High-Surface-Area ZIF-67 Nanoflowers: Synthesis and Application Toward Enhanced CH₄/N₂ Separation in Mixed Matrix Membranes

Abstract

Under elevated loading conditions, the aggregation of fillers emerges as a pivotal factor driving the degradation of separation performance in mixed matrix membranes. The two-dimensional (2D) modification of fillers, aimed at enhancing interfacial contact with polymers, has been recognized as an effective strategy to improve interphase compatibility and increase filler loading capacity. However, it is worth noting that the BET surface area of 2D fillers is typically relatively low. In this study, a two-step approach was developed. First, a “diffusion-mediated” process was combined with a solvent optimization strategy based on first-principles (DFT) calculations, achieving a 20-fold suppression in ZIF-67 nucleation-crystallization rate. This enabled the successful synthesis of a 2D amorphous nanoflower structure. Subsequently, the processing parameters were fine-tuned to enhance the specific surface area of ZIF-67 to 403 m²/g while preserving its 2D structural integrity. Ultimately, the as-prepared 2D ZIF-67 was incorporated into a hydrogenated styrene-butadiene block copolymer (SEBS) matrix to fabricate a mixed matrix membrane. Remarkably, at a filler loading of 20 wt%, the CH₄ permeability coefficient increased significantly from 11.7 barrer to 35.3 barrer, while the CH₄/N₂ selectivity was maintained at 3.21, indicating minimal interfacial defects and demonstrating the feasibility and effectiveness of the proposed methodology.

1. Introduction

Methane (CH₄) stands as the second most copious greenhouse gas, possessing a Global Warming Potential (GWP) that is 84 times higher than that of CO₂ within the 20-year time frame. Among the principal sources of methane emissions, the oil and gas industry discharged 72 million tonnes of CH₄ in 2020, as per the database of the International Energy Agency [1]. In the process of oil and gas production, CH₄ is frequently emitted in conjunction with N₂. The separation of CH₄ from N₂ poses a significant challenge due to their comparable physical characteristics and minimal disparity in molecular diameters [2,3].

Membrane-based separation has emerged as an innovative and environmentally friendly technology, characterized by low energy consumption, high efficiency, and operational simplicity. In recent years, it has gained significant attention as an advanced separation strategy, appealing to both academic researchers and industrial practitioners alike [4,5,6,7]. However, during the commercialization process, the inherent denseness of polymer materials typically limits the gas processing capacity of separation membranes. The incorporation of metal–organic frameworks (MOFs) into polymer matrices to fabricate mixed matrix membranes (MMMs) has been identified as an effective approach to address this limitation [8]. Nevertheless, conventional MOF synthesis methods predominantly produce isotropic crystals with micrometer-scale dimensions. These crystals are prone to spontaneous aggregation within the polymer matrix, leading to interfacial defects that severely compromise the separation performance of the membranes.

To enhance interfacial compatibility, Gascon et al. [9] synthesized a 2D nanosheet CuBDC MOF (ns-CuBDC) with an aspect ratio exceeding 20, which was applied to CO₂/CH₄ separation. With the ns-CuBDC loading increased to 8.2 wt%, the CO₂/CH₄ selectivity improved from 59.8 for the pristine PI membrane to 78.7. However, a significant decline in the permeability of both CO₂ and CH₄ was observed, primarily attributed to the extremely low BET surface area of ns-CuBDC, which was only 53 m²/g. Zeolitic imidazolate frameworks (ZIFs), particularly exemplified by ZIF-67, have emerged as promising MOFs owing to their energy-efficient synthesis protocols, demonstrating exceptional separation capabilities for challenging gas pairs including He/CH₄ [10], H₂/CH₄ [11], CO₂/CH₄, and CO₂/N₂ [12]. Kim et al. [13] synthesized 2D ZIF-L and incorporated it into PI matrices to fabricate MMMs for H₂/CH₄ separation. However, due to the relatively low surface area of ZIF-L (161 m²/g), even at a loading level of 20 wt%, the H₂ permeability coefficient exhibited only a marginal increase from 219.7 barrer to 260.4 barrer. In recent years, extensive research has been devoted to 2D MOF-based MMMs to address these limitations and explore their potential in various separation applications [14,15,16,17]. Nonetheless, achieving filler architectures with adequate thinness and porosity remains a substantial and unresolved challenge.

Liu et al. [18] employed direct time-resolved in situ transmission electron microscopy (TEM) to investigate the nucleation and initial growth stages of ZIF-8, revealing the presence of a lamelliform amorphous cluster structure prior to crystallization. However, the amorphous clusters transformed into granular structures within 45 s, making it challenging to capture this transient state using conventional synthesis techniques. To address this challenge, the present study first employs solvent selection based on first-principles calculations, combined with a “diffusion-mediated” strategy, to significantly suppress the crystallization process of ZIF-67. Subsequently, process parameters are optimized to enhance the specific surface area of 2D ZIF-67 to 403 m²/g. Finally, SEBS MMMs with a 2D ZIF-67 loading of 20 wt% are fabricated, resulting in a remarkable increase in the CH₄ permeability coefficient from 11.7 barrer to 35.3 barrer, while maintaining the CH₄/N₂ selectivity at 3.21.

2. Materials and Methods

2.1. Computational Details

The crystal structure of ZIF-67 was obtained from the Cambridge Crystallographic Data Centre (CCDC). DFT simulations were executed using Gaussian 16W [19]. For the geometry optimizations, all systems studied were optimized at the PBE0-D3(BJ) [20]/6-311G(d) [21] level, and the IEFPCM model [22] was employed to take into account the interactions between the solvent and the solute. All single-point energies were calculated using the DSD-PBEP86-D3(BJ) function [23] and def2-TZVPP basis set [24], and the SMD solvent model [25] was employed. Multiwfn3.7 [26] was used to analyze the electrostatic potential (ESP) and averaged local ionization energy (ALIE) information. All figures were plotted using VMD1.9.3 [27].

The entire process of a molecular reaction can be systematically categorized into two distinct stages. In the initial stage, the reactants gradually approach one another from a considerable distance. During this period, the electronic distribution within the molecules undergoes negligible change, and the long-range electrostatic interaction plays a crucial and decisive role. Subsequently, when the van der Waals surfaces of two molecules are on the verge of coming into contact, they continue to draw even closer. The chemical reaction is ultimately accomplished through the complex process of electronic structure rearrangement, with bond formation and bond cleavage being prominent features.

To accurately assess the two reaction processes of ZIF-67 molecules in various solvent environments, ESP and ALIE analysis are meticulously carried out, respectively. Here, ESP precisely describes the interaction energy between a unit positive charge located at a specific point r and the existing system. Regions with a relatively more positive (or more negative) electrostatic potential are generally regarded as more likely to attract the attack of nucleophilic (or electrophilic) reagents, thus effectively initiating a reaction. ALIE, on the other hand, refers to the average energy required for a molecule to lose an electron at a particular position. Notably, the lower this value is, the weaker the binding of electrons at this location becomes, and consequently, the more likely these electrons are to participate in electrophilic reactions. Both ESP and ALIE can be accurately expressed by Equations (1) and (2), respectively.

$$V_{tot}(\mathit{r})=V_{nuc}(\mathit{r})+V_{ele}(\mathit{r})={\sum }_A{\displaystyle \frac{Z_A}{\left|\mathit{r}-{\mathit{R}}_A\right|}}-\int {\displaystyle \frac{\rho ({\mathit{r}}^{\prime })}{\left|\mathit{r}-{\mathit{r}}^{\prime }\right|}}d{\mathit{r}}^{\prime }$$

(1)

where Z_A is the charge on nucleus A, located at R_A, and ρ(r) is the electron density.

$$\overline{I}(\mathit{r})={\displaystyle \frac{\sum _{i\in OCC}{\rho }_i(\mathit{r})\left|{\epsilon }_i\right|}{\rho (\mathit{r})}}$$

(2)

where ρ represents the total density, ρ_i represents the electron density corresponding to the i-th occupied molecular orbital and represents the energy of the i-th molecular orbital, $\overline{I}(\mathit{r})$ reflects the ionization energy of electrons at a local position.

2.2. Materials

The chemical and polymer were utilized in the study, including Cobalt(II) nitrate hexahydrate [Co(NO₃)₂·6H₂O, 99.99% purity], 2-Methylimidazole [2-MeIm, 98% purity], N,N-Dimethylformamide [DMF, 99.5% purity], SEBS [Kraton G1652, styrene/rubber: 30/70], Tetrabutylammonium bromide [TBAB, 99% purity], Toluene [99.5% purity], Ethanol [99.7% purity].

2.3. Synthesis of 2D Nanoflower ZIF-67 (ZIF-67_nf)

The preparation procedure for the 2D nanoflower-like ZIF-67 was adapted from the synthesis of CuBDC nanosheets by Gascon et al. [6]. Initially, m₁ g of Co(NO₃)₂·6H₂O was completely dissolved in 15 mL of solvent to prepare Solution A. Subsequently, 1 g of 2-MeIm was thoroughly dissolved in 15 mL of solvent to obtain Solution B. In parallel, m₂ g of TBAB was fully dissolved in V₁ mL of solvent to prepare Solution C.

Following this, Solution A was transferred into a beaker. Solution C was then slowly added dropwise to Solution A, followed by the gradual dropwise addition of Solution B onto Solution C, thereby establishing a vertically layered solution system. The system was allowed to stand at room temperature (RT) for t₁ hours, during which Co(NO₃)₂·6H₂O and 2-MeIm slowly diffused into Solution B under the influence of a concentration gradient, resulting in the formation of ZIF-67. The resulting purple products were collected via centrifugation and washed with ethanol. The diffusion-mediated layered solution systems at different reaction durations are illustrated in Figure 1.

2.4. Preparation of MMMs

ZIF-67 was ultrasonically dispersed in 5 g of toluene to form Solution D. Meanwhile, SEBS was completely dissolved in 6.33 g of toluene under vigorous stirring to generate Solution E. Subsequently, Solutions D and E were combined and stirred continuously for 0.5 h, resulting in Solution F. After ultrasonic dispersion, Solution F was placed in a vacuum oven to eliminate air bubbles. The MMMs were then cast onto a glass plate and covered for 48 h to minimize solvent evaporation. Finally, the MMMs were dried under vacuum to remove residual solvent. In this process, the total mass of ZIF-67 and SEBS was fixed at 2 g, with ZIF-67 loadings set at 5 wt%, 10 wt%, 15 wt%, and 20 wt%, respectively. For comparative purposes, MMMs incorporating a three-dimensional (3D) filler morphology with identical loading levels were also prepared.

2.5. Characterization

A Panalytical X’Pert PRO MPD Diffractometer (Cu Kα, 40 kV, 40 mA) was used to obtain X-ray diffraction (XRD) patterns, ranging from 5° to 45° with a step size of 0.016° [28]. The ZIF-67 nanoflower morphologies were examined using a Hitachi S4800 scanning electron microscope (SEM). To investigate the surface area and pore size distribution, Brunauer–Emmett–Teller (BET) and Barrett-Joyner-Halenda (BJH) were tested on Micromeritics ASAP 2460. In order to evaluate the separation performance of the prepared MMMs in our work, the gas separation capabilities were assessed using pure CH₄ and N₂ permeation experiments. Measurements were conducted at 25 °C under an applied pressure difference of 100 kPa, employing a differential pressure gas permeameter (Labthink G2/110) with a standardized membrane area of 1 cm². Real-time analysis of permeate flux was performed via online gas chromatography, with data acquisition initiated upon establishment of steady-state permeation conditions.

3. Results and Discussion

3.1. Aggregation and Reactivity

ESP and ALIE analyses were conducted to evaluate the aggregation and reactivity characteristics of the ZIF-67 unit cell. Figure 2 depicts the distributions of ESP and ALIE on the surface of ZIF-67 molecules under vacuum and in various solvent environments. Figure 3 provides a quantitative analysis of the surface areas of ZIF-67 molecules corresponding to different ESP and ALIE distribution intervals. As shown in Figure 3, in the ethanol environment, both the negative ESP and the positive ALIE occupy the smallest molecular surface areas, indicating that ZIF-67 is least prone to aggregation and reaction in this environment. Therefore, ethanol is an optimal reaction solvent for ZIF-67, as it minimizes the crystallization rate and facilitates the formation of 2D aggregate structures.

3.2. SEM, XRD and BET

Figure 4, Figure 5 and Figure 6 systematically illustrate the morphological evolution, crystallinity, and gas adsorption characteristics of ZIF-67 synthesized under varying reaction durations. As depicted in Figure 4a, ZIF-67 synthesized with a reaction duration of 0.25 h (denoted as ZIF-67_0.25h) exhibits a well-defined 2D nanoflower morphology with ultrathin petal-like structures. However, prolonging the reaction duration to 0.5 h and 1.0 h triggers a morphological transition to 3D granular aggregates (Figure 4b,c), consistent with the nucleation-crystallization mechanism proposed for ZIF-8 by Liu et al. [12]. This structural evolution is corroborated by XRD analysis (Figure 5), where ZIF-67_0.25h displays no discernible diffraction peaks, confirming its amorphous nature. In contrast, ZIF-67_1.0h exhibits sharp Bragg reflections matching the simulated ZIF-67 pattern, indicative of high crystallinity.

The CO₂ adsorption–desorption isotherms (Figure 6) further quantify the impact of structural evolution on porosity. ZIF-67_0.25h demonstrates negligible gas uptake due to its non-porous 2D architecture, yielding a BET surface area of only 18.2 m²/g. Conversely, ZIF-67_0.5h and ZIF-67_1.0h exhibit Type I isotherms characteristic of microporous materials, with BET surface areas increasing dramatically to 434.1 m²/g and 1011.5 m²/g, respectively. This enhancement correlates with the development of 3D granular structures, which introduce hierarchical porosity and accessible adsorption sites. Notably, the steep adsorption at low relative pressures (P/P₀ < 0.1) for ZIF-67_1.0h underscores its dominant microporosity, while the hysteresis loop at higher pressures (0.4 < P/P₀ < 0.9) suggests the coexistence of mesopores formed during particle aggregation.

Collectively, these findings highlight the critical role of reaction duration in governing ZIF-67’s structural and textural properties. The amorphous 2D nanoflower morphology (ZIF-67_0.25h) offers minimal porosity but serves as a precursor for subsequent crystallite growth. By contrast, extended reaction durations promote crystallization and 3D granular formation, significantly enhancing surface area and pore volume. This structural-property relationship underscores the necessity of precise kinetic control in tailoring ZIF-67 for specific applications, such as gas separation membranes, where balanced crystallinity and porosity are essential for optimizing performance.

Figure 7, Figure 8, Figure 9 and Figure 10 present SEM images of ZIF-67 synthesized under varying parameters, including molar ratios of precursors (m₁), transition layer volumes (V₁), ethanol/water solvent ratios, and template agent (TBAB) loadings. As shown in Figure 7a, ZIF-67 synthesized with m₁ = 0.66 exhibited a well-defined 2D nanoflower morphology. However, reducing the precursor ratio (m₁ = 0.44 and 0.22) led to irregular granular structures (Figure 7b,c), highlighting the critical role of stoichiometric control in morphology development. Similarly, Figure 8, Figure 9 and Figure 10 demonstrate that optimizing the transition layer volume (V₁ = 15), solvent composition (V_C2H5OH:V_H2O = 50:50), and TBAB loading (m₂ = 0.5 g) significantly enhanced structural uniformity and porosity.

The BET surface area and pore structure of the optimized ZIF-67_nf were evaluated through CO₂ adsorption–desorption isotherms (Figure 11) and pore size distribution analysis (Figure 12). The isotherm of ZIF-67_nf displays a Type IV hysteresis loop, indicative of mesoporous materials, with a steep adsorption slope at low relative pressures (P/P₀ < 0.1), confirming the presence of micropores. The BET surface area of ZIF-67_nf reached 403 m²/g, a substantial improvement compared to the non-porous 2D ZIF-67_0.25h (18.2 m²/g, Figure 6). The pore volume increased to 0.197 cm³/g, with a narrow pore size distribution centered at ~0.9 nm (Figure 12), suggesting a homogeneous mesoporous structure. These structural enhancements are attributed to the synergistic effects of solvent optimization, templating agents, and diffusion-mediated synthesis, which suppressed rapid crystallization and promoted controlled growth of the 2D nanoflower architecture.

3.3. MMMs Performance

To evaluate the enhancement effect of ZIF-67_nf on the membrane separation performance, MMMs with different loadings were prepared. Meanwhile, for the purpose of comparison, MMMs with the same loadings using 3D ZIF-67 as the filler were also prepared. The CH₄/N₂ separation performance is shown in Figure 13. When the ZIF-67_nf loading in the MMMs reaches 20 wt%, the CH₄ permeability coefficient increases steadily from 11.7 barrer to 35.3 barrer, while maintaining a CH₄/N₂ selectivity above 3. In contrast, MMMs incorporating 3D ZIF-67 particles show a more pronounced increase in CH₄ permeability; however, this is accompanied by a sharp decline in CH₄/N₂ selectivity, highlighting the presence of significant interfacial defects in the membrane structure. This indicates that 2D-structured fillers possess significant potential advantages in enhancing interfacial compatibility within MMMs.

Table 1. Comparative of CH₄/N₂ separation performance of MMMs.

MMMs	Solvent	Temperature	Permeability (Barrer)		CH4/N2 Selectivity
			CH4	N2
Matrimid®/MOF-5 [29]	chloroform	RT	0.45	0.52	0.87
Matrimid® 5218/MOP-18 [30]	chloroform		0.95	0.60	1.58
SBS/[Ni3(HCOO)6] [31]	Tetrahydrofuran		77.7	26.7	2.9
SBS/Ni-MOF-74 [32]	Tetrahydrofuran		74.7	32.2	2.3
SEBS/ZIF-8 [33]	Tetrahydrofuran		84.9	41.2	2.1
Pebax@1657/glycerol/PEG200/NiFe2O4 [34]	1-butanol	80 °C	4.07	0.98	4.13
Pebax1657/Ag/[BMIM][BF4] [35]	1-butanol	110 °C	2.95	0.96	3.07
SEBS/ZIF-67nf (this work)	toluene	RT	35.3	11.0	3.21

and Figure 14 compare the CH₄/N₂ separation performance of the polymeric membranes fabricated in this work with those reported in other studies, demonstrating promising potential when considering both permeability and selectivity.

4. Conclusions

In summary, to obtain ZIF-67 with a high aspect ratio, we successfully reduced the crystallization rate of ZIF-67 by more than one order of magnitude using a solvent screening strategy based on first-principles calculations combined with a diffusion-mediated process, resulting in the formation of a 2D amorphous nanoflower structure. Furthermore, by introducing a templating agent into the reaction system, the formation of pore structures was promoted to some extent, increasing the BET surface area from 18.2 m²/g to 403 m²/g while preserving the 2D nanoflower morphology. Finally, MMMs with a 20 wt% loading of this 2D nanoflower ZIF-67 were fabricated. Due to the synergistic effect of high filler loading and the relatively high specific surface area, the CH₄ permeability coefficient increased from 11.7 barrer to 35.3 barrer, representing a 201% improvement, while maintaining a CH₄/N₂ selectivity above 3, demonstrating excellent interfacial compatibility. Although existing 2D MOF-based MMMs achieve high filler loadings, their gas separation performance remains critically limited by the ultra low specific surface area of the fillers. Future research must prioritize developing 2D MOFs with engineered pore architectures to enable breakthrough enhancements in membrane permeability/selectivity combinations beyond current Robeson upper bounds.