Hydrogen drying and purification

What leaves the stack during operation is moisture and trace oxygen. Nitrogen is not made by the stack; it enters from purge gas and air left in the pipework.

Moisture

The outlet gas is saturated. This is most of what the purity figure measures

Oxygen

Trace crossover from the anode. Fuel cell and refuelling specifications set a limit on it

Nitrogen

Enters from purge and residual air. It shows up at start-up and after maintenance, and does not grow during operation

99.999 percent means total impurities, water and oxygen and nitrogen together, at or below 10 ppmv. Dew point describes water alone. Using one figure in place of the other produces the wrong specification. The conversions below are on an atmospheric basis. At -70 °C water uses only a quarter of the total allowance, so it is compatible with 99.999 percent and oxygen removal covers the rest. At -60 °C water alone is already past 10 ppmv: removing every trace of oxygen still would not reach 99.999 percent, so an application with that requirement has to dry further.

Dew point -40 °C

About 128 ppmv — enough to stop condensation, short of 99.999 percent

Dew point -60 °C

About 10.7 ppmv — water alone is already past the 10 ppmv allowance, short of 99.999 percent

Dew point -70 °C

About 2.6 ppmv — compatible with 99.999 percent

-40 °C at atmospheric pressure and -40 °C at 10 barg are entirely different conditions. If you need a figure at line pressure we will convert it. One more reference point has to be agreed. The dryer outlet dew point is not the dew point at your point of use: moisture held on the pipe walls desorbs downstream and the reading there is higher. The longer the run, and the longer the system has stood idle, the wider that gap.

The downstream train is not a choice between methods so much as a stack built to the requirement. The layers are below. An application that only needs a lower dew point stops after the first one or two; an application with a purity figure adds the later ones. Of these, the only layer we can source today at this scale is palladium membrane purification. The rest either go into the HXD-V1 design or have no market product this small. That is why the table below is short.

Cooling and condensation

Chills the gas so water drops out as liquid. The simplest and cheapest arrangement, limited to a dew point near 0 °C by the freezing point of water. It earns its place by reducing the load on whatever follows

Adsorption, pressure swing (PSA)

Water is held on an adsorbent, then driven off by dropping the pressure, with two vessels alternating. Regeneration needs no heat and is quick, but some product gas leaves with the purge, so hydrogen is lost

Adsorption, temperature swing (TSA)

The same adsorption regenerated with heat instead of pressure. Regeneration takes longer and the heater draws power, but almost no product gas is discarded. This is the side to look at when hydrogen loss matters at small scale

Catalytic deoxo

Reacts the remaining oxygen with hydrogen to form water. The oxygen becomes moisture, so drying always follows. The reaction gives off heat, so cooling comes into the picture when the inlet oxygen is high

Getter (chemical absorption)

A heated metal chemically binds oxygen, moisture and nitrogen. It reaches ppb level and discards no hydrogen. What it binds it does not release, so the unit is replaced outright at end of life

Palladium membrane purification

Only hydrogen atoms cross a heated palladium membrane, leaving everything else behind and taking total impurities to ppb level. The membrane has to be held at around 400 °C, and gas that does not cross is vented, so recovery is lower

Polymer membrane separation

Separates on the difference in how fast species cross a membrane. Easy to run continuously and free of consumables, but the separation is incomplete so the achievable purity is limited. It belongs where continuity matters more than purity