As one of the primary nuclear fuels, deuterium (D₂) plays an irreplaceable role in controlled nuclear fusion and is also widely used in non-radioactive isotope tracing, chemical reaction mechanism tracing, as well as medicine and life sciences [1-4]. However, the separation and purification of D₂ from H₂ isotopic mixture is a challenging task due to their similar sizes and chemical properties. Conventional D₂/H₂ separation methods on industrial plant scale, such as cryogenic distillation process and Girdler sulfide process, are highly energy- and time-intensive. Furthermore, these technologies can only provide very low separation factors, making it difficult for extensive application [5-10]. Therefore, it is still an urgent and daunting task to explore new alternative methods for D₂/H₂ separation.

Recently, the strategy of separating hydrogen isotopes based on the kinetic quantum sieving (KQS) and chemical affinity quantum sieve (CAQS) effects of porous materials has attracted considerable attention. Based on KQS effect, which means that different diffusion rates can separate hydrogen isotopes in restricted pores [11], many reports have studied hydrogen isotope separation behavior of porous materials such as porous carbons [12-15], zeolites [16-18], metal-organic frameworks (MOFs) [19-23], covalent organic frameworks (COFs) [24] and porous organic cages [25]. However, the KQS effect is only obvious at extremely low temperatures (as low as 20 K), increasing separation costs dramatically. Unlike the KQS effect, CAQS effect can effectively separate hydrogen isotopes in porous materials under higher temperatures (≥ 77 K) because hydrogen isotopes demonstrate different adsorption capacity and enthalpies under low pressure [26]. Based on CAQS effect, heavier D₂ preferentially adsorbed on strong active sites, achieving high D₂/H₂ selectivity. Compared to porous carbons, zeolites and COFs, MOFs exhibit more advantages for D₂/H₂ separation because their pore sizes and open metal sites (OMSs) are controllable and further optimize gas uptake and adsorption enthalpy, which are critical factors for D₂/H₂ separation based on CAQS effect. In recent years, MOFs have been widely used in gas separation [27-30]. However, very few MOFs, such as Cu(Ⅰ)-MFU-4L [31, 32], Co-MOF-74 [33] and FJI-Y11 [34], are applied to D₂/H₂ separation based on obvious CAQS effect. Low-temperature thermal desorption spectroscopy indicated that Cu(Ⅰ)-MFU-4L only showed high H₂/D₂ selectivity at very low temperature (20 K). Furthermore, it is well known that Cu(Ⅰ)-MOFs are always unstable in air and can be easily oxidized by oxygen, limiting its practical application. Therefore, we devote to screening stable materials with strong binding sites and high D₂/H₂ uptakes for D₂/H₂ separation under more mild conditions.

Compared to low-temperature thermal desorption spectroscopy, the breakthrough experiment used for D₂/H₂ separation is closer to simulated industrial separation processes. Recently, our group carried out breakthrough experiments to explore D₂/H₂ separation performance of the famous MOF-74 series frameworks with a high density of OMSs, especially Co-MOF-74 that exhibited a satisfying D₂ retention time of 300 min/g (H₂/D₂/Ne: 1/1/98) at 77 K [35]. As structural isomers of MOF-74, M₂(m-dobdc) (M = Co, Ni, Mg, Mn) frameworks possess a narrower pore size of 9.8 Å than that of MOF-74 (11 Å). The smaller pore size may be more conducive to mass transfer process of hydrogen isotope gas molecules in the channels. Based on the above considerations, we evaluate the D₂/H₂ separation ability of M₂(m-dobdc) frameworks by considering the competitive and synergistic effect between adsorption enthalpy and capacity during the breakthrough process.

M₂(m-dobdc) (M = Co, Ni, Mg, Mn) were synthesized by solvothermal method (Fig. S1 in Supporting information) according to previous reports [36]. M₂(m-dobdc) frameworks feature a three-dimensional honeycomb structure constructed by one-dimensional [M(μ−COO)(μ−OH)]_n chains. The calculated pore size of M₂(m-dobdc) is 9.8 Å, smaller than that of MOF-74 (11 Å). This facilitates mass transfer process of hydrogen isotope molecules in the channels. Since the hydroxyl oxygen atom and the carboxyl oxygen atom in m-dobdc ligands are involved in coordination, this series of materials can maintain their original frameworks under high vacuum at 180 ℃. The removal of coordinated solvent molecules leaves large number of OMSs, with a calculated density of 4.6 OMSs/nm⁻³ in M₂(m-dobdc), much higher than those of HKUST-1 (2.6 OMSs/nm⁻³) [37] and PCN-16 (1.2 OMSs/nm⁻³) [37]. The purity of prepared frameworks was confirmed by their PXRD patterns (Fig. 1a), and typical type-Ⅰ N₂ sorption isotherms (Fig. 1b) at 77 K indicate their microporous characteristics.

To better understand the adsorption behavior of hydrogen isotope for M₂(m-dobdc), sorption isotherms of H₂ and D₂ at 77 K and 87 K were obtained, respectively. Among these four materials, Co₂(m-dobdc) shows the highest adsorption capacity as 9.62 mmol/g for H₂ and 10.7 mmol/g for D₂ at 800 mmHg and 77 K, higher than Ni₂(m-dobdc) (H₂: 8.24 mmol/g, D₂: 9.24 mmol/g), Mg₂(m-dobdc) (H₂: 7.97 mmol/g, D₂: 8.75 mmol/g) and Mn₂(m-dobdc) (H₂: 7.88 mmol/g, D₂: 8.56 mmol/g) (Fig. 2). As for the low-pressure part (1 kPa, a partial pressure of H₂ and D₂ during the breakthrough process) of the adsorption curve, the D₂(H₂) adsorption capacity of Co₂(m-dobdc) and Ni₂(m-dobdc) sharply increase to 5.49(5.22) and 4.32(3.96) mmol/g, respectively, which are contrast to the near-flat growth of adsorption curve for Mg₂(m-dobdc) (3.07(2.70) mmol/g) and Mn₂(m-dobdc) (3.29(2.90) mmol/g), which can be attributed to the strong binding force between OMSs and hydrogen isotope molecules in the framework. The results further indicate their great potential for D₂/H₂ separation. The results demonstrate that metal centers significantly influence host-guest interactions, resulting in different performance in the selective adsorption of H₂ and D₂.

Adsorption enthalpy contributes towards understanding the interaction strength between gas molecules and adsorbates. Herein, the adsorption enthalpies of H₂ and D₂ are calculated through the Clausius-Clapeyron equation. As shown in Fig. 3a, among these four materials, Ni₂(m-dobdc) shows the highest adsorption enthalpy of H₂ and D₂ at zero coverage (H₂: 11.5 kJ/mol, D₂: 13.0 kJ/mol), slightly higher than Co₂(m-dobdc) (H₂: 10.7 kJ/mol, D₂: 11.8 kJ/mol). Such high adsorption enthalpy implies that Co₂(m-dobdc) and Ni₂(m-dobdc) may be good candidates for D₂/H₂ separation.

To predict the D₂/H₂ separation ability of M₂(m-dobdc), the separation performance for D₂/H₂(50/50) mixtures was evaluated through ideal adsorbed solution theory (IAST) [38] (Figs. 3b-e). At 77 K, the D₂/H₂(50/50) selectivity of Co₂(m-dobdc) and Ni₂(m-dobdc) can reach 4.3 and 5.5 at zero coverage, respectively, higher than some famous reported porous materials such as CuBOTf (1.3) [39], Fe-MOF-74 (2.5) [26] and Co-MOF-74 (3.2) [26]. When the pressure rises to 1 kPa, these two values still maintain at 4.22 and 5.39, respectively. According to previous reports, materials with such high IAST selectivity at 77 K were rarely reported, meaning the potential ability of Co₂(m-dobdc) and Ni₂(m-dobdc) for separating hydrogen isotopes.

Breakthrough experiments were performed to check the actual separation performance of M₂(m-dobdc). Herein, we simulated two hydrogen isotope mixtures with different compositions (H₂/D₂/Ne: 1/1/98) and (H₂/D₂: 50/50) and evaluated the actual separation ability of M₂(m-dobdc) adsorbents through breakthrough experiments at 77 K and 87 K, respectively. For these four M₂(m-dobdc) adsorbents, when the gas mixture (H₂/D₂/Ne: 1/1/98) flowed over the packed column with a flow rate of 15 mL/min at 77 K, H₂ always flowed out first due to its lower adsorption capacity and weaker binding force with adsorbent than D₂. After the hydrogen broke up, D₂ can still be retained in the packed column filled with Co₂(m-dobdc) for 180 min/g, higher than those of Ni₂(m-dobdc) (150 min/g), Mg₂(m-dobdc) (80 min/g) and Mn₂(m-dobdc) (3 min/g) (Fig. 4). When breakthrough experiments were conducted with H₂/D₂(50:50), effective separation of D₂/H₂ can still be achieved with a breakthrough time of 4.5 min/g for Co₂(m-dobdc) and 3 min/g for Ni₂(m-dobdc), while Mg₂(m-dobdc) and Mn₂(m-dobdc) showed no obvious separation ability. Surprisingly, although Ni₂(m-dobdc) has the highest D₂ and H₂ adsorption enthalpy, Co₂(m-dobdc) showed the longest D₂ retention time. As for Mg₂(m-dobdc) and Mn₂(m-dobdc), they have similar adsorption capacity, but Mn₂(m-dobdc) has inferior separation effect due to its low adsorption enthalpy. The similar rule can also be found in our previous D₂/H₂ separation research on MOF-74. On the other hand, although M₂(m-dobdc) (M = Co, Ni, Mg) exhibit similar hydrogen isotope adsorption enthalpies with M₂(dobdc) (M = Co, Ni, Mg), M₂(m-dobdc) always showed shorter breakthrough time due to their narrower pore sizes and lower adsorption capacities of D₂ and H₂. Based on the above comparison, although adsorption enthalpy and adsorption capacity are both important to breakthrough experiment performance, adsorption capacity seems to play a more decisive role in D₂/H₂ separation because D₂/H₂ separation conducted by breakthrough experiments requires MOFs to display high D₂/H₂ uptakes as well as significant difference of D₂/H₂ uptakes to extend D₂ retention time. The regeneration of adsorbents is also an important evaluation criterion in industrial applications, and cyclic experiments revealed that during three cycles, the breakthrough times of D₂ in Co₂(m-dobdc) and Ni₂(m-dobdc) remained (Fig. 4e), indicating their excellent cycling stability for D₂/H₂ separation.

In conclusion, D₂/H₂ separation performance of M₂(m-dobdc) is explored by breakthrough experiments, in which two critical factors, including gas uptake and adsorption enthalpy, are considered to investigate the competitive and synergistic effect between adsorption enthalpy and capacity in D₂/H₂ separation ability of M₂(m-dobdc) frameworks. The results showed that for MOF materials with approximate adsorption enthalpies, MOF with higher gas uptake exhibits greater D₂/H₂ separation performance. Therefore, among these materials, Co₂(m-dobdc) exhibits the second-highest hydrogen isotope adsorption enthalpy and the highest gases uptake (5.22 mmol/g for H₂ and 5.49 mmol/g for D₂) under 1 kPa, making it the best material for D₂/H₂ separation with D₂ retention time of 180 min/g for (H₂/D₂/Ne: 1/1/98) at 77 K. Compared to Co-MOF-74 (Table S1 in Supporting information), Co₂(m-dobdc) has lower adsorption enthalpy and capacity, resulting in shorter retention time of D₂. The results demonstrate that the adsorption capacity shows a more significant impact than adsorption enthalpy on D₂/H₂ separation performance. This work demonstrates a noteworthy design consideration to develop materials with excellent hydrogen isotope separation ability.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

This work was financially supported by the Strategic Priority Research Program of Chinese Academy of Sciences (No. XDB20000000), the Key Research Program of Frontier Sciences, Chinese Academy of Sciences (No. QYZDB-SSW-SLH019) and the National Natural Science Foundation of China (Nos. 21771177, 51603206 and 21203117).

Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.cclet.2021.02.063.