JOURNAL OF APPLIED PHYSICS VOLUME 87, NUMBER 9 1 MAY 2000 Neutron scattering on magnetic thin films: Pushing the limits invited... A. Schreyer,a) T. Schmitte, R. Siebrecht, P. Bo¨deker, and H. Zabel Institut fu¨r Experimentalphysik/Festko¨rperphysik, Ruhr-Universita¨t Bochum, D-44780 Bochum, Germany S. H. Lee, R. W. Erwin, and C. F. Majkrzak National Institute of Standards and Technology, Gaithersburg, Maryland 20899 J. Kwo and M. Hong Lucent Technologies, Bell Laboratories, Murray Hill, New Jersey 07974 Neutron scattering has been the scattering technique of choice for the analysis of magnetic structures and their dynamics for many decades. The advent of magnetic thin film systems has posed new challenges since such samples have inherently small scattering volumes. By way of examples, recent progress in the application of neutron scattering for the study of both magnetic structure and dynamics in magnetic thin film systems will be presented. First, a combined high angle neutron scattering and polarized neutron reflectivity investigation of the magnetic order of Cr and its influence on the exchange coupling between the Fe layers in Fe/Cr superlattices is discussed. It is shown that in the whole thickness range up to 3000 Å, the magnetic structure is governed by frustration effects at the Fe/Cr interfaces. Second, it is demonstrated that it is now possible to investigate the dynamic properties of magnetic thin films with neutron scattering. Unlike, e.g., Brillouin light scattering, inelastic neutron scattering provides access to large portions of the Brillouin zone. First results on spin wave excitations in a Dy/Y superlattice are presented. © 2000 American Institute of Physics. S0021-8979 00 92908-6 I. INTRODUCTION France have been built, which can provide access to small and high Q in one experiment. Neutron scattering has been the scattering technique of The use of strong dynamical scattering effects, like the choice for the analysis of magnetic structures and their dy- strong reflectivity increase towards total reflectivity at very namics for many decades. Due to advances in preparation small Q Fresnel reflectivity , can effectively increase the techniques in the last decade, research in magnetism has scattering signal in PNR. At high scattering angles, no such moved more and more towards the investigation of layered effects occur. The scattering is simply proportional to the magnetic thin film structures. A wealth of effects has been number of scattering atoms. Thus, if information on mag- discovered which modify the magnetic structure, the dynam- netic order on atomic length scales is required, even more ics, and the transport properties in magnetic thin films com- attention must be given to the use of samples with sufficient pared to bulk.1 This development has posed new challenges scattering volume. for neutron scattering since such samples have inherently small scattering volumes. In this article, we will outline a II. ELASTIC SCATTERING few examples of recent progress which has been made in the application of neutron scattering techniques for research in The magnetic structure of Cr in thin films and superlat- magnetic thin film structures. tices is such a problem. Cr exhibits a complicated spin den- Polarized neutron reflectometry PNR clearly is a very sity wave magnetic structure in bulk.3 Historically, epitaxial important technique in the field. However, since it is covered Fe/Cr structures have played a key role in thin film magne- by Gian Felcher in this volume, we will focus on comple- tism. The oscillatory exchange coupling, the giant magne- mentary recent developments. PNR is performed at small toresistance effect and the noncollinear coupling have been scattering vectors Q 4 / sin up to 0.4 Å 1. Here, is discovered in Fe/Cr.4 Fe/Cr differs from most other systems, the neutron wavelength and is the scattering angle. This which also show these properties, due to the inherent Cr scattering vector range is well suited to investigate magneti- antiferromagnetism. Thus, the question is how any magnetic zation profiles of thin film structures which may exhibit pe- order of the Cr correlates with the exchange coupling be- riodicities on the order of tens of Å. However, PNR does not tween the Fe layers. In the following, we review a recent provide information on the magnetic order on the atomic combined high angle neutron scattering and PNR investiga- tion of the magnetic order of Cr and its influence on the scale. To obtain such information, scattering vectors on the exchange coupling between the Fe layers. It will be shown order of a few Å 1 are required, i.e., classical high angle that the details of the magnetic Cr structure and the exchange scattering techniques must be used. Recently, instruments coupling are governed by frustration at the Fe/Cr interfaces such as ADAM2 at the Institut Laue Langevin in Grenoble/ over the whole thickness range investigated. Since Cr has only a small magnetic moment of about a Electronic mail: andreas.schreyer@uni-bochum.de 0.55 B , the magnetic neutron scattering cross section is 0021-8979/2000/87(9)/5443/6/$17.00 5443 © 2000 American Institute of Physics Downloaded 27 Mar 2001 to 148.6.178.13. Redistribution subject to AIP copyright, see http://ojps.aip.org/japo/japcr.jsp 5444 J. Appl. Phys., Vol. 87, No. 9, 1 May 2000 Schreyer et al. quite weak. This requires special efforts to maximize the scattering volume for high angle neutron experiments. We have grown a series of Fe/Cr samples with a fixed Fe thick- ness of 20 Å and the Cr thickness varying between 10 and 3000 Å by molecular-beam epitaxy MBE methods on a Nb buffer using large 2 in. 2 in. sapphire substrates.6­8 To ob- tain sufficient scattering volume, Fe/Cr superlattice samples with up to 250 bilayer repeats were used for the smallest Cr thicknesses tCr . For tCr 1000 Å, a single Cr film between the Fe layers was sufficient. Bulk Cr is antiferromagnetic, i.e., the moments at the center of the Cr bcc structure are aligned opposite to those on the corners. This simple structure is called a commensurate spin density wave CSDW . Superimposed on this structure is a sinusoidal modulation of the magnitude of the antiferro- magnetically aligned Cr moments with a period FIG. 1. Neutron scans along L through 001 as a function of temperature T ISDW . Since for a sample of composition Cr80 /Fe19 133 . The scattering vector Q is ISDW is incommensurate with the Cr lattice this mag- plotted in reciprocal lattice units r.l.u. and denoted according to the corre- netic structure is classified as an incommensurate spin den- sponding axis of the reciprocal lattice. For the 250 K data, the distance of sity wave ISDW . ISDW is about 21 lattice constants at T the two fitted broad peaks from the central peak corresponds to Cr,AF 0 and increases with temperature.3 The incommensurabil- 2 SL Fe,NC . The inset depicts the model obtained from the fitted data. ity is ascribed to a nesting vector along the 100 directions of the Cr Fermi surface. The wave vector q defines the di- rection of propagation of the ISDW. At lowest temperatures This complicated magnetic structure mediates a noncollinear a longitudinal ISDW LSDW forms, i.e., is parallel to q. coupling between the Fe layers, as is shown by PNR. A Above the spin-flip transition temperature, TSF 123 K, is schematic picture is provided in the inset of Fig. 1. A lateral perpendicular to q, forming a transverse ISDW TSDW . Cr thickness variation on average leads to a lateral coexist- Above TN 311 K bulk Cr is paramagnetic. ence of even and uneven number of Cr layers. The coupling The incommensurate modulation of the antiferromag- of the Cr to the Fe is assumed to be the same at all top and netic AF spin structure by the ISDW causes two satellite bottom Fe/Cr interfaces, respectively. Then, partial spirals of peaks to occur around the 1,0,0 positions,3 e.g., at (0,0,1 opposing sense of rotation are the topologically simplest way ), (1 ,0,0), and (0,1 ,0) which can be investigated to accommodate the Cr between the noncollinear Fe layers. by neutron scattering. Here, 1 q with q in reciprocal Since the Cr tends to be a collinear antiferromagnet, each of lattice units, a/ ISDW , and the Cr lattice constant a. In the spirals applies a torque. However, due to the opposing the first case, the position of the satellites indicates q being sense of rotation of the laterally coexisting spirals, the torque oriented along the out of the film plane L axis, the two latter is frustrated. Neither can an even number of Cr layers rotate cases occur for either direction of in-plane propagation along the Fe layers antiparallel, nor can an uneven number of Cr H or K. In addition, the polarization of the ISDW i.e., layers rotate the Fe layers parallel. Instead, the Fe layers TSDW or LSDW can be obtained by making use of the choose a compromise. They align noncollinearly at roughly selection rules for magnetic neutron scattering. It requires a 90°. This spiral magnetic structure had originally been pro- component of the magnetization vector to be perpendicu- posed by Slonczewski within his proximity magnetism lar to the scattering vector Q . Thus, a LSDW propagating model.9 Above the Ne´el temperature of this magnetic Cr or- along L, i.e., out-of-plane, will generate no intensity at der, the frustration vanishes. Consequently, the noncollinear (0,0,1 ). However, satellites will occur at (1,0,0 ). On order also vanishes.6 the other hand, a TSDW propagating along L, will produce It is important to point out that so far we have consid- peaks at (0,0,1 ), whereas no intensity will occur at ered a case where tCr ISDW . The observation of a com- (1,0,0 ). A CSDW phase, on the other hand, will yield a mensurate SDW instead of an ISDW can be attributed to the single peak of purely magnetic origin exactly at the Cr 001 fact that the Cr is too thin to support an ISDW. This will positions due to the commensurability with the Cr lattice. obviously change at larger Cr thicknesses. In Fig. 1, neutron Thus, with neutron scattering we can uniquely distinguish 00L scans through 001 are shown for a Cr80 /Fe19 133 CSDW and ISDW magnetic order as well as ISDW propa- sample as a function of temperature. Unlike at tCr 42 Å, 6 gation and polarization. For a more detailed discussion, see, the form of the spectra changes drastically as a function of e.g., Ref. 8. temperature. For T 20 K, a double peak structure around 001 is found, as is typical for a TSDW propagating along A. Thin Cr films the film growth direction (L). In the present case, tCr is on First, we briefly review the results for small tCr , where the order of ISDW , i.e., one ISDW period fits into each Cr the Cr mediates an exchange coupling between the Fe layer along the growth axis. Thus, it is plausible that an layers.4 For tCr 42 Å, Cr is found to exhibit CSDW order. ISDW can form. The ISDW period of ISDW 79 Å deter- Superimposed on this magnetic order we find a spiral modu- mined from the peak positions at (0,0,1 ) is somewhat lation which is induced by lateral Cr thickness variations.6 enhanced compared to bulk values.8 Also, weak second- Downloaded 27 Mar 2001 to 148.6.178.13. Redistribution subject to AIP copyright, see http://ojps.aip.org/japo/japcr.jsp J. Appl. Phys., Vol. 87, No. 9, 1 May 2000 Schreyer et al. 5445 order harmonics at (0,0,1 2 ) are observed, which do not occur in bulk3 indicating a more square waveform of the ISDW than in bulk. It is tempting to attribute this effect to the finite Cr thickness. The squaring can actually be viewed as a precursor of the transition from ISDW to CSDW which occurs as we reduce tCr . With increasing temperature, a dramatic change of the spectrum is observed in Fig. 1: The line shape changes from a predominantly double peak to a broad peak at 001 at T 250 K which is very much reminiscent of the data for tCr 42 Å.6 As is shown in the figure, the data fits well to the same model, as the tCr 42 Å data, i.e., to a sharp central Gaussian and two broad superlattice satellite peaks. Their distance to the central peak corresponds to twice the super- lattice periodicity SL and is equal to the periodicity Fe,NC of the noncollinearly coupled Fe layers, i.e., FIG. 2. Kerr rotation in remanence normalized by Kerr rotation in saturation Cr,AF 2 SL as a function of temperature for the same sample as in Fig. 1. In the insets, Fe,NC . Inspection of the figure in the inset shows that this complete hysteresis loops are shown for the temperatures indicated by the relation holds for the depicted model structure. As detailed in arrows. The data indicate a transition from uncoupled behavior to NC cou- Ref. 6, the disorder along the growth direction of the model pling with increasing temperature. structure induces the observed broadening of the satellite peaks. Structure factor calculations and polarized neutron measurements confirm that the frustrated spiral structure ferent kind of substrate than in the present case, it is to be shown in the inset causes the observed scattering. Thus, we expected that in their case the distance between the steps is find a transition from an ISDW at low temperatures to the much smaller. Any antiferromagnetic order in the Cr layers frustrated spiral CSDW at elevated temperatures.10 This tran- would still be highly frustrated. However, the correlation sition can be explained by the increase of the ISDW period length of this frustrated structure could be so small, that it with temperature, as it occurs in bulk: For a given t might not yield sufficiently strong peaks which are observ- Cr , a single ISDW period can form below a certain temperature. If able with neutron scattering. Nevertheless, such a frustrated the temperature rises above this value, the LRO changes to a structure could still mediate the observed noncollinear cou- CSDW. pling between Fe layers. An obvious question at this point is if the exchange cou- The absence of coupling between the Fe layers in the pling between the Fe layers changes with the transition from ISDW phase at low temperatures can be understood in the the ISDW to the frustrated CSDW Cr structure. In Fig. 2, following way. Apparently, the ISDW minimizes the frustra- temperature-dependent magneto-optical Kerr effect MOKE tion of the Fe/Cr interlayer coupling, which is induced by data is shown. At low temperatures, the hysteresis loop is imperfect Fe/Cr interfaces, by moving its nodes to the Fe/Cr nearly square, indicating very weak or no coupling. Above a interfaces.12 The small magnetic Cr moments at the ISDW broad transition region,10 a hysteresis curve typical of non- nodes then can effectively decouple the ISDW Cr from the collinear coupling is found. PNR measurements have con- ferromagnetic Fe, strongly reducing any interlayer exchange coupling. firmed these results. Thus, we observe a transition from es- Shi and Fishman have studied the magnetic structure of sentially no coupling between the Fe layers to NC coupling, Cr in Fe/Cr theoretically.13 For ideal Fe/Cr interfaces, they as the Cr structure transforms from ISDW to the frustrated predicted the observed ISDW to CSDW transition as a func- spiral CSDW for a sample with tCr 80 Å. tion of temperature and Cr thickness. In a more recent The detailed functional form of the region of existence paper,14 Fishman considered more realistic interfaces with of the ISDW phase as a function of tCr and temperature has steps. He finds helical SDWs HSDW corresponding to the been mapped out by Fullerton et al.11,12 using MOKE hys- spiral magnetic structure proposed by Slonczewski, confirm- teresis loops, resistance measurements, and neutron scatter- ing the frustration effect as the cause of the noncollinear ing. The phase boundary which they obtain is fully consis- coupling between the Fe layers. Interestingly, he predicts for tent with our data. As in our case, they also find that the t coupling between the Fe layers vanishes, as soon as the Cr 80 Å that the HSDW performs a 270° twist instead of only a 90° rotation as shown in the inset of Fig. 1. So far our ISDW occurs. However, they do not observe the occurrence data do not confirm this effect. of the strong CSDW signal as the ISDW vanishes with in- creasing temperature. This is most probably due to a differ- ence in the structure of the Fe/Cr interfaces. It has been B. Proximity effect shown by Slonczewski that the average distance between Most recently, we have extended our study to tCr steps in the Fe/Cr interface, which effectively change the Cr 42 Å.15 We obtain strong evidence that the Ne´el tempera- thickness, must be on the order of 100 Å, consistent with our ture TN of the frustrated CSDW increases as we reduce tCr . data.6 Otherwise, the specific frustrated structure observed This behavior is contrary to expectation, since scaling theory here does not form. Since the samples of Fullerton et al. had would suggest a reduction of TN with tCr .16 We attribute this been prepared by a different technique sputtering on a dif- to a proximity effect between the Fe and the Cr. The Fe Downloaded 27 Mar 2001 to 148.6.178.13. Redistribution subject to AIP copyright, see http://ojps.aip.org/japo/japcr.jsp 5446 J. Appl. Phys., Vol. 87, No. 9, 1 May 2000 Schreyer et al. layers have a much higher ordering temperature TC in bulk layers. With increasing tCr , however, the domain wall en- than Cr and remain ferromagnetic at TFe 20 Å, as is seen in ergy of such domains increases so much that the system our experiments. In this case, Fe is expected to polarize the chooses a different way to accommodate the frustration. In- Cr atoms near the Fe/Cr interface and to induce magnetic stead it effectively breaks the antiferromagnetic coupling be- order, which would not exist otherwise.9 In this scenario, it tween Cr and Fe at the Fe/Cr interface and reorients the Cr can be expected that the Cr ordering temperature will in- moments perpendicular to the Fe moments. crease as we reduce tCr , since the influence of the Fe will increase strongly toward the Fe/Cr interfaces. This proximity D. Grazing incidence diffraction effect was such a basic assumption in Slonczewski's prox- imity magnetism model9 that he named the model accord- Interestingly, during the reorientation transition with in- ingly. Only this assumption enabled him to postulate the ex- creasing tCr an additional CSDW phase occurs with the mo- istence of a specific magnetic structure as the origin of NC ments parallel to the TSDW, i.e., out of the film plane. To coupling in layered systems consisting of FM and AF mate- establish, if the CSDW and TSDW phases coexist laterally in rials. Our observation of magnetic order and an increasing the film plane or stacked in layers, grazing incidence diffrac- T tion GID was performed with neutrons on the EVA diffrac- N with reduced tCr strongly supports the existence of a proximity effect in Fe/Cr layered systems experimentally. tometer at the Institut Laue Langevin in Grenoble/France.22 Another way to test this proximity effect is to vary the By varying the incident angle in the regime of total external ordering temperature T reflection, the penetration depth of the neutrons into the C of the FM material, and to study the effect on the ordering temperature T sample can be tuned.23 Thus, GID can distinguish if one of N of the AF material. One would then expect that T the two phases alone exists near the sample surface, indicat- N is reduced with TC . We have performed such a study using the system Fe ing a layering of the two phases. The measurements were 1 xCrx /Cr. For increasing x between 0 and 0.8, the T carried out on a 3000 Å thick Cr film with 20 Å Fe on top at C of bulk Fe1 xCrx drops from the value of pure Fe to nearly zero. Above x T 50 K with the in-plane scattering vector aligned either 0.8, the system enters a spin glass state.17 Since Fe with the commensurate 100 peak or with an incommensu- 1 xCrx , Fe and Cr have the same lattice structure and nearly equal rate 0.953 0 0 ISDW reflection. An ISDW peak originating lattice constants Fe from near the sample surface was found, whereas an equiva- 1 xCrx /Cr layered structures can be grown in similar quality as Fe/Cr, making the system an lent scan at the CSDW 100 position yielded no such ideal candidate for such a study of the proximity effect. The peak.22 Therefore, we can conclude that near the Fe/Cr inter- experiments were performed as a function of x on a series of face the ISDW prevails. The commensurate reflections ob- samples with t served with conventional high angle neutron scattering must Cr 30 Å and tFeCr 30 Å. TC(Fe1 xCrx) was determined via measurements of the magnetization using come from deep within the film. Consequently the occur- PNR and MOKE,18 whereas T rence of the commensurate phase in this system appears not N(Cr) was measured using high angle neutron scattering on the AF Cr peaks. We find to be induced by the top Fe/Cr interface, but rather by the that T bottom Cr/Nb interface. N(Cr) drops with TC(Fe1 xCrx) as would be expected for a proximity effect. The weaker the ferromagnetism of the As is discussed in detail in Ref. 8, the large lattice misfit proximity Fe between Cr and Nb at the Cr/Nb interface induce strain and a 1 xCrx layer is, the less it can polarize the Cr, reducing its T very small crystalline coherence in the film plane. Appar- N(Cr). The data and more details of the mea- surements on the Fe ently, this small in-plane coherence prevents the formation of 1 xCrx /Cr system will be reported elsewhere.19,20 an in-plane TSDW similar to the suppression of the ISDW at small tCr . Instead, a CSDW forms. Further away from the Cr/Nb interface the crystalline quality improves allowing the C. Thick Cr films formation of a TSDW. For out-of-plane propagation of the TSDW for 45 Å t Last but not least, we briefly review our results for very Cr 250 Å, on the other hand, the out-of- plane crystalline coherence length is found to be much larger large tCr . As is shown above, we observe a TSDW for tCr than along the in-plane direction. Thus, no concomitant 80 Å. Neutron scattering data show that with increasing CSDW forms. This argument also holds for those samples tCr , the TSDW reorients itself between 250 Å tCr where the Cr does not directly touch the Nb due to an inter- 2000 Å from propagating along the film growth axis to vening Fe layer. Since the Fe always was only 20 Å thick, it in-plane propagation with the Cr moments oriented out of the cannot relieve the strain induced by the Nb. film plane.7,8 This reorientation transition again is caused by steps at the Fe/Cr interface. These steps frustrate the system since the ferromagnetic coupling within the Fe, the antifer- E. Conclusions romagnetic coupling within the Cr and the antiferromagnetic In conclusion, we have discussed results obtained on the coupling between the Fe and the Cr at the Fe/Cr interface21 magnetic structure of Cr by neutron scattering methods. To cannot be satisfied at the same time. be able to perform neutron experiments, it was important to Calculations within a classical Heisenberg model repro- maximize the scattering volume by using large substrates duce the observed reorientation transition as soon as steps in and, for small tCr , many Fe/Cr bilayer repeats. Thus, the the Fe/Cr interface are assumed.7 For small tCr , the frustra- preparation of samples specially designed for neutron scat- tion can be accommodated by forming domain walls con- tering is important if one wants to push the limits of neutron necting the steps at opposing Fe/Cr interfaces through the Cr scattering in thin film research. This has enabled us to ob- Downloaded 27 Mar 2001 to 148.6.178.13. Redistribution subject to AIP copyright, see http://ojps.aip.org/japo/japcr.jsp J. Appl. Phys., Vol. 87, No. 9, 1 May 2000 Schreyer et al. 5447 serve a novel frustrated spiral CSDW order in Cr films, which are only 42 Å thick. This magnetic order causes the noncollinear coupling between the Fe layers. In this regime, we also observe a proximity effect between the Fe and the Cr. As we increase tCr , the Cr layers become thick enough to support a TSDW propagating out of the film plane. At high temperatures, this TSDW transforms into the frustrated CSDW, causing noncollinear coupling between the Fe lay- ers. As we increase tCr further, the frustration at the Fe/Cr interfaces causes a reorientation of the TSDW propagation to in-plane. A concomitant CSDW is observed, which is shown to be limited to the region near the bottom Cr/Nb interface using neutron GID. This latter technique operates at the lim- its of what neutrons can do. Nevertheless, if the samples are optimized accordingly, meaningful results can be extracted. FIG. 3. Difference spectrum of inelastic neutron scattering data taken at 75 III. INELASTIC SCATTERING and at 10 K with Gaussian fits and the resulting dispersion curve plotted in the Q-E plane. So far, we have only considered elastic neutron scatter- ing on magnetic thin films. However, inelastic neutron scat- tering has historically been just as important an application in condensed matter science. For neutron wavelengths in the The inelastic measurements were performed on a super- Å range, the neutron energies happen to be well matched, lattice of the layer sequence Y500 Å Dy43 Å /Y28 Å 350 / e.g., to phonon and magnon excitation energies in condensed Y2340 Å /Nb2000 Å /Al2O3 substrate , which was grown by matter. However, the cross sections of this inelastic scatter- MBE methods29 along the c-axis of the hcp structure of Dy ing are orders of magnitudes smaller, than in the elastic case. and Y. Again, the amount of Dy in the sample was maxi- Thus, if elastic neutron scattering already poses a challenge mized by growing 350 bilayers and using a rather large in thin film research, inelastic neutron scattering will be even 1 in. 0.5 in. substrate. Nevertheless, the Dy in the sample harder. However, in the following, we demonstrate that even amounts to only 10 mg. Below its bulk Nee´l temperature of inelastic measurements on thin film systems are possible. 179 K, Dy exhibits a helical phase which transforms into a Up to now, information on magnetic excitations in mag- ferromagnetic structure below 89 K. In Dy/Y superlattices, a netic thin films and superlattices has come from inelastic coherent helical phase occurs which extends over many bi- light scattering Brillouin scattering, BS 24,25 and ferromag- layers whereas the ferromagnetic phase is suppressed due to netic resonance FMR experiments.26 FMR can probe spin magnetoelastic effects.30 The coherent helical phase is medi- waves at the center of the Brillouin zone BZ , and at mag- ated by the paramagnetic Y interlayers via RKKY exchange non wave vectors qm inversely proportional to the film thick- coupling. Elastic neutron scattering experiments have con- ness. With Brillouin scattering, wave vectors on the order of firmed that the present Dy/Y sample exhibits the same long- the wave vector of light are accessible, i.e., also close to the range helical order observed previously. BZ center. Thus, it has not been possible to determine the The neutron scattering experiments were performed at dispersion of spin waves in magnetic thin film systems in the the NIST Center for Neutron Research on the cold neutron whole BZ. spectrometer NG-5. A focusing analyzer was used for the With inelastic neutron scattering INS , on the other inelastic measurements to maximize intensity while sacrific- hand, the dispersion of spin waves in much larger portions of ing resolution in Q. the BZ is, in principle, accessible. Smaller wavelength The inelastic measurements were performed at 75 K to higher qm spin waves on the order of the superlattice peri- create a sufficiently large spin-wave scattering cross odicity would become accessible which can be dominated by section.31 Equivalent scans at 10 K, where the spin-wave the magnetic exchange coupling between the magnetic lay- cross section is negligible, serve as a quasi-background. The ers. Thus, INS would open up the possibility to study the intensity difference of these scans then is a good measure of dynamic behavior of the exchange coupling in magnetic su- the spin-wave cross section. Counting for about 30 min per perlattices, which was not possible so far. point was necessary to obtain the data summarized in Fig. 3, Thin film samples made from rare-earth materials are the where the difference between the 75 and 10 K data sets is prime candidates for a study of magnetic excitations with plotted together with Gaussian fits and the projections of the inelastic neutron scattering, since the rare earths exhibit the peaks onto the Q-I and Q-E planes. We find a spin-wave largest magnetic moments of all elements, causing the largest excitation, which is well-localized in energy. The resulting possible cross-section for magnetic neutron scattering. Rare- dispersion curve is shown in the Q-E plane. Its extrapolation earth superlattices have been studied intensely over the last to higher Q yields an elastic peak at 1.97 Å 1 as its origin. decade using elastic neutron scattering to determine their This peak is caused by the helical long-range magnetic order magnetic structures.27,28 in the superlattice. Comparison with the data of bulk Dy32 Downloaded 27 Mar 2001 to 148.6.178.13. Redistribution subject to AIP copyright, see http://ojps.aip.org/japo/japcr.jsp 5448 J. Appl. Phys., Vol. 87, No. 9, 1 May 2000 Schreyer et al. shows that the dispersion curve observed here is essentially 7 P. Bo¨deker, A. Hucht, A. Schreyer, J. Borchers, F. Gu¨thoff, and H. Zabel, consistent with the bulk curve. Phys. Rev. Lett. 81, 914 1998 . These data demonstrate that it is possible to measure 8 P. Bo¨deker, A. Schreyer, and H. Zabel, Phys. Rev. B 59, 9408 1999 . 9 spin wave excitations in magnetic superlattices. A number of J. C. Slonczewski, J. Magn. Magn. Mater. 150, 13 1995 . 10 According to the more complete data shown here, the transition region interesting questions can now be investigated. In a superlat- between the ISDW and CSDW phases extends roughly from 150 to 250 tice the additional superlattice periodicity should lead to a K. Thus, the transition region is somewhat smaller than shown in Fig. 1 of folding of the Brillouin zone.33 Furthermore, it is expected, Ref. 6. that the coupling between the Dy layers via the Y modifies 11 E. E. Fullerton et al., Phys. Rev. Lett. 75, 330 1995 . 12 the dispersion in a certain q E. E. Fullerton, S. D. Bader, and J. L. Robertson, Phys. Rev. Lett. 77, m range. Gaps could exist at the 1382 1996 . boundaries of the folded Brillouin zone. Most recent higher 13 Z. P. Shi and R. S. Fishman, Phys. Rev. Lett. 78, 1351 1997 . resolution measurements provide strong evidence for these 14 R. S. Fishman, Phys. Rev. Lett. 81, 4979 1998 . effects. These results will be reported elsewhere.34 15 T. Schmitte, A. Schreyer, V. Leiner, R. Siebrecht, K. Theis-Bro¨hl, and H. Zabel, Europhys. Lett. 48, 692 1999 . 16 IV. CONCLUSIONS K. Binder, in Phase Transitions and Critical Phenomena, edited by C. Domb and J. L. Lebowitz Academic, London, 1983 , pp. 1­144. In conclusion, we have presented not only elastic but 17 Landolt-Bo¨rnstein, Zahlenwerte und Funktionen aus Naturwissenschft und also inelastic neutron scattering measurements on magnetic Technik, Gruppe 3: Kristall- und Festko¨rperphysik. N-S.; Magnetische thin film samples. The latter was not possible up to now. The Eigenschaften von Metallen Springer, Berlin, 1986 . 18 R. Siebrecht, A. Schreyer, T. Schmitte, W. Schmidt, and H. Zabel, key is to provide samples of sufficient scattering volume. In Physica B 267­268, 207 1999 . combination with modern neutron instrumentation at high 19 R. Siebrecht, Ph.D. thesis, Ruhr-Universita¨t Bochum, Germany in prepa- flux sources, this will provide new insights into the magnetic ration . 20 structure and dynamics which may not be obtainable R. Siebrecht, A. Schreyer, T. Schmitte, W. Schmidt, and H. Zabel un- otherwise. published . 21 See, e.g., F. U. Hillebrecht et al., Europhys. Lett. 19, 711 1992 . 22 P. Bo¨deker, A. Schreyer, P. Sonntag, Ch. Sutter, G. Gru¨bel, and H. Zabel, ACKNOWLEDGMENTS Phys. Rev. B 248, 115 1998 . 23 H. Dosch, Critical Phenomena at Surfaces and Interfaces, Springer Tracts The authors acknowledge financial support from the in Modern Physics Springer, Berlin, 1992 , Vol. 126. Bundesministerium fu¨r Bildung und Forschung 03- 24 P. Gru¨nberg, in Light Scattering in Solids V, edited by M. Cardona and G. ZA5BC20 and the Deutsche Forschungsgemeinschaft SFB Gu¨ntherodt Springer, Berlin, 1989 . 166 . One of the authors A.S. is indebted to the Alexander 25 B. Hillebrands and G. Gu¨ntherodt, in Ref. 1, Vol. II, pp. 258­277. 26 von Humboldt Foundation for a Feodor Lynen fellowship B. Heinrich, in Ref. 1, Vol II, pp. 195­257. 27 C. F. Majkrzak et al., Adv. Phys. 40, 99 1991 . during which the work on the inelastic measurements was 28 D. F. McMorrow et al., Physica B 192, 150 1993 . begun. 29 J. Kwo, in Thin Film Techniques for Low Dimensional Structures, edited by R. F. C. Farrow, S. S. S. Parkin, P. J. Dobson, N. H. Neaves, and A. S. 1 See, e.g., Ultrathin Magnetic Structures I II, edited by B. Heinrich and Arrott, NATO ASI Series B, Physics V13 Plenum, New York, 1988 , p. J. A. C. Bland Springer, Berlin, 1994 . 337; M. Hong, R. M. Fleming, J. Kwo, L. F. Schneemeyer, J. V. Wasz- 2 A. Schreyer, R. Siebrecht, U. Englisch, U. Pietsch, and H. Zabel, Physica czak, J. P. Mannaerts, C. F. Majkrzak, D. Gibbs, and J. Bohr, J. Appl. B 248, 349 1998 . Phys. 61, 4052, 1987 . 3 30 For a review, see, e.g., E. Fawcett, Rev. Mod. Phys. 60, 209 1988 . M. B. Salomon et al., Phys. Rev. Lett. 56, 259 1986 . 4 For an overview and a list of references, see, for example, Ref. 5. 31 See, e.g., G. L. Squires, Introduction to Thermal Neutron Scattering 5 A. Schreyer, J. F. Ankner, Th. Zeidler, H. Zabel, M. Scha¨fer, J. A. Wolf, Cambridge University Press, Cambridge, New York, 1978 . P. Gru¨nberg, and C. F. Majkrzak, Phys. Rev. B 52, 16066 1995 . 32 R. M. Nicklow et al., Phys. Rev. Lett. 26, 140 1971 . 6 A. Schreyer, C. F. Majkrzak, Th. Zeidler, T. Schmitte, P. Bo¨deker, K. 33 See, for example, J. Sapriel and B. Djafari Rouhani, Surf. Sci. Rep. 10, Theis-Bro¨hl, A. Abromeit, J. A. Dura, and T. Watanabe, Phys. Rev. Lett. 189 1989 . 79, 4914 1997 . 34 A. Schreyer et al. unpublished . Downloaded 27 Mar 2001 to 148.6.178.13. Redistribution subject to AIP copyright, see http://ojps.aip.org/japo/japcr.jsp