Hydrogen production from ammonia by the plasma membrane reactor

A new plasma membrane reactor (PMR) was developed to efficiently produce hydrogen from NH₃ with the use of atmospheric pressure plasma and a hydrogen separation membrane. The generation of high-purity hydrogen from NH₃ was also examined. First, hydrogen gas flowing into the PMR revealed the effect of the PMR on hydrogen separation. Hydrogen separation depends on the partial pressure of hydrogen gas supplied (P_{in, H2}) and permeated (P_{out, H2}) when P_{in, H2}^0.5 − P_{out, H2}^0.5 > 0. Second, NH₃ gas flowing into the PMR revealed its hydrogen production characteristics: the maximum hydrogen conversion rate of a typical plasma reactor (PR) is 13%, whereas the PMR converted 24.4%. Hydrogen obtained by hydrogen separation was approximately 100% pure. A hydrogen generation rate of 20 mL/min was stably obtained.     

Improvement of hydrogen-separating performance by on-stream catalytic cracking of silane over hollow fiber MFI zeolite membrane

MFI zeolite membranes were synthesized on porous α-alumina hollow fibers by in-situ hydrothermal synthesis. The membranes were further modified for H₂ separation by on-stream catalytic cracking deposition of methyldiethoxysilane (MDES) in the zeolitic pores. The separation performance of the modified membranes was characterized by separation of H₂/CO₂ gas mixture at 500 °C. Activation of MFI zeolite membranes by air at 500 °C was found to promote catalytic cracking deposition of silane in the zeolitic pores effectively, which resulted in significant improvement of H₂-separating performance. The H₂/CO₂ separation factor of 45.6 with H₂ permeance of 1.0 × 10⁻⁸ mol m⁻² s⁻¹ Pa⁻¹ was obtained at 500 °C for a modified hollow fiber MFI zeolite membrane. The as-made membranes showed good thermochemical stability for the separation of H₂/CO₂ gas mixture containing H₂O and H₂S, respectively.     

Hydrogen production by reforming of liquid hydrocarbons in a membrane reactor for portable power generation—Experimental studies

One of the most promising technologies for lightweight, compact, portable power generation is proton exchange membrane (PEM) fuel cells. PEM fuel cells, however, require a source of pure hydrogen. Steam reforming of hydrocarbons in an integrated membrane reactor has potential to provide pure hydrogen in a compact system. Continuous separation of product hydrogen from the reforming gas mixture is expected to increase the yield of hydrogen significantly as predicted by model simulations. In the laboratory-scale experimental studies reported here steam reforming of liquid hydrocarbon fuels, butane, methanol and Clearlite^® was conducted to produce pure hydrogen in a single step membrane reformer using commercially available Pd–Ag foil membranes and reforming/WGS catalysts. All of the experimental results demonstrated increase in hydrocarbon conversion due to hydrogen separation when compared with the hydrocarbon conversion without any hydrogen separation. Increase in hydrogen recovery was also shown to result in corresponding increase in hydrocarbon conversion in these studies demonstrating the basic concept. The experiments also provided insight into the effect of individual variables such as pressure, temperature, gas space velocity, and steam to carbon ratio. Steam reforming of butane was found to be limited by reaction kinetics for the experimental conditions used: catalysts used, average gas space velocity, and the reactor characteristics of surface area to volume ratio. Steam reforming of methanol in the presence of only WGS catalyst on the other hand indicated that the membrane reactor performance was limited by membrane permeation, especially at lower temperatures and lower feed pressures due to slower reconstitution of CO and H₂ into methane thus maintaining high hydrogen partial pressures in the reacting gas mixture. The limited amount of data collected with steam reforming of Clearlite^® indicated very good match between theoretical predictions and experimental results indicating that the underlying assumption of the simple model of conversion of hydrocarbons to CO and H₂ followed by equilibrium reconstitution to methane appears to be reasonable one.     

Operation of solid oxide fuel cell on biomass product gas with tar levels >10 g Nm−3

This work assesses experimentally the feasibility of feeding a high tar load product gas from biomass gasification to a planar solid oxide fuel cell (SOFC) for renewable electricity generation. The SOFC had a nickel gadolinium-doped ceria anode (Ni-GDC) and the gasifier was a pilot scale circulating fluidized bed, employing hot gas-cleaning to remove particulates, HCl and H₂S. The SOFC operated for several hours on either pre-reformed gas (reduced tar levels < 0.5 g Nm^?3) as well as on high tar-laden wood gas (tar levels > 10 g Nm^?3) i.e. with no pre-reforming of tars. The tests were carried out at low fuel utilization U_f of around 20% at a current density j = 130 mA cm^?2. In all cases stable continuous SOFC performance was established. Post experimental examination of the SOFC showed that the anode was not affected by carbon deposition or other impurity accumulation.     

Further performance enhancement of a DME-fueled solid oxide fuel cell by applying anode functional catalyst

In order to increase the coking resistance of SOFCs operating on DME fuel, a Pt/Al₂O₃–Ni/MgO mixture catalyst was investigated for internal partial oxidation of DME. Catalytic test demonstrated the mixture catalyst has higher activity for DME partial oxidation and lower CH₄ selectivity than the individual Pt/Al₂O₃ and Ni/MgO catalysts. O₂-TPO analysis demonstrated that the mixture catalyst also had much higher coke resistance than sintered Ni-YSZ anode, especially at high O₂ to DME ratio. Raman spectroscopy of the carbon-deposited catalysts demonstrates that the graphitization degree of carbon is reduced with introducing O₂ into DME, and the carbon deposited on the mixture catalyst is almost in amorphous structure. Two operation modes of the mixture catalyst for indirect internal partial oxidation of DME, i.e, directly depositing on the anode surface and locating in the anode chamber were tried. The performance of the cells operating on DME fuel through both operation modes was studied by I–V polarization test and EIS characterization. The cells delivered attractive peak power density of around 750 mW cm⁻² by operating on DME-O₂ mixture gas, modestly lower than 1012–1065 mW cm⁻² operating on pure hydrogen fuel at 700 °C. The direct deposition of Pt/Al₂O₃–Ni/MgO onto anode surface to perform as a functional layer and a DME to O₂ ratio of 2:1 in the mixture gas is preferred to minimize coke formation and maximize power output for the cell to operate on DME fuel.     

Concerning adsorbed and absorbed hydrogen on and in ferrous metals

Fe specimens containing 0.2% carbon were chosen to minimize D_H and maximize c_H. 0.05 M H₂SO₄ solutions were used. Additives were naphthonitrile, naphthalene, benzonitrile, H₂S, NaH₂PO₄.Measurements: (1) Steady state hydrogen evolution as a function of potential in the absence and presence of additives; (2) constant current transients with the initial potential in the hydrogen evolution region and a region positive to the corrosion potential; (3) laser desorption of hydrogen—a Neodynium-Yag laser was used to introduce micro-cavities into the metal after its exposure to hydrogen from solution at various fugacities. The H was measured in a quadrupole mass spectrometer, and analyses of the P_H/time transient yields the H concentration (c_H) in the region of metal struck by the laser.The results obtained are the Tafel parameters; the fraction (θ_η) of the surface occupied by H at various η's (±50%), and the concentration of absorbed hydrogen (±10%). Tafel parameters in the additive show b > 2RT/F; (i_o) additive < (i_o) pure. d lnθ/dη > 4RT/F. In c_H is linear with η in the non-additive case. Extrapolation to η = 0 gives $\text{c}{}_{\mathrm{<mtext>H</mtext>}}\text{?}$ that determined by other approaches. The equilibrium constant, c_H/θ, depends parabolically upon potential; inflection occurs at 400 mV NHE. Phosphate and benzonitrile reduced θ most. $\text{θ ? 1–10\%}$ referred to the whole surface. All additives increase c_H, and lower θ_η in some potential regions.The HER mechanism is slow discharge followed by coupled atomic combination. The rate determining step takes place on the regions between carbide and ferritic phases. Results are consistent with Tempkin kinetics. The Tafel slope in the presence of additives is a result of this, together with the effect of the variation of θ_org with potential. θ/η relations are interpreted on the basis of the hypothesis that ΔG^ads_θ=θ=ΔG^ads_θ=θ + r ln θ The fugacity/η relation is consistent with Tempkin kinetics in the pure case; complexities intervene, in that with additives, the relation of c_H to θ as a function of η shows fair numerical agreement with experiment, discrepancies of which are due to trapping. What determines a substance to be a promoter or inhibitor of H in the metal is analyzed.     

Note on a real gas model for OWC performance

Oscillating water column (OWC) are devices for wave energy extraction equipped with turbines for energy conversion. The purpose of the present work is to study the thermodynamic of a real gas flow through the turbine and its differences with respect to the ideal gas hypothesis, with the final goal to be applied to OWC systems. The effect of moisture in the air chamber of the OWC entails variations on the atmospheric conditions near the turbine, modifying its performance and efficiency. In this work we study the influence of humid air in the performance of the turbine. Experimental work is carried out and a real gas model is asserted, in order to take a first approach to quantify the extent of influence of the air–water vapour mixture in the turbine performance. The application of a real gas model and the experimental study confirmed the deviations of the turbine performance from the expected values depending on flow rate, moisture and temperature.     

Determination of equilibrium isotope effect at Pd/alkaline solution (regular and heavy water) interfaces by the phase-shift method and its comparison with other Pt-group metals

The Frumkin isotherm provides information about the atomic behaviors of hydrogen (H) and deuterium (D) on the electrocatalyst surface. However, it cannot be obtained easily by the conventional electrochemical methods, owing to the similar electrochemical behaviors of the under- and over-potential deposition of H and D (UPD H, OPD H, UPD D, and OPD D) and the numerous overlapping factors at the electrocatalyst interface. In this study, the Frumkin isotherms for the discharge reactions of the OPD H and OPD D in the Volmer steps, isotopic shift of these isotherms, and equilibrium isotope effect (EIE) at the interfaces of Pd/alkaline (regular and heavy water: H₂O and D₂O) solutions are precisely determined by the phase-shift method. The interaction parameter (ɡ) and equilibrium constant (K) are 6.6 and 9.7 × 10^?6 exp (?6.6θ) mol^?1, respectively, for OPD H and 8.7 and 1.1 × 10^?6 exp (?8.7θ) mol^?1, respectively, for OPD D, where θ is the fractional coverage (0 ≤ θ ≤ 1). In the range of θ (0 ≤ θ ≤ 1), the standard Gibbs energy (ΔG_θ⁰) of OPD D is 5.4–10.6 kJ mol^?1 higher than that of OPD H. The EIE values increase from 8.8 to 72.1 with increasing θ and are extraordinarily high compared to those at other Pt-group metal interfaces. The EIE values at the interfaces of Pt-group metal/alkaline (H₂O and D₂O) solutions decrease in the following order: Pd >> Pt–Ir alloy > Pt > Ru > Ir and Rh. We expect that the Frumkin isotherms of OPD H and OPD D and the related EIE established in this study will be a useful and effective tool for studying the isotope effects at various electrocatalyst interfaces.     

Numerical analysis of convective flow and thermal stratification in a cryogenic storage tank

In this study, the liquid–vapor mixture model was used for a numerical study of natural convective flow in a cryogenic tank with a capacity of 4.9?m³ under various conditions of heat flux and filling level to understand the early stages of convective flow phenomena and the consequent thermal stratification of cryogenic liquid. Two cryogens—liquefied natural gas (LNG) and liquefied nitrogen (LN₂)—were compared to observe their effects. LN₂ exhibited faster vaporization owing to its lower heat of vaporization. It was observed that higher heat flux and lower filling level led to faster vaporization and relatively vigorous heat transfer, showing early thermal stratification.     

Characterization of internal wetting in polymer electrolyte membrane gas diffusion layers

Capillary pressure vs. saturation (P_C(S_L)) curves are fundamental to understanding liquid water transport and flooding in PEM gas diffusion layers (GDLs). P_C(S_L) curves convolute the influence of GDL pore geometry and internal contact angles at the three-phase liquid/solid/gas boundary. Even simple GDL materials are a spatially non-uniform mixture of carbon fiber and binder, making a Gaussian distribution of contact angles likely, based on the Cassie–Baxter equation. For a given Gaussian contact angle distribution with mean (θ_Mean) and standard deviation (σ), a realistic P_C(S_L) curve can be computed using a bundle of capillaries model and GDL pore size distribution data. As expected, computed P_C(S_L) curves show that θ_Mean sets the overall hydrophilic (θ_Mean < 90°) or hydrophobic (θ_Mean > 90°) character of the GDL (i.e., liquid saturation level at a given capillary pressure), and σ affects the slope of the P_C(S_L) curve. The capillary bundle model also can be used with (θ_Mean, σ) as unknown parameters that are best-fit to experimentally acquired P_C(S_L) and pore size distribution data to find (θ_Mean, σ) values for actual GDL materials. To test this, pore size distribution data was acquired for Toray TGP-H-090 along with hysteretic liquid and gas intrusion capillary pressure curve data. High quality best-fits were found between the model and combined datasets, with GDL liquid intrusion showing fairly neutral internal surface wetting properties (θ_Mean = 92° and σ = 10°) whereas gas intrusion displayed a hydrophilic character (θ_Mean = 52° and σ = 8°). External liquid advancing and receding contact angles were also measured on this same material and they also showed major hysteresis. The new methods described here open the door for better understanding of the link between GDL material processing and the wetting properties that affect flooding.     

Production of syngas from steam gasification of three different ranks of coals and related reaction kinetics

The steam gasification properties of three different ranks of coals, Shengli lignite (SL), Shenhua subbituminous coal (SH), and Tavan Tolgoi anthracite (TT), were investigated using a lab-scale fixed-bed reactor, and the thermodynamic equilibrium constant and kinetics of the reaction were analyzed. The results showed that the aromaticity and condensation of aromatic structures in SL, SH, and TT became higher, and the maturity of organic substance became lower. The steam gasification reaction showed that the syngas from low-rank SL had a high H₂/CO molar ratio, while the syngas from high-rank TT had relatively high CO content. The direct carbon gasification reactions for these three different ranks of coals were far from in equilibrium; the water gas shift reaction of SL was near equilibrium, and the degree of reaction for SL was higher than that of SH and TT. We studied a random pore model (RPM), shrinking core model (SCM), and hybrid model (HM), and the hybrid model was found to be the most suitable model of the three for fitting the steam gasification reactions of the three types of coal. It had high fitting correlation coefficient R² values (ranging from 0.9939 to 0.9990) and small average error θ values (ranging from 0.009 to 0.016). The apparent activation energy E values of SL, SH, and TT fitted by HM were 179.10, 48.14, and 63.06 kJ/mol, respectively, and the corresponding pre-exponential factor k₀ values were 3.14 × 10⁷, 1.01, and 1.22 min⁻¹, respectively. This study finds that the steam gasification of SL, SH, and TT coal samples consists of homogeneous phase reaction and shrinking core reaction.     

Saturation boiling of HFE-7100 from a copper surface, simulating a microelectronic chip

Experiments are performed, which investigated the effect of inclination angle, θ, on saturation pool boiling of HFE-7100 dielectric liquid from a smooth, 10×10 mm copper surface, simulating a microelectronic chip. For θ?90° and surface superheats, ΔT_sat>20 K, nucleate boiling heat flux decreases with increased θ, but increases with θ for ΔT_sat<20 K. Similarly, at higher inclinations and ΔT_sat>13 K, nucleate boiling heat flux decreases with increased inclination, but at lower surface superheats the trend is inconclusive. The developed nucleate boiling correlation is within ±10% of the data and the developed correlations for critical heat flux (CHF) and the surface superheat at CHF are within ±3% and ±8% of the data, respectively. Results show that CHF decreases slowly from 24.45 W/cm² at 0° to 21 W/cm² at 90°, then decreases fast with increased θ to 4.30 W/cm² at 180°. The surface superheat at CHF also decreases with θ, from 31.7 K at 0° to 19.9 K at 180°. Still photographs are recorded of pool boiling at different heat fluxes and θ=0°, 30°, 60°, 90, 120°, 150° and 180°. The measured average departure bubble diameter from the photographs taken at the lowest nucleate boiling heat flux of ∼0.5 W/cm² and θ=0° is 0.55±0.07 mm and the calculated departure frequency is ∼100 Hz.     

Energy requirement of HI separation from HI–I2–H2O mixture using electro-electrodialysis and distillation

The separation of HI from HI–I₂–H₂O mixture is an essential subsection of the Iodine–Sulfur (IS) process for thermochemical hydrogen production. The energy requirement of the separation determines, to a large extent, the hydrogen production efficiency of the IS process. In order to examine duty of the separation using electro-electrodialysis (EED) and distillation, a process simulation study was carried out using an analytical model of EED based on ideal membrane properties and properties of the reported EED experiments using a Nafion^® membrane and graphite electrodes. For both of the ideal-membrane case and Nafion-membrane case, effects of the operating parameters on heat duty were estimated, which comprised column pressure, HI molality in the column feed, and the flow rate ratio of the input from Bunsen section to distillate rate. Low column pressure, and high HI molality in the column feed were preferable for the ideal-membrane case; column pressure of 1.0 MPa and optimized HI molality in the column feed were desired for the Nafion-membrane case. The flow rate ratio had little effect on the minimum heat duty in the ideal-membrane case; a value in the vicinity of the lower limit of the flow rate ratio was optimal for the Nafion-membrane case. The difference of the inclination of parameters resulted from the fractional vaporization of the column feed in the ideal-membrane case and weight of the EED cell duty on the total duty due to the membrane voltage drop. The optimization of these parameters was also carried out. The minimum total heat duty of the Nafion-membrane case was 3.07 × 10⁵ J/mol-HI, and that of the ideal-membrane case was 12.5% of this value.     

Hydrogen isotopes separation using frontal displacement chromatography with Pd–Al2O3 packed column

Separation or purification of tritium isotopes is one of the key technologies in ITER. A set of frontal displacement chromatography (FDC) device was designed and constructed for hydrogen isotopes separation using palladium loading on/in alumina (Pd–Al₂O₃) as the separation material. The hydrogen isotopes separation experiments were carried out. It was found that deuterium abundance of the product was up to 98.5% and the average separation factor was as high as 64, under the condition of 273 K column temperature and 15 mL(NTP)/min flow rate, for a feed gas of 5%H₂-5%D₂-90%Ar. The deuterium recovery ratio was 42% in this separation test. The results showed that the separation performance of our FDC device was good by using Pd–Al₂O₃ as separation materials, and it suggested considerable potential for the applicability of FDC in hydrogen isotopes separation.     

Improved yield parameters in catalytic steam gasification of forestry residue; optimizing biomass feed rate and catalyst type

The catalytic gasification (900 °C) of forestry industry residue (Eucalyptus saligna) was laboratory-studied. Biomass feed rate and type and amount of catalyst were assayed for their effect on the gasified product composition and the overall energy yield of the gasification reaction. The use of a calcined dolomite catalyst resulted in a combustible gas mixture of adequate calorific power (10.65 MJ m^?3) for use as fuel, but neither the product gas composition nor the energy yield varied significantly with widely different amounts of the catalyst (2 g and 20 g). The use of NiO-loaded calcined dolomite catalysts did not affect the product gas composition significantly but led to a 30% increase in the total product gas volume and to a reduction in the rate of tar and char formation. The catalyst loaded with the smallest amount of NiO studied (0.4 wt%. Ni/Dol) led to the highest energy yield (21.50 MJ kg^?1 on a dry-wood basis) based on the use of the gasified product as fuel. The gasified product was found to have an adequate H₂/CO molar ratio and H₂ content for use as synthesis gas source and partial source of H₂.     

Hydrogen isotopes separation using frontal displacement chromatography: The influences of column temperature and gas flow rate

Based on the previous study in frontal displacement chromatography (FDC) packed with Pd-Al₂O₃, two groups of separation experiments were conducted to derive the rules how the two most significant factors, temperature and gas flow rate, to influence the separation performance. Separations of the first group were carried out at the feed gas flow rate of 15 mL/min and temperature of 303–213 K, and the second group at 253 K and 10–100 mL/min, with the identical composition of feedstock ((5 ± 0.1)%H₂-(5 ± 0.1)%D₂-(90 ± 0.1)%Ar). The results indicate the derived rules are consistent with those from references: 1) the rules of temperature effects on separation efficiency lie in two aspects that the lower the temperature is, the larger the thermodynamic separation effect is, and the higher the temperature is, the quicker the hydrogen isotopes exchange dynamics becomes. As to FDC using palladium, 263–213 K will be an appropriate temperature range to have excellent separation performance achieved with the insight into these two aspects. 2) the rule of the influence of gas flow rate basically obeys the van Deemter equation, which means that it does exist an theoretically optimal gas flow rate, ū_opt, at a certain temperature and for a certain composition of feedstock, and considering the theoretics and efficiency, separations conducted at the gas flow rate of an suitable range that higher than ū_opt but less than 10ū_opt can derive good separation performance. The results and discussion have verified the imperative impact of temperature and gas flow rate on the separation performance of this FDC method, and the derived desirable temperature and gas flow rate ranges would supply valuable supports and references for future applications of FDC in hydrogen isotopes separation and tritium recovery in fusion reactors.     

Long term and performance testing of NaMg double salts for H2/CO2 separation

This work investigates the synthesis and performance of double salts for H₂/CO₂ separation. A series of NaMg double salts were prepared based on xMg(NO₃)₂: yNa₂CO₃: zH₂O and characterised. The best sorbents reached CO₂ uptake of 17.9 wt% at 0.62 MPa and 375 °C. The NaMg double salts preferentially sorbed CO₂ as determined by breakthrough tests. The NaMg double salts were packed in a sorbent bed and tested for H₂/CO₂ separation at the back end of a water gas shift reactor. The space velocity had the largest impact on the performance of the sorbent bed, as increasing the space velocity from 2.16 × 10^?3 to 9.51 × 10^?3 s^?1 sped up the breakthrough time by 84%. Increasing the feed gas pressure from 0.3 to 0.6 MPa reduced the breakthrough time by ～45%. The NaMg double salt sorbents were exposed for over 1000 h of continuous temperature including 28 cycles of sorption and desorption, and proved to be stable during changes of operating conditions such as flow rates and pressures.     

Effects of biomass-generated producer gas constituents on cell growth,product distribution and hydrogenase activity of Clostridium carboxidivorans P7T

In our previous work, we demonstrated that biomass-generated producer gas can be converted to ethanol and acetic acid using a microbial catalyst Clostridium carboxidivorans P7^T. Results showed that the producer gas (1) induced cell dormancy, (2) inhibited H₂ consumption, and (3) affected the acetic acid/ethanol product distribution. Results of this work showed that tars were the likely cause of cell dormancy and product redistribution and that the addition of a 0.025 μm filter in the gas cleanup negated the effects of tars. C. carboxidivorans P7^T can adapt to the tars (i.e. grow) only after prolonged exposure. Nitric oxide, present in the producer gas at 150 ppm, is an inhibitor of the hydrogenase enzyme involved in H₂ consumption. We conclude that significant conditioning of the producer gas will be required for the successful coupling of biomass-generated producer gas with fermentation to produce ethanol and acetic acid.     

Research and development of a low-BTU gas-driven engine for waste gasification and power generation

Combustion characteristics of low-BTU gases (about 1000 kcal/N m³) were experimentally investigated in order to develop engine generators for waste gasification and power generation systems. Two simulated low-BTU gases, obtained from one-step high temperature gasification (hydrogen rich) and two-step pyrolysis/reforming gasification (methane rich), as well as natural gas, were tested in a small-scale spark ignition engine. Compared to the natural gas driven engine, the hydrogen rich low-BTU gas driven engine showed similar thermal efficiency but with significantly lower NO_x and hydrocarbon emissions and wider equivalence ratio range for stable engine operation. On the other hand, the methane rich low-BTU gas engine showed narrower equivalence ratio range for stable operation. The test results show engine performance more depends on combustion characteristics than on the heating value of the fuel gas. For better engine performance, hydrogen rich fuel gas is desirable.     

CFD modeling of hydrogen separation through Pd-based membrane

Hydrogen separation through palladium-based membranes is one of the most promising technologies to produce H₂ gas. The main purpose of this work is to comprehensively study the impact of different parameters on hydrogen diffusion flux (J_H2) through a Pd–Ag membrane. The effect of implementing sweep gas on J_H2 is investigated along with two methods of applying pressure difference, namely pressurized method, and vacuum method. Also, the effect of species mole fraction for three binary mixtures (H₂/N₂, H₂/CO_2, and H₂/CO) is examined. A CFD model is developed and used to perform the study. Experimental data from the literature are used to validate the CFD model, and the results showed good agreement with the experimental data. The results revealed that implementing sweep-gas could significantly improve the hydrogen diffusion flux (by 25%) at the law-pressure difference. Moreover, it turns out that the vacuum method is more effective than the pressurized method, where it results in J_H2 greater than the pressurized method by 15–36%. Furthermore, the CFD results showed that more hydrogen gas can be extracted from a binary mixture of H₂/N₂ (0.93 mol/m².s) than of CO (0.90 mol/m².s) and H₂/CO₂ (0.81 mol/m².s).