**2. Cell feed contributions to mass transport losses**

#### **2.1.1 Fuel cells**

It has been determined that when a PEMFC is operating under dilute hydrogen feed streams (as low as 40% H2 and a high utilization up to 90%) stack power losses should not exceed 10% of the power achieved with neat hydrogen (Springer, Rockward et al. 2001). When carbon monoxide was present in the dilute feed stream, the power losses were amplified significantly over the neat hydrogen feed stream. The authors suggest this problem may not be solved alone by changing anode catalysts and that a method such as air bleeding may need to be employed to achieve necessary power limits.

This work has been confirmed both experimentally and theoretically (Bhatia and Wang 2004) by other groups. The feed gases tested contained hydrogen contents as low as 40% and as high as 100%. The authors noted that the poisoning of CO was a quick process, taking less than 10 minutes for effects to be seen on the polarization curves for the fuel cell, taking two hours to reach steady state conditions. Yet, the poisoning process was reversible by feeding the cell with pure hydrogen for 2 hours. It was noted that CO preferentially adsorbs on the catalyst surface and when hydrogen is present in a dilute feed stream CO slows hydrogen adsorption even further, resulting in polarization losses. The hydrogen purity standard of the gas will depend dramatically on the dilution level of hydrogen.

The influence of ammonia on PEMFCs as been analyzed by only a few groups (Uribe, Gottesfeld et al. 2002; Soto, Lee et al. 2003; Halseid, Vie et al. 2006). In general, it was found that ammonia exposure has detrimental effects on the fuel cell performance. There was a steady loss of performance associated with the increase in current density and an overall increase in cell resistance. When the exposure was studied exposure from 1-30ppm NH3, it was found that the poisoning process was slow, up to 24 hours (Halseid, Vie et al. 2006). This poisoning was also reversible in most cases, but only after exposure to neat hydrogen for several days, while exposure to as low as 1ppm was found to have had detrimental effects on the fuel cell system performance. Ammonia was highly soluble in the membrane, but had no significant adsorption on the gas diffusion layer. This adsorption on the membrane by ammonia impurity was determined to have the largest effect on the oxygen reduction reaction, requiring an increase in power to drive the reaction to occur. The authors suggest that all ammonia must be removed from the feed stream before hydrogen can be used as a fuel and that the nitrogen content is closely monitored to prevent formation by metal-hydride alloys for hydrogen storage. The effects of ammonia at ppm and sub-ppm concentrations have been studied by (Martinez-Rodriguez, Fox et al. 2011) In their testing it was demonstrated during hydrogen pump experiments and electrochemical impedance spectroscopy that at concentrations of 10 ppm the effects of ammonia not only affect the solid electrolyte membrane, but at high current densities the resistance by the ionomer in the electrode is significantly higher than on the membrane. On the other hand during fuel cell testing, at 0.1 ppm the performance is unaffected by the ammonia.

An investigation on the effects of slightly higher concentrations of ammonia on PEMFC performance (Uribe, Gottesfeld et al. 2002), found the damage to the fuel cell to be irreversible, unlike previous results (Halseid, Vie et al. 2006). Even at 30ppm levels it was found the cell performance to drop considerably after several hours of exposure. The authors were able to successfully trap the ammonia using an ion exchange resin and continue use of the fuel cell without further damage.

Fuel cell systems are even more sensitive to sulfur containing compounds, yet few systematic studies have been completed on the phenomenon. Mohtadi et al. found that exposure to 5ppm of H2S would cause a 96% performance loss in a Pt catalyst based PEMFC (Mohtadi, Lee et al. 2005). This rate of poisoning was approximately 69% lower at 50oC than at 90°C. There was also evidence that sulfur crossed over at the cathode and affected the oxygen reduction reaction.

Recent research by Ballard Power Systems on a commercial stack suggests that not only is the source of a hydrogen impurity important, but it's point of induction also (Knights, Jia et al.). The following impurities were found to effect cell performance in decreasing order: H2S in fuel >SO2 in air > NO2 in air > NH3 in air > CO in fuel > NH3 in fuel. This suggests that the control of environmental air pollutants is as important for PEMFC operation as a high purity hydrogen standard. The changes in air quality could result in up to 30mV performance loss, which was most noticeable on cold, clear days. In order to address problems such as performance loss due to impurity effects, new catalysts or membranes are being developed.

Recent studies have been investigating the effect of trace halide contaminants on performance (Martínez-Rodríguez, Fox et al. 2011). The study of tetrachloroethylene, a common cleaning and degreasing agent, found that even at levels equal to the current ISO standards for hydrogen purity (ISO under development) detrimental impacts on fuel cell performance occur. At overpotentials above 0.2V, cell performance was fully recoverable. Poisoning that occurred at lower potentials was recoverable either by purging the cell or by changing the operating voltage.

### **2.1.2 Electrolyzers**

312 Mass Transfer - Advanced Aspects

impurities on fuel cell performance can be devastating. Trace impurities arising from different hydrogen production processes include carbon monoxide, carbon dioxide, ammonia, water, sulfur, hydrocarbons, oxygen, helium, nitrogen, argon, formaldehyde, formic acid and halogenates. The effect of the impurities can alter the catalytic activity of the catalyst, the ohmic resistance due to poisoning on the solid electrolyte and changes in the hydrophobicity of the pores affecting the water management in the system, which in turn affects the mass transport. Figures 7 and 8 shows a simplified schematic of the losses on the

It has been determined that when a PEMFC is operating under dilute hydrogen feed streams (as low as 40% H2 and a high utilization up to 90%) stack power losses should not exceed 10% of the power achieved with neat hydrogen (Springer, Rockward et al. 2001). When carbon monoxide was present in the dilute feed stream, the power losses were amplified significantly over the neat hydrogen feed stream. The authors suggest this problem may not be solved alone by changing anode catalysts and that a method such as air bleeding may

This work has been confirmed both experimentally and theoretically (Bhatia and Wang 2004) by other groups. The feed gases tested contained hydrogen contents as low as 40% and as high as 100%. The authors noted that the poisoning of CO was a quick process, taking less than 10 minutes for effects to be seen on the polarization curves for the fuel cell, taking two hours to reach steady state conditions. Yet, the poisoning process was reversible by feeding the cell with pure hydrogen for 2 hours. It was noted that CO preferentially adsorbs on the catalyst surface and when hydrogen is present in a dilute feed stream CO slows hydrogen adsorption even further, resulting in polarization losses. The hydrogen purity standard of

The influence of ammonia on PEMFCs as been analyzed by only a few groups (Uribe, Gottesfeld et al. 2002; Soto, Lee et al. 2003; Halseid, Vie et al. 2006). In general, it was found that ammonia exposure has detrimental effects on the fuel cell performance. There was a steady loss of performance associated with the increase in current density and an overall increase in cell resistance. When the exposure was studied exposure from 1-30ppm NH3, it was found that the poisoning process was slow, up to 24 hours (Halseid, Vie et al. 2006). This poisoning was also reversible in most cases, but only after exposure to neat hydrogen for several days, while exposure to as low as 1ppm was found to have had detrimental effects on the fuel cell system performance. Ammonia was highly soluble in the membrane, but had no significant adsorption on the gas diffusion layer. This adsorption on the membrane by ammonia impurity was determined to have the largest effect on the oxygen reduction reaction, requiring an increase in power to drive the reaction to occur. The authors suggest that all ammonia must be removed from the feed stream before hydrogen can be used as a fuel and that the nitrogen content is closely monitored to prevent formation by metal-hydride alloys for hydrogen storage. The effects of ammonia at ppm and sub-ppm concentrations have been studied by (Martinez-Rodriguez, Fox et al. 2011) In their testing it was demonstrated during hydrogen pump experiments and electrochemical impedance spectroscopy that at concentrations of 10 ppm the effects of ammonia not only affect the solid electrolyte membrane, but at high current densities the resistance by the ionomer in the

**2. Cell feed contributions to mass transport losses** 

need to be employed to achieve necessary power limits.

the gas will depend dramatically on the dilution level of hydrogen.

performance.

**2.1.1 Fuel cells** 

PEM electrolyzers have a thermoneutral voltage of 1.48V, below which H2 or O2 cannot be generated. Testing of single cell PEM electrolyzers, operated at 75°C, have produced cell efficiencies of 82% at 1 A/cm2 and 69% at 2 A/cm2 (Badwal, Giddey et al. 2006). Results indicate that the voltage losses experienced are ohmic in nature, or the voltage drop is the resistance of electron flow across the electrodes and interconnects of the cell. The cell was found to have better performance with thinner membranes, but these membranes have a shorter lifetime and are more fragile. The optimal operating current density of a water electrolyzer is between 0.5-1 A/cm2 , where resistances are minimized (Wendt and Imarisio 1988). Minimizing the ohmic resistance of the cell is important due to the high internal resistance and overvoltages experienced during operation. Cell efficiency will increase with decreasing resistance. Cell voltages will decrease with increasing cell temperature due to the decrease in overpotential and resistive losses (Onda, Murakami et al. 2002). If the individual cell is upgraded to small stacks of approximately fourteen cells, enough heat is generated due to internal resistive losses to make the cell thermally self-sustaining (Badwal, Giddey et al. 2006).

Mass Transport Limitations in Proton Exchange Membrane Fuel Cells and Electrolyzers 315

The efficiency of PEME are greatly affected by the water content of the feed. The mass transfer model presented in previous work (Fox, Greenway et al. 2008) predicted that the current for an anode liquid water feed electrolyzer would be around 8 times larger than the limiting current in a cathode water vapor feed electrolyzer. The difference in the cell polarization between these two feed configurations is shown in Figure 9. The mass transfer limiting current density for the cathode water vapor feed system is around 92 mA/cm2 while the current for the cathode water vapor feed system is near 475 mA/cm2 at 1.8V at 30°C. This current density could most likely be increased to 1000-1400 mA/cm2 if the a higher current density is desired and if it was determined that the higher voltage did not significantly affect the lifetime of the MEAs. Therefore, the cell current density produced by the anode liquid water feed system and thus the water reaction rate could be between 5-8 times larger than a similarly sized cathode water vapor feed system. To get an equivalent water processing rate between the two systems, either the reaction area of the cathode water vapor feed system or the number of cells used for processing the water would need to be

Membrane development is of particular interest due to the limitations of current Nafion® membranes such as temperature restrictions due to dehydration and subsequent loss of conductivity. In order to meet these demands researchers have attempted to improve the membrane by doping or by investigating new polymer membranes. These alternate routes may also be used to increase fuel cell performance in the presence of gas impurities such as

For example, the effects of carbon monoxide on alternative membranes such as poly(2,5 benzimidazole) have been investigated (Krishnan, Park et al. 2006). These polymers, doped with phosphoric acid, had the ability to be operated at temperatures up to 210°C with 1% CO without performance losses, which are higher temperatures and higher carbon monoxide concentrations than conventional MEA configurations are tolerant. Other investigations involve using alternatives such as glass papers to support organic membranes (Tezuka, Tadanaga et al. 2005). The membranes cast from 3-glycidoxypropyltrimethoxysilane and tetraalkoxysilane would otherwise be too thick and have to high of a resistance for viable fuel cell use. These membranes were able to achieve a maximum power density of

Other methods of development include (Jalani, Dunn et al. 2005) impregnating Nafion® to create more stable composite materials. The authors found that when Nafion® was impregnated with ZrO2, SiO2 and TiO2 the composite membranes has better water retention and thermal stability than Nafion® alone. ZrO2 impregnated Nafion® had the best performance overall and this is believed to be due to the increased acidity and surface area of the membrane. ZrO2 impregnated Nafion® was the only modified membrane that showed increased conductivity over Nafion®. Leading the authors to conclude that the distribution of water between the surface and bulk of a system is as important as the

Other than membrane development, an alternative method of improving the MEA is through catalyst development. A current area of interest is the use of non-precious metals or new binary catalysts for the oxygen reduction reaction at the cathode. Presently Pt and Pt alloys are widely used as anode and cathode materials in Proton Exchange Membrane

increased proportionally to the difference in current density.

**3. Mitigation of mass transport losses** 

80mW/cm2 at 130°C and 7% relative humidity.

amount of water absorbed.

carbon monoxide.

Uniform current density is an important factor to the lifetime of the membrane. A uniform current density is important to prevent the formation of hotspots, which can further decrease the performance of the electrodes and membranes. If the electrolyzer ratio (flow rate of electrolyzed water divided by the flow rate of the feed water) is less than 10%, the current density distribution in the cell is uniform in the cell (Onda, Murakami et al. 2002). At higher ratios, the current density will increase upstream where there is sufficient water and the current density will decrease downstream where there is insufficient water for the reaction.

The performance of PEM electrodes has been found to be sensitive to differential pressures as low as 20 mbar between the cathode and anode sides of the cell (Millet, Andolfatto et al. 1996). If high operating pressures are used, the differential pressure must be controlled. Stainless steel pipes in contact with deionized water can cause a steady decrease in cell voltage. Low concentrations of Fe, Ni and Cr from the stainless steel became concentrated in the membrane, limiting cell performance. On-line deionizers were found to give a more stable performance, but can impose limitations when used with high operating pressures and temperatures.

Demonstration electrolyzer plants, rated at 100kW, have been successfully run for up to 15,000 hours (Stucki, Scherer et al. 1998). One plant was shut down after 15,000 hours due to hydrogen concentrations in the oxygen off-gas of higher than 3 percent. A second demonstration plant, run for only 2300 hours with 50,000 hours of standby operation, was shut down for the same reason. During the standby period, a protective polarization current of 0.34 mA/cm2 was applied in order to prevent corrosion of the current collectors at the cathode. Post mortem analyses of membranes from both plants indicate that stack failure was due to thinning of the Nafion® 117 membrane. The non-uniform membrane thinning coincided with an observed decrease in cell voltage. The ion exchange capacity of the membranes remained consistent throughout operation.

Fig. 9. Cell polarization comparison for anode liquid feed and cathode vapor feed PEM electrolyzer cells at 30°C. () Anode liquid feed (U) Cathode vapor feed. Reproduced from (Greenway, Fox et al. 2009)

The efficiency of PEME are greatly affected by the water content of the feed. The mass transfer model presented in previous work (Fox, Greenway et al. 2008) predicted that the current for an anode liquid water feed electrolyzer would be around 8 times larger than the limiting current in a cathode water vapor feed electrolyzer. The difference in the cell polarization between these two feed configurations is shown in Figure 9. The mass transfer limiting current density for the cathode water vapor feed system is around 92 mA/cm2 while the current for the cathode water vapor feed system is near 475 mA/cm2 at 1.8V at 30°C. This current density could most likely be increased to 1000-1400 mA/cm2 if the a higher current density is desired and if it was determined that the higher voltage did not significantly affect the lifetime of the MEAs. Therefore, the cell current density produced by the anode liquid water feed system and thus the water reaction rate could be between 5-8 times larger than a similarly sized cathode water vapor feed system. To get an equivalent water processing rate between the two systems, either the reaction area of the cathode water vapor feed system or the number of cells used for processing the water would need to be increased proportionally to the difference in current density.
