- Email: [email protected]
XAFS studies of CoCr films
Journal of Magnetism and Magnetic Materials 155 (1996) 222-224
4~
Journalof magnetism 4~H and magnetic materials
ELSEVIER
XAFS studies of Co-Cr films Isuke O u c h i a,*, Ikuo N a k a i a H i r o n o b u M a e d a b Facult)" of Engineering, Tottori Universib', Tottori, Japan 680 b Faculty qf'Science, Okayama UnicersiO,. Okayama. Japan 700
Abstract The XAFS spectra of C o - C r films prepared at different substrate temperatures have been compared. The patterns of XANES and EXAFS of the films with high He1 (1150 Oe) and low He1 (200 Oe) were slightly different. On the assumption that C o - C r films consist of Co- and Cr-rich phases, an EXAFS analysis of nearest-neighbour atoms was made for various possible fractional ratios of these phases; the reliability factor (R-factor) was smaller for the two-phase model than for the one-phase model in the high H~ • films, and vice versa in the low Hc i films.
1. Introduction The microstructure of C o - C r films have attracted interest because of its relevance to magnetic domain structures which restrict the ultimate recording density. Earlier electron microscopy observations of grain structures suggested the segregation of Cr atoms into grain boundaries [1]. Such an interpretation was supported by Auger electron profiles [2], but was disputed by considering magnetostatic and other effects [3]. On the other hand, a model in which C o - C r films consist of Cr- and Co-rich phases has been reinforced by various evidence such as electron microscopic observations of chrysanthemum patterns after chemical etching [4], spin-echo NMR [5], small-angle neutron scattering [6], and atomic probe field ion microscopy [7]. The temperature dependence of the saturation magnetization, M~, of C o - C r films, which deviates from that of the bulk toward higher My values [8], and thermomagnetic analysis of sputtered films [3] were explained by the inhomogeneous structure of the film. The size of these inhomogeneities is considered to be on the order of 10 rim. Several possible mechanisms for this compositional separation have also been discussed [9]. For further confirmation of the model and to clarify local structures on an atomic scale, we have attempted to apply X-ray absorption fine structure (XAFS) techniques to this problem. XAFS would supply information on atomic distance, coordination numbers and positional fluctuations of atoms. A preliminary study did not find affirmative evidence of the existence of Cr layers at the grain boundaries [10].
More recently, we tried to find differences in the XAFS patterns of various C o - C r films together with Co and Cr foils and C o - C r - X (X = Mo, Nb) films [11]. As long as C o - C r films had the hcp structure, Fourier transforms of their EXAFS (extended XAFS) at the K-edge of Co were basically similar to those of the Co foil, and were not so different from those in C o - C r films prepared at various substrate temperatures, although their X-ray absorption near-edge structures (XANES) were somewhat different. The situation was similar for the K-edge of Cr. In the continuation of the above work, in this paper we report EXAFS analyses of C o - C r films with different H c ± with the assumption of a two-phase model. 2. Experimental 2.1. Samples
C o - C r films formed on a polyethylene terephthalate (PET) web by facing target sputtering were obtained from the Thin Film Research Laboratory of Teijin Ltd. The values of the coercive force H~± were 1150 Oe for sample A, 700 Oe for B and 200 Oe for C. It was confirmed beforehand that chrysanthemum patterns were observed in sample A, but not in C, by transmission electron microscopy (TEM) after chemical etching of the film. Some more samples of C o - C r and Co were formed on a PET sheet (7 X 7 cm) by dc magnetron sputtering. C o - C r powder was obtained by filing a C o - C r ingot prepared from 99.9 purity Co and Cr under an Ar atmosphere. The composition of all the samples was Co80Cr20. 2.2. XAFS measurements
Corresponding author. Fax: ouchi @ele.tottori-u.ac.jp.
+ 81-857-31-5664; email:
The XAFS measurements were performed on BL-6B and 7C at the Photon Factory, National Laboratory for
0304-8853/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved SSD10304-8853(95)00641-9
I. Ouchi et al. / Journal of Magnetism and Magnetic Materials 155 (1996) 222-224 High Energy Physics. The sample temperature was changed from 10 K to room temperature. Diffracted light of higher order from the S i ( l l l ) double crystals was eliminated by detuning. 3. Results and discussion
3.1. XANES spectra XANES spectra are sensitive to the surrounding environment of light absorbing atoms; the XANES of a Co foil was distinguished from C o - C r with less than 10% Cr [1 I]. When the film was placed obliquely to the incident beam by as much as 60 °, the spectra of the Co foil and the sputtered Co film retained their shapes, but others changed slightly. The XANES spectra of C o - C r films of high, medium and low H~ • (samples A, B and C, respectively) together with C o - C r powder (sample D) are shown for normal incidence in Fig. I(a) and for oblique incidence in Fig. l(b). The absorption peaks at 7723-7730 eV change slightly from sample A to D. Hence, it is possible that the films with high H~ ± have somewhat different local structures from those with low H~. • .
3.2. EXAFS analysis An oscillatory part of X-ray absorption, EXAFS, is expressed as a function of the wave vector k of the photoemitted electron:
x(k) = E {(UJkR~)ti(k)exp(-2RJh) !
• s i n [ Z k R i + 2 3 , + 0j(k)] exp(-2k2o-/2)},
(1)
where j specifies the type of backscattering atoms, h is the elastic mean free path of the photoelectrons, and 6~ is the phase shift of the photoelectron on the absorbing atoms. Hence, x ( k ) is determined by the radial distance Rj from the absorbing atom, coordination number N i, backscatter-
1.0
~ 0.5
B C D 0.0
ing amplitude t i(k) from the jth atom, phase shift Oi(k) of the jth atom, and by the Debye-Waller t e r m O'i2 which represents the positional fluctuation. The analysis is performed by fitting the calculated k3x(k) from an assumed model onto the measured curve of k3x(k). C o - C r is hcp with a = 2.512 ,~ and c = 4.070 A, if the Cr content is less than 40%; the atomic distance between six nearest neighbours is 2.499 A and between six secondnearest neighboursoiS 2.512 ~,. However, since a difference of less than 0.01 A is generally difficult to distinguish by EXAFS analysis, we treated these 12 atoms as nearest neighbours and attributed the difference of these distances to 0". In this work, we tried to analyze EXAFS data assuming a two-phase model in which the C o - C r films are composed of Co-rich (Co > 80%) and Cr-rich (40% > Cr > 20%) portions. In the actual analyses, we dealt with two cases of Cr concentration in the Cr-rich phase: 30% and 38%. The fractional weight of the Cr-rich phase was varied from 50% to 0.5%; if the Cr-rich phase occupies about 50% of the entire volume, then the Cr atoms in the Co-rich area are mostly depleted for the case of 38% of Cr in the Cr-rich phase. The Cr atom percentage in the Co-rich phase is determined by any given (but less than 50%) percentage of the Cr-rich phase in the C o - C r film. We performed a 'best fit' analysis for the cases in which the Cr-rich phase occupies 50, 30, 20, 10, 5, and 0.5% of the C o - C r film: this was repeated for each sample with two cases (38% and 30% Cr in the Cr-rich phase). In a series of analyses, parameters such as the atomic distance and mean free path were fixed and only the Debye-Waller factor 0-i was freed. For the evaluation of the fitting, we used the reliability factor (R-factor), defined as
E R
=
,
B
0= 0°
7.70
.
.
.
i
7.75 Energy / keV
,
,
" " . (k,,,)-K,,,X . . . . . . .
' . . tK,,, .... H
I
' ,!\
,,. '..'/: .',', -
Co80Cr2o Co K edge
"' ,"
0
=
00"
D "
300 K .
I
C
/
I
'
(b)
A
Co K edge
--J/
x
Ill
~ 0.5
CosoCr2o
Jt '/
{k;:,
m . . Y', k;:,x o b s zt~,,,~
[.0
A ---J////
223
0.11
'
300 K ~
7.70
. '
.
.
.
I
,
,
7.75 Energy /' keV
Fig. 1. XANES spectra of Co-Cr films: sample A (H c ~ = 1150 Oe); sample B (H, ± = 700 Oe): sample C ( Hc • = 200 Oe) and sample D (Co-Cr powder). (a) Normal incidence; (b) oblique incidence (60°).
224
L Ouchi et al. / Journal of Magnetism and Magnetic Materials 155 (1996) 222-224
Table 1 R-factors in the best fits for samples A ( H~ ± = 1150 Oe), B ( H¢ a. = 700 Oe) and C ( Hc ± = 200 Oe) Fraction of Cr-rich phase in Co-Cr 0.5%
5%
10%
20%
30%
50%
0.125 0.11 o 0.09 i
0.11 I
0.10 s 0.11 s 0.097
0.100 0.14 2 0.10 s
0.110 0.13 o
0.10 s 0.154 0.12j
Case 1. Cr concentration in the Cr-rich phase: 30%
Sample A 0.138 Sample B 0.114 Sample C 0.09 o Case 2. Cr concentration in Cr-rich phase: Sample A 0.14 o Sample B 0.110 Sample C 0.090
0.13, O.ll o 38% 0.120 0.091
0.120 0.117 0.095
where m denotes the number of data points. Fittings were made for the Cr K-edge. As starting parameters, we utilized the data obtained from the one-phase analysis made for the Co K-edge: nearest-neighbour distance R = 2.508 ,~, O-co_co = 0.101 A, and o'co cr = 0.055 '~. The resulting R-factors are listed in Table 1. The R-factor is smallest for sample A when the fraction of Cr-rich (30%) phase is 50%, whereas it is smallest for sample C when Cr-rich phase is zero. For sample B, the smallest R-factor is found for no Cr-rich (38%) phase and for less than 20% Cr-rich (30%) phase. Hence, sample A which has H~ ± as high as 1150 Oe and in which chrysanthemum patterns had been observed, fits the two-phase model better than the one-phase model. Sample C ( H c ± = 200 Oe) fits the one-phase model better than the two-phase model. Sample B (/4~ ± = 700 Oe) could have a small fraction of Cr-rich phase. More rigorous examinations may be necessary to confirm the above interpretation because the obtained differences in the R-factors were very small. Nevertheless, we believe that such a possibility was shown in the above. Recently, Kemner et al. reported on the EXAFS of C o - C r films, using normal and 80 ° obliquely incident light [12]. They analyzed the EXAFS, and found that the surplus of Cr concentration above the level of bulk C o - C r was greater in the film plane than normal to it; they suggested the presence of a thin Cr-rich region along the film plane between Co-rich platelets. They seemed to have analyzed only one kind of film; their stoichiometric data were not necessarily definite. On the other hand, we inclined the film only 60 °, and found small differences in XANES, but not in EXAFS. Hence, a detailed comparison may not be appropriate at this stage.
4. Conclusions We used C o - C r films with different H c± and found slight differences among them. From the EXAFS analysis,
0.099
a C o - C r film with H c ± = 1150 Oe could consist of 50% Co-rich and 50% Cr-rich (30% Cr) phases.
Acknowledgements The authors are indebted to Dr M. Hirasaka and other researchers at Teijin Limited for the series of C o - C r samples with magnetization and TEM characterization. The EXAFS measurements were performed on BL-6B and 7C at the Photon Factory of the National Laboratory for High Energy Physics under project No. 93-G145.
References [1] K. Ouchi and S. Iwasaki, IEEE Trans. Magn. 18 (1982) I110. [2] R. Sugita, IEEE Trans. Magn. 20 (1984) 687. [3] J.E. Snyder and M.H. Kryder, J. Appl. Phys. 73 (1993) 5551, [4] Y. Maeda, S. Hirono and M. Akashi, Jpn. J. Appl. Phys. 24 (1985) L 951. [5] K. Yoshida, H. Kakibayashi and H. Yasuoka, J. Appl. Phys. 68 (1990) 705. [6] K. Takei, J. Suzuki, Y. Maeda and S. Funahashi, Jpn. J. Appl. Phys. 32 (1993) 2265. [7] K. Hono, Y. Maeda, J.-L. Li and T. Sakurai, J. Magn. Magn. Mater. 110 (1992) L 254. [8] M. Sagoi, R. Nishikawa and T. Suzuki, IEEE Trans. Magn. 22 (1986) 1335. [9] Y. Maeda, D.J. Rogers and K. Takei, J. Magn. Soc. Jpn. 18 (1994) Supl. 1, 27. [10] I. Ouchi, I. Nakai and H. Maeda, Proc. 2nd Int. Symp. Phys. Magn. Mater. 1 (1992) 339. [11] I. Ouchi, I. Nakai and H. Maeda, Proc. Magn, Soc. Jpn. 18 (1994) Suppl. 23. [12] K.M, Kemner, V.G. Harris, W.T. Elam and C.L. Lodder, IEEE Trans. Magn. 30 (1994) 4017.
4~
Journalof magnetism 4~H and magnetic materials
ELSEVIER
XAFS studies of Co-Cr films Isuke O u c h i a,*, Ikuo N a k a i a H i r o n o b u M a e d a b Facult)" of Engineering, Tottori Universib', Tottori, Japan 680 b Faculty qf'Science, Okayama UnicersiO,. Okayama. Japan 700
Abstract The XAFS spectra of C o - C r films prepared at different substrate temperatures have been compared. The patterns of XANES and EXAFS of the films with high He1 (1150 Oe) and low He1 (200 Oe) were slightly different. On the assumption that C o - C r films consist of Co- and Cr-rich phases, an EXAFS analysis of nearest-neighbour atoms was made for various possible fractional ratios of these phases; the reliability factor (R-factor) was smaller for the two-phase model than for the one-phase model in the high H~ • films, and vice versa in the low Hc i films.
1. Introduction The microstructure of C o - C r films have attracted interest because of its relevance to magnetic domain structures which restrict the ultimate recording density. Earlier electron microscopy observations of grain structures suggested the segregation of Cr atoms into grain boundaries [1]. Such an interpretation was supported by Auger electron profiles [2], but was disputed by considering magnetostatic and other effects [3]. On the other hand, a model in which C o - C r films consist of Cr- and Co-rich phases has been reinforced by various evidence such as electron microscopic observations of chrysanthemum patterns after chemical etching [4], spin-echo NMR [5], small-angle neutron scattering [6], and atomic probe field ion microscopy [7]. The temperature dependence of the saturation magnetization, M~, of C o - C r films, which deviates from that of the bulk toward higher My values [8], and thermomagnetic analysis of sputtered films [3] were explained by the inhomogeneous structure of the film. The size of these inhomogeneities is considered to be on the order of 10 rim. Several possible mechanisms for this compositional separation have also been discussed [9]. For further confirmation of the model and to clarify local structures on an atomic scale, we have attempted to apply X-ray absorption fine structure (XAFS) techniques to this problem. XAFS would supply information on atomic distance, coordination numbers and positional fluctuations of atoms. A preliminary study did not find affirmative evidence of the existence of Cr layers at the grain boundaries [10].
More recently, we tried to find differences in the XAFS patterns of various C o - C r films together with Co and Cr foils and C o - C r - X (X = Mo, Nb) films [11]. As long as C o - C r films had the hcp structure, Fourier transforms of their EXAFS (extended XAFS) at the K-edge of Co were basically similar to those of the Co foil, and were not so different from those in C o - C r films prepared at various substrate temperatures, although their X-ray absorption near-edge structures (XANES) were somewhat different. The situation was similar for the K-edge of Cr. In the continuation of the above work, in this paper we report EXAFS analyses of C o - C r films with different H c ± with the assumption of a two-phase model. 2. Experimental 2.1. Samples
C o - C r films formed on a polyethylene terephthalate (PET) web by facing target sputtering were obtained from the Thin Film Research Laboratory of Teijin Ltd. The values of the coercive force H~± were 1150 Oe for sample A, 700 Oe for B and 200 Oe for C. It was confirmed beforehand that chrysanthemum patterns were observed in sample A, but not in C, by transmission electron microscopy (TEM) after chemical etching of the film. Some more samples of C o - C r and Co were formed on a PET sheet (7 X 7 cm) by dc magnetron sputtering. C o - C r powder was obtained by filing a C o - C r ingot prepared from 99.9 purity Co and Cr under an Ar atmosphere. The composition of all the samples was Co80Cr20. 2.2. XAFS measurements
Corresponding author. Fax: ouchi @ele.tottori-u.ac.jp.
+ 81-857-31-5664; email:
The XAFS measurements were performed on BL-6B and 7C at the Photon Factory, National Laboratory for
0304-8853/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved SSD10304-8853(95)00641-9
I. Ouchi et al. / Journal of Magnetism and Magnetic Materials 155 (1996) 222-224 High Energy Physics. The sample temperature was changed from 10 K to room temperature. Diffracted light of higher order from the S i ( l l l ) double crystals was eliminated by detuning. 3. Results and discussion
3.1. XANES spectra XANES spectra are sensitive to the surrounding environment of light absorbing atoms; the XANES of a Co foil was distinguished from C o - C r with less than 10% Cr [1 I]. When the film was placed obliquely to the incident beam by as much as 60 °, the spectra of the Co foil and the sputtered Co film retained their shapes, but others changed slightly. The XANES spectra of C o - C r films of high, medium and low H~ • (samples A, B and C, respectively) together with C o - C r powder (sample D) are shown for normal incidence in Fig. I(a) and for oblique incidence in Fig. l(b). The absorption peaks at 7723-7730 eV change slightly from sample A to D. Hence, it is possible that the films with high H~ ± have somewhat different local structures from those with low H~. • .
3.2. EXAFS analysis An oscillatory part of X-ray absorption, EXAFS, is expressed as a function of the wave vector k of the photoemitted electron:
x(k) = E {(UJkR~)ti(k)exp(-2RJh) !
• s i n [ Z k R i + 2 3 , + 0j(k)] exp(-2k2o-/2)},
(1)
where j specifies the type of backscattering atoms, h is the elastic mean free path of the photoelectrons, and 6~ is the phase shift of the photoelectron on the absorbing atoms. Hence, x ( k ) is determined by the radial distance Rj from the absorbing atom, coordination number N i, backscatter-
1.0
~ 0.5
B C D 0.0
ing amplitude t i(k) from the jth atom, phase shift Oi(k) of the jth atom, and by the Debye-Waller t e r m O'i2 which represents the positional fluctuation. The analysis is performed by fitting the calculated k3x(k) from an assumed model onto the measured curve of k3x(k). C o - C r is hcp with a = 2.512 ,~ and c = 4.070 A, if the Cr content is less than 40%; the atomic distance between six nearest neighbours is 2.499 A and between six secondnearest neighboursoiS 2.512 ~,. However, since a difference of less than 0.01 A is generally difficult to distinguish by EXAFS analysis, we treated these 12 atoms as nearest neighbours and attributed the difference of these distances to 0". In this work, we tried to analyze EXAFS data assuming a two-phase model in which the C o - C r films are composed of Co-rich (Co > 80%) and Cr-rich (40% > Cr > 20%) portions. In the actual analyses, we dealt with two cases of Cr concentration in the Cr-rich phase: 30% and 38%. The fractional weight of the Cr-rich phase was varied from 50% to 0.5%; if the Cr-rich phase occupies about 50% of the entire volume, then the Cr atoms in the Co-rich area are mostly depleted for the case of 38% of Cr in the Cr-rich phase. The Cr atom percentage in the Co-rich phase is determined by any given (but less than 50%) percentage of the Cr-rich phase in the C o - C r film. We performed a 'best fit' analysis for the cases in which the Cr-rich phase occupies 50, 30, 20, 10, 5, and 0.5% of the C o - C r film: this was repeated for each sample with two cases (38% and 30% Cr in the Cr-rich phase). In a series of analyses, parameters such as the atomic distance and mean free path were fixed and only the Debye-Waller factor 0-i was freed. For the evaluation of the fitting, we used the reliability factor (R-factor), defined as
E R
=
,
B
0= 0°
7.70
.
.
.
i
7.75 Energy / keV
,
,
" " . (k,,,)-K,,,X . . . . . . .
' . . tK,,, .... H
I
' ,!\
,,. '..'/: .',', -
Co80Cr2o Co K edge
"' ,"
0
=
00"
D "
300 K .
I
C
/
I
'
(b)
A
Co K edge
--J/
x
Ill
~ 0.5
CosoCr2o
Jt '/
{k;:,
m . . Y', k;:,x o b s zt~,,,~
[.0
A ---J////
223
0.11
'
300 K ~
7.70
. '
.
.
.
I
,
,
7.75 Energy /' keV
Fig. 1. XANES spectra of Co-Cr films: sample A (H c ~ = 1150 Oe); sample B (H, ± = 700 Oe): sample C ( Hc • = 200 Oe) and sample D (Co-Cr powder). (a) Normal incidence; (b) oblique incidence (60°).
224
L Ouchi et al. / Journal of Magnetism and Magnetic Materials 155 (1996) 222-224
Table 1 R-factors in the best fits for samples A ( H~ ± = 1150 Oe), B ( H¢ a. = 700 Oe) and C ( Hc ± = 200 Oe) Fraction of Cr-rich phase in Co-Cr 0.5%
5%
10%
20%
30%
50%
0.125 0.11 o 0.09 i
0.11 I
0.10 s 0.11 s 0.097
0.100 0.14 2 0.10 s
0.110 0.13 o
0.10 s 0.154 0.12j
Case 1. Cr concentration in the Cr-rich phase: 30%
Sample A 0.138 Sample B 0.114 Sample C 0.09 o Case 2. Cr concentration in Cr-rich phase: Sample A 0.14 o Sample B 0.110 Sample C 0.090
0.13, O.ll o 38% 0.120 0.091
0.120 0.117 0.095
where m denotes the number of data points. Fittings were made for the Cr K-edge. As starting parameters, we utilized the data obtained from the one-phase analysis made for the Co K-edge: nearest-neighbour distance R = 2.508 ,~, O-co_co = 0.101 A, and o'co cr = 0.055 '~. The resulting R-factors are listed in Table 1. The R-factor is smallest for sample A when the fraction of Cr-rich (30%) phase is 50%, whereas it is smallest for sample C when Cr-rich phase is zero. For sample B, the smallest R-factor is found for no Cr-rich (38%) phase and for less than 20% Cr-rich (30%) phase. Hence, sample A which has H~ ± as high as 1150 Oe and in which chrysanthemum patterns had been observed, fits the two-phase model better than the one-phase model. Sample C ( H c ± = 200 Oe) fits the one-phase model better than the two-phase model. Sample B (/4~ ± = 700 Oe) could have a small fraction of Cr-rich phase. More rigorous examinations may be necessary to confirm the above interpretation because the obtained differences in the R-factors were very small. Nevertheless, we believe that such a possibility was shown in the above. Recently, Kemner et al. reported on the EXAFS of C o - C r films, using normal and 80 ° obliquely incident light [12]. They analyzed the EXAFS, and found that the surplus of Cr concentration above the level of bulk C o - C r was greater in the film plane than normal to it; they suggested the presence of a thin Cr-rich region along the film plane between Co-rich platelets. They seemed to have analyzed only one kind of film; their stoichiometric data were not necessarily definite. On the other hand, we inclined the film only 60 °, and found small differences in XANES, but not in EXAFS. Hence, a detailed comparison may not be appropriate at this stage.
4. Conclusions We used C o - C r films with different H c± and found slight differences among them. From the EXAFS analysis,
0.099
a C o - C r film with H c ± = 1150 Oe could consist of 50% Co-rich and 50% Cr-rich (30% Cr) phases.
Acknowledgements The authors are indebted to Dr M. Hirasaka and other researchers at Teijin Limited for the series of C o - C r samples with magnetization and TEM characterization. The EXAFS measurements were performed on BL-6B and 7C at the Photon Factory of the National Laboratory for High Energy Physics under project No. 93-G145.
References [1] K. Ouchi and S. Iwasaki, IEEE Trans. Magn. 18 (1982) I110. [2] R. Sugita, IEEE Trans. Magn. 20 (1984) 687. [3] J.E. Snyder and M.H. Kryder, J. Appl. Phys. 73 (1993) 5551, [4] Y. Maeda, S. Hirono and M. Akashi, Jpn. J. Appl. Phys. 24 (1985) L 951. [5] K. Yoshida, H. Kakibayashi and H. Yasuoka, J. Appl. Phys. 68 (1990) 705. [6] K. Takei, J. Suzuki, Y. Maeda and S. Funahashi, Jpn. J. Appl. Phys. 32 (1993) 2265. [7] K. Hono, Y. Maeda, J.-L. Li and T. Sakurai, J. Magn. Magn. Mater. 110 (1992) L 254. [8] M. Sagoi, R. Nishikawa and T. Suzuki, IEEE Trans. Magn. 22 (1986) 1335. [9] Y. Maeda, D.J. Rogers and K. Takei, J. Magn. Soc. Jpn. 18 (1994) Supl. 1, 27. [10] I. Ouchi, I. Nakai and H. Maeda, Proc. 2nd Int. Symp. Phys. Magn. Mater. 1 (1992) 339. [11] I. Ouchi, I. Nakai and H. Maeda, Proc. Magn, Soc. Jpn. 18 (1994) Suppl. 23. [12] K.M, Kemner, V.G. Harris, W.T. Elam and C.L. Lodder, IEEE Trans. Magn. 30 (1994) 4017.