The deposition of vanadium metal on SrTiO3 results in the spontaneous scavenging of oxygen ions from SrTiO3 to oxidize vanadium to VOx, where x = 0.3–1.1, depending on the deposition temperature. At temperatures above 700 °C, an epitaxial film of the rock salt compound vanadium monoxide is formed on both SrTiO3 (111) and SrTiO3 (100) substrates. Surprisingly, oxygen scavenging and epitaxy persist for thicknesses over 800 Å with no sign of degradation. We describe the growth process and layer-by-layer characterization of the films using in situ reflection high-energy electron diffraction and x-ray photoelectron spectroscopy, as well as ex situ in-plane and out-of-plane x-ray diffraction and cross-sectional scanning transmission electron microscopy. This easy method of growing vanadium monoxide with controlled stoichiometry can provide new opportunities to study this normally hard to synthesize material known for the highly correlated nature of its electronic structure.
Vanadium monoxide (VOx) is one of the many oxides of vanadium, although somewhat less known than vanadium pentoxide or vanadium dioxide due to its relative chemical instability and its lack of significant metal-to-insulator transition or the ability to act as a catalyst in redox reactions.1 However, VOx still possesses interesting properties that made it the subject of experimental and theoretical research in the 1970s.2–4 Interest in these properties tends to stem from its relatively simple cubic rock salt structure (space group Fm-3m) and its highly correlated nature rather than any practical application. These properties also vary somewhat over the allowed values of x, which range from roughly 0.8 to 1.3. Despite its nominal cubic rock salt structure, VOx is generally known to have a relatively high percentage of anion and cation vacancies: up to 15% even for x = 1. These vacancies subsequently affect resistivity and other electronic and transport properties. In fact, resistivity has a variable dependence on the temperature depending on the relative stoichiometry. For x < 1, the resistivity is largely unaffected by even large changes in temperature up to 300 K. However, for x near to and greater than 1, over the same change in temperature, the resistivity can change on the order of a factor of 108. This change is, however, not a phase transition but a gradual crossing over from a metallic to semiconducting behavior controlled mainly by the changing stoichiometry.1 Additionally, the Seebeck coefficient, dictating the thermoelectric voltage, undergoes a shift from positive to negative around x = 1, indicating a change in the dominant carrier type from holes to electrons. More recently, interest in VOx as a material has resurfaced utilizing modern electronic structure theory to better understand the physical properties and the mechanisms behind them.5
Growth of VOx thin films is not a trivial process, owing largely to its tendency to oxidize to V2O3, but studies published in the early 2000s by Rata et al. revealed that single-phase, epitaxial VOx can be achieved by an appropriate choice of the substrate and growth conditions using an ultrahigh vacuum molecular beam epitaxy system (MBE). They determined that MgO (100) was an optimum choice of substrate due to its matching crystal structure with a relatively low lattice mismatch of 3%. The desired oxide was achieved by evaporating pure vanadium metal onto the room temperature substrate, which was simultaneously oxidized by a molecular beam of O2 provided to the system. The total background pressure during the deposition was 2 × 10−8 mbar or lower. The exact stoichiometry of the oxide was tuned by adjusting the O2 buffer pressure, which controlled the arriving oxygen flux.6–8 An alternative approach is the oxidation of vanadium metal deposited on an ordered metallic surface, as exemplified by the study of Petukhov et al.9 where they oxidized vanadium metal on Pt(111) by controlled exposure to water vapor. In both approaches, one needs a very high degree of control of the oxygen source.
We have previously described how the relative ease of forming oxygen vacancies in SrTiO3 (STO) causes many metals deposited on it to steal oxygen and form a metal oxide on top of oxygen-deficient STO.10 Oxygen outdiffusion from STO substrates occurs even when growing oxide films,11–13 as definitively demonstrated by Schneider et al. using 18O tracer diffusion.11 In some cases, the metal oxide formed after depositing a metal on STO can persist for a significant thickness and also form as an epitaxial layer if the temperature is sufficiently high to provide surface mobility.14,15 Vanadium is one such metal that readily takes oxygen from STO and forms an ordered vanadium oxide overlayer. The oxidation of V by an oxide substrate was first reported by Della Negra et al. who deposited vanadium metal on rutile TiO2 (110) at room temperature and subsequently annealed the film and showed some epitaxy using x-ray photoelectron diffraction.16 Subsequent work with more precisely controlled oxygen content was able to achieve single-phase monoxide.17 In this work, we show that epitaxial vanadium monoxide layers can be grown without the need for oxygen control by simply depositing vanadium metal on STO at elevated temperatures. We show how the oxidation state of vanadium varies as a function of layer thickness and how the overall vanadium oxide stoichiometry changes with the deposition temperature.
STO substrates with (111) orientation were utilized for this study. Substrates with dimensions 10 × 10 × 0.5 mm3 were degreased by sonication in acetone, isopropanol, and de-ionized water for 5 min each then dried with N2 gas. The substrates were then loaded and outgassed in a customized DCA Instruments M600 MBE chamber at 700 °C for 10 min prior to vanadium metal deposition. Vanadium metal was evaporated from an electron beam evaporation source operated at an accelerating voltage of 7.75 kV and an emission current of 55–60 mA. This produces a vanadium metal flux of about 2–4 Å/min, which is measured by a quartz crystal microbalance just before each deposition run. In situ reflection high-energy electron diffraction (RHEED) and x-ray photoelectron spectroscopy (XPS) were used to characterize each deposition run of vanadium on STO. RHEED images were obtained using 21 keV electrons at a glancing angle of 4–5°. XPS was measured using a VG Scienta R3000 electron energy analyzer with monochromated Al Kα x rays as the excitation source. The valence band and combined O 1s-V 2p spectra were taken for each run. The binding energy of the XPS instrument was calibrated such that the Ag 3d5/2 core level is at 368.28 eV. No charging was observed in any of the samples.
For the layer thickness dependence study, vanadium metal was deposited on an STO substrate at 800 °C at a rate of 2.8 Å/min for an amount of time needed to obtain 1, 2, 4, 8, 16, 32, 56, and 96 monolayers (MLs), where one ML is defined to be 2.2 Å of the metal.18 For the deposition temperature dependence study, either 20 or 30 nm of vanadium were deposited at rates of 2–4 Å/min on the outgassed STO substrate at 200, 400, 600, and 800 °C. In situ RHEED and XPS measurements were performed for each vanadium deposition run.
Ex situ x-ray diffraction and cross-sectional scanning transmission electron microscopy (STEM) were performed on select samples to confirm the resulting film crystalline structure. A thin 10 Å capping layer of amorphous Si was deposited for such samples to prevent oxidation in ambient conditions. X-ray diffraction (using Cu Ka radiation) was performed both in the out-of-plane and in-plane directions. Out-of-plane scans were taken with an Ni filter to remove the Cu Kβ contribution. In-plane scans were taken parallel to the STO [1−10] azimuth and were taken without any filtering to maximize the signal. Residual peaks due to W Lα x rays can be seen in all the scans. Annular dark field (ADF) STEM imaging was performed using an atomic resolution STEM (JEOL USA, ARM 200F) equipped with a spherical aberration (cs) corrector. The STEM samples were sectioned both along STO [1−10] and STO [11−2] using a dual-beam focused ion beam system (FEI Nova 200) and nanomanipulator (Oxford Instruments Omniprobe 400).
III. RESULTS AND DISCUSSION
A. Vanadium thickness dependence
Figure 1(a) shows in situ XPS spectra of the combined oxygen 1s and vanadium 2p region for various layer thicknesses of vanadium metal deposited on STO (111) substrates at 800 °C. Spectra for 0, 1, 2, 4, 8, 16, 32, and 96 MLs are plotted together. From 1 to 4 ML, a very broad feature for V 2p3/2 centered around 516 eV and about 3 eV wide can be seen emerging and getting stronger as more vanadium is deposited. This feature is attributed to the V3+ oxidation state, consistent with the analysis reported by Hryha et al.19 At 8 ML, a sharper feature centered around 513 eV begins to dominate, which is attributed to the V2+ oxidation state. This feature continues to grow and get sharper with additional vanadium deposition. The asymmetric metallike shape of the V 2p3/2 core level of VO is consistent with the study of Hryha et al. on stoichiometric vanadium oxides.19
There is also a shift in the O 1s binding energy observed as vanadium metal is progressively deposited. The STO oxygen binding energy is around 530.8 eV, with the first few monolayers of vanadium not significantly shifting its position. Oxygen binding energy shifts noticeably toward 531.4 eV at 8 ML, at the same time as the vanadium oxidation state shifts, as shown by the emergence of the 513.7 eV feature in the V 2p spectrum [Fig. 1(b)]. The oxygen intensity also remains relatively unchanged with metal deposition, indicating the presence of oxygen in the accumulating film. Figure 2 plots the O 1s integrated area (after Shirley background subtraction) from the vanadium thickness series of the samples. One can see that the O 1s signal increases from 344 units at 0 ML to 414 units at 4 ML and then settles down to 406 units at 96 ML, about 18% stronger signal than pure STO. This is consistent with VOx accumulating on top of STO rather than V metal burying the oxygen signal. STO has three oxygen atoms per formula unit and one formula unit per unit cell with cell volume of 0.0595 nm3, resulting in an atomic density of 50.4 oxygen atoms per nm3. VO has one oxygen atom per formula unit with four formula units per rock salt fcc unit cell, which has a volume of 0.0684 nm3, resulting in an atomic density of 58.5 oxygen atoms per nm3. Thus, one should expect a 16% stronger XPS O 1s signal from pure VO, compared to pure STO. The initial signal peak around 4 ML is consistent with the formation of interfacial V2O3, which has an even higher oxygen atomic density of 61 per nm3 in its bulk corundum phase.
Looking at the Ti 2p signal of STO for thin vanadium layers (<32 ML), we also note that there is a Ti3+ shoulder on the Ti 2p3/2 signal, indicating a reduction of the Ti in STO. STO is well-known to support significant amounts of oxygen vacancies while still retaining its ideal cubic perovskite crystalline structure. Studies on bulk single crystal and ceramic show that up to 1/6 of the oxygen atoms can be removed without change of crystal structure (resulting in composition of SrTiO2.5).20–22 In Fig. 3, we plot the Ti 2p spectrum of the STO substrate for different vanadium metal overlayer thicknesses. The Ti 2p3/2 peak at 459 eV shows a shoulder at lower binding energy (∼457.7 eV) that is associated with the Ti3+ oxidation state. The proportion of the Ti3+ shoulder to the overall Ti 2p3/2 signal grows from 8.7% at 1 ML of vanadium to 12.6% at 4 ML, and to 22.4% at 16 ML of vanadium. Beyond this thickness, the Ti 2p spectrum is buried. STO in which 22% of Ti is in the 3+ oxidation state implies a composition of SrTiO2.89, which is well within the capacity of single crystal STO to sustain.
Figure 4 shows the evolution of the RHEED pattern at selected various vanadium thicknesses. Figure 4(a) shows the clean STO(111) substrate along the [1−10] zone axis, showing the sharp dots along an arc characteristic of atomically flat surfaces. Figures 4(b)–4(d) show the RHEED pattern of the surface after deposition of 4, 16, and 56 ML of vanadium metal, respectively, at 800 °C. One can see that the pattern remains streaky throughout the deposition indicating a well-ordered surface.
B. Deposition temperature dependence
Figure 5(a) shows the combined O 1s-V 2p XPS spectra for either 20 or 30 nm vanadium deposited on STO(111) at different temperatures. Either of these thicknesses is sufficiently large to completely bury any photoelectron signal from the STO substrate. The spectra for both 200 and 400 °C samples are virtually identical with the spectrum from the higher temperature slightly stronger but with the positions and peak ratios about the same. For the sample deposited at 600 °C, the V 2p3/2 binding energy shifts from 512.35 to 512.75 eV.23 The oxygen has also increased relative to vanadium. Finally, for 800 °C, the V 2p3/2 binding energy shifts even higher to 513.05 eV, and the oxygen signal increases significantly as well. Figure 5(b) shows the binding energies of the V 2p3/2 peak plotted as a function of deposition temperature.
The composition of the VOx films can be determined from the relative areas of the O 1s component and the V 2p components of the XPS spectrum. For calculating the O:V ratio, we use relative sensitivity factors of 0.78 for O 1s and 2.5 for V 2p. The integrated area of each region is then divided by the sensitivity factors to determine the relative composition. These values were determined by modifying the Wagner relative sensitivity factors24 to produce the correct stoichiometry for a V2O5 film grown separately. We plot the calculated stoichiometry from XPS as a function of temperature in Fig. 6. The composition is overlaid against the low oxygen phase diagram published by Davydov and Rempel.25 At 200 and 400 °C, we see that the so-called β phase (V16O3) is formed, which grows as 3D islands as observed by RHEED (Fig. 7). At 600 °C, RHEED appears somewhat blurry, consistent with the obtained composition being located in a two-phase region with incommensurate structures. At 800 °C, the film is highly epitaxial and smooth as observed by RHEED (Fig. 7) and is consistent with the rock salt vanadium monoxide phase. We can also deduce from Fig. 6 that, because the monoxide is stable for O:V ratios as low as 0.8 , the monoxide should start to form at temperatures of about 700 °C.
C. Structural characterization of VO films
From the thickness dependence studies described previously, we know that the monoxide phase is formed when vanadium metal is deposited on STO at a temperature of 800 °C. We deposit 20 nm of V on both STO (111) and STO (100) at 800 °C and characterize the resulting films by ex situ x-ray diffraction and cross-sectional STEM. In order to protect the VOx film from oxidation in the ambient, a thin (10 Å) Si layer26 was deposited at room temperature, which is expected to transform into SiO2 when exposed to ambient conditions. Figure 8 shows cross-sectional ADF-STEM images of a sample grown on STO (111) taken along two perpendicular directions of the STO substrate. The figure also shows the corresponding atomic model of STO and VO besides the STEM image for each projection. It can clearly be seen that the VO layer is epitaxial with respect to the substrate, with an atomic structure consistent with the atomic models. It should also be noted in Fig. 8(b) that there is a visible 3 ML interfacial region in the [1−10] zone axis where a slightly different atomic arrangement is present, which coincides with the V2O3 phase observed in XPS up to 4 ML. The presence of the interfacial layer is not obvious in the [11−2] zone axis. It should be noted that ultrathin V2O3 layers are reported to have interesting topological properties.27 Figure 9 shows an energy dispersive x-ray (EDX) analysis of the VO film showing that the oxygen to vanadium ratio is constant throughout the film thickness, indicating uniform film composition.
X-ray diffraction (both in-plane and out-of-plane) was also performed on the films, as shown in Fig. 10. Out-of-plane diffraction for samples grown on STO (111) and STO (100) are shown in Figs. 10(a) and 10(b), respectively. Only the substrate peak and the corresponding orientation of VO can be observed in the scans—VO(111) grows on STO(111) and VO(100) grows on STO(100). The calculated lattice constant of the VO film is 4.08 Å for STO(111) and 4.09 Å for STO(100). In-plane diffraction (coupled 2θχ − ϕ scans) taken at a fixed incidence angle of 0.4° along the STO azimuth for films grown on STO(111) and STO(100) are shown in Figs. 10(c) and 10(d), respectively. As with the out-of-plane scans, only a single in-plane film orientation is observed that matches with the substrate orientation, confirming epitaxy as observed in RHEED. The extra peaks at ∼1.5°–2.5° lower angle are artifacts due to W Lα x rays coming from evaporated filament material on the Cu target. These peaks are indicated in Fig. 10 as well. We have also confirmed with RHEED that VO films remain epitaxial for thicknesses exceeding 80 nm (not shown). For such a VO film, which takes ∼3 h to grow and causes oxygen depletion in STO resulting in a stoichiometry of SrTiO2.89, about 1.4 μm of reduced STO is required to be formed, which is well within the diffusion length of STO at 800 °C for this amount of time, which can be on the order of 10 μm.28,29 This shows that oxygen scavenging from STO is a viable way to grow stoichiometric vanadium monoxide without the need to precisely control oxygen flow.
We have demonstrated the growth of rock salt VOx films on STO without the need to precisely control the amount of added oxygen by using the phenomenon of oxygen scavenging of STO. The resulting film composition can be adjusted by simply changing the temperature at which vanadium metal is deposited. The films are highly epitaxial, following the substrate orientation closely, and the film composition is uniform throughout the film thickness, with a thin (3 ML) interfacial layer with V2O3 composition. The relatively easy way of obtaining vanadium monoxide may allow other researchers to study this highly correlated material, which has not been well-studied due to the difficulty in synthesizing the material as a single-phase layer.
This work was supported by the Air Force Office of Scientific Research under Grant No. FA9550-18-1-0053. TEM analysis was supported, in part, by the Louis Beecherl, Jr., Endowment Funds.
The data that support the findings of this study are available from the corresponding author upon reasonable request.