- Email: [email protected]
carbonate slurry
Journal of Power Sources 329 (2016) 567e573
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
Impact of gas products around the anode on the performance of a direct carbon fuel cell using a carbon/carbonate slurry Hirotatsu Watanabe*, Daisuke Umehara, Katsunori Hanamura Department of Mechanical Engineering, School of Engineering, Tokyo Institute of Technology, NE-6, 2-12-1, Ookayama, Meguro-ku, Tokyo, 152-8552, Japan
h i g h l i g h t s An in situ observation around the anode is performed during the discharge of DCFC. Gas products prevents the carbon particles and ions from reaching the anode. The performance is influenced by the gas products around the anode.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 1 June 2016 Received in revised form 26 August 2016 Accepted 29 August 2016 Available online 4 September 2016
This paper investigates the impact of gas products around the anode on cell performance via an in situ observation. In a direct carbon fuel cell used this study, the anode is inserted into the carbon/carbonate slurry. The current-voltage (I-V) curves are measured before and after a long discharge in the constant current discharge mode. An in situ observation shows that the anode is almost completely covered by gas bubbles when the voltage becomes nearly 0 V in the constant current discharge at 40 mA/cm2; this indicates that gas products such as CO2 prevent the carbon particles and ions from reaching the anode. Meanwhile, the long discharge at 20 mA/cm2 is achieved for 30 min, even though the anode is covered by the CO2 bubbles at 15 min. The I-V curves at 1 min after the termination of the long discharge at 20 mA/ cm2 are lower than those prior to the long discharge. The overpotential significantly increases at higher current densities, where mass transport becomes the limiting process. The cell performance is significantly influenced by the gas products around the anode. © 2016 Elsevier B.V. All rights reserved.
Keywords: DCFC In situ observation Gas products Slurry
1. Introduction Highly efficient energy conversion systems for converting solid fuels such as coal and biomass to electricity are required for clean energy production. Recently, the integrated gasification and fuel cell (IGFC) system, which incorporates a solid oxide fuel cell (SOFC), has attracted much attention [1,2]. After the pyrolysis of solid fuels, volatile matter and char are produced; the char is then gasified to produce H2 in the IGFC system. However, a large exergy loss was observed in char gasification at 1100 C [2]. Conversely, direct carbon fuel cells (DCFCs) directly convert the chemical energy in solid carbon into electricity without the need for a gasification process. The DCFC is an important device for minimizing the exergy
* Corresponding author. Department of Mechanical Engineering, School of Engineering, Tokyo Institute of Technology, NE-6, 2-12-1, Ookayama, Meguro-ku, Tokyo, 152-8552, Japan. E-mail address: [email protected] (H. Watanabe). http://dx.doi.org/10.1016/j.jpowsour.2016.08.122 0378-7753/© 2016 Elsevier B.V. All rights reserved.
loss in energy conversion systems that use solid fuels. For example, a DCFC/SOFC combined system, in which volatile matter is used in the SOFC after reforming and the char directly used in the DCFC, has the potential to achieve very high efficiency. Although a wide range of designs and concepts has been tested, the overall efforts to develop DCFC technology have been relatively minor in comparison to the efforts devoted to other major fuel cell technologies [3]. Ordowich et al. forecasted coal and natural gas power generation technologies through 2050 [4]. In their model, IGFC becomes the dominant power generation technologies from 2030 onward, and DCFC technologies would likely become an important technology past 2050. Although there are many barriers to develop a practical DCFC such as lower power density, ash accumulation and continuous supply of carbon fuel [5,6], further investigation is necessary to overcome these issues. An important difference in DCFCs from other fuel cells is that the fuel is solid carbon particles, which possess a disadvantage in terms of reactivity and fluidity. A number of researchers have addressed
568
H. Watanabe et al. / Journal of Power Sources 329 (2016) 567e573
this issue by developing DCFCs that use a carbon/electrolyte slurry, in which carbon particles are dispersed into the electrolyte to form a triple-phase boundary (carbon, electrolyte, and anode) [7e18]. Molten carbonates have often been used as an electrolyte because they offer many advantages such as high ionic conductivity [3]. The anode (R1), cathode (R2), and overall (R3) reactions in this system are expressed as follows: C þ 2CO2 3 / 3CO2 þ 4e
(R1)
O2 þ 2CO2 þ 4e / 2CO2 3
(R2)
C þ O2 / CO2
(R3)
The desired reaction is complete oxidation, which releases four electrons per carbon, as given by R1. The carbon wetting is an important factor in the formation of triple phase boundary [12,19]. Our previous study showed the carbon wetting has a significant impact on the continuous power generation capacity of the DCFC [12]. When the carbon particles are well-wetted by the molten carbonate, the anodic reaction process is described as follows: (1) carbon and carbonate ions are provided to the anode surface, (2) the anode reaction (R1) takes place at a triple-phase boundary, (3) CO2 and electrons are produced. The transport of the carbon and the ion around the anode plays an important role in the power generation of the DCFC. In addition, it has been pointed out that CO2 bubbles, which evolve from the electrochemical oxidation of solid carbon (R1), can potentially inhibit the cell operation and current stability [20]. Our previous study showed that bubbles were produced around the anode of the DCFC during discharge by using the single carbon pellet [13]. Although the impact of gas products around the anode on the cell performance should be addressed, little effort has been made to perform an in situ observation around the anode during discharge. Further investigations are still necessary to study the reaction and transport phenomena around the anode in order to improve the DCFC performance. This study aims to study impact of gas products around the anode on the DCFC performance. The transport phenomena around the anode are studied through in situ observation and currentvoltage (I-V) characteristics.
2.2. DCFC setup Fig. 1 shows a schematic of a DCFC with a molten carbonate electrolyte, which was developed in our laboratory. Although further study of the electrode material is necessary for the high performance DCFC, in this study, the working electrode (WE), counter electrode (CE), and reference electrode (RE) were made from gold sheet which is immune to corrosion in the molten carbonate [21]. The area in contact with the carbon was 1.0 cm2. The RE and CE were in alumina tubes, and the WE was in a porous alumina tube (for separation of carbon particles, as described later). Each gold sheet was spot welded to the gold wire and extended to the other end of the tube to provide a connection to the potentiostat/ galvanostat (HAL-3001, Hokuto Denko) for cell parameter measurements. The DCFC used in this study is similar to the one used in previous studies [11e13]. The part that differs is the quartz window attached to the plug above the anode for the in situ observation. A zoom camera (PowerShot SX50 HS, Canon) was used to record the phenomena around the anode as seen through the quartz windows. A total of 240 g of dry ternary carbonate powder without carbon was placed in an alumina crucible with an inner diameter of 80 mm and a height of 70 mm. In addition, the mixed carbon/carbonate powder was contained in a porous alumina tube with an inner diameter of 15 mm. The average pore diameter of the porous tube
2. Experimental section 2.1. Carbon characterizations Commercially available activated carbon (activated charcoal, Kantokagaku) was used in this study. Table 1 shows an elemental analysis (dry-base) and the ash content of the samples. The size distribution of the activated carbon particles, which was measured by laser diffraction (LMS-2000e, Seishin Enterprise Co. Ltd.), was shown in the supplemental material. By sieving commercially activated carbon, a size distribution ranging from 2 to 60 mm was obtained. The average size of the activated carbon particles was 16.7 mm. The crystalline parameters of the activated carbon were as follows: d002 (interplanar distance), Lc, (the layer dimension perpendicular to the basal plane) and La (the layer dimension parallel to the basal plane) were 0.376 nm, 0.987 nm, and 3.22 nm, respectively.
Table 1 Elemental analysis of the activated chars [wt%, dry]. C
H
N
S
O (diff)
Ash
92.3
0.17
3.04
0.10
4.39
0.8
Fig. 1. Schematic diagram of DCFC with quartz window for the in situ observation.
H. Watanabe et al. / Journal of Power Sources 329 (2016) 567e573
was 120 nm. Therefore, carbon particles larger than this size did not pass through the porous alumina tube. This allowed contact between the molten carbonate electrolyte and WE, CE, and RE. In fact, the porous alumina tube worked as a separator between the carbon/carbonate slurry and the molten carbonate. The DCFC developed in this study was of a cartridge type, and the cell stack was packages by using several anode compartments consisting of porous alumina tubes and carbon/carbonate slurry. In addition, containing the carbon/carbonate slurry in the porous alumina tube was advantageous because the slurry was easily exchanged by using new porous alumina tube with the slurry when ash accumulates with the use of coal and biomass chars. In this study, the activated carbon was used as fuel. The mixed carbon/carbonate powder consisted of 10 g of dry ternary carbonate powder and 0.1 g of activated carbon. The carbon fuel content in the carbonates was 1.0 wt%, and the carbonate molar compositions (Li2CO3/Na2CO3/ K2CO3) were set to 16.9/25/58.1 mol%. The quartz reactor, with an inner diameter of 95 mm and length of 200 mm, was heated using an electric furnace, and Ar was introduced into the reactor compartment during heating. The gases O2 (50 mL/min) and CO2 (100 mL/min) were introduced to both the cathode and reference electrodes. The carbon/carbonate slurry was stirred using bubbling Ar gas with a flow rate of 50 mL/min. The Ar gas was released from two holes in the gas line tube, each with a diameter of 2 mm. When bubbling was not used, the Ar gas was instead introduced from the top of the porous tube at a rate of 100 mL/min. The furnace temperature was set to 1073 K. Details are given in Refs. [11,12].
569
during the long discharge at 20 mA/cm2 for 30 min are 0.14 ml and 25 ml, respectively. The amount of CO2 produced during the measurement of the I-V curves is thus insignificant compared to that produced during the long discharge. This DCFC comprised the WE, CE, RE, and reactor compartments. In this apparatus, the reactor compartment was a space in the quartz reactor, excluding the WE, CE, and RE compartments. Anode off-gases from the WE were able to pass through the porous alumina tube and diffuse to the reactor compartments. Therefore, after reaching steady state, the CO2 and CO concentrations in the anode off-gases from the WE and reactor compartments (XCO2,WE, XCO2,reactor, XCO,WE, XCO,reactor) were measured with a flame ionization detector (FID) equipped with a Porapack-Q column and a methanizer. The outlet gas flow rates (QWE, and Qreactor) were also measured, using a wet gas meter (W-NK-1, Shinagawa Co. Ltd). The CO2 and CO production rates (QCO2,product and QCO,product) were determined by the sum of the differences between CO2 and CO flow rates during discharge in the constant current mode and those at the open circuit voltage (OCV) condition, given as:
QCO2 ;product ¼
X
XCO2 ;i Qi discharge
i
XCO2 ;i Q i OCV
ði ¼ WE; reactor compartmentÞ (2)
QCO;product ¼
X
XCO;i Qi discharge
i
XCO;i Q i OCV
2.3. Measurement of current-voltage curves and anode off-gases The long discharge was performed at a constant current density of either 20 or 40 mA/cm2. The in situ observation and anode offgas measurements were also performed during the long discharge. As shown in Fig. 2, the I-V curves were measured at a scanning rate of 100 mV/s before and after the long discharge for 30 min to study the impact of the gas products during the long discharge. Measurement of I-V curve is completed within approximately 10 s. Based on R1, the CO2 production rate during discharge is calculated as follows:
QCO2 ;product ¼
3IVm 4F
(1)
where F is the Faraday constant (96,450 C mol1), Vm is the molar volume of CO2 and I(A) is the constant current flow. When the current density is 20 mA/cm2, CO2 production rate is 1.55 107 mol/s which corresponds to 0.22 ml/min at 293 K. At 1073 K where the gas is produced on the anode, CO2 production rate is 0.82 ml/min and the theoretical amount of CO2 produced during the measurement of the I-V curves and that produced
Fig. 2. Experimental procedure for I-V curves measurement.
ði ¼ WE; reactor compartmentÞ (3)
Anode off-gases were measured prior to discharge, confirming that the O2 concentration was negligible in the anode and reactor compartments. 3. Results and discussion Fig. 3 shows I-V and the current-power (I-P) curves prior to the long discharge with and without Ar bubbling. In this experiment, the I-V curves with bubbling are similar to those without bubbling, except for around the maximum current density near 100e120 mA/ cm2. This was because the concentration polarization appeared earlier, caused by the bubbling that inhibited contact between the carbon and anode. The maximum power density was reached at approximately 50 mW/cm2. Meanwhile, our previous study showed that Ar bubbling improves the current discharge curves during long discharge [11]. During long discharge, an amount of gas is expected to be produced around the anode, whereas that is insignificant in the measurement of I-V curves with the scanning rate of 100 mV/s. This will be discussed later. Fig. 4 shows the constant current discharge curves (a) and in situ observations around the anode (b) at 40 mA/cm2 without bubbling (Qbub ¼ 0 ml/min). The marks (A), (B), and (C) on the constant current discharge curve (Fig. 4a) refer to the images of in situ observation (Fig. 4b). In Fig. 4a, the voltage at 0 min corresponds to the OCV. The initial voltage drop is mainly caused by activation losses arising as a result of the kinetics at the electrode. Subsequently, the voltage decreases to 0.6 V at 2 min and finally reaches 0 V at 3.5 min. These voltage drops can be explained from the in situ observation around the anode in Fig. 4b. In the observation, the upper half of the anode plate is clearly seen at 0 min ((A) OCV, before discharge). During discharge, electrochemical oxidation (C þ 2CO2 3 / 3CO2 þ 4e (R1)) is progressed on the lower half of the anode embedded into the slurry. Due to bubble formation, the
570
H. Watanabe et al. / Journal of Power Sources 329 (2016) 567e573
(a) I-V
(b) I-P
Fig. 3. Current-voltage (I-V) and current power (I-P) curves of the DCFC with and without bubbling before the long discharge.
(A, OCV)
(B) (C)
(a) Constant current discharge curve
2 min (B)
0 min (A)
Upper half of the anode is seen
1.0 mm
Carbon slurry
The anode boundary becomes cloudy due to bubble formation
3.5 min (C)
1.0 mm
The anode is hidden by bubbles
1.0 mm
(b) In-situ observation around the anode Fig. 4. Current discharge curves and in situ observation around the anode at I ¼ 40 mA/cm2 without bubbling. (a) Constant current discharge curve. (b) In situ observation around the anode.
anode boundary becomes cloudy at 2 min (B) after discharge and then is difficult to observe from the top at 3.5 min (C). In fact, gas bubbles, which are the products of electrochemical oxidation of carbon, cover the anode as the discharge proceeds. The finding in this case is that the voltage becomes 0 V in the constant current discharge mode when the anode is hidden by gas bubbles at 3.5 min (C). This is because the gas bubbles around the anode prevent the carbon particles and ions from reaching the anode. The details are discussed later in this study. The discharge period of 3.5 min is too short to measure the gas products with the FID. The IV curves following the long discharge at 40 mA/cm2 were not measured in this study. Fig. 5 shows the constant current discharge curve (a), the CO2
and CO production rate (b), the in situ observation around the anode (c), and a magnified image of the in situ observation at 15 min (d) at 20 mA/cm2 with bubbling (Qbub ¼ 50 ml/min). The marks (A), (B), and (C) in Fig. 5a refer to the images in Fig. 5c. In Fig. 5a, at the steady state, the voltage reaches approximately 0.75 V, which is lower than that in the I-V curves at 20 mA/cm2, as shown in Fig. 3. This implies that the impact of gas products is not appeared in the I-V curves measured at the scanning rate of 100 mV/s. In Fig. 5b, only CO2 is produced from the discharge. The average total CO2 flow rates at 20 mA/cm2 and OCV condition are 0.62 ml/min and 0.43 ml/min, respectively. The average measured CO2 production rate (0.19 ml/min) almost corresponds to the theoretical rate (¼ 0.22 ml/min), indicating that the carbon is
H. Watanabe et al. / Journal of Power Sources 329 (2016) 567e573
571
(A, OCV) (B)
(C)
(a) Constant current discharge curve
(b) CO2 and CO production rate The anode is hidden by bubbles
15 min (B)
0 min (A)
22 min (C)
Many bubbles form
Upper half of the anode is seen
1.0 mm
Carbon slurry
(d)
1.0 mm
1.0 mm
(c) In-situ observation around the anode during discharge 0.5 mm
(d) Magnified image of (c, 15 min (B)) Fig. 5. Current discharge characteristics at I ¼ 20 mA/cm2 with bubbling. (a) Constant current discharge curve (b) CO2 and CO production rate. (c) In situ observation around the anode during discharge. (d) Magnified image of (c) (15 min (B)).
consumed by the electrochemical oxidation expressed as R1. With bubbling, CO is not detected at both the discharge (20 mA/cm2) and OCV. However, without bubbling, CO is formed at the discharge (20 mA/cm2), whereas CO is not formed at OCV. CO is supposed to be formed through the partial electrochemical oxidation (Cþ 1/2 CO2 3 / 3/2 CO þ e ) which is discussed in our previous study [11,13]. Compared with Fig. 4b, CO2 bubbles are more clearly seen near the anode in Fig. 5c at 15 min and 22 min of discharge because the longer discharge produces larger amount of CO2. The constant current discharge at 20 mA/cm2 is achieved for 30 min even though the anode is almost completely covered by small CO2 bubbles after 15 min of discharge, as shown in Fig. 5c. This indicates that triple phase boundaries are able to form at lower current density even through the anode is covered by bubbles at the macroscopic view. In Fig. 5d, which is the magnified image of the box in Fig. 5c at 15 min (B), bubbles with a diameter of nearly 0.3 mm are seen near the anode. During the discharge at 20 mA/cm2, the CO2 production rate is 0.82 ml/min at 1073 K. If the CO2 bubble size is assumed to be 0.3 mm for all bubbles, the bubble generation rate is approximately 58,000 bubbles/min, which is sufficient to cover the anode after
15 min of discharge. Meanwhile, a small quantity of bubbles is also observed in the image at 0 min (A) because it was taken after the measurement of I-V curves at the scanning rate of 100 mV/s. A very small amount of CO2 is produced during the measurement of the IV curves. Fig. 6 is a schematic illustration depicting the impact of gas products on the DCFC performance during discharge. Prior to the long discharge (A), the carbon and carbonate ions can approach the Au surface without encountering any obstacles. During discharge, the anode reaction (R1) takes place at a triple phase boundary, and CO2 bubbles are produced on the anode. The CO2 bubbles prevent the carbon particles and ions from reaching the Au surface, and they reduce the effective surface area for the reaction. After the long discharge (B), the CO2 bubbles cover a sufficient area on the reaction site to reduce the cell performance. Fig. 7 shows the I-V and I-P curves before and after the long discharge at 20 mA/cm2 for 30 min with bubbling (Qbub ¼ 50 ml/ min). The I-V curves are measured before the long discharge and at 1 min and 10 min after the termination of the long discharge at 20 mA/cm2. The I-V curve at 1 min after termination of the long
572
H. Watanabe et al. / Journal of Power Sources 329 (2016) 567e573
Fig. 6. Schematic illustration depicting the CO2 bubble production around the anode during discharge in the DCFC.
(a) I-V
(b) I-P
Fig. 7. I-V and I-P curves before and after the long discharge for 30 min (Qbub ¼ 50 mL/min, scanning rate ¼ 100 mV/s).
discharge was lower than those prior to the long discharge, indicating that the overpotential significantly increases due to the CO2 bubbles around the anode. The overpotential particularly increases at higher current density, where the mass transport is the limiting process, indicating that CO2 bubbles around the anode prevent the carbon particles and ions from making contact with the anode. At 10 min after the termination, the I-V curve approaches to that before the long discharge. This indicates that the CO2 bubbles around the anode have been released into the atmosphere and that
(a) I-V
carbon particles and ions can move to the anode again. Theoretically, 1.1 mg of activated carbon is consumed through R1 during discharge at 20 mA/cm2 for 30 min. Because the initial amount of carbon is 100 mg, the change on carbon content before and after the long discharge is insignificant. Fig. 8 shows the I-V and I-P curves before and after the long discharge at 20 mA/cm2 without bubbling (Qbub ¼ 0 ml/min). The OCVs without bubbling dramatically diminish to 1.0 V at 1 min after the termination of the long discharge unlike those with bubbling as
(b) I-P
Fig. 8. I-V and I-P curves before and after the long discharge for 30 min (Qbub ¼ 0 mL/min, scanning rate ¼ 100 mV/s).
H. Watanabe et al. / Journal of Power Sources 329 (2016) 567e573
shown in Fig. 7. Our previous study showed that the carbon content around the anode influences the reaction scheme of solid carbon, leading to a change in OCV [11]. The CO2 bubbles have the ability to change the local carbon content around the anode. When bubbling is used, the reduction of OCVs at 1 min after the termination is less significant (Fig. 7) because Ar bubbling is supposed to remove gas bubbles around the anode, and enhance the formation of reaction site during the short period. This also explains our previous result that bubbling improves the current discharge curves during long discharge which produces the amount of gas [11]. It is suggested that the role of bubbling is not only stirring the slurry but also removing gas bubbles around the anode. The I-V curve at 10 min after the termination of the long discharge without bubbling is similar to that with bubbling. This indicates that the CO2 bubbles are released into the atmosphere after sufficient time elapses regardless of the presence of Ar bubbling. As shown in Fig. 7, the stirring by Ar bubbling is still insufficient to restore back the carbon/carbonate slurry characteristics prior to the long discharge. 4. Conclusion In this study, the impact of gas products around the anode on the cell performance was studied. The in situ observation was performed during discharge. In addition, the I-V curves were measured before and after the long discharge. The in situ observation showed that the anode was almost completely covered by gas bubbles when the voltage became nearly 0 V in the constant current discharge mode at 40 mA/cm2. Meanwhile, the long discharge at 20 mA/cm2 was achieved for 30 min, although the anode was covered by the CO2 bubbles at 15 min. This indicated that triple phase boundaries were able to form at the lower current density when the anode was covered by bubbles at the macroscopic view. The I-V curves measured 1 min after termination of the long discharge were lower than the curves prior to the long discharge. In particular, the overpotential significantly increased at the higher current density, where the mass transport became the limiting process. This indicates that the CO2 bubbles prevented the carbon particles and ions from reaching the anode, resulting in lower DCFC performance. As time elapsed after the termination of the long discharge, the I-V curve became closer to that before the long discharge, indicating that the CO2 bubbles were released into the atmosphere and the carbon particles and ions were able to contact the anode again. It was shown that the cell performance was significantly influenced by gas products around the anode of the DCFC. Acknowledgment This study was partly supported by J-POWER.
573
Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2016.08.122. References [1] J.P. Trembly, R.S. Gemmen, D.J. Bayless, The effect of IGFC warm gas cleanup system conditions on the gas-solid partitioning and form of trace species in coal syngas and their interaction wth SOFC anodes, J. Power Sources 163 (2007) 986e996. [2] S. Chen, N. Lior, W. Xiang, Coal gasification integration with solid oxide fuel cell and chemical looping combustion for high-efficiency power generation with inherent CO2 capture, Appl. Energy 146 (2015) 298e312. [3] S. Giddey, S.P.S. Badwal, A. Kulkarni, C. Munnings, A comprehensive review of direct carbon fuel cell technology, Prog. Energy Combust. Sci. 38 (2012) 360e399. [4] C. Ordowich, J. Chase, D. Steele, R. Malhotra, M. Harada, K. Makino, Applying learning curves to modeling future coal and gas power generation technologies, Energy & Fuels 26 (2012) 753e766. [5] N.J. Cherepy, R. Krueger, K.J. Fiet, A.K. Jankowski, J.F. Cooper, Direct conversion of carbon fuels in a molten carbonate fuel cell, J. Electrochem. Soc. 152 (2005) A80eA87. [6] S. Giddey, A. Kulkarni, C. Munnings, S.P.S. Badwal, Composite anodes for improved performance of a direct carbon fuel cell, J. Power Sources 284 (2015) 122e129. [7] D.G. Vutetakis, D.R. Skidmore, H.J. Byker, Electrochemical oxidation of molten carbonate-coal slurries, J. Electrochem. Soc. 134 (1987) 3027e3035. [8] X. Li, Z.H. Zhu, R.D. Marco, A. Dicks, J. Bradley, S. Liu, G.Q. Lu, Factors that determine the performance of carbon fuels in the direct carbon fuel cell, Ind. Eng. Chem. Res. 47 (2008) 9670e9677. [9] J.F. Cooper, J.R. Selman, Analysis of the carbon anode in direct carbon conversion fuel cells, Int. J. Hydrogen Energy 37 (2012) 19319e19328. [10] C.Q. Wang, J. Liu, J. Zeng, J.L. Yin, G.L. Wang, D.X. Cao, Significant improvement of electrooxidation performance of carbon in molten carbonates by the introduction of transition metal oxides, J. Power Sources 233 (2013) 244e251. [11] H. Watanabe, T. Furuyama, K. Okazaki, Enhancing the efficiency of direct carbon fuel cells by bubbling Ar gas in carbon/carbonate slurry, J. Power Sources 273 (2015) 340e350. [12] H. Watanabe, A. Kimura, K. Okazaki, Impact of ternary carbonate composition on the morphology of the carbon/carbonate slurry and continuous power generation by direct carbon fuel cells, Energy Fuels 30 (2016) 1835e1840. [13] H. Watanabe, T. Furuyama, K. Okazaki, Electrochemical oxidation of activated carbon and coal chars in a direct carbon fuel cell using carbon/carbonate slurry stirred by Ar bubbling, J. Chem. Eng. Jpn. 49 (2016) 237e242. [14] C. Li, H. Yi, D. Lee, On-demand supply of slurry fuels to a porous anode of a direct carbon fuel cell: attempts to increase fuel-anode contact and realize longterm operation, J. Power Sources 309 (2016) 99e107. [15] K. Pointon, B. Lakeman, J. Irvine, J. Bradley, S. Jain, The development of a carbon-air semi fuel cell, J. Power Sources 162 (2006) 750e756. [16] L. Deleebeeck, K.K. Hansen, HDCFC performance as a function of anode atmosphere (N2-CO2), J. Electrochem. Soc. 161 (2014) F33eF46. ndez, K.K. Hansen, Hybrid direct carbon [17] L. Deleebeeck, A. Arenillas, K. Mene fuel cell anode setup, Int. J. Hydrogen Energy 40 (2015) 1945e1958. [18] L. Deleebeeck, D. Ippolito, K.K. Hansen, Enhancing hybrid direct carbon fuel cell anode performance using Ag2O, Electrochim. Acta 152 (2015) 222e239. [19] C.C. Chen, T. Maruyama, P.H. Hsieh, J.R. Selman, Wetting behavior of carbon in molten carbonate, J. Electrochem. Soc. 159 (2012) D597eD604. [20] T.M. Gür, Critical review of carbon conversion in “carbon fuel cells”, Chem. Rev. 113 (2013) 6179e6206. [21] M. Wyatt, J.M. Fisher, Control of corrosion in molten carbonate fuel cells, Platin. Met. Rev. 32 (1988) 200e203.
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
Impact of gas products around the anode on the performance of a direct carbon fuel cell using a carbon/carbonate slurry Hirotatsu Watanabe*, Daisuke Umehara, Katsunori Hanamura Department of Mechanical Engineering, School of Engineering, Tokyo Institute of Technology, NE-6, 2-12-1, Ookayama, Meguro-ku, Tokyo, 152-8552, Japan
h i g h l i g h t s An in situ observation around the anode is performed during the discharge of DCFC. Gas products prevents the carbon particles and ions from reaching the anode. The performance is influenced by the gas products around the anode.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 1 June 2016 Received in revised form 26 August 2016 Accepted 29 August 2016 Available online 4 September 2016
This paper investigates the impact of gas products around the anode on cell performance via an in situ observation. In a direct carbon fuel cell used this study, the anode is inserted into the carbon/carbonate slurry. The current-voltage (I-V) curves are measured before and after a long discharge in the constant current discharge mode. An in situ observation shows that the anode is almost completely covered by gas bubbles when the voltage becomes nearly 0 V in the constant current discharge at 40 mA/cm2; this indicates that gas products such as CO2 prevent the carbon particles and ions from reaching the anode. Meanwhile, the long discharge at 20 mA/cm2 is achieved for 30 min, even though the anode is covered by the CO2 bubbles at 15 min. The I-V curves at 1 min after the termination of the long discharge at 20 mA/ cm2 are lower than those prior to the long discharge. The overpotential significantly increases at higher current densities, where mass transport becomes the limiting process. The cell performance is significantly influenced by the gas products around the anode. © 2016 Elsevier B.V. All rights reserved.
Keywords: DCFC In situ observation Gas products Slurry
1. Introduction Highly efficient energy conversion systems for converting solid fuels such as coal and biomass to electricity are required for clean energy production. Recently, the integrated gasification and fuel cell (IGFC) system, which incorporates a solid oxide fuel cell (SOFC), has attracted much attention [1,2]. After the pyrolysis of solid fuels, volatile matter and char are produced; the char is then gasified to produce H2 in the IGFC system. However, a large exergy loss was observed in char gasification at 1100 C [2]. Conversely, direct carbon fuel cells (DCFCs) directly convert the chemical energy in solid carbon into electricity without the need for a gasification process. The DCFC is an important device for minimizing the exergy
* Corresponding author. Department of Mechanical Engineering, School of Engineering, Tokyo Institute of Technology, NE-6, 2-12-1, Ookayama, Meguro-ku, Tokyo, 152-8552, Japan. E-mail address: [email protected] (H. Watanabe). http://dx.doi.org/10.1016/j.jpowsour.2016.08.122 0378-7753/© 2016 Elsevier B.V. All rights reserved.
loss in energy conversion systems that use solid fuels. For example, a DCFC/SOFC combined system, in which volatile matter is used in the SOFC after reforming and the char directly used in the DCFC, has the potential to achieve very high efficiency. Although a wide range of designs and concepts has been tested, the overall efforts to develop DCFC technology have been relatively minor in comparison to the efforts devoted to other major fuel cell technologies [3]. Ordowich et al. forecasted coal and natural gas power generation technologies through 2050 [4]. In their model, IGFC becomes the dominant power generation technologies from 2030 onward, and DCFC technologies would likely become an important technology past 2050. Although there are many barriers to develop a practical DCFC such as lower power density, ash accumulation and continuous supply of carbon fuel [5,6], further investigation is necessary to overcome these issues. An important difference in DCFCs from other fuel cells is that the fuel is solid carbon particles, which possess a disadvantage in terms of reactivity and fluidity. A number of researchers have addressed
568
H. Watanabe et al. / Journal of Power Sources 329 (2016) 567e573
this issue by developing DCFCs that use a carbon/electrolyte slurry, in which carbon particles are dispersed into the electrolyte to form a triple-phase boundary (carbon, electrolyte, and anode) [7e18]. Molten carbonates have often been used as an electrolyte because they offer many advantages such as high ionic conductivity [3]. The anode (R1), cathode (R2), and overall (R3) reactions in this system are expressed as follows: C þ 2CO2 3 / 3CO2 þ 4e
(R1)
O2 þ 2CO2 þ 4e / 2CO2 3
(R2)
C þ O2 / CO2
(R3)
The desired reaction is complete oxidation, which releases four electrons per carbon, as given by R1. The carbon wetting is an important factor in the formation of triple phase boundary [12,19]. Our previous study showed the carbon wetting has a significant impact on the continuous power generation capacity of the DCFC [12]. When the carbon particles are well-wetted by the molten carbonate, the anodic reaction process is described as follows: (1) carbon and carbonate ions are provided to the anode surface, (2) the anode reaction (R1) takes place at a triple-phase boundary, (3) CO2 and electrons are produced. The transport of the carbon and the ion around the anode plays an important role in the power generation of the DCFC. In addition, it has been pointed out that CO2 bubbles, which evolve from the electrochemical oxidation of solid carbon (R1), can potentially inhibit the cell operation and current stability [20]. Our previous study showed that bubbles were produced around the anode of the DCFC during discharge by using the single carbon pellet [13]. Although the impact of gas products around the anode on the cell performance should be addressed, little effort has been made to perform an in situ observation around the anode during discharge. Further investigations are still necessary to study the reaction and transport phenomena around the anode in order to improve the DCFC performance. This study aims to study impact of gas products around the anode on the DCFC performance. The transport phenomena around the anode are studied through in situ observation and currentvoltage (I-V) characteristics.
2.2. DCFC setup Fig. 1 shows a schematic of a DCFC with a molten carbonate electrolyte, which was developed in our laboratory. Although further study of the electrode material is necessary for the high performance DCFC, in this study, the working electrode (WE), counter electrode (CE), and reference electrode (RE) were made from gold sheet which is immune to corrosion in the molten carbonate [21]. The area in contact with the carbon was 1.0 cm2. The RE and CE were in alumina tubes, and the WE was in a porous alumina tube (for separation of carbon particles, as described later). Each gold sheet was spot welded to the gold wire and extended to the other end of the tube to provide a connection to the potentiostat/ galvanostat (HAL-3001, Hokuto Denko) for cell parameter measurements. The DCFC used in this study is similar to the one used in previous studies [11e13]. The part that differs is the quartz window attached to the plug above the anode for the in situ observation. A zoom camera (PowerShot SX50 HS, Canon) was used to record the phenomena around the anode as seen through the quartz windows. A total of 240 g of dry ternary carbonate powder without carbon was placed in an alumina crucible with an inner diameter of 80 mm and a height of 70 mm. In addition, the mixed carbon/carbonate powder was contained in a porous alumina tube with an inner diameter of 15 mm. The average pore diameter of the porous tube
2. Experimental section 2.1. Carbon characterizations Commercially available activated carbon (activated charcoal, Kantokagaku) was used in this study. Table 1 shows an elemental analysis (dry-base) and the ash content of the samples. The size distribution of the activated carbon particles, which was measured by laser diffraction (LMS-2000e, Seishin Enterprise Co. Ltd.), was shown in the supplemental material. By sieving commercially activated carbon, a size distribution ranging from 2 to 60 mm was obtained. The average size of the activated carbon particles was 16.7 mm. The crystalline parameters of the activated carbon were as follows: d002 (interplanar distance), Lc, (the layer dimension perpendicular to the basal plane) and La (the layer dimension parallel to the basal plane) were 0.376 nm, 0.987 nm, and 3.22 nm, respectively.
Table 1 Elemental analysis of the activated chars [wt%, dry]. C
H
N
S
O (diff)
Ash
92.3
0.17
3.04
0.10
4.39
0.8
Fig. 1. Schematic diagram of DCFC with quartz window for the in situ observation.
H. Watanabe et al. / Journal of Power Sources 329 (2016) 567e573
was 120 nm. Therefore, carbon particles larger than this size did not pass through the porous alumina tube. This allowed contact between the molten carbonate electrolyte and WE, CE, and RE. In fact, the porous alumina tube worked as a separator between the carbon/carbonate slurry and the molten carbonate. The DCFC developed in this study was of a cartridge type, and the cell stack was packages by using several anode compartments consisting of porous alumina tubes and carbon/carbonate slurry. In addition, containing the carbon/carbonate slurry in the porous alumina tube was advantageous because the slurry was easily exchanged by using new porous alumina tube with the slurry when ash accumulates with the use of coal and biomass chars. In this study, the activated carbon was used as fuel. The mixed carbon/carbonate powder consisted of 10 g of dry ternary carbonate powder and 0.1 g of activated carbon. The carbon fuel content in the carbonates was 1.0 wt%, and the carbonate molar compositions (Li2CO3/Na2CO3/ K2CO3) were set to 16.9/25/58.1 mol%. The quartz reactor, with an inner diameter of 95 mm and length of 200 mm, was heated using an electric furnace, and Ar was introduced into the reactor compartment during heating. The gases O2 (50 mL/min) and CO2 (100 mL/min) were introduced to both the cathode and reference electrodes. The carbon/carbonate slurry was stirred using bubbling Ar gas with a flow rate of 50 mL/min. The Ar gas was released from two holes in the gas line tube, each with a diameter of 2 mm. When bubbling was not used, the Ar gas was instead introduced from the top of the porous tube at a rate of 100 mL/min. The furnace temperature was set to 1073 K. Details are given in Refs. [11,12].
569
during the long discharge at 20 mA/cm2 for 30 min are 0.14 ml and 25 ml, respectively. The amount of CO2 produced during the measurement of the I-V curves is thus insignificant compared to that produced during the long discharge. This DCFC comprised the WE, CE, RE, and reactor compartments. In this apparatus, the reactor compartment was a space in the quartz reactor, excluding the WE, CE, and RE compartments. Anode off-gases from the WE were able to pass through the porous alumina tube and diffuse to the reactor compartments. Therefore, after reaching steady state, the CO2 and CO concentrations in the anode off-gases from the WE and reactor compartments (XCO2,WE, XCO2,reactor, XCO,WE, XCO,reactor) were measured with a flame ionization detector (FID) equipped with a Porapack-Q column and a methanizer. The outlet gas flow rates (QWE, and Qreactor) were also measured, using a wet gas meter (W-NK-1, Shinagawa Co. Ltd). The CO2 and CO production rates (QCO2,product and QCO,product) were determined by the sum of the differences between CO2 and CO flow rates during discharge in the constant current mode and those at the open circuit voltage (OCV) condition, given as:
QCO2 ;product ¼
X
XCO2 ;i Qi discharge
i
XCO2 ;i Q i OCV
ði ¼ WE; reactor compartmentÞ (2)
QCO;product ¼
X
XCO;i Qi discharge
i
XCO;i Q i OCV
2.3. Measurement of current-voltage curves and anode off-gases The long discharge was performed at a constant current density of either 20 or 40 mA/cm2. The in situ observation and anode offgas measurements were also performed during the long discharge. As shown in Fig. 2, the I-V curves were measured at a scanning rate of 100 mV/s before and after the long discharge for 30 min to study the impact of the gas products during the long discharge. Measurement of I-V curve is completed within approximately 10 s. Based on R1, the CO2 production rate during discharge is calculated as follows:
QCO2 ;product ¼
3IVm 4F
(1)
where F is the Faraday constant (96,450 C mol1), Vm is the molar volume of CO2 and I(A) is the constant current flow. When the current density is 20 mA/cm2, CO2 production rate is 1.55 107 mol/s which corresponds to 0.22 ml/min at 293 K. At 1073 K where the gas is produced on the anode, CO2 production rate is 0.82 ml/min and the theoretical amount of CO2 produced during the measurement of the I-V curves and that produced
Fig. 2. Experimental procedure for I-V curves measurement.
ði ¼ WE; reactor compartmentÞ (3)
Anode off-gases were measured prior to discharge, confirming that the O2 concentration was negligible in the anode and reactor compartments. 3. Results and discussion Fig. 3 shows I-V and the current-power (I-P) curves prior to the long discharge with and without Ar bubbling. In this experiment, the I-V curves with bubbling are similar to those without bubbling, except for around the maximum current density near 100e120 mA/ cm2. This was because the concentration polarization appeared earlier, caused by the bubbling that inhibited contact between the carbon and anode. The maximum power density was reached at approximately 50 mW/cm2. Meanwhile, our previous study showed that Ar bubbling improves the current discharge curves during long discharge [11]. During long discharge, an amount of gas is expected to be produced around the anode, whereas that is insignificant in the measurement of I-V curves with the scanning rate of 100 mV/s. This will be discussed later. Fig. 4 shows the constant current discharge curves (a) and in situ observations around the anode (b) at 40 mA/cm2 without bubbling (Qbub ¼ 0 ml/min). The marks (A), (B), and (C) on the constant current discharge curve (Fig. 4a) refer to the images of in situ observation (Fig. 4b). In Fig. 4a, the voltage at 0 min corresponds to the OCV. The initial voltage drop is mainly caused by activation losses arising as a result of the kinetics at the electrode. Subsequently, the voltage decreases to 0.6 V at 2 min and finally reaches 0 V at 3.5 min. These voltage drops can be explained from the in situ observation around the anode in Fig. 4b. In the observation, the upper half of the anode plate is clearly seen at 0 min ((A) OCV, before discharge). During discharge, electrochemical oxidation (C þ 2CO2 3 / 3CO2 þ 4e (R1)) is progressed on the lower half of the anode embedded into the slurry. Due to bubble formation, the
570
H. Watanabe et al. / Journal of Power Sources 329 (2016) 567e573
(a) I-V
(b) I-P
Fig. 3. Current-voltage (I-V) and current power (I-P) curves of the DCFC with and without bubbling before the long discharge.
(A, OCV)
(B) (C)
(a) Constant current discharge curve
2 min (B)
0 min (A)
Upper half of the anode is seen
1.0 mm
Carbon slurry
The anode boundary becomes cloudy due to bubble formation
3.5 min (C)
1.0 mm
The anode is hidden by bubbles
1.0 mm
(b) In-situ observation around the anode Fig. 4. Current discharge curves and in situ observation around the anode at I ¼ 40 mA/cm2 without bubbling. (a) Constant current discharge curve. (b) In situ observation around the anode.
anode boundary becomes cloudy at 2 min (B) after discharge and then is difficult to observe from the top at 3.5 min (C). In fact, gas bubbles, which are the products of electrochemical oxidation of carbon, cover the anode as the discharge proceeds. The finding in this case is that the voltage becomes 0 V in the constant current discharge mode when the anode is hidden by gas bubbles at 3.5 min (C). This is because the gas bubbles around the anode prevent the carbon particles and ions from reaching the anode. The details are discussed later in this study. The discharge period of 3.5 min is too short to measure the gas products with the FID. The IV curves following the long discharge at 40 mA/cm2 were not measured in this study. Fig. 5 shows the constant current discharge curve (a), the CO2
and CO production rate (b), the in situ observation around the anode (c), and a magnified image of the in situ observation at 15 min (d) at 20 mA/cm2 with bubbling (Qbub ¼ 50 ml/min). The marks (A), (B), and (C) in Fig. 5a refer to the images in Fig. 5c. In Fig. 5a, at the steady state, the voltage reaches approximately 0.75 V, which is lower than that in the I-V curves at 20 mA/cm2, as shown in Fig. 3. This implies that the impact of gas products is not appeared in the I-V curves measured at the scanning rate of 100 mV/s. In Fig. 5b, only CO2 is produced from the discharge. The average total CO2 flow rates at 20 mA/cm2 and OCV condition are 0.62 ml/min and 0.43 ml/min, respectively. The average measured CO2 production rate (0.19 ml/min) almost corresponds to the theoretical rate (¼ 0.22 ml/min), indicating that the carbon is
H. Watanabe et al. / Journal of Power Sources 329 (2016) 567e573
571
(A, OCV) (B)
(C)
(a) Constant current discharge curve
(b) CO2 and CO production rate The anode is hidden by bubbles
15 min (B)
0 min (A)
22 min (C)
Many bubbles form
Upper half of the anode is seen
1.0 mm
Carbon slurry
(d)
1.0 mm
1.0 mm
(c) In-situ observation around the anode during discharge 0.5 mm
(d) Magnified image of (c, 15 min (B)) Fig. 5. Current discharge characteristics at I ¼ 20 mA/cm2 with bubbling. (a) Constant current discharge curve (b) CO2 and CO production rate. (c) In situ observation around the anode during discharge. (d) Magnified image of (c) (15 min (B)).
consumed by the electrochemical oxidation expressed as R1. With bubbling, CO is not detected at both the discharge (20 mA/cm2) and OCV. However, without bubbling, CO is formed at the discharge (20 mA/cm2), whereas CO is not formed at OCV. CO is supposed to be formed through the partial electrochemical oxidation (Cþ 1/2 CO2 3 / 3/2 CO þ e ) which is discussed in our previous study [11,13]. Compared with Fig. 4b, CO2 bubbles are more clearly seen near the anode in Fig. 5c at 15 min and 22 min of discharge because the longer discharge produces larger amount of CO2. The constant current discharge at 20 mA/cm2 is achieved for 30 min even though the anode is almost completely covered by small CO2 bubbles after 15 min of discharge, as shown in Fig. 5c. This indicates that triple phase boundaries are able to form at lower current density even through the anode is covered by bubbles at the macroscopic view. In Fig. 5d, which is the magnified image of the box in Fig. 5c at 15 min (B), bubbles with a diameter of nearly 0.3 mm are seen near the anode. During the discharge at 20 mA/cm2, the CO2 production rate is 0.82 ml/min at 1073 K. If the CO2 bubble size is assumed to be 0.3 mm for all bubbles, the bubble generation rate is approximately 58,000 bubbles/min, which is sufficient to cover the anode after
15 min of discharge. Meanwhile, a small quantity of bubbles is also observed in the image at 0 min (A) because it was taken after the measurement of I-V curves at the scanning rate of 100 mV/s. A very small amount of CO2 is produced during the measurement of the IV curves. Fig. 6 is a schematic illustration depicting the impact of gas products on the DCFC performance during discharge. Prior to the long discharge (A), the carbon and carbonate ions can approach the Au surface without encountering any obstacles. During discharge, the anode reaction (R1) takes place at a triple phase boundary, and CO2 bubbles are produced on the anode. The CO2 bubbles prevent the carbon particles and ions from reaching the Au surface, and they reduce the effective surface area for the reaction. After the long discharge (B), the CO2 bubbles cover a sufficient area on the reaction site to reduce the cell performance. Fig. 7 shows the I-V and I-P curves before and after the long discharge at 20 mA/cm2 for 30 min with bubbling (Qbub ¼ 50 ml/ min). The I-V curves are measured before the long discharge and at 1 min and 10 min after the termination of the long discharge at 20 mA/cm2. The I-V curve at 1 min after termination of the long
572
H. Watanabe et al. / Journal of Power Sources 329 (2016) 567e573
Fig. 6. Schematic illustration depicting the CO2 bubble production around the anode during discharge in the DCFC.
(a) I-V
(b) I-P
Fig. 7. I-V and I-P curves before and after the long discharge for 30 min (Qbub ¼ 50 mL/min, scanning rate ¼ 100 mV/s).
discharge was lower than those prior to the long discharge, indicating that the overpotential significantly increases due to the CO2 bubbles around the anode. The overpotential particularly increases at higher current density, where the mass transport is the limiting process, indicating that CO2 bubbles around the anode prevent the carbon particles and ions from making contact with the anode. At 10 min after the termination, the I-V curve approaches to that before the long discharge. This indicates that the CO2 bubbles around the anode have been released into the atmosphere and that
(a) I-V
carbon particles and ions can move to the anode again. Theoretically, 1.1 mg of activated carbon is consumed through R1 during discharge at 20 mA/cm2 for 30 min. Because the initial amount of carbon is 100 mg, the change on carbon content before and after the long discharge is insignificant. Fig. 8 shows the I-V and I-P curves before and after the long discharge at 20 mA/cm2 without bubbling (Qbub ¼ 0 ml/min). The OCVs without bubbling dramatically diminish to 1.0 V at 1 min after the termination of the long discharge unlike those with bubbling as
(b) I-P
Fig. 8. I-V and I-P curves before and after the long discharge for 30 min (Qbub ¼ 0 mL/min, scanning rate ¼ 100 mV/s).
H. Watanabe et al. / Journal of Power Sources 329 (2016) 567e573
shown in Fig. 7. Our previous study showed that the carbon content around the anode influences the reaction scheme of solid carbon, leading to a change in OCV [11]. The CO2 bubbles have the ability to change the local carbon content around the anode. When bubbling is used, the reduction of OCVs at 1 min after the termination is less significant (Fig. 7) because Ar bubbling is supposed to remove gas bubbles around the anode, and enhance the formation of reaction site during the short period. This also explains our previous result that bubbling improves the current discharge curves during long discharge which produces the amount of gas [11]. It is suggested that the role of bubbling is not only stirring the slurry but also removing gas bubbles around the anode. The I-V curve at 10 min after the termination of the long discharge without bubbling is similar to that with bubbling. This indicates that the CO2 bubbles are released into the atmosphere after sufficient time elapses regardless of the presence of Ar bubbling. As shown in Fig. 7, the stirring by Ar bubbling is still insufficient to restore back the carbon/carbonate slurry characteristics prior to the long discharge. 4. Conclusion In this study, the impact of gas products around the anode on the cell performance was studied. The in situ observation was performed during discharge. In addition, the I-V curves were measured before and after the long discharge. The in situ observation showed that the anode was almost completely covered by gas bubbles when the voltage became nearly 0 V in the constant current discharge mode at 40 mA/cm2. Meanwhile, the long discharge at 20 mA/cm2 was achieved for 30 min, although the anode was covered by the CO2 bubbles at 15 min. This indicated that triple phase boundaries were able to form at the lower current density when the anode was covered by bubbles at the macroscopic view. The I-V curves measured 1 min after termination of the long discharge were lower than the curves prior to the long discharge. In particular, the overpotential significantly increased at the higher current density, where the mass transport became the limiting process. This indicates that the CO2 bubbles prevented the carbon particles and ions from reaching the anode, resulting in lower DCFC performance. As time elapsed after the termination of the long discharge, the I-V curve became closer to that before the long discharge, indicating that the CO2 bubbles were released into the atmosphere and the carbon particles and ions were able to contact the anode again. It was shown that the cell performance was significantly influenced by gas products around the anode of the DCFC. Acknowledgment This study was partly supported by J-POWER.
573
Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2016.08.122. References [1] J.P. Trembly, R.S. Gemmen, D.J. Bayless, The effect of IGFC warm gas cleanup system conditions on the gas-solid partitioning and form of trace species in coal syngas and their interaction wth SOFC anodes, J. Power Sources 163 (2007) 986e996. [2] S. Chen, N. Lior, W. Xiang, Coal gasification integration with solid oxide fuel cell and chemical looping combustion for high-efficiency power generation with inherent CO2 capture, Appl. Energy 146 (2015) 298e312. [3] S. Giddey, S.P.S. Badwal, A. Kulkarni, C. Munnings, A comprehensive review of direct carbon fuel cell technology, Prog. Energy Combust. Sci. 38 (2012) 360e399. [4] C. Ordowich, J. Chase, D. Steele, R. Malhotra, M. Harada, K. Makino, Applying learning curves to modeling future coal and gas power generation technologies, Energy & Fuels 26 (2012) 753e766. [5] N.J. Cherepy, R. Krueger, K.J. Fiet, A.K. Jankowski, J.F. Cooper, Direct conversion of carbon fuels in a molten carbonate fuel cell, J. Electrochem. Soc. 152 (2005) A80eA87. [6] S. Giddey, A. Kulkarni, C. Munnings, S.P.S. Badwal, Composite anodes for improved performance of a direct carbon fuel cell, J. Power Sources 284 (2015) 122e129. [7] D.G. Vutetakis, D.R. Skidmore, H.J. Byker, Electrochemical oxidation of molten carbonate-coal slurries, J. Electrochem. Soc. 134 (1987) 3027e3035. [8] X. Li, Z.H. Zhu, R.D. Marco, A. Dicks, J. Bradley, S. Liu, G.Q. Lu, Factors that determine the performance of carbon fuels in the direct carbon fuel cell, Ind. Eng. Chem. Res. 47 (2008) 9670e9677. [9] J.F. Cooper, J.R. Selman, Analysis of the carbon anode in direct carbon conversion fuel cells, Int. J. Hydrogen Energy 37 (2012) 19319e19328. [10] C.Q. Wang, J. Liu, J. Zeng, J.L. Yin, G.L. Wang, D.X. Cao, Significant improvement of electrooxidation performance of carbon in molten carbonates by the introduction of transition metal oxides, J. Power Sources 233 (2013) 244e251. [11] H. Watanabe, T. Furuyama, K. Okazaki, Enhancing the efficiency of direct carbon fuel cells by bubbling Ar gas in carbon/carbonate slurry, J. Power Sources 273 (2015) 340e350. [12] H. Watanabe, A. Kimura, K. Okazaki, Impact of ternary carbonate composition on the morphology of the carbon/carbonate slurry and continuous power generation by direct carbon fuel cells, Energy Fuels 30 (2016) 1835e1840. [13] H. Watanabe, T. Furuyama, K. Okazaki, Electrochemical oxidation of activated carbon and coal chars in a direct carbon fuel cell using carbon/carbonate slurry stirred by Ar bubbling, J. Chem. Eng. Jpn. 49 (2016) 237e242. [14] C. Li, H. Yi, D. Lee, On-demand supply of slurry fuels to a porous anode of a direct carbon fuel cell: attempts to increase fuel-anode contact and realize longterm operation, J. Power Sources 309 (2016) 99e107. [15] K. Pointon, B. Lakeman, J. Irvine, J. Bradley, S. Jain, The development of a carbon-air semi fuel cell, J. Power Sources 162 (2006) 750e756. [16] L. Deleebeeck, K.K. Hansen, HDCFC performance as a function of anode atmosphere (N2-CO2), J. Electrochem. Soc. 161 (2014) F33eF46. ndez, K.K. Hansen, Hybrid direct carbon [17] L. Deleebeeck, A. Arenillas, K. Mene fuel cell anode setup, Int. J. Hydrogen Energy 40 (2015) 1945e1958. [18] L. Deleebeeck, D. Ippolito, K.K. Hansen, Enhancing hybrid direct carbon fuel cell anode performance using Ag2O, Electrochim. Acta 152 (2015) 222e239. [19] C.C. Chen, T. Maruyama, P.H. Hsieh, J.R. Selman, Wetting behavior of carbon in molten carbonate, J. Electrochem. Soc. 159 (2012) D597eD604. [20] T.M. Gür, Critical review of carbon conversion in “carbon fuel cells”, Chem. Rev. 113 (2013) 6179e6206. [21] M. Wyatt, J.M. Fisher, Control of corrosion in molten carbonate fuel cells, Platin. Met. Rev. 32 (1988) 200e203.