Colossal Dielectric Perovskites of Calcium Copper Titanate (CaCu3Ti4O12) with Low‐Iridium Dopants Enables Ultrahigh Mass Activity for the Acidic Oxygen Evolution Reaction

Abstract Oxygen evolution reaction (OER) under acidic conditions becomes of significant importance for the practical use of a proton exchange membrane (PEM) water electrolyzer. In particular, maximizing the mass activity of iridium (Ir) is one of the maiden issues. Herein, the authors discover that the Ir‐doped calcium copper titanate (CaCu₃Ti₄O₁₂, CCTO) perovskite exhibits ultrahigh mass activity up to 1000 A gIr −1 for the acidic OER, which is 66 times higher than that of the benchmark catalyst, IrO2. By substituting Ti with Ir in CCTO, metal‐oxygen (M‐O) covalency can be significantly increased leading to the reduced energy barrier for charge transfer. Further, highly polarizable CCTO perovskite referred to as “colossal dielectric”, possesses low defect formation energy for oxygen vacancy inducing a high number of oxygen vacancies in Ir‐doped CCTO (Ir‐CCTO). Electron transfer occurs from the oxygen vacancies and Ti to the substituted Ir consequentially resulting in the electron‐rich Ir and ‐deficient Ti sites. Thus, favorable adsorptions of oxygen intermediates can take place at Ti sites while the Ir ensures efficient charge supplies during OER, taking a top position of the volcano plot. Simultaneously, the introduced Ir dopants form nanoclusters at the surface of Ir‐CCTO, which can boost catalytic activity for the acidic OER.


Synthesis of HTO nanobelts
1.88 g of TiO2 was dispersed in 91 ml of 10 M NaOH and stirred for 24h. Then the mixed solution was relocated to a Teflon-lined autoclaves which was hydrothermally reacted at 240ºC for 24h. The resulting sodium titanate nanobelts were washed several times with DIW and then soaked in a 0.2M HCl solution and stirred for 4h. During this step, ion exchange between sodium and hydrogen can be taken palace resulting in hydrogenated titanate (H2Ti3O6, HTO) nanobelts. Finally, the resulting precipitate was obtained after washing several times with DIW and drying at 70ºC overnight.

Synthesis of CaCu3Ti4O12 (CCTO) nanobelts
To begin, 0.2 g HTO NWs, 0.396 g Cu(NO3)2.2.5H2O and 1.17 g Ca(NO3)2.4H2O (99%) were dispersed in 60 ml C2H5OH anhydrous. The solution was then sonicated for 30 min and saturated with nitrogen. Then, this solution was heated in an oven at 155ºC for 24h. After hydrothermal process was complete, the precipitate was collected, washed several with DIW, and dried at 70ºC overnight. The powder was transferred to furnace, and calcined in air at 500°C/600ºC/700°C/800°C for 2h with heating rate of 3ºC/min. After cooling to room temperature, the as-obtained powder was treated in 0.2M HCl for 1h to remove the copper (II) oxide (CuO). Finally, the remaining precipitate was washed three times with DIW and twice times with ethanol, and dried at 70ºC overnight.

Synthesis of Ir-CCTO nanobelts
Ir-doped CCTO NWs was synthesized by hydrothermal method. With magnetic stirring, a mixed of 0.15g CCTO and 0.2/1/5/10 % of IrCl3.xH2O was dispersed in 60 ml DIW. Then, the solution was transferred into a Teflon and heated at 150ºC for 24h. Finally, the product was dried in an oven at 70ºC overnight after washing three times with H2O to obtain Ir-CCTO NWs.

Synthesis of IrO2-CCTO composite
IrO2-CCTO nanocomposite was synthesized by hydrothermal method. With magnetic stirring, a mixed of 0.15g CCTO and 5 % of IrO2 was dispersed in 60 ml DIW. Then, the solution was transferred into a Teflon and heated at 150ºC for 24h. Finally, the product was dried in an oven at 70ºC overnight after washing three times with H2O to obtain IrO2-CCTO nanocomposite.

Synthesis of IrO2-TiO2 composite
IrO2-TiO2 nanocomposite was synthesized by hydrothermal method. With magnetic stirring, a mixed of 0.15g TiO2 and 5 % of IrO2 was dispersed in 60 ml DIW. Then, the solution was transferred into a Teflon and heated at 150ºC for 24h. Finally, the product was dried in an oven at 70ºC overnight after washing three times with H2O to obtain IrO2-TiO2 nanocomposite.

Synthesis of Ir-TiO2 composite
The as-prepared HTO (0.2 g) and 0.037 g of IrCl3 were mixed in 100 ml of DI water by stirring for 10 h followed by 20 min of sonication. The mixed solution was transferred into a Teflon-lined autoclave for a hydrothermal reaction at 200 °C for 24 h. The resulting powder was washed with DI water theree times, collected via centrifugation, and dried at 70 °C.

Characterizations
Scanning electron microscopy SEM (model S4800; Hitachi) was performed on the samples to investigate the structural parameters.

Transmission electron microscopy
TEM (Talos F200X; Thermo Fisher Scientific) was employed to collect high-resolution TEM images with elemental distributions using an equipped energy dispersive X-ray (EDX) spectroscopy.
Dual-beam focused ion beam (AURIGA CrossBeam Workstation; Carl Zeiss) was employed to acquire atomic-scale scanning TEM (STEM) images with a EDX chemical mapping information using JEOL-EDX in the STEM imaging mode, and each detector has an effective detection area of 100 mm 2 . The sample drift during the acquisition was tried to eliminate by tracking the reference atom position which was determined at the starting of the measurement.

Powder X-ray diffraction
XRD data were collected with Cu Kα radiation under the operating set-up of 40 kV and 100 mA using a X-ray diffractometer (Rigaku D/Max 2550).

X-ray photoelectron spectroscopy
XPS (VG ESCALAB 200i; Themo Fisher Scientific) was employed to track the samples' surface electronic structure. The spectrometer energy calibration was calibrated using the C peak position. Pass energies of 100 eV and 20 eV were utilized for survey and high-resolution scans, respectively.

Inductively coupled plasma-optical emission spectroscopy
Ir amount in the samples were calculated via inductively coupled plasma-optical emission spectroscopy (ICP-OES; Thermo Scientific; iCAP6500 Duo). RF power and the wavelength were set as 1350 W and 214.423 nm, respectively. The samples were firstly dissolved in a mixed solution of nitric and hydrochloric acid and heated at 150°C for about 30 min. Then, hydrofluoric acid was added to fully dissolve all materials and heated at 150°C until the solution becomes transparent.

X-ray absorption spectroscopy
XAS were measured at the 1D-PAL-KIST beamline of the Pohang Accelerator Laboratory (PAL). A double crystal monochromator (DCM) Si(111) was employed at 1D beamline and measurements were taken with a ring current of 300 mA. The harmonics of the incident X-ray beam were detuned by DCM. The energy was calibrated using reference metallic foils prior to measurements. The obtained XAS data were acquired in transmission mode and the Ti, Cu, and Ir metallic foils were used as reference. The acquired XAS data were analyzed using ATHENA software.[1-2] In order to observe the wavevector and interatomic distance (R) data in three dimensions, Continuous Cauchy wavelet transform (CCWT) analysis were accomplished with k 2 -weighted signals with a k-space range of 2.0 -12.0 Å -1 . Near edge Xray absorption fine structure (NEXAFS) spectra of O K-edge and Ti L3,2 edge were measured at 10D beamline with bending magnet at PAL. NEXAFS measurements were taken at room temperature with a resolution of 0.01 eV and the spectra were acquired in a total electron yield (TEY) mode under a base pressure of 3×10 -10 Torr.

Electrochemical Measurements Preparation of catalyst inks and electrodes
The catalyst ink was prepared through a dispersion of 5 mg of the catalyst in 750 µL of DIW and 250 µL of ethanol containing 20 µL of Nafion 117 solution. After 30 min sonication, 5 µL of the catalyst ink was coated on a clean surface of glassy carbon electrode (GCE) with diameter of 3 mm yielding a loading amount of approximately 0.35 mg cm -2 . The GCE was then dried at room temperature under air.

Evaluation of catalytic activity for the acidic OER
The electrocatalytic properties of the samples were measured in 0.1M HClO4 electrolyte using a three-electrode cell connected to a potentiostation (model Autolab PGSTAT; Metrohm). A rotating disk electrode (RDE) was utilized to investigate the electrochemical properties of the catalysts. A typical three-electrode setup (Pt as counter and Ag/AgCl as reference electrodes) was employed for electrochemical tests. The recorded potentials were re-calculated against the reversible hydrogen electrode (RHE). Electrolyte (0.1 M HClO4) was purged with Ar2 gas for approximately 30 minutes. In addition, 50 cycles of cycling voltammetry (CV) scans were performed in the OER potential window to electrochemically stabilize the catalysts' surface.
Then, linear sweep voltammetry (LSV) curves were recorded at a 5 mV s −1 scan rate. In addition, the electrochemical impedance spectroscopy (EIS) was performed at 1.4VRHE in a frequency range from 10 0000 Hz to 0.1 Hz. All LSV polarization curves were iR-corrected using the solution resistance (Rs) measured by EIS. Tafel plots were derived from the iRcorrected LSV polarization curves where the Tafel slopes were calculated from the equation: η = b log j + a where (b: Tafel slope, j: current density, η: overpotential).

Determination of electrochemical double layer capacitance (Cdl)
CVs were measured in the non-Faradic potential range (e.q., 0.83 ~ 0.93 VRHE) with a different scan rates. The difference of anodic and cathodic currents (J = Janodic-Jcathodic) at the middle of potential (0.88 VRHE) was plotted against the scan rate in which the slope corresponds to twice the Cdl of the catalyst.

Determination of electrochemical surface area (ECSA)
The ECSAs of sample was obtained from the measured Cdl. Notably, the charging of double layer is originated from the non-Faradaic currents which has a linear relationship with the active surface area; the 1cm 2 of flat surface area has a specific capacitance which is equal to Cdl value of 40 μF cm -2 . [3] Therefore, the Cdl is directly related with the ECSA as: ECSA = Cdl of catalyst (mF cm -2 )/0.04 (mF cm -2 ).

Determination of Ir mass activity for OER
For the mass activity calculations, the current density at a certain potential in LSV polarization curves was normalized with total mass of Ir loaded on GCE electrode which was determined from the ICP-OES results.

Determination of turn over frequency (TOF)
TOF was calculated according to the equation: , where j is the current density at a certain potential (Acm -2 ), A is surface area of the working electrode (cm -2 ), F is Faraday constant 96,458 C mol -1 , and Ns is concentration of active sites (mol cm -2 ). [4] The value of Ns was determined by CV measurements where the oxidative peak currents generated by Ir has a linear relationship with a different scan rate. Here, the slope in linear plot is equal to: slope = n 2 F 2 AN S /4RT, in which n, F,A, Ns, R, T are the number of electrons transferred, Faradic constant, the surface area of the electrode, the surface concentration of active sites, the ideal gas constant, and the temperature, respectively.

Determination of Faradaic efficiency using rotating ring disk electrode (RRDE)
First, the collection efficiency for our RRDE system was estimated in 0.05 M Na2SO4 dissolved with 4 mM of potassium ferricyanide, K3Fe(CN)6 , electrolyte. It can be noted that the ferrocyanide/ferricyanide half reaction occurs thorugh a simple and single-electronic reaction, and thus often used as the standard mean for estimating collection efficiency of a certain RRDE equipment. [5] Fe(CN) 6 3− + e − → Fe(CN) 6 4− (reduction of ferricyanide) Fe(CN) 6 4− + e − → Fe(CN) 6 3− (oxidation of ferrocyanide) Initially, CV of bare GCE of RRDE was conducted using a scan rate of 100 mV/s without rotation. Then, LSV was perfomred on the GCE by applying potentials from 0 to 1. 23

Density Functional Theory (DFT) Calculations
Spin-polarized DFT calculations were conducted by using the Vienna ab initio simulation package (VASP) [6][7][8][9]. The projector augmented wave (PAW) method and exchangecorrelation functional of the generalized gradient approximation (GGA) followed by the work of Perdew-Burke-Ernzerhof (PBE) were used [10][11][12]. The H 1s 1 ,O 2s 2 2p 4 , Ca 3s 2 3p 6 4s 2 , Ti 3s 2 3p 6 4s 2 3d 2 , Cu 3p 6 4s 1 3d 10 , Ir 5d 7 6s 2 electrons were explictly treated as valence states. The energy cutoff of 500 eV was adopted. The criteria of electronic energy convergence and geometry optimization convergence were 10 -6 eV and 0.01 eV·Å -1 , respectively. The Hubbard correction scheme [13] was applied to explicitly correct the Cu 3d and Ir 5d states with Ueff of 4 eV and 2 eV, respectively, taken from references [14,15]. The atomic structures of cubic-based ABO3 type perovskite oxides with 2 × 2 × 2 supercells were used to simulate the bulk CaCu3Ti4O12 system. The Monkhorst-Pack scheme with 5 × 5 × 5 k-points meshes was adopted to sample the Brillouin zone for bulk systems [16]. The computational methods used in this work well-reproduced the lattice constant and the magnetic moment on CuO4 in CCTO compared with previous studies (Table S7). The oxygen vacancy formation energy, ( ) was calculated from Equation (1): where indicates the total energy of an oxygen-deficient or perfect system. and represent the number of oxygen vacancies and the chemical potential of oxygen, respectively.
To calculate the adsorption free energy (∆ ) of oxygen-intermediates for the OER pathway, (001) surface which is the most stable surface for cubic-based ABO3 perovskite oxides was chosen [17][18][19][20]. The slabs were simulated with seven layers. The bottommost two layers were fixed maintaining the optimal bulk structures, whereas the top five layers were fully relaxed.

( ) and
( ) indicate the total energy of the substrate and adsorbate in a vacuum, respectively.
∆ , , and ∆ represent the difference in zero-point energy between the adsorbed state and the gas phase, temperature and change of entropy, respectively, and these values for gasphase molecules at 300 K were directly taken from thermodynamic tables. The zero-point energies for adsorbants were obtained by vibrational frequencies calculations. , , and denote the number of electron transferred, electric charge, and applied potentail, respectively.