Utilization of Robust Zr-Based Metal–Organic Framework for Efficient N₂/H₂ Separation

Abstract

The development of energy-efficient separation technologies for nitrogen (N₂)–hydrogen (H₂) mixtures under conventional industrial conditions remains a critical challenge. Adsorptive separation emerges as a promising strategy, contingent upon the design of high-performance adsorbents with tailored physicochemical properties. In this study, three isoreticular zirconium-based metal–organic frameworks (UiO-66, UiO-66-CH=CH₂, and UiO-66-F₄) are systematically evaluated for the purification of industrial N₂-H₂ mixed gases. By integrating static adsorption isotherms, dynamic breakthrough experiments, and molecular dynamics (MD) simulations, we comprehensively characterize the adsorption mechanisms and diffusion kinetics. The results demonstrate that UiO-66-F₄ exhibits optimal performance in capturing N₂ from N₂-H₂ mixtures under simulated industrial conditions, owing to its balanced binding affinity, high nitrogen uptake, and favorable micro-/mesoporous diffusion dynamics. This work offers a promising alternative for N₂-H₂ separation in industrial applications.

1. Introduction

As a kind of essential fuel and energy carrier, hydrogen has been widely applicated in modern industry and emerging energy sources due to having the advantage of zero emissions and being pollution-free [1,2,3,4,5]. Industrial sources of hydrogen include coal gasification, methane reformation, water electrolysis and industrial by-products such as biomass, chlor-alkali, and synthetic ammonia [6,7,8,9,10]. For example, recovering hydrogen from ammonia purge gas is an economic and environmentally benign method [10]. However, the nonpolar N₂/H₂ pair is one of the most difficult to be separated due to their similar molecular sizes (N₂: 3.64 Å, H₂: 2.89 Å), similar quadrupole moments (N₂: 4.7 × 10⁻⁴⁰ C·m², H₂: 0 C·m²), and similar physicochemical properties [8,11]. Compared with other feasible gas separation and purification methods such as condensation, cryogenic rectification, and membrane separation, the technology of adsorptive separation possesses advantages of high efficiency and low consumption, simple technology, and lower industrial-scale construction costs considering the present technical state [12,13,14,15]. Adsorbents with excellent adsorption properties are crucial to the adsorptive separation technique [7,16,17]. As a result, the development of novel adsorptive separation materials, which are more energy efficient than existing materials, will have a significant impact on the economic and environmental costs associated with global H₂ production.

Metal–organic frameworks (MOFs) have become a class of representative new porous materials in the fields of molecular capture/adsorptive separation due to their characteristics of flexible and tunable structure, functionally modified pore space, high specific area, and permanent high porosity [18,19,20,21,22,23,24]. Despite extensive research on MOFs, the vast majority focus on synthesis, with limited reports addressing their performance for hydrogen purification, particularly the separation of N₂-H₂ mixtures, and even fewer investigating this process under dynamic flow conditions [25,26,27,28,29]. Furthermore, the substantial adsorbent quantities required for industrial applications necessitate frameworks that combine excellent adsorption properties with straightforward synthesis to offset operational costs. Zirconium-based MOFs represented by UiO-66 (UiO for University of Oslo) are a fascinating family of robust porous materials with excellent stability, rapid adsorption, low production cost, and ease of regeneration [12,30,31,32]. This makes these frameworks highly attractive for commercial applications. Accordingly, we selected several isostructural UiO-66-type MOFs as candidates to evaluate their performance in separating N₂-H₂ mixtures. Previous research suggests that the affinity between the frameworks and H₂ molecules can be enhanced by exposing a high concentration of open metal sites (OMS) and expanding the π-conjugated system formed by ligand [25,28,33]. Complementing this, our prior work demonstrates that the affinity between the frameworks and N₂ molecule was enhanced by introducing F atoms with strong electron-withdrawing ability to the F-pillared-layer frameworks [8]. This enhancement arises because N₂ molecules possess a denser electron cloud distribution compared to H₂ molecules, leading to stronger induction interactions between N₂ molecules and F atoms of the framework [8,34]. Building upon these insights, three analogs with interpenetrative structure, namely UiO-66, UiO-66-CH=CH_2, and UiO-66-F4 were selected for adsorptive separating N₂ from H₂ molecules.

Herein, the adsorptive separation of N₂-H₂ mixture over Zr-based MOFs (UiO-66, UiO-66-CH=CH_2, and UiO-66-F₄) was investigated from a thermodynamics and kinetics perspective by combining adsorption experiments and molecular dynamic (MD) simulation. The performance of these hierarchically porous materials was characterized through pure component N₂/H₂ adsorption isotherms and high-throughput dynamic breakthrough experiments with binary mixtures. Experimental and simulation results jointly demonstrate that UiO-66-F₄ exhibits the longest breakthrough time and moderate N₂ adsorption capacity, rendering it highly suitable for N₂/H₂ separation. Its superior separation performance stems from the combined effects of binding site affinity, guest molecule adsorption capacity, and diffusion kinetics within its micro-mesoporous structure. Furthermore, UiO-66-F₄’s excellent cyclic stability and recyclability underscore its potential for industrial hydrogen purification from N₂-H₂ mixtures.

2. Experimental Section

2.1. Materials and Characterization

All raw materials and reagents used for synthesis were purchased from commercial sources and used without further purification. The used gases had the following ultra-high purity: H₂ (>99.999%), N₂ (>99.999%), and He (>99.999%) was purchased from BaiYan Co., Ltd. (Zibo, China). Powder X-ray diffraction (PXRD) data were collected at room temperature by using a conventional high-resolution (θ–2θ) diffractometer (D8-ADVANCED, Bruker, Germany) with Cu-Kα radiation at a scanning rate of 0.2° s⁻¹ from 3° to 50°. The relevant pore characteristics of the adsorbents (BET specific surface area, pore volume, pore size) are obtained from N₂ adsorption at 77 K and relative pressure (P/P₀) up to 1 using a Micromeritics 3Flex surface characterization analyzer (Micromeritics Instrument Corporation, Norcross, GA, USA). FT-IR spectra of the samples were recorded by a Nicolet 5700 spectrometer (Thermoelectric, Waltham, MA, USA) in the range of 50–7800 cm⁻¹.

2.2. Synthesis of UiO-66 Type Zirconium-Based MOFs

Isostructural UiO-66 samples were synthesized via a modified solvothermal protocol based on established methodologies [30,35,36,37,38], utilizing ZrCl₄ as the metal precursor (Scheme 1). Specifically, HBr acid-mediated modulation was incorporated during synthesis, resulting in UiO-66-nHBr variants (where n denotes the molar ratio of HBr to ZrCl₄). Two isostructural MOF derivatives with distinct pore architectures were obtained: UiO-66-1HBr (U₁) and UiO-66-5HBr (U₅). Additionally, UiO-66-CH=CH₂ (Uvinyl) and UiO-66-F₄ (UF₄) were prepared by introducing 2-vinylterephthalic acid and 2,3,5,6-tetrafluoroterephthalic acid ligands, respectively. All as-synthesized materials underwent post-synthetic activation via vacuum treatment at 423 K for 12 h prior to characterization and adsorption performance evaluation.

2.3. Pure-Component Adsorption Experiments

Pure-component adsorption measurements were performed at pressures spanning from 0 to 4000 kPa using a self-assembled static volumetric apparatus (equipped with key components from Shanghai Huotong Experimental Instrument Co., Ltd., Shanghai, China), as illustrated in Figure 1. Approximately 500 mg of sample was loaded into the adsorption slot, after which the entire system was hermetically sealed. Prior to the determination of adsorption isotherms, the free-space volume was measured utilizing helium at room temperature [39]. The gas uptake was calculated based on the material balance principle when expanding a known amount of gas from the reference slot to the adsorption slot. The criterion for achieving adsorption equilibrium was defined as a pressure change of less than 0.5% within 600 s. Subsequently, the gas adsorption amounts on the sample at various pressures were calculated [40,41]. Before each test, the entire sealed system was evacuated for 30 min.

2.4. Isotherm Fitting

Based on the calculation-derived characteristics of experimental data, optimized fitting parameters were systematically determined for curve fitting. This study employed the Langmuir (Equation (1)) [42] and Freundlich (Equation (2)) [43] models to fit the single-component adsorption isotherms of N₂ and H₂, respectively. Through comparative analysis of the plotted adsorption isotherms and fitting curves, the influence patterns of modified materials on the adsorption performance of H₂ and N₂ were systematically elucidated.

$$q_{eq}={\displaystyle \frac{q_{m}p}{p+\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$b$}\right.}}$$

(1)

$$q_{eq}=k{p}^t$$

(2)

where q_eq is the gas uptake (mmol/g), q_m is Langmuir volume reflecting the maximum uptake of the adsorbent (mmol/g), p is pressure (kPa), b (kPa⁻¹) and k (mmol g⁻¹ kPa^−t) are constants associated with temperature and adsorption heat, and t is a constant related to temperature, pore distribution, and adsorption intensity.

2.5. Isosteric Heat (Qst) and Ideal Adsorption Selectivity (SN₂/H₂)

To determine the coverage-dependent isosteric heat of adsorption (Q_st), the pure-component adsorption isothermal data were modeled with the Virial expansion method [28,44]. The adsorption equilibrium was described by the following equation:

$$llnP=\mathrm{l}\mathrm{n}\mathrm{N}+{\displaystyle \frac{1}{T}}{\displaystyle {\sum }_{i=0}^m}{a}_i{N}^i+{\displaystyle {\sum }_{j=0}^n}{b}_j{N}^j$$

(3)

where P (kPa) represents the equilibrium pressure, N (mg/g) denotes the gas uptake, T (K) is the temperature, and a_i and b_i are temperature-independent virial parameters [44,45]. The parameters m and n were optimized to ensure a robust fit to the experimental isotherm data. Subsequently, the Q_st was calculated via the following equation:

$$Q_{st}=-\mathrm{R}\times {\displaystyle {\sum }_{i=0}^m}{a}_i{N}^i$$

(4)

where R is the gas constant (8.314 J mol⁻¹ K⁻¹). This approach enables the extraction of energy landscapes governing adsorbate–adsorbent interactions under varying coverage conditions.

To predict competitive adsorption behavior in multicomponent systems, the Ideal Adsorbed Solution Theory (IAST) was employed to evaluate the adsorption separation selectivity of N₂-H₂ mixture (S_N2/_H2) on Zr-based MOFs. The IAST model integrates the pure-component adsorption isotherms derived from the Langmuir and Freundlich models, with fitting parameters obtained from the virial expansion analysis. The selectivity was quantified using the following expression:

$$S_{N_2/H_2}={\displaystyle \frac{x_{N_2}/x_{H_2}}{y_{N_2}/y_{H_2}}}={\displaystyle \frac{q_{N_2}/q_{H_2}}{p_{N_2}/p_{H_2}}}$$

(5)

where q_i and p_i are the molar loading and partial pressure of component i in equilibrium, respectively. This framework provides a theoretical basis for evaluating the separation efficiency of Zr-MOFs under mixed-gas conditions, bridging single-component adsorption data to practical gas separation scenarios.

2.6. Dynamic Column Breakthrough Experiments

Dynamic column breakthrough experiments were conducted on a commercial multicomponent gas adsorption system (Mixsorb S, 3P INSTRUMENT, Germany) integrated with a mass spectrometer (RGA, MKS, USA) as the detection unit. The cylindrical adsorption column featured an inner diameter of 4.57 mm and an effective length of 45 mm. Prior to experimentation, the adsorbent underwent thermal activation under He flowing (10 mL·min⁻¹) at 473 K (external heater setting, corresponding to column temperature ≈ 413 K) for 6 h. Subsequent breakthrough tests were performed under three distinct feed conditions: (1) H₂/N₂/He (10/10/80, vol%; carrier gas: He) at 30 mL·min⁻¹ total flow rate; (2) H₂/N₂/He (10/20/70, vol%) at 30 mL·min⁻¹; (3) H₂/N₂/He (20/10/70, vol%) at 25 mL·min⁻¹. All experiments were executed under controlled pressure and temperature regimes. Post-test regeneration of the adsorbent was achieved through He purge (10 mL·min⁻¹) at 453 K (external heater setting, column temperature ≈ 393 K) for 3 h prior to each successive run. The experimental sequence ensured complete desorption of previously adsorbed species between cycles, maintaining adsorbent integrity for comparative analysis.

2.7. Molecular Dynamics (MD) Simulations

Molecular dynamics (MD) simulations were performed to explore the confined diffusion behavior of gas molecules within UiO-type MOFs. First, the Vienna ab initio Simulation Package (VASP 6.4.2) [46] was utilized for the geometric optimization of the Cambridge Crystallographic Data files of U₁ and U_F4. The Perdew–Burke–Ernzerhof (PBE) density function [47] in combination with D3 dispersion correction [48] was used in this optimization procedure. The density cut-off was set at 500 Ry, and the overall energy converged to within 10⁻⁶ eV. To calculate the atomic charges of the frameworks, the Density Derived Electrostatic and Chemical (DDEC6) [49,50] method was adopted. Subsequently, Grand Canonical Monte Carlo (GCMC) simulations were performed by the RASPA code [51] using the polarizable force field parameters based on DREDING and Universal Force Field (UFF) [52]. In the GCMC simulation, both the frameworks and guest molecules were treated as rigid molecules. Specifically, H₂ and N₂ molecules were modeled as three-site rigid molecules. All atomic or metal ionic polarizabilities were obtained from references [53,54]. To ensure the accuracy of the force field, global scale factor ζ (ζ = 0.09) and scaling factors λ were induced to adjust the polarizability and the Lennard-Jones energy parameters respectively. This adjustment was crucial to ensure that the experimental isothermal adsorptive data were in good agreement with the simulated results. All MD simulations were performed within the NVT ensemble and were run for a total duration of 1 ns with a time step of 1 fs. The simulations consisted of 1 × 10⁶ cycles, including 2000 initialization cycles and 2000 equilibration cycles.

3. Results

3.1. Structural Characterization of Isostructural UiO-66 Samples

The crystalline integrity of the synthesized isostructural UiO-66 samples was first evaluated via powder X-ray diffraction (PXRD) analysis (Figure 2a). The observed diffraction peaks matched well with literature-reported values for isostructural UiO-66 frameworks [12,31,37], confirming the successful fabrication of high-purity crystalline materials via the solvothermal synthesis route. The main diffraction peaks are located at 2θ = 7.3° and 8.5°, corresponding to the (111) and (200) planes, respectively. Notably, the absence of extraneous peaks further attests to the absence of secondary phases, underscoring the structural homogeneity of the samples.

Complementary Fourier-transform infrared spectroscopy (FT-IR) analysis (Figure 2b) provided molecular-level insights into the framework composition. UiO-66-type materials show complete disappearance of the broad O-H stretch (2500–3300 cm⁻¹), replaced by asymmetric/symmetric COO⁻ vibrations at 1579 cm⁻¹ and 1396 cm⁻¹, confirming Zr⁴⁺-carboxylate coordination. All spectra lack free carboxylic acid peaks, confirming deprotonation and framework formation. U_vinyl retains vinyl C=C stretch at 1606 cm⁻¹ and =CH₂ deformation at 1012 cm⁻¹ and 928 cm⁻¹, while its COO⁻ bands shift to 1579 cm⁻¹ (ν_as) and 1410 cm⁻¹ (ν_s), indicating minimal electronic perturbation from conjugation. The adsorption band at 1750 cm⁻¹ is also observed in the spectrum of U_vinyl, which we assign to either a combination band of δ(H₂O) from trace adsorbed water with framework vibrations, or the C=O stretching vibration of dimers from uncoordinated -COOH groups [55,56]. U_F4 exhibits a high-frequency shifted ν_as(COO⁻) at 1620 cm⁻¹ (due to F-induced electron withdrawal) and intense C-F stretches (1271 cm⁻¹), suggesting a slight alteration in the coordination mode of the carboxylate group with Zr. The distinct splitting of the aromatic C=C skeletal vibration observed at 1473 cm⁻¹ and 1412 cm⁻¹ can be attributed to the strong electron-withdrawing inductive effect induced by the tetrafluoro substitution pattern. This substitution-induced electronic perturbation disrupts the symmetry of the benzene ring, resulting in vibrational mode separation that manifests as two resolved absorption peaks.

As summarized in

Table 1. Adsorbents properties: BET specific surface area (SSA), pore volume (V_p), HK Median pore width (MPW), and BJH desorption average pore width (APW).

Adsorbent	Sample Designation	SSA (m2/g)	Vp (cm3/g)	MPW (Å)	APW (Å)
UiO-66-1HBr	U1	1527	0.63	5.1	42.6
UiO-66-5HBr	U5	1303	0.55	5.4	38.2
UiO-66-CH=CH2	Uvinyl	801	0.45	4.8	43.7
UiO-66-F4	UF4	702	0.41	4.2	79.2

and Figures S1–S4, the isostructural UiO-66-type samples exhibit gradient characteristics in pore structures. U₁ (SSA = 1527 m²/g, Vp = 0.63 cm³/g) and U₅ (SSA = 1303 m²/g, Vp = 0.55 cm³/g) exhibited microporous-dominated structures (MPW = 5.1~5.4 Å), with their ultra-high surface area and pore volume making them ideal candidates for gas storage. U_vinyl (SSA = 801 m²/g, APW = 43.7 Å) showed a comparable average pore width to U₁ (APW = 42.6 Å) but with significantly enhanced mesoporous contribution. Despite reduced specific surface area (702 m²/g) and pore volume (0.41 cm³/g), U_F4 (APW = 79.2 Å) exhibits a shift toward larger pore size distribution. This structural transition is corroborated by the characteristic H₄-type hysteresis loop in N₂ adsorption–desorption isotherms (Figure S4), indicative of mesoporous structure rearrangement. U_F4 shows a much larger BJH average pore width than U₁, indicating the presence of significant mesoporosity. This mesoporosity mainly comes from ligand defect-induced pore structures. The isostructural UiO-66 materials displayed continuous evolution from microporous to micro-mesoporous structures, with U₁ achieving an optimal balance between adsorption capacity and mass transfer efficiency, while U_F4’s structural features may be suitable to address specific applications after weighing relevant considerations.

3.2. Adsorption Isotherms

Adsorption isotherms are crucial for understanding the interaction between adsorbates and adsorbents. To this end, the gas isothermal equilibrium adsorption data of pure N₂ and H₂ were measured at 278 K and 298 K, and the results are plotted in Figure 3. Across all experimental conditions, the adsorption capacities of N₂ were consistently higher than those of H₂ within the 0–4000 kPa pressure range. It can be attributed to the higher induced adsorption potential formed between zirconium metal and N₂ molecules via back-bonding interaction [29,57]. Additionally, molecular quadrupole moments and molecular sizes also play significant roles [11]. Both N₂ and H₂ uptake showed a downward trend with increasing temperature, indicating that these adsorption processes are typical thermodynamically controlled physisorption. Over the tested pressure and temperature ranges, the order of N₂ uptake from high to low was: U₅, U₁, U_vinyl, and U_F4. For hydrogen, U₅ exhibited the highest adsorptive capacity, followed by U_vinyl, U_1, and U_F4. At 278 K and approximately 4000 kPa (Figure 3a), the experimental adsorption capacities of U₅, U₁, U_vinyl and U_F4 reached 3.68 (N₂) and 1.13 (H₂), 2.56 (N₂) and 0.61 (H₂), 0.79 (N₂) and 0.81 (H₂), 0.71 (N₂), and 0.34 (H₂) mmol/g, respectively. As shown in Figure 3b, at 298 K and 4000 kPa, the adsorption capacity of U₅ for N₂ and H₂ decreased to 3.09 and 0.80 mmol/g, respectively. The N₂ adsorption capacities of U_vinyl and U_F4 were less remarkable compared to U₁ and U₅. This may be because U₁ and U₅ have larger pore volumes than U_F4 and U_vinyl (Table 1). In fact, at high pressures, pore volume is a key driving force for adsorption [33]. Moreover, the relatively large adsorption capacity of U_vinyl for H₂ is consistent with the conclusion proposed by Yaghi et al., which states that an expanded π-conjugated system will improve the H₂ adsorption capacity [58]. Nevertheless, the H₂ adsorption capacity was still lower than that of N₂. U_F4 had the lowest uptake and the smallest slope, which makes it easier to be regenerated. In the low-pressure range, the adsorption isotherms were irregular. This was due to the interference of atmospheric gases that entered the cell when it was transferred from the regeneration station to the analysis port, even though the amounts of gases were extremely small [33]. Given these results, it is challenging to intuitively determine whether the combination of these properties is favorable for the adsorptive separation of N₂-H₂ mixture using UiO MOFs. The potential performance of these Zr-based analogs in separation requires further investigation through a combination of theoretical analysis and dynamic breakthrough experiments.

By analyzing the original isothermal adsorption data, the ideal adsorption selectivity of each adsorbent for the N₂-H₂ mixture can be assessed based on a specific adsorption model. As depicted in Figure 3 and Figures S13–S16, within the measurement range, the Langmuir model was employed to fit all the experimental isotherms of N₂, whereas the Freundlich model was used for those of H₂. The fitted isotherms showed excellent agreement with experimental data (R² ≥ 0.99), and the detailed fitting parameters are presented in Table S1.

3.3. Isosteric Heat of Adsorption (Qst) and Ideal Selectivity

Apart from adsorption capacity, the isosteric heat of adsorption (Q_st) is another crucial parameter that helps to understand the trend, state, and spontaneity of the adsorption process, and to evaluate the interaction strength between adsorbate and adsorbent. The relevant parameters for fitting the adsorption heat using the Virial expansion method are detailed in the attached Figures S5–S12. The differences in Q_st for UiO MOFs are shown in Figure 4 and compared at nearly zero coverage (with a loading of 0.002 mmol/g).

As shown in Figure 4, on the given UiO MOFs at a loading of 0.002 mmol/g, the Q_st of N₂ adsorption was consistently higher than that of H₂. Among these frameworks, N₂ binds most strongly to U_F4, followed by U₁, U_5, and U_vinyl, with initial adsorption enthalpies of 23.4, 23.0, 20.8, and 19.9 kJ/mol, respectively. For H₂, the Q_st values in descending order are: U_vinyl (16.5 kJ/mol), U_F4 (16.1 kJ/mol), U₁ (14.7 kJ/mol), and U₅ (14.3 kJ/mol), indicating that the preferential adsorption of H₂ molecules by U_vinyl and U_F4 was more significant. U_vinyl and U₅, especially U_vinyl, showed relatively small differences in the initial adsorption heat of H₂ and N₂, suggesting weak differential adsorption effects. U₁ and U_F4, by contrast, exhibited stronger differential adsorption effects. Overall, the combined effects of pore size changes and the introduction of functional groups on the initial adsorption enthalpies may outweigh the negative effects of increased coordination defects on the adsorption heat. Naturally, the main interaction between the material and adsorbate is Coulomb attraction, resulting from the synergy of open metal sites (OMS), functionalized pores, and controlled preparation [28].

Figure 5 exhibits the IAST selectivity of N₂-H₂ mixture (N₂/H₂, 2/1, v/v) on Zr-based MOFs at 278 K and 298 K. As depicted, the N₂/H₂ selectivity values of the UiO MOFs at 278 K were higher than those at 298 K. U₁ and U₅ had the highest S_N2/H2 values, reaching about 6.5 at 278 K, and remaining above 4.0 even when the pressure approached 4000 kPa. Notably, the S_N2/H2 values of U₁ increased with pressure, peaking at 6.4 at 2000 kPa and 298 K. This may be related to nonnegligible adsorbate–adsorbate interactions at low coverage [45]. The experimental adsorption selectivity value was close to the simulated value (about 8.0 at 20 bar and 298 K, N₂/H₂, 3/7, v/v) in reference [7]. However, from an economic perspective of the adsorption process, selectivities greater than 4.0 have limited practical significance [17]. For U_F4, the N₂/H₂ separation selectivity was about 3 across the entire 0~4000 kPa pressure range, with higher selectivity at low pressures. Compared with other Zr-based MOFs, U_vinyl exhibited a low predicted selectivity of about 1.0, which may be attributed to the fact that the enlarged π-conjugated system increases the H₂ adsorbed amount (nH₂), and this was consistent with the adsorption heat and nN₂/nH₂. For a separation factor less than 2.0, it is challenging to design a satisfactory adsorption separation process [17]. Notably, since the adsorption amount of N₂ and H₂ on Zr-based MOFs is low at low pressure, especially U_F4 and U_vinyl, there are some uncertainties in the single-component adsorption isotherm fitting.

3.4. Dynamic Column Breakthrough Experiments

Based on the preceding discussion, U₁ and U_F4 were selected for multicomponent dynamic column breakthrough measurements to evaluate their N₂/H₂ separation performance at 298 K and 5 bar. As shown in Figure 6, U₁ exhibited limited N₂/H₂ separation capability across all tested gas mixtures (N₂/H₂/He ratios: 10/10/80, 20/10/70, 10/20/70, vol%). The breakthrough profiles revealed that N₂ displacement occurred before H₂ reached adsorption equilibrium, with gas concentration variations only marginally affecting the retention time difference (Δτ = 1.1–2.5 min/g). Increasing H₂ concentration to 20% induced a roll-up effect in its breakthrough curves (Figure 6c), suggesting weak H₂-framework interactions and low H₂ adsorption capacity. While the distinctive hump in Figure 6c,d hinted at potential kinetic separation under 10/20/70 vol% conditions [59,60], pressure elevation to 15 bar eliminated this feature (Figure 6e), demonstrating the impracticality of dynamic separation at elevated pressures. The distended breakthrough curves (Figure 6) indicated intra-crystalline diffusion limitations [61], confirming incomplete thermodynamic equilibrium in the fixed-bed system. In contrast, U_F4 demonstrated superior separation performance under identical conditions (N₂/H₂/He, 10/20/70, vol%; 25 mL/min, 298 K, 5 bar). As depicted in Figure 7, U_F4 achieved clear N₂/H₂ separation with N₂ retention time reaching 3.5 min/g, enabling pure H₂ recovery during the initial elution phase. The ratio of breakthrough time differential to operating pressure (Δτ/ΔP = 0.7 min·g⁻¹·bar⁻¹) exceeds the PSA process threshold of 0.5, confirming superior pressure-responsive behavior for industrial-scale gas separation. This performance difference arises not only from the affinity between the adsorbent and adsorbate but also from differential gas molecule diffusion rates within the U₁ and U_F4 frameworks.

3.5. MD Simulations

In GCMC simulations, the N₂-to-H₂ molar ratio in both U₁ and U_F4 materials is 1:2, consistent with the dynamic breakthrough experiments. The polarizability was calibrated using a global scale factor (ζ = 0.09), while Lennard-Jones energy parameters were scaled by λ to minimize experimental-simulation discrepancies. All polarizable force field parameters and atomic/ionic polarizabilities are provided in Table S2. Figure 8 compares simulated and experimental isothermal adsorption data, showing deviations within ±15% across most pressure conditions.

MD simulations reveal distinct gas diffusion behaviors within the adsorbents (Figure 9). The relatively fast diffusion of H₂ in U₁ is attributed to its high surface area and pore volume (Table 1), which minimizes steric hindrance, and its weak H₂-framework interaction (Q_st = 14.7 kJ/mol, Figure 4), which reduces surface residence time. The calculated self-diffusion coefficients for H₂ and N₂ in material U₁ were 3.87 × 10⁻⁴ cm²/s and 4.27 × 10⁻⁵ cm²/s, respectively, yielding a diffusion selectivity (D_H2/D_N2) of 9.05. In contrast, material U_F4 exhibited significantly lower diffusivities of 1.48 × 10⁻⁴ cm²/s for H₂ and 3.66 × 10⁻⁶ cm²/s for N₂, resulting in a markedly higher diffusion selectivity of 40.38. This superior H₂ diffusivity drives its preferential occupation of pore space during the initial adsorption phase, which aligns with competitive breakthrough experiments demonstrating an H₂ elution time advantage (Δτ = 3.5 min·g⁻¹). The pronounced difference in gas diffusivity, particularly within U_F4, underscores its kinetic preference for H₂ during adsorption initiation. This simulation result also shows good agreement with the dynamic penetration experiment.

However, thermodynamic equilibrium favors N₂ adsorption due to its stronger interactions with the framework, as evidenced by the difference in isosteric heats of adsorption (ΔQ_st (N₂-H₂): 8.3 kJ·mol⁻¹, U₁; 5.3 kJ·mol⁻¹, U_F4). This contrast highlights U_F4’s superior kinetic selectivity towards H₂ over N₂ compared to U₁. Crucially, although both adsorbents exhibit higher thermodynamic selectivity for N₂, the MD results confirm that the N₂-H₂ separation mechanism in these micro/mesoporous Zr-MOFs is primarily kinetically controlled.

The efficient N₂/H₂ separation achieved by U_F4 stems from a synergistic mechanism: strong micropore confinement effectively captures N₂ (reflected in a van der Waals interaction energy), while the mesoporous structure and the fluorinated pore facilitate accelerated H₂ transport. This combination enables U_F4 to meet the kinetic performance requirements (Δτ > 3 min·g⁻¹) for industrial PSA processes. Furthermore, U_F4’s enhanced N₂/H₂ adsorption selectivity, validated by competitive dynamic adsorption experiments, establishes it as a viable adsorbent for practical N₂/H₂ separation applications.

3.6. Cyclic Adsorption Performance Assessment

For practical separation applications, adsorbents must exhibit robust regeneration capability and structural stability. To evaluate the cyclic performance of U_F4, breakthrough experiments employing a N₂/H₂/He gas mixture (10/20/70, vol%; 25 mL/min) were conducted for 5 consecutive cycles at 298 K and 5 bar. Between each breakthrough run, the adsorption column was regenerated under 10 mL/min of He gas flow at 453 K (setpoint of the external heater, estimated bed temperature ~393 K) for 3 h.

As illustrated in Figure 10, the breakthrough curves and the characteristic breakthrough time for both N₂ and H₂ showed negligible variation across the 5 cycles. Furthermore, as confirmed by the PXRD spectra (Figure 2), the framework structure and characteristic morphology of U_F4 remained intact after the cyclic testing. This exceptional stability and reproducible performance under repeated adsorption-regeneration conditions strongly supports the practical viability of U_F4 for gas separation applications.

4. Conclusions

The present study demonstrates the potential application of UiO-66-F₄ (U_F4) for the selective separation of N₂-H₂ gas mixtures. This material exhibits excellent stability and strong adsorbate binding affinity. To evaluate its practical performance, breakthrough experiments were combined with molecular dynamics (MD) simulations. The breakthrough results reveal that U_F4 possesses higher selectivity towards N₂ over H₂, enabling the production of high-purity hydrogen gas from N₂-H₂ mixtures under specific conditions. The MD simulation results corroborated the breakthrough experimental findings for N₂/H₂ separation. Furthermore, this work highlights the promising prospects of breakthrough separation technology for addressing the N₂/H₂ separation challenge. However, research on N₂/H₂ separation using MOFs as adsorbents, particularly under high-flux conditions and employing breakthrough experiments, remains limited, and significant work is still required.