Optimization of Platinum Nanoparticles for Proton Exchange Membrane Fuel Cells using Pulse Electrochemical Deposition
Ryan Lindeborg, AJ Swoboda
Harvard College ’16, Princeton College ’15
Jonathan Burk, Steve Buratto
University of California, Santa Barbara, Department of Chemistry and Biochemistry
With current energy sources rapidly depleting, the need for clean, efficient energy is greater than ever before. Recently, major research has been conducted to identify energy sources for buildings and automobiles that would also be free of carbon dioxide, the result of the greenhouse effect of fossil fuel use (Litster and McLean, 2004). Among these is the proton exchange membrane fuel cell (PEMFC). This cell is an alternative clean energy source that has the potential to be an important part of the solution to our current and future energy demands. Interestingly, the first use of fuel cells was in the Gemini space flights in the 1960s, where the cells were used as an auxiliary power supply (Kim et al., 2004). However, the widespread use of this technology was limited due to the cost of the cells.
Presently, PEMFCs continue to be too expensive for the commercial market because of the high price of the platinum (Pt) catalyst. As of 2013, the cost is approximately $1,576/troy ounce. Identifying a technique to minimize Pt use in the cell or Pt waste during cell preparation is the primary focus of research today because it can lead to a major switch in energy sources (Kim et al., 2004).
A catalyst such as Pt is crucial in a PEMFC for the oxidation of hydrogen (H2) at the anode and the reduction of oxygen (O2) at the cathode. (Fig. 1) An ideal PEM, such as Nafion, consists of a fluorocarbon backbone and sulfonate sidechains, which phase separate into hydrophilic channels and a surrounding hydrophobic domain (Chou et al., 2006). When oxidation occurs, H2 decomposes on Pt into H+s and electrons, and the H+s traverse from anode to cathode through the hydrophilic domains of the PEM, while the electrons travel through an electronic load producing electricity. At the same time, O2 adsorbs onto Pt and combines with the electrons and H+s to form H2O.
Although direct deposition of platinum may be less efficient than deposition through a Nafion membrane (Kim et al., 2004; Chou et al., 2006), it is more efficient than the conventional powder-type membrane electrode assembly (MEA) and utilizes less of the Pt catalyst to obtain comparable power outputs (Kim and Popov, 2004). It is important to consider the disadvantages of not selectively depositing the Pt particles in membrane channels though. First, if the catalyst is present in the hydrophobic domain of the membrane, it is useless because the protons have no channel to cross through. Secondly, if there is a platinum particle in the hydrophilic domain that is not located at the terminal ends of the anode and cathode, then the catalyst is wasted. Lastly, if Pt is located at one terminal end and not the other of the hydrophilic domain, then the platinum is also wasted. Therefore, it is important to localize Pt nanoparticles at or near the hydrophilic domains that traverse the membrane. (Fig. 2) This study focused on how various parameters affected the electrochemical deposition of Pt from H2Pt(OH)6 plating solution. In the future, these parameters can hopefully be applied with deposition methods through the Nafion membrane to produce an even more efficient fuel cell.
The MEA is the central part of the PEMFC and consists of a proton exchange membrane (Nafion®) that is sandwiched between two catalyst-containing gas diffusion electrodes (typically Pt/carbon) as shown in Figure 1. A well-constructed MEA will have an evenly distributed volume in the catalyst layer for each transport media so that when protons, electrons and water travel through it, a minimal amount of transportation loss will occur (Litster and McLean, 2004). The catalyst layer is applied either to the gas diffusion layer or to the membrane. In this experiment, it is applied onto the microporous layer of the gas diffusion electrode. This allows the platinum to be in contact with both the membrane and the GDL. The gas diffusion layer provides the fuel and oxidant a pathway to the catalyst layer and conducts electrons. It also helps to regulate the amount of water that the membrane has access to, in order to keep it hydrated.
The amount of platinum required for a PEMFC has decreased by over one hundred fold since its initial development in 1960, decreasing from 4 mg cm-2 to 0.014 mg cm-2 (Litster and McLean, 2004). Less platinum usage does not necessarily correlate with higher energy output of the PEMFC. It may indicate, however, better performance and efficiency for the amount of platinum that is deposited in the fuel cell. Although these improved fuel cells are less expensive, they do not necessarily produce as much energy as commercial PEMFCs currently available. Precatalyzing the carbon cloth with Pt particles and combining it with a PTFE binder has proved to be less efficient in terms of performance per platinum loading. This is because the binder covers some platinum nanoparticles when the gas-diffusion electrode is being created, rendering them useless (Maoka, 1988). Catalyzation from a dilute electrolyte, performed after the membrane electrode assembly was completed, was found to waste fewer platinum particles (Verbrugge, 1994). Multiple methods of deposition have been investigated, and in the future these factors can hopefully be used to improve efficiency with these methods. To optimize parameters in this experiment, platinum will first be deposited onto the microporous layer of a cheap FTO electrode that is easy to characterize.
In order to minimize Pt use even further while maintaining energy output, it is better to create a large number of small Pt particles rather than a small number of large particles.
This is because smaller particles have a much greater surface area to volume ratio than larger particles. With more surface area available, there are more active sites to catalyze the reaction. Additionally, larger particles may have more inner pores coursing through the center of the particle, theoretically increasing the measured surface area. However, these sites are less efficient for catalyst reactions. The optimal PEM has a high amount of nucleation, or loading of individual Pt particles, which are small in particle size and diffusely spread out. The purpose of this investigation was to increase Pt loading and particle surface area by testing various conditions during the electrochemical deposition of platinum at the polymer-carbon interface. With a defined method to produce a less expensive fuel cell, the commercialization and implementation of this technology as part of the solution to the current energy crisis is an exciting prospect.
Preparation of the FTO Linear Electrode
The optimal Pt loading parameters for the fuel cell must be measured in order to obtain optimum utilization of the Pt catalyst. This was done using a fluorine-doped tin oxide (FTO) planar electrode, which serves as an easy-to-characterize replacement in order to conserve the materials necessary for the proton exchange membrane fuel cell.
First, the FTOs were placed in a beaker and a 50:50 ethanol and deionized (DI) water mixture was poured into the beaker until the mixture covered the FTOs. They were sonicated for 15 minutes, taken out, rinsed with DI water, and sonicated for another 15 minutes. The FTO samples were placed back in their storage container and dried with nitrogen gas on both sides. Once the FTO was clean, the voltmeter was used to test for the conductive side, where copper tape was placed. The area in which the platinum was deposited was restricted to allow for a comparison of the distribution of platinum and the particle size for the same area. Silicon tape was placed over the FTO and only a little more than 1 cm2 was available for the deposition of Pt nanoparticles. The FTO was then ready for the deposition process.
Deposition of platinum nanoparticles onto the FTO
Multiple precursor solutions were used to determine which solution was the most optimal for Pt deposition. The precursor solutions used were 5mM H2Pt(OH)6 (platinic acid) in 1.5 M, 2.0 M, and 2.5 M H2SO4 (sulfuric acid) and 5mM H2Pt(OH)6 in 20% 2:1 HNO3 (nitric acid)/H2SO4. In various samples, the precursor solution was also purged with an inert gas for 20 minutes. The precursor solution was placed in a 50mL beaker that was cut to 30mL and the FTO electrode was placed in an alligator clip as the working electrode. A Pt mesh was used as the counter electrode and a reference electrode was evenly spaced between the two. The reference electrode was cleaned with KCl/AgCl (potassium chloride/silver chloride) prior to deposition. The three electrodes were lowered into the solution and were connected to the potentiostat for the electrodeposition process. The computer program EC-Lab Express was used to modify the different parameters. Such parameters included the on and off time, the current and potential range, and the number of cycles that were performed. This study used the current range of 1A, the potential range of 2.5V, the on and off time at 50μs, and the bandwidth of 7-high speed. With each solution, the deposition was performed at 0.1V, 0.25V, 0.5V, and 1.0V at different deposition times of 5, 10, and 15 minutes. Many depositions were repeated using the same solution, but after an hour, the solution was purged again to ensure that no impurities were present when the platinum was being deposited onto the FTO. The OCV (open circuit voltage) was taken before deposition to ensure that the precursor solution had not touched the copper tape.
Preparation of nafion-coated carbon cloth
To prepare the ELAT A-10 carbon cloth for the MEA, copper tape was applied to a piece of carbon cloth, 1cm in width and 2cm in length. Keeping the length of copper tape consistent with each carbon cloth, platinum was deposited using the same preparation techniques as depositing onto the FTO except that now deposition was onto a carbon cloth. Next, the carbon cloths were rinsed in boiling DI H2O for 2 hours and then dried for 24 hours. A pipette was used to mix 10μL of the Nafion solution (5% wt) with 100μL of isopropanol. After, the hot plate was preheated, the solution was dropcasted in increments of 20μL onto the carbon cloth electrode, which sat on the hot plate. Before dropcasting occurred, the electrode was placed on aluminum foil because it is a good heat conductor. As the solution was dropcasted, another 20μL of isopropanol was added to the Nafion solution to remove the residue that remained. The pipette was then used to dropcast the rest of the solution in increments of 5μL so that the Nafion was spread over the entire electrode. This created a total solution of 130μL of Nafion on the carbon cloth.
The electrode was then put in an oven for 2 hours at 80°C. These steps were then performed for the other electrode, which allowed construction of the MEA.
Membrane electrode assembly
A small piece of the Nafion membrane, about 1.75 by 2.5 inches in size, was cut for the MEA. The metal clamp was cleaned with isopropanol and the hot-press was heated to a temperature of 266°F. The Nafion membrane was cleaned, pressed for approximately 5 minutes, and then the protective layers were peeled off. The Nafion membrane was sandwiched between the two carbon cloths, with the platinum facing towards the membrane. It was lined up so that the gas diffusion layer (carbon cloth) was in contact with the catalyst, which had access to the Nafion membrane. A Teflon backbone with a fiberglass covering was placed on the outside of both sides of the MEA, to prevent the gases from escaping. This assembly was pressed at a pressure of 250psi for 3 minutes and cooled for 10 minutes in air.
The cyclic voltammogram (CV) scans were performed in sulfuric acid that had been purged with nitrogen gas for 20 minutes. The purpose of these scans was to find the hydrogen adsorption and desorption peaks. By observing that one hydrogen atom binds to one platinum particle, it can be assumed that the oxidation and reduction peaks will give an accurate representation of the surface area of the catalyst. The preparation of the CV scans were exactly the same as the deposition preparations except that sulfuric acid was used in place of the precursor solution. To rid the sulfuric acid of impurities, it was purged with nitrogen gas for 20 minutes. Since all of the impurities may not be eliminated during the purging process, many CV scans were performed until the graph became constant. Roughly 30-60 scans were performed before the solvent was purged of impurities and an accurate reading of the catalyst performance was obtained. This number is constant throughout all FTOs because the same solvent (H2SO4) was used.
Scanning Electron Microscopy
Scanning electron microscopy (SEM) was performed on all the FTOs to produce a visual representation of the Pt nanoparticles deposited on the surface of the FTO. A focused beam of electrons was scanned across the surface of each sample and changes in reflected energy were measured to generate an image. The image provided insight into Pt distribution, particle size, and whether nucleation was occurring on new sites or on preexisting nucleation sites.
Fuel Cell Testing and Performance Curves
Once the complete fuel cell was ready for testing, it was sandwiched between two lead plates (anode and cathode) and connected to gas tanks of hydrogen and oxygen. With this test station, performance curves of the fuel cell were measured. These curves represented the performance of the fuel cell as if it were actually being used as a power source. The fuel cell test station incrementally decreased the voltage (V) and then measured the current density (mA/cm2) of the fuel cell. The performance curve reflects the various conditions within the fuel cell. First, fuel crossover, or electron conduction, occurs. Next, oxygen is reduced while the PEM resistance increases. Finally, gas transport is occurring, where the reactants are being consumed faster than they are being supplied. This is expected because the catalyst reactions are going to take place faster than the reactants can be supplied, so the reactions will slow down. This will, in turn, decrease the performance of the fuel cell.
Using electrodeposition at the polymer-carbon interface, various experiments were performed in order to minimize total Pt used and maintain performance of the fuel cell. A variety of steps were taken in order to optimize the parameters for the construction of the hydrogen fuel cell. The following variable conditions were modified during this study: precursor solution, solvent concentration, voltage, deposition time, purity of the solution, and type of solvent.
The precursor solution platinic acid, H2Pt(OH)6, was compared to chloroplatinic acid, H2PtCl6, a common supporting electrolyte used in the production of fuel cells for its properties of ideal electronic and ionic percolations and high-Pt-mass fractions (Antoine and Durand, 2006). H2PtCl6 caused Pt particle formation of extreme size. SEM images show only a few, porous, large particles present on the FTO with the supporting electrolyte of H2PtCl6 (a), while there are many nanoparticles with the precursor solution H2Pt(OH)6 (b), yielding a higher surface area of the catalyst. (Fig. 3) Also, the concentration of 5mM H2Pt(OH)6 resulted in greater Pt loading than 1mM H2Pt(OH)6.
Solvent concentrations of 1.5, 2.0, and 2.5M H2SO4 were investigated. The SEM images indicate that as the solution concentration is decreased, the platinum loading is increased. (Fig. 4).
In 1.5M H2SO4 (a), there are many small particles and nucleation sites, as opposed to the 2.0M H2SO4 (b), where nucleation has occurred on other particles and decreased the surface area available for catalyst reactions. The 2.5M H2SO4 image (c) shows the least amount of platinum deposition onto the FTO. The oxidation and reduction peaks of the CV graphs of 5mM H2Pt(OH)6 in 1.5M H2SO4 are larger than the peaks of 2.0M H2SO4, signifying that there is more Pt surface area on the surface of the sample. (Fig. 5) However, when the CV was taken in 2.5M H2SO4, the oxidation and reduction peaks were substantially larger, with the maximum current being greater than both 2.0M and 1.5M. The particles prepared in 2.5M H2SO4 were the largest in size with the smallest total exterior surface area, but they had the largest number of inner pores that gave a false high reading with CV.
Voltages applied across the electrolyte were varied. A trend was observed that the lower the voltage, the greater the amount of platinum deposited, and the lower the chance the platinum nanoparticles would nucleate on top of each other. First, voltages of ±1.0V and ±0.5V were compared. The CV graph of ±0.5V (a) has bigger oxidation and reduction peaks than the peaks of ±1.0V (b). (Fig. 6).
These graphs support the trend that lower voltages deposit more platinum nanoparticles. When the voltage was decreased down to ±0.1V, a limit to the trend was identified. SEM images showed that as the voltage was decreased from ±1.0V (a) to ±0.5V (b), and then to ±0.25V (c), more surface area of the catalyst was available for reactions. (Fig. 7) Once the voltage was decreased to ±0.1V (d), nucleation occurred on other particles, creating large platinum particles that contained little surface area. Therefore, it can be concluded that the trend applies to all the voltages, with the lowest threshold at ±0.25V.
The independent variable, deposition time, was tested using the potentiostat and the three-electrode system. Depositions were performed onto the FTO at 5, 10, and 15 minutes. As the times increased, the Pt loading increased, but also the nanoparticles were more inclined to nucleate on each other. Depositions of 15 minutes still proved to have the best performance because even though nucleation on other particles was common, platinum nanoparticles were forming on many new nucleation sites as well. SEM images of three FTO substrates deposited with Pt under different deposition times were obtained. (Fig. 8).
The 5 minute deposition (a) clearly had the least Pt loading. A greater number of particles are apparent with a 10 minute deposition (b) because nucleation sites are distributed throughout the surface of the sample. With the 15 minute deposition (c), platinum nanoparticles are abundant and there is more surface area than either of the other two samples. This set of data demonstrates that as deposition time increases, the Pt loading increases. The CV graph confirms the deposition time trend. (Fig. 9) The oxidation and reduction peaks of hydrogen gradually become larger as the time increases. The peaks also become more defined as deposition time increases, allowing for a more exact reading of the surface area of the catalyst.
The purity of the precursor solution was also a variable that was tested in the study. It was postulated that if the supporting electrolyte were purged with inert gas before deposition, the solution would be free from impurities. This would allow easier deposition and possibly improve the amount of deposited platinum nanoparticles onto the sample. It was found that there was no significant difference when comparing a deposition with purged solution versus one without purging. The CV graph of the purged solution was almost identical to the CV graph of the non-purged solution. This would mean that both depositions had similar surface areas for catalyst reactions, making each fuel cell no more significantly efficient than the other. Similarly, SEM images of the purged and non-purged samples were almost identical. Although the Pt deposition in the purged sample appeared to be greater, the difference was too minute to form a definite trend.
Two different solvents were used in this study, H2SO4 and 20% 2:1 by volume HNO3/ H2SO4. The platinum was more easily dissolved in the 20% 2:1 HNO3/ H2SO4 acid, making it easier for deposition onto the sample. It was observed with CVs and SEM images that the platinum was more abundant and more evenly distributed throughout the sample when deposited with this new acid as compared to H2SO4.
Fuel Cell Testing
Once a complete fuel cell was constructed, it was connected to the fuel cell test station for performance testing and various performance curves were obtained. (Fig. 10) The performance curve of a commercial fuel cell demonstrates a very high performance, but at the expense of a large amount of wasted platinum in the cell. The ratio of performance per platinum usage is actually significantly better for the fuel cells produced during this study, as shown by the performance curves. The fuel cells were constructed using deposition parameters of ±1.0V, 25 minutes, with 5mM H2Pt(OH)6 in 1.5M H2SO4, 2.5M H2SO4, or 20% 2:1 HNO3/H2SO4 acid. It was found that although the solvent of 20% 2:1 HNO3/H2SO4 acid achieved better platinum deposition onto the carbon cloth than any concentration of H2SO4, the performance of the cell undergoing Pt deposition with 1.5M H2SO4 was substantially higher, producing the most energy.
Each of the parameters varied in this study showed clear trends regarding efficient deposition of Pt on the FTO. The new precursor solution H2Pt(OH)6 was clearly superior to the traditional solution, H2PtCl6 , because there was a substantially greater amount of Pt deposited with less waste. The Cl- ions often poisoned the Pt particles and prevented them from making new nucleation sites, as confirmed by Schmidt et al., who detailed their corrosive and degrading effects (2001). The Pt tended to nucleate on itself, creating large particles that contributed to less efficient fuel cells. When the solvent concentration of H2SO4 was decreased, the SEM images demonstrated improved Pt loading. The CVs of these samples were consistent with this trend except with the 2.5M solution. In this circumstance, the CV result was spurious because the particles created with the 2.5M solution had large inner pores. Though they were involved in oxidation and reduction in CV scans, these Pt particles would not be as efficient in catalyst reactions in a working fuel cell. When the voltage applied across the electrolyte was decreased, a trend was observed that the lower the voltage, the greater the Pt was deposited to a limit of ±0.25V. At ±0.1V, Pt nucleation occurred on other particles, creating large Pt particles that contained a low surface area to volume ratio. In this circumstance the voltage was not sufficient to effectively deposit Pt. Increasing Pt deposition time clearly showed a trend with improved Pt loading. Increasing the time Pt was exposed to the FTO improved the chances for deposition. However, there is a limit, because as the time increases, the Pt accumulates on itself and creates large particles that may not be effective considering the increased amount of platinum used. These findings were consistent with Chou et al., who observed that the density of Pt depositions increased as deposition time increased (2006). Attempting to purge the solution prior to deposition did not improve Pt loading, but using a new solvent solution of 20% by volume mixture of 2:1 HNO3/H2SO4 did. It produced large even particles with a CV curve larger than all solutions except the 2.5M H2SO4 solution, a solution with a spurious CV curve. When performance testing was done on electrodes utilizing the best preparation parameters as identified in this study, the 5 mM H2Pt(OH)6 in 1.5M H2SO4 fuel cell ultimately proved to be the best. This cell had similar performance to that of a commercial fuel cell, but used much less Pt during the preparation of the MEA. This is most likely observed because all the optimized parameters in this fuel cell work together to complement each other through unique mechanisms to produce the most platinum loading with the highest surface area to volume ratio. Further investigation might include determining why the fuel cell using 1.5M H2SO4 solution yielded a higher power output than the fuel cell that used the 20% by volume mixture of 2:1 HNO3/H2SO4, considering CV graph peaks and SEM images seemed to suggest otherwise.
Optimizing the parameters for the hydrogen fuel cell proved to lower the cost substantially and maintain adequate performance. Using pulse electrodeposition allowed for plating of Pt particles on the nanometer (nm) scale, consistent with the findings of Kim and Popov, who detailed the potential of pulse electrodeposition to deposit Pt particles as small as 5nm (2004). Various trends were identified from the CV graphs, SEM images, and performance curves. It was found that depositing platinum with 5mM H2Pt(OH)6 was superior to a previous supporting electrolyte of H2PtCl6. Also, 5mM of the precursor solution deposited more platinum than 1mM H2Pt(OH)6. The new solvent of 20% by volume mixture of 2:1 HNO3/H2SO4 was found to yield more platinum surface area than the H2SO4 acid. The trend did not coincide with the performance graph though. The fuel cell made with 1.5M H2SO4 acid proved to have a better performance curve than a fuel cell made using the 20% acid. A trend was identified that as the solvent concentration of H2SO4 acid decreased from 2.5M to 2.0M to 1.5M, the Pt loading involved in catalytic reactions increased. Also, during pulse voltage deposition it was found that as the voltage decreased, new nucleation sites would form more often, creating more surface area. When deposition with ±0.1V was tested, SEM images indicated that the platinum particles were porous and were not as abundant on the FTO sample. It can be concluded that down to ±0.1V, as the voltage decreases, the Pt loading increases. It was found that as deposition time increased, the Pt loading increased and nucleation was occurring on new sites as well. Lastly, it was concluded that purging the supporting electrolyte prior to deposition did not make a difference in the amount of platinum surface area that was available for catalyst reactions.
In this study, parameters for optimization of Pt loading were investigated via direct deposition of platinum onto the microporous layer of the gas diffusion electrode rather than deposition through the Nafion membrane. This was done to conserve expensive materials required in the fuel cell, such as the Nafion membrane and carbon cloth electrodes. At this time, the high price of the hydrogen fuel cell is the major obstacle preventing its use in the commercial market. These results will spur future research to utilize these optimized parameters to create a more efficient proton exchange membrane fuel cell.
- Litster, S.; McLean, G. PEM Fuel Cell Electrodes. J. Power Sources, 2004, 130, 61-76.
- Kim, H.; Subramanian, N. P.; Popov, B. N. Preparation of PEM Fuel Cell Electrodes using Pulse Electrodeposition. J. Power Sources, 2004, 138, 14-24.
- Chou, J.; Jayaraman, S.; Ranasinghe, A. D.; McFarland, E. W.; Buratto, S. K.; Metiu, H. Efficient Electrocatalyst Utilization. J. Phys. Chem. B, 2006, 110, 7119-7121.
- Kim, H.; Popov, B. N. Development of Novel Method for Preparation of PEMFC Electrodes. Electrochem. Solid St., 2004, 7, A711-A74.
- Maoka, T. Electrochemical Reduction of Oxygen on Small Platinum Particles Supported on Carbon in Concentrated Phosphoric Acid-I. Effects of Platinum Content in the Catalyst Layer and Operating Temperature of the Electrode. Electrochim. Acta, 1988, 33, 371-377.
- Verbrugge, M. W. Selective Electrodeposition of Catalyst within Membrane- Electrode Structures. J. Electrochem. Soc., 1994, 141, 46-53.
- Antoine, O.; Durand, R. In Situ Electrochemical Deposition of Pt Nanoparticles on Carbon and Inside Nafion. Solid St., 2001, A55-A58.
- Schmidt, T. J.; Paulus, U. A.; Gasteiger, H. A.; Behm R. J. The Oxygen Reduction Reaction on a Pt/Carbon Fuel Cell Catalyst in the Presence of Chloride Anions. J. Electroanal. Chem., 2001, 508, 41-47.