Highly Efficient MOF-Driven Silver Subnanometer Clusters for the Catalytic Buchner Ring Expansion Reaction.

The preparation of novel efficient catalysts─that could be applicable in industrially important chemical processes─has attracted great interest. Small subnanometer metal clusters can exhibit outstanding catalytic capabilities, and thus, research efforts have been devoted, recently, to synthesize novel catalysts bearing such active sites. Here, we report the gram-scale preparation of Ag20 subnanometer clusters within the channels of a highly crystalline three-dimensional anionic metal-organic framework, with the formula [Ag20]@AgI2NaI2{NiII4[CuII2(Me3mpba)2]3}·48H2O [Me3mpba4- = N,N'-2,4,6-trimethyl-1,3-phenylenebis(oxamate)]. The resulting crystalline solid catalyst─fully characterized with the help of single-crystal X-ray diffraction─exhibits high catalytic activity for the catalytic Buchner ring expansion reaction.

Introduction

The preparation, stabilization, and characterization of subnanometer metal clusters (SNMCs) has been a main challenge for chemists during the last years.[1−7] Indeed, the preparation of such ultrasmall entities is highly complex and often requires, for instance, the use of stabilizing blocking ligands[8] that prevent their agglomeration into larger metal nanoparticles (MNPs) but, in turn, may worsen their catalytic properties. Moreover, the characterization of such tiny SNMCs is really not an easy task and requires the use of high-resolution microscopy techniques—for example, high-angle annular dark-field–scanning transmission electron microscopy (HAADF–STEM)—whose electron beam can, with time, degrade the sample that is being observed.[9] However, despite such complications, the outstanding medical, optical, and/or catalytic properties of SNMCs make it well worth the effort.[3,4,10,11]

Focusing on the catalytic properties of SNMCs, stabilizing capping ligands of SNMCs dramatically reduces their catalytic activity as they prevent a proper contact with reactants and trigger their decomposition under reaction conditions. Therefore, it seems clear that “naked” SNMCs, with all metal atoms exposed, are called to offer the best catalytic performances.[12] In this context, a proper strategy to obtain ligand-free SNMCs consists in supporting these metal species in porous solids such as zeolites or organic porous materials.[13] More recently, another type of porous materials, the so-called metal–organic frameworks (MOFs),[14] have emerged as a suitable platform not only to host SNMCs but also to be used as chemical reactors for the in situ chemical synthesis of the SNMCs.[15,16]

MOFs are crystalline porous materials that have attracted significant attention in the past 2 decades due to the myriad applications they can be used in.[17] Moreover, aiming at encapsulating/synthesizing SNMCs, or even single-atom catalysts,[18−22] MOFs offer clear advantages compared to other porous materials such as a fine control of the functionalities decorating the channels—which allows us to retain and align metals in specific positions and controlled stoichiometries—and the possibility to use single-crystal X-ray crystallography[23−25] to unveil the crystal structure of these ultrasmall metal species.

In this context, we recently used,[21] as a chemical reactor, a highly robust anionic three-dimensional MOF, with the formula NiII2{NiII4[CuII2(Me3mpba)2]3}·54H2O [Me3mpba4– = N,N′-2,4,6-trimethyl-1,3-phenylenebis(oxamate)], for the MOF-driven preparation of ligand-free tetranuclear [Pd4]2+ clusters after two consecutive post-synthetic steps consisting of first replacing the Ni2+ cations hosted within its channels by Pd2+ ones and the concomitant reduction to form the final tetranuclear species within the empty space of the MOF. The resulting host–guest material had the following formula: [Pd4]0.5@Na3{NiII4[CuII2(Me3mpba)2]3}·56H2O (Figure ). Overall, the anionic nature of the MOF allowed a fine control of the number of inserted Pd2+ cations in the first step, which were occupying specific positions by interacting with the carboxylate oxygens from the network. Then, the confined space, as well as the mentioned controlled stoichiometry, allowed the formation of the small [Pd4]2+ clusters, homogeneously distributed within the walls of the MOF (Figure ). As expected, such naked clusters, possessing all four metal atoms outwardly exposed, exhibited outstanding catalytic activity, outperforming state-of-the-art metal catalysts in carbene-mediated reactions, also showing high yields (>90%) and turnover numbers (up to 100,000). However, considering the high cost of palladium, it would be highly desirable to achieve highly performing SNMCs with lower prices.

Figure 1

Crystal structure of the previously reported [Pd4]0.5@Na3{NiII4[CuII2(Me3mpba)2]3}·56H2O MOF.[21] Atoms constituting the framework and Pd atoms are represented by light yellow and blue spheres, respectively.

Results and Discussion

Herein, aiming at expanding these results to more affordable metals, we report the two-step post-synthetic preparation of Ag20 nanoclusters using the same MOF NiII2{NiII4[CuII2(Me3mpba)2]3}·54H2O (1) as the host matrix (Figure a). First, nickel(II) cations, located in the pores of 1, are exchanged by Ag+ ones, yielding the novel compound AgI4{NiII4[CuII2(Me3mpba)2]3}·51H2O (2) (Figure b). Then, after introducing NaBH4, the reduction process occurs to give the final compound [Ag20]@AgI2NaI2{NiII4[CuII2(Me3mpba)2]3}·48H2O (3) (Figure c). The whole process could be followed by single-crystal X-ray diffraction (SCXRD), unveiling certain details about the nanocluster formation, which constitutes one of the very few examples of MOF-hosted silver subnanometer clusters[26] and the first whose crystal structure could be elucidated.

Figure 2

Design approach showing the crystal structures of 1 (a), 2 (b), and 3 (c) showing the two-step post-synthetic process consisting of the exchange of the NiII cations in the pores of 1 by AgI ones to yield 2 and the reduction process to form the Ag20 clusters in 3. Virtual diameters of larger octagonal pores are 2.0 nm for MOFs 1–3, respectively. Copper and nickel atoms from the network are represented by cyan and orange polyhedra, respectively, whereas organic ligands are depicted as gray sticks. Orange, yellow, and blue spheres represent Ni, Na, and Ag atoms, respectively.

The nature of the final hybrid material, 3, containing Ag2 clusters (together with unreduced Ag+ ions) has been further confirmed by the combination of a variety of techniques including inductively coupled plasm–mass spectrometry (ICP–MS) (Table S1, Supporting Information), elemental mapping, powder X-ray diffraction (PXRD), thermogravimetric analysis (TGA), X-ray photoelectron spectroscopy (XPS), and scanning electron microscopy (SEM). The N2 adsorption isotherms at 77 K confirmed the permanent porosity of 2 and 3 (see below). Finally, as previously mentioned, SCXRD with synchrotron radiation allowed the resolution of the crystal structure of 2 and 3 (Table S2, Supporting Information) and the observation of ultrasmall silver dinuclear entities and surroundings within the solid in 3 (Figures and S1).

The anionic NiII4CuII6 open-framework structures in both 2 and 3 are isoreticular and crystallize in the P4/mmm space group of the tetragonal system. Compound 2 exhibits the Ag+ cations situated within the walls of the hydrophilic octagonal pores (virtual diameter of 2.0 nm), where they are stabilized by noncovalent interactions involving oxamate oxygen atoms [Ag+···Ooxamate of 2.72(1)–2.79(1) Å], with no evidence of previously Ni2+ cations of 1, thus indicating that they are completely exchanged by Ag+ ones (Figure b). Ag+ ion surroundings unveil interacting oxygen atoms likely belonging to nitrate anions (the whole fragments were not found from the ΔF map, see Supporting Information, Figures S1–S3) or solvent water molecules [Ag+···O distance range of 2.38(3)–2.56(3) Å], together with Ag+···Ag+ with a separation of 2.74(2) Å, which is shorter than the van der Waals contact distance, and they might be considered as precursors of the Ag20 dimers observed in 3. On the other hand, the crystal structure of 3 reveals the stabilization provided by the nano-confined space of the MOF on the as-synthesized Ag20 dimers, constricted into the walls of the hydrophilic octagonal channels of the anionic Ni42+Cu62+ open-framework net (Figures c, S4, and S5), together with unreduced Ag+···Ag+ dimers in smaller square pores (Figures c and S5). Further hydrated charge-counterbalancing alkali Na+ cations are retained in the preferential cationic sites, which stabilize the final material, showing an outstanding robustness (Figures S4 and S5). The poorer accessibility to the small square pores for solvated NaBH4 is most likely the reason for still unreduced Ag+···Ag+ dimers [blocked by Ag+···Ooxamate interactions at a distance of 2.84(1) Å] (Figure S5). Figure shows that Ag20 dimers [intradimer Ag···Ag distance of 3.19(1) Å] are well-fixed and stabilized inside the walls of the largest pores of the network by means of supramolecular interactions involving oxamate ligands [Ag···Ooxamate distance range of 2.93(1)–3.05(1) Å] and very weak connections with solvent molecules [Ag0···Owater distance of 3.25(1) Å].

Figure 3

One single channel of 3, showing supramolecular interactions involving oxamate ligands of the network (distances are reported in angstroms).

SEM coupled with energy-dispersive X-ray spectroscopy (EDX) measurements of 2 and 3 are given in Figures S6 and S7. EDX elemental mappings for Cu, Ni, Ag, and Na (3) elements show a heterogeneous spatial distribution of Ag atoms always located next to Cu and Ni atoms. Moreover, aberration-corrected HAADF–STEM (AC-HAADF–STEM) images are shown in Figure S8. They allow a direct visualization of both Ag2 dimers together with Ag1 species—most likely to silver atoms residing in smaller square channels.

TGA of 2 and 3 (Figure S9) established the solvent contents for both materials, which are reflected in their chemical formulas. PXRD patterns of 2 and 3 (Figure S10) indicate that the bulk samples are crystalline and pure, with no typical peaks of Ag0 nanoparticles. Indeed, experimental diffraction patterns of 2 and 3 are identical to the theoretical ones extracted from the SCXRD data. XPS spectra of compounds 2 and 3 are depicted in Figure S11. For 2, only possessing Ag+ cations, two bands at 367.6 and 373.6 eV, ascribed to Ag 3d5/2 and Ag 3d3/2 binding energies, respectively,[27] are observed (Figure S11a). In turn, for MOF 3—where SCXRD and elemental analyses suggest that Ag+ cations and Ag20 nanoclusters coexist—apart from the same Ag 3d5/2 and Ag 3d3/2 bands at 367.6 and 373.6 eV, respectively, indicative of Ag+, two additional peaks at 368.4 and 374.4 eV can be observed, which are attributed to reduced Ag0 atoms, with a 1:1 ratio respect to Ag+ (Figure S11b). CO-probe diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was conducted on MOF 3, run at 77 K to avoid any in situ reduction of Ag+ and to observe potential Ag0–CO species (Figure S12). The results show three main peaks, one at 1938 cm–1, consistent with CO bridged-bonded to Ag0 atoms,[28] a second at 2059 cm–1, attributable to Ag(CO)+ species,[29] and a last peak at 2043 cm–1, corresponding to free CO, after saturation. It is known that the adsorption of CO on Ag0 is lower than Ag+;[30] thus, the lower intensity of the former makes sense and could very well correspond to a 1:1 ratio between Ag oxidation states. These results strongly support that 50% of Ag+ present in 2 are reduced by NaBH4 forming Ag20 nanoclusters, whereas 50% of Ag+ cations remain untouched, occupying inaccessible sheltered interstitial positions where the reducing agent cannot accede (see the structural description). The N2 and CO2 adsorption isotherms for 1–3 confirm their permanent porosity (Figures S13 and S14). N2 adsorption isotherms for 1–3, with calculated Brunauer–Emmett–Teller[31] surface areas of 974, 1013, and 625 m2/g, respectively, indicate a very similar permanent porosity for 1 and 3, which is in agreement with their identical estimated virtual diameters of 2.0 nm. In turn, MOF 2 exhibits lower N2 adsorption despite having the same virtual diameter (2.0 nm), which could be due to a partial collapse of the structure upon solvent evacuation treatment. Remarkably, CO2 adsorption isotherms show a 66% uptake increase for 3, suggesting quadrupole interactions between CO2 molecules and Na+ cations.

The Buchner ring expansion reaction was attempted with catalytic amounts of 3. The results show that the reaction between toluene (4) and ethyl diazoacetate (5) (EDA) proceeds rapidly (30 min) in a very high yield, under standard reaction conditions (Figure top).[21,32] Blank experiments without any catalyst gave a 8% conversion, and the use of MOFs 1 and 2 as catalysts showed the lower catalytic activity of these MOFs than that of 3, with a 3 times lower initial rate for the former (Figure S15). Commercial Ag NPs on alumina only gave a 16% conversion, and remarkably, the state-of-the art catalyst for this reaction, that is, Rh2(OAc)4, gave a lower result than MOF 3 in this study, under identical reaction conditions (63% after the addition of 5 at once). An optimum >95% yield of product 6 was obtained after maintaining a low concentration of 5 during the reaction, which was achieved by adding a solution of 5 (in dichloromethane) into the reaction mixture using a syringe pump, instead of adding it at once. Otherwise, the unwanted dimerization reaction of 5 occurs. It is worth noting here that product 6 corresponds to the typical mixture of cycloheptatriene isomers, in accordance with previous results.[21,32] A hot filtration test, where the solid MOF catalyst 3 is removed from the reaction mixture at the reaction temperature (60 °C) at an early conversion (∼30%), shows that the catalytic active species are not present in solution within the experimental error (<10%, Figure bottom), which supports the relative stability of the solid catalyst. In accordance with this result, MOF 3 can be recovered at the end of the reaction by centrifugation, washed, and reused six times, maintaining a good catalytic activity (Figure S16). However, the catalytic yield of MOF 3 decreased to 30% after six uses, which could be due to the progressive (although minor) leaching of active species during the reaction.

Figure 4

Top: results for the Buchner ring expansion reaction between toluene 4 and ethyl diazoacetate 5 catalyzed by MOF 3. Bottom: Hot filtration test after adding 5 at once, error bars account for 5% uncertainty. (a) GC yield after syringe pump addition of 5, with the result after the addition of 5 at once given in parentheses; the result with Rh2(OAc)4 as a catalyst (5 mol %) is 63%.

The Buchner ring expansion reaction catalyzed by 3 could be expanded to other aromatic substrates (Table ). Halogenated (products 13, 14, and 17), cyano (product 15), methoxy (product 16), and ortho-disubstituted (product 18) aromatic compounds react with 5 in good to excellent yields and still in short reaction times (<2 h). These results should be remarked upon as this is difficult to find in the open literature Ag-catalyzed Buchner ring expansion reactions.[32−35] The slightly lower results obtained for the bigger substrates 11 and 12 can be explained by the size discrimination associated to the microporous structure of MOF 3. To check this, 1,3,5-triisopropylbenzene 19 was tested as a substrate for the reaction, and product 20 was not found (Figure S17). Besides, diffusion tests with varying stirring speeds confirmed that the initial reaction rate is dependent on the stirring, which confirms that the reaction occurs inside the MOF’s channels (Figure S18). Indeed, 1,3,5-triisopropylbenzene 19 reacted when the Rh2(OAc)4 salt was used as a catalyst (3 mol %, 69% yield, Figure S17). Therefore, for MOF- or zeolite-stabilized Pd clusters,[21,36] the formation of Ag2 clusters in micro-structured solids enables not only the catalytic use of this metal in this complex organic synthetic reaction but also its recovery and reuse.[36,37] Interstitial Ag+ cations must be excluded as catalytically active species since reagents cannot access these sites; besides, leaching does not occur under the reaction conditions employed. Finally, the integrity of MOF 3 is ensured by PXRD (Figure S19) and XPS (Figure S20) after catalytic experiments.

Table 1

Results for the Buchner Ring Expansion Reaction between Different Aromatics 7–12 and 5, Catalyzed by MOF 3a

entry	aryl substrate	substituent(s)	product	yield (%)a
1	7	Cl	13	66.5
2	8	Br	14	64.9
3	9	CN	15	92.8
4	10	OMe	16	82.6
5	11	CH2Br	17	50.2
6	12	Me-ortho-F	18	72.7

GC yields after syringe pump addition of 5.

Conclusions

In conclusion, ligand-free Ag20 clusters have been prepared, stabilized and characterized within an MOF and used as efficient and recoverable catalysts for the Buchner ring expansion reaction. These results expand the toolkit of readily affordable Ag species for heterogeneous catalysis in organic synthesis.

Experimental Section

Preparation of AgI4{NiII4[CuII2(Me3mpba)2]3}·51H2O (2)

Well-formed deep green prisms of 2, which were suitable for XRD, were obtained by immersing crystals of 1 (ca. 0.0015 mmol) for 48 h in 5 mL of a AgNO3 aqueous solution (0.004 mmol), which was replaced three times. A multigram scale procedure was also carried out by using the same synthetic procedure but with greater amounts of both, a powder sample of compound 1 (ca. 20 g, 5.8 mmol) and AgNO3 (2.38 g, 14.0 mmol), with the same successful results and a very high yield (20.33 g, 96%). Anal.: calcd (%) for Cu6Ni4Ag4C78H162N12O87 (3707.7): C, 25.27; H, 4.40; N, 4.53. Found: C, 25.34; H, 4.37; N, 4.59. IR (KBr): ν = 3008, 2961 and 2926 cm–1 (C–H), 1601 cm–1 (C=O).

Preparation of [Ag2]AgI2NaI2{NiII4[CuII2(Me3mpba)2]3}·48H2O (3)

Both crystals (ca. 5 mg) and a powder polycrystalline sample of 2 (ca. 10 g) were suspended in 50 mL of a H2O/CH3OH (1:2) solution, to which an excess of NaBH4, divided in 26 fractions (each fraction consisting of 1 mol of NaBH4 per mole of 2 to give a final NaBH4/MOF molar ratio of 26 or a NaBH4/Ag atom molar ratio of 13, which is the same), was added progressively in the space of 72 h. After each addition, the mixture was allowed to react for 1.5 h. After this period, samples were gently washed with a H2O/CH3OH solution and filtered on paper, giving high yields (ca. 98%). Anal. Calcd (%) for Cu6Ni4Ag4Na2C78H156N12O84 (3699.61): C, 25.32; H, 4.25; N, 4.54. Found: C, 25.28; H, 4.17; N, 4.59. IR (KBr): ν = 3011, 2971 and 2928 cm–1 (C–H), 1605 cm–1 (C=O).

Gas Adsorption

The N2 and CO2 adsorption–desorption isotherms at 77 and 273 K were obtained on polycrystalline samples of 2 and 3 using a BELSORP-mini-X instrument. Samples were first activated with methanol and then evacuated at 348 K during 19 h under 10–6 Torr prior to their analysis.

Microscopy Measurements

SEM elemental analysis was carried out for 2 and 3 using a HITACHI S-4800 electron microscope coupled with an EDX detector. Data were analyzed using QUANTAX 400.

HAADF–STEM characterization for 3 was performed using an HAADF-FEI-TITAN G2 electron microscope. 5 mg of the material was re-dispersed in 1 mL of absolute EtOH. Carbon-reinforced copper grids (200 mesh) were submerged into the suspension 30 times and then allowed to dry on air for 24 h.

PXRD Measurements

Polycrystalline samples of 2 and 3 were introduced into 0.5 mm borosilicate capillaries prior to being mounted and aligned on an Empyrean PANalytical powder diffractometer, using Cu Kα radiation (λ = 1.54056 Å). For each sample, five repeated measurements were collected at room temperature (2θ = 2–60°) and merged in a single diffractogram.

XPS Measurements

Samples of 2 and 3 were prepared by sticking, without sieving, the samples onto a molybdenum plate using a scotch tape film, followed by air-drying. Measurements were performed on a K-Alpha XPS system using a monochromatic Al K(alpha) source (1486.6 eV). As an internal reference for the peak positions in the XPS spectra, the C 1s peak has been set at 284.8 eV.

DRIFTS of Adsorbed CO

DRIFTS using CO as a probe molecule was used to evaluate the electronic properties of MOF 3. The experiments have been carried out in a homemade IR cell able to work in the high and low (77 K) temperature ranges. Prior to CO adsorption experiments, the sample was evacuated at 298 K under vacuum (10–6 mbar) for 1 h. CO adsorption experiments were performed at 77 K in the 0.2–20 mbar range. Spectra were recorded once complete coverage of CO at the specified CO partial pressure was achieved. Deconvolution of the IR spectra has been performed in the Origin software using Gaussian curves where the full width at half-maximum of the individual bands has been taken as a constant. The peak areas are normalized to the sample weight.

X-ray Crystallographic Data Collection and Structural Refinement

Crystals of 2 and 3 with ca. 0.06 × 0.08 × 0.08 and 0.08 × 0.12 × 0.12 mm dimensions, respectively, were selected and mounted on a MITIGEN holder in Paratone oil and very quickly placed on a liquid nitrogen stream cooled at 90 K to avoid the possible degradation upon dehydration. Diffraction data for 2–3 were collected using synchrotron at the I19 beamline of the DIAMOND at λ = 0.6889 Å. Crystallographic details can be found in the Supporting Information

General Catalytic Reaction Procedure

MOF 3 (9.5 mg, 10 mol % Ag) was weighed in a 2 mL vial with a magnetic stirrer, and the aromatic substrate (0.8 mL) was added. Then, the vial was placed in a pre-heated oil batch at 60 °C, and ethyl diazoacetate 5 (0.1 mmol) was added, either at once or using a syringe pump (solution in dichloromethane). The mixture was allowed to react for 0.5–2 h. After the reaction was complete, filtration was carried out to separate the solid catalyst. The reaction mixture was analyzed by gas chromatography (GC) and GC–MS. Further details can be found in the Supporting Information.