Structure and Transport in High Pressure Oxygen Sputter-deposited Basno3−δ Low Pressure Oxygen Annealing at High Temperature for High Quality Sputtered Yba 2 Cu 3 O 7−δ Films

Articles you may be interested in Dopant-site-dependent scattering by dislocations in epitaxial films of perovskite semiconductor BaSnO3 Electron transport and defect structure in highly conducting reactively sputtered ultrathin tin oxide films Appl. Oxygen pressure-tuned epitaxy and optoelectronic properties of laser-deposited ZnO films on sapphire Appl.

Semiconducting oxides with predominantly s-orbital-derived conduction bands are attractive due to their potential for low electron effective mass and high room temperature mobility.ZnO is an illustrative example, with electron mobility up to 430 cm 2 V −1 s −1 at ambient, 1,2 increasing to 700 000 cm 2 V −1 s −1 at cryogenic temperatures, 3 enabling observation of the integer, 2 and fractional 4 quantum Hall effects.Particularly, in wide band gap cases like ZnO, such semiconductors are also of interest as transparent conductive oxides (TCOs), as part of the drive to displace indium tin oxide. 5While much exploration of binary and ternary wide gap semiconducting oxides has been thus stimulated, 5 relatively little, in this regard, has been done with perovskite oxides.This is despite the remarkable chemical flexibility and functionality of perovskites, which could potentially include high mobility wide gap semiconductors.Realizing high room temperature mobility in perovskites, potentially via s-bands rather than d-derived bands, is important for several reasons.Primary among these is the need, for the advancement of oxide electronics, [6][7][8] to translate the extraordinary diversity in functionality that has been demonstrated in perovskite oxide heterostructures [6][7][8] to room temperature.A high room temperature mobility perovskite semiconductor could enable, for instance, ambient temperature high mobility oxide two-dimensional electron systems, oxide transistors, and all-perovskite spintronic devices.
Here, we report on a study of the growth, structure, and stoichiometry of epitaxial BSO deposited via high pressure oxygen sputter deposition.This is a proven method for oxides, where a high pressure (O(1 Torr)) O 2 plasma provides aggressive oxygenation, thermalized ions, and minimized resputtering.It is shown here to produce single-phase, relaxed, close to stoichiometric BSO, as characterized by high resolution wide-angle X-ray diffraction (WAXRD), grazing-incidence X-ray reflectivity (GIXR), atomic force microscopy (AFM), and scanning transmission electron microscopy (STEM).This is demonstrated on both STO(001) and MgO(001), the latter providing smaller lattice mismatch (2.3% tensile vs. 5.1% compressive) but with symmetry mismatch.Significantly, vacuum annealing of 300-Å-thick films on MgO(001) at 850 • C is shown to readily reduce BSO, the resulting V O density generating a Hall electron density of 5 × 10 19 cm −3 , 6 mΩ cm resistivity, and 20 cm 2 V −1 s −1 mobility.The transport is degenerate (in fact metallic) with positive temperature coefficient of resistivity down to 100 K.While this mobility is lower than prior thin film reports with other dopants, there remain numerous promising routes to further mobility optimization.
RF sputter deposition was done from 2" BSO targets synthesized by solid-state reaction of BaCO 3 and SnO 2 (99.9% purity) in air at 1200 • C, followed by cold-pressing then sintering at 1400 • C. Prior to growth, the substrates were annealed for 15 min in 1.9 Torr of O 2 at 900 • C. Films were subsequently deposited at 76 W with 550-850 • C substrate temperatures (T dep ) and 1.5-2.2Torr O 2 pressures (P O2 ), the rate (0.4-1.5 Å/min) being controlled by the source-tosubstrate distance.Post-growth cooling was done in 600 Torr of O 2 .WAXRD (including rocking curves (RCs) and reciprocal space maps (RSMs)) and GIXR were performed with Cu K α radiation in a Panalytical X'Pert system.GenX 29 was used for GIXR refinement.Contact mode AFM was carried out in a Bruker Nanoscope V Multimode 8, while cross-sectional high-angle annular dark-field (HAADF) STEM was performed on an aberration-corrected monochromated FEI Titan G2 60-300 at 300 kV.The semi-convergent angle of the incident beam was 24.5 mrad, HAADF images were obtained with a detector angle of 50-200 mrad, and samples were tilted to a [100] zone axis.Vacuum annealing was done at 600-900 • C for 4 h at pressures <10 −7 Torr.Subsequent electronic transport measurements (AC and DC, 2-300 K, 0-9 T) utilized In contacts in a van der Pauw geometry.
To illustrate the basic structural features of high pressure oxygen sputter-deposited BSO, WAXRD for 230-Å-thick films on STO(001) and MgO(001) is shown in Figs.1(a) and 1(b).These films were deposited at 750 • C, 10,14,26 in 1.5 Torr of O 2 (typical for this method).Figs.1(a) and 1(b) reveal only the BSO, STO, and MgO 00l reflections, establishing phase purity (within detection limits), and out-of-plane epitaxy.Grazing incidence diffraction (supplementary material 30 Fig. S1) confirms in-plane cube-on-cube epitaxy.As can be seen from the 002 close-ups in the insets to Figs. 1(a) and 1(b), finite-size fringes are observed on STO(001), but not on MgO(001), indicating that films on STO are smoother on short lateral length scales.As shown in Fig. 1(c), however, BSO films on both substrates are smooth on the long lateral scales probed in GIXR; the fits refine the film RMS (root-mean-square) surface roughnesses to 6.8 Å on STO and 5.4 Å on MgO, i.e., the 1-2 unit cell level.
Data of the type shown in Figs.1(a) and 1(b) also provide the out-of-plane lattice parameter (a op ), a probe of stoichiometry and strain state.In Fig. 2, this is plotted vs. P O2 , deposition rate, and T dep for films of approximate thickness (t) 100-150 Å on both STO(001) and MgO(001).The values of the deposition parameters that are held constant are shown at the top of each panel.There are two notable features in these data.First, a op is similar on STO and MgO, in almost all cases lying between 4.125 and 4.140 Å, despite the very different lattice mismatch.These values are close to, but greater than (by 0.01-0.02Å), the bulk lattice parameter 9,11,16 (horizontal line).Second, there is a remarkable insensitivity to growth conditions over the probed parameter space.One simple interpretation is that, in this thickness range, the films on both substrates are already strain relaxed.The slightly elevated a op could then indicate some level of cation non-stoichiometry (potentially from the ceramic target) or non-negligible defect density.
Thickness-dependent strain relaxation was thus probed (Fig. 3) using BSO deposited on STO(001) at 750 • C in 1.5 Torr of O 2 .In this figure, panels (a) through (c) plot a op , the Scherrer length Λ (i.e., the 002 WAXRD peak width converted to a length via the Scherrer equation) and the 002 film RC full-width at half-maximum (FWHM).Two values of the latter are plotted, as t-dependent 002 RCs (supplementary material 30 Fig. S2) reveal a sharp peak superimposed on a broader one.To facilitate the discussion of strain state, Fig. 3(a) also shows the bulk lattice parameter, as well as the critical thickness for strain relaxation (t crit ) from the Matthews-Blakeslee formula, using the literature parameters. 18Although this t crit is rather approximate for perovskites, it is nevertheless instructive to note that it is as low as 25 Å, due to the large mismatch.The data in Fig. 3(a) are indeed consistent with low t crit ; a op decreases with increasing t from the lowest thicknesses probed, levelling off at a op ≈ 4.13 Å for t ≥ 150 Å.It is noted (i) that this a op is consistent with Fig. 2 and (ii) that no reliable lattice parameters could be extracted below 80 Å.That films on STO above t ≈ 150 Å are indeed relaxed is confirmed by asymmetric (013) RSMs (Fig. 3(d (t) = 450 Å), where the expected reflections from relaxed and fully strained (i.e., pseudomorphic) films are marked.The BSO peak is close to the fully relaxed position, but with minor lattice expansion.This is ascribed to minor cation non-stochiometry or defect formation, as returned to below.The t dependence of a op was also studied on MgO (supplementary material 30 Fig. S3), full strain relaxation occurring by ∼150 Å.
The t-dependence of the Scherrer length, Λ (Fig. 3(b)) further supports very low t crit on STO(001).Specifically, Λ is found to almost exactly coincide with the thickness (the solid line in Fig. 3(b) is Λ = t), meaning that the WAXRD widths are finite size dominated, with negligible broadening due to microstrain.This is consistent with the majority of the depth of such films being relaxed, i.e., very low t crit .The RC FWHMs in Fig. 3(c) are also consistent with this, as they reveal little t dependence, the narrow contribution being barely broadened over the STO(001) substrate.Further work will be required to pin down the origin of the broader RC contribution, which is clearly evident (Fig. S2 30 ) above t ≈ 100 Å.This could be related to the expanded a op , potentially reflecting a t-dependent density of defects such as misfit and threading dislocations.
To further probe defects, cross-sectional STEM was performed.Fig. 3(f) shows a representative HAADF STEM image of a 35-Å-thick BSO film, again on STO(001).A smooth, continuous film is evident even at this thickness (∼8 unit cells), the major feature of interest being the quite evenly spaced dislocations, marked by arrows.A magnified view of one such dislocation is shown, from which it can be seen that these are misfit edge dislocations with Burger's vector along [100].The spacing of these dislocations, averaged over multiple images, is 70-80 Å, close to the spacing expected (75 Å) if such misfit dislocations relax the full 5.1% lattice mismatch.This is again consistent with the diffraction observations of very low t crit .STEM imaging of MgO(001)/BSO (supplementary material 30 Fig. S4(a)) reveals a similar picture but with larger dislocation spacing due to the smaller mismatch.As a final comment, it is noted that the low surface roughness from Having established the microstructure of these BSO layers, vacuum annealing was pursued to attempt reduction and n-doping.Note that as-deposited films displayed no measurable conduction at any P O2 or cooling atmosphere.Vacuum annealing was done for 4 hrs at temperatures from 600 to 900 • C, using MgO(001)/BSO(t = 300 Å) films; no such experiments were performed on STO, due to anticipated complications with V O formation in the substrates.Consistent with earlier work, 9 no measurable conduction was induced below 600 • C. Above this, finite resistivity was detected, decreasing rapidly with annealing temperature.We focus here, as an illustrative example, on 850 • C reduction.As shown in Fig. 4(a), the WAXRD data are largely unchanged after this vacuum annealing, confirming, within the limits of WAXRD, that no secondary phases form.This is strengthened by analytical STEM using energy dispersive X-ray analysis (supplementary material 30 4(d), the deduced mobility (µ) is around 20.5 cm 2 V −1 s −1 at 300 K, increasing to 23.5 cm 2 V −1 s −1 at 50-100 K, before decreasing to 22.5 cm 2 V −1 s −1 at low T. Indications of phonon scattering contributions to the mobility are thus found, although scattering from defects such as dislocations is likely also important.In addition to establishing that V O formation is facile in BSO and that it enables n-doping without impurities, these results point to numerous viable approaches to improve mobility.Substrate choice, thickness, and reduction temperature are obvious optimization variables, with the potential to elucidate the range of n obtainable by reduction, the resulting µ variation, the quantitative influence of misfit and threading dislocations, and, ultimately, the relative impact of the various defects on mobility.
In summary, we have demonstrated the growth of epitaxial, phase pure, relaxed, close to stoichiometric BSO films on STO(001) and MgO(001) using high pressure oxygen sputter deposition.A detailed picture of the microstructure is provided, particularly with respect to strain relaxation via misfit dislocations.Importantly, vacuum annealing at 850 • C is found to induce oxygen vacancies at significant densities, without secondary phase formation, enabling facile n-doping.Mobilities of 20 cm 2 V −1 s −1 at 5 × 10 19 cm −3 are obtained in films only 300 Å thick, with much scope for improvement.Ultimately, this will enable a comparison of electronic properties obtainable by intrinsic vs. extrinsic doping in BSO, an important step for room temperature applications.
FIG. 2. Out-of-plane lattice parameter, a op , vs. (a) O 2 growth pressure (P O2 ), (b) deposition rate, and (c) deposition temperature (T dep ), for ∼100-150-Å-thick BSO films on STO(001) (red) and MgO(001) (blue).In each panel, the growth parameter(s) held fixed is labeled at the top.Error bars are comparable to the symbol size.The bulk BSO lattice parameter is shown by the dashed-dotted line.Dashed lines through the data are guides to the eye.

FIG. 4 .
FIG. 4. (a) High resolution WAXRD from a 300-Å-thick MgO(001)/BSO film before and after vacuum annealing at 850 • C. Temperature dependence of (b) the (log scale) film resistivity ρ (c) the Hall electron density n, and (d) the Hall electron mobility µ.Dashed lines are guides to the eye.The insets to (b) show the conductivity, σ, vs. temperature below 50 K, and the zero-field-offset-subtracted 300 K Hall resistivity vs. magnetic flux density (B), respectively.The growth conditions (750 • C, 1.9 Torr) are labeled at the top.Note that this film was sputtered at 66 W.
Figs. S4(b)-S4(e)), which also detects no chemical inhomogeneity after annealing.The lattice parameter does change though, increasing by 0.009 Å, consistent with V O formation.Electronic transport reveals large changes in resistivity (ρ) and electron density (n) after annealing, implying reduction to BaSnO 3−δ and facile n-doping.As shown in Fig. 4(b), room temperature ρ values are ∼6 mΩ cm after 850 • C annealing, in comparison to unmeasurably high values (>15 kΩ cm) as-deposited.ρ decreases on cooling to ∼100 K, below which it undergoes a weak upturn, to 6.5 mΩ cm at 2 K.The weak temperature (T) dependence suggests degenerate doping and in fact, the T → 0 extrapolation of the conductivity, σ, is finite (inset to Fig. 4(b)), implying metallic behavior.Hall effect measurements further elucidate this.A clear, field-linear Hall signal due to electrons is observed (inset to Fig. 4(c)), n varying only weakly with T (Fig. 4(c)), consistent with degenerate doping.Additionally, although uncertainty in the effective mass and dielectric constant results in uncertainty in the Mott prediction for the critical electron density in BSO, our upper bound is ∼10 18 cm −3 .The observed n (5 × 10 19 cm −3 , Fig. 4(c)) exceeds this, consistent with the metallicity deduced from the inset to Fig. 4(b).Both this observation and the field-linearity of the Hall voltage are consistent with relatively modest compensation.As shown in Fig.
S. Bose, M. Manno, T. Wang, A. Prakash, X. Zhang, and D. Phelan are thanked for assistance.Work supported primarily by the NSF through the UMN MRSEC under Nos.DMR-0819885 and DMR-1410888.J.S.J. and K.A.M. acknowledge C-SPIN, a STARnet/SRC program sponsored by MARCO and DARPA.Part of this work was carried out in the UMN Characterization Facility, which receives equipment funding through the UMN NSF MRSEC.