One‐Pot Cooperation of Single‐Atom Rh and Ru Solid Catalysts for a Selective Tandem Olefin Isomerization‐Hydrosilylation Process

Abstract Realizing the full potential of oxide‐supported single‐atom metal catalysts (SACs) is key to successfully bridge the gap between the fields of homogeneous and heterogeneous catalysis. Here we show that the one‐pot combination of Ru1/CeO2 and Rh1/CeO2 SACs enables a highly selective olefin isomerization‐hydrosilylation tandem process, hitherto restricted to molecular catalysts in solution. Individually, monoatomic Ru and Rh sites show a remarkable reaction specificity for olefin double‐bond migration and anti‐Markovnikov α‐olefin hydrosilylation, respectively. First‐principles DFT calculations ascribe such selectivity to differences in the binding strength of the olefin substrate to the monoatomic metal centers. The single‐pot cooperation of the two SACs allows the production of terminal organosilane compounds with high regio‐selectivity (>95 %) even from industrially‐relevant complex mixtures of terminal and internal olefins, alongside a straightforward catalyst recycling and reuse. These results demonstrate the significance of oxide‐supported single‐atom metal catalysts in tandem catalytic reactions, which are central for the intensification of chemical processes.

: Raman spectra for CeO 2 and selected Pt/CeO 2 catalysts with various metal contents obtained by oxidative redispersion at 1073 K. Spectra have been normalized to the Raman signal for the triply degenerate F2g mode of the CeO 2 lattice (band at ca. 462 cm -1 ).
The Raman spectrum for pure CeO 2 shows a prominent band at a Raman shift of 462 cm -1 , which corresponds to the triply degenerate F2g mode of the fluorite-type CeO 2 lattice. A notably weaker and broader band at ca. 590-600 cm -1 can be ascribed to Frenkel-type oxygen vacancies in CeO 2 . [11] On incorporation of Pt (at metal contents ≤1.0 Pt at nm -2 ), a broad signal developed, which can be deconvoluted into two bands peaking at 660 and 700 cm -1 , respectively. In this region, active Raman modes for Pt-O-Ce are to be expected. [12] Hence the Raman results provide evidence for the creation of Pt-O-Ce linkages upon annealing. S12 Figure S2: Raman spectra for CeO 2 and selected Rh/CeO 2 catalysts with various Rh contents obtained by oxidative redispersion at 1073 K. Spectra have been normalized to the Raman signal for the triply degenerate F2g mode of the CeO 2 lattice (band at ca. 462 cm -1 ).
The Raman spectrum for pure CeO 2 shows a prominent band at a Raman shift of 462 cm -1 , which corresponds to the triply degenerate F2g mode of the fluorite-type CeO 2 lattice. A notably weaker and broader band at ca. 590-600 cm -1 can be ascribed to Frenkel-type oxygen vacancies in CeO 2 . [11] On incorporation of Rh (at metal contents ≤1.0 Rh at nm -2 ), a broad signal developed, which can be deconvoluted into two bands peaking at 574 and 649 cm -1 , respectively, i.e. the spectral region for Raman active modes associated to Rh-O-Ce species. [13] Hence the Raman results provide evidence for the creation of Rh-O-Ce linkages upon annealing. S13 Figure S3: Raman spectra for CeO 2 and selected Ru/CeO 2 catalysts with various Ru contents obtained by oxidative redispersion at 1073 K. Spectra have been normalized to the Raman signal for the triply degenerate F2g mode of the CeO 2 lattice (band at ca. 462 cm -1 ).
The Raman spectrum for pure CeO 2 shows a prominent band at a Raman shift of 462 cm -1 , which corresponds to the triply degenerate F2g mode of the fluorite-type CeO 2 lattice. A notably weaker and broader band at ca. 590-600 cm -1 can be ascribed to Frenkel-type oxygen vacancies. [11] On incorporation of Ru (at metal contents ≤1.0 Ru at nm -2 ), a weak band developed at ca. 709 cm -1 which can be ascribed to Ru-O-Ce species. [14] Hence, the Raman results provide evidence for the creation of Ru-O-Ce linkages upon annealing. S15 Figure S5: X-ray diffractograms (MoK 1 radiation) for a series of Rh/CeO 2 catalysts, with various Rh surface coverages, obtained by oxidative redispersion at 1073 K. Star labels denote diffractions corresponding to Rh 2 O 3 (ICDD: 00-041-0541).   For Pt/CeO 2 catalysts, a Pt4f 7/2 binding energy (BE) of 72.6 eV was determined for metal coverages ≤2 Pt nm -2 , indicating the presence of Pt(II) as the single species. Increasing the Pt content to 5.0 Pt nm -2 led to the detection of additional contributions at BEs of 74.4 eV and 70.8 eV, corresponding to Pt(IV) oxide species and metallic Pt(0), respectively. The fact that no diffraction peaks were detected for PtO 2 by XRD at any metal content suggests that these PtO 2 species are either amorphous or they exhibit very small crystalline domains (<2 nm).  Additonal experiments, showed this contrbution to be an artifact, as Rh was reduced in situ un der the UHV of the spectrometer chamber (see text below).
For Rh/CeO 2 catalysts, Rh3d 5/2 contributions with BEs of 308.6 eV and 307.7 eV, respectively, were determined at all studied metal loadings (0. .0 Rh at nm -2 ). The former species can be assigned to Rh(III), whereas the latter corresponds to metallic Rh(0). However, additional experiments showed the contribution of the latter species to vary with the experimental settings. Hence, the development of metallic Rh was inferred to occur by metal reduction under X-ray irradiation in the UHV conditions of the spectrometer chamber. Independent XANES experiments (see Figure S10) confirmed cationic Rh as the solely metal species in all Rh/CeO 2 catalysts.  The XANES spectra for the entire series of Rh/CeO 2 catalysts was consistent with the presence of exclusively oxidic Rh species. No signs for metallic Rh(0) could be detected. These results verified that the contribution from metallic Rh(0) detected in the XPS spectra for the same samples was an artifact due to the in situ reduction of rhodium species under the UHV conditions of the XPS chamber. S21 Figure S11: k 3 -weighted phase-uncorrected χ(k) EXAFS function in k-space for) Pt/CeO 2 catalysts with different surface metal contents (M at nm -2 ). The corresponding spectra for bulk metal oxide and metallic foil have also been included for reference in the bottom panel. The scale-marker along the y-axis is identical for both panels. S22 Figure S12: k 3 -weighted phase-uncorrected χ(k) EXAFS function in k-space for) Rh/CeO 2 catalysts with different surface metal contents (M at nm -2 ). The corresponding spectra for bulk metal oxide and metallic foil have also been included for reference. The scale-marker along the y-axis is identical for both panels. ). The corresponding spectra for bulk metal oxide and metallic foil have also been included for reference. The scale-marker along the y-axis is identical for both panels. Figure S15: a-c) Representative C s -HAADF-STEM micrographs for 5.0Pt/CeO 2 . d) C s -HAADF-STEM micrograph and e,f) corresponding EDX compositional maps for Rh and Ce, respectively for 10Rh/CeO 2. In both cases, metal clustering and agglomeration is evident, in line with EXAFS results (see main text). In the case of Pt/CeO 2 , the higher Z-contrast of Pt (oxide) species enables the direct visualization of the metal clusters on the CeO 2 surface. In the case of Rh/CeO 2 , the limited Z-contrast of Rh 2 O 3 requires EDX analysis to identify metal oxide nanoparticles as nanoscale regions of high local Rh concentration (see panel e).

Octene isomers
Organosilane product Figure S17: Fourier-Transform Infrared (FTIR) spectra (collected at T=110 K) after adsorption of CO at increasing dosages on the surface of 0.2Rh/CeO 2 , 1.0Rh/CeO 2 and 5.0Rh/CeO 2 catalysts. The legend color code for CO partial pressures applies to all panels.
FTIR spectroscopy coupled to CO as surface probe was applied to study the nature of surface Rh species on selected Rh/CeO 2 catalysts. For the catalyst with the lowest metal content of 0.2 Rh at nm -2, only a band peaking at 2160 cm -1 emerged upon increasing the CO partial pressure in the cell above 0.11 mbar. This band is ascribed to CO linearly bond to cus Ce 4+ Lewis centers on the CeO 2 surface. [15] No bands in the spectral region 2000-2100 cm -1 could be observed, where Rh x+ -carbonyl vibrations are expected. This result indicates the absence of accessible Rh atoms on the catalyst surface. On the contrary, very prominent bands peaking at 2008 and 2075 cm -1 , respectively, could be observed already from P CO <0.05 mbar in the case of 1.0Rh/CeO 2 . These bands are known to correspond to the asymmetric and S28 symmetric stretching modes, respectively, of Rh + (CO) 2 gem-dicarbonyl species and are a fingerprint for atomically dispersed Rh complexes on oxide carriers. [16] The detection of these signals from very low CO partial pressures and their prominent intensity suggest that the single Rh atoms are abundantly and readily accessible on the catalyst surface. At higher CO dosages, the band at 2160 cm -1 , corresponding to Ce 4+ centers on the CeO 2 support emerges. For a catalyst with a higher metal content (5.0Rh/CeO 2 ), the doublet of bands assigned to Rh + (CO) 2 gem-dicarbonyl species become noticeable only after the CO dosage had reached 8-fod that required in the case of 1.0Rh/CeO 2 . Moreover, these bands clearly showed a lower relative intensity with respect to the Ce 4+ -CO band at the same CO dosage level (0.26±0.02 mbar). Although a direct quantification of individual species is not possible, a qualitative comparison of the bands arising from surface Rh atoms and the ceria support suggests a lower density of surface-exposed Rh x+ centers in spite of the higher overall Rh content. It is inferred from these results that the metal oxide (RhO x ) agglomerates which develop on the catalyst surface at surface metal contents in excess to 1-2 Rh at /nm 2 (observed by XRD ( Figure S5) and EXAFS (Figure 2b)) might also deplete the catalyst surface from atomically dispersed species, possibly via an Ostwald ripening mechanism during the high-temperature catalyst synthesis, compared to catalysts with lower metal contents and no metal agglomerates, e.g. 1.0Rh/CeO 2 . Figure S18: a) Evolution of the olefin conversion (circles) and the selectivity to olefin isomers (diamonds) in the hydrosilylation of 1-octene with Et 3 SiH employing as catalyst 1.0Rh/CeO 2 as-prepared (full black symbols) and after a reduction treatment at 623 K in 20%H 2 /N 2 (open red symbols). Reaction conditions: 1-octene (5 mmol ), triethylsilane (5 mmol), catalyst (2 μmol, metal basis), P=10 bar ( N 2 , 99.999% purity). b) Evolution of the normalized XANES spectra for the 1.0Rh/CeO 2 catalyst with the temperature during the in situ XAS-monitored reduction under flow of 20%H 2 /N 2 . Reference spectra for bulk-type Rh 2 O 3 and metallic Rh(0) foil are also included in the plot (dashed lines). Linear deconvolution of the XANES spectrum after reduction at 623 K, as a linear combination of the spectra for the as-prepared (unreduced catalyst) and the metallic Rh foil, respectively, determined the Rh(0) content in the as-reduced catalyst to be 61%, confirming the partial reduction of rhodium species. -weighted χ(k) EXAFS function (radial distances not phase-corrected) for the 1.0Rh/CeO 2 single-atom catalyst in slurry phase after exposure to the reaction temperature in n-dodecane solvent (blank test, full black symbols), in 3 mmol Et 3 SiH + 3 mmol n-dodecane (full red symbols) and in 3 mmol Et 3 SiH + 3 mmol 1-dodecene, hence fully mimicking reaction conditions (open green symbols) for 1 hour. b) Gas chromatrogram showing confirming the partial conversion of 1-dodecene to the corresponding 1,1,1-trimeyl-1dodecylsilane and hence the completion of the catalysis induction period under the reaction conditions applied.
After exposure of 1.0Rh/CeO 2 to olefin hydrosilylation conditions for a reaction time of 1 hour, sufficient to complete the catalysis induction period, the spectrum showed a decrease in the amplitude for the |FT| of the EXAFS function at the radial distance of ~1.59 Å, corresponding to the first coordination shell around the Rh atoms (Fig. S19a). This is compatible with the cleavage of Rh-O bonds during the induction period and the substitution of oxygen by lighter, e.g. carbon, atoms in the Rh most direct coordination sphere during the development of the hydrosilylation-active metal sites. The spectrum for the catalyst after the catalysis induction period does not reveal any scattering contribution at radial distances >2.4 Å which could be ascribed to either first-shell Rh-Rh or second shell Rh-O-Rh scattering contributions in dimeric (or larger agglomerate) rhodium (oxide) species. Figure S20: Relationship between the catalysis onset time (duration of the induction period) in the hydrosilylation of 1-octene with Et 3 SiH and the average Rh-O coordination number (as derived by fitting of the EXAFS spectra for the as-prepared catalysts) for the series of Rh/CeO 2 catalysts. Reaction conditions: 1-octene (5 mmol), triethylsilane (5 mmol), catalyst (2 μmol, Rh metal basis), P=10 bar (N 2 , 99.999% purity), T=393 K. The dotted line is added as a guide to the eye. Figure S21: a) Scheme showing the major reaction pathways for allyl alcohol under hydrosilylation conditions. b) (i) 1 H NMR spectrum for a) pure allyl alcohol (prop-2-en-1-ol) reactant; (ii) the crude product after the isomerization reaction in the absence of Et 3 SiH using 1.0Rh/CeO 2 as catalyst; and (iii) the crude product after the hydrosilylation reaction in the presence of Et 3 SiH using 1.0Rh/CeO 2 as catalyst. Reaction conditions: allyl alcohol (10 mmol ), triethylsilane (10 mmol), catalyst (4 μmol, metal basis), P=10 bar ( N 2 , 99.999% purity), reaction time 18 h.
In order to assess whether olefin hydrosilylation or isomerization proceeded faster under reaction conditions on different SACs, 2-propen-1-ol (allyl alcohol) was applied as olefin substrate. In this case, the tautomerization equilibrium established between the doublebond isomerization product, 1-propen-1-ol, and the thermodynamically most stable propanal, provides an energy "sink" which inhibits double-bond back-migration and it thus enables a direct assessment of the relative forward reaction rates for olefin isomerization and hydrosilylation, respectively ( Figure S21a). As shown in Table S5, in the absence of Et 3 SiH, 1.0Rh/CeO 2 led to a 7% olefin conversion after 18 h, exclusively via isomerization. In presence of the silane reagent, however, an essentially full conversion of the olefin was attained, with a selectivity >90% to the terminal 1-silyl-propan-2-ol product. These results prove that olefin hydrosilylation proceeds notably (>30 times) faster than isomerization on isolated Rh sites. Platinum sites on 1.0Pt/CeO 2 led to a lower overall olefin conversion (82%) and similar selectivities to both isomerization and hydrosilylation products, evidencing comparable forward reaction rates along both pathways.1.0Ru/CeO 2 , albeit essentially inactive in the absence of the Et 3 SiH reactant, showed to be very selective for olefin isomerization.  Figure S22: a) Time-resolved evolution of the olefin conversion under reaction conditions for the hydrosilylation of 1-octene with Et 3 SiH employing Ru/CeO 2 catalysts synthesized with different surface metal content. For clarity, only data points for selected catalysts of those tested are displayed. Olefin conversion was essentially to isomers. Reaction conditions: 1-octene (5 mmol), triethylsilane (5 mmol), catalyst (2 μmol, Ru metal basis), P=10 bar (N 2 , 99.999% purity), T=393 K. b) |FT| of the k 3 -weighted χ(k) EXAFS function (radial distances not phase-corrected) for the 1.0Ru/CeO 2 single-atom catalyst after exposure to the reaction temperature in n-dodecane solvent (blank test, full symbols) or in an excess of Et 3 SiH in n-dodecane (open symbols).
No induction period was observed to precede 1-octene conversion (to isomers) with Ru/CeO 2 catalysts (Fig. 22a). This indicates that the development of the isomerization-active Ru species, which takes place only in the presence of Et 3 SiH, is kinetically facile. This is in line with the observation of a distortion (decrease in amplitude) for the |FT| of the EXAFS function at the radial distance of ~2.03 Å, corresponding to the first coordination shell around the Ru atoms (Fig. S22b), short after the catalyst is exposed to excess Et 3 SiH and brought to reaction temperature (393 K), which is compatible with a fast cleavage of Ru-O bonds and the formation of Ru hydride species in situ, which are known to be active sites for olefin isomerization. [17] Figure S23: Schematic illustration of the stepped CeO 2 surface considered in the DFT calculations. A 211surface, that connects two (111) facets has been used to simulate a type II CeO 2 step-edge where single-atom metal centers are stabilized. Atom color code: gray (Rh), pink (H), red (O), green (Ce).
As illustrated in Fig. S28, the tandem cooperation of Rh 1 /CeO 2 and Ru 1 /CeO 2 single-atom catalysts achieves a remarkable performance on an internal olefin such as 3-octene, for which either of the two catalysts individually is barely active. The optimal reactivity is attained for a blend of catalysts with a Ru/Rh ratio of 4 (i.e. 80% Ru in an overall metal basis).. Figure S29: Catalyst stability and recyclability. a) Olefin conversion and selectivity to 1,1,1-triethyl-1-octylsilane in five consecutive catalytic runs in a tandem 2-octene isomerization-hydrosilylation process via the single-pot integration of 1.0Rh/CeO 2 and 1.0Ru/CeO 2 single-atom catalysts. Reaction conditions: 2-octene (5 mmol), triethylsilane (5 mmol), catalyst (2 μmol Rh, 2 μmol Ru), P=10 bar (N 2 , 99.999% purity). |FT| of the k Figure S30: C s -HAADF-STEM and corresponding EDX compositional nanoscale maps for (a-f) 1.0Rh/CeO 2 and (gl) 1.0Ru/CeO 2 catalysts recovered after the tandem isomerization-hydrosilylation of 1-octene. Tables   Table S1: Summary of optimized parameters by fitting EXAFS data recorded for as-synthesized Pt/CeO 2  catalysts with different metal contents at room temperature. EXAFS spectra fits were performed on the first coordination shell over the FT of the k 3 -weighted χ(k) function in the Δk= 3-12 Å -1 interval. The amplitude reduction factor S 0 2 was fixed to 0.82 relative to the pure metal. a Average coordination number. b Mean square variation in path length. c Energy shift. The r-factor represents the goodness of fit.  The evolution of the first-shell M-O average coordination number (CN), derived from the fitting of the EXAFS spectra, was assessed as a function of the surface metal content (Tables S1-S3). The results reveal first a decrease in CN as δ increases up to 1.0 M at nm -2 , followed by an augment in CN for increasingly higher metal contents beyond this value. This result suggests that, at any metal content, a fraction of the metal atoms incorporates into sub-surface positions within the CeO 2 structure, being coordinatively saturated and thus contributing to a higher average CN at very low metal contents, i.e. 0.2 M at nm -2 , at which their abundance is highest. On increasing the overall metal content, the solid-state solubility of metal cations in the CeO 2 lattice is likely exceeded, and metal species are consequently "spelled" onto the CeO 2 surface, according to our EXAFS results, as isolated (coordinatively unsaturated) metal cations up to a surface coverage of 1.0-2.0 M at nm -2 . Further rise in metal content leads to partial aggregation of surface metal species into oxide nanoparticles, as proved by EXAFS and XRD, and thus the average 1 st -shell CN is set to increase again. This trend is observed regardless of the nature of the supported metal, and it thus reflects a generalized behavior. Excellent yields and selectivities to terminal alkyl silanes were obtained with linear 1-olefin reactants of different hydrocarbon chain lengths. Styrene, which is a particularly challenging substrate given that the bulkiness of the phenyl substituent is known to favor β-hydride elimination and thus dehydrogenative hydrosilylation pathways, [19] was converted under standard reaction conditions with remarkably good yields to the anti-Markovnikov terminal hydrosilylation product. A remarkable 87% selectivity was achieved at 80% conversion with 1.0Rh/CeO 2 . In sharp contrast a 25% olefin conversion with a marginal 22% selectivity was obtained applying the 1.0Pt/CeO 2 analog under identical reaction settings, owing to the massive production of unsaturated silanes (dehydrogenative hydrosilylation) and ethylbenzene (hydrogenation) sideproducts. Moreover, 1.0Rh/CeO 2 showed a remarkable tolerance to various functional groups