- Email: [email protected]
Magneto-optic multilayers: Fundamental and technological aspects
~
ELSEVIER
Journal of Magnetism and Magnetic Materials 175 (1997} 90-98
Journalol magnetism J H and magnetic materials
Magneto-optic multilayers: Fundamental and technological aspects R. Krishnan a'*, M. Nyvlt b, S.
Visnovsky b
Laboratoire de Magnbtisme et d'Optique, University de Versailles, Bdtiment Fermat, CNRS URA 1531, 45 Avenue des Etats-U)lis, 78035 Versailles cedex, France blnstitute o['Physics, Charles Universi~, Ke Karlovu 5, 12 116 Prague 2, Czech Republic
Abstract We describe our magnetic and magneto-optic (MO) studies on some multilayers based on Ag, Au and Pt, such as Fe/Ag, Fe/Pt, Co-Ni/M (M = Pt, Au) prepared by evaporation under UHV conditions. In Co-Ni/M, the surface anisotropy increases linearly with the addition of Co verifying the single ion model. The Curie temperature can be tailored by the addition of Co without affecting adversely the properties of interest for MO storage applications. For instance, in Fe/Ag, the MO effect shows enhancement due to Ag plasma resonance and in Fe/Pt the effect is due to interaction of Pt. The spectra can be explained by a phenomenological model though quantitative agreement is not obtained. In Au-based multilayers, the MO Kerr spectra show the effect of plasma resonance of Au, particularly for thinner layers. Some of the results could be modelled by electromagnetic model. The Kerr spectra for Co-Ni alloy films could also be modelled with success using effective medium theory which also confirms the absence of any effect from hybridisation between Co and Ni a result which supports our conclusions on the surface anisotropy.
Keywords: Multilayers; Anisotropy; Magneto-optics; Information storage
1. Introduction Magnetic multilayers are a new class of materials with specific novel properties, such as, enhancement of magnetic moment, surface anisotropy, giant magnetostriction, etc. The application potential of these materials gave a big boost to intense research activity in this field. In this paper, we will
* Corresponding author. Tel.: + 33 1 3925 4658; fax: + 33 1 3925 4652; e-maih [email protected].
focus our attention to the anisotropy and the magneto-optical (MO) effects. Indeed, one of the most promising applications of magnetic multilayers (ML) is for the high-density M O information storage in the blue wavelengths and we will also discuss this aspect in this paper. The typical examples are Co/Pt and Co/Pd, which by virtue of their strong uniaxial anisotropy and good magneto-optical effects, have attracted much attention from several research groups [1-3]. Despite several advantages of these systems, over the current rare-earth transition-metal amorphous films, there is a serious
0304-8853/97/$17.00 ~(" 1997 Elsevier Science B.V. All rights reserved PII S 0 3 0 4 - 8 8 5 3 ( 9 7 ) 0 0 1 5 2 - 2
R. Krishnan et al. / Journal of Magnetism and Magnetic Materials 175 (1997) 90- 98
problem for practical applications which is the following. The Curie temperature of these multilayers even for Co layers as thin as 0.5 nm is in the range of 35(~400°C. Thermomagnetic writing at such temperatures not only necessitate high-power laser, but also would cause interdiffusion causing the destruction of the multilayers structure. Therefore, it is necessary to find new class of materials with lower Tc. Multilayers based on (CoxNil x)/Pt instead of pure Co/Pt have been suggested recently by some authors and by us as well [-4 6] which show similar properties as Co/Pt but with a relatively lower Curie temperature. We also found that the surface anisotropy (Ks) in this system varied linearly with the Co concentration which in our opinion is of fundamental importance [5]. We therefore wanted to study, in detail, this aspect and extended our study also to (CoxNil_x)/Au. We chose the system based on Au for the following reasons. Co/Au has been intensely studied by many in the past [7, 8] and it shows perpendicular anisotropy for thin Co layers. We had shown that in Ni/Au multilayers K~ is negative [9] and it is, therefore, interesting to see if the addition of Co would make Ks positive. For the sake of completeness, we will also describe the magneto-optical (MO) Kerr spectra of Fe/Ag and Fe/Pt before discussing those of material of practical interest for MO information storage technology.
91
hess t(Pt, Au) was kept constant at 15 ,~ and that of Co Ni varied in the range 4 30A. The growth parameters will be designated as {t(Co, Ni), t(Pt, Au)} x N, where N stands for the number of bilayers which was in the range 1~25. The top layer was either Pt or Au layer 20 A thick which also served as a protective layer. Fe/Ag and Fe/Pt multilayers were prepared by RF sputtering. The details of the preparation can be found in Ref. [10]. Fe/A~ the multilayer was grown on Ag buffer layer 60 A, thick. We consider here three samples with t(Ag) = 60 ~, and t(Fe) = 110, 36 and 6 A, respectively. The top layer was Ag. Fe/Pt multilayers were grown on a buffer layer of Pt 1 0 0 k thick, t(Pt) was fixed at 15A and t(Fe) = 2.4, 8, 18 and 36 A, respectively. The top layer was 15 ,~ thick Pt. Low- and high-angle X-ray diffraction studies were made to verify the periodic structure and to calculate the layer thicknesses. Magnetization was measured using a vibrating sample magnetometer (VSM) and the anisotropy by a home-made sensitive torque meter in the temperature range 6 300 K. The M H and magneto-optical loops were also taken at the laser wavelength of 633 nm. Magneto-optical polar Kerr spectra were studied at 295 K in the photon energy range 1.5 5.2 eV.
3. Results and discussions 2. Experimental details
3.1. Magnetic anisotropy
The (CoxNi~_x)/Pt,Au multilayers were prepared by sequential evaporation using a dual ebeam under ultra-high-vacuum conditions ( < 5 x 10-9 Torr during deposition). The details of the deposition can be found in Ref. [2]. For depositing the CoxNil x alloy layers, an alloy ingot was used as a source. Subsequently, the composition was determined on a film of CoxNi~-x about 150 m thick. No difference in the composition between the source and the layer could be detected. Both glass and silicon were used as substrates. First, Pt, Au buffer layer 100 A. thick was deposited at room temperature and then annealed at 200°C for 2 h. The multilayers were then grown on this buffer layer at room temperature. The Pt, Au layer thick-
Let us first discuss the results on the anisotropy (CoxNil _x)/Pt,Au. The effective anisotropy Kerr in the multilayers can be expressed by the well-known phenomenological model by the relation, Keff = Kv + 2Ks~t, where, Kv and K~ represent the bulk and surface anisotropies and t the magnetic layer thickness. The term Kv contains the contributions from the demagnetization energy 2rcM 2 and the crystalline (Kcr) and magnetoelastic anisotropy (Kme) energies. Plotting K~ff x t as a function of t yields both Kv and K~. The results were analysed on the basis of the above model. Fig. 1 shows the linear variation of the product Kef f × t as a function of t for the composition Coo.lNio.9 at 295 and 5 K which is in accordance
R. KHshnan et al. / Journal ( f Magnetism and Magnetic Materials 175 (1997) 90 98
92
0.5 0.5.
0.4
"'~&
E © v o')
"'"
0 295K
o
~"~
0.3
.~
0.1
I
.d 0 "7
T=5K
O
to9
-0.2 0.2
]
I
I
i
0.4
0.6
0.8
_
1
Co (x) 0.5
Fig. 3. The variation of K~as a functionof content in Co Ni/Au at5K.
Fig. 1. The v a r i a t i o n of the p r o d u c t K~ff x t as a function o f t at 295 and 5 K.
0.7 - -
i
[
i
0.6 0.5 O
0.4
.9
=
0.3 0.2
T=5K
0.1 0
__
i
I
1
I
0.2
0.4
0.6
0.8
I
1
Co (x) Fig. 2. The variation of K, as a function of Co content in Co Ni/Pt at 5 K.
with model cited above. It can be seen that the values of K e f f increase considerably at 5 K indicating the proximity of Tc to room temperature. Therefore, to avoid such effects we will consider all the results at 5 K. Let us now discuss the Co concentration dependence of K~. Fig. 2 shows the variation of Ks as a function of Co (x) concentration in C o - N i / P t at 5 K. The experimental points show a linear variation of Ks with the Co content. The straight line is calculated based on the relation, assuming one ion contribution to anisotropy, which can be written as, K an°y = KC°(x) + K y i ( 1 - X). Here K c° and K~ ~ stand for the surface anisotropy of Co and Ni
in Co/Pt and Ni/Pt multilayers, respectively. We experimentally determined these values in Co/Pt and Ni/Pt multilayers, as 0.6 and 0.17 erg cm 2, respectively. So taking the above values we calculated the straight line in Fig. 2. The excellent agreement confirms our assumption. Fig. 3 shows a similar result for C o - N i / A u system. It is noteworthy that here, although K, is negative for x = 0, it starts becoming more and more positive with the addition of Co. The increase in K~ could be modelled as before and we can write, K2"°Y= KC°(x) + K~i(1 - x), where, K c° and K~ i are now surface anisotropies found in Co/Au and Ni/Au multilayers. KNi=~ --0.1 e r g c m 2 at 5 K from our earlier work on Ni/Au [9] and K ~ ° = + 0.52 erg cm -2 at 295 K (Tc being very high, this value is expected to be the same at 5 K) from Ref. [8]. The straight line in Fig. 3 calculated using the above values explain the experimental points very well and confirm the validity of the one ion model that we have assumed. These results also show that the contribution from magneto-elastic effects to Ks is negligible if not absent. Because the magnetostriction in Co Ni alloys vary in a complex manner and even changes sign [11], if there is any contribution from this, then one can hardly expect a linear variation. So we are led to conclude that the surface anisotropy in these systems arises mainly from crystal field effects (N6el type) and can be treated as single ion anisotropy. It implies that the modification of the band structure of Ni by the addition of Co and vice versa does not
R. Krishnan et al. / Journal of Magnetism and Magnetic Materials 175 (1997) 90-98
93
®K
H
He = 1.2 kOe Fig. 4. Kerr loop for Coo.7Nio.3/Ptsample (4.5,/15A)x 10.
seem to influence the contribution of the surface atoms to the anisotropy. More theoretical work in this area would probably be welcome.
3.2. M.O. loops In (Co,Nil _x)/Pt,Au systems, with the addition of Co and with decreasing layer thickness, the loops get more and more rectangular and the coercivity also increases• Figs. 4 and 5 show the typical loop for the Coo.3Nio.7/Pt sample (4.5/15 A)x 10 and Coo.TNio.g/Au sample (5.4/15 A) x 16. However, for the same modulation when the Au buffer layer thickness was increased to 300 A, the coercivity more than doubled and increased to 800 Oe. Detailed structural studies using X-ray diffraction also showed that the interface quality improved significantly when the buffer layer thickness was increased from 100 to 300A [12]. So some optimisation is needed to improve further the loop characteristics.
3.3. Magneto-optical studies First, let us describe our results on Fe/Ag. These spectra were taken with a field of 0.45 T and hence, not the saturation values. We have, therefore, used appropriate scaling factors for the calculation of the spectra to allow for this. Fig. 6 shows the Kerr
0.05 ,
llOA 36A
• •
bulk Fe
bO
.......
0.00 °~ °0 ~ 0
-0.05
(D
f
}.,
I" / -0.10
1.5
2.0
2.5
Photon
3.0
3.5
4.0
Energy
4.5
5.0
(eV)
He = 0.45 kOe
Fig. 5. Kerr loop for Coo.~Nio.3/Au sample (5.4/15 ,~) × 16.
Fig. 6. Polar Kerr rotation spectra for three Fe/Ag multilayer samples.
94
R. Krishnan et al. /Journal of Magnetism and Magnetic Materials 175 (1997) 90 98
spectra for the three samples. Here, the bulk Fe data has been scaled down by 0.2. Anomalous line shapes are observed in the region of plasma resonance of Ag near 3.9 eV. The effect is predominant particularly in the thinnest sample with t ( F e ) = 6 A. The observed spectra can be explained by a phenomenological model proposed by one of us [-13]. The model assumes step like profile near the interfaces and assumes the optical constants of the bulk metals. Agreement was best for the thickest sample with t(Fe) = 110 ,~ as shown in Fig. 7 [14]. Fig. 8 shows the result for the sample with t(Fe) = 36 ,~. It is seen that the line shapes are well reproduced but the computed amplitude of rotation is considerably smaller than the experimental one. This is due to some over simplification of the model as it is now recognized that the physical properties of Fe at monolayer thickness range may differ significantly from that of bulk Fe. For the thinnest sample, due to the fact that it saturates at a field of about 1.2 T we used a reduction in the
0,05 i
""
t (Fe) = 36
,_ ", '~ ",
v'* v
0.00
~D "~':~)'~ i -0.05 '~'•"'*"**,~, • Ii.
~" ,~
-0.10 eomp. rotation comp. elliptieity . . . . . . . . . . exp. rotation exp. ellipticity t.5
I
i
~
2.o
~.5
3.o
• •
L
3.~
n
4.o
Energy
I
,.5
r
5.0
(eV)
t ( F e ) -- 1 1 0 i t
0,00 "k..\ "'~x.,, k \
•. , .
....
".. :'.,-,:_,.. ~
T.,~ "*¢
e-
•
): ~
o
o -0.10
eomp. rotation comp. elliptieity . . . . . . . . . . exp. rotation exp. ellipticity n
1.5
~: ?' 2:"
,, .; ;
Fig. 8. Experimental and calculated polar Kerr spectra for the sample Fe/Ag with t(Fe)= 36 A. The calculated spectra are reduced bya factor of 0.2 to account for the lower magnetic field used.
0.05
-0.05
~.* :"
M
Photon
t
...........
i
2.0
2.5
Photon
n
3.0
•
I
3.5
n
4.0
Energy
t
I
4.5
5.0
(eV)
Fig. 7. Experimental and calculated polar Kerr effect spectra for the sample Fe/Ag with t(Fe) ~ 110 A..
off-diagonal tensor element spectra of 0.4 instead of 0.2. A better agreement between the experimental and calculated spectra was obtained for the sample with t(Fe) = 6 A. The result is shown in Fig. 9. It is noteworthy that in this case the ellipticity and rotation spectra are dominated by the features of Ag. Four samples of Fe/Pt were studied where t(Pt) was fixed at 15 A and that of t(Fe) were 2.4, 8, 18 and 36 ]~, respectively. For the thinnest sample perpendicular magnetization could be stabilised and the Kerr rotation is saturated at the applied field of 1.07 T. For the other two samples the rotation obtained has been scaled appropriately to get the saturated value. Fig. 10 shows the polar Kerr rotation spectra of the four samples along with that of the bulk iron. For the thickest sample the spectrum resembles that of Fe but for thinner layers it is transformed. The thinnest sample can be considered as F e - P t alloy and the peak near 4.5 eV arises from the interaction of Pt-Fe.
R. Krishnan et al. /Journal of Magnetism and Magnetic Materials 175 (1997) 90-98 0.1
'
i
'
95
I
'
q~2~, ellipticity t (Fe)
=
'
'
J ~/
~
l
6/~
~z~ 1 . 4 n m HDO0~
0.0
/
....
o
o.o ,
,~
-" ,!"
Ill
, . ~.
I1\
"
O;
% ,,x
J
N -o.2
ID
rotation "%_
_~-%
-0.! ,
eomp. r o t a t i o n comp. ellipticity . . . . . exp. r o t a t i o n exp. ellipticity -0.2
1.5
2.0
2.5
, ;
2
3.0
3.5
4.0
Energy
4.5
5.0
i
Fe/Pt
','~'ax'~.,,•A
o '~
s_
M
(8/15) tz
• •
•
,~'"m"'="====,,
"~
i
- ( 2 . 4 / l a ) 22
.~
( 1B /
z,
15
)
15
~,,• 0 (36/15) 1°
-0.2
""
•
-0.a
-0.4
"="'--
/ ~ F e ,
I
2
,
3
,
I
,
4
p
5
I
3
,
Fig. 11. M O spectra in Coo 3Nio.7/Pt for t = 14 and 4.5 A. Insert: P K R loop for 4.5 A.
(eV)
i
i
-0.1
r
•
Fig. 9. Experimental and calculated polar Kerr effect spectra for the sample Fe/Ag with t(Fe) = 6 A. See text for details of calculation.
,,
0,45
Photon energy in eV
•
Photon
0.0
~
I
4
,
I
5
Photon energy in eV Fig. 10. Polar Kerr rotation in Fe/Pt multilayer samples. The spectrum of bulk Fe is also shown for comparison.
Fig. 11 shows the Kerr spectra in Coo.3Nio.7/Pt for t = 14 and 4.5 ~,. One can notice the large negative peak in the rotation near 4 eV. This peak has been attributed to the effect arising from Co-Pt interactions in Co/Pt and also been observed by us in Ni/Pt [15]. The peak position shifts to lower energy as the Ni concentration increases. It is noteworthy that the rotation of 0.28 ° at 3 eV (wave length of interest for high-density magneto-optical information storage) of the thinner sample is much higher than that of even Co/Pt with similar magnetic layer thickness and, furthermore, the Tc is close to 200°C. All these characteristics make this material a very interesting candidate for application. Figs. 12 and 13 demonstrate the effect of the Co content (x) on the magneto-optical polar Kerr rotation (PKR) and ellipticity (PKE) spectra in Co~Nil_x/Au multilayers for various t(Co-Ni). In the case of thinner magnetic layer (approx. 3 atomic layers) the optical contribution from the interband transition in gold situated near 2.75 eV is clearly manifested especially for Ni/Au. A scaling factor of 6.45 has been employed in order to account for the low Tc of this sample. The addition of Co leads to an increase in both Tc and MO activity of the
R. Krishnan et al. / Journal of Magnetism and Magnetic Materials 175 (1997) 90 98
96 I
o.o5
I
- ~
I
I
ell¢~tiei~y
0.05
I
'
I
I
I
ell£~ic{~Y 0.00
O.OO
.=_ "6
W%.. f l
-0.05
-O.IO
-0.I~
;2
-0.20
*,~.,.~
h
I
,
I
2
,
ro~a~on I
3
,
4
-0.25
'
I
'
I
'
I
~d
-0.2
rotation 2
3
I
~
i
I
i
I
4
Photon energy in eV
I
Fig. 12. M O spectra in C o - N i / A u (6.4/15), x = 0, N = 25 (A) and x = 0.7, N = 16 (@).
I
2
-r.o ~ a t { o r L ,
5
Photon energy in eV
•
I
4
5
Photon energy in eV Fig. 13. M O spectra for x = 0 . 7 , (18/15) x 9 (circles); x = 0.35, (18/15) x l0 (triangles), x = 0, (27/15) (squares).
magnetic layer and as a result the spectral amplitudes are increased. At higher Co Ni layer thicknesses the Au interband transition effect practically disappears from the spectra as seen in Fig. 14,
Fig. 14. Effect of t(Co-Ni) on M O spectra for Co0.vNio.3/Au. Full circles: (6.3/15)x 16; open circles: (9/15)x 12; full triangles: (18/15) x 9.
which also illustrates the effect of t(Co-Ni) on the spectra when both t(Au) and x-parameter are fixed. As expected the negative peak in PKR originating from Au gradually shifts to lower photon energies as t(Co-Ni) increases. This optical effect can be well understood as a reduced radiation penetration to the Au layers situated below magnetic Co-Ni layers. Therefore, one observes more Co-Ni-like behaviour. As expected, the spectra display an overall increase in amplitude with x and approach that of pure Co. It is noteworthy that for the sample with x = 0 . 7 and t(Co N i ) = 1 8 A , the PKR reaches 0.23 ° at 1.8 eV. Recent MO studies of Co/Au interfaces in sandwiches and multilayers provided detailed information on the nature of the Au-Co interfaces [-16, 17]. Thanks to the sharpness of the interfaces the observed spectra in Co/Au and Ni/Au, ultra thin film systems could be satisfactorily simulated using an electromagnetic model employing bulk metal (or thick film) optical data and step-like composition profile [18]. This approach should also be useful in the case of Co-Ni/Au. For this purpose we have studied some CoxNil x alloy films and looked for practical expressions which would account for the effect of the x-parameter on the MO Kerr spectra.
R. Krishnan et al. / Journal of Magnetism and Magnetic Materials 175 (1997) 90-98 0.2
i
i
i
i
rotation o.1 ~o .~
e llip tici ty 0.0
.:'~//
".
./'"
•~
""
j"
". .... /t././,':" ......--.~allo
i%V.,"/
CO
"'~" ..... ""
x
/
-0,4 ..°'°'° --0.6
'
I
2
,
I
3
,
I
4
:
\,
-
t
I
5
Photon energy in eV Fig. 15. PKE spectra of Cox-Ni 1 x alloy thick films. Experimental for x = 0.5 (0, O); calculated curves, x = 0.5 (-); x = 0.7 (-.-).
Starting from the Maxwell G a r n e t t theory (effective m e d i u m model) [19] modified to a c c o u n t for circularly polarized proper modes in M O medium we have obtained the expression for the x-dependence of the off-diagonal permittivity tensor element. Fig. 15 shows the Kerr effect spectra of bulk Co and Ni, that of Coo.sNio.s alloy film and the calculated curves for x = 0.5 and 0.7, respectively. The measured polar Kerr spectra of Coo.sNio.s alloy film and the modelled one show very g o o d agreement. We remind that the model is appropriate for the materials which can be described as a mixture of non-interacting isotropic particles [18]. It can therefore be inferred that the hybridisation of C o and Ni has no effect on the M O properties. Incidentally, this also lends support to the one ion model that we have proposed for interpreting the surface anisotropy.
4. Conclusion
In conclusion, we have described the magentic and magneto-optical properties of some multilayer systems. In Fe/Ag, the M O spectra show the en-
97
h a n c m e n t in the rotation due to plasma resonance in Ag. In C o - N i / P t , A u multilayers prepared by evaporation, some interesting results have been found. The C o concentration dependence of the surface anisotropy can be explained by a single ion model. The AUo system sample with x = 0.7 and t(Co-Ni) = 5.4 A, shows a rectangular perpendicular loop with coercivity as high as 800 Oe and with polar Kerr rotations close to 0.14 ° near 2.8 eV. However, in the Pt system the multilayer with x = 0.3 and t(Co-Ni) = 4.5 A shows a perfect rectangular loop with He = 1.2 kOe, a Kerr rotation of 0.28 ° at 3 eV and a Tc close to 200°C. These characteristics make this material a potential candidate for magneto-optical storage applications in the blue light. In the M O spectra for thin C o - N i layers the interband transition of Au can be seen but for thicker layers the behaviour is C o - N i like. The spectra can be well explained in terms of the model proposed.
Acknowledgements
We would like to thank H. Lassri, M. Seddat, M. Porte and M. Tessier for their contribution to this work.
References
I-1] W.B. Zepper, J.A.M. Greidanus, P.F. Carcia, P.R. Fincher, J. Appl. Phys. 65 (1989) 4971. I-2] R. Krishnan, M. Porte, M. Tessier, IEEE Trans. Magn. 26 (1990) 2727. [3] H. Takahashi, S. Fukatsu, S. Tsunashima, S. Uchiyama, J. Magn. Magn. Mater. 93 (1991) 469. [4] M. Mes, J.C. Lodder, T. Takahata, I. Moritani, N. Imamura, J. Magn. Soc. Japan 17 (Suppl. S1) (1993) 44. [-5] R. Krishnan, H. Lassri, M. Seddat, M. Porte, M. Tessier, Appl. Phys. Lett. 64 (1994) 2312. 1-6] S. Hashimoto, J. Appl. Phys. 75 (1994) 438. [7] C. Chappert, P. Bruno, J. Appl. Phys. 64 (1988) 5736. [8] F.J.A. den Broeder, D. Kuiper, A.P. van de Moesselaer, W. Hoving, Phys. Rev. Let. 60 (1988) 2771. 1-9] R. Krishnan, H. Lassri, M. Porte, M. Tessier, IEEE Trans. Magn. 29 (1993) 3388. [10] R. Krishnan, M. Tessier, J. Appl. Phys. 67 (1990) 5392. [-11] R.M. Bozorth, Ferromagnetism, Van Nostrand, New York, 1951.
98
R. Krishnan et al. / Journal of Magnetism and Magnetic Materials 175 (1997) 90 98
[12] D. Rafaja, M. Chladek, V. Valvoda, M. Seddat, H. Lassri, R. Krishnan, Thin Solid Films 292 (1997) 61. [13] S. Visnovsky, Czech J. Phys. B 36 (1991) 67. [14] R. Krishnan, T. Sikora, S. Visnovsky, J. Magn. Magn. Mater. 118 (1993) 52. [15] R. Krishnan, M. Nyvlt, V. Prosser, M. Seddat, Z. Smetana, M. Tessier, S. Visnovsky, J. Magn. Magn. Mater. 148 (1995) 283.
[16] M. Nyvlt, V. Prosser, S. Visnovsky, J. Ferr+, D. Renard, R. Krishnan, J. Magn. Magn. Mater. 148 (1995) 287. [17] A. Bounouh, P. Beauvillain, C. Chappert, R. Megy, P. Veillet, J. Magn. Magn. Mater. 148 (1995) 273. [18] R. Krishnan, H. Lassri, M. Nyvlt, V. Prosser, D. Rafaja, V. Valvoda, S. Visnovsky, J. Magn. Magn. Mater. 148 (1995) 285. [19] D.E. Aspnes, Spectroscopic Characterization Techniques for Semiconductor Technology, SPIE, Vol. 452, 1983, p. 60.
ELSEVIER
Journal of Magnetism and Magnetic Materials 175 (1997} 90-98
Journalol magnetism J H and magnetic materials
Magneto-optic multilayers: Fundamental and technological aspects R. Krishnan a'*, M. Nyvlt b, S.
Visnovsky b
Laboratoire de Magnbtisme et d'Optique, University de Versailles, Bdtiment Fermat, CNRS URA 1531, 45 Avenue des Etats-U)lis, 78035 Versailles cedex, France blnstitute o['Physics, Charles Universi~, Ke Karlovu 5, 12 116 Prague 2, Czech Republic
Abstract We describe our magnetic and magneto-optic (MO) studies on some multilayers based on Ag, Au and Pt, such as Fe/Ag, Fe/Pt, Co-Ni/M (M = Pt, Au) prepared by evaporation under UHV conditions. In Co-Ni/M, the surface anisotropy increases linearly with the addition of Co verifying the single ion model. The Curie temperature can be tailored by the addition of Co without affecting adversely the properties of interest for MO storage applications. For instance, in Fe/Ag, the MO effect shows enhancement due to Ag plasma resonance and in Fe/Pt the effect is due to interaction of Pt. The spectra can be explained by a phenomenological model though quantitative agreement is not obtained. In Au-based multilayers, the MO Kerr spectra show the effect of plasma resonance of Au, particularly for thinner layers. Some of the results could be modelled by electromagnetic model. The Kerr spectra for Co-Ni alloy films could also be modelled with success using effective medium theory which also confirms the absence of any effect from hybridisation between Co and Ni a result which supports our conclusions on the surface anisotropy.
Keywords: Multilayers; Anisotropy; Magneto-optics; Information storage
1. Introduction Magnetic multilayers are a new class of materials with specific novel properties, such as, enhancement of magnetic moment, surface anisotropy, giant magnetostriction, etc. The application potential of these materials gave a big boost to intense research activity in this field. In this paper, we will
* Corresponding author. Tel.: + 33 1 3925 4658; fax: + 33 1 3925 4652; e-maih [email protected].
focus our attention to the anisotropy and the magneto-optical (MO) effects. Indeed, one of the most promising applications of magnetic multilayers (ML) is for the high-density M O information storage in the blue wavelengths and we will also discuss this aspect in this paper. The typical examples are Co/Pt and Co/Pd, which by virtue of their strong uniaxial anisotropy and good magneto-optical effects, have attracted much attention from several research groups [1-3]. Despite several advantages of these systems, over the current rare-earth transition-metal amorphous films, there is a serious
0304-8853/97/$17.00 ~(" 1997 Elsevier Science B.V. All rights reserved PII S 0 3 0 4 - 8 8 5 3 ( 9 7 ) 0 0 1 5 2 - 2
R. Krishnan et al. / Journal of Magnetism and Magnetic Materials 175 (1997) 90- 98
problem for practical applications which is the following. The Curie temperature of these multilayers even for Co layers as thin as 0.5 nm is in the range of 35(~400°C. Thermomagnetic writing at such temperatures not only necessitate high-power laser, but also would cause interdiffusion causing the destruction of the multilayers structure. Therefore, it is necessary to find new class of materials with lower Tc. Multilayers based on (CoxNil x)/Pt instead of pure Co/Pt have been suggested recently by some authors and by us as well [-4 6] which show similar properties as Co/Pt but with a relatively lower Curie temperature. We also found that the surface anisotropy (Ks) in this system varied linearly with the Co concentration which in our opinion is of fundamental importance [5]. We therefore wanted to study, in detail, this aspect and extended our study also to (CoxNil_x)/Au. We chose the system based on Au for the following reasons. Co/Au has been intensely studied by many in the past [7, 8] and it shows perpendicular anisotropy for thin Co layers. We had shown that in Ni/Au multilayers K~ is negative [9] and it is, therefore, interesting to see if the addition of Co would make Ks positive. For the sake of completeness, we will also describe the magneto-optical (MO) Kerr spectra of Fe/Ag and Fe/Pt before discussing those of material of practical interest for MO information storage technology.
91
hess t(Pt, Au) was kept constant at 15 ,~ and that of Co Ni varied in the range 4 30A. The growth parameters will be designated as {t(Co, Ni), t(Pt, Au)} x N, where N stands for the number of bilayers which was in the range 1~25. The top layer was either Pt or Au layer 20 A thick which also served as a protective layer. Fe/Ag and Fe/Pt multilayers were prepared by RF sputtering. The details of the preparation can be found in Ref. [10]. Fe/A~ the multilayer was grown on Ag buffer layer 60 A, thick. We consider here three samples with t(Ag) = 60 ~, and t(Fe) = 110, 36 and 6 A, respectively. The top layer was Ag. Fe/Pt multilayers were grown on a buffer layer of Pt 1 0 0 k thick, t(Pt) was fixed at 15A and t(Fe) = 2.4, 8, 18 and 36 A, respectively. The top layer was 15 ,~ thick Pt. Low- and high-angle X-ray diffraction studies were made to verify the periodic structure and to calculate the layer thicknesses. Magnetization was measured using a vibrating sample magnetometer (VSM) and the anisotropy by a home-made sensitive torque meter in the temperature range 6 300 K. The M H and magneto-optical loops were also taken at the laser wavelength of 633 nm. Magneto-optical polar Kerr spectra were studied at 295 K in the photon energy range 1.5 5.2 eV.
3. Results and discussions 2. Experimental details
3.1. Magnetic anisotropy
The (CoxNi~_x)/Pt,Au multilayers were prepared by sequential evaporation using a dual ebeam under ultra-high-vacuum conditions ( < 5 x 10-9 Torr during deposition). The details of the deposition can be found in Ref. [2]. For depositing the CoxNil x alloy layers, an alloy ingot was used as a source. Subsequently, the composition was determined on a film of CoxNi~-x about 150 m thick. No difference in the composition between the source and the layer could be detected. Both glass and silicon were used as substrates. First, Pt, Au buffer layer 100 A. thick was deposited at room temperature and then annealed at 200°C for 2 h. The multilayers were then grown on this buffer layer at room temperature. The Pt, Au layer thick-
Let us first discuss the results on the anisotropy (CoxNil _x)/Pt,Au. The effective anisotropy Kerr in the multilayers can be expressed by the well-known phenomenological model by the relation, Keff = Kv + 2Ks~t, where, Kv and K~ represent the bulk and surface anisotropies and t the magnetic layer thickness. The term Kv contains the contributions from the demagnetization energy 2rcM 2 and the crystalline (Kcr) and magnetoelastic anisotropy (Kme) energies. Plotting K~ff x t as a function of t yields both Kv and K~. The results were analysed on the basis of the above model. Fig. 1 shows the linear variation of the product Kef f × t as a function of t for the composition Coo.lNio.9 at 295 and 5 K which is in accordance
R. KHshnan et al. / Journal ( f Magnetism and Magnetic Materials 175 (1997) 90 98
92
0.5 0.5.
0.4
"'~&
E © v o')
"'"
0 295K
o
~"~
0.3
.~
0.1
I
.d 0 "7
T=5K
O
to9
-0.2 0.2
]
I
I
i
0.4
0.6
0.8
_
1
Co (x) 0.5
Fig. 3. The variation of K~as a functionof content in Co Ni/Au at5K.
Fig. 1. The v a r i a t i o n of the p r o d u c t K~ff x t as a function o f t at 295 and 5 K.
0.7 - -
i
[
i
0.6 0.5 O
0.4
.9
=
0.3 0.2
T=5K
0.1 0
__
i
I
1
I
0.2
0.4
0.6
0.8
I
1
Co (x) Fig. 2. The variation of K, as a function of Co content in Co Ni/Pt at 5 K.
with model cited above. It can be seen that the values of K e f f increase considerably at 5 K indicating the proximity of Tc to room temperature. Therefore, to avoid such effects we will consider all the results at 5 K. Let us now discuss the Co concentration dependence of K~. Fig. 2 shows the variation of Ks as a function of Co (x) concentration in C o - N i / P t at 5 K. The experimental points show a linear variation of Ks with the Co content. The straight line is calculated based on the relation, assuming one ion contribution to anisotropy, which can be written as, K an°y = KC°(x) + K y i ( 1 - X). Here K c° and K~ ~ stand for the surface anisotropy of Co and Ni
in Co/Pt and Ni/Pt multilayers, respectively. We experimentally determined these values in Co/Pt and Ni/Pt multilayers, as 0.6 and 0.17 erg cm 2, respectively. So taking the above values we calculated the straight line in Fig. 2. The excellent agreement confirms our assumption. Fig. 3 shows a similar result for C o - N i / A u system. It is noteworthy that here, although K, is negative for x = 0, it starts becoming more and more positive with the addition of Co. The increase in K~ could be modelled as before and we can write, K2"°Y= KC°(x) + K~i(1 - x), where, K c° and K~ i are now surface anisotropies found in Co/Au and Ni/Au multilayers. KNi=~ --0.1 e r g c m 2 at 5 K from our earlier work on Ni/Au [9] and K ~ ° = + 0.52 erg cm -2 at 295 K (Tc being very high, this value is expected to be the same at 5 K) from Ref. [8]. The straight line in Fig. 3 calculated using the above values explain the experimental points very well and confirm the validity of the one ion model that we have assumed. These results also show that the contribution from magneto-elastic effects to Ks is negligible if not absent. Because the magnetostriction in Co Ni alloys vary in a complex manner and even changes sign [11], if there is any contribution from this, then one can hardly expect a linear variation. So we are led to conclude that the surface anisotropy in these systems arises mainly from crystal field effects (N6el type) and can be treated as single ion anisotropy. It implies that the modification of the band structure of Ni by the addition of Co and vice versa does not
R. Krishnan et al. / Journal of Magnetism and Magnetic Materials 175 (1997) 90-98
93
®K
H
He = 1.2 kOe Fig. 4. Kerr loop for Coo.7Nio.3/Ptsample (4.5,/15A)x 10.
seem to influence the contribution of the surface atoms to the anisotropy. More theoretical work in this area would probably be welcome.
3.2. M.O. loops In (Co,Nil _x)/Pt,Au systems, with the addition of Co and with decreasing layer thickness, the loops get more and more rectangular and the coercivity also increases• Figs. 4 and 5 show the typical loop for the Coo.3Nio.7/Pt sample (4.5/15 A)x 10 and Coo.TNio.g/Au sample (5.4/15 A) x 16. However, for the same modulation when the Au buffer layer thickness was increased to 300 A, the coercivity more than doubled and increased to 800 Oe. Detailed structural studies using X-ray diffraction also showed that the interface quality improved significantly when the buffer layer thickness was increased from 100 to 300A [12]. So some optimisation is needed to improve further the loop characteristics.
3.3. Magneto-optical studies First, let us describe our results on Fe/Ag. These spectra were taken with a field of 0.45 T and hence, not the saturation values. We have, therefore, used appropriate scaling factors for the calculation of the spectra to allow for this. Fig. 6 shows the Kerr
0.05 ,
llOA 36A
• •
bulk Fe
bO
.......
0.00 °~ °0 ~ 0
-0.05
(D
f
}.,
I" / -0.10
1.5
2.0
2.5
Photon
3.0
3.5
4.0
Energy
4.5
5.0
(eV)
He = 0.45 kOe
Fig. 5. Kerr loop for Coo.~Nio.3/Au sample (5.4/15 ,~) × 16.
Fig. 6. Polar Kerr rotation spectra for three Fe/Ag multilayer samples.
94
R. Krishnan et al. /Journal of Magnetism and Magnetic Materials 175 (1997) 90 98
spectra for the three samples. Here, the bulk Fe data has been scaled down by 0.2. Anomalous line shapes are observed in the region of plasma resonance of Ag near 3.9 eV. The effect is predominant particularly in the thinnest sample with t ( F e ) = 6 A. The observed spectra can be explained by a phenomenological model proposed by one of us [-13]. The model assumes step like profile near the interfaces and assumes the optical constants of the bulk metals. Agreement was best for the thickest sample with t(Fe) = 110 ,~ as shown in Fig. 7 [14]. Fig. 8 shows the result for the sample with t(Fe) = 36 ,~. It is seen that the line shapes are well reproduced but the computed amplitude of rotation is considerably smaller than the experimental one. This is due to some over simplification of the model as it is now recognized that the physical properties of Fe at monolayer thickness range may differ significantly from that of bulk Fe. For the thinnest sample, due to the fact that it saturates at a field of about 1.2 T we used a reduction in the
0,05 i
""
t (Fe) = 36
,_ ", '~ ",
v'* v
0.00
~D "~':~)'~ i -0.05 '~'•"'*"**,~, • Ii.
~" ,~
-0.10 eomp. rotation comp. elliptieity . . . . . . . . . . exp. rotation exp. ellipticity t.5
I
i
~
2.o
~.5
3.o
• •
L
3.~
n
4.o
Energy
I
,.5
r
5.0
(eV)
t ( F e ) -- 1 1 0 i t
0,00 "k..\ "'~x.,, k \
•. , .
....
".. :'.,-,:_,.. ~
T.,~ "*¢
e-
•
): ~
o
o -0.10
eomp. rotation comp. elliptieity . . . . . . . . . . exp. rotation exp. ellipticity n
1.5
~: ?' 2:"
,, .; ;
Fig. 8. Experimental and calculated polar Kerr spectra for the sample Fe/Ag with t(Fe)= 36 A. The calculated spectra are reduced bya factor of 0.2 to account for the lower magnetic field used.
0.05
-0.05
~.* :"
M
Photon
t
...........
i
2.0
2.5
Photon
n
3.0
•
I
3.5
n
4.0
Energy
t
I
4.5
5.0
(eV)
Fig. 7. Experimental and calculated polar Kerr effect spectra for the sample Fe/Ag with t(Fe) ~ 110 A..
off-diagonal tensor element spectra of 0.4 instead of 0.2. A better agreement between the experimental and calculated spectra was obtained for the sample with t(Fe) = 6 A. The result is shown in Fig. 9. It is noteworthy that in this case the ellipticity and rotation spectra are dominated by the features of Ag. Four samples of Fe/Pt were studied where t(Pt) was fixed at 15 A and that of t(Fe) were 2.4, 8, 18 and 36 ]~, respectively. For the thinnest sample perpendicular magnetization could be stabilised and the Kerr rotation is saturated at the applied field of 1.07 T. For the other two samples the rotation obtained has been scaled appropriately to get the saturated value. Fig. 10 shows the polar Kerr rotation spectra of the four samples along with that of the bulk iron. For the thickest sample the spectrum resembles that of Fe but for thinner layers it is transformed. The thinnest sample can be considered as F e - P t alloy and the peak near 4.5 eV arises from the interaction of Pt-Fe.
R. Krishnan et al. /Journal of Magnetism and Magnetic Materials 175 (1997) 90-98 0.1
'
i
'
95
I
'
q~2~, ellipticity t (Fe)
=
'
'
J ~/
~
l
6/~
~z~ 1 . 4 n m HDO0~
0.0
/
....
o
o.o ,
,~
-" ,!"
Ill
, . ~.
I1\
"
O;
% ,,x
J
N -o.2
ID
rotation "%_
_~-%
-0.! ,
eomp. r o t a t i o n comp. ellipticity . . . . . exp. r o t a t i o n exp. ellipticity -0.2
1.5
2.0
2.5
, ;
2
3.0
3.5
4.0
Energy
4.5
5.0
i
Fe/Pt
','~'ax'~.,,•A
o '~
s_
M
(8/15) tz
• •
•
,~'"m"'="====,,
"~
i
- ( 2 . 4 / l a ) 22
.~
( 1B /
z,
15
)
15
~,,• 0 (36/15) 1°
-0.2
""
•
-0.a
-0.4
"="'--
/ ~ F e ,
I
2
,
3
,
I
,
4
p
5
I
3
,
Fig. 11. M O spectra in Coo 3Nio.7/Pt for t = 14 and 4.5 A. Insert: P K R loop for 4.5 A.
(eV)
i
i
-0.1
r
•
Fig. 9. Experimental and calculated polar Kerr effect spectra for the sample Fe/Ag with t(Fe) = 6 A. See text for details of calculation.
,,
0,45
Photon energy in eV
•
Photon
0.0
~
I
4
,
I
5
Photon energy in eV Fig. 10. Polar Kerr rotation in Fe/Pt multilayer samples. The spectrum of bulk Fe is also shown for comparison.
Fig. 11 shows the Kerr spectra in Coo.3Nio.7/Pt for t = 14 and 4.5 ~,. One can notice the large negative peak in the rotation near 4 eV. This peak has been attributed to the effect arising from Co-Pt interactions in Co/Pt and also been observed by us in Ni/Pt [15]. The peak position shifts to lower energy as the Ni concentration increases. It is noteworthy that the rotation of 0.28 ° at 3 eV (wave length of interest for high-density magneto-optical information storage) of the thinner sample is much higher than that of even Co/Pt with similar magnetic layer thickness and, furthermore, the Tc is close to 200°C. All these characteristics make this material a very interesting candidate for application. Figs. 12 and 13 demonstrate the effect of the Co content (x) on the magneto-optical polar Kerr rotation (PKR) and ellipticity (PKE) spectra in Co~Nil_x/Au multilayers for various t(Co-Ni). In the case of thinner magnetic layer (approx. 3 atomic layers) the optical contribution from the interband transition in gold situated near 2.75 eV is clearly manifested especially for Ni/Au. A scaling factor of 6.45 has been employed in order to account for the low Tc of this sample. The addition of Co leads to an increase in both Tc and MO activity of the
R. Krishnan et al. / Journal of Magnetism and Magnetic Materials 175 (1997) 90 98
96 I
o.o5
I
- ~
I
I
ell¢~tiei~y
0.05
I
'
I
I
I
ell£~ic{~Y 0.00
O.OO
.=_ "6
W%.. f l
-0.05
-O.IO
-0.I~
;2
-0.20
*,~.,.~
h
I
,
I
2
,
ro~a~on I
3
,
4
-0.25
'
I
'
I
'
I
~d
-0.2
rotation 2
3
I
~
i
I
i
I
4
Photon energy in eV
I
Fig. 12. M O spectra in C o - N i / A u (6.4/15), x = 0, N = 25 (A) and x = 0.7, N = 16 (@).
I
2
-r.o ~ a t { o r L ,
5
Photon energy in eV
•
I
4
5
Photon energy in eV Fig. 13. M O spectra for x = 0 . 7 , (18/15) x 9 (circles); x = 0.35, (18/15) x l0 (triangles), x = 0, (27/15) (squares).
magnetic layer and as a result the spectral amplitudes are increased. At higher Co Ni layer thicknesses the Au interband transition effect practically disappears from the spectra as seen in Fig. 14,
Fig. 14. Effect of t(Co-Ni) on M O spectra for Co0.vNio.3/Au. Full circles: (6.3/15)x 16; open circles: (9/15)x 12; full triangles: (18/15) x 9.
which also illustrates the effect of t(Co-Ni) on the spectra when both t(Au) and x-parameter are fixed. As expected the negative peak in PKR originating from Au gradually shifts to lower photon energies as t(Co-Ni) increases. This optical effect can be well understood as a reduced radiation penetration to the Au layers situated below magnetic Co-Ni layers. Therefore, one observes more Co-Ni-like behaviour. As expected, the spectra display an overall increase in amplitude with x and approach that of pure Co. It is noteworthy that for the sample with x = 0 . 7 and t(Co N i ) = 1 8 A , the PKR reaches 0.23 ° at 1.8 eV. Recent MO studies of Co/Au interfaces in sandwiches and multilayers provided detailed information on the nature of the Au-Co interfaces [-16, 17]. Thanks to the sharpness of the interfaces the observed spectra in Co/Au and Ni/Au, ultra thin film systems could be satisfactorily simulated using an electromagnetic model employing bulk metal (or thick film) optical data and step-like composition profile [18]. This approach should also be useful in the case of Co-Ni/Au. For this purpose we have studied some CoxNil x alloy films and looked for practical expressions which would account for the effect of the x-parameter on the MO Kerr spectra.
R. Krishnan et al. / Journal of Magnetism and Magnetic Materials 175 (1997) 90-98 0.2
i
i
i
i
rotation o.1 ~o .~
e llip tici ty 0.0
.:'~//
".
./'"
•~
""
j"
". .... /t././,':" ......--.~allo
i%V.,"/
CO
"'~" ..... ""
x
/
-0,4 ..°'°'° --0.6
'
I
2
,
I
3
,
I
4
:
\,
-
t
I
5
Photon energy in eV Fig. 15. PKE spectra of Cox-Ni 1 x alloy thick films. Experimental for x = 0.5 (0, O); calculated curves, x = 0.5 (-); x = 0.7 (-.-).
Starting from the Maxwell G a r n e t t theory (effective m e d i u m model) [19] modified to a c c o u n t for circularly polarized proper modes in M O medium we have obtained the expression for the x-dependence of the off-diagonal permittivity tensor element. Fig. 15 shows the Kerr effect spectra of bulk Co and Ni, that of Coo.sNio.s alloy film and the calculated curves for x = 0.5 and 0.7, respectively. The measured polar Kerr spectra of Coo.sNio.s alloy film and the modelled one show very g o o d agreement. We remind that the model is appropriate for the materials which can be described as a mixture of non-interacting isotropic particles [18]. It can therefore be inferred that the hybridisation of C o and Ni has no effect on the M O properties. Incidentally, this also lends support to the one ion model that we have proposed for interpreting the surface anisotropy.
4. Conclusion
In conclusion, we have described the magentic and magneto-optical properties of some multilayer systems. In Fe/Ag, the M O spectra show the en-
97
h a n c m e n t in the rotation due to plasma resonance in Ag. In C o - N i / P t , A u multilayers prepared by evaporation, some interesting results have been found. The C o concentration dependence of the surface anisotropy can be explained by a single ion model. The AUo system sample with x = 0.7 and t(Co-Ni) = 5.4 A, shows a rectangular perpendicular loop with coercivity as high as 800 Oe and with polar Kerr rotations close to 0.14 ° near 2.8 eV. However, in the Pt system the multilayer with x = 0.3 and t(Co-Ni) = 4.5 A shows a perfect rectangular loop with He = 1.2 kOe, a Kerr rotation of 0.28 ° at 3 eV and a Tc close to 200°C. These characteristics make this material a potential candidate for magneto-optical storage applications in the blue light. In the M O spectra for thin C o - N i layers the interband transition of Au can be seen but for thicker layers the behaviour is C o - N i like. The spectra can be well explained in terms of the model proposed.
Acknowledgements
We would like to thank H. Lassri, M. Seddat, M. Porte and M. Tessier for their contribution to this work.
References
I-1] W.B. Zepper, J.A.M. Greidanus, P.F. Carcia, P.R. Fincher, J. Appl. Phys. 65 (1989) 4971. I-2] R. Krishnan, M. Porte, M. Tessier, IEEE Trans. Magn. 26 (1990) 2727. [3] H. Takahashi, S. Fukatsu, S. Tsunashima, S. Uchiyama, J. Magn. Magn. Mater. 93 (1991) 469. [4] M. Mes, J.C. Lodder, T. Takahata, I. Moritani, N. Imamura, J. Magn. Soc. Japan 17 (Suppl. S1) (1993) 44. [-5] R. Krishnan, H. Lassri, M. Seddat, M. Porte, M. Tessier, Appl. Phys. Lett. 64 (1994) 2312. 1-6] S. Hashimoto, J. Appl. Phys. 75 (1994) 438. [7] C. Chappert, P. Bruno, J. Appl. Phys. 64 (1988) 5736. [8] F.J.A. den Broeder, D. Kuiper, A.P. van de Moesselaer, W. Hoving, Phys. Rev. Let. 60 (1988) 2771. 1-9] R. Krishnan, H. Lassri, M. Porte, M. Tessier, IEEE Trans. Magn. 29 (1993) 3388. [10] R. Krishnan, M. Tessier, J. Appl. Phys. 67 (1990) 5392. [-11] R.M. Bozorth, Ferromagnetism, Van Nostrand, New York, 1951.
98
R. Krishnan et al. / Journal of Magnetism and Magnetic Materials 175 (1997) 90 98
[12] D. Rafaja, M. Chladek, V. Valvoda, M. Seddat, H. Lassri, R. Krishnan, Thin Solid Films 292 (1997) 61. [13] S. Visnovsky, Czech J. Phys. B 36 (1991) 67. [14] R. Krishnan, T. Sikora, S. Visnovsky, J. Magn. Magn. Mater. 118 (1993) 52. [15] R. Krishnan, M. Nyvlt, V. Prosser, M. Seddat, Z. Smetana, M. Tessier, S. Visnovsky, J. Magn. Magn. Mater. 148 (1995) 283.
[16] M. Nyvlt, V. Prosser, S. Visnovsky, J. Ferr+, D. Renard, R. Krishnan, J. Magn. Magn. Mater. 148 (1995) 287. [17] A. Bounouh, P. Beauvillain, C. Chappert, R. Megy, P. Veillet, J. Magn. Magn. Mater. 148 (1995) 273. [18] R. Krishnan, H. Lassri, M. Nyvlt, V. Prosser, D. Rafaja, V. Valvoda, S. Visnovsky, J. Magn. Magn. Mater. 148 (1995) 285. [19] D.E. Aspnes, Spectroscopic Characterization Techniques for Semiconductor Technology, SPIE, Vol. 452, 1983, p. 60.