P-stereogenic Phosphines for the Stabilisation of Metal Nanoparticles. a Surface State Study

Palladium and ruthenium nanoparticles have been prepared following the organometallic precursor decomposition methodology, under dihydrogen pressure and in the presence of borane protected P-stereogenic phosphines. NMR (Nuclear Magnetic Resonance) monitoring of the corresponding syntheses has permitted to determine the optimal metal/ligand ratio for leading to small and well-dispersed nanoparticles. Exchange ligand reactions of the as-prepared materials have proven the strong interaction of the phosphines with the metal surface; only oxidative treatment using hydrogen peroxide could release the phosphine-based stabiliser from the metal surface. Pd and Ru nanoparticles have been evaluated in hydrogenation reactions, confirming the robustness of the stabilisers, which selectively permitted the hydrogenation of exocyclic C=C bonds, preventing the coordination of the aromatic rings and as a result, their hydrogenation.

Herein we describe the synthesis of ruthenium and palladium nanoparticles stabilised by P-stereogenic phosphines and their coordination to the metallic surface.A study of hydrogenation of C=C bonds was carried out with the purpose to understand the interaction of these stabilisers with the metal surface.
Herein we describe the synthesis of ruthenium and palladium nanoparticles stabilised by Pstereogenic phosphines and their coordination to the metallic surface.A study of hydrogenation of C=C bonds was carried out with the purpose to understand the interaction of these stabilisers with the metal surface.
With the aim of determining the optimal metal/ligand ratio to give well-dispersed nanoparticles, a multi-nuclear NMR monitoring study of the synthesis of these MNPs was carried out.For this purpose, a mixture of the corresponding organometallic complex and the corresponding ligand in 1 mL of THF-d8 was prepared and pressurised with dihydrogen (3 bar; NMR tube with PTFE (polytetrafluoroethylene) Young valve), in the presence of dodecane as internal standard.With the purpose to quantify the amount of capped ligand at the MNP surface, an excess of ligand was used (a metal/L ratio of 1/1).The intensity decrease of the t-Bu doublet signal (1.2 ppm; 3 JH-P = 12 Hz) in the corresponding 1 H NMR spectra, was used as probe to determine the percentage of ligand adsorbed at the metallic surface (see Figures S2-S5 in the Supplementary Materials).The mixtures were monitored until unchanged 1 H NMR spectra were recorded.For PdNPs, 0.4 equivalent of free ligand remained in the solution for both PdL1 and PdL2, meaning that the Pd/L ratio in the formed PdNPs was 1/0.6 (Figure 2).TEM (Transmission Electron Microscopy) analyses after the NMR monitoring The as-prepared materials were analysed by TEM, evidencing the formation of small and well-dispersed ruthenium nanoparticles, RuL1 and RuL2 (mean size, ca.1.2 nm; Figure 1).In contrast, PdNPs led to more agglomerated systems (Figure 1).Infrared spectra of these materials revealed the absence of borane, meaning that BH 3 , protecting group of the phosphines, was removed during the synthesis of the corresponding MNPs (see Figure S1 in the Supplementary Materials).
With the aim of determining the optimal metal/ligand ratio to give well-dispersed nanoparticles, a multi-nuclear NMR monitoring study of the synthesis of these MNPs was carried out.For this purpose, a mixture of the corresponding organometallic complex and the corresponding ligand in 1 mL of THF-d 8 was prepared and pressurised with dihydrogen (3 bar; NMR tube with PTFE (polytetrafluoroethylene) Young valve), in the presence of dodecane as internal standard.With the purpose to quantify the amount of capped ligand at the MNP surface, an excess of ligand was used (a metal/L ratio of 1/1).The intensity decrease of the t-Bu doublet signal (1.2 ppm; 3 J H-P = 12 Hz) in the corresponding 1 H NMR spectra, was used as probe to determine the percentage of ligand adsorbed at the metallic surface (see Figures S2-S5 in the Supplementary Materials).The mixtures were monitored until unchanged 1 H NMR spectra were recorded.For PdNPs, 0.4 equivalent of free ligand remained in the solution for both PdL1 and PdL2, meaning that the Pd/L ratio in the formed PdNPs was 1/0.6 (Figure 2).TEM (Transmission Electron Microscopy) analyses after the NMR monitoring showed the exclusive formation of nanoparticles of size 1.4 nm for both ligands (Figure 2), in contrast with the agglomerates obtained using a Pd/L ratio of 1/0.2 (Figure 1).showed the exclusive formation of nanoparticles of size 1.4 nm for both ligands (Figure 2), in contrast with the agglomerates obtained using a Pd/L ratio of 1/0.2 (Figure 1).In the case of ruthenium, only 0.4 equivalent of ligand was necessary for the formation of RuNPs, RuL1 and RuL2 (Figure 3).Contrary to the behaviour of the formation of PdNPs, the decomposition showed the exclusive formation of nanoparticles of size 1.4 nm for both ligands (Figure 2), in contrast with the agglomerates obtained using a Pd/L ratio of 1/0.2 (Figure 1).In the case of ruthenium, only 0.4 equivalent of ligand was necessary for the formation of RuNPs, RuL1 and RuL2 (Figure 3).Contrary to the behaviour of the formation of PdNPs, the decomposition In the case of ruthenium, only 0.4 equivalent of ligand was necessary for the formation of RuNPs, RuL1 and RuL2 (Figure 3).Contrary to the behaviour of the formation of PdNPs, the decomposition of [Ru(COD)(COT)] was slower (the NMR tube was regularly shaken in order to observe evolution of the decomposition of the organometallic precursor).TEM analyses of the particles after the NMR monitoring exhibited nanoparticles of size 1.2 nm for both ligands (Figure 3), analogously to those obtained using a Ru/L ratio of 1/0.2 (Figure 1). of [Ru(COD)(COT)] was slower (the NMR tube was regularly shaken in order to observe evolution of the decomposition of the organometallic precursor).TEM analyses of the particles after the NMR monitoring exhibited nanoparticles of size 1.2 nm for both ligands (Figure 3), analogously to those obtained using a Ru/L ratio of 1/0.2 (Figure 1).
The decomposition of the metallic precursors, [Pd2(dba)3] and [Ru(COD)(COT)], was also checked by the concomitant formation of dbaH4 (dbaH4 = 1,5-diphenylpenta-3-one) and cyclooctane, respectively.The formation of PdNPs was also monitored by the intensity decrease of the corresponding 31 P and 11 B NMR signals, in addition to the disappearance of the 1 H NMR ligand signals (see Figure S6 in the Supplementary Materials); other signals corresponding to molecular palladium species could be also observed, together with the formation of free BH4 − , which indicates that the ligand adsorbed on the surface corresponds to the BH3-free phosphine.In the case of the formation of RuNPs, the differences in intensity of 31 P and 11 B NMR signals were less perceptible (see Figure S7 in the Supplementary Materials).
With this information, PdNPs and RuNPs were synthesised at a larger scale, using now a metal/ligand ratio of 1/0.6 and 1/0.4,respectively, preserving the conditions previously described (see above, Scheme 1).Under these conditions, PdNPs with a mean diameter of ca.1.5 nm for both ligands were obtained (Figure 4).Electronic diffraction analyses of these PdNPs showed a fcc structure as expected for bulk Pd(0) and the corresponding EDX (Energy dispersive X-ray spectroscopy) data evidenced the presence of ligand in the as-prepared nanoparticles (see Figure S8 in the Supplementary Materials).Elemental analyses indicated that the Pd/L ratio found on the isolated PdNPs was: Pd/L1 = 1/0.27and Pd/L2 = 1/0.50(corresponding to a ratio Pd/L at the surface PdL10.4 and PdL20.8).Concerning RuNPs, small nanoparticles were also obtained, with a mean diameter of 1.0 and 1.2 nm for RuL1 and RuL2, respectively.Their electronic diffraction analyses exhibited a hcp arrangement characteristic of bulk Ru(0); the corresponding EDX data confirmed the presence of the ligand in the as-prepared materials.Elemental analyses indicated that the Ru/L ratio found on the isolated RuNPs was: Ru/L1 = 1/0.56 and Ru/L2 = 1/0.46(corresponding to a ratio Pd/L at the surface PdL10.4 and PdL20.8).
The synthesis of MNPs was also run with borane-free L2 with both metal precursors, [Pd2(dba)3] and [Ru(COD)(COT)] (under the same conditions than those described above using a metal/ligand The formation of PdNPs was also monitored by the intensity decrease of the corresponding 31 P and 11 B NMR signals, in addition to the disappearance of the 1 H NMR ligand signals (see Figure S6 in the Supplementary Materials); other signals corresponding to molecular palladium species could be also observed, together with the formation of free BH 4 − , which indicates that the ligand adsorbed on the surface corresponds to the BH 3 -free phosphine.In the case of the formation of RuNPs, the differences in intensity of 31 P and 11 B NMR signals were less perceptible (see Figure S7 in the Supplementary Materials).With this information, PdNPs and RuNPs were synthesised at a larger scale, using now a metal/ligand ratio of 1/0.6 and 1/0.4,respectively, preserving the conditions previously described (see above, Scheme 1).Under these conditions, PdNPs with a mean diameter of ca.1.5 nm for both ligands were obtained (Figure 4).Electronic diffraction analyses of these PdNPs showed a fcc structure as expected for bulk Pd(0) and the corresponding EDX (Energy dispersive X-ray spectroscopy) data evidenced the presence of ligand in the as-prepared nanoparticles (see Figure S8 in the Supplementary Materials).Elemental analyses indicated that the Pd/L ratio found on the isolated PdNPs was: Pd/L1 = 1/0.27and Pd/L2 = 1/0.50(corresponding to a ratio Pd/L at the surface PdL1 0.4 and PdL2 0.8 ).Concerning RuNPs, small nanoparticles were also obtained, with a mean diameter of 1.0 and 1.2 nm for RuL1 and RuL2, respectively.Their electronic diffraction analyses exhibited a hcp arrangement characteristic of bulk Ru(0); the corresponding EDX data confirmed the presence of the ligand in the as-prepared materials.Elemental analyses indicated that the Ru/L ratio found on the isolated RuNPs was: Ru/L1 = 1/0.56 and Ru/L2 = 1/0.46(corresponding to a ratio Pd/L at the surface PdL1 0.4 and PdL2 0.8 ).ration of 1/0.6 and 1/0.4 for Pd and Ru, respectively).Small MNPs of the approximatively 1.0 nm were obtained for both metals (see Figure S9 in the Supplementary Materials).In a previous work, we could prove that ligands can be modified during the synthesis of metal nanoparticles [40].In order to recover the stabiliser after synthesis of the corresponding MNPs, ligand exchange reactions were carried out [41].Therefore, taking advantage of the affinity of noble-based metal surfaces by thiols and carbon monoxide, dodecanethiol and CO (up to 3 bar of pressure) were added to the solution of MNPs dispersed in THF-d8, with the aim of releasing the corresponding phosphine from the metallic surface.In any case, ligand exchange occurred (monitored by 1 H and 31 P NMR during one week at room temperature).Only oxidative conditions allowed the observation of free stabilisers.Therefore, treatment of PdNPs with hydrogen peroxide showed the presence of the corresponding oxides of L1 and L2, by both 31 P NMR and mass spectrometry (see Figures S10 and The synthesis of MNPs was also run with borane-free L2 with both metal precursors, [Pd 2 (dba) 3 ] and [Ru(COD)(COT)] (under the same conditions than those described above using a metal/ligand ration of 1/0.6 and 1/0.4 for Pd and Ru, respectively).Small MNPs of the approximatively 1.0 nm were obtained for both metals (see Figure S9 in the Supplementary Materials).
In a previous work, we could prove that ligands can be modified during the synthesis of metal nanoparticles [40].In order to recover the stabiliser after synthesis of the corresponding MNPs, ligand exchange reactions were carried out [41].Therefore, taking advantage of the affinity of noble-based metal surfaces by thiols and carbon monoxide, dodecanethiol and CO (up to 3 bar of pressure) were added to the solution of MNPs dispersed in THF-d 8 , with the aim of releasing the corresponding phosphine from the metallic surface.In any case, ligand exchange occurred (monitored by 1 H and 31 P NMR during one week at room temperature).Only oxidative conditions allowed the observation of free stabilisers.Therefore, treatment of PdNPs with hydrogen peroxide showed the presence of the corresponding oxides of L1 and L2, by both 31 P NMR and mass spectrometry (see Figures S10 and S11 in the Supplementary Materials).This behaviour evidenced that both ligands interact strongly with the palladium surface.The same oxidative treatment was applied to preform RuNPs, but unfortunately the detection of the corresponding oxides was not possible.

Hydrogenation Reactivity
As benchmark reactions, we chose to study the catalytic behaviour of the prepared PdNPs and RuNPs in the hydrogenation of both styrene (3) and 4-vinylpyridine (4), because of the different functions to be hydrogenated (arene and vinyl moieties) and the different ability to coordinate to the metal surface (phenyl versus pyridyl group) (Scheme 2 and Table S1 Supplementary Materials) [42].For both substrates using any of the Ru-or Pd-based catalysts, we observed the exclusive formation of ethylbenzene ( 5) and 4-ethylpyridine (6), under relative smooth conditions, 3 bar H 2 pressure at 50 • C (Scheme 2).PdNPs were more active than RuNPs, achieving full conversion after 1 h and 16 h, respectively.With the aim of triggering the reduction of the aromatic group, hydrogenations at higher pressures (up to 40 bar) using RuNPs as catalysts were carried out.However no arene-hydrogenation was favoured; only under 40 bar H 2 , we could observe the presence of traces of ethylcyclohexane and 4-ethylpiperidine (ca.5%), in addition to the corresponding ethyl derivatives 5 and 6.A similar catalytic trend has been reported for achiral phosphines [43].Unfortunately, RuNPs were not active in the hydrogenation of the carbonyl group of acetophenone (under 40 bar H 2 at 50 S11 in the Supplementary Materials).This behaviour evidenced that both ligands interact strongly with the palladium surface.The same oxidative treatment was applied to preform RuNPs, but unfortunately the detection of the corresponding oxides was not possible.

Hydrogenation Reactivity
As benchmark reactions, we chose to study the catalytic behaviour of the prepared PdNPs and RuNPs in the hydrogenation of both styrene (3) and 4-vinylpyridine (4), because of the different functions to be hydrogenated (arene and vinyl moieties) and the different ability to coordinate to the metal surface (phenyl versus pyridyl group) (Scheme 2 and Table S1 Supplementary Materials) [42].For both substrates using any of the Ru-or Pd-based catalysts, we observed the exclusive formation of ethylbenzene ( 5) and 4-ethylpyridine (6), under relative smooth conditions, 3 bar H2 pressure at 50 °C (Scheme 2).PdNPs were more active than RuNPs, achieving full conversion after 1 h and 16 h, respectively.With the aim of triggering the reduction of the aromatic group, hydrogenations at higher pressures (up to 40 bar) using RuNPs as catalysts were carried out.However no arenehydrogenation was favoured; only under 40 bar H2, we could observe the presence of traces of ethylcyclohexane and 4-ethylpiperidine (ca.5%), in addition to the corresponding ethyl derivatives 5 and 6.A similar catalytic trend has been reported for achiral phosphines [43].Unfortunately, RuNPs were not active in the hydrogenation of the carbonyl group of acetophenone (under 40 bar H2 at 50 °C overnight).These results certainly evidence that both phosphines remain robustly linked to the surface (as proven by the ligand exchange reactions performed; see above), hindering the approach of the aromatic group of the substrates 3 or 4 at the metal surface; only the vinyl group is consequently able to be reduced.Furthermore, this assumption is supported by the fact that [Ru(COD(COT)] used as catalytic precursor in the absence of any ligand, gave, under 20 bar H2 for 16h of reaction, a mixture of ethylbenzene and ethylcyclohexane (64% and 36%, respectively).The behaviour observed using RuL1 and RuL2 contrasts with that described for RuNPs stabilised by phosphines, for which the reduction of the aromatic fragments occurs [28,[44][45][46][47][48].In our case, L1 and L2 can be coordinated to the metal surface in a bidentate mode by the dative M-P bond and a secondary interaction: π interaction by the naphthyl group (L1) or interaction by the alcohol function or phenyl group (L2), decreasing the vacant coordination sites on nanoparticles.This coordination mode means that the compact faces on nanoparticles are occupied, avoiding the coordination of arene substrates [49].
Therefore, the lack of hydride transfer from the metal surface to the aromatic cycles is most likely due to a high coverage of the surface by the phosphines (L1, L2).Thus, these saturated surfaces favour the chemisorption of C=C vinyl bond (adsorption enthalpy, ca.−20 kJ/mol) in relation to the chemisorption of an aromatic cycle (adsorption enthalpy, ca.−70 kJ/mol) [50].This presumption is in agreement with the non-coordinated mechanism proposed by Sautet and co-workers for the hydrogenation of exocyclic C=C bonds [49].
We also analysed by ICP-AES the product 6 obtained by hydrogenation of 4-vinylpyridine (4) using both PdL2 and RuL2; the metal content was lower than 1 ppm (for Ru: 0.7 ppm; for Pd: 0.08 ppm), what indicates an irrelevant metal leaching.With the aim of elucidating the catalyst nature, we added mercury to the hydrogenation of 4 using both PdL2 and RuL2 after 5 min of reaction at room temperature under 1 bar H2 pressure (conversion: 16% for PdL2; 5% for RuL2); the reaction was then These results certainly evidence that both phosphines remain robustly linked to the surface (as proven by the ligand exchange reactions performed; see above), hindering the approach of the aromatic group of the substrates 3 or 4 at the metal surface; only the vinyl group is consequently able to be reduced.Furthermore, this assumption is supported by the fact that [Ru(COD(COT)] used as catalytic precursor in the absence of any ligand, gave, under 20 bar H 2 for 16h of reaction, a mixture of ethylbenzene and ethylcyclohexane (64% and 36%, respectively).The behaviour observed using RuL1 and RuL2 contrasts with that described for RuNPs stabilised by phosphines, for which the reduction of the aromatic fragments occurs [28,[44][45][46][47][48].In our case, L1 and L2 can be coordinated to the metal surface in a bidentate mode by the dative M-P bond and a secondary interaction: π interaction by the naphthyl group (L1) or interaction by the alcohol function or phenyl group (L2), decreasing the vacant coordination sites on nanoparticles.This coordination mode means that the compact faces on nanoparticles are occupied, avoiding the coordination of arene substrates [49].
Therefore, the lack of hydride transfer from the metal surface to the aromatic cycles is most likely due to a high coverage of the surface by the phosphines (L1, L2).Thus, these saturated surfaces favour the chemisorption of C=C vinyl bond (adsorption enthalpy, ca.−20 kJ/mol) in relation to the chemisorption of an aromatic cycle (adsorption enthalpy, ca.−70 kJ/mol) [50].This presumption is in agreement with the non-coordinated mechanism proposed by Sautet and co-workers for the hydrogenation of exocyclic C=C bonds [49].
We also analysed by ICP-AES the product 6 obtained by hydrogenation of 4-vinylpyridine (4) using both PdL2 and RuL2; the metal content was lower than 1 ppm (for Ru: 0.7 ppm; for Pd: 0.08 ppm), what indicates an irrelevant metal leaching.With the aim of elucidating the catalyst nature, we added mercury to the hydrogenation of 4 using both PdL2 and RuL2 after 5 min of reaction at room temperature under 1 bar H 2 pressure (conversion: 16% for PdL2; 5% for RuL2); the reaction was then heated at 50 • C under 3 bar overnight, obtaining full conversion of 4-vinylpyridine in both cases.These results point to a homogeneous catalytic behaviour, in contrast with the ICP analyses of the hydrogenated products.This behaviour seems to evidence that preformed MNPs act as a reservoir of homogeneous active species, being again aggregated at the end of the reaction in agreement with the insignificant content of metal on the isolated products.
Taking into account that L1 and L2 are optically pure, we decided to test the hydrogenation of prochiral substrates, in particular styrene derivatives (7, 8) using Pd-based catalysts (Scheme 3).For substrate 7, the same behaviour than that observed for styrene was observed: full conversion towards 9 after 1 h of reaction.However the hydrogenation of 8 was less efficient (only 25% conversion after 5 h of reaction), probably due to its higher steric-demanding effect, and led to a mixture of 10 and the hydroxylamine 11 (the oxime tautomer was not observed by 1 H NMR). Unfortunately no asymmetric induction was observed in any case; this absence of enantioselectivity cannot be attributed to the racemization of the chiral ligands, as proven by the control experiment using L1 as substrate under catalytic conditions [see experimental section].
Catalysts 2016, 6, 213 7 of 11 heated at 50 °C under 3 bar overnight, obtaining full conversion of 4-vinylpyridine in both cases.These results point to a homogeneous catalytic behaviour, in contrast with the ICP analyses of the hydrogenated products.This behaviour seems to evidence that preformed MNPs act as a reservoir of homogeneous active species, being again aggregated at the end of the reaction in agreement with the insignificant content of metal on the isolated products.
Taking into account that L1 and L2 are optically pure, we decided to test the hydrogenation of prochiral substrates, in particular styrene derivatives (7, 8) using Pd-based catalysts (Scheme 3).For substrate 7, the same behaviour than that observed for styrene was observed: full conversion towards 9 after 1 h of reaction.However the hydrogenation of 8 was less efficient (only 25% conversion after 5 h of reaction), probably due to its higher steric-demanding effect, and led to a mixture of 10 and the hydroxylamine 11 (the oxime tautomer was not observed by 1 H NMR). Unfortunately no asymmetric induction was observed in any case; this absence of enantioselectivity cannot be attributed to the racemization of the chiral ligands, as proven by the control experiment using L1 as substrate under catalytic conditions [see experimental section].

General
Unless otherwise stated, all chemical reagents were obtained from commercial suppliers and used without further purification.Optically pure borane-protected phosphines, (S)-tertbutylmethyl(2-(2-naphthyl)ethyl)phosphine-borane (L1-BH3) and (R)-(2,2-diphenyl-2hydroxyethyl)-P-methyl-t-butyl-phosphine-borane (L2-BH3) were synthesised as previously described [35,36].All manipulations were performed using standard Schlenk techniques under argon atmosphere.NMR spectra were recorded on a Bruker Avance 400 spectrometer (Bruker, Billerica, MA, USA) at 293 K (400 MHz for 1 H NMR; 162 MHz for 31 P NMR; 128 MHz for 11 B NMR). GC analyses were carried out on a GC Perkin Elmer Clarus 500 with ionization flame detector (Waters, Waltham, MA, USA), using a SGE BPX5 column composed by 5% phenylmethylsiloxane and a Perkin Elmer Clarus MS560 mass detector.The determination of enantiomeric excesses was carried out on a UPLC Acquity Waters at 40 °C with a UV PDA detector (Waters, Milford, Massachusetts, United States) and a cyclobond I2000HP RSP15cm/4.6 5 um column with a mixture of H2O/CH3CN = 60/40.TEM images of particles in the solid state and dispersed in glycerol were obtained from transmission electron microscopes JEOL JEM (JEOLn Akishima, Tokyo, Japan) 1400 running at 120 kV and HR-TEM from JEOL JEM 2100F running at 200 kV equipped with X PGT (detection of light elements, resolution 135 eV).The nanoparticles size, distribution and average diameter were determined from TEM images with Image-J software (Image processing, Public domain, https://imagej.nih.gov/ij/)associated to a Microsoft Excel macro developed by Christian Pradel.IR spectra were recorded in the range of 4000-400 cm −1 on a Varian 640-IR FTIR Spectrometer (Varian, Salt Lake City, United States).High-pressure reactions were carried out in a Top Industrie Autoclave suitable from 0 to 50 bar and from 15 to 150 °C (Top Industrie, Vaux Le Pénil, France).Elemental and ICP-AES analyses were carried out at the

General
Unless otherwise stated, all chemical reagents were obtained from commercial suppliers and used without further purification.Optically pure borane-protected phosphines, (S)-tert-butylmethyl (2-(2-naphthyl)ethyl)phosphine-borane (L1-BH 3 ) and (R)-(2,2-diphenyl-2-hydroxyethyl)-P-methyl-t -butyl-phosphine-borane (L2-BH 3 ) were synthesised as previously described [35,36].All manipulations were performed using standard Schlenk techniques under argon atmosphere.NMR spectra were recorded on a Bruker Avance 400 spectrometer (Bruker, Billerica, MA, USA) at 293 K (400 MHz for 1 H NMR; 162 MHz for 31 P NMR; 128 MHz for 11 B NMR). GC analyses were carried out on a GC Perkin Elmer Clarus 500 with ionization flame detector (Waters, Waltham, MA, USA), using a SGE BPX5 column composed by 5% phenylmethylsiloxane and a Perkin Elmer Clarus MS560 mass detector.The determination of enantiomeric excesses was carried out on a UPLC Acquity Waters at 40 • C with a UV PDA detector (Waters, Milford, Massachusetts, United States) and a cyclobond I2000HP RSP15cm/4.6 5 um column with a mixture of H 2 O/CH 3 CN = 60/40.TEM images of particles in the solid state and dispersed in glycerol were obtained from transmission electron microscopes JEOL JEM (JEOLn Akishima, Tokyo, Japan) 1400 running at 120 kV and HR-TEM from JEOL JEM 2100F running at 200 kV equipped with X PGT (detection of light elements, resolution 135 eV).The nanoparticles size, distribution and average diameter were determined from TEM images with Image-J software (Image processing, Public domain, https://imagej.nih.gov/ij/)associated to a Microsoft Excel macro developed by Christian Pradel.IR spectra were recorded in the range of 4000-400 cm −1 on a Varian 640-IR FTIR Spectrometer (Varian, Salt Lake City, United States).High-pressure reactions were carried out in a Top Industrie Autoclave suitable from 0 to 50 bar and from 15 to 150 • C (Top Industrie, Vaux Le Pénil, France).Elemental and ICP-AES analyses were carried out at the "Service d'Analyse" of Laboratoire de Chimie de Coordination (Toulouse) using a Perkin Elmer 2400 series II analyser (Perkin Elmer, Waltham, MA, USA) and by Antellis respectively.As a standard procedure, the organometallic complex (37 mg of [Pd 2 (dba) 3 ] (0.04 mmol) or 25 mg [Ru(COD)(COT)] (0.08 mmol)) was dissolved in a solution of 80 mL of THF with 0.6 eq. of L-BH 3 (14 mg of L1-BH 3 or 15 mg for L2-BH 3 ) in a Fischer-Porter bottle.The mixture was then pressurised under dihydrogen (3 bar) at room temperature.The vigorous stirring and dihydrogen pressure were maintained for 18h at room temperature, leading to black and homogeneous colloidal solutions.The solvent was then evaporated.The particles as black solids were washed with pentane (3 × 10 mL) and dried under vacuum.The nanoparticles were characterised by TEM, IR and elemental analysis.

Synthesis of
PdL1

Hydrogenation Reactions Catalysed by PdNPs and RuNPs
2 mg of RuNPs (0.01 mmol Ru) or 3 mg of PdNPs (0.01 mmol Pd) and the corresponding substrate (1 mmol of 3 (105 mg), 4 (104 mg), 7 (236 mg) or 8 (163 mg)) were dissolved in 20 mL heptane at room temperature, in the presence of 142 mg of decane as internal standard.The catalytic mixture was stirred, pressurized under the desired pressure (3-40 bar) and heated at 50 • C overnight in an autoclave or in a Fisher-Porter bottle.The system was then depressurized, the solution was filtered on celite and analyzed by GC.

Hydrogenation of L1
3 mg of PdNPs (0.01 mmol Pd) and L1-BH 3 (15 mg, 0.055 mmol) were dissolved in 10 mL heptane at room temperature.The catalytic mixture was stirred, pressurized under 3 bar and heated at 50 • C overnight in Fisher-Porter bottle.The system was then depressurized, the solution was filtered on celite and analyzed by 1 H, 11

Conclusions
In summary, P-stereogenic phosphines L1 and L2 have proven their efficiency to stabilise small and well-dispersed palladium and ruthenium nanoparticles, depending on the metal/ligand ratio used.The NMR monitoring of the syntheses of MNPs together with the TEM analyses of the resulting colloidal solutions, allowed us knowing the appropriate amount of ligand-to-metal in order to get non agglomerated materials.The coordination strength of the stabilisers at the metal surface is crucial to understand their aptitude as catalysts.With this purpose in mind, we studied exchange ligand reactions between the as-prepared nanoparticles and ligands such as thiols and carbon monoxide, demonstrating that L1 and L2 are robust, and only oxidative treatments (reaction of MNPs with hydrogen peroxide) permitted to de-coordinate the phosphines from the surface.This ligand robustness can be explained by a bidentate coordination mode, through both the phosphorus centre and the aryl or alcohol functions present in the phosphine structure.This behaviour could be also reflected in the hydrogenation of vinyl-aromatic substrates catalysed by the MNPs.In particular, Ru-based catalysts exclusively hydrogenated the corresponding vinyl groups, because the aromatic rings of the substrates cannot come close to the metal surface, due to the coverage of the faces of nanoparticles by the phosphines.Unfortunately, no enantioselectivity could be induced.

Figure 2 .
Figure 2. Graphical representations of the data corresponding to the consumption of ligand and decomposition of Pd precursor from the 1 H NMR monitoring of the formation PdNPs with L1 and L2 (top); TEM images of PdL1 and PdL2 after total consumption of [Pd2(dba)3] (bottom).

Figure 2 .
Figure 2. Graphical representations of the data corresponding to the consumption of ligand and decomposition of Pd precursor from the 1 H NMR monitoring of the formation PdNPs with L1 and L2 (top); TEM images of PdL1 and PdL2 after total consumption of [Pd2(dba)3] (bottom).

Figure 2 .
Figure 2. Graphical representations of the data corresponding to the consumption of ligand and decomposition of Pd precursor from the 1 H NMR monitoring of the formation PdNPs with L1 and L2 (top); TEM images of PdL1 and PdL2 after total consumption of [Pd 2 (dba) 3 ] (bottom).

Figure 3 .
Figure 3. Graphical representations of the data corresponding to the consumption of ligand and [Ru(COD)(COT)], and formation of cyclooctane (top); TEM images of RuL1 and RuL2 after total consumption of [Ru(COD)(COT)] (bottom).

Figure 3 .
Figure 3. Graphical representations of the data corresponding to the consumption of ligand and [Ru(COD)(COT)], and formation of cyclooctane (top); TEM images of RuL1 and RuL2 after total consumption of [Ru(COD)(COT)] (bottom).