Solid Source Metal-Organic Molecular Beam Epitaxy of Epitaxial RuO2

A seemingly simple oxide with a rutile structure, RuO2 has been shown to possess several intriguing properties ranging from strain-stabilized superconductivity to a strong catalytic activity. Much interest has arisen surrounding the controlled synthesis of RuO2 films but, unfortunately, utilizing atomically-controlled deposition techniques like molecular beam epitaxy (MBE) has been difficult due to the ultra-low vapor pressure and low oxidation potential of Ru. Here, we demonstrate the growth of epitaxial, single-crystalline RuO2 films on different substrate orientations using the novel solid-source metal-organic (MO) MBE. This approach circumvents these issues by supplying Ru using a pre-oxidized solid metal-organic precursor containing Ru. High-quality epitaxial RuO2 films with bulk-like room-temperature resistivity of 55 micro-ohm-cm were obtained at a substrate temperature as low as 300 C. By combining X-ray diffraction, transmission electron microscopy, and electrical measurements, we discuss the effect of substrate temperature, orientation, film thickness, and strain on the structure and electrical properties of these films. Our results illustrating the use of novel solid-source MOMBE approach paves the way to the atomic-layer controlled synthesis of complex oxides of stubborn metals, which are not only difficult to evaporate but also hard to oxidize.


INTRODUCTION
RuO2 has gained considerable attention for the rich material properties it exhibits. Additionally, RuO2 also serves as a precursor to the growth of more complex materials such as perovskite SrRuO3 and Sr2RuO4, which are shown to exhibit itinerant ferromagnetism and unconventional superconductivity, respectively.19][20][21][22][23][24] Molecular beam epitaxy (MBE) is one of these thin film techniques, which has been used for the growth of epitaxial single crystalline RuO2 [10][11][18][19][20] . In geneal, the growth of Ru-based oxides in oxide MBE is challenging due to the ultra-low vapor pressure and low oxidation potential of Ru.Temperatures exceeding 2000°C are needed to achieve a suitable Ru vapor pressure for growth which is why reports of MBE-grown RuO2 have all used electron-beam (e-beam) evaporators.Furthermore, the low oxidation potential makes stabilizing the Ru 4+ in RuO2 difficult and has led to ozone, a highly oxidizing source, being the preferred oxidant in most reports.While the use of e-beam source and ozone has facilitated synthesizing epitaxial RuO2 or other Ru-based oxides, they possess several challenges associated with the issues related to the instability of fluxes, and source oxidation in the presence of ozone.
In this paper, we demonstrate a novel solid source metal-organic molecular beam epitaxy (MOMBE) approach using a detailed RuO2 growth study that addresses these synthesis issues by using a solid metal-organic precursor as the metal Ru source. 25This allows for supplying a "preoxidized" metal with orders of magnitude higher vapor pressure than the elemental metal at a particular temperature.

EXPERIMENTAL SECTION
RuO2 films were grown using the solid source MOMBE approach which has been described in more detail elsewhere. 25Here, an effusion cell source temperature of 100°C was used for Ru(acac)3 and oxygen was supplied with a radio-frequency inductively coupled plasma source.An oxygen background pressure of ~ 10 -6 -10 -5 Torr was used.Substrate temperatures were 300°C, unless stated otherwise.Films were grown on r-plane sapphire r-Al2O3, TiO2 (101), TiO2 (110), TiO2 (001), and TiO2 (100) substrates.
Film surfaces were monitored before, during, and after growth using reflection high-energy electron diffraction (RHEED, Staib Instruments).Surface morphologies were imaged post-growth using atomic force microscopy (AFM, Bruker).Structural characterization and determination of film thickness was carried out using high-resolution X-ray diffraction (HRXRD), reciprocal space mapping (RSM), and grazing incidence X-ray reflectivity (GIXR, Rigaku SmartLab XE).
Thickness was also alternatively determined by finite thickness fringes if present in the HRXRD scans.X-ray photoelectron spectroscopy (XPS, Physical Electronics VersaProbe III) was used for determine the Ru oxidation state.Four-probe resistivity measurements were performed in the Van der Pauw geometry as a function of temperature (Quantum Design DynaCool Physical Property Measurement System).Ohmic contacts were made using aluminum wire bonds.
Cross-sectional Scanning Transmission Electron Microscopy (STEM) samples were prepared by a Focused Ion Beam (FIB) method using FEI Helios Nanolab G4 dual-beam FIB, where lamellae were cut and thinned using 30 kV Ga-ion beam and further polished with 2 kV Gaion beam.Low-magnification high-angle annular dark-field STEM (HAADF-STEM) images were acquired using Thermofischer Talos F200X, and atomic-resolution HAADF-STEM images and Energy Dispersive X-ray (EDX) elemental maps were obtained using aberration-corrected FEI Titan G2 60-300 equipped with a super-X EDX detector.The microscopes were operated at 200 keV and screen current was ~25 pA.Probe semi convergent angle was 10.5 mrad and 17.3 mrad for Talos and Titan microscopes, and detector angles for HAADF-STEM images were in the range of 55-110 mrad.

RESULTS AND DISCUSSION
Figure 2a shows the vapor pressure of Ru metal in comparison to the precursor used for Ru in this study, Ru(acac)3. 26We also illustrate in figure 2b the important factors relevant to thin film synthesis by comparing them between e-beam-assisted MBE, hybrid MBE, and the solid-source MOMBE.Clearly, besides the advantages of high vapor pressure and the pre-oxidized state of the metal, the solid-source MOMBE also does not use a liquid precursor like hybrid MBE. 27The solid metal-organic precursor can be sublimed in a conventional low temperature effusion cell directly in the vacuum system instead of a relatively complex gas inlet system.The low temperature effusion cell is also significantly less expensive, less complicated, and safer to operate than an ebeam source.The Ru(acac)3 precursor itself comes with an additional source of oxygen bonded in the ligands, is air stable, and non-toxic, removing some safety concerns that can come with the use of metal-organic precursors like hexamethylditin (HMDT) in hybrid MBE growth of Sncontaining compounds. 28ing solid-source MOMBE approach, we first examined the effect of substrate temperature on the growth of RuO2.RuO2 films were grown on r-Al2O3 with substrate temperatures (Tsub) from 300°C to 850°C, for a fixed growth time.All films were epitaxial and phase pure with a single peak corresponding to RuO2 (101) orientation, the common epitaxial orientation for rutile films on r-Al2O3 as shown in the HRXRD scans in figure 3a.. 29 As Tsub was increased, the growth rate increased, figure 3b, which led to differences in thickness of the films from 7 -17 nm.As will be discussed it later, the films grown at Tsub = 750°C and 850°C had surfaces too rough to determine a reliable thickness using GIXR.However, an estimate of the growth rates, and therefore thicknesses, is given in Figure 3b, obtained from the peak broadening of (101) film peak using the Scherrer formula. 30While the Scherrer analysis can give a poor approximation of the film thickness, thicknesses obtained here agreed well with those from GIXR and HRXRD thickness fringes for T ≤ 650 °C.
The initial increase and later saturation of the growth rate with increasing Tsub indicates a change of growth mechanism from a reaction-limited to a flux-limited regime. 31This suggests that below 650°C the growth rate is limited by the thermal decomposition of the Ru(acac)3 precursor.Above 650°C, the relatively constant growth rate is typical of being limited by the amount of precursor being supplied, or the flux. 31No desorption-limited growth regime, i.e. a decrease in growth rate with increasing temperature, was observed. With increasing Tsub, an increase in the out-of-plane spacing of (101) planes (d(101)) was seen reaching toward the expected bulk value.The change in d(101) with Tsub is most likely due to the strain relaxation.To determine whether strain relaxation was due to the growth rate or film thickness, thicker films were grown at a constant growth rate, by keeping Tsub = 300°C (figure S1).
In this case, even as thickness was increased up to 48 nm, d(101) did not reach the bulk value.For instance, film thickness of 48 nm yielded d(101) = 2.538 ± 0.002 Å, which is significantly less than that of the 17 nm film (d(101) = 2.544 ± 0.002 Å) grown at higher temperatures (and at higher growth rate).This results thus suggest that the strain relaxation is more dominant effect at higher substrate temperature which is also accompanied by the higher growth rates.
Consistent with strain relaxation with increasing temperature, the full width at half maximum (FWHM) of the film (101) rocking curves increased by about an order of magnitude from 450°C to 550°C, as shown in figure 3d.RHEED images taken before growth, 10 minutes into growth, and after growth and cool down in oxygen (Tsub = 200°C), as well as the post-growth AFM images, are shown in figures 3e -3f.From the AFM images, it can be clearly seen that the increase in FHWM was also accompanied by a roughening of the film surface.The difference in the surface morphologies was confirmed by RHEED to be a result of a change in the growth mode during growth.At 10 minutes of growth, considerable differences in the RHEED patterns can be seen for these films grown at higher temperatures, with a change to an island growth mode.
Having identified the optimal substrate temperatures of 300°C -450°C, RuO2 films were grown at 300°C on TiO2 substrates with different orientations.HRXRD scans, figure 4a, confirm phase pure, epitaxial, single crystalline films on all these substrates.Finite thickness fringes are present in all cases, although not very well defined in the case on TiO2 (001), attesting to the high structural quality on a short lateral length scale.To investigate the structure of these films on an atomic scale, we performed STEM imaging of a representative RuO2 film grown on TiO2 (101) along [1 # 01] and [010] zone axes.Consistent with the HRXRD data, phase pure, epitaxial film is seen with a sharp film/substrate interface with no misfit dislocations.The lack of dislocations signifies coherent growth, which agrees well with the strained d101 = 2.51 ± 0.002 Å obtained from HRXRD.EDX elemental maps further attest to a uniform distribution of Ru in the film.
Interestingly, STEM images also reveal an atomically smooth surface along the [010] zone axis whereas a significantly rougher morphology was observed when viewed along [1 # 01] zone axis (figure 4b).As shown in the zoom-in image of figure 4b, the rough surface was found to be terminated not only at the expected (101) plane parallel to the (101) TiO2 substrate but also other plane consistent with (111) face.While the origin of this unusual surface morphology is unclear, and remains a subject of future study, we argue that it may be related to the significantly different strain mismatch of +0.04% and +2.3%, along [1 # 01] and [010] direction, respectively.
As a next step, we investigated the strain relaxation behavior of t nm RuO2 film/TiO2 (110) with t = 3 -26 nm.Theoretically, RuO2 on TiO2 (110) has a relatively large lattice mismatch of about -4.7% and + 2.3% along the [001] and [11 # 0] directions, respectively, indicating coherently strained growth may be challenging on TiO2 (110).Figure 5a shows HRXRD scans of t nm RuO2 film/TiO2 (110) with t = 3 -15 nm revealing thickness fringes and a film (110) peaks being partially overlapped with that of the substrate.The well-defined Kiessig fringes again attest to the highquality film on a short lateral length scale.Upon analysis of the film peak position, 26 nm RuO2 film/TiO2 (110) yielded d(110) = 3.204 ± 0.002 Å, which is larger than the bulk value of 3.176 Å suggesting partially strained films.To examine the strain state of these films along in-plane [001] and [11 # 0] directions, RSMs were taken.Figures 5(b-e) show RSMs around (332) and (310) reflections for two representative films with t = 6 nm and 26 nm.As was expected based on the value of d( 110), the 26 nm sample was partially relaxed with both in-plane spacings, along the [001] and [11 # 0] directions, falling between the expected fully strained and fully relaxed values (figures 5(b-d)).Interestingly, the 6 nm sample showed the same in-plane spacing as the 26 nm sample along the [001] direction (the [11 # 0] direction could not be determined due to overlap with the substrate).These results suggest the strain relaxation begins to occur at t as small as 6 nm or less for film grown on TiO2 (110).Consistent with the strain-relaxation behavior, a broadening of the (220) RuO2 film rocking curve was seen with increasing t.Figures 5f and 5g show the rocking curves of 26 nm and 6 nm film, respectively.Rocking curves were fitted using two Gaussian peaks, marked as a broad and a narrow peak.The results of this fitting are shown in figure 5h and 5i.
Figure 5h shows the narrow peak remained at a relatively constant FWHM of ~ 0.07° while the broad peak FWHM decreased.Taking the ratio of the peak intensity of the broad component (Ibroad) to the total intensity of the two (Itotal = Ibroad + Inarrow), figure 5i, revealed an almost linear increase in this ratio with increasing film thickness.Results from the identical analysis of t nm RuO2 film/r-Al2O3 are also included in figure 5i, which compare well with films grown on TiO2 (110) substrates.
The origin of the broad component can be thought of as being caused by the disorder induced due to the strain relaxation.We know the strain relaxation process has begun by at least 6 nm based on the RSM results, and likely at an even smaller thickness because of the presence of the broad component in the rocking curves of those films as well.As thickness is increased, the volume fraction of film that was influenced by the relaxation increases and, therefore, the intensity of the broad component does as well.Similar results were seen in films grown on r-Al2O3, however, with a faster increase in the intensity ratio with increasing t.This observation is again consistent with faster strain relaxation expected from a larger lattice-mismatch and difference in symmetry between RuO2 and r-Al2O3.
Finally, we turn to the discussion of electrical property of these films revealing a clear correlation between film thickness, strain relaxation and electrical resistivity.Figure 6a shows the temperature-dependent resistivity (ρ) for t nm RuO2 film/TiO2 (110) with t = 3 -15 nm.Films with t > 15 nm showed a large resistance anisotropy between the two in-plane directions for reliable four-terminal resistivity measurements, which is consistent with the prior results 10 .All films showed metallic behavior with increasing resistivity with decreasing t.The room-temperature ρ of 15 nm was 56 μΩ•cm, closest to the 35 μΩ•cm of bulk RuO2 among the films grown on TiO2 (110) 34 .Figure 6b shows the residual resistivities (ρ0 = ρ at 1.8 K) revealing an exponential-like decrease with increasing t, saturating at ~ 13 μΩ•cm.Increasing film thickness can influence the electronic properties by way of finite size effects, such as effects from the film-substrate interface, dimensionality as well as the defects arising from the strain relaxation process, such as misfit and/or threading dislocations.With regards to the latter, as strain relaxation occurs and more defects are formed, the residual resistivity should likely increase as it is generally dependent on these structural crystalline defects.Here we see the opposite trend, implying the increase in thickness has a much larger effect on ρ than strain relaxation-related defect formation.To this end, we plot ρ0 vs. the film rocking curve intensity ratio we defined earlier in Figure 6c.This shows, once again, the opposite trend of what should be seen if strain relaxation was the critical factor.These data establish an important role of film dimensionality on the electrical transport properties of RuO2 films.

CONCLUSIONS
In summary, we have shown the growth of epitaxial RuO2 films of different orientations using different substrates by a novel solid-source MOMBE.Single crystalline films with low film rocking curve FWHMs were grown on r-Al2O3 with substrate temperatures between 300°C -450°C.At higher temperatures, films showed a significant increase in structural disorder.Using this approach and by keeping substrate temperature of 300°C, films were then grown on TiO2 substrates with different orientations.STEM results confirmed phase pure, epitaxial films free of strain-relaxation-related defects when grown on TiO2 (101).However, films on TiO2 (110) were found to relax at thickness as low as 6 nm.Films with resistivity similar to that of the bulk RuO2 single crystals was obtained for 15 nm RuO2/TiO2 (110).Finally, with increasing film thickness, we revealed an important role of film thickness on electrical properties.This work establishes the solid-source MOMBE technique for the growth of high-quality RuO2 in a much simpler, and costeffective manner when compared to conventional MBE approaches.For instance, Ru metal was supplied by subliming a solid metal-organic precursor, Ru(acac)3 in a low temperature effusion cell operating at 100°C as opposed to several thousand using e-beam in conventional MBE.nm and (e) 6 nm film.Film (220) rocking curves for (f) 26 nm and (g) 6 nm film.Gaussian fits are shown for the "narrow" and "broad" peaks.In RSMs, str.and rlx.correspond to a fully-strained and a fully-relaxed position, respectively.Thickness-dependent (h) FWHM of the narrow and broad fits and (i) intensity ratio of broad peak intensity to total intensity for (220) film peaks.

Figure 4 .
Figure 4. (a) HRXRD patterns of RuO2 films grown on a variety of TiO2 substrate orientations,