Nuclear magnetic resonance spin‐echo studies in evaporated Co/Pt multilayers

(1)

Nuclear magnetic resonance spinecho studies in evaporated Co/Pt multilayers

K. Le Dang, P. Veillet, R. Krishman, and H. Lassri

Citation: Applied Physics Letters 61, 994 (1992); doi: 10.1063/1.108471 View online: http://dx.doi.org/10.1063/1.108471

View Table of Contents: http://scitation.aip.org/content/aip/journal/apl/61/8?ver=pdfcov Published by the AIP Publishing

Articles you may be interested in

Morphology of porous media studied by nuclear magnetic resonance line shapes and spin–echo decays J. Chem. Phys. 114, 3258 (2001); 10.1063/1.1333759

Diffusion in porous systems and the influence of pore morphology in pulsed gradient spinecho nuclear magnetic resonance studies

J. Chem. Phys. 97, 651 (1992); 10.1063/1.463979

Magnetization and nuclear magnetic resonance (spinecho) studies on an amorphous CoFeMnBSi alloy J. Appl. Phys. 57, 1394 (1985); 10.1063/1.334494

UHF spinecho spectrometer for the study of impurities in ferromagnetic compounds by nuclear magnetic resonance

Rev. Sci. Instrum. 45, 759 (1974); 10.1063/1.1686730

Effects of Diffusion in Nuclear Magnetic Resonance SpinEcho Experiments J. Chem. Phys. 34, 2057 (1961); 10.1063/1.1731821

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

137.149.200.5 On: Wed, 17 Dec 2014 19:44:56

(2)

Nuclear magnetic resonance spin-echo studies in evaporated Co/Pt multilayers

K. Le Dang and P. Veillet

Institut d’EI&tronique Fondamentale, L&iment 220, 9140.5 Orsay, France R. Krishman and H. Lassri

Laboratoire de Maghisme et Mathiaux Magnktiques, C.N. R.S. 9219.5 Meudon, France (Received 4 March 1992; accepted for publication 8 June 1992)

Multilayered Co/Pt films with Co layer thickness ranging from 10 to 26 A have been studied by the NMR spin-echo method at 2 K. The rapid increase of the signal around 210 MHz as the Co layer thickness is decreased indicates the alloy formation at the interfaces. This signal coincides with the resonance spectrum of CosPt alloy. The latter corresponds to a disordered alloy and is coherent with the shifts of the satellite lines observed in a dilute alloy. The layer thickness of Co forming CosPt alloy at each interface is estimated to be about 4 A.

Co/Pt multilayers are promising candidates for magneto-optic information storage of future generation and hence has attracted the attention of several laborato- ries.i3 Besides the interest for applications these materials also raise some fundamental questions as to the origin of the surface anisotropy. It is well known that Co and Pt readily form an alloy and therefore there is need for some information on the state of the interfaces such as the nature of the alloy formed and its thickness, etc. Among various techniques one can think of, the technique of NMR spin echo is quite interesting.“,’ We have shown the utility of this technique to probe for instance the interfaces in Co/

Mn multilayers. In samples with Co layers thinner than about 35 A, Mn nuclear signal was observed which arises from the Mn atoms diffused into the Co layer.5 Therefore we wanted to use this technique to explore the interfaces in Co/Pt multilayers and describe our results in this letter.

The Co/Pt multilayers were prepared by sequential evaporation under ultrahigh vacuum conditions, the pres- sure during deposition was below 6X lo-’ Torr. The deposition rate and the final thickness were controlled by pre- calibrated quartz monitors. The growth parameters of the samples reported in this letter are designated by ( ta, tr,)“, where tco tpt, are the Co and Pt layer thickness respectively, and N the number of Co layers. The samples (26,32),” ( 14,20),i5 and (10,20) l6 were deposited on glass substrates held at 300 K. The first and the last layers were Pt. No special Pt buffer layer was used. The details of the preparation, the characterization, and the properties have been published by us before.3 Magnetic hysteresis loops were studied in order to characterize the samples. The NMR measurements were carried out at 2 K using a vari- able frequency spin-echo apparatus where the frequency could be swept at desired rate. The samples were mounted in a dewar tail around which the exciting coil was fitted.

The surface area of the sample was about 1 cm’.

The magnetic studies indicated that for the above samples the magnetization was lying in the plane. The Co spin- echo spectra of three samples at 2 K in zero field are shown in Fig. 1. For the thickest Co layer of 26 A, the cusp frequency near 222 MHz is characteristic of the hcp phase.‘?’ As the Co thickness is decreased the signal

around 210 MHz increases strongly. In order to interpret this behavior we prepared two alloys of composition CopsPta and CosPt and studied them. The first alloy was annealed at 600 “C in vacuum for about 4 h in order to reduce the hcp phase and then was used to measure the shift of Co line due to one Pt atom nearest neighbor. The fee and hcp phases are illustrated by the cusps around 216.5 and 224 MHz, respectively, as shown in Fig. 2. The low frequency tail shows two lines near 209 and 200 MHz which are associated with fee Co atoms having one and two Pt nearest neighbors, respectively. These satellite lines occur at regular intervals from the main line as in Co-Cu alloy.7 However, the shift of the first satellite line Sl of about 8 MHz which is much smaller than 18 MHz found in the Co-Cu alloy. This small shift is related to the fact that the saturation magnetization of the Co-Pt alloy de- creases more slowly than from simple Co dilution.’ Fol- lowing Kobayashi et al9 we assume that the Co hyperfine field is due to its own moment r(l, and the moment of each nearest neighbor p,,. Thus the hyperflne field can be written asz9

H,=aps+ b xi%, nearest

neighbors

(1)

where the constants a and b are equal to - 39 kOe/p, and -7.3 kOe/pB, respectively. It is recalled that the Co resonance frequency in MHz is equal to 1.0054 H,, (kOe) .

The Co moment, nearest to an impurity neighbor is deduced from the rate of change d,u/dc of the average magnetic moment with the nonmagnetic impurity concentration. In the case of a simple Co dilution, the Co moment is unagected and the resonance frequency of the first satellite line is expected to decrease by 12.5 MHz. Our present result shows that the shift is only 8 MHz. This result could be explained in two ways as follows. The average magnetic moment in the Co r -,Pt, alloy can be written as p = ( 1 -x) pco+~pp,. If ycO is assumed to have the bulk value of 1.71 ,uLg then we can, using the data of Ref. 8, calculate yr,=O.7 IL& Assuming that Eq. ( 1) is also valid for Pt a decrease in the magnitude of H,, due to one Pt nearest neighbor is simply given by 7.3 X 1.0117.4 kOe. Alternatively if pPt

994 Appl. Phys. Lett 61 (8), 24 August 1992 0003-6951/92/330994-03$03.00 @ 1992 American institute of Physics 994 This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:

137.149.200.5 On: Wed, 17 Dec 2014 19:44:56

(3)

4 2 d 2 .4 i.l -.

4 a .S z a 0 ‘u -8

b

/,A

// *_----___

/

” . . ..,,\,,

‘\ ‘.

/ -‘\ /,/-;-~ /.---..,,~~

,)-pp..L- - L--

190 210 230 F(MHz)

FIG. 1. Co spin-echo spectra at 2 K in Co/Pt multilayer samples (a) (26,32),’ (b) (14,20),” (c) ( 10,20).‘6 The dash-dotted lines repre- sent the contributions from hcp Co and the dotted ones ColPt alloy.

assumed to be zero then the moment of a Co atom nearest neighbor to a Pt atom is deduced to be 1.77 pB, resulting in a decrease in the resonance frequency of the first satellite line of 8.4 MHz. Thus the two limiting cases, namely, ,uPt =0 and ppt =0.7 yB, both lead to values close to 8 MHz observed.

The Co spin-echo spectrum in Co,Pt alloy is centered near 208 MHz which is rather high (Fig. 3). This behavior again reflects the small line shifts due to Pt nearest neighbors. The broad spectrum results from different Co hyperfine fields, depending upon the number it of Pt nearest neighbors. Since the actual shift of the hcp satellite line is not known, we used a slightly different value namely 7 MHz instead of 8 MHz to explain the experimental results.

We assumed a random distribution of atoms. Thus the

190 200 ²¹⁰ ²²⁰ 230 F(MHz)

FIG. 2. Co spin-echo spectra at 77 K in Cog8Ptz alloy. .S, and Sz are satellite lines arising from environments with one and two Pt nearest neighbors, respectively. The signal near 224 M H z is due to residual hcp phase.

.Y- 5 d 2 N” 1 u.

0 -8 s 2 c T:

.f 8 ,a

0

180 ~~ 200 220 F(MHz)

PIG. 3. Co spin-echo spectra at 4.2 K in Co,Pt alloy. The dotted line is the calculated curve arising from individual Gaussian lines corresponding to different Co environments, with a second moment M2:: 144 MHz.’ Fa is the reference frequency for n=O.

statistical weight of Co environments assumed as 12, with II Pt nearest neighbors is given by the relation

P,' 1"! Cn(l--C)l2--n

( 1 (2)

where (;.J is the binomial coefficient and c the Pt concentration. Each resonance line S, corresponding to P, is assumed to be Gaussian and shifted away from the main line

(n=O) by -7n MHz. The calculated curve, shown as dotted line in Fig. 3 is in better agreement with experiments by taking the effective concentration c as 0.23 instead of 0.25. Similar calculations for the CoPt alloy (c

=0.5) show (Fig. 4) that the spectrum intensity is maximum for n = 6 that is at 182 MHz, well below the observed and the calculated cusp frequencies for CosPt.

For such a high concentration of c=O.5, the total width of the spectrum is mainly determined by the statistical weights P,,. The effect of the second moment on the Gauss- ian line is apparent only in the tails as shown in Fig. 4.

The observed Co,Pt spectrum fits reasonably well the low frequency tails (dotted lines) of the spectra of Co/Pt multilayers [Fig. 1). The proportion of Co atoms com-

---

x .%

P 1 - 2 4 I*I

0

140 160 180 200 220 F(MHz)

FIG. 4. Calculated resonance spectra for CoPt alloy assuming random distribution of Co and Pt atoms. The second moments of the individual Gaussian lines are taken to be 144 MHz* (solid line) and 36 MHz*

(dotted line). Fe is the reference frequency for n=O.

995 Appt. Phys. Lett., Vol. 61, No. 8, 24 August 1992 Le Dang et al. 995

137.149.200.5 On: Wed, 17 Dec 2014 19:44:56

(4)

bined with Pt can be obtained by comparing the integrated signal intensity of Co,Pt with that from the sample (Fig.

1). The proportions are then calculated to be (within 30%), 23%: 55%, and 90% for the samples with t,=26, 14, and 10 A, respectively. This means that at each interface, a Co layer of about 4 A is alloyed. It is interesting to mention that by studying the variation of Faraday rotation with t(Co) one of us has evaluated” the interface layer thickness to be about 3 A. When one examines the spectrum corresponding to the inner layers it is seen that they are broadened probably due to some Pt diffusion.

echo study, the nature of the interface in Co/Pt multilayers prepared by evaporation in ultrahigh vacuum. It is shown that the alloy formed at the interfaces due to interdiffusion corresponds to Co,Pt and that its thickness is about 4 A.

The presence of this interfacial alloy should play a role in controlling the anisotropy.

For another sample ( 11,20) 22 deposited on a substrate held at 483 K the easy direction of magnetization is found to be normal to the film plane indicating a strong uniaxial anisotropy. Owing to this strong anisotropy, in this sample only a very small signal could be observed in the range 215-224 MHz. It is recalled here that the enhancement factor occurring in NMR experiments is inversely propor- tional to the magnetic anisotropy field. The high pulse am- plitude needed in this experiment would suggest an effective anisotropy field of about 6 kOe which is comparable to 5 kOe obtained from torque measurements. On the other hand as the hypertine field H, is antiparallel to the magnetization, the demagnetizing field is added to Hr This explains why the center of this spectrum is at a higher frequency than that observed for the sample ( 10,16),16 by about 10 MHz.

Partial support of this work by Brite Euram action Contract No. 0153-C is gratefully acknowledged. The In- stitut d’Electronique Fondamentale is a Unite de Recher- the AssociC au Centre National de la Recherche Scienti- fique, No. 22.

‘W. B. Zeper, F. J. A. Greidanus, P. F. Garcia, and P. R. Fincher, J.

Appl. Phys. 65, 4971 (1989).

‘Y. Ochiai, S. Hashimoto, and K. Aso, Jpn. J. Appl. Phys. 28, L659 (1989).

‘R Krishnan, M. Porte, and M. Tessier, IEEE Trans. Mag. 26, 2727 (i990).

4K. Takanashi, H. Yasuoka, K. Takahashi, N. Hosoito, T. Shinjo, and T.

Takada, J. Phys. Sot. Jpn. 53, 2445 (1984).

‘K. Le Dang, P. Veillet, H. Sakakima, and R. Krishnan, J. Phys. F: Met.

Phys. 16, 93 (1986).

‘C. Ceasri, J. P. Faure, G. Nihoul, ft. Le Dang, P. Veillet, and D.

Renard, J. Magn. Magn. Mater. 78, 296 (1989).

‘K. Le Dang, P. Veillet, Hui He, F. J. Lamelas, C. H. Lee., and R.

Clarke, Phys. Rev. B. 41, 12902 (1990).

*F. Bolzoni, F. Leccabue, R. Panizzieri, and L. Pareti, IEEE Trans.

Mag. MAG-20, 1625 (1984).

9S. Kobayashi, K. Asayama, and J. Itoh, J. Phys. Sot. Jpn. 21, 65 (1966).

In conclusion we have investigated, by the NMR spin- ‘OR. Krishnan, Solid State Commun. 77, 499 (1991).

996 Appl. Phys. Lett., Vol. 61, No. 6, 24 August 1992 Le Dang ef al. 996

137.149.200.5 On: Wed, 17 Dec 2014 19:44:56