- Email: [email protected]
Cu multilayers at the first antiferromagnetic maximum
Journal of Magnetism and Magnetic Materials 177-181 (1998) 1183-1185
Journal of magnetism and magnetic materials
ELSEVIER
Discontinuous Co layer in Co/Cu multilayers at the first antiferromagnetic maximum E. Jedryka ~'*, M. Wojcik a, S. Nadolski% D.J. Kubinski b, H. Holloway b aInstitute o/'Physics, Polish Academy of Sciences, AI. Lotnikow 32/46, 02 668 Warszawa, Poland b Ford Research LaboratoFv SRL/MD 3028, P.O. Box 2053, Dearbonz, MI 48121-2053, USA
Abstract S9Co NMR spectroscopy was used to study the structure of [Co(tco)/Cu(9 A)]20 nmltilayers with 4 A ~ tco ~< 13 A. Between 4 and 10 A, Co is found to form discontinuous islands embedded in the alloy-like CoCu mixture, similarly to the structure reported previously for [Co(tco)/Cu(20 A)]2o multilayers. However, the restoring field is significantly different for samples with tcu = 9 A from that in samples with tc~ = 20 A. G~ 1998 Elsevier Science B.V. All rights reserved. Keywords: Multilayers - metallic; NMR-spin echo; Granular systems; 3d metals; Magnetoresistance - multilayers
Island growth of Co on Cu in Co/Cu multilayers has been suggested from the NMR studies [1, 2] as well as from the X-ray (EXAFS and XANES) spectroscopy [3]. In a previously studied series of multilayers with a Cu thickness (tc,) of 20 A, i.e. where the antiferromagnetic coupling between the Co layers reaches its second maximum [4, 5] (further on referred to as the second antiferromagnetic maximum, 2nd AFM) we have shown by the NMR analysis that the initial growth of Co is discontinuous up to the thickness of deposited cobalt (tco) equal to 10 A. The purpose of the present paper is to describe a similar NMR study on samples with a Cu thickness of 9 A, i.e. at the first antiferromagnetic maximum (lst AFM) in order to find out whether the Cu thickness influences the granular growth of Co. A series of samples with t c u = 9 A and 4 A /co ~< i3 A have been prepared by the magnetron sputtering. Details of the sample preparation are described in Refs. [-6, 7]. Sgco NMR experiments have been carried out at 4.2 K using an automated, frequency-swept spinecho spectrometer [8]. The NMR spectra recorded from the whole series of samples under investigation are presented as an inset in Fig. 1. They are practically identical to the NMR spectra recorded from the 2nd AFM
*Corresponding author. Fax: + 48 22 43 09 26; e-maih [email protected].
series [2]. To exemplify this, a superposition of the spectra for samples with tco= 6 A at the 1st and at the 2nd AFM is presented in Fig. 1 in a blown-up scale. The NMR spectra for samples with other Co thicknesses corresponded equally well to those reported for the 2nd AFM series 1-2] and we have applied the same procedure to analyze them. Fig. 2 shows the spectrum decomposition for tco = 6 A. Obviously, since the NMR spectra are almost identical in the two series of samples, so are the
10A00 8Aco
.~
v
~>,
"~
~4
-8 LLI
.
6-
50
1oo 15o 2 Frequency (MHz)
" 2-
. . ~ [ C o 6 A / C u 9A]=o
O"
-~-{Co6AJCu20A]=
'5.
03
50
100 150 Frequency (MHz)
200
Fig. l. 59Co NMR spectra recorded at 4.2 K in [Co (6 A)/Cu (tcu)]2o multilayers for tcu = 9 and tc, = 20 A. Inset: SgCo NMR spectra in [Co (tco)/Cu(9 A)]2o multilayers for 4 A ~
0304-8853/98/$19.00 1998 Elsevier Science B.V. All rights reserved PII S 0 3 0 4 - 8 8 5 3 ( 9 7 ) 0 0 6 3 8 - 0
1184
E. Jedryka et al. / Journal of Magnetism and Magnetic Materials 177-18l (1998) 1t83-1185
2500firstAFM ~
4
2000-
O o
.E 0
0.00 0,05 0.10 0.15 0.20 0.25 0.30
50
100 150 200 Frequency (MHz)
Fig. 2. Experimental NMR spectrum in [Co (6 A)/Cu (9 ~-)]2o multilayer (full circles) and its decomposition into contributions from the particular atomic layers based on a model described in Ref. [2]. Inset shows a schematic representation of the resulting bilayer structure. A dashed line represents the sum of the theoretical subspectra. analytical results which confirm that the structure of both series of samples is the same in the thickness range 4 A ~ tco ~< 10 ~, and that Co exists in the form of separated clusters embedded in the alloy-like CoCu mixture. With growing thickness an increasing amount of Co is found to exist in the separated clusters and at tco= 10 .~ a continuous Co layer with a strained FCC structure is formed. Both series show that the interface does not change appreciably as the Co thickness increases. However, the structure of the continuous Co layer formed for tco > 10 ~, is different for the two series. As we have already shown, the 1st A F M series contains a much higher proportion of HCP stacking than the series at the 2nd A F M [9~. Another difference found between the two series of samples concerns the restoring field, determined from the optimum NMR excitation conditions 1-10]. In magnetic materials the RF field ht applied to induce the N M R signal acts, first of all, on the electronic magnetization which experiences a certain restoring field H~e~ as a result of magnetic anisotropy, coercive field and exchange coupling between the magnetic layers. The oscillations of electronic magnetization due to ht are transferred to the hyperfine field Hlaf, inducing its oscillating transverse component H~.... that is directly responsible for the nuclear transitions H,,.../H.,
= h , / H .....
(1)
The general condition to optimize the NMR signal (i.e. to obtain the highest spin-echo amplitude), is that the nuclear spins experience a turn angle of 17/2 as a result of the RF field acting on them 17/2 = ~Hopt z, where 7 is the gyromagnetic ratio and r the pulse length used in the experiment [1 t"]. In a ferromagnet this optimum condition has to be fulfilled by Ht,~ns = Ho~t, induced by the
Fig. 3. Restoring field determined from the NMR optimum excitation field as a function of the inverse Co layer thickness for the series ofCo/Cu multilayers at the 1st and at the 2nd antiferromagnetic maximum, respectively.
optimum external R F field hlopt, which is directly measured in the experiment. Therefore, in optimum excitation conditions, Eq. (1) determines the value of Hr,st: Hves~ = Hhfhloi, t/Hopt.
(2)
The NMR experiment makes it thus possible to compare the stiffness of magnetic moments (strength of H~es~)between different samples. Such comparison is presented in Fig. 3 for the 1st and the 2nd A F M series. For the 2nd A F M the linear dependence of H~o~, versus 1/tco means that the restoring torque acting on magnetic moment of the Co tayer is constant. This, combined with the fact that the plot extrapolates to zero, implies that the origin of the restoring field is related mainly to the antiferromagnetic coupling which is tCo independent. However, the results for the 1st A F M cannot be explained in this way, considering that the antiferromagnetic coupling is much stronger at the 1st A F M than at the 2nd A F M [4, 5] and one would expect a linear dependence with a slope even larger than in case of the 2nd A F M series. This suggests that in the 1st A F M series there is an easy rotation of the net magnetization under the action of RF fietd with a small amplitude. Several mechanisms can he evoked to account for this, e.g. the presence of pinholes in the Cu separator layer, but our experiment does not give a direct explanation of the particular mechanism. In addition, Fig. 3 suggests that for tco< 10 ,~ in both series of samples the data points depart from the linear behaviour. This would imply that another mechanism contributes to the restoring field in the region where Co layers are discontinuous. The anisotropy of Co in form of small grains is larger than that of the continuous layer, rising the restoring field as observed. In conclusion, we have shown that in Co/Cu multilayers for 4 A ~< tco <~ 10.~ the Co environments are identical at the 1st and 2nd AFM. In this thickness range the Co layers consist of Co islands embedded in a CoCu
E. Jedr3,ka et al, / Journal of Magnetism and Magnetic Materials 177-181 (1998) 1183-1185
a11oy. The restoring field appears to be slightly modified in the region where the Co layer is discontinuous. The origin of the restoring mechanism is found to be different between the ist and the 2nd A F M series. This research has been supported partially by a grant from Ford Motor Company to the N M R group at the Institute of Physics, Polish Academy of Sciences.
References [11 C. Meny, P. Panissod, P. Humbert, J.P. Nozieres, V.S. Speriosu, B.A. Gurney, R. Zehringer, J. Magn. Magn. Mater. 121 (1993) 406. [2"1 E. Jedryka, M. Wojcik, S. Nadolski, D.J. Kubinski, H. HolIoway, J. Magn. Magn. Mater. 165 (1997) 292.
i185
[31 C. Prieto, R. Castaner, J.L. Martinez, A. de Andres, J. Trigo, J.M. Sanz, J. Magn. Magn. Mater. 161 (1996) 31. [4"1 D.H. Mosca, F. Petroff, A. Fert, P.A. Schroeder, W.P. Pratt Jr., E. Loloee, J. Magn. Magn. Mater. 94 (1991) L1. [5"1 S.S.P. Parkin, R. Bhadra, K.P. Roche, Phys. Rev. Lett. 66 (1991) 2152. [61 D.J. Kubinski, H. Holloway, J. Appl. Phys. 79 (1995) 1661. [7"1 D.J. Kubinski, H. Holloway, J. Appl. Phys. 79 (1995) 7395. I-8"1 S. Nadolski, M. Wojcik, E. Jedryka, K. Nesteruk, J. Magn. Magn. Mater. 140-144 (1993) 2187. [91 E. Jedryka, M. Wojcik, S. Nadolski, D.J. Kubinski, H. HolIoway, P. Panissod, J. Appl. Phys. 81 (1997) 4776. 110"1 P. Panissod, C. Meny, M. Wojcik, E. Jedryka, in: J. Tobin (Ed.), Proc. MRS Meeting, San Francisco, 1997, in print. 1-11"1 A. Abragam, The Principles of Nuclear Magnetism, Clarendon Press, Oxford, 1962.
Journal of magnetism and magnetic materials
ELSEVIER
Discontinuous Co layer in Co/Cu multilayers at the first antiferromagnetic maximum E. Jedryka ~'*, M. Wojcik a, S. Nadolski% D.J. Kubinski b, H. Holloway b aInstitute o/'Physics, Polish Academy of Sciences, AI. Lotnikow 32/46, 02 668 Warszawa, Poland b Ford Research LaboratoFv SRL/MD 3028, P.O. Box 2053, Dearbonz, MI 48121-2053, USA
Abstract S9Co NMR spectroscopy was used to study the structure of [Co(tco)/Cu(9 A)]20 nmltilayers with 4 A ~ tco ~< 13 A. Between 4 and 10 A, Co is found to form discontinuous islands embedded in the alloy-like CoCu mixture, similarly to the structure reported previously for [Co(tco)/Cu(20 A)]2o multilayers. However, the restoring field is significantly different for samples with tcu = 9 A from that in samples with tc~ = 20 A. G~ 1998 Elsevier Science B.V. All rights reserved. Keywords: Multilayers - metallic; NMR-spin echo; Granular systems; 3d metals; Magnetoresistance - multilayers
Island growth of Co on Cu in Co/Cu multilayers has been suggested from the NMR studies [1, 2] as well as from the X-ray (EXAFS and XANES) spectroscopy [3]. In a previously studied series of multilayers with a Cu thickness (tc,) of 20 A, i.e. where the antiferromagnetic coupling between the Co layers reaches its second maximum [4, 5] (further on referred to as the second antiferromagnetic maximum, 2nd AFM) we have shown by the NMR analysis that the initial growth of Co is discontinuous up to the thickness of deposited cobalt (tco) equal to 10 A. The purpose of the present paper is to describe a similar NMR study on samples with a Cu thickness of 9 A, i.e. at the first antiferromagnetic maximum (lst AFM) in order to find out whether the Cu thickness influences the granular growth of Co. A series of samples with t c u = 9 A and 4 A /co ~< i3 A have been prepared by the magnetron sputtering. Details of the sample preparation are described in Refs. [-6, 7]. Sgco NMR experiments have been carried out at 4.2 K using an automated, frequency-swept spinecho spectrometer [8]. The NMR spectra recorded from the whole series of samples under investigation are presented as an inset in Fig. 1. They are practically identical to the NMR spectra recorded from the 2nd AFM
*Corresponding author. Fax: + 48 22 43 09 26; e-maih [email protected].
series [2]. To exemplify this, a superposition of the spectra for samples with tco= 6 A at the 1st and at the 2nd AFM is presented in Fig. 1 in a blown-up scale. The NMR spectra for samples with other Co thicknesses corresponded equally well to those reported for the 2nd AFM series 1-2] and we have applied the same procedure to analyze them. Fig. 2 shows the spectrum decomposition for tco = 6 A. Obviously, since the NMR spectra are almost identical in the two series of samples, so are the
10A00 8Aco
.~
v
~>,
"~
~4
-8 LLI
.
6-
50
1oo 15o 2 Frequency (MHz)
" 2-
. . ~ [ C o 6 A / C u 9A]=o
O"
-~-{Co6AJCu20A]=
'5.
03
50
100 150 Frequency (MHz)
200
Fig. l. 59Co NMR spectra recorded at 4.2 K in [Co (6 A)/Cu (tcu)]2o multilayers for tcu = 9 and tc, = 20 A. Inset: SgCo NMR spectra in [Co (tco)/Cu(9 A)]2o multilayers for 4 A ~
0304-8853/98/$19.00 1998 Elsevier Science B.V. All rights reserved PII S 0 3 0 4 - 8 8 5 3 ( 9 7 ) 0 0 6 3 8 - 0
1184
E. Jedryka et al. / Journal of Magnetism and Magnetic Materials 177-18l (1998) 1t83-1185
2500firstAFM ~
4
2000-
O o
.E 0
0.00 0,05 0.10 0.15 0.20 0.25 0.30
50
100 150 200 Frequency (MHz)
Fig. 2. Experimental NMR spectrum in [Co (6 A)/Cu (9 ~-)]2o multilayer (full circles) and its decomposition into contributions from the particular atomic layers based on a model described in Ref. [2]. Inset shows a schematic representation of the resulting bilayer structure. A dashed line represents the sum of the theoretical subspectra. analytical results which confirm that the structure of both series of samples is the same in the thickness range 4 A ~ tco ~< 10 ~, and that Co exists in the form of separated clusters embedded in the alloy-like CoCu mixture. With growing thickness an increasing amount of Co is found to exist in the separated clusters and at tco= 10 .~ a continuous Co layer with a strained FCC structure is formed. Both series show that the interface does not change appreciably as the Co thickness increases. However, the structure of the continuous Co layer formed for tco > 10 ~, is different for the two series. As we have already shown, the 1st A F M series contains a much higher proportion of HCP stacking than the series at the 2nd A F M [9~. Another difference found between the two series of samples concerns the restoring field, determined from the optimum NMR excitation conditions 1-10]. In magnetic materials the RF field ht applied to induce the N M R signal acts, first of all, on the electronic magnetization which experiences a certain restoring field H~e~ as a result of magnetic anisotropy, coercive field and exchange coupling between the magnetic layers. The oscillations of electronic magnetization due to ht are transferred to the hyperfine field Hlaf, inducing its oscillating transverse component H~.... that is directly responsible for the nuclear transitions H,,.../H.,
= h , / H .....
(1)
The general condition to optimize the NMR signal (i.e. to obtain the highest spin-echo amplitude), is that the nuclear spins experience a turn angle of 17/2 as a result of the RF field acting on them 17/2 = ~Hopt z, where 7 is the gyromagnetic ratio and r the pulse length used in the experiment [1 t"]. In a ferromagnet this optimum condition has to be fulfilled by Ht,~ns = Ho~t, induced by the
Fig. 3. Restoring field determined from the NMR optimum excitation field as a function of the inverse Co layer thickness for the series ofCo/Cu multilayers at the 1st and at the 2nd antiferromagnetic maximum, respectively.
optimum external R F field hlopt, which is directly measured in the experiment. Therefore, in optimum excitation conditions, Eq. (1) determines the value of Hr,st: Hves~ = Hhfhloi, t/Hopt.
(2)
The NMR experiment makes it thus possible to compare the stiffness of magnetic moments (strength of H~es~)between different samples. Such comparison is presented in Fig. 3 for the 1st and the 2nd A F M series. For the 2nd A F M the linear dependence of H~o~, versus 1/tco means that the restoring torque acting on magnetic moment of the Co tayer is constant. This, combined with the fact that the plot extrapolates to zero, implies that the origin of the restoring field is related mainly to the antiferromagnetic coupling which is tCo independent. However, the results for the 1st A F M cannot be explained in this way, considering that the antiferromagnetic coupling is much stronger at the 1st A F M than at the 2nd A F M [4, 5] and one would expect a linear dependence with a slope even larger than in case of the 2nd A F M series. This suggests that in the 1st A F M series there is an easy rotation of the net magnetization under the action of RF fietd with a small amplitude. Several mechanisms can he evoked to account for this, e.g. the presence of pinholes in the Cu separator layer, but our experiment does not give a direct explanation of the particular mechanism. In addition, Fig. 3 suggests that for tco< 10 ,~ in both series of samples the data points depart from the linear behaviour. This would imply that another mechanism contributes to the restoring field in the region where Co layers are discontinuous. The anisotropy of Co in form of small grains is larger than that of the continuous layer, rising the restoring field as observed. In conclusion, we have shown that in Co/Cu multilayers for 4 A ~< tco <~ 10.~ the Co environments are identical at the 1st and 2nd AFM. In this thickness range the Co layers consist of Co islands embedded in a CoCu
E. Jedr3,ka et al, / Journal of Magnetism and Magnetic Materials 177-181 (1998) 1183-1185
a11oy. The restoring field appears to be slightly modified in the region where the Co layer is discontinuous. The origin of the restoring mechanism is found to be different between the ist and the 2nd A F M series. This research has been supported partially by a grant from Ford Motor Company to the N M R group at the Institute of Physics, Polish Academy of Sciences.
References [11 C. Meny, P. Panissod, P. Humbert, J.P. Nozieres, V.S. Speriosu, B.A. Gurney, R. Zehringer, J. Magn. Magn. Mater. 121 (1993) 406. [2"1 E. Jedryka, M. Wojcik, S. Nadolski, D.J. Kubinski, H. HolIoway, J. Magn. Magn. Mater. 165 (1997) 292.
i185
[31 C. Prieto, R. Castaner, J.L. Martinez, A. de Andres, J. Trigo, J.M. Sanz, J. Magn. Magn. Mater. 161 (1996) 31. [4"1 D.H. Mosca, F. Petroff, A. Fert, P.A. Schroeder, W.P. Pratt Jr., E. Loloee, J. Magn. Magn. Mater. 94 (1991) L1. [5"1 S.S.P. Parkin, R. Bhadra, K.P. Roche, Phys. Rev. Lett. 66 (1991) 2152. [61 D.J. Kubinski, H. Holloway, J. Appl. Phys. 79 (1995) 1661. [7"1 D.J. Kubinski, H. Holloway, J. Appl. Phys. 79 (1995) 7395. I-8"1 S. Nadolski, M. Wojcik, E. Jedryka, K. Nesteruk, J. Magn. Magn. Mater. 140-144 (1993) 2187. [91 E. Jedryka, M. Wojcik, S. Nadolski, D.J. Kubinski, H. HolIoway, P. Panissod, J. Appl. Phys. 81 (1997) 4776. 110"1 P. Panissod, C. Meny, M. Wojcik, E. Jedryka, in: J. Tobin (Ed.), Proc. MRS Meeting, San Francisco, 1997, in print. 1-11"1 A. Abragam, The Principles of Nuclear Magnetism, Clarendon Press, Oxford, 1962.