Polymorphism and Its Influence on Catalytic Activities of Lanthanide–Glutamate–Oxalate Coordination Polymers

Abstract

To study the relationship between polymorphism and catalytic activities of lanthanide coordination polymers in the cycloaddition reactions of CO₂ with epoxides, the monoclinic and triclinic polymorphs of [Ln^III(NH₃–Glu)(ox)]·2H₂O, where Ln^III = La^III (I), Pr^III (II), Nd^III (III), Sm^III(IV), Eu^III (V), Gd^III (VI), Tb^III (VII), and Dy^III (VIII), NH₃–Glu^– = NH₃⁺ containing glutamate, and ox^2– = oxalate, were synthesized and characterized. Factors determining polymorphic preference, the discrepancy between the two polymorphic framework structures, potential acidic and basic sites, thermal and chemical stabilities, active surface areas, void volumes, CO₂ sorption/desorption isotherms, and temperature-programmed desorption of NH₃ and CO₂ are comparatively presented. Based on the cycloaddition of CO₂ with epichlorohydrin in the presence of tetrabutylammonium bromide under solvent-free conditions and ambient pressure, catalytic activities of the two polymorphs were evaluated, and the relationship between polymorphism and catalytic performances has been established. Better performances of the monoclinic catalysts have been revealed and rationalized. In addition, the scope of monosubstituted epoxides was experimented and the outstanding performance of the monoclinic catalyst in the cycloaddition reaction of CO₂ with allyl glycidyl ether under ambient pressure has been disclosed.

Short abstract

Monoclinic and triclinic polymorphs of lanthanide–glutamate–oxalate coordination polymers were synthesized and characterized. Single-crystal structures were included. Factors determining polymorphic preference are discussed. Based on the CO₂ cycloaddition reactions with a scope of monosubstituted epoxides, a relationship between polymorphism and catalytic performance has been established.

1. Introduction

To mitigate the rising concentration of atmospheric CO₂ and its adverse effects, CO₂ capture and utilization has emerged as one of the most sustainable outlooks.¹ The use of CO₂ as a C1 feedstock to produce commodity chemicals is particularly appealing, as they comply with the 100% atom economy.^2,3 Among various possibilities, the production of cyclic carbonates from cycloaddition reactions of CO₂ with epoxides has been extensively explored. Because CO₂ is thermodynamically stable and kinetically inert, majority of the research work is on the development of high-performance catalysts for use at or near ambient temperature and pressure.⁴⁻⁷ The recently emerging trend is the research on bifunctional catalysts.⁸ In terms of practicality, thermal and chemical stability as well as an ability to be recycled are also demanded.⁹⁻¹²

Coordination polymers (CPs), which are also known as metal–organic frameworks (MOFs), have been well acknowledged for their catalytic potential.¹³⁻¹⁵ Through appropriate choices of inorganic and organic building units, their structures and properties can be tailor-made at the molecular level. CPs of the transition metals and lanthanides (Ln^III-CPs) have been extensively explored, the latter of which tend to provide better performance at ambient pressure.¹⁶⁻¹⁹ This is partially due to the hard acidic nature of the vacant coordination sites of Ln^III, which can be generated simply through thermal treatment or the so-called activation step.²⁰ At slightly elevated temperatures, the small solvent ligands can be removed to provide the vacant coordination sites. These sites can interact with the epoxide O atom to promote the subsequent nucleophilic attack and ring opening.²¹ As the nucleophile is essential, the presence of nucleophile sources, e.g., tetrabutylammonium bromide (TBABr), tetrapropylammonium bromide (TPABr), and tetramethylammonium bromide (TMABr), in the catalytic reaction is unavoidable.¹⁹ Thus far, reports on the use of Ln^III-CPs without prior activation are noticeably limited and the reported performances are moderate (Table S1).²²⁻²⁶ Ligand substitution is reportedly responsible for the exemption of the prior activation.²²⁻²⁶ It was only recently that the noteworthy performance manifested through the turnover number (TON) and turnover frequency (TOF) of 7142 and 1786 h^–1, respectively, could be achieved from [Eu^III₂(Habtc)₂(DMSO)₄]·DMSO·H₂O (H₄abtc = 3,3′,5,5′-azobenzenetetracarboxylic acid, DMSO = dimethyl sulfoxide).¹⁸ Regarding the organic building units, they also play important role in promoting the catalytic efficiency of CPs. With respect to Ln^III-CPs, the majority of the organic building units are polycarboxylates, e.g., 2-sulfoterephthalic acid, 2-aminoterephthalic acid, and 3,3′,5,5′-azobenzenetetracarboxylic acid.^8,27,28 The introduction of acidic and basic sites, e.g., –SO₃H, –NH₂, –N=N–, and –COOH,^8,18 can be achieved through both appropriate choice of molecular backbones and postmodification.

The catalytic performance of CPs depends not only on the choices of structural building units but also the arrangement of these units in the framework structures.^19,29−31 Contrary to the transition metal-based CPs, the arrangement of these units in Ln^III-CPs can vary significantly depending on subtle changes in the synthesis conditions.^19,29−31 Polymorphism or framework isomerism, which is the ability of a compound to display different framework structures,³² is therefore frequently observed in Ln^III-CPs, although their polymorphic phase transition is often undetected.³²⁻³⁵ Examples of polymorphism in Ln^III-CPs include the 2D and 3D polymorphs of [Ln(Tz*)₃] (Ln = Gd–Lu and Tz*^– = 1,2,3-triazolate),³⁶ the low- and high-temperature polymorphs of [Eu(BDC)(NO₃)(DMF)₂]_n (BDC^2– = benzenedicarboxylate, DMF = dimethylformamide),³² and the α-, β-, and γ-phases of [Ln₂(hfipbb)₃] (Ln = Sc, Y, La–Yb and hfipbb^2– = hexafluoroisopropylidenebisbenzoate).³⁵ In view of crystal engineering, polymorphism allows the investigation of the structure–property relationship, which is crucial to the rational design of crystal structures. Literature reporting the relationship between polymorphism and catalytic activities is nonetheless rare. One of the few examples is the different catalytic activities of two polymorphic hcp-UiO-66 and fcc-UiO-66.³⁷ The better activity of hcp-UiO-66 than fcc-UiO-66 in alcoholysis of styrene oxide was reported to originate from a dissimilar framework assembly. The other example is the higher electrocatalytic activities of the two-dimensional Co-MOF (CTGU-5) than its polymorphic three-dimensional CTGU-6.³⁸ The difference was rationalized by the more open framework structure of CTGU-5 and the presence of coordinating water molecules. Noticeably, there is no previous report on the relationship between polymorphism and catalytic activities of Ln^III-CPs, particularly in cycloaddition reactions of CO₂.

Herein, the monoclinic (m) and triclinic (t) polymorphs of [Ln^III(NH₃–Glu)(ox)]·2H₂O (Ln^III = La^III (I), Pr^III (II), Nd^III (III), Sm^III(IV), Eu^III (V), Gd^III (VI), Tb^III (VII), and Dy^III (VIII), NH₃–Glu^– = NH₃⁺ containing glutamate, ox^2– = oxalate) were synthesized and characterized. Glutamate was chosen due to the presence of the –NH₂ group, which should be beneficial to the catalysis, whereas oxalate, which is a small dicarboxylate with a tendency to adopt a chelating mode of coordination, was selected to promote framework stability. Their single-crystal structures were elucidated from which the distinct structural differences have been revealed. Thermal and chemical robustness, Brunauer–Emmett–Teller (BET) active surface areas, framework void volumes, CO₂ sorption/desorption isotherms, and temperature-programmed desorption of NH₃ and CO₂ were studied. Their catalytic activities in the cycloaddition reactions of CO₂ and a scope of monosubstituted epoxides under solvent-free conditions and atmospheric CO₂ pressure were then explored. Founded on experimental evidence, the relationship between polymorphism and catalytic performance has been established. In addition, a substantial catalytic performance in the reaction with allyl glycidyl ether (AGE) was demonstrated.

2. Experimental Section

Details of materials, characterization methods, experimental setting, and single-crystal structure determination³⁹⁻⁴⁴ are provided as the Supporting Information. Crystallographic data and refinement details are summarized in Table S2.

2.1. Synthesis of I–VIII

The corresponding Ln^III(NO₃)₃·6H₂O (0.100 mmol) was dissolved in 3.00 mL of deionized water to give solution A. Separately, monosodium glutamate (NaHGlu, 0.100 mmol, 16.9 mg) and di-ammonium oxalate [(NH₄)₂ox, 0.100 mmol, 14.2 mg] were dissolved in 7.00 mL of deionized water to give solution B. Solutions A and B were then mixed in a 10 mL glass vial, closed carefully, and kept in a hot-air oven at a specified temperature and time as summarized in Table 1.

Table 1. Summary of Synthesis Conditions and Corresponding Products with Their Percentage Yields (% Yield) and CHN Analysis Results.

	synthesis conditions						elemental analysis
	60 °C, 4 days		80 °C, 4 days		80 °C, 1 days		C (% wt)		H (% wt)		N (% wt)
LnIII source (weight used in synthesis)	product	% yield	product	% yield	product	% yield	Calcda	Expb	Calcda	Expb	Calcda	Expb
LaIII(NO3)3·6H2O (43.3 mg)	It	76	LaIII-oxc		LaIII-oxc		20.6	20.7	2.70	2.90	3.43	3.43
PrIII(NO3)3·6H2O (43.5 mg)	IIt	79	IIt	77	IIt	77	20.5	20.8	2.68	2.86	3.41	3.46
NdIII(NO3)3·6H2O (43.8 mg)	IIIt + IIIm		IIIt	76	IIIt	76	20.3	20.8	2.66	2.87	3.39	3.45
SmIII(NO3)3·6H2O (44.4 mg)	IVm	78	IVt	75	IVt	78	20.0	21.1	2.62	3.21	3.34	3.77
EuIII(NO3)3·6H2O (44.6 mg)	Vm	76	Vt + EuIII-oxc		Vt	78	19.9	19.8	2.61	3.11	3.32	3.37
GdIII(NO3)3·6H2O (45.1 mg)	VIm	78	GdIII-oxc		VIt	76	19.7	19.7	2.56	3.06	3.28	3.52
TbIII(NO3)3·6H2O (45.3 mg)	VIIm	79	TbIII-oxc		VIIt	77	19.6	19.4	2.57	2.94	3.27	3.44
DyIII(NO3)3·6H2O (45.7 mg)	VIIIm	79	DyIII-oxc		VIIIt + DyIII-oxc		19.4	17.9	2.55	2.63	3.24	3.07

^aTwo crystallizing water molecules per formula, i.e., [Ln^III(NH₃–Glu)(ox)]·2H₂O, which is the most consistent with the CHN analysis results, were assumed in the calculations.

^bThe experiments were carried out using pure samples.

^cBased on PXRD experiments, they were isostructure to [Nd^III₂(ox)₃(H₂O)₂]·H₂O.⁴⁵

2.2. Catalytic Experiments

The crystals were ground manually using a mortar and pestle and used as is without particle size control. Typically, the tested catalyst was weighed into a 10 mL two-neck round-bottom flask into which epichlorohydrin (ECH, 20 mmol) and TBABr (0.25 mmol) were subsequently added. The flask was then sealed and equipped with a CO₂ balloon. The catalytic reaction was carried out under atmospheric CO₂ pressure and solvent-free conditions at 80 °C for 3.5–4 h with continuous stirring and regular filling of CO₂. To stop the reaction, the flask was cooled down to room temperature using an ice bath, and the remaining CO₂ was slowly released. Progress of the reactions was followed by using ¹H NMR spectroscopy. The triplicate results were produced for every experiment unless specified otherwise. Yields of the reactions were determined using mesitylene as an internal standard. In addition to ECH, 1,2-epoxyoctane (1,2-EO), glycidol (Gly), butyl glycidyl ether (BGE), benzyl glycidyl ether (BnGE), 1,2-epoxy-3-phenoxypropane (EPP), styrene oxide (SO), and AGE were examined.

Recyclability and robustness of the catalyst were additionally studied using Vm and ECH as the modeled catalyst and epoxide. To regenerate the catalyst, it was recovered from the reaction by filtration followed by washing with CHCl₃ several times, drying in air, and regrinding. Several batches were carried out for each cycle, particularly the earlier ones to ensure that the recovered catalysts from each cycle were sufficient for further experiments until the 10th cycle. Stability of the samples was assured using powder X-ray diffraction (PXRD) experiments.

3. Results and Discussion

3.1. Synthesis and Single-Crystal Structures of I–VIII

According to the PXRD experiments (Figure 1) and single-crystal X-ray diffraction data (Table S2), the triclinic (t) and monoclinic (m) polymorphs of I–VIII could be synthesized as pure phases, although their purity depended on the synthesis temperature and time. At 60 °C and 4 days, the monoclinic phases were mostly yielded except for La^III (I), Pr^III (II), and Nd^III (III), which provided the triclinic phrases. Increasing the temperature from 60 to 80 °C at the same synthesis time of 4 days resulted in the formation of the triclinic phases for almost all Ln^III except II (Pr^III), III (Nd^III), and IV (Sm^III), which provided the oxalate frameworks isostructural to [Nd^III₂(ox)₃(H₂O)₂]·H₂O.⁴⁵ Shortening of the synthesis time from 4 days to 1 day at 80 °C led to the formation of the triclinic phases for also almost every Ln^III except La^II (I). According to these results, the monoclinic frameworks can be considered as the low-temperature phases while the triclinic and the oxalate frameworks are the high-temperature counterparts. Notably, the crystals of monoclinic samples were substantially larger than those of the triclinic samples (Figure S1). In addition, the synergy between the synthesis temperature and time with lanthanide contraction can be assumed. While the light Ln^III, i.e., La^III (I), Pr^III (II), and Nd^III (III), preferred the triclinic structures, the heavier Ln^III, i.e., Sm^III(IV), Eu^III (V), Gd^III (VI), Tb^III (VII), and Dy^III (VIII), showed a tendency to provide both the low- and high-temperature phases.

Figure 1.

PXRD patterns of the as-synthesized I–VIII yielded from the reactions conducted at (a) 60 °C for 4 days, (b) 80 °C for 4 days, and (c) 80 °C for 1 day, compared with the simulated patterns of It, Vm, and [Nd^III₂(ox)₃(H₂O)₂]·H₂O⁴⁵

3.1.1. Description of Triclinic Structures

The single-crystal structure is described based only on the data of It because the crystals of IIt–VIIt were inappropriate for full data collection. Their structures were alternatively concluded from a preliminary check of the unit cell parameters and PXRD experiments.

The asymmetric unit of It comprises one crystallographically unique La^III, two halves of independent ox^2–, and a molecule of glutamate with its amine group protonated (NH₃–Glu^–) (Figure 2). The protonation of the –NH₂ group has been expected because of the acidic synthesis condition. The presence of the Bronsted –NH₃ group should be beneficial to the catalysis. The N atom of NH₃–Glu^– is disordered over two crystallographic sites with approximately equal site occupancy. The carbon skeleton of NH₃–Glu^– is arranged with approximately identical ∠C1–C2–C3–C4 (168°) and ∠C2–C3–C4–C5 (169°) torsion angles and the C1···C5 distance of ca. 4.9 Å (Figure S2). Viewed along the molecular skeleton, the two terminal carboxyl groups almost coincide with each other, reflected through the small ∠O1–C1–C5–O4 (9.0°) and ∠O2–C1–C5–O3 (4.8°) torsion angles. All atoms of ox^2– are notably aligned in the molecular plane.

Figure 2.

Views of (a) an extended asymmetric unit, the (b) local coordination environment of the edge-shared TPRS-{La^IIIO₈}₂ dimer and coordination modes of the organic linkers, the (c) three-dimensional framework of It, and the (d) corresponding net topology (hydrogen atoms are omitted for clarity. Symmetry operations: (i) −x, 1 – y, 1 – z; (ii) −x, −y, 1 – z; (iii) −x, 1 – y, −z; (iv) −1 + x, y, z).

The La^III is 9-fold coordinated to two ox^2– and four NH₃–Glu^–. The O atoms define a distorted tricapped trigonal prismatic coordination geometry and therefore the TPRS-{La^IIIO₉} coordination entity (Figure 2). The two ox^2– are positioned next to each other, making a dihedral angle of ca. 81°. Through the –O– and –OCO– bridges of four NH₃–Glu^–, every two TPRS-{La^IIIO₉} coordination entities are fastened into an edge-shared TPRS-{La^IIIO₈}₂ dimer. Each of the dimeric units is further linked to the other six equivalents via ox^2– (μ₂:η¹η¹η¹η¹) and NH₃–Glu^– (μ₄:η²η¹η¹η¹). The linkage results in a 3D framework of [Ln^III(NH₃–Glu)(ox)]. As anticipated, ox^– stabilizes the framework by using its commonly adopted chelating coordination mode. The framework contains ca. 17% void calculated using PLATON.⁴⁶ This corresponds to the 1D channel with an opening of 2.6 × 3.9 Ų. Within the channel, two crystallizing water molecules (O1w and O2w) are located. These water molecules show site disordering attributable to their involvement in the hydrogen bonding interactions with –NH₃⁺ of NH₃–Glu^– and the carboxyl O atoms of both NH₃–Glu^– and ox^2– (Table S3).

3.1.2. Description of Monoclinic Structures

The single-crystal data of IVm–VIIIm were elucidated (Table S2), revealing their isostructures and similar asymmetric structures to It. Their asymmetric structures comprise one unique Ln^III, two halves of independent ox^2–, and a molecule of NH₃–Glu^– (Figure 3). The structural building units are also identical to those of It, which are the edge-shared TPRS-{Ln^IIIO₈}₂ dimer and the organic linkers of the same coordination modes. In addition, the derived 3D frameworks are closely similar to It from a quick glance.

Figure 3.

Views of an (a) extended asymmetric unit, the (b) local coordination environment of the edge-shared TPRS-{Eu^IIIO₈}₂ dimer and coordination modes of the organic linkers, (c) three-dimensional framework of Vm, and the (d) corresponding net topology (hydrogen atoms are omitted for clarity). Symmetry operations: (i) 1 – x, 2 – y, 1 – z; (ii) 1 – x, y, 0.5 – z; (iii) 1.5 – x, 1.5 – y, 1 – z; (iv) x, 1 – y, −0.5 + z; (v) 1.5 – x, 0.5 + y, 1.5 – z).

Discrepancies between the two polymorphic framework structures are, however, distinct and principally originate from the organic linkers. First, the terminal carboxyl groups of NH₃–Glu^– in Vm (as a representative of IVm–VIIIm) do not coincide with each other (Figure S2). The ∠O1–C1–C5–O4 (39.4°) and ∠O2–C1–C5–O3 (31.7°) torsion angles are large. This is contrary to It case although the arrangement of their molecular skeleton is similar, i.e., ∠C1–C2–C3–C4 = 180°, ∠C2–C3–C4–C5 = 176°, and the C1···C5 distance = ca. 5.0 Å in Vm. Second, the carboxyl O6 atom of one ox^2– is not coplanar with the other atoms but slightly projecting out of the molecular plane (Figure S2). Third, the dihedral angle between two ox^2– of Vm is ca. 75° (Figure 3a), which is slightly smaller than that of It (Figure 2a). Fourth and probably the most important is the spatial arrangement of four NH₃–Glu^– linkers about the TPRS-{Ln^IIIO₈}₂ dimer. Each dimer of It is linked to six equivalents (Figure 2b) whereas that of Vm (Figure 3b) is connected to eight. This difference is reflected explicitly in the simplified framework topologies. Using the TPRS-{Ln^IIIO₈}₂ dimer as a node, the framework structure of It can be simplified to a 6-connected pcu or sqc1 net with a 4¹².6³ point symbol (Figure 2d) while that of Vm to an 8-connected sqc117 net with a 3⁶.4¹².5⁸.6² point symbol (Figure 3d). It is therefore conclusive that the structural discrepancy between the two polymorphic frameworks is genuine and driven by dissimilarity in the spatial conformation of the organic linkers and their linking patterns.

Calculated using PLATON,⁴⁶ the framework void of Vm is ca. 19%. The 1D channel with an aperture of 4.2 × 5.9 Ų is accountable for this void. Noticeably, these values are slightly larger than those of It. The crystallizing water molecules residing inside the channel are extremely disordered and could not be precisely located from the single-crystal data. This is also contrary to the It case. The treatments of the single-crystal data using SQUEEZE routine⁴⁷ were necessary, resulting in the estimation of 2.2 to 6.6 water molecules per formula (Table S2). These numbers were, however, larger than the numbers derived from the elemental analyses (ca. 2 water molecules). The thermogravimetric/differential thermal analysis (TG/DTA) experiments of It and Vm (as the representatives of the two polymorphs) were therefore conducted (Figures S3 and S4) from which the assumption of 2 water molecules per formula was concluded. Identical chemical formulas of both the triclinic and monoclinic structures, i.e., [Ln^III(NH₃–Glu)(ox)]·2H₂O, were accordingly decided. The Fourier transform infrared (FT-IR) spectra displayed the characteristic vibrational features of the expected functional groups (Figure S5 and Table S4).

3.2. Thermal Robustness of It and Vm

Based on the TG/DTA experiments, similar weight losses were observed for both It and Vm, which were used as the representatives of the two polymorphs (Figures S3, S4). The patterns showed two major weight losses alike. The first weight loss occurred over the 150–220 °C range, signifying the removal of the crystallizing water molecules, followed by the second substantial loss at ca. 350 °C. Disregarding similarity in the weight loss patterns, the PXRD experiments suggested otherwise (Figure S6). The framework structure of It remained intact even after heating at 350 °C whereas that of Vm started to transform to other phases after heating at 220 °C. Better thermal robustness of the triclinic framework, whose structural building units are organized more orderly, can therefore be assumed. At 80 °C, which would be the temperature used in the catalysis experiments, the stability of the two polymorphs was nonetheless certain because the framework structure of Vm remained stable up to 150 °C.

3.3. Chemical Robustness of Vt and Vm

Since chemical robustness is an important property of catalysts, the stabilities of Vt (Eu^III) and Vm (Eu^III), which would be employed in the catalysis experiments to avoid any influence from different Ln^III, in water and common organic solvents were studied. The experimented organic solvents were toluene, tetrahydrofuran (THF), hexane, methyl alcohol (MeOH), ethyl alcohol (EtOH), ethyl acetate (EtOAc), acetonitrile, chloroform (CHCl₃), dichloromethane (CH₂Cl₂), benzene, and acetone. After soaking the ground crystals of Vt (Eu^III) and Vm (Eu^III) in the solvents for 1 day at room temperature, their PXRD patterns were collected and showed the retaining of all the diffraction peaks (Figure S7). The robustness of both Vt and Vm toward these common solvents at room temperature is therefore conclusive.

3.4. Sorption/Desorption Isotherms of N₂ and CO₂ of Vt and Vm

The N₂ sorption/desorption isotherms of Vt and Vm are similar (Figure S8), exhibiting the characteristic profiles of microporous frameworks. However, the BET surface area and void volume of Vm (5.037 m²·g^–1 and 1.157 cm³·g^–1) were approximately 3 times larger than those of Vt (1.566 m²·g^–1 and 0.3599 cm³·g^–1) based on the measurement at 77 K. These results are seemingly contradicting with the slight difference in the calculated voids of the two polymorphs. This contradiction, on the other hand, implies the better access of N₂ into the framework of Vm than Vt. This implication is in good accordance with the larger channel aperture of Vm (4.2 × 5.9 Ų) relative to that of It (2.6 × 3.9 Ų). With respect to the kinetic diameter of N₂ (3.64 Å),⁴⁸ N₂ should not be able to enter the It framework.

Intriguingly, the CO₂ uptake capacities of Vm (1.5163 mL·g^–1 at 195 K and 1.4419 mL·g^–1 at 298 K) were also approximately 3 times larger than those of Vt (0.39601 mL·g^–1 at 195 K and 0.33431 mL·g^–1 at 298 K) (Figure S9). Considering the kinetic diameter of CO₂ (3.30 Å),⁴⁸ these substantial differences clearly demonstrate the molecular sieve property of the frameworks. Furthermore, both polymorphs showed the retention of CO₂ upon the desorption (Figure S9), suggesting the strong interaction between CO₂ and the frameworks. The interaction may occur at both the acidic –NH₃⁺ group and the lone-pair electron-containing O atoms of the carboxyl groups. If the desorption isotherms at 195 and 298 K are compared, then the dissimilarity of the two polymorphs is magnified. While the retention of CO₂ at 298 K of Vt was significantly less than that of 195 K, the retention of CO₂ of Vm at both temperatures was approximately equal. This suggests a stronger interaction of CO₂ with Vm than Vt. In line with these results, a better catalytic performance should be expected from Vm.

3.5. Framework Acidity and Basicity

The acidity and basicity of Vt and Vm were additionally evaluated using the temperature-programmed desorption of NH₃ (NH₃-TPD) and CO₂ (CO₂-TPD) (Figure S10). The NH₃-TPD profile of Vt showed a vivid desorption peak at a lower temperature (ca. 147 °C) than that of Vm, which showed only a small shoulder at ca. 194 °C. A tendency to desorb at temperatures higher than 200 °C can however be assumed. Complying with these results, stronger interaction with NH₃ and therefore stronger acidity should be expected from Vm. This is considered to be beneficial to its catalytic function.

The CO₂-TPD experiments did not provide any significant information as the desorption of CO₂ commenced at ca. 140 °C, which is close to the limitation of the employed instrument (200 °C). The desorption of CO₂ from Vt was nonetheless more significant than Vm up to 200 °C. Projecting to temperatures higher than 200 °C, the release of CO₂ from both Vt and Vm can be hypothesized. This implies strong interaction between CO₂ and the frameworks especially in the Vm case, which are in good accordance with the CO₂ sorption/desorption isotherms.

3.6. Catalytic Activities of Vt and Vm

Catalytic activities of Vt and Vm were evaluated based on the CO₂ cycloaddition with ECH in the presence of TBABr (Table 2). The catalysts were used as powders and without prior activation unless stated otherwise. The amount used in each batch of the experiments was notably diminutive, i.e., 0.025 mol % (ca. 2.1 mg) with respect to the ECH substrate, and 1/50 of TBABr (1.25 mol % or 80.5 mg). As references, the catalytic activities of TBABr (without the catalysts, entry 1), the catalysts (without TBABr, entries 2 and 3) and Eu(NO₃)·6H₂O (entry 4) were also determined. Based on the integrated areas of ¹H NMR spectra, the performance of the catalysts was evaluated via the percentage conversion of ECH to chloromethyl ethylene carbonate (%conversion) and selectivity (%selectivity). The byproduct from the reactions was 1-bromo-3-chloro-2-propanol. Exemplified ¹H NMR spectra are provided as the Supporting Information (Figures S11–S27). Percentage yields of the products (%yield) were also determined by using mesitylene as an internal standard. The TON and TOF values were then calculated based on either % conversion and % selectivity or % yield.

1

2

3

Table 2. Catalytic Performance of Non-activated Catalysts (unless Specified Otherwise) in Catalyzing CO₂ Cycloaddition with ECH at 80 °C for 3.5 h under Ambient CO₂ Pressure and Solvent-less Conditions.

entry	catalyst	% conversiona	% yield	% selectivitya	TONb	TOF (h–1)b
1	None (use only TBABr)	53(±1)		96
2	Vm (without TBABr)	trace
3	Vt (without TBABr)	trace
4	Eu(NO3)3·6H2O	79	74	96	2960	846
5	IVm	80	77	97	3080	880
6	Vm	84	84	95	3360	960
7	Vmc	81	81	95	3240	926
8	Vm crystal	79	77	95	3080	880
9	VIm	81	79	97	3160	903
10	VIIm	81	81	97	3240	926
11	VIIIm	82	82	94	3280	937
12	It	76	74	95	2960	846
13	IIt	76	74	95	2960	846
14	IIIt	75	73	96	2920	834
15	IVt	77	74	95	2960	846
16	Vt	77	74	94	2960	846
17	Vtc	77	73	95	2920	834
18	VIt	79	77	95	3080	880
19	VIIt	78	77	95	3080	880

^aAveraged from triplicate results with ±1 to ±3 standard deviation (based on ¹H NMR results).

^bCalculated from % yield:

^cWith prior activation (soaking in MeOH overnight and then heating at 80 °C for 30 min under N₂).

At 80 °C and 3.5 h, IVm–VIIIm (entry 5, 6, 9–11) provided comparable activities, i.e., 80–84% conversion with 94–97% selectivity. The %yields of 77–84 were achieved, and the corresponding TON and TOF were 3080–3360 and 880–960 h^–1, respectively. Similarly, It–VIIt (entry 12–16, 18, 19) showed indistinguishable results, i.e., 75–79% conversion, 94–96% selectivity, and 73–77% yields, which correspond to the TON and TOF of 2920–3080 and 834–880 h^–1, respectively. The absence of lanthanide contraction can thus be presumed. Intriguingly, these values are in the same range as the values obtained from Eu(NO₃)₂·6H₂O. The salt was, however, completely soluble under the experimented condition and therefore acted as a homogeneous catalyst. As the existing species of Eu(NO₃)₂·6H₂O in an aqueous medium is [Eu^III(H₂O)_n]³⁺, the catalysis may involve directly with Eu^III through the substitution of the hydrating water molecules with the epoxide,^18,22−26 which is contrary to the Ln^III-CPs reported herein.

Since all of the catalysts were experimented with without prior activation and there is no small ligand coordinating with Ln^III, the catalysis should proceed without any direct involvement with Ln^III. This assumption is consistent with the fact that closely similar results were observed from the same polymorphic catalysts. Additionally, the attempt to activate the catalyst before use (entries 7 and 17) did not lead to any significant differentiation. Accordingly, the important functional groups in the catalysis should be the Bronsted acidic –NH₃⁺ group of NH₃–Glu^– and the lone-pair electron-containing O atoms of the organic linkers (Figures 4 and S28). The epoxide O atom may interact with the Bronsted acidic –NH₃⁺ group of NH₃–Glu^–, inducing the attack of the nucleophilic Br^– provided from TBABr. The CO₂ bearing both negative (on O atoms) and positive (on C atom) partial charges can interact with both the nearby carboxylate O atom and the Bronsted acidic –NH₃⁺. The proximity of the activated intermediate and CO₂ can assumingly promote the subsequent CO₂ addition and ring closing.

Figure 4.

Locations of potential acidic (blue) and basic (red) sites in the frameworks of (a) Vt and (b) Vm and their neighboring environment.

Because the active surface area can be influential to catalytic activities, the catalytic activity of Vm crystals (entry 8) was assessed. The crystals of Vm were selected owing to their large sizes. Compared with the activity of the ground crystals (entry 6), it is apparent that the conversion yielded from the crystals was slightly inferior. The effect of surface area has therefore manifested. Contrarily, this effect was suggested to be insignificant from the small standard deviations of the averaged conversions and selectivity in the cases of the ground crystals although their sizes were not carefully controlled (Figure S1).

If the two polymorphs are compared (Figure 5), the monoclinic polymorphs evidently offer slightly better activities. The performance of Vm (entry 6, 84% yield) was, for instance, better than that of Vt (entry 16, 74% yield). This can be accounted for by the larger surface area and void volume as well as the stronger acidity and basicity of Vm. Since these differences are brought about through structural dissimilarity between the two polymorphs, the relationship between polymorphism and catalytic activity has been demonstrated.

Figure 5.

Comparison of (a) percentage yields and (b) TOF of two polymorphic catalysts based on the CO₂ cycloaddition reactions with ECH at 80 °C for 3.5 h under ambient CO₂ pressure.

3.7. Catalytic Performance of Vm

Because Vm showed the best performance in catalyzing the reaction with ECH, further experiments were performed using this catalyst. The prolongation of the reaction from 3.5 to 4 h at 80 °C (Table 3) led to the improved conversion and TON value although the TOF value (863 h^–1) was slightly reduced as a penalty of the longer reaction time. Compared with the other Ln^III-CP-based catalysts (Table S1), the performances of Vm were nonetheless remarkable.

Table 3. Catalytic Performance of Non-activated Vm in the CO₂ Cycloaddition with Various Epoxides at 80 °C for 4 h under Ambient CO₂ Pressure and Solvent-less Conditions.

^aCould not be calculated due to the unidentified byproduct.

^bAverage values from triplicate results.

^cCalculated from the average % conversion and % selectivity:

Vm could also be easily recovered and reused. Over 10 successive cycles of catalysis and regeneration, the conversions and selectivity provided by Vm were approximately unaltered, varying between 83–85 and 94–96%, respectively (Figure S29a). Its robustness was demonstrated via the preservation of the framework structure and crystallinity according to the PXRD experiments (Figure S29b).

In addition to ECH, Vm was additionally attempted with the other monosubstituted epoxides (Table 3 and Figures S30–S36). While poor to moderate results were observed from the other epoxides, 67% conversion with an absolute selectivity could be yielded from AGE. The corresponding TON and TOF values of 2680 and 670 h^–1, respectively, were inspiring. The superior catalytic efficiency in the reaction of AGE to those of SO, EPP, and BnGE may be attributed to steric hindrance imparted from the bulky substituted chains of SO, EPP, and BnGE. Compared with 1,2-EO and BGE possessing similar structures, the presence of electron-rich moieties on the substituted chain of AGE may be responsible for its superiority in a similar manner to the presence of halide in ECH.^49,50

The attempt to elevate the reaction temperature from 80 to 90 °C led to a leap of the conversion to 84% with the improved TON and TOF values to 3360 and 840 h^–1, respectively. With reference to mesitylene, an excellent yield of 83% was also affirmed. Compared with the other Ln^III-CPs operated under similar conditions (Table S5), the performance of Vm was assuredly remarkable and superior to the other previously reported Ln^III-CPs. It was literally comparable to those of the heterometallic 3d-4f CPs, e.g., [Dy₂Zn₄(μ₃-OH)₂L₄(AcO)₂(NO₃)₂(DMF)₂]·2(CH₃OH) (TOF = 783 h^–1),⁵¹ Tb₂Zn₂(μ₃-OH)₂L₄(NO₃)₄ (TOF = 700 h^–1), and Tb₂Zn₄(μ₃-OH)₂L₄(OAc)₆(NO₃)₂ (TOF = 668 h^–1).⁵² This is a rare example of the efficient catalysis of the CO₂ cycloaddition with AGE, which does not involve the vacant coordination sites of the metal but the other acidic and basic sites on the organic linkers, i.e., the Bronsted–NH₃⁺ group of NH₃–Glu^– and the basic O atoms of the carboxyl groups in this study.

4. Conclusions

The low-temperature monoclinic (m) and high-temperature triclinic (t) polymorphs of I–VIII were synthesized and characterized. The synergistic influences of the lanthanide contraction and synthesis parameters on the formation of these polymorphic frameworks were described. The thermodynamically stable oxalate frameworks were identified, in addition to the two polymorphs. Based on the single-crystal data, similarities and dissimilarities of the two polymorphic framework structures were presented. These dissimilarities were apparently manifested in diverse thermal stability but not chemical robustness. With a larger aperture into the framework void, the monoclinic polymorph (represented by Vm) allowed better accessibility to both N₂ and CO₂ and therefore larger BET surface area and effective void volume than the triclinic polymorph (represented by Vt). Stronger interactions with NH₃ and CO₂ implying stronger acidity and basicity could also be concluded for the monoclinic polymorph. The as-described structural features and properties of the monoclinic polymorphs evidently led to superiority in catalytic performance to the triclinic polymorphs. The Bronsted acidic–NH₃⁺ and Lewis basic O atoms of the organic linkers are responsible for effective catalysis. This assumption is supported by the independence of the catalytic activities on the lanthanide contraction and the prior activation. Based on the experiments using Vm, the catalyst could be reused for at least 10 cycles without deterioration in its crystallinity and performance. In addition to ECH, Vm also exhibited inspiring catalytic activity in the reaction of AGE.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.inorgchem.4c00095.

• Details on materials and methods, crystallographic data, and other results including additional crystallographic presentations, FT-IR spectra and band assignments, ¹H NMR spectra, TD/DTA results, hydrogen bonding interactions, sorption/desorption isotherms of N₂ and CO₂, NH₃-TPD, CO₂-TPD, PXRD spectra, and additional catalysis data (PDF)

The authors declare no competing financial interest.