Uncompensated magnetization and exchange-bias field in La 0.7 Sr 0.3 MnO 3 /YMnO 3 bilayers: The influence of the ferromagnetic layer
C. Zandalazini, P. Esquinazi, G. Bridoux, J. Barzola-Quiquia, H. Ohldag, E. Arenholz
aa r X i v : . [ c ond - m a t . m t r l - s c i ] S e p Uncompensated magnetization and exchange-bias field in La . Sr . MnO /YMnO bilayers: The influence of the ferromagnetic layer C. Zandalazini, ∗ P. Esquinazi, † G. Bridoux, ‡ and J. Barzola-Quiquia Division of Superconductivity and Magnetism, University of Leipzig, D-04103 Leipzig, Germany
H. Ohldag
Stanford Synchrotron Radiation Lightsource, Stanford University, Menlo Park, CA 94025, USA
E. Arenholz
Advanced Light Source, Lawrence Berkeley National Laboratory, Berkeley, CA 94720, USA (Dated: June 19, 2018)We studied the magnetic behavior of bilayers of multiferroic and nominally antiferromagnetico-YMnO (375 nm thick) and ferromagnetic La . Sr . MnO and La . Ca . MnO (8 . . .
225 nm),in particular the vertical magnetization shift M E and exchange bias field H E for different thicknessand magnetic dilution of the ferromagnetic layer at different temperatures and cooling fields. Wehave found very large M E shifts equivalent to up to 100% of the saturation value of the o-YMOlayer alone. The overall behavior indicates that the properties of the ferromagnetic layer contributesubstantially to the M E shift and that this does not correlate straightforwardly with the measuredexchange bias field H E . I. INTRODUCTION
In bilayers composed of antiferromagnetic (AFM) anda ferromagnetic (FM) phases a “horizontal” shift in thefield axis of the hysteresis loops is generally observed af-ter cooling them in a field applied at temperatures be-tween the N´eel T N and Curie T C temperatures . This“exchange-bias field” H E has been studied in differentsystems due to its fundamental importance as well as itstechnological relevance in spin-valve sensors, actuatorsand in high-density recording media and some details ofthe origin of H E are still a matter of discussion. Less studied is the shift in the magnetization axis,i.e. the “vertical” M E shift in the hysteresis loop, prob-ably because of its rather small relative values andits dependence on the cooling field H FC . Recently, amaximum shift of 16% of the saturation magnetizationwas found in Fe x Ni − x F /Co bilayers, which appeared tohave an exchange bias field of its own. It was proposedthat M E is related to uncompensated moments (UCM) atthe AFM/FM interface and should have a direct correla-tion to H E . Element specific x-ray magnetic studies ofFeF /Co and CoO/Fe layered structures confirmedthe existence of this M E shift and revealed its relation tospecific UCM in the AFM material.Due to the limited number of studies on the M E ef-fect it is of general interest to find systems with largermagnetization shifts, not only because of its fundamen-tal interest but also because this shift provides a newdegree of freedom in the hysteresis loop that may be wellhave some applicability in future devices. In this workwe studied the exchange-bias shifts H E and M E of thehysteresis loops as a function of temperature T and H FC for three AFM/FM bilayers having the same AFM layerbut different thickness and dilution of the FM layer. Weobserved an unusually large uncompensated magnetiza- tion shift M E that is not simply correlated with H E anddoes not originate only from the AFM layer but from theFM one. II. SAMPLE PREPARATION DETAILS ANDX-RAY CHARACTERIZATION
We prepared bilayers composed of a FMLa . Sr . MnO (LSMO) layer (selected for itsweak anisotropy and small coercivity) covering anAFM orthorhombic o-YMnO (YMO) layer grown on(100) SrTiO substrates of area 5 × for samples Aand B and 6 × for sample C. For the depositionsa KrF excimer laser (wavelength 248 nm, pulse duration25 ns) was used and the optimal parameters foundfor o-YMO were 1.7 J/cm with 5 Hz repetition rate,800 ◦ C and 0.10 mbar for the substrate temperatureand oxygen pressure during preparation. We havemeasured three bilayers, all of them with the same375 nm thick o-YMO layer on STO substrates preparedalways under the above mentioned conditions. Tocheck the reproducibility of the found effects we haveprepared a fourth bilayer with identical thickness as insample A but instead of the LSMO FM layer we usedLa . Ca . MnO (LCMO) deposited YMO and thislast one on a (100)LSAT substrate.For the FM LSMO layer, deposited immediately afterthe o-YMO one, the parameters were 10 Hz repetitionrate and 0.35 (0.38) mbar oxygen pressure, 8 (30) nmthickness and at the same laser fluency and substratetemperature, for sample A (B). In order to corroboratethe contribution of the FM layer in the M E -shift we havedecreased further the oxygen concentration to deposit theLSMO film in sample C (oxygen pressure 0.10 mbar) witha larger thickness of 225 nm decreasing in this way itscoercivity. For the fourth LCMO/YMO bilayer the YMOlayer was grown under similar conditions as before butthe LCMO layer under an oxygen pressure of 0.55 mbar;all other conditions as for the LSMO layers.The epitaxial growth in the 00 l direction for the o-YMO and l
00 for LSMO phases was confirmed by x-raydiffraction using Cu-K α line. As an example we show inFig. 1 the the x-ray spectrum of the single o-YMO layeron STO. The preferential growth of the (00 l ) planes ofthe orthorhombic phase YMO is clearly seen. Within theexperimental resolution no maxima due to the hexagonalphase are observed. Figure 2 shows the x-ray spectrumobtained for sample B. The main diffraction peaks fromthe LSMO layer are observed as a weak shoulder near theSTO main maxima. Magnetization measurements wereperformed with a superconducting quantum interferencedevice (SQUID) from Quantum Design in the tempera-ture range between 5 K and 350 K.In addition, we performed soft x-ray absorption andcircular dichroism measurements using the bending mag-net beamline 6.3.1 at the Advanced Light Source inBerkeley, CA (USA) and the elliptical undulator beam-line 13.1 at the Stanford Synchrotron Radiation Light-source, Stanford, CA (USA). For these measurements thesample was mounted between the poles of an electromag-net so that the x-rays are incident on the sample undera grazing angle of 30 ◦ parallel to the direction of the ap-plied magnetic field. The x-ray absorption intensity wasmonitored using the electron yield method. Hysteresisloops were acquired by sweeping the external field whilemonitoring the electron yield at the Mn L and L ab-sorption resonance ( ≈
640 eV). This approach is surfacesensitive and in general it yields information only on thefirst ∼ III. RESULTSA. Single YMnO layers According to literature the o-YMO phase is AFMwith N´eel temperature T N = 42 ± ∼
31 K. In spite of its low T N thismaterial has several advantages for exchange bias stud-ies. It belongs to the family of the perovskite manganiteRMnO and the magnetic and electrical properties canbe changed by cation substitution keeping similar lat-tice constants and therefore without drastic changes inits structural properties. On the other hand, o-YMO isa phase that was not thoroughly studied yet and the in-fluence of its ferroelectric behavior, in spite of the lowtemperature, might be used as a paradigm for potentialapplications in magnetoelectric devices. Figure 3(a) shows the magnetization loop of single o-YMO layer. The hysteresis loop indicates a magnetiza-
20 30 40 5010 Y M O ( ) Y M O ( ) S T O ( ) K + K S T O ( ) K + K I ( c p s ) S T O ( ) K FIG. 1. X-ray spectrum of the single YMO AFM layer onSTO substrate.
20 30 40 50 L S M O ( ) L S M O ( ) Y M O ( ) Y M O ( ) S T O ( ) K + K S T O ( ) K + K I ( c p s )
2 (degree) S T O ( ) K FIG. 2. X-ray spectrum of the bilayer sample B. The labelsindicate the corresponding the main diffraction peaks. tion at saturation of 1.8 emu/cm at 5 K and at appliedfields µ H > . Figure 3(b) shows the magnetic moment of a single o-YMO layer (6 × × .
375 10 − mm ) on STO measuredas a function of temperature in ZFC and FC states atan applied field of 0.05 T. A clear increase in m ( T ) de-creasing temperature is observed at T ≃
42 K. An hys-teresis between ZFC and FC is observed already below T ∼
60 K. As was shown in earlier studies on YMO wemay expect to have persistent spin waves at temperaturesabove T N . From the hysteresis loop shown in Fig. 3(a) one may -1.0 -0.5 0.0 0.5 1.0 -2.4-1.8-1.2-0.60.00.61.21.82.4
Applied Field (cid:181) o H (T) M agne t i z a t i on M ( e m u / c m ) YMO on STOT = 5 K (a) (b) M agne t i c M o m en t m ( (cid:181) e m u ) Temperature T(K)
YMO on STO (cid:181) H = 0.05 T
FIG. 3. (a) Hysteresis loop of the magnetization at 5K forthe 375 nm thick YMO layer on STO. The error bars indi-cate the maximum error due to the SQUID and geometrymeasurements. (b) Temperature dependence of the magneticmoment of a single YMO layer on STO in ZFC and FC statesat an applied field of 0.05 T. An error bar of ± . µ emu is theexpected maximum error from our SQUID measurements. speculate that the YMO film behaves as a ferro- or ferri-magnet and not as an antiferromagnet. In fact, a recentstudy suggests a change of the usual bulk antiferromag-netic state to a strain-dependent non-collinear magneticone in thinner ( .
120 nm) o-YMO films. Taking intoaccount that our YMO layers are much thicker and showa different m ( T ) behavior (at ZFC and low applied fieldsthe measured m ( T ) of our YMO films alone resemblespractically the usual T − dependence found for antiferro-magnets) as those reported in Ref. 18 we remark that themagnetic behavior of the o-YMO layers may correspondto the one observed in diluted antiferromagnets in exter-nal magnetic field (DAFF). It is well known that DAFFdevelop a domain state when cooled below T N (some-times with a spin-glass-like behavior) and this leads toa net magnetization, which couples to the external field,see e.g. Refs. 4, 7, 19–21.From the measured temperature dependence of themagnetic moment and the observed scaling of the ex- change bias field H E with the inverse of the thickness ofthe LSMO layer for samples A and B, see section III B,and the quantitative agreement of the obtained H E and M E shifts for the fourth sample (similar to sample A butwith LCMO instead of LSMO) we may conclude thatYMO behaves as an AFM or DAFF layer for the ex-change bias effects. Whatever the real magnetic equilib-rium state of our o-YMO films is, we may expect to seeexchange bias effects when these films are coupled to aferromagnet. Further examples for exchange bias effectsin heterostructures with different ferro- or ferrimagnetscan be seen in Refs. 22 and 23 and H E effects, positive aswell as negative, has been also observed in ferrimagneticbased bilayers. B. La . Sr . MnO /YMnO bilayers Figure 4 shows the remanent moment for samples Aand B measured increasing temperature at zero field, af-ter cooling them to 5 K in a field of 0.1 T applied in-plane, i.e. a or b direction. Changes in slope of theremanence moment are observed near the N´eel tempera-ture onset T N ∼
50 K of the o-YMO layer. This increaseof ∼ T N might be related to the an exchange-bias or strain effect. An anomaly is also observedat T ∼
20 K, as shown in Fig. 3(b), and already reportedin the literature.
The temperature dependence of theremanence measured in sample B shows a clear change ofslope near the Curie temperature of the LSMO layer. Incontrast, due to the smaller LSMO thickness the rema-nent moment of sample A does not show a clear anomalyat T C ; similarly for sample C (not shown). For sample Cwe show in Fig. 4 the field cooled (FC) curve at 0.1 T;the absence of a marked anomaly at T C and the smoothdecrease of the magnetic moment with T demonstratesthe expected strong magnetic dilution of the LSMO film.The existence of the FM state in this layer was confirmedthrough hysteresis loop measurements up to its ferromag-netic onset at T C ∼
300 K. The FC results presented be-low were obtained always after cooling the samples from
T > T C at zero field and after applying an in-plane field H FC at 100 K > T N .Figures 5(a) and (b) show the hysteresis loops for ZFCand FC measurements at 5 K for samples A and B. Aremarkable M E shift of the same order of the saturationmagnetic moment m s is observed for sample A after FCfrom 100 K at µ H FC = 0 . M E shift is also clearer measured but it is smaller relative to m s . The sign of the M E -shift changes when the direc-tion of H FC changes, i.e. it has the same sign as thatof H FC . This indicates that the effective UCM layer ispinned in the direction of the cooling field, which meansa ferromagnetic coupling.In the determination of the M E and H E shifts we tookspecial care to rule out effects due to minor hysteresisloops . Studying the behavior of the loops at different H FC we conclude that no minor loops and a clear sat- Temperature T(K)
FC- Sample C
Sample A
Sample B M agne t i c m o m en t m ( (cid:181) e m u ) FIG. 4. Temperature dependence of the remanence for sam-ples A and B measured at zero field after cooling them to 5 Kat 0.1 T in-plane field. Also shown is the field cooling curveat 0.1 T for sample C. Note the difference scales of the y − axisfor each sample. uration behavior of the magnetic moment are obtainedfor µ H FC ≥ . T ≥ µ H FC < . µ H FC ≥ . H c vs. H FC indicates (see Fig. 7 below).We note that the value of m s obtained from the hys-teresis loops depends on the applied H FC . As examplewe show this effect for sample B where the hysteresis loopwas measured after cooling the sample at µ H FC = 2 T,see Fig. 5(b). This effect is due to the LSMO layer andindicates that the number of aligned domains can bechanged with H FC . In this case we expect that the M E effect will be strongly influenced by the FM layer since,as in the case of a diluted AFM layer , the formationand number of its domains that take part in the exchangebias coupling with the AFM layer can be enhanced lead-ing to an increase of M E . Note however that the M E effect is expected to decrease with H FC , i.e. M E → H FC → ∞ .Note the opening of ∼ µ emu of the hysteresis at theend of the loop at 0.5 T for sample A, see Fig. 5(a).A similar opening is measured for all samples in agree-ment with the numerical results obtained with the do-main state model for exchange bias proposed by Nowaket al. . The fact that the loops do not close indi-cates that uncompensated spins - pinned earlier duringthe field cooling - rotate and remain pinned in the op-posite direction during the field sweep loop, reducing thefinal saturation moment. We note that in all three bi-layers this opening remains of the order of 1 . . . µ emu,i.e. several times smaller than the M E shift, as we show -0.4 -0.2 0.0 0.2 0.4-300-200-1000100200300-20-1001020 M agne t i c m o m en t m ( (cid:181) e m u ) Applied Field (cid:181) o H (T)
Sample B ZFC FC 0.3T FC 2T
Sample A ZFC FC 0.5T (a) (b)
FIG. 5. Hysteresis loops at 5 K measured for samples A(a) and B (b) after zero-field cooled (ZFC) and field cooled(FC) states at the fields shown in the figures. The arrowsindicate the sweeping field direction starting the loop alwaysfrom positive fields. below.To characterize quantitatively the exchange bias M E effect and for a direct comparison with the saturationmagnetic moments of each of the layers we define it as m shift = ( m + s + m − s ) /
2, where m + s and m − s are the satura-tion moments at positive and negative fields. The shift inthe field axis is defined as H E = ( H + c + H − c ) /
2, where H + c and H − c are the coercive fields in upward and descendingloop branches, respectively. We note that the H E valueswere obtained only after centering the hysteresis loop,subtracting the upward M E shift.Figure 6 shows the coercivity H c (a), the exchange-bias H E (b) and the vertical shift in magnetic moment m shift (c) as a function of T
80 K for sample B, measuredafter µ H FC = 0 . H C and H E showan anomaly at T .
20 K, in agreement with the behav-ior found in the remanence curve, see Fig. 4, suggestingthat the transition at that temperature influences the ex-change interaction. At T &
35 K H E crosses zero andchanges to positive. This sign change of H E from nega-tive to positive increasing temperature was observed alsoin CoO/Co bilayers and suggests a change from direct( J interface >
0) to indirect ( J interface <
0) interface inter- (a)
Temperature T(K) H c ( O e ) H E ( O e ) m s h i ft ( (cid:181) e m u ) (c)(b) FIG. 6. Temperature dependence of the coercivity (a) andexchange bias (b) fields and of the shift in magnetic moment m shift due to the M E effect (c) for sample B after cooling itin a field of 0.3 T. action. As expected, H E ( T ) as well as m shift vanish at T & T N . In contrast to H E ( T ) no anomalous behavior isobserved in m shift ( T ) at T < T N , with exception of theslope change at T ∼
20 K, see Fig. 6(c).Figure 7 shows the H FC -dependence of H c , m shift and H E for the three samples measured at 5 K. The decreaseof H E from samples A to B agrees with the expectedinverse proportionality of H E with the thickness of theFM layer. According to this thickness dependence sam-ple C should show nearly one order of magnitude smaller H E than for sample B, in clear disagreement with theobtained result, see Fig. 7(c), suggesting that the mag-netic dilution of this sample is responsible for the largeobserved H E field.Regarding the M E effect and in agreement with the re-sults in Co/CoO bilayers we observe a vanishing effectat zero and at large enough values of H FC , see Fig. 7(b).Under the assumptions done in Refs. 9 and 19 the M E shift is mainly due to the AFM layer. According to thismodel, the largest m shift expected from our o-YMO layer,assuming complete saturation in the whole 375 nm thicklayer, would be m YMO = 17 µ emu and 24 . µ emu fromsamples A or B and C, respectively. To estimate thosenumbers we have taken into account the measured mag-netization at saturation of the single layers. The normal-ized m shift by the corresponding m YMO , see Fig. 7(b), (a)CA B H c ( k O e ) (c) H E ( O e ) (cid:181) H FC (T) B C B m s h i ft / m Y M O A (b) FIG. 7. Dependence of the coercive field (a), shift in magneticmoment m shift (b) and exchange bias field H E (c) on thecooling field H FC for the three measured bilayers at 5 K. In(b) we plot m shift normalized by the maximum saturationmoment m YMO of the o-YMO layer, i.e. m YMO = 17 µ emu forsamples A and B and 24 . µ emu for sample C. Note that thevalues of m shift ∼ H FC = 0 were obtained using maximumfields between 0.3 and 0.5 T for the hysteresis loops. For allthe other points the maximum field of the loops coincides with H FC . would indicate that it is necessary that 50% to 70% ofthe YMO layer should be responsible for the measured m shift at H FC ∼ . . ≤ H FC ≤ . / Ca / MnO layer, which ori-gin remains unclear. In our case the large m shift values –actually a giant M E effect – indicate that a large contri-bution should come from the FM layer. Taking into ac-count the saturation moments of the LSMO layers alone,we estimate for example that a thickness of the LSMOlayer of less than 1.3 nm for sample B and <
10 nmfor sample C should be enough to produce the observed m shift at H FC = 0 . C. La . Ca . MnO /YMnO bilayer Further evidence for the reproducibility and robustnessof the effects observed in the three LSMO/YMO bilayersreported in the last section are provided by the results ofa LCMO/YMO bilayer with similar geometry and prepa-ration conditions as sample A. Figure 8(a) shows the re-manent magnetic moment of this bilayer after cooling thesample at 1 T applied field. The transition at the N´eeltemperature of the YMO layer is clearly seen as well asthe change of slope at ∼
20 K. In Fig. 8(b) the hysteresisloops for three field cooled states at fields H FC = ± H F C fields the exchangebias M E - and H E -effects are clearly observed whereas athigh enough fields the M E effect vanishes, see Fig. 8(b).Figure 9 shows the H FC -dependence for the three charac-teristics parameters. The observed m shift at H FC . H c ( H FC )and m shift ( H FC ), see Fig. 7, one may expect some cor-relation between them in case of a bilayer with a verythin (and diluted) FM layer. This may be so if we takeinto account the amount of the FM layer that remainspinned at the interface. In this case the smaller the ef-fective thickness of the remained unpinned ferromagneticlayer the smaller might be H c . Apparently this is ob-served in the (thin)LCMO/(thick)YMO bilayer. Indeed,the results shown in Fig.9 indicate that when m shift de-creases at H FC > . H c increases. IV. DISCUSSION AND CONCLUSION
To further corroborate our conclusion that the ob-served vertical shift is mainly due to the FM and its inter-face region with the AFM layer we show the hysteresisloops acquired using x-ray magnetic circular dichroismin Fig. 10. For sample A we find a shift of about 5%using the surface sensitive approach measuring the re-sponse of the Mn ions within the LSMO FM layer only.The observed vertical shift is a clear indication that theFM layer is contributing to the M E effect and that the R e m anen t m agne t i c m o m en t m ( (cid:181) e m u ) Temperature T(K)
LCMO/YMO/LSAT (a) -2 -1 0 1 2 -150-100-50050100150 M a gn e t i c m o m e n t m ( e m u ) Applied Field H (T) H FC = 2 T H FC = 1 T H FC = -1 T T = 5 K (b)
FIG. 8. (a) Temperature dependence of the zero field rema-nent magnetic moment measured after field cooled at 1 Tof a bilayer La . Ca . MnO (8 nm) / YMnO (375 nm),similar to sample A, but the YMO layer first deposited on a(100)LSAT substrate. (b) Hysteresis loops at 5 K measuredfor the same sample after field cooled (FC) at the fields shownin the figure. shift is not confined to the bulk of the AFM. Assum-ing that 95% of the secondary electrons detected in ourexperiment originate from the top 6 nm we can con-clude that the interfacial region of the FM/AFM layercontributes significantly more to the m shift compared tothe surface layers of the FM. This result agrees with theestimate from the bulk SQUID measurements that oneneeds about 1 nm thick FM layer (for samples A as wellas B) to account for the observed m shift . Taking intoaccount the previous statement that it is highly unlikelythat the entire AFM bulk contributes to the shift we canconclude that the excess magnetization is produced pre-dominately at the FM interface during the field coolingprocess due to interfacial exchange coupling between theAFM and the FM as shown previously for the case ofCo/FeF .Using similar arguments on the importance of the mag-netic dilution of the AFM layer , we argue that in oursystem the dilution of the FM layer may play a mayorrole in the M E shift. In other words, the robust AFM -300-200-1000 (cid:181)oH FC (T) H C ( k O e ) (a) H E ( O e ) (c)(b) m s h i ft / m Y M O FIG. 9. Similar to Fig. 7 but for the YMO/LCMO bilayer:Dependence of the coercive field (a), shift in magnetic moment m shift normalized by the maximum saturation moment m YMO of the o-YMO layer alone (b), and exchange bias field H E (c),on the cooling field H FC at 5 K. layer influences the magnetic behavior of the FM one,within a certain thickness from the interface. We notethat some kind of M E -shift were recently reported for fer-rimagnetic very thin hard/soft (3nm/12nm) DyFe /YFe heterostructures. However, in that work the M E effectis in opposite direction to that of the applied H FC , incontrast to our observations.Furthermore, a comparison between the overall behav-ior obtained for m shift ( H FC ) and H E ( H FC ) indicates thatthere is no simple correlation between the two exchangebias effects. Note that H E decreases strongly from sam-ple A to B, whereas m shift increases. Although elementselective x-ray magnetic measurements would help to de-termine the penetration depth of the UCM in each of thelayers, it is clear from our SQUID measurements that theo-YMO layer alone cannot be the reason for the observedgiant M E effect, this is the main message of our work.In conclusion, our studies on LSMO/o-YMO bilayersand on a single LCMO/o-YMO bilayer found large un- compensated M E shifts, whose sign correlates with thedirection of the cooling field H FC . Both, the exchange-bias H E and M E effects, vanish near T N of the YMO -1.0 -0.5 0.0 0.5 1.0-1.5-1.0-0.50.00.51.01.5 X M CD d i ff e r en c e [ a r b . un i t s ] Applied Field (cid:181) H (T) -0.5 T +0.5 T
FIG. 10. Hysteresis loops of sample A acquired at 15 K aftercooling in a field of either +0.5 T or -0.5 T using x-ray mag-netic circular dichroism and the Mn L-absorption resonance.The loops exhibit a horizontal loop shift H E of 140 Oe as wellas a vertical shift m shift ≃
5% of the saturation value. layer. The large m shift values indicate that the AFM layercannot be the only responsible but a certain thickness ofthe FM layer near the interface. This behavior can beactually understood taking similar arguments as thoseused for the AFM layer in the domain state exchange-bias model of Refs. 9 and 19. Tuning the thickness andmagnetic dilution of the FM layer one should be ableto obtain large M E shifts making it an effect worth tostudy in systems with T N >
300 K. The different behav-iors of H E and M E with temperature, cooling field andFM layer thickness indicate that these two phenomenaare not correlated in a simple way. ACKNOWLEDGMENTS
We thank M. Ziese for fruitful discussions and forpointing us Ref. 29 and A. Setzer for technical assistance.This work was supported by the DFG within the Col-laborative Research Center (SFB 762) “Functionality ofOxide Interfaces”. One of us (C.Z.) was supported bythe S¨achsisches Staatsministerium f¨ur Wissenschaft undKunst under 4-7531.50-04-0361-09/1. SSRL and ALS arenational user facilities supported by the Department ofEnergy, Office of Basic Energy Sciences. SSRL is oper-ated by Stanford University and ALS is operated by theUniversity of California. ∗ Permanent address: FaMAF-CLCM at University of Cor-doba, Medina Allende S/N, 5000 Cordoba, Argentina. † [email protected] ‡ Present address: Institut Catala de Nanotecnologia (ICN),Universitat Autonoma de Barcelona, E-08193 Bellaterra,Spain W. Meiklejohn and C. Bean, Phys. Rev. , 904 (1957). J. Nogu´es and I. K. Schuller, J. Magn. Magn. Mat. ,203 (1999). V. Skumryev, S. Stoyanov, Y. Zhang, G. Hadjipanayis,D. Givord, and J. Nogu´es, Nature (London) , 850(2003). P. Milt´enyi, M. Gierlings, J. Keller, B. Beschoten,G. G¨untherodt, U. Nowak, and K. D. Usadel, Phys. Rev.Lett. , 4224 (2000). J.-I. Hong, T. Leo, D. J. Smith, and A. E. Berkowitz, Phys.Rev. Lett. , 117204 (2006). J. Nogu´es, C. Leighton, and I. K. Schuller, Phys. Rev. B , 1315 (2000). J. Keller, P. Milt´enyi, B. Beschoten, G. G¨untherodt,U. Nowak, and K. D. Usadel, Phys. Rev. B , 014431(2002). M. Cheon, Z. Liu, and D. Lederman, J. Appl. Phys. ,095E503 (2007). U. Nowak, K. D. Usadel, J. Keller, P. Milt´enyi,B. Beschoten, and G. G¨untherodt, Phys. Rev. B ,014430 (2002). H. Ohldag, A. Scholl, F. Nolting, E. Arenholz, S. Maat,A. T. Young, M. Carey, and J. St¨ohr, Phys. Rev. Lett. ,017203 (2003). H. Ohldag, H. Shi, E. Arenholz, J. St¨ohr, and D. Leder-man, Phys. Rev. Lett. , 027203 (2006). M. Gruyters and D. Schmitz, Phys. Rev. Lett. , 077205(2008). C. C. Hsieh, T. H. Lin, H. C. Shih, C.-H. Hsu, C. W. Luo,J.-Y. Lin, K. H. Wu, T. M. Uen, and J. Y. Juang, J. Appl.Phys. , 103912 (2008). J. Kim, S. Jung, M. S. Park, S.-I. Lee, H. D. Drew,H. Cheong, K. H. Kim, and E. J. Choi, Phys. Rev. B ,052406 (2006). V. Laukhin, V. Skumryev, X. Mart´ı, D. Hrabovsky,F. Sanchez, M. Garc´ıa-Cuenca, C. Ferrater, M. Varela,U. L¨uders, J. F. Bobo, et al., Phys. Rev. Lett. , 227201(2006). S. Z. Li, Z. B. Yan, T. Wei, S. J. Luo, B. Liu, K. F. Wang,and J. M. Liu, Appl. Phys. A , 975 (2009). F. Demmel and T. Chatterji, Phys. Rev. B , 212402(2007). X. Marti, V. Skumryev, A. Cattoni, R. Bertacco,V. Laukhin, C. Ferrater, M. V. Garc´ıa-Cuenca, M. Varela,F. S´anchez, and J. Fontcuberta, J. Magn. Magn. Mater. , 1719 (2009). U. Nowak, A. Misra, and K. D. Usadel, J. Magn. Magn.Mat. , 243 (2002). F. Radu, M. Etzkorn, R. Siebrecht, T. Schmitte, K. West-erholt, and H. Zabel, Phys. Rev. B , 134409 (2003). F. Radu and H. Zabel,
Magnetic Heterostructures , vol. 227of
STMP (Springer Berlin, Heidelberg, 2008). M. R. Fitzsimmons, C. Dufour, K. Dumesnil, J. Dou, andM. Pechan, Phys. Rev. B , 144425 (2009). X. Ke, M. S. Rzchowski, L. J. Belenky, and c. B. Eom,Appl. Phys. Lett. , 5458 (2004). F. Canet, S. Mangin, C. Bellouard, M. Pi´ecuch, andA. Schuhl, J. Appl. Phys. , 6916 (2001). J. van Lierop, K.-W. Lin, J.-Y. Guo, H. Ouyang, and B. W.Southern, Phys. Rev. B , 134409 (2007). P. J. van der Zaag, Y. Ijiri, J. A. Borchers, L. F. Feiner,R. M. Wolf, J. M. Gaines, R. W. Erwin, and M. A. Ver-heijen, Phys. Rev. Lett. , 6102 (2000). H. Shi, D. Lederman, K. V. O’Donovan, and J. A.Borchers, Phys. Rev. B , 214416 (2004). I. Fina, X. Mart´ı, L. F`abrega, F. S`anchez, and J. Fontcu-berta, Thin Solid Films (2010), to be published. L. Pi, S. Zhang, S. Tah, and Y. Zhang, Appl. Phys. Lett. , 102502 (2006), see also the comment by L. Klein, idem , 036101 (2006) and the Erratum by the authors in idem , 039902 (2006). A. Hoffmann, S. J. May, S. G. E. te Velthuis, S. Park,M. R. Fitzsimmons, G. Campillo, and M. E. G´omez, Phys.Rev. B80