An insight into separating H₂ from natural gas/H₂ mixtures using Mg-based systems

Abstract

The use of light, abundant, and relatively cheap Mg-based systems arouses great interest in hydrogen-economy-related applications such as hydrogen and heat storage. So far, MgH₂, capable of storing large amounts of H₂ (7.6 wt%), has been scarcely evaluated for its H₂ separation potential, which may be crucial for H₂ recovery from various H₂-containing gas mixtures. Herein, we reveal and discuss the ability of Mg-based systems to separate H₂ from CH₄-rich gas mixtures. Mg-Ni and Mg-Fe systems can separate ∼5.5 wt% of H₂ during the hydrogenation process and release pure H₂ (at least 99.9%) within the dehydrogenation process. Pure H₂ can, therefore, be obtained in a one-step separation system. In this study, we discuss the selection of the hydrogenation/dehydrogenation processes catalyst (Ni, Fe) as well as the optimal separation process temperature. The tested systems show satisfactory performance stability during cyclic H₂ separation from CH₄/H₂ and natural gas/H₂ gas mixtures. We also present the first investigation of the Mg-based systems (with Ni or Fe catalyst) after the cycled separation processes. The results of complementary techniques revealed H₂ separation-induced chemical and phase segregation in the studied materials. Moreover, we report the observation of networked MgH₂ microstructure formation. This research points out the potential of metal-hydrides in the H₂ separation sector as well as the challenges facing their application – especially those related to the presence of CO₂ impurity in the gas mixture. The unique and detailed description of processes taking place in a reactor during the separation process will significantly impact the design of future metal-hydrides-based scaled-up systems for H₂ separation.

---

Introduction

Climate change, environmental pollution, and biodiversity loss are forcing humankind, the initiator of critical changes on Earth, to change its current way of life. Since the warnings contained in the Club of Rome report regarding the negative impact of CO₂ on the Earth's atmosphere, many Climate Change Conferences focus on putting pressure on countries worldwide to defossilize their economies. One option that is perceived as a game changer for the global economy is hydrogen (H₂).

The transformation to a hydrogen economy and its associated challenges can be followed by the example of Germany – a country transitioning to a zero-emission economy.¹ Considering the long-range H₂ transport sector in Germany, which connects the H₂ production sites with the end-use destinations, the total length of pipelines is planned to reach 9700 km by 2032 (Fig. S1†).² Approximately 60% of the mentioned total H₂ pipeline length will operate using currently available natural gas pipelines.² For this reason, to fulfill the transformation goals during the energy transition (towards the hydrogen economy), H₂ will be transported together (as a stable blend) with natural gas (NG). The transfer of NG/H₂ gas mixtures from North Africa to Europe utilizing the existing pipelines is also under consideration.³ Intermediate solutions are also expected in many technology sectors, such as gas turbines, which require further development to enable the full transformation from operating on NG to pure H₂ (Fig. S2, Chapter S1 of the ESI†).

In most cases, the transport of H₂ blended in NG must be combined with its recovery at the end of the pipeline. Therefore, apart from the key sectors of the hydrogen economy (H₂ production, transport, storage, and end use), H₂ separation and purification are exceptionally important technologies that allow the merging of all the economic sectors. They not only enable the efficient H₂ production and safe operation of H₂-containing gas mixtures but are also crucial to supplying high-purity H₂ to the sectors and devices where the H₂ purity is critical – e.g., fuel cells.^4–6

There are many established H₂ separation/purification techniques: pressure swing adsorption (PSA),^4,5,7 temperature swing adsorption (TSA),^8–10 vacuum swing adsorption (VSA),^8,11 cryogenic separation,¹² membrane techniques,^13–16 and electrochemical compressors/separators (EHC/S).^17–20 All these methods have their advantages and disadvantages. For example, the PSA technique, which separates >99% pure H₂ is limited to operating at relatively high initial H₂ concentrations (75–90%).²¹ Moreover, PSA, TSA, and VCA operate discontinuously. The cryogenic methods, whereas, are relatively expensive and deliver H₂ of only <98% purity.^12,22 The use of Pd-membranes, which show very good H₂ separation properties, is limited by the high cost of Pd. Moreover, these membranes suffer from low resistance to poisoning (by CO and H₂S) and high operating temperature (>300 °C).^13,23 Perovskite-based proton-conducting membranes are characterized by a low H₂ flux, are sensitive to gas impurities (CO₂ and H₂O), and their performance largely depends on their production method.²² The H₂ separation selectivity of the polymeric membranes is limited,^24–26 and can only be improved using a complex two-step system combining polymeric membrane and PSA.^24,27 The EHC/S systems, can operate continuously and separate H₂ of relatively high purity (100–500 ppm CH₄ in H₂ on the cathode site). However, their restraint is the low H₂ yield (<90%) when a gas mixture with a low H₂ concentration is supplied at the anode site.²⁰

An alternative to the H₂ separation methods discussed above are metals and alloys that can react with H₂, forming hydrides. The principle of metal hydride operation is based on the selective absorption of H₂ from the gas mixture used. While H₂ is chemically stored in the metallic structure, the other gas components remain unreacted in the initial gas mixture.²⁸ Separation using metal hydride-based systems does not involve a significant pressure drop, so no additional energy is needed to compress the residual gas. The selectivity of the metal hydride can be tuned by the selection of chemical composition, crystal structure, and operating temperature.

Several metallic systems (mostly from low-temperature, AB₅-type intermetallic compounds) have already been used to separate H₂ from various gas mixtures.^29,30 For example, MmNi_5−xAl_x (where Mm stands for mishmetal) can effectively recover high-purity H₂ (grade 5.0) from the synthetic ammonia production plant's pre-treated purge gas (with lower ammonia and moisture content).³¹ T. Saitou et al. showed that another low-temperature system – FeTi_0.95Mm_0.08 can also separate highly pure H₂ (grade 7.0) from CH₄-containing gas mixtures.³² In another work, F. R. Block et al. revealed that LaNi₅ and FeTi could absorb H₂ from gas mixtures. However, different gas impurities such as CO, CO₂ and H₂S strongly poisoned the active material.³³ Currently, the most serious restraint of the use of metal hydrides in separation applications is the gaseous poisoning of the material initiated by gases like O₂, CO₂, or H₂O vapor, which leads to the partial or complete deactivation of the material (mostly through its surface passivation).^34–36

Compared to low-temperature alloys (for example AB₅- and AB-type systems), high-temperature hydrides exhibit higher gravimetric H₂ storage capacities. The most studied representative of this group is MgH₂, which is considered cheap (3000 $/t Mg powder),³⁷ abundant, light, and capable of storing 6.0–7.6 wt% H₂.³⁸ Despite the potential of MgH₂ in H₂ separation applications being predicted in 1987, research on this topic is very limited.³⁹ S. Ono proved that Mg (catalyzed by Ni) separates large amounts of pure H₂ (grade 6.0) from H₂-containing gas mixtures.⁴⁰ In our recent study, we presented that a Mg-based system is capable of absorbing all of the provided H₂ from a limited volume of CH₄/H₂ (90 : 10) mixture.⁴¹ Although B. Bogdanović et al. demonstrated the possibility of using Mg to separate H₂ from H₂/CH₄ gas mixtures, the experiments were never extended to NG/H₂ (especially to gas mixtures rich in NG).³⁹ In addition, the detailed analysis of the Mg-based material after the separation experiments, which is necessary for the final materials implementation, has never been presented.

Herein, we demonstrate and discuss, for the first time, the H₂ separation from CH₄-rich gas mixtures (including NG-based blends) using Mg catalyzed by different metals (Ni or Fe). To the best of our knowledge, the effect of Fe catalyst on Mg separation properties (in any H₂-containing blends) has never been addressed. Ni and Fe proved to be effective catalysts for hydrogenation/dehydrogenation reactions in Mg-based systems (when pure H₂ is used in the reaction).⁴² Transition metals facilitate hydrogenation by accelerating the dissociation of H₂ molecules into the H atoms and dehydrogenation promoting H atoms diffusion between the subsurface and the surface (in the case of Mg-Ni system) and weakening the Mg-H interaction strength (in the case of Mg-Fe system).^43,44

The influence of different processing temperatures on the system performance was also demonstrated. This work contains a detailed analysis of the metal hydride systems after cyclic separation processes (to our knowledge, this has not been reported for any system so far), providing unique insights into the processes occurring during H₂ recovery. Furthermore, since CH₄ does not react with Mg and Mg-based systems, it is crucial for the feasibility of this hydride-based system to check the impact of NG pollutants on the durability of the active material.^40,45 Therefore, in this study, we specifically focused on the influence of CO₂, which, according to the literature, can be adsorbed on the Mg particles' surface and then be reduced to CH₄, simultaneously causing Mg oxidation.⁴⁰ The highlighted advantages and challenges of Mg-based systems will be useful for the design of metal hydride-based systems for H₂ separation and purification.

Experimental

Materials and gases

Pure Mg (Non-Ferrum GmbH, <45 μm, 99.8% purity), Ni (Honeywell Riedel-de Haën, <45 μm, 99.8% purity), and Fe (Sigma Aldrich, <45 μm, 99.5% purity) powders were used in this study. The morphology of these powders is presented in Fig. S3.† Depending on the experiment, Mg-Ni or Mg-Fe powder mixtures were used. Ni and Fe were applied as catalysts to improve the Mg hydrogenation/dehydrogenation kinetics.³⁹ The weight ratio of Mg to the second element equaled 95 : 5. The starting powders were mixed for one hour under Ar using SPEX 8000M mill (without milling balls). The mixing resulted in a fine blending of the elements and not in an alloy formation or amorphization of the shaken materials (Fig. S4†). To avoid oxidation, all the handling with materials before and after mixing or any experiments was done in the MBraun glovebox filled with purified argon atmosphere (O₂ and H₂O <1 ppm).

High-purity H₂, N₂, Ar, and CO₂ gases were used in all experiments (Air Liquide, grade 5.0). The CH₄ (grade 4.5) and natural gas – NG (Natural gas H) were purchased from Air Liquide. The composition of NG was evaluated/confirmed by mass spectrometry (MS) and gas chromatography (GC) – see Table S1, Fig. S5, and Chapter S2 in the ESI†. The CH₄/H₂, NG/H₂, and H₂/CO₂ gas mixtures were prepared on-site.

The CH₄/H₂ gas composition calibration curves were determined based on the results of GC performed on the gas mixtures well-determined in terms of composition (prepared on-site using the calibrated EL-FLOW Bronkhorst mass flow meters, with the accuracy of 0.01%) – see Fig. S6 and S7.† The linear and exponential fits were used for 5–15% and 75–95% CH₄ concentration in the CH₄/H₂ gas mixtures, respectively. The H₂ or CH₄ concentration calculations were based on the ratios between H₂ and CH₄ integrals. The calibration curves were used to evaluate the composition of the prepared on-site gas mixtures and the CH₄ concentration in the desorbed H₂.

Separation of H₂ from CH₄/H₂, NG/H₂ gas mixtures

The H₂ separation experiments were performed using the setup presented in Fig. S8.† Its detailed description can be found in Chapter S4 of the ESI†. A 100 mL autoclave and 500 mL reservoir were used in the gas separation experiments. The temperature of the reservoir was kept constant at 30 °C.

Before any experiments, the setup with a loaded autoclave (with one gram of Mg-Ni or Mg-Fe system) was evacuated several times under dynamic vacuum and purged with pure Ar at room temperature. The Mg-Ni and Mg-Fe systems were firstly hydrogenated (about ten times) under pure H₂ (20 bar H₂). These hydrogenation cycles were treated as activation of the material – to obtain a stable hydrogen storage capacity and good hydrogenation kinetics (the details are published elsewhere).⁴¹ Next, the systems were used to separate H₂ from CH₄/H₂ and NG/H₂ gas mixtures under 100 bar of gas mixture and 20 bar of partial H₂ pressure (the total pressure of the gas mixture was 100 bar to fix the H₂ partial pressure at the level of 20 bar of H₂ in all the experiments).

The hydrogenation was performed under isothermal conditions at 350 °C (for three h), followed by cooling to 215 or 195 °C (in 30 minutes). The only exception (regarding the hydrogenation time) was the first hydrogenation, where 24 h absorption was performed. Fig. S9† shows that the main part of the hydrogenation process is completed within half an hour. Nevertheless, to ensure the completion of the reaction, hydrogenation lasted for at least 3 hours in each hydrogenation/dehydrogenation cycle. The cooling was carried out by only heat losses to the environment, and no active cooling system was used. At each cycle, the hydrogenation was accomplished with the autoclave opened to the gas bottle for the entire process – which ensured an unlimited supply of H₂ necessary to reach full hydrogen storage capacity. Carrying out hydrogenation with the autoclave opened to the gas bottle also reflects the constant hydrogen content that can be expected in H₂ separation flow reactors.

After each hydrogenation process (when the autoclave was cooled to 215 or 195 °C), the gas remaining in the setup was vented, and the setup was five times purged using an alternately dynamic vacuum and 5 bar Ar. The autoclave with the hydrogenated material was then closed (through the valve), and the rest of the system was evacuated under a dynamic vacuum for one hour.

The dehydrogenation was performed in a part of the setup restricted to the autoclave and reservoir. In this case, the presence of a reservoir allowed for a significant increase in the system's volume during the H₂ desorption. It ensures that the rise of the pressure induced by the desorbed gas will not be too great, and the final pressure will not get close to the equilibrium one (at the dehydrogenation temperature) before dehydrogenation is completed. At the same time, the overall volume of the setup (composed of the autoclave and reservoir) is not too big, so the pressure change in the system is measurable with good precision – which allows us to calculate the amount of the desorbed gas and, therefore, the hydrogen storage capacity. The dehydrogenation process consisted of heating to 350 °C (in 30 minutes), followed by three hours of isothermal heating at the final temperature. The amount of the desorbed H₂ was calculated based on the change of the pressure in the known volume of the temperature-stabilized setup. After finishing the H₂ desorption, the dehydrogenated gas was expanded to the empty gas sampling bag for GC experiments.

After each dehydrogenation and before the following hydrogenation, the entire system was vented and degassed using a dynamic vacuum. After the final separation experiments, the autoclave was cooled to room temperature, and the materials in the hydrogenated state were transferred inside the autoclave to the glovebox to be further analyzed. All the samples were then studied using X-ray diffractometry (XRD), thermogravimetry (TG), differential scanning calorimetry (DSC), scanning electron microscopy (SEM), and energy dispersive X-ray spectroscopy (EDX).

Separation of H₂ from H₂/CO₂ gas mixture

Mg-Ni and Mg-Fe systems (0.3 g) were used to separate H₂ from the H₂/CO₂ gas mixture (99 : 1). In these experiments, a simplified setup was used (no gas reservoir; 26 mL autoclave). The material handling, degassing of the system, and hydrogenation/dehydrogenation temperatures were the same as described above. Before the separation experiments, each material was activated five times under pure H₂. The hydrogenation in the H₂/CO₂ gas mixture was performed under 50 bar total pressure in a closed autoclave (at approx. 0.5 bar CO₂ partial pressure). Due to the slow kinetics of the process, the hydrogenation lasted at least 36 hours until the pressure in the autoclave was equilibrated. The hydrogenation/dehydrogenation cycles in an H₂/CO₂ gas mixture were performed four times for both studied systems. The gas left after hydrogenation and the desorbed gases were tested by GC.

GC

The gas samples collected at different stages of the gas separation process (leftover gas mixtures after absorption or desorbed gas) were studied with a gas chromatograph Agilent 6850 Network GC System with a thermal conductive detector. The chromatograph was equipped with a 27 m long PLOT capillary pillar with a diameter of 0.53 mm made of fused silica. The carrier and detector gas was N₂. To study the gas composition, the inlet of the GC was heated up to 220 °C, and the carrier inner gas flow was 1.7 mL min⁻¹. The average velocity inside the pillar (at 80 °C) was kept for 5 min at a level of 21 cm s⁻¹. The gas samples were transported inside the gas sampling bag and measured immediately after the sampling. For each gas mixture, three measurements were done.

XRD

The crystal structures and phase compositions of the starting materials, as-prepared and hydrogenated Mg-Ni and Mg-Fe systems, were evaluated using an X-ray powder diffractometer. The diffraction patterns were detected on an STOE STADI P transmission diffractometer. In all experiments, Mo radiation (0.7093 Å) was used. The instrument was equipped with a primary Ge (111) monochromator (MoKα₁) and a position-sensitive Mythen1K detector. The data were collected between 5 and 70° 2θ with a step width of 0.015° 2θ. Measuring times per step was 20 s. For each measurement, eight scans were collected and summed after data collection. For the measurements, the samples were filled into borosilicate glass capillaries (0.3 or 0.5 mm outer diameter, 0.01 mm wall thickness) inside a glovebox and sealed afterwards. The measured diffractograms were compared with simulated data using the crystal structure data from the ICSD database.⁴⁶ The Rietveld refinement was performed using TOPAS Version 6 (Bruker AXS, Karlsruhe, Germany).⁴⁷

DSC/TG/MS

The desorption of H₂ from the hydrogenated Mg-Ni and Mg-Fe systems (after a series of H₂ separation experiments) was studied using Mettler Toledo TGA/DSC 1 STARe System thermogravimetric differential scanning calorimetry. All experiments were conducted on ∼15 mg of material placed inside a 40 μL Al sample crucible and run under 50 mL min⁻¹ Ar gas flow. The heating was performed at 10 °C min⁻¹ up to 480 °C. In several cases, the DSC/TG gas outlet was connected to the ThermoStar GSD 300 T2 mass spectrometer (MS) to record the gas desorption profiles (H₂ and CH₄) during the heating process. MS was also used to evaluate the composition of used NG.

SEM/EDX

The hydrogenated Mg-Ni and Mg-Fe systems were examined by scanning electron microscopy (SEM) – TM3030 (with an accelerating voltage of 15 kV) and Hitachi S-5500 (with an accelerating voltage of 30 kV). The chemical composition of the sample and distribution of elements were evaluated using Energy Dispersive X-ray Analysis (EDX) – Xplore Compact 30 (Oxford) and Thermo Scientific UltraDry (SSD). The samples were placed on a copper grid with a lacey (carbon) film or an aluminum stub with a Leit-Tab (carbon and sulfur). EDX was primarily used to determine the concentrations and distributions of Mg, Fe, and Ni, but it also allowed us to detect F, Si, and Ca traces (which were the impurities coming from the starting materials; not shown here).

Results and discussion

Separation of H₂ from CH₄-rich gas mixtures

Mg-Ni and Mg-Fe systems were used to separate H₂ from the CH₄/H₂ and NG/H₂ gas mixtures. In the first place, the accomplishment of the separation was evaluated based on the amount of desorbed H₂.

The evolution of the hydrogen storage capacity of the Mg-Ni system in the cycles (hydrogenation/dehydrogenation) of separation from different gas mixtures and at different starting desorption temperatures is presented in Fig. 1a. The hydrogen storage capacity measured after hydrogenation in pure H₂ was determined at the level of 5.41(13) wt%. Further experiments proved that the Mg-Ni system can separate H₂ well from CH₄/H₂ and NG/H₂ gas mixtures, reaching 5.44(17) and 4.86(35) wt% H₂, respectively.

Fig. 1 H₂ storage capacity (measured during the desorption starting at 215 or 195 °C) as a function of hydrogenation/dehydrogenation cycles performed on (a) Mg-Ni (b) Mg-Fe systems. The hydrogenation has been performed under 20 bar H₂ (labeled as H₂), 100 bar of CH₄/H₂ gas mixture (20 bar of partial H₂ pressure, labeled as CH₄/H₂), and 100 bar of natural gas/H₂ gas mixture (20 bar of partial H₂ pressure, labeled as NG/H₂). The accuracy of measured capacities is close to 0.05 wt%. (c and d) results of GC made on the natural gas/H₂ gas mixture (20% H₂, labeled as NG/H₂), reference CH₄/H₂ gas mixture (95% H₂, labeled as CH₄/H₂), gas desorbed from the Mg-Ni and Mg-Fe systems after hydrogenation under 100 bar of NG/H₂ gas mixture (20 bar of partial H₂ pressure, labeled as (Mg-Ni)_dehydr and (Mg-Fe)_dehydr, respectively). The dotted lines indicate the position of H₂ and CH₄ peaks. The blue area in (c) points out the region of patterns enlarged in (d).

The hydrogen desorption capacity related to the separation of H₂ from the CH₄/H₂ gas mixture is maintained on the same level as in experiments with pure H_2. It means that the Mg-Ni system can effectively separate H₂ from the gas mixtures dominated by CH₄ (an 80 : 20 gas mixture was used here). Moreover, surrounding the material with CH₄ does not hinder H₂ absorption or cause any side reactions that affect the separation capability.

Despite good separation abilities, a slight reduction of H₂ storage capacity and capacity deviations were observed for the experiments employing the NG/H₂ gas mixture. There are at least two plausible reasons for this phenomenon. The first is the partial poisoning of the Mg-Ni system by NG contaminations (CO₂ and H₂S – see Table S1 and Fig. S5†). The second is related to the partial instability of the Mg-Ni-based hydride at 215 °C. It is possible that because of the cycled hydrogenation/dehydrogenation, the kinetic of H₂ desorption was improved, and part of the stored H₂ was removed from the material during the system's purging (which was always performed after hydrogenation). The amount of H₂ desorbed from Mg-Ni system under isothermal conditions is presented in Fig. S10 and discussed in Chapter S5 of the ESI†.

To check whether the second explanation is at least partially responsible for the H₂ capacity reduction, we performed a series of dehydrogenation experiments starting at a reduced temperature of 195 °C (Fig. 1a). The results show that by lowering the starting desorption temperature (and therefore the purging temperature), the H₂ desorption capacity was restored at the level of 5.57(16) wt% H₂ after hydrogenation in pure H₂. The following separation of H₂ from the NG/H₂ gas mixture did not cause a reduction in the storage capacity, which was recorded over cycling at the level of 5.59(6) wt%. This equals to ∼59 mg H₂ desorbed from Mg-Ni system. It proves that purging/evacuating temperatures, which are directly related to the transition temperature between hydrogenation and dehydrogenation, are an essential factor to consider when designing the overall gas separation process.

Taking advantage of the optimization of the processing temperature for the Mg-Ni system, the separation properties of the Mg-Fe system were evaluated at the desorption starting temperature equaled to 195 °C. Similarly to Mg-Ni, the Mg-Fe system, exhibited good H₂ separation properties (Fig. 1b). The hydrogen storage capacity was maintained after switching the gas from pure H₂ to the CH₄/H₂ gas mixture. The capacities equaled 5.97(11) and 5.90(12) wt% for H₂ and CH₄/H₂ gas mixture, respectively. The Mg-Fe system was also able to separate 4.86(35) wt% H₂ when exposed to NG/H₂ gas mixture. This equals to ∼51 mg H₂ desorbed from Mg-Fe system. The decrease in H₂ storage capacity observed after changing the gas mixture from CH₄/H₂ to NG/H₂ should be associated with partial poisoning of the material with NG impurities (as discussed below).

The purity of separated H₂ was assessed using GC, based on the gas samples dehydrogenated from Mg-Ni and Mg-Fe systems. Fig. 1c and d presents the GC curves of the above-mentioned gases compared to the NG/H₂ (80 : 20) and standard CH₄/H₂ (5 : 95) gas mixtures (which were used as reference ones). Both gas samples exhibit a high H₂ purity of 99.9% (according to the calibration curves in Fig. S7†). The GC detected only traces of CH₄. To determine if the CH₄ molecules are adsorbed at the metal surface, we performed the same absorption/desorption experiment (the same temperatures, pressures, and heating/cooling regimes) under pure CH₄. After this experiment, we did not detect any CH₄, indicating the necessity of H₂ absorption/desorption processes in capturing the CH₄ traces. In another series of experiments, we tried to determine at what point, of MgH₂ to Mg transition, CH₄ is desorbed (e.g., only at the beginning or at the end of the process). However, our analysis always showed traces of CH₄, indicating CH₄ desorption associated with the release of H₂. The complete removal of CH₄ traces was impossible by Ar purging and degassing under a dynamic vacuum.

To summarize, we prove that independently of the applied catalyst (Ni or Fe), Mg-based systems can effectively separate significant amounts of highly pure H₂ from CH₄/H₂ or NG/H₂ gas mixtures. Moreover, the importance of the processing temperature was highlighted. To further elucidate the processes that occurred during the H₂ separation, the hydrogenated Mg-Ni and Mg-Fe systems (after the last separation of H₂ from NG/H₂ gas mixtures) were transferred to the glovebox to pursue detailed examination.

Mg-Ni system after separation of H₂ from CH₄/H₂ and NG/H₂ gas mixtures

The macroscopic observation of the hydrogenated Mg-Ni system revealed significant material inhomogeneity. To facilitate further analysis, we determined five material layers/samples, which are graphically presented in a simplified manner in Fig. 2a. The layers' thicknesses shown in Fig. 2a (and later in Fig. 4a and 5e) do not reflect the actual ones. Moreover, despite the distinct nature of each layer, there were also gradual transitions between them. The differences in the color of the samples collected from different layers are presented in Fig. S12 and S13.†

Fig. 2 (a) Scheme of the arrangement of the Mg-Ni system material after cycles of hydrogenation/dehydrogenation experiments under 20 bar H₂, 100 bar of CH₄/H₂ gas mixture (20 bar of partial H₂ pressure), and 100 bar of natural gas/H₂ gas mixture (20 bar of partial H₂ pressure). The numbers depict the positions of the layer/sample examined by different techniques; (b) BS-SEM image of the sample of 1st layer collected from the Mg-Ni system after cyclic experiments (at 15 kV). Scale bar: 100 μm; (c) the average concentration of Mg and Ni calculated based on the EDX results obtained for samples of different layers' collected from the Mg-Ni system after cyclic experiments; (d) hydrogen storage capacity determined from TG experiments on samples of different layers collected from the Mg-Ni system after cyclic experiments. Purple and blue bars show the amount of H₂ desorbed in the first (up to 305 °C) and jointly second and third steps of H₂ desorption (above 305 °C), respectively; (e) concentrations of different phases identified (via Rietveld refinement) for samples collected from various layers of the Mg-Ni system after cyclic experiments. The sample numbers in (c), (d), (e) correspond to those presented in scheme (a).

Light grey granules formed the first layer. The granules were easy to fragment with a spatula. At the same time, they exhibit an elastic tendency to stay bulk and not disintegrate into a fine powder (which is typical for brittle metal hydrides). We also observed dark red spots that were present mainly on the granules' surface and rarely in their inner parts.

All subsequent layers were in the form of powder. The second layer had a slightly darker grey color than the first and only a few dark red spots were observed therein. The third layer, which consisted of the largest part of the Mg-Ni system material, had a greater share of the dark red spots. We separated some dark red spots from the third layer and labeled them fourth sample. The last, the fifth layer was located on the bottom of the autoclave and was composed of very dark red powder. Additionally, we observed very thin and brittle white flakes attached to the autoclave lid (resulted from Mg sublimation and condensation in the coolest spot of the autoclave).

The layers described above were further examined using SEM and EDX. Surprisingly, we discovered that the granules (first layer) mostly comprise networked microstructure (Fig. 2b and S14†). The interconnected rod-like structures were several μm in diameter and tens of μm in length. The detailed SEM analysis shows that they were composed of nanosized grains (Fig. S15†). This networked structure must be responsible for the aforementioned elastic nature of the granules. The previous studies demonstrated that Mg-MgH₂ system is represented by many nano and microstructures.⁴⁸ One of the examples are nanowhiskers formed during the hydrogen induced disproportionation of Mg₂₄Y₅.⁴⁹

BS-SEM micrographs show the presence of additional, denser particles (bright part of the micrograph in Fig. 2b) of few μm in size. EDX analysis revealed that the networked structure consisted mainly of Mg, while the denser particles consisted of Mg and Ni (Fig. S16†).

The systematic analysis of SE- and BSE-SEM micrographs indicates that the share of Mg-Ni particles gradually increases with the number of subsequent layers to dominate over Mg particles in the fifth layer (Fig. S14†). Moreover, the networked microstructure visible in the first and second layers is mostly replaced by irregular Mg-based particles in the following layers. The change in the chemical composition of the following layers is clearly visible on EDX maps (Fig. S17†) and was summarized in Fig. 2c. The Ni concentration increased significantly from 2 to 27% between the first and fifth layers.

To investigate the influence of the observed chemical composition inhomogeneity on the hydrogen storage properties, the samples from all the layers were dehydrogenated using the DSC/TG apparatus. The DSC and TG curves show a multistep H₂ desorption, manifested by endothermic events and corresponding materials weight loss (Fig. S18†). According to the literature, the first event (up to 305 °C) is related to the first step of Mg₂NiH₄ dehydrogenation.^50,51 The following, at least three, overlapping endothermic events (above 305 °C) correspond to further desorption of H₂ from Mg₂NiH₄ phase (the second and third step of its decomposition) and dehydrogenation of the MgH₂ phase.^50,52 The initial temperature of the second and main part of hydrogen desorption corresponds well to the H₂ desorption kinetic curve, which showed the main dehydrogenation occurs above 285 °C (Fig. S11†). The DSC curves analysis, revealing the presence of the Mg₂NiH₄ phase, which in a lattice distorted (microtwinned) form is characterized by an orange-rust color (which turns dark red upon the surface oxidation), corresponds well with macroscopic and SEM/EDX observations.⁵⁰ A decidedly different DSC curve shape, with a sharp endothermic event was observed for the sample from the fifth layer. According to E. Rönnebro et al., this is related to the transformation of the distorted Mg₂NiH₄ phase from the low-temperature to high-temperature form, followed by rapid H₂ desorption (manifested in the DSC curve by the high-temperature shoulder of the peak).⁵⁰ The H₂ desorption of the fifth sample is completed by dehydrogenation of the MgH₂ phase (on-set temperature around 375 °C).

The total amount of H₂ desorbed during the DSC/TG experiments (4.5–5.93 wt% H₂; excluding the fifth layer, Fig. S18b†) fits well with the capacities reported in Fig. 1a. Fig. 2d summarizes the hydrogen desorption capacities of samples of different layers' in detail. The graph clearly shows that the overall H₂ storage capacity is decreased for the following layers. Moreover, the amount of H₂ desorbed from the material within the first endothermic event is increased from the first to the fourth sample. This trend is consistent with all previous observations of chemical separation: The MgH₂ phase, capable of storing up to 7.6 wt% H₂, is gradually replaced in the following layers of the material by less stable Mg₂NiH₄ phase capable of storing ∼3.6 wt% H₂. This phenomenon ultimately results in the reduction of the hydrogen storage capacity and partial destabilization of the hydride. Moreover, the observed lower temperature stability of the Mg-Ni hydride (caused by the presence of the Mg₂NiH₄ phase) supports our conclusion of the necessity of lowering the starting dehydrogenation/system purging temperature (Fig. 1a).

As we presented in Fig. 1c and d the desorption of H₂ is accompanied by the desorption of CH₄ traces. To determine the period of CH₄ liberation, we analysed the gas desorbed from the DSC/TG apparatus with MS (Fig. S19†). The obtained curves show that CH₄ is desorbed at the same time as H₂, which confirms our previous observations. The amount of desorbed CH₄ is proportional to desorbed H₂.

CH₄ was not adsorbed on the metal-hydride particle surface as it could be easily removed from it by applying degassing under a dynamic vacuum and purging with Ar (see the discussion above). A mechanism based on the “hydride breathing” phenomenon is more likely. According to it, a small amount of a gas mixture (mainly CH₄) can be trapped in the pores of the MH particles, which close during hydrogenation (swelling of the active material) and open during dehydrogenation (contraction of the active material), releasing the trapped gas. Such a release process would strongly correlate to the H₂ desorption observed in this study. The change of the metal hydride porosity induced by the hydrogenation/dehydrogenation process was suggested by D. O. Dunikov et al. for La_0.9Ce_0.1Ni₅ alloy.⁵³

The SEM analysis of the dehydrogenated sample (after DSC/TG experiments) from the first layer of the Mg-Ni system showed the complete stability of the Mg networked structure after the desorption process. This proves that hydrogen is not essential for stabilizing this microstructure (Fig. 3).

Fig. 3 SE-SEM (a and c) and BS-SEM (b and d) of the dehydrogenated (under TG/DSC experiment, 10 °C min⁻¹) 1st layer’ sample (Fig. 2a) collected from the Mg-Ni system after cyclic separation experiments (15 kV). Scale bars: 100 μm (a and b), 30 μm (c and d).

The further analysis focused on investigating the correlation of chemical inhomogeneity with the phase composition of subsequent material layers. It was determined based on the Rietveld refinement of the XRD patterns obtained for each layer's sample (Fig. 2e and S20†). The results revealed that most of the Mg-Ni system formed MgH₂, two Mg₂NiH₄ (monoclinic-m and cubic-c), and Mg₂NiH_0.3 hydride phases upon hydrogenation under the gas mixtures. In subsequent layers of the material, a tendency for MgH₂ to be replaced by both Mg₂NiH₄ phases was observed. This agrees with the discussion of the SEM/EDX and DSC/TG results and is consistent with the observed change in the color of following layers of the material. Chemical and phase segregation occurring in successive layers of the material result from differences in phase densities (1.74 and 3.28 g cm⁻³ for Mg and Mg₂Ni, respectively) and is induced by repeated severe volume expansion and contraction of the Mg-Ni system during the hydrogenation/dehydrogenation cycling (∼30%).⁵⁴ The material volume change accompanied by gas flows resulted in particle mobility. This caused the heavier Ni-containing particles to gradually settle to the bottom of the autoclave.

Approximately 9 wt%. of MgO was detected in all hydrogenated Mg-Ni system layers. The MgO phase is formed due to the side reaction of Mg with NG impurities (as discussed below), and its presence negatively affects the system's hydrogen storage capacity. The Rietveld analysis did not detect any carbon-containing compounds, which proves that the Mg-Ni system did not react with CH₄ (the main component of NG). The same conclusion was reached in the past by B. Bogdanović et al.³⁹

In summary, the results of the complementary techniques proved the chemical and phase composition inhomogeneity in the hydrogenated Mg-Ni system after H₂ separation cycling. The relatively light, Mg-rich surface layer is gradually enriched in denser Ni-containing phases while moving to lower material layers. The described phase and chemical composition gradients also affect the hydrogen storage capacity of each material layer. The inhomogeneity and partial material poisoning (formation of the MgO phase) do not affect the system's ability to separate pure H₂ from gas mixtures.

Mg-Fe system after separation of H₂ from CH₄/H₂ and NG/H₂ gas mixtures

After the series of gas separation experiments, the hydrogenated Mg-Fe system was evaluated in the same way as the Mg-Ni system. The macroscopic observation revealed material inhomogeneity and the formation of four material layers, which are graphically presented in Fig. 4a and S21.† The upper layer consisted of several light grey and brittle granules, which, did not exhibit the elastic nature of the granules observed in the Mg-Ni system. In subsequent layers, the light grey powder gradually darkened, and rare reddish material spots appeared, the concentration of which continuously increased. Moreover, unlike the Mg-Ni system, no brittle flake layer attached to the lid was observed in the Mg-Fe system.

Fig. 4 (a) Scheme of the arrangement of the Mg-Fe system material after cycles of hydrogenation/dehydrogenation experiments under 20 bar H₂, 100 bar of CH₄/H₂ gas mixture (20 bar of partial H₂ pressure), and 100 bar of natural gas/H₂ gas mixture (20 bar of partial H₂ pressure). The numbers depict the positions of the layer/sample examined by different techniques; (b) the average concentration of Mg and Fe calculated based on the EDX results obtained for samples of different layers' collected from the Mg-Fe system after cyclic experiments; (c and d) BS-SEM images of the sample of 3rd layer collected from the Mg-Fe system after cyclic experiments (all at 15 kV). Scale bars: 100 μm (c) and 10 μm (d); (e) hydrogen storage capacity determined from TG experiments on samples of different layers collected from the Mg-Fe system after cyclic experiments. Purple and blue bars show the amount of H₂ desorbed in the first (up to 350 °C) and in the second step of H₂ desorption (above 350 °C), respectively; (f) concentrations of different phases identified (via Rietveld refinement) for different layers' samples collected from the Mg-Fe system after cyclic experiments.

The EDX and SEM results showed a gradual increase in Fe concentration in subsequent layers of the material (Fig. 4b, S22, and S23†). However, the differences in Fe concentration between the layers are less explicit than in the case of Ni in the Mg-Ni system. SEM images revealed that the hydrogenated Mg-Fe system consists of irregular Mg-based particles of tens of μm in diameter and denser Fe-rich particles of a few μm in diameter (brighter parts in BS-SEM images) – see Fig. 4c and S23.† The detailed SEM analysis showed that the Mg-based particles were intensely fractured after the cycled separation experiments (Fig. 4d and S23†). Since no networked Mg-based microstructure was observed in the Mg-Fe system, its formation must be associated with the presence of Ni in the Mg-Ni system.

The DSC/TG analysis shows two well-separated endothermic dehydrogenation events for each sample collected from the hydrogenated Mg-Fe system (Fig. S24†). The first event, in the range of 250–350 °C, corresponds to the desorption of H₂ from Mg₂FeH₆.⁵⁵ The second event, with an onset temperature of approximately 350 °C, manifested by a clearly visible endothermic reaction, is related to the dehydrogenation of the MgH₂. The share of the first event in the overall dehydrogenation storage capacity is increased for the subsequent layers. However, the differences in the capacity between layers are not as pronounced as in the Mg-Ni system, suggesting better resistance to chemical and phase segregation in the Mg-Fe system than in the Mg-Ni system (Fig. 4e). The dehydrogenation storage capacities (5.16–6.10 wt% H₂, Fig. 4e) corresponds well to the ones presented in Fig. 1b. The DSC/TG results agree with SEM/EDX and macroscopic observations which revealed a higher Fe concentration in subsequent layers and a darkening of the powder color. Mg₂FeH₆ particles, which are green, mixed with light grey MgH₂ particles cause a darkening of the powder, the final shade of which depends on the proportion of both hydride phases.

XRD analysis of samples from different layers confirmed that most of the Mg in the Mg-Fe system formed the MgH₂ phase upon separation experiment (Fig. 4f and S25†). Moreover, the Rietveld analysis confirmed that Fe only slightly separated between the layers. The larger amount of Mg₂FeH₆ (detectable by XRD) was confirmed only for the third and fourth layers. This fits well with the larger weight loss associated with the first dehydrogenation event (up to 350 detected °C) in the DSC/TG curves. However, most of the Fe did not react with Mg to form a hydride phase and remains metallic Fe. The Rietveld refinement also pointed to the relatively high concentration of MgO (10–20 wt%), which, as in the case of the Mg-Ni system, originates from poisoning inducted by NG impurities (as discussed below). The rare reddish spots observed macroscopically may be related to the formation of Fe₂O₃, the presence of which, however, due to its low concentration, was not confirmed by XRD analysis.

Although the results of complementary techniques showed chemical and phase segregation in the Mg-Fe system (after the cycled H₂ separation experiments), the inhomogeneity was not as severe as in the case of the Mg-Ni system. Moreover, the detected inhomogeneity and partial material poisoning did not affect the ability of the Mg-Fe system to separate pure H₂ from gas mixtures.

The influence of CO₂ impurity on the hydrogen storage properties

The NG impurities considered harmful to metal hydrides include O₂, CO, CO₂, H₂O, SO₂, and H₂S. The impact of these impurities on low-temperature hydrides (AB₅- and AB-type systems) was addressed by G. D. Sandrock et al.²⁸ Our study showed that the separation of H₂ from NG/H₂ leads to poisoning of Mg-based systems, resulting in partial Mg oxidation and loss of hydrogen storage capacity. Therefore, the following part of our research focused on understanding the distinct impact of CO₂ impurity (according to Table S1 and Fig. S5,† the main pollutant of NG used here) on the hydrogen storage ability of Mg-based systems. For this reason, the activated Mg-Ni and Mg-Fe systems were hydrogenated four times in H₂/CO₂ gas mixture (99 : 1). To emphasize the possible poisoning phenomenon, a relatively high CO₂ concentration was chosen (two orders of magnitude higher than in used here NG).

The analysis of the pressure change associated with hydrogenation of both systems under H₂/CO₂ gas mixture shows that the presence of CO₂ significantly slows down the reaction kinetics (Fig. 5a). The total hydrogenation time was increased from three (used in experiments with NG/H₂ gas mixture) to 30–60 hours. Moreover, the hydrogenation took place in two steps. A decrease in hydrogenation kinetics without a change in hydrogen storage capacity is called retardation and was observed to be induced, for example, by an NH₃-containing gas mixture.²⁸ Our observation differs from the results reported by B. Bogdanović et al., who observed no detrimental effect of 1% CO contamination (which reacts even more active with the metallic surface than CO₂) on the Mg-Ni system hydrogenation/dehydrogenation kinetics.³⁹

Fig. 5 (a) Change of the pressure during hydrogenation of activated Mg-based systems under ∼50 bar of H₂/CO₂ gas mixture (0.5 bar of partial CO₂ pressure); (b–d) results of GC made on the: H₂/CO₂ gas mixture (1% CO₂, labeled as H₂/CO₂), leftover gas after the hydrogenation process (labeled as hydr), gas desorbed from the Mg-based systems after hydrogenation under ∼50 bar of H₂/CO₂ gas mixture (0.5 bar of partial CO₂ pressure, labeled as dehydr). The dotted lines indicate the position of H₂, CH₄, and CO₂ peaks. The blue and purple areas in (b) point out the regions of patterns enlarged in (c) and (d), respectively; (e) scheme of the arrangement of the materials of Mg-based systems after four cycles of hydrogenation/dehydrogenation experiments under ∼50 bar of H₂/CO₂ gas mixture (0.5 bar of partial CO₂ pressure). * indicates that flakes were detected only in the case of the Mg-Ni system; (f) XRD pattern and Rietveld refinement plot of the hydrogenated Mg-Ni system powder sample. The concentrations of the main phases are presented in wt%; (g) DSC and TG curves obtained for the hydrogenated Mg-based system powder samples (10 °C min⁻¹).

The analysis of the gas leftover after the hydrogenation process revealed that all CO₂ present in the initial gas mixture was reduced to CH₄ (Fig. 5b–d). This reaction was catalyzed by Ni and Fe catalysts according to the following Sabatier reaction:

CO₂ + 4H₂ ⇄ CH₄ + 2H₂O (1)

The catalytic activity of Ni and Fe towards the methanation of CO and CO₂ has been studied elsewhere.⁵⁶

As a result of this reaction, twice as many H₂O molecules are produced than CH₄. At the hydrogenation temperature (350 °C) used herein, the presence of H₂O vapor causes the observed oxidation of Mg. The heterogeneous CO₂ reduction reaction may also be partly responsible for the observed reduced hydrogenation kinetics. When Ni and Fe are used not only in the catalysis of the hydrogenation reaction but also in the reduction of CO₂, the hydrogenation rate is decreased. The following analysis of the gas desorbed from the metal hydride (Fig. 5b and c after dehydrogenation) confirmed that highly pure H₂ (with only traces of CH₄) is recovered from the material despite its partial poisoning.

After a series of hydrogenation experiments in H₂/CO₂ gas mixtures, the hydrogenated Mg-Ni and Mg-Fe systems were further studied by XRD and DSC/TG techniques. The macroscopic observations revealed that both Mg-based systems were homogeneous light gray powders (Fig. S26–S28). No dark red spots were observed. For the Mg-Ni system, an additional layer of light gray flakes was observed on the autoclave lid (the flakes were easy to fracture but did not immediately fall apart when touched).

The XRD analysis proved the assumed oxidation of the Mg in both systems (Fig. 5f and S29†). The Rietveld refinement of the XRD patterns showed that >60(4) wt% of the systems consist of the MgO phase, which indicates a severe impact of the poisoning process. As expected, the formation of stable MgO affected the hydrogen storage capability, resulting in a reduction in storage capacity that was approximately proportional to the MgO phase content in the hydrogenated system. The hydrogenated Mg-Ni and Mg-Fe system samples desorbed 2.51 and 2.76 wt% H₂, respectively (Fig. 5g). The decomposition of metal hydrides (conducted using the DSC method) was similar to the processes observed for the samples after H₂ separation experiments (Fig. 5g, S18 and S24†).

The formation of a surface MgO layer on the Mg-based particles at the beginning of the hydrogenation reaction is also partly responsible for the hydrogenation retardation. As observed for the 2LiNH₂ + MgH₂ system, the presence of a surface oxide layer prevents diffusion of the atomic hydrogen into the metallic bulk.⁴²

The overall analysis of the results of experiments using an H₂/CO₂ gas mixture shows that 1% CO₂ in H₂ is fatally hazardous to the hydrogenation storage properties of the Mg-based system (affecting the hydrogen storage capacity and reaction kinetics). The poisoning, however, does not affect the system's ability to separate pure H₂ (even if the amount of separated H₂ is significantly lowered).

Conclusions and outlook

In conclusion, we demonstrated that Mg-based systems can effectively separate large quantities of pure H₂ from CH₄-rich gas mixtures. The purity of recovered H₂ is not affected by observed material poisoning. Both tested hydrogenation catalysts, Ni and Fe, presented similar performance. We highlighted also the importance of proper selection of the processing temperature on the amount of recovered H₂ during dehydrogenation.

Moreover, this research provided unprecedented insights into processes taking place during H₂ separation – undescribed so far for any metal hydride. The results of complementary techniques exhibited partial chemical and phase segregation in the Mg-based system, which was, less prominent in the Mg-Fe than in the Mg-Ni system. Our study showed that homogenizing the chemical distribution of the active material before gas separation does not ensure its maintenance during this process. Nevertheless, no negative impact of observed segregation on the abilities of the Mg-based systems to separate pure H₂ was observed. The particle mobility during cycled separation experiments can be significantly reduced by employing a binding material such as graphite. The addition of graphite not only immobilizes the metal particles but also increases the thermal conductivity of the material (which translates into more dynamic system operation).⁵⁷ The disadvantage of this solution is the reduction of the volumetric H₂ storage density of the separation system.

The awareness of the mentioned segregation, which certainly involves a change in the mass distribution inside the tested container, may be crucial when designing full-scale systems based on metal hydrides. This study shows that detailed post-processing analysis of the materials employed for H₂ separation/purification is crucial, even in systems as simple as Mg-Ni or Mg-Fe.

Interestingly, we presented the MgH₂ networked microstructure formation. We proved that this microstructure is stable after dehydrogenation.

Finally, we indisputably proved the fatal impact of CO₂ impurity on the hydrogen storage properties of the Mg-based systems. The formation of H₂O vapors accompanying the reduction of CO₂ to CH₄ causes the deactivation of Mg through its, at least surface, oxidation.

The use of MHs for the separation of H₂ from various gas mixtures undoubtedly requires solving this major issue. The origin of the poisoning may be related to a chemical reaction involving the active material, the catalyst, or both. An example of an active material-related origin of poisoning was discussed by F. Sun et al., who reported a decrease in the H₂ desorption capacity and a decline of process kinetics in KF doped (2LiNH₂ + MgH₂) material upon hydrogenation in H₂/CO (1 mol% CO). In that case, carbon from CO reacted with Li⁺, N³⁻, and K⁺ to cause permanent loss of the NH₂ groups of the active material and inactivation of the KF catalyst.⁵⁸ In contrast, in our study, the Mg-based system poisoning is caused by the presence of applied catalysts (Ni, Fe).

Prevention of poisoning in metal hydrides can be achieved through various approaches, offering potential solutions to this significant issue. One such approach is to use a different catalyst that enables the hydrogenation process to be carried out at a lower temperature and does not catalyze the CO₂ reduction reaction.⁵⁹ For instance, the CO₂ poisoning effect, although observed, was not severe in the case of AB₅-type systems operating below 100 °C.^29,30 The strong dependence of the CO₂ reduction reaction rate and its selectivity on the presence or absence of a catalyst, as well as the reaction parameters (temperature, time, etc.), was demonstrated by G. Amica et al. for the Mg-Co system.⁶⁰ Alternatively, the Mg-based system could be operated without a catalyst, but this would adversely affect the hydrogenation kinetics.

A promising approach is to modify the metal hydride surface with a protective layer that can separate gaseous species based on their molecular size (for example, employing the fluoride layer and aminosilane functionalization).^61,62 The improvement of the resistivity against CO and CO₂ poisoning for Mg-rich alloy was described by H. Wang et al.⁶³ Such a protective layer could be, however, ruptured during particle volume expansion and contraction that accompanies cyclic hydrogenation/dehydrogenation, resulting in an exposure of fresh, non-coated surface that could be poisoned.

Another solution involves an additional pre-purification system that eliminates severe pollutants (CO₂, O₂, H₂O, H₂S, CH₃SH) from the gas steam. The so-called natural gas sweetening plants are used as a standard purification unit eliminate sour compounds of natural gas such as CO₂ and H₂S.^64,65 The above-described phenomenon of Mg poisoning also requires looking at it from a different perspective. Because of the high CO₂ reactivity and full conversion of CO₂ into the CH₄, the Mg-based systems can be considered as potential candidates for the CO₂ capture and subsequent CH₄ conversion units.⁶⁶

The further development of the Mg-based system should also lead to achieving an H₂ storage capacity close to the MgH₂ theoretical one. The robust MH-based system could be employed not only to separate H₂ from CH₄-rich gas mixtures but also, for example, to recover H₂ from waste gases from the semiconductor production,⁶⁷ from tail gases of NH₃ synthesis,⁶⁸ as well as to purify H₂ produced from NH₃ decomposition or biomass hydrothermal gasification.^69,70 Moreover, some alkaline fuel cells are considered to operate well when fed with CH₄/H₂ gas mixtures transported through NG pipelines.¹⁹ The H₂ concentration in such gas mixtures could be adjusted by employing MHs.

Data availability

All the data supporting this article have been included as part of the ESI.†

Author contributions

M. Balcerzak: conceptualization, data curation, formal analysis, investigation, methodology, supervision, validation, visualization, writing – original draft, writing – review & editing. R. Urbanczyk: funding acquisition, supervision, writing – review & editing. F. Lange: data curation, investigation F. A. Helm: data curation, investigation J. Ternieden: investigation, data curation, writing – review & editing M. Felderhoff: funding acquisition, supervision, writing – review & editing. All authors have approved the final version of the manuscript.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This project (IGF Vorhaben 20761 N) was funded by the German Federal Ministry of Economics and Energy on the basis of a resolution of the German Bundestag. The authors gratefully acknowledge the help of Dr. Özgül Agbaba, who provided standard gas mixtures for GC calibration, Felipe Marques, who provided SEM images of pure elements, and the scientists from the Chemical Crystallography and Electron Microscopy Department (SEM and EDX analysis) and the High-Pressure Laboratory. Open access funded by DEAL Project. Open Access funding provided by the Max Planck Society.

References

1. F. Kullmann, J. Linßen and D. Stolten, Int. J. Hydrogen Energy, 2023, 48, 38936–38952.
2. Homepage of german Vereinigung der Fernleitungsnetzbetreiber Gas e.V. (FNB Gas), 14 March 2024, https://fnb-gas.de/en/hydrogen-core-network/.
3. S. Timmerberg and M. Kaltschmitt, Appl. Energy, 2019, 237, 795–809.
4. Z. Du, L. Congmin, Z. Junxiang, G. Xiuying, X. Yalin, S. Wei and H. Guangli, Catalysts, 2021, 11, 393.
5. M. Yáñez, F. Relvas, A. Ortiz, D. Gorri, A. Mendes and I. Ortiz, Sep. Purif. Technol., 2020, 240, 116334.
6. T. Yang, Y. Xiao and T.-S. Chung, Energy Environ. Sci., 2011, 11, 4171–4180.
7. R. Kumar, D. E. Guro and W. P. Schmidt, Gas Sep. Purif., 1995, 9, 271–276.
8. A. Golmakani, S. Fatemi and J. Tamnanloo, Sep. Purif. Technol., 2017, 176, 73–91.
9. R. R. Sadhankar, C. R. Aelick, D. L. Burns and K. Marcinkowska, Can. J. Chem. Eng., 2000, 78, 1087–1095.
10. Z. Wu, Z. Zhang and M. Ni, Energy Convers. Manage., 2018, 174, 802–813.
11. L. Dehdari, P. Xiao, K. Gang Li, R. Singh and P. A. Webley, Chem. Eng. J., 2022, 450, 137911.
12. D. Werner, Chem. Ing. Tech., 1981, 53, 73–81.
13. G. C. Bandlamudi, M. Wetergrove, L. N. Warr, J. Wartmann and A. Kruth, Energy Technol., 2024, 2301023.
14. H. Fan, A. Mundstock, A. Feldhoff, A. Knebel, J. Gu, H. Meng and J. Caro, J. Am. Chem. Soc., 2018, 140, 10094–10098.
15. S. Kämnitz, A. Simon, H. Richter, M. Weyd, U. Lubenau, T. Geisler and I. Voigt, Chem. Ing. Tech., 2022, 94, 49–55.
16. D. V. Strugova, M. Yu Zadorozhnyy, E. A. Berdonosova, M. Yu Yablokova, P. A. Konik, M. V. Zheleznyi, D. V. Semenov, G. S. Milovzorov, M. Padaki, S. D. Kaloshkin, V. Yu Zadorozhnyy and S. N. Klyamkin, Int. J. Hydrogen Energy, 2018, 43, 12146–12152.
17. K. Wagner, P. Tiwari, G. F. Swiegers and G. G. Wallace, Adv. Energy Mater., 2018, 8, 1702285.
18. A. Abdulla, K. Laney, M. Padilla, S. Sundaresan and J. Benziger, AIChE J., 2021, 57, 1767–1779.
19. K. Wagner, P. Tiwari, G. F. Swiegers and G. G. Wallace, Energy Environ. Sci., 2018, 11, 172–184.
20. S. Mrusek, M. Blasius, F. Morgenroth, S. Thiele and P. Wasserscheid, Int. J. Hydrogen Energy, 2024, 50A, 526–538.
21. D. Dunikov, V. Borzenko and S. Malyshenko, Int. J. Hydrogen Energy, 2012, 37, 13843–13848.
22. S. S. Hashim, M. R. Somalu, K. S. Loh, S. Liu, W. Zhou and J. Sunarso, Int. J. Hydrogen Energy, 2018, 43, 15281–15305.
23. S. Yun and S. T. Oyama, J. Membr. Sci., 2011, 375, 28–45.
24. W. Liemberger, M. Gross, M. Miltner, H. Prazak-Reisinger and M. Harasek, Chem. Eng. Trans., 2016, 52, 427–432.
25. P. S. O. Patrício, J. A. de Sales, G. G. Silva, D. Windmöller and J. C. Machado, J. Membr. Sci., 2006, 271, 177–185.
26. D. Wang, W. K. Teo and K. Li, J. Appl. Polym. Sci., 2002, 86, 698–702.
27. B. Li, G. He, X. Jiang, Y. Dai and X. Ruan, Front. Chem. Sci. Eng., 2016, 10, 255–264.
28. G. D. Sandrock and P. D. Goodell, J. Less-Common Met., 1984, 104, 159–173.
29. A. Kumar and P. Muthukumar, Int. J. Hydrogen Energy, 2023, 48, 37774–37783.
30. N. Hanada, T. Nakagawa, H. Asada, M. Ishida, K. Takahashi, S. Isobe, I. Saita, K. Asano, Y. Nakamura, A. Fujisawa and S. Miura, J. Alloys Compd., 2015, 647, 198–203.
31. M. Au, C. Chen, Z. Ye, T. Fang, J. Wu and O. Wang, Int. J. Hydrogen Energy, 1996, 21, 33–37.
32. T. Saitou and K. Sugiyama, J. Alloys Compd., 1995, 231, 865–870.
33. F. R. Block and H.-J. Bahs, J. Less-Common Met., 1983, 89, 77–84.
34. S. Bouaricha, J. Huot, D. Guay and R. Schulz, Int. J. Hydrogen Energy, 2002, 27, 909–913.
35. D. E. Dedrick, R. Behrens and R. W. Bradshaw, Sandia Report, SAND2007-4960, Sandia National Laboratories. s.l., 2007. https://www.osti.gov/biblio/920782.
36. C.-C. Shen and H.-C. Li, Int. J. Hydrogen Energy, 2015, 40, 3277–3282.
37. 13 March 2024, https://www.argusmedia.com/en/news-and-insights/latest-market-news/2547490-eu-magnesium-prices-enter-grey-area-below-3-000-t.
38. B. Bogranović, US Pat., Nr. 4,749,558, 1988.
39. B. Bogdanović and B. Spliethoff, Int. J. Hydrogen Energy, 1987, 12, 863–873.
40. S. Ono, Y. Ishida, E. Akiba, K. Jindo, Y. Sawada, I. Kitagawa and T. Kakutni, Int. J. Hydrogen Energy, 1986, 11, 381–387.
41. A.-L. Woeste, M. Balcerzak, R. Urbanczyk and M. Felderhoff, Energy Technol., 2021, 9, 2001079.
42. G. Liang, J. Huot, S. Boily, A. Van Neste and R. Schulz, J. Alloys Compd., 1999, 292, 247–252.
43. H. Liang, H. Zhang, Y. Zong, H. Xu, J. Luo, X. Liu and J. Xu, J. Alloy Compd., 2022, 905, 164279.
44. L. Zhang, L. Ji, Z. Yao, N. Yan and Z. Sun, Int. J. Hydrogen Energy, 2019, 44, 21955–21964.
45. F. Sun, M.-Y. Yan, X.-P. Liu, J.-H. Ye and Z.-N. Li, Int. J. Hydrogen Energy, 2015, 40, 6173–6179.
46. D. Zagorac, H. Muller, S. Ruehl, J. Zagorac and S. Rehme, J. Appl. Crystallogr., 2019, 52, 918–925.
47. A. A. Coehlo, J. Appl. Crystallogr., 2018, 51, 210–218.
48. C. Zhu, S. Hosokai, I. Matsumoto and T. Akiyama, Cryst. Growth Des., 2010, 10, 5123–5128.
49. C. Zlotea, J. Lu and Y. Andersson, J. Alloys Compd., 2016, 426, 357–362.
50. E. Rönnebro and D. Noréus, Appl. Surf. Sci., 2004, 228, 115–119.
51. J. Zhang, D. W. Zhou, L. P. He, P. Peng and J. S. Liu, J. Phys. Chem. Solids, 2009, 70, 32–39.
52. J. Čermák and L. Král, Acta Mater., 2008, 56, 2677–2686.
53. D. O. Dunikov, D. V. Blinov, A. M. Bozieva, A. N. Kazakov, A. A. Krapivina, I. A. Romanov, E. V. Zadneprovskaya and S. I. Allakhverdiev, Int. J. Hydrogen Energy, 2024, 51, 375–387.
54. K. Nogita, X. Q. Tran, T. Yamamoto, E. Tanaka, S. D. McDonald, C. M. Gourlay, K. Yasuda and S. Matsumura, Sci. Rep., 2015, 5, 8450.
55. M. Polanski, T. K. Nielsen, Y. Cerenius, J. Bystrzycki and T. R. Jensen, J. Hydrogen Energy, 2010, 35, 3578–3582.
56. S. De, J. Zhang, R. Luque and N. Yan, Energy Environ. Sci., 2016, 9, 3314–3347.
57. M. Lau, O. Ehrensberger, T. Weißgärber and F. Heubner, Int. J. Hydrogen Energy, 2024, 71, 562–570.
58. F. Sun, M.-Y. Yan, J.-H. Ye, X.-P. Liu and L.-J. Jiang, J. Alloys Compd., 2014, 616, 47–50.
59. F. J. Antiqueira, D. R. Leiva, G. Zepon and W. J. Botta, Int. J. Hydrogen Energy, 2022, 47, 470–489.
60. G. Amica, S. Rozas Azcona, S. Aparicio and F. C. Gennari, Phys. Chem. Chem. Phys., 2020, 22, 14720–14730.
61. M. V. Lototsky, M. Williams, V. A. Yartys, Ye. V. Klochko and V. M. Linkov, J. Alloys Compd., 2011, 509S, S555–S561.
62. M. Lototskyy, K. D. Modibane, M. Williams, Ye. Klochko, V. Linkov and B. G. Pollet, J. Alloy. Compd., 2013, 580, S382–S385.
63. H. Wang, Y. Liu and J. Zhang, Adsorption, 2022, 28, 85–95.
64. M. Mohajeri, M. Panahi and A. Shahsavand, Comput. Chem. Eng., 2024, 184, 108631.
65. A. Baudot, Natural Gas Sweetening, in Encyclopedia of Membranes, ed. E. Drioli and L. Giorno. Berlin Heidelberg: Springer-Verlag, 2012.
66. M. L. Grasso, J. Puszkiel, F. C. Gennari, A. Santoru, M. Dornheim and C. Pistidda, Phys. Chem. Chem. Phys., 2020, 4, 1944–1952.
67. L. Rochlitz, M. Steinberger, R. Oechsner, A. Weber, S. Schmitz, K. Schillinger, M. Wolff and A. Bayler, Int. J. Hydrogen Energy, 2019, 44, 17168–17184.
68. R. T. Abudu, Q. Sun, Z. Xu, X. Guo and L. Yang, J. Chem. Eng. Data, 2020, 65, 1715–1720.
69. L. Lin, Y. Tian, W. Su, Y. Luo, C. Chen and L. Jiang, Sustainable Energy Fuels, 2020, 4, 3006–3017.
70. M. Holewa, R. Dürrschnabel, N. Boukis and P. Pfeifer, Int. J. Hydrogen Energy, 2018, 43, 13294–13304.

---

Footnote

† Electronic supplementary information (ESI) available: Supporting figures and tables, additional characterization data and discussion. See DOI: https://doi.org/10.1039/d4ta05654j

---