Growth of SrVO₃ thin films by hybrid molecular beam epitaxy

Transition metal oxides with strongly correlated d electrons provide the foundation for materials with unconventional electronic and magnetic states, such as Mott-insulators, high-temperature superconductivity, and colossal magnetoresistance.^1,2 Due to the correlation effects of the conduction electrons,³ small modifications, such as doping, epitaxial strain, and dimensional confinement,^4,5 or alteration of external parameters such as temperature or electric field, can drastically change the system's electronic properties. Correlation effects are ubiquitous in transition metal oxides with the ABO₃ perovskite structure, where A is an alkaline Earth or rare Earth element and B is a 3d transition metal, such as titanium,⁶ vanadium,⁷ nickel,^8,9 or manganese.^4,5 These materials exhibit rich electronic phase diagrams, originating from the sensitive interplay of the charge, orbital, and spin degrees of freedom, which affects the electronic configuration of the transition metal cation. Disorder of any form, such as the formation of precipitant phases, extended structural imperfections, or point defects to accommodate unintentional deviation from an ideal composition, can potentially upset the delicate balance of these long range correlation phenomena, dramatically altering the stability of different ground states and thus directly affecting the electronic properties in these strongly correlated materials.¹⁰ 

The complex oxide perovskite explored here, SrVO_3, has a d¹ electronic configuration with a single electron in the partially filled t_2g band. It is metallic due to its moderate bandwidth and relatively weak onsite Coulombic repulsion.¹¹ Experiments have shown that a metal to insulator transition can be induced in SrVO₃ through dimensional confinement: reducing the bandwidth can drive SrVO₃ into the Mott insulating state.^12–15 However, as the dimensionality is reduced, defects and disorder at interfaces may play a significant role in governing or suppressing the material's intrinsic electronic properties.¹⁰ In other strongly correlated systems such as VO₂, it has been shown that the quality of the transition is critically dependent on film quality.¹⁶ Therefore, exploring and harnessing the novel physical phenomena in correlated electron systems requires a growth method capable of producing high quality films with minimal defects.

Epitaxial SrVO₃ thin films have been grown by numerous methods, including laser molecular beam epitaxy,¹⁷ pulsed laser deposition,¹⁸ chemical vapor deposition,¹⁰ and pulsed electron beam deposition.¹⁵ Here, we report in detail on the synthesis and characterization of SrVO₃ films grown by hybrid molecular beam epitaxy (HMBE).¹⁹ In this process the A-site cation, strontium, is supplied with a standard thermal effusion cell. The B-site vanadium was supplied as a metalorganic precursor vanadium (V) oxytriisopropoxide (VTIP) through a gas inlet system, which thermally decomposes on the growing surface into volatile by-products and a reduced vanadium oxide with a lower than V⁵⁺ valence state that is incorporated into the film.²⁰ Additional oxygen was supplied in molecular form using a gas inlet system to avoid potential oxygen deficiency in the film. The HMBE approach applies growth kinetics different from conventional methods, and enables the growth of SrVO₃ in an adsorption assisted manner, where the high vapor pressure of the metalorganic VTIP provides an additional volatile component, and may assist in providing the superior film quality for a given ability to control the V:Sr flux ratio.²¹ Furthermore, it is a low energy deposition technique which helps limit defect formation due to the high impact energy of species ablated from targets, which is present in some other techniques. When grown by HMBE, SrVO₃ has a resistivity that is several orders of magnitude lower¹¹ than films from different growth methods reported in literature, indicating that the combination of low energy deposition and good stoichiometry control during growth can lead to high quality thin films with much lower active defect density.

SrVO₃ films were grown on (001) (LaAlO₃)_0.3(Sr₂AlTaO₆)_0.7 (LSAT) substrates in a DCA M600 MBE system with a base pressure of 5 × 10⁻¹⁰Torr. The strontium (4N, Sigma Aldrich) was supplied from a conventional effusion cell, while VTIP (4N trace metal basis, MULTIVALENT Laboratory) was supplied with a gas injector. VTIP flux was controlled by throttling a vapor source with a linear leak valve to maintain a constant gas pressure upstream of the injector, which was controlled by a capacitance manometer. The Sr flux was calibrated using a quartz crystal microbalance. In addition, beam equivalent pressure (BEP) measurements of both VTIP and Sr fluxes were performed using a beam flux ion gauge. Both flux monitoring probes were inserted at the growth position. A series of films were grown using different VTIP:Sr BEP flux ratios in the range from 14.92 to 17.49. The BEP values were determined from calibration curves. The physical flux of Sr, measured by the quartz crystal monitor, and the inlet pressure of the VTIP gas injector, measured by a capacitance manometer, were related to their respective BEP values. A dramatically improved repeatability in the flux ratio determination was found throughout the set of experiments by using the calibration curves rather than direct measurements by only the beam flux monitor.

Prior to growth, LSAT substrates were prepared by slowly heating to 1200 °C in a tube furnace while flowing oxygen with LaAlO₃ mounted 0.7 mm above the surface to obtain atomically smooth surfaces with an A-site surface termination.²² In the growth chamber, substrates were heated to 800 °C, as measured by a thermocouple mounted behind the substrate. Before film deposition, the substrates were annealed at the growth temperature, first for 20 min in UHV, then for 20 min in 5 × 10⁻⁶Torr BEP oxygen flux generated by a plasma source. Films were then grown at 800 °C with a 2 × 10⁻⁷Torr molecular oxygen BEP.

Film growth was monitored using reflection high-energy electron diffraction (RHEED). After growth, the films were cooled to room temperature at 30 °C per minute under the same oxygen flux. Film surface morphology was measured using a Bruker Icon atomic force microscope (AFM) in peak force tapping mode. Out-of-plane lattice parameter measurements were performed using a Phillips X'Pert MRD Pro four circle x-ray diffractometer. Samples for scanning transmission electron microscopy (STEM) were wedge polished in plan view or cross section with an Allied Multiprep before ion milling with a Fischione Model 1050. Energy dispersive x-ray spectroscopy (EDS) was conducted on a probe-corrected FEI Titan G2 60–300 kV equipped with an X-FEG and a SuperX EDS system. High angle annular dark field (HAADF) STEM was conducted at 200 kV using either a probe corrected FEI Titan G2 60–300 kV or with a JEOL 2010F.

During growth, RHEED was observed along the [110] direction for each film. RHEED diffraction images captured throughout the growth are shown in Fig. 1. The LSAT substrate exhibited a two-fold reconstruction [Fig. 1(a)]. Immediately after growth initialization, the intensity of the (0–1) reflection decreased. The twofold reconstruction almost completely disappeared within the first minute of growth, i.e., the deposition of ∼2 monolayers of SrVO₃, shown in Fig. 1(b). Within the growth of the next 3–5 monolayers, the diffraction pattern intensity increased and revealed a pronounced twofold reconstruction that persisted throughout the growth [see Fig. 1(c)]. The RHEED pattern remained streaky, indicating a smooth 2D film growth, while the measured intensity slightly decreased. No RHEED intensity oscillations were observed, which indicated that films grew in a step-flow mode. As shown in Fig. 1(d), the diffraction images taken after the growth displayed sharp diffraction spots that were pointlike rather than streaky and organized along the Laue rings, indicating very smooth film surfaces, consistent with AFM measurements.²³ 

A comparison of the RHEED patterns taken after growth for different VTIP:Sr ratios is shown in Fig. 2. Away from the optimum VTIP:Sr ratio, additional features were observed for the electron beams along both the [110] and the [100] direction. For comparison, pristine LSAT substrates are shown in Figs. 2(a) and 2(b). Films grown under Sr-rich conditions displayed more diffuse streaks and exhibited a twofold reconstruction along the [110] direction, whereas no additional reflections were observed along the [100] direction [see Figs. 2(c) and 2(d)]. As shown in Figs. 2(e) and 2(f), RHEED patterns for the film grown under stoichiometric conditions showed intense and crisp diffraction patterns with minimal background intensity. Similar to the Sr-rich films, a twofold reconstruction was observed only along the [110] direction and not in the [100] direction. In contrast, the RHEED for films grown under V-rich conditions shown in Figs. 2(g) and 2(h) exhibited a tenfold reconstruction along the [110], while a three-fold reconstruction was observed along [100].

Selected four circle x-ray 2θ-ω out-of-plane diffraction of Sr-rich, stoichiometric, and V-rich films are shown in Fig. 3. As shown in Fig. 3(a), the films grown at the lowest VTIP:Sr flux ratio showed no additional phases in the on-axis 2θ-ω scans. The quality of the thickness fringes was slightly inferior to those of films grown under optimal VTIP:Sr ratios [Fig. 3(b)], and the film peak was shifted toward smaller 2θ values, due to a larger out-of-plane lattice parameter. The measured out-of-plane film lattice parameter of 3.821 Å determined for the film grown under optimal conditions corresponded to that expected for stoichiometric SrVO₃ coherently strained by the LSAT substrate (0.73% tensile strain).¹⁹ Here, the pronounced thickness fringes indicated a 50 nm thick film with very smooth surface and sharp interfaces. With increasing VTIP:Sr ratios beyond the optimal conditions the thickness fringes began to vanish. Figure 3(c) shows a XRD scan of a film grown at VTIP:Sr = 17.49. No thickness fringes were found, and the film peak was broad and shifted to smaller 2θ values. An additional peak was found at 2θ = 40.74°, which indicated the presence of an additional phase in the film. The origin of this parasitic phase, however, could not be unambiguously attributed to a specific V-rich phase in the SrO-VO₂-V₂O₅ phase diagram.²⁴ Hence, a possible intergrowth of V–O compounds have been considered as well. The additional peak at 2θ = 40.74° is close to the VO₂ (111) peak (2θ = 40.3°), V₃O₇ (−408) peak (2θ = 41.3°), and V₂O₅ (002) peak (2θ = 41.5°). Detailed TEM investigation discussed below revealed that this peak can be attributed to rocksaltlike VO_x intergrowths found in films grown under V-rich conditions.

The film surface morphology varied with the VTIP:Sr flux ratio, which is summarized in Fig. 4. For the growth at the lowest VTIP:Sr ratio (14.9), a terracelike structure was observed with large islands, see Fig. 4(a). Typical island heights were ∼5 nm and the lateral dimensions varied up to ∼100 nm × 1 μm oriented with the long direction along [110] and revealed a more regular shape suggesting that the agglomerated islands had a crystalline phase. With increasing VTIP:Sr ratio toward the optimized conditions, the terrace structure remained but the number and size of islands was reduced, as shown in Fig. 4(b). In contrast, Fig. 4(c) shows that for films grown at optimal flux ratios, the surface was dominated by atomically smooth terraces with unit cell step heights, and generally devoid of islands. For films grown at even higher VTIP:Sr flux ratios, shown in Fig. 4(d), atomic terracing of the surface was reduced; however, the films remained flat with a root-mean-square roughness around ∼1 nm. Closer inspection revealed a wavy pattern of the surface height, where atomic terraces were still obtained, but terrace edges were rough, and small crystallites were found. Although the unit-cell step height remained, Fig. 4(e) shows that there is considerable interruption in the long range sample smoothness for films grown in more VTIP rich conditions.

For further analysis, HAADF-STEM was used to explore the local structure of SrVO₃ thin films as a function of VTIP. Representative images of the films are shown in Fig. 5. The Sr-rich films exhibited vertically oriented Ruddlesden–Popper defects, similar to those reported in Sr-rich SrTiO₃.^25–27 Films grown under optimal VTIP:Sr ratio were found to be virtually free of structural defects, indicating the growth of phase-pure and stoichiometric SrVO₃.²⁸ 

Similar to the Sr-rich films, the V-rich sample also contained defective regions that extended throughout the entire film thickness. Characterization of these V-rich defects was provided through plan view imaging. In plan view orientation, the high density of the V-rich defects is revealed, which is shown in Fig. 6(a). Composition analysis by EDS [Fig. 6(a)] showed that the defects were rich in V and deficient in Sr. Furthermore, as shown in Fig. 6(b), the V-rich defects were crystalline and embedded within the perovskite matrix. Based on the projected atom positions, the VO_x inclusions were determined to exhibit rock salt structure. These results contrast with Ti-rich SrTiO₃, which forms amorphous TiO_x regions throughout the films.^25,26 Neither inclusion, VO_x in SrVO₃ nor TiO in SrTiO₃, would be considered low strain. When comparing the strain for Ti and V monoxides with rock salt structure, the strain for VO in SVO remained considerable at 4.3%, much lower than 7.3% in the case of TiO_x in SrTiO₃.²⁹ The formation of epitaxially oriented rocksalt VO_x inclusions may be stabilized by a reduced lattice mismatch with SrVO₃ compared to that of TiO and SrTiO₃.

The cosupply of elemental Sr and a metalorganic precursor for V in the presence of a molecular oxygen flux has resulted in the growth of high quality SrVO₃. Films grown at varying VTIP:Sr fluxes on LSAT were comprehensively analyzed by RHEED, XRD, HAADF-STEM, STEM-EDS, and AFM. Phase-pure and stoichiometric SrVO₃ were obtained at a VTIP:Sr ratio of 16.08. Films grown at VTIP:Sr flux ratios deviating from the ideal value revealed the incorporation of Sr-rich and V-rich defects into the films, as well as a segregation of islands on the surface to accommodate the excess cation supply. A means to grow and minimize defects in SrVO₃ provides a method for engineering and designing novel functional properties of SrVO₃ and related systems.

C.E. and J.A.M. acknowledge funding provided by ONR Grant No. N00014-11-1-06655 and the National Science Foundation through the Penn State MRSEC Program DMR-1420620 for film growth, data analysis and preparation of the manuscript. M.B. acknowledges support from the Department of Energy (Grant DE-SC0012375) for data analysis and preparation of the manuscript. EDG was supported by the National Science Foundation through a Graduate Research Fellowship (Award No. DGE-1252376). The authors acknowledge the use of the Analytical Instrumentation Facility (AIF) at North Carolina State University, which was supported by the State of North Carolina and the National Science Foundation.