Microstructure and constituents distribution analysis

The energy dispersive spectroscopy analysis for AD and PA LSnMO films were as illustrated in Fig 3.5a and 5b. It reveal there are no difference between these two sample, suggesting the amounts of the four elements (La, Sn, Mn, O) do not change after the post annealing process. Fig. 3.8 compares the θ−2θ scan of x-ray diffraction (XRD) results for AD, ArPA and PA LSnMO films. As is evident from the results, all the films are c-axis oriented with no observable impurity phases. The slight split of the (00l) peaks between the ArPA and PA-films and the substrates, however, indicates that significant strain relaxation may have occurred after prolonged annealing. To further delineate the possible subtle changes in the film microstructure due to annealing, L-scan and φ-scan measurements were performed. The typical results are illustrated in Fig. 3.9－3.11, respectively. In Fig. 3.9, it is evident that, in addition to the sharp Bragg peak from the LAO substrate, well-resolved (003)-reflection peaks of the LSnMO films were observed for the AD, ArPA, and PA samples with the correspondent c-axis lattice constant being 3.908Å, 3.893Å and 3.878 Å, respectively. Since the LSnMO film on LAO substrate is expected to experience an in-plane compressive stress due to the smaller substrate lattice constants, the progressive shortening of the c-axis lattice constant exhibited in the ArPA and PA films is indicative of the occurrence of annealing-induced strain relaxation, which in turn drives the film toward its bulk characteristics (The pseudocubic lattice parameter of La1-xMnO3 ranges from 0.3866 nm-0.3880 nm for x ＝ 0 – 0.33 [25, 28].). We note

that the in-plane lattice parameters also exhibited noticeable shrinkage upon annealing with a ＝ 3.913Å, 3.907Å and 3.900Å for AD, ArPA, and PA films, respectively.

Valencia et al. [32] pointed out, in their study of LCMO/STO films, that the existence of oxygen vacancies compensates the excessive elastic energy in coherently strained films.

Thus, the shrinkage of the unit cell volume may reflect the elimination of oxygen vacancies and associated strain energy. The full width at the half maximum (FWHM) of the (0 –1 3) peak of AD-LSnMO is about 4.4°, which is much larger than that of LAO (103) (0.1°). This can be either due to the strain or fine grain size effects. The in-plane grain size estimated from the line width of K-scan across the LSnMO (0 –1 3) reflection is approximately 100 Å. On the other hand, for the PA film, the FWHM of LSnMO (103) is about 1.9°, which, though still much larger than that of LAO (103) (∼0.036°), is significantly smaller than that observed in AD-film. The in-plane grain size as estimated from the line width of the H-scan across the LSnMO (103) reflection is approximately 135Å. This difference strongly suggests the crystallinity of LSnMO films was significantly improved upon annealing while the in-plane epitaxial relations remain largely intact. The φ-scans were taken along the (0 –1 3) Bragg peak of the LAO substrate and LSnMO films. As is evident from Figs. 3.10 and 3.11, both AD- and PA-LSnMO films display well-aligned ab-plane epitaxy with the LAO substrate. Again, the FWHM of the diffraction peaks for the AD-film is larger than that of the PA-film consistent with the arguments aforementioned for the XRD results.

To further delineate the evolving film/substrate relations due to annealing, Fig. 3.12

－3.14 compares the reciprocal space maps of AD-, ArPA- and PA-LSnMO films. The decreasing peak position offset in q between LAO and LSnMO in the reciprocal space

maps clearly shows the occurrence of strain relaxation effect after annealing process. It is also evident that the lattice constants of LSnMO are significantly larger than that of LAO, consistent with the abovementioned results. For comparison, we show, in Fig. 3.15, the similar plot obtained for the PA LSnMO/STO (001). Since the lattice constant of LSnMO falls between STO and LAO and is closer to that of STO, LSnMO grown epitaxially on STO and LAO would experience a tensile/compressive average in-plane strain, respectively. In the case of LSnMO/STO, the films are fully coherent to the substrate, as indicated by the nearly perfect alignment between the center of the STO(113) and LSnMO(113) peak contours. Nevertheless, in both bases, we observe shrinkage of both the c-axis lattice constant and unit cell volume of LSnMO upon post-annealing; in LSnMO/STO case, a-b remains invariant but in the case of LSnMO/LAO, a-b also decreases slightly. It appears that some of the strain, originally compensated by oxygen vacancy incorporation [32], was released through the reduction of average unit cell volume. Alternatively, it is also possible that the change of average lattice parameters upon PA is coupled with change of composition, or even second-phase segregations. In any case, the lattice constant change associated relaxation of the strain in the films, though might originate from very different mechanisms, do intimately correlate to the enhanced CMR properties in a consistent manner. Unfortunately, the x-ray analyses seemed inadequate to precisely answer the question about the role played by Sn.

In order to further examine the distribution of Sn and the possibly formed Sn-compounds that are not discernable by using x-ray diffraction alone, we performed the cross-sectional transmission electron microscopy (X-TEM) analyses to reveal the microstructure and the element distribution of the LSnMO films. Fig. 3.16 and Fig. 3.17

show the bright-field TEM images and selective area diffraction (SAD) pattern of AD LSnMO(001) film grown on LAO(001) substrate. As can be seen from Fig. 3.16, the film microstructure appears to be rather homogeneous and there is no trace of any additional compound existing within the interface of film and substrate. In addition, the absence of extra diffraction spots in Fig. 3.17 indicates that the AD-LSnMO film grown on LAO substrate is indeed single phase with well-organized epitaxial relations. The rod-like diffraction spots and the streaky patterns appeared in the columnar grains of the LSnMO phase suggest that there exists a significant amount of strain in the film. We believe that both the slight lattice mismatch between the film and substrate and the large ionic size difference between La³⁺ and Sn⁴⁺ may contribute. For the PA-LSnMO film, however, the results are quite different. The bright field image shown in Fig. 3.18 apparently displays two somewhat separated regions. The grains in the near surface upper region appear to be more “equi-axial” with much less strain-induced streaky patterns, suggesting significant recrystallization may have occurred due to annealing. In the “lower” region (region close to the substrate interface) the features are more like that observed in the AD-film.

Although it is still not obvious to conclude whether or not the recrystallized upper region containing any newly formed phases from the SAD results, it is, nevertheless, interesting to find that the oxygen annealing affects only the upper part of the film. With the about 4 hours of annealing time and 100 nm of the affected depth, the estimated oxygen diffusion rate at 800 °C is about 25 nm/hr.

Figure 3.5a: The energy dispersive spectroscopy analysis (EDX) for AD LSnMO film.

Figure 3.5b: The energy dispersive spectroscopy analysis (EDX) for PA LSnMO film.

Figure 3.8: XRD results for the as-deposited, Ar-annealed and post-annealed LSnMO films grown on LAO substrates at Ts＝780^oC.

Figure 3.9: L-scan of AD, ArPA and PA films across (003) Bragg peak.

Figure 3.10: The φ-scans for the LAO (103) and AD LSnMO (103) Bragg peaks, respectively.

Figure 3.11: The φ-scans for the LAO (103) and PA LSnMO (103) Bragg peaks, respectively.

Figure 3.12: Reciprocal space maps of AD LSnMO films grown on LAO substrate.

Figure 3.13: Reciprocal space maps of ArPA LSnMO films grown on LAO substrate.

Figure 3.14: Reciprocal space maps of PA LSnMO films grown on LAO substrate.

Figure 3.15: Reciprocal space map of PA LSnMO film grown on STO substrate.

From the above discussion, the strain relaxation effect is consistent with what observed in x-ray diffraction analyses and hence account for part of the magneto-transport properties obtained. The XAS results, however, remain to be clarified.

In order to delineate the effect of annealing on the composition distribution of the doped element in the LSnMO film, the electron energy loss spectroscopy (EELS) mapping was performed on the TEM samples. Figs. 20(a1)-(a5) show the zero-loss image of the AD-LSnMO film and the elemental maps of La, Sn, Mn, O, respectively. As revealed by the series of the images, the four constituents distribute uniformly over the entire AD-LSnMO film, indicating that they are presumably situated within the parent perovskite structure. The results are largely consistent with the data discussed above.

Nevertheless, the elemental maps of the four constituent elements in the PA-LSnMO film display rather different results. As illustrated in Figs. 20(b1)-(b5), there are clear evidence of local segregation of Sn and O in the “upper” region of the PA-LSnMO film (Fig. 20(b3) and (b5)). More interestingly, in these Sn-O segregated areas, both La and Mn are absent (Fig. 20(b2) and (b4)), indicating that the clusters formed are some kind of Sn-O compound. (From the previous XAS results it should be SnO2.) This strongly implies that the oxygen annealing affected regions may indeed induce the formation of the La-deficient manganite, which, in turn, accounts for the marked enhancement in Tc and TIM for the PA-LSnMO films described above.

Figure 3.16: The bright-field TEM image for the AD LSnMO (001) film.

Figure 3.17: The electron diffraction pattern for the AD LSnMO (001) film.

Figure 3.18: The bright-field TEM image for the PA LSnMO (001) film. Notice that the near-surface upper part of the PA LSnMO film displays apparent recrystallization

signature upon prolonged annealing.

Figure 3.19: The electron diffraction pattern for the PA LSnMO (001) film.

Figure 3.20: (a) The electron energy loss spectroscopy of the zero-loss image (a1) and the elemental maps of La (a2), Sn (a3), Mn (a4), and O (a5) of the AD LSnMO film. (b) The electron energy loss spectroscopy of the zero-loss image (b1) and the elemental maps of La (b2), Sn (b3), Mn (b4), and O (b5) of the PA LSnMO film.