The value of concentrated hydrogen is higher then the caloric value of unconcentrated hydrogen. Therefore it can be interesting to win back the hydrogen. A potential source could be the off-gas stream of an ammonia plant. Possible processes to win back the hydrogen are cryogene separation, pressure swing adsorption and membranes.

An alternative way to win back hydrogen is selective adsorption using metalhydrids, followed by a gas-solid separation. A second reactor is needed to desorb the (almost pure) hydrogen. Suitable metalhydrids include FeTi, LaNi₅ (see picture) and CaNi₅. For this study LaNi_4.7Al_0.3 was selected. This metalhydrid is very expensive, but it has very good adsorption and desorption characteristics at convenient pressures and temperatures. Besides it's very resistant against poisoning.

It appeared that heat transfer is the rate determing step in this process. Therefore a reactor with good heat-transfer is needed, like a bubbling or circulating fluidized bed. In this study a bubbling bed was selected. The small particles (about 7 mu) can be bounded to Nickel-particles of 20 µm to get better fluidisation. The main problem was estimating the heat transfer for a bubbling bed containing these pellets with a diameter of 20 µm and a very high density of over 8000 kg/m³. Further investigation was therefore recommended.

The process should be operated at a pressure of about 8 bar. If 90 % of the hydrogen has to be recovered the adsorption should take place at 20 °C and the desorption at 90 °C. For an economic attractive process heat integration between the adsorber and the desorber was recommended.