Journal of Electron Spectroscopy and Related Phenomena184 (2011) 287–290
Contents lists available atScienceDirect
Journal of Electron Spectroscopy and Related Phenomena
j o u r n a l h o m e p a g e :w w w . e l s e v i e r . c o m / l o c a t e / e l s p e c
Layer-selective studies of an anti-ferromagnetically coupled multilayer by resonant magnetic reflectivity in the extreme ultraviolet range
P. Grychtol
∗, R. Adam, A.M. Kaiser, S. Cramm, D.E. Bürgler, C.M. Schneider
Institute of Solid State Research IFF-9 & JARA-FIT, Research Center Jülich, D-52425 Jülich, Germany
a r t i c l e i n f o
Article history:
Available online 10 October 2010
Keywords:
T-MOKE EUV
Mabsorption edges Layer-selectivity
a b s t r a c t
A multilayer comprising a 5 nm Ni80Fe20 and a 10 nm Co40Fe60layer separated by a 0.6 nm Cr layer was investigated by resonant magnetic reflectivity measurements of horizontally polarized light in the extreme ultraviolet spectral range (EUV). By exploiting the transversal magneto-optical Kerr effect (T- MOKE) at theMabsorption edges of iron, cobalt and nickel (54 eV, 60 eV and 67 eV) a magnetic contrast as large as 30% can be obtained near a Brewster angle of about 45◦. Energy dependent scans of the magnetic asymmetry as well as magneto-optical hysteresis loops were recorded to study the magneto- optical response and to determine whether the switching behavior of individual layers in the strongly anti-ferromagnetically coupled multilayer system can be probed.
© 2010 Elsevier B.V. All rights reserved.
1. Introduction
Exploiting the magneto-optical Kerr effect (MOKE) in a reflec- tivity experiment is a well-established technique for investigations of static and dynamic processes in magnetism[1]. At wavelengths above 250 nm, however, the effect is weak in most materials – less than a millirad of Kerr rotation upon a full magnetization rever- sal. Furthermore, the wavelengths employed in conventional MOKE experiments are large compared to the resolution required to image cutting-edge magnetic structures on the nanometer scale[2].
In the last decades these shortcomings have been overcome by performing magnetic investigations in the soft X-ray region. How- ever, this increases the system complexity, since large synchrotron radiation facilities are needed for this purpose. By exploiting X- ray resonant magnetic scattering and X-ray magnetic dichroism at theLedges of transition metals at photon energies above 500 eV, not only a superior magneto-dichroic contrast of up to several tens of percent and a lateral resolution down to a few nanometers can be attained, but also element-selectivity can be gained by means of resonant excitations[3–5]. An entire suite of X-ray magneto-optical phenomena allows for magnetic investigations of individual con- stituents of heterogeneous ferromagnetic systems on a nanometer and on a femtosecond scale – the latter of which requires suit- able light sources such as free electron lasers[6]or appropriate techniques such as femto-slicing[7].
In the last years several approaches to explore resonant mag- netic reflectivity and magnetic dichroism in the extreme ultraviolet
∗Corresponding author: Tel.: +49 2461 61 2258.
E-mail address:p.grychtol@fz-juelich.de(P. Grychtol).
range (EUV) for magnetic contrast generation by addressing the Medges of transition metals have proven to be equally success- ful, attaining a magnetic contrast of up to 100% at photon energies around 50 eV[8–12]. Because the majority of synchrotron beam- lines at third generation storage rings are optimized for photon energies in the soft X-ray region, investigating magnetic proper- ties in the EUV region has not been in the focus of the magnetism community. However, recent developments justify a closer look into EUV magneto-optics, as laser-based light sources are able to produce photons with energies of up to 100 eV with moderate effort and a reasonable flux density. Advancements in laser ampli- fier technology have brought about reliable table-top light sources, which provide coherent and ultrashort EUV pulses by means of higher harmonic generation (HHG). They may serve as compact tools for element-selective investigations of magnetic properties on the femtosecond and nanometer scale in a laboratory environment [13–17].
Since only little attention has been paid to the EUV region, it is the purpose of this work to explore its potential as a mag- netic contrast mechanism by tuning the photon energy to theM absorption edges of Fe, Co and Ni in a transversal MOKE reflectiv- ity experiment that focuses on layer-selective investigations of an anti-ferromagnetically coupled multilayer.
2. Sample
The sample consisted of a magnetic bilayer structure and was prepared by thermal evaporation at a base pressure of 5×10−11mbar[18]. In a first step a substrate system for the mag- netic multilayer was manufactured comprising a 150 nm thick Ag(0 0 1) buffer layer[19], which was grown on an iron precovered 0368-2048/$ – see front matter© 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.elspec.2010.09.012
288 P. Grychtol et al. / Journal of Electron Spectroscopy and Related Phenomena184 (2011) 287–290
Fig. 1a.Hysteresis loop of the Ni80Fe20/Cr/Co40Fe60multilayer measured with a SQUID at room temperature.
Fig. 1b.PEEM images of the top Ni80Fe20(left) and the bottom Co40Fe60(right) layer taken in remanence at theL2edge of Ni (854 eV) and Co (780 eV). The grey scale in both images represents the magnetic contrast in the plane of the sample.
GaAs(0 0 1) wafer at a temperature of 380 K. After postannealing at a temperature of 570 K, a 10 nm Co40Fe60layer followed by a 0.6 nm Cr interlayer were deposited onto the buffer at room temperature.
A 5 nm Ni80Fe20layer was deposited on top of that and finally this multilayer was capped by 3 nm Al to prevent oxidation. For refer- ence purposes single Co40Fe60as well as Ni80Fe20layers of 5 nm thickness were manufactured on the same buffer system.
The magnetic switching behavior in the plane of the multi- layer system was first characterized by MOKE measurements in the visible range[20]. Recording hysteresis loops in a longitudinal MOKE (L-MOKE) setup for various angles of all three samples with respect to the external magnetic field revealed a fourfold magneto- crystalline anisotropy in the Co40Fe60layer exhibiting an easy axis along the1 0 0direction and a hard axis along the1 1 0direction.
Only a negligible magneto-crystalline anisotropy could be found in the presumably polycrystalline Ni80Fe20layer. The anisotropy of both single layers was measured to be translated into the multilayer system.
In the following, we investigated the hysteretic behavior of all samples by applying a magnetic field only along their easy axes for the sake of simplicity. We employed a SQUID magnetometer in order to account for magneto-optical effects by referring to the absolute magnetic moment. Both single layers showed a rectan- gular hysteresis loop with a coercivity of less than 10 mT. The multilayer, on the other hand, produced a hysteresis loop which could be attributed to the anti-ferromagnetic coupling of the bot- tom Co40Fe60layer and the top Ni80Fe20layer, as can be clearly seen inFig. 1a.
By taking into account the thicknesses of the top and bot- tom layer as well as the saturation magnetization derived from
SQUID measurements of the individual layers, we can understand the switching behavior of the multilayer in detail. Above 50 mT the multilayer is saturated and both layers point into the same direction. As we decrease the magnetic field the top layer starts to reverse its magnetization until an anti-parallel alignment is reached around 10 mT. Once the magnetic field exceeds the coer- cive field of the bottom layer around−10 mT, its magnetization flips into the field direction followed by an immediate reversal of the top layer to maintain the preferred anti-parallel configuration.
When the magnetic field is increased further, the top layer starts to reverse again until the multilayer saturates in a parallel state below−50 mT. We confirmed the anti-ferromagnetic configuration in remanence by XMCD images taken with a photoemission elec- tron microscope (PEEM) tuned to theL2absorption edges of Ni and Co[21], which are displayed inFig. 1b.
The complex domain structure indicates that the magnetization reversal proceeds mainly via domain wall nucleation and motion rather than magnetization rotation.
3. Experimental setup
We performed resonant magnetic reflectivity measurements in a T-MOKE geometry at the undulator beamline UE56/1-SGM of the synchrotron radiation facility BESSY II. The reflectivity of lin- earlyp-polarized EUV light was measured across theMabsorption edges of Fe, Co and Ni from 52 eV to 72 eV with an energy reso- lution of 0.1 eV and a degree of linear polarization exceeding 99%.
Focusing mirrors enabled the synchrotron beam to be concentrated into a spot size of approximately 100×100m2. The multilayer stack was placed into a dedicated UHV reflectometer allowing for −2 scans in a horizontal plane with the angle of incidence ranging from 0◦ to 90◦. The intensity of the EUV light reflected off the sample was detected by a Schottky-type GaAsP photodiode (Hamamatsu G1127) directly connected to a sensitive ampereme- ter (Keithley 6517A). A set of coils was used to generate a static magnetic field of up to±140 mT along the vertical axis of the chamber.
4. Theoretical considerations
In contrast to longitudinal or polar Kerr effects, where usually the change of polarization is analyzed to obtain information about the magnetic state of the sample, it is sufficient as well as conve- nient to record the change of reflectivity in a T-MOKE geometry.
This is particularly true in the EUV regime due to the magnitude of the magneto-dichroic signal. Its strength originates from a large resonant enhancement of the reflected light if the energy of the inci- dence beam is tuned to the absorption edge of the material under investigation[22]. This strongly resonant behavior involves low order electric multipole transitions between core levels and unoc- cupied states of the valence band which results in both element selectivity and magnetic sensitivity in the presence of spin-orbit coupling and exchange splitting. For 3d transition metals, enhanced magnetic resonances occur at the M2,3 absorption edges in the range of 50–75 eV by involving mainly 3p→3dtransitions.
The strength of the T-MOKE signal is commonly denoted as the normalized difference of the reflected intensityIfor two inverted directions of the sample magnetization, here referred to as ↑ and↓. The latter can be achieved by an external magnetic field applied perpendicular to the scattering plane. This so-called mag- netic asymmetryA is introduced to separate the non-magnetic from the magnetic contribution in the magneto-optical response of the sample. It is related to the Fresnel reflection coefficientsrpp, which describes the influence of a material on p-polarized light within the classical magneto-optical formalism[23]. If this mate-
P. Grychtol et al. / Journal of Electron Spectroscopy and Related Phenomena184 (2011) 287–290 289
Fig. 2.Magnetic asymmetry spectra for a single Ni80Fe20 layer (5 nm), a single Co40Fe60 layer (5 nm) and the Ni80Fe20/Cr/Co40Fe60 multilayer stack (5 nm/0.6 nm/10 nm) taken at a fixed incidence angle of 45◦.
rial carries a magnetic moment,rppcan be written as the sum of a non-magneticrpp0 and a magneticppcomponent, the former of which is eliminated in the evaluation ofAleaving only the magnetic contribution:
A=I↑−I↓
I↑+I↓=|r↑pp|2− |rpp↓|2
|r↑pp|2+ |rpp↓|2withrpp↑↓=r0pp±pp.
5. Magneto-optical response in the EUV
In a first step of the synchrotron T-MOKE experiment, we scanned the photon energy across the Fe, Co and Ni edge from 52 eV to 72 eV for our three samples in order to determine the con- ditions for a maximum of the element-specific magneto-dichroic signal. In this energy range we measured the magnetic asymmetry Aat a fixed incidence angle of 45◦with a magnetic field revers- ing between±100 mT, i.e. magnetically saturating each sample in opposite directions.
From this result, which is displayed inFig. 2, we can see that the magneto-optical response of the single layers show distinct extrema of the magnetic asymmetry located around their respec- tive absorption edges, whose approximate positions are indicated by vertical lines in the same figure, whereas the response of the multilayer can approximately be considered a superposition thereof. The magnetic asymmetry of the single Ni80Fe20layer (black line) has a maximum of 20% at 54 eV accompanied by a mini- mum of−10% at 58 eV both located around the Fe absorption edge, and a maximum of 25% at 66 eV with a small minimum of
−5% at 69 eV located around the Ni absorption edge. The single Co40Fe60layer (red dots) shows maxima of the magnetic asymme- try at the Fe (30%) and Co (10%) absorption edges which seem to
share a common minimum of−7% at 58 eV. The merging of several bipolar shaped resonances of the magnetic asymmetry, which are located around an absorption edge, can be understood by taking into account the energetic proximity and width of the absorption edges causing a spread of the magneto-optical constants and thus response over several electron volt[12]. The variation in height can most likely be explained by the unequal mixing ratios of the Fe and the Co/Ni forming the respective alloys. The magnetic asymmetry of the multilayer system (green dotted line) resembles the response of both single layers. It is striking that not only the amplitude of the extrema at the Fe and Co absorption edges have changed, but also have they moved by about 1 eV to higher photon energies, whereas the maxima and minima at the Ni edge barely moved, changing in amplitude only. These phenomena can be attributed to thickness dependent interferences in the EUV, which are dominated by the response of buried layers, as has been shown previously[24].
6. Magnetic switching behavior
After having identified the maxima of the magneto-dichroic sig- nal at theMabsorption edges of Fe, Ni and Co, the photon energy can be tuned to probe the magnetic switching behavior element- selectively, which in the case of a multilayer system is supposed to translate into a layer-specific response. To this end, we have recorded the magnetic field dependence of the T-MOKE reflectivity resulting in hysteresis like loops which only indirectly reflect the switching behavior of the sample magnetization, as the recorded signal comprises both magnetic and magneto-optical contribu- tions. We recorded a set of thesemagneto-optical loopsin steps of 1 eV from 52 eV to 72 eV in order to discern the magnetic switching behavior from the magneto-optical response. This discrimination is achieved by comparing the recorded loops with the loops taken by the above-mentioned SQUID magnetometer (Fig. 1) and L-MOKE setup in the visible range (Fig. 3a).
In order to elaborate on how to exploit the magneto-optical response of the multilayer to attain a layer-selective response, we display selected magneto-optical loops inFig. 3b–d. They were recorded at 58 eV for the response of the bottom layer –Fig. 3b, at 69 eV for the response of the top layer –Fig. 3d and at 61 eV showing the response of the entire multilayer system –Fig. 3c. It is important to note that we had to specifically choose these photon energies, because we observed that for some photon energies, and even at local extrema of the magnetic asymmetry, the recorded magneto-optical loops feature a strong superposition caused by magneto-optical crosstalk between neighboring elements. There- fore, there is no simple recipe how to chose the photon energy in order to isolate the response from a single element or layer, respectively.
The magneto-optical loop shown inFig. 3b mainly exhibits a rectangular shape reflecting an instant switching of the sample magnetization around±10 mT previously associated with the flip- ping of the bottom Co40Fe60layer.Fig. 3d, on the other hand, shows
Fig. 3.Magneto-optical hysteresis loops of the Ni80Fe20/Cr/Co40Fe60multilayer taken in the visible range and at photon energies of 58 eV, 61 eV and 69 eV.
290 P. Grychtol et al. / Journal of Electron Spectroscopy and Related Phenomena184 (2011) 287–290
a magneto-optical loop which exemplifies the multiple reversal of the top Ni80Fe20layer. It begins to reverse its magnetization around 50 mT with a decreasing magnetic field until an anti-parallel align- ment is reached at about 10 mT. Once the magnetic field exceeds the coercive field of the Co40Fe60layer around−10 mT, its magne- tization flips to maintain the preferred anti-parallel configuration.
When the magnetic field is decreased even further, this layer starts to reverse again until it saturates in parallel to the bottom layer below−50 mT.
Both magneto-optical loops shown inFig. 3a and c basically illus- trate the switching behavior of the entire multilayer and fall into place if compared withFig. 1. Because the SQUID measurement is only sensitive to absolute magnetic moments, the amplitude of the recorded signal is not superimposed by a magneto-optical response which may lead to a misrepresentation of the switching behavior. Depending on the strength and direction of the magneto- optical effect and the penetration depth of the light, the response of one layer can dominate the recorded signal. This explains why the amplitude ratio between the signal from the top and bottom layers change from 1:3 in the case of the SQUID, to 1:1 in the case of the magneto-optical loop taken in the visible range and finally to about 3:1 in the case of the magneto-optical loop taken at 61 eV.
Here, the spread of the magneto-optical constants of Fe, Co and Ni in the EUV region over several electron volt can result in a magneto- optical crosstalk which affects a layer-selective response, as the reflected signal of both layers can be resonantly enhanced by the same order of magnitude[12,24]. This may also explain the bumps in the magneto-optical loop shown in Fig. 3b around ±30 mT.
Nonetheless, we were able to demonstrate that the response of the multilayer system can be probed layer-selectively, if the photon energy is carefully tuned to energies near theMabsorption edge of the element contained in the layer under investigation.
7. Summary and conclusion
We explored the potential for layer-selective measurements of the magnetization in magnetic multilayers by tuning to theM absorption edges in the EUV range. Our results as well as the future employment of a table-top soft X-ray source pave the way for ultrafast element- as well as layer-selective investigations of the magnetization on the nanometer scale in a laboratory environment.
Acknowledgments
We would like to thank T. Jansen, F.-J. Köhne, B. Küpper, J. Lauer, H. Pfeiffer, N. Schnitzler, and R. Schreiber for their relentless techni- cal support, without which this work would not have been possible.
We also are grateful to the BESSY staff in assisting us to make the beamline run 24/7. The financial support by the BMBF (Project 05KS7UK1) and the DFG (SFB 491) is gratefully acknowledged.
References
[1] B. Hillebrands, Spin Dynamics in Confined Magnetic Structures, vols. 1, 2, 3, Springer, Berlin, 2004.
[2] A. Zvezdin, V. Kotov, Modern Magnetooptics and Magnetooptical Materials, Institute of Physics Publishing, Bristol, 1997.
[3] C. Kao, J.B. Hastings, E.D. Johnson, D.P. Siddons, G.C. Smith, G.A. Prinz, Phys. Rev.
Lett. 65 (1990) 373.
[4] H.-C. Mertins, D. Abramsohn, A. Gaupp, F. Schäfers, W. Gudat, O. Zaharko, H.
Grimmer, P.M. Oppeneer, Phys. Rev. B 66 (2002) 184404.
[5] F. Nolting, A. Scholl, J. Stohr, J.W. Seo, J. Fompeyrine, H. Siegwart, J.-P. Locquet, S. Anders, J. Luning, et al., Nature 405 (2000) 767.
[6] C. Gutt, L.-M. Stadler, S. Streit-Nierobisch, A.P. Mancuso, A. Schropp, B. Pfau, C.M. Gunther, R. Konnecke, J. Gulden, B. Reime, et al., Phys. Rev. B 79 (2009) 212406.
[7] C. Stamm, T. Kachel, N. Pontius, R. Mitzner, T. Quast, K. Holldack, S. Khan, C.
Lupulescu, E.F. Aziz, M. Wietstruk, et al., Nat. Mater. 6 (2007) 740.
[8] F.U. Hillebrecht, T. Kinoshita, D. Spanke, J. Dresselhaus, C. Roth, H.B. Rose, E.
Kisker, Phys. Rev. Lett. 75 (1995) 2224.
[9] M. Pretorius, J. Friedrich, A. Ranck, M. Schroeder, J. Voss, V. Wedemeier, D.
Spanke, D. Knabben, I. Rozhko, H. Ohldag, et al., Phys. Rev. B 55 (1997) 14133.
[10] M. Sacchi, G. Panaccione, J. Vogel, A. Mirone, G. Laan, Phys. Rev. B 58 (1998) 3750.
[11] M. Hecker, P. Oppeneer, S. Valencia, H.-C. Mertins, C.M. Schneider, J. Electron Spectrosc. Relat. Phenom. 144–147 (2005) 881.
[12] S. Valencia, A. Gaupp, W. Gudat, H.-C. Mertins, P.M. Oppeneer, D. Abramsohn, C.M. Schneider, New J. Phys. 8 (2006) 254.
[13] H.C. Kapteyn, M.M. Murnane, I.P. Christov, Phys. Today 58 (2005) 39.
[14] X. Zhou, R. Lock, N. Wagner, W. Li, H.C. Kapteyn, M.M. Murnane, Phys. Rev. Lett.
102 (2009) 073902.
[15] T. Popmintchev, M. Chen, A. Bahabad, M. Gerrity, P. Sidorenko, O. Cohen, I.P.
Christov, M.M. Murnane, H.C. Kapteyn, PNAS 106 (2009) 10516.
[16] C. La-O-Vorakiat, M. Siemens, M.M. Murnane, H.C. Kapteyn, S. Mathias, M.
Aeschlimann, P. Grychtol, R. Adam, C.M. Schneider, J.M. Shaw, et al., Phys. Rev.
Lett. 103 (2009) 257402.
[17] R.L. Sandberg, D.A. Raymondson, C. La-O-Vorakiat, A. Paul, K.S. Raines, J.
Miao, M.M. Murnane, H.C. Kapteyn, W.F. Schlotter, Opt. Lett. 34 (2009) 1618.
[18] R.R. Gareev, D.E. Bürgler, M. Buchmeier, D. Olligs, R. Schreiber, P. Grünberg, Phys. Rev. Lett. 87 (2001) 157202.
[19] D.E. Bürgler, C.M. Schmidt, J.A. Wolf, T.M. Schaub, H.J. Güntherodt, Surf. Sci. 366 (1996) 295.
[20] M. Buchmeier, R. Schreiber, D.E. Bürgler, C.M. Schneider, Phys. Rev. B 79 (2009) 064402.
[21] C.M. Schneider, A. Kaiser, C. Wiemann, C. Tieg, S. Cramm, J. Electron Spectrosc.
Relat. Phenom. 181 (2010) 159.
[22] J.P. Hannon, G.T. Trammell, M. Blume, D. Gibbs, Phys. Rev. Lett. 61 (1988) 1245.
[23] P. Oppeneer, Handbook of Magnetic Materials, vol. 13, Elsevier, Amsterdam, 2001.
[24] P. Grychtol, R. Adam, S. Valencia, S. Cramm, D.E. Bürgler, C.M. Schneider, Phys.
Rev. B 82 (2010) 054433.
