Cation non-stoichiometry in pulsed laser deposited Sr_2+yFe_1+xMo_1-xO₆ epitaxial films

Sr₂FeMoO₆ (SFMO) is an attractive material for spintronic devices due to a combination of highly spin polarized transport and a Curie temperature well above room temperature (T_C = 420 K (Ref. 1)). However, the magnetic properties of SFMO are highly dependent upon the ordering of the B-site cations, Fe and Mo, as well as the stoichiometry. The ordering of the cations is highly sensitive to the synthesis conditions, particularly the temperature and oxygen partial pressure.^2,3

Pulsed laser deposition (PLD) is an attractive film growth method for compositionally complex materials due to the assumed stoichiometric transfer from a target material of interest to the substrate. Over the past 14 years, there have been numerous reports describing deposition of SFMO films.^4–17 Reconciling the variations in properties reported from one paper to the next can be challenging due to the variations in the types of characterization methods used on those films. Most reports do not attempt to quantify the ordering of Fe and Mo, and even fewer characterize the stoichiometry of the films. Yet, from studies of corresponding bulk materials, the magnetic and magnetotransport properties are known to be very sensitive to both parameters.¹⁸ 

In order to realize the potential of double perovskite materials such as SFMO for spintronic applications, systematic studies are needed to understand how growth parameters impact cation stoichiometry and ordering of Fe and Mo cations, which in turn dictate the magnetic properties of SFMO films. To accurately portray the reproducibility of conditions, films have been grown in two different chambers to understand the role of temperature, pressure, and substrate-to-target distance on the composition and properties of SFMO thin films.

The SFMO target was prepared via traditional solid state synthesis methods. Stoichiometric amounts of SrCO₃, MoO₃, and Fe₂O₃ were ground and heated twice at 1050 °C in a flowing 95% Ar + 5% H₂ gas mixture for 10 h. The sample was heated two more times at 1100 °C for 5 h with intermittent grinding in a mortar and pestle. A pellet with an approximate diameter of 25 mm and thickness of 5 mm was pressed and sintered at a temperature of 1200 °C for 2 h, which led to a density that was approximately 64% of the theoretical value. Excellent phase purity was confirmed using X-ray diffraction (XRD). The films were deposited on SrTiO₃ (111)- and (100)-oriented substrates using two different PLD chambers that we refer to hereafter as chambers I and II. The differences between these two chambers are quantified in Table I. For all of the films, a 248-nm Lambda Physik^© KrF Excimer laser was used with the laser fluence optimized to 2 J/cm², and a pulse rate of 3 Hz. The optimal growth conditions discussed for chambers I and II were determined by varying the pressure, temperature, and laser fluence until phase pure samples were obtained. In all cases, a base pressure of 10⁻⁷Torr was achieved prior to film growth, and the target was pre-ablated with 1000 pulses. The substrate-to-target distance was measured from the front edge of the target to the edge of the substrate. A value of 38 mm was used for chamber II and 89 mm for chamber I.

XRD scans in a θ–2θ mode were collected for each film and used to assess the crystallinity and phase purity of each film. The out-of-plane lattice parameters were determined using a Bruker D8 Advance diffractometer equipped with an incident beam Ge (111) monochromator and a LynxEye^TM position sensitive detector. Rocking curves were obtained from a Bruker D8 Discover equipped with a Ge (220) 4-bounce monochromator and a variable slit Pathfinder detector. Refinement of the lattice parameters, phase purity, and ordering were completed using the Rietveld method via the software package TOPAS.¹⁹ 

The stoichiometry and thicknesses of the films were determined using Rutherford backscattering spectrometry (RBS) at Rutgers University using a standard ORTEC Si surface barrier detector. The scattering angle was 163° with a detector energy resolution around 17 keV. The RBS data were analyzed using the program SIMNRA.²⁰ Magnetic measurements were obtained using a Quantum Design^© Superconducting Quantum Interference Device (SQUID).

Atomic resolution, high-angle annular dark-field scanning transmission electron microscope (HAADF-STEM) images were obtained utilizing a probe aberration corrected FEI Titan™ 80–300 at The Ohio State University Center for Electron Microscopy and Analysis (CEMAS). STEM samples were prepared using an FEI Helios NanoLab™ 600 DualBeam™ focused ion-beam (FIB). Crystallographic orientations of the sample were determined using electron backscattered diffraction (EBSD) and the specimen was sectioned normal to the 〈110〉 direction in order produce a STEM sample suitable for imaging the double perovskite atomic ordering.

Initial attempts to grow high quality, phase pure films in chamber I were carried out in a pure argon atmosphere at a pressure of 50 mTorr with a substrate-to-target distance of 89 mm and temperature range of 800–850 °C. Growth pressures both higher and lower than 50 mTorr were also tested, but those conditions led to films with poor Fe/Mo order and secondary phases and are not reported here. The θ–2θ XRD patterns of these films are shown in Fig. 1, while the structural parameters and saturation magnetization (M_sat) values are given in Table II. An important parameter is the degree of cation ordering which was quantified in the (111)-oriented films from the refinements of the θ–2θ XRD patterns. The long range order parameter, ξ, is defined as 100(1 − 2q), where q is the fraction of Fe atoms occupying the Mo-site (antisite defects) in the double perovskite unit cell. For a perfectly ordered crystal q = 0 and ξ = 1, while for a crystal with a random distribution of Fe and Mo q = 0.5 and ξ = 0.

Despite the fact that we were able to grow highly crystalline SFMO films on both (100)- and (111)-oriented SrTiO₃ substrates, the films have several undesirable characteristics. First, the presence of epitaxial, crystalline face-centered cubic iron, γ-Fe, in five of the six films is disconcerting. Not only is the presence of multiple phases not ideal for practical applications but also it complicates interpretation of the magnetic properties and attempts to characterize the compositions of the perovskite phase. It is interesting to see that the metastable γ-Fe forms instead of the more stable body-centered cubic form. We attribute this to epitaxial stabilization on the SrTiO₃ perovskite substrate and/or the surface of the growing SFMO films. The tendency to form epitaxial Fe appears to be more pronounced on the (111)-oriented films.

Although the presence of metallic iron in the films complicates interpretation of the magnetic properties, it is still clear that the M_sat values fall well short of the 4 μ_B/f.u. expected for ideal SFMO samples. The M_sat values are particularly low for the films grown on (111)-oriented SrTiO₃ substrates. This can be attributed at least in part to high levels of Fe/Mo disorder. We were not able to directly measure the cation order parameter ξ on the (100)-oriented films, but, based on the larger values of M_sat, we speculate that these films are more highly ordered.

The out-of-plane lattice parameters for these films are considerably larger than the bulk value of 7.890 Å, particularly for the (100)-oriented films. The expansion of the out-of-plane lattice parameters could result from a compression of the in-plane lattice parameter due to the lattice strain imposed by the SrTiO₃ substrate (2 a = 7.810 Å). However, the in-plane lattice parameter of the (100)-oriented films, ranging from 7.926(8) to 7.934(3) Å, were also larger than both the substrate and bulk values. The observation that the in-plane lattice parameter does not match the substrate indicates a relatively rapid relaxation of the film. Based on previous studies of SFMO films, the nearly 100 nm thicknesses of these films should be sufficient for full relaxation.¹² The fact that the lattice parameters in both directions are larger than the values seen in phase-pure SFMO bulk samples suggests deviations from the ideal Sr₂FeMoO₆ stoichiometry. For example, studies of bulk samples have demonstrated an increase in the lattice parameters for the Mo-rich compositions in the Sr₂Fe_1+xMo_1−xO₆ series to values as high as ∼7.92 Å for x = −0.5.¹⁸ Therefore, the expansion in the lattice parameter with respect to bulk SFMO may be in part attributable to non-stoichiometry of the perovskite phase in these films.

To determine the compositions of these films, selected samples were analyzed using RBS, and the spectra are shown in Fig. 2. Only the films grown at 800 °C were analyzed because they had the lowest levels of γ-Fe impurities. The raw Sr:Fe:Mo cation ratios were determined to be 0.198:0.090:0.111 for the (100)-oriented film and 0.198:0.100:0.121 for the (111)-oriented film. It is interesting to observe that both films are Mo-rich. This observation coupled with the fact that no impurities containing molybdenum are seen in the XRD indicates that the perovskite phase is Mo-rich. From the RBS results, we estimate the stoichiometry of the (100)-oriented film to be approximately Sr_2.0Fe_0.9Mo_1.1O₆. The presence of γ-Fe makes it difficult to accurately estimate the composition of the perovskite phase in the (111)-oriented film. However, the fact that the RBS results show a larger Mo-excess in this film coupled with the fact that some of the Fe can be attributed to the γ-Fe phase seen in the XRD pattern, allows us to say that x ≥ 0.2 if the stoichiometry is written Sr₂Fe_1−xMo_1+xO₆.

Given the difficulty in obtaining phase pure SFMO films with high levels of cation ordering discussed earlier, we opted to change the atmosphere from pure Ar to a 95% Ar/5% H₂ mixture. Once again, films were grown at temperatures ranging from 800 °C to 850 °C, but unlike the films grown in pure argon the phase purity improved as the temperature increased. The θ–2θ diffraction patterns of films grown on both (100)- and (111)-oriented substrates in the 95% Ar/5% H₂ atmosphere are illustrated in Fig. 3. The (100)-oriented film grown at 850 °C is now phase pure, but the presence of a small amount of Fe-impurity in the (111)-oriented film grown at the same temperature could not be avoided.

The out-of-plane lattice parameters for the (100)- and (111)-oriented films were determined to be 7.9438(6) Å and 7.9290(1) Å, respectively, values that are 0.7% and 0.5% larger than the bulk value of 7.890 Å. The in-plane lattice parameter of the (100)-oriented film, 7.940(2) Å, is once again larger than both the SrTiO₃ substrate (by 1.7%) and the bulk SFMO lattice parameter (by 0.6%). These values suggest that the films grown in 95% Ar/5% H₂ atmospheres exhibit similar symptoms of non-stoichiometry as the films grown in pure argon.

The (100)-oriented film had a rocking curve with a FWHM of 0.14°, while the (111)-oriented film had a much broader rocking curve with a FWHM of 0.68°. The larger FWHM of the (111)-oriented film could be due to a number of factors. The presence of γ-Fe could serve as a nucleation center for various misfit dislocations leading to a greater mosaicity. Furthermore, the nucleation and growth of the (111)-oriented films likely follow a different mechanism than the (100)-oriented films. A previous study illustrated smooth terraces for the (100)-oriented SFMO films, while screw dislocations likely driven by the 3-fold symmetry of the 〈111〉 growth direction were visible in the (111)-oriented films.⁴ 

The compositions were analyzed by RBS, and the spectra are shown in Fig. 4. The Sr:Fe:Mo ratios of 2.05:0.70:1.30 and 2.05:0.70:1.25 for the (100) and (111) oriented films (respectively) once again reveal Mo-rich/Fe-poor films. In fact, the non-stoichiometry is even more pronounced than it was for the films grown in pure argon. Although there is a small difference in composition between the two films, the deviation is within the resolution of the RBS instrument. A comparison of the stoichiometric simulation (Sr:Fe:Mo = 2:1:1) with the experimental data clearly illustrates a deficiency in iron and an excess of molybdenum. The presence of excess molybdenum in SFMO films is significant and helps to explain the expansion of the lattice parameters noted above.^18,21 Previous arguments for an expanded in-plane lattice in relaxed SFMO films suggested thermal expansion or strain upon cooling to be the cause.¹² While this may contribute, these studies did not report the stoichiometry of the films, which must also be a factor in the expansion of the lattice parameters. The effect of excess Sr on the structure and properties of SFMO is a relatively unexplored topic, but will be discussed more thoroughly when the films grown in chamber II are discussed.

The non-stoichiometry also limits the extent to which Fe and Mo can adopt a perfect three-dimensional ordering. A Rietveld refinement of the (111)-oriented sample deposited at 850 °C was carried out with a restriction that the overall stoichiometry match the value determined from RBS (Sr₂Fe_0.7Mo_1.3O₆). From this refinement, occupancies were determined to be Sr₂(Fe_0.65Mo_0.35)(Fe_0.05Mo_0.95)O₆. The observation that the Mo-site contains only 5% iron indicates that these films are approaching the maximum order that is possible given the non-stoichiometry. Thus, one can conclude that these growth conditions facilitate ordering of the B-site cations, but unfortunately, they do not lead to stoichiometric transfer of the elements from the target to the substrate.

These films exhibit an extreme case of non-stoichiometry which undoubtedly has an effect upon the magnetic properties. The M_sat was determined to be 2.2 μ_B/f.u. for the (111)-oriented film and 1.7 μ_B/f.u. for the (100)-oriented film, as illustrated in Fig. 5. These values are substantially lower than the 4 μ_B/f.u. expected for fully ordered and stoichiometric SFMO. However, given the degree of non-stoichiometry, the reduction in M_sat is expected. The M_sat of the 100-oriented film, 1.7 μ_B/f.u., is in good agreement with bulk studies for a Sr₂Fe_1−xMo_1+xO₆ sample with x = −0.30, while the (111)-oriented film has a larger M_sat than expected.¹⁸ 

It is important to revisit the small Fe secondary phase visible in the XRD pattern of the (111)-oriented film. The larger M_sat in this film may best be explained by contributions of the magnetic Fe impurity, which has also been observed in bulk samples.²² In an effort to enhance the homogeneity of films, ex-situ post-annealing at 900 °C for 4 h in 95% Ar/5% H₂ atmosphere was completed on a (111)-oriented film containing the Fe secondary phase. Following the post-annealing process, peaks associated with the secondary phase disappeared from the XRD pattern, and the M_sat increased from 2.8 μ_B/f.u. to 3.3 μ_B/f.u. However, the ordering and lattice parameters of the perovskite phase remained unchanged. This behavior was reproducible. This observation suggests that the enhanced magnetization is not due to changes in the double perovskite phase, but rather to a larger contribution from the magnetic secondary phase(s).

The disappearance of the Fe peak in the XRD pattern is thought to correspond not with its elimination, but rather with its change in morphology leading to a loss of epitaxy with the substrate. Changes in SFMO film morphology due to post-annealing have been previously observed.⁷ Compared to the substrate and the epitaxial SFMO phase, any non-epitaxial secondary phase will have exceedingly weak peaks in the XRD. This result shows that extreme caution must be taken when using M_sat of SFMO films as a measure of the quality of the films. The use of magnetization data alone is misleading because of contributions to the M_sat from magnetic, non-epitaxial phases, some of which may not even be visible in the XRD spectrum.

In an effort to improve the film stoichiometry as well as test the transferability of growth conditions from one laboratory to another, SFMO films were grown in a second PLD growth chamber. We were able to obtain phase pure films in chamber II with a pure argon atmosphere, so this atmosphere was used exclusively for all films grown in chamber II. In order to directly monitor the ordering, film growth in this chamber was carried out primarily on SrTiO₃ (111)-oriented substrates, though structure and stoichiometry data for a (100)-oriented film are included for comparison.

Films were grown at various temperatures and pressures until the optimal parameters were realized. Compared to the optimal films in chamber I, the best SFMO film in chamber II was grown at a lower temperature (680 °C), a higher pressure (225 mTorr), and with a shorter substrate-to-target distance (38 mm). Diffraction patterns of the films (labeled S1–S5) are illustrated in Fig. 6, and a summary of their structural and magnetic data are listed in Table III. This set of growth conditions contrasts sharply with those optimized by Besse et al., where low pressures (P < 5 × 10⁻⁶Torr) and high temperatures (865 < T < 930 °C) were necessary to produce high quality films.⁴ Interestingly, films grown in chamber II under the conditions that were optimal in chamber I—low pressure and high temperature—exhibited very low Fe/Mo ordering (ξ = 26%), and several extra peaks in the XRD patterns that indicate the presence of secondary phases. It is not obvious why these conditions led to such different results between the chambers, though it is likely related to the optics of the laser and energy density distribution within the laser spot.²³ This stresses the point that factors other than growth temperature and pressure must also be considered when attempting to reproduce results.

Fig. 6 shows the XRD patterns of films grown in chamber II. The presence of sharp peaks, strong (111) and (333) ordering peaks, and no visible secondary phases, all speak to the high structural quality of these films. A quantitative measure of the Fe/Mo ordering was obtained from Rietveld refinements of the diffraction patterns. As shown in Table II, the order parameter ξ varies from 44(3)% to 85(2)% and is quite sensitive to changes in the temperature and pressure. Note from Table III how M_sat scales with ξ.

Unlike the films grown in chamber I, here, there are no spurious peaks associated with Fe-impurity phases. The out-of-plane lattice parameters for these films are close to the bulk value of 7.890 Å, which suggests the non-stoichiometry is less problematic in these films. The rocking curve FWHMs for these films ranged from 0.5° to 0.8°, values which are similar to those obtained for the (111)-oriented films in chamber I. However, the (100)-oriented film grown in chamber II had a FWHM near 0.5°, which is larger than the (100)-oriented film grown in chamber I. The faster growth rates of 0.26 to 0.94 Å/pulse associated with this chamber may account for the larger FWHM value, as there is less time for the atoms to form a perfect crystal.

RBS analysis of the films confirms that these films are closer to the desired stoichiometry. A Sr:Fe:Mo ratio of 2.05:0.97:1.03 was found for the film with the highest ordering and largest M_sat, S5. The near-stoichiometric SFMO is also observed for the (100)-oriented film with a cation ratio of 2.02:0.98:1.02. These RBS spectra are illustrated in Fig. 7. In each case, the Fe:Mo ratio is close to stoichiometric SFMO. As mentioned earlier, the error bars associated with the cation ratios determined by RBS are not easily estimated. Within the uncertainties of the RBS method, we can consider these films to be stoichiometric. However, this result in conjunction with the electron microscopy data given later provides evidence for a small excess of Sr.

The magnetic hysteresis loops for each film listed in Table III are plotted in Fig. 8(a). Each sample displays ferrimagnetic ordering and similar coercivities, but the M_sat ranges from 1.2 to 2.4 μ_B/f.u. As shown in Fig. 8(b), the M_sat values increase as the Fe/Mo order increases, reaching a maximum value for the film grown at the highest pressure (225 mTorr) and temperature (680 °C). The correlation between M_sat and ξ is expected from prior studies of bulk samples and theoretical predictions.^3,21,24 However, it is surprising to see that the M_sat values are consistently less than bulk samples with similar values of ξ, Fig. 8(b).

The magnetic hystereses tell only a portion of the story. Examination of the magnetic susceptibility as a function of temperature is a useful way to assess the homogeneity of the samples, as well as to look for the effects of secondary magnetic phases like Fe. A comparison of the zero-field cooled magnetic susceptibility versus temperature for (111)-oriented films grown under conditions optimized in chamber I and chamber II is shown in Fig. 9. The susceptibility curve for the film grown in chamber II, S5, reveals a Curie temperature of 345 K, which is lower than the 420 K observed in bulk samples; however, a reduction of T_C in thin films as compared to their bulk counterparts is not uncommon in magnetic double perovskites.^25,26

In contrast, susceptibility magnetization versus temperature for a (111)-oriented film containing Fe impurities grown in chamber I is visibly inhomogeneous at low temperatures, and, more importantly, the susceptibility does not go to zero, even at 400 K. This indicates the presence of a magnetic phase with an ordering temperature larger than 400 K. The susceptibility versus temperature curve for this film after ex-situ post-annealing, which was shown to produce a “pure-phase” XRD pattern, exhibits very similar features. This observation, coupled with the fact that annealing did not result in any changes in the perovskite peaks in the XRD pattern, suggests that the annealing does not change the perovskite phase but does cause the Fe impurity to lose epitaxy with the substrate. The smoothly decreasing magnetization seen for the sample grown in chamber II attests to the homogeneity of the films.

In Fig. 10(a), STEM-HAADF images are provided which illustrate areas of order interrupted by linear defects. HAADF intensities are directly proportional to the atomic number of elements raised to the power of x (x ∼ 2) for each element. Along the 〈110〉 direction of SFMO, all-Fe and all-Mo columns can be identified (least intense for Fe columns, most intense for Mo columns, and intermediate for Sr columns). Disordered regions most certainly contribute to the degraded M_sat, but this has already been taken into account in the Rietveld refinement. The linear defects shown in Figs. 10(b) and 10(c) provide evidence that helps to explain the reduction in M_sat observed for this sample. Spectrum imaging using energy dispersive X-ray (EDX) analysis was performed to verify the elemental composition of the defect regions which revealed an excess of Sr. These defects have been identified as Sr-rich Ruddlesden-Popper (RP) intergrowths in which an extra layer of SrO is present in the stacking sequence of the Fe-O and Mo-O layers. While ordered regions surround several of these planar defects, we suspect the extra SrO layer disrupts the strong ferrimagnetic coupling present between the Fe and Mo atoms which would otherwise lead to the high M_sat values. SrO-rich RP phases are not uncommon in bulk or thin film perovskites and have been reported in HRTEM images of the perovskite Sr_1+yCeO₃ for non-stoichiometry levels as small as y = 0.02.^27,28

It is important to assess any potential magnetic contributions of common RP phases such as Sr₃FeMoO₇ or Sr₄FeMoO₈. Sher et al. have completed both neutron diffraction and magnetic measurements on these two RP phases.²⁹ Both systems exhibit a ferromagnetic-type hysteresis curve with reported M_sat values below 0.8 μ_B/f.u. Neutron diffraction measurements reveal spin-glass behavior for Sr₃FeMoO₇ and canted antiferromagnetism for Sr₄FeMoO₈. Therefore, the presence of RP intergrowths would be expected to decrease M_sat.

The verification of Sr-rich intergrowths by STEM-HAADF supports the Sr non-stoichiometry determined from RBS. It also provides the best explanation for the observation that for a given value of the cation order parameter, ξ, the films have a lower than expected M_sat (see Fig. 8(b)). This suspicion is validated by a study done by Sanchez et al., where the ordering and M_sat of presumably Sr-stoichiometric films and bulk materials were in good agreement with one another.³⁰ This stresses the importance of quantifying not only the Fe and Mo stoichiometry but also the Sr content.

The results presented above for chambers I and II raise a number of interesting questions. Why are the optimal conditions so different in the two chambers? Why do the growth conditions used in chamber II—lower temperature, higher pressure and shorter substrate-to-target distance—create more stoichiometric films than those used in chamber I? Why is the magnetization for stoichiometric films lower than the expected value based on the Fe/Mo order parameter?

One of the most intriguing questions to consider is why the conditions in chamber I lead to the deposition of highly non-stoichiometric films. Growth rate is one possible factor since it has been shown to impact the B-site ordering and lattice parameters of SFMO films.³⁰ The growth rate for chamber I was 0.07–0.10 Å/pulse compared to rates of 0.26–0.94 Å/pulse in chamber II. The fact that all of the films grown in chamber II were nominally stoichiometric, even though the growth rates varied substantially would imply that growth rate and non-stoichiometry are not closely correlated. The lower growth rate in chamber I may be influenced by not only pressure, temperature, and substrate-to-target distance (which do contribute) but also the laser fluence.³¹ However, the laser fluence was calibrated to 2 J/cm² in both chambers which suggests that pressure, temperature, and substrate-to-target distance are the variables most closely linked to the film stoichiometry.

To understand the relationships between growth conditions and stoichiometry, we need to consider the characteristics of the plume under different growth environments. Studies have demonstrated that the PLD plasma plume is quite sensitive to changes in the pressure and temperature.^32–35 For example, the plume is more diffuse when the pressure is low and/or the substrate temperature is high, while the opposite conditions favor a compact plume. The general impact of pressure and temperature upon the plume dynamics of the two different chambers are illustrated in Fig. 11.^32–34 A shift in plume intensity (red portion) is depicted schematically as a function of substrate temperature and growth pressure.

Given the lower pressure of 50 mTorr, it is perhaps not surprising that the long substrate-to-target distance was necessary in chamber I in order to minimize damage caused by the highly energetic species being deposited. In chamber II, the shorter substrate-to-target distance was offset in part by the increased pressure, which helps to decrease the impact and velocity of the deposited species upon the substrate. This led to well-ordered and nearly stoichiometric films. The inverse correlation between the optimal pressure and substrate-to-target distance is in good agreement with the scaling law described by Kwok et al.³⁵ 

Speaking qualitatively the conditions in chamber I (50 mTorr, 800–850 °C) correspond to the large plume shown in Fig. 11(d), while those in chamber II (160–225 mTorr, 640–680 °C) correspond to the compact plume shown in Fig. 11(a). If the distribution of elements is not uniform in the plume, for example if lighter elements are preferentially scattered to the periphery of the plume (i.e., Fe), the diffuse plume obtained in chamber I may not facilitate a stoichiometric transfer of elements from the target to the substrate. We hypothesize that when a more well-defined plume front with more focused intensity is observed, as is the case in chamber II, the contents of the plume are more uniform, and stoichiometric transfer is maintained. For depositing compositionally complex materials containing elements with considerably different atomic masses, such as SFMO, a compact plume with a front edge appears to be desirable.

We have shown that phase purity, stoichiometry, magnetic properties, and ordering of SFMO films are highly dependent upon the growth conditions. Films grown in a high temperature, low pressure growth regime (chamber I) were highly non-stoichiometric and prone to contain elemental iron as a secondary phase. The potential presence of metallic iron highlights the fact that low temperature saturation magnetization can be a misleading indicator of film quality when used alone. Growth in a 5% H₂/95% Ar atmosphere suppressed the formation of elemental iron in these films, but increased the non-stoichiometry of the perovskite phase.

Films grown in a second regime of high pressure and low temperature (chamber II) were nearly stoichiometric, and under the right conditions, it was possible to obtain Fe/Mo order parameters as high as 85(2)%. However, the M_sat values of these films were consistently lower than the bulk samples with similar order parameters. This is believed to be due to the presence of Sr-rich Ruddlesden-Popper intergrowths, which disrupt the ferrimagnetic coupling between Fe and Mo. While temperature and pressure play a direct role in tailoring the plume size and shape, the substrate-to-target distance must also be optimized to ensure that the correct part of the plume is hitting the substrate, enabling better stoichiometric transfer. The M_sat values of these films were found to be highly correlated with the stoichiometry of the films.

We would like to thank Dr. Leszek Wielunski for performing the RBS measurements at Rutgers University. Funding for this research was provided by the Center for Emergent Materials at the Ohio State University, an NSF MRSEC (Award No. DMR-0820414), the US DOE, Office of Science, High Energy Physics Program and the State of Ohio.