Carbon Nanofibers Synthesized by Catalytic Decomposition of

Energy & Fuels 2009, 23, 3721–3731

3721

Carbon Nanofibers Synthesized by Catalytic Decomposition of Methane and Their Electrochemical Performance in a Direct Carbon Fuel Cell Xiang Li,† Zhonghua Zhu,*,† Roland De Marco,‡ John Bradley,† and Andrew Dicks† DiVision of Chemical Engineering, School of Engineering, UniVersity of Queensland, Brisbane 4072 Australia and Department of Applied Chemistry, Curtin UniVersity of Technology, Perth, Australia ReceiVed March 7, 2009. ReVised Manuscript ReceiVed May 10, 2009

Various carbon nanofibers (CNFs) have been synthesized by catalytic decomposition of methane over Ni-Al2O3 and Ni-Cu-Al2O3 catalysts at 500-700 °C. On the basis of the physical and chemical characterization of these CNFs, their electrochemical reactivities are evaluated in a novel direct carbon fuel cell (DCFC). It is found that the microstructures and electrochemical reactivities of CNFs are highly dependent on the decomposition temperature, catalyst composition and operation times in the DCFC. Decreasing decomposition temperature and introducing Cu into Ni-Al2O3 catalyst can produce electrochemically reactive CNFs, due to their enlarged surface area and decreased degrees of graphitic structure. The electrochemical reactivity of synthesized CNFs can be further promoted by the preoxidation in the high temperature molten carbonates, owing to an obvious increase in the oxygen functional groups on the surface of the CNFs.

1. Introduction The direct carbon fuel cell (DCFC), the only fuel cell capable of converting solid carbon into electricity without a reformation process, is gaining increased attention due to its high conversion efficiency. The solid carbon fuels in the DCFC can be produced easily from various resources, including coal, natural gas, petroleum, biomass carbon (wood, grass, charcoal, etc.) and even organic waste.1,2 Compared with other hydrogen-based fuel cells, the DCFC has the significant thermodynamic advantage of a near zero entropy change at high temperature, which means the theoretical electrochemical efficiency of the DCFC (∆G/∆H) is almost 100%.1-4 Even under practical conditions, about 80% efficiency can be reached in the DCFC system. Moreover, the activities (chemical potentials) of both reactant carbon and the product carbon dioxide are fixed, resulting in a stable carbon anode potential during practical operation.1 Last but not least, the DCFC has lower emissions compared with conventional power plants. In principle, the off-gas can be pure carbon dioxide, which can be directly collected for industrial use or sequestration.1,2 Various electrolytes such as molten carbonates,1,5,6 molten hydroxides,4,7,8 and yttria stabilized zirconia (YSZ)-based solid * Corresponding author. E-mail: [email protected]; phone: 61 7 3365 3528; fax: 61 7 3365 4199. † University of Queensland. ‡ Curtin University of Technology. (1) Cherepy, N. J.; Krueger, R.; Fiet, K. J.; Jankowski, A. F.; Cooper, J. F. J. Electrochem. Soc. 2005, 152, A80–A87. (2) Cao, D.; Sun, Y.; Wang, G. J. Power Sources 2007, 167, 250–257. (3) Antal, M. J.; Nihous, G. C. Ind. Eng. Chem. Res. 2008, 47, 2442– 2448. (4) Zecevic, S.; Patton, E. M.; Parhami, P. Carbon 2004, 42, 1983– 1993. (5) Vutetakis, D. G.; Skidmore, D. R.; Byker, H. J. J. Electrochem. Soc. 1987, 134, 3027–3035. (6) Weaver, R. D.; Nanis, L. J. Electrochem. Soc. 1980, 127, C410– C410. (7) Hackett, G. A.; Zondlo, J. W.; Svensson, R. J. Power Sources 2007, 168, 111–118.

electrolytes9-12 have been used in the DCFC systems. One of the latest developments in DCFC technology is to utilize highly reactive carbon particulates dispersed in a molten carbonate electrolyte, which flows between the anode and cathode at high temperature.1,5,6 The anode and cathode reactions may be expressed as eq 1 and eq 2. The overall reaction is given by eq 3.1,5 Anode reaction: C + 2CO23 f 3CO2 + 4e

(1)

Cathode reaction: O2 + 2CO2 + 4e- f 2CO23

(2)

Overall reaction: C + O2 f CO2

(3)

where the anode potential is given by eq 4:5 3 2 (w)] + (RT/4F)ln[PCO (r)PO2(r)] Ecell ) E° - (RT/4F)ln[PCO 2 2 (4)

and E° is the anode potential at standard condition, R is the universal gas constant, and T is the cell temperature (K). PCO2(w) is the CO2 partial pressure at the working electrode, whereas PCO2(r) and PO2(r) are the partial pressures of CO2 and O2 at the reference electrode, respectively. Although different DCFC systems have been successfully developed, this technology is still a bench-scale system at present. One of the major challenges in scaling up the DCFC is to find suitable carbon fuel sources with low impurities such (8) Nunoura, T.; Dowaki, K.; Fushimi, C.; Allen, S.; Meszaros, E.; Antal, M. J. Ind. Eng. Chem. Res. 2007, 46, 734–744. (9) Gur, T. M.; Huggins, R. A. J. Electrochem. Soc. 1992, 139, L95– L97. (10) Pointon, K.; Lakeman, B.; Irvine, J.; Bradley, J.; Jain, S. J. Power Sources 2006, 162, 750–756. (11) Lee, A. C.; Li, S.; Mitchell, R. E.; Gur, T. M. Electrochem. SolidState Lett. 2008, 11, B20–B23. (12) Ihara, M.; Hasegawa, S. J. Electrochem. Soc. 2006, 153, A1544– A1546.

10.1021/ef900203h CCC: $40.75  2009 American Chemical Society Published on Web 06/05/2009

3722

Energy & Fuels, Vol. 23, 2009

as mineral matter and sulfur, which have potential side-effects on the anodic electrochemical reaction and the lifetime of the DCFC.1,2,5 In this regard, catalytic chemical vapor deposition (CCVD) has been accepted as one of the most suitable methods to produce large-scale and economical carbons with low impurities, such as carbon blacks, carbon nanotubes, and carbon nanofibers (CNFs).13,14 The principle of the method is the decomposition of a gaseous precursor (such as hydrocarbons) on the substrate forming the solid carbon deposition. Particularly, the catalytic decomposition of methane under relatively mild conditions will not only produce large-scale CNFs but also COXfree hydrogen.15-18 These almost pure CNFs can be used in the DCFC as fuel directly, and the byproduct of hydrogen is a highly desirable clean fuel for other hydrogen-oxygen fuel cells or chemical processes. It is well-known that the properties of synthesized CNFs are highly dependent on the catalyst composition, decomposition temperatures, and the carbon precursors. Until now, Ni-based catalysts are the most effective catalysts to produce CNFs by the CCVD method, due to their lower activation temperature and more stable properties.13 A number of publications13,16,17,19 have reported that synthesized CNFs may reach more than 100 g of carbon per gram of catalyst over coprecipitated Ni-Al2O3 during the decomposition of methane at 450-650 °C. With an elevation in decomposition temperature, the conversion of methane into CNFs and hydrogen may increase and reach its highest values at around 650 °C. Above this temperature, the catalysts would be deactivated quickly due to the encapsulation of the metal active sites by in the graphitic layers that are formed. However, doping of Cu into these Ni catalysts can improve the methane conversion by 10-20% and increase the yields of CNFs by 2-5 times at even higher temperature.15,20,21 Previous electrochemical studies on CNFs were mainly focused on their use as structural components (such as bipolar plates and catalysts supports) in fuel cells and the electrode materials in lithium-ion batteries due to their high porosity and conductivity.22-24 To our knowledge, no study has been conducted on using CNFs as fuels consumed in the DCFC. In this paper, four CNFs were synthesized by catalytic decomposition of methane at 500-700 °C on coprecipitated Ni-Al2O3 and Ni-Cu-Al2O3 catalysts. On the basis of the characterization of microstructure and properties of the CNFs, we compared systematically their electrochemical reactivities in our DCFC system. Particularly, the relationship between the surface (13) Choudhary, T. V.; Aksoylu, E.; Goodman, D. W. Catal. ReV. s Sci. Eng. 2003, 45, 151–203. (14) Rodriguez, N. M. J. Mater. Res. 1993, 8, 3233–3250. (15) Chen, J.; Li, Y.; Li, Z.; Zhang, X. Appl. Catal., A 2004, 269, 179– 186. (16) Li, Y.; Chen, J.; Qin, Y.; Chang, L. Energy Fuels 2000, 14, 1188– 1194. (17) Ermakova, M. A.; Ermakov, D. Y.; Kuvshinov, G. G. Appl. Catal., A 2000, 201, 61–70. (18) Choudhary, T. V.; Goodman, D. W. Catal. Today 2002, 77, 65– 78. (19) Zhou, J. H.; Sui, Z. J.; Li, P.; Chen, D.; Dai, Y. C.; Yuan, W. K. Carbon 2006, 44, 3255–3262. (20) Reshetenko, T. V.; Avdeeva, L. B.; Ismagilov, Z. R.; Chuvilin, A. L.; Ushakov, V. A. Appl. Catal., A 2003, 247, 51–63. (21) Li, Y.; Chen, J.; Chang, L.; Qin, Y. J. Catal. 1998, 178, 76–83. (22) Dicks, A. L. J. Power Sources 2006, 156, 128–141. (23) Yoon, S.-H.; Park, C.-W.; Yang, H.; Korai, Y.; Mochida, I.; Baker, R. T. K.; Rodriguez, N. M. Carbon 2004, 42, 21–32. (24) Zou, G.; Zhang, D.; Dong, C.; Li, H.; Xiong, K.; Fei, L.; Qian, Y. Carbon 2006, 44, 828–832.

Li et al. Table 1. Composition and Preparation Conditions of Catalysts catalysts

composition (weight ratio)

calcination reduction temperature (°C) temperature (°C)

Ni-Al2O3 Ni/Al ) 81:19 Ni-Cu-Al2O3 Ni/Cu/Al ) 54:27:19

450 450

700 700

Table 2. Nomination and Synthesis Conditions of All CNFs

CNFs

catalyst composition (weight ratio)

A500 A600 A700 B500 B600 B700 B700-1 B700-2

Ni-Al2O3 Ni-Al2O3 Ni-Al2O3 Ni-Cu-Al2O3 Ni-Cu-Al2O3 Ni-Cu-Al2O3 Ni-Cu-Al2O3 Ni-Cu-Al2O3

CH4 CNF decomposition decomposition yields temperature (°C) time (h) (g/gcatalyst) 500 600 700 500 600 700 700 700

75 60 35 60 55 50

128.3 106.7 8.4 15.6 254.3 223.8

properties of CNFs and their electrochemical performance in the DCFC was explored. 2. Experimental Section 2.1. Preparation of CNFs. The synthetic CNFs were prepared by catalytic chemical vapor deposition in a horizontal tubular furnace. The catalysts Ni-Al2O3 and Ni-Cu-Al2O3 were prepared by coprecipitation from a mixed aqueous solution of nitrates with sodium carbonate (as a precipitate). Details on the preparation are given elsewhere.15,25 The formed precipitates were washed in distilled water, dried at 110 °C, and calcined in air at 450 °C for 10 h. Table 1 summarizes the composition and preparation conditions for the two catalysts used in this work. Each time, 0.5 g of calcined catalyst (200-250 meshes particle size) was initially reduced in hydrogen (99.99% pure) at 700 °C for 2 h, and the reactor temperature was adjusted to a prescribed value in an argon atmosphere. Finally, methane gas (99.99% pure) flowed through the reactor (100 mL/min) for 30-75 h until the catalyst was almost deactivated (which can be defined by an extent of methane conversion of less than 3%). Table 2 shows the synthesis conditions for the CNFs using the two catalysts at different methane decomposition temperatures. The synthesized CNFs are denoted as A500, A600, A700, B500, B600, and B700, where A and B stand for the Ni-Al2O3 and Ni-Cu-Al2O3 catalysts, respectively; and the number is the methane decomposition temperature. Notably, the samples A700 and B500 have only 8.4 g/gcatalyst and 15.6 g/gcatalyst CNF yields, which means that they contain much higher levels of impurities than the other CNFs. Therefore, these two CNFs were not used in further characterization and DCFC tests. On the contrary, the B700 has the highest CNF yield (223.8 g/gcatalyst) in all CNFs. Thus, it was selected as a starting sample to study the effect of pre-electrochemical oxidation of molten carbonate electrolytes in the DCFC on the structural properties of CNFs. After the first test of B700 in DCFC, this sample was washed by 1 M HCl solution to remove residual carbonate salts and designated as B700-1. After characterization, B700-1 was tested in the DCFC again. The residual sample was washed by 1 M HCl and designated as B700-2 for the final characterization and DCFC tests. 2.2. Characterization of CNFs and Catalysts. The morphology of CNFs was characterized by a JEOL 6400 field emission scanning electron microscopy (SEM), with an accelerating voltage of 15.0 kV at room temperature. All samples were mounted on aluminum stubs and coated with platinum to ensure good conduction. The morphology of catalysts was performed on a JEOL JEM-100 CXII transmission electron microscopy (TEM). The catalyst samples were prepared under ethanol and suspended on a carbon film for examination. X-ray diffraction characterization of the CNFs was performed on a Rigaku Miniflex X-ray diffractometer (40 kV, 30 mA) with (25) Li, Y.; Chen, J.; Chang, L. Appl. Catal., A 1997, 163, 45–57.

CNF Production from Methane Decomposition

Energy & Fuels, Vol. 23, 2009 3723

Co KR radiation (λ ) 0.17902 nm) at a scanning rate of 2°/min in the 2θ range from 10° to 90°. The average size of carbon crystallite was calculated from the Debye-Scherrer equation:

L)

Kλ β cos θ

(5)

where λ is the wavelength of the X-rays, θ is the diffraction angle, K is the shape factor, and β is the peak width at half-maximum intensity. The value of K ) 0.89 and 1.84 were used for Lc and La, respectively. The crystallite size perpendicular to the basal plane, Lc, is obtained from the (002) reflection, whereas the crystallite size parallel to the basal plane, La, is calculated using the (100) reflection corresponding to the unit cell parameter. For the characterization of Ni-Al2O3 and Ni-Cu-Al2O3 catalysts, a D/MAX 2038 X-ray diffractometer with Fe KR (λ ) 0.19373 nm) or Cu KR (λ ) 0.15418 nm) radiation was employed to get the XRD patterns of the precipitated precursor, calcined, and reduced samples. Electrical conductivities were measured using a frequency response analyzer (Solartron SI1260). First, approximately 200 mg of carbon material was pressed into a small pellet (13 mm diameter) at 150 kg/cm2. This sample pellet was placed between two goldplated blocking electrodes of 0.5 cm2 area in a chamber, and the impedance of the sample was measured over a frequency range 10 MHz to 1 Hz using a sinusoidal excitation amplitude of 10 mV root-mean-square (rms). Finally, the bulk conductivity σ (S/cm) was calculated using eq 6:

σ)

l (r - r0)A

(6)

where l is the sample pellet thickness (cm), r is the tested sample resistance (Ω) taken as the impedance at zero phase angle, r0 is the rig short circuit resistance (Ω), and A is the electrode contact area (0.5 cm2). N2 adsorption/desorption experiments were carried out in a Quadrasorb adsorption analyzer (Quantachrome, USA) at - 196 °C. The specific surface areas (SBET) of the CNFs and catalysts were calculated by the multiple point Brunauer-Emmett-Teller (BET) method in the relative pressure range P/Po ) 0.05-0.25. The total pore volume (Vtotal) is derived using the adsorption parameters at a relative pressure of P/Po ) 0.99. The average pore diameters (Dpore) of samples were calculated by the available software(QuadraWinV2.0),whichappliestheBarrett-Joyner-Hallenda (BJH) method. Prior to the N2 adsorption/desorption measurements, all samples were degassed at 200 °C overnight. Temperature programmed oxidation (TPO) measurements were conducted in air-flow of 80 mL/min in a thermogravimetric analyzer (Shimadzu TGA-50). The CNFs were loaded into a platinum pan and heated under nitrogen atmosphere from room temperature to 200 °C, and held for 1 h to remove the adsorbed water, and then the temperature was further increased to 900 °C in air (80 mL/ min) with a heating rate of 10 °C/min. The X-ray photoelectron spectroscopy (XPS) measurements were conducted using a Kratos Axis Ultra XPS system incorporating a 165 mm hemispherical electron energy analyzer. The incident radiation was monochromatic Al KR X-rays (1486.6 eV) at 150 W (15 kV, 10 mA). Survey scans were taken at an analyzer pass energy of 160 eV and performed over a 1200 eV binding energy range using a 1.0 eV step and a dwell time of 100 ms. The mass titration method of Noh and Schwarz26 was used to estimate the point of zero charge (PZC) on the surfaces of CNFs, at which the net total (external and internal) surface charge of CNFs is zero. Three different initial pH solutions (pH 3, 6, and 11) were prepared using HNO3 (0.1 M) and NaOH (0.1 M). Sodium nitrate was used as the background electrolyte. For each initial pH, six containers were filled with 20 mL of the solution and different amounts of CNFs were added (0.05, 0.10, 0.50, 1.00, and 10% by (26) Noh, J. S.; Schwarz, J. A. Carbon 1990, 28, 675–682.

Figure 1. Schematic diagram of the DCFC.

weight). The equilibrium pH was measured after 24 h. The plot of pH versus mass fraction showed a plateau, and the PZC is identified as the point at which the change in pH was negligible. The PZC was taken as the average of the three asymptotic pH values. Temperature-programmed desorption (TPD) experiments were carried out in a vertical tube furnace with Ar (80 mL/min) as the carrier gas. Each time, 0.5 g sample was placed in a quartz tube reactor, heated to 110 °C, held for 60 min, and then ramped at 5 °C/min to 900 °C. The gases evolved were analyzed using a gas chromatograph (Shimadzu GC-17A) equipped with a thermal conductivity detector and a Carbosphere column. 2.3. Evaluation of CNFs in the DCFC. Figure 1 shows the schematic diagram of the DCFC system. The working electrode (WE) typically consists of a solid gold rod (serving as a current connector, 99.9% purity, and 3.2 mm diameter) that was cemented to the end of an alumina tube (7 mm diameter). A gold wire was spot-welded to the gold rod and extended to the other end of the alumina tube for connecting to the potentiostat. The total exposed surface area of the gold rod was 3.0 cm2. The counter electrode (CE) was made from a gold sheet (with 1.0 cm2 surface area) spotwelded to a gold wire, and the gold parts were sheathed in a closedbottom alumina tube (12 mm diameter). A 1.5 mm hole at the bottom of the alumina sheath allowed contact of the electrolytes between the CE and the WE. The reference electrode (RE) was constructed from an alumina sheath (12 mm diameter) containing a gold wire in contact with the electrolyte melts, and a 0.05-0.1 mm pinhole (by a laser drill) at the bottom of the alumina sheath offered the contact of electrolytes between the RE and WE. The particulate carbon fuels with the ternary molten carbonate electrolyte (32% Li, 34% Na, 34% K eutectic) were contained in a large alumina crucible. To keep the entire cell in a gastight environment, the alumina crucible was placed at the bottom of an inconel canister and sealed by a water-cooled brass lid. The three electrodes and gas feed tubes were supported by the brass lid and inserted into the WE anode compartment. An inconel stirring rod with a half circle impeller (3 cm diameter) and a type-K thermocouple was also placed in the molten carbonate electrolyte. Prior to each experiment, 250 g of dry ternary carbonate powders were mixed with the 12.5 g of CNFs (5 wt % carbon fuels to total carbonate electrolytes). The gold parts of the three different electrodes were washed by aqua regia for 10 s and rinsed with distilled water, followed by acetone. After the fuel cell was assembled and sealed gastight, it was heated by a crucible furnace (Lindberg Blue/M, at a heating rate of 3 °C/min) to the appropriate operating temperature of 600, 700, or 800 °C. During the heat-up

3724


Li et al.

stage, Ar (150 mL/min) was purged into the WE compartment, and CO2 (50 mL/min) was purged into the CE and RE compartments. Once the required operating temperature was attained, the Ar and CO2 purge rates were decreased to 70 mL/min and 15 mL/ min, respectively. At the same time, air (35 mL/min) was introduced into the CE and RE compartments. Finally, the carbon anode halfcell measurements were performed with a potentiostat (Autolab PGSTAT302), using the GPES software package (Version 4.9). For the linear sweep voltammetry measurements, the anodic polarization started from the open circuit voltage (OCV, which is the potential difference between the WE and RE under open circuit conditions) to 0.0 V (relative to RE) at a scan rate of 1 mV/s. Note that under open circuit conditions the potential of the reference electrode is the same as the cathode and therefore the potential difference between the anode and reference is taken as the OCV. To obtain an estimate of power produced by the DCFC at a given current density, the anode potential (WE relative to RE) was multiplied by the corresponding current density to calculate the theoretical power density. Here we focus on the maximum theoretical power density (Pmax), that is, the maximum value of the theoretical power density for a certain anode potential versus current density curve (or V-I curve). It should be pointed out that the real fuel cell power density is determined by the cell voltage (i.e., the potential difference between anode and cathode) and not just the anode potential. Because the reactivity of the anode was the main focus of this work, the cathode potential was not measured, and cell voltages are therefore not reported. In the online measurements of anodic off-gas (evolved from the WE), the gas flow rate was measured by a soap film buret, and the gas compositions were analyzed by gas chromatography (Shimadzu GC-8A) with H2 as the carrier gas. In the tests, the off-gas composition was initially analyzed by gas chromatography with the current switched off to give a baseline and with the current switched on to measure the gases produced by electrochemical reaction. The current value was controlled by the potentiostat under potentiostatic conditions. Consequently, the net electrochemical CO2 evolution rate (QCO2) and total net electrochemical carbon oxides (CO2 and CO) evolution rate (Q(CO2+CO)) are calculated by the difference between the outlet flow rates with current switched on and the flow rates with current switched off, as eqs 7 and 8:5

QCO2 ) (XCO2QTotal)I-on - (XCO2QTotal)I-off

(7)

Q(CO2+CO) ) [(XCO2 + XCO)QTotal]I-on [(XCO2 + XCO)QTotal]I-off

(8)

in which the XCO2 and XCO are the concentrations of CO2 and CO in anodic off-gas by gas chromatography analysis, respectively. QTotal is the total anodic off-gas evolution rate measured using the bubble-meter. Accordingly, the electrochemical CO2 yield (YCO2) and total electrochemical product gas yield (Y(CO2+CO)) can be calculated by eqs 9 and 10:5

YCO2 )

Y(CO2+CO) )

QCO2F VmI Q(CO2+CO)F VmI

(9)

(10)

where F (96485 coulomb/mol) is the Faraday constant, Vm (mL/ mol) is the molar volume of CO2 at standard temperature and pressure, and I (A) is the current drawn from the DCFC during the off-gas measurements. The carbon efficiency (ECarbon) is another useful parameter showing the ratio of electrochemical oxidation of carbon via anodic reaction as per eq 1 to the total (chemical and electrochemical) carbon consumption, which can be calculated by eq 11:5

ECarbon )

1 /3QCO2 1 /2(XCOQTotal)I-on

+ 1/3QCO2

× 100%

(11)

where the coefficients are determined by the molar ratios of the gas products to one mole of carbon reactant consumed in eq 1 and the Boudouard reaction (eq 12) for CO2 and CO, respectively.

CO2(g) + C(s) ) 2CO(g)

(12)

3. Results 3.1. Structure and Property of Catalysts. Figures 2 and 3 show the XRD patterns of Ni-Al2O3 and Ni-Cu-Al2O3 catalysts at different stages during the preparation, including the coprecipitated precursors (by XRD with Fe KR radiation), the calcined precursors at 450 °C (with Fe KR radiation), and final reduced catalysts at 700 °C (with Cu KR radiation). The precursors of the two catalysts have a typically hydrotalcitelike anionic clay structure (also known as the Feitknecht compound).15,21 Compared with that of Ni-Al2O3, the crystallization degree of the precursor of Ni-Cu-Al2O3 is fairly low due to an increase in Cu proportion in the precursor.15,27 After calcination at 450 °C, the difference between the XRD patterns of these two catalysts is negligible. However, only the intensive NiO peaks and no CuO or Al2O3 peaks are identified in the calcined Ni-Cu-Al2O3 catalyst, as shown in Figure 3. This feature indicates that CuO and Al2O3 have mixed well with the bulk of NiO in the hydrotalcite-like anionic clay structure. After reduction in hydrogen at 700 °C, the intensive peaks of metallic Ni are detected in both catalysts. For the reduced Ni-Al2O3, a small amount of NiO was detected (see Figure 2) that should be formed in the passivation process. For the reduced Ni-Cu-Al2O3 catalyst, however, there is a Cu-Ni alloy phase (Cu3.8Ni) identified between the Cu metal phase and the Ni metal phase as shown in Figure 3. The TEM morphology of two reduced catalysts, Ni-Al2O3 and Ni-Cu-Al2O3, are shown in Figure 4. In the Ni-Al2O3 catalyst, the size of Ni particles (dark spots) have been roughly estimated to fall within the range of 5-10 nm with aggregation of large irregular bulks, due to the high loading of Ni in the catalysts (see Table 1). Compared with the Ni-Al2O3 catalyst, the Ni-Cu-Al2O3 catalyst shows larger particle size, approximately in the range of 10-20 nm. The difference between the two catalysts is attributable to the different degrees of crystallization of their coprecipitated precursors. As shown in the Figure 4b, the dark Ni-Cu alloy particles are dispersed fairly well on the Al2O3 films (the shallow matters), which is different from the Ni-Al2O3 catalysts in Figure 4a. This may be caused by the hydrotalcite-like anionic clay structure of the Ni-Cu-Al2O3 coprecipitated precursors, in which the octahedral coordinated bivalent (Ni2+ and Cu2+) and trivalent (Al3+) cations are uniformly distributed in the brucite-like layers. After calcination and reduction, the precursors result in a well-mixed paracrystalline metal phase with irreducible domains such as Al2O3.15,21,25 The N2 adsorption/desorption isotherms of Ni-Al2O3 and Ni-Cu-Al2O3 catalysts before and after reduction are presented in Figure 5. All of these catalysts principally show the type-IV characteristics, with a hysteresis loop starting from the medium relative pressures and closing near P/P0 ) 1. However, the reduced catalysts show a downward shift in the volumes, (27) Avdeeva, L. B.; Goncharova, O. V.; Kochubey, D. I.; Zaikovskii, V. I.; Plyasova, L. M.; Novgorodov, B. N.; Shaikhutdinov, S. K. Appl. Catal., A 1996, 141, 117–129.


Figure 2. XRD patterns of Ni-Al2O3 catalyst: (1) Ni6Al2(OH)16CO3 · 4H2O, (2) NiO, and (3) Ni.


Figure 5. N2 adsorption/desorption isotherms of Ni-Al2O3 and NiCu-Al2O3 catalysts before and after reduction. Table 3. Textural Properties of Ni-Al2O3 and Ni-Cu-Al2O3 Catalysts before and after Reduction catalysts Ni-Al2O3 Ni-Cu-Al2O3

Figure 3. XRD patterns of Ni-Cu-Al2O3 catalyst: (1) Ni6Al2(OH)16CO3 · 4H2O, (2) NiO, (3) Cu, (4) Ni, and (5) Cu3.8Ni.

before reduction reduced at 700 °C before reduction reduced at 700 °C

SBET (m2/g) Vtotal (mL/g) Dpore (nm) 143 49 156 58

0.467 0.212 0.489 0.238

7.4 13.1 6.3 12.4

the sizes of reduced Ni catalyst particles (5-10 nm), as shown in Figure 4. This difference indicates the aggregation of small Ni particles into big ones during the decomposition of methane. Compared with A500 and A600, more twisted carbon filaments with smaller diameter distribution (20-50 nm) are presented in B600 and B700. It may be explained by the quasiliquid state of Ni-Cu particles in the reaction at 600-700 °C, which were readily cut into small particles by the growing carbon layers,15 or owing to the formation of “octopus” type CNFs on the Ni-Cu-Al2O3 catalysts as reported by other investigators.19-21,27 The oxidation of B700 in high temperature electrolytes ultimately results in short and sintered filamentous fragments, as shown in Figure 6, panels e and f. No obvious differences in

Figure 4. TEM images of reduced catalysts (a) Ni-Al2O3 and (b) Ni-Cu-Al2O3.

probably due to the shrinkage of pores during reduction at 700 °C. The derived values of BET surface area, total pore volume, and average pore diameter of these catalysts are summarized in the Table 3. At the same preparation conditions, the catalyst Ni-Al2O3 shows slightly lower surface area and pore volume than Ni-Cu-Al2O3. However, the surface area and pore volume of the two catalysts dramatically decreased after the reduction at 700 °C, which is consistent with other reports.20,27 3.2. Morphology of CNFs. As shown in Figure 6, the SEM morphologies of the CNFs are highly affected by the catalyst types and the electrochemical oxidation conditions, rather than the synthesis temperature. For A500 and A600, most of the CNFs are typical bent filaments with diameter distributions of 50-100 nm, and each of them grows from one catalytic particle. However, the diameters of these CNFs are not consistent with

Figure 6. SEM images of CNFs (a) A500, (b) A600, (c) B600, (d) B700, (e) B700-1, and (f) B700-2.

3726


Li et al.

Figure 9. N2 adsorption/desorption isotherms of four as-synthesized CNFs.

Figure 7. XRD patterns of four as-synthesized CNFs.

Figure 8. XRD patterns of B700, B700-1, and B700-2. Table 4. Crystalline Parameters and Electrical Conductivities of All Samples XRD carbon fuels

d002 (nm)

Lc (nm)

La (nm)

electrical conductivity (S/cm)

A500 A600 B600 B700 B700-1 B700-2

0.345 0.343 0.343 0.341 0.339 0.339

4.8 5.6 5.4 7.1 8.7 8.8

9.1 9.8 8.5 8.6 8.8 9.0

3.1 3.9 3.4 4.1 4.2 4.5

macroscopic structure can be seen between B700-1 and B700-2 due to the low resolution of the SEM images. However, it will be shown that the oxidation time affects the structure and surface chemistry as well as the concomitant reactivity of the B700 in the following sections. 3.3. Graphitic Structures of CNFs. XRD patterns (with Co KR radiation) of four as-synthesized CNFs are shown in Figure 7. The (002) diffraction peak of CNFs was clearly observed around 30.5°, and the broad (100) diffraction peak can be found around 50°. In addition, a weak (004) diffraction peak at about 63.5° was also visible. These peaks indicate that all of CNFs are highly graphitical. Interestingly, nearly no diffraction peaks corresponding to metal catalysts are observed, due to the high yields (106.7-254.3 g/gcatalyst) of the CNFs on the catalysts as shown in Table 2. The XRD patterns of B700 before and after electrochemical tests in the DCFC (see Figure 8) shows that there are almost no changes in their XRD patterns, indicating that the graphite-like structure of B700 was stable during the electrochemical oxidation in high temperature molten carbonates. Table 4 compares the quantitative crystalline parameters of all CNFs, including the interplanar spacing (d002), the average diameter (La) and the stacking height (Lc). The d002 values of these CNFs vary in a narrow range of 0.339-0.345 nm, which are barely higher than that of ideal graphite (0.335 nm). The

Lc and La values are in the ranges of 4.8-8.8 nm and 8.5-9.8 nm, respectively, indicating the typical turbostratic graphitelike structure for these CNFs. It is well-known that the microstructures of CNFs are highly dependent on the catalyst type and synthesis temperature.13,28 With a rise in the methane decomposition temperature, both La and Lc of the CNFs are elevated, whereas the d002 values are correspondingly lowered. At the same decomposition temperature of 600 °C, the A600 has a slightly larger crystallite size than B600, which should be caused by the changes in both orientations and sizes of the grown graphitic layers when doping the Cu into Ni-Al2O3 catalyst systems.20,28,29 It is also found that electrochemical oxidation of B700 in the DCFC leads to a decrease in d002 and a slight increase in the crystallite size (Lc and La), due to the preferential burning off of the disordered carbon. To further confirm the degree of graphitic structure of the CNFs, electrical conductivity measurements were also undertaken. It is well-known that ordered carbons have a higher electrical conductivity due to their more perfect two-dimensional (2D) crystalline structure. The bulk conductivity values of all CNFs are listed in the last column of Table 4. The CNFs synthesized at higher temperatures such as B700 show high conductivity due to their higher degree of graphitic structure. At the same decomposition temperature of 600 °C, B600 shows slightly lower conductivity than A600, in agreement with the XRD results. As expected, the conductivity of B700-2 is slightly higher than B700 due to the electrochemical oxidation of the disordered carbon. 3.4. Textural Properties of CNFs. As shown in Figure 9, the N2 adsorption/desorption isotherms of four as-received CNFs show type-IV characteristics, which are consistent with the isotherms of their catalysts as shown in Figure 5. The patterns of the hysteresis loops seem to be type H3, which is often associated with slit-shape mesopores. The derived values of BET surface area, total pore volume, and average pore diameter of these CNFs are summarized in Table 5. It is found that the BET surface area varies over a wide range between 76 and 286 (m2/ g), depending on the catalyst type and the methane decomposition temperature. The highest surface area value is obtained on sample B600, which is attributed to changes in the microstructure (i.e., crystallographic orientation and morphology) of CNFs after using a Ni-Cu-Al2O3 catalyst as described previously. For the CNFs synthesized on the same catalyst, an elevation on the synthesis temperature can decrease the BET surface area (28) Reshetenko, T. V.; Avdeeva, L. B.; Ismagilov, Z. R.; Chuvilin, A. L.; Fenelonov, V. B. Catal. Today 2005, 102-103, 115–120. (29) Reshetenko, T. V.; Avdeeva, L. B.; Ismagilov, Z. R.; Pushkarev, V. V.; Cherepanova, S. V.; Chuvilin, A. L.; Likholobov, V. A. Carbon 2003, 41, 1605–1615.



Table 5. Textural Properties of All CNFs sample

SBET (m2/g)

Vtotal (mL/g)

Dpore (nm)

A500 A600 B600 B700 B700-1 B700-2

88 76 286 194 146 132

0.222 0.159 0.485 0.384 0.278 0.266

8.4 9.3 7.8 8.1 8.3 8.4

and the total pore volume, but increase the average pore size. This is ascribed to the production of ordered CNFs at higher temperatures.20,29 Figure 10 compares the N2 adsorption/desorption isotherms of B700 before and after usage in the DCFC. The three samples principally show the same type-IV characteristic, with hysteresis loop starting from medium relative pressures and closing near P/P0 ) 1. The derived textural values are summarized in the last three rows of Table 5. Compared with as-synthesized B700, both B700-1 and B700-2 have a decreased surface area and pore volume. This may be caused by the preferential burning off of highly reactive disordered carbon or the collapse of some pore walls during electrochemical oxidation of the CNFs in the DCFC. 3.5. Chemical Reactivity of CNFs in Air. Thermogravimetric analysis in an oxidative atmosphere is an efficient method to evaluate the relative oxidation activity of carbon materials. Figure 11 presents the variation of weight as a function of temperature for the four as-synthesized CNFs. It can be seen that there is a small amount of weight loss below 100 °C due to desorption of physisorbed water. Significant weight loss starts at 450-550 °C and ends around 740-850 °C, which demonstrates the oxidation of the carbonaceous material. The lowest on-set oxidation temperature in these four CNFs was found in sample A500 at about 450 °C. The on-set of carbon oxidation

Figure 10. N2 adsorption/desorption isotherms of B700, B700-1, and B700-2.

Figure 11. TGA weight loss curves of four as-synthesized CNFs.

Figure 12. TGA weight loss curves of four B700, B700-1, and B700-2.

temperature increases with an elevation in the CNF synthesis temperature. For example, on-set oxidation temperature of A600 is higher than that of A500. It is also noticed that the B series of samples (B600 and B700) show higher oxidation rates compared to the A series of samples (A500 and A600), which may be attributed to the lower degree of graphitic structure and higher surface area of the B series compared to the A series. Figure 12 shows the weight loss curves of B700 series before and after electrochemical tests in the DCFC. The on-set of oxidation of as-synthesized B700 starts at 550 °C and ends around 800 °C. Both the on-set and end temperatures for B700-1 in TGA are significantly lower than those of B700, but slightly higher than those of B700-2, indicating that electrochemical oxidation in the DCFC makes the CNFs more reactive. 3.6. Surface Oxygen Groups on CNFs. TPD is one of the most convenient techniques to investigate the stability and nature of surface oxygen groups on carbon samples. During the TPD process, the surface oxygen complexes would release CO2 and CO at different temperatures depending on their stabilities. Figure 13 presents the CO2 and CO evolution profiles of all CNF samples. In the CO2 evolution spectra, the B700-2 and B700-1 generate the highest levels of CO2 in all CNFs with a maximum peak at 300 °C and a shoulder peak at around 600 °C. On the contrary, A600 presents the lowest evolutions of CO2. A500, B600, and B700 yield a comparable evolution of CO2 with a broad peak at around 200-500 °C, and B600 produces the most CO2 in the four as-synthesized CNFs. In the CO evolution curves, there is no obvious CO peak until 450 °C for all CNFs. Interestingly, the order of the CO production is nearly the same as CO2 production, but the differences in CO production by different samples is much smaller compared to CO2 production. It is known that a charge on the surface of carbon can arise from the interaction between the carbon and an aqueous solution, and this can be used as a convenient index for the surface acidity of carbons. The acidity of the carbon surface is usually related to the preponderance of oxygen functional groups such as carbonyls, phenols, lactones, and anhydrides30,31 that give rise to the evolution of CO2 and CO during TPD. The surface charge distribution of carbons are most often assessed by two methods, mass titration, and electrophoresis, both of which provide a good indication of the character of surface oxygen complexes.32 According to the procedure described by Noh and Schwarz,26 (30) Boehm, H. P. Carbon 2002, 40, 145–149. (31) Wang, S.; Lu, G. Q. Carbon 1998, 36, 283–292. (32) Radovic, L. R.; Rodriguez-Reinoso, F. Chem. Phys. Carbon 1997, 25, 243–358.

3728


Li et al.

Figure 14. Polarization curves of 5% CNFs at 800 °C with 600 rpm stirring conditions. Table 7. Electrochemical Data of All CNFs at Different Temperatures with 600 rpm Stirring operation conditions

A500 A600 B600 B700 B700-1 B700-2

600 °C 600 rpm OCV (V) -0.90 -0.89 -0.92 -0.91 3 6 4 I at -0.8 V (mA/cm2) 5 14 17 18 I at -0.5 V (mA/cm2) 16 I at -0.2 V (mA/cm2) 24 23 29 26 PMax (mW/cm2) 8 7 11 9

Figure 13. TPD data of all CNFs: (a) CO2 evolution profiles and (b) CO evolution profiles. Table 6. XPS and PZC Analyses of Surface C1s and O1s Concentrations of All Carbon Samples sample

O1s (%)

C1s (%)

O/C

PZC

A500 A600 B600 B700 B700-1 B700-2

1.33 0.94 1.88 1.47 2.05 3.63

98.67 99.16 98.12 98.53 97.95 96.37

0.013 0.009 0.019 0.015 0.021 0.038

9.56 9.88 7.52 7.92 4.91 4.17

the PZC values in the mass titration can be shown to change systematically with the extent of oxidation of carbon. Usually, the more oxidized the carbon, the lower is its PZC value. As shown in the last column of Table 6, the PZC values of all CNFs decrease in the order of A600 > A500 > B700 > B600 > B700-1 > B700-2, which is in agreement with the amount of surface oxygen functional groups in the TPD tests. The atomic concentrations of C1s and O1s on the surface of CNFs measured by survey scans in XPS are also shown in Table 6. The highest surface oxygen values were found in B700-2 and B700-1. This agrees well with the TPD results that much more CO and CO2 are desorbed after the B700 was electrochemically oxidized in DCFC. For the four as-synthesized CNFs, B600 has the highest surface oxygen content of 1.88%, whereas the A600 has the lowest value, only 0.94%. The A500 and B700 surfaces have moderate surface oxygen amounts of 1.33% and 1.47%, respectively. On the basis of the TPD, PZC, and XPS results, we may conclude that the amount of surface oxygen complexes on these CNFs surface follow an order: B700-2 > B700-1 > B600 > B700 > A500 > A600. 3.7. Electrochemical Reactivity of CNFs in the DCFC. Figure 14 shows the anodic polarizations curves of various CNFs with 5% loading operated at 800 °C in the DCFC. Generally, the polarization curves of these CNFs are similar in shape. The curves all drop steeply from OCV to around -0.8 V due to activation losses. Following that, a more stable linear region

-0.98 8 20 40 10

-1.00 8 19 43 11

OCV (V) I at -0.9 V (mA/cm2) I at -0.6 V (mA/cm2) I at -0.3 V (mA/cm2) PMax (mW/cm2)

700 °C 600 rpm -1.12 -1.07 -1.14 10 5 8 24 19 29 32 27 36 15 11 18

-1.11 7 25 38 16

-1.16 11 29 55 18

-1.21 10 26 57 20

OCV (V) I at -0.9 V (mA/cm2) I at -0.6 V (mA/cm2) I at -0.3 V (mA/cm2) PMax (mW/cm2)

800 °C 600 rpm -1.22 -1.16 -1.23 17 16 19 45 32 55 67 48 77 27 20 34

-1.21 14 46 59 29

-1.28 26 62 84 38

-1.30 32 66 82 42

appears at middle range of potential (ca. from -0.8 to -0.4 V), which indicates that anodic polarization is under significant ohmic resistance control. Finally, the potential decreases sharply at high current density as fuel is consumed faster than it is supplied to the electrode, which is known as the mass transport limitation. Table 7 provides quantitative comparisons of OCV, current density (i) at a given potential and the maximum theoretical power density (PMax) of all CNFs at 600-800 °C in the DCFC. For the four as-synthesized CNFs, B600 always shows the best electrochemical reactivity, which is demonstrated by the most negative OCV, highest current density, and highest PMax value. In contrast, the A600 shows the lowest electrochemical reactivity at different given potentials and different temperatures. The A500 and B700 samples are equally reactive, but the former shows slightly higher current densities than the latter at the high anode potential region (from OCV to -0.8 V), probably due to the more disordered microstructure of A500. It is also noticed that the discharge rate of B700 can be effectively increased after electrochemical oxidation in the DCFC. At 800 °C especially, the current density of B700-2 is 30-50% higher than the as-synthesized B700 at a given potential, and the PMax value increases to 42 mW/cm2. In summary, the electrochemical reactivities of all CNFs increase in an order: A600 < B700 ≈ A500 < B600 < B700-1 < B700-2. 3.8. Anodic Off-gas Analysis. To study the anodic electrode kinetics, Table 8 compares the anodic off-gas (from the WE (33) Sasaki, K.; Kunai, A.; Sada, T. Denki Kagaku 1980, 48, 311–314.



Table 8. Anodic Off-gas Data of All CNFs in the DCFC at 600-800 °C with 600 rpma XAr (%) 600 °C 700 °C 800 °C 600 °C 700 °C 800 °C 600 °C 700 °C 800 °C a

I off I ) 75 mA I off I ) 75 mA I off I ) 100 mA I off I ) 75 mA I off I ) 75 mA I off I ) 100 mA I off I ) 75 mA I off I ) 75 mA I off I ) 100 mA

A500 98.33 97.66 93.29 92.48 92.63 90.35 A600 98.37 97.83 93.42 92.86 92.77 90.57 B600 98.35 97.64 93.40 92.64 92.69 90.56

XCO (%)

XCO2 (%)

0.33 0.36 3.74 3.78 6.54 8.50

1.34 1.98 2.97 3.74 0.63 0.95

0.33 0.35 3.66 3.70 6.41 8.33

1.31 1.82 2.91 3.44 0.62 0.90

0.33 0.36 3.60 3.48 6.48 8.30

1.32 2.00 3.00 3.88 0.63 0.94

XAr (%) 600 °C 700 °C 800 °C 600 °C 700 °C 800 °C 600 °C 700 °C 800 °C

I off I ) 75 mA I off I ) 75 mA I off I ) 100 mA I off I ) 100 mA I off I ) 100 mA I off I ) 150 mA I off I ) 100 mA I off I ) 100 mA I off I ) 150 mA

B700 98.40 97.73 93.45 92.76 92.90 90.35 B700-1 97.88 97.09 93.89 92.60 92.37 90.22 B700-2 98.11 97.18 93.92 92.41 92.78 90.24

XCO (%)

XCO2 (%)

0.32 0.37 3.69 3.65 6.29 8.54

1.28 1.90 2.86 3.59 0.61 0.91

0.35 0.38 3.30 3.57 6.86 8.71

1.77 2.53 2.81 3.83 0.67 0.98

0.35 0.37 3.37 3.49 6.43 8.53

1.54 2.45 2.72 4.00 0.69 1.13

XAr, XCO, and XCO2: Concentration of Ar, CO, and CO2, respectively, in anodic off-gas by gas chromatography analysis.

Figure 15. Electrochemical CO2 yield (YCO2) of 5% CNFs at 600-800 °C.

Figure 16. Electrochemical CO2 and CO yield (Y(CO2+CO)) of 5% CNFs at 600-800 °C.

compartment) data of all CNFs at different temperatures with a 600 rpm stirring rate in the DCFC. The off-gas composition was analyzed by gas chromatography with the current switched off to give a baseline, and with the current switched on to measure the gases produced by the electrochemical reaction. Only the purge gas argon (Ar) and the carbon oxides (CO and CO2) were found at measurable levels using gas chromatography. It was found that the CO concentration (XCO) increases obviously with an increase of fuel cell temperature; however, the CO2 concentration (XCO2) increases slightly from 600 to 700 °C and decreases sharply at 800 °C. This phenomenon may be caused by the Boudouard reaction eq 12, which can easily occur at temperatures above 700 °C.5,6 As shown in Figures 15 and 16, both the electrochemical CO2 yields (YCO2) and the sum of CO2 and CO yields (Y(CO2+CO)) of all CNFs are highly dependent on the operational temperature of the DCFC. At 600 °C, the YCO2 is around 0.40-0.55 and Y(CO2+CO) ranges from 0.50 to 0.60 for the CNFs. These values are slightly lower than the theoretical value YCO2 ) 0.75, as shown in the main anodic reaction eq 1. This may be caused by an insufficient anodic electrochemical reaction of these CNFs with high degrees of graphitic structure. At 700 °C, the YCO2 and Y(CO2+CO) increase to 0.55-0.80 and 0.70-0.90, respectively. These data agree well with the theoretical YCO2 ) 0.75 and indicate that the main product for electrochemical oxidation of carbon at 700 °C is CO2 rather than CO. When the cell temperature is elevated to 800 °C, the YCO2 decreases sharply to 0.10-0.20 and Y(CO2+CO) increases to 0.90-1.60. This may

be caused by the Boudouard reaction (eq 12), or possibly the electrochemical oxidation of carbon at 800 °C, and may be explained by the following reactions:6,33 C + CO23 ) CO2 + CO + 2e

(13)

2C + CO23 ) 3CO + 2e

(14)

in which the theoretical values of Y(CO2+CO) are 1.0 and 1.5, respectively. At the same operational conditions, B600 presents the highest YCO2 value in the four as-synthesized CNFs, which means B600 has the highest electrochemical reactivity and carbon efficiency as will be discussed in the following test. On the contrary, the lowest YCO2 value at different temperatures is found in A600, implying its lowest electrochemical reactivity in these four CNFs. It is also noticed that B700-1 has a higher YCO2 value than that of B700, but lower than that of B700-2, further confirming that pretreatment by electrochemical oxidation makes CNFs more reactive in the DCFC. In Figure 17, the carbon efficiencies (ECarbon) of all CNFs are calculated by eq 11, which indicates the ratio of electrochemical oxidation of carbon via the proposed anodic reaction (eq 1) to the total carbon consumption. With an increase of temperature from 600 to 800 °C, the ECarbon values decrease obviously for all CNFs due to the chemical loss of carbon by the Boudouard reaction at higher temperature as described earlier. Especially at 800 °C, the ECarbon values are less than 5%, indicating more than 95% of carbon fuels were heavily consumed by the

3730


Figure 17. Carbon efficiencies (ECarbon) of 5% CNFs at 600-800 °C with 600 rpm stirring.

Boudouard reaction. In the four as-synthesized CNFs, the highest ECarbon values are found in the B600 series at different temperatures, and the lowest ECarbon values belong to the A600 series. However, A500 and B700 show comparable ECarbon values. The ECarbon can be increased after B700 was electrochemically oxidized in the DCFC. These results further confirm the order of electrochemical reactivities of various CNFs, as shown in Section 3.7. 4. Discussion

Li et al.

As discussed in our previous reports, the nature of carbon fuels plays an important role in the anodic performance of the DCFC.36,37 Usually, a higher surface area (especially the mesoporous area) or pore volume in carbon fuel can effectively improve its electrochemical reactivity by increasing the interaction between the carbon particles and the molten carbonate electrolyte. Similarly, a carbon fuel with smaller particle size also shows a higher electrochemical reactivity due to its comparative high surface area. On the contrary, a fuel with a higher degree of graphitic carbon results in a lower electrochemical reactivity in the DCFC due to the less reactive sites (such as edges and defects) on carbon surface, although it may exhibit better electrical conductivity. Furthermore, the quantity of the surface oxygen functional groups directly affects the electrochemical discharge rate of carbon fuels in the DCFC, which is consistent with the assumed anodic electrochemical mechanism as shown below:1 22CO2Dissociation of carbonate salts 3 f 2CO2 + 2O (15)

CRS + O2- f CRSO2- First adsorption

(16)

CRSO2- f CRSO- + e- Fast discharge

(17)

CRSO- f CRSO + e- Fast discharge

(18)

CRSO + O2- f CRSO22 Slow adsorption

(19)

CRSO22 f CRSO2 + e Fast discharge

(20)

From the experimental results, we found that the microstructure and electrochemical reactivities of CNFs are highly dependent on their processing conditions, such as catalyst compositions and synthesis temperatures. Compared with the A series CNFs, the B series shows relatively higher reactivities in the DCFC tests. This can be explained by significant differences in the physical and chemical properties (such as morphologies, crystalline structures, and textual properties) of the B series of CNFs by using Cu doped into the Ni-Al2O3 catalyst. As shown in Table 2, the primary role of Cu additives is to increase the yield of CNFs (almost twice the carbon yield than those on a Ni-Al2O3 catalyst) during the methane decomposition. It was suggested that Cu has a high affinity with the carbon layer and may reduce the encapsulation rate of the Ni-Cu-Al2O3 catalyst by carbon layers.14,15 Consequently, more CNFs are formed on the Ni-Cu-Al2O3 catalyst without deactivation for long periods. More importantly, it is noted that the Ni-Cu-Al2O3 catalyst essentially influences the morphologies and textural properties of B600 and B700, which show twisted carbon filaments with smaller diameter distributions than those of A500 and A600 (see Figure 6). These morphologies of the B series samples further determine the higher surface areas and pore volumes compared to those of the A series, as shown in Table 5. In addition, the crystallite sizes of the CNFs are slightly decreased by the Ni-Cu alloy catalyst due to changes in the orientation and growth mechanism of carbon layers.20,28,29 Meanwhile, the higher decomposition temperature results in more ordered CNFs formed on the catalysts, which are demonstrated by the growth of graphitic structures with decreasing surface area (or pore volume) in XRD and N2 adsorption/desorption tests. The number of surface oxygen groups on the CNFs can be also limited by the higher synthesis temperature, which is consistent with other reports.13,34,35

where the CRS is the a reactive site on the carbon fuel surface such as edge or defect, and the O2- is a free oxide ion dissociated from the molten carbonate electrolytes at higher temperature. Obviously, more CO2- or CO-yielding functional groups provide a higher degree of reactive sites, making the carbon fuels more reactive in the DCFC. In the four as-synthesized CNFs, B600 shows the highest electrochemical reactivity in the DCFC, which is attributed to its highest surface area and largest number of surface oxygen functional groups in these four CNFs. In contrast, A600 has the lowest electrochemical reactivity in these samples, which is associated with its lowest surface area and smallest number of surface oxygen functional groups as shown in Tables 5 and 6. These results are in good agreement with our previous reports as mentioned previously.36,37 The improvement of electrochemical reactivities of B700-1 and B700-2 can be explained by a dramatic increase in surface oxygen functional groups after treatment in high temperature alkali carbonate melts in the DCFC. As shown in Table 6, the surface atomic oxygen concentration of B700-1 and B700-2 increase around 50-150%, although their surface area decreases about 20-30% (see Table. 5). In the TPD tests, the two of treated B700 samples produce larger amounts of CO2 and CO than as-synthesized B700, which indicates more free reactive sites have been produced on the surface of CNFs. Interestingly, the order of the CO2- or CO-yield functional groups on the surfaces of all

(34) Zhou, J. H.; Sui, Z. J.; Zhu, J.; Li, P.; De, C.; Dai, Y. C.; Yuan, W. K. Carbon 2007, 45, 785–796. (35) Li, J.; Vergne, M. J.; Mowles, E. D.; Zhong, W.-H.; Hercules, D. M.; Lukehart, C. M. Carbon 2005, 43, 2883–2893.

(36) Li, X.; Zhu, Z. H.; De Marco, R.; Dicks, A.; Bradley, J.; Liu, S. M.; Lu, G. Q. Ind. Eng. Chem. Res. 2008, 47, 9670–9677. (37) Li, X.; Zhu, Z. H.; Chen, J. L.; De Marco, R.; Dicks, A.; Bradley, J.; Lu, G. Q. J. Power Sources 2009, 186, 1–9.

CRSO2 f CO2(g) + e Fast discharge and evolution

(21)


CNFs is consistent with the order of the electrochemical reactivities of the CNFs in the DCFC (see Table 7), which further confirms that the amount of surface oxygen functional groups on CNFs plays a dominant role in the anodic electrochemical reaction in the DCFC. In this study, the preoxidation effect of molten carbonates on changing the electrochemical reactivity of carbon fuels is similar to that of HNO3 pretreatment as given elsewhere.37 However, it is also known that various alkali metals (especially lithium) have a catalytic effect on the gasification of carbon due to the formation of the surface active intermediates.14 5. Conclusion Graphitic CNFs were prepared by catalytic decomposition of methane over the Ni-Al2O3 and Ni-Cu-Al2O3 catalysts at 500-700 °C. It was found that the microstructures and electrochemical reactivities of CNFs are highly dependent on the decomposition temperature and catalyst compositions. Lower decomposition temperatures lead to a decrease in the carbon


crystallite size but an increase of surface areas (or pore volumes) as well as the number of surface oxygen functional groups. As a result, the electrochemical reactivities of CNFs can be effectively improved in the DCFC tests. Introducing Cu into the Ni-Al2O3 catalyst not only increases the yields of CNFs but also improves their electrochemical reactivities, due to the higher surface area and larger amounts of surface oxygen functional groups on these CNFs. Therefore, it is important to develop highly active catalysts to synthesize CNFs with large surface area and surface oxygen functional groups at a relatively low decomposition temperature. The electrochemical reactivities of B700-1 and B700-2 are highly promoted by surface oxidation in the high temperature molten carbonates, further confirming the importance of the surface oxygen groups on carbons to their electrochemical performances in DCFC. Acknowledgment. Financial Support from ARC (Australian Research Council) discovery project is greatly appreciated. EF900203H

Carbon Nanofibers Synthesized by Catalytic Decomposition of

Recommend Documents