Membrane Reactors for Hydrogen Production

A. BRUNETTI, A. CARAVELLA, E. DRIOLI AND G. BARBIERI

1.1 Introduction

In the last decade, the energy demand has grown by 1.2% a year and fossil fuels still maintain a production share of ca. 75%. However, the ever stricter problems connected to sustainable growth and lower environmental impact lead to the conclusion that the times of easy oil consumption are over. Nowadays, the necessity to produce energy from oil and natural gas as primary energy sources is becoming more and more pressing. Indeed, more generally, the diversification of said sources in order to ensure a constant supply makes the interest in membrane reactor (MR) technology more urgent. Moreover, the increasing efforts dedicated to the reduction of environmental problems has recently led to the development of clean technologies, designed to enhance both the efficiency and environmental acceptability of energy production, storage, and use, in particular for power generation. Among these technologies, the exploitation of light hydrocarbons is surely the main realistic energy source, since they allow both power generation and environmentally friendly fuel production. Specific reference should be made to hydrogen in this context.

At present, the global hydrogen production relies mainly on processes that extract hydrogen from fossil fuel feedstocks. About 96% of hydrogen is directly produced from fossil fuels and about 4% is produced indirectly using electricity generated through them.The stream coming out from a reformer or a coal gasification plant contains around 50% hydrogen (on a dry basis) that must be recovered and between 40–45% CO that is usually reduced in an upgrading stage, producing more hydrogen at the same time. In traditional applications (Figure 1.1), the upgrading of reformate streams is performed using a multi-stage CO-shift process based on a series of catalytic reactors: the first one operates at high temperatures (about 350–400 °C) and takes advantage of the high reaction rate, converting a large portion of CO into hydrogen and CO2; the other one operates at lower temperature (around 220–300 °C) and refines the carbon monoxide conversion, thus allowing a lower final concentration of CO (less than 1% molar). This H2-rich stream coming out from the last reactor is fed to a pressure swing adsorption (PSA) unit for H2 separation from other gases. It should be pointed out that the new utilization of H2 as feed in fuel cells for mobile power sources requires the anode inlet gas to have a CO concentration below 10–20 ppmin order to avoid catalyst poisoning with subsequent drops in fuel cell efficiency. Hence, the purification step for the H2 produced from hydrocarbons must be very efficient to fulfil said fuel cell requirements. Because of this, in some cases, another reaction unit is added to oxidize CO into CO2.

One of the main challenges in the next few years will be the identification of new technologies able to provide better exploitation of fossil fuels, e.g., hydrocarbons, in order to improve the yield, energy savings, and so on. The reduction of the number of reaction/separation/purification stages, which translates into a lower footprint area occupied by the whole plant, fewer auxiliary devices, reduction of the energetic load, and so forth, is a fundamental issue to consider when redesigning hydrogen production processes. A promising approach for concretizing these technological aspects in the field of hydrogen production is the use of MRs, combining the reaction and H separation by means of selective membranes. Many studies are now focused on the analysis of MR performance, where light hydrocarbon reforming or water–gas shift (WGS) reactions are carried out. In these cases, for both reactions, the presence of a membrane allows the recovery of a hydrogen-rich stream that does not require further separation/purification. Moreover, the removal of H, the reaction product, from the reaction volume shifts the reaction toward further conversion. This means the possibility of having an intensified process with a reduced plant size and higher yield. The traditional process can thus be redesigned in a more compact and efficient manner (Figure 1.2), following the logic of the Process Intensification Strategy, which is an innovative methodology for process and plant design proposing a new design philosophy to achieve significant reductions (by factors of 10 to 100 or more) in plant volume at the same production capacity or to improve the overall efficiency.

Figure 1.2 shows an integrated membrane system constituted by fewer reaction/separation units than the conventional one (Figure 1.1). A first MR can be used to carry out the reforming of light hydrocarbons and another reactor for the WGS reaction.

The presence of a membrane in both reactors allows the separation of a hydrogen-rich stream from the two reaction volumes, as well as improvements in the conversion of the two stages. Obviously, the H2 purity level strictly depends on the membrane type used in each MR. In fact, membranes can be distinguished by their selectivity, which can be infinite or finite. The first ones, traditionally Pd-based, allow a pure hydrogen stream to be obtained, whereas the others provide a hydrogen-rich stream of variable purity. If the recovered H stream does not have the purity required, the latter can be increased by adding another purification unit depending on the final use of the H2 stream. Selective CO oxidation is known as an interesting and economical approach for CO removal from H2-rich gas streams. Also in this field, new studies proposed in the literature have demonstrated how the use of MRs can improve the process by increasing the CO conversion as well as the purity of the hydrogen stream.

In this context, membrane engineering plays a fundamental role in the integration of these units into a single plant and, at the same time, in the definition of the knowledge necessary to drive the process by maximizing the gains, both in terms of efficiency and plant size reduction. The synergic effects offered by MRs by combining reaction and separation processes in the same unit, their simplicity, and the possibility of advanced levels of automation and control offer an attractive opportunity to redesign industrial processes.

1.2 Membranes for Hydrogen Production

MRs represent the most significant class of the so-called multifunctional reactors, which integrate reaction and separation processes in the same unit. Membranes for hydrogen separation should exhibit high selectivity toward hydrogen and high flux, being in the meantime highly mechanically and chemically stable. These aspects are very well addressed in Chapter 4 of this volume.

Most of the membranes used in hydrogen production allow the selective removal of Hfrom the reaction volume under the effect of a driving force. This is a function of the species partial pressure on each membrane side and can be created by means of an inert sweep gas in the permeate compartment (nitrogen, helium, water, etc.), or by application of a pressure difference between the retentate and permeate sides.

For Pd-alloy membranes, Sieverts' law (eqn (1.1)) is used worldwide for the mathematical description of H permeating fluxes in this type of membranes. Accordingly, the hydrogen permeating flux is a linear function of the permeability and driving force and a reverse function of the membrane thickness. The permeation driving force in Sieverts' law is the difference of the square root of the hydrogen partial pressure on both membrane sides.

[MATHEMATICAL EXPRESSION OMITTED] 1.1

The removal of a product such as hydrogen from the reaction volume implies a series of advantages:

• conversion enhancement of equilibrium-limited reactions,

• depletion of undesired secondary reactions,

• recovery of concentrated rich streams: pure H2 in the permeate, CO2 concentrated and compressed in the retentate,

• coupling of two or more reactions, e.g., dehydrogenation (endothermic) with hydrogenation (exothermic) on the two membrane sides,

• more desirable operating conditions (e.g., temperature).

The thermodynamic equilibrium limit of a traditional reactor (TR) can be exceeded owing to the removal of the product from the reaction volume, obtaining higher conversion under analogous operating conditions. In other words, for endothermic reactions, this allows the MR to achieve the same conversion of a TR at significantly lower temperatures. Another interesting aspect of MR usage is the positive effect that the reaction pressure can have on the process, also for reactions taking place without mole number variation (e.g., WGS) or with a mole number increase (e.g., methane steam reforming, SMR).

In hydrogen production, dense or microporous membranes can be used depending on the role of the membrane, whether it is aimed at H separation or purification. Most studies reported in the open literature show that membranes can be separated into dense metallic Pd-based membranes and ceramic membranes (silica, zeolite, etc.). The former show permselective transport governed by a solution-diffusion mechanism. Microporous ceramic membranes can present both permselective and non-permselective transport, depending on the size of the permeating molecules with respect to the membrane pore size as well as the chemical nature of the permeating molecules and the membrane material.

1.3 MR Configurations for Hydrogen Production

In the past, MRs were studied by carrying out several gaseous phase reactions with different membrane types, in particular for high temperature operations, as firstly proposed by Prof. Gryaznov in the late 1960s. Since then, many papers have been published on the use of MRs for hydrogen production via various reactions.

Most of the studies carried out to date on MRs have focused on equilibrium- limited reactions, where the permeation of the product enhances the conversion with respect to that of a TR. Other new applications propose the use of membranes as contactors between catalysts and reactants. However, even though MR studies on pilot plants have returned promising results supporting the wide- ranging potential of this technology, there are currently no large-scale applications of MRs. Different types of MRs for hydrogen production have been proposed in the literature. Most works refer to packed bed MRs; however, other configurations such as fluidized bed MRs and micro-MRs have also been recently introduced.

1.3.1 Packed Bed MRs

Usually, packed bed MRs have a tubular configuration where the outer tube is the shell side and the inner tube is the membrane. The catalytic bed can be confined in the core of the membrane or in the annulus between the two tubes (Figure 1.3), while the permeate stream is recovered on the other side of the membrane. In the case of multi-tubular configurations, the catalyst is packed in the shell side both for construction reasons as well as for reduced heat and mass transfer limitations. One of the mains drawbacks claimed for packed bed MRs is their external mass transfer limitations, such as the limitations to hydrogen transport between the bulk of the catalytic bed (where hydrogen is produced) and the membrane wall, especially for high flux membranes.

Most recently, Caravella et al. investigated the concentration gradient distributions in Pd-based MRs for the WGS reaction by considering a 3.6 µm-thick membrane. The already developed and validated multicomponent-based permeation model was updated to account for the presence of the particle (catalyst) bed. It was demonstrated that the velocity field between particles and membranes contributes to the enhancement of the mass transfer toward the membrane surface and that the particle size does not provide an appreciable contribution toward changing the concentration polarization level in the reactor, at least for mono-disperse particles. The simulation results indicated that the maximum concentration polarization in the reactor was ca. 20%. This high value, present at the reactor end, is caused by the low hydrogen concentration, which implies a larger resistance to mass transport owing to non-permeating species. However, the weight of this reactor section on the overall concentration polarization was not so high, ca. 10.5% in average, which is significantly lower than the maximum value.

1.3.2 Fluidized Bed MRs

A typical fluidized MR for hydrogen production consists of hydrogen-selective membranes immersed in a catalytic bed operated in the bubbling or turbulent regime. The main advantage of this reactor is the negligible pressure drop, which allows using small particle sizes resulting in no internal mass and heat transfer limitations. Moreover, fluidized beds are also suitable for isothermal operations even if a highly exothermic reaction is occurring, as demonstrated by Deshmukh et al., who carried out the oxidative dehydrogenation of methanol in lab-scale membrane fluidized bed reactors. This important aspect allows the auto-thermal reforming of methane (and other hydrocarbons) by feeding oxygen directly into the MR, preventing the formation of hot spots and subsequent damage to the membranes. An example of a fluidized bed MR is shown in Figure 1.4.

1.3.3 Micro-MRs

Membrane micro-reactors or micro-MRs can be defined as micro-reactors reinforced by membrane separation/purification or MRs miniaturized into characteristic dimensions of 1–1000 µm, combining the advantages of both MRs and micro-reactors, leading to greatly intensified operation units. The improvement of mass/heat transfer owing to the reduction of the scale length and the enhancement of the surface area-to-volume ratio owing to the extremely high intensification are the main advantages of micro-MRs.

A lot of research is currently devoted to the study of micro-MRs with hydrogen separation function as they have found a number of applications, such as hydrogen production from the water–gas shift (WGS) reaction, hydrogen production from the methanol steam reforming reaction, on-board fuel processing for portable PEMFCs (Polymer Electrolyte Fuel Cells), production of moisture-free formaldehyde by the dehydrogenation of methanol, and dehydrogenation of cyclohexane to benzene. Three different configurations can be found for these reactors: planar, hollow-fiber, and monolithic.

Microchannels in a planar configuration based on microelectromechanical systems were the first micro-MRs investigated. Mejdell et al. constructed a microchannel MR in a planar configuration from thin defect-free Pd/23 wt% Ag membranes. As shown in Figure 1.5, this microchannel MR consisted of a stainless steel feed channel plate with six parallel channels with dimensions of 1 mm × 1 mm × 13 mm. The Pd/23 wt% Ag membrane was placed between the channel housing and a stainless steel plate with apertures corresponding to the channel geometry. Such stainless steel plate was employed for mechanical support.

In the hollow fiber configuration, the diameter of a membrane tube is reduced to below 1000 µm. Catalysts can be coated on the inner surface of the hollow fibers or impregnated inside the porous wall, whilst the separation can be achieved by the porous hollow fibers themselves or by a membrane formed on the outer surface of the hollow fibers.

Honey-comb or straight-channel monoliths provide an inexpensive and rapid means for constructing scalable two-dimensional arrays of identical square microchannels with diameters of 500–5000 µm and wall thicknesses of 200–2000 µm. This kind of structures can be prepared from a variety of porous ceramic materials, such as cordierite, mullite, and alumina, which afford large networks of micro-MRs. Monolithic micro-MRs provide much better mechanical stability than hollow-fiber micro-MRs and much higher intensification than planar microchannel MRs (Figure 1.6).