Supplementary Information Interface energetics and atomic structure of epitaxial La1-xSrxCoO3 on Nb:SrTiO3 Quentin Van Overmeere1,*, John D. Baniecki2,*, Takashi Yamazaki2, Dan Ricinschi3, Hiroyuki Aso2, Yusuke Miyata4, Hiroaki Yamada4, Norifumi Fujimura4, Yuji Kataoka2, Yoshihiko Imanaka2 1 Institute of Mechanics, Materials and Civil Engineering, Université catholique de Louvain, B-1348 Louvain-la-Neuve, Belgium 2 Fujitsu Laboratories, Atsugi, Kanagawa-ken, Japan 3 Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology, 4259-G2-25 Nagatsuta, Midori-ku, Yokohama, 226–8502, Japan 4 Graduate School of Engineering, Osaka Prefecture University, 1-1 Gakuen-cho, Naka-ku, Sakai, Osaka 599-8531, Japan Corresponding authors: quentin.vanovermeere@uclouvain.be and john.d.baniecki@jp.fujitsu.com Composition of the films and phase purity ~100 nm-thick films deposited on either LaAlO3 ((La,Sr)CoO3) or SrTiO3 (LaCoO3) substrates were dissolved in aqua regia and the atomic ratios determined by inductively coupled plasma-optical emission spectroscopy (ICP-OES). The film compositions corresponded to the desired stoichiometry within instrumental resolution. The compositions were also determined for thinner films by Rutherford backscattering spectroscopy (RBS), although the uncertainty is higher than for ICP-OES. ICP-OES and RBS on LaCoO3 (LCO) yielded La/Co ratios of 1.04 ± 0.03 and 1.01 ± 0.06 respectively. No secondary phases were detected from x-ray diffractograms for thin films (Figure S1(a)) and even for thicker films, as shown in Figure S2. Further support for the phase purity and cation stoichiometry of the films was inferred from transport measurements, shown in Figure S3 and S4. Room-temperature resistivity values were relatively close to values reported previously for bulk polycrystalline La1-xSrxCoO3. The higher resistivities of our films on (001) SrTiO3 than on (La0.3Sr0.7)(Al0.65Ta0.35)O3 (LSAT) could be due to a depletion layer effect or strain-dependent resistivity for the films deposited on STO. LaCoO3 and La89Sr11CoO3 films with Sr concentrations below the insulator-metal transition (18 at% Sr for bulk La1-xSrxCoO3) had a semiconductor-like resistivity-temperature dependence, while La70Sr30CoO3 films had a metallic-like resistivity-temperature dependence, as shown in Figure S4. 1 Figure S1. Structural characterization of single phase epitaxial LaCoO3 (LCO), La89Sr11CoO3 (LSr11CO), and La70Sr30CoO3 (LSr30CO) ~15 nm-thick films grown on Nb-doped SrTiO3(001) single crystal substrates (NSTO). (a,b) X-ray symmetric -2 diffraction patterns. Arrows indicate Kiessig fringes, associated with low roughness of the films. (c) Reciprocal space maps of the (013) reflection of ~15 nm-thick LCO, LSr11CO and LSr30CO films. Figure S2. Symmetric X-ray diffraction on ~100 nm-thick LaCoO3 on Nb-doped SrTiO3(001) show no evidence for secondary phases. The small reflection at 33.3° corresponds to the most intense reflection for polycrystalline LaCoO3 (ICDD #00-048-0123), possibly caused by the formation of droplets for longer deposition times. 2 Figure S3. Resistivity of La1-xSrxCoO3 as a function of doping for bulk ceramics (black squares),1 bulk polycrystals (gray squares),2 bulk single crystals (orange and magenta stars)3,4 and our epitaxial thin films on LSAT (red losanges) and SrTiO3 (blue triangles). Figure S4. Resitivity vs. temperature for LaCoO3, La89Sr11CoO3 and La70Sr30CoO3 thin films on LSAT (001). 3 Figure S5. HAADF STEM image of (a) LaCoO3 (LCO) (b) La89Sr11CoO3 (LS11CO) (c) La70Sr30CoO3 (LS30CO) thin films on Nb:SrTiO3(001) (NSTO) substrates. Atomic ordering of dark planes at thicknesses larger than ~3 unit cells is visible for all film compositions. High resolution photoelectron spectroscopy The core level evolution with LCO film thickness are visible in Figure S6. For all film compositions, the intensity attenuation of the substrate peaks are predicted by Cumpson & Seah’s attenuation length formula and indicate a layer-by-layer growth mode, facilitating thickness-dependent analyses (Figure S7). The Co 2p and La 3d spectra have similar characteristics to those reported for bulk LCO. For all peaks, the full width at half-maximum increases slightly during formation of the first 2 nm, likely due to the built-in potential at the interface since interfacial reactions could be ruled out from STEM analyses. The La 3d peak shows spin-orbit splitting between La 3d3/2 and La 3d5/2 and since the former overlaps with the La Auger LMM transition, further analysis is focused to the La 3d5/2 peak which is multiplet split in three distinct contributions.5 The multiplet splitting separation of 4.10 ± 0.03 eV is consistent with previous values reported for bulk LaCoO3.6 The main peak, La 3d5/2 cf0, relates to photoemission without charge transfer to the La 4f orbital while two satellite peaks La 3d5/2 cf1L are associated to the final state with charge transfer from the ligand VB to La 4f.5 The La 3d5/2 peak was deconvoluted into these three components, and the La 3d5/2 cf0 is subsequently used for accurate core-level determination for valence band offset determination. The O 1s core level originates from two contributions with the higher binding energy contribution corresponding to oxygen in NSTO, while the lower binding energy contribution originates from the LCO film. A small satellite at higher binding energy is also visible for the thickest film. Although this contribution is frequently reported to correspond to hydroxyl groups from adsorbed water on the film surface, recent reports indicated it may be an intrinsic feature related to different oxygen coordination at oxide surfaces.5 O 1s spectra acquired at variable emission angle on the thickest LaCoO3 indicated that the high binding energy component is indeed a surface contribution, although surface hydroxyl formation cannot be ruled out despite the chamber pressure being lower than 6 10-8 Pa. The La 3d, Co 2p and O 1s core levels are affected by doping as shown in Figure S8. The La 3d intensity decreases with increase of the dopant concentration as expected for substitutional doping of Sr on La sites. The La 3d and O 1s core level peaks shift to smaller binding energies by 0.1 eV for La89Sr11CoO3 and 0.4 eV for La70Sr30CoO3 compared to LaCoO3 and reflect the shift in chemical potential with Sr doping, similar to bulk LCO.7 The Co 2p peak 4 shift by 0.2 and 0.5 eV to higher binding energy (BE) and broaden on the high BE side, since Sr introduces high spin (HS) Co4+.8 The broader and asymmetric Co 2p and introduction of Co4+ upon Sr doping would lead to larger uncertainty on the band offset value than when the La 3d transition is considered. Comparing these shifts to the shift in the valence band spectra with doping (Figure 3(b)), the lowest BE peak (0.8 eV for LCO) shifts by 0.1 and 0.3 eV while the second lowest BE peak (3.0 eV for LCO) shifts by 0.2 and 0.7 eV, respectively for La89Sr11CoO3 and La70Sr30CoO3. Since Fermi level shifts of 0.18 and 0.24 eV are obtained by workfunction measurements with UPS (Figure S8 and Table S1), these peak shift values indicate that a rigid band model is not valid. Figure S6. X-ray induced photoelectron spectra of epitaxial LaCoO3 films deposited in stepwise sequences on Nb:SrTiO3(001) substrate. Figure S7. Attenuation of Ti 2p substrate photoelectron core level as a function of the La1-xSrxCoO3 film thickness on Nb:SrTiO3 substrates. The continuous line models attenuation for layer-by-layer (Frank – van der Merwe) growth, with calculated attenuation length 1.65 nm. 5 Figure S8. Background-subtracted X-ray induced photoelectron spectra of the La 3d, Co 2p and O 1s regions of ~15 nm-thick LaCoO3 (LCO), La89Sr11CoO3 (LSr11CO) and La70Sr30CoO3 (LSr30CO) films. Figure S9. He(I) induced photoelectron spectra of ~15 nm-thick LaCoO3 (LCO), La89Sr11CoO3 (LSr11CO) and La70Sr30CoO3 (LSr30CO) films and of Nb:SrTiO3 (NSTO) substrate. DFT calculations First-principles calculations have been done using the density functional theory under the local density approximation, including the correction for the strongly correlated electrons of d bands (LDA+U) as implemented in the ABINIT package.9-11 For all chemical elements, projected-augmented wave pseudo-potentials with electronic wavefunction plane-waves truncated at about 400 eV (15 Ha) were used. Both valence and semicore states, specifically the La 5s, 5p, 5d and 6s electrons, the Co 3d and 4s electrons, the O 2s and 2p electrons were included into the calculations. In order to account for the exchange and correlation of localized electrons, the full localized limit corrections of Lichtenstein et al.12 were used with Ueg = 8 eV, Jeg = 1 eV for the Co atoms with filled eg orbitals and Ut2g = 5 eV, Jt2g = 1 eV for Co atoms that have filled t2g orbitals only. The influence of ordered spins states with and without planes of oxygen vacancies on the atomic and electronic structure of LCO was studied using DFT for 3x1x2 supercells with antiferromagnetic (AFM) stacking of rows of a HS Co atom and two LS Co atoms (HLL) (Figures S10(a) and (b)) and a ferromagnetic structure with an oxygen vacancy plane (Figure S10(c)). Without oxygen vacancies the La-La spacing for the HS Co3+ block is dilated to 4.05 Å (bulk psudeocubic LCO a = 3.85 Å) and slightly contracted to 3.82 Å for the LS Co3+ blocks. The calculated spacings are close to previous calculations but significantly different 6 than observed in the STEM data of Figure 2.13,14 Introducing oxygen vacancies yielded La-La distances closest to experimental values for all investigated ordered spin states. The presence of the ordered vacancy planes results in significant changes in both atomic and electronic structure. A larger La-La dilation of 4.4 Å is observed in the HS Co2+ block with the vacancy plane as well as larger contractions of 3.7 Å and 3.6 Å between La atoms in the LS Co 3+ and LS Co2+ blocks compared to the oxygen-stoichiometric superstructure. For the stoichiometric supercell of Figure S10(a) the total DOS exhibits a four peak structure. As revealed by the mresolved PDOS of the Co atoms the LS Co3+ t2g PDOS exhibits two peaks separated by ~ 4 eV while the HS Co3+ atom contributes a peak at lower energies, as well as a peak of t2g character in the conduction band. The top of the VB has thus significant Co 3d t2g and O 2p character, with the second peak of predominantly O 2p character and the third peak again of Co 3d t2g character. These features are similar to those predicted for MS bulk LCO with rhombohedral symmetry.15 In presence of O vacancies, the PDOS structure changes as indicated in the main text. Figure S10. DFT calculations. 3x1x2 supercells of (a) stoichiometric LCO with antiferromagnetic (AFM) stacking of rows of a HS Co3+ atom (dark blue spheres) and two LS Co3+ atoms (light blue spheres), (b) oxygen deficient LCO with AFM stacking of rows of a HS Co2+ atom (purple spheres), LS Co3+ atom, and LS Co2+ atom (magenta spheres) ,and (c) a oxygen deficient ferromagnetic structure. Interplanar spacings are in Å and yellow arrows represent the spins states. (d) DFT calculated total density of states (TDOS) and summed projected density of states (PDOS); the OPDOS and LaPDOS includes all O and La atoms in the supercell while the Co PDOS is summed over the two Co atoms with similar valence in the AFM or FM structures. 7 Sample contamination Prior to film growth, the substrates were in situ heated to 923 K in 500 mTorr O2 for 1 h to remove chemisorbed surface species. XPS survey spectra of the in situ heated NSTO surface (cooled in vacuum) and that of a 15 nm LaCoO3 film (cooled in 500 mTorr O2) are featureless near 285 eV indicating the surfaces are free from carbon contaminants (Figure S11). Figure S11. X-ray induced photoelectron survey spectra of the Nb:SrTiO3 substrate (NSTO) and ~15 nm-thick LaCoO3 film on SrTiO3. No C-contamination is detected within the detection limit of XPS. References 1 Y. Wang, Y. Sui, P. Ren, L. Wang, X. Wang, W. Su, H.J. Fan, Inorg. Chem. 49, 3216 (2010). 2 J. Wu, C. Leighton, Phys. Rev. B 67, 174408 (2003). 3 J. Wu, H. Zheng, J.F. Mitchell, C. Leighton, Phys. Rev. B 73, 020404R (2006). 4 H.M. Aarbogh, J. Wu, L. Wang, H. Zheng, J.F. Mitchell, C. Leighton, Phys. Rev. B 74, 134408 (2006). 5 M.F. Sunding, K. Hadidi, S. Diplas, O.M. Løvvik, T.E. Norby, A.E. Gunnæs, J. Electron Spectrosc. Rel. Phen. 184, 399 (2011). 6 R.P. Vasquez, Phys. Rev. B 54, 14938 (1996). 7 T. Saitoh, T. Mizokawa, A. Fujimori, M. Abbate, Y. Takeda, M. Takano, Phys. Rev. B 56, 1290 (1997). 8 A. Chainani, M. Mathew, D.D. Sarma, Phys. Rev. B 46, 9976 (1992). 9 X. Gonze, B. Amadon, P.-M. Anglade, J.-M. Beuken, F. Bottin, P. Boulanger, F. Bruneval, D. Caliste, R. Caracas, M. Côté, T. Deutsch, L. Genovese, Ph. Ghosez, M. Giantomassi, S. Goedecker, D.R. Hamann, P. Hermet, F. Jollet, G. Jomard, S. Leroux, M. Mancini, S. Mazevet, M.J.T. Oliveira, G. Onida, Y. Pouillon, T. Rangel, G.-M. Rignanese, D. Sangalli, R. Shaltaf, M. Torrent, M.J. Verstraete, G. Zerah, and J.W. Zwanziger, Comput. Phys. Commun. 180, 2582 (2009). 10 F. Bottin, S. Leroux, A. Knyazev, G. Zerah, Comput. Mater. Sci. 42, 329 (2008). 11 B. Amadon, F. Jollet, M. Torrent, Phys. Rev. B 77, 155104 (2008). 12 A.I. Lichtenstein, V.I. Anisimov, J. Zaanen, Phys. Rev. B 52, R5467 (1995). 8 13 J.H. Kwon, W.S. Choi, Y.-K. Kwon, R. Jung, J.-M. Zuo, H.N. Lee, M. Kim, Chem. Mater. 26, 2496 (2014). 14 N. Biškup, J. Salafranca, V. Mehta, M.P. Oxley, Y. Suzuki, S.J. Pennycook, S.T. Pantelides, M. Varela, Phys. Rev. Lett. 112, 087202 (2014). 15 S. Mukhopadhyay, M.W. Finnis, N.M. Harrison, Phys. Rev. B 87, 125132 (2013). 9