Metal–organic frameworks (MOFs) are porous crystalline materials constructed by stitching metals and organic linkers together by strong bonds [1-3]. Recently, MOFs have shown a variety of potential applications, such as gas storage and separation [4, 5], catalysis [6], luminescence [7, 8] and biomedicines [9]. Over the developments in last two decades, most of the MOF structures reported were based on the coordination bonds between metals and carboxylates [10] or azolates [11, 12], as seen in the cases of MOF-5 [13] and ZIF-8 [14]. Meanwhile, there were also several examples in which phenolic hydroxyl groups worked with carboxylates together for the MOF construction, such as MOF-74 [15]. Recently, MOFs that were purely based on pyrogallol moieties (three phenolic hydroxyl) [16, 17] or catechol moieties (two phenolic hydroxyl) [18] were synthesized, and they have shown promising properties in efficient CO₂ photoreduction [17] and high ion conductivities [18]. On the other hand, using both carboxylates and pyrogallates at the two ends of the organic linkers for MOFs is relatively rare. Limited examples include metal gallates [19-21] by Cheetham and Devic et al. Further exploring the chemistry of MOFs using gallic acid (H₄GAL, Fig. 1a) with mixed coordination moieties will definitely introduce novel secondary building units (SBUs) [22] in the assembly, and provide new MOF structures with designed properties.

The linkers with different functionalities in the MOFs have been demonstrated to affect luminescence of the frameworks, and thus have been studied extensively for their application in molecular sensing and adsorption [23-25]. On the other hand, lanthanide MOFs, in view of their unique characteristics, show intrinsic luminescence properties, such as long lifetime, high quantum yields, and well-defined line emissions [26]. Therefore, lanthanide MOFs have been studied for the application of sensing [27], systematic color tuning [28], and optical transition [29, 30], etc. Combining the luminescence properties from the lanthanide metals and the distinct linkers with specific coordination configurations will help us further explore the chemistry of lanthanide MOFs. Herein, we took advantage of the gallic acid and Ce(Ⅲ) to construct a new MOF named Ce-FDM-50 (FDM = Fudan Materials). In Ce-FDM-50, both carboxylate and pyrogallate moieties were coordinated with Ce(Ⅲ) to form a new kind of paddle wheel SBU. These SBUs were further joined together by the phenyl rings to form a two-dimensional (2D) framework in sql topology. With the concept of reticular chemistry [31], we further used five different lanthanide metals to construct isoreticular M-FDM-50 (M = Nd, Eu, Gd, Tb, Yb) with the same configuration. In addition, the successful synthesis of Tb_xEu_1-x-FDM-50 demonstrated that the framework was capable of incorporating more than two kinds of lanthanide metals into a single framework in one-pot reaction, and provided an ideal platform to fine tune the metal compositions and resulting properties in the structure. We used Tb-FDM-50 as an example to illustrate their luminescence properties in solid state. The FDM-50 series showed that using mixed coordinating moieties from the linker, we can enrich the structures with new SBUs. Combining them with the lanthanide metals and practicing the multi-variate strategy [32], emerging luminescence properties are expected.

We describe the synthesis of Ce-FDM-50 as an example of preparation procedures of FDM-50 series. Specifically, H₄GAL (170.0 mg, 1.0 mmol) and CeCl₃ 7H₂O (372.6 mg, 1.0 mmol) were dissolved in 5 mL N, N-dimethylformamide (DMF) in a 25-mL vial. The vial was then capped tightly and placed in an oven at 120 ℃ for 12 h.Brown block crystals were collected and rinsed with 15 mL DMF for five times. Formula of Ce-FDM-50: CeC₁₃H₁₈N₂O₇Cl = Ce (H₂GAL)(DMF)₂Cl. FT-IR: (KBr, 450–4000 cm^-1): 3438(w), 2865(w), 2711(w), 2633(w), 2567(w), 2517(w), 1673(s), 1595(w), 1546(s), 1392(s), 1345(s), 1275(s), 1104(w), 1038(s), 906(w), 834(s), 790(s), 740(s), 663(s), 580(w). For the synthesis of M-FDM-50 (M = Nd, Eu, Gd, Tb, Yb), the same procedure was applied, except that MCl₃·xH₂O (M = Nd, Eu, Gd, Tb, Yb) with the same mole amount was used instead of CeCl₃·7H₂O.

In addition to the FDM-50 structures which are based on one kind of linker only, we explored the possibility of including mixed linkers into one single framework. 3, 4-Dihydroxybenzoic acid (31.0 mg, 0.2 mmol), H₄GAL (136.0 mg, 0.8 mmol), and CeCl₃·7H₂O (372.6 mg, 1.0 mmol) were dissolved in 5 mL DMF in a 25-mL vial. The vial was capped tightly and placed in an oven at 120 ℃ for 12 hours. The brown crystals, named Mix1-FDM-50, were harvested and then washed by DMF for several times. Mix2-FDM-50 was synthesized with the same protocol, except that 3, 4-dihydroxybenzoic acid was replaced by 3, 5-dihydroxybenzoic acid. On the other hand, the incorporation of mixed metals was demonstrated in the case of Tb_xEu_1-x-FDM-50 (x = 0.9, 0.7, 0.5, 0.3, 0.1), which were synthesized by employing a similar synthetic procedure with that for M-FDM-50, except that TbCl₃·6H₂O and EuCl₃·6H₂O, with the molar ratio of x:(1-x), were used as the starting metal source.

Previous research on the construction of MOFs showed that deliberate selection of organic linkers plays an essential role in the formation of particular SBUs and the resulting network structures [33]. In this study, we picked a ditopic linker, H₄GAL, as a fascinating linker with two kinds of distinctive functional groups in the ends. As shown in Fig. 1a, one end of H₄GAL features carboxyl moiety and the other end has pyrogallol moiety. Each linker possesses five O atoms that could be coordinated to the transition metal, and could possibly construct new SBU which was not observed in the MOFs with carboxylates or phenol groups only. Specifically, by reacting CeCl₃·7H₂O and H₄GAL in DMF at 120 ℃ for 12 h, brown transparent block crystals were collected. The crystal structure, named Ce-FDM-50, was solved by single crystal X-ray diffraction (SXRD) (Table S1 in Supporting Information). It revealed the Ce-FDM-50 crystallizes in the orthorhombic space group Cmca, with a = 19.000(7) Å, b = 14.474(6) Å, and c = 12.867(5) Å.

In each unit cell, one Ce, one GAL linker, two DMF molecules, and one Cl were located. Each Ce was eight coordinated with two O from the carboxyl of GAL linkers, two O from meta-phenolic hydroxyl, two O from para-phenolic hydroxyl, and two O from DMF molecules (Fig. 1b). Determined by the SXRD, the bond lengths of the Ce-O bonds were at the range from 2.42 Å to 2.61 Å, which were similar to those reported in the literatures [34, 35].

The carboxylate and pyrogallate have brought two Ce(Ⅲ) together, forming a new type of SBU (Fig. 1b) which can be viewed as an analogue to the famous Cu₂(COO)₄ paddle wheel SBU in HKUST-1 [36]. However, in this SBU, the spatial distance between two neighboring metals was 3.92 Å, which was larger than that in HKUST-1 (2.63 Å). The two neighboring Ce atoms were connected by two carboxylates from two diagonal directions, and by two pyrogallates from the other two directions. The unsymmetrical linker has made the SBU in a slightly distorted square planer geometry, as indicated by the distorted dihedral angle between the two paddles (82.3). This new SBU could serve as a 4-connected node in the MOF assembly. Connecting the SBUs together by the phenyl rings (Fig. 1c) led to the formation of a 2D layered Ce-FDM-50 with sql topology. In addition, DMF molecules occupied the remaining coordination sites of Ce(Ⅲ) as terminal ligands and they were not involved in the connections between the SBUs.

Furthermore, the 2D layers sat along the [010] plane, and a clear ABA packing of the layers could be seen along the [010] direction (Fig. 1d). Each layer was separated by the interlayer distance of 5.59 Å. This relatively large interlayer distance was due to the coordinated DMF molecules occupying partial interlayer space. To compensate the charge balance of the framework, a Cl atom was bonded to the meta-phenolic hydroxyl group through hydrogen bond interaction (O-H...Cl), and the length was 2.95 Å. Overall, Ce-FDM-50 had a formula of Ce(H₂GAL)(DMF)₂Cl. The successful synthesis of Ce-FDM-50 proves that the GAL linker with two distinct coordinating moieties could form a new type of SBU, and interesting MOFs could be obtained by strategically connecting the SBUs into extended networks.

To validate the phase purity of Ce-FDM-50, powder X-ray diffraction (PXRD) of the MOF crystals were examined. The powder pattern of Ce-FDM-50 was in good agreement with the simulated pattern derived from the single crystal structure, indicating the purity of Ce-FDM-50 (Fig. 2a). In addition, the high-resolution X-ray photoelectron spectroscopy (XPS) of Ce 3d binding energy indicated the presence of only Ce³⁺ in Ce-FDM-50. Specifically, the peaks at 879.4 and 883.4 eV were associated with the Ce 3d_5/2, and the peaks around 898.1 and 901.9 eV were assigned to Ce 3d_3/2. On the other hand, the characteristic peak of Ce⁴⁺ at~914.0 eV [37] was absent in the spectrum (Fig. 2b). By comparing the FT-IR spectra of the H₄GAL linker and the as-synthesized Ce-FDM-50 (Fig. S2 in Supporting information), the peak disappearance at 3494 cm^-1, which accounts for the intramolecular hydrogen bond (v_O-H) in the pyrogallol group, confirmed the deprotonation of the para-phenolic hydroxyl in GAL when they were incorporated into the extended framework.

By examining the single crystal structure of Ce-FDM-50, we concluded that the intrinsic porosity of this MOF was blocked by the coordinated DMF molecules in the framework. If successful removal of DMF could be achieved without losing the integrity of the framework, pores inside Ce-FDM-50 would be accessible to incoming guests. To explore the possible activation methods for this MOF, the material was firstly solvent-exchanged with dichloromethane for three days, followed by vacuum under various temperatures (room temperature, 150 ℃, or 200 ℃) for ten hours. To check whether the DMF molecules were removed under these conditions, we performed the solution ¹H NMR of the DCl-digested MOF samples after different activation methods. As presented in Fig.]FID$T53[S3 in Supporting information, for Ce-FDM-50 activated at room temperature or at 150 ℃, the peaks at 2.71, 2.87 and 7.93 ppm clearly indicated the presence of DMF in the MOF. Calculating from the peak integration, the molar ratio between DMFand GAL (7.93, 6.92 ppm) is 1.94:1, and it matched with the theoretical value (2:1). On the other hand, when the sample was heated at 200 ℃, the molar ratio between DMF and GAL dropped down to 1.36:1. Although less DMF molecules were present in the ¹H NMR, 68% of the DMF molecules remained in the framework. Furthermore, by investigating the PXRD patterns of the samples after different methods of activation (Fig. S1 in Supporting information), we can see that the material held its structure until 150 ℃. However, further heating the material up to 200 ℃ under vacuum destroyed the framework, as evidenced by the complete phase transformation in the PXRD pattern. The thermogravimetric analysis (TGA) curve (Fig. S4 in Supporting information) of Ce-FDM-50 exhibited that the material did not lose the coordinated DMF molecules under 190 ℃. Above 190 ℃, the framework lost weight gradually, indicating possible structural collapse. Overall, removing the DMF without destroying the framework was not successful. This conclusion was also in agreement with our adsorption studies on the Ce-FDM-50, in which no meaningful uptake was observed in the 77 K nitrogen adsorption isotherm.

The concept of reticular chemistry allows us to explore a whole series of isoreticular structures based on the same linker, but with different metals sharing similar coordination geometries. Lanthanide metals tend to have the same coordination behaviors, thus we further explored the adaptability of different Ln(Ⅲ) into this new paddle wheel SBU. As a result, under similar reaction parameters, five new different MOFs with different Ln(Ⅲ) metals, named M-FDM-50 (M = Nd, Eu, Gd, Tb, Yb), were achieved. All five structures were obtained in the single crystal form, and the absolute structures were solved by SXRD (Table S1). In addition, their PXRD patterns matched well with the simulated pattern, confirming the phase purities of these isoreticular structures (Fig. 2a). The unit cell volumes of these MOFs decreased slightly along the increase of the atomic numbers. For example, the unit cell volume of Ce-FDM-50 was 3549 ų, while that of Yb-FDM-50 dropped down to 3416 ų. This phenomenon is a classical indicator of the lanthanide contraction due to the decrease of the ionic radii from 1.14 Å for Ce to 0.99 Å for Yb.

The similarity between the lanthanides allowed us to construct multi-variate MOF (MTV-MOF) [32] structures, in which two kinds of metals or linkers could occupy the crystallographic equivalent positions, and the ratio between the mixed components could be precisely tuned. The additional metals and/or linkers could enrich the properties of the MOF structures, such as fine-tuned luminescence. To realize this strategy preliminarily, we intended to construct bimetallic Tb_x Eu_1-x -FDM-50, in which both Tb and Eu ions were located precisely and arranged periodically in the framework. Specifically, different MOFs with five different metal ratios, with the feeding ratios between Tb and Eu being 9:1, 7:3, 5:5, 3:7 and 1:9, were successfully synthesized by one-pot reaction. These MOFs, named Tb_xEu_1-x-FDM-50 (x = 0.9, 0.7, 0.5, 0.3, 0.1), show the same PXRD patterns with the simulation (Fig. S6 in Supporting Information). Their compositional molar ratio between Tb and Eu were accurately determined by inductively coupled plasma (ICP) spectroscopy, and they were listed in Table 1. The molar ratios between the two metals in Tb_xEu_1-x-FDM-50 were roughly in agreement with the feeding ratios, suggesting similar reaction kinetics in the coordination assembly of the frameworks using different lanthanide nodes.

The MTV-MOF construction clearly demonstrated that the MOF with this particular GAL linker was quite tolerate towards different kinds of metals. To further examine its tolerance towards different organic linkers with slightly different functional groups, we picked two organic compounds, 3, 4-dihydroxybenzoic acid (with one less -OH than H₄GAL in the meta- position) and 3, 5-dihydroxybenzoic acid (one less -OH than H₄GAL in the para- position), as the "fragmented linker" of H₄GAL. When the fragmented linkers were mixed with H₄GAL in the reaction solution, there was possibility that the fragmented linkers, in which the particular phenyl hydroxyl group were missing, participated in the assembly, and potentially defect MOFs (named Mix1-FDM-50 and Mix2-FDM-50, respectively) could be formed. Performing the MOF construction with both H₄GAL and the fragmented linker with different ratios in one-pot with Ce(Ⅲ), we found that the same MOFs could be constructed, as validated by the PXRD measurement (Fig. S5a in Supporting information). However, based on the ¹H NMR of the digested MOFs, no fragmented linkers were present in the structure (Fig. S5b in Supporting information). In other words, the FDM-50 had very low tolerance towards fragmented linkers, and it tended to form a perfect crystal structure without any defect. This experiment also illustrated the significance of pyrogallol groups in FDM-50.

The successful construction of Ln-FDM-50 has provided us a platform to investigate their characteristic luminescence. Here, we use Tb-FDM-50 as an example. The UV–vis absorption spectra of the free H₄GAL linker and the resulting Tb-FDM-50 in solid state were recorded at room temperature (Fig. 3a). The H₄GAL linker showed strong absorbance below 329 nm, while Tb-FDM-50 showed a strong peak at 355 nm. In the solid-sate excitation and emission spectroscopy, when excited at 311 nm, the Tb-FDM-50 exhibited the characteristic luminescence with sharp and well-separated emission bands at 489, 544, 588, and 621 nm (Fig. 3b). These peaks can be assigned to the ⁵D₄ → ⁷F_J (J = 6, 5, 4, and 3) transitions of Tb ions [38]. With the reticular chemistry concept and MTV-MOF concept shown in the FDM-50, interesting luminescence properties could be furtherly designed and finetuned.

In summary, by utilizing the two coordination moieties from gallic acid (carboxylates and pyrogallates), fascinating paddle wheel SBUs could be formed with six different kinds of lanthanide ions (Ce, Nd, Eu, Gd, Tb, Yb). The resulting frameworks have expanded the concept of SBUs, which were usually based on only one kind of coordination moiety, into mixing carboxylates and pyrogallates. The obtained FDM-50 series featured isoreticular 2D square networks, with the SBUs as the nodes and the phenyl rings as the edges, in ABA packing mode. The successful synthesis of six MOFs with different metals was confirmed by the single crystal X-ray diffractions. The MTV-MOF strategy further extended the frameworks to bimetallic MOFs in which the two metals (Tb and Eu) could be precisely located and distributed. The study has shown that by mixing two distinct coordination moieties in the organic linker, MOF structures with intriguing SBUs could be designed and synthesized. With the capability of incorporating different metals into the parent frameworks, emerging properties, such as fluorescence and sensing, are anticipated.

This work was supported by the National Natural Science Foundation of China (Nos. 21571037, 21733003, 21961132003), the National Key Research and Development Project of China (No. 2018YFA0209400), National Top-Notch Talent Program, and the Science & Technology Commission of Shanghai Municipality (Nos. 16520710100, 17JC1400100).

Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.cclet.2019.05.055.