Epitaxial thin film growth
Epitaxial SrFeO2.5 thin films were grown on a SrTiO3 (001) substrates using pulsed laser epitaxy at 700 °C under 1 mTorr of oxygen partial pressure. For the electrical measurements, metallic Nb-doped (0.5 wt%) SrTiO3 substrate was employed as the bottom electrode. KrF excimer laser of 248 nm wavelength (IPEX 864, Lightmachinery, Canada) with an energy fluence of 1.3 J cm−2 and a repetition rate of 4 Hz was used.
Epitaxial CaFeO2.5 thin films were grown on a LaAlO3 (001) substrates (CrysTec GmbH, Germany) using pulsed laser deposition equipped with a KrF excimer laser (λ = 248 nm). A pellet was prepared by mixing CaCO3 (99.95%) and Fe2O3 (99.9%) powders (Sigma-Aldrich) and forming. The pellet was first heated at 850°C to eliminate CO2 and then crushed, ground, and pressurized to make a 1-inch-diameter button-shaped target. It was calcined at 850°C for 8.5 hours at an ambient condition. After the calcination, the pellet proceeded with the same previous processes. Then, it was sintered at 900°C for 9 hours at an ambient condition. The epitaxial CaFeO2.5 thin films were grown at a heater temperature of 650°C in an oxygen environment of 0.05 torr. Laser fluence and repetition rate were set to be ~1 J cm−2 and 10 Hz. All films were cooled down to room temperature at a rate of 10°C min−1 under an oxygen pressure of 500 torr.
In-situ TEM / STEM characterizations
Cross-sectional TEM samples were prepared by dual-beam focused ion beam system (Helios NanoLab G3 CX, FEI, USA) along the [101]O zone axis of both samples. 30 kV Ga ion beam followed by 5 kV Ga ion beam was used to fabricate the thin samples. After making the samples, we notched the samples into three parts isolating top electrode and film from bottom electrode for obtaining several electrode/film/electrode regions. To minimize the Ga ions contaminations and do thinning, we further milled the samples with a 0.1 keV Ar ion beam (PIPS II, Gatan, USA). Also, deposited Pt layers by focused ion beam system were totally milled for effective electric field biasing on the samples.
In-situ electrical biasing experiments were performed with an in-situ holder (STM-TEM, Nanofactory, USA) and a sourcemeter (6430, Keithley, USA). Tungsten tip equipped with the holder was piezo-controlled and made a proper contact with top electrode, Pt. Bottom electrodes, Nb-doped SrTiO3 and (Pr,Ca)MnO3, was attached to copper TEM grid connected to the holder ground using silver paste. After contact between the tip and top electrode, gradual DC voltage from 0 to 9 V was applied on the film by the sourcemeter. BF-TEM images were acquired before and after applying bias with a 200 kV field emission TEM (JEM-ARM200F, JEOL, Japan) at the Materials Imaging & Analysis Center of POSTECH, Republic of Korea. For the strong contrast depending on the ferroelectric switching, two-beam condition was used where the pseudocubic (110) reflection was excited.
Atomic scale imaging was conducted using 5th order aberration corrector (ASCOR, CEOS GmbH, Germany) equipped STEM (JEM-ARM200F, JEOL, Japan) at a 200 kV. Convergence semi-angle of the electron beam was 28 mrad for STEM imaging. Inner and outer angles of detectors were 45 mrad and 180 mrad for HADDF-STEM, respectively; and 10 mrad and 20 mrad for ABF-STEM, respectively. After acquiring atomic scale images, atomic peak position analysis was conducted using lab-made codes based on MATLAB (MathWorks, USA)36. To calculate the electric polarizations, atomic displacements were calculated
Each unit cell was defined by four neighboring A-site ions, Sr and Ca, so atomic displacements of irons were calculated by the difference between center of unit cell and irons peak position. The negative charge positions were disregarded under the assumption that the oxygen ions shifts are opposite to positive charge shifts, irons. Then, electric polarization was calculated by the following equation, where is the effective charge and displacements of atom i, V is the unit cell volume, respectively. Also, A-site buckling angles were calculated using two horizontally nearer A-site ions, and bottom angle between two vectors is indicated on the images. Finally, lattice ratio was calculated by height over distance between neighbour A-site atoms, and each ratio of the unit cell was indicated on the B-site. For the better understanding of the material structures, STEM simulation was performed by the multi-slice method based software (Dr. Probe, Ernst Ruska-Centre, Germany)37.
DFT calculations
Our first-principles density functional theory (DFT) calculations were performed using the Perdew Burke Ernerhof of generalized gradient approximation (GGA-PBE)38 and the projector-augmented wave (PAW) method with a plane-wave basis39, as implemented in Vienna Ab Initio Simulation Package (VASP)40. For the Brillouin zone integration, the kinetic cut-off energy for the plane wave basis set to 500 eV and Γ-centered k-point grid was chosen for the brownmillerite structures of CaFeO2.5 and SrFeO2.5, comprised of 36 atoms. The structures were fully relaxed until the forces were less than 10-2 eV/Å with an energy convergence of 10-6 eV/cell. To treat the localized d electrons of Fe, we employ the effective Hubbard interaction parameters of U = 5 eV and 3 eV on Fe of CaFeO2.5 and SrFeO2.5, respectively, within G-type antiferromagnetic order.
The energy barrier for the polarization switching was estimated using nudged elastic band (NEB) method41. The polarization in tetrahedral chains was calculated using the born effective charge tensors obtained from Berry phase method42. For phonon calculations, anti-polar Pnma CaFeO2.5 and polar I2bm SrFeO2.5 structures were fully optimized in energy to 10-8 eV/cell until the forces were less than 10-3 eV/Å and phonon dispersions were calculated using the small displacement method as implemented in the PHONOPY code43.
PFM measurements
Ferroelectric up and down domains with a width of 700 nm were created by scanning areas of interest in the SrFeO2.5 and CaFeO2.5 thin films where +8 V for down domains and -8 V for up domains were applied to the conductive SPM tips (EFM, Nanosensors, for the SrFeO2.5 thin film, and Multi75E-G, Budget Sensors, for the CaFeO2.5 thin film) while the bottom was grounded. The written domains were visualized by PFM using the same SPM tips with an ac bias voltage of 1 V at around the contact resonant frequencies of 340 kHz (EFM) or 370 kHz (Multi75E-G).
References for the Methods section
36 Han, J., Go, K.-J., Jang, J., Yang, S. & Choi, S.-Y. Materials property mapping from atomic scale imaging via machine learning based sub-pixel processing. npj Computational Materials 8, 196 (2022).
37 Barthel, J. Dr. Probe: A software for high-resolution STEM image simulation. Ultramicroscopy 193, 1-11 (2018).
38 Perdew, J. P., Burke, K. & Wang, Y. Generalized gradient approximation for the exchange-correlation hole of a many-electron system. Phys. Rev. B 54, 16533-16539 (1996).
39 Blöchl, P. E. Projector augmented-wave method. Phys. Rev. B 50, 17953-17979 (1994).
40 Kresse, G. & Furthmüller, J. Efficient iterative schemes for ab initio total-energy calculations using a plane-wave basis set. Phys. Rev. B 54, 11169-11186 (1996).
41 Henkelman, G., Uberuaga, B. P. & Jónsson, H. A climbing image nudged elastic band method for finding saddle points and minimum energy paths. J. Chem. Phys. 113, 9901-9904 (2000).
42 King-Smith, R. D. & Vanderbilt, D. Theory of polarization of crystalline solids. Phys. Rev. B 47, 1651-1654 (1993).
43 Togo, A., Oba, F. & Tanaka, I. First-principles calculations of the ferroelastic transition between rutile-type and CaCl2-type SiO2 at high pressures. Phys. Rev. B 78, 134106 (2008).