One‐Step Separation of Propylene From Ternary C₃ Mixtures in Amino Acid Functionalized Metal–Organic Frameworks

ABSTRACT

The low‐cost C₃H₈/C₃H₆/C₃H₄ separation for high‐grade C₃H₆ is of paramount importance but extremely difficult in the chemical industry. Here we report an economic strategy of functionalizing Zr‐based metal–organic frameworks (MOFs) with amino acids for one‐step C₃H₆ purification from ternary C₃ mixtures. The resulting UiO‐67‐AA exhibits exceptionally high C₃H₈ and C₃H₄ uptakes and benchmark selectivities for C₃H₈/C₃H₆ and C₃H₄/C₃H₆ separations, rivaling reported ternary C₃ separating MOFs. Breakthrough experiments confirm its excellent performance in producing high‐purity C₃H₆ in one step. Crucially, the estimated over 2250 times the cost reduction among reported C₃H₈/C₃H₄‐selective MOFs makes UiO‐67‐AA an ideal candidate for industrial C₃H₆ purification. In situ infrared spectroscopy, in situ neutron powder diffraction, and theoretical calculations indicate that amino acids offer selective binding to dually strengthen both C₃H₈ and C₃H₄ affinity over that of C₃H₆, with the amino group dictating the supramolecular interactions. Systematic evaluation across UiO‐67 with diverse nitrogen moieties establishes a monotonic relationship between the separation performance and the specific nitrogen electronegativity, providing an element‑specific and quantitative electronegativity‐driven design rule for next‐generation MOF adsorbents for industrial olefin purification.

Amino acid functionalized UiO‐67 enables highly efficient one‐step C₃H₆ separation from C₃H₈/C₃H₆/C₃H₄ mixtures, reducing cost by over 2250 times. A monotonic relationship between the separation performance and the specific nitrogen electronegativity is established, offering a practical adsorbent design rule for industrial olefin purification.

1. Introduction

Propylene (C₃H₆) is a fundamental feedstock in the chemical industry and is widely used for the production of many petrochemicals, polymers, and speciality chemicals [1, 2]. At present, C₃H₆ is mainly produced via hydrocarbon cracking or propane dehydrogenation, and trace impurities such as propane (C₃H₈) and propyne (C₃H₄) are inevitably introduced [3, 4]. These impurities pose severe problems for subsequent processes; thus, there is a great need for their deep removal [5]. Traditionally, C₃H₈ is removed by cryogenic distillation [6], and C₃H₄ is selectively hydrogenated using precious‐metal catalysts at high temperatures and pressures [7]. These methods are of extremely high cost. The low‐cost C₃H₈/C₃H₆/C₃H₄ separation for high‐grade C₃H₆ has been highlighted as one of the most important industrial separation tasks [8]. Adsorptive separation using metal–organic frameworks (MOFs) is a cost‐friendly alternative [9, 10, 11, 12]. Many MOFs have shown excellent separation of C₃H₆ from binary C₃H₈/C₃H₆ and C₃H₄/C₃H₆ mixtures [13, 14, 15, 16, 17]. However, only a handful of them enable simultaneous C₃H₈ and C₃H₄ removal from C₃H₈/C₃H₆/C₃H₄ [18, 19, 20], which would facilitate the direct production of high‐purity C₃H₆ at the outlet via a single adsorption step in one column, yielding substantial operational cost savings. Furthermore, these ternary C₃ separating MOFs are prohibitively costly to synthesise, hindering their industrial application.

To address this challenge, we turned to Zr‐based MOFs (Zr‐MOFs), which are among the most promising MOF candidates for large‐scale industrial applications [21, 22]. Although they have been investigated for C₃H₈/C₃H₆ separation [23, 24, 25], performance enhancement often relies on expensive functionalized ligands, which contradicts the goal of cost‐effective separation. Zr‐MOFs are well known for their defect‐rich structures [26, 27, 28, 29]. The effect of defects on their CO₂, NH₃, and other adsorption behaviors has been demonstrated previously [30, 31]. These defect sites not only act as adsorption sites but also offer opportunities to easily introduce functional groups instead of ligand modification [32, 33, 34, 35]. Therefore, defect engineering provides us with a more practical way to functionalize Zr‐MOFs for industrial separation of C₃H₈/C₃H₆/C₃H₄. Inspired by previous work on nitrogen‐functionalized MOFs for separating ternary C₂ mixtures (C₂H₆/C₂H₄/C₂H₂) [36], we hypothesized that introducing widely available, nitrogen‐containing amino acids into the defect sites of Zr‐MOFs could provide a cost‐effective platform for one‐step separation of C₃H₆ from the C₃H₈/C₃H₆/C₃H₄ mixture. Through a comprehensive screening of Zr‐MOFs, we chose UiO‐67 as the fundamental framework backbone to verify our hypothesis because of its non‐polar pore surfaces and large pore sizes [37]. We demonstrate that incorporating amino acids into stable UiO‐67 via one‐pot synthesis provides selective binding to dually enhance C₃H₈ and C₃H₄ affinity over C₃H₆. These amino acid functionalized UiO‐67 (UiO‐67‐AA) display not only record‐high C₃H₈ and C₃H₄ uptakes (6–10 and 7–11 mmol·g⁻¹) but also benchmark C₃H₈/C₃H₆ (50/50) and C₃H₄/C₃H₆ (50/50) selectivities (1–1.5 and 2–17) at ambient conditions. The performance of UiO‐67‐AA rivals that of reported ternary C₃ separating MOFs. One‐step removal of C₃H₈ and C₃H₄ from C₃H₈/C₃H₆/C₃H₄ mixtures to produce polymer‐grade C₃H₆ was evidenced by breakthrough experiments on UiO‐67‐AA at 298 K and 1 bar. Crucially, we estimate a cost reduction of over 2250 times compared to the reported C₃H₈/C₃H₆ and/or C₃H₄/C₃H₆ separating MOFs, highlighting the outstanding industrial potential of UiO‐67‐AA. Our economical and effective strategy paves the way for low‐cost industrial C₃H₆ purification.

With the assistance of in situ infrared (IR) spectroscopy, in situ neutron powder diffraction (NPD), and density functional theory (DFT) calculations, we revealed that the amino acids in UiO‐67‐AA contribute to the separation through a synergistic mechanism and offer selective binding to dually strengthen both C₃H₈ and C₃H₄ molecular/network affinity over that of C₃H₆, with the amino group dictating the supramolecular interactions. The electronegative nitrogen atoms of amino groups enhance supramolecular interactions between the porous surface and C−H bonds in C₃H₈ and C₃H₄ over those in C₃H₆, while the electropositive hydrogen atoms intensify supramolecular interactions with C≡C bonds in C₃H₄; besides, C₃H₄ forms molecular networks over amino groups, making its packing density significantly increase in UiO‐67‐AA. Furthermore, through comparative analysis of UiO‐67 with diverse nitrogen moieties exhibiting varying electronic properties, we successfully established a monotonic relationship between the separation performance and the specific nitrogen electronegativity. It is suggested that the strategic incorporation of high‐electronegativity functionalities directly enhances separation efficiency. This provides an element‑specific and quantitative electronegativity‐driven design rule for next‐generation MOF adsorbents for industrial olefin purification.

2. Results and Discussion

2.1. Synthesis and Characterization

In the solvothermal synthesis of Zr‐MOFs, acidic modulators are often used to control product crystallinity [38]. During the reaction, modulators compete with carboxylic acid ligands for coordination with Zr ions. This prevents the coordination reaction between Zr ions and ligands from being too drastic, thus ensuring the integrity of the crystals. It is worth noting that modulators are often not completely replaced by ligands during crystal growth and usually remain in the framework at defect sites. Therefore, it is possible to introduce specific functional groups using modulators in one‐pot synthesis [39]. Inspired by this functionalization strategy, we employed widely available amino acids as modulators in the synthesis of UiO‐67 and expected them to introduce nitrogen functional groups (–NH₂) to the framework, in order to enhance its capability to separate the C₃H₈/C₃H₆/C₃H₄ mixture. UiO‐67 modulated by 16 types of amino acids (UiO‐67‐AA, Figure 1) were synthesized based on a previously reported procedure [40].

FIGURE 1.

(a) Structure of amino acid functionalized UiO‐67 (UiO‐67‐AA). Zr: blue, C: grey, O: red, N: green. H is omitted for clarity. (b) Synthetic scheme for introducing amino acids to UiO‐67 (AA: amino acids; bpdc: 4,4’‐biphenyldicarboxylic acid; SBU: secondary building unit). (c) Structural formulae of amino acids used and their abbreviations.

Powder X‐ray diffraction (PXRD) data demonstrate that UiO‐67‐AA samples have the same topology as UiO‐67 and have relatively high purity and crystallinity (Figure S1). The N₂ sorption isotherms at 77 K were measured to evaluate the permanent porosity of activated UiO‐67‐AA (Figure S2). They exhibit reversible type I sorption behavior without hysteresis, which confirms their microporous structures. The Brunauer‐Emmett‐Teller (BET) surface area was calculated to be about 2000 m²·g⁻¹ on average (Table S1), in line with literature‐reported values [40]. The specific surface area of UiO‐67‐AA is slightly reduced to varying degrees compared to that of primitive UiO‐67. This is perhaps because introducing amino acids reduces the connectivity of the framework. ¹H nuclear magnetic resonance (NMR) measurements were carried out on all digested UiO‐67‐AA samples to confirm the presence of amino acids in these MOFs (Figure S3–S19). The thermal stability of UiO‐67‐AA was tested by thermogravimetric analysis (TGA) (Figure S20). In general, our samples are stable up to 800 K. TGA also assisted us in calculating the chemical formulae of UiO‐67‐AA (Table S2). It is estimated that amino acids replace 1/12–1/3 of the ligands. The percentage of amino acids is not high, but as will be revealed later, they make a big difference in the efficient separation of the C₃H₈/C₃H₆/C₃H₄ mixture.

2.2. Adsorption and Separation Performance

Single‐component adsorption isotherms of C₃H₈, C₃H₆, and C₃H₄ for UiO‐67‐AA were collected at 298 K. For the three gases, UiO‐67‐AA exhibits type I isotherms (Figure S21–S37). In general, the C₃H₈ adsorption capacity of UiO‐67‐AA is 6–10 mmol·g⁻¹ (1 bar), and their C₃H₄ adsorption capacity is 7–11 mmol·g⁻¹ (1 bar). As a prominent example, the L‐threonine functionalized UiO‐67 (UiO‐67‐L‐Thr) can uptake 8.5 mmol·g⁻¹ of C₃H₈ or 11 mmol·g⁻¹ of C₃H₄ at 1 bar (Figure 2a). The C₃H₈ and C₃H₄ adsorption capacity of UiO‐67‐AA generally exceeds that of reported C₃H₈/C₃H₄‐selective MOFs. The C₃H₈ and C₃H₄ adsorption capacities of UiO‐67‐AA are both consistently higher than their C₃H₆ adsorption capacity over the whole pressure range. It indicates that there is a good affinity between the framework and C₃H₈/C₃H₄ molecules, and UiO‐67‐AA may be a potential adsorbent to simultaneously capture both C₃H₈ and C₃H₄ from the C₃H₈/C₃H₆/C₃H₄ mixture.

FIGURE 2.

(a) Single‐component adsorption isotherms of C₃H₈/C₃H₆ and C₃H₄/C₃H₆ for UiO‐67‐L‐Thr at 298 K. (b) IAST selectivity curves of UiO‐67‐AA for C₃H₈/C₃H₆ (50/50) mixtures at 298 K. (c) Comparison of C₃H₈ uptakes and C₃H₈/C₃H₆ (50/50) selectivities at 1 bar. UiO‐67‐AA: pink, UiO‐67: yellow, reported ternary C₃ separating MOFs: blue, reported binary C₃ separating MOFs: black. 1: UiO‐67‐L‐Thr, 2: UiO‐67‐L‐Phe, 3: UiO‐67‐L‐Val, 4: UiO‐67‐L‐Ile, 5: UiO‐67‐L‐DL‐Ala, 6: UiO‐67‐L‐Ala, 7: UiO‐67‐Gly, 8: UiO‐67‐L‐Pro, 9: UiO‐67‐L‐Glu, 10: UiO‐67‐L‐Tyr, 11: UiO‐67‐L‐Leu, 12: UiO‐67‐L‐Trp, 13: UiO‐67‐L‐Cys, 14: UiO‐67‐L‐Gln, 15: UiO‐67‐L‐Met, 16: UiO‐67‐L‐Ser. (d) IAST selectivity curves of UiO‐67‐AA for C₃H₄/C₃H₆ (50/50) mixtures at 298 K. (e) Comparison of C₃H₄ uptakes and C₃H₄/C₃H₆ (50/50) selectivities at 1 bar. (f) IAST selectivity curves of four N‐functionalized UiO‐67 MOFs for C₃H₈/C₃H₆ (50/50) mixtures at 298 K. (g) IAST selectivity curves of four N‐functionalized UiO‐67 MOFs for C₃H₄/C₃H₆ (50/50) mixtures at 298 K. (h) Comparison of C₃H₈ and C₃H₄ uptakes, C₃H₈/C₃H₆ (50/50) and C₃H₄/C₃H₆ (50/50) selectivities among four N‐functionalized UiO‐67 MOFs at 1 bar. (i) The Q _st of C₃H₈, C₃H₆, and C₃H₄ over four N‐functionalized UiO‐67 MOFs.

To evaluate the potential of UiO‐67‐AA for separating C₃H₆ from ternary C₃ mixtures, we calculated the adsorption selectivity of UiO‐67‐AA for C₃H₈/C₃H₆ and C₃H₄/C₃H₆ mixtures, based on the ideal adsorbed solution theory (IAST) (Figure S38) [41]. At 298 K and 1 bar, the calculated IAST selectivities of UiO‐67‐AA for the C₃H₈/C₃H₆ (50/50) mixture are 1–1.5 (Figure 2b), which is comparable to many reported C₃H₈‐selective MOFs, such as NUM‐7a (1.7) [42], WOFOUR‐1‐Ni (1.6) [43], Ni(ADC)(TED)_0.5 (1.6) [44], JNU‐9‐CH₃ (1.5) [19], BUT‐10 (1.4) [25], HIAM‐402 (1.4) [8], CPM‐734c (1.4) [45], NKU‐301 (1.4) [18], ZIF‐8 (1.3) [46], and MOF‐801 (1.3) [47] (Figure 2c). For the C₃H₄/C₃H₆ (50/50) mixture, the IAST selectivities of UiO‐67‐AA were calculated to be 2–17 at 298 K and 1 bar (Figure 2d), comparable to several recently reported C₃H₄‐selective MOFs like ZNU‐2 (16.2) [13], NTU‐100‐NH₂ (11.3) [17], BUT‐309 (5.2) [48], Ni‐DMOF‐DM (3.0) [20], and JNU‐9‐CH₃ (2.1) [19] (Figure 2e). It is noted that the L‐proline functionalized UiO‐67 (UiO‐67‐L‐Pro) has a high C₃H₄/C₃H₆ selectivity of 16.9 and a decent C₃H₄ adsorption capacity of 9.7 mmol·g⁻¹, both of which make UiO‐67‐L‐Pro among the top‐performing C₃H₄‐selective MOFs. More importantly, the C₃H₈/C₃H₆ (50/50) and C₃H₄/C₃H₆ (50/50) selectivities of UiO‐67‐AA rival those of reported ternary C₃ separating MOFs such as Ni‐DMOF‐DM (2.7 and 3.0) [20], JNU‐9‐CH₃ (1.5 and 2.1) [19], and NKU‐301 (1.4 and 1.9) [18] (Figure 2c,e). Therefore, UiO‐67‐AA is an excellent candidate for the one‐step purification of C₃H₆ from the ternary mixture of C₃.

Comparing the low‐pressure regions of adsorption isotherms for UiO‐67‐L‐Thr and UiO‐67 suggests that the incorporation of amino acids enhances guest–host interactions to varying degrees for different gases, which underpins the observed selectivity (Figure S39–S41). To demonstrate this conclusion, we calculated the heat of adsorption (Q _st) using the Clausius‐Clapeyron equation based on the adsorption isotherms at 273 and 298 K. For UiO‐67‐L‐Thr (Figure S42–S43 and Table S3), the Q _st of C₃H₈ and C₃H₄ were estimated to be 35.19 and 34.48 kJ·mol⁻¹ at near‐zero adsorption, which are higher than the Q _st of C₃H₆ (27.96 kJ·mol⁻¹). This provides strong evidence that the material has the strongest binding to C₃H₈ and C₃H₄ and the weakest binding to C₃H₆. In addition, the Q _st of C₃H₈ and C₃H₄ are in the low‐to‐medium heat of adsorption range (20–60 kJ·mol⁻¹), indicating that the material can easily desorb C₃H₈ and C₃H₄ to enable regeneration. We also calculated the Q _st for UiO‐67 (Figure S44–S45 and Table S3). It is noted that UiO‐67 has a similar Q _st of C₃H₆ (27.22 kJ·mol⁻¹) to UiO‐67‐L‐Thr (27.96 kJ·mol⁻¹). But UiO‐67 has smaller Q _st of C₃H₈ (34.47 kJ·mol⁻¹) and C₃H₄ (31.54 kJ·mol⁻¹) than UiO‐67‐L‐Thr (35.19 and 34.48 kJ·mol⁻¹). This verifies that UiO‐67‐AA mainly strengthens the binding to C₃H₈ and C₃H₄.

The synthesis of UiO‐67‐AA represents a specialized strategy for incorporating nitrogen functional groups into the UiO‐67 framework. To further investigate the influence of nitrogen functionalities on the separation of C₃H₈/C₃H₆/C₃H₄ mixtures, we chose a few UiO‐67 comparison samples whose modulators or ligands have various nitrogen functional groups. To be specific, they are UiO‐67‐Aba whose modulator is 3‐aminobenzoic acid, UiO‐67‐Py whose ligand is 2,2’‐bipyridine‐5,5’‐dicarboxylic acid, and UiO‐67‐(NH₂)₂ whose ligand is 4,4’‐diamino‐[1,1’‐biphenyl]‐2,2’‐dicarboxylic acid. These materials were synthesized following procedures reported in the literature [36]. They were subjected to basic characterization, and the results can be found in Figure S46. We measured their single‐component adsorption isotherms of C₃H₈, C₃H₆, and C₃H₄ at 298 K (Figure S47–S49). In general, comparison samples exhibit noticeably lower uptakes than UiO‐67‐L‐Thr under identical conditions. More significantly, their adsorption preferences deviate from the desired behavior. Both UiO‐67‐Aba and UiO‐67‐Py prefer C₃H₆ adsorption at relative pressures above P/P₀ = 0.6, while UiO‐67‐(NH₂)₂ shows a consistent adsorption preference for C₃H₆ across the entire pressure range. These results suggest a limitation in their ability to selectively capture C₃H₈ and C₃H₄ over C₃H₆. To quantitatively evaluate separation performance, we calculated the IAST selectivities of the above comparison samples for C₃H₈/C₃H₆ and C₃H₄/C₃H₆ mixtures. The IAST selectivity curves (Figure 2f,g) confirm the inferior separation potential of these materials compared to UiO‐67‐L‐Thr. Figure 2h once again indicates that none of the comparison samples can achieve the same C₃H₈/C₃H₆ and C₃H₄/C₃H₆ separation efficiency as UiO‐67‐L‐Thr. Adsorption heat analysis in Figure 2i and Figure S50–S55 provides further insight. The comparison materials generally have smaller Q _st of C₃H₈ and C₃H₄ than UiO‐67‐L‐Thr, indicating their weaker binding affinity. Therefore, not all nitrogen functional groups can enhance the separation efficiency of UiO‐67. The mere presence of such groups does not lead to the performance achieved by the amino acids.

It is generally believed that pore partition could also affect the adsorption behavior of MOFs for light hydrocarbons. Our TGA results have pointed out that amino acids only replace about 1/8 of the ligands in UiO‐67‐L‐Thr (Table S2). Therefore, the framework is not much different from that of UiO‐67, meaning that the contribution from pore partition may be minute to the adsorption behavior of UiO‐67‐AA. The same holds for UiO‐67‐Aba, in which 3‐aminobenzoic acid only replaces about 1/6 of the ligands (Table S4), exerting a minor change to the framework. Both UiO‐67‐Py and UiO‐67‐(NH₂)₂ were synthesized through complete ligand replacement. The former does not exhibit pore partition in comparison with UiO‐67, as its pyridine rings do not induce structural deviation. UiO‐67‐(NH₂)₂ may possess a pore partition effect. Its inferior separation performance compared to UiO‐67 suggests that pore partition does not contribute positively to C₃H₆ purification efficiency.

2.3. Mechanistic Study

To understand the guest–host interactions in UiO‐67‐AA, we first employed in situ IR spectroscopy to follow C₃H₈, C₃H₄, and C₃H₆ adsorption in UiO‐67‐L‐Thr and UiO‐67 (Figure 3). For C₃H₈ (Figure 3a,b), the ν(N–H) bands at 3227 cm⁻¹ in UiO‐67‐L‐Thr exhibited a redshift upon gas exposure, indicating the weak bonding between –NH₂ and C₃H₈. The ν(C–H) bands at around 3000 cm⁻¹ in two samples were present before and after gas exposure, so this region contains signals from C–H stretching of both C₃H₈ and the MOF frameworks, making it unusable to explain guest–host interactions. For C₃H₄ (Figure 3c,d), the peaks at 3346–3321 cm⁻¹ appeared only after gas exposure, so they are assigned to the multiple ν(C–H) bands of C₃H₄. They exhibited a more rapid intensification and higher relative intensity in UiO‐67‐L‐Thr, suggesting stronger C₃H₄ binding sites [36]. The emergence of a C–H bending peak of C₃H₄ at 630 cm⁻¹ in UiO‐67‐L‐Thr further confirms the stronger binding to C₃H₄. More importantly, the intensified ν(C–N) bands near 1100 cm⁻¹ indicate that –NH₂ offers additional binding to C₃H₄ in UiO‐67‐L‐Thr. For C₃H₆ (Figure 3e,f), the bands at 3100–2800 cm⁻¹ (ν(C–H) of C₃H₆) and 1650 cm⁻¹ (ν(C = C) of C₃H₆) showed more rapid changes in UiO‐67‐L‐Thr, indicating stronger framework–C₃H₆ interactions. Furthermore, the appearance and variation of the δ(N–H) peak at 1607 cm⁻¹ in UiO‐67‐L‐Thr indicate that –NH₂ likely serves as an adsorption site. These results collectively reveal the critical role of –NH₂ groups of amino acids in enhancing C₃H₈, C₃H₄, and C₃H₆ adsorption over UiO‐67‐AA.

FIGURE 3.

In situ IR spectra of UiO‐67‐L‐Thr and UiO‐67 exposed to C₃H₈, C₃H₄, or C₃H₆ at room temperature. (a) UiO‐67‐L‐Thr exposed to C₃H₈. (b) UiO‐67 exposed to C₃H₈. (c) UiO‐67‐L‐Thr exposed to C₃H₄. (d) UiO‐67 exposed to C₃H₄. (e) UiO‐67‐L‐Thr exposed to C₃H₆. (f) UiO‐67 exposed to C₃H₆.

We then used in situ NPD to visualize the guest–host interactions by revealing the adsorption structure of C₃H₈, C₃H₄, and C₃H₆ in UiO‐67‐L‐Thr [49, 50, 51]. We first obtained the framework structure of UiO‐67‐L‐Thr and verified the presence and content of L‐threonine (Figure S56–S57 and Table S5). Then, we studied C₃H₈ adsorption (Figure S58–S59 and Table S6). Three adsorption sites were found (Figure S60). Two sites are near the benzene ring of the biphenyl dicarboxylate ligand, sites 1 and 3. They adsorb ∼68% of all C₃H₈ molecules. It is reasonable that most C₃H₈ are near the ligands because only a small portion of the ligands was substituted by L‐threonine in UiO‐67‐L‐Thr (∼13%). By analysing the adsorption structure, we noticed that the main interactions over sites 1 and 3 are the van der Waals forces between the hydrogen atoms of C–H in C₃H₈ and the π‐electron cloud of the phenyl group in the ligand (C–H···π interactions). Site 2 is where C₃H₈ interacts with the amino acid through C–H···N interactions (C(H)···N = 1.80 Å) (Figure 4a). Although the content of L‐threonine is low in UiO‐67‐L‐Thr (∼13% of all linkers), there is a significant amount of C₃H₈ adsorbed over site 2 (∼32% of all C₃H₈). This proves that the –NH₂ group of amino acids plays an important role in enhancing the adsorption of C₃H₈. Next, we studied C₃H₄ adsorption (Figure S61–S62 and Table S7). Seven adsorption sites were found (Figure S63). They were divided into three groups: sites near the benzene ring of the ligand, sites near L‐threonine, and molecular networks formed by mutual attraction between gas molecules. Two sites are near the benzene ring of the ligand, sites 3 and 5. They adsorb ∼60% of all C₃H₄ molecules. The main interactions over these sites are the C–H···π interactions between C₃H₄ and the ligand. Again, there is a significant amount of C₃H₄ adsorbed by the amino acid (∼30% of all C₃H₄). There are four adsorption sites: 1, 4, 6, and 7. Sites 1 and 7 are distributed near the –NH₂ group (Figure 4b,c, respectively), and electrostatic attractions were found between the hydrogen atoms of C–H in C₃H₄ and the nitrogen atom of –NH₂ in L‐threonine (C–H···N interactions, C(H)···N = 3.01 Å at site 1 and 3.08 Å at site 7). Site 4 is also near the –NH₂ group (Figure 4d). However, C₃H₄ molecules over site 4 form interactions through the carbon atoms of their C≡C bond with the hydrogen atoms of –NH₂ or through the antibonding orbital of their C≡C bond with the nitrogen atom of –NH₂ (C≡C···(H–)N interactions, C···(H)N = 3.74 Å). C₃H₄ molecules at site 6 (∼2% of all C₃H₄) are adsorbed by the –OH group in L‐threonine via C–H···O interactions (C(H)···O = 2.37 Å). C₃H₄ molecules over site 2 are close to neither the ligand nor the amino acid. They are stabilized by C₃H₄ molecules over site 1 through C–H···C≡C interactions (C(H)···C = 3.52 Å). We believe that C₃H₄ molecules at sites 1 and 2 form a molecular network, which leads to a significant increase in the stacking density of C₃H₄ in UiO‐67‐L‐Thr (by ∼10% of all C₃H₄). Our previous work has demonstrated the formation of NH₃ molecular networks in functionalized UiO‐67 [51]. Although the polarity of C₃H₄ is smaller than that of NH₃, they can still form dipole–dipole interactions or van der Waals forces, leading to molecular networks in a confined space. Finally, we investigated C₃H₆ adsorption (Figure S64–S65 and Table S8). Four adsorption sites were identified (Figure S66). Sites 2 and 3, located near the benzene ring of the ligand, account for the majority (∼81%) of adsorbed C₃H₆ molecules and are stabilized by C–H···π interactions. In contrast, sites 1 and 4 involve C–H···N interactions over the amino acid moiety (C(H)···N = 4.16 and 4.53 Å) (Figure 4e,f). Compared to the analogous sites in C₃H₈ and C₃H₄ adsorption, these two sites adsorb a smaller proportion of C₃H₆ (∼19%) and are situated further from the nitrogen atom of –NH₂, indicating weaker binding to C₃H₆. Overall, our NPD studies made it clear that amino acids contribute to the separation through a synergistic mechanism and offer selective binding to dually strengthen both C₃H₈ and C₃H₄ molecular/network affinity over that of C₃H₆, with the –NH₂ group dictating the supramolecular interactions.

FIGURE 4.

Selected adsorption structures of C₃H₈, C₃H₄, and C₃H₆ in N‐functionalized UiO‐67 MOFs, revealed by NPD. (a) C₃H₈ in UiO‐67‐L‐Thr with C–H···N interactions. (b, c) C₃H₄ in UiO‐67‐L‐Thr with C–H···N interactions. (d) C₃H₄ in UiO‐67‐L‐Thr with C≡C···(H–)N interactions. (e, f) C₃H₆ in UiO‐67‐L‐Thr with C–H···N interactions. (g) C₃H₄ in UiO‐67‐Py with C–H···N interactions. Hydrogen atoms are not included in distance measurements because their positions have a non‐negligible degree of uncertainty caused by thermal effects. See Supporting Information for details. Zr: blue, C of framework: grey, O: red, N: green, C of C₃H₈: pink, C of C₃H₄: orange, C of C₃H₆: purple. H is omitted for clarity. (h) The monotonic relationship between $E_{\mathbf{a}|\mathbf{b}}^{\mathbf{a}}$ (including $E_{{\mathbf{C}}_3{\mathbf{H}}_6|{\mathbf{C}}_3{\mathbf{H}}_8}^{{\mathbf{C}}_3{\mathbf{H}}_6}$ and $E_{{\mathbf{C}}_3{\mathbf{H}}_6|{\mathbf{C}}_3{\mathbf{H}}_4}^{{\mathbf{C}}_3{\mathbf{H}}_6}$) and the charge of nitrogen atoms in four N‐functionalized UiO‐67 MOFs. The red and blue lines represent the linearly fitted curves.

DFT calculations were carried out to further study the differences in the interactions between C₃H₈/C₃H₆/C₃H₄ and UiO‐67‐L‐Thr. As described in Supplementary Discussions (Figure S67–S68 and Table S9–S10), results indicate that N of –NH₂ predominantly contributes to the separation and the interactions between C₃H₈/C₃H₆/C₃H₄ and the N site are mainly the C–H···N interactions. We compared the C–H···N interactions between three types of gas molecules and the –NH₂ group by analysing distances between the hydrogen atoms of C–H in C₃H₈/C₃H₆/C₃H₄ and the nitrogen atom of –NH₂. In general, C₃H₆ is the farthest away from the –NH₂ group (C–H···N = 2.68 Å in the octahedral pore and 2.87 Å in the tetrahedral pore), indicating its weakest binding. C₃H₈ is closest to the –NH₂ group (C–H···N = 2.47 Å in the octahedral pore and 2.59 Å in the tetrahedral pore), indicating its strongest binding. C₃H₄ has a modest interaction with the –NH₂ group (C–H···N = 2.58 Å in the octahedral pore and 2.85 Å in the tetrahedral pore). The interaction strength follows C₃H₆< C₃H₄< C₃H₈, which well matches the trends in our experimental Q _st and NPD‐derived interaction strengths. We notice that C₃H₄ uptake is generally higher than C₃H₈ uptake in single‐component adsorption isotherms. This is probably because C₃H₄ uptake not only depends on single‐site adsorption evaluated by Q _st and DFT but also is influenced by multiple‐site adsorption and molecular networks revealed by NPD.

We further used NPD to investigate the guest–host interactions in other N‐functionalized UiO‐67 MOFs discussed previously, particularly focusing on UiO‐67‐Py with C₃H₄ adsorbed (Figure S69–S70 and Table S11). Three adsorption sites were identified (Figure S71). The π‐electron cloud of the pyridine group in the ligand was found to form van der Waals forces with the hydrogen atoms of C–H in C₃H₄ (C–H···π interactions), leading to C₃H₄ adsorption at sites 1 and 3. Site 2 is where weak interactions form between the nitrogen atom of the pyridine group and the hydrogen atoms of C–H in C₃H₄ (C–H···N interactions, C(H)···N = 4.90 Å) (Figure 4g). It is noticed that C₃H₄ is further away from the nitrogen atom in UiO‐67‐Py than from that in UiO‐67‐L‐Thr (C(H)···N = 3.01 and 3.08 Å). This means that C–H···N interactions in UiO‐67‐L‐Thr are stronger than those in UiO‐67‐Py. The stronger C–H···N interactions in UiO‐67‐L‐Thr are also the main driving force of C₃H₄ molecular network formation. These molecular networks, which are not found in UiO‐67‐Py, play a crucial role in enhancing C₃H₄ adsorption in UiO‐67‐L‐Thr. All in all, our experimental studies suggest that the binding strength to nitrogen functional groups may be a dominating factor that determines the C₃H₈/C₃H₆/C₃H₄ separation performance of UiO‐67 MOFs.

The binding strength to nitrogen functional groups is probably determined by the electronegativity of or charge distribution on nitrogen atoms, influenced by their chemical environment. In UiO‐67‐AA, nitrogen atoms are bonded to saturated carbon atoms of amino acids, whereas in comparison samples, they are attached to unsaturated carbon atoms of phenyl/pyridine carboxylic acid ligands/modulators. The saturated carbon atoms may exert an electron‐donating effect on nitrogen atoms, enhancing their electronegativity and thereby strengthening interactions with C₃H₈, C₃H₆, and C₃H₄. DFT calculations were carried out to provide a quantitative relation between the separation performance and the electronegativity of nitrogen atoms. To quantify the separation performance, a quantity, $E_{\mathbf{a}|\mathbf{b}}^{\mathbf{a}}$, is defined. It is calculated as the average adsorption energy of gas a minus that of gas b. It can be used to evaluate the performance of an MOF with multiple sites or a given site in the separation of gas a from a mixture consisting of components a and b. An elevated $|E_{\mathbf{a}|\mathbf{b}}^{\mathbf{a}}|$ will result in superior separation performance. The calculated values of $E_{{\mathbf{C}}_3{\mathbf{H}}_6|{\mathbf{C}}_3{\mathbf{H}}_8}^{{\mathbf{C}}_3{\mathbf{H}}_6}$ and $E_{{\mathbf{C}}_3{\mathbf{H}}_6|{\mathbf{C}}_3{\mathbf{H}}_4}^{{\mathbf{C}}_3{\mathbf{H}}_6}$ of four N‐functionalized UiO‐67 MOFs are summarized in Table S12. UiO‐67‐L‐Thr gives the highest $E_{{\mathbf{C}}_3{\mathbf{H}}_6|{\mathbf{C}}_3{\mathbf{H}}_8}^{{\mathbf{C}}_3{\mathbf{H}}_6}$ and $E_{{\mathbf{C}}_3{\mathbf{H}}_6|{\mathbf{C}}_3{\mathbf{H}}_4}^{{\mathbf{C}}_3{\mathbf{H}}_6}$, thereby indicating that it has the best performance in the separation of C₃H₆ from the ternary mixture, in line with experimental results. The calculated $E_{{\mathbf{C}}_3{\mathbf{H}}_6|{\mathbf{C}}_3{\mathbf{H}}_8}^{{\mathbf{C}}_3{\mathbf{H}}_6}$ and $E_{{\mathbf{C}}_3{\mathbf{H}}_6|{\mathbf{C}}_3{\mathbf{H}}_4}^{{\mathbf{C}}_3{\mathbf{H}}_6}$ are plotted against the charge of the nitrogen atom (q _N) (Figure 4h) [52]. The monotonic relationship between $E_{{\mathbf{C}}_3{\mathbf{H}}_6|{\mathbf{C}}_3{\mathbf{H}}_8}^{{\mathbf{C}}_3{\mathbf{H}}_6}$ or $E_{{\mathbf{C}}_3{\mathbf{H}}_6|{\mathbf{C}}_3{\mathbf{H}}_4}^{{\mathbf{C}}_3{\mathbf{H}}_6}$ and q _N indicates that a large negative charge of nitrogen atoms is instrumental in enhancing the separation performance. Although the calculated $E_{\mathbf{a}|\mathbf{b}}^{\mathbf{a}}$ cannot fully predict the experimental separation performance due to other factors, such as pore partition, the key role of nitrogen charge is still well elucidated. A detailed analysis of the function of N sites and other adsorption sites can be found in Supplementary Discussions (Figure S72–S73, Tables S9 and S13–S18). Notably, introducing nitrogen atoms into the MOF framework not only provides direct adsorption sites that dominate the separation but also modulates the electronic properties and the degree of conjugation of adjacent benzene (pyridine) rings, thereby influencing their interactions with gas molecules. Overall, these findings establish an element‑specific and quantitative electronegativity‐driven design principle, valuable for the development of next‐generation MOF adsorbents for industrial olefin purification. A schematic illustrating this concept can be found in Figure S74.

2.4. Evaluation of Practical Application

To evaluate the practical separation capability of UiO‐67‐AA for C₃H₆ from C₃H₈/C₃H₆/C₃H₄ mixtures, we performed dynamic column breakthrough experiments on a packed column of UiO‐67‐L‐Thr. For a C₃H₈/C₃H₆ (50/50, v/v) mixture (Figure S75), breakthrough occurred at 277 min·g⁻¹ for C₃H₆ and 287 min·g⁻¹ for C₃H₈, giving a penetration interval of 10 min·g⁻¹. This interval increased to 18 min·g⁻¹ for the 10/90 (v/v) mixture (Figure S76). For the C₃H₈/C₃H₆/C₃H₄ mixture, we first used a composition of 9/90/1 (v/v/v), mimicking typical industrial streams from steam crackers or fluidized catalytic cracking units (C₃H₈: 5%–15%, C₃H₆: 80%–95%, C₃H₄: 0.5%–5%). As shown in Figure 5a, C₃H₆ broke through at 280 min·g⁻¹, while C₃H₈ and C₃H₄ broke through at 284 and 310 min·g⁻¹, resulting in a C₃H₆/C₃H₄ penetration interval of 30 min·g⁻¹. Gas chromatography‐mass spectrometry (GC‐MS) analysis of the effluent gas confirmed the high purity of the collected C₃H₆ (Figure S77). Further evaluation with a 1/98/1 (v/v/v) mixture demonstrated the robustness of UiO‐67‐L‐Thr, which maintained excellent separation performance—24 min·g⁻¹ penetration interval for C₃H₆/C₃H₈ and 50 min·g⁻¹ penetration interval for C₃H₆/C₃H₄—while consistently delivering high‐purity C₃H₆ (Figure 5b). Cyclic breakthrough and adsorption tests revealed that the separation performance remained stable over at least five consecutive cycles (Figure 5c and Figure S78–S81). Regeneration via an N₂ purge was proven to be a more energy‑efficient alternative (Figure S82). PXRD and BET data show that UiO‐67‐L‐Thr fully retained its crystallinity and porosity after the dynamic separation tests, underscoring its excellent structural stability (Figure S83). Notably, UiO‐67‐L‐Thr also exhibited excellent long‐term stability. After six months of storage in air at 60%–80% relative humidity (RH), the material retained its structural integrity with no phase transition or crystallinity loss (Figure 5d and Figure S84). Moreover, no significant degradation in C₃H₈ and C₃H₄ adsorption capacity was observed compared to freshly synthesized material. This implies that UiO‐67‐AA can be stored conveniently for practical use.

FIGURE 5.

Breakthrough curves of UiO‐67‐L‐Thr for C₃H₈/C₃H₆/C₃H₄ (9/90/1, v/v/v) (a) and (1/98/1, v/v/v) (b) mixtures at 298 K and 1 bar with a flow rate of 2 mL·min⁻¹. (c) Cyclic breakthrough tests of UiO‐67‐L‐Thr for a C₃H₈/C₃H₆/C₃H₄ (9/90/1, v/v/v) mixture at 298 K and 1 bar with a flow rate of 2 mL·min⁻¹. (d) PXRD patterns of UiO‐67‐L‐Thr before and after being exposed to air (60%–80% RH) for 6 months, with a comparison of C₃H₈ and C₃H₄ adsorption isotherms at 298 K. (e) PXRD patterns of UiO‐67‐L‐Thr synthesized at different scales. The slight peak broadening observed in the large‐scale samples is attributed to a modest increase in crystallite size distribution due to minor variations in crystallization kinetics, which is common and expected upon scaling. (f) Comparison of C₃H₈, C₃H₆, and C₃H₄ adsorption isotherms at 298 K for large‐scale and small‐scale synthesized UiO‐67‐L‐Thr. (g) Cost comparison of reported MOFs for C₃H₈/C₃H₆ and/or C₃H₄/C₃H₆ separation. UiO‐67‐AA: pink, reported ternary C₃ separating MOFs: blue, reported binary C₃ separating MOFs: grey. (h) Radar chart for comparing the comprehensive performance of UiO‐67‐AA with some representative binary or ternary C₃ separating MOFs.

The production of scalable and low‐cost MOFs is critical for achieving industrial implementation. To evaluate the potential of large‐scale synthesis, we scaled up our synthesis from 15 mL to 500 mL vessels using identical synthetic protocols. PXRD analysis confirmed the good phase purity of the large‐scale synthesized UiO‐67‐L‐Thr (Figure 5e). Moreover, gas adsorption over the large‐scale and small‐scale synthesized materials showed no significant difference (Figure 5f), confirming consistent separation performance. Cost analysis reveals transformative economics. The synthesis of UiO‐67‐AA with naturally occurring and widely available amino acids provides a significant cost advantage over other MOFs that need more expensive or scarce precursors. Through the established MOFs cost calculation method [53, 54, 55], a comparison of material costs was conducted (Figure 5g). A detailed cost breakdown, including assumptions and raw material prices, is provided in Table S19. Table S20 presents a sensitivity analysis that examines how variations in key parameters influence overall cost reduction. Compared with literature‐reported C₃H₈/C₃H₆ and/or C₃H₄/C₃H₆ separating MOFs, UiO‐67‐L‐Thr could reduce the cost by over 2250 times at an estimated cost of $293 per kilogram. Scaling considerations detailed in Table S19 point out that the scale‐up of UiO‐67‐AA would involve lower technical uncertainty and potentially lower overall process costs. The balance between selectivity and cost was evaluated by calculating the cost of producing 1 L of C₃H₆ (see notes to Table S19). UiO‐67‐L‐Thr could reduce this cost by over 690 times, demonstrating a considerable economic benefit. Therefore, the substantial cost advantage of UiO‐67‐AA can more than compensate for a moderate numerical selectivity, especially since this selectivity is already adequate to achieve target product purity. Overall, as shown in Figure 5h, UiO‐67‐AA exhibits excellent comprehensive performance among the reported C₃H₈ and/or C₃H₄ MOF adsorbents in terms of separation efficiency (selectivity, uptake), operational robustness (regeneration, stability), scalable manufacturing, and most importantly, low cost, making it a promising candidate for industrial C₃H₆ purification.

3. Conclusion

In summary, we have developed a highly economical and effective strategy for one‐step C₃H₆ separation from C₃H₈/C₃H₆/C₃H₄ by functionalizing industrially promising Zr‐MOFs with widely available amino acids. The resulting UiO‐67‐AA exhibits not only exceptionally high C₃H₈ and C₃H₄ uptakes but also benchmark selectivities toward both C₃H₈/C₃H₆ and C₃H₄/C₃H₆ separation. Breakthrough experiments proved that UiO‐67‐AA can concurrently remove C₃H₈ and C₃H₄ from ternary C₃ mixtures, producing polymer‐grade C₃H₆ in one step. Crucially, an estimated over 2250 times cost reduction among reported C₃H₈ and/or C₃H₄ MOF adsorbents makes UiO‐67‐AA an ideal candidate for industrial C₃H₆ purification. Assisted by in situ IR spectroscopy, in situ NPD, and DFT calculations, we revealed that amino acids offer selective binding to dually strengthen both C₃H₈ and C₃H₄ molecular/network affinity over that of C₃H₆, with the amino group dictating the supramolecular interactions. Furthermore, through a systematic evaluation across UiO‐67 with diverse nitrogen moieties exhibiting varying electronic properties, we established a monotonic relationship between the separation performance and the specific nitrogen electronegativity. This element‑specific and quantitative electronegativity‐driven rule is valuable for designing next‐generation MOF adsorbents for industrial olefin purification.

Funding

This work was supported by the National Natural Science Foundation of China (22303059, 92477104, and 22473042), Jiangsu Provincial Department of Science and Technology (BK20230472), Suzhou Municipal Department of Science and Technology (ZXL2024393), Suzhou Key Laboratory of Spent Lithium Battery Recycling (SZS2025005), and Guangdong Innovative & Entrepreneurial Research Team Program (2021ZT09C539).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: smll74568‐sup‐0001‐SuppMat.pdf.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.