PHYSICAL REVIEW B VOLUME 57, NUMBER 1 1 JANUARY 1998-I Resonant magnetic scattering of polarized soft x rays: Specular reflectivity and Bragg diffraction from multilayers Maurizio Sacchi Laboratoire pour l'Utilisation du Rayonnement Electromagne´tique, Centre Universitaire Paris-Sud, 91405 Orsay, France and Istituto Nazionale di Fisica della Materia, Sezione di Modena, Via Campi 213/a, 41100 Modena, Italy Coryn F. Hague Laboratoire de Chimie Physique­Matie re et Rayonnement, Universite´ Pierre et Marie Curie, 11 rue Pierre et Marie Curie, 75231 Paris Cedex 05, France and Laboratoire pour l'Utilisation du Rayonnement Electromagne´tique, Centre Universitaire Paris-Sud, 91405 Orsay, France Eric M. Gullikson and James H. Underwood Center for X-ray Optics, Materials Sciences Division, Ernest Orlando Lawrence Berkeley National Laboratory, University of California, Berkeley, California 94720 Received 11 June 1997 We have measured resonant magnetic scattering of elliptically polarized soft x rays from magnetically ordered Fe/Co multilayers, tuning the photon energy across the 2p edges of Fe and Co. Specular reflectivity was measured for a series of angles of incidence as a function of the photon energy. Bragg diffraction was measured performing /2 scans for several photon energies. In both cases, large magnetic signals were observed, up to 20% peak to peak in the asymmetry ratio. An estimate of the variation of the real part of the refractive index through the Fe L3 edge is derived from the Bragg peak displacement versus energy. S0163-1829 98 00401-9 I. INTRODUCTION sample, but at the same time we also wanted to test the feasibility of resonant magnetic scattering experiments using Magnetic effects on elastic x-ray scattering Bragg dif- elliptically polarized soft x rays from a bending magnet fraction, specular reflectivity or diffuse scattering are a well- known phenomenon.1­3 They represent a powerful tool for beamline. The measurements were performed on the Ad- investigating magnetic materials since, as has been vanced Light Source beamline 6.3.2 at Berkeley.18 The shown,4­6 they are strongly enhanced when the photon en- beamline, based on a Hettrick-Underwood design, has no ergy is tuned across an absorption edge resonant process . entrance slits to the monochromator, which uses a varied The resonant enhancement of the magnetic scattering has line-space grating, and features several mechanical solutions mainly been investigated at high photon energies7­10 in order which guarantee high stability and ease of operation. In order to match the Bragg law for the typical lattice spacings of to change the polarization state of the light without affecting crystals. In the soft x-ray range, even larger effects are ex- either the optical alignment of the beamline or the calibration pected, working, for instance, at the 2p edges of transition of the energy scale, we simply modified the position of the metals of the first row or at the 3d edges of rare earths vertical jaws that define the angular acceptance at the en- 300­1500 eV , but the corresponding long wavelengths pre- trance of the monochromator. Linearly polarized light is ob- vent the use of single crystals. Two approaches have been tained when selecting a vertical accepted angle symmetric adopted recently in this energy range: 1 The study of the with respect to the orbit plane of the electrons in the ring. To Bragg diffraction from artificial structures of appropriate 2d have elliptically polarized light of positive negative helic- spacing.11,12 2 The analysis of the specular reflectivity, ity, only the portion of the beam emitted above below the which contains analogous information but has no constraints orbit plane should go through the monochromator. A good related to the lattice spacing.13­17 Both approaches have their compromise between flux and polarization rate was found own specific advantages: for instance, working under Bragg accepting the beam within the (0.17 0.05) mrad angular conditions provides information about the magnetic period- range above the orbit plane. In these conditions, the circular icity in ordered structures, while resonant reflectivity can polarization rate of the collected photons not affected by the easily be related to electronic properties and absorption spec- grazing incidence monochromator is calculated to be about tra. An important aspect common to all the resonant x-ray- 60% in the energy range that we used 650­850 eV . The scattering techniques is the element selectivity which is in- exit slits were set at 50 m, for a resolving power of 1200 herent to working at a specific absorption edge: under these and a flux of 1010 photons s 1 on the sample at the Fe 2p conditions, x-ray scattering in fact becomes a spectroscopy. edges. Measurements were performed on ex situ deposited me- II. EXPERIMENT tallic layers, as well as on multilayers and crystals. The The aim of our experiment was to compare the magnetic samples were magnetized along the intersection between the signal in reflectivity and diffraction from a given multilayer surface and scattering planes by a permanent magnet placed 0163-1829/98/57 1 /108 4 /$15.00 57 108 © 1998 The American Physical Society 57 BRIEF REPORTS 109 FIG. 3. /2 scans at different photon energies close to the Fe FIG. 1. Reflectivity curves left panel taken at different angles L3 edge a . Continuous and dotted lines are for opposite magneti- of incidence as a function of the photon energy, over a range in- zation directions. The difference curve b for h 704 eV and the cluding the 2p absorption edges of both Fe and Co. The sample is asymmetry ratio c are also given. a 11.5 Å Fe / 20 Å Co, Fe-terminated multilayer. The right panel shows the corresponding asymmetry, i.e., the ratio between the dif- asymmetry curves is related to the interference between the ference and the sum of the reflectivity curves measured for opposite real and imaginary parts of the refractive index through helicity/magnetization orientations. Fresnel's equations. It can also be noted in Fig. 1 that for behind the sample holder generating a field of approximately 1° the reflectivity is almost constant away from the Fe 800 G at the sample position. The magnet could be rotated in 2p energy region and the signal from Co second layer can vacuum around the axis normal to the sample surface, using hardly be detected. Figure 2 shows in more detail the reflec- a stepper motor.19 In this way we could change the relative tivity spectrum measured at 5° for a different Fe/Co orientation between photon helicity and sample magnetiza- multilayer 21.5 Å Fe / 29.5 Å Co : the curves for opposite tion for each scan without affecting the sample alignment. In magnetizations and their difference are reported in the left the end station of line 6.3.2 the scattering plane is vertical panel, while on the right we have the asymmetry ratio. The i.e., orthogonal to the plane of the ring . The sample and the fine structure that can be clearly observed in between the L3 detector can be rotated around the same axis in an indepen- and L2 iron edges indicates that, when working on metals, a dent or coupled /2 mode. We performed energy scans at resolving power of 1 103 is sufficient for a detailed fixed scattering angles and also /2 scans at fixed photon analysis of both the reflectivity and its magnetization depen- energies. is measured relative to the sample surface. dence. On the same sample, we also measured the Bragg diffrac- III. RESULTS tion from the periodic structure of the multilayer, when the photon energy approaches the 2p resonances. Figure 3 a Figure 1 reports the magnetization-averaged reflectivity shows a few examples of a series of /2 scans taken for curves for an Fe/Co multilayer 11.5 Å Fe / 20 Å Co, Fe- terminated measured at different angles of incidence over a photon energy range including both Fe and Co L2,3 edges. The right panel of Fig. 1 shows the corresponding magnetic part of the resonant scattering, presented as the asymmetry ratio (I I )/(I I ), where I is the intensity for the photon helicity parallel or antiparallel to the magnetization. The strong angular dependence of both reflectivity and FIG. 2. Resonant reflectivity curves for opposite magnetization FIG. 4. Photon energy dependence of the effective refractive directions in a 21.5 Å Fe / 29.5 Å Co multilayer. The difference index neff value for the 21.5 Å Fe / 29.5 Å Co multilayer, as ob- ( 5) and the asymmetry ratio right panel are also given. tained from the Bragg peak position. 110 BRIEF REPORTS 57 various photon energies just below the Fe 2p3/2 edge. For tivity and Bragg diffraction at the 2p resonances. Specular each energy, two curves are reported corresponding to oppo- reflectivity has the advantage of being free from constraints site magnetization/helicity orientations. Apart from an in- on the existence of a periodic structure and on the value of its creased intensity of the diffraction peak, there is also an evi- lattice parameter. The preliminary measurements that we dent enhancement of the variation with the magnetic field performed to characterize our experimental setup were con- when the photon energy gets closer to the edge. The result of ducted on a simple Fe film deposited on silicon and in a a /2 scan taken at 704 eV is shown together with the previous work a Ni single crystal was studied: in both cases, corresponding asymmetry ratio Figs. 3 b and 3 c , respec- no Bragg peak would have been available at the 2p reso- tively . The displacement with the photon energy of the nances. Figure 1 also shows that the probing depth of reso- Bragg peak on the angle scale is certainly related to the nant reflectivity can be roughly tuned to enhance near sur- change in wavelength, but it persists even when the curves face contributions. Finally, working at grazing angles gives are plotted versus the scattering vector 4 sin / . This en- ergy dependence originates from the sharp variations in the high reflectance and, consequently, fast data collection real part of the refractive index n close to an absorption about 20 min to measure one atomic layer of nickel on edge. The effective n value for the multilayer as a function copper20 . On the other hand, resonant Bragg diffraction is a of the photon energy is plotted in Fig. 4. These values were unique tool to investigate the periodicity and interface rough- obtained from the Bragg peak position ness of multilayers, both in terms of structure and magne- B according to the approximate relation tism. Moreover, Fig. 4 shows another interesting application of this technique, namely it provides access to a direct deter- mination of the real part of the index of refraction through an n e f f 2d sin sin2 B cos2 B absorption resonance. B 2dsin B cos2 B . In conclusion, both specular reflectivity and Bragg dif- For the multilayer spacing d we took the nominal value of 51 fraction represent important tools for characterizing mag- Å. The curve in Fig. 4 should still vary for higher photon netic materials, especially in the form of thin films and mul- energies, and larger magnetic effects are expected at the ex- tilayers. Given their absolute compatibility in terms of act 2p3/2 edge position about 707 eV for Fe , but we could experimental setup, they should be performed together not see a diffraction peak there. The reason is that for our sample a photon incident at about 9° has to travel through whenever possible. The maximum magnetic asymmetry ratio approximately 275 Å of Fe to be reflected at the first Fe/Co obtainable depends on the adopted geometry, but it is easily interface and come out again. This distance is larger than the as high as observed in absorption or even higher.13­17 To- absorption length at the maximum of the Fe L gether with its intrinsic photon-in­photon-out character, this 3 edge, hence photons of 707­710 eV can hardly feel the periodic structure makes the technique perfectly suited for semiquantitative of the multilayer. In general, the Bragg peak is bound to get magnetometry element specific hysteresis loops, orientation broader at resonance since the increased absorption reduces of easy axes, etc. . In this experiment, we have also shown the number of planes scattering in phase. If the Bragg peak that the monochromator and the end station of beamline remains measurable low oscillator strength and/or reduced 6.3.2 at the Advanced Light Source are perfectly suited for thickness of the absorbing element for a given 2d of the resonant magnetic scattering experiments in the soft x-ray multilayer , data can easily be corrected for absorption and range. The location of the beamline on a bending magnet still give very useful information.11,12 source means that a wide energy range may be covered with The enhanced surface sensitivity in the proximity of a only smooth variations of the incoming intensity, albeit at resonance is also indicated by the appearance in Fig. 3 a of the price of a lower flux. Together with easy tunability of the a second Bragg peak. Combined absorption measurements in polarization state, this represents a major advantage over an total electron yield mode probing depth 20­30 Å indi- insertion device delivering high flux in narrow energy bands, cate that a partial oxidation of the top layer might be at the for this kind of spectroscopy. origin of the altered Bragg peak position through a change of the actual top layer thickness and/or of its refractive index. ACKNOWLEDGMENT IV. CONCLUSIONS We have performed resonant magnetic scattering experi- The authors would like to thank the staff of the Advanced ments on Fe/Co multilayers, measuring both specular reflec- Light Source for efficient help. 1 O. Klein and Y. Nishina, Z. Phys. 52, 853 1929 . 6 E. D. Isaacs, D. B. McWhan, C. Peters, G. E. Ice, D. P. Siddons, 2 P. M. Platzman and N. Tzoar, Phys. Rev. B 2, 3556 1970 . J. B. Hastings, C. Vettier, and O. Vogt, Phys. Rev. Lett. 62, 3 F. de Bergevin and M. Brunel, Phys. Lett. 39A, 141 1972 . 1671 1989 . 4 D. Gibbs, D. R. Harshman, E. D. Isaacs, D. B. McWhan, 7 P. Carra, M. Altarelli, and F. de Bergevin, Phys. Rev. B 40, 7324 D. Mills, and C. Vettier, Phys. Rev. Lett. 61, 1241 1988 . 1989 . 5 J. P. Hannon, G. T. Trammell, M. Blume, and D. Gibbs, Phys. 8 D. Gibbs, G. Gru¨bel, D. R. Harshman, E. D. Isaacs, D. B. Rev. Lett. 61, 1245 1988 . 57 BRIEF REPORTS 111 McWhan, D. Mills, and C. Vettier, Phys. Rev. B 43, 5663 Smith, and G. A. Prinz, Phys. Rev. Lett. 65, 373 1990 . 1991 . 14 C. C. Kao, C. T. Chen, E. D. Johnson, J. B. Hastings, H. J. Lin, 9 C. C. Tang, W. G. Stirling, G. H. Lander, D. Gibbs, W. Herzog, G. H. Ho, G. Meijs, J.-M. Brot, S. L. Hubert, Y. U. Idzerda, and P. Carra, B. T. Thole, K. Mattenberger, and O. Vogt, Phys. Rev. C. Vettier, Phys. Rev. B 50, 9599 1994 . B 46, 5287 1992 . 15 M. Sacchi, J. Vogel, and S. Iacobucci, J. Magn. Magn. Mater. 10 F. de Bergevin, M. Brunel, R. M. Gale´ra, C. Vettier, E. Elkai¨m, 147, L11 1995 . M. Bessie re, and S. Lefe bvre, Phys. Rev. B 46, 10772 1992 . 16 V. Chakarian, Y. U. Idzerda, C.-C. Kao, and C. T. Chen, J. Magn. 11 J. M. Tonnerre, L. Se ve, D. Raoux, G. Soullie´, B. Rodmacq, and Magn. Mater. 165, 52 1997 . P. Wolfers, Phys. Rev. Lett. 75, 740 1995 . 17 M. Sacchi, Surf. Rev. Lett. 4, 343 1997 . 12 C. F. Hague, J.-J. Gallet, J.-M. Mariot, and M. Sacchi, in Raman 18 J. H. Underwood et al., Rev. Sci. Instrum. to be published . Emission by X-rays, edited by D. Ederer and J. H. McGuire 19 C. F. Hague, J.-M. Mariot, and J.-J. Gallet, Appl. Phys. A to be World Scientific, Singapore, 1996 , p. 137. published . 13 C. C. Kao, J. B. Hastings, E. D. Johnson, D. P. Siddons, G. C. 20 M. Sacchi, S. Iacobucci, and J. Rife unpublished .