JOURNAL OF APPLIED PHYSICS 99, 08C903 共2006兲

Effect of ferromagnetic/antiferromagnetic interfaces on the magnetic properties of La2/3Sr1/3MnO3 / Pr2/3Ca1/3MnO3 superlattices D. Niebieskikwiata兲 Department of Physics and Frederick Seitz Materials Research Laboratory, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801

L. E. Hueso Department of Materials Science, University of Cambridge, Cambridge CB2 3QZ, United Kingdom

M. B. Salamon Department of Physics and Frederick Seitz Materials Research Laboratory, University of Illinois at Urbana-Champaign, Urbana, Illinois 61801

N. D. Mathur Department of Materials Science, University of Cambridge, Cambridge CB2 3QZ, United Kingdom

共Presented on 31 October 2005; published online 21 April 2006兲 The magnetic properties of a series of ferromagnetic/antiferromagnetic 共FM/AFM兲 superlattices are studied. The multilayers are composed of five La2/3Sr1/3MnO3 / Pr2/3Ca1/3MnO3 共LSMO/PCMO兲 bilayers, grown by pulsed laser deposition using SrTiO3 as the substrate. The evolution of the magnetic moment with the AFM PCMO layer thickness 共0 艋 tA 艋 7.6 nm兲 is analyzed and is found to be strongly affected by the strain fields introduced by the FM/AFM interfaces. The effect on the LSMO is the appearance of a magnetic dead layer of ⬃1.5 nm, close to the interface with the PCMO. On the contrary, the magnetic moment of PCMO is remarkably enhanced for thicknesses tA around 3.5 nm, where nanometric FM droplets would find the optimal strain conditions to be accommodated within the AFM PCMO spacer. The implications of these findings on the magnetoresistance response are discussed. © 2006 American Institute of Physics. 关DOI: 10.1063/1.2170058兴 During the last decade, colossal magnetoresistance has been among the most studied physical phenomena in condensed-matter physics.1,2 In the celebrated manganites of nominal composition Ln1−xAxMnO3 共Ln= La, Pr, Nd; A = Sr, Ca, Ba兲, a wide variety of electronic phases emerges due to the complexity of the interactions among the spin, charge, orbital, and lattice degrees of freedom.1,2 In particular, for a fixed doping level x ⬃ 1 / 3, ferromagnetic 共FM兲 metallic, FM insulating, and charge-ordered/antiferromagnetic 共CO/AFM兲 phases can be accomplished by appropriately replacing the constituent cations. In the most interesting cases, such as, for example, Pr0.7Ca0.3MnO3, all these phases can occur at the same time,3 coexisting in a magnetic state known as electronic phase separation 共PS兲.2 Because of the large resistivity of the CO/AFM state, large magnetoresistance values are obtained when this phase is transformed into the FM metallic one through the application of a magnetic field. However, in bulk materials, very large magnetic fields are usually required to produce such an insulator-to-metal 共IM兲 transition.1 On the other hand, magnetoresistive devices consisting of FM-metallic/ nonmagnetic-insulating multilayers have been built to exploit the tunneling magnetoresistance phenomenon.4,5 However, the poor magnetic properties at the interfaces limit the applicability of these devices. In this context, some moderately successful attempts have been made in order to fabria兲

Author to whom correspondence should be addressed; electronic mail: [email protected]

0021-8979/2006/99共8兲/08C903/3/$23.00

cate artificial heterostructures with the desired magnetotransport properties. These structures consist of all-manganite FM-metallic/AFM-insulating superlattices,6–9 where the internal field of the FM material would help to produce an IM transition in the AFM spacer, and the manganite-manganite interfaces would ideally keep the required magnetic properties. The previous studies have been mainly focused on the transport properties of these samples.6–9 In the present work we center our attention on the effect of the FM/AFM interfaces as well as the effect of the thickness of the AFM layers on the magnetic properties of La2/3Sr1/3MnO3 / Pr2/3Ca1/3MnO3 multilayers. High-quality epitaxial superlattices made of La2/3Sr1/3MnO3 共LSMO兲 and Pr2/3Ca1/3MnO3 共PCMO兲 were grown on atomically flat SrTiO3 共001兲 cubic substrates by pulsed laser deposition. A KrF laser 共repetition rate of 1 Hz and fluence of 2.0 J / cm2兲 was used to ablate the polycrystalline LSMO/PCMO targets 共Praxair, USA兲. The films were grown at a substrate temperature of 750 ° C in a flowing oxygen ambient of 15 Pa. After deposition, the films were annealed at the same temperature in 60 kPa of oxygen for 1 h. Multilayers involving five repetitions of LSMO/PCMO bilayers were grown in situ using a multistage carousel, using always LSMO as the starting layer. We prepared seven different films, where the thickness of the LSMO layers is in all cases 15 nm 共a stacking of ⬃40 MnO6 octahedra兲, while the thickness of the PCMO is tA = 0, 0.8, 1.7, 2.8, 3.5, 5.1, and 7.6 nm, respectively 共for tA = 0, a single 15 nm layer of LSMO was grown兲.

99, 08C903-1

© 2006 American Institute of Physics

Downloaded 04 May 2006 to 192.167.164.69. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

08C903-2

Niebieskikwiat et al.

FIG. 1. X-ray-diffraction patterns 共␻-2⌰ scans兲 for selected LSMO/PCMO multilayers, showing the typical superlattice reflections of high-quality films.

X-ray-diffraction analysis using a four-angle diffractometer shows very clear superlattice reflections in the ␻-2⌰ scans 共see Fig. 1兲, indicating epitaxial growth and the high quality of the films. For all the studied multilayers, the calculated out-of-plane lattice parameters for LSMO and PCMO are 3.85 and 3.76 Å, respectively, independent of the thickness of the PCMO layers. These values are smaller than the bulk values 共3.88 and 3.83 Å, respectively兲, confirming the tensile stress imposed by the substrate 共the in-plane lattice parameter is 3.90 Å for all SrTiO3, LSMO, and PCMO兲. Magnetization 共M兲 measurements were performed in a superconducting quantum interference device 共SQUID兲 magnetometer for temperatures T between 5 and 355 K and magnetic fields H of up to 70 kOe. The magnetic field was applied parallel to the plane of the films. The M共T兲 data of the superlattices show the FM transition of the LSMO layers at TC ⬃ 345 K. In bulk samples, Pr2/3Ca1/3MnO3 is a CO/AFM insulator with the chargeordering transition at TCO ⬃ 220 K and the magnetic transition to the AFM state at TN ⬃ 140 K.1 On the contrary, it is well known that in manganite thin films the CO state is not realized.7,8,10 Due to the strong electron-phonon coupling, strain fields play a fundamental role in the CO state. Therefore, the absence of charge ordering is likely to be related to the lattice strain and interface clamping in thin-film samples. However, the AFM phase has been shown to be robust enough, and it can be observed even in thin films.6,10,11 In our multilayers, the dominant FM contribution that develops at TC hinders the observation of the AFM transition of the PCMO layers at lower temperatures 共only a very minor hint is observed at ⬃136 K for the samples with thicker PCMO sheets兲. Although at low T all the studied multilayers present a similar FM behavior, as shown in Fig. 2, a quantitative analysis of the magnetization data reveals an important evolution of the magnetic moment with the PCMO thickness. In Fig. 3共a兲 we show the spontaneous magnetic moment M 0 共the extrapolation to zero field from the high-field region兲 as a function of tA. For the pure LSMO sample 共tA = 0兲, the

J. Appl. Phys. 99, 08C903 共2006兲

FIG. 2. Magnetization vs magnetic-field loops at T = 5 K for the LSMO/ PCMO multilayers, for selected PCMO thicknesses, as labeled. Upper and lower insets: Low-field region of the M vs H loops for the samples with tA = 1.7 and 7.6 nm, respectively. These loops were measured after cooling the samples with zero field 共ZFC兲 and with an applied magnetic field of 70 kOe 共FC兲.

M 0 ⬃ 9.8⫻ 10−4 emu/ cm2 corresponds to ⬃3.9 ␮B / Mn, close to the expected saturation for LSMO. However, as soon as the PCMO layers are intercalated the magnetic moment rapidly drops to ⬃8.0⫻ 10−4 emu/ cm2 per bilayer, and M 0

FIG. 3. 共a兲 Spontaneous magnetic moment at 5 K of the La2/3Sr1/3MnO3 / Pr2/3Ca1/3MnO3 superlattices as a function of the PCMO thickness. 共b兲 Magnetic moment at 5 K of the PCMO layers only in the LSMO/PCMO multilayers as a function of the PCMO thickness.

Downloaded 04 May 2006 to 192.167.164.69. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp

08C903-3

J. Appl. Phys. 99, 08C903 共2006兲

Niebieskikwiat et al.

remains always below the pure LSMO value, independent of the increase in the amount of PCMO in the superlattices. It is clear that an AFM coupling between different LSMO layers cannot be responsible for this behavior, since at high magnetic fields a rotation of the opposite layers is not observed at all, and full saturation is not reached even for the samples with thicker PCMO layers. Therefore, it is the inclusion of the LSMO/PCMO interfaces that is inducing the significant decrease of the magnetic moment in the LSMO layers. Obviously, magnetization measurements cannot distinguish between a surface effect and a homogeneous reduction in the magnetic moment of LSMO. However, the M共H兲 curves do not show any sign of exchange bias when the samples are cooled with an applied magnetic field of 70 kOe 共see insets in Fig. 2兲. This likely indicates the presence of a magnetically dead layer in the LSMO at the interface with the PCMO layers, where the magnetization is suppressed. From the moment reduction 共⬃20% 兲 we estimate a thickness of ⬃1.5 nm 共4 f.u.兲 for the magnetically dead layer. Indeed, suppression of ferromagnetism at similar length scales has also been observed in other systems,5,12,13 and may be related to strain, disorder, or oxygen deficiency. Our most relevant result shows that for increasing amounts of PCMO material the magnetic moment increases again and exhibits a peak for tA ⬃ 3.5 nm. Since the thickness of LSMO is kept constant, this peak is related to the changes in the magnetic contribution of the PCMO layers. In Fig. 3共b兲 we present the spontaneous magnetic moment M * of the PCMO only, i.e., after the contribution from the LSMO 共⬃8 ⫻ 10−4 emu/ cm2 per bilayer兲 was subtracted from M 0. At low temperatures pure PCMO samples, either bulk or thin films, display a weak FM moment 共⬍0.5 ␮B / Mn兲.9,14,15 This moment is related to nanometric FM domains that appear at ⬃115 K and occupy a small fraction of the volume 共⬃10% 兲.1–3,9,14–16 These domains coexist with the predominant AFM background, i.e., phase separation occurs. Even though the PCMO in the superlattices also exhibits a spontaneous moment, its evolution with tA denotes a remarkable difference with respect to the pure PCMO samples. For small tA, the moment of PCMO is ⬃0.1– 0.4 ␮B / Mn. As tA increases above ⬃2 nm the magnetic moment notably increases, reaching over 2 ␮B / Mn for tA = 3.5 nm and decreasing again for larger tA. This peak in the magnetic moment has a direct correlation with the observation of an optimal thickness of the AFM insulating spacer for the magnetoresistance 共MR兲 response of similar superlattices.7 In La2/3Ca1/3MnO3 / Pr2/3Ca1/3MnO3 multilayers, the optimal MR thickness of PCMO is ⬃2 nm. In this kind of system, the enhanced MR was shown to be induced by the metallization of the insulating spacer.7–9 In our multilayers, the increase of the magnetic moment demonstrates the same tendency towards the metallization of the PCMO spacer. In phase-separated compounds, the MR phenomenon occurs because of the expansion of the FM metallic domains when the magnetic field is applied.2,15 The enlarged metallic volume then forms conductive paths, producing a large drop of the electrical resistivity. Similarly for the multilayers, for

thicknesses tA ⬃ 3 nm where the metallic phase is favored, the application of a magnetic field could be responsible for building the percolative paths in the PCMO spacer, connecting the LSMO metallic layers at both sides. The goal of these LSMO/PCMO multilayers is to change the magnetic properties of the PCMO insulating spacer, such that for the optimal thickness the FM metallic phase is greatly enhanced. But the open question is what is the reason for such increase in the magnetic moment of the PCMO? The proximity effect of the FM LSMO layers could certainly act as an internal field.6,10 Besides this, the absence of charge ordering and exchange bias and the maximum of M * for tA ⬃ 3.5 nm suggest that strain effects play a major role. Many recent works point to the fact that strain fields at phase boundaries play a crucial role in the growth of minority phases inside the AFM matrix.3,17 Multiphase coexistence in manganite thin films has also been linked to strain effects.13 In all these cases, the minority phases appear at the nanometer scale. In particular, for bulk samples of Pr1−xCaxMnO3 with x ⬃ 1 / 3, direct evidence shows FM nanodomains with characteristic sizes ⬃2 – 3 nm.3,16 We believe that the maximum magnetic moment of the PCMO in our multilayers for tA ⬇ 3.5 nm could be related to an overpopulation of FM domains, favored by the accommodation of the nanometric droplets in a strain field occurring at a similar length scale. In summary, we studied high-quality La2/3Sr1/3MnO3 / Pr2/3Ca1/3MnO3 multilayers and found that strain effects have a profound impact on the magnetic properties of this system. On one hand, a 1.5-nm-thick magnetic dead layer was deduced to appear in the FM LSMO, at the interface with the mainly AFM PCMO. This would inhibit the observation of exchange bias. On the other hand, the evolution of the low-T spontaneous magnetic moment of the PCMO with its layer thickness presents a peculiar peak for tA ⬇ 3.5 nm, reaching more than 2 ␮B / Mn, several times larger than the value for pure PCMO. This maximum appears to be correlated with an optimal thickness for magnetoresistance in similar FM/AFM multilayers. Accommodation of FM nanodroplets inside the PCMO layers, induced by the strain fields, would be responsible for the enlarged FM moment. Y. Tokura and Y. Tomioka, J. Magn. Magn. Mater. 200, 1 共1999兲. E. Dagotto et al., Phys. Rep. 344, 1 共2001兲. 3 P. G. Radaelli et al., Phys. Rev. B 63, 172419 共2001兲. 4 M.-H. Jo et al., Appl. Phys. Lett. 75, 3689 共1999兲. 5 K. Dörr et al., J. Appl. Phys. 89, 6973 共2001兲. 6 I. N. Krivorotov et al., Phys. Rev. Lett. 86, 5779 共2001兲. 7 H. Li et al., Appl. Phys. Lett. 80, 628 共2002兲. 8 A. Venimadhav et al., J. Phys. D 33, 2921 共2000兲. 9 R. Cheng et al., Appl. Phys. Lett. 72, 2475 共1998兲. 10 I. Panagiotopoulos et al., J. Appl. Phys. 85, 4913 共1999兲. 11 M. Izumi et al., Phys. Rev. B 61, 12187 共2000兲. 12 F. Giesen et al., Phys. Rev. B 69, 014421 共2004兲; J. Z. Sun et al., Appl. Phys. Lett. 74, 3017 共1999兲. 13 M. Bibes et al., Phys. Rev. Lett. 87, 067210 共2001兲. 14 V. N. Smolyaninova et al., Phys. Rev. B 65, 104419 共2002兲; M. S. Reis et al., ibid. 71, 144413 共2005兲. 15 D. Niebieskikwiat et al., J. Magn. Magn. Mater. 237, 241 共2001兲. 16 S. Mercone et al., Phys. Rev. B 68, 094422 共2003兲. 17 J. Tao et al., Phys. Rev. Lett. 94, 147206 共2005兲. 1 2

Downloaded 04 May 2006 to 192.167.164.69. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp