Mechanism for Major Improvement in SOFC Electrolyte Conductivity When Using Lithium Compounds as Anode

: Recent studies indicate that the electrolyte ionic conductivity in an SOFC can be considerably increased by using lithium compounds as the electrode. We found that the ionic conductivity of Gd 0.1 Ce 0.9 O 1.95 electrolyte in the Ni 0.8 Co 0.15 Al 0.05 LiO 2 anode cell was 10.1 mS/cm only at 550 ° C without H 2 , but it increased to 44.6 mS/cm after feeding H 2 to the anode. It was found that LiOH/Li 2 CO 3 moved into the GDC electrolyte from the NCAL anode and formed a three-phase composite electrolyte. A space charge region with a high oxygen vacancy concentration is formed around the interface of LiOH/Li 2 CO 3 and GDC, which increases the ionic conductivity.


INTRODUCTION
Solid oxide fuel cells (SOFCs) are all-solid-state energy conversion devices that directly convert the chemical energy of fuel and oxidants into electrical energy. 1 However, the high operating temperature (750−1000°C) weakens the long-term stability of SOFC and leads to higher costs limiting its commercialization. In recent years, a new type of fuel cell with lithium oxide as electrode and oxide or carbonate and oxide composite as electrolyte has resulted in a much lower operating temperature (400−600°C) but still high electrochemical performance. 2−7 SOFCs with nanocomposite electrolyte composed of doped ceria and salts (LiCl, SrCl 2 , NaOH, MCO 3 ; M = Li, Na, K, Ca, Ba, Sr, etc.) have been reported. 8,9 The ionic conductivity of the composite electrolyte was between 0.01 and 1 S/cm at 400−700°C with a maximum power density of 100−800 mW/cm 2 . 8,9 Lithium-containing oxide electrodes such as LiNiO 2 , LiNiCuO x , LiNi 0.2 Fe 0.65 -Cu 0.15 O 3 , Li 0.3 Ni 0.6 Cu 0.07 Sr 0.03 O 2−δ , and LiNiCuZnFeO x exhibit good electrode catalytic activity as cathode and anode for lowtemperature SOFCs. 10,11 In our previous work, we found that a device with traditional SOFC electrolyte materials (Ba-Ce 0.9 Y 0.1 O 3 (BCY), Gd-doped CeO 2 (GDC)) of 0.5 mm thickness and classical semiconductor materials (La 0.25 Sr 0.75 TiO 3 (LST), SrTiO 3 (STO), etc.) as electrolytes, and NCAL coated foam Ni (foam Ni-NCAL) as symmetrical electrode reached high electrochemical performance at low temperature, or 736 mW/cm 2 at 550°C. 12−15 In a comparative study, the performance of a single-layer fuel cell increased from 357 to 801 mW/cm 2 at 550°C, when Au layers were replaced with NCAL layers. 16 It is generally known that the ionic conductivity of a classical GDC electrolyte is 0.011−0.025 S/ cm at 500−600°C, while with foam Ni-NCAL as electrode the conductivity increased significantly. 13,17 Cells with a structure of foam-Ni-NCAL/STO/foam-Ni-NCAL and foam-Ni-NCAL/LST/foam-Ni-NCAL nearly acted as insulators when air was introduced on both sides of the cell before the fuel cell performance test, but the ionic conductivity increased rapidly to 0.2 S/cm when H 2 was introduced on the anode. 14,15 This improvement may be attributed to formation of a 1−4 nm thin amorphous layer with carbonate on the surface of STO and LST electrolyte particles, but the composition and formation mechanism of the carbonate in the amorphous layer has remained open, which is the main subject of this paper.
The main aim of this paper is to verify that the improvement in ionic conductivity in a fuel cell with lithium-containing oxide as electrode can be related to hydrogen on the anode side. Previous studies indicate that the ionic conductivity of a GDC electrolyte sintered at 1550°C significantly increases when an NCAL anode is used, and therefore, the same electrolyte was used in the present work. 13 Porous Pt and NCAL were used as the anode for an electrolyte-supported fuel cell test configuration, in which porous Pt was used as the cathode. The focus of the paper is to explore and understand the mechanisms that increase the electrolyte conductivity when H 2 is introduced into the anode. This information would be important to develop further the new fuel cell system, e.g., its stability and performance. Figure 1a shows the IV−IP curves of a cell with the Pt and NCAL anode operated at 550°C with air. The open circuit voltage (OCV) of the cell with the NCAL anode is 0.96 V, and the maximum power density is 37 mW/cm 2 , while the OCV of the cell with Pt anode is 0.79 V and the maximum power density is 1.2 mW/cm 2 only. Figure 1b and Figure S1 show that the ohmic resistance (R o ) of the cell with the Pt anode before and after introduction of H 2 is 3.79 and 3.72 Ω cm 2 , respectively. Figure 1d,c indicates that the R o of the cell with the NCAL anode before and after introduction of H 2 is 4.97 and 1.12 Ω cm 2 . Figure S2 shows the equivalent circuit used in this study to which the impedance spectra of the cells with NCAL and the Pt anode in the H 2 −air atmosphere were fitted. The fitting results are shown in Table S1. NCAL is an electronic conductor with a conductivity of 9.8 S/cm, which is much higher than the ionic conductivity of the GDC electrolyte. Therefore, the ohmic resistance of the cells prepared in this study is mainly contributed by the GDC electrolyte. 18 The conductivity of the GDC electrolyte in the cell with Pt as anode is about 13.4 mS/cm at 550°C, which is consistent with the conductivity of the GDC electrolyte reported in the literature. 13,17 The conductivity of the GDC electrolyte in the NCAL anode cell is 10.1 mS/cm with air on both sides of the cell before the fuel cell performance test. This is close to the results when Pt is used as an anode. However, when H 2 is fed to the NCAL anode, the conductivity of the GDC electrolyte increased to 44.6 mS/cm, which is 4.4 times higher than without H 2 (as indicated in Figure 1e). As the conductivity of the electrolyte is an intrinsic property of the material, the increase in the conductivity of the GDC electrolyte should be related to the change of NCAL in H 2 . Figure 2 shows the back scattered electron diffraction scanning electron microscope (BSED SEM) images of the cross sections of the cell with the NCAL anode. The NCAL anode has a porous structure with spherical particles of an average size of ca. 12 μm. The thickness of the anode is ca. 50 μm. The GDC electrolyte is dense, and though it contains some closed holes, the internal fuel leakage should be negligible. The color of the NCAL particles change to silverwhite surrounded by a gray phase after H 2 is introduced and   the electrochemical performance is tested (Figure 2b). The gray phases in the GDC electrolyte are distributed around the grain boundaries of the GDC grains ( Figure 2c); sometimes, it almost covers the whole GDC grain. SEM images with larger magnification of the NCAl/GDC interface and GDC electrolyte before and after the electrochemical performance test are shown in Figure S3. Figure 2d,e shows the energy disperse Xray spectroscopy (EDS) of the positions in the GDC electrolyte with and without the deep gray phase after performance testing, where no dark gray phase represents the GDC electrolyte before the performance test. Table S2 gives its composition. In addition to Ce, Gd, and O, there is also carbon present. The carbon content in the dark gray phase is much higher than that in the gray phase of the GDC electrolyte. To determine the specific chemical composition of carbon, the C 1s and O 1s X-ray photoelectron spectroscopy (XPS) spectra of the GDC electrolyte pellet of NCAL anode cell before and after the electrochemical performance test are shown in Figure 3. The XPS spectra of other elements (Li 1s, Gd 4d, Ce 3d) are shown in Figure S4. In the C 1s XPS spectrum, the peak at 284.8 eV represents hydrocarbons, and the peak at 289.3 eV represents CO 3 2− . 19 It should be noted that the peak at 284.48 eV in the C 1s XPS spectrum should represent the pollution carbon signal, which is mainly caused by dust adsorbed on the surface of the sample when it is exposed to the air. In addition, the existence of the hydrocarbon bond may indicate the presence of HCO 3 − , but this would need more investigation which was outside the scope of this work. In the O 1s XPS spectrum, the peak at 528.8 eV represents the lattice oxygen on the surface of the GDC grains, and the peak at 531.8 eV represents the oxygen with high binding energy such as CO 3 2− , OH groups, oxygen in oxygen vacancies, or a water molecule adsorbed on the surface of material. 19 Actually, this peak significantly increased after the cell performance test.

RESULTS AND DISCUSSION
To further investigate the composition of the dark gray phase entering GDC, we characterized the GDC electrolyte pellet in powder form after the performance with inductively coupled plasma optical emission spectrometry (ICP-OES). The analysis results are shown in Table S3, which shows that 3.2% lithium appears in the GDC electrolyte after the performance test. On the basis of the EDS and XPS results, it seems very likely that the dark gray phase entering the GDC electrolyte contains lithium carbonate. The Li 2 CO 3 enters the electrolyte from NCAL anode through the anode/electrolyte interface. There are several possible sources of carbon. First, NCAL raw material can react with carbon dioxide in the air to produce some lithium carbonate during long-term exposure. 15,20 Another source is terpineol, a binder and solvent/ dispersant used in the electrode preparation process, which decomposes into carbon dioxide during the anode sintering process and reacts with NCAL to produce lithium carbonate. NCAL may also react with CO 2 from air during electrode sintering. The amount of Li 2 CO 3 produced in this way is, however, very small as the CO 2 content in air is very low.
Another interesting question is the following: How does the lithium carbonate containing dark gray phases get into the electrolyte? For this purpose, we reduced NCAL raw powder and NCAL mixed with terpineol binder in dry H 2 and humidified H 2 with different water vapor content (H 2 + 12% H 2 O) for 4 h at 550°C and finally cooled them in reducing atmosphere. Then, the reduced NCAL powder was characterized by XRD, FTIR, and XPS. XRD patterns shown in Figure S5 indicate that NCAL has been reduced to Ni, LiOH, and a small amount of oxide peaks that may contain Co and Al. No peak for lithium carbonate was found, which may be because Li 2 CO 3 exists in an amorphous state. 21 Figure 3c,d shows the C 1s and O 1s XPS spectra of the raw NCAL powder, the reduced NCAL powder, and the NCAL powder with terpineol in humidified H 2 (H 2 + 12% H 2 O), respectively. In the C 1s XPS spectra, the peak of the binding energy center is at 289.5 eV, which represents CO 3 2− . After H 2 reduction, the CO 3 2− peak in NCAL was significantly enhanced but strongest in NCAL with terpineol. According to the XRD results in Figure S5, NCAL will generate LiOH after being reduced by H 2 , and carbon dioxide will be generated after the decomposition of terpineol. These two products easily react to generate lithium carbonate according to the chemical reaction 2LiOH + CO 2 → Li 2 CO 3 + H 2 O. In addition, LiOH will react with carbon dioxide in the air to form Li 2 CO 3 and cover the NCAL surface, which should be the reason why there is a large amount of CO 3 2− on the surface of reduced NCAL without the addition of terpineol. In the O 1s XPS spectra shown in Figure 3d, compared to the raw NCAL powder, the peak of lattice oxygen in the reduced NCAL completely disappears, and the peak of oxygen (oxygen in CO 3 2− , OH groups, oxygen vacancies) with high binding energy of 531.5 eV was significantly enhanced. The intensity of oxygen with a high binding energy peak in reduced NCAL with terpineol was the strongest, which indicates that there are not many oxides on the surface of reduced NCAL particles, but they are covered by amorphous Li 2 CO 3 and LiOH. FTIR results of Li 2 CO 3 powder, reduced NCAL without terpineol, and reduced NCAL with terpineol powders in humidified hydrogen (H 2 + 12% H 2 O) shown in Figure S6 indicated that most of the lithium carbonate is generated on the surface of reduced NCAL with terpineol powder after reduction, which is consistent with the XPS results. These indicate that the binder and solvent/ dispersant containing terpineol are important sources of carbon in the cell.
To summarize, a part of the lithium carbonate was formed on the NCAL surface during the preparation of the NCAL anode due to the mixed terpineol. In the performance test of  the fuel cell, NCAL is reduced to Ni, LiOH, Li 2 CO 3 , and a small amount of oxides containing CoAl in the H 2 atmosphere. Ni is mainly responsible for the conduction of electrons and the catalytic function of the hydrogen oxidation reaction (HOR). The LiOH, Li 2 CO 3 , and CoAl oxides should play an important role in the oxygen ion conduction and the reduction of activation energy of HOR. TG-DSC curves of LiOH·H 2 O and Li 2 CO 3 measured in a N 2 atmosphere shown in Figure  S7a,b indicated that Li 2 CO 3 is in a solid state and LiOH is decomposed into Li 2 O at 550°C. However, H 2 O is continuously generated on the anode side of the fuel cell, which in turn reacts with Li 2 O to form LiOH. Therefore, LiOH should be in the molten state at this temperature. We found that when hydrogen was introduced on both sides of the cell and no fuel cell voltage was formed, LiOH and Li 2 CO 3 complexes also diffused into the electrolyte, which can be observed in Figure S8. It can be concluded that the LiOH and Li 2 CO 3 complexes are diffused from the anode to the GDC electrolyte under the driving force of the chemical potential difference. When H 2 was injected into the NCAL anode, the electrolyte turned into a composite composed of LiOH/ Li 2 CO 3 and GDC, which resulted in the conductivity enhancement. Unlike in a traditional GDC electrolyte, where bulk diffusion and grain boundary diffusion dominate ionic conduction, interfacial conduction plays a leading role in a GDC-LiOH/Li 2 CO 3 composite electrolyte. 17 The associated mechanism is linked to the formation of a space charge layer between the LiOH/Li 2 CO 3 complex and the GDC electrolyte, which can be observed in Figure 4. The space charge region refers to the region where the charge carrier concentration between the two-phase interfaces changes. This concept was first proposed by Wagner to explain the conductive effect of a semiconductor two-phase interface. 22 Later, Jow et al. used it to explain the interface phenomena in an ionic conductor system. 23 In the present paper, due to the difference in chemical potential between the LiOH/Li 2 CO 3 complex and the GDC electrolyte, Li + in the LiOH/Li 2 CO 3 complex diffuses into the two-phase interface, and Li + accumulates near the electrolyte side of the GDC to form a space charge layer. According to the principle of electric neutrality, in order to compensate for the loss of positive charge, a region with high oxygen vacancy concentration will be formed around it. The electron paramagnetic resonance (EPR) result shown in Figure  S9 proved that the oxygen vacancy concentration in GDC containing the LiOH/Li 2 CO 3 complex was higher than that in the pure GDC electrolyte. The continuous phase interface network formed in the region with high oxygen vacancy concentration provides a high-speed ion conduction channel. This should be the main explanation for increased ionic conductivity in the GDC electrolyte.

CONCLUSIONS
In this paper, the mechanisms for improved electrolyte conductivity when using an NCAL anode in an SOFC-type fuel cell were investigated. For this purpose, two NCAL/ GDC/Pt and Pt/GDC/Pt structures were employed in test cells. The NCAL anode cell yielded a maximum power density of 37 mW/cm 2 at 550°C in H 2 , which was 32 times higher than that of the Pt anode cell. The electrolyte ionic conductivity of the cell with a NCAL anode was 4.4 times higher than that with Pt. Through detailed characterization, we were able to identify formation of LiOH and Li 2 CO 3 in the NCAL anode under H 2 atmosphere, which entered into the electrolyte through diffusion driven by the chemical potential difference covering the GDC grains to form a composite electrolyte. We believe that a space charge layer with cation enrichment was formed at the interface of the two phases, with a region with a high oxygen vacancy concentration around it, which may form a channel of high-speed ionic transport in the composite electrolyte. As a result, the ionic conductivity of the electrolyte is significantly improved. This observation is well in line with recent literature, which shows that NCAL as a current collector significantly improves the performance of a singlecomponent fuel cell employing a mixed-oxide and ionic material composition. 12  . Schematic diagram of the major improvement mechanism in ionic conductivity of the GDC electrolyte in a cell with an NCAL anode.