Self-sustainable protonic ceramic electrochemical cells using a triple conducting electrode for hydrogen and power production

The protonic ceramic electrochemical cell (PCEC) is an emerging and attractive technology that converts energy between power and hydrogen using solid oxide proton conductors at intermediate temperatures. To achieve efficient electrochemical hydrogen and power production with stable operation, highly robust and durable electrodes are urgently desired to facilitate water oxidation and oxygen reduction reactions, which are the critical steps for both electrolysis and fuel cell operation, especially at reduced temperatures. In this study, a triple conducting oxide of PrNi0.5Co0.5O3-δ perovskite is developed as an oxygen electrode, presenting superior electrochemical performance at 400~600 °C. More importantly, the self-sustainable and reversible operation is successfully demonstrated by converting the generated hydrogen in electrolysis mode to electricity without any hydrogen addition. The excellent electrocatalytic activity is attributed to the considerable proton conduction, as confirmed by hydrogen permeation experiment, remarkable hydration behavior and computations.

In-situ high-temperature XRD in hydration process at 600 ○ C. The chemical expansion is observed by switching the exposed gas from dry air to wet air.          The measurements were carried out at different temperature (500~600 ○ C) and steam pressure (15% and 40%). In the hydrogen electrode side, dry 5% H2 was used as feed gas. Gas chromatography was used to monitor the hydrogen concentration change at different constant current densities. The ratio of experimental and theoretical hydrogen production amounts is calculated as the efficiency.   The oxygen non-stoichiometry was determined by iodometric titration method combining with TGA result. As temperature increased, the oxygen deficiency was slightly reduced, indicating the slow oxidation upon the oxidizing condition (Supplementary Figure 3). As PCEC is operated between 500 ○ C and 600 ○ C in this work, the oxygen non-stoichiometry is about ~0.08. The chemical compatibility between PNC electrode and BCZYYb electrolyte was confirmed by XRD after calcining the mixture powder with 50:50wt% ratio at 1000 ○ C in air for 5 h. There were not any observable impurity peaks in the pattern (Supplementary Figure 4). The surface of PNC particles was examined by X-ray photoelectron spectroscopy (XPS) to find out the co-existence of Pr, Ni, Co and O elements (Supplementary Figure 5a). The valences of each metals were detected and analyzed. Specifically for Ni dopant, there are two main +2 and +3 valences found with major contributing to +2, indicating the formation of intrinsic oxygen vacancy with considerable concentration (Supplementary Figure 5b-5d).

Supplementary Note 2: DFT calculations
In the orthorhombic Pbnm structure of PCO before doping, Pr and Co atoms occupy the  Eperfect are the total energy of the 160-atom supercell after and before oxygen removal, respectively. EO2 is the energy of a spin polarized O2 molecule in its ground state triplet state.
Our final GGA+U results are reported in Supplementary Figure 6c.
With 50% replacement of Co sites with Ni atoms, the oxygen vacancy formation energy can be significantly decreased, e.g., from 3.40 eV to 1.61 eV at O1 sites and from 3.49 eV to 1.72 eV at O2 sites. Importantly, since the formation of oxygen vacancies is a prerequisite for water dissociative incorporation into the defective lattice, the formation energy reduction may induce essential water hydration. One should stress that LSCF oxide is an oxygen vacancy-rich material.
Evidently, a large oxygen deficiency is not the only requirement to obtain an efficient proton  Figure 2E in the main text), our spin polarized DFT calculations also suggest a reduction of proton migration energy due to Ni doping, although to a lesser degree.

Supplementary Note 3: Hydration behavior of PNC oxide observed in high-temperature Xray diffraction and thermogravimetry analysis.
The phase structure of PNC electrode powder sample was monitored during the process of changing air humidity from dry to wet (~3% H2O) to observe the chemical expansion due to insertion of water into the crystal structure. The proton defects ( . ) are formed in the oxides by the Wagner hydration mechanism (Equation 1) in which water molecule combines with oxygen vacancy to generate two mobile protons in the anion sublattice.
The chemical expansion due to hydration process can be observed by examining the lattice parameter with XRD ( Supplementary Figure 9). At 600 ○ C, when the gas was switched from dry air to wet air, the clear shift of diffraction peaks towards left side can be observed, indicating the expansion of crystal structure.
The hydration behavior of the electrode in wet air condition was also observed in TGA examination, as shown in Supplementary Figure 10. The powder was firstly heated up to 950 ○ C with ramping rate of 5 ○ C/min in dry air to remove the surface residues. After cooling to 500 ○ C and reaching the equilibrium, the wet air was flushed into the chamber to start the hydration process. The weight increase was recorded as function of hydration time as total time is 15 h.
The weight gain after hydration for PNC is found to be highest ~0.055%, comparing to 0.048% for PBSCF, and 0.031% for LSCF, respectively.
Results from hydrogen permeation experiment are meaningful to demonstrate the presence of mixed proton and electron conduction. As well known, dense membranes for hydrogen permeation based on high temperature proton conductors have been studied for many years due to mixed electron and proton conductivity in ceramic-metal or dual ceramic-ceramic composites.
About the concern of hydrogen permeation through grain boundary, it is common to see that grainboundary diffusion could contribute an appreciable amount to the overall permeation rate in palladium and its alloys, particularly in the case of a nanostructured membrane ( (5). Contrary to these findings, for the ceramics LaCoO3-δ, La0.3Sr0.7CoO3-δ, La0.6Sr0.4Co0.2Fe0.8O3-δ, Ba0.5Sr0.5Co0.8Fe0.2O3-δ and CaTi0.8Fe0.2O3-δ, the increase of grain size leads to an enhanced permeation (6)(7)(8)(9)(10). In this work, we agree that hydrogen permeation through grain boundary is possible, as we can see very small amount of hydrogen flux can be measured in dry 3% H2, e.g., 2.6×10 -8 mol/cm 2 s at 400 ○ C. However, the hydrogen flux was significantly increased by 5 times when PNC is exposed to wet gas, which indicates the hydration process introduced effective proton conductivity that facilitated permeation flux.

Supplementary Note 4: Fourier-transform infrared spectroscopy (FTIR) and temperature programmed desorption (TPD) techniques for detecting proton defects in PNC.
To further demonstrate the presence of protons in the PNC lattice, two other techniques were employed to measure the water signal during the dehydration process. For FTIR examination, the PNC powder sample was firstly hydrated at 600 ○ C in wet condition and then cooled down to room temperature. After flushing with dry Ar for 1 hour, the background spectrum was collected. The FTIR spectra were then collected every 50 ○ C from 50 ○ C to 600 ○ C in dry Ar with a ramping rate of 10 ○ C/min. The negative OH peak intensity (3000-3500 cm -1 ) of PNC increased monotonously with temperature (Supplementary Figure 11) (11), indicating the continuous loss of hydroxyl groups on the PNC surface. The negative peak intensity increased rapidly before 400 ○ C and then smoothly after 400 ○ C, which indicates two different temperature zones for the desorption of surface chemisorbed hydroxyl/water (<400 ○ C) and lattice protons (>400 ○ C) respectively.
Temperature programmed desorption (TPD) is rarely used to investigate desorbed water molecules from surface and dehydration process in the bulk material when the temperature is increased (11). It is considered that the desorption of water on the particle surface can occur at temperature around 100~300 ○ C and the water released at higher temperature range (> approximate 400 ○ C) is from the dehydration process. Therefore, this PNC material is examined by TPD method. The powder was firstly hydrated at 600 ○ C for 2 h in ~10% H2O/air) and then cooled down to room temperature. The PNC powder was then reheated in dry air and mass spectroscopy was used to monitor the real-time water signal. As shown in Supplementary Figure   12, two samples were examined to reproduce the result. As expected, the residual water started to desorb at low temperature (<300 ○ C) and when the temperature reached the range of 500 to 700 ○ C a weak but discernable broad peak appeared which could be attributed to the water formation from the dehydrated proton defects in the bulk.

Supplementary Note 5: Chemical stability in hydrogen permeation experiment and reducing condition.
The chemical stability of PNC oxide after hydrogen permeation experiment was examined (Supplementary Figure 13a). The XRD pattern indicated no any impurities after pellet sample exposing to 3% H2 at 500 ○ C. In addition, the chemical stability under more reducing condition is evaluated at different hydrogen concentration (3%, 20%, 50%, and 100%) and temperature (500~700 ○ C) (Supplementary Figure 13b). As can be seen, PNC is stable against diluted 3% H2 when the temperature is increased up to 700 ○ C; however, when hydrogen centration is increased to 20%, PNC can decompose into oxides at 700 ○ C. The decomposition also occurs in the higher H2 concentration. Therefore, there is some limitation on atmosphere when this material is applied for hydrogen permeation membrane.

Supplementary Note 6: Electrochemical impedance spectra at reduced temperatures (400~600 ○ C).
The impedance spectra in fuel cell mode under open circuit conditions are shown in Supplementary Figure 14a. When the operating temperature was decreased from 600 ○ C to 500 ○ C, both ohmic and electrode polarization resistances increase correspondingly. The activation energy for electrolyte resistance was obviously smaller than that of interfacial electrode polarization. In electrolysis mode, when the operating temperature of the cell was decreased to lower-temperature range, the electrode polarization resistance was 0.35 Ω cm 2 at 450 ○ C and 0.6 Ω cm 2 at 400 ○ C, respectively (Supplementary Figure 14b). The total cell resistance at 400 ○ C was even smaller than recent reported BCZYYb-based electrolysis cell (1.32 Ω cm 2 ) at higher temperature of 500 ○ C with 3D self-architectured PBSCF as steam electrode.

Supplementary Note 7: Hydrogen production in different conditions.
Supplementary Figure 15 shows the dependence of electrolysis performance on oxygen partial pressure at oxygen electrode side and hydrogen concentration at hydrogen electrode side at 500 ○ C. When the humid air was switched to oxygen, the cell showed higher current density at the same electrolysis voltage. The results are consistent with the finding in the study of symmetric cell, whereas higher oxygen partial pressure can improve electrode polarization resistance (Supplementary Figure 16). In contrast, it exhibited much poor performance in argon which is opposite to the expectation that low partial pressure can promote water splitting reaction by removal of product. At hydrogen electrode side, the less concentrated hydrogen gas enhanced the hydrogen production, and in pure argon the cell showed the highest current density.

Supplementary Note 8: Long-term stability of material structure and electrode activity.
Supplementary Figure 17 shows the impedance spectra for the cell before and after the long-term stability testing shown in Figure 3E. The high frequency ohmic resistance and electrode polarization resistance are both decreased after the testing, e.g., from 0.46 Ω cm 2 to 0.445 Ω cm 2 . The result demonstrates the material and interface stability over electrolysis reaction. The direct evidences were obtained from the chemical stability of PNC electrode and BCZYYb electrolyte and activity stability of symmetric cell (Supplementary Figure 18) and 500 ○ C to observe the degradation over time (Supplementary Figure 19). At the beginning, the electrolysis current density decreased slightly but then stabilized over the next 480 hours.
The operation at lower temperature of 500 ○ C is more favorable for stable electrolysis because the materials and interfaces are more stable.

Supplementary Note 9: Faradaic efficiency in BCZYYb4411-based PCEC at different operating conditions.
The Faradaic efficiencies at different temperatures (500~600 ○ C) and steam concentrations were measured by on-line GC to monitor the change of hydrogen concentration in the gas flow of hydrogen electrode. It is evident that the efficiency was affected by the operating temperature and steam concentration. As operating temperature was decreased, at the same concentration of 15% the Faradaic efficiency was improved slightly at the fixed electrolysis voltage. It is attributed to the increased proton transfer number at lower temperature while the hole conductivity tends to be eliminated and the reduction of cerium ions are also diminished. In addition, when the steam concentration was increased, the efficiency can be increased. The higher steam concentration can increase the hydration of electrolyte and decrease the material oxidation of creating more holes.  Figure 21 shows the cross-sectional view of the cell with nanofiber structured electrode. As can be seen, the single layer of PNC mesh was adhered strongly to electrolyte thin film by PNC ink. The use of PNC ink can improve the adhesion to enhance the interfacial polarization while the high porosity is not affected. The open space between the mesh bundles can allow the direct gas diffusion without any obstruction which significantly benefits the steam/oxygen transport for reactions at entire electrode surface.

Supplementary Note 11: Performance comparisons.
To better show the comparison of water electrolysis performances, the performances of different material systems (electrolyte and steam electrode) were summarized in Supplementary Supplementary Note 12: Self-sustainable reversible operation between electrolysis mode and fuel cell mode.
Supplementary Figure 22 shows the voltage response of the reversible cell when it was operated between electrolysis (-1.2 A cm -2 ) and fuel cell mode (0.4 A cm -2 ) transiently at 550 ○ C.
The performance in electrolysis mode was stable after each fuel cell cycle, indicating nearly constant hydrogen production to supply fuel for electricity generation. In fuel cell mode, the cell also showed decreased discharging voltage similar with the case at 500 ○ C due to fuel deficit caused by limitation of experiment apparatus for the cell to fully react with generated hydrogen.

Supplementary Note 13: Microscopy characterization of the PCEC after test.
Supplementary Figure 23 shows the microstructure view of the cell after a series of electrochemical testing by SEM and FIB/TEM. Firstly, the cross-sectional image shows the electrode-supported cell with respective three layers. The thicknesses of PNC electrode and BCZYYb4411 electrolyte are about 50 μm and 15 μm, respectively. Both the cathode and anode frame bond well to the electrolyte without any cracks at the interfaces after the test in high water vapor, indicating good mechanical bonding and material stability in this condition, which is consistent with the stable electrolysis current densities under different applied voltages. The PNC electrode particles did not show obvious growth or agglomeration after the long-term operation.
The FIB technique was used to cut a very thin piece of the three-layer sample for element mapping, and the result clearly shows the uniform distribution and no signal of element diffusion, which indicates there is no chemical reaction or migration during the long-time testing process and the electrode/electrolyte interface is robust against high steam conditions and oxygen evolution reaction at this electrode side.