The Effect of Fuel and Oxidant Contaminants on the Performance of PEM Fuel Cells

Transcription

1 WHEC 6 / 3-6 June 26 Lyon France The Effect of Fuel and Oxidant Contaminants on the Performance of PEM Fuel Cells Robert Benesch, Sumaeya Salman, Tracey Jacksier, American Air Liquide, 523 S East Ave, Countryside, IL 6525, USA robert.benesch@airliquide.com, sumaeya.salman@airliquide.com, tracey.jacksier@airliquide.com ABSTRACT: The performance of a Proton Exchange Membrane Fuel Cell (PEMFC) system can be affected by various chemical compounds that may enter the system in the fuel or oxidant gases. These compounds or impurities can degrade the fuel cell s performance by interfering with the normal physical processes that occur on or in the fuel cell s membrane electrode assembly (MEA). Hydrogen can contain several impurities that are known to affect the performance of the PEMFC. The type and amount of impurity depends on the particular process used in the production of the hydrogen. Ambient air also contains impurities which can vary significantly depending on the location, time of day and season. In order to examine the impact of impurities on the performance of the fuel cell, experiments were performed in which the impurities were introduced over a period of approximately hours. Cell voltage loss was used as the primary evaluation method to judge the impact of the impurity. The ability of the cells to recover after exposure to the impurity was also examined by running the fuel cell with pure gas following the exposure. The results of CO, H 2 S and NH 3 impurities evaluation in hydrogen as well as the evaluation of CO in air are presented. KEYWORDS : Proton Exchange Membrane Fuel Cell, Impurities. Introduction Proton Exchange Membrane Fuel Cell (PEMFC) Fuel cells are electrochemical devices that convert the chemical energy stored in the bonds of molecules into electrical energy. There are several different types of fuel cells; one type of fuel cell that is beginning to be introduced in the automotive and small scale power generation markets is the proton exchange membrane fuel cell or PEMFC. These markets are currently dominated by the internal combustion engine (ICE) which has a proven track record of over years. A PEMFC has several benefits over an ICE which will aid in its proliferation into these markets. The benefits include: reduced emissions, higher well to wheel efficiencies and a greater simplicity (there are almost no moving parts). One drawback to PEMFCs is that they can experience fuel cell poisoning. Fuel cell poisoning occurs when the performance, in terms of power output from the fuel cell, is reduced following the exposure to harmful chemicals. These harmful chemicals are often introduced into the fuel cell as impurities in the fuel and oxidant gases. In order to overcome the fuel cell poisoning issue, a better understanding of which types of chemicals can act as poisons needs to be obtained. Also, an understanding of performance losses as a function of exposure times or exposure levels is an important issue. This knowledge should help overcome one hurdle for the successful implementation of PEMFCs in the marketplace. Impurities can affect the various physical and chemical processes that occur within the PEMFC. Most of these processes occur on or in the main component of the PEMFC, which is the membrane electrode assembly (MEA). A MEA typically consists of a proton conducting membrane sandwiched between two electrodes. The membrane s primary function is to conduct protons from one electrode to the other; it is similar to the electrolyte found in a typical battery. Certain impurities can physically absorb into the membrane and can lower the conductivity of the membrane, thus reducing the PEMFC performance. The electrodes, which sandwich the membrane, are made from a catalyst that is supported on a conductive media. The role of the catalyst is to facilitate the adsorption of the reactant species and the electrochemical reactions. Platinum is typically used as /

2 WHEC 6 / 3-6 June 26 Lyon France the catalyst, which is well known to be susceptible to poisoning by contaminants. It is anticipated that some impurities will have different effects depending on whether they are in the fuel or oxidant stream as a result of the differing conditions which exist at each electrode. The performance losses attributed to specific impurities can often be temporary and the performance can recover after the exposure has been terminated, while others have permanent effects. Origin of impurities in the reactant gases The origin of impurities in hydrogen is dependent upon the method used to produce and purify it as well as the feedstock used. There are currently several industrial processes to produce hydrogen, these include: steam methane reforming (SMR), autothermal reforming (ATR), partial oxidation (POX), purification of hydrogen rich streams from petrochemical plants and water electrolysis. The most utilized method for the production of H 2 is SMR []. Methane is typically used as the hydrocarbon feed stock. In this process, methane and steam are passed over a catalyst at elevated temperatures [2]. Under these conditions, the amount of CO and CO 2 produced is on the order of 25%. Carbon monoxide is typically separated from hydrogen by utilizing various purifications processes, such as a pressure swing adsorber (PSA), however trace impurities of CO can exist in the purified hydrogen. Additionally, sulfur compounds and heavier hydrocarbons, which are originally present in the hydrocarbon feed stock, may carry through the purification steps. Some impurities which have been reported in SMR produced hydrogen include: ammonia [3], formaldehyde, formic acid, methanol and methane [4]. The impurity origins for the oxidant stream are easy to understand. The burning of fossil fuels generates volatile organic carbons (VOC s), hydrocarbons, nitrogen oxides (NO x ), sulfur oxides (SO x ), as well as other compounds. Not surprisingly, these impurity levels tend to be the greatest in cities or near heavily industrialized areas. Table gives the maximum hourly concentration that was detected at a Houston (Texas, US) air monitoring facility in 25 [5]. Other impurities that were analyzed included: benzene, toluene, ethylbenzene,,3 butadiene, hexane, 2,2,4 methyl pentane and methyl ethyl ketone. The levels of these hydrocarbons tended to be in the single digit parts per billion level. Impurity Maximum Hourly Average (ppm) CO 3 NO 2.65 SO 2.37 O 3.9 Table : Maximum hourly average concentration of air impurities detected at a Houston air monitoring facility. Experimental Impurity exposure tests were conducted by exposing a cell stack to reactant gases with the added impurities. The cells were configured such that there were 5 groups with two cells in each group. The 5 groups utilized separate piping for the feed gases thus allowing 5 different impurity levels to be tested simultaneously. The first set of cells is utilized as an experimental control with no added impurities. The experimental configuration for the hydrogen impurity experiments is illustrated in Figure. 2/

3 WHEC 6 / 3-6 June 26 Lyon France Figure : Experimental configuration of impurity evaluation in hydrogen The PEM membranes utilized in this study were Gore PRIMEA Series 55 (Gore FC Technologies, Elkton, MD) with an active electrode area of cm 2. The catalyst was Pt/Pt (anode/cathode) with a loading of.4 mg/cm 2 for each electrode. Carbel CL series gas diffusion layers were utilized (Gore). The bipolar plates used (Fuel cell Technologies, Inc., Alburquerque, NM) in the experiments were made of POCO graphite. The channeling consisted of quad serpentine flow channels for both the anode and cathode plates. The compressed air was generated from ambient air using an oilless diaphragm compressor. The air was passed through an activated carbon filter bed (not shown) to remove trace impurities. The air flow rate was operated at a stoichiometic ratio of 2.2. A bubbler type humidifier was used to humidify the air stream which was controlled at % relative humidity at a temperature of 6 C. The temperature was controlled by supplying heat to the humidifier. The regulation of the incoming air was ± 2 C, under normal operating conditions. Cell temperatures were regulated using 4 cooling fans that were placed below the cells. The temperature of the middle cell was used as the controlling parameter. The regulation of the temperature was ± 2 C for the middle cells. Because of the single parameter used to control temperature of the entire stack, a temperature gradient of 7 C existed across the stack, with the outer cells being the coolest. The hydrogen circuit was operated in a closed loop. The hydrogen stream was dead-ended using a purge valve at the exit of the cells. Under normal operation, water and nitrogen diffuse across the membrane from the air side of the fuel cell to the hydrogen side. These gases build up in the hydrogen circuit and if allowed to remain they can dilute the hydrogen in the cells. As a result, the hydrogen circuit is purged periodically to remove these gases. The circuit was purged every 45 seconds with a purge duration of 2 seconds. Impurities were blended into the gas cylinders used for the feed hydrogen. The cylinders were prepared on-site using known standards with diluted Alphagaz 2 Hydrogen (Air Liquide America LP, Houston TX). The standards were analyzed to verify their concentration for all compounds except ammonia. The pressure just prior to the cell entrance was regulated via pressure regulators located on the gas cylinders. The pressure was regulated to 4.4 psig (±.2 psig). A programmable electronic load (TDI Dynaload, Hackettstown, NJ) was utilized to draw power from the fuel cell stack. The electronic load, flow meters, flow controllers, valves and data acquisition were all interfaced and controlled using Labview (National Instruments, Austin, TX). In order to gain a more fundamental understanding of the effect of the impurities, the exiting gas stream was analyzed. A gas chromatograph (GC) was utilized (Model 34, Varian, Inc., Walnut Creek, CA). The GC was equipped with 2 different detectors which were chosen depending on the species to be analyzed. A pulse discharge helium ionization detector (PDHID) (VICI, Houston, TX) was used to analyze for CO and CO 2. A sulfur chemiluminscence detector (SCD) (Antek Instruments, Houston, TX) was used to analyze the sulfur compounds. At the time of these experiments, an analyzer for ammonia was not available. A ft, /8 3/

4 WHEC 6 / 3-6 June 26 Lyon France diameter molecular sieve column was used to measure CO and a ft, /8 A Haysep Q column was used for CO 2 (both columns were from VICI). A 3 meter.32 diameter Silicaplot column (Varian) was used for the sulfur compounds. A phase separator was used between the fuel cell exit and the analyzer to remove most of the water that was present in the gas. Testing An impurity s effect was evaluated in 2 ways, similar to the methodology described by Watanabe et al [6]. The primary method utilized was to observe the cell voltage drop over time, see Figure 2. Here, the fuel cell performance loss is illustrated by the loss in the individual cell voltage as a function of exposure time. For these experiments, the impurity was continually added to the cell until a threshold voltage, typically set at.3 V, was reached. If the cell voltage dropped below this voltage, the impurity would be automatically shut down and the specific group of cells would run on pure hydrogen. In this manner, the reversibility of the impurity action on the FC could be evaluated. Cell Voltage [V] Hydrogen containing impurity H Time [h] VDR Figure 2: Illustration of voltage drop (VD) over time The second method utilized involved conducting polarization curves. A polarization curve is a graph of the cell voltage versus the current density (Figure 3). These are conducted periodically during the primary testing. When conducting the polarization curve, the current density is changed stepwise from the normal operating value of.5 A/cm 2 to values that range from. to.8 A/cm 2, starting at.. The stack is held for 5 minutes at each current density interval; voltage data from the last 3 seconds of the interval is averaged and plotted. Cell Voltage [V] VD Current Density [ma/cm2] Pure Hydrogen Hydrogen Containing Impurity Figure 3: Polarization curve illustrating the voltage drop (VD) between pure H 2 and H 2 containing an impurity 4/

5 WHEC 6 / 3-6 June 26 Lyon France Hydrogen Impurities Carbon Monoxide The effect of CO concentration was evaluated. The four concentrations which were tested were,.52,., 4.5 and 9.2 ppm, see Figures 4-6. Shortly after exposure to 9.2 ppm, the cell voltage quickly decreased (Figure 4). At 3.4 hours, the voltage decreased below the threshold value and the cells were allowed to recover under pure hydrogen Switch to pure H 2 (3.4 hrs) Figure 4: Reversibility of catalyst poisoning after exposure to 9.2 ppm CO The voltages for the. and 4.5 ppm exposed cells (Figure 5), decreased to steady state values of.5 and.37 volts, respectively. These decreases corresponded to a 2 % decrease and a 42 % decrease, respectively. The voltage for.52 ppm also decreased (Figure 6), although at a very slow rate. After 76 hours there was a decrease of 4.3 %.9 Cell Voltages (V) ppm CO. ppm CO Figure 5: Exposure to. and 4.5 ppm carbon monoxide 5/

6 WHEC 6 / 3-6 June 26 Lyon France.9 Cell Voltages (Volts) Reference H 2.52 ppm CO.45.4 Figure 6: Exposure to and.52 ppm carbon monoxide The polarization curve for the 4.5 ppm CO exposure is given in Figure 7. As expected, the polarization curves show a decrease in cell voltage with time and eventually reach steady values. The decrease in cell voltage is observed after about hours, although a significant effect in the polarization curve after 6.5 hours is noticed Hours.75 Hours 6.5 Hours 25.6 Hours 5.2 Hours 7.5 Hours 96. Hours Current Density (A/cm 2 ) Figure 7: Polarization curve of cell exposed to 4.5 ppm of carbon monoxide in hydrogen. Carbon monoxide is said to affect the performance of the fuel cell by adsorbing to the platinum catalyst sites and thus preventing hydrogen adsorption and oxidation. Baschuk et al [7] summarized the work of Vogel et al [8], and stated that the CO chemisorbs on the platinum sites and excludes hydrogen. The CO is more strongly bonded to platinum because the sticking probability is 5 times higher for CO than hydrogen on platinum. If the overpotential becomes large enough, the CO can be oxidized to CO 2 which readily desorbs from the platinum site; this frees the site and allows hydrogen to adsorb. A mechanism can be described [9] as illustrated in Reactions and 2: CO adsorption: CO g + Pt s Pt CO ads Reaction CO oxidation: Pt CO ads + H2O Pt s + CO 2 g + 2 H + + 2e - Reaction 2 In order to validate this proposed mechanism, the exit stream of the anode was sent to a gas chromatograph equipped with a PDHID detector. This detector was selected because it was able to analyze the gases in the expected concentrations. Referring to Figure 8, an input concentration of 9.2 ppm CO was used for this experiment. Within approximately 2.5 hours, the voltage reached a steady state value of approximately.4 V. Prior to the introduction of the CO into the H 2 stream, the CO 2 concentration was measured and found to be 2. ppm. The source of the background CO 2 is unclear, as it was not present in the source H 2. Initially, it 6/

7 WHEC 6 / 3-6 June 26 Lyon France appears that the CO being introduced was adsorbed onto the catalyst surface. After a period of time, the number of available catalyst sites had been reduced to a point at which the hydrogen oxidation reaction was being affected and the cell voltage began to decrease. At this point, CO was detected in the exit stream. Subsequently, the over-potential of the cell became large where the adsorbed CO species was oxidized to CO2. The CO2 subsequently desorbed and was detected in the exit stream. After approximately 4 hours, the concentrations and the voltage reached an equilibrium. At this point the total carbon entering the system was approximately equal to the carbon exiting the system in the form of CO and CO2 molecules CO CO2 Impurity Concentration (ppm) Run Figure 8: Exit stream analysis of carbon monoxide and carbon dioxide Hydrogen Sulfide The ability of low concentrations of H2S to affect the fuel cell performance was evaluated by exposing the cells to H2S at various concentrations (Figure 9). Under the conditions evaluated, after initial exposure of 5 hours, the voltage decreased somewhat linearly for the cells exposed to 2.,, and.5 ppm H2S. From the data that was collected for hours of exposure, it is difficult to determine if the voltage for the.25 ppm concentration is decreasing. It is clear that a longer testing period is required before a conclusion can be drawn on the effect of.25 ppm..9. ppm.8.5 ppm ppm ppm ppm. Pure H Figure 9: Exposure to,.25,.5,., 2. ppm H2S 7/ 8

8 WHEC 6 / 3-6 June 26 Lyon France The cell voltage for the concentrations greater the.5 ppm, decrease to the point where the impurity was removed and the cells were allowed to recover. The recovery of the cells following the switch to pure hydrogen resulted in a minimal recovery. This recovery seemed to be favored when a polarization curve was performed. As an example, the recovery of the 2. ppm cells was initiated at approximately 3 hours. There was only a small observable recovery until about 49 hours, when a polarization curve was performed. Following the polarization curve, there was a step change in the cell voltage. This phenomenon can be a result of the cell potential increasing on the anode and partially oxidizing the platinum sulfur species and thus freeing up a catalytic site. A mechanism for hydrogen sulfide (H 2 S) poisoning on platinum was proposed by Farooque et al []. They proposed that H 2 S adsorbs onto platinum catalyst. This physically blocks the catalyst site and thus limits the amount of catalyst available for hydrogen adsorption and electrochemical oxidation. The adsorbed H 2 S-Pt species subsequently undergoes an electrochemical oxidation to form a platinum sulfur species. The result of the oxidation is a platinum sulfur species which is difficult to remove. Chang et al [,2] determined that sulfur induced platinum agglomeration and migration. The analysis of the exiting hydrogen stream was performed on the cell exposed to 2. ppm. The data is given in Figure. The GC equipped with a sulfur specific detecor has the ability to analyze for both H 2 S and other sulfur compounds. However, the only observable sulfur species was H 2 S. The exiting H 2 S concentration showed a gradual increase as the cell voltage decreased. At 25 hours, there was a 5 minute shut down of the test bench; afterward there was subsequent drop in the H 2 S concentration. At 32.5 hours, the cell was switched to pure hydrogen and thus the concentration decreased Cell Voltage H 2 S concentration (2. ppm) H 2 S Concentration (ppm) Figure : Exit stream analysis of H 2 S. Ammonia The effect of NH 3 concentration was next evaluated. The concentrations of ammonia that were evaluated were:,.5,., 9. and 44.7 ppm. During the experiment, the test bench shut down at 84 hours into the experiment and it was restarted 9 hours later. Three hours after the restart, the impurities were shut off and the cells were allowed to recover under pure hydrogen for a period of 2 hours. There was some variability in the voltage of the control cell; as such, the cells were compared to the control cell and not to their initial performance. There did not appear to be a difference between the control cell and the.5 and. ppm concentration, Figure. The data for the., 9. and 44.7 ppm is given in Figure 2. The difference between the control and the 9. ppm cell was 2.8 % at 8 hours. The difference between the control and the 44.7 ppm was approximately 2 % during the period from 4 to 83 hours. 8/

9 WHEC 6 / 3-6 June 26 Lyon France.8 ppm Cell Voltage (volts) ppm. ppm Time (hours) Figure : Exposure to,.5 and. ppm ammonia.8 ppm Cell Voltage (volts) ppm ppm Time (hours) Figure 2: Exposure to, 9. and 44.7 ppm ammonia There was no significant recovery of the cells exposed to 9. ppm. It is not clear whether a longer time would result in full recovery. The cells exposed to 44.7 ppm showed a partial recovery; the difference between the control cell and the 44.7 ppm was 5.8 %, following the 2 hours of recovery. The polarization curves for the 44.7 ppm concentration is given in Figure 3. The curves show a continued decrease from to 74 hours. The recovery can be seen by comparing the polarization curve at 8 hours to the curve at 96 hours. 9/

10 WHEC 6 / 3-6 June 26 Lyon France Cell Voltage (V) hours 2 hours 22 hours 47 hours 74 hours 96 hours 8 hours Current Density (A/cm 2 ) Figure 3: Polarization curve of cell exposed to 44.7 ppm ammonia in hydrogen Air Impurities Carbon monoxide It has been previously mentioned that the level of CO in the ambient air may be at levels of approximately 3 ppm in urban areas. It has already observed that CO can affect the anode at similar concentrations. The concentrations that were tested were,.44,.9, 8.5 and 63.2 ppm. The results are given in Figure ppm.44 ppm.9 ppm 8.47 ppm 63.2 ppm Figure 4: Exposure to,.44,.9, 8.47, 63.2 ppm carbon monoxide Initially, there was some variability in cell voltages. This was observed after the conditioning of the cells and it is not attributable to the impurity. During the experiment, all of the voltages decreased slightly from their initial values. As a result of this, it is difficult to determine if there is any effect of carbon monoxide at any of the concentration levels. This is likely a result of the fact that carbon monoxide is easily oxidized at voltages between.6 and.9 volts [3]. There was no differences in the polarization curves that were performed. /

11 WHEC 6 / 3-6 June 26 Lyon France Conclusions Several impurities have been tested to determine the effect on the performance of a fuel cell. Carbon monoxide in hydrogen resulted in performance loses for all of the levels that were tested. The performance losses do not appear to be cumulative but eventually reach a steady state depending upon the CO concentrations. This is a result of the CO being converted to CO 2. The fuel cell can easily recover from CO poisoning. Hydrogen sulfide poisoning appears to be cumulative and not easily recoverable. Continued exposure results in catastrophic failure of the fuel cell. Because of the cumulative effects, it would be logical to conclude that even low levels (<.25 ppm) will poison a fuel cell; longer exposure times should be tested to verify this fact. For the time period of testing, ammonia did not appear to affect the fuel cell at concentrations less than ppm, longer testing should also be considered. At concentrations greater than 9 ppm, there was a loss of performance that did not appear to be easily recoverable. Carbon monoxide in air appears to have a negligible effect on the fuel cell at levels less than 63 ppm. References [] C. Koroneos, A. Dompros, G. Roumbas, N. Moussiopoulos, Life cycle assessment of hydrogen fuel production processes, International Journal of Hydrogen Energy Vol 29, pp , 24 [2] J. Rostrup-Nielsen, Conversion of hydrocarbons and alcohols for fuel cells, Physi. Chem., Chem., Vol 3 pp , 2 [3] F. A. Uribe, T. Zawodzinski, Jr. and S. Gettesfeld, Effect of ammonia as a possible fuel impurity on PEM fuel cell performance, Electrochemical Society Proceedings, Vol 98-27, p 229, 999 [4] K. Narusawa, Study on fuel cell poisoning resulting from hydrogen fuel containing impurities, Presented at the World Automotive Congress, Barcelona, Spain, paper F24F397, 24 [5] Environmental Protection Agency website [6] S. Watanabe, M. Tatsumi, M. Akai Hydrogen quality standard for fuel cell vehicles, Presented at the 24 Fuel Cell Seminar, San Antonio, TX, Nov 24 [7] J.J. Baschuk, Xianguo Li, Carbon monoxide poisoning of proton exchange membrane fuel cells, International Journal of Energy Research, , 2 [8] W. Vogel, J Lundquist, P. Ross, P. Stonehart, Reaction pathways and poisons-ii. The rate controlling step for electrochemical oxidation of hydrogen on Pt in acid and poisoning of the reaction by CO, Electrochimica Acta, 2 () 79-93, 975 [9] J. Baschuyuk and X. Li, Carbon monoxide poisoning of proton exchange membrane fuel cells, Internal Journal of Energy Research, 25, 695, 2 [] M. Farooque, T.Z. Fahidy, Low potential oxidation of hydrogen sulfide on a rotating tripolar wiper blade electrode via continuous anode reactivation, Journal of the Electrochemical Society, 24, 9, 977 [] J.R. Chang, S.L. Chang and B.T. Line gamma-alumina Supported Pt Catalysts for Aromatics Reduction: A Structural Investigation of Sulfur Poisoning Catalyst Deactivation, Journal of Catalysts, 69, 338, 997 [2] J.R. Chang, S.L. Chang Catalytic Properties of gamma-alumina supported Pt catalysts for tetralin hydrogenation, Journal of Catalysts, 76, 42,998 [3] A.M. de Becdelievre, J. de Becdelievre, J. Clavilier Electrochemical oxidation of adsorbed carbon monoxide on platinum spherical single crystals. Effect of anion adsorption Journal of Electroanalytical Chemistry and Interfacial Electrochemistry 294 (-2) 97-, 99 /