cobalt multilayers

Journal of Magnetism and Magnetic Materials 148 (1995) 148-149

Journal of magnetism and magnetic materials

ELSEVIER

Magnetic anisotropy studied with PAC: magnetic hyperfine fields on 111Cd in bcc iron/cobalt multilayers B. Swinnen *, J. Dekoster, G. Langouche, M. Rots Instituut voor Kern en StralingsfyMca, K.U. Leuven, 3000 Leuven, Belgium

Abstract The technique of perturbed angular correlation (PAC) is introduced. It is shown how this technique ca~n contribute to the study of magnetic anisotropy in films and multilayers using its sensitivity to the absolute orientation of the hyperfine field. As an egample, experiments are presented that demonstrate a [110] orientation of the magnetic hyperfine fields of bcc Fe and bcc Co in epitaxial F e / C o multilayers.

1. Principles of PAC and orientation sensitivity Perturbed angular correlation (PAC) is a nuclear technique that allows one to study local field strength and anisotropy as well as structural features in the environment of a certain probe atom. (For a formal description of the method, the reader is referred to Ref. [1].) In our measurements 11l ln(l,lCd) the nuclear probe is incorporated into the sample by ion implantation. We measure, during the lifetime (,,, 100 ns) of a particular nuclear state, the 'Larmor precession' of the nuclear spin orientation as induced by the hyperfine interaction between the nuclear moments and extranuclear electromagnetic fields from the environment of the probe. This technique is in many respects similar to the better known p, SR method. As other nuclear methods PAC also is essentially a 'fingerprint' method identifying through the hyperfine interaction parameters the microscopic surrounding of the nuclear probe. Those parameters are the magnitude, symmetry and erientation o f the magnetic (hyperfine) field or electric field gradient. Essentially a number of lifetime measurements are performed simultaneously with four detectors configured as in Fig. 1. A slight modulation will be superimposed on the exponential decay curves as a consequence of the spin precession of the intermediate state. From these modulated exponential decay spectra the anisotropy function R(t), whicla gives the anisotropy of the radiation as a function of time, is calculated. Tlhe curve obtained is characteristic for the present hyperf'me field (IN-f) d~str~bution. It can be decomposed into a n~mber of harmonic oscillations: each magnetic

* Corresponding author, Fax: + 32-16-2919 59.

field existing at a probe's site is at the origin of a characteristic frequency that eventually may occur in the R ( t ) function together with its second harmonic (quadrupole interactions are omitted for simplicity). The ratio of the amplitudes of the first and second harmonics is determined by the absolute orientation of the hff relative to the detectors. In general, for an arbitrary orientation of the field, both first and second harmonics contribute to the spectrum. The orientation of the fields can be determined by fitting R(t) curves for different sample orientations censistently. Flewever for some geometries the R ( t ) function is simplified so that determination of the orientation is strongly simplified. If the field considered is oriented in tile detector # a r e at 45 ° between two detectors (as in Fig. l(a)), only the first harmonic will be observed in the spectrum. Art orientation of the field perpendicular to the detector plane (Fig. l(b)) will only contribute through the second llarnaonic. Finally when the field points in the direction of one o f the detectors (Fig. 1(c)), the total contribution to the R(t) spectrum will be nearly a constant: only a small amount of second harmonic will be present.

2. Sample preparation and experiments The experiments presented below were performed on an epitaxial F e / C o multilayer of the following form: (001) M g O / F e 50 A°/(Co 24 A / F e 21 ~t)t0. It was grown as

•,

g

b, !]

o, g

g

g

g

Fig. 1. Different orientations of the hypeffine field that give rise to strongly simplified R(t) curves.

0304-8853//95/$09.50 © 1995 Elsevier Science B.V. All rights reserved SSDI 0 3 0 4 , 8 8 5 3 ( 9 5 ) 0 0 1 8 5 - 9

149

B. Swinnen et al. /Journal of Magnetisra and Magnetic Materials 148 (1995) 148-149

R(0 -0.00,

-000 -0.02

....

:'.".

-0.02.

,, ,,'~ ,

'

~, ~

e

¥

5

.0.04 -0.04.

.

,

-0.06 -0.06

-.0.08

-0.08, 0

50

!00

150

200

6

250

Time (tin)

3

4

!

2

I112

AmP

5b 3

30.

4

2oh

z~b

1 4--" I • I

30-

31

20-

16o 1~0 Time (m)

311

20.

|0.

~ , ~

10-

I I I i I I

| I I I

I I O.

0-

6

5~

t~ 15o Frequency (Mite)

26O

0

Frequency (MHz)

Fig. 2. R(t) spectra and Fourier transforms on (001) MgO/Fe 50 A / ( C o 24. A / F e 21 A)t0: (a) Samples normal perpendicular to the detector plane. [110] ages at 45 ° between the detectors. (b) Samples normal and one [110] axis at 45 ° between the detectors. The other equivalent [110] axis is perpendicular to the detector plane (see text for details). described in Ref. [2]. After growth, a room-temperature implantation o f 1014 a t / e r a 2 n l l n was performed at an energy o f 80 keV. To avoid channeling during the implantation, the samples normal was fixed to an angle of 7 ° with respect to the incident ion beam. X-ray diffraction confirmed that the multilayer structure is conserved in this procedure. Room-temperature P A C experiments were done with an equipment equivalent to the one described in Ref.

[3]. In a first experiment the sample was mounted with its normal perpendicular to the detector plane and with the [110] axes at 45 ° between two detectors. The recorded an~sotropy function and a fit o f the data are shown in Fig. 2(a) together with a Fourier transform o f the experimental spectrum. The four probe environments that account for the data have been previously identified [2] and the positions o f the fitted frequencies and their harmonics are indicated in the figure. The contributions o f probe atoms in the pure metal layers are at 89.3(3) M H z for bee Fe (labeled 1 in Fig. 2) and at 38.4(5) M H z for bee Co (3 in Fig. 2). Sites 2 and 4 are due to probe atoms in a Fex_xCo x and Co~_xFe x environment, respectively. For sites 1 and 3 only the first harmonics were fitted, suggesting a [110] orientation of the bee Co and the bee Fe hffs: since the second harmonies are completely absent a perpendicular component o f the hff is excluded. Because in this first experiment all equivalent [110] axes were oriented at 45 ° between two detectors, one cannot decide whether the layers are magnetized in one

[110] direction or whether all equivalent [110] orientations of the hff are present in the sample. Therefore a second experiment is set up in which the samples normal and one [110] axes were oriented at 45 ° between the detectors; the other [110] axis is now perpendicular to the detector plane. If only one [110] orientation is present one should observe either only the first (if the magnetization is as in Fig. l(a)) or only the second harmonic (ff the magnetization is as in Fig. I(b)). If on the contrary the hff lies along all equivalent [110] axes in the plane of the multilayer one expects both harnlonics to be present in the spectrum. The data are shown in Fig. 2(b). First and second harmonic o f site 1 are clearly present. The second harmonic o f the site 3 frequency is observed as a shoulder on the left o f the bee Fe frequency while its first harraonie is obvious. Moreover when starting from this position the sample is turned 90 ° around its normal, the spectrum is not altered. We m a y thus conclude that the bee Co and the bee Fe hffs in the layers considered lie along all equivalent [110] axes in the plane of the sample. References

[1] Th. Wichert and E. Recknagel, in: Microscopic Methods in Metals, ed. U. Gonser (Spinger, Berlin, 1986) p. 317. [2] B. Swinnen, J. Dekcster, G. Langouche and M. Rots, Phys. Rev. B, submitted. [3] A. Bartos, K. Scher~merling, Th. WerLzel and M. Uhrmaeher, Nue]. Instr. and Metlu A 330 (1993) 132.