A square lattice topology coordination network that exhibits highly selective C2H2/CO2 separation performance

Abstract

C₂H₂/CO₂ separation is an industrially important process that remains challenging because of the similar physicochemical properties of C₂H₂ and CO₂. We herein report that the new square lattice (sql) coordination network [Cu(bipy‐xylene)₂(NO₃)₂]_n, sql‐16‐Cu‐NO₃, 16 = bipy‐xylene = 4,4′‐(2,5‐dimethyl‐1,4‐phenylene)dipyridine, exists in at least three forms, as‐synthesised (α), activated (α′) and hydrated (β). The activated phase, sql‐16‐Cu‐NO₃‐α′, is an ultramicroporous material that exhibits high selectivity towards C₂H₂ over CO₂ as revealed by dynamic gas breakthrough experiments (1:1, C₂H₂/CO₂) that afforded 99.87% pure CO₂ in the effluent stream. The separation selectivity at 298 K and 1 bar, 78, is the third best value yet reported for C₂H₂ selective physisorbents whereas the mid‐loading performance sets a new benchmark. The performance of sql‐16‐Cu‐NO₃‐α′ is attributed to a new type of C₂H₂ binding site in which CH···ONO₂ interactions enable moderately strong sorbent‐sorbate binding (Q_st (C₂H₂) = 38.6 kJ/mol) at low loading. Conversely, weak CO₂ binding (Q_st (CO₂) = 25.6 kJ/mol) at low loading means that (ΔQ_st)_AC [Q_st (C₂H₂)–Q_st (CO₂)] is 13 kJ/mol at low coverage and 11.4 kJ/mol at mid‐loading. Analysis of in situ powder X‐ray diffraction and modelling experiments provide insight into the sorption properties and high C₂H₂/CO₂ separation performance of sql‐16‐Cu‐NO₃‐α′.

INTRODUCTION

Addressing the downstream processing of commodity chemicals with energy-efficient sorbent materials is considered an urgent global challenge.¹ Thanks to its utility both as a fuel and a building block/feedstock for several industrially relevant processes, acetylene (C₂H₂) is an important chemical product.² Production of C₂H₂ generally affords CO₂ as an impurity.³ Therefore, selective separation of C₂H₂ from C₂H₂/CO₂ mixtures is industrially relevant but challenging process because of their similar molecular sizes and physical properties. Further, with an extremely high flammability range of 2.5–81%, C₂H₂ poses an immediate fire and explosion hazard at concentrations >2.5%. When coupled with the high reactivity of C₂H₂ to afford undesired derivatives during downstream catalytic processes, and the large energetic and economic costs of industrial-scale distillation processes, the need to develop energy-efficient sorbents that can selectively capture C₂H₂ is apparent.⁴

Among the recovery and purification methods that offer potential for a low energy footprint, porous coordination networks that can function as physisorbents have emerged as benchmark materials for separation of C₂H₂ from CO₂.⁵ Specifically, physisorbents such as metal organic materials (MOMs),⁶ commonly known as porous coordination polymers⁷ or metal-organic frameworks (MOFs),⁸ have attracted attention for separating C2 light hydrocarbon gas mixtures, either through capture/desorption of the product gas or via selective adsorption of impurities.^9,10 In particular, a subclass of MOMs, hybrid ultramicroporous materials (HUMs; pore size <0.7 nm), currently offer benchmark C2 separation performances including for C₂H₂ capture from C₂H₄ and CO₂.^11–14 Fine tuning of pore size and pore chemistry enabled by systematic crystal engineering studies of HUMs has thus far resulted in C₂H₂ selectivity values higher than classical physisorbents by one or two orders of magnitude.^5,12

Ultramicroporous square lattice (sql) topology coordination networks also offer promise for C₂H₂ selective sorption,⁵ as exemplified by UTSA-300a¹⁵ and NCU-100a.¹⁶ sql coordination networks that exhibit layered structures were first reported by the group of Fujita et al. in 1994¹⁷ and represent the most common type (45.09%)¹⁸ of reported 2D coordination networks. Nets of sql topology also play a role in the crystal engineering of other common network topologies such as pillared pcu networks.¹⁹ sql coordination networks can be readily prepared by the self-assembly of four-connected metal centres and ditopic (e.g., dipyridyl) linker ligands. Thanks to their modular composition (metal, linker ligand, anion and guest can all be substituted), their inherent amenability to crystal engineering guided design is long known.^17,20–22 Whereas interpenetration in sql coordination networks is possible,^20,21,23 noninterpenetrated square grid networks can stack in laminated fashion and when the sql layers separate to intercalate guests in the interlayer space they are, in effect, clay mimics.^22,24 That the prototypal sql network ELM-11, [Cu(4,4‘-bipyridine)₂(BF₄)₂]_n, exhibits phase switching in the presence of CO₂, N₂, O₂, CH₄, C₂H₂, and n-butane^25,26 at different pressures raises the issue of whether or not ELM-11 variants might also exhibit strong gas separation performance. Indeed, ELM-11,²⁷ ELM-12 [Cu(4,4′-bipyridine)₂(OTf)₂]_n,²⁸ and ELM-13 [Cu(4,4′-bipyridine)₂(CF₃BF₃)₂]_n,²⁹ have been studied for C₂H₂/C₂H₄, C₃H₄/C₃H₆, and CO₂/N₂ separations, respectively.

Linker ligands not only play an important role in the design of sql coordination networks, by lining the pore or cavity surfaces they often emerge as the key to controlling sorbent-sorbate molecular recogniton.^20,30 N-donor based linker ligands have been widely studied for this purpose.^20,31,32 A literature survey of sql networks reported in TTO and CCDC revealed²⁸ that, whereas there are hundreds of distinct N-donor linkers that have been reported to form sql topology nets, only 15 such nets have been studied for gas/vapour sorption (Table S1). Seven bipyridine based linker ligands form sql networks that exhibit switching behaviour induced by exposure to gas or vapor (Table S2). Two of these sql networks have been studied for binary C₂H₂/CO₂ separation: UTSA-300a¹⁵ and UTSA-83a.^5,33 Banglin Chen's group reported that UTSA-300a,¹⁵ [Zn(SiF₆)(dp_s)₂]_n (4,4'-dipyridylsulfide [dps]), is an ultramicroporous sql coordination network which exhibits partial molecular sieving for C₂H₂ over CO₂ and benchmark ideal adsorbed solution theory (IAST) selectivity. Herein we report that the linker ligand bipy-xylene, 4,4'-(2,5-dimethyl-1,4-phenylene)dipyridine, 16, forms an ultramicroporous sql network with high affinity for C₂H₂ over CO₂. To the best of our knowledge, 16 has only been used once before to sustain an ultramicroporous coordination network, namely Cr₂O₇²⁻ pillared Co(II) and Ni(II) HUMs.³⁴

Single crystals of as-synthesised sql-16-Cu-NO₃-α were obtained by solvent diffusion of bipy-xylene and copper nitrate in iPrOH (isopropyl alcohol) and water (see Supporting Information for full procedure). sql-16-Cu-NO₃-α transforms to its β-phase ([Cu(bipy-xylene)₂(NO₃)(H₂O)]_n), in which one of the axial nitrates is replaced by a water molecule, when exposed to ambient humidity for a week (Figure 1A). Crystallographic studies revealed that sql-16-Cu-NO₃-α and sql-16-Cu-NO₃-β crystallized in the monoclinic space groups C2/c and P2₁/c respectively (Table S3). Porosity in both structures occurs thanks to one-dimensional channels along the crystallographic c-direction (Figure 1D and S1). In sql-16-Cu-NO₃-α, the pore limiting diameter³⁵ was determined to be 4.6 Å, while the maximum channel diameter was found to be 5.0 Å (Figure S1). In sql-16-Cu-NO₃-β, the channel openings are constricted to a limiting diameter of 3.8 Å, while the maximum channel diameter is 4.8 Å, close to that in sql-16-Cu-NO₃-α. It is noteworthy that the NO₃⁻ ions present in sql-16-Cu-NO₃-α and sql-16-Cu-NO₃-β occupy these channels only minimally (Figure S1). Rather, the axial NO₃⁻ ions interdigitate between adjacent undulating layers in both sql-16-Cu-NO₃-α and sql-16-Cu-NO₃-β (Figures 1B,C, S2, and S3). Bulk phase purity of sql-16-Cu-NO₃-α was verified by powder X-ray diffraction (PXRD; Figures S4 and S5). Thermogravimetric analysis of sql-16-Cu-NO₃-α suggests thermal stability up to 573 K (Figure S11). Exposure of sql-16-Cu-NO₃-α to air at room temperature resulted in a mixture of α and β phases (Figures S6 and S7) and hence, although a single crystal of sql-16-Cu-NO₃-β was isolated, we were unable to isolate pure bulk samples for further study. A variable-temperature PXRD (VT-PXRD) study revealed that the mixed phase of α and β converted to desolvated α′-phase after heating at 373 K in air (Figure S8). While PXRD patterns of α and α′ phases are similar, the appearance of two new peaks at 11.2° and 11.5° 2θ in the α′ phase indicated that there are subtle structural differences between these two phases (Figures S9 and S10). Fourier-transform infrared (FT-IR) spectroscopy was used to study sql-16-Cu-NO₃-α and sql-16-Cu-NO₃-α′ to assess insight that might come from spectroscopic differences between these two phases (Figure S12). The FT-IR spectra for sql-16-Cu-NO₃-α and sql-16-Cu-NO₃-α′ were found to exhibit only subtle differences, consistent with our observation concerning their PXRD patterns (Figure S9).

Figure 1. Synthesis and crystal structures of sql-16-Cu-NO3-α and sql-16-Cu-NO3-β, (A) Synthetic steps to obtain single crystals of sql-16-Cu-NO3-α and sql-16-Cu-NO3-β. Layer arrangements in (B) sql-16-Cu-NO3-α and (C) sql-16-Cu-NO3-β. (D) Connolly surface of sql-16-Cu-NO3-α with a probe radius of 0.7 Å, revealing a 1D channel along the crystallographic c-direction

To assess the porosity of sql-16-Cu-NO₃-α′, low pressure CO₂ and C₂H₂ adsorption isotherms were collected. Before conducting gas sorption measurements, sql-16-Cu-NO₃-α was activated at 353 K under vacuum to generate sql-16-Cu-NO₃-α′. CO₂ sorption at 195 K revealed a type I isotherm (Figure S13) with a Brunauer–Emmett–Teller surface area of 179 m²/g. A pore width of 4.1 Å for sql-16-Cu-NO₃-α′ was determined by the Horvath–Kawazoe method applied to the CO₂ isotherm at 195 K (Figure S16). Conversely, the C₂H₂ isotherm at 195 K is a stepped type F-II isotherm,¹⁵ which we attribute to C₂H₂ induced phase transformation as seen for several other sorbents.^27,36,37 Such a gating effect has been seen for other sql coordination networks but remains a rarity among physisorbents in general.²⁹ Switching sql coordination networks are known to exhibit type F-IV³⁸ or F-II³⁸ isotherms.^30,31

The sorbate induced phase transformation of sql-16-Cu-NO₃-α′ under cryogenic conditions prompted us to conduct in situ C₂H₂ sorption PXRD experiments³⁹ at 195 K and 298 K from 0 to 1 bar to gain insight into the structural changes that accompany C₂H₂ sorption. At 195 K (Figure 2A, left), initial C₂H₂ uptake of 52 cm³/g occurred below 0.3 bar, at which point a phase transformation resulted in a gate-opening effect. Saturation C₂H₂ uptake of 81 cm³/g was measured at 1 bar. In situ PXRD (Figure 2A, right) confirmed a structural change upon C₂H₂ adsorption above the “gate-opening” 0.3 bar pressure evidenced by a new peak at 11.7° 2θ (representative pattern D, Figure 2A, right). Upon desorption, this peak is retained until ca. 0.05 bar (representative pattern E, Figure 2A, right). Unlike the 195 K data, at 298 K sql-16-Cu-NO₃-α′ exhibited a type I isotherm for C₂H₂ (Figure 2b, left) with no new peaks in the in situ PXRD patterns (Figure 2b, right). Further, the unit cell parameters obtained in a batch Pawley profile fit of PXRD data changed only 1.4% (Figure 2b, left and Figure S17), suggesting framework rigidity upon C₂H₂ sorption at 298 K. Further, the C₂H₂ adsorption isotherm at 273 K and CO₂ adsorption isotherms at 273 K and 298 K were found to exhibit type I isotherms (Figure 2C). At 1 bar, sql-16-Cu-NO₃-α′ adsorbs 40 cm³/g and 34.7 cm³/g of C₂H₂ at 273 K and 298 K, respectively. In contrast, it adsorbs only 24 cm³/g and 16.7 cm³/g of CO₂ at 273 K and 298 K, respectively (Figure 2C). Isosteric enthalpies of adsorption (Q_st) were calculated from sorption isotherms at 273 K and 298 K. Low loading Q_st(C₂H₂, 38.6 kJ/mol) and Q_st(CO₂, 25.6 kJ/mol; Figure 2D) values correlate well with the different uptakes of C₂H₂ and CO₂. IAST⁴⁰ selectivities (S_AC) were calculated by fitting the single-component isotherms to the dual-site Langmuir-Freundlich equation (see Supporting Information for details, Table S4), considering binary gas mixtures of C₂H₂/CO₂ (1:1 v/v) at 1 bar and 298 K. The calculated S_AC (1:1) at 1 bar is 27.8 for sql-16-Cu-NO₃-α′ (Figure S22), which is higher than leading C₂H₂-capture sorbents including ZJU-196 (25),⁴¹ FeNiM′MOF (24),⁴² [Ni₃(HCOO)₆] (22),⁴³ DICRO-4-Ni-i (18.2),⁴⁴ TCuCl (16.0),⁴⁵ HOF-3 (14.0),⁴⁶ MIL-100(Fe) (12.5),⁴⁷ TIFSIX-2-Cu-i (10),¹¹ ZJUT-2a (10),⁴⁸ TCuBr (9.1),⁴⁵ SSNU-45 (8.5),⁴⁹ FJU-22a (7.1),⁵⁰ ZJU-60a (6.7),⁵¹ UTSA-83a (6.2),²³ and MUF-17 (6)⁵² (Table S5). The high S_AC (1:1) at 1 bar derived from single-component isotherms and Q_st differences are indicative of potential suitability for equimolar C₂H₂/CO₂ binary mixture separations under dynamic conditions.

Figure 2. In-situ coincident PXRD/C2H2 sorption measured for sql-16-Cu-NO3-α′ at 195 K (A) and 298 K (B): (A, B, left) adsorption (closed symbols) and desorption (open symbols); (A, B right) representative PXRD patterns (λ = 1.54178 Å) at different C2H2 adsorption/desorption loadings, each pattern corresponds to the point labelled on the left. (C) C2H2 and CO2 adsorption isotherms at 273 K and 298 K. (D) Qst(C2H2), and Qst(CO2) for sql-16-Cu-NO3-α′. PXRD, powder X-ray diffraction

Pure gas adsorption kinetics for C₂H₂ and CO₂ were studied using activated samples of sql-16-Cu-NO3-α exposed to a constant flow of 10 cm³/min C₂H₂ or CO₂ at 303 K and 1.0 bar (Figure S23). The slope of the kinetic curve is steeper for C₂H₂ versus CO₂, indicating faster adsorption kinetics for C₂H₂. The kinetic curves level off at 2.9 wt.% (25 cm³/g) for C₂H₂ and 1.6 wt.% (8 cm³/g) for CO₂. Higher uptake of C₂H₂ is in agreement with the volumetric pure gas sorption experiments. Regeneration was achieved by heating at 373 K under N₂ flow in ca. 1 h (flow rate: 60 cm³/min).

The promising isotherms and kinetic studies prompted us to study the C₂H₂/CO₂ separation performance of sql-16-Cu-NO₃-α′ through dynamic column breakthrough (DCB) experiments⁵³ that mimic typical process conditions² with an inlet gas mixture composition of 1:1 (v/v) C₂H₂:CO₂. The C₂H₂/CO₂ gas mixture was passed through a fixed bed column (8 mm diameter) filled with ca. 0.5 g of sql-16-Cu-NO₃-α′ with a flow rate of 1 cm³/min, at 1 bar and 298 K. The preactivated fixed bed of sql-16-Cu-NO₃-α was first heated at 373 K in a 20 cm³/min flow of He for 4 h to ensure activation. The sample was then cooled to room temperature before being subjected to DCB experiments. Eluted components were continuously monitored through gas chromatography (GC; Figure S24, see Supporting Information for details). Figure 3A reveals that CO₂ breakthrough occurred at 55 min/g, well before that of C₂H₂ (142 min/g). During the time lag before breakthrough, GC data revealed that the C₂H₂ level in the effluent gas stream was ≤1276 ppm. These data show that, until C₂H₂ breakthrough, CO₂ purity in the effluent stream was >99.87% for sql-16-Cu-NO₃-α′, a purity much higher than the commercial specification (N2.0, 99%). For sql-16-Cu-NO₃-α′, C₂H₂ saturation uptake calculated from the DCB profile is 29.4 cm³/g, in good agreement with the single-component C₂H₂ isotherm-based uptake at half-coverage (31.8 cm³/g under ambient equilibrium conditions). sql-16-Cu-NO₃-α′ maintained the same retention time and acetylene uptake capacity as the initial DCB cycle in subsequent cycles (Figure 3B). The separation selectivity (α_AC) for sql-16-Cu-NO₃-α’, calculated from the DCB experiment is 78 (Figure 4A). In terms of α_AC for the studied 1:1 binary gas mixture, α_AC(1:1), sql-16-Cu-NO₃-α′ outperforms TCuI,⁴⁵ UTSA-74a,⁵⁴ JXNU-5,⁵⁵ JCM-1,⁵⁶ FJU-89a,⁵⁷ SSNU-45,⁴⁹ NKMOF-1-Ni,⁵⁸ FJU-6-TATB,⁵⁹ FJU-36a,⁶⁰ HOF-3,⁴⁶ FJU-22a,⁵⁰ and FeNiM′MOF.⁴² Only TCuBr,⁴⁵ and TCuCl⁴⁵ compare favourably to sql-16-Cu-NO₃-α′ with respect to α_AC(1:1), making sql-16-Cu-NO₃-α′ the third best-performing C₂H₂/CO₂ selective adsorbent with respect to DCB derived 1:1 separation selectivity (Figure 4A and Table S5).

Figure 3. Binary C2H2/CO2 mixture (v/v = 1:1) based DCB experimental curves at 298 K and 1 bar for sql-16-Cu-NO3-α′: (A) one cycle (B) three cycles. DCB, dynamic column breakthrough

Figure 4. Comparison of leading C2H2/CO2 selective sorbents with sql-16-Cu-NO3-α′, (A) separation selectivity, αAC (B) (ΔQst)AC at half loadings

The high α_AC for sql-16-Cu-NO₃-α′ can be partially attributed to weak CO₂ binding versus C₂H₂, as reflected in the difference between the low coverage adsorption enthalpies. (ΔQ_st)_AC for sql-16-Cu-NO₃-α′ (13) is comparable to SNNU-45 (12.8) and only below NKMOF-1-Ni (19.4) kJ/mol (Figure S20 and Table S5). However, the higher (ΔQ_st)_AC for NKMOF-1-Ni is driven by its high Q_st(C₂H₂) value of 60.3 kJ/mol,⁵⁸ which then declines to 46 kJ/mol at half loading and is <20 kJ/mol at saturation (Figure S21). Comparison of (ΔQ_st)_AC values at half loading to evaluate the preferential binding of C₂H₂ versus CO₂ in a 1:1 mixture is given in Figure 4B. Figure 4B reveals that sql-16-Cu-NO₃-α′ exhibits benchmark (ΔQ_st)_AC at half loading of 11.4 kJ/mol, followed by NKMOF-1-Ni (9.5 kJ/mol)>TCuCl (8.2 kJ/mol)>TCuI (7.7 kJ/mol)>TIFSIX-2-Cu-i (7.4 kJ/mol)>UTSA-74a (6.0 kJ/mol)>TCuBr (5.5 kJ/mol) > SSNU-45 (4.1 kJ/mol)>JXNU-5 (3.4 kJ/mol)>FJU-22a (3.0 kJ/mol)>FeNiM′MOF (2.1 kJ/mol)>FJU-89a (1.4 kJ/mol).

Modelling studies on sql-16-Cu-NO₃ were conducted using the experimentally determined crystal structure of sql-16-Cu-NO₃-α. C₂H₂ and CO₂ binding site distributions (see Supporting Information for other plausible binding sites, Figure S25) were determined by Monte Carlo techniques employing the UFF-force field.⁶¹ In line with our experimental sorption data, these binding sites (Figure 5) indicate preference for C₂H₂ versus CO₂ that can be attribute to hydrogen bonding between C₂H₂ and NO₃^‒ ions. At low coverage, the simulated Q_st values for C₂H₂ and CO₂ are within ±3.5 kJ/mol of the experimental Q_st values (see Supporting Information for details).

Figure 5. Binding sites of CO2 and C2H2 in sql-16-Cu-NO3-α as determined by molecular modelling calculations detailed in SI: (A) View of the CO2 binding sites and (B) C2H2 binding sites in sql-16-Cu-NO3-α (CO2 and C2H2 molecules are shown in space-filling mode)

The stability of sql-16-Cu-NO₃-α′ towards humidity was evaluated using an accelerated stability protocol adopted by the pharmaceutical industry (test at 313 K and 75% relative humidity [RH], see Supporting Information for details)⁶² and water vapour sorption experiments. sql-16-Cu-NO₃-α′ transformed to a mixture of sql-16-Cu-NO₃-α and sql-16-Cu-NO₃-β when exposed to humidity for 1 day and retains this state after 30 days under humid conditions (Figure S27). Figures S28 and S29 reveal that sql-16-Cu-NO₃-α′ transformed to a mixed phase of sql-16-Cu-NO₃-α and sql-16-Cu-NO₃-β after water vapour sorption. The accelerated stability tests and water vapour sorption experiments indicated that sql-16-Cu-NO₃ retained structural integrity and that it has a shelf-life of ≥4 months as an α + β mixture (stable at 313 K and 75% RH over ≥30 days).

In summary, we report efficient C₂H₂/CO₂ separation using an ultramicroporous sql coordination network, sql-16-Cu-NO₃-α′, a new member of the still understudied sql class of sorbents. sql-16-Cu-NO₃-α′ exhibits a high C₂H₂/CO₂ (1:1) IAST selectivity derived from single-component isotherms and the third best experimentally derived equimolar separation selectivity. The C₂H₂/CO₂ separation performance of sql-16-Cu-NO₃-α′ is credited to its thermodynamic preference for C₂H₂ versus CO₂ binding, exemplified by benchmark (ΔQ_st)_AC at half loading driven by hydrogen bonding between C₂H₂ molecules and NO₃^‒ ions. This new type of binding site for C₂H₂ is likely to be present in other sorbent materials with similar pore size and chemisty.

ACKNOWLEDGEMENTS

Michael J. Zaworotko acknowledges the support of the Science Foundation Ireland (SFI Awards 13/RP/B2549 and 16/IA/4624) and the Irish Research Council (IRCLA/2019/167). Mohana Shivanna and Susumu Kitagawa thank the financial support of KAKENHI, Grant-in-Aid for Scientific Research (S) (JP18H05262), and Early-Career Scientists (JP19K15584) from the Japan Society of the Promotion of Science (JSPS).

CONFLICT OF INTERESTS

The authors declare that there are no conflict of interests.