Two-Effects-in-One: Kinetic Quantum and Molecular Sieving Effect in a Robust-Flexible [Th6Co2] Cage for both Isotope and Isomer Separation

Although molecular and quantum sieving (MS and QS) effects have been realized and employed to execute separation tasks, however, little is known about the coexistence of both effects in a compound and its corresponding structure. In this work, we show the rst observation of coexistence of both quantum and molecular sieving effects in a [Th6Co2] cage-based compound (Th-Co-Cage-1). The [Th6Co2] cage shows 0.78 nm aperture but with an irregular crescent-like window. This allows [Th6Co2] cage to give ultrahigh uptake for hydrogen isotope and selectivity towards D2 over H2, leading to complete hydrogen isotope separation, as evidenced by experimental breakthrough test. When the size of guest molecule is more than the crescent-like window, a highly rare robust-exible adsorption was observed, consequently leading to complete isomer separation for C4 and C6 isomers through molecular sieving effect. High thermal, water and chemical stability further supports the materials for practical separation application. The results open up a gate of material with coexistence of QS and MS effects and its fundamental design for superior separation application.


Introduction
Cages involved in organic cages and coordination-driven cages containing inherent inner porosity present a highly desired porous material for host-guest recognization and separation. [1][2][3][4][5][6][7] The pore of cages is also very special and composed of two major factors, viz. inner aperture that generally related to the accommodating capacity for guest molecules, and window size that determines the entrance size for guest molecules. Accordingly, cages could be a good candidate to perform molecular sieving effect for separation. When the window size becomes small enough to generate a con nement system, then the kinetic quantum sieving (KQS) effect was observed. 8,9 Separation is an important chemical industrial process and now provides almost all purity matters for us from corresponding mixture. 10 The mature technique for separation is based on boiling-point-different distillation method, however, which not only requires huge capital investment, but also consume extensive energy, accounting for 10-15% of energy consumed worldwide. 10 By contrast, porous adsorbent based pathway is relatively low-cost and energy-saving and is now receiving increasing attention. [11][12][13][14][15][16][17] And several porous adsorbents involving in traditional zeolite and porous carbon, and newly developed metal-organic framework, cage compounds, hydrogen-bonded framework, and covalent organic frameworks have been attested to be effective for separating certain mixtures such as CO 2 /N 2 , C 2 H 2 /CO 2 , C 2 H 2 /C 2 H 4 , C 2 H 4 /C 2 H 6 , SO 2 /CO 2 , and Xe/Kr. [18][19][20][21][22][23][24] Among these separations, both isotope and isomer separation are highly desirable but extremely challenging, due to their urgent and plentiful demand and nearly identical physicochemical properties. For example, deuterium (D), an isotope isomer of H, is of prime importance in utilizing nuclear fusion reactors, whereas isomer separation such C 6 separation is of extreme importance in upgrading gasoline.
Both of them play an important role in resolving energy crisis.
As suggested by theoretical and experimental results, production of D from D/H separation upon porous adsorbents can be overcome through both kinetic quantum sieving (KQS) and chemical a nity quantum sieving (CAQS) effect. 8,9,[25][26][27][28][29][30][31][32] KQS just occurs within a very small con ned space or entrance, which can then lead to different transport between D and H, due to their difference in de Broglie wavelength. Correspondingly, CAQS is generally dominated by open-metal site that affords distinct a nity towards D and H. Due to the extremely small size of D 2 and H 2 (dynamic diameter 2.89 Å), KQS often requires ultra ne pore with aperture or entrance around 3 Å (Scheme 1), however, this typically leads to materials with low pore volumes, and consequently low D 2 adsorption capacity, thus making such processes di cult in industrial application.
On the other hand, isomer separation in principle can be gured out upon the discrepancy in molecular size of isomers through molecular sieving (MS) effects of porous adsorbents. 33 As we know, the smallest molecule with isomer phenomenon is C 4 molecules, for example, n-butane vs. iso-butane, where n-butane and iso-butane give dynamic diameter of 4.7 Å and 5.3 Å, respectively. Thereby, generally isomer separation via MS effect requires microporous pore with aperture or entrance more than 4.5 Å (Scheme 1). Interestingly, when the pore meets the criterion of MS for isomer separation, however it will largely reduce and even eliminate KQS, and vice versa, and thereby, there brings an interestingly scienti c concern of whether KQS and MS can coexist in one material.
To answer this concern, we synthesized a [Th 6 Co 2 ] cage compound, which shows a special crescent-like window with continuous varying size, but big inner aperture, meeting the dual standard for both KQS and MS, as well as industrial practice demand (big pore volume along with high adsorption capacity). Both single component adsorption tests and breakthrough experiments con rm its superior application in D/H, and C 4 , C 6 isomer separation.

Results
Th-Co-Cage-1 ([Th 2 Co 2 L 6 ], L 2-=5-(1H-imidazol-1-yl)isophthalate) was synthesized by the solvothermal reaction of Th(NO 3 ) 4 , Co(NO 3 ) 2 , and L in DMF with present of HNO 3 at 120°C. The yield is up to 90% based on Th. Single crystal X-ray diffraction reveals that Th-Co-Cage-1 crystallizes in the rhombohedral crystal system with R-3c space group. In the asymmetry unit there is one crystallography-independent Th and Co site. The Th(IV) site is coordinated by nine carboxylate oxygen atoms from six L 2ligands, while the Co(II) site holds the common octahedral geometry nished by three carboxylate oxygen atoms from three L 2ligands and three nitrogen atoms from other three L 2ligands. Each L 2ligand connects to two Th(IV) ions and two Co(II) ions with one carboxylate in the chelate mode and the other one in the bridging mode. The secondary building unit in Th-Co-Cage-1 is the unique [Th 6 Co 2 ] nanocage ( Figure 1a, 1b, 1c), showing an elongated hexagonal bipyramid con guration, nished by six Th(IV) ions creating hexagon, and two Co(II) ions as vertex up and down. The effective aperture excluded the van der Waals Radii of atoms is ca. 7.8 Å. Each cobalt vertex connects to the middle hexagon through three L 2ligands. Due to the non-symmetrical L 2ligands plus big distortion of imidazol fragment of L 2ligands (dihedral angle of ca. 39° between imidazol and phenyl sections), an irregular crescent-like window showing continuous varying size was observed for this [Th 6 Co 2 ] cage. The three-dimensional structure was generated through co-sharing two Th(IV) ions among two adjacent cages ( Figure 1c). The topology matrix can be obtained by considering each CoTh dinuclear (connected by three carboxylate groups of L 2ligands) and L 2ligand to be nine-connected node ( Figure S1) and three-connected node ( Figure S2), respectively, resulting in an overall ftw-type topology (Figure 1d). 34 The solvent-accessible volume is estimated to be 41.5% of the cell volume. 35 Impressively, as discussed above, the unique structure feature in this [Th 6 Co 2 ] cage showing big cavity but irregular crescent-like window with continuous varying size strongly suggests its potential with coexistence of both KQS and MS and thus superior applications in both isotope and isomer separation.
The thermal stability of Th-Co-Cage-1 was initially investigated by thermogravimetric analysis (TG), giving the loss of solvent molecules before 340 °C ( Figure S3). Notably, this MOF displays ultrahigh thermal and chemical stability (Figure 2), for example calcination at 350 °C for 24 h, immersion in water or boiling water for 7 d, as well as immersion in acidic or alkaline solution at pH=3-12 for 24 h. After methanol exchange the activated samples were obtained by 150°C under vacuum. A typical type I pro le of N 2 adsorption at 77 K was observed for the activated samples ( Figure S4). The BET surface area and pore volume is 717 m 2 /g and 0.29 cm 3 /g. A narrow pore size distribution at 0.76 nm is observed, well consistent with the aperture (0.78 nm) of [Th 6 Co 2 ] cage estimated from structure.
To con rm our claim, we rst explored isotope separation such as D 2 /H 2 separation. As shown in Figure 3a, Th-Co-Cage-1 enables ultrahigh D 2 adsorption capacity up to 208 cm 3 /g at 1 bar and 77 K, respectively, exceeding benchmark porous adsorbents for D/H separation such as zeolite 5A (91.8 cm 3 /g), 36 porous organic cage of cocryst1 (107 cm 3 /g), 8,9 and MFU-4I (186 cm 3 /g). 37 By contrast, H 2 uptake at similar condition is 186 cm 3 /g, lower than D 2 uptake about 22 cm 3 /g, indicative of selective adsorption of D 2 than H 2 , mainly due to that the heavier isotope of D 2 has lower zero energy and smaller de Broglie wavelength as compared to H 2 . Notably, this extremely high adsorption capacity will be very important for industrial large-scale D 2 /H 2 separation. To evaluate the a nity between cage and hydrogen isotopes, we further carried out calculation of adsorption enthalpies (Q st ) for D 2 28 To further con rm the D 2 /H 2 separation ability, dynamic breakthrough experiments upon Th-Co-Cage-1 bed were performed at 77 K and 1 bar for a H 2 /D 2 /Ne (10/10/80, vol.%) mixture. Impressively, H 2 was rst to eluted through the packed column, whereas the retention time for D 2 is about 10 min/g, suggesting complete D 2 /H 2 separation (Figure 3b). The separation ability is comparable with HKUST-1 (10 min/g) at the same conditions, 28 where the separation mechanism results from the CAQS effect from open Cu(II) site. As discussed above, within this [Th 6 Co 2 ] cage, both Th and Co ions are fully coordinated by L 2-ligands, excluding possibility of providing an open metal site for host-guest recognization and thus CAQS effect. Accordingly, KQS effect could be responsible for the D 2 /H 2 separation in Th-Co-Cage-1, and from the viewpoint of structural feature, this claim is reasonable, as this [Th 6 Co 2 ] cage shows an irregular crescent-like window with continuous varying size indeed involved in a very narrow window around 2.9 Å that meets the criterion to occur of KQS effect.
After con rming the validity of isotope separation by Th-Co-Cage-1, we next explore the use of this material for isomer separation. Firstly, C 4 of n-butane and iso-butane was selected based on the consideration of their dynamic diameter of 4.7 Å and 5.3 Å, respectively, where the size of n-butane is comparable with the big window of [Th 6 Co 2 ] cage, whereas the size of iso-butane is bigger than the big window of [Th 6 Co 2 ] cage, thus making a possibility of MS effect for isomer separation. As expected, the single-component adsorption experiments show that the adsorption capacity of n-butane is as high as 3.5 mmol/g, whereas iso-butane is somewhat excluded from [Th 6 Co 2 ] cage with just low uptake of 1.2 mmol/g at 1 bar and 298 K (Figure 3c). The n-butane uptake is far exceeding most reported porous adsorbents such as ZU-36-Co (2.1 mmol/g), 41 Y-fum-fcu-MOFs (2.0 mmol/g), 42 and commercial shaped 5A zeolite (1.3 mmol/g). 41 Seem form the adsorption isotherms, n-butane adsorbed steeply at low pressure below 0.02 bar, then an abrupt junction was observed, and following adsorption also becomes steep, fully agreeing with the recently de ned robust-exible adsorption. 43,44 However, this phenomenon is not observed for iso-butane adsorption. As we konw, n-butane gives dynamic diameter of 4.7 Å, slightly larger than the window of [Th 6 Co 2 ] cage (big window of 4.4Å), and thus occurrence of robust-exible adsortion is reasonable. While for branched isomer of iso-butane, it owns bigger dynamic diameter of 5.3 Å, far exceeding the window of [Th 6 Co 2 ] cage, resulting in normal MS effect and consequently low uptake. To further con rm this robust-exible adsortion, we next measured the adsorption of C 3 H 6 that also owns a dynamic diameter of 4.7 Å. Similarly, Figure S8 shows the typical robust-exible adsorption for C 3 H 6 . First a steep adsorption at 0-2.5 kPa was observed, then a step adsorption indicating a gateopening behavior was followed. The overall C 3 H 6 adsorption agrees well with the reported de nition of robust-exible MOF such as UTSA-300 and ELM-12. 37 To our best of knowledge, only extremely few robust-exible compounds were reported until now. Our case should present the rst actinides-based compounds with robust-exible adsorption. For both n-butane and C 3 H 6 , they hold slight larger kinetic diameter than the window of [Th 6 Co 2 ] cage, thus rationally resulted in the gate-opening behavior. In this regard, we can draw a conclusion that robust-exible adsorption will be absent for molecule smaller than the window of the [Th 6 Co 2 ] cage. As expected, Th-Co-Cage-1 preforms high C 2 H 4 uptake of 3.67 mmol/g at 1 bar and 298 K, however the adsorption isotherm of C 2 H 4 is distinct from that for n-butane and C 3 H 6 , excluding the robust-exible nature ( Figure S8), owing to C 2 H 4 molecule with a smaller kinetic diameter of 4.2 Å.
The separation ability of C 4 isomer was initially re ected on the high n-butane/iso-butane selectivity, giving n-butane/iso-butane selectivity up to 130.5 at 1 bar and 298 K for an equimolar n-butane/isobutane mixture ( Figure S9). The results imply its outstanding potential in n-butane/iso-butane separation. Whereafter, to assert the actual separation potential, breakthrough experiments upon Th-Co-Cage-1 bed was carried out for an equimolar n-butane/iso-butane mixture at 298 K ( Figure 3d). As expected, isobutane breaks out the column immediately, suggesting that the iso-butane adsorption was completely excluded by Th-Co-Cage-1, owing to MS effect, well consistent with the adsorption results and structural analysis. By contrast, n-butane can be retained in Th-Co-Cage-1 column for a very long time up to 70 min/g. The n-butane adsorption capacity is as high as 3.25 mmol/g, equal to 97% of the theoretic adsorption capacity (3.35 mmol/g at 0.5 bar). This value is far exceeding the benchmark material of ZU-36-Co (0.89 mmol/g). 41 The production of pure iso-butane (>99.9%) is as high as 68 mL/g, also far exceeding the benchmark material of ZU-36-Co (17.5 mL/g). 41 The results strongly suggest its superior application in C 4 isomer separation.
Furthermore, separation of C 6 isomer was carried out. For C 6 isomer separation, complete separation of linear, monobranched and dibranched isomers are still a challenging task. 33 Regarding the kinetic diameter of C 6 isomers shows the order of n-HEX (n-hexane, 4.3Å)<3MP(3-methylpentane, 5.0 Å)=2MP(2methylpentane, 5.0 Å)<23DMB (2,3-dimethylbutane, 5.6 Å) <22DMB(2,2-dimethylbutane, 6.2 Å), in conjunction with the robust-exible adsorption in this [Th 6 Co 2 ] cage, complete separation of linear, monobranched and dibranched isomers by Th-Co-Cage-1 can be expected. Accordingly, we then measured C 6 vapour adsorption at 30 °C and the results were shown in Figure 3e. Ultrahigh n-HEX adsorption capacity up to 209 mg/g was observed, exceeding most top level materials for this task [45][46][47][48][49][50][51][52][53] such as Al-bttotb (151 mg/g), 49 Ca-tcpb (150 mg/g), 48 MIL-53(Fe)-(CF 3 ) 2 (32 mg/g), 47 and Fe 2 (BDP) 3 (113 mg/g) 46 . By contrast, no adsorption and very low adsorption of 10.7 mg/g was observed for the dibranched isomers of 22DMB and 23DMB, owing to MS effect, since the kinetic diameter of dibranched isomers is far larger than the window of [Th 6 Co 2 ] cage. Impressively, moderate uptake of 93.2 mg/g and 93.8 mg/g for monobranched 3MP and 2 MP was observed, most likely due to the unique robust-exible nature in this material, as both 3 MP and 2 MP owns comparable kinetic diameter with C 3 H 6 and nbutane molecules that shows typical robust-exible adsorption as discussed above. The results suggest potential complete separation among linear, monobranched, and dibranched isomers. We further carried out the adsorption kinetics. As shown in Figure S10, Th-Co-Cage-1 shows high n-HEX uptake, moderate 2 MP and 3 MP uptake, low 23DMB uptake, and no uptake for 22DMB, which is well consistent with the adsorption isotherm. The adsorption equilibrium time is within 100 s for these isomers. Impressively, the adsorption kinetics also shows some difference for monobranched isomers of 2MP and 23DMB, or dibranched isomers of 23DMB and 22DMB, implying an outstanding potential for total separation of C 6 isomers.
To con rm the practical separation capability, we next carried out a series of batch experiments from a binary, or ternary, or quaternary mixture in liquid phase (Table 1, Figure S11 -20). In the literature, all reports are focused on the separation of C 6 vapour via multicomponent column breakthrough measurements. [45][46][47][48][49][50][51][52][53] From the viewpoint of saving energy, the direct separation of C 6 isomer from liquid mixture is more desirable, as these C 6 isomers are essentially liquid at room temperature with boiling point more than 58°. In each case, the activated Th-Co-Cage-1 samples were rst immersed in commercially binary mixture. Then their contents were extracted by n-butyl acetate from the samples after drying naturally. Impressively, 100% pure liner n-HEX or monobranched 2MP/3MP can be obtained from the binary linear/monobranched, or linear/dibranched, or monobranched/ dibranched mixture. More importantly, 100% pure 3MP can also be obtained from the 2MP/3MP mixture. Similar trend is also observed for the ternary mixture, yielding the 100% pure n-HEX or 3MP from the n-HEX/3MP/2,2DMB or 3MP/2MP 2,2DMB mixture, respectively. Besides, 100% n-HEX can be generated from the quaternary n-HEX/3MP/2MP/2,2DMB mixture. All the above results strongly suggest complete separation of C 6 isomer with 100% pure linear or monobranched product. Moreover, the production ability is also attractive. For example, 160 mg/g, 140 mg/g, and 110 mg/g pure n-HEX can be yielded from the n-HEX/3MP, or n-HEX/3MP/2,2DMB, or n-HEX/3MP/2MP/ 2,2DMB mixture, while 46 mg/g pure 3MP was produced from a 3MP/2,2DMB mixture. Even for 3MP/2MP mixture, a high production for 3MP up to 58 mg/g was observed. To further con rm the practical separation capability, we also for the rst time carried out the direct separation via multicomponent column breakthrough measurements with the liquid phase mixture (Figure 3f). The results show that Th-Co-Cage-1 can well separate the C 6 mixture into individual components. 22DMB breaks rstly from Th-Co-MOF bed after 24 s/g, then the retention time for 2MP, 3MP and n-HEX is 42 s/g, 58 s/g, and 72 s/g, respectively.   The breakthrough experiments upon Th-Co-Cage-1 bed for C6 isomer separation at 303 K.