Metastable honeycomb SrTiO3/SrIrO3 heterostructures (2024)

T. J. Anderson;

S. Ryu

;

H. Zhou;

L. Xie;

J. P. Podkaminer

;

Y. Ma;

J. Irwin

;

X. Q. Pan;

M. S. Rzchowski;

C. B. Eom

Article history

Advances in the design and growth of complex oxides have led to the discovery of complicated physical phenomena and the realization of intricate engineering functionality, often accessible only in metastable materials.1 Of these oxide systems, iridates of pyrochlore A₂Ir₂O₇ (A = metal) and Ruddlesden–Popper Sr_n+1Ir_nO_3n+1 phases have generated considerable excitement due to a plethora of predicted and observed exotic physical properties believed to arise from the interplay between crystal field interactions, localized electron–electron correlations, and strong spin–orbit coupling.2–11 Of particular interest to us is the perovskite phase of SrIrO₃, which has attracted significant attention in the community due to recent theory predictions that it exhibits non-trivial band topologies when sandwiched in between (111) perovskite SrTiO₃ layers.7,8 Because these calculations are based on ultra-thin cubic perovskite SrIrO₃ rather than its bulk monoclinic phase, fabrication of metastable phase thin films is the necessary gateway to probe these phenomena.

However, the equilibrium crystal structure of SrIrO₃ complicates the fabrication of (111) perovskite SIO thin films. In bulk,12 SrIrO₃ assumes a monoclinic structure of space group C2/c that is energetically favorable at room temperature and atmospheric pressure (a = 5.604 Å, b = 9.62 Å, c = 14.17 Å, β = 93.65°), shown in Figure 1(a). However, a perovskite orthorhombic structure (a = 5.60 Å, b = 5.58 Å, c = 7.89 Å) of space group Pbnm is stable at high temperature and pressure, shown in Figure 1(b). A crucial distinction between these two bulk phases is that the monoclinic structure exhibits both face and corner IrO₆ octahedra sharing, whereas the orthorhombic perovskite structure only exhibits corner sharing. Given this exclusive corner octahedral sharing in the orthorhombic case, it is helpful to redefine the unit cell as a pseudocubic13 cell with a lattice parameter of ∼3.95 Å, which is close to those of conventional ABO₃ perovskite-structure [Figure 1(c)] substrates. This similar structural relationship has allowed synthesis of SrIrO₃ on perovskite (001)-oriented single-crystal substrates such as LSAT (3.87 Å), SrTiO₃ (3.90 Å), DyScO₃ (∼3.94 Å), GdScO₃ (∼3.96 Å), and NdScO₃ (∼3.99 Å) in single-layer films11,14–16 and superlattices.17 However, growth on (111) SrTiO₃ has produced (001) monoclinic SrIrO₃.18,19 This is possibly explained by the fact that (001) monoclinic SrIrO₃ is equivalent to a distorted 6H hexagonal structure12 with in-plane hexagonal lattice parameters similar to the buckled hexagonal in-plane lattice of (111) perovskites [Figure 1(d)]. While the fabrication of perovskite iridate (111) films has been demonstrated by tuning the iridate lattice constant by doping the A-site cation,19 pure perovskite (111) SrIrO₃ remains unreported. The (111) stacking of perovskites consists of two planes, AO₃ and B, that together constitute the (111) surface-normal repeating unit [Figure 1(e)], which we will hereafter refer to as a “bilayer.” Two of these bilayers (with spacing of ∼2.28 Å and ∼2.25 Å for perovskite SrIrO₃ and SrTiO₃, respectively) form the buckled honeycomb symmetry in (111) perovskites. Since monoclinic SrIrO₃ is evidently more stable on (111) perovskite substrates, any critical thickness of coherent perovskite SrIrO₃ on (111) SrTiO₃ is expected to be quite thin. Therefore, our approach was to grow ultra-thin films on the thickness order of several bilayers to ensure that our SrIrO₃ would be perovskite phase. Finally, we capped the SrIrO₃ with SrTiO₃ to impose symmetric structural boundaries on the SrIrO₃ films. Our efforts to stabilize perovskite (111) SrIrO₃ thicker than 6–7 bilayers produced films with mixed amorphous and polycrystalline phases, so it would indeed seem that the critical thickness of single crystal perovskite SrIrO₃ is quite thin on (111) perovskite substrates compared to (001).

The SrTiO₃/SrIrO₃ thin film heterostructures were grown using pulsed laser deposition (PLD) with a KrF 248 nm excimer laser. The growth was monitored in situ with high-pressure reflection high energy electron diffraction (RHEED). Before deposition, the substrate surfaces were carefully treated by immersion in a buffered HF acid solution followed by annealing at 1000 °C for 6 h in a quartz tube furnace with flowing oxygen followed by a second immersion in the buffered HF. This assured that our substrates were atomically smooth with single-atom termination at the surface. The SrIrO₃ ceramic target used for the film growth was manufactured commercially by the spark-plasma sintering method, and the SrTiO₃ films were grown from a single-crystal commercial target. The SrIrO₃ films were grown at a substrate temperature of 550 °C in an O₂ partial pressure of 10⁻¹ mbar, and a target-to-substrate distance of 58 mm. The laser energy density was 1.5–2 J/cm² and the pulse repetition rate was 3 Hz. The capping SrTiO₃ bilayers were deposited in the same environment as the SrIrO₃, with a repetition rate of 5 Hz. This was done to prevent damage to the SrIrO₃ film quality by changing the chamber environment.

Figure 2 summarizes the growth of both bare SrIrO₃ films and SrTiO₃/SrIrO₃ heterostructures. The clear in situ RHEED specular spot intensity oscillations [Figure 2(a)] indicate layer-by-layer growth of our SrIrO₃ films. A single oscillation corresponds to one bilayer of SrIrO₃. Thus, the data in Figure 2(a) represent a 3 bilayer SrIrO₃ film. The RHEED image of the SrIrO₃ film surface in the right insets of Figure 2(a) exhibits streaky patterns over the substrate diffraction spots, also an indication of layer-by-layer growth. The left inset in Figure 2(a) shows the RHEED image of our SrTiO₃ substrate right before growth. The AFM surface morphology for the bare SrIrO₃ film in Figure 2(c) shows clear substrate step-terrace preservation with step heights on the order of a single perovskite bilayer spacing. As shown in Figure 2(b) and its insets, the SrTiO₃ growth does not show layer-by-layer growth on a similarly ultra-thin SrIrO₃ film, as no oscillations were observed and the RHEED pattern in the right inset shows spots rather than streaks, indicating a rougher capping SrTiO₃ surface than the underlying SrIrO₃ surface. This is likely attributable to the fact that we grew our SrTiO₃ at the same conditions as the SrIrO₃, which differ from those used for epitaxial SrTiO₃ growth. Nonetheless, the step-terrace structure from the underlying substrate and bare SrIrO₃ film is still preserved. Examination of the RHEED images of either film reveals no noticeable half-order peaks between the primary diffraction spots that would suggest a lower symmetry than that of the cubic SrTiO₃ substrate. It should also be noted that the terrace step height of the SrTiO₃/SrIrO₃ heterostructure is smaller than that of the bare SrIrO₃ film, which is expected given that the SrTiO₃ (111) bilayer spacing is smaller than that of SrIrO₃.

We determined the symmetry and epitaxial arrangement of the heterostructures using both scanning transmission electron microscopy (STEM) and synchrotron surface X-ray diffraction. First, STEM experiments on a 3 bilayer SrIrO₃ film capped with 5 bilayers of SrTiO₃ were performed. Figure 3(a) shows the filtered cross-sectional STEM high-angle annular dark field (HAADF) image of the heterostructure. The contrast of the STEM-HAADF image is approximately proportional to the atomic number Z, where brighter colors represent heavier elements. Here, we see that the brightest atoms, iridium (highest Z), are confined to the B-site columns in 3 bilayers. This indicates sharp SrIrO₃ interfaces with the SrTiO₃ substrate and the capping SrTiO₃ film. Additionally, these layers have the same stacking as those in the SrTiO₃ substrate, more easily seen in the magnified image and structural cartoon in Figure 3(b). This layering should be immediately distinguishable from the layering of monoclinic SrIrO₃ depicted in Figure 1(a). From this image, we extracted the FFT diffraction pattern, shown in Figure 3(c), which shows the symmetry of the entire film–film–substrate heterostructure. The FFT shows a pattern that is characteristic of the much thicker SrTiO₃ substrate with no signature of additional lower-symmetry diffraction spots from the film layers. This suggests good coherence of the SrTiO₃/SrIrO₃ heterostructure with the underlying SrTiO₃ substrate, which would indicate equivalent in-plane symmetry for the entire 3 layers. However, given how thin our SrIrO₃ films were, we could not reliably confirm their symmetry solely from STEM due to the resolution limits of the FFT. Thus, from STEM, we were able to determine that our heterostructure exhibited sharp interfaces and shared the same epitaxial arrangement as the SrTiO₃ substrate, but more work was needed to confirm the heterostructure symmetry.

To unambiguously determine the film symmetry, we carried out surface X-ray diffraction measurements in the 33-ID-D beamline at the Advanced Photon Source with the samples mounted on a Kappa 4 + 2 circle diffractometer along with a Pilatus 100 K 2D pixel area detector. Our approach was to start with an ansatz of 3-fold in-plane symmetry for the entire heterostructure (what we believed from our STEM work) and scan 2 sets of crystal truncation rod (CTR) families carefully chosen to exhibit in-plane 3-fold symmetry under our ansatz. From Figure 4(a), one CTR family, {11L}, intersects the reciprocal-space nodes of the hexagonal perovskite (111) system, whereas the other, {21L}, intersects the nodal midpoints. If the heterostructure indeed had 3-fold symmetry in-plane, the CTR intensity features at each reciprocal space point for all 3 symmetry equivalent CTRs in the family would be identical. If not, the 3 spectra would exhibit distinct intensity differences at equivalent reciprocal space points due to diffraction points originating from a crystal structure different from that of the cubic perovskite substrate. This would indicate lattice anisotropy in the SrIrO₃ layers, and hence, a different symmetry. Since these CTRs are quite sensitive to atomic layering, any deviations from the expected buckled honeycomb structure would be easily discernable from the intensity features, and any different phases present would be thus distinguishable. Thus, by choosing these 2 CTR families, the in-plane symmetry of our heterostructures would be unequivocally evident.

In Figures 4(b) and 4(d), we see the 3 symmetry equivalent CTRs superimposed on the same plot of the {11L} and {21L} families, respectively. Overall, the intensity spectra for the 3 rods in each family are virtually indistinguishable and, thus, indicate no structural difference along each CTR. The most deviation between equivalent CTRs (but still within experimental error) is observed in the mid-zone region halfway between the Bragg rods, but this region is more sensitive to surface atomic structure rather than the entire heterostructure and its in-plane symmetry. Of the utmost importance for our study is the CTR intensity at the reciprocal space points near the substrate Bragg peaks, which are structural indications of the overall heterostructure. From the magnified regions of these spectra [Figures 4(c) and 4(e)], we see that it is impossible to distinguish any difference between the 3 CTRs for each family near the substrate Bragg peak. Since the footprint of the SrTiO₃ substrate, known to have the (111) buckled honeycomb symmetry we were looking for (as previously discussed), is present in these scans, we can conclude that the SrTiO₃/SrIrO₃ films have that equivalent in-plane structure and layering, and no other symmetries or phases exist. Therefore, the measured spectra of these CTR families give strong constraints on the overall in-plane pseudocubic perovskite symmetry of our heterostructures.

In conclusion, we have demonstrated the synthesis of metastable pseudocubic perovskite SrTiO₃/SrIrO₃ heterostructures on (111) SrTiO₃ substrates using pulsed laser deposition with in situ high pressure RHEED. Our STEM and synchrotron surface X-ray diffraction results demonstrate the ability to stabilize honeycomb SrIrO₃ on (111) SrTiO₃. Now that perovskite (111) SrIrO₃ synthesis has been confirmed, it may be interesting to examine these heterostructures with surface-sensitive angle resolved photoemission spectroscopy (ARPES) to discern if the predicted non-trivial band topologies are realized. Additionally, the role of electron correlations on the electronic properties of (111) SrIrO₃ bilayers has recently been predicted to be significant,20 so this can also be studied now that these metastable structures can be synthesized. Thus, our work has overcome the initial obstacle of (111) perovskite SrIrO₃ synthesis and serves as the groundwork to proceed with further experiments on this material system.

This research was supported by the National Science Foundation under DMREF Grant No. DMR-1234096.

This research used resources of the Advanced Photon Source, a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory under Contract No. DE-AC02-06CH11357.

References