The production of CO₂ gas at the DMFC anode leads to dramatic increases in pumping power requirements and reduced power output because of mass transfer limitations as bubble trains form in the channels of larger stacks. Experimental observations taken in a 5 cm² DMFC test cell operated at 60ºC, 1 atm, and with a methanol/water fuel flow rates of 5-10 cc min^-1 indicate that the rate of bubble formation can be reduced by increasing the fuel flow because more liquid is available for the CO₂ to dissolve in. Further observations indicate that KOH and LiOH added to the fuel eliminates CO₂ gas formation in situ at low concentrations because of the greatly increased solubility that results.

A mathematical model for the volumetric rate of CO₂ gas production that includes effects of temperature and solubility is developed and extended to include the effects of hydroxide ions in solution. The model is used to predict the onset location of gas formation in the flow field as well as the void fraction at any point in the flow field. Predictions from the model agree very well with our experiments. Model predictions explain differences in the initial location of bubble formation for fuel solutions pre-saturated with CO₂ as opposed to CO₂-free solutions. Experiments with KOH and LiOH added to fuel solutions confirm the validity of the model extension that includes solubility that is enhanced by chemical reaction.

Experiments with LiOH, KOH, and ammonium hydroxide show that the long-term durability of standard Pt-Ru/Nafion/Pt membrane electrode assemblies is compromised because of the presence of lithium, potassium, and ammonium cations that interact with the Nafion membrane and result in increasing the ohmic limitations of the polymer electrolyte membrane. Experiments with Ca(OH)₂, while reducing gas formation, precipitate the product CaCO₃ out of solution too rapidly for downstream filtering, blocking channels in the flow field.