An electrolyte-supported button cell was used. The disk-type electrolyte was composed of YSZ (8 mol% Y2O3-stabilized ZrO2). The diameter and thickness of the cell were 20 mm and 300 mm, respectively. To manufacture the anode, a mixture of NiO/YSZ powder, ethyl cellulose, α-terpineol, a dispersant and a plasticizer was prepared and coated on one side of the disk (Ni:YSZ = 50:50 vol%). The disk was dried at 90 oC for 12 h and sintered at a temperature of 1300 oC. The fabricated cell was put into a quartz reactor. The temperature near the cell was measured by K-type thermocouple. The cell was heated in the reactor to 900 oC by an electric furnace. CH4 and Ar were supplied to the cell at 10 and 90 ml/min, respectively. The mixture gas impinged on the cell. During the experiment, H2 concentration in off-gas from the reactor was measured with GC. Before the cell was pulled out from the quartz reactor for ex-situ observation, the gas line was switched from CH4/Ar to Ar, and the reactor was rapidly cooled for reaction quench. Then, the carbon deposition on the cell was performed by SEM-EDX and off-gas analysis. The carbon was supposed to form on Ni surface. Thus, the one-dimensional stagnation flow model in which CH4/Ar flow impinged on the Ni plate like experimental setup was used in the calculation. Catalytic heterogeneous reactions were modeled by elemental step based reaction mechanisms developed by Maier et al. (Top. Catal., 54 (2011) 845-858). The 42 surface reactions involving 12 surface-adsorbed species were used. The transient surface coverage of carbon was calculated, and the carbon formation pathways were computationally studied.
As a result, the amount of carbon was increased as time proceeded in the experiment. The amount of deposited carbon measured with the balance corresponded to that estimated from H2 concentration, assuming that the carbon was formed through CH4 decomposition (CH4 → C + 2 H2). It was shown that the mass was well balanced in the experimental system. SEM-EDX analysis showed that carbon was deposited on not YSZ but Ni surface, indicating Ni acted as catalyst for carbon deposition as reported by previous studies. 20 min after CH4/Ar was supplied, the Ni/YSZ layer started to expand. A swelling ratio based on the initial cell thickness became over 100 % after 60 min. SEM observation showed that the porous anode structure was destroyed by the carbon growth. In fact, the recovery of the anode structure was difficult once the anode was expanded even if the carbon was removed by reforming. Calculation showed that the surface coverage of carbon became over 0.9 within 10 sec when CH4/Ar was supplied to Ni substrate at the same condition as the experiment. Carbon was formed on the surface subsequent to CH4 adsorption on Ni surface and dissociation reactions. CH4 adsorption reaction on Ni surface showed a high sensitivity against the carbon formation. Calculation also showed that the carbon was not formed when S/C (steam carbon ratio) was less than 1.0 where the carbon was formed in equilibrium condition, indicating kinetics of surface reaction played an important role in the carbon deposition than equilibrium.