Challenging Thermodynamics: Hydrogenation of Benzene to 1,3‐Cyclohexadiene by Ru@Pt Nanoparticles

Since the earliest reports on catalytic benzene hydrogenation, 1,3‐cyclohexadiene and cyclohexene have been proposed as key intermediates. However, the former has never been obtained with remarkable selectivity. Herein, we report the first partial hydrogenation of benzene towards 1,3‐cyclohexadiene under mild conditions in a catalytic biphasic system consisting of Ru@Pt nanoparticles (NPs) in ionic liquid (IL). The tandem reduction of [Ru(COD)(2‐methylallyl)2] (COD=1,5‐cyclooctadiene) followed by decomposition of [Pt2(dba)3] (dba=dibenzylideneacetone) in 1‐n‐butyl‐3‐methylimidazolium hexafluorophosphate (BMI⋅PF6) IL under hydrogen affords core–shell Ru@Pt NPs of 2.9±0.2 nm. The hydrogenation of benzene (60 °C, 6 bar of H2) dissolved in n‐heptane by these bimetallic NPs in BMI⋅PF6 affords 1,3‐cyclohexadiene with an unprecedented 21 % selectivity at 5 % benzene conversion. Conversely, almost no 1,3‐cyclohexadiene was observed when using monometallic Pt0 or Ru0 NPs under the same reaction conditions and benzene conversions. This study reveals that the selectivity is related to synergetic effects of the bimetallic composition of the catalyst material as well as to the performance under biphasic reaction conditions. It is proposed that colloidal metal catalysts in ILs and under multiphase conditions (“dynamic asymmetric mixtures”) can operate far from the thermodynamic equilibrium akin to chemically active membranes.


Introduction
Benzene hydrogenation is one of the most investigated reactions when using metal nanoparticles (NPs) in view of its industrial applicationsa nd appealing basic surfaces cience. [1][2][3][4][5][6][7][8] It is assumedt hat the hydrogenation of benzene proceeds stepwise, that is, coordination of the aromatic to the metal surface followed by its reduction to 1,3-cyclohexadiene (CHD) then to cyclohexene( CHE) and finally to the thermodynamic cyclohexane (CHA) product. [1] Although the partial hydrogenation of benzene to CHE is quite ac hallenge, [9] it can be performed on an industrial scale with relativelyh igh selectivities by employing modifiedR un anocatalysts. [10] However,t he preparation of mono-hydrogenated products (CHDs) is still ac hallenge even in terms of detection under catalytic hydrogenation conditions, as under standard conditions the reactiono fb enzene and H 2 to CHDs is 13 kcal mol À1 uphill in free energy.T his catalytic reaction is therefore thermodynamically impossible. In this vein, first-principles DFT calculations of benzene hydrogenation over Pt(111)s uggest that CHD and CHE are expected to be at best minor products,asthey are not formed along the dominant reaction path. The only product that can desorb is CHA, and the most-abundantr eaction mixture contains benzene and hydrogen. [11,12] Nonetheless,C HE and CHDs have been obtained duringh ydrogenation of benzene promoted by lanthanide NPs in ammonia, albeit in virtually stoichiometric conditions. [13] Interestingly,C HD is also usually observed as an intermediate duringt he dehydrogenation of CHE or CHA by Pt catalysts. [14,15] Others [16] and our group [17] have already demonstrated that Ru NPs modified by ionic liquids( ILs) are quite effective and selectivec atalysts for the partial hydrogenationo fb enzene to CHE, but no CHDs have been detected so far.W ee nvisioned that the selectivity of this reaction may be improved by controlling the electronic properties of the NPs, the atomic geometry of the NPs' surface atoms in the ILs, the reactions conditions (temperature,p ressure and benzene concentration) and by using multiphase conditions, fore xample, by extracting the formed CHE from aN Ps/IL/benzene phase. Therefore, working far from the thermodynamic equilibrium,i deally in non-equilibrium thermodynamic conditions, it will be possible to achieve higherC HE selectivities, at least at the very early stages of benzene hydrogenation.
Since the earliest reports on catalytic benzene hydrogenation, 1,3-cyclohexadiene andc yclohexene have been proposed as key intermediates. However,t he former hasn everb een obtained with remarkable selectivity.H erein, we report the first partial hydrogenation of benzene towards 1,3-cyclohexadiene under mild conditions in ac atalytic biphasic system consisting of Ru@Ptnanoparticles (NPs) in ionic liquid (IL). The tandem reductiono f[ Ru(COD)(2-methylallyl) 2 ]( COD = 1,5-cyclooctadiene) followed by decomposition of [Pt 2 (dba) 3 ]( dba = dibenzylideneacetone) in 1-n-butyl-3-methylimidazolium hexafluorophosphate (BMI·PF 6 )I Lu nder hydrogen affords core-shell Ru@Pt NPs of 2.9 AE 0.2 nm. The hydrogenation of benzene (60 8C, 6bar of H 2 )d issolved in n-heptaneb yt hese bimetallic NPs in BMI·PF 6 affords 1,3-cyclohexadiene with an unprecedented 21 %s electivity at 5% benzene conversion. Conversely,a lmost no 1,3-cyclohexadiene was observedw hen using monometallic Pt 0 or Ru 0 NPs under the same reactionc onditions and benzene conversions. This study reveals that the selectivity is related to synergetic effects of the bimetallic composition of the catalystm ateriala sw ell as to the performance under biphasic reaction conditions. It is proposed that colloidal metal catalysts in ILs and under multiphase conditions ("dynamic asymmetric mixtures") can operate far from the thermodynamic equilibrium akin to chemically active membranes.
Activation of the Pt catalytic centre (using which, CHD is formed duringd ehydrogenation of CHE) may be induced by the introductionofRuasthe second component in abimetallic NP. [18,19] Analogous observations in terms of electronic modifications have been made with Pd@Au NPs in dehalogenation reactions, [20] hydrogenation reactions [21] and Pt@Co [22] and Pd@Ag NPs for decomposition of formic acid. [23] We reporth erein that, indeed, the use of an extracting phase with the bimetallic Ru@PtN Ps in 1-n-butyl-3-methylimidazolium hexafluorophosphate (BMI·PF 6 )a llows not only the formationo fC HE but also 1,3-CHDw ith unprecedented selectivity for the partial hydrogenation of benzene.

Results and Discussion
The most commonly used syntheticm ethod for the generation of bimetallic core@shell metal NPs is the reduction of as econd metal onto pre-formed cores, but this generally leads to larger NPs (> 10 nm). [24] By using the "template" method,s mall Ru@Pt NPs have been easily prepared in BMI·PF 6 .T he Pt 0 precursor was decomposed over ap reviously synthesised ruthenium core, while self-nucleation of Pt was inhibited by keeping the temperature above the nucleation temperature of Pt. [18] Thus, the reduction of [Ru(COD)(2-methylallyl) 2 ]( COD = 1,3-cyclooctadiene)i nB MI·PF 6 at 75 8Cf or 18 hu nder 5bar of hydrogen affords Ru 0 nanoparticles. [25] The addition of [Pt 2 (dba) 3 ]( dba = dibenzylideneacetone) to the thus-prepared Ru 0 nanoparticles in IL followed by treatment with molecular hydrogen (4 bar) at 75 8Cf or 24 ha ffords ab lack solution containing Ru@Ptn anoparticles of 2.9 AE 0.1 nm (Figure 1). These nanoparticles were characterised by Rutherford backscattering (RBS), transmission electron microscopy (TEM), scanning transmission electron microscopy (STEM), energy-dispersiveX -ray spectroscopy (EDS), X-ray diffraction (XRD) and X-ray photoelectrons pectroscopy (XPS).T he RBS analysiss hows a2 :1 Pt/Ru composition (Figure S1 in the Supporting Information), indicating that half of the initial Ru was not incorporated in the bimetallic structure. Indeed, the Ru speciesn ot incorporated were extracted during the purification step (extraction with ethanol, benzene and pentane) as determined by inductively coupled plasma optical emissions pectroscopy (ICP-OES) analysiso ft he organic phase (Table S1 in the Supporting Information).
The mean size diameter of Ru@PtN Ps determinedf rom TEM (120 kV) and high-angle annular dark field (HAADF)-STEM (300 kV) micrographs ( Figure 1) are 2.8 AE 0.1 and 2.9 AE 0.1 nm, respectively.T he NPs are well distributed and stable in the IL, as seen in Figure1a, b. High-resolution HAADF-STEMi mages of isolated particles are also shown (insets). High Z-contrast allied to fast Fourier transform (FFT) analyses may indicatea ne xcess of Pt surrounding the Ru@Pt particles. Typical face-centred cubic (fcc) features of the Pt arrangement sometimes prevail in the HAADF-STEMa nalyses. In some images, Pt species are also easily observed outsidet he particles. The single-particle driftcorrected EDS profile was also acquired, supporting the observation of aP t-richR u@Pts urface ( Figure 2). The mean particle size of the Ru@PtN Ps (2.8-2.9 nm) is larger than those of the monometallic Ru 0 and Pt 0 (both approximately 2.5 nm) prepared in ILs. The Ru@Pt NPs in the IL are stable and do not show any sign of agglomeration/aggregation after one week ( Figure S2 in the Supporting Information). Note that density Figure 1. Micrographs of the Ru@Pt NPs in BMI·PF 6 ;a)HAADF-STEM (300 kV);b)TEM (120 kV);size distribution histogramsc )before and d) after catalysis. HAADF-STEMi mages of isolatedR u@Pt NPs are shown as insets. ChemCatChem 2017, 9,204 -211 www.chemcatchem.org 2017 Wiley-VCH Verlag GmbH &Co. KGaA, Weinheim functional theoryc alculations have been already reported to characterise the interactions between BMI·PF 6 IL and Ru@Pt model nanoclusters. [26] The metal-metal distances calculated from the XRD diffractogram of isolatedR u@PtN Ps are shifted compared with those of pure metals ( Figure S3 in the SupportingI nformation). The obtainedd iffractogramsm atch with the one simulated for Ru@Pt ( Figure S3 in the Supporting Information) rather than with the one calculated for Pt/Rua lloy and the monometallic Pt + Ru aggregate mixture. [19,27] The highest convergence found with aR u@Pt core-shell structurew ith ah exagonal close packed (hcp) Ru core and af ace-centred cubic (fcc) Pt shell has an average size of 3.4 nm, slightly larger than that determined by TEM ( Figure 1). However,t he reflection for Pt(2 00)c ould not be detected. It is assumed that the intensity is lowereda nd shifted to lower degrees, resultingi na no verlapping with the Ru(1 01)r eflection akin to that observed earlier. [19] Also, the Pt(2 20)i ss hiftedt ol ower degrees, indicating at hicker platinum shell ( Figure S3 in the Supporting Information). The determined lattice parameters for ruthenium are in good agreement with theoretical values, and the refinedl attice parameterf or the platinum shell gives ac ell parameter of 4.028 ,w hich is slightly stretched compared with pure Pt (3.9231 ), [28] indicating interactions between the Ru core and Pt shell.T his is probably due to the poorlyc rystalline Ru core and/or to the partial incorporation of Ru into thePtshell. [19] Furthermore, the surfacec omposition of the Ru@PtN Ps was investigated by XPS with two incident photon energies (1840 eV and 3000 eV). The wide-scan XPS spectrum indicates the presence of Pt, Ru, O, C, N, Fa nd Pa toms ( Figure S4 in the Supporting Information). The presence of the oxygen atom is probablyd ue to the oxidation of the metal surface during experimental manipulation and the fluorine, nitrogen and phosphorusa toms are from the IL (no metal fluoride was observed). Figure 3s hows the XPS spectra in the Pt 4f and C1s+ Ru 3d regions measuredw ith incident photon energies E ph of 1840 and 3000 eV before the catalytic reaction. The Pt 4f region (Figure 3a)r eveals the presence of three different chemical states: Pt 0 at 70.9 eV,P t II at 72.1 eV andP t IV at 74.9 eV. [27,29] The Ru 3d region (Figure 3b)i sc lose in energy to the C1sr egion where it is possible to observe an overlap between both regions.A s the carbon signal comesf rom different sources,i ti sh ard to analyse the Ru 3d region in this case. It is possible to observe that the Ru 3d signal is composed of 11 %R u 0 in the as-prepared sample with the remainings ignal coming from Ru in higher oxidation states. [16] The change in the incident photon energy allows us to determine the atomic distribution inside the nanoparticles. [18] To perform this study,t he ratio was mea-  . ChemCatChem 2017, 9,204 -211 www.chemcatchem.org sured between the intensities of the Pt 4f to Ru 3p 3/2 XPS regions normalised by the corresponding differentialc ross section and incident flux. [30] The inelastic mean free path of photoelectrons ejectedo wing to an incident photon energy of 1840 eV is around 18 (Pt 4f and Ru 3p 3/2 regions), and for an incident photon energy of 3000 eV,i ti sa round 27 (Pt 4f region) and 29 (Ru 3p 3/2 region). [31] In this way,b yc omparing the XPS intensitiesf rom these two regions, it is possible to probe the Ru and Pt atoms at essentially the same depth of the sample. The ratio of intensities of Pt 4f/Ru 3p 3/2 changes from 2.2 (E ph = 1840 eV) to 0.8 (E ph = 3000 eV). As this ratio decreasesw hen increasing the probed depth,i ti sc onsistent with an increasei nt he intensity of the Ru 3p 3/2 region compared with the intensity of the Pt 4f region at E ph = 3000 eV. This result is evidence for the existence of aR u-rich core and aP t-rich shell region.
For comparison purposes,P t 0 NPs ( % 2.5 nm in mean diameter) [32] and Ru 0 ( % 2.6 nm in mean diameter) [17] were prepared in BMI·PF 6 by using knownp rocedures. It is known that the hydrogenation of benzene at 75 8C, under 4bar of hydrogen by Pt 0 in BMI·PF 6 affords only CHA, [18] whereas CHE was detected at very low benzene conversion in the reaction performed with Ru 0 NPs in BMI·PF 6 . [17] However,w efound that by increasing the H 2 pressure to 6bar and reducingt he temperature to 60 8C, partially hydrogenatedp roducts (CHE and1 ,3-CHD, Scheme 1) could be detected (Table S2 in the Supporting Information).
Only CHA was detectedi nt he reaction performed with Pt 0 NPs, whereas in that performed with the Ru 0 NPs, CHE was observed with 12 %s electivity and, to our delight, 1,3-CHDw ith 23 %s electivity at 1% benzene conversion (Table S2 in the Supporting Information). In the case of bimetallic Ru@Pt NPs, the selectivity was 34 %f or CHE (at 1% benzene conversion) and no CHDs were detected. This is ac lear indication that the Ru core changes the Pt shell properties as the reactionu sing monometallic Pt nanoparticles has no selectivity for the partial hydrogenation products, whereas the Ru@Pt shows relatively high selectivity for CHE (compare entries 1a nd 3, Table S2 in the SupportingI nformation).
When the hydrogenation reaction was performed in the presenceo f2mL of n-heptane using the same reactions conditions, 1,3-CHD was formed (see entries 4-6, Ta ble S2 in the Supporting Information) even in the case of Pt 0 .H owever,1 ,3-CHD was obtained with 27-33 %s electivity when Ru 0 or Ru@Pt was used (entries 5a nd 6, Table S2 in the Supporting Information), although at very low benzene conversion (1 %). Most impressive, high selectivities (up to 21 %) for1 ,3-CHDw ere achieved in the case of Ru@Pt NPs even at 5% benzene conversions (Figure 4).
It is worth noting that no significant changes on the mean diameter and size distribution of the NPs wereo bserved by TEM and STEM after catalysis (Figures 1c and d, and Figure S5 in the Supporting Information). Moreover,t he recovered catalytic IL dispersion could be re-used atl east three times with only as mall drop in the 1,3-CHD selectivity,f rom 21 %t o1 7% after the third cycle ( Figure S6 in the Supporting Information).
XPS analysis of the Ru@PtN Ps after catalysis revealed that the intensity of the components in the Pt 4f region (E ph = 1840 eV) changed from 9% (Pt 0 ), 25 %( Pt II )a nd 66 %( Pt IV )i n the as-prepared sample to 43 %( Pt 0 ), 46 %( Pt II )a nd 11 %( Pt IV ) after the catalytic reaction (Figure 5a nd Figure S7 in the Supporting Information).
The Ru 0 fractiona lso changes from 11 %b efore to 31 %a fter the reaction. It is clear from the XPS measurements that there is an enrichment of the skin layers of the nanoparticlesw ith Pt atomsa fter catalysis ( Figure 5). For af ixed photon incident energy of 1840 eV,t he Pt 4f/Ru 3p 3/2 ratio increases from 2.2 (before)t o5 .5 (after the catalytic reaction). These results provide evidence that the surface composition changes through surfaces egregation as nanoparticles are exposed to hydrogenation conditions, as usually observed in bimetallic NPs. [22,33,34] Therefore, the dropin1 ,3-CHD may be relatedt othe structural changes of the metal surface.  At this stage of our investigations, there are at least two possible explanations concerning the "apparent" fail on the standard principle of traditional catalysis in which ac atalyst does not alter the final thermodynamic equilibrium of ar eaction. First, the reaction is not catalytic, that is, the benzene hydrogenation is stoichiometricv is-à-vist he metal NPs at such lower benzene conversions.H owever,t his hypothesis can be discarded as turnover numbers( TONs) > 30 were obtained. Note that the catalytic reactions have been repeated several times by three different chemists in two different laboratories (Brazil and UK) using distinct experimental setups.
The second and most probably explanation is that under these asymmetricd ynamic conditions (metalN Ps/ILs/organics/ H 2 )t he reactionp roceeds far from equilibrium [35] and the reaction surface is better considered as as eparate thermodynamic system [36][37][38][39] ("2D reaction surface"). [36] Indeed, chemical reactions far from equilibrium may exhibit various phenomena of temporal and spatial self-organisation [40] and complex transitory structures, [41] as fore xample at the Pt surface. [40] Moreover, the viscoelastic model is probably the most adequate to describe the dynamic asymmetric mixture (metalN Ps/IL/organic substrates and products)a si ti sc omposed of fast and slow components, as in colloidal suspensions. [42] Therefore, the IL/ NPs catalytic system can be regarded as ac hemically active membrane [43] (confined space) [44,45] in which benzene is retained in the catalytic phase and the hydrogenated products (CHD, CHE and CHA) are expelled akin to that observed in the partial hydrogenation of dienes by Pd-based catalysts in neat ILs [46][47][48] or supported in hybrid-IL materials. [49] Hence, the catalytic system metal NPs/IL/n-heptane provides the conditions in which the 1,3-CHDf ormed is removedf rom the catalytic site probably by the benzene/n-heptane mixture. Indeed, in the reaction performed with the isolatedR u@Pt NPs, that is, without BMI·PF 6 ,t he 1,3-cyclohexadiene selectivity drops from2 1% in IL to 5% withoutI La t5%b enzene conversion (Figure6). Moreover, the reaction performed by employing co-solvents such as dichloromethane,w hich is miscible with the IL, gave very low selectivity for the partial benzene hydrogenation products.
Moreover,b enzene is at least 20 times more soluble in IL than the alkenes [17] and thus can displace the relatively stable CHD from the metal surface. As expected, the selectivityo ft he partial benzene hydrogenation products (1,3-CHD and CHE) decreases significantly with increasingb enzene conversion, from 32 %a t1 %c onversion to 21 %a t5 %c onversion ( Figure 4). No 1,4-CHDw as observed in any experiment using Pt 0 ,R u 0 or Ru@Ptn anoparticles, indicating that the hydrogenation proceeds preferentially through the 1,3-CHDi ntermediate.
The hydrogenation of benzene and 1,3-CHD is as tructuresensitiver eaction, [18] whereas the hydrogenation of monoalkene is not.T he hydrogenation of benzene and 1,3-CHD should, in principle, occur only at specific sites on the metal surface, whereas the hydrogenation of CHE occurs on almost all of the surfacem etal active sites. Therefore, the Ru@PtN Ps in BMI·PF 6 /n-heptane provides the right environment to achieve higher selectivities for the mono-hydrogenation of benzene, that is, reducing the contact of CHD with specific catalytic sites, mainly the surfacea toms at whichi tw as formed. This activation is very likely ar esult of the modification of the geometric/electronic structure of the Pt surface induced by the presence of subsurface Ru atoms. This modification accelerates the benzene partial reduction through as ubstantial stabilisation of the reactive intermediate (1,, which is expelled  . Selectivity of the products in the benzene hydrogenationcatalysed by Ru@Pt NPs in the presence and absenceo ft he IL BMI·PF 6 (1 mL). Reaction conditions:6 7mmol metal NPs, benzene/metal (molarr atio) = 670, 60 8C and 6bar of H 2 ;2mL of n-heptane. ChemCatChem 2017, 9,204 -211 www.chemcatchem.org by benzene and removedf rom the IL catalytic phase by the hydrocarbon phase before it is hydrogenated to CHE. The origin of the Ru subsurface effect is very likely related to the distorteda nd more compressed fcc Pt shells of the Ru@Pt NPs compared with bulk fcc Pt, as already observed for sub-5.0 nm Ru@Pt NPs. [18] Note that the larger size of the Ru@Pt( 2.9 nm) NPs compared with those of monometallic Pt 0 NPs (2.5 nm) implies only as mall reduction in face-to-cornera nd edge-surface atoms (from 55 %t o4 8%,a ssumingf cc Pt in both cases). [50] Moreover,the Pt surface is more electrophilic in the Ru@Pt NPs than in Pt 0 NPs as observed by XPS (only 9% of Pt 0 in Ru@NPs against 37 %i nP t 0 NPs). [27] It is known that the affinity of benzene for its active site increases with decreasing the electron density of the surface metals. [1] In turn, the electron-deficient NPs surface increases the stabilisation of the 1,3-CHD intermediate, and then allows its desorption.

Conclusions
We have demonstrated that small ( % 2.9 nm), stable and welldistributed BMI·PF 6 soluble Ru@Pt NPs can be easily prepared by at andem reduction of Ru II followed by decomposition of Pt 0 organometallic precursors under hydrogen. The Pt 0 surface of these core-shell like Ru@Pt NPs display unprecedented catalytic properties towards the hydrogenation of benzene to 1,3-CHD (21 %a t5%b enzene conversion) that is completely different from monometallicP t 0 or Ru 0 ,w hich show "very low selectivities" for partially hydrogenated products under the same reactionc onditions. The Pt electronic modificationi sv ery likely due to its higher electron deficiency provoked by the subsurface Ru atoms in which the "encapsulated" substrate/intermediate molecule can only adopt specific conformations as it has to adjust to the geometric and electronic features of the IL/NPs container. The fine-tuning of the reactionc onditions, in particular the use of n-heptane, in the hydrogenation of benzene allows the formation of the partial hydrogenation product 1,3-CHD in relatively high selectivity (21 %), although at low substrate conversion (5 %) by using Ru@PtN Ps in BMI·PF 6 . These resultsi ndicatet hat metal NPs in highly organised ILs can operate far from equilibrium( similart oachemically active membrane) and thus opens an ew window of opportunities for the developmento fm ore selective "soluble" heterogeneousc atalysts.

Experimental Section General information
All manipulations involving the metal complexes were carried out under an argon atmosphere by using Schlenk or glovebox techniques. [Ru(COD)(2-methylallyl) 2 ]w as obtained from Sigma-Aldrich and used without further purification. The BMI·PF 6 , [51] [Pt 2 (dba) 3 ] [52] and the Ru 0 [25] and Pt 0 [32] NPs were prepared according to reported procedures. Benzene was degassed and stored under argon prior to use. All of the other chemicals were purchased from commercial sources and used without further purification. NMR spectra were recorded with aV arian VNMR spectrometer (300 MHz).

Preparation of Ru@Pt NPs
In as tandard reaction, aF ischer-Porter bottle was loaded in the glovebox with the precursor [Ru(COD)(2-methylallyl) 2 ]( 64.5 mg, 0.2 mmol) and BMI·PF 6 (6 mL). The system was stirred under vacuum for 20 min and heated to 75 8C. Then, 5bar hydrogen was added to the system, which was kept reacting for 18 ha t7 5 8C. The obtained black suspension was evacuated to remove the volatiles. Then, to the formed Ru nanoparticles, as olution of [Pt 2 (dba) 3 ] (110 mg, 0.1 mmol) in acetonitrile (10 mL) was added. All volatile compounds were removed under reduced pressure at 75 8Ca nd 4bar of hydrogen was added. After 24 h, the black solution was washed with benzene (3 20 mL), ethanol (3 10 mL) and pentane (3 30 mL). Again, the system was evacuated to remove all volatile compounds. The formed nanoparticles were stored under argon at À20 8C. The nanoparticles were analysed by transmission electron microscopy (TEM), scanning transmission electron microscopy (STEM), energy-dispersive X-ray spectroscopy (EDS), X-ray photoelectron spectroscopy (XPS) and electron and X-ray diffraction (XRD). For XRD analysis, the NPs were isolated by centrifugation with the addition of THF (10 mL) and washed with CH 2 Cl 2 (10 10 mL), ethanol (10 10 mL) and pentane (3 10 mL) and dried under reduced pressure.

Hydrogenation of benzene
As ag eneral procedure, as olution of benzene (4 mL, benzene/catalyst = 670) with ac o-solvent, n-heptane (2 mL), was added to aF ischer-Porter reactor that contained the appropriate amount of catalyst (67 mmol of metal NPs) dissolved in BMI·PF 6 (1 mL). The reactor was pressurised with 6bar of H 2 at 60 8C. As ample was taken from the reaction mixture every 10 min. After the desired reaction time, the reactor was cooled to room temperature and depressurised. GC and GC-MS analysis (Figures S8-S10 in the Supporting Information) of the samples were used to determine conversions and selectivities.

RBS
Measurements were carried out in a3MV Ta ndetron accelerator using aH e + ion beam of 1.5 MeV at IF/UFRGS. The Si surface-barrier detector was positioned at ascattering angle of 1658.

TEM and STEM
TEM analysis was performed by using aJ EOL JEM 1200 ExII operating at 80 kV.T EM samples were prepared by dropping the acetone-diluted solution of the isolated Ru@Pt nanoparticles onto acopper TEM grid. The ruthenium and platinum contents were determined by EDS by using aN ORAM Pioneer spectrometer with ab eam energy of 200 kV.S TEM and high-resolution TEM (HRTEM) were performed by using aX FEG Cs-corrected FEI Titan 80/300 mi- ChemCatChem 2017, 9,204 -211 www.chemcatchem.org croscope at INMETRO operated at 80 and 300 kV.H igh Z-contrast images were acquired through STEM by using ahigh-angle annular dark field detector (HAADF) and as emi-convergence angle of 27.4 mrad. Spatially correlated EDS profile experiments were carried out by using the Ka nd Ll ines from Ru and Pt. The typical lateral resolution was greater than 0.1 nm.

XPS
For the XPS measurements, the powder of the Ru@Pt nanoparticles was spread out over the carbon tape and introduced into the analysis chamber at the D04A-SXS beam-line endstation [6] at LNLS. The sample was investigated by using the long scan Ru 3d, Ru 3p 3/2 , Pt 4f,O 1s and C1ss can regions. The spectra were collected by using an InSb (111)d ouble crystal monochromator at fixed photon energies of 1840 and 3000 eV.T he hemispherical electron analyser (PHOIBOS HSA3500 150 R6) was set at ap ass energy of 30 eV,a nd the energy step was 0.1 eV,w ith an acquisition time of 100 ms/ point. The overall resolution was around 0.3 eV.T he base pressure used inside the chamber was around 1.0 10 À9 mbar.T he monochromator photon energy calibration was done at the Si Ke dge (1839 eV). An additional calibration of the analyser's energy was performed by using as tandard Au foil (Au 4f 7/2 peak at 83.8 eV). We also considered the C1sp eak value of 284.5 eV as the reference to verify possible charging effects. The XPS measurements were obtained at a458 take off angle at room temperature.

XRD
The XRD patterns were recorded for a2q range of 20 to 908 with a0 .058 step size and measurement time of 1sper step with Cu Ka radiation (l = 1.54 )a nd monochromator of graphite. Data processing was performed by the Rietveld method by using FullProf software. The instrumental resolution function (IRF) of the diffractometer was obtained from the LaB 6 standard.