Magnetic domain structures of La 0.67 Sr 0.33 MnO 3 thin films with different morphologies

Transcription

1 Magnetic domain structures of La 0.67 Sr 0.33 MnO 3 thin films with different morphologies P. Lecoeur and P. L. Trouilloud IBM T. J. Watson Research Center, Yorktown Heights, New York Gang Xiao Department of Physics, Brown University, Providence, Rhode Island A. Gupta a) IBM T. J. Watson Research Center, Yorktown, Heights, New York G. Q. Gong and X. W. Li Department of Physics, Brown University, Providence, Rhode Island Received 28 March 1997; accepted for publication 17 July 1997 Using a wide-field Kerr microscope, we have studied the magnetic domain structures of epitaxial and polycrystalline La 0.67 Sr 0.33 MnO 3 thin films as well as a film having thermally induced 110 microcracks. The epitaxial film on a 001 SrTiO 3 substrate has different magnetic domain behaviors for in-plane fields applied along the 100 and 110 directions. Magnetic domain orientation and contrast suggest a biaxial magnetic anisotropy with 110 easy axes. Defects such as microcracks and grain boundaries have a strong perturbing effect on the local magnetization and can lead to an enhanced and controllable spin-dependent scattering American Institute of Physics. S I. INTRODUCTION The discovery of the colossal magnetoresistance CMR effect in manganites 1,2 is of significance not only for the better understanding of strongly correlated electron systems, but also for their potential in field-sensing applications. The manganite materials exhibit magnetoresistance MR values of unprecedented magnitude. However, obstacles for applications are abundant. In particular, the CMR cannot be attained without a strong field in the Tesla range. 1 3 Recently, some progress has been made to reduce the field scale. 4 6 In magnetic tunneling junctions whose electrodes are made of La 0.67 Sr 0.33 MnO 3 LSMO, the MR at 4.2 K reaches 83% in only tens of Oe. 4 There exist two resistive states corresponding to the parallel and antiparallel magnetization configurations between the two electrodes. In polycrystalline LSMO either bulk or thin films, there is also a large MR component at a field scale of tens of Oe. 5,6 Absent in single crystal films, the low field MR response has been attributed to the specific magnetization configurations or magnetic domain structures in the polycrystalline materials. 5,6 Unfortunately, very little is known with regard to the magnetic domain structures in the epitaxial or polycrystalline films. Yet, magnetization distribution and domain formation are relevant to magnetotransport. For the manganites which tend to have a large spin polarization, 7,8 the spin-dependent electron scattering is expected to be appreciable across the magnetic domain or grain boundaries. In this article, we will present magnetic domain structures, obtained from a wide-field magnetic Kerr microscope, of LSMO thin films with three different morphologies. In epitaxial thin films, we have evidence of a biaxial magnetic anisotropy and of dispersion in the directions of the easy a Electronic mail: agupta@watson.ibm.com axes. In the film with stress-induced microcracks, the domain structures are found to be strongly influenced by the cracks. In the polycrystalline film, most grains are observed to reverse independently. 5 These results provide useful micromagnetic basis for the MR transport behavior. II. EXPERIMENT We have prepared three types of LSMO thin films using the pulsed-laser deposition technique as described in Ref. 9. The films Å in thickness deposited on 001 oriented SrTiO 3 substrates are epitaxial, which was confirmed by transmission-electron microscopy TEM. 4 The cation stoichiometry, determined by Rutherford backscattering spectroscopy, was within 5% of the nominal target compositions. In the second type of films, we have taken advantage of the mismatched thermal expansion between LSMO and the substrate. A thick single crystal film ( 3600 Å) develops postdeposition microcracks along the 110 directions. The last type of films were grown on polished polycrystalline SrTiO 3 substrates. 5 The polycrystalline structure is well replicated by the LSMO films with an average grain size of about 14 m. Cross-sectional TEM images revealed the local epitaxial growth of LSMO on individual SrTiO 3 grains, and the disordered regions about 10 Å wide along the grain boundaries. A wide-field Kerr microscope 10,11 was used to image the distribution of magnetization M in the thin films at room temperature. The polarized-light microscope was set to detect the changes in the polarization of light upon reflection from the magnetic surface due to the longitudinal Kerr effect. We did not find any evidence of polar Kerr contributions. The longitudinal Kerr rotation of the polarization plane is proportional to M p, the in-plane component of M projected onto the plane of incidence of the light. By rotating the plane 3934 J. Appl. Phys. 82 (8), 15 October /97/82(8)/3934/6/$ American Institute of Physics

2 FIG. 1. Temperature dependence of the spontaneous magnetization (M s ) and the coercivity (H c ) of an epitaxial La 0.67 Sr 0.33 MnO 3 thin film. The insets show two magnetic hysteresis loops at 5 and 290 K measured using a SQUID magnetometer. of incidence and adjusting the polarizers a different in-plane M component can be detected. Optical images were captured for different field values while cycling the applied field through a major loop from to 200 Oe and back. In order to eliminate the background optical signal, the Kerr images show the difference between optical images obtained for symmetric points on the loop. To increase the contrast, the Kerr signal was averaged over many loops. The Kerr images, (I 2 I 1 ), therefore reveal the changes in M p which are brought about by varying the external field. Under the assumption that the magnetic states for symmetrically applied fields are mirror images of each other, the black/white contrast shows the direction and magnitude of M p averaged over many loops. III. RESULTS AND DISCUSSION We have measured the magnetic properties of an asprepared epitaxial LSMO film by using a superconducting quantum interference device SQUID magnetometer. Figure 1 shows the temperature (T) dependence of the magnetic coercivity (H c ), obtained from a hysteresis loop, and the spontaneous magnetization (M s ), extrapolated from a M H curve. The thin film has a sharp magnetic transition temperature (T c ) of 347 K and a M s of 622 emu/cm 3 at 4.2 K. The film is magnetically soft, having a small H c of 49 Oe at 4.2 K and only 9.5 Oe at 300 K. The inserts in Fig. 4 show two representative hysteresis loops, measured at 5 and 290 K, from which we obtained H c. Figure 2 shows two images of the remanent state of an epitaxial LSMO film about 500 Å in thickness. These are contrast images for the remanent state after saturation. The field (H) was applied in-plane and along the 100 direction x axis. Figure 2 a shows the M y component perpendicular to H, and the contrast displays a distinctive pattern of vertical black and white striations with an average width of about 2 m. Thus M y takes alternating / signs across the black and white regions, as depicted below the Kerr image in the schematic representations of the M y components for the two remanent states. The dashed lines represent walls between areas of different magnetic orientation. On the other hand, J. Appl. Phys., Vol. 82, No. 8, 15 October 1997 Lecoeur et al. 3935

3 FIG. 2. a Room temperature magnetic Kerr image of the M y component along the field (H) applied along the 100 direction x axis of an epitaxial LSMO film in the remanent state (H 0). b Kerr image of the M x component in the same state. The associated schematics show the deduced local magnetic domain structures and the M components along the Kerr sensitivity axes. Fig. 2 b shows the M x component along the H direction. The white contrast is essentially uniform, meaning that M x is almost constant over the whole image area. The interpretation of the two images is that, as H is reduced from 200 Oe to zero, M rotates symmetrically away from the 100 direction and settles towards either 110 or 11 0 directions on both sides of 100. The two remanent states are described in Fig. 2 c. Since these images were the averages of over 64 cycles of the M H loop, the strong contrast in Fig. 2 a indicates that at a given spot M repeatedly rotates towards the same 110 or 11 0 axis. The local selection of either 110 or 11 0 by M is due to a slight dispersion of the direction of easy axes and the magnetostatic and exchange interactions. The result is a buckled magnetization pattern for the remanent state as described schematically in Fig. 2 c. This behavior is consistent with 100 being hard axes. While M rotates in order to reduce the anisotropy energy, the M x component is kept uniform in order to minimize the magnetostatic energy. The strength of the M y contrast, normal to the walls, suggests 110 as the easy axes. This is, however, not conclusive since the magnetization rotates only partially towards these directions in order to reduce the exchange energy contained in the buckled structure. By averaging the contrast over an image, we can obtain the average Kerr signal which is proportional to M p along the direction of Kerr sensitivity. Figure 3 shows a LSMO hysteresis loop measured using this method, where H and M p are along the 100 direction. The Kerr hysteresis loop, having a H c of 7 Oe, resembles that obtained from the SQUID magnetometer see Fig. 1. The remanent state is similar to that described in Fig. 2. Three magnetic images are also shown in Fig. 3; they are obtained at locations A, B, and C along the loop. The magnetization reversal is via a nucleation process as indicated by the blacks regions in image A. The black areas expand with increasing negative field image B, and finally permeate through the entire area image C. However, even in the nearly saturated state image C, there are still white spots and lines where M remains unreversed. Again, this is related to dispersion in the direction and strength of the magnetic anisotropy. For a field applied close to 110, striations also appear in the Kerr images, but the striations are at 45 to the axis of the applied field Fig. 4. This corresponds to domains of 110 and 11 0 magnetization whose orientation minimize the magnetostatic energy. While these striations are small in the remanent state, they grow and become more pronounced near the coercive field. As in Fig. 3, images A and C of Fig. 4 are on both sides of the coercive field while B is very close to it. At the coercive field, a striking cross-hatched pattern is revealed. These domain patterns confirm that the underlying anisotropy is biaxial with 110 easy axes directions. Next we will look at the domain formation in thin films with a high number of defects. In the thin film with thermally induced 110 microcracks, the magnetic domain structure is controlled by the presence of cracks. Figure 5 a shows a 3936 J. Appl. Phys., Vol. 82, No. 8, 15 October 1997 Lecoeur et al.

4 FIG. 3. Kerr hysteresis loop of the epitaxial LSMO film shown in Fig. 2 and three magnetic images taken at locations A, B, and C along the loop. H is along the ~100! direction. Kerr M H loop. As in the previous loop, each point has a corresponding Kerr image which shows the difference between the magnetic states at the specified field and at the opposite field ~with a full excursion of 1/2200 Oe!. Figure 5~b! is an image taken at remanence, and Fig. 5~c! near H c. Tracing the loop, we find that the sample is saturated at 100 Oe, having a uniform white contrast except at the location of the cracks. At remanence, the sample remains mostly satu- FIG. 4. Kerr images of the epitaxial LSMO film shown in Fig. 2 for H applied along the ~110! easy axis. The three magnetic images A, B, and C were taken at fields smaller ~7 Oe!, close to ~8 Oe!, and larger ~9 Oe! than the coercive field. J. Appl. Phys., Vol. 82, No. 8, 15 October 1997 Lecoeur et al. 3937

5 FIG. 5. Kerr hysteresis loop of an LSMO film with thermally induced ^110& microcracks and two magnetic images obtained at locations H r (H50) and H c on the loop. H is along the ~110! direction. rated, but the gray contrast at the cracks has spread into a broader line about 3 mm in width ~revealing the locations of the cracks! much wider than the physical width of the cracks (,1000 Å). This indicates spreading demagnetization near the crack areas. Near the average H c, though the average contrast is close to zero, the image itself is a combination of white, black, and gray areas. If we look at one of the large rectangular areas defined by the cracks and oriented parallel to the applied field, the image at H c shows a black reverse domain that nucleated from the right and left edges and occupies the center of the rectangle. The area very close to the horizontal cracks is grey, demagnetized as described previously. The white area between them has not switched yet, but will be reversed by wall motion and expansion of the black domain at higher negative fields. The above results demonstrate that the presence of microcracks can effectively manipulate the domain structure in LSMO films. Finally, in the polycrystalline sample, the formation of domain structure is controlled by the grain structure as discussed in Ref. 5. The average H c of about 22 Oe is more than twice as large as that of a single crystal film. A more detailed look shows that the grains switch mostly independently; each in a narrow range of fields but that the distribu3938 J. Appl. Phys., Vol. 82, No. 8, 15 October 1997 tion of switching fields is quite large leading to a higher macroscopic coercivity. Even at 200 Oe, some of the grains appear not to have switched. Within most of the grains, particularly the smaller ones, M is rather uniform. However, in larger grains, walls are apparent in the reversal process. IV. CONCLUSIONS Our results show that structural defects such as microcracks and grain boundaries are effective at creating nonuniform magnetization distributions in LSMO thin films. One may take advantage of this fact, and artificially create a high density of areas with enhanced spin-dependent electron scattering, and strategically distribute these areas along the current path. This would lead to a large magnetoresistance at a field scale corresponding to the small coercivity. So far, we have indeed observed enhanced low field MR component in polycrystalline LSMO films,5 which we attribute to the nonuniform magnetization near the grain boundaries. A similar low field MR effect has been reported by Hwang et al. in bulk polycrystalline LCMO materials.6 The large low-field magnetotunneling effect observed in LSMO junction by Lu Lecoeur et al.

6 et al. 4 is also due to electron transport across two separated magnetic domains with reversed orientations. ACKNOWLEDGMENT We thank Yu Lu, T. R. McGuire, W. J. Gallagher, J. Slonczewski, P. R. Duncombe, and J. Z. Sun for helpful discussions, and Y. Y. Wang and V. P. Dravid for TEM measurement of some samples. This work was supported by National Science Foundation Grants Nos. DMR and DMR , and by IBM. 1 R. von Helmolt, J. Wecker, B. Holzapfel, L. Schults, and K. Samwer, Phys. Rev. Lett. 71, S. Jin, T. H. Tiefel, M. McCormack, R. A. Fastnacht, R. Ramesh, and L. H. Chen, Science 264, G. Q. Gong, C. Canedy, G. Xiao, J. Z. Sun, A. Gupta, and W. J. Gallagher, Appl. Phys. Lett. 67, Y. Lu, X. W. Li, G. Q. Gong, G. Xiao, A. Gupta, P. Lecoeur, J. Z. Sun, Y. Y. Wang, and V. P. Dravid, Phys. Rev. B 54, R A. Gupta, G. Q. Gong, G. Xiao, P. R. Duncombe, P. Lecoeur, P. Trouilloud, Y. Y. Wang, V. P. Dravid, and J. Z. Sun, Phys. Rev. B 54, R H. Y. Hwang, S.-W. Cheong, N. P. Ong, and B. Batlogg, Phys. Rev. Lett. 77, W. E. Pickett and D. J. Singh, Phys. Rev. B 53, P.-G. de Gennes, Phys. Rev. 118, A. Gupta, T. R. McGuire, P. R. Duncombe, M. Rupp, J. Z. Sun, W. J. Gallagher, and G. Xiao, Appl. Phys. Lett. 67, B. E. Argyle, B. Petek, and D. A. Herman, Jr., J. Appl. Phys. 61, P. Trouilloud, B. Petek, and B. E. Argyle, IEEE Trans. Magn. 30, J. Appl. Phys., Vol. 82, No. 8, 15 October 1997 Lecoeur et al. 3939