Novel synthesis approach for “stubborn” metals and metal oxides

As a demonstration of the potential of this synthesis technique for the growth of ultra-low vapor pressure metals and metal-containing oxides, we chose to grow a variety of materials, including the simple metal Pt, binary oxide RuO₂, and complex oxide SrRuO₃, using the metal-organic precursors platinum(II) acetylacetonate, Pt(acac)₂, and ruthenium(III) acetylacetonate, Ru(acac)₃. Pt films have been useful as electrodes for dielectric materials due to the high work function, and recent interest has also included the study of the spin Hall effect, where films low in defects are desired (24). RuO₂ has been studied for its low resistivity and high chemical and thermal stability for applications such as electrodes in supercapacitors (25), catalysts in reactions like the oxygen evolution reaction (26), and, recently, for the discovery of superconductivity when strained (10). Finally, SrRuO₃ has been a material of significant interest due to its itinerant ferromagnetism, presence of the anomalous Hall effect, and for its use as electrodes in oxide electronics (27). Our work opens up a method for tackling the highly rewarding problem of stabilizing challenging metals like these for further use and study of these exciting phenomena and applications.

The metal-organics Pt(acac)₂ and Ru(acac)₃ were identified as suitable precursors due to their vapor pressures falling in the desired range at low temperatures, about 10⁻³ Torr at 100 °C (28). In fact, we were able to grow films here with source temperatures of only 65 to 85 °C for Pt and 100 °C for RuO₂. Fig. 1B shows the vapor pressure of Pt(acac)₂ and Ru(acac)₃ compared with metallic elements Ba and Sr and volatile metal-organics vanadium(V) oxytriisopropoxide, titanium(IV) tetraisopropoxide, and hexamethylditin, commonly used in MBE. Pt(acac)₂ and Ru(acac)₃ are also solids at these temperatures, with a melting point of 240 °C (29) and 260 °C (30), respectively, which is important for their ease of use in an effusion cell. Finally, Pt(acac)₂ and Ru(acac)₃ are thermally stable until about 210 °C and 230 °C (28), respectively, indicating thermal decomposition does not occur at typical growth temperatures. A schematic of the technique setup is shown in Fig. 1C, involving a low-temperature effusion cell for the metal-organic, oxygen plasma [needed for (001) crystal orientation control in Pt (31)], and an additional conventional effusion cell for expansion to more complex materials, such as using Sr for the SrRuO₃ grown here.

Pt films were grown on SrTiO₃ (001) substrates using a Pt(acac)₂ source temperature of only 65 °C. The low-temperature sublimation led to films with structural characteristics very similar to previous reports of Pt films grown on SrTiO₃ (32–34). Phase-pure, epitaxial (001) oriented films were obtained for substrate temperatures ranging from 520 to 930 °C, shown in SI Appendix, Fig. S1, with the (111) orientation occurring at 400 °C. However, we did not find atomically smooth surfaces in these films. In fact, growing atomically smooth Pt films on dielectric substrates such as SrTiO₃ has historically been difficult due to the favorable agglomeration of metals on dielectric materials (35). Growing atomically smooth films, however, is desirable for heterostructure applications. Instead, here, we found that by changing to a conducting substrate, Nb-doped SrTiO₃, we were able to grow Pt films with these desired surfaces. We attribute it to the change in growth kinetics owing to an enhanced surface diffusivity of Pt adatoms on a conducting substrate. A 70-nm Pt film with a single peak in high-resolution X-ray diffraction (HRXRD) is shown in Fig. 2A. Steps are seen on the surface from atomic force microscopy (AFM) images (Fig. 2B), with a step height of one Pt unit cell, showing this technique is able to grow films with an atomically smooth surface. Similar reflection high-energy electron diffraction (RHEED) patterns were obtained as those films grown at high substrate temperatures on undoped SrTiO₃ (SI Appendix, Fig. S1) but with more well-defined streaks, attesting to the high crystalline quality and smooth surface.

To test the electronic properties of the Pt films, a 70-nm film was grown at an intermediate SrTiO₃ substrate temperature of 760 °C along with a source temperature of 85 °C to increase the growth rate from 9 to 35 nm/h, thereby showing a scalable growth rate. Temperature-dependent resistivity measurements were performed in van der Pauw geometry (36) down to 1.8 K, as can be seen in Fig. 2C. A room-temperature resistivity (ρ) of 15 μΩ·cm was obtained, slightly larger than the bulk Pt value of ∼11 μΩ·cm. The residual resistivity ratio (RRR), defined here as ρ(300 K)/ρ(1.8 K), was 27. The resistivity versus temperature behavior was then fit with the Bloch–Grüneisen equation (37),whereThe Bloch–Grüneisen equation models the resistivity of metals considering contributions from electron–electron scattering (ρ_ee), interband scattering (ρ_sd), and intraband scattering (ρ_ss) in addition to the residual resistivity (ρ₀). We found the intraband and electron–electron scattering term to be negligible. Excellent fits were obtained using only the ρ₀ and contribution from ρ_sd terms. Debye temperature Θ_D of 232 K was determined from the fitting, agreeing well with the bulk value of 240 K (38). These observations attest to the bulk-like electronic properties and high quality of Pt thin films grown using this approach.

To show that this technique does not only apply to simple metals like Pt, thin films of the metallic binary oxide RuO₂ were grown using Ru(acac)₃ with a source temperature of only 100 °C. Epitaxial and phase-pure films were obtained at the low substrate temperatures of 300 to 400 °C on TiO₂ (110) and TiO₂ (101) substrates. HRXRD patterns seen in Fig. 3 A and B show the single crystalline (110) or (101) orientation, respectively, for the RuO₂ films grown at 300 °C, with a thickness of 12 nm and 16 nm, respectively, obtained from the finite-thickness fringes. Streaky RHEED patterns (SI Appendix, Fig. S2) were seen along with atomic steps from AFM for films grown on TiO₂ (110), once again displaying the atomically smooth films this growth process allows for. Temperature-dependent resistivity measurements as shown in Fig. 3C indicate metallic behavior and resistivities approaching that of bulk RuO₂, with a lower resistivity for films grown on TiO₂ (101). The 16-nm film on TiO₂ (101) shown here had ρ₀ = 6 μΩ·cm and RRR = 8.

As a function of thickness, RuO₂ films grown on TiO₂ (110) revealed a decreasing resistivity with increasing thickness, a common result for thin metallic films that is likely due to finite size effects such as defects at the film/substrate interface or film’s surface. As film thickness was increased beyond 18 nm, a large anisotropy occurred between the resistance along the [001] and $\left[1\overline{1}0\right]$ directions, making four-terminal resistivity measurements difficult. The reason for this anisotropy is likely due to cracking of the film under tensile strain, as was observed in the previous report (10). Although both RuO₂ and TiO₂ have the rutile crystal structure, there is a large lattice mismatch of +2.3% in the $\left[1\overline{1}0\right]$ direction, causing a significant tensile strain on RuO₂. Regardless, RuO₂/TiO₂ (110) showed record-low residual resistivities, comparable to recently reported e-beam MBE-grown films (10, 39), with values approaching 13 μΩ·cm (Fig. 3D). We do recognize, however, a small difference in the residual resistivities is likely due to the difference in the epitaxial strain.

Finally, applying the solid source MOMBE technique to more complex materials, SrRuO₃ thin films were grown on SrTiO₃ (001) substrates. At a substrate temperature of 665 °C and, again, a Ru(acac)₃ cell temperature of 100 °C, phase-pure and epitaxial films were obtained. In Fig. 3E, the HRXRD pattern of a 16-nm film can be seen with a single film peak and finite-thickness fringes. Reciprocal space maps around the (103) peak showed a coherently strained film at this thickness consistent with the fully strained orthorhombic phase.

Metallic behavior was present for the entire temperature range (1.8 to 300 K) in this SrRuO₃ film, with a kink in the slope occurring at 149 K (Fig. 3F), around the temperature (∼160 K in bulk) where ferromagnetism is expected to appear (27). The onset of ferromagnetism was confirmed by a nonzero magnetic moment below 149 K (Fig. 3F) and a well-defined ferromagnetic hysteresis loop below this temperature, such as that seen in SI Appendix, Fig. S3 at 75 K. The saturation magnetization of ∼0.36 μ_B/Ru achieved on cooling in a magnetic field of 2 T (SI Appendix, Fig. S3) along the in-plane [100] azimuth of the substrate is comparable to previously reported values from SrRuO₃ films (40). The ordinary and anomalous Hall effect were also present, as expected, and can be seen in the hysteretic Hall resistance as a function of the applied magnetic field at 75 K in Fig. 3G. Comparing with SrRuO₃ films grown by PVD techniques, the RRR of 18 obtained here is larger than any reported PLD-grown (41) or sputtered films (42) but not quite as large as the highest reported MBE films (43) or bulk single crystals (44). As these films were grown with a precursor purity of only 97%, we believe there is still room for improvement with further growth optimization and by decreasing point defects with a higher purity starting material.

The ability to synthesize atomically controlled materials continues to drive modern technology and fundamental study. Andre Geim and Kostya Novoselov were awarded the 2010 Nobel Prize in Physics “for groundbreaking work regarding the two-dimensional material graphene.” It was the simplicity of exfoliation as a synthesis method that made it possible for them and for a wide scientific community to access this material experimentally. This aspect of their exfoliation approach has resulted in the exponential rise in the work involving graphene with over 200,000 publications so far. To put the potential broader impact of the solid source MOMBE approach into context, we first show in Fig. 4A the number of publications (blue bars) on different materials containing Pt, Ru, Ir, or W. A large number of publications exceeding tens of thousands highlights their relevance for both fundamental study and technological applications. In contrast, however, only a few thousand are relevant to thin films (green bars) and, even more surprisingly, only a few tens of them are on films grown using MBE. Therefore, this statistic clearly illustrates a large gap between material demand and our capability to synthesize them in an atomically controlled fashion. We recognize this difference is due to these materials containing elements with ultra-low vapor pressures. Furthermore, many of these low vapor pressure elements also possess low oxidation potential (or high reduction potential), which makes their oxidation nontrivial, especially in low-pressure PVD techniques. Fig. 4B shows standard reduction potential (E₀) of these select elements as a function of temperature at which their vapor pressure is 10⁻³ Torr (i.e., suitable for their MBE growth). Our solid source MOMBE technique therefore aims to close this publication gap by allowing for low-temperature sublimation (at T < 100 °C) instead of the extremely high temperatures (typically 2,000 to 3,000 °C) in e-beam evaporation along with the added benefits of their preoxidized state in the metal-organic precursor. As an example of the latter, we have already shown the growth of epitaxial RuO₂ and SrRuO₃, where Ru in its preoxidized 3+ state in Ru(acac)₃ can be oxidized to the desired 4+ state using oxygen plasma, a weaker oxidant than ozone, which is typically used.

Finally, by being such a low-temperature process, the complexity and cost of supplying the metal is decreased while maintaining the quality common to MBE-grown materials. For example, keeping in mind that there are differences in substrates and thicknesses, Fig. 4C shows RRR reports for Pt films grown by PVD techniques. The solid source MOMBE-grown Pt film has the largest RRR while only requiring relatively inexpensive and simple low-temperature sublimation compared with the process of sputtering a Pt target, controlling an electron beam for evaporation, or managing a laser in PLD. We believe even larger RRR values can easily be achieved using this technique by increasing the purity of the source materials, as the ones used here were only 97% pure, and the source purity can greatly affect the residual resistivity.

In summary, the solid source MOMBE approach we have developed is a unique route to supply “stubborn” metals such as Pt or Ru at operating temperatures less than 100 °C, which can be achieved in a low-temperature effusion cell as opposed to the several thousand degrees Celsius needed using e-beam evaporators. To illustrate the capability of the solid source MOMBE approach, we have demonstrated the growth of single crystalline Pt and RuO₂ films as well as the more complex SrRuO₃ films, all containing an ultra-low vapor pressure metal. Pt films with a record-high RRR value were achieved, whereas RuO₂ films revealed a record-low residual resistivity, attesting to the ability of our approach to grow high-quality materials in an atomically controlled fashion. Likewise, SrRuO₃ films with only 97% pure Ru precursor yielded an RRR value of 18, which is already higher than those grown using PLD and sputtering approaches. We argue that the solid source MOMBE approach may potentially open up new PVD pathways (not just MBE) for “stubborn” materials such as delafossites, iridates, and tungstates containing elements that have low vapor pressures and are hard to oxidize in low-pressure PVD approaches.

An effusion cell (E-Science, Inc.) was used for the low-temperature, 65 to 85 °C, sublimation of Pt(acac)₂ (97%; MilliporeSigma) and 100 °C sublimation of ruthenium(III) acetylacetonate, Ru(acac)₃ (97% for RuO₂ and SrRuO₃; MilliporeSigma, and 99.99% for RuO₂; American Elements). Sr was also supplied by an effusion cell for the growth of SrRuO₃. The powder precursors were placed directly in a pyrolytic boron nitride crucible (E-Science, Inc.) inside an effusion cell. A beam equivalent pressure of ∼2 × 10⁻⁷ Torr was measured for Pt at 65 °C and ∼1 × 10⁻⁷ Torr for Ru at 100 °C. Both materials were grown in an oxygen environment, as it has been shown necessary for the stabilization of the (001) epitaxial orientation (31) in Pt and to ensure complete oxidation of RuO₂ and SrRuO₃. Oxygen was supplied at a pressure of 5 × 10⁻⁶ Torr by a radio frequency plasma source (Mantis) operated at 250 W and with charge deflection plates. Pt films were grown on SrTiO₃ (001) and Nb-doped SrTiO₃ (001) single-crystal substrates (Crystec GmbH). RuO₂ films were grown on TiO₂ (110) and TiO₂ (101) (MTI Corporation). SrRuO₃ films were grown on SrTiO₃ (001) (Shinkosha). Substrate temperatures were 760 °C for Pt, 300 °C for RuO₂, and 665 °C for SrRuO₃. Following growth, the films were cooled in an oxygen plasma environment. The use of oxygen plasma during growth prevented the formation of oxygen vacancies in SrTiO₃ substrate. It was confirmed by performing a controlled experiment by subjecting a bare SrTiO₃ substrate under the same thermal cycle in a vacuum showing no measurable conductivity. By placing ohmic contacts on the backside of the SrTiO₃ substrate, we further ensured the substrate remained insulated after growth.

RHEED (Staib Instruments) was carried out in situ during and after growth to monitor the film surfaces. Film surfaces after growth were also characterized using AFM (Bruker). HRXRD and reciprocal space maps were performed using a SmartLab XE X-ray diffractometer (Rigaku). Temperature-dependent four-terminal resistivity measurements were performed down to 1.8 K, using indium as an ohmic contact for Pt and aluminum for RuO₂ and SrRuO₃ in a DynaCool Physical Property Measurement System (PPMS, Quantum Design). Hall measurements were performed using a four-quadrant sweep of magnetic field between −9 T and +9 T normal to the film surface in the DynaCool PPMS. Temperature and magnetic field–dependent magnetization measurements were performed with a magnetic field applied along the in-plane [100] azimuth of the SrTiO₃ substrate using the Vibrating Sample Magnetometer option in an EverCool Physical Property Measurement System (Quantum Design). For temperature-dependent magnetization measurements, sample was cooled under zero applied magnetic field while the data were taken upon warming.

We thank Fengdeng Liu for help with the MBE schematic and Darrell Schlom for helpful discussion. This work was primarily supported by the US Department of Energy (DOE) through Grant DE-SC002021. A.K.M. and T.K.T. acknowledge support from the Air Force Office of Scientific Research through Grants FA9550-19-1-0245 and FA9550-21-1-0025 and partially through NSF Grant DMR-1741801. J.Y. and A.R. acknowledge support from the US DOE through the University of Minnesota Center for Quantum Materials under Award DE-SC0016371. Parts of this work were carried out at the Characterization Facility, University of Minnesota, which receives partial support from NSF through the Materials Research Science and Engineering Centers (MRSEC) program under Award DMR-2011401.