Although nowadays hydrogen is mainly produced on a large-scale, small-scale delocalized hydrogen production from methane will have an important role in future. In fact, large-scale processes, such as steam reforming, suffer of problems regarding elevated gradients of temperature between the catalytic bed and the walls of the reactor. In addition, large and expensive hydrogen purification units, namely PROXY and PSA, to eliminate co-produced CO and CO2 are needed. Finally, hydrogen storage and transport is challenging.
Small-scale plants can overcome these issues. However, the integration between small-scale production and separation has to be improved. For this reason, the processes studied in this work are the methane oxy-reforming and the membrane reactor unit for the water gas shift reaction and hydrogen separation. These two have characteristics that make them suitable for an integrated process, such as the possibility to run them on low contact times and reduce some of the limitations found on large-scale. In fact, the oxy-reforming process, which consists in the combination of Steam Reforming (SR) and Catalytic Partial Oxidation (CPO) of methane in a configuration that provides low amount of oxygen (O/C=0.21) and steam (S/C=0.7-1.5) and short residence times (36-150 ms), helps to reduce some of the heat transfer issues found in classical SR and CPO However, the low amount of reagents employed disfavour methane conversion and results in deactivation processes such as carbon formation and sintering. For this reason, an efficient catalyst with good properties must be used to fulfil high hydrogen production for prolonged time. On the opposite, the low amount of reagents employed in the oxy-reforming make it a process suitable for an integration with a hydrogen-separating membrane reactor. In fact, the driving force of this separation is hydrogen partial pressure, which would be badly affected by an over-stoichiometric amount of reagents.
Thus, the process was integrated with a membrane reactor based on an efficient WGS Pt-catalyst and a selective Pd-based membrane that allowed hydrogen separation at a high rate, which was close to the hydrogen production rate over the catalyst.
- Results and discussion
A Ce0.5Zr0.5O2 support for both oxy-reforming and water gas shift was synthesized by reverse microemulsion technique1. This support was impregnated with Rh for the oxy-reforming catalyst and with Pt for the water gas shift one and provided stable and efficient catalyst in both cases. The Rh-based catalyst provided methane conversions close the equilibrium in all of the oxy-reforming conditions, showing a high stability, namely a decrease of conversion of only 3% after about 30 h of operation. The best performances were achieved at T=750°C, P=5 atm GHSV=100000 h-1 and S/C=1.
To increase hydrogen production and provide it in a pure stream, a hydrogen selective Pd-based membrane filled with a Pt-Ce0.5Zr0.5O2 catalyst for the water gas shift reaction was integrated downstream the oxy-reforming. The Pt/Ce0.5Zr0.5O2, good activity and its performances in the membrane reactor were compared with those of a fixed bed charged with the same catalyst. In the fixed bed operation, the catalyst proved to be selective toward water gas shift reaction, with low methanation activity, even at low S/C and H/C ratios. Though, hydrogen was produced in a good yield and CO was mainly converted through WGS. Decreasing the contact time also helped to reduce the methanation activity, as the WGS reaction was favoured in these conditions. A decrease in conversion was found, but this was to be attributed to a lower consumption of CO in the methanation reaction rather than a decrease of WGS occurrence. In fact, H2 yield was increased concurrently.
However, several advantages were obtain by substituting the fixed bed with the membrane reactor setup. In particular, the methanation reaction was completely avoided in all the conditions, even at low S/C ratios due to hydrogen removal from the reaction environment. Moreover, for the same reason the equilibrium was shifted, resulting in increased hydrogen yields and carbon monoxide conversions that were higher than those calculated at the equilibrium for the fixed bed process in the same conditions. The membrane provided total hydrogen selectivity and a high permeance of 126 ml/(cm2atm-0.5min) which was comparable to the hydrogen production rate and thus allowed to recover most of the hydrogen produced thanks to the good integration of the oxy-reforming and WGS membrane reactor. The best performances were obtained at T=400°C, P=10 atm, GHSV=100000 h-1 and S/C=1,5 and provided a hydrogen recovery of 90%, with a high permeate flux (731.44 ml/min) of pure hydrogen and a lower one (404.65 ml/min) in the retentate. As the membrane potential was not completely exploited, with some hydrogen left in the retentate, a sweep gas may be applied in order to further enhance hydrogen separation.
At lower GHSV, The driving force of the separation approached 0, indicating the complete membrane exploitation. However, lower fluxes and recoveries were obtained and a higher amount of unseparated hydrogen. Lowering the pressure to 5 atm, allowed for a better exploitation of the membrane as well but only 325.14 ml/min were found in the permeate, much lower than 690.38 ml/min of the retentate. However, when the results were compared with those obtained in a conventional WGS reactor at same conditions, an improvement of hydrogen production was observed, with values which overtook the equilibrium.
Pure hydrogen can be produced from methane in a two-step process consisting of a first oxy-reforming unit followed by a membrane reactor in which the water gas shift reaction and hydrogen separation are carried out simultaneously. This setup proved to be optimal for a small-scale integrated process and performed better than an oxy-reforming unit followed by a fixed bed water gas shift reactor. In particular, by linking oxy-reforming and membrane reactor pure hydrogen was produced in higher yields, with an increased CO conversion that was found to be over the equilibrium value calculated for the fixed bed in the same conditions. Finally, methanation reaction, which consumes the hydrogen produced in the reforming, could be avoided.
- A. Martínez-Arias, et al., Langmuir 1999, 15, 4796–4802.