Physica B 297 (2001) 180}184

Probing magnetic structures by neutron re#ectometry:O!-specular scattering from interfaces and domains in

FeCoV/TiZr multilayers

R.W.E. van de Kruijs��*, V.A. Ul'yanov�, M.Th. Rekveldt�,H. Fredrikze�, N.K. Pleshanov�, V.M. Pusenkov�, V.G. Syromyatnikov�,

A.F. Schebetov�, S. Langridge�

�Interfacultair Reactor Instituut, Delft University of technology, Mekelweg 15, 2629 JB Delft, The Netherlands�Petersburg Nuclear Physics Institute, Gatchina, 188300, Russia

�Rutherford Appleton Laboratory, Chilton, Didcot, Oxfordshire, OX11 0QX, UK

Abstract

Polarized neutron re#ectometry (PNR) is commonly used to characterize the depth-dependent nuclear density andmagnetization pro"les of thin "lms and multilayers. While specular re#ectivity is connected with characteristicsperpendicular to the sample surface, in-plane features cause o!-specular scattering. In this work we present time-of-#ightPNR results on (FeCoV/TiZr) multilayers with applied "elds parallel to the in-plane easy axis of magnetization. Ourmeasurements at low applied "elds clearly show di!use scattering. We attribute this scattering to randomly orientedmagnetic domains, with no coupling between the magnetic FeCoV layers. With increasing applied "eld the domainscattering decreases. At the same time, o!-specular scattering with a well-de"ned relation between the wavelength andthe scattering angle increases with increasing applied "eld. This type of scattering can be attributed to correlatedinterfaces. � 2001 Elsevier Science B.V. All rights reserved.

PACS: 03.75.Be; 61.12.Ex; 61.12.Ha; 75.25.#z; 75.70.Ak.; 75.70.Cn; 75.70.Kw

Keywords: Polarized neutron re#ectometry; Magnetic multilayers; O!-specular scattering; Interfacial roughness

1. Introduction

Magnetic thin "lms and multilayers are com-monly used in devices for magnetic recording andneutron beam-line optics. To develop and optimizesuch devices, there is an increasing demand to char-

*Corresponding author. Fax: #31-15-278-8303.E-mail address: kruijs@iri.tudelft.nl (R.W.E. van de Kruijs).

acterize the structural and magnetic properties ofthese systems. The depth-dependent characteristicsare often obtained by polarized neutron re#ec-tometry (PNR) because of its unique ability toobtain simultaneously the nuclear density and themagnetization pro"les. Recent advances in beam-line optics, neutron #ux and neutron detectionequipment have made it possible to detect o!-specular scattered neutrons that are linked to per-turbations of densities in the plane of the "lm.

0921-4526/01/$ - see front matter � 2001 Elsevier Science B.V. All rights reserved.PII: S 0 9 2 1 - 4 5 2 6 ( 0 0 ) 0 0 8 3 4 - 6

Fig. 1. Contour map of spin-up (a) and spin-down (b) neutronscattering from a (FeCoV/TiZr) multilayer, with an applied "eldof 5.1 kOe along the easy axis.

Magnetic o!-specular neutron scattering is there-fore useful for receiving information about changesin the magnetic domain structure and about cor-relations in interfacial roughness in magnetic multi-layers [1,2].

We have studied the behavior of 60�(10 nmFe

��Co

��V

�/ x Ti

��Zr

��), magnetron sputtered on

a 5 mm thick glass substrate, with x"10 nm andx"30 nm. (FeCoV/TiZr) multilayers are goodcandidates for neutron polarization devices be-cause of the contrast in density for spin-upand spin-down neutrons [3]. Although actualpolarizing devices are designed with varyinglayer thickness throughout the multilayer, cons-tant layer thickness simplify the interpretationof the PNR results. In this work, we show o!-specular PNR measurements on (FeCoV/TiZr)multilayers and discuss the observed o!-specularscattering.

2. Results and discussion

Previously, we have reported on specular PNRmeasurements on 60�(FeCoV/TiZr) multilayers[4]. It was shown that the multilayers exhibitclear Bragg re#ections resulting from thebilayer repetition. The spin-dependent magneticcontrast between FeCoV and TiZr resulted inhigh intensity Bragg re#ections for one spinstate and vanishing Bragg re#ections for the other.In the current study we used the CRISP re#ec-tometer at ISIS, which has a neutron wavelengthrange of �"0.1 } 0.6 nm for PNR experiments.The combination of a high-brilliance pulsed neu-tron beam and the presence of a one-dimensionalposition sensitive detector rapidly yields a contourmap of scattering angle versus wavelength at a "xedincident beam angle. In this paper, we de"ne thehorizon of the sample at detector angle �"0. Forthe PNR measurements on 60�(30 nm FeCoV/10 nm TiZr) we used an incident angle of���

"!4 mrad. At this incident angle, the avail-able wavelength region includes the total re#ectionregion and the "rst two Bragg re#ections. Allspectra presented in this work have been nor-malized to intensities measured by a monitor in theincident beam.

We performed measurements with applied "eldin-plane along the easy axis, starting with theremnant magnetization after saturating the mag-netization in a negative "eld and then taking dataup to "elds of 5 kOe. From magnetization measure-ments we expect the multilayer to be magneticallysaturated above 500 Oe. Fig. 1a shows the contourmap for spin-up neutrons at an applied "eld of5.1 kOe. The specularly re#ected beam is recogniz-able at �

���"4 mrad. The transmitted beam (to-

gether with that part of the incident beam that doesnot illuminate the sample) is visible at detectorangle �

���"�

��"!4 mrad. O!-specular scatter-

ing is also visible in the form of lines that showa clear relation between �

���and �. Such lines are

the result of scattering from correlated rough

R.W.E. van de Kruijs et al. / Physica B 297 (2001) 180}184 181

Fig. 2. Contour map of spin-up neutron scattering froma (FeCoV/TiZr) multilayer, with an applied "eld of 120 Oe alongthe easy axis.

Fig. 3. Field dependence of spin-up (a) and spin-down (b) neu-tron scattering. All intensities are averaged for 0.4 nm; 0.5 nm.The inset shows the easy axis hysteresis loop.

interfaces, and are characterized by

k����

!k���

"n2�

d, (1)

where k������

"(2�/�)sin������

, n is the order ofthe Bragg re#ection and d is the bilayer thickness.

For a re#ectometer with constant incident angleand variable wavelength, we can write for smallangles �

���K�

��#n�/d. With �

��"constant"

!4 mrad, this translates into the straight linesmarked (1), converging at the incident beam posi-tion �

���"�

��for �P0. In addition to this inter-

facial scattering connected to the incident beam,also visible is a set of o!-specular straight linesmarked (2), that converge at �

���"4 mrad for �P

0. We attribute these lines to o!-specular interfacialscattering caused by the specularly reyected beam.One special feature we would like to note is thatFig. 1a also shows o!-specular scattering below thesample horizon. These lines have a negative slopeand therefore (see Eq. (1)) have negative values of n.

We would like to stress that Eq. (1) is only validfor su$ciently high values of k

������, when refrac-

tion inside the layers is negligible. Corrections forrefraction can be taken into account, but ultimatelyfail at high wavelengths where the wavevector ofthe incident and/or scattered beam can be imagi-nary inside a layer. For such cases, a distortedwave born approximation (DWBA) is more suc-cessful in predicting the o!-specular scattering pat-terns [5,6].

Fig. 1b shows the contour map for spin-downneutrons, again at an applied "eld of 5.1 kOe. Theo!-specular interfacial scattering is much less in-tense than in the case of spin-up neutrons. How-ever, even if the interfacial roughness would be fullynon-magnetic in origin, the o!-specular scatteringwould still be spin-dependent due to the spin-de-pendent neutron density distribution throughoutthe multilayer, as predicted by DWBA calculations.The presence of magnetic roughness may bestudied by a rigorous spin-analysis of the o!-specu-lar scattered neutrons.

The magnetization measurements (see insetFig. 3b) indicate a coercive "eld H

�+120 Oe.

Fig. 2 shows the contour map for spin-up neutronsfor an applied "eld of 120 Oe. Besides the directbeam and the specularly re#ected beam, neutrons

182 R.W.E. van de Kruijs et al. / Physica B 297 (2001) 180}184

are scattered over a wide angular range at allwavelengths. We attribute this to small-angle scat-tering from uncorrelated magnetic domains, withno coupling between magnetic layers throughoutthe multilayer stack. Coupled magnetic domainsinside multilayers are reported to show increasedo!-specular scatting close to the specular Braggconditions where �"(2d�

����)/n [7]. The decrease

in o!-specular intensity observed at the "rst orderBragg re#ection (�+0.3 nm) is explained by thefact that most of the re#ected intensity is comingfrom the top bilayers at that wavelength. Modelcalculations for this system show that for six bi-layers the "rst-order specular Bragg re#ection isalready close to unity. Therefore, the neutron beamdoes not penetrate deeply into the multilayer andencounters less domains and interfaces, causing lesso!-specular scattering. This decrease in o!-specularscattering is also observable in Fig. 1a as gaps inthe lines of o!-specular scattering. For spin-downneutrons the spectrum looks almost identical tothat for spin-up neutrons, con"rming that the aver-age magnetization is close to zero and indeedH+H

�.

The contour map of � versus � can be trans-formed into a contour map of Q

�versus Q

�, with

Q�

and Q�

the momentum transfer, respectively,parallel and perpendicular to the "lm plane.At constant Q

�, the width of the scattering pro"le

as a function of Q�

is inversely retated to thein-plane correlation length [8,9]. This correlationlength, estimated at 5}10 �m, is connected to theaverage magnetic domain size and is con"rmedby neutron depolarization studies on similarsystems [10].

From the specular re#ectivity at ����

"4 mradwe obtain that total re#ection occurs from thesubstrate for �'0.4 nm. Similarly, total re#ectionoccurs from the average multilayer potential for�'0.5 nm. Fig. 3a shows the spin-up intensity asa function of �

���, averaged for 0.4 nm(�(

0.5 nm. In this region, the neutrons are able topenetrate the entire multilayer, before being totallyre#ected from the substrate [4]. The direct andspecular re#ected beam are easily recognizable at����

"!4 mrad and ����

"4 mrad. At low applied"elds the presence of randomly oriented magneticdomains shows as a broad peak in the small-angle

scattering. At higher "elds, the domains grow andthis results in a decrease of this kind of o!-speculardomain scattering. Finally, only interfacial scatter-ing is visible at the highest "elds.

We have already observed (Fig. 1b) that almostno interfacial scattering occurs at high applied"eld for spin-down neutrons. Fig. 3b indeedonly shows a continuous decrease in domainscattering with increasing applied "eld. It is inter-esting to note that the multilayer is clearly notsaturated at 500 Oe: the magnetization in the layersis still not completely aligned and causes o!-specu-lar scattering.

3. Conclusion

We have clearly observed o!-specular scatteringfrom interfaces and magnetic domains in periodic60�(FeCoV/TiZr) multilayers. At high "elds weobserve spin-dependent interfacial scattering,caused by correlated interface roughness. At lower"elds this interfacial scattering is overshadowed byspin-independent scattering from uncorrelated do-mains. The domain scattering is shown to be verysensitive to small deviations from the total align-ment of the domains.

Acknowledgements

The authors wish to thank B.G. Peskov and A.V.Zaitsev for preparation of the samples. This workwas "nancially supported by the Netherlandsorganization for scienti"c research (NWO).

References

[1] M. Takeda, Y. Endoh, A. Kamijo, J. Mizuki, Physica B 248(1998) 14.

[2] V.G. Syromyatnikov, A. Menelle, B.P. Toperverg, Z.N.Soroko, A.F. Schebetov, Physica B 267}268 (1999) 190.

[3] A. Schebetov, N. Pleshanov, V. Syromyatnikov, V. Pusen-kov, B. Peskov, G. Schmelev, Z. Soroko, V. Ul'yanov, J.Phys. Soc. Japan 65 (Suppl. A.) (1996) 195.

[4] R.W.E. van de Kruijs, H. Fredrikze, M.Th. Rekveldt, V.A.Ul'yanov, L.A. Akselrod, V.M. Pusenkov, N.K. Pleshanov,Physica B 267}268 (1999) 181.

R.W.E. van de Kruijs et al. / Physica B 297 (2001) 180}184 183

[5] S.K. Sinha, E.B. Sirota, S. Garo!, H.B. Stanley, Phys. Rev.B 38 (1988) 2297.

[6] R. Pynn, Phys. Rev. B 45 (1992) 602.[7] S. Langridge, J. Schmalian, C.H. Marrows, D.T. Dekad-

jevi, B.J. Hickey, J. Appl. Phys. 87 (2000) 5750.[8] W. Hahn, M. Loewenhaupt, G.P. Felcher, Y.Y. Huang,

S.S.P. Parkin, J. Appl. Phys. 75 (1994) 3564.

[9] J.A. Borchers, J.A. Dura, J. Unguris, D. Tulchinsky,M.H. Kelley, C.F. Majkrzak, Phys. Rev. Lett. 82 (1999)2796.

[10] W.H. Kraan, M.Th. Rekveldt, V.A. Ul'yanov, L.A. Aksel-rod, G.P. Gordeev, V.M. Pusenkov, Physica B 267}268(1999) 75.

184 R.W.E. van de Kruijs et al. / Physica B 297 (2001) 180}184