Literature review has been performed in order to identify recent publications on hydrogen separation methods, hydrogen solubility, materials and concepts in research institutes and laboratories. The aim of the performed literature survey was to monitor recent worldwide literature and find out whether some of the developed and reported solutions might possibly help to improve existing hydrogen separation concept in PDh system, enabling efficient complete separation of hydrogen from all unwanted hydrocarbons.
Literature survey on hydrogen separation technique
Basically there are four important methods applied to the separation of gases in the industry: absorption, adsorption, cryogenic and membranes.
Pressure swing adsorption (PSA) is a gas purification process consisting of the removal of impurities on adsorbent beds. The usual adsorbents and gases adsorbed are molecular sieves for carbon monoxide, activated carbon for CO2, activated alumina or silica gel. Industrial PSA plants consist of up to 12 adsorbers and along with the number of valves required this makes the system rather complicated and complex. The PSA process is usually a repeating sequence of the following steps: adsorption at feed pressure, co-current depressurisation to intermediate pressure, counter-current depressurisation to atmospheric pressure usually starting at 10 % to 70 % of the feed pressure, counter-current purge with hydrogen enriched or product gas at ambient pressure, co-current pressure equalisation and finally, co-current pressurisation with feed or secondary process gas. For hydrogen purification by PSA hydrogen purity is high but the amount of rejected hydrogen is also relatively high (10 – 35 %). It seems also that cryogenic technology might not be applicable for PDh process gas separation. Cooling down the mixture will finally end in a solid jet fuel and a gas phase. Handling the solid is more difficult when compared with liquid. During the survey it became evident that membrane technology is the most popular, used and still investigating for the improvement process for hydrogen separation therefore the focus of the study is mainly on this technique.
The membrane separation process involves several elementary steps, which include the solution of hydrogen and its diffusion as atomic hydrogen through the membrane bulk material. Nowadays, membrane technologies are becoming more frequently used for separation of wide varying mixtures in the petrochemical related industries. According to Sutherland it is estimated that bulk chemicals and petrochemicals applications represented about 40% of the membrane market in the whole chemicals industry or about $ 1.5 billions, growing over 5 % per year. Membrane gas separation is attractive because of its simplicity and low energy cost.
The advantages of using membrane gas separation technologies could be summarized as following:
- Continuous and clean process, membranes do not require regeneration, unlike the adsorption or the absorption processes, which require regeneration step leading to the use of two solid beds or a solvent regeneration unit. Required filtration system is simple and inexpensive.
- Compared with conventional techniques, membranes can offer a simple, easy-to-operate, low-maintenance process.
- Membrane process is simple, generally carried out at atmospheric conditions which, besides being energy efficient, can be important for sensitive applications in pharmaceutical and food industry.
- The recovery of components from a main stream using membranes can be done without substantial additional energy costs.
Membrane is defined essentially as a barrier, which separates two phases and restricts transport of various chemicals in a selective manner. A membrane can be homogenous or heterogeneous, symmetric or asymmetric in structure, solid or liquid; can carry a positive or negative charge or be neutral or bipolar. Transport through a membrane can be affected by convection or by diffusion of individual molecules, induced by an electric field or concentration, pressure or temperature gradient. It takes place when a driving force is applied to the components in the feed. In most of the membrane processes, the driving force is a pressure difference or a concentration (or activity) difference across the membrane. Another driving force in membrane separations is the electrical potential difference. This driving force only influences the transport of charged particles or molecules.
The hydrogen separation factor is sometimes used to specify membrane quality. It is defined as following:
where ni stands for moles of species i transferred through the membrane and ?pi stands for the partial pressure difference of species i through the membrane.
The membrane thickness may vary from as small as 10 microns to few hundred micrometers. Basic types of membranes are presented in Figure 4.
Membranes in petrochemical industry are mainly used for concentration, purification and fractionation however they may be coupled to a chemical reaction to shift the chemical equilibrium in a combination defined as a membrane reactor. Using a membrane is adding costs to any process, therefore in order to overcome the cost issue another advantages must overcome the added expenses like material with a very good separation factor, high flux, high quality membrane materials (stable during many months of operation). In a membrane separation reactor both organic and inorganic membranes can be used. Many industrial catalytic processes involve the combination of high temperature and chemically harsh environments favouring therefore inorganic membranes due to their thermal stability, resistance to organic solvents, chlorine and other chemicals. Some promising applications using inorganic membranes include certain dehydrogenation, hydrogenation and oxidation reactions like formation of butane from dehydrogenation of ethyl benzene, styrene production from dehydrogenation of ethyl benzene, dehydrogenation of ethane to ethane, oxidative coupling of methane etc. In membrane reactor two basic concepts can be distinguished as can be seen in Figure 5.
- reaction and separation combined in one reactor (catalytic membrane reactor)
- reaction and separation are not combined and the reactants are recycled along a membrane system (membrane recycle reactor)
Catalytic membrane reactor concept is used especially with inorganic membranes (ceramics, metals) and polymeric membranes where the catalyst is coupled to the membrane. Membrane recycle reactor can be applied with any membrane process and type of membranes. Most of the chemical reactions need catalyst to enhance the reaction kinetics. The catalyst must be combined with the membrane system and various arrangements are possible, as can be seen in Figure 6. The advantage of the catalyst located inside the bore of the tube is simplicity in preparation and operation. When needed the catalyst could be easily replaced. In case of top layer filled with catalyst and membrane wall, the catalyst is immobilized onto the membrane.
Palladium has been known to be a highly hydrogen permeable and selective material since the 19th century. The existing Pd-based membranes can be mainly classified into two types according to the structure of the membrane as (i) self-supporting Pd-based membranes and (ii) composite structures composed of thin Pd-based layers on porous materials. Most self-supporting Pd-based membranes are commercially available in the forms that are easily integrated into a separation setup. However these membranes are relatively thick (50 mm or more) and therefore the hydrogen flux through them is limited. Thick palladium membranes are expensive and rather suitable for use in large scale chemical production. For practical use it is necessary to develop separation units with reduced thickness of the layer. An additional problem is that in order to have adequate mechanical strength, relatively thick porous supports have to be used. In the last decade a significant research has been carried out to achieve higher fluxes by depositing thin layers of Pd or Pd alloys on porous supports like ceramics or stainless steel. A submicron thick and defect-free palladium-silver (Pd-Ag) alloy membrane was fabricated on a supportive microsieve by using microfabrication technique and tested by Tong et al. The technique also allowed production of a robust wafer-scale membrane module which could be easily inserted into a membrane holder to have gas-tight connections to outside. Fabricated membrane had a great potential for hydrogen purification and in application like dehydrogenation industry. One membrane module was investigated for a period of ca. 1000 hours during which the membrane experienced a change in gas type and its concentration as well as temperature cycling between 20 – 450 °C. The measured results showed no significant reduction in flux or selectivity, suggesting thus very good membrane stability. The authors carried out experiments with varying hydrogen concentration in the feed from 18 to 83 kPa at 450 °C to determine the steps limiting H2 transport rate. It is assumed that the fabricated membrane may be used as a membrane reactor for dehydrogenation reactions to synthesize high value products although its use may be limited due to high pressures of tens of bars. Schematic drawing of the hydrogen separation setup is presented in Figure 7. The membrane module was placed in a stainless steel holder installed in a temperature controlled oven to ensure isothermal operation. The H2/He feed (from 300 to 100 ml/mol) was preheated in spirals placed in the same oven. The setup was running automatically for 24 h/day and could handle 100 recipes without user intervention.
Tucho et al. performed microstructural studies of self-supported Pd / 23 wt. % Ag hydrogen separation membranes subjected to different heat treatments (300/400/450 °C for 4 days) and then tested for hydrogen permeation. It was noted that changes in permeability were dependent on the treatment atmosphere and temperature as well as membrane thickness. At higher temperatures significant grain growth was observed and stress relaxation occurred. Nam et al. were able to fabricate a highly stable palladium alloy composite membrane for hydrogen separation on a porous stainless steel support by the vacuum electrodeposition and laminating procedure. The membrane was manufactured without microstructural change therefore it was possible to obtain both high performance (above 3 months of operation) and physical and morphological stability of the membrane. It was observed that the composite membrane had a capability to separate hydrogen from gas mixture with complete hydrogen selectivity and could be used to produce ultra-pure hydrogen for applications in membrane reactor. Tanaka et al. aimed at the improved thermal stability of mesoporous Pd-YSZ-g-Al2O3 composite membrane. The improved thermal stability allowed operation at elevated temperature (> 500 °C for 200 hours). This was probably the result of improved fracture toughness of YSZ-g-Al2O3 layer and matching thermal expansion coefficient between palladium and YSZ. Kuraoka, Zhao and Yazawa demonstrated that pore-filled palladium glass composite membranes for hydrogen separation prepared by electroless plating technique have both higher hydrogen permeance, and better mechanical properties than unsupported Pd films. The same technique was applied by Paglieri et al. for plating a layer of Pd and then copper onto porous ?-substrate. Zahedi et al. developed a thin palladium membrane by depositing Pd onto a tungsten oxide WO3 modified porous stainless steel disc and reported that permeability measurements at 723, 773 and 823 K showed high permeability and selectivity for hydrogen. The membrane was stable with regards to hydrogen for about 25 days. Certain effort has been performed for improving hydrothermal stability and application to hydrogen separation membranes at high temperatures. Igi et al. prepared a hydrogen separation microporous membranes with enhanced hydrothermal stability at 500 °C under a steam pressure of 300 kPa. Co-doped silica sol solutions with varying Co composition (Co / (Si + Co) from 10 to 50 mol. %) were prepared and used for manufacturing the membranes. The membranes showed increased hydrothermal stability and high selectivity and permeability towards hydrogen when compared with pure silica membranes. The Co-doped silica membranes with a Co composition of 33 mol. % showed the highest selectivity for hydrogen, with a H2 permeance of 4.00 x 10-6 (m3 (STP) × (m × s × kPa)-1) and a H2/N2 permeance ratio of 730. It was observed that as the Co composition increased as high as 33 %, the activation energy of hydrogen permeation decreased and the H2 permeance increased. Additional increase in Co concentration resulted in increased H2 activation energy and decreased H2 permeance. Due to high permselectivity of Pd membranes, high purity of hydrogen can be obtained directly from hydrogen containing mixture at high temperatures without further purification providing if sufficient pressure gradient is applied. Therefore it is possible to integrate the reforming reaction and the separation step in a single unit. A membrane reformer system is simpler, more compact and more efficient than the conventional PSA system (Pressure Swing Adsorption) because stem reforming reaction of hydrocarbon fuels and hydrogen separation process take place in a single reactor simultaneously and without a separate shift converter and a purification system. Gepert et al. have aimed at development of heat-integrated compact membrane reformer for decentralized hydrogen production and worked on composite ceramic capillaries (made of ?-Al2O3) coated with thin palladium membranes for production of CO-free hydrogen for PEM fuel cells by alcohol reforming. The membranes were tested for pure hydrogen and N2 as well as for synthetic reformate gas. The process steps comprised the evaporation and overheating of the water/alcohol feed, water gas shift combined with highly selective hydrogen separation. The authors have focused on the step concerned with the membrane separation of hydrogen from the reforming mixture and on the challenges and requirements of that process. The challenges encountered with the development of capillary Pd membranes were as following: long term temperature and pressure cycling stability in a reformate gas atmosphere, the ability to withstand frequent heating up and cooling down to room temperature, avoidance of the formation of pin-holes during operation and the integration of the membranes into reactor housing. It was observed that palladium membranes should not be operated at temperatures below 300 °C and pressures lower than 20 bar, while the upper operating range is between 500 and 900 °C. Alloying the membrane with copper and silver extend their operating temperature down to a room temperature. The introduction of silver into palladium membrane increases the lifetime, but also the costs when compared with copper. Detailed procedure of membrane manufacturing, integration into reformer unit and testing is described by the authors. Schematic of the concept of the integrated reformer is shown in Figure 8. The membrane was integrated in a metal tube embedded in electrically heated copper plates. Before entering the test tube, the gases were preheated to avoid local cooling of the membrane. Single gas measurements with pure N2 and H2 allowed the testing of the general performance of the membrane and the permselectivity for the respective gases to be reached. Synthetic reformate gas consisting of 75 % H2, 23.5 % CO2 and 1.5 % CO was used to get information about the performance. The membranes were tested between 370 – 450 °C and pressures up to 8 bar. The authors concluded that in general the membranes have shown good performance in terms of permeance and permselectivity including operation under reformate gas conditions. However, several problems were indicated concerning long-term stability under real reforming conditions, mainly related to structural nature (combination of different materials: ceramic, glaze, palladium resulted on incoherent potential for causing membrane failure). At operation times up to four weeks the continuous Pd layer remained essentially free from defects and pinholes.
Han et al. have developed a membrane separation module for a power equivalent of 10 kWel. A palladium membrane containing 40 wt. % copper and of 25 mm thickness was bonded into a metal frame. The separation module for a capacity of 10 Nm3 h-1 of hydrogen had a diameter of 10.8 cm and a length of 56 cm. Reformate fed to the modules contained 65 vol. % of hydrogen and the hydrogen recovery through the membrane was in the range of 75 %. Stable operation of the membrane separation was achieved for 750 pressure swing tests at 350 °C. The membrane separation device was integrated into a methanol fuel processor. Pientka et al. have utilized a closed-cell polystyrene foam (Ursa XPS NIII, porosity 97 %) as a membrane buffer for separation of (bio)hydrogen. In the foam the cell walls formed a structured complex of membranes. The cells served as pressure containers of separated gases. The foam membrane was able to buffer the difference between the feed injection rate and the rate of consumption of the product. Using the difference in time-lags of different gases in polymeric foam, efficient gas separation was achieved during transient state and high purity hydrogen was obtained. Argonne National Laboratory (ANL) is involved in developing dense hydrogen-permeable membranes for separating hydrogen from mixed gases, particularly product streams during coal gasification and/or methane reforming. Novel cermet (ceramic-metal composite) membranes have been developed. Hydrogen separation with these membranes is non-galvanic (does not use electrodes or external power supply to drive the separation and hydrogen selectivity is nearly 100 % because the membrane contain no interconnected porosity). The membrane development at ANL initially concentrated on a mixed proton/electron conductor based on BaCe0.8Y0.2O3-d (BCY), but it turned to be insufficient to allow high non-galvanic hydrogen flux. To increase the electronic conductivity and thereby to increase the hydrogen flux the development focused on various cermet membranes with 40-50 vol. % of metal or alloy dispersed in the ceramic matrix. Balachandran et al., described the development performed at ANL. The powder mixture for fabricating cermet membranes was prepared by mechanical mixing Pd (50 vol. %) with YSZ, after that the powder mixture was pressed into discs. Polished cermet membranes were affixed to one end of alumina tube using a gold casket for a seal (as can be seen in Figure 9). In order to measure the hydrogen permeation rate, the alumina tube was inserted into a furnace with a sealed membrane and the associated gas flow tubes.
Hydrogen permeation rate for Pd/YSZ membranes has been measured as a function of temperature (500-900 °C), partial pressure of hydrogen in the feed stream (0.04-1.0 atm.) and membrane thickness (» 22-210 mm) as well as versus time during exposure to feed gases containing H2, CO, CO2, CH4 and H2S. The highest hydrogen flux was » 20.0 cm3 (STP)/min cm2 for » 22- mm thick membrane at 900 °C using 100 % hydrogen as the feed gas. These results suggested that membranes with thickness < 22 mm should give a higher hydrogen flux. The hydrogen flux of Pd/YSZ membranes showed no degradation in a simulated synthesis gas atmosphere and was stable for » 270 hours in atmospheres containing up to ca. 400 ppm of hydrogen sulphide. Hardy et al.have focused on developing cermet membranes made from nanoscale precursor powders for which both barium cerate-based proton conducting ceramic and the nickel oxide were co-sythesized in a single combustion reaction. Among the barium cerate composition investigated, the 30 % Zr and 15 % Nd-doped material exhibited the best combination of chemical stability in CO2 and conductivity in hydrogen environments.
In the last decade Matrimid 5218 (Polyimide of 3,3',4,4'-benzophenone tetracarboxylic dianhydride and diamino-phenylindane) has attracted a lot of attention as a material for gas separation membranes due to the combination of relatively high gas permeability coefficients and separation factors combined with excellent mechanical properties, solubility in non-hazard organic solvents and commercial availability. Shishatskiy et al. have developed asymmetric flat sheet membranes for hydrogen separation from its mixtures with other gases. The composition and conditions of membrane preparation were optimized for pilot scale membrane production. The resulting membrane had a high hydrogen flux (1 m3 (STP)/m2h*bar) and selectivity of H2/CH4 at least 100, close to the selectivity of Matrimid 5218, material used for asymmetric structure formation. The hydrogen flux through the membranes increased with the decrease of polymer concentration and increase of non-solvent concentration. In addition, the influence of N2 blowing over the membrane surface (0, 2, 3, 4 Nm3 h-1 flow rate) was studied and it was proved that the selectivity of the membrane decreased with increase of the gas flow. The SEM image of the membrane supported by Matrimid 5218 is shown in Figure 10.
The stability against hydrocarbons was tested by immersion of the membrane into the mixture of n-pentane/n-hexane/toluene in 1:1:1 ratio. Stability tests showed that the developed membrane was stable against mixtures of liquid hydrocarbons and could withstand continuous heating up to 200 °C for 24 and 120 hours and did not lose gas separation properties after exposure to a mixture of liquid hydrocarbons. The polyester non-woven fabric used as a support for the asymmetric membrane gave to the membrane excellent mechanical properties and allowed to use the membrane in gas separation modules.
Interesting report on development of compact hydrogen separation module called MOC (Membrane On Catalyst) with structured Ni-based catalyst for use in the membrane reactor was presented by Kurokawa et al. In the MOC concept a porous support itself had a function of reforming catalyst in addition to the role of membrane support. The integrated structure of support and catalyst made the membrane reformer more compact because the separate catalysts placed around the membrane modules in the conventional membrane reformers could be eliminated. In that idea first a porous catalytic structure 8YSZ (mixture of NiO and 8 mol. % Y2O3-ZrO2 at the weight ratio 60:40) was prepared as the support structure of the hydrogen membrane. The mixture was pressed into a tube closed at one end and sintered then in air. Slurry of 8YSZ was coated on the external surface of the porous support and heat-treated for alloying. Obtained module of size 10 mm outside and 8 mm inside diameter, 100 ~ 300 mm length and the membrane thickness was 7 ~ 20 mm were heated in flowing hydrogen at 600 °C for 3 hours to reduce NiO in the support structure into Ni before use (the porosity of the support after reduction was 43 %). A stainless steel cap and pipe were bonded to the module to introduce H2 into the inside of the tubular module. Figure 11 presents the conceptual structure design of the MOC module as compared with the structure of the conventional membrane reformer.
The sample module in the reaction chamber was placed in the furnace and heated at 600 °C, pre-heated hydrogen (or humidified methane) was supplied inside MOC at the pressure of 0.1 MPa and the permeated hydrogen was collected from the outside chamber around the module at ambient pressure. The 100 ~ 300 mm long modules with 10 mm membrane showed hydrogen flux of 30 cm3 per minute per cm2 which was two times higher than the permeability of the conventional modules with palladium based alloy films. Membrane On Catalyst modules have a great potential to be applied to membrane reformer systems. In this concept a porous support itself has a function of reforming catalyst in addition to the role of membrane support. It seems that Membrane On Catalyst modules have a great potential to be applied to membrane reformer systems.
Amorphous alloy membranes composed primarily of Ni and early transition metals (ETM) are an inexpensive alternative to Pd-based alloy membranes, and these materials are therefore of particular interest for the large-scale production of hydrogen from carbon-based fuels. Catalytic membrane reactors can produce hydrogen directly from coal-derived synthesis gas at 400°C, by combining a commercial water-gas shift (WGS) catalyst with a hydrogen-selective membrane. Three main classes of membrane are capable of operating at the high temperatures demanded by existing WGS catalysts: ceramic membranes producing pure hydrogen via ion-transfer mechanism at ³ 600 °C, alloy membranes which produce pure hydrogen via a solution-diffusion mechanism between 300 – 500 °C and microporous membranes, typically silica or carbon, whose purity depends on the pore size of the membrane and which operate over a wide temperature range dependent on the membrane material. In order to explore the suitability of Ni-based amorphous alloys for this application, the thermal stability and hydrogen permeation characteristics of Ni-ETM amorphous alloy membranes has been examined by Dolan et al. Fundamental limitation of these materials is that hydrogen permeability is inversely proportional to the thermal stability of the alloy. Alloy design is therefore a compromise between hydrogen production rate and durability. Amorphous Ni60Nb(40-x)Zr(x) membranes have been tested at 400°C in pure hydrogen, and in simulated coal-derived gas streams with high steam, CO and CO2 levels, without severe degradation or corrosion-induced failure. The authors have concluded that Ni-Nb-Zr amorphous alloys are therefore prospective materials for use in a catalytic membrane reactor for coal-derived syngas. Much attention has been given to inorganic materials such as zeolite, silica, zirconia and titania for development of gas- and liquid- separation membranes because they can be utilized under harsh conditions where organic polymer membranes cannot be applied. Silica membranes have been studied extensively for the preparation of various kinds of separation membranes: hydrogen, CO2 and C3 isomers.
Kanezeashi have proposed silica networks using an organo-inorganic hybrid alkoxide structure containing the organic groups between two silicon atoms, such as bis(triethoxysilyl)ethane (BTESE) for development of highly permeable hydrogen separation membranes with hydrothermal stability. The concept for improvement of hydrogen permeability of silica membrane was to design a loose-organic-inorganic hybrid silica network using mentioned BTESE (to shift the silica networks to a larger pore size for an increase in H2 permeability). A hybrid silica layer was prepared by coating a silica-zirconia intermediate layer with a BTESE polymer sol followed by drying and calcination at 300°C in nitrogen. A thin, continuous separation layer of hybrid silica for selective H2 permeation was observed on top of the SiO2-ZrO2 intermediate layer as presented in Figure 12. Hybrid silica membranes showed a very high H2 permeance, ~ 1 order of magnitude higher (~ 10-5 mol m-2 s-1 Pa-1) than previously reported silica membranes using TEOS (Tetraethoxysilane). The hydrothermal stability of the hybrid silica membranes due to the presence of Si-C-C-Si bonds in the silica networks was also confirmed.
Nitodas et al. for the development of composite silica membranes have used the method of chemical vapour deposition (CVD) in the counter current configuration from TEOS and ozone mixtures. The experiments were conducted in a horizontal hot-wall CVD quartz reactor (Figure 13) under controlled temperature conditions (523 – 543 K) and at various reaction times (0 -15 hours) and differential pressures across the substrate sides using two types of substrates: a porous Vycor tube and alumina (g-Al2O3) nanofiltration (NF) tube. The permeance of hydrogen and other gases (He, N2, Ar, CO2) were measured in a home-made apparatus (able to operate under high vacuum conditions 10-3 Torr, feed pressure up to 70 bar) and the separation capability of the composite membranes was determined by calculating the selectivity of hydrogen over He, N2, Ar, CO2. The in-situ monitoring of gas permeance during the CVD development of nanoporous membranes created a tool to detect pore size alterations in the micro to nanometer scale of thickness. The highest permeance values in both modified and unmodified membranes are observed for H2 and the lowest for CO2. This indicated that the developed membranes were ideal candidates for H2/CO2 separations, like for example in reforming units of natural gas and biogas (H2/CO2/CO/CH4).
Moon et al. have studied the separation characteristics and dynamics of hydrogen mixture produced from natural gas reformer on tubular type methyltriethoxysilane (MTES) silica / ?-alumina composite membranes. The permeation and separation of CO pure gas, H2/CO (50/50 vol. %) binary mixture and H2/CH4/CO/CO2 (69/3/2/26 vol. %) quaternary mixture was investigated. The authors developed a membrane process suitable for separating H2 from CO and other reformate gases (CO2 or CH4) that showed a molecular sieving effect. Since the permeance of pure CO on the MTES membrane was very low (CO » 4.79 – 6.46 x 10-11 mol m-2 s-1 Pa-1), comparatively high hydrogen selectivity could be obtained from the H2/CO mixture (separation factor: 93 – 110). This meant that CO (which shall be eliminated before entering fuel cell) can be separated from hydrogen mixtures using MTES membranes. The permeance of the hydrogen quaternary mixture on MTES membrane was 2.07 – 3.37 x 10-9 mol m-2 s-1 Pa-1 and the separation factor of H2 / (CO + CH4 + CO2) was 2.61 – 10.33 at 323 – 473 K (Figure 14). The permeation and selectivity of hydrogen were increased with temperature because of activation of H2 molecules and unfavourable conditions for CO2 adsorption. Compared to other impurities, CO was most successfully removed from the H2 mixture.
The MTES membranes showed great potential for hydrogen separation from reforming gas with high selectivity and high permeance and therefore they have good potential for fuel cell systems and for use in hydrogen stations. According to the authors, the silica membranes are expected to be used for separating hydrogen in reforming environment at high temperatures.
Silica membranes prepared by the CVD or sol-gel methods on mesoporous support are effective for selective hydrogen permeation, however it is known that hydrogen-selective silica materials are not thermally stable at high temperatures. Most researchers reported a loss of permeability of silica membranes even 50 % or greater in the first 12 hours on exposure to moisture at high temperature. Much effort has been spent on the improvement of the stability of silica membranes. Gu et al. have investigated a hydrothermally stable and hydrogen-selective membrane composed of silica and alumina prepared on a macroporous alumina support by CVD in an inert atmosphere at high temperature. Before the deposition of the silica-alumina composite multiple graded layers of alumina were coated on the alumina support with three sols of decreasing particle sizes. The resulting supported composite silica-alumina membrane had high permeability for hydrogen (in the order of 10-7 mol m-2 s-1 Pa-1) at 873 K. Significantly the composite membrane exhibited much higher stability to water vapour at the high temperature of 873 K in comparison to pure silica membranes. The introduction of alumina into silica made the silica structure more stable and slowed down the silica disintegration process. As mentioned, silica membranes produced by sol-gel technique or by CVD applied for gas separation, especially for H2 production are quite stable in dry gases and exhibit high separation ratio, but lose the permeability when used in the steamed gases because of sintering or tightening. This could be a factor limitating its use in harsh environmental conditions. Zeolites, alumino-silicates crystalline composed of Al2O4 and SiO2 tetrahedra possess high thermal and chemical stability. On the other hand, their application on industrial level is restricted due to high manufacturing costs, reproducibility problems in the manufacturing stage and cracks developing in the intercrystalline phase. Wach et al. have reported the usage of polycarbosilane and its blend with polyvinylsilane of 20 wt. % as precursor polymers to produce silicon carbide (SiC)-based membranes for gas separation. Excellent H2/N2 perm-selectivity of over 150 and 250 at 523 K was achieved for PCS and PCS/PVS membranes respectively. Suda et al. have focused in their study on the optimisation of silicon carbide membranes. The authors investigated the factors that might have a great effect on pore size distribution and gas separation performance of the products. The factors included the effect of additives dispersed within membranes and the effect of the condition for pyrolysis and pre/post pyrolysis. It was fund that the addition of polystyrene, appropriate oxidation of the membrane and the low cross-linking contribute to improved gas permeation performance (H2 permeance and H2/N2 permselectivity). Novel cobalt-doped amorphous silica composite (Si-Co-O) membranes were fabricated on a mesoporous anodic alumina capillary (MAAC) tubes by Mori et al. It was suggested that cobalt and cobalt oxide nanoparticles could enhance hydrogen permeance through Si-Co-O membrane and cobalt ions could contribute to the densification of the Si-Co-O membrane. Gu et al. have modified ?-alumina supported MFI zeolite membranes by on-stream catalytic thermal cracking of methyldiethoxysilane (MDES) molecules inside the zeolitic channels during the separation of H2/CO2 gas mixture at 450°C and atmospheric pressure for enhancing hydrogen separation process. Modified membranes exhibited a significant increase in hydrogen selectivity over CO2. They also showed good performance and stability in separation of H2/CO2 gas mixture containing up to 28.4 % water vapour at 450 °C and atmospheric pressure. Figure 15 shows the schematic diagram of the experimental apparatus used for membrane modification and gas separation.
The zeolite membrane was mounted in a stainless steel cell with the membrane surface facing the feed stream. During membrane modification H2/CO2 gas mixture bubbled through the MDES at room temperature before entering the membrane cell. The modification of the membrane was performed at 450 °C and atmospheric pressure. Separation of equimolar H2/CO2 dry gas mixture was conducted as a function of the temperature and the obtained results are shown in Figure 16. The membrane became H2-selective after modification even at 23 °C where the separation factor of 2.56 was obtained. This indicated that the modified membrane possessed certain size-selectivity between hydrogen and CO2. Compared to the fresh, unmodified membrane the modified one had only a moderate loss of 40% in H2 permeance when the H2/CO2 separation factor increased from 1.85 to 10.8 at 450 °C.
Carbon molecular sieves have been shown to achieve excellent performance with respect to hydrogen permeability and selectivity, in the separation of hydrogen from light hydrocarbons such as methane. Grainger and Hägg have developed and evaluated carbon membranes derived from cellulose and optimised them for hydrogen/methane separation. Copper (II) nitrate was added to the precursor resulting in increased H2/CH4 perselectivity. Carbonisation temperature was varied from 400 to 700 °C with the best performance for membranes between 550 and 650 °C. Mixed gas tests with H2, CH4, C1-C4 and N2 showed that these membranes tolerate light hydrocarbons and separated hydrogen with a permeability of about 480 Barrer. The separation performance of membranes was found out to be strongly dependent on the exposure to air. One of the problems confronting the use of polymeric membrane is the strong trade-off between permeability and selectivity. For example, the permeation flux through polymeric membranes is considerably reduced as the gas selectivity increases depending on the nature of the polymer. It has been reported that in general rubbery polymers show higher permeability while glassy polymers show higher selectivity for gas transport in polymeric membranes16,. Hosseini et al. demonstrated successful implications of bleeding technique with chemical modification for fabrication of high performance polymeric membranes for gas separation applications. They found that augmentation in PBI (poly[2,2'-(1,3-phenylene)-5,5'-bibenzimidazole]) composition results in enhancement in gas separation performance mainly due to the effect of diffusivity selectivity. Acharya et al.  investigated hydrogen separation in blend glassy polymer membranes of polysulphone (PSF) and polycarbonate (PC) in different concentrations ratios as well as FeCl3 doped PC membranes. The 20% FeCl3 doped PC was found to be a new material for membrane separation. Bossard and Mettes have presented fuel processing optimizing an integrated steam reforming and palladium based membrane separation system. Schematic of setup divided in three sections: steam reforming and burner, hydrogen separation and raffinate analysis is shown in Figure 17.
In case of hydrogen separation section, the separator removes part of the hydrogen from the incoming reformate stream. The heat exchanger passes the in- and outgoing separator streams. Once started up, the auxiliary electric heater of the separator only has to provide a relatively small amount of energy to compensate for thermal losses (through insulation, piping and radiation). Separated hydrogen can either exit through a vacuum pump or directly into atmospheric pressure through a mass flow meter (MFM). A vacuum pump was used to remove essentially all of the hydrogen produced. Results on individual system components have shown that a significant improvement can be achieved in hydrogen generation in the reformate gas by wall- catalyzed micro-channel reactor.
Other options which possibly might be considered as a potential solution for gas separation in PDh process are: process using artificial gravity by rotating system (centrifuge) or microchannels (sponge), which uses capillary forces to collect the liquid. Centrifugal process works based on the effect of simulating gravity, where the acceleration forces are used for separation of molecules according to their mass and can be applied to liquid and gaseous substances. Forces are applied by placing the mixture or substance inside a mechanism that rotates the substance at a high speed. An advantage of the centrifuge process is its low energy consumption. Limited papers have examined the application of gas centrifuges to gas separation outside the usual isotope enrichment area of interest. A gas centrifuge is basically a rotating cylinder filled with a gas mixture. A mass of gas with two components of different molecular weight is spun up. A pressure gradient develops nearly immediately with a concentration gradient for each component. Diffusion occurs along this concentration gradient until the centrifugal force is balanced. Schematic of operational principles is shown in Figure 18.
Batch centrifugation is not very interesting for industrial applications, but due to its simplicity could be used for small systems, mainly to investigate the influence of parameters like pressure, temperature and speed. Van Wissen et al. have worked on the determination the order of magnitude of the maximum achievable separation for decontamining a natural gas by separating CO2 and methane in continuous counter-current gas centrifuges. An example of counter-current centrifuge is shown in Figure 19.
Based on the performed experiments is was concluded that the centrifugation of a contaminated gas mixture into a gas product (CH4) and a waste gas (CO2) stream could not be carried fast enough to achieve complete separation in one or a small number of units. Increasing the separation rate required boosting the mass transfer rates and spatial separation rate of product and waste.
In microchannel based distillation processes, thin vapour and liquid films are contacted in small channels where mass transfer is diffusion-limited. This process has been already applied at PNNL (Pacific Northwest National Laboratory) with several of microchannel devices for gas-liquid processing, including phase separation, partial condensation, absorption, desorption and distillation. For many of these units, the primary resistance to mass transfer occurred due to diffusion in the liquid phase. PNNL has tested and applied microchannel process for production a light fraction of JP-8 fuel with reduced sulphur content for use as feed to produce fuel-cell grade hydrogen. The applied JP-8 microchannel distillation was the first step in sulphur reduction (light sulphur compounds such as thiols and sulphides as well as heterocyclic sulphurs increasing boiling point like tiophene, benzothiophene, dibenzothiophene and their alkyl substituted drivatives) in the fuel process preparation (Figure 20).
A microchannel distillation unit patented at PNNL was used as a rectifying section to separate a low sulphur fraction from raw JP-8 fuel. JP-8 was vaporized and fed to the device at one end. Vapour product was removed and condensed as the distillate product. A portion of this condensate was refluxed back to the device as liquid feed and a heavy fraction liquid was removed from the feed end as the residual product. The microchannel unit was designed in a way that the direction of all its internal vapour and liquid flows was horizontal. The distillate product sulphur content was 329 ppmw, compared to the 1107 ppmw of sulphur in the raw JP-8. TeGrotenhuis et al. have described a novel approach for gas-liquid processing in microchannels. Results are reported about an investigation of microchannel phase separation in a transparent, single-channel device. Afterwards heat exchanger was integrated with the micro-channel wick in order to create a partial condenser that also separated the condensate and finally the unit was scaled-up to multi-channel phase separator. Figure 21 presents the schematic of microchannel phase separator with liquid flow along wick from left to right to liquid stream outlet and test device view of the single-channel, microchannel phase separator.
Experiments were performed at increasing liquid flow rate at constant gas flow and pressure difference between the gas and the liquid outlets. At low liquid flows the recovery of liquid from the flowing gas stream was complete and very little hold-up of the liquid water was observed in the liquid channel. The liquid breakthrough was dependent on several parameters like the gas flow rate, the pressure difference across the pores and the properties of the liquid. The microchannel architecture was scaled up by stacking the planar channels to a set of parallel channels in order to increase the reaction rates, as can be seen in Figure 22. Such construction provided the possibility to integrate heat exchanger into it. An aluminium gas-liquid phase separator was tested as a partial condenser and operated effectively in normal gravity and reduced gravity.
In addition, scale up of microchannel gas-liquid processing has been pursued for the application of recovering water from the cathode of PEM fuel cell. Schematic showing the operation of multichannel phase separator is shown in Figure 23. Good flow distribution between the channels was achieved by designing the pressure drop in the entrance header to be small compared to the pressure drop in each channel, allowing the flow to distribute evenly in the header before entering the channels.
Literature Survey on Hydrogen Solubility
Solubility is a quantity of particular substance that can dissolve in a particular solvent. In the partial dehydrogenation process, there is a risk of hydrogen dissolving heavily in Kerosene Jet A-1. This is not favourable due to the fact that the process produces hydrogen as fuel for the fuel cell. Therefore, it is vital to know how much hydrogen is lost in Kerosene Jet A-1 (hydrogen solubility), so that it can be factored in for the processes such as system optimization, process understanding, safety, conversion, etc. It is widely known that solubility can be influenced thermodynamically through temperature, pressure and composition (both the solvent and the solute). In general, hydrogen solubility (in terms of mole fraction) increases with hydrocarbon number. Literature study has shown that knowledge of hydrogen solubility is important for design and operation of industrial processes (such as desulphurization plants).
In the partial dehydrogenation (PDh) process, Kerosene Jet A-1 is only partially dehydrogenated in the liquid phase, therefore every molecule of hydrogen is required for the concept to work – that is to produce sufficient hydrogen for fuel cell application. In the PDh process, Kerosene Jet A-1 is reacted over a catalyst to yield hydrogen & hydrocarbon gases (C1 – C6), as well as dehydrogenated Kerosene Jet A-1 (and rest of un-reacted Kerosene Jet A-1). From the moment hydrogen is produced from the reaction, the hydrogen gas is in immediate contact with liquid Kerosene Jet A-1 (or dehydrogenated Kerosene Jet A-1). This would mean that if hydrogen solubility in Kerosene Jet A-1 is high, large amount of hydrogen produced is lost in the liquid, rather than in the gas phase which is channelled into the fuel cell. This would influence the hydrogen production rate calculation and the conversion rate calculation. Below is a basic schematic showing where hydrogen is in contact with Kerosene Jet A-1 (both dehydrogenated and un-reacted), this is highlighted in the red boxes.
Determining Hydrogen Solubility
There are several known methods for measuring hydrogen solubility in liquids, whereby commonly used techniques can be grouped into two methods – direct and indirect methods. In direct measurement methods, a known quantity of liquid at equilibrium is sampled into a gas chromatography, where hydrogen solubility can be determined. Indirect methods are measured through calculating pressure drop in a tank of known volume (liquid & gas) and temperature. In general, direct method is more accurate however proper sampling has to be practised to avoid certain uncertainties (sampling errors/bias). By contrast, the indirect methods do not involve sampling, and measurements can often be carried out at high temperature and pressure. In the sections below, two forms of hydrogen solubility determination is described. Analytical method is a form of direct measurement method, and Synthetic method is an example of indirect measurement. It has to be noted, that the literature survey performed for this report yielded no information regarding hydrogen solubility for Kerosene Jet A-1.
The analytical methods used to measure gas solubility tend to be polargraphic and electrochemical procedures, in addition to gas chromatographic procedures (as described above). A recent study (2009) on this subject was found to investigate the impact of trace oxygen and other dissolved gases in thermally stressed jet fuels, this investigation utilized an in-line analytical instrumentation for measuring trace level of oxygen, hydrocarbon, etc. The analytical instrument was a gas chromatography that uses a three member tandem separation column; this was developed for the purpose of the study. The figure below highlights the separation column.
The first column contains highly porous material (diatomaceous earth) that could be easily wetted by inserted liquid (i.e. hydrocarbon fuel). This packed bed is highly permeable so that gaseous species could readily pass downstream of the columns. The second column is a Porapak Q, which strongly retains hydrocarbons that are C8 and larger (at room temperature). The third column constituted the analytical separation column, a 5-Å molecular sieve plot, which rapidly separate hydrogen, oxygen, nitrogen, methane and carbon monoxide.
Detection of separated gaseous solute is accurately detected using several detection devices. In this study, discharge ionization and the helium ionization detection were considered, as well as coupled mass spectrometry. But it was experimentally determined that series arrangement involving modulated thermal conductivity sensor and a hydrogen flame ionization device would meet sensitivity and low-noise output signal requirements.
This system is an advantage since it requires only minor maintenance, short time for analytical separation (6,000 analyses have been performed).
Determination of hydrogen solubility in liquids can also has been carried out using a synthetic method, with a setup seen in Figure 27. This setup is built to calculate hydrogen solubility in aromatic, cyclic hydrocarbons, and their mixtures, this is derived through measurements (and changes in) of pressure and temperature to determine the number of mole of hydrogen. 
The apparatus above consists of the following parts: (1) constant (2) agitator (3) glass cell (4) magnetic stirrer (5) heater (6) chiller unit (7) thermometer (8) mercury manometer (9) mercury reservoir (10) pressure sensor (11) oil reservoir (12) pressure generator (13) vacuum pump (14) z-axis slider. 
In this setup, hydrogen is fed into an evacuated cell (3), and the pressure and temperature is determined to measure the amount of moles of hydrogen within the cell (assuming ideal gas state). In parallel, liquid samples are loaded into the cell, and the hydrocarbon fuel concentration is determined. The samples are then moved over to a water bath and connected to a pressurized line (for pressurizing the cell). With this temperature and pressure can be controlled for specific conditions for hydrogen solubility determination.
Another synthetic method of measurement is the X-ray view cell shown in the figure below. In order to measure hydrogen solubility, the hydrogen is fed into the panoramic view cell (Pyrex glass), and the pressure and the temperature were measured to determine the mole amount of hydrogen assuming ideal gas state.
The setup contains two parts, a view cell and a gas loading reservoir. The view cell is virtually transparent to X-rays since it is made out of beryllium, which can also sustain a broad range of conditions: vacuum up to 30 MPa, and up to 450 °C. The cell can contain ca. 50 ml of liquid, and gas can be added on demand into the cell through the gas loading reservoir. In addition, the view cell was developed to contain opaque hydrocarbon fluids, using x-rays to study phase densities and behaviour of individual composition.
Calculating the hydrogen solubility is done by using the following equation:
Where Z is an approximate unit derive from a function of temperature and pressure (determined from a compressibility chart modified for hydrogen), and is the amount of hydrogen dissolved in the liquid, derived from calculations involving volume, pressure and temperature from the system.
Kerosene Jet A-1 composition
Kerosene composition was reported on earlier reports within this project. However, since there are no recent literature studies about hydrogen solubility within Kerosene Jet A-1, it is best to review briefly on Kerosene Jet A-1 composition to help better understand its impact.
Unfortunately, the exact content of Kerosene Jet A-1 is virtually impossible to identify, due to the complexity of the mixture, and the content varies from different supplier and batch. Below is a GC-MS analysis of Kerosene Jet A-1:
This spectrum shows the distribution of the peaks from GC-MS, and the main peaks are circled. The circled peaks are in blue showing hydrocarbon from C10 to C14. The smaller peaks correspond to other linear, cyclic and aromatic compounds in the range of C8 to C16. As it can be seen, the spectrum is complex and it is difficult to quantify and identify each individual hydrocarbon compounds, therefore it is easier to group Kerosene Jet A-1 composition into chemical classes rather than individual compounds. Figure # below is an approximate distribution of hydrocarbon class within Kerosene Jet A-1.
The greatest percentage volume of Jet A-1 is composed of straight chain (25 %) and mono-cyclic paraffins (30 %), with smaller amounts of mononuclear aromatics (16 %), branched chain paraffins (11 %) and di-cycloparaffins (12 %). Using these chemical classes, it is then possible to compare with literature study, and attempt to determine the scale of hydrogen solubility in Kerosene Jet A-1.
Potential Solubility of Hydrogen in PDh Process
As there was no direct investigation relating hydrogen solubility in Kerosene Jet A-1 was found within the scope of the literature survey, the only comparable method was to use the knowledge of Kerosene Jet A-1's chemical class to compare with literature investigations.
According to T. Tsuji et al  who used synthetic apparatus method (Figure 27) to determine hydrogen solubility in cyclic, aromatic hydrocarbons and their mixtures at 303.15 K (figures below), cyclo-hydrocarbons and cyclic-hydrocarbons affects the solubility of hydrogen. Below in Figure # are the experimental findings from their investigation.
Cyclic- and cyclo-hydrocarbons can also be found in Kerosene Jet A-1, roughly 21 % are cyclic-hydrocarbons and 43 % cyclo-hydrocarbons are found in Kerosene Jet A-1. Cyclohexane (from the figure above) is able to accept more hydrogen into itself compared to benzene, and through an equimolar mixture of the two did not yield an "average" of the two, but it tends to be more similar to cyclohexane in terms of solubility than benzene. Tiny triangles are data obtained from literature study made by the authors, comparing their experimental findings with two other investigators on similar subjects (Herskowitz et al – solid triangle, and Ronze et al – empty triangle). Ironically, T. Tsuji et al.'s data correlate better with Ronze et al.'s data, even though they are operating at two different temperatures, 303.15 K compared with 304 K respectively. On paper, it would suggest that T. Tsuji et al.'s data should be more similar to Herskowitz et al. rather than Ronze et al. Perhaps the difference is with varying in instruments used to investigate, measure, and collect the data. But even with the possible variance between the three investigators, their data all showed linear pressure dependence which follows Henry's Law. Therefore, using the figure above, it can be seen that hydrocarbon with a higher molecular weight encourages hydrogen solubility. This is in agreement with another investigation with methylcyclohexane and toluene using the same apparatus. The result of the investigation can be seen in the figure below:
The results are similar when compared to benzene and cyclohexane. Here an additional CH4 was added onto the molecule to give methylcyclohexane and toluene, and again the results shows that molecule with the higher molecular weight have higher mole fraction of hydrogen dissolve in the respective hydrocarbon liquid. The data can also be compared between benzene and toluene, here toluene has the higher mole fraction of hydrogen dissolve in itself, and the same conclusion is derived with a comparison with cyclohexane and methylcyclohexane.
In terms for Kerosene Jet A-1, whereby generally it has a high molecular weight (average between C11H24 – C12H26), this would mean solubility is definitely present with relative degree. It is also interesting to note, methylcyclohexane is a known surrogate compound used for modelling surrogate Kerosene Jet A-1.
Investigations performed by A. Ghosh et al.  were done with alkenes such as 1-octene and 1-heptene. Below figure are some of the findings:
Octene could be found in Kerosene Jet A-1 as one (or more) of its 18 structural isomers, and octane could easily be part of the composition, or even part of the dehydrogenated-Kerosene Jet A-1 (after undergoing partial dehydrogenation). The graph above shows the result of hydrogen solubility in 1-octene as a function of temperature. Hydrogen mole fraction found solubilised in 1-octene is in some agreement with the investigation found above, while octane has a higher molecular weight than methylcyclohexane, the trend still seems to fit. Here, it can be seen that hydrogen solubility increases with increasing temperature. As well as increase in pressure, while the current partial dehydrogenation process functions at 10 bar, it is good to note that operating under higher pressures will encourage hydrogen dissolving in Kerosene Jet A-1 further.
Similarly for heptene, C7H14, which could exsist in Kerosene Jet A-1 or at least in dehydrogenated-Kerosene Jet A-1, the relation here shows the same as with octane. Though with a lower molecular weight, the amount of hydrogen found seems to be similar, however it is difficult to tell due to the scale of the graph.
Decane is another hydrocarbon molecule that is known as a Kerosene Jet A-1 surrogate, therefore taking a look at hydrogen solubility in decane might reveal further information. M.R. Riazi et al. displayed the effects of pressure on hydrogen solubility in decane at various at two extreme conditions of 283 K and 433 K, and these are shown in the figures below:
The graph shows at 433 K that hydrogen solubility increases more with small amount of pressure being added compared to 233 K. This means that hydrogen solubility is greater at 433 K then 233 K. Once again showing that increasing pressure has an adverse effect on hydrogen solubility, even though it might be advantageous for the partial dehydrogenation process, also temperature increase shows that more hydrogen is dissolve, probably accounting for the increase volatility of the liquid hydrocarbon.
From the literature studies it has been found that heavy molecular weight hydrocarbon show higher measurement of hydrogen solubility, but these measurements have not been measured in Kerosene Jet A-1.
At the current operation of partial dehydrogenation process which is at 10 bar at 350 °C, it is possible to use the data obtain from T. Tsuji et al. to have a rough estimation.
Comparison should be made with other hydrocarbon mixtures like Kerosene Jet A-1, for their hydrogen solubility. H.Y Cai et al.  performed experiments with other fractions such as HVGO, LVGO, heavy oil (GUDAO), and Athabasca bitumen vacuum bottoms (ABVB), using the X-ray view cell method. However, the heavier hydrocarbon fractions in its investigation are not comparable with Kerosene Jet A-1, but fortunately LVGO has several similarities with Kerosene Jet A-1 for comparison, the hydrogen solubility results for LVGO can be found below.
Light virgin gas oil (LVGO) share similar properties, such as they both contain sulphur, contain similar amounts of aromatic carbon, similar carbon weight percentage, and they are derived from relatively same distillate (184 – 454°C in LVGO and in Kerosene Jet A-1 180 – 300 °C). This could mean the hydrogen solubility in LVGO could be taken as an example for understanding hydrogen solubility in Kerosene Jet A-1. The graph shows the hydrogen solubility of LVGO at different temperatures. At the highest temperature 380 °C hydrogen solubility is ca. 1.3 mol/kg and at lower temperature its 0.3 mol/kg. Currently, the partial dehydrogenation process operates at 350 °C at 10 bar, which taken this graph into account, the current process has roughly 0.1 mol / kg.
H. Y Cai et al. other experimental results can be found in the table below:
Over 400 data points from 13 different systems have been taken from various sources is shown in Table 3 above. The carbon number range is from 8 to 46 and pressure range is from 1 to 160 bar. Kerosene Jet A-1 hydrogen solubility may fall in between Naphtha reformate and LVGO, because these fuels has similar composition as Kerosene Jet A-1, distillation range, and boiling point. This means that for Kerosene Jet A-1 average molecular weight (C12H26) to be 170 g/mol hydrogen solubility might range 0.064 – 14.3 mol%. However, actual experimental test needs to be performed to accurately determine the hydrogen solubility value.
- S. Specchia, F. W. A. Tillemans, P. F. Van den Oosterkamp, G. Saracco, J. Power Sources, 145 (2005), 683-690
- Ken Sutherland, Filtration + Separation, July/August 2008
- M. Takht Ravanchi, T. Kaghazchi, A. Kargari, Desalination, 235, 2009, 199-244
- H. D. Tong, F. C. Gielens, J. G. E. Gardeniers, H. V. Jansen, J. W. Berenschot, M. J. de Boer, J. H. de Boer, C. J. M. Van Rijn, M. C. Elwenspoek, Journal of Microelectromechanical Systems 14 (2005) 113-124
- W. M. Tucho, H. J. Venvik, J. C. Walmsley, M. Stange, A. Ramachandran, R. H. Mathiesen, A. Borg, R. Holmestad, Journal of Materials Scencei 44 (2009) 4429-4442
- S.-E. Nam, Y.-K. Seong, J. W. Lee, K.-H. Lee, Desalination 236 (2009) 51-55
- D. A. P. Tanaka, M. A. Llosa Tanco, J. Okazaki, Y. Wakui, Journal of Membrane Science 320 (2008) 436-441
- A. Kuraoka, H. Zhao, T. Yazawa, Journal of Materials Science 39 (2004) 1879-1881
- S. N. Paglieri, S. A. Birdsell, R. C. Snow, F. M. Smith, C. R. Tewell, Advanced Materials for Energy Conversion, (2004), 219-224
- M. Zahedi, B. Afra, M. Deghani-Mobarake, M. Bahmani, Journal of Membrane Science 333 (2009) 45-49
- R. Igi, T. Yoshioka, Y. H. Ikuhara, Y. H. Ikuhara, Y. Iwamoto, T. Tsuru, A. Am. Soc. 91, 9 (2008) 2975-2981
- V. Gepert, M. Kilgus, T. Schiestel, H. Brunner, G. Eigenberger, C. Merten, Fuel Cells, 6 (2006), 472-481
- J. Han, I.-S. Kim, K.-S. Choi, International Journal of hydrogen Energy, 27 (2002), 1043-1047
- Z. Pientka, P. Pokorny, K. Belafi-Bako, Journal of Membrane Science 304 (2007) 82-87
- U. Balachandran, T. H. Lee, L. Chen, S. J. Song, S. E. Dorris, Proc. 9th Int. Conf. On Inorganic Membranes, Lillehammer-Norway, June 25-29, 2006
- U. Balachandran, T. H. Lee, L. Chen, S. J. Song, J. J. Picciolo, S. E. Dorris, Fuel 85 (2006) 150-155
- J. S. Hardy, E. C. Thomsen, N. L. Cranfield, J. V. Crum, K. S. Weil, L. R. Pederson, International Journal of Hydrogen Energy 32 (2007) 3631-3639
- S. Shishatskiy, C. Nistor, M. Popa, S. P. Nunes, K. V. Pienemann, Advanced Engineering Materials, 8 (2006) 5, 390-397
- H. Kurokawa, T. Tsuneki, Y. Shiarasaki, T. Shimamori, H. Shgaki, H. Tanaka, K, Mitsuya, Fuel Cells: On the Path to Energy Independence, 31 (2007) 235-238
- M. Dolan, N. Dave, L. Morpeth, R. Donelson, D. Liang, M. Kellam, S. Song, J. Membrane Science, 2 (2009) 549-555
- M. Kanezeashi, K. Yada, T. Yoshioka, T. Tsuru, Journal of the American Chemical Society, 131 (2009) 2, 414-415
- S. F. Nitodas, E. P. Favvas, G. E. Romanos, M. A. Papadopoulou, A. C. Mitropoulos, N. K. Kannelopoulos, J. Porous Mater (2008) 15, 551-557
- J-H. Moon, J-H Bae, Y-S. Bae, J-T. Chung, C-H. Le, Journal of Membrane Science 318 (2008) 45-55
- Y. Gu, P. Hacarlioglu, S. T. Oyama, Journal of Membrane Science 310 (2008) 28-37
- R. A. Wach, M. Sugimoto, A. Idesaki, M. Yoshigawa, Material Science and Engineering B 140 (2007) 81-89
- H. Suda, H. Yamaguchi, Y. Uchimaru, I. Fujiwara, K. Haraya, Desalination 193 (2006) 252-255
- H. Mori, T. Nagano, S. Fujisaki, T. Sumino, Y. Iwamoto, Materials Science 6 Technology, Materials Science 2 (2006) 655-662
- X. Gu, Z.Tang, J. Dong, Microporous and Mesoporous Materials 111 (2008) 441-448
- S. M. Saufi, A. F. Ismail, Carbon 42 (2004) 241
- D. Grainger, M-B. Hägg, Journal of Membrane Science 306 (2007) 307-317
- 1 Barrer = 3.348 x 10-19 kmol m / (m2 s Pa)
- T-H. Weng, H-H. Tseng, M-Y. Wey, International Journal of Hydrogen Energy 33 (2008) 4178-4182
- M. Lopez-Gonzalez, V. Compan, E. Saiz, E. Riande, J. Guzman, Journal of Membrane Sciences, 253 (2005) 175-181
- S. S. Hosseini, M. M. Teoh, T. S. Chung, Polymer 49 (2008) 1594-1603
- N. K. Acharya, V. Kulshretha, K. Awasthi, A. K. Jain, M. Singh, Y. K. Vijay, International Journal of Hydrogen Energy 33 (2008) 327-331
- P. R. Bossard and J. Mettes, Fuel Cell Seminar & Exposition ; Phoenix, Arizona, October 27 - 30, 2008, pp. 309-315
- R. Van Wissen, M. Golombok, J.J.H. Brouwers, Chemical Engineering Science (2005) 4397-4407
- F. Zheng, V. S. Stenkamp, W. E. TeGrotenhuis, B. Q. Roberts, X. Huang, D. L. King, Catalysis Today, 136, 3-4 (2008) 291-300
- W. TeGrotenhuis, S. Stenkamp, A. Twitchell, "Gas-Liquid Processing in Microchannels" in Microreactor Technology and Process Intensification, American Chemical Society Symposium Series, 2005, Vol. 914 Chapter 22, pp 360–377, ISBN: 9780841220324