All reagents were obtained from commercial vendors and were used without further purification. Water was distilled and deionized prior to use. Hydrogen 2-(trimethylsilyl)ethyl 1,4-benzenedicarboxylate (1) was prepared according to the literature.¹¹

CHN elemental analyses were carried out using a Perkin-Elmer 2400 CHNS Elemental Analyzer II. Room temperature infrared (IR) spectra were recorded with a Jasco 4100 FT-IR spectrometer using a KBr disk. Thermogravimetric/differential thermal analyses (TG/DTA) were acquired using a Rigaku Thermo Plus 2 series TG/DTA TG 8120 instrument. Nuclear magnetic resonance (NMR) data were collected using a Jeol JNM-EX 400 FT-NMR spectrometer with a Jeol EX-400 NMR data processing system. X-ray powder diffraction (XRPD) was carried out on a MacScience M18XHF material analysis and characterization instrument using Cu Kα radiation. Nitrogen adsorption isotherms were measured using a Micromeritics ASAP2010 volumetric adsorption apparatus. The samples of RhCu-DMF and RhZn-DMF were immersed in THF for several days to exchange all of the included nonvolatile solvates (DMF and H₂O), then evacuated at rt for 5 h. Only the RhCu-EtOH sample was evacuated at rt for 24 h. Inductively coupled plasma atomic emission spectroscopy (ICP-AES) was conducted on a PII NT SPS3500. Samples (3–5 mg) were digested in concd HNO₃ and heated at 120 ℃ until the solution became almost clear. This concentrated acid solution was diluted with deionized H₂O. However, the compositional amount of Rh, Cu, and Zn in the solid could not be quantitatively determined by ICP-AES due to their poor solubility; therefore, only the ratios of Rh:Cu/Zn were determined.

Hydrogen 2-(trimethylsilyl)ethyl 1,4-benzenedicarboxylate (1) (328 mg, 1.23 mmol) and [Rh₂(OAc)₄] (0.118 mg, 0.267 mmol) were dissolved in chlorobenzene (30 mL) in a Schlenk tube with an argon atmosphere. The mixture was heated at 150 ℃ for 1 day and the solvent was then removed in vacuo. The resulting material was purified by flash chromatography on silica gel with MeOH/CH₂Cl₂ = 1/19 (v/v) as the eluent to yield compound 2 as a light greenish blue solid (320 mg, 0.253 mmol, 95%). R_f = 0.72 (MeOH/CH₂Cl₂ = 1/19 (v/v)). ¹H NMR (CDCl₃, 400 MHz): δ 8.17 (d, J = 8.4 Hz, 8H), 8.13 (d, J = 8.4 Hz, 8H), 4.46 (t, J = 8.4 Hz, 8H), 1.16 (t, J = 8.4 Hz, 8H), 0.10 (s, 36H, TMS). Anal. Calcd for C₅₂H₆₈O₁₆Rh₂Si₄ or [Rh₂(2-TMS-ethyl-1,4-bdc)₂] (M_r = 1267.24): C, 49.28; H, 5.41%. Found: C, 49.50; H, 5.63%. Selected IR (KBr): ν (cm⁻¹) = 2955(m), 2898(w), 1721(s), 1685(m), 1605(m), 1561(m), 1508(w), 1400(vs), 1276(vs), 1175(w), 1143(w), 1104(m), 1018(m), 930(w), 839(s), 737(s), 566(m).

Tetrabutylammonium fluoride (1 mol L⁻¹ in 1.8 mL THF, 1.8 mmol) was added to a solution of [Rh₂(2-TMS-ethyl-1,4-bdc)₄] (2) (320 mg, 0.253 mmol) in THF (20 mL) and the resulting solution was stirred at rt for 1 h. Saturated NH₄Cl(ag) (20 mL) was added to this solution and only THF was evaporated. The resulting precipitate was separated by filtration and washed with H₂O to give the complex (130 mg, 0.144 mmol, 57%) as a greenish blue powder. Partial deprotonation of carboxy group occurred, i.e., one of four carboxy groups in BDC exists as carboxylate anion, and requires a counter cation such as tetrabutylammonium (TBA) to balance charge. Therefore, the accurate formula is (TBA)[Rh₂(1,4-Hbdc)₃(1,4-bdc)(H₂O)₂]. This deprotonation does not inhibit the subsequent reactions. ¹H NMR (DMSO-d₆, 400 MHz): δ 7.85 (s, 16H), 3.17 (t, J = 8.4 Hz, 8H), 1.57 (m, 8H), 1.31 (m, 8H), 0.94 (t, J = 7.3 Hz, 12H). Anal. Calcd for C₄₈H₅₉NO₁₈Rh₂ or (TBA)[Rh₂(1,4-Hbdc)₃(1,4-bdc)(H₂O)₂] (M_r = 1143.79): C, 50.40; H, 5.20; N, 1.22%. Found: C, 50.39; H, 5.21; N, 1.18%. Selected IR (KBr): ν (cm⁻¹) = 2959(m), 1717(s), 1600(s), 1561(m), 1507(m), 1396(vs), 1275(s), 1144(m), 1105(m), 1018(m), 877(m), 841(m), 786(m), 739(m), 565(m).

Cu(NO₃)₂·3H₂O (8.2 mg, 0.034 mmol) was dissolved in EtOH (20 mL) containing discrete Rh paddle-wheel complex 3 (15 mg, 0.0131 mmol). The resulting solution was placed in a Teflon-lined stainless steel vessel, and heated without stirring in a 100 ℃ oil bath for 48 h. The product was separated by filtration and washed with EtOH and acetone to give the complex (13 mg, 0.0121 mmol, 92%) as a green powder. Anal. Calcd for C₃₄H₂₆Cu₂O₁₉Rh₂ or [Rh₂Cu₂(1,4-bdc)₄]·2H₂O·EtOH (M_r = 1071.46): C, 38.11; H, 2.45%. Found: C, 38.04; H, 2.56%.

Cu(NO₃)₂·3H₂O (58 mg, 0.240 mmol) was dissolved in dry DMF (25 mL) containing discrete Rh paddle-wheel complex 3 (131 mg, 0.115 mmol). The resulting solution was heated without stirring in a 60 ℃ oil bath for 2 h. The product was separated by filtration and washed with DMF to give the complex (82 mg, 0.0640 mmol, 56%) as a green powder. Anal. Calcd for C₄₄H₄₄N₄Cu₂O₂₀Rh₂ or [Rh₂Cu₂(1,4-bdc)₄]·4DMF (M_r = 1281.74): C, 41.23; H, 3.46; N, 4.37%. Found: C, 41.34; H, 3.08; N, 4.43%. Rh: Cu ratio estimated by ICP-AES was 1.0:0.9. TG/DTA under atmospheric conditions (Figure S1): a weight loss of 23.76% below 298.9 ℃ was observed; calcd 22.81% for 4DMF of [Rh₂Cu₂(1,4-bdc)₄]·4DMF. Selected IR (KBr): ν (cm⁻¹) = 1652(vs), 1625(s), 1596(s), 1508(m), 1437(m), 1394(vs), 1300(w), 1256(w), 1156(m), 1106(m), 1063(w), 1016(m), 883(m), 830(m), 751(m), 674(m), 584(m).

Zn(NO₃)₂·6H₂O (22 mg, 0.0739 mmol) was dissolved in dry DMF (20 mL) containing discrete Rh paddle-wheel complex 3 (40 mg, 0.0350 mmol). The resulting solution was heated without stirring in an 80 ℃ oil bath for 3 h. The product was separated by filtration and washed with DMF to give the complex (17 mg, 0.0134 mmol, 38.2% yield based on 3) as a grayish green powder. Anal. Calcd for C_43.4H_42.6N_3.8O_19.8Rh₂Zn₂ or [Rh₂Zn₂(1,4-bdc)₄]·3.8DMF (M_r = 1270.85): C, 41.02; H, 3.38; N, 4.19%. Found: C, 40.89; H, 3.09; N, 4.28%. Rh:Zn ratio estimated by ICP-AES was 1.0:1.0. TG/DTA under atmospheric conditions (Figure S2): a weight loss of 24.64% below 304.5 ℃ was observed; calcd 21.86% for 3.8DMF of [Rh₂Zn₂(1,4-bdc)₄]·3.8DMF. Selected IR (KBr): ν (cm⁻¹) = 1650(vs), 1594(s), 1506(m), 1437(m), 1394(vs), 1300(w), 1256(w), 1154(w), 1106(m), 1062(w), 1016(m), 884(m), 828(m), 751(m), 693(w), 672(w), 576(m).

In order to prepare the discrete dirhodium(II) tetra(1,4-benzenedicarboxylate) (3), 1,4-benzenedicarboxylic acid monoester, in which one of the two carboxy groups of 1,4-benzenedicarboxylic acid was protected by esterification, was synthesized (Scheme 1). The residual carboxylic acid was then reacted with dirhodium(II) tetraacetate, resulting in the discrete dirhodium(II) tetra(1,4-benzenedicarboxylate) tetraester (2). Deprotection of the four esters of 2 gave the discrete dirhodium(II) tetra(1,4-benzenedicarboxylate) (3). The key reaction in this synthetic route is the de-esterification step of the discrete dirhodium(II) tetra(1,4-benzenedicarboxylate) tetraester (2). The de-esterification step requires an appropriate choice of ester that can be deprotected under mild conditions, to avoid decomposition of the paddle-wheel dirhodium moiety. 2-(Trimethylsilyl)ethyl ester (TMS ethyl ester) was adopted, which gives carboxylic acid in the presence of fluoride ion through a β-elimination reaction. Hydrogen 2-(trimethylsilyl)ethyl 1,4-benzenedicarboxylate (1) was prepared according to the literature.¹¹ The discrete dirhodium(II) tetra(1,4-benzenedicarboxylate) tetraester (2) was prepared by ligand-exchange from reaction between 1 and dirhodium(II) tetraacetate. The discrete dirhodium(II) tetra(1,4-benzenedicarboxylate) (3) was obtained following the deprotection of TMS ethyl ester with tetrabutylammonium fluoride (TBAF). Initially, the methyl ester was used as the protecting group; however, 3 could not be obtained due to decomposition of the Rh paddle-wheel complex under the basic conditions. The rhodium–copper porous complexes RhCu-DMF and RhCu-EtOH were synthesized by the reaction of 3 with Cu(NO₃)₂·3H₂O in N,N-dimethylformamide (DMF) at 60 ℃ or ethanol at 100 ℃ (solvothermal reaction), whereas the rhodium–zinc complex (RhZn-DMF) was prepared by reacting 3 with Zn(NO₃)₂·6H₂O in DMF at 80 ℃. Although the synthesis of the original dirhodium(II) tetra(1,4-benzenedicarboxylate) [Rh₂(1,4-bdc)₂] requires high temperature (>160 ℃),⁹ these hetero-metal complexes were obtained under relatively lower reaction temperature (60–100 ℃). The lower synthesis temperature suppresses the formation of metallic rhodium as a by-product, because the reduction of Rh₂ units is avoided. Therefore, high-purity rhodium-containing MOFs were successfully obtained. Elemental analyses (CHN) of as-synthesized samples of RhCu-DMF, RhCu-EtOH, and RhZn-DMF indicated that these complexes included Rh and Cu/Zn in 1:1 ratios. In addition, evaluation of the Rh:Cu and Rh:Zn ratios of RhCu-DMF and RhZn-DMF by inductively coupled plasma atomic emission spectroscopy (ICP-AES) confirmed that Rh:Cu and Rh:Zn were 1.0:0.9 and 1.0:1.0, respectively.

We have previously reported the syntheses of rhodium-based carboxylate MOFs and their catalytic activities for various reactions.^9,10 However, these rhodium-based MOFs showed extremely broad X-ray powder diffraction (XRPD) patterns, which indicates no long-range ordering and low crystallinity, and therefore the structural characterization of rhodium-based MOFs was very difficult. Similar results were obtained for the ruthenium-based 1,4-benzenedicarboxylate.¹² In contrast, XRPD patterns of RhCu-DMF, RhCu-EtOH, and RhZn-DMF were sharply defined, which indicated high crystallinity (Figures 2a, 2b, and 2c). The diffraction patterns of both RhCu-DMF and RhZn-DMF were almost identical, and these diffraction patterns were very similar to the predicted and experimental XRPD patterns reported for [Cu₂(1,4-bdc)₂] prior to desolvation (Figure 2d).¹³ These results indicate that terephthalate ligands are coordinated in a bidentate bridging fashion to Rh₂ and Cu₂/Zn₂ paddle-wheel nodes and the resulting 2D sheets are bonded through weak stacking interactions, similar to the original [Cu₂(1,4-bdc)₂]. Each axial site of paddle-wheel node is coordinated to a molecule of DMF. In contrast, RhCu-EtOH was isotypic with [Cu₂(1,4-bdc)₂] re-solvated with EtOH.¹³ The structure of [Cu₂(1,4-bdc)₂] re-solvated with EtOH loses its crystallinity compared with that of as-synthesized [Cu₂(1,4-bdc)₂], which also occurred for the presently synthesized RhCu-EtOH. In these complexes, no peaks corresponding to the (111) and (200) planes of metal rhodium (2θ = 41 and 48°), were observed.¹⁴ The reported dirhodium(II) tetra(1,4-benzenedicarboxylate) MOF can be synthesized only at high temperature (>160 ℃); therefore, under the described conditions (60–100 ℃), the ligand-exchange reaction of the paddle-wheel Rh₂ unit did not occur and the dirhodium(II) tetra(1,4-benzenedicarboxylate) MOF product was not obtained. These results indicate that two types of paddle-wheel clusters, i.e., Rh₂ and Cu₂/Zn₂ units, alternately form vertices of a 2D square grid without any scrambling in the RhCu-DMF, RhCu-EtOH, and RhZn-DMF complexes (Figure 3).

We anticipate this approach to be applicable in a broader context, as other multi-carboxylate ligands are also potentially accessible, such as trimesic acid, biphenyl-3,3',5,5'-tetracarboxylic acid or polyphenyl tetracarboxylate ligands. Preliminary experiments have shown that trimesic acid (H₃btc) can be used as a building block for the preparation of the {[Rh₂(btc)_4/3][Cu₂(btc)_4/3]₂}·nSolvent rhodium-containing hetero bi-paddle-wheel MOF using this method, and its structure is very similar to that of the original [Cu₂(btc)_4/3], known as HKUST-1¹⁵ (see Supporting Information).

N₂ gas adsorption measurements at 77 K were performed for RhCu-DMF, RhCu-EtOH, and RhZn-DMF (Figure 4). The BET surface areas for RhCu-DMF, RhCu-EtOH, and RhZn-DMF were estimated to be 297, 649, and 297 m² g⁻¹, respectively. The BET surface areas of RhCu-DMF and RhZn-DMF synthesized in DMF were less than half of that of RhCu-EtOH synthesized in EtOH, which had the highest surface area of 649 m² g⁻¹. This value compares favorably with those of 545 and 625 m² g⁻¹ for [Cu₂(1,4-btc)₂] synthesized in methanol and in DMF, as reported by Seki et al.¹⁶ and Carson et al.,¹³ respectively. The pore diameter distributions of RhCu-DMF, RhCu-EtOH, and RhZn-DMF calculated by the Saito–Foley model¹⁷ from the adsorption isotherms using N₂ at 77 K showed narrow peaks at 8.0, 8.1, and 8.0 Å, respectively (Figure S4). Many hetero-bi-metallic MOFs synthesized utilizing metal-complex ligands as the building blocks have been reported to date, due to the approach of coordinatively unsaturated metal site introduction into the frameworks.¹⁸ However, in many cases the corresponding products did not possess permanent porosity upon solvent removal. On the other hand the present hetero bi-paddle-wheel MOFs prepared by the described method are isostructural with the original Cu–BDC MOFs already known to exhibit permanent porosity. Therefore, the hetero bi-paddle-wheel MOFs synthesized here exhibit permanent porosity and good gas adsorption properties. The persistence of permanent porosity after solvent evacuation is essential for gas-phase catalysis; therefore, incorporation of noble metals into MOFs by the method discussed in this paper is very useful for catalytic applications.