Applications of Carbon Nanostructures to Fuel Cell Technology

Transcription

1 Applications of Carbon Nanostructures to Fuel Cell Technology Fuel Cell Technology Jason Killgore, Zachary Budiselic Fuel cells are not a new technology as some have come to believe. The concept of a battery-like system consisting of separate chambers of hydrogen and oxygen has existed since the middle 19 th century. In fact, the first demonstration of a fuel cell was by William Grove in 1839 [1]. In his demonstration, Grove showed an apparatus with separate chambers, each of which was half filled with water and immersed in a water bath. A platinum electrode was also placed in each chamber and connected to a current source. Then an electrical current was passed through the cathodes, which led to the water being electrolyzed into hydrogen and oxygen gas. Next, when the current source was removed and replaced by an ammeter, an electrical current was produced as the hydrogen and oxygen gasses combined and turned back into water. Of course, over the last 150 years this technology has been refined to its present state, a much more compact and efficient system. Along with the basic hydrogen fuel cell came other variations on this method of producing electricity. Among the most common and highly researched fuel cells are the PEM fuel cell, solid oxide fuel cell, phosphoric acid fuel cell, molten carbonate fuel cell, and alkaline fuel cell. These types of fuel cells are explained in the next few paragraphs. The solid oxide fuel cell (SOFC) utilizes a similar structure to the PEM fuel cell except that the electrolyte is an ion-conducting ceramic material and the ion crossing the electrolyte is oxygen instead of a proton. One advantage to this fuel cell is that it only deals with gas and solid whereas the PEM must account for liquid, gas, and solid as water passes through the polymer membrane as well. Another advantage is that the SOFC can process carbon monoxide as a fuel source instead of just hydrogen. Although one may think of the SOFC as a better fuel cell system, two factors lead to problems. First, the SOFC runs at extremely high temperatures between 800 and 1100 o C and secondly, the fuel cell is very large and thus is not feasible for automobile use. It is mainly used for power generation in place of other sources. The molten carbonate fuel cell (MCFC) utilizes a molten mixture of alkali metal carbonates as an electrolyte and also runs at high temperatures between 600 and 700 o C. This fuel cell functions by allowing carbonate ions to travel across the electrolyte. These fuel cells are also used for power generation purposes and can be found in some cogeneration plants.

2 The phosphoric acid fuel cell is similar to a PEM cell in that is uses a proton-conducting electrolyte and the reactions occur in the anode and cathode. Platinum is also used for the catalyst. The main difference is that the electrolyte is an inorganic acid, concentrated phosphoric acid (100%). Unfortunately, this type of fuel cell will not run at the low temperature of a PEM cell and functions best at temperatures above 150 o C. These are mainly used for power generation. The alkaline fuel cell uses a mobile electrolyte, which is pumped around the cell and facilitates the reaction of the hydrogen and oxygen by transferring hydroxide ions from anode to cathode. An interesting use for this fuel cell was in the first manned spacecraft sent up by NASA. The PEM was the first choice, but the alkaline fuel cell was the first functional fuel cell and thus it was used instead. The PEM, proton exchange membrane, fuel cell is of great interest to automobile manufacturers. It utilizes a compact sandwich of a polymer electrolyte between the anode and cathode plates and a platinum catalyst. This polymer allows the proton from a reacted hydrogen atom to travel from the anode to cathode thus releasing an electron and giving rise to an electrical current. When the proton reaches the cathode side, it combines with an oxygen molecule and the lost electron to form a water molecule. shows a schematic of a PEM fuel cell. The most recent push towards making small fuel cells has led to the increased research on the PEM fuel cell. Several reasons for the use of Figure 1: Schematic of a PEM fuel cell. Courtesy of [2] this particular fuel cell exist. It is possible to make a PEM fuel cell small enough to fit in an economy car chassis and even smaller applications. PEM cells run at a much lower temperature (80 o C) than other fuel cells (150 o C and above). This advantage is mainly due to the unique nature of the polymer membrane currently used. Another advantage of the PEM fuel cell is that it does not require potentially toxic chemicals in its system in order to function and the main exhaust product is water.

3 ChemE 554 Applications of Carbon Nanostructures to Fuel Cell Technology Killgore/Budiselic Carbon Nanotubes In order for fuel cell technology to reach the mainstream, a number of important obstacles must be overcome. Many researchers believe that nanotubes could be a way of overcoming certain obstacles. From the time carbon nanotubes were first synthesized, scientists and engineers have sought practical uses for these unique new molecules. Proposed applications for nanotubes have ranged from semiconductor technology to aerospace structures. Along the way, considerable attention has been paid to the potential role of nanotubes in fuel cell applications. In order to understand the functions that carbon nanotubes (CNTs) might play in fuel cells, it is useful to first examine what exactly CNTs are. Carbon nanotubes were developed as a variation on the C60 Buckminsterfullerene molecule first discovered in by Krätchmer in 1985 [3]. When C60 molecules were discovered, they invigorated new research into all carbon molecules. C60 molecules consist of 12 pentagonal faces bonded to a total of 20 hexagonal faces. If the C60 molecule is cut in half, it forms two semispheres. If these semispheres are joined by a band of hexagonal carbons, a tube is formed. The smallest tubes could have only a single band, while most have many bands, yielding lengths on the micrometer scale. All Figure 2: A number of different carbon allotropes have been structures exhibiting a synthesized, each showing certain unique properties. Courtesy of [3] combination of hexagonal and pentagonal carbon faces create the fullerene family. Fullerenes were originally discovered as an accidental byproduct of electric arc graphite evaporation. As the aspect ratio of the tube increases, properties range from insulator to semiconductor to metallic. Single molecule fullerene tubes are referred to as single wall nanotubes. Concentric layers of carbon nanotubes can also be synthesized, yielding multi-wall nanotubes. Multi walled nanotubes exhibit far more imperfections than single walled nanotubes. The broad range of unique and tailorable properties has spurred research into many seemingly dissimilar areas of science and engineering.

4 One major attraction of fuel cells is their promise of outstanding efficiency. Nanotubes are being explored to enhance and aid in this efficiency. The unique properties of nanotubes could have numerous applications in fuel cells, although two applications have dominated fullerene based fuel cell research. The first is the use of nanotubes as catalyst supports, and the second is the use of nanotubes as a hydrogen storage medium. In both areas, nanotubes have exhibited combinations of desirable properties unattainable with other materials. Nanotubes as Catalyst Supports Carbon nanotubes offer themselves as good catalyst supports for a variety of platinum based catalysts. The extremely small size, chemical stability, and high surface area to volume ratio all lead to a much smaller electrolyte size for PEM fuel cells and methanol fuel cells. Carbon nanotubes (CNT) range from 10 to 30 nm diameter and several micrometers in length and consist of highly graphitized multilayer walls. [4] This extremely small size, see Figure x, means that a PEM membrane of the same dimensions may be constructed with the catalyst deposited on the walls of the CNT to facilitate the anode-cathode assembly all in one layer instead of several layers as in present PEM fuel cells. CNTs also exhibit good chemical stability when subjected to various chemical agents. When nanotubes are subjected to any liquids with surface tensions higher than mn/m, no metals or other elements are deposited on the surface. [4] This suggests that the highly graphite nature of CNTs may remain stable when it comes in contact with substances typically used in fuel cells such as the water and gasses associated with PEM fuel cells. This would lead to less risk associated with membrane tearage. Figure 3: TEM image of purified carbon nanotubes [4]. The dimensions of the photo are nm each side Another attribute of CNTs is the high surface area to volume ratio. For example, a carbon nanotube with a 10 nm diameter and length of 4 micrometers would have a surface area to volume ratio of Surface area = 2πrL = 2π(0.5D)L = 126 µm 2, where r is the radius of the nanotube, in nm, D is the diameter in nm, L is the length in nm. Volume = πr 2 L = 314 µm 3. Hence: surface area/volume =

5 Thus, for this size of nanotube, the surface area is almost half of the volume of the tube. This large ratio means that almost half of the nanotube is available for deposition of the catalyst assuming that only the outer part of the tube is used for the catalyst due to the difficulty with inserting catalyst inside the nanotube. Methods for catalyst deposition are currently being researched. Several different methods exist presently as found by Liu, Lin, et al, and Rajesh, Karthik, et al. The methods of Liu, Lin, et al are used to deposit Au, Pt, and Ag nanoclusters on the outer surface of CNTs. The CNTs are first pretreated with a solution of SnCl 2 /HCl and next placed in a platinum electroless plating solution. Finally, the CNTs are ultrasonicated with PtCl 2 and washed in distilled water. Figure x shows different levels of Pt coating according to the age of the pretreating solution. The black spots are platinum particles. The Liu et al experiment found that platinum nanoparticles 1-5 nm in diameter were deposited on multiwalled CNTs by this plating procedure. The aging time of the sensitizing solution had a large impact on the Figure 4: TEM images of Pt coated on CNTs by the twostep process using sensitizing solutions of different ages, (upper left) 1 day, (upper right) 3 days, (lower left) 15 days, (lower right) 15 days, but with fresh SnCl 2 -H 2 O added before use. [4] percentage of platinum nanoparticles deposited with an optimum aging period of 3 days and 15 days with added tin chloride hydrate. They also discovered that, using this method, the Pt/CNT electrocatalysts contained 67.3 % of Pt(0) and 32.7 % of Pt(IV) and that the CNTs with this plating showed high electrocatalytic activity for oxygen reduction in a single stack PEM fuel cell.[4] Figure x shows the polarization curve for the PEM fuel cell. As expected, the sensitized CNTs showed a higher cell voltage to current density than the direct-coated CNTs.

6 Rajesh et al used a different method to actually place the catalysts on the inside of the CNTs. First, the CNTs are immersed in hexachloroplatinic acid and ruthenium chloride. Then the ions are allowed to reduce to the corresponding metal by exposure to hydrogen gas at 823 K. The result is a Pt or Pt-Ru nanocluster loaded CNT. In order to load the inside of the CNTs with WO 3, which later will be bonded to Pt for lower activation energy in direct methanol fuel cells, the CNT is immersed in peroxotungstic acid. This immersion allows the peroxo-tungstic acid to penetrate inside the pores of the CNTs. Finally, the WO 3 /CNT structure is immersed in hexachloroplatinic acid, dried, and reduced with hydrogen gas at 623 K. Thus, the result is a CNT with platinumtungsten-oxide nanoclusters inside. Several conclusions were made from this research. First, the higher activity of the platinum tungsten oxide compared to that of commercial Pt and Pt-Ru catalysts suggests that the higher electrochemical surface are of the carbon nanotube has been effectively used for the dispersion of catalytic nanoparticles which results in a higher activity for methanol oxidation in direct methanol fuel cells. [5] Another conclusion is that the differing amounts of nitrogen in the carbon nanotube affect the hydrophilicity of the nanotube. Thus, water may be better controlled through the catalyst. Several benefits are associated with using nanotubes for catalyst deposition and thus in fuel cell applications. These include smaller overall fuel cells, increased reliability from the stronger nanotube structure, high surface area to volume for catalyst deposition, and thus more sites for reaction to occur and possibly lower activation energies in the case of the platinum-tungsten-oxide nanotubes. Hydrogen Storage in Carbon Nanotubes Figure 5: Polarization curve of different Pt/CNT electrocatalysts Courtesy [5] One of the obstacles that must be overcome in the commercialization of PEM fuel cells is determining a means to store hydrogen. Hydrogen forms a very low density gas under typical atmospheric conditions. With a density on the order of only.1 kg m -3, prohibitively large storage tanks would be required to fuel practical devices. This limitation has sparked considerable amounts of research into methods of storing

7 hydrogen in a more condensed form. A number of ultra-high surface area to volume ratio (SA:Vol) materials have been investigated as potential matrices for hydrogen adsorption. Different metrics are commonly used to describe the storage of hydrogen. Mass and volume percentage are the most commonly referenced metrics. A better measure of effective storage capacity is the desorption capacity. Unlike gravimetric and volumetric storage capacities, desorption capacity is a measure of the amount of hydrogen that can actually be utilized. Figure 6 shows the hydrogen desorption capacity of various materials currently under investigation. Figure 6: Storage capacity of various hydrogen storage materials. Courtesy of Hijikata et al. [6] Ideal materials should allow for high storage capacities at moderate temperatures and pressures. Three general material classes are identified in the figure, metal hydrides, complex aluminum hydrides, and carbon materials. Of these classes, carbon materials display the most attractive characteristics. Not only do carbon materials exhibit the highest desorption capacities, but they are able to accomplish this at low temperatures. Hynek et al. [7] have investigated the adsorption of hydrogen on 10 different traditional carbon surfaces. Carbon sorption is obtained by placing carbon powders, pellets, or granules inside a suitable storage vessel. Under proper conditions, the mass and volume displaced by the carbon can be offset by the additional storage capacity introduced by the high surface area adsorbent. Adsorption occurs as a result of field forces between a solid surface, the adsorbent, and a gas or vapor, the adsorbate. Unlike the purely physical storage of hydrogen by liquefaction, carbon sorption relies on both chemical and physical processes. In an adsorbed state, the hydrogen condenses to a density of approximately 71 kg m -3 [10]. The benefits of carbon sorption are only realized up to a certain critical pressure (Figure 7). Beyond this critical pressure, the adsorbate density becomes so high that it offsets the benefits obtained by carbon sorption. In addition to pressure, as expected with any phase behavior, adsorption properties also depend heavily on temperature.

8 Exceptional adsorbtion properties have been exhibited by two major classes of carbon materials. Both of these materials exhibit critical dimensions on the nanometer and subnanometer scale. It is believed that hydrogen molecules are able to enter these confined areas where a lack of mobility maintains condensed properties even at temperatures above the boiling point. The first class of carbon materials exhibiting these valuable properties is nanostructured graphite [9]. Figure 7: Comparison of hydrogen storage with and without carbon sorption. Courtesy of [7] Nanostructured graphite fibers are comprised of graphite platelets stacked in a variety of directions. Stacking of the platelets can be planar, longitudinal, or diagonal, with a minimum separation of 3.35Å. Hydrogen has a kinetic diameter of 2.89Å, allowing the molecules to penetrate and adsorb between the layers. Under very large pressures, Chambers et al. have reported storage capacities of nearly 70 wt% in nano-graphite. It should be noted however that a number of subsequent studies have questioned these results, obtaining maximum storage capacities an order of magnitude lower [7]. Even at the more typical experimental values, nano-graphite offers exceptional promise, matching very closely with the more extensively studied carbon nanotubes. The perceived benefit of carbon nanotubes lies in their hollow structure. Similar to the inter-platelet absorption in graphite, CNTs are believed to confine hydrogen within their hollow centers. Very little is known about the precise mechanisms by which hydrogen adsorption has been so efficient in CNTs. To better understand the process, considerable research on CNTs has been performed using both theoretical and experimental bases. Research into hydrogen adsorption on carbon nanotubes was sparked by a 1997 report by Dillon et al. [14]. Cobalt and graphite were co-evaporated by electric arc to produce single walled nanotubes. The process led to a mixture of 1.2 nm diameter tubes and about 20 wt% 5-50nm cobalt particles. Temperature programmed desorption (TPD) spectroscopy was used to examine the release of hydrogen from the carbon/cobalt system. Studies were performed that isolated the carbon nanotubes as the key factor in improving storage capacity. Dillon found that whereas hydrogen has a heat of adsorption of 4 kj mol -1 on planar carbon, the CNTs exhibited energies of nearly 20 kj mol -1. This tight binding reportedly allowed for gravimetric storage densities between 5 and 10 wt%. By performing theoretical studies, researchers are able to investigate the adsorption of hydrogen on CNTs without drawing possibly erroneous inferences from macroscopic experiments. The trade off comes in the specific assumptions that must be made about the nature of the interaction. Theoretical studies have sought to determine which factors

9 influence storage capability, as well as seek optimal combinations of those factors. Specific studies have focused on tube diameter, orientation, spacing, dispersion forces and surface chemistry. The effects of tube diameter have been investigated by Rzepka et al. [11] and Darkrim et al. [12]. Rzepka has used grand canonical ensemble Monte Carlo (GCMC) simulation to optimize pore geometry of both nanotubes and nano-graphite. The interaction between hydrogen and carbon was modeled using a 6-12 Lennard-Jones potential (1) with literature values for the constants σ σ U() s 4ε = s s (1) Figure 8 shows the results of the simulation in regards to volumetric and gravimetric storage capacity. Both the graphite and the nanotubes exhibit a distinct maximum volumetric storage capacity at approximately 7Å. Aside from a local maxima at 7Å, gravimetric storage capacity increases monotonically over the range studied. At a separation of 7Å, the hydrogen particle is exposed to surface forces from multiple walls. This leads to an increased densification not experienced at wider separations which allow multiple hydrogen layers to be formed. Figure 8: Predicted storage capacity for nanotubes and nano-graphite structures as a function of pore size. Courtesy of [11] The study by Darkrim has taken a very similar approach, also using Monte Carlo simulation with a Lennard Jones potential. Darkrim s simulation predicted an optimum tube diameter of nm. This was in contrast to Rzepka s results which indicated that above 10 nm two or more hydrogen layers would form, thus decreasing the efficiency of the medium. This discrpency might by explained by the fact that Darkrim used slightly different interaction parameters, although it is also possible that the differences are attributed to the specific program used for the computations. Both simulations were run at the same pressure, although Darkrim s calculations were performed at 293K instead of Rzepka s 300K.

10 In addition to tube diameter, Darkrim s research also probed the effect of tube spacing on storage capacity. Using his predetermined optimum diameter of nm, Darkrim varied the spacing between tubes from.5 to.9 nm. Here it was found that a tube spacing of.7 nm yielded the most efficient storage system. One might think that a close packed arrangement would be most effective, however close packing wastes a considerable amount of adsorption area. A much more comprehensive study of tube spacing as well as packing orientation has been performed by Wang et al [13]. Wang has used a Silvera-Goldman potential to model the hydrogen, with a Crowell-Brown potential to model the hydrogen-carbon interaction. Both triangular and square array configurations have been studied at a variety of separations, again using GCMC. Calculations were performed at 77K and 298K at a pressure of 50 atm. Figure 9: Volumetric (top) and gravimetric (bottom) storage capacities as a function of tube separation distance. Hollow labels represent triangular geometry, solid labels represent square array. Figure on left simulates adsorption at 298K and figure on right simulates adsorption at 77K. Courtesy of [13] Figure 9 shows the results of the simulation by Wang. The data agrees very closely with Darkrim s results, indicating an optimal spacing of 5-7 nm at 298K and 5-9 nm at 77K. The square array yielded the highest volumetric adsorption at low separation distances which are typical in experimentally observed samples. At optimal separations, a triangular array with (9,9) lattice parameters yielded the most productive configuration. Although substantial gains can be obtained by tailoring tube diameter and arrangement, the author cautions that experimentally producing such ideal systems has not yet been possible. The majority of the theoretical studies performed have assumed that the nanotubes are perfectly structured. Volpe 10] has considered how surface imperfections might affect the adsorption chemistry. Two different processes resulting from imperfection were explored. The presence of defects can lead to chemisorption of the hydrogen. The chemisorped hydrogen can not be extracted from the tube, thus reducing useful strorage capacity. The alternative outcome of the defect is that it can lead to surface oxidation. Oxidation creates additional surface area in the defect zone, thus allowing for increased

11 storage capacity. This behavior is shown in figure 5, where especially at low temperatures, oxidation substantially increases storage capacity. Weight efficiency is improved more than 30% at 77K by the inclusion of an oxidized surface. Figure 10: Adsorption isotherms showing pure samples (solid labels) and 40% oxidized samples (hollow labels) at three temperatures. The dashed line represents compressed hydrogen without carbon. Courtesy of [10] The theoretical approaches have suggested a number of critical areas that must be addressed if nanotube based hydrogen storage systems are going to be viable. Unfortunately, the ability to precisely test most of the theoretical predications does not currently exist. Still, these studies do suggest paths for future experiments, as well as explanations for some of the complex phenomena involved in the hydrogen adsorption process. Even in light of the successful studies on improving hydrogen storage capabilities, mainstream use of hydrogen fuels is still a long way off. Bunger and Zittel have comprised a detailed summary of the commercial status of hydrogen storage in carbon nanostructures [15]. In order to achieve fuel tanks comparable in size to those in internal combustion vehicles, 20 wt% efficiencies will be required. Although Chambers demonstrated this capability, it is yet to be repeated, and very far from production scale. As of 2001, commercial nanotubes could be obtained for $ 60/g. In order to store sufficient hydrogen to power a single vehicle, $3 million worth of tubes would be required. Enormous strides in the mass production of nanotubes will therefore be necessary before any chance commercial success exists. In the short term, it is much more likely that devices will have to rely on traditional compressed gas storage. Conclusion When fullerenes were first discovered, researchers promised remarkable properties that would revolutionize numerous branches of science. Fullerenes have been slow to deliver on this promise, although fuel cell applications do appear promising. The role of carbon

12 nanotubes as catalyst supports has provided substantial benefits compared to traditional carbon supports. For use as hydrogen storage devices, carbon nanostructures have broke new ground, unattainable with other technologies. Perhaps the biggest obstacle in commercial nanotubes use is cost. Even for smaller electronic devices, the cost still hinders any near future potential applications. It will likely be a long time before commercial fuel cells make use of nanotubes, yet by studying and understanding nanotubes, researchers may be able to obtain similar benefits with other less expensive materials. 1. Larminie, James and Dicks, Andrew. Fuel Cell Systems Explained. Wiley Publishing, New York Pp. 1-12, 61-62, , current as of 5/20/ current as of 5/12/ Z. Liu, X. Lin, J.Y. Lee, W. Zhang, M. Han, L.M. Gan, Langmuir 18 (2002) B. Rajesh, V. Karthik, S. Karthikeyan, K.R. Thampi, J.M. Bonard, B. Viswanathan, Fuel 81 (2002) T. Hijikata, Intl. Journal of Hydrogen Energy 27 (2002) S. Hynek, W. Fuller, J. Bentley, Intl. Journal of Hydrogen Energy 22 (1997) A. Zuttel, P. Sudan, Intl. Journal of Hydrogen Energy 27 (2002) A. Chambers, C. Park, R. Terry, J. Phys. Chem. B 102 (1998) M. Volpe, F. Cleri, Chemical Physics Letters 371 (2003) M. Rzepka, P. Lamp, M.A. de la Casa-Lillo, J. Phys. Chem. B 102 (1998) F. Darkrim, D. Levesque, Journal of Chemical Physics 109 (1998) Q. Wang, J. K. Johnson, J. Phys. Chem. B 103 (1999) A. C. Dillon, K. M. Jones, T. A. Bekkedahl, C. H. Kiang, D.S. Bethune, M.J. Heben, Nature 386 (1997) U. Bunger, W. Zittel, Applied Physics A 72 (2001)