Carbon Monoxide Promotes the Catalytic Hydrogenation on Metal Cluster Catalysts

Size effect plays a crucial role in catalytic hydrogenation. The highly dispersed ultrasmall clusters with a limited number of metal atoms are one candidate of the next generation catalysts that bridge the single-atom metal catalysts and metal nanoparticles. However, for the unfavorable electronic property and their interaction with the substrates, they usually exhibit sluggish activity. Taking advantage of the small size, their catalytic property would be mediated by surface binding species. The combination of metal cluster coordination chemistry brings new opportunity. CO poisoning is notorious for Pt group metal catalysts as the strong adsorption of CO would block the active centers. In this work, we will demonstrate that CO could serve as a promoter for the catalytic hydrogenation when ultrasmall Pd clusters are employed. By means of DFT calculations, we show that Pdn (n = 2‐147) clusters display sluggish activity for hydrogenation due to the too strong binding of hydrogen atom and reaction intermediates thereon, whereas introducing CO would reduce the binding energies of vicinal sites, thus enhancing the hydrogenation reaction. Experimentally, supported Pd2CO catalysts are fabricated by depositing preestablished [Pd2(μ-CO)2Cl4]2- clusters on oxides and demonstrated as an outstanding catalyst for the hydrogenation of styrene. The promoting effect of CO is further verified experimentally by removing and reintroducing a proper amount of CO on the Pd cluster catalysts.


Preparation of TiO2-EG and TiO2
Ultrathin TiO2-EG nanosheets were synthesized corresponding to the reported method. [1] 2 mL TiCl4 was carefully introduced into 60 mL EG and stirred under ultrasonic until homogeneous light yellow solution was obtained, then the solution was transferred into 100 mL Teflon-lined stainless-steel autoclave and heated at 150 °C for 4h. The obtained white colloids were centrifuged and washed with water for 3 times.
The EG-free TiO2 that was obtained by heating TiO2-EG at 350 °C for 2 h under air atmosphere in muffle with a ramp rate of 1 °C/min. Preparation of Pd2CO/TiO2, Pd2/TiO2-cal, Pd2CO/Al2O3. 10 μL H2PdCl4 (1 M) was introduced into 1 mL THF in a glass bottle, and the solution was kept stirred under 0.2 MPa CO at room temperature till the color of the solution turned into bright yellow.
Then the solution was introduced dropwise into 20 mL THF dispersions of the supports (500 mg TiO2 or Al2O3) under stirring, then the solvent was removed by centrifuge and dried under vacuum at room temperature, the as-obtained catalysts were denoted as Pd2CO/TiO2 and Pd2CO/Al2O3.
The releasing of CO2 upon introducing TiO2 into the solution of H2[Pd2(μ-CO)2Cl4] was recorded by MX6 iBrid Gas Detector (Industrial Scientific) equipped with a CO2 detector. The UV-vis spectrum of the H2[Pd2(μ-CO)2Cl4]/THF solution and as-obtained Pd2CO/TiO2 were performed in dilute and recorded by Shimadzu UV2600 instrument.
In order to remove the adsorbed CO, Pd2/TiO2-cal was obtained by heating Pd2CO/TiO2 at 350 °C for 1h (ramp rate, 2 °C/min) under static air atmosphere. Before applying in catalysis, Pd2/TiO2-cal was treated with H2 at room temperature for 15 min.
Preparation of Pd1/TiO2-EG and Pd1/TiO2-cal The single-atom dispersed Pd catalysts were synthesized following the procedures reported in our group previously. [2,3]

Preparation of Pd nanosheets and Pd nanocubes
The colloidal Pd nanosheets (Pd NSs) and Pd nanocubes (Pd NCs) with preferential (111) and (100) exposed surface respectively were prepared following the procedures reported previously in our group. [4,5] For Pd nanosheets, 30 μL 1 M H2PdCl4 was added into 10 mL DMF and stirred under 0.1 MPa CO atmosphere at room temperature for 15 min, then 1 mL H2O was introduced and stirred for another 15 min without CO. The colloidal Pd nanosheets were centrifuged and washed with EtOH and acetone for 3 times. For Pd nanocubes, 8 mL aqueous solution containing poly(vinylpyrrolidone) (PVP, MW= 55,000, 105 mg), L-ascorbic acid (60 mg), KBr (200 mg) were placed in a 20 mL vial, and pre-heated in air under magnetic stirring at 80 °C for 10 min. Then the aqueous solution of Na2PdCl4 (57 mg in 3 mL H2O) was added into the vial and kept at 80 °C for another 3 h before it was cooled down to room temperature. The product was collected by centrifugation and washed with water-acetone mixture.
Preparation of 0.5 wt% Pd/Al2O3 200 mg Al2O3 was dispersed in 6 mL acetone, then Pd(OAc)2/Acetone solution was introduced, the Pd mass loading was controlled to be 0.5 wt%. The samples dried at 40 °C overnight, the obtained 4 / 39 powders were then calcined (ramp rate, 2 °C/min) in static air at 300 °C for 2h. The catalysts were reduced in H2 at 100 °C for 1 h before applied in hydrogenation.

Catalysis tests
For styrene hydrogenation, a proper amount of catalyst was introduced in 10 mL EtOH and stirred at 30 °C and 0.1 MPa H2 atmosphere for 10 min, then 0.55 mL (5 mmol) styrene was added and started to measure the conversion of styrene on a KB-WAX chromatographic column equipped gas chromatograph (GC). The ratio of substrate to catalyst (S/C) was controlled, for 0.2 wt% Pd2CO/TiO2 Pd2/TiO2-cal, Pd1/TiO2, Pd1/TiO2-cal and Pd nanosheets S/C=50,000. The activation energy was measured by adjusting the temperature to 20 °C 30 °C, 40 °C and 50 °C. In order to evaluate the effect of the different amount of CO on Pd2/TiO2-cal, Pd1/TiO2-EG and Pd1/TiO2-cal, different amount of dilute CO was introduced to the bottle before styrene hydrogenation was carried out. The effect of CO on the colloid Pd NSs and Pd NCs was evaluated by soaking them in 5% CO/Ar in EtOH for 15 min and then flushing with H2 for 15 min, then styrene was introduced. The mass specific activity was calculated based on the total amount of Pd in the corresponding catalyst as a conversion level lower than 20%.
For the gas-powder phase ethylene hydrogenation. 2 mg 0.5 wt% Pd/Al2O3 diluted with 20 mg Al2O3 was loaded in a glass tube (8 mm diameter). The sample was reduced at 100 °C with 30 mL/min H2 for 30 min before cooled down to 30 °C. Then, the catalytic hydrogenation was carried out with the feed gas flow of 25 mL/min C2H4 and 75 mL/min H2, the conversion of C2H4 was determined by online GC. For CO adsorption, the catalyst was treated with 5% CO/Ar (30 mL/min) at 30 °C for 15 min, then flushed with the feed gas at 60 °C for 30 min before cooled down to 30 °C. For CO desorption, the catalyst was treated with feed gas at 150 °C for 30 min, then cooled down to 30 °C again.
The production of H2O2 was performed following the procedure reported in the literature. [6] Typically, 100 mg 0.2 wt% Pd2CO/Al2O3 were dispersed in 10 mL TOP and then 2.4 g 2-eAQ diluted in 10 mL toluene were added. The hydrogenation of 2-eAQ was performed under 0.2 MPa H2 at 30 °C. 1 mL of the solution was taken out and introduced into 10 mL of 1 M H3PO4 every 30 min, the obtained liquid was stirred in air for 30 min at room temperature. 5 mL of 20 % H2SO4 was added before the amount of H2O2 was titrated by 0.02 M KMnO4.

Transmission Electron Microscope (TEM) characterization
For TEM characterization, the samples were dispersed in EtOH and dropping onto 300-mesh carboncoated copper grids and the solvent was evaporated in air subsequently. TEM characterization and energy dispersive X-ray spectroscopy (EDX) was carried out on a TECNAI F30 transmission electron microscope operating at 300 kV.

HAADF-STEM characterizations
High-resolution transmission electron micrographs (HRTEM) were performed on JEOL 200F transmission electron microscope operated at 200 keV. Both annular-bright-field (ABF) and highangle annular-dark-field (HAADF) images were acquired with the illumination semi-angle of 25 mrad and probe current of 100 pA. The dwell time for image acquisition was set at 10 micro second per pixel to ensure desirable signal to noise ratio. The attainable spatial resolution of microscope was 78 pm with a probe spherical-aberration corrector. The collection angles for the ABF and HAADF images were fixed at 12-25 mrad and 90-250 mrad, respectively.

Powder X-ray diffraction (XRD) characterizations
The XRD experiments were carried out on Rigaku Ultima IV using Cu Kα radiation. The operation voltage was set at 40 kV, the current was set at 30 mA. The scan speed was set at 10 °/min

Zeta-potential characterizations
Zeta-potential experiments were tested on Nano-ZS zetasizer (Malvern Instruments, UK). 2.0 mg TiO2 was dispersed into 10 mL water with varied pH value (2 -12) and 3 times tests were repeated for every pH point.

N2 adsorption and desorption experiments.
The BET (Brunauer-Emmett-Teller) surface area of oxides was measured by N2 adsorption-desorption experiments on Micromeritics ASAP2020 at liquid nitrogen temperature. The samples were degassed at 200 °C for 3h.

Temperature Programmed Desorption-mass spectrometry (TPD-MS) characterization
The TPD-MS experiment was performed on a home-made TPD-TOF analyzer. 3 mg of 1 wt% Pd2CO/TiO2 was added into a small tube which will be heated by the surrounded heating coil. A K-type thermocouple was put inside the sample tube, and insulated from samples to measure the temperature. The heating coil was powered by a precise electric source, and adjusted at interval of 10 mV. The temperature of sample tube was ramping from room temperature to 450 °C with a rate of 5 K/min. The desorbed spices were ionized by a UV lamp at position very close to the sample tube, with phonon energy of 10.6 eV, and then transferred to TOF analyzer by an ion optical system. The TOF analyzer had a resolution of more than 5000, and the sensitivity of ppb level. All those steps were processed in high vacuum of about 3×10 -5 Pa. The mass spectrum and sample temperature were acquired and recorded every second. Each spectrum was an accumulation of 10000 spectra gathered at interval of 100 s.

X-ray absorption spectroscopy (XAS) measurements and data processing
The X-ray absorption experiments were carried out at the XAS station (BL14W1) of the Shanghai Synchrotron Radiation Facility (SSRF). The electron storage ring was operated at 3.5 GeV. Si (311) double-crystal was used as the monochromator, and the data was collected using solid-state detector under ambient conditions. The beam size was limited by the horizontal and vertical slits with the area of 1 ×4 mm 2 during XAS measurements. The X-ray absorption of Pd foil at Pd K-edge of was measured for energy calibration. All the samples were sealed in N2 atmosphere before taking to the station, and the data was recorded under air atmosphere. The obtained XAFS data was processed in Athena (version 0.9.25) for background, pre-edge line and post-edge line calibrations. Then Fourier transformed fitting was carried out in Artemis (version 0.9.25). [7] The k 2 weighting, k-range of ~3 -12 Å -1 and R range of 1-3 Å were used. The model of bulk Pd and PdO were used to calculate the simulated scattering paths. The coordination number of Pd-Pd for Pd foil was fixed at 12 to determine the amplitude reduction factor (S0 2 =0.87). Then the four parameters, coordination number, bond length, Debye-Waller factor and E0 shift (CN, R, σ 2 , ΔE0) were fitted without anyone was fixed, constrained, or correlated.

Pd dispersion determination
Pd dispersion was measured by CO titration carried out on a Micromeritics Auto Chem II 2920 chemical adsorption instrument equipped with a TCD detector. The samples were reduced at 100 °C in H2 for 1h before cooled down to 50 °C. Then flashed with He for 1h and CO titration was performed with 5% CO/He. The Pd dispersion was calculated based on the consumed CO molecules and ratio between Pd and CO was 2:1.

Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS)
DRIFTS was carried out on ThermoFisher IS50 Fourier Transform Infrared spectrometer equipped with MCT detector. For Pd2CO/TiO2, the spectrum was recorded under ambient condition, and KBr powder was used as the background. For 0.5 wt% Pd/Al2O3, the sample were loaded in an in-situ chamber (Harrick) and treated with 5% H2/Ar (30 mL/min) at 100 °C for 1 h, then cooled down to room temperature and the background spectrum was collected before treated with 5% CO/Ar for 20 min. The chamber was flushed with Ar for 20 min. The chamber was heated up to 30, 60 and 100 °C with 25 mL/min C2H4 and 75 mL/min H2 for ~10 min, then cooled down to room temperature and flushed with Ar for 10 min before recording. For Pd2/TiO2-cal, after flushing with Ar and recording the background spectrum, the sample was treated with 5% CO/Ar for 20 min and then flushing with Ar before recording the spectrum.
Electron Paramagnetic Resonance (EPR) X-band EPR spectra were recorded by a Bruker EMX-10/12 microspectrometer at 90 K, with an operation frequency of 9.45 GHz and a microwave power of 19.9 mW. In a typical measurement, 50 mg sample was used.

Temperature programed desorption
AutoChem 2920 II equipped with TCD detector was used to perform these measurements. For H2-TPD, 0.5 wt% Pd/Al2O3 was first reduced with 5% H2/Ar at 100 °C for 1h after calcined in static air.
The sample was pretreated under 200 °C with Ar flow for 1 h before cooling down to 50 °C. The sample was further treated with 5% H2/Ar and Ar for 30 min before the recording, respectively. CO-TPD was performed following a similar procedure except 5% CO/Ar was applied after cooling down to 50 °C.
For the H2-TPD with pre-adsorbed CO, the sample treated with 5% CO/Ar and 5% H2/Ar for 30 min stepwise before flushing with Ar and recording.

Computational details
Spin-polarized calculations were carried out with the Vienna ab initio simulation package (VASP). [8,9] The electron exchange and correlation were treated with the generalized gradient approximation using PBE functional. [10] The valence electrons were described by plane wave basis sets with a cutoff energy of 400 eV, and the core electrons were replaced by the projector augmented wave pseudopotential. [11,12] Geometries of minima and transition states (TSs) were converged to a 8 / 39 residual force smaller than 0.03 eV/Å. The TSs were determined using the nudged elastic band (NEB) approach, [13] with a subsequent quasi-Newton optimization to refine the TS' structures and energies.
All the local minima and TSs were verified by vibrational frequency calculations.
For Pdn clusters (n=2, 3, 4, 7, 13, 55, 147)，the geometry structures with the highest symmetry were chosen, [14,15] as shown in Figure 1. To avoid image interaction，the shortest distances between the image clusters were set to be more than 10 Å . In these cases, the Gamma point only calculations were performed. For the Pd(111) and Pd(100) surfaces, (34) supercells with five atomic layers were used.
The vacuum regions between the slabs were set to 15 Å, and the k-points sampling was generated following the Monkhorst-Pack procedure with a 3×3×1 mesh. For the Pd surface models, the bottom two layers were fixed at a bulk truncated position, while the top three layers and the adsorbates were allowed to be fully relaxed.
For the practical use, the Pd clusters would be loaded on the oxide surfaces, such as anatase TiO2(010). Computationally, to model the anatase TiO2 (010) surface, a five-layer p(1×4) slab was used and the utmost surface was fully hydroxylated ( Figure S1). Since GGA was not able to correctly describe the electronic structure of Ti 4+ , we adopted the GGA+U approximation with the Dudarev                   Figure S23). The peaks with wavelength over 400 nm were related to the metal-metal d-d electron transfer. [19,20] The wavelength of this feature is related to the nuclearity/size of the cluster and the distance between Pd atoms. The predominant peak at ~420 nm indicated that the small Pd clusters PdnCO with n close to 2-3 were dominated in the as-obtained catalyst. [21,22]         Pd2-cal were hard to be measured due to the low mass loading, the dispersions of Pd on these catalysts were estimated to be close to 1. The calculated activity and TOF demonstrated that the supported Pd2CO clusters were as active as the surface Pd in large NPs. More importantly, the high Pd dispersion made the supported Pd2CO exhibit 4-6 times higher mass-specific activity.          Pd/Al2O3. The H2 TPD for blank Al2O3 (black) was also shown for comparison. The shoulder peak with desorption temperature of 175 °C in H2 TPD was the desorption of H2 from Pd, the higher-temperature desorption should be related to the dehydration of the support. For the CO TPD, despite the desorption of CO was hard to discriminate from the background, no desorption feature was observed below 150 °C.