Adaptive Throat‐Sieving Gate in a Metal Azolate Framework Enabling Synergistic Equilibrium‐Kinetic Separation of Propylene and Propane

ABSTRACT

Developing porous adsorbents for propylene/propane (C₃H₆/C₃H₈) separation faces the challenge of integrating high adsorption capacity with fast adsorption kinetics. Herein, we address this challenge through precise pore control in a slightly flexible metal azolate framework NUM‐27a, enabling synergistic equilibrium‐kinetic separation of C₃H₆/C₃H₈. The periodically expanded throat gate enables effective impeding the diffusion of C₃H₈, while a large pocket‐shaped cavity decorated with exposed oxide groups facilitates exceptional C₃H₆ capture at low pressures. Specifically, NUM‐27a achieves exceptional low‐pressure C₃H₆ capture (89.61 cm³ cm⁻³ at 0.1 bar) and a record C₃H₆ packing density (310.0 g L⁻¹ at 0.01 bar). Kinetic analysis further reveals effective diffusion coefficient for C₃H₆ is 1.57 × 10⁻⁴ s⁻¹ at 298 K. Gas‐loaded single crystal X‐ray diffraction analysis coupled with computational simulations elucidate that the intrinsic pore geometry underpins the unique adsorption and separation performance. Breakthrough experiments validate the outstanding separation performance of NUM‐27a for C₃H₆/C₃H₈ mixtures. Moreover, the great stability, recyclability, and low‐cost precursors of NUM‐27a underline its potential as a reliable adsorbent for C₃H₆/C₃H₈ separation. The work unveils the adaptive throat‐sieving gate strategy with optimal separation performance for challenging gas separations.

We report a new micro‐flexible MOF, NUM‐27a, featuring an adaptive throat‐sieving gate and pocket‐like cavities that enable the synergistic equilibrium‐kinetic separation of C₃H₆/C₃H₈ mixtures. This unique structure endows NUM‐27a with an exceptionally low‐pressure C₃H₆ capture capacity and a record C₃H₆ packing density. The excellent stability, recyclability, and low‐cost precursors of NUM‐27a underline its potential as a reliable adsorbent for C₃H₆/C₃H₈ separation.

1. Introduction

Propylene (C₃H₆) stands as the primary olefin feedstock in the petrochemical industry, with the global demand for C₃H₆ projected to reach 170 million tons in 2025 [1, 2]. The pivotal role of C₃H₆ in the petrochemical industry is underscored by its indispensable position in meeting the surging demand for key chemicals, including but not limited to polypropylene, acrylonitrile, and isopropanol [3, 4]. In current industrial practice, C₃H₆ is predominantly produced through naphtha steam cracking or gas oil catalytic cracking, with propane (C₃H₈) concurrently generated as a major by‐product [5, 6]. The C₃H₈ impurity can severely impact the polymerization of C₃H₆, making its separation essential for achieving polymer‐grade purity (>99.5%) [7, 8]. The conventional separation of C₃H₆/C₃H₈ mixture relies on a cyclic distillation‐compression process conducted in massive splitter towers, which can reach heights of up to 300 feet and house more than 200 trays—under operating conditions of approximately 243 K and 0.3 MPa [5]. Evidently, such a separation process remains one of the most capital‐intensive and energy‐demanding operations in the industry [9, 10]. Therefore, the development of energy‐efficient separation technologies, such as adsorptive separation using porous materials, is crucial for potentially reducing the cost and the energy consumption [11].

Developing specialized adsorbents is the key to the success of adsorptive technologies [12, 13]. Among the emerging adsorbent materials, metal‐organic frameworks (MOFs) have been extensively explored as promising candidates for challenging gas separations, owing to the high tunability and versatile functionality at the molecular level [14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30]. Particularly, molecular sieving with infinite selectivity is recognized as the most ideal separation approach [4, 31, 32, 33, 34, 35, 36, 37, 38, 39]. Several sieving MOFs with tailored pore size and shape to selectively exclude the larger C₃H₈ molecules from the binary C₃H₆/C₃H₈ mixtures have been reported [3, 5, 40, 41, 42, 43, 44, 45]. The conventional molecular sieving approach mostly relies on single narrow channels that allow the complete separation of C₃H₆ from C₃H₈ based on molecular size or shape cut‐off, avoiding the co‐adsorption of C₃H₈ (Figure 1a). However, since the kinetic diameter difference of C₃H₆ and C₃H₈ is less than 0.4 Å, precise regulation of the pore size/shape matching is very challenging [46]. In addition, this approach typically restricts the diffusion rate and diminishes the host‐guest interaction between the framework and C₃H₆ due to limited pore volume [22, 47, 48]. On the other hand, quasi‐molecular sieving has been proven beneficial by accelerating C₃H₆ diffusion within the channel, while providing an extended adsorption domain to accommodate high C₃H₆ loading (Figure 1b) [46, 49, 50]. Yet it suffers from the drawback of co‐adsorption of C₃H₈, which further reduces the separation selectivity and the actual work production capacity. Thus, it is particularly challenging to optimize the adsorption kinetics and host‐guest interaction for sieving materials because of the difficulty of precise pore configuration control at the sub‐angstrom scale.

FIGURE 1.

Schematic illustration of sieving behaviors in different pore systems. (a) Conventional molecular sieving occurs when the target molecule A diffuses in a single narrow channel, and the impurity molecule B is completely blocked. (b) Quasi‐molecular sieving occurs when target molecule A and impurity molecule B exhibit discrepant diffusion rates while permeating through the narrow channels. (c) Adaptive throat‐gate sieving, where the expanded throat‐gate acts as a sieve, blocking the entry of impurity molecule B. The adjacent pocket‐shaped space provides an ample “adsorption domain” to accommodate target molecule A. Rose pink ball: target molecule A, the diameter is D_A; Coral orange ball: impurity molecule B, the diameter is D_B.

In light of this, a highly promising strategy for the challenging C₃H₆/C₃H₈ separation involves developing a pore‐engineered framework that integrates both an adaptive throat sieving gate and large cavities. In this rational design, the throat sieving gate exhibits conformational adaptation to C₃H₆, with its structural flexibility enabling subtle adjustments in aperture size and geometry upon interaction with C₃H₆ molecules. These adjustments optimally match the dynamic diameter of C₃H₆ while remaining a steric barrier to the larger C₃H₈ (Figure S1). The adjacent large cavities provide an ample “adsorption domain”, where internal volume not only accommodates a high loading of C₃H₆ molecules but also minimizes diffusion resistance to allow the adsorbed C₃H₆ to migrate freely (Figure 1c). This integrated pore architecture endows distinct thermodynamic and kinetic behaviors: the smaller C₃H₆ effortlessly permeates the gate for rapid diffusion, while engaging in strong host‐guest interactions with the exposed functional groups within the aperture. In contrast, C₃H₈ is sterically blocked by the throat‐sieving gate (kinetic exclusion) and thus excluded from the framework. This synergistic interplay between equilibrium and kinetics can overcome the long‐standing trade‐off between adsorption capacity and diffusion dynamics in sieving separation.

With this in mind, we present a new and micro‐flexible MOF, Zn₂(ATZ)₂(SO₄) (ATZ = 3‐amino‐1,2,4‐triazolate), termed NUM‐27a, which features a periodically porous configuration that aligns with the “adaptive throat‐sieving gate” model. As hypothesized, the narrow throat gate effectively impedes the diffusion of C₃H₈ and the pocket‐shaped space, which provides sufficient oxygen affinity sites for C₃H₆. NUM‐27a shows an ultra‐high ideal adsorbed solution theory (IAST) selectivity for equimolar C₃H₆/C₃H₈ mixtures (exceeding 3 × 10⁴). Additionally, kinetic analysis using an effective diffusion coefficient (D’) reveals fast adsorption kinetics for C₃H₆. Breakthrough experiments confirm the successful separation of a binary C₃H₆/C₃H₈ mixture as well as excellent cycling stability over multiple cycles of tests. Gas‐loaded single crystal X‐ray diffraction (SCXRD) analysis and modeling simulations further provide valuable insights into the separation mechanism at the molecular level. These studies provide in‐depth insights into the unique interactions and diffusion dynamics within the framework, furnishing a comprehensive understanding of its remarkable sieving separation behavior.

2. Result and Discussion

2.1. Synthesis and Characterization

A new sulfate‐pillared metal azolate framework, termed NUM‐27 [Zn₂(ATZ)₂SO₄·4H₂O], was synthesized for the first time by the hydrothermal reaction of zinc sulfate heptahydrate and ATZ in a mixed solution of water and trifluoroacetic acid at 453 K for 48 h. The roles of solvent and reaction temperature are crucial, as using more commonly used solvent systems or lower reaction temperatures would form different products, such as HAF‐1 [6] and Zn₂(aTz)₂SO₄ [51] (Figures S2, S3, and Table S1). Single crystal X‐ray diffraction analysis demonstrates that NUM‐27 crystallizes in the Pc space group (Table S2). Specifically, its simplest asymmetric unit consists of two Zn²⁺, two ATZ ligands, and one SO₄ ²⁻. Each Zn²⁺ center is coordinated with three nitrogen atoms from three ATZ ligands and one oxygen atom from the SO₄ ²⁻, adopting a four‐coordinate geometry (Figure S4). Two Zn²⁺ centers are bridged by two ATZ ligands to form a [Zn₂(ATZ)₂] secondary building unit (SBU). The adjacent [Zn₂(ATZ)₂] SBU undergoes a rotation of approximately 90°, positioning itself in two mutually perpendicular planes. This rotational effect causes the out‐of‐plane distortion of ATZ rings, which in turn drives the coordinated deformation of Zn─N coordination bonds, constructing a wavy 2D [Zn₂(ATZ)₂]_n layer. When these layers stack closely together, SO₄ ²⁻ serves as a bridge, facilitating the formation of a distinct 3D framework (Figure S5). Notably, immersing NUM‐27 in MeOH for several minutes will trigger a distinct single‐crystal‐to‐single‐crystal transformation. The new phase, termed NUM‐27‐MeOH [Zn₂(ATZ)₂SO₄(MeOH)·MeOH], crystallizes in the monoclinic P2₁ space group. Structural comparison reveals that Zn²⁺ centers exhibit two distinct coordination geometries: half of the Zn²⁺ centers coordinate with MeOH molecules, forming a distorted tetrahedral geometry, while the other half adopt a distorted trigonal bipyramidal geometry in NUM‐27‐MeOH (Figures S6–S8). However, NUM‐27‐MeOH is not a stable final phase. The unbalanced coordination environment formed by the involvement of MeOH molecules within the framework serves as a thermodynamic driving force, further triggering structural rearrangement. Upon guest removal, the cleavage of Zn─N coordination bonds occurs, accompanied by the conformational rotation of ATZ rings driven by thermodynamics. Ultimately, the coordination mode between Zn^{2 +} centers and ATZ rings is completely reconfigured, and MeOH molecules dissociate from the coordination environment, leading to the formation of the activated phase NUM‐27a [Zn₂(ATZ)₂SO₄] (Figure 2a,b, and Figure S9). The skeleton of NUM‐27a incorporates a regular pocket‐shaped space interconnected by a narrow throat‐sieving gate (Figure 2c). Notably, the throat gate aperture is constructed by four Zn²⁺ centers, two ATZ rings, and two SO₄ ²⁻, with the shortest distance between adjacent SO₄ ²⁻ being 4.34 Å. The pocket‐like cavity (5.97 × 6.03 × 5.75 ų) possesses adequate volume to accommodate both C₃H₆ and C₃H₈ guest molecules, whereas the throat gate size (4.34 Å) is slightly narrower than C₃H₈. Moreover, the degrees of freedom provided by the tridentate ATZ rings and SO₄ ²⁻ groups facilitate the deformation of the wavy 2D [Zn₂(ATZ)₂]_n layer when adapting to external changes, giving the structure a certain extent of flexibility to allow target molecules to pass through.

FIGURE 2.

Illustration of the framework structure and pore geometry of NUM‐27 and NUM‐27a. (a) The single‐crystal‐to‐single‐crystal transformations. (b) Framework of NUM‐27 and NUM‐27a. (c) Pore channels of NUM‐27a with a large pocket cavity and a narrow throat gate viewed along a representative direction. Color code: C, gray; N, light blue; H, white; O, red; S, yellow; Zn, blue‐gray.

From the optical microscope image (Figure S10), it is distinctly seen that NUM‐27 exhibits a colorless block. Powder X‐ray diffraction (PXRD) analysis verified the phase purity of the NUM‐27 and NUM‐27a (Figures S11 and S12). However, certain diffraction peaks (such as 100 and 011 planes) of NUM‐27a exhibit a shift toward higher angles relative to those of NUM‐27 (Figure 5a), indicating structural changes. The simulated PXRD patterns of HAF‐1, Zn₂(aTz)₂SO₄, and NUM‐27a further corroborate the structural differences among these three frameworks (Figure S13). Thermogravimetric analysis (TGA) reveals an initial weight loss of 12.6% observed from 25°C to 171°C, which is attributed to the removal of four water molecules in NUM‐27. This value is in agreement with the 15.5% weight loss calculated from the SCXRD data. Subsequently, a sharp and steep weight drop occurs upon further heating up to 400°C, indicating the onset of framework collapse. Notably, NUM‐27a exhibits remarkable thermal stability, maintaining its structural integrity under argon atmosphere at temperatures up to 400°C (Figure S14). The permanent porosity of the NUM‐27a was investigated using both N₂ and CO₂ as probe gases (Figure 3a). Due to its narrowed channel, a significant diffusion barrier for N₂ at 77 K was observed in NUM‐27a. As a result, CO₂ was chosen as the preferred probe. Notably, a subtle stepwise CO₂ adsorption behavior emerged in NUM‐27a: at low relative pressures (ranging from near zero up to P/P₀ = 0.017), the CO₂ uptake increases gradually from approximately 10 to 60 cm³ g⁻¹. Subsequently, a distinct plateau region emerges between P/P₀ = 0.017 and 0.03. Beyond this threshold (P/P₀ >0.03), the uptake curve undergoes a steep, exponential ascent, reaching 120 cm³ g⁻¹ at P/P₀ = 0.1. The Brunauer–Emmett–Teller (BET) surface area and pore volume are 502.88 m² g⁻¹ and 0.16 cm³ g⁻¹, respectively, aligning closely with the theoretical value (Figures S15 and S16). And the pore size was estimated to be 4.58–6.28 Å (Figure S17). This sequential behavior reflects responsive structural dynamics of the framework, where initial low‐pressure interactions gradually trigger a cooperative pore‐opening event at the critical pressure point.

FIGURE 5.

Characterization of crystalline, along with evaluation of dynamic breakthrough performance and cycling stability of NUM‐27a. (a) PXRD patterns of NUM‐27 and NUM‐27a. Dynamic breakthrough curves for equimolar C₃H₆/C₃H₈ on NUM‐27a at 298 and 313 K with a total flow of (b) 3.5 mL min⁻¹, (c) 1.5 mL min⁻¹. (d) Cycled breakthrough experiments for equimolar C₃H₆/C₃H₈ with a total flow of 3.5 mL min⁻¹ at 298 K. (e) The comprehensive comparison with other MOF benchmark materials.

FIGURE 3.

Gas sorption properties of NUM‐27a. (a) N₂ adsorption–desorption isotherm at 77 K and CO₂ adsorption–desorption isotherm at 195 K for NUM‐27a. (b) The adsorption isotherms of C₃H₆ and C₃H₈ on NUM‐27a at 298 K (Inset: the isotherm is presented in semilogarithmic scale to point out details in the low‐pressure region). (c) Cycling test of C₃H₆ adsorption measurements on NUM‐27a at 298 K from 0 to 1.0 bar. (d) The adsorption kinetic profiles of C₃H₆ and C₃H₈ on NUM‐27a at 298 K. (e) The fitting of the kinetic curve for C₃H₆ at 298 and 313 K. (f) IAST selectivity for equimolar C₃H₆/C₃H₈ mixture at 298 K. (g) Isosteric heat of adsorption of C₃H₆ for NUM‐27a. (h) Comparison of the zero‐coverage heat of adsorption of C₃H₆ and uptake at 298 K and 0.01 bar with those of typical MOF materials for C₃H₆/C₃H₈ sieving separation. (i) Comparison of C₃H₆ packing density at 298 K and 0.01 bar with those of top‐performing MOF materials.

2.2. Gas Sorption Analysis

Inspired by its intriguing pore configuration, the single‐component adsorption isotherms of C₃H₆ and C₃H₈ on NUM‐27a were measured at various temperatures over a pressure range of 0–1.0 bar (Figure 3b and Figures S18 and S19). The steep C₃H₆ isotherm and near saturation at low pressures indicate a strong affinity of the framework for C₃H₆ at all tested temperatures. Additionally, the negligible adsorption capacity of C₃H₈ was observed (8.67 cm³ cm⁻³ at 298 K), validating the effective exclusion of C₃H₈ due to the optimal aperture of NUM‐27a. However, as the temperature increases, the adsorption capacity of C₃H₈ gradually rises, which is presumably attributed to the synergistic effect of the enhanced diffusion kinetics of C₃H₈ and the expanded throat gate size at elevated temperatures. Specifically, at 298 K, NUM‐27a demonstrated a volumetric uptake of 46.36 cm³ cm⁻³ at 0.01 bar, which increased to 89.61 cm³ cm⁻³ at 0.1 bar, and further rose to 94.53 cm³ cm⁻³ at 1.0 bar, superior to those of well‐known molecular sieve MOFs (Figure 3h and Table S3) including ZU‐609 (0.052/12.43/76.19 cm³ cm⁻³) [4], Y‐abtc (3.52/∼53.85/65.27 cm³ cm⁻³) [52], Co‐gallate (∼1.56/∼19.54/66.61 cm³ cm⁻³) [42], UTSA‐400 (4.45/∼46.3/92.1 cm³ cm⁻³), etc.

Gas diffusion is inherently coupled to the adsorption–desorption efficiency for porous materials, where suboptimal diffusion behavior precipitates a degradation in separation performance [53, 54]. Herein, time‐dependent gas adsorption measurements for pure‐component gases at 298 and 313 K yielded profound insights into the diffusion dynamics of C₃H₆ and C₃H₈ within the channel of NUM‐27a. As shown in Figure 3d, the saturated adsorption capacity of C₃H₆ stands at 91.94 cm³ cm⁻³ at 298 K. Notably, the diffusion time constant (D′ = D/R² ) was calculated by fitting the time‐dependent adsorption curves to a micropore diffusion model based on Fickian diffusion assumptions [55, 56, 57]. Here, R denotes the equivalent spherical radius of NUM‐27a crystals, derived from the crystal size distribution via the equivalent volume principle for block‐shaped particles (Figure S20). The D’ for C₃H₆ at 298 K is calculated to be 1.57 × 10⁻⁴ s⁻¹ (Figure 3e and Figures S21 and S22), surpassing most reported molecular sieve MOFs (Table S4). Meanwhile, the D’ for C₃H₈ at 298 K is calculated to be 8.28 × 10⁻⁷ s⁻¹ (Figure S23), yielding an ultrahigh C₃H₆/C₃H₈ kinetic selectivity of 189.4 for NUM‐27a. Moreover, the dynamic saturated adsorption capacity of C₃H₆ stands at 89.35 cm³ cm⁻³ at 313 K, and the value of D’ for C₃H₆ at 313 K is markedly elevated to 2.0 × 10⁻⁴ s⁻¹ (Figures S24 and S25). Compared to HAF‐1, which features a structurally analogous framework, NUM‐27a exhibits an approximately fivefold enhancement in C₃H₆ diffusion rate at 298 K. Key structural differences between the two materials (including throat size, cavity volume, and diffusion model) that underpin this kinetic superiority are summarized in Table S5. This pronounced disparity arises from the fact that NUM‐27a provides a sufficiently unobstructed pathway for C₃H₆ transport and promotes effective translational migration. The synergetic interplay between equilibrium and kinetics fundamentally underlies efficient separation of C₃H₆ and C₃H₈ in NUM‐27a.

To quantify the separation potential, ideal adsorbed solution theory (IAST) [58] calculations were performed using the IAST++ software [59] to assess the selectivity toward the C₃H₆/C₃H₈ mixture (Figures S26 and S27 and Table S6). As shown in Figure 3f, NUM‐27a exhibited an ultra‐high IAST selectivity of exceeding 3 × 10⁴ for equimolar C₃H₆/C₃H₈ at 298 K and 1.0 bar. Therefore, the effective balance between C₃H₆/C₃H₈ selectivity and C₃H₆ uptakes positions NUM‐27a as a highly promising C₃H₆/C₃H₈ separation agent. The adsorption enthalpy (Q _st) for C₃H₆ at zero coverage in NUM‐27a was calculated to be 28.99 kJ mol⁻¹ (Figure 3g). Notably, this Q _st value of C₃H₆ is markedly lower than those of molecular sieving MOFs with benchmark C₃H₆/C₃H₈ separation performance (Table S7), such as HAF‐1 (65.88 kJ mol⁻¹) [6], KAUST‐7 (57.4 kJ mol⁻¹) [3], Y‐dbai (55.0 kJ mol⁻¹) [52], etc. Such a relatively low Q _st value for C₃H₆ enables NUM‐27a mild regeneration, a trait that benefits energy‐efficient C₃H₆/C₃H₈ separation.

Packing density acts as a quantitative gauge of the efficient utilization of pore space, a critical factor for enhancing both the gravimetric and volumetric capacities of adsorbents [60, 61, 62]. The occurrence of extremely high packing densities in MOFs has been noted in prior literature [40]. We further calculated the packing density of C₃H₆ under different pressures for NUM‐27a (Table S7). Remarkably, based on the determined pore volume and C₃H₆ uptake, the C₃H₆ packing density was determined to be 310.0 g L⁻¹ at 0.01 bar, being the record‐high value in comparison with reported MOF materials (Figure 3i), along with far exceeding other top‐performing materials including but not limited to Co‐aip‐pyz (∼219.3 g L⁻¹) [40], Y‐dbai (206.0 g L⁻¹) [52], and NCU‐20 (∼130.0 g L⁻¹) [41]. Furthermore, at 298 K and 1.0 bar, the C₃H₆ packing density in NUM‐27a reaches a high value of 632.0 g L⁻¹, which is 370 times higher than that of the gaseous C₃H₆ density (1.707 g L⁻¹) [44].

2.3. Gas‐Loaded SCXRD Study

To directly visualize the location of the C₃H₆ molecule within the NUM‐27a framework, SCXRD experiments were performed on NUM‐27a after C₃H₆ loading at ambient temperature. In comparison to NUM‐27a (with unit cell parameters: a = 8.0388(2) Å, b = 9.9939(2) Å, c = 9.7799(2) Å, β = 108.904(2)°, and V = 743.33(3) ų), C₃H₆@NUM‐27a also crystallized in monoclinic P2₁ featuring an expanded unit cell (a = 8.2719(3) Å, b = 9.9777(3) Å, c = 9.7705(3) Å, β = 109.773(4)°, and V = 758.86(5) ų). NUM‐27a exhibits slight structural flexibility with a unit cell volume expansion of 2.1% upon C₃H₆ loading, lower than that of highly flexible MOFs such as NTU‐99‐NO₂ (volume expansion of 4.6% upon C₃H₆ loading) [63] and TYUT‐17 (volume expansion of 5.2% upon ethylene loading) [21]. SCXRD analysis further revealed that each unit cell of NUM‐27a can accommodate two C₃H₆ molecules, being identical to the adsorption isotherms. C₃H₆ molecules exhibit a preferential localization pattern (Figure 4b and Figure S28): they strategically occupy the central region demarcated by four adjacent ATZ ligands, and are further confined by four SO₄ ²⁻, which not only spatially encapsulate C₃H₆ but also establish a localized electrostatic microenvironment. After C₃H₆ loading, the ATZ ligands undergo a rotation by a certain angle (Figures S29 and S30), resulting in each network being perfectly tailored to the size of C₃H₆ molecules, thereby imposing a pore confinement on C₃H₆. This structural adjustment endows the necessary conformational flexibility to facilitate C₃H₆ access into the channels. In more detail, each C₃H₆ molecule within NUM‐27a is bound by the oxygen atoms from three SO₄ ²⁻ with multiple C─H···O interactions with distances of 2.29–3.16 Å. Furthermore, a short‐range C─H···N interaction (1.32 Å) is formed by H atoms from C₃H₆ and N atoms from the amino groups of ATZ ligands. Concurrently, the robust C─H···π (2.96 Å) are also observed (Figure 4a), collectively underscoring the strong binding affinity between C₃H₆ and the framework. Such a synergistic arrangement of organic ligands and inorganic anions could construct highly specific recognition and confinement sites for C₃H₆.

FIGURE 4.

Investigation of the separation mechanism. (a) SCXRD results of C₃H₆@NUM‐27a. (b) The corresponding C₃H₆ packing diagram along the pore channel. DFT‐calculated energy profiles and relative energies in NUM‐27a for (c) C₃H₆ and (d) C₃H₈ diffusions.

2.4. Modeling Simulation Studies

To elucidate the underlying diffusion behavior at the atomic level, theoretical simulations were employed. As depicted in Figure 4c,d, density functional theory (DFT) calculations involved diffusing a C₃H₆ or C₃H₈ molecule from outside into the cavity. The relative energy at different positions was calculated, and the energy barrier (ΔE) from state II to state III was used to evaluate the difficulty of guest molecules entering the framework. The results showed the energy barrier for C₃H₈ (97.24 kJ mol^{− 1}) is significantly higher than that for C₃H₆ (29.25 kJ mol^{− 1}). This high energy barrier accounts for the extremely low uptake of C₃H₈, indicating that its diffusion through the narrow channel is kinetically hindered. In contrast, C₃H₆ can diffuse smoothly into the cavity with minimal resistance. Furthermore, ab initio molecular dynamics (AIMD) simulations were performed at 300 K for a total duration of 1 ps to investigate the time‐resolved diffusion behavior of gas molecules within the NUM‐27a framework. The migration barrier profiles derived from AIMD simulations revealed that the difference in migration barriers for C₃H₈ (90.70 kJ mol^{− 1}) is much larger than that for C₃H₆ (16.40 kJ mol^{− 1}), providing a complementary insight into the diffusion kinetics in NUM‐27a (Figure S31). The theoretical calculations align remarkably well with the results of the adsorption data, further supporting the hypothesis of the sub‐angstrom scale regulation of its throat‐sieving gate, thereby establishing a clear structure‐performance relationship at the atomic level.

2.5. Breakthrough Experiments and Material Stability

Fixed‐bed column breakthrough experiments were conducted on NUM‐27a to evaluate the separation performance for an equimolar C₃H₆/C₃H₈ mixture. As expected, C₃H₈ undergoes rapid breakthrough within the column, indicating a negligible co‐adsorption. The sharp separations for the equimolar C₃H₆/C₃H₈ mixture were realized at various flow gas rates (Figure 5b,c, and Figure S32), exhibiting a substantial separation window. Specifically, at 298 K and a gas flow rate of 3.5 mL min⁻¹, C₃H₆ was retained for a longer time of 20.8 min g⁻¹. To further quantify the practical separation performance, we measured the dynamic capture capacities of the two gases within the breakthrough equilibrium time and calculated the separation factor, which is a core metric for evaluating the practical separation performance of adsorbents. The results demonstrate that the dynamic capture capacity of NUM‐27a for C₃H₆ reaches 65.36 cm³ cm⁻³, while that for C₃H₈ is only 0.89 cm³ cm⁻³, corresponding to a high separation factor of 73.44 (Figure S33). This value is notably higher than that of many reported MOF adsorbents for C₃H₆/C₃H₈ separation, such as TYUT‐23 (15.3) [46], Zn₂(BDC‐Cl)₂(Py₂TTz) (5.2) [64], and Zn₃(OH)₂(pzdc)(atz) (3.2) [65]. Notably, the dynamic capture capacity of C₃H₆ is much higher than most representative MOFs, including Co‐gallate (36.48 cm³ cm⁻³) [42], Co‐aip‐pyz (43.13 cm³ cm⁻³) [40], and HIAM‐301 (59.17 cm³ cm⁻³) [66]. Furthermore, the separation performance was further investigated at higher temperatures. As anticipated, at 313 K, the breakthrough times were reduced; however, effective and clean separation was still preserved.

The durability of adsorbents under operational conditions constitutes a critical determinant of their practical applicability [67]. Consequently, the application potential of NUM‐27a was comprehensively assessed. Primarily, the C₃H₆ and C₃H₈ adsorption cycle stability of NUM‐27a was evidenced by the absence of a significant decline in its adsorption capacity after five adsorption–desorption cycles (Figure 3c and Figure S34). Furthermore, cycling breakthrough experiments for an equimolar C₃H₆/C₃H₈ mixture were conducted (Figure 5d). Upon activation of NUM‐27a, the breakthrough curves manifest reproducible performance over at least six consecutive cycles, without significant performance decline. This observation unequivocally validates the outstanding recyclability of NUM‐27a for practical applications. The PXRD patterns reveal that the samples retain excellent crystallinity under a series of rigorous stability tests: exposure to air for 3 months, after adsorption and breakthrough experiments (Figure S35). Furthermore, both the CO₂ adsorption capacities at 195 K as well as the BET surface areas of NUM‐27a manifest negligible changes after cycling adsorption experiments (Figure S36). These findings collectively validate the superior stability of the NUM‐27a under operational conditions. Notably, endowed with low‐cost precursors, low adsorption heat, and high C₃H₆ packing density, NUM‐27a exhibits substantial potential for application in the petrochemical industry. (Figure 5e and Table S8).

3. Conclusion

In conclusion, this work demonstrates the pivotal role of the adaptive throat‐sieving gate model in enabling the synergistic equilibrium‐kinetic separation for C₃H₆/C₃H₈ mixtures within NUM‐27a. Ultra‐high IAST selectivity (exceeding 3 × 10⁴) for equimolar C₃H₆/C₃H₈ mixtures could be realized due to a narrow throat‐sieving gate. Through the incorporation of a large pocket‐like cavity functioning as a rapid‐transport channel with high‐affinity adsorption sites, NUM‐27a achieves a record‐high C₃H₆ packing density with 310.0 g L⁻¹ at 298 K and 0.01 bar, which is unprecedented in the C₃H₆/C₃H₈ separation. Kinetic analyses revealed a diffusion time constant of 1.57 × 10⁻⁴ s⁻¹ for C₃H₆ and 8.28 × 10⁻⁷ s⁻¹ for C₃H₈. Such a marked discrepancy underscores the preferential adsorption kinetics of NUM‐27a toward C₃H₆ relative to C₃H₈. Column breakthrough experiments confirmed the excellent separation capability on NUM‐27a, with no performance degradation observed over multiple cycles. Gas‐loaded SCXRD analysis and computational simulations provided insights into the origins of preferential C₃H₆ adsorption and diffusion. Our findings demonstrate the potential of engineering a narrow adaptive throat‐sieving gate in large pore space to tailor pore architecture for structurally similar molecule separation. Given its satisfactory separation performance, moderate Q _st of C₃H₆, and low‐cost precursors and robust framework, NUM‐27a holds great promise for practical industrial separation of C₃H₆ and C₃H₈.

4. Experimental Section

Experimental details are given in the Supporting Information. CCDC 2493331, 2476267, 2476268, and 2493369 contain the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

Funding

This work was financially supported by the National Natural Science Foundation of China (Nos. 22275102 and 22575128).

Conflicts of Interest

The authors declare no conflict of interest.

Supporting information

Supporting File 1: adma73459‐sup‐0001‐SuppMat.docx.

Supporting File 2: adma73459‐sup‐0002‐Data.zip.

Data Availability Statement

The data that support the findings of this study are available in the supplementary material of this article.