One perovskite oxide, ASnO3 (A = Sr, Ba), is a candidate for use as a transparent conductive oxide with high electron mobility in single crystalline form. However, the electron mobility of films grown on SrTiO3 substrates does not reach the bulk value, probably because of dislocation scattering that originates from the large lattice mismatch. This study investigates the effect of insertion of bilayer BaSnO3 / (Sr,Ba)SnO3 for buffering this large lattice mismatch between La:BaSnO3 and SrTiO3 substrate. The insertion of 200-nm-thick BaSnO3 on (Sr,Ba)SnO3 bilayer buffer structures reduces the number of dislocations and improves surface smoothness of the films after annealing as proved respectively by scanning transmission electron microscopy and atomic force microscopy. A systematic investigation of BaSnO3 buffer layer thickness dependence on Hall mobility of the electron transport in La:BaSnO3 shows that the highest obtained value of mobility is 78 cm2V−1s−1 because of its fewer dislocations. High electron mobility films based on perovskite BaSnO3 can provide a good platform for transparent-conducting-oxide electronic devices and for creation of fascinating perovskite heterostructures.
I. INTRODUCTION
Exploration of transparent conductive films (TCFs) possessing high electrical conductivity and a wide band gap higher than 3 eV, has been studied intensively for numerous fundamental applications such as electrodes for optical devices and interconnections of flat panel displays, solar cells, and light-emitting diodes.1,2 The criteria for TCFs generally include resistivity of less than 10−3Ωcm and transparency higher than 80% in the visible light region. Many attempts have been undertaken to find highly conductive materials including oxides, nitrides, and borides. A degree of success has been achieved with transparent conductive oxide (TCO) films.2 In fact, TCOs with high charge carrier density have been demonstrated for Sn:In2O3 (ITO),3 ZnO,4 SnO2,5 TiO2,6 and their compounds.7 Among them, the ITO is an exemplary TCO with long stability: it has already been applied for practical use with 10−4 Ω cm.
In addition to these binary oxides, barium stannate perovskite (BaSnO3) is a good candidate for use as a TCO material. (i) Based on first-principle calculation, the electron conduction band is mainly constituted by the Sn 5s orbital.8 Consequently, stannates show good potential for superior conductivity based on high electron mobility because of the small effective electron mass of about 0.2 m0.8 In fact, the actual values of electron mobility for single crystals has been evaluated experimentally as 320 cm2V−1s−1.9–11 Improvement of electron mobility in BaSnO3 films is especially important for technical use with the low carrier concentration regime as σ = μen with σ, μ, e, and n respectively representing conductivity, mobility, the elementary charge and carrier concentration. (ii) Furthermore, BaSnO3 possesses a direct band gap of 3.1 eV.12 Its transparency is experimentally valid for a wide range up to ultraviolet light.9 (iii) Thermal stability is so robust that transport properties do not change much after thermal treatment cycles at 530 °C.10 Furthermore, substitution of the Ba site with Sr enlarges the band gap (4.1 eV for SrSnO312), which is expected to provide a good arena for quantum heterostructures based on band-engineered perovskite stannates.13 In addition, substituting either cation site with heterovalent ions enables charge carrier doping.14 Consequently, the interface of BaSnO3 with Sr-doped BaSnO3 or other transition metal perovskite oxides can be a new platform for the emergence of high electron mobility through the charge transfer into BaSnO3.13
For actual application of perovskite stannates as a transparent semiconductor, many attempts have been undertaken to increase the mobility of La-doped BaSnO3 films by tuning the stoichiometry of Ba/Sn and oxygen and by minimizing the unintentional impurity concentration in films using oxide molecular beam epitaxy.15 Optimization of growth conditions for the films on SrTiO3 substrates16 and selecting proper substrates such as PrScO3, SmScO3, BaSnO3, and TbScO315,17–20 have been investigated to reduce the number of threading dislocations. Especially, the dangling bond in a threading dislocation acts as a Coulomb scattering center.21 In fact, it has been experimentally reported that the electron mobility of BaSnO3 films at room temperature is proportionally dependent on the root square of carrier concentration, implying that the main scattering origin for electron transport is such a dislocation.16 Therefore, the reduction of film dislocations should be pursued to improve the electron mobility of the films towards the value of the single crystal. With this study, we have demonstrated improvement of mobility of La-doped BaSnO3 (La:BaSnO3) films by introducing a BaSnO3 / (Sr0.5,Ba0.5)SnO3 buffer layer, although the highest mobility does not reach the bulk value. The effect of buffer layer has been validated properly for ferroelectrics,22 semiconductors,23–25 and cuprate superconducting films.26
II. EXPERIMENTAL PROCEDURE
For film deposition, we used sintered ceramic targets of BaSnO3, Sr0.5Ba0.5SnO3, and LaxBa1−xSnO3 with x = 0.5, 2, and 5 at% as source materials, which were synthesized by grounding and sintering stoichiometric composites of BaCO3, SrCO3, SnO2 and La2O3 powders at 1250 °C for 6 hours in air. Then the pelletized targets were sintered at 1400 °C for 24 hours in air. Film growth was achieved using pulsed layer deposition (PLD) with KrF excimer laser at substrate temperature of 900 °C and oxygen pressure of 100 mTorr. The typical deposition rate is 1–3 nm/min. The investigated films for characterization of electrical properties are constituted by La:BaSnO3 (80 nm)/ BaSnO3 (80 nm) / BaSnO3 (t nm) / Sr0.5Ba0.5SnO3 (10 nm) / SrTiO3 substrate. These structures were grown in the following way: the buffer layer BaSnO3 (t nm) / Sr0.5Ba0.5SnO3 (10 nm) was first deposited on SrTiO3 (001) substrate. Then the film was annealed in air at 1200 °C for 1 h. Finally, the film was re-installed to the PLD chamber to deposit La:BaSnO3 (50 or 80 nm) / BaSnO3 (50 or 80 nm) layer. Thickness t of BaSnO3 in the buffer layer was varied from 25 nm to 410 nm to investigate the buffer thickness dependence of lattice constants and transport properties. Each layer thickness was calibrated as follows. The total film thickness and deposition rate per laser pulse were determined by the XRD Laue fringe (see Fig. 2(a)) and the total number of pulses used for deposition. Then, each layer’s thickness was calculated using the number of pulses with the estimated deposition rate. We assumed an equal deposition rate per pulse for all stannate targets. To tune the carrier concentration in La:BaSnO3, the La concentration was controlled by the laser pulse ratio ablating on non-doped BaSnO3 and La:BaSnO3 targets. The respective bulk values of the lattice constants for BaSnO3 and Sr0.5Ba0.5SnO3 are 4.117 Å27 and 4.076 Å28 in the cubic lattice. Those respective lattice mismatches to the SrTiO3 substrate (a = 3.905 Å) are 5.4 and 4.4%. To evaluate the film surface morphology and threading dislocations, we used atomic force microscopy (AFM) and cross-sectional scanning transmission electron microscopy (STEM). The electrical transport properties were investigated using the temperature dependence of longitudinal and Hall resistances in a standard four-terminal method. For control experiments, we prepared non-doped (280 nm for STEM) and 2% La-doped BaSnO3 (70 nm for electrical measurement) films deposited directly on SrTiO3 (001) substrates. All layers except for the La-doped layer retain their insulating properties, supplying no parasitic electrical conduction.
(a) X-ray diffraction patterns of BaSnO3 (t nm) / (Sr,Ba)SnO3 (10 nm) / SrTiO3 buffer structures with t varying from 25 nm (top panel) to 410 nm (bottom panel). (b) Out-of-plane c (closed red circles) and in-plane a (closed blue squares) lattice constants of the whole BaSnO3 layer after deposition of La:BaSnO3 conducting layer as a function of BaSnO3 buffer layer thickness t. (c) Ratio of c to a as a function of t. (d) Buffer BaSnO3 thickness dependence of the root mean square roughness (RMS) extracted from 25 × 25 μm2 atomic force micrographs.
(a) X-ray diffraction patterns of BaSnO3 (t nm) / (Sr,Ba)SnO3 (10 nm) / SrTiO3 buffer structures with t varying from 25 nm (top panel) to 410 nm (bottom panel). (b) Out-of-plane c (closed red circles) and in-plane a (closed blue squares) lattice constants of the whole BaSnO3 layer after deposition of La:BaSnO3 conducting layer as a function of BaSnO3 buffer layer thickness t. (c) Ratio of c to a as a function of t. (d) Buffer BaSnO3 thickness dependence of the root mean square roughness (RMS) extracted from 25 × 25 μm2 atomic force micrographs.
III. RESULTS AND DISCUSSIONS
Thin film growth and structural characterization
Figures 1(a) and 1(b) respectively depict AFM images of the annealed BaSnO3 (t = 130 nm) / Sr0.5Ba0.5SnO3 on SrTiO3 and 280-nm-thick BaSnO3 grown directly on SrTiO3 substrate. As might be readily apparent, the surface morphology of BaSnO3 / Sr0.5Ba0.5SnO3 has a clear step and terrace structure, indicating that its surface is atomically flat. On the other hand, the directly-grown BaSnO3 film has large precipitates. Its surface roughness is on the order of tens of nanometers. We characterized the precipitates as the consequence of three-dimensional growth of BaSnO3 attributable to the large lattice mismatch between the film and substrate. Figures 1(c) and 1(d) respectively present cross-sectional bright-field STEM images of the films with and without a Sr0.5Ba0.5SnO3 buffer layer. The layer stacking geometry in Fig. 1(c) was determined from the deposition rate and the number of pulses used for the deposition of each layer. The vertical white and black contrast shown in both images resembles that observed in previous studies.16,29 The contrast was identified as threading dislocations based on etch pit method.16,29 It should be noted that diffraction contrasts other than dislocations, such as thickness fringes, bend contours, and dynamical contrast in defect images, are largely reduced in STEM, while these contrasts usually appear significantly in TEM.30 The dislocations in the film reaching the layer surface from substrate in Fig. 1(d) were counted as 11 (indicated by black arrows) within 760 nm in the image. The resultant areal density of the dislocation can therefore be estimated roughly as (11/760 nm)2 = 2.1 × 1010 cm−2 under the assumption that the dislocation densities are identical between along the width and thickness of the TEM samples, which is consistent with previous work examining BaSnO3 / SrTiO3.16 The observed spot contrasts in Fig. 1(d) show local distribution of grains having slightly different crystallographic direction, which probably originates from the large mismatch between the film and substrate. However, the chemically sensitive high-angle annular dark-field (HAADF) STEM image (see right inset in Fig. 1(d)) shows no such spot contrast, implying that the spot area has the same composition as that of the surrounding region. In addition to threading dislocations, such misoriented grains might degrade transport properties. Therefore the insertion of Sr0.5Ba0.5SnO3 also reduces defects originated from the large lattice mismatch between the film for interest and SrTiO3 substrate. As Fig. 1(c) shows, the number of dislocations reaching the film surface from the substrate is reduced by inserting the Sr0.5Ba0.5SnO3 layer, although the thicknesses of whole layers in both films are comparable. The density of dislocations in Fig. 1(c) is estimated as (5/760 nm)2 ∼ 4.3 × 109 cm−2. The areal density of dislocation is reduced by a factor of five by the introduction of 10 nm Sr0.5Ba0.5SnO3 layer. It should be mentioned that the threading dislocations in principle do not terminate within the crystal; hence it is presumed that the dislocations are redirected horizontally and glide away to become in-plane dislocations when the step edges overgrow the surface outcrop of the dislocations during step-flow growth of 80-nm BaSnO3 on annealed buffer layer.31 Then, the horizontally-trailing dislocations are invisible in the cross-sectional STEM images when the in-plane dislocations satisfy invisible criterion (g ⋅ b = 0, where g and b denote a reflection concerned and Burgers vector, respectively).32 For the film without buffer layer some of the vertical contrasts also terminate in Fig. 1(c) because of the same reason. However, the number of the threading dislocations remaining in surface region of the film without buffer layer is apparently much larger than that of the film with buffer. As a result, the number of threading dislocations are suppressed in top La:BaSnO3 layer. To interpret the imperfect reduction of misfit dislocation in La:BaSnO3, we take an example of the growth of (Sr,Ba)TiO3 (180 nm) / BaTiO3 (12 nm) on SrTiO3 substrate, where the thinner BaTiO3 buffer layer was used to reduce misfit dislocations.22 The lattice constant of BaTiO3 is largest among the three materials. By post-growth annealing, the buffering effect within the bottom partially strained BaTiO3 layer plays a crucially important role in relaxing the lattice constant of the (Sr,Ba)TiO3 layer. As a result, misfit dislocations in the (Sr,Ba)TiO3 layer are dissolved completely. In our study, the thin Sr0.5Ba0.5SnO3 layer is inserted between BaSnO3 and SrTiO3 layers to sever the misfit dislocations. However, the in-plane lattice constant of Sr0.5Ba0.5SnO3 is 1.0% smaller than that of BaSnO3, which is an opposite situation to that of the (Sr,Ba)TiO3 / BaTiO3 case. Moreover, the lattice mismatch of 5.4% between Sr0.5Ba0.5SnO3 and SrTiO3 is too large to keep coherency, resulting in high density of dislocations before post-growth annealing, even at 10 nm thickness. Consequently, threading dislocations still survive in BaSnO3 because of residual strain at the interface between BaSnO3 and Sr0.5Ba0.5SnO3. Figure 1(e) shows x-ray diffraction (XRD) patterns of the buffered La:BaSnO3 structure with t = 130 nm presented in Fig. 1(c). The appearance of only (00l) diffraction peaks in Fig. 1(e) confirms no secondary phase. The inset in Fig. 1(e) shows a rocking curve for (002) BaSnO3 diffraction peak. The full-width at half-maximum (FWHM) in the rocking curve is as narrow as 0.008, representing the enlarged grain size and reduction of dislocations.33 Figure 1(f) shows reciprocal space mapping around (103) peaks of BaSnO3 and SrTiO3. The Q[001] and Q[100] axes respectively represent reciprocal vectors for out-of-plane and in-plane axes. The extracted lattice constants of BaSnO3 for in-plane (denoted as a) and out-of-plane (denoted as c) are 4.106 Å and 4.122 Å, respectively. Comparison with bulk values a = c = 4.117 Å (marked as a black cross) shows that the BaSnO3 lattice is compressed slightly in the plane. In addition, a tail of BaSnO3 peak along the +Q[100] direction contains the diffraction intensity from the (Sr,Ba)SnO3 film. If so, the diffraction peak from (Sr,Ba)SnO3 film shifts to smaller Q[100] from the bulk value (marked as a white cross), indicating a larger in-plane lattice constant of (Sr,Ba)SnO3. Although this shift can be interpreted by the concept in the example of (Sr,Ba)TiO3 / BaTiO3, dislocations remain in BaSnO3.
(a) and (b) Atomic force micrographs of BaSnO3 (130 nm) / (Sr,Ba)SnO3 (10 nm) annealed buffer layer and 280-nm-thick BaSnO3 grown directly on SrTiO3, respectively. (c,d) Cross-sectional bright-field (BF) images of scanning transmission electron microscopy of (c) 0.5 at% La-doped BaSnO3 (80 nm) / BaSnO3 (80 nm) on the ex-situ annealed BaSnO3 (130 nm) / (Sr,Ba)SnO3 (10 nm) buffer layer and (d) BaSnO3 (280 nm) layer deposited directly on SrTiO3 substrate presented in Fig. 1(b). (Left inset in (d)) A magnified BF image around spot contrast shown by a white square in the main panel. (Right inset in (d)) An HAADF image of the same region. (e) X-ray diffraction pattern in 2θ − ω scan of La:BaSnO3 (80 nm) / BaSnO3 (80 nm) / BaSnO3 (130 nm) / (Sr,Ba)SnO3 (10 nm) / SrTiO3 buffered structure, identical to that presented in (c). Inset shows a rocking curve around the BaSnO3 (002) diffraction peak. (f) Corresponding reciprocal space mapping around (103). Lattice constants of BaSnO3 are 4.122 Å in the growth direction and 4.106 Å in the plane. Black and white crosses respectively represent the expected peak positions for bulk values of BaSnO3 and (Sr,Ba)SnO3.
(a) and (b) Atomic force micrographs of BaSnO3 (130 nm) / (Sr,Ba)SnO3 (10 nm) annealed buffer layer and 280-nm-thick BaSnO3 grown directly on SrTiO3, respectively. (c,d) Cross-sectional bright-field (BF) images of scanning transmission electron microscopy of (c) 0.5 at% La-doped BaSnO3 (80 nm) / BaSnO3 (80 nm) on the ex-situ annealed BaSnO3 (130 nm) / (Sr,Ba)SnO3 (10 nm) buffer layer and (d) BaSnO3 (280 nm) layer deposited directly on SrTiO3 substrate presented in Fig. 1(b). (Left inset in (d)) A magnified BF image around spot contrast shown by a white square in the main panel. (Right inset in (d)) An HAADF image of the same region. (e) X-ray diffraction pattern in 2θ − ω scan of La:BaSnO3 (80 nm) / BaSnO3 (80 nm) / BaSnO3 (130 nm) / (Sr,Ba)SnO3 (10 nm) / SrTiO3 buffered structure, identical to that presented in (c). Inset shows a rocking curve around the BaSnO3 (002) diffraction peak. (f) Corresponding reciprocal space mapping around (103). Lattice constants of BaSnO3 are 4.122 Å in the growth direction and 4.106 Å in the plane. Black and white crosses respectively represent the expected peak positions for bulk values of BaSnO3 and (Sr,Ba)SnO3.
Details of the thickness dependence of lattice constant of BaSnO3 are presented in Fig. 2. Figure 2(a) depicts XRD patterns around the BaSnO3 (002) peak with different thickness from t = 25 nm to 410 nm. Clear Laue fringes, which were used for estimation of the film thickness, were observed for all films. With increasing buffer layer thickness, the (002) diffraction peak approaches the bulk value (vertical dashed line). Figure 2(b) presents a summary the lattice constants of a and c axes of the films as a function of thickness, as evaluated using the asymmetric diffraction peak of the (103) plane. The values of a and c clearly approach the bulk value (4.117 Å) with increasing t, indicating that the compressed strain is relaxed. This fact rules out the effect of diffusion of Sr into the BaSnO3 buffer layer during annealing. Actually, 410 nm thickness is necessary to relax the 1% mismatch between Sr0.5Ba0.5SnO3 and BaSnO3. Furthermore, the residual strain effect is expected to remain with a similar density of dislocations, as presented in Fig. 1(c) in our samples. Correspondingly, the ratio of c/a asymptotically approaches unity in Fig. 2(c). However, the surface morphology of the fully relaxed 410 nm buffer layer sample becomes rough, as presented in Fig. 2(d), which plots the root mean square (RMS) of film surface roughness as a function of BaSnO3 thickness in the buffer layer. Therefore, in terms of surface morphology, the buffer layer thickness of BaSnO3 (t nm) / (Sr,Ba)SnO3 (10 nm) is preferable to be less than 300 nm.
Electrical transport properties
A stark contrast is apparent between the electrical transport properties of materials with and without a buffer layer. The temperature dependences of Hall mobility of 2 at% La-doped BaSnO3 film on the buffer layer and on SrTiO3 substrate are presented in Fig. 3. The carrier density is slightly different between two films while they were fabricated by the identical growth conditions with the same La-doped target. Although it is difficult to identify the origin for it, globules shown in Fig. 1 and/or point defects may play a role for the difference. The Hall mobility is improved by a factor of 2. As a consequence of STEM and electrical properties, the number of threading dislocations is reduced effectively by the insertion of the buffer layer. As a simple test of transparency, the inset in Fig. 3 shows an optical micrograph of the La:BaSnO3 (90 nm) / BaSnO3 (90 nm) on BaSnO3 (140 nm) / (Sr,Ba)SnO3 (10 nm) buffer layer on SrTiO3 film, which shows superior transparency.
Hall mobility μ of 2 at% La-doped BaSnO3 (50 nm) / BaSnO3 (50 nm) / BaSnO3 (90 nm) / (Ba,Sr)SnO3 (10 nm) / SrTiO3 (closed red circles) and 2 at% La-doped BaSnO3 (70 nm) / SrTiO3 without the buffer layer (closed blue squares) as a function of temperature. The inset shows a picture of La:BaSnO3 (90 nm) / BaSnO3 (90 nm) on ex-situ annealed BaSnO3 (140 nm) / (Sr,Ba)SnO3 (10 nm) / SrTiO3.
Hall mobility μ of 2 at% La-doped BaSnO3 (50 nm) / BaSnO3 (50 nm) / BaSnO3 (90 nm) / (Ba,Sr)SnO3 (10 nm) / SrTiO3 (closed red circles) and 2 at% La-doped BaSnO3 (70 nm) / SrTiO3 without the buffer layer (closed blue squares) as a function of temperature. The inset shows a picture of La:BaSnO3 (90 nm) / BaSnO3 (90 nm) on ex-situ annealed BaSnO3 (140 nm) / (Sr,Ba)SnO3 (10 nm) / SrTiO3.
Figure 4 presents the buffer BaSnO3 thickness t dependence of electrical transport properties of 0.5 at% La-doped BaSnO3 top layers. The longitudinal resistivity ρ, the carrier concentration n3D extracted by Hall effect measurements, and the calculated Hall mobility μ are shown as a function of temperature in Fig. 4(a). All films show metallic behavior. Moreover, the carrier concentrations are independent of temperature. Judging from rough estimation of the Mott criterion with effective electron mass of 0.2 m08 and dielectric constant of 40,34 the metallic transport indicates a degenerated regime for such La concentrations. The activation ratio for films of 0.5 at% La doping slightly exceeds 100%. This behavior is more pronounced in the thinner t sample. Although the reason for this behavior remains unclear, we were able to attribute this behavior to inhomogeneous La distribution within the films and/or to the ceramic targets, or to point defects in the film29 such as oxygen deficiency or unintentional impurity. The Hall mobility is increased at low temperature, which, because of the reduction of phonon scattering, is typical behavior for semiconductor systems.35,36 Indeed, the ionized impurity scattering at the low-temperature region is suppressed by the degenerated situation. The room-temperature mobility for the films is shown as a function of the BaSnO3 buffer layer thickness t and c/a ratio, respectively, in Figs. 4(b) and 4(c). Hall mobility increases with film thickness up to 250 nm as a consequence of the reduced dislocation density and associated enlargement of grains and/or the residual strain effect.37 Further increase in t, however, degrades the mobility, probably because of surface roughness as presented in Fig. 2(d). Therefore, the optimum buffer layer structure in this study is inferred to be BaSnO3 (50 < t < 250 nm) / (Sr,Ba)SnO3 (10 nm).
(a) Longitudinal resistivity, carrier concentration and calculated Hall mobility of 0.5 at% La-doped BaSnO3 (80 nm) / BaSnO3 (80 nm) on BaSnO3 (t nm) / (Sr,Ba)SnO3 (10 nm) / SrTiO3 buffer layers as a function of temperature. The value of t varies: 25 (closed blue triangles), 230 (closed red circles), and 410 nm (closed green squares). (b) Buffer thickness t and (c) the value of c/a dependence of the Hall mobility at room temperature.
(a) Longitudinal resistivity, carrier concentration and calculated Hall mobility of 0.5 at% La-doped BaSnO3 (80 nm) / BaSnO3 (80 nm) on BaSnO3 (t nm) / (Sr,Ba)SnO3 (10 nm) / SrTiO3 buffer layers as a function of temperature. The value of t varies: 25 (closed blue triangles), 230 (closed red circles), and 410 nm (closed green squares). (b) Buffer thickness t and (c) the value of c/a dependence of the Hall mobility at room temperature.
Finally, we compare the Hall mobility of our investigated La:BaSnO3 films on the BaSnO3 (80 nm) / BaSnO3 (t = 60–230 nm) / (Sr,Ba)SnO3 (10 nm) buffer layer, i.e., optimal thickness condition, as presented in Fig. 4(b), (red squares) with that of previously reported BaSnO3 single crystals (blue circles)11 and films directly deposited on SrTiO3 substrates (green triangles)10 shown in Fig. 5. The La doping concentration is controlled by the laser pulse ratio for ablating 0.5 at% La-doped and non-doped BaSnO3 targets. In the low carrier concentration regime, although the values of mobility are a bit scattered, the film mobility on the buffer layer is approximately four times greater than that of films deposited directly on the SrTiO3 substrate. Nevertheless, the mobility decreases concomitantly with decreasing carrier concentration below 2 × 1020 cm−3, which follows the n1/2 (black dotted line in Fig. 5). By comparison with single crystals, the mobility of the buffered films still does not reach their values. These results reflect the fact that residual threading dislocations left in the top La:BaSnO3 as presented in Fig. 1(c) still serve a role in electron scattering.4,35,36 Further reduction of dislocations will be possible using, for instance, lattice matched substrate and/or BaSnO3 / (Sr,Ba)SnO3 superlattice as a buffer layer. However, we leave those as subjects for future investigations. In a high electron concentration (>1020 cm−3) regime, mobilities with and without the buffer layer are comparable because screening suppresses grain boundary scattering. However, the resistivity of those films satisfies the first criterion of 10−3 Ω cm (shown as a black dashed line in Fig. 5) required for actual application as TCFs. In this region, the mobility slightly decreases concomitantly with increasing carrier concentration, implying ionized impurity scattering.4,38,39 To apply La:BaSnO3 as a transparent conductive oxide, further investigation is needed to achieve the difficult criterion for low resistivity below 10−4 Ω cm (black dashed dotted line in Fig. 5) with rich La doping concentration. Very recently, improvement of electron mobility in BaSnO3 films up to 100 and 150 cm2 V−1 s−1 at room temperature has been reported using modified oxide molecular beam epitaxy with SrTiO3 and PrScO3 substrates.15 By introducing a buffer layer, the management of strain and dislocations can be expected to increase mobility further.
Hall mobilities of La:BaSnO3 films on BaSnO3 (80 nm) / BaSnO3 (t = 60–230 nm) / (Sr,Ba)SnO3 (10 nm) / SrTiO3 buffer layer described as a result of this study (closed red squares) incorporated with previously reported values in BaSnO3 single crystals (closed blue circles)11 and BaSnO3 films grown directly on SrTiO3 substrate (closed green triangles)10 as a function of the carrier concentration. The carrier concentration of La:BaSnO3 is tuned by the La concentration. For reference, those of La:BaSnO3 films on SrTiO3 fabricated in this study are presented as open green squares. The dotted lines are a guide showing the root square dependence of mobility on the carrier concentration.
Hall mobilities of La:BaSnO3 films on BaSnO3 (80 nm) / BaSnO3 (t = 60–230 nm) / (Sr,Ba)SnO3 (10 nm) / SrTiO3 buffer layer described as a result of this study (closed red squares) incorporated with previously reported values in BaSnO3 single crystals (closed blue circles)11 and BaSnO3 films grown directly on SrTiO3 substrate (closed green triangles)10 as a function of the carrier concentration. The carrier concentration of La:BaSnO3 is tuned by the La concentration. For reference, those of La:BaSnO3 films on SrTiO3 fabricated in this study are presented as open green squares. The dotted lines are a guide showing the root square dependence of mobility on the carrier concentration.
IV. CONCLUSION
In conclusion, we demonstrate that the insertion of the BaSnO3 / (Sr,Ba)SnO3 buffer layer on SrTiO3 substrate reduces the density of dislocations considerably, thereby improving surface smoothness and electrical transport. The systematic investigation of buffer layer thickness dependence reveals that a certain condition to maximize the electron mobility of about 80 cm2 V−1 s−1 exists for a wide range of electron densities because of the strain effect. This maximization might further increase the conductivity of perovskite stannate films and increase the applicable functionality for transparent electronics. The optimal characterization of the BaSnO3 buffer layer paves the way for perovskite band engineering by mimicking the conventional GaAs/AlGaAs heterostructure such as quantum well and high-electron mobility transistor (HEMT) based on modulation doping technique.
ACKNOWLEDGMENTS
This work was supported by JSPS Grant-in-Aid for Scientific Research (A) No. 15H02022. We thank K. Fujiwara for fruitful discussion and thank Cooperative Research and Development Center for Advanced Materials, Institute for Materials Research, Tohoku University for fabrication of ceramic targets.
REFERENCES
After the submission of this paper, it has been reported that the mobility of La-doped BaSnO3 reaches 100 cm2 V−1s−1 probably owing to the reduction of unintentional impurity concentration by applying molecular beam epitaxy15 and the reduction of number of dislocations by homoepitaxy on BaSnO3 substrate.18
The lattice constant of (Srx, Ba1−x)SnO3 is obtained by assuming that SrSnO3 has an ideal cubic lattice with a = 4.034 Å and that the lattice constant changes linearly with x from BaSnO3 with a = 4.117 Å. The actual structure of SrSnO3 deviates slightly from the cubic lattice. However, this deviation does not alter our discussion to a great degree.