline Materials;

Magnetic Films: Anisotropy; Magneto-optical Disks:

Thermal Effects; Magneto-optic Recording: Total

Film Stack, Layer Conﬁguration

Bibliography

Buschow K H J 1985 Research on amorphous alloys. Philips

Tech. Rev. 42, 48–57

Hansen P 1988 Numerical data and functional relationships in

science and technology. Crystalline and amorphous ﬁlms

with rare earth and 3d transition elements In: Wijn H P J

(ed.) Landolt–Bo

rnstein. Springer, Berlin, New Series III 19,

pp. 136–287

Kissinger H E 1957 Reaction kinetics in differential thermal

analysis. Anal. Chem. 29, 1702–6

Klahn S, Bentin H, Kno

rr B, Heitmann H, Wilting H 1990

Oxidation kinetics and morphology of dc magnetron sput-

tered and evaporated amorphous GdTbFe ﬁlms. J. Appl.

Phys. 67, 1442–8

Klahn S, Hartmann M, Witter K, Heitmann H 1987 Structural

relaxation and surface induced aging effects of amorphous

794

Magneto-optic Recording Materials: Chemical Stability and Life Time

GdTbFe ﬁlms. In: Hernando A, Madurga V, Sanche

Trujillo M C, Vazquez M (eds.) Magnetic Properties of

Amorphous Metals. Elsevier, Amsterdam, pp. 176–8

Mergel D, Heitmann H, Hansen P 1993 Pseudocrystalline

model of the magnetic anisotropy in amorphous rare earth

transition metal thin ﬁlms. Phys. Rev. B 47, 882–91

Raasch D 1993 Recording characteristics of Dy–FeCo-based

magneto-optical disks in comparison to other MO materials.

IEEE Trans. Magn. 29, 34–40

Raasch D 1996 Magnetic and recording properties of amor-

phous TbFeCo–In thin ﬁlms. J. Appl. Phys. 80, 2964–7

Raasch D, Bernards J P C, Clausen C, Much G 1995 Magnetic

and magneto-optical properties of Tb–FeCo–Cr layers.

J. Magn. Magn. Mater. 148, 359–60

Raasch D, Hansen P, Mergel D 1994a Bloch wall energy in (Tb,

Nd)–(Fe, Co) thin ﬁlms. J. Appl. Phys. 76, 8022–4

Raasch D, Reck J, Mathieu C, Hillebrands B 1994b Exchange

stiffness constant and wall energy density of amorphous

GdTb–FeCo thin ﬁlms. J. Appl. Phys. 76, 1145–9

Saito N, Togami Y, Morishita T, Sato R 1986 Changes of

structure and magnetization of amorphous rare earth transi-

tion metal ﬁlms by annealing. Jpn. J. Appl. Phys. 25,L850–2

Weast R C 1976 Handbook of Chemistry and Physics. CRC

Press, Cleveland, OH

D. Raasch

Philips Research Laboratories, Aachen, Germany

Magneto-optic Recording: Overwrite and

Associated Problems

In the thermomagnetic write process, the magnetic

storage layer is locally heated to its Curie temperature

) by a focused laser beam. A magnetic domain is

formed during cooling with its magnetization parallel

to the direction of the external write ﬁeld. This pro-

cess is assisted by the demagnetizing ﬁeld. Two modes

to write information are applied: magnetic ﬁeld

modulation (MFM) and laser intensity modulation

(LIM). With MFM, the laser power is kept constant

and the magnetic ﬁeld is switched according to the

bit stream. The main issue here is to get sufﬁciently

fast ﬁeld switching (Lin 1990). With LIM, the laser

intensity is modulated under constant external mag-

netic ﬁeld. The data rate is higher than with MFM.

However, only new domains are written. The old

domains have to be erased before, in a separate run,

so that there is no direct overwrite.

Two main schemes have been proposed to achieve

direct overwrite with laser intensity modulation

(LIM–DOW). Both use specifically designed storage

media, i.e., a single layer with suitable magnetic

properties and exchange-coupled multiple layers

(ECMLs), respectively. Circular domains in an ex-

change-coupled double layer, ECDL, are sketched in

Fig. 1.

1. Force Equilibrium at the Domain Wall

The total energy of a cylindrical domain is given by

the sum of three energies: wall energy, Zeeman, and

demagnetizing energy (Mergel 2000). The radial

forces on the domain wall are obtained by differen-

tiating with respect to the radial coordinate r. The

pressure on the wall, i.e., the force f

per unit area of

domain wall, is obtained by dividing the radial force

by 2prh, where h is the height of the domain, i.e., in

general the thickness of the corresponding layer:

¼2m

½H

ðrÞþH

ðrÞþðH

ðrÞÞ







ð1Þ

The resulting expressions have the dimension of

energy density. The ﬁrst square brackets enclose the

external write ﬁeld, H

, and the demagnetizing ﬁelds,

, due to the domain in one layer, and H

if there

is a second layer. H

favors parallel orientation of

the magnetization in the two layers. The demagnet-

izing ﬁeld, H

, always acts expanding and is given

by:

ðrÞ¼M

1 þ 3r=2h



ð2Þ

Under LIM–DOW conditions, the applied ﬁeld,

, always represents an expanding force. Domain

erasure is driven by the contracting wall forces

summarized in the second set of square brackets in

Eqn. (1). The last term in Eqn. (1), (s

/h), arises

from an interface wall and has only to be taken into

account in exchange-coupled layers.

2. Single-layer DOW

In a single layer, Eqn. (1) applies without the terms

(r)) and (s

/h). The collapse of domains is achi-

eved by heating to a temperature where the contract-

ing domain wall forces overcome the sum of the

Figure 1

Circular magnetic domains in an exchange-coupled

double layer consisting of a write layer w and a memory

layer m. Top view and cross-section. The numbers

designate the three steps in DOW.

795

Magneto-optic Recording: Over write and Associated Problems

expanding forces, due to the demagnetizing and ex-

ternal write ﬁelds, and the coercive ﬁeld. This is

possible if the coercive forces are only moderate, e.g.,

for GdTb–FeCo, but not for Tb–FeCo, the standard

composition. Domain erasure is fostered when the

underlayers that are deposited onto the disk before the

MO layer are polished because then the wall pinning

strength is reduced. Furthermore, the demagnetizing

ﬁeld must be small. The best performance (CNR ¼

32 dB) in a standard recorder was obtained for 24–25

at.% rare-earth content yielding low magnetization at

the erase temperature (Schultz et al. 1993).

The principle of single-layer DOW leads to a di-

lemma. For erasure, on the one hand, the domain

walls have to move. During the write process, on the

other hand, they must not move because moving do-

main walls become jagged and increase the write

noise (Mergel 2000). This dilemma leads to contra-

dicting requirements for the different steps of record-

ing which could not be solved. For easy and complete

domain erasure, materials with low coercive ﬁeld

and low magnetization are needed. Then, however, a

write ﬁeld higher than speciﬁed for standard record-

ers, 200 Oe (16 kAm

1

), is needed in order to write

domain patterns with a high signal-to-noise ratio.

The high ﬁeld is necessary to assure a force equilib-

rium with f

¼0 during domain formation in the write

process. If f

a0 and the coercive ﬁeld is low, the do-

main walls move. However, moving domains become

jagged. The write noise increases and the storage

density is reduced. Therefore, attempts to achieve a

sufﬁciently high signal-to-noise ratio under realistic

recording conditions failed and single-layer DOW

has not been implemented on practical disks.

3. DOW in an Exchange-coupled Double Layer

Domains in an exchange-coupled double layer

(ECDL) are thermomagnetically created in the three

steps illustrated in Fig. 1. An ECDL for direct over-

write consists of a write layer, w, into which domains

are thermomagnetically written during the write

process (step 1) and a memory layer, m, into which

the domains are copied during cooling below its T

(step 2). Now common domains extend through both

layers. The original domains in the write layer are

then erased at room temperature by an initializing

ﬁeld, H

ini

, of several kOe provided by a permanent

magnet (step 3). The reversal of the magnetization of

the write layer creates magnetic interface walls.

Domain erasure is supported by the existence of an

interface wall. The term ( s

/h) in Eqn. (1) represents

the strongest contracting force due to the small value

of h as compared to r. The write temperature is above

the T

of the memory layer, m, such that there is no

interface wall during the write process. The contract-

ing forces are much smaller than during erasure such

that the dilemma of single-layer overwrite is avoided.

Such a write mechanism is possible if there is a

cross-over of the H

¼H

(T) curves of the two layers,

as illustrated in Fig. 2 (Saito 2000). At room tem-

perature, the initializing ﬁeld, H

ini

, must be between

the critical switching ﬁelds for the two layers. The

creation and annihilation of the interface walls modi-

ﬁes the limits of the write (H

write

) and initializing

ini

) ﬁelds as will be discussed later.

Direct overwrite is achieved by switching the laser

power between an erase and a write level. At the erase

level, the magnetic layers are heated to a temperature

where the coercive energy of the storage layer is re-

duced below the interface wall energy. The domains

are annihilated by outside-in wall motion. At the

write level, new domains are written into the write

layer.

In order to control the exchange coupling, a thin

intermediate layer is interposed between the storage

and the write layer. A material with small anisotropy

constant, K

, is chosen such that the wall energy,

s ¼4O(AK

), becomes smaller than in the adjacent

layers. Then, the interface wall is located the inter-

mediate layer. This allows a wider range of compo-

sitions of the other layers. The thickness of the

intermediate layer is about 10 nm corresponding to

typical wall thicknesses in RE–TM alloys (Mergel

2000).

4. Exchange-coupled Multilayers with a

Permanent Magnetic Layer

The external initializing magnet in step 3 can be re-

placed by an additional permanent magnetic layer, p.

The essential parameters of a typical layer stack for

such a case are summarized in Table 1 Tsutsumi

Figure 2

Coercive ﬁeld of the two layers, write layer, w, and

memory layer, m, in an ECDL that allows direct

overwrite. The temperature regions of the three steps of

DOW are indicated below the T axis. The open circles

refer to a P-ECDL and the closed circles refer to an

A-ECDL (compare Fig. 4).

796

Magneto-optic Recording: Overwrite and Associated Problems

(1993). The layer, p, is separated from the write layer,

w, by a switching layer, s, that couples w and p when

room temperature is approached. The switching layer

has a low T

so that p is decoupled from w during the

write process. There are constraints on the thickness

of the layers. The total stack should not be thicker

than 150 nm to assure sufﬁcient write sensitivity. The

thickness of the memory layer should be bigger than

30 nm in order for the MO readout signal not to be

affected by the other layers.

5. Reversal Mechanisms in Exchange-coupled

Layers

So far, the principal mechanisms of LIM–DOW have

been explained. For a discussion of associated prob-

lems, the phenomena occurring in exchange-coupled

multilayers must be discussed in more detail (Mergel

1993, 2000, Tsunashima 2000, Kobayashi et al. 1981).

The reversal mechanisms in ECDLs are compli-

cated. A great variety of hysteresis loops are ob-

served. They are determined by the interface wall

energy density, s

, the magnetization, M

, the coer-

cive ﬁeld, H

, and the thickness, d

, of the layers. The

different layers are denoted by the index, z. A mag-

netic reversal takes place when the balance of coer-

cive, Zeeman, and wall energy densities yields a

negative value:

042m

ðH

þ v7H

7ÞM

þ ws

; ðv; w ¼ 71Þð3Þ

04ðH

þ v7H

7ÞþwH

ð4Þ

ð5Þ

with v ¼þ1or1 when the magnetization is rotated

against [ag] or into [in] the external ﬁeld, respectively,

and with w ¼þ1or1 when an interface wall is cre-

ated [c] or annihilated [a] by the process, respectively.

The abbreviations in the square brackets [a, c,

ag, in] are used to designate magnetic reversal pro-

cesses. Eqns. (3) and (4) can only be fulﬁlled when v

or w or both are negative. Three processes are, there-

fore, allowed and all play a role in LIM–DOW (Mer-

gel 1993)

(i) [c, in, z] Creation of an interface wall by rota-

tion of M

into the external ﬁeld.

(ii) [a, in, z] Annihilation of the interface wall with

rotation of M

of layer z into the external ﬁeld.

(iii) [a, ag, z] Annihilation of the interface wall

with rotation of M

against the external ﬁeld.

An example of a hysteresis loop of an ECDL is

given in Fig. 3(c). The isolated layers are represented

by (a) and (b). When the two layers are exchange

coupled the following changes may be observed:

(i) Reversal of layer 1 by [a, in, 1] may occur by

motion of the interface wall. In such a case, the hys-

teresis curve becomes sheared.

(ii) The reversal ﬁelds are shifted to lower or high-

er values because domain walls are annihilated or

created, respectively.

(iii) Minor loops are not symmetric to zero ﬁeld

but are centered around a value given by

ð6Þ

where M

and d

are magnetization and thickness,

respectively, of the reversing layer, z. The ﬁrst two

effects have to be taken into account when predicting

recording properties of ECDLs from the properties of

the isolated layers. The interface wall energy can be

determined from Eqn. (6).

In ferrimagnetic RE–TM alloys, the TM moments

alone determine the exchange-coupling whereas the

magnetization, M

, that interacts with the external

magnetic ﬁeld, is given by the difference of the TM

and RE submagnetizations (M

M

). This

deﬁnes two types of ECDLs (see Fig. 4), A and P.

In P-type double layers, M

is parallel or antiparallel

to M

in both layers. In A-type double layers, M

Table 1

Composition and properties of a LIM–DOW multilayer.

Layer Composition

Thickness

(nm)

(1C)

(kOe)

m, memory Tb

30 200 12

i, intermediate Gd

w, writing Dy

30 280 10

s, switching Tb

10 170 2

p, permanent,

initializing

30 4300 8

Source: Tsutsumi (1993).

Figure 3

Hysteresis loop of an exchange-coupled double layer (b).

Single layers (a) layer 1 and (c) layer 2.

797

Magneto-optic Recording: Over write and Associated Problems

parallel to M

only in one layer and interface walls

exist at inﬁnite external ﬁeld.

We now apply the reversal systematics to derive

the conditions for the LIM–DOW mechanism in

ECDLs.

Writing: Avoid [a, ag, w]

Copying: Enable [a, in, m] for P or [a, ag, m]

for A, respectively.

Initialization: Enable [c, in, w]: H

ini

þH

Stable domains after initialization:

Avoid [a, in, m], H

ini

H

for P, inside a

circular domain

Avoid [c, in, m], H

ini

þH

for A, outside

a circular domain.

The restrictions for the initialization ﬁeld are indi-

cated in Fig. 2 by circles. The upper limit is higher for

A-ECDLs.

6. Media Design

In order to achieve reliable operation, the composi-

tion and the thickness of the various layers in a LIM–

DOW medium have to be chosen appropriately. In

the following, the requirements for the design of a

ﬁve-layer stack and their implications for the chemical

composition of suited RE–TM alloys are schematic-

ally summarized. For a typical example see Table 1.

In the memory layer, m, the domains must remain

stable under the high initializing ﬁeld. This implies a

high H

which is achieved by choosing terbium as the

dominant rare-earth element. Terbium has a strong

spin-orbit coupling that leads to a large anisotropy

constant, K

. H

increases with K

with a power be-

tween 0.5 and 4. The T

in the memory layer, m,

must be lower than in the write layer, w. Cobalt has a

stronger exchange-coupling than iron and leads to a

higher T

in RE–TM alloys. Therefore, w, contains

more cobalt than m.

The intermediate layer must have a small wall

energy in order for the interface wall to be situated

in this layer. This is achieved by a high gadolinium

content. Due to lack of spin orbit coupling, gadolin-

ium leads to a low anisotropy constant, K

, and a low

interface wall energy, s

¼ 4O(AK

In the write layer, the force on the domain wall

must be small during the write process in order

to avoid domain wall motion that leads to jagged

domain boundaries. Dy–FeCo is easily optimized to

meet these requirements (Raasch 1993).

The initializing layer must keep a permanent mag-

netization. A high H

and a high T

are needed. They

are obtained by high concentrations of terbium and

cobalt.

7. Write Pulse Modulation

The temperature proﬁle on the disk is inﬂuenced by

heat ﬂow from neighboring marks written before or

after the considered mark. Therefore, the mark shape

depends on the information pattern, and more do-

main length jitter is generated. This decreases the

signal-to-noise ratio and consequently the storage

density. The problem could be solved by modifying

the modulation scheme of the laser power as shown

in Fig. 5 (Saito 2000, Sect. 8.3.5). An assisting power

is applied, heating the track to a medium tempera-

ture, and shutoff for some time after a domain has

been written. This reduces the additionally intro-

duced heat when a domain is written. The write pulse

is replaced by a sequence of short pulses such that the

write process takes place at a lower temperature.

8. Concluding Remarks

The principle of single-layer DOW leads to contra-

dicting requirements for the different steps of record-

ing. For erasure, on the one hand, the domain walls

have to move. During the write process, on the other

hand, they should not move because moving domain

walls become jagged and increase the write noise.

This dilemma could not be solved and the attempts to

achieve a sufﬁciently high signal-to-noise ratio under

realistic recording conditions failed.

A typical layer stack for DOW in exchange-coupled

layers comprises ﬁve layers: a write, a memory, and

an initializing layer together with two intermediate

layers. The initializing layer has a permanent magnet-

ization. Domain erasure is supported by an interface

wall between the write and the memory layer. Dur-

ing the write process, the write layer is magnetically

Figure 4

Two types of ECDLs, A and P, based on ferrimagnetic

materials. The thin and bold arrows refer to the TM

moments and the total magnetization, respectively.

Figure 5

Laser power modulation pattern for DOW on ECMLs

in order to avoid thermal interference.

798

Magneto-optic Recording: Overwrite and Associated Problems

decoupled from both the initializing and the memory

layer and the dilemma of single-layer DOW is

avoided.

The relevant magnetic properties like T

and mag-

netic wall energy density can be tuned separately

for every layer by the composition of the RE–TM

alloys. Therefore, overwrite with ECMLs can be im-

plemented in practical disks and yields a good re-

cording performance.

See also: Magneto-optical Disks: Thermal Effects;

Magneto-optic Multilayers; Magneto-optic Record-

ing: Total Film Stack, Layer Conﬁguration; Magneto-

optical Effects, Enhancement of; Magnetic Recording

Technologies: Overview; Super-resolution: Optical

and Magnetic Techniques

Bibliography

Gambino R J, Suzuki T (eds.) 2000 Magneto-optical Recording

Materials. IEEE Press, Piscataway, NJ

Kobayashi T, Tsuji H, Tsunashima S, Uchiyama S 1981 Mag-

netization process of exchange-coupled ferrimagnetic double-

layered ﬁlms. Jpn. J. Appl. Phys. 20, 2089

Lin C J 1990 Critical assessment of the physics underlying di-

rect overwrite in magneto-optic recording. J. Appl. Phys. 67,

4409–14

Mergel D 1993 Magnetic reversal processes in exchange-

coupled double layers. J. Appl. Phys. 74, 4072

Mergel D 2000 Domain dynamics and recording physics. In:

Gambino R J, Suzuki T (eds.) Magneto-optical Recording

Materials. IEEE Press, Piscataway, NJ

Raasch D 1993 Recording characteristics of Dy–FeCo-based

magneto-optical disks in comparison to other MO materials.

IEEE Trans. Magn. 29,34

Saito J 2000 Light-intensity modulation direct-overwrite. In:

Gambino R J, Suzuki T (eds.) Magneto-optical Recording

Materials. IEEE Press, Piscataway, NJ, Chap. 8

Schultz M D, Kryder M H, Sekiya M, Chiba K 1993 True

direct overwrite in single layer magneto-optical recording.

IEEE Trans. Magn. 32, 3772

Tsunashima S 2000 Exchange-coupled ﬁlms. In: Gambino R J,

Suzuki T (eds.) Magneto-optical Recording Materials. IEEE

press, Piscataway, USA, Chap. 7

Tsutsumi K 1993 Direct overwrite materials for magneto-

optical recording. J. Magn. Magn. Mater. 120, 247–53

D. Mergel

Universita

t Essen, Germany

Magneto-optic Recording: Total Film

Stack, Layer Conﬁguration

The currently used rare-earth transition-metal (RE–

TM) recording media suffer from two main disad-

vantages: the RE elements oxidize easily and the

polar Kerr effect and consequently the signal strength

of the magneto-optic readout decrease significantly at

shorter wavelengths. Pt/Co multilayers have advan-

tages relative to RE–TM media in both aspects.

The larger Kerr effect of Pt/Co below a wavelength

of 450 nm promises a higher storage density in future

recording with blue lasers. However, the recording

performance is also determined by several other

properties:

the write sensitivity within the speciﬁed ranges

of the magnetic ﬁeld and the laser power

the magnetic stability of the written domains

the noise level to get a high signal-to-noise ratio

and low error rates

degradation of the disk after repeated read/

write cycles (see Magneto-optic Recording Materials:

Chemical Stability and Life Time).

In this article, the recording-relevant properties of

Pt/Co multilayers are discussed and compared with

those of RE–TM media.

1. Magnetic Properties of Pt/Co Multilayers in

Comparison with those of RE–TM Films

The saturation magnetization M

and the coercive

ﬁeld H

of Pt/Co multilayers are displayed in Fig. 1 as

a function of temperature together with M

of a typi-

cal amorphous RE–TM ﬁlm.

Pt/Co multilayers generally exhibit a higher T

than amorphous RE–TM ﬁlms. They need a higher

laser power for the thermomagnetic write process and

their signal-to-noise ratio of written domains de-

grades more rapidly with the number of write/erase

Figure 1

Saturation magnetization M

and coercive ﬁeld H

of a

typical (Pt/Co) multilayer as a function of the

temperature. For comparison, M

(T) of a typical

RE–TM composition is also shown.

799

Magneto-optic Recording: Total Film Stack, Layer Conﬁguration