Jianjun
Ma§
,
Cairong
Jiang§
,
Paul A.
Connor
,
Mark
Cassidy
and
John T. S.
Irvine
*
School of Chemistry, University of St Andrews, The Purdie Building, St Andrews, Fife, Scotland, UK KY16 9ST. E-mail: jtsi@st-andrews.ac.uk; Fax: +44 (0) 133 4463808; Tel: +44 (0) 133 4463817
First published on 18th August 2015
Solid oxide fuel cells (SOFCs) afford an opportunity for the direct electrochemical conversion of biogas with high efficiency; however, direct utilisation of biogas in nickel-based SOFCs is a challenge as it is subject to carbon deposition. A biogas composition representative of a real operating system of 36% CH4, 36% CO2, 20% H2O, 4% H2 and 4% CO used here was derived from an anode recirculation method. A BaZr0.1Ce0.7Y0.1Yb0.1O3−δ (BCZYYb) infiltrated Ni-YSZ anode was investigated for biogas conversion. The infiltration of BCZYYb significantly promoted the electrochemical reactions and the cells exhibited high power output at the operational temperatures of 850, 800 and 750 °C. At 800 °C, supplied with a 20 ml min−1 biogas, the cell with a BCZYYb-Ni-YSZ anode, generated 1.69 A cm−2 at 0.8 V with an optimal amount of 0.6 wt% BCZYYb, whereas only 0.65 A cm−2 was produced with a non-infiltrated Ni-YSZ in the same conditions. At 750 °C, a maximum power density of 1.43 W cm−2 was achieved on a cell with a BCZYYb-Ni-YSZ anode, a 3 μm dense YSZ film electrolyte, a Gd0.1Ce0.9O2 (GDC) buffer layer and a La0.6Sr0.4Co0.2Fe0.8O3–Gd0.1Ce0.9O2 (LSCF-GDC) composite cathode. The cell remained stable, while operating at 0.8 V for 50 hours with a current density of 1.25 A cm−2. A well-designed cell structure and selected components made it possible to obtain excellent performance at good fuel utilisation. The analysis of gases in open-circuit conditions or under various current loads suggested that the prevalent reaction was reforming of methane without coking. This study demonstrates that the BCZYYb-Ni-YSZ is a promising electrode for carbon-containing fuel.
Solid oxide fuel cells (SOFCs) are a new technology for converting the chemical energy of hydrocarbon or biogas into electricity with high efficiency and fuel flexibility.6 SOFCs can operate within a wide temperature range from low temperatures (400–600 °C) to high temperatures (over 1000 °C) by selecting suitable electrolyte and electrode materials. Although operation of solid oxide fuel cells at low temperature is desirable in terms of cell life and cost, intermediate or even high temperature operation can be beneficial to the internal reforming reactions of methane or other hydrocarbon fuels.5,7–10 In addition, the direct heat exchange between the endothermic internal reforming reactions and the exothermic electrochemical reactions in the fuel cells increases the total system efficiency.11,12
In general, steam reforming and dry reforming reactions are considered as the main reactions in a biogas-fuelled cell.
CH4 + H2O → CO + 3H2 | (1) |
CH4 + CO2 → 2CO + 2H2 | (2) |
There are, however, other reaction paths, such as the water-gas shift reaction.
CO + H2O ↔ CO2 + H2 | (3) |
Depending on the reaction conditions, the equilibrium for the water-gas shift can be pushed in either direction. In fuel cell systems, the dominant reactions will be dependent on the gas concentration, temperatures and system design. The task for researchers is to find good electrode materials that can operate in biogas fuel and generate good power output with good durability.
Nickel is a commonly used catalyst13 due to its excellent catalytic activity for direct oxidation or reforming reactions of hydrocarbon fuel. Therefore it is still considered to be a good catalyst for biogas conversion despite of extensive research on perovskite, La0.75Sr0.25Cr0.5Mn0.5O3 (ref. 14) and Sr2Mg1−xMnxMoO6−δ,15 Cu-based electrode (Cu-YSZ, Cu-Ceria),13,16 precious metal (Rh, Ru, Ir, Pd, Pt)17–20 or precious metal impregnated supports.16,21,22 However, when a nickel electrode is used in a biogas of 60% CH4 and 40% CO2, carbon coking occurs resulting in degradation of the cell performance, or even deactivation of the nickel electrode due to a build-up of carbon on the nickel surface.23–25 In order to avoid carbon deposition an additional reforming agent is therefore required. This could be air, steam or anode recycling containing CO2 and steam. It was reported that a high ratio of steam/methane (i.e., 2)26 or carbon dioxide/methane (i.e., 1.5)10,27 was needed for steam reforming and dry reforming, respectively. This dilutes the fuel concentration and subsequently reduces the fuel utilisation and electrical efficiency. Since SOFCs with anode gas recirculation have higher electrical and thermal efficiencies than non-recycling SOFCs,28,29 it is an effective strategy to combine dry reforming and steam reforming, because carbon dioxide and steam are available from partial anode exhaust gas recycling.30 In comparison with external steam generation, anode exhaust gas recycling provides an internal steam circuit, which eliminates the need for a waste heat recovery steam generator such as the boiler, thus reducing the capital cost of the system. At the same time, the steam content of the exhaust gas is reduced and this elevates efficiency of the total system, by increasing the sensible thermal energy of the exhaust gas.
Viana showed that good reverse water-gas shift activity was often associated with proton conducting oxide perovskites giving an early indication that these might be beneficial in anodes for hydrocarbon fuelled SOFC.31 Proton conductors, BaZr0.9Yb0.1O3−δ,32 BaZr0.1Ce0.7Y0.2O3−δ33 and Ni–BaZr0.1Ce0.7Y0.1Yb0.1O3−δ34 were also reported to have excellent resistance to sulphur poisoning. Cells with those proton conductors have been running in hydrocarbon fuels without coking and have been operating in high sulphur hydrogen without failure. A NiO-4% Zn, doped with BaZr0.8Y0.2O3 was used the anode for the CH4/CO2 = 1:
1 biogas with 3% H2O and the cell displayed good durability even though carbon formation was thermodynamically favoured using this CH4/CO2 stoichiometric ratio, shown in the C–H–O equilibrium diagram.35 The application of BaZr0.1Ce0.7Y0.1Yb0.1O3−δ in biogas fuels has not yet been reported.
Apart from the carbon deposition, another important parameter to be evaluated in the use of biogas in SOFC, is the power output. Some researchers have performed some experiments to enhance the anode performance. The cell of Ni-BZYZn/BZYZn/Pt developed by Luisetto et al.35 generated a maximum power density of 20 mW cm−2 at 750 °C, which was limited by a large ohmic resistance, and/or possibly by the poor catalytic activity of the electrode. Wang et al. reported a Ni-based cell generated 424 mW cm−2 at 800 °C in a simulated biogas of 52.2% CH4, 46.3% CO2 and 1.5% N2, but carbon deposition caused quick degradation of cell performance.36 This was improved by introduction of Al2O3, however, some carbon was still observed by SEM-EDX test after operation at 750 °C for over 120 hours.
At the operational temperatures of the solid oxide fuel cells, the cathode polarisation resistance is a major loss as well.37–39 This can be improved by choosing materials with high catalytic activity or by tailoring the microstructure.40,41 (La, Sr)(Co, Fe)O3 (LSCF), a mixed ion-electron conductor, was reported to have good electronic and ion conductivity and was suitable for intermediate temperature ranges (700 to 900 °C).42 Especially if low temperatures are desirable for the application.
In this study, we aim to develop biogas fuelled cells that they are coking free and have good durability combined with good fuel utilisation. Here a simple route will be developed to prepare a modified Ni-YSZ anode with BaZr0.1Ce0.7Y0.1Yb0.1O3−δ (BCZYYb) infiltration. Gas components will be detected in open circuit conditions and under current load to explore possible reaction mechanism.
A biogas of 63% CH4 and 37% CO2, which has a typical amount of methane for different feedstocks, was selected as an original biogas.2 This biogas is prone to form carbon at the operational temperatures (e.g. 750–850 °C) of fuel cells (no. 1 biogas, in Fig. 1). We anticipate that a practical small system will utilise anode recirculation to minimise carbon formation and best utilise thermal energy and this would mean a modified biogas composition needs to be considered in solid oxide fuel cells. Starting from the original biogas composition (no. 1) methane will be converted into CO, H2 according to reforming reactions and then oxidised. A typical overall fuel conversion rate of 80% was chosen and 25% of the anode exhaust could be recycled. An iterative method was used to calculate a steady-state composition. It was assumed that the gases were at thermodynamic equilibrium and the gas composition was calculated with HSC chemistry software. A single-pass fuel utilisation was 73.3%. 25% of the anode exhaust gas was added into the original gas. Then the cycle described above was repeated until a steady-state composition of gases was obtained. The overall fuel utilisation was 80%. This slightly higher overall fuel utilisation than a singly-pass fuel utilisation of 73.3% matched the report by Powell et al.44 Fuel utilisation will be discussed in Section 3.3. The final steady-state biogas is defined as a recirculated biogas and its composition was 36% CH4, 36% CO2, 20% H2O, 4% H2 and 4% CO. The details of the calculation were shown elsewhere.45 This recirculated biogas (no. 2, Fig. 1) is in the safe region and should not result in carbon deposition above 720 °C. In the following, we will be using the recirculated biogas of 36% CH4, 36% CO2, 20% H2O, 4% H2 and 4% CO to do further research.
After the nickel oxide had been completely reduced to nickel, the anode gas was connected to a bubbler to humidify the hydrogen. It was then switched to a steam generator to gradually increase the steam content, until 20% H2O was obtained. The water content during testing was monitored by a dewpoint sensor (Vaisala). A biogas with desired compositions was switched on. The gases were mixed to the correct ratio using mass flow controllers (Bronkhorst), with a total flow rate of a 300 ml min−1, consisting of 285 ml min−1 of the biogas composition, balanced with argon. The cells were characterised in dry hydrogen, wet hydrogen, steam hydrogen and the recirculated biogas mentioned above. In the biogas, the cell was kept in open-circuit conditions for 2 hours to record the anode exhaust gas. The electrochemical polarisation curves and the EIS of the cells were tested from 850 °C to 750 °C, at a 50 °C interval, using various flow rates. To simplify the experiments, the optimisation of infiltration BCZYYb was carried out on the cell with a LSM-YSZ cathode rather than a LSCF-GDC cathode, to avoid the extra work of the screen-printing of the GDC interlayer.
There should be no change in the carbon content in the original gas and in the anode exhaust gas as long as there is no carbon deposition.
Carbon content (CH4 + CO + CO2)input = Carbon content (CH4 + CO + CO2)output |
The flow rate of argon in input is equal to the flow rate of argon from output.
f(Ar(input)) = f(Ar(output)) |
The content of CH4, CO and CO2 was recalculated by subtracting hydrogen and argon from the GC results.
The flow rate of hydrogen from output was calculated using the following equation:
f(H2 out) = 2 × (f(CH4 in) − f(CH4 out)) − f(H2 in) |
Conversion rate of CH4 was calculated in (f(CH4 in) − f(CH4 out))/f(CH4 in) × 100%.
Fig. 3b shows examples of the performance of the cells tested in the recirculated biogas at 800 °C with infiltrated BCZYYb. For the non-infiltrated Ni-YSZ cell, the current density was 0.65 A cm−2 at 0.8 V. After the infiltration, the cell performance increased dramatically and the current density reached 1.69 A cm−2 when 0.6 wt% BCZYYb was infiltrated into the electrode. Any further increase in the infiltration amount resulted in a drop in the cell performance, but these were still better than the performance of non-infiltrated Ni-YSZ.
The impedance spectra of the cells are shown in Fig. 3c. The polarisation resistance Rp of the cell with the non-infiltrated Ni-YSZ anode was 0.275 Ω cm2. After infiltration with 0.3 wt% BCZYYb, the Rp decreased to 0.206 Ω cm2. Further increasing the infiltration amount to 0.6 wt%, the lowest Rp of 0.131 Ω cm2 was obtained. It seemed that this was an optimal amount as additional increase in the infiltration amount would not further improve the polarisation resistance. The Rp of the cell infiltrated with 1.0 wt% and 1.6 wt% BCZYYb was 0.225 and 0.265 Ω cm2. From the simulation of the impedance resistance, electrochemical reactions at both high-frequency (assigned to charge transfer resistance) and low-frequency processes (assigned to diffusion resistance) were promoted.50 In the presence of an optimal amount of BCZYYb, the water-gas shift reaction and electrochemical reaction were enhanced. However, further increase in the infiltration will block the nickel active site for electrochemical reaction. The series resistance of the cells was very small in a range from 0.093 to 0.029 Ω cm2. The cells in this case were taken from the same batch of the experiments. The thickness of the YSZ thin film was around 7 (±1) μm. So the theoretical ohmic resistance from YSZ was 0.0179 Ω cm2. The difference between observed and theory was thought to be due to leads and especially contacts. The slightly smaller series resistance of the infiltrated cells might be due to the more conductive electrode, or better current collection.
Current | 10 ml min−1 | 20 ml min−1 | 30 ml min−1 | 40 ml min−1 |
---|---|---|---|---|
2A | 99.9 | 50.0 | 33.3 | 25.0 |
1A | 50.0 | 25.0 | 16.7 | 12.5 |
Following on from this, various flow rates, 10, 20, 30 and 40 ml min−1, were used and the current density–voltage and current density–power density curves of the cell at 750 °C are shown in Fig. 4a. The cathode flow rate was 200 ml min−1 to ensure an adequate supply of gas for the cathode. With a 10 ml min−1 biogas flow, good current density was obtained (up to 1 A cm−2). This was very impressive current density with this low flow rate. The cell suffered serious fuel starvation at a current density above 1 A cm−2. This result matched well with Table 1. When the biogas flow rate increased to 20 ml min−1, the current density of the cell was above 2 A cm−2. Further improvement in the current density was observed with a 30 ml min−1 biogas flow. The cell had enough fuel supply and no severe fuel starvation was recorded. However, further increasing the flow rate to 40 ml min−1 was not found efficient in terms of performance improvement and fuel utilisation. We used a 20 ml min−1 flow rate in order to get reasonable fuel utilisation and good electrochemical power output.
We attributed the high power generation to a well-designed cell structure and selected components. In the presence of BCZYYb, apart from oxygen ions transportation, proton transportation played a role as well. The complex reactions of dry reforming, steam reforming and water-gas shift reactions at the anodes were promoted by the BCZYYb. The cathode response measured in a three-electrode mode of this cell, shown in Fig. S3,† indicated the cell performance could be improved by using other cathode in the future work.
The influence of the cathode air flow rate on the cell performance was investigated and the results are shown in Fig. 4b. Different flow rates, 0, 50, 100, 150, 200 ml min−1 were applied. The air flow rate at the cathode had significant influence on the cell performance, in particular at a high current density. The cell displays oxygen starvation when there was no air flow at the cathode. In this study, a 200 ml min−1 air supply was used to ensure enough air supply at the cathode. At 750 °C, the cell generated a current of 2.2 A cm−2 when the voltage was 0.65 V. At a fuel cell's operational voltage (generally a hydrogen fuel cell operates 0.7 V), a current of 1.67 A cm−2 was obtained. It showed better performance at 850 °C (Fig. S4†), a current of 2.2 A cm−2 was achieved at 0.8 V. To our best knowledge, these high current densities with biogas fuel were not previously reported in the literature. The maximum power densities in biogas were comparable to those in pure hydrogen, wet hydrogen and steam hydrogen (Fig. S4†). The main difference was seen in the OCV values of each gas (Fig. S5, S6 and S7†). The OCV of the cells will be discussed in Section 3.4.1. The maximum power density of the cell with a 20 ml min−1 biogas at 750 °C was 1.43 W cm−2, as shown in Fig. 4a. From Fig. 4b, in some cases we cannot identify the maximum power densities, because of the maximum current limitation of the current source (2.5 (A)). From a rough linear extrapolation, the maximum power density could be ∼1.9 W cm−2 at 750 °C supplied with 200 ml min−1 air to the cathode.
It seemed that the cells were stable over a wide temperature range and at different current densities (0.1 A cm−2, 0.5 A cm−2 and 1 A cm−2, Fig. S8†). The durability of the cell at 750 °C was examined as shown in Fig. 4c. The current density was 1.25 A cm−2 at 0.8 V. The cell was stable for 50 hours with slight fluctuation during the day being possibly due to changes in room temperature and hence actual furnace temperature. The current–voltage curves were collected at a flow rate of 20 ml min−1 biogas. This appeared to be the optimal flow rate, since any further increase, produced no significant improvement in the cell performance (Fig. 4a), but wasted fuel. From Table 1, we can see that the fuel utilisation was 31.2% if the current was 1.25 A (the active area was 1 cm2). Of course, high fuel utilisation can be obtained if the cell is operating at higher current density. An image of this cell after this test inserted in Fig. 4c shows no any obvious carbon deposition was found after operation for 50 hours at 750 °C.
The open-circuit voltage of the cell operating in wet hydrogen at 850 °C was 1.13 V (Fig. S5†), which was close to theoretical value of 1.15 V, showing that the YSZ membrane was gas tight. The OCV changed to 1.06 V when the pure hydrogen was humidified (Fig. S6†). When the gas was changed into 20% steam in hydrogen, the OCV decreased to 0.97 V (Fig. S7†), which was close to the theoretical value of 0.99 V. The open-circuit voltages in the biogas at 850, 800 and 750 °C were 1.00 V, 1.01 V and 1.03 V, respectively (Fig. 5b, pink line), which matched well with the EMFs in theory in Fig. 5b (blue line). Along with the theoretical OCV values in the biogas at different temperatures, the theoretical OCV values using H2 or CH4 as fuel were plotted against temperature. The calculation was done, assuming the H2 or CH4 to be directly oxidised. If the dominant reaction was the oxidation of CH4, there would be no significant change of OCV with temperature (Fig. 5b, red line). The experimental open-circuit voltage increased with the decrease in testing temperature, showing that hydrogen was the primary reactant, since the OCV of a hydrogen-fuelled cell increases with the decreasing temperature (Fig. 5b, black line).
H2 + 2O2− → H2O + 4e− | (4) |
Similarly, carbon dioxide increased and carbon monoxide decreased with any increase in the current level. This indicated that carbon monoxide was being oxidised into carbon dioxide by a reaction with H2O producing CO2 and H2 (eqn (3)), with the hydrogen being further oxidised into H2O.
The anode exhaust gas composition also showed that gas almost reached equilibrium in most cases, which suggested excellent water-gas shift domination. The only deviation from equilibrium was observed at high current (i.e. 1 A cm−2). At 1 A cm−2, 3.4 ml min−1 CO2 and 4.2 ml min−1 H2O produced, which were the theoretical products for this current (the active cathode area is 1 cm2). With an increase in the current, there was more H2O produced than CO2. This indicated that as more oxygen was driven to oxidise hydrogen, the oxidation of hydrogen exceeded the rate of catalytic water-gas shift. Therefore the water-gas shift deviated from equilibrium. The gas utilisation obtained Fig. 6a was 24.3%, which was close to the theoretical value list in Table 1.
The CH4 content remained at the same level, during the test period, slightly changed with cell configurations and components and current load (Fig. 6b). This indicated that CH4 had mostly reformed in open-circuit conditions, and that direct oxidation of CH4 was not a prevalent reaction. This was in consistent with the OCV results in Fig. 5b. At current load, more products of CO2 and H2O promoted the dry reforming and steam reforming reactions, therefore less CH4 was found. It had been shown in the literature, that the conversion rate of methane increased with the S/C ratio, and that with a S/C ratio of 0.5 the methane conversion was 32%.51 The result in this research, however, showed only 1 ml min−1 of methane in all products, which would indicate that most of the methane had been reformed into hydrogen and carbon monoxide. It should, nevertheless, be noticed that the S/C ratio was only 0.55, and this value was much lower than the required steam amount in a single steam reforming system. This suggested that the participation of carbon dioxide in the reforming reaction was very effective.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ta06421j |
‡ The research data supporting this publication can be accessed at http://dx.doi.org/10.17630/408942b8-b397-41ef-a932-ffc3bc6ced49 |
§ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2015 |