Soluble/MOF-Supported Palladium Single Atoms Catalyze the Ligand-, Additive-, and Solvent-Free Aerobic Oxidation of Benzyl Alcohols to Benzoic Acids

Metal single-atom catalysts (SACs) promise great rewards in terms of metal atom efficiency. However, the requirement of particular conditions and supports for their synthesis, together with the need of solvents and additives for catalytic implementation, often precludes their use under industrially viable conditions. Here, we show that palladium single atoms are spontaneously formed after dissolving tiny amounts of palladium salts in neat benzyl alcohols, to catalyze their direct aerobic oxidation to benzoic acids without ligands, additives, or solvents. With this result in hand, the gram-scale preparation and stabilization of Pd SACs within the functional channels of a novel methyl-cysteine-based metal–organic framework (MOF) was accomplished, to give a robust and crystalline solid catalyst fully characterized with the help of single-crystal X-ray diffraction (SCXRD). These results illustrate the advantages of metal speciation in ligand-free homogeneous organic reactions and the translation into solid catalysts for potential industrial implementation.


Experimental Section
Materials. All chemicals were of reagent grade quality. They were purchased from commercial sources and used as received.
The reaction mixture was washed with acetone (150 mL) and diethyl ether (100 mL) and further concentrated, under reduced pressure, to afford the methyl ester derivative of the (S)-methyl-(L)-cysteine amino acid, which was used in the next step without further purification.   A multigram scale procedure was also developed by using the same synthetic procedure but using a higher amount of a polycrystalline sample of 3 (2 g, 1.15 mmol), which were suspended a H2O/CH3OH  Gas adsorption. The N2 adsorption-desorption isotherms at 77 K were carried out on crystalline samples of 3, 4 and 5 with a Micromeritics ASAP2020 instrument.
Samples were evacuated at 70 ºC during 24 hours under 10-6 Torr prior to their analysis.
X-ray Powder Diffraction Measurements. Polycrystalline samples of 3, 4 and 5 were introduced into 0.5 mm borosilicate capillaries prior to being mounted and aligned on a 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. A polycrystalline sample of 5 was also measured after catalysis following the same procedure.

X-ray photoelectron spectroscopy (XPS) measurements. Samples 4 and 5
were prepared by sticking, without sieving, the MOF onto a molybdenum plate with scotch tape film, followed by air drying. Measurements were performed on a K-Alpha™ X-ray Photoelectron Spectrometer (XPS) System using a monochromatic Al S6 K(alpha) source (1486.6 eV). As an internal reference for the peak positions in the XPS spectra, the C1s peak has been set at 284.8 eV.

Microscopy measurements.
Electron microscopy studies were performed on a FEI Titan Themis 60-300 Double Aberration Corrected microscope working at 200kV.
The aberrations of the condenser lenses were corrected up to fourth-order using the Zemlin tableau to obtain a sub-Angstrom electron probe. A condenser aperture of 50 µm yielding an electron probe with a convergence angle of 20 mrad was used. In order to avoid sample modification under the electron probe a beam current of 0.025 nA was used. Chemical XEDS maps were collected at medium magnifications (c.a. 100k to 200k) by STEM-XEDS using the high-efficiency SuperX G2 detection system equipped in the microscope, which integrates four windowless detectors surrounding the sample and high performance signal-procedure hardware. STEM-XEDS was used to map the spatial distribution Pd species. To limit the damage by the electron beam, a fast image recording protocol was used by combining a beam current of 25 pA, and 4s dwell time and an automated fine-tuning alignment of A1 and C1 using the OptiSTEM software.
X-ray crystallographic data collection and structure refinement. Crystals of 3, 4 and 5 with 0.14 x 0.10 x 0.10, 0.06 x 0.05 x 0.05, and 0.05 x 0.05 x 0.05 as dimensions were selected and mounted on a MiTeGen MicroMount in Paratone oil and very quickly placed on a liquid nitrogen stream cooled at 100 K, to avoid the possible degradation upon dehydration or exposure to air. Diffraction data for 3 were collected on a Bruker-Nonius X8APEXII CCD area detector diffractometer using graphite-monochromated Mo-Kα radiation ( = 0.71073 Å), whereas data for 4 and 5 using synchrotron radiation at I19 beamline of the Diamond Light Source at = 0.6889 Å. The data were processed through SAINT 1 reduction and SADABS 2 multi-scan absorption (3) or xia2 3 (4 and 5) software. The structures were solved with the SHELXS structure solution program, S7 using the Patterson method. The model was refined with version 2018/3 of SHELXL against F 2 on all data by full-matrix least squares. 3 As reported in the main text, the robustness of the 3D network, allowed the resolution of the crystal structure of both 4 and 5 adsorbates, being their crystals suitable for X-ray diffraction, even over one-and two-step process, after a crystal-to-crystal transformation. For these reasons it is reasonable to observe a diffraction pattern sometimes affected by expected internal imperfections of the crystals [likely at the origin of some Alert level B for 4-5 in checkcifs related to U(eq) value of some atoms] and thus a quite expected difficulty to perform a perfect correction of anisotropy, mainly affected by highly flexible thioether chains as terminal moiety (vide infra).
In all samples, all non-hydrogen atoms of the MOF net, except the dimethyl thioether chains from the mecysmox ligand, and Pd atoms (4 and 5), were refined anisotropically. The use of some C-C and C-S bond lengths restrains as well as Pd-S (5), during the refinements, has been reasonable imposed and related to extraordinary flexibility of dimethyl thioether chains from the methylcysteine residues, that are dynamic components of the frameworks (see Figures S8 and S9). In the refinement of crystal structures some further restrains, to make the refinement more efficient, have been applied, for instance ADP components have been restrained to be similar to other related atoms, using SIMU 0.04 for disordered sections or EADP for group of atoms expected to have essentially similar ADPs. All the hydrogen atoms of the net were set in calculated position and refined isotropically using the riding model. Disordered sites for atoms C4S in refinement of 4 and 5, belonging to the dimethyl thioether chains, reside in special position resulting statistically disordered for symmetry reason. S8 The occupancy factors of Pd atoms have been defined in agreement with SEM and ICP-MS results (see Table S2). The high thermal factors are most likely related to the porosity of the net hosting Pd 2+ ions/Pd 0 atoms within very large pores.
The solvent molecules in 3-5 were highly disordered and have not been found from the F map neither the NH3 molecules and Clanions in 4-5 belonging to the mononuclear complex [Pt(NH3)4]Cl2 inserted within pores. The quite large channels featured by this series of MOFs likely account for that [see Figures S9 and S11].
Consequently, in 3-5, the contribution to the diffraction pattern from the highly disordered water and NH3 molecules/Clanions located in the voids was subtracted from the observed data through the SQUEEZE method, implemented in PLATON. 5 A summary of the crystallographic data and structure refinement for the three compounds is given in Table S3. The comments for the alerts A and B are described in the CIFs using the validation reply form (vrf). CCDC reference numbers are 1995182-1995184 for 3-5, respectively.
The final geometrical calculations on free voids and the graphical manipulations were carried out with PLATON 5 implemented in WinGX, 6 and CRYSTAL MAKER 7 programs, respectively. Energy scale has been calibrated by measuring a Pd foil. Several scans were acquired in continuous mode to ensure spectral reproducibility and good signal-to-noise ratio. Data reduction has been done using the Demeter program suite: 13 raw data has been normalized by subtracting and dividing pre-edge and post-edge backgrounds as low order polynomial smooth curves. The local structure of the sample has been than refined using the EXAFS signal in the k range 3:10 Å -1 . Corresponding Fourier transformed curved are shown in Figure 2. Theoretical models for further data analysis have been obtained from palladium NPs and PdS structures using the FEFF6 code and fitted 14 to the EXAFS spectra by adjusting the structural parameters (i.e. coordination number, distances, disorder factors).

S11
Typical procedure for the oxidation of benzyl alcohol 1a in open air. In a 10 ml glass vial equipped with a stirring bar, benzyl alcohol 1a (1.96 mmol) was charged with the palladium catalyst, and the vial was connected to a condenser. The mixture was placed in a pre-heated metal heating plate at 150°C and stirred at 450 rpm for the indicated time. Aliquots were extracted from the reaction at different times, mesitylene (3 ml) was added as an external standard, and the mixture was analysed by GC. S12    Table S6. DFT data for the interaction of CO with Pd(0) atoms and PdCl2 within the MOF, and with isolated Pd13 and PdCl2 (gas) for comparison. Pd0-A-C refers to the different interaction models presented in Figure 7 in the main text.                             Figure S31. Initial rate-concentration plots for the reaction catalyzed by MOF 5. Reaction orders for MOF 5 and benzyl alcohol 1a are similar (+1), while the reaction order for O2 is likely 0.