Magnetic Fe2O3–SiO2–MeO2–Pt (Me = Ti, Sn, Ce) as Catalysts for the Selective Hydrogenation of Cinnamaldehyde. Effect of the Nature of the Metal Oxide

The type of metal oxide affects the activity and selectivity of Fe2O3–SiO2–MeO2–Pt (Me = Ti, Sn, Ce) catalysts on the hydrogenation of cinnamaldehyde. The double shell structure design is thought to protect the magnetic Fe2O3 cores, and also act as a platform for depositing a second shell of TiO2, SnO2 or CeO2 metal oxide. To obtain a homogeneous metallic dispersion, the incorporation of 5 wt % of Pt was carried out over Fe2O3–SiO2–MeO2 (Me = Ti, Sn, Ce) structures modified with (3-aminopropyl)triethoxysilane by successive impregnation-reduction cycles. The full characterization by HR-TEM, STEM-EDX, XRD, N2 adsorption isotherm at −196 °C, TPR-H2 and VSM of the catalysts indicates that homogeneous core-shell structures with controlled nano-sized magnetic cores, multi-shells and metallic Pt were obtained. The nature of the metal oxide affects the Pt nanoparticle sizes where the mean Pt diameter is in the order: –TiO2–Pt > –SnO2–Pt > –CeO2–Pt. Among the catalysts studied, –CeO2–Pt had the best catalytic performance, reaching the maximum of conversion at 240 min. of reaction without producing hydrocinnamaldehyde (HCAL). It also showed a plot volcano type for the production of cinnamic alcohol (COL), with 3-phenyl-1-propanol (HCOL) as a main product. The –SnO2–Pt catalyst showed a poor catalytic performance attributable to the Pt clusters’ occlusion in the irregular surface of the –SnO2. Finally, the –TiO2–Pt catalyst showed a continuous production of COL with a 100% conversion and 65% selectivity at 600 min of reaction.


Introduction
Because of the versatility in composition and synthesis, materials with core-shell structure architecture have attracted particular attention in fields in which nanotechnology has a central role such as electronics [1], biomedicine [2], environmental remediation [3] and heterogeneous catalysis [4][5][6]. These materials in a nano-or micro-size structure produce a large surface-to-volume ratio, increasing the dominance of surface atoms and predominance of quantum effects, thereby improving their performance in various applications [7]. Efforts have been focused on the application of these materials in heterogeneous catalysts, specifically to increase their activity, operational stability and selectivity [8].
The selective catalytic hydrogenation of organic substrates containing a number of unsaturated functional groups is an important step in the industrial preparation of fine chemicals, increasing their relevancy for fundamental research in catalysis [9]. The heterogeneous selective hydrogenation, Among the transition metals, platinum (Pt) is the most-reported active phase in the hydrogenation of CAL. The selectivity of this active phase towards the formation of unsaturated alcohols has been attributed to the geometric and electronic defects of the metal [13] and can be improved by the incorporation of a second metal [14], a promoter [15], modification of the particle size [16] or by using a non-inert metal oxides in support [10,17] such as TiO2 [11], SnO2 [18], CeO2 [19], ZnO [20], among others. Moreover, the well-known strong metal-support interaction (SMSI) effect promoted by partially reducible metal oxide [21][22][23][24] is observed in metal-supported catalysts submitted to a reducing atmosphere at temperatures above 300 °C. At this condition, the metal oxide partially covers the metallic phase, promoting C=O catalytic hydrogenation performance.
In the field of heterogeneous catalysis, the use of core-shell catalysts as single unit with different components has been previously reported [25][26][27]. For CAL hydrogenation, Song et al. has reported the use of a Pt@CeO2 nanocatalyst achieving over 95% conversion with 87% selectivity to HCAL in 5 h under 1 atm H2 pressure [28]. In the same way, Liu et al. used MOF@Pt@MOF (MOF: Fe III -based MIL-100) as a catalyst to produce COL at 96 % selectivity and 95 % of conversion of CAL in 4 h of reaction. However, the preparation process of MOF catalysts is relatively cumbersome, uneconomical and inapplicable in mass production.
One way to produce improved Pt-based catalysts is to employ a core-shell configuration, including a magnetic core isolated by a layer of oxide shell, which could act as support for Pt-metal nanoparticles (Pt NPs). The magnetic core-shell structure catalysts could provide the additional property of easy removal of the nanostructured catalyst by an external magnetic field [28]. The magnetic particles can be synthesized by different chemical methods; the polyol process is an effective sol-gel route for nano-and micro-particle synthesis with a controlled shape and size [29]. Moreover, is possible to modify the core materials to produce single-shell, multi-shell, or porous-Scheme 1. Cinnamaldehyde (CAL) hydrogenation pathways.
Among the transition metals, platinum (Pt) is the most-reported active phase in the hydrogenation of CAL. The selectivity of this active phase towards the formation of unsaturated alcohols has been attributed to the geometric and electronic defects of the metal [13] and can be improved by the incorporation of a second metal [14], a promoter [15], modification of the particle size [16] or by using a non-inert metal oxides in support [10,17] such as TiO 2 [11], SnO 2 [18], CeO 2 [19], ZnO [20], among others. Moreover, the well-known strong metal-support interaction (SMSI) effect promoted by partially reducible metal oxide [21][22][23][24] is observed in metal-supported catalysts submitted to a reducing atmosphere at temperatures above 300 • C. At this condition, the metal oxide partially covers the metallic phase, promoting C=O catalytic hydrogenation performance.
In the field of heterogeneous catalysis, the use of core-shell catalysts as single unit with different components has been previously reported [25][26][27]. For CAL hydrogenation, Song et al. has reported the use of a Pt@CeO 2 nanocatalyst achieving over 95% conversion with 87% selectivity to HCAL in 5 h under 1 atm H 2 pressure [28]. In the same way, Liu et al. used MOF@Pt@MOF (MOF: Fe III -based MIL-100) as a catalyst to produce COL at 96% selectivity and 95% of conversion of CAL in 4 h of reaction. However, the preparation process of MOF catalysts is relatively cumbersome, uneconomical and inapplicable in mass production.
One way to produce improved Pt-based catalysts is to employ a core-shell configuration, including a magnetic core isolated by a layer of oxide shell, which could act as support for Pt-metal nanoparticles (Pt NPs). The magnetic core-shell structure catalysts could provide the additional property of easy removal of the nanostructured catalyst by an external magnetic field [28]. The magnetic particles can be synthesized by different chemical methods; the polyol process is an effective sol-gel route for nanoand micro-particle synthesis with a controlled shape and size [29]. Moreover, is possible to modify the core materials to produce single-shell, multi-shell, or porous-shell structures to improve catalytic activity, selectivity and structural stability [3]. An alternative approach for the protection of magnetic cores is the use of the Stöber method to produce silica-coated shells on magnetic NPs via the facile hydrolysis of tetraethyl orthosilicate (TEOS). It has been reported that uniform and robust silica shells have been formed by the complete hydrolysis of TEOS followed by the condensation of silicic acid, giving a network of tetrahedral SiO 4 units with shared vertices [30].
The surface physiochemical properties of small crystallites change dramatically at sizes lower than 5.0 nm, where the metal or metal oxide particles have low-coordinated surfaces, improving the catalytic activity [31]. The development of core-multi-shell nano-supports allows the shell characteristics to be modified in a controlled way in terms of the particle size, shape or crystallinity to improve their catalytic effect in terms of activity, selectivity and operational stability.
The aim of this work is to provide new magnetic multifunctional Fe 2 O 3 -SiO 2 -MeO 2 -Pt (Me = Ti, Sn, Ce) nanomaterials as catalysts to be used in the selective hydrogenation of CAL. One of the most important steps to obtain the desired nanocatalyst structured microspheres is the coating process with TEOS for the formation of a silica shell as a protection layer for the magnetic core and a growth platform to provide for a homogeneous dispersal of a metal oxide shell on the Fe 2 O 3 -SiO 2 surface. The synthesis of the catalysts, while employing stepwise impregnation-reduction deposition, provides highly dispersed Pt catalysts in a core-shell structure. The effect of the nature of the metal oxide MeO 2 (Me = Ti, Sn, Ce) on well-stabilized magnetic core double shell structures in the selective hydrogenation of CAL was also described in this experiment.

Fe 3 O 4 -SiO 2 Structure
The Fe 3 O 4 -SiO 2 structures were synthesized starting from Fe 3 O 4 -NPs coated with TEOS using the Stöber method in basic media [32]. The iron precursor (FeCl 3 ·6H 2 O) was dissolved in an ethylene glycol and PVP-K30 mixture, and sodium acetate was added as a nucleating agent under vigorous magnetic stirring for 5 h. The solution was transferred to a Teflon-autoclave reactor (Hydrion Scientific Instruments Company Ltd., Baltimore, MD, USA) and heated at 200 • C for 8 h to produce a homogeneous and stabilized Fe 3 O 4 -NPs dispersion. Using magnetic separation, the dispersion was washed several times with an ethanol-water mixture before the coating procedure with Fe 3 O 4 -NPs and TEOS as silica precursor. Then, 2 mL of TEOS was added drop-wise to 300 mg of Fe 3 O 4 -NPs dispersed in a 180 mL ethanol, 60 mL water and 10 mL NH 3 mixture, while stirring with mechanical agitation for 6 h at room temperature. The solid was separated by magnetization, washed three times with an ethanol-water mixture and dried at 50 • C for 12 h to produce Fe 3 O 4 -SiO 2 nanostructures. The coating process with titanium and tin was carried out with a modified Stöber method [33], with 300 mg of Fe 3 O 4 -SiO 2 particles dispersed in 200 mL of ethanol and 1 mL ammonium solution, which was stirred for 1 h at 300 rpm under mechanical agitation. The condensation of the respective alkoxides was carried out by the addition drop-by-drop of 40 mL of ethanol with 5% by volume of the (Ti or Sn) alkoxide precursor to the dispersion and heating to 60 • C for 4 h. The coating process with Ce was carried out using a solvothermal methodology, as previously reported [34], with 300 mg of Fe 3 O 4 -SiO 2 particles dispersed in 300 mL of absolute ethanol sonicated for 30 min at room temperature. The addition drop-by-drop of 60 mL of an aqueous dissolution of 1.2 g urea, 6.0 g D-xylose and 1.0 g To better-control metallic dispersion, the incorporation of Pt was carried out on functionalized APTMS structures. The functionalized procedure that was carried out contacted the Fe 2 O 3 -SiO 2 -MeO 2 (Me = Ti, Sn and Ce) structures with APTMS under reflux in toluene for 24 h, was washed with a 3:1 mixture of acetone:toluene and dried at 50 • C for 12 h. Then, the deposition of Pt-NPs was carried out using an aqueous dissolution of K 2 PtCl 6 as a precursor and NaBH 4 as a reducing agent by successive impregnation-reduction steps until reaching a metallic loading of nominally ≈5 wt % Pt.

Characterization
The synthesis of the core-shell structures was studied by high-resolution transmission electron microscopy (HR-TEM) measurement. Morphology was examined using a FEI Tecnai G2 F20 S-Twin microscope (FEI, Hillsboro, OR, USA.), and the microanalysis was performed using a scanning transmission electron microscope (FEI, Hillsboro, OR, USA) and energy-dispersive X-ray spectroscopy (STEM-EDS) (FEI, Hillsboro, OR, USA). The total Pt and Fe content was determined by inductively coupled plasma-optical emission spectrometry (ICP-OES, Perkin Elmer optima 2100 DV ICP, (PerkinElmer, Waltham, MA, USA)) after dissolving the samples in 1:3 HNO 3 :HCl mixtures and diluting them with doubly distilled water. The surface and pore size distributions by nitrogen adsorption isotherms were obtained using a Micromeritics apparatus Model ASAP 2010 at −196 • C (Norcross, GA, U.S.A). X-ray powder diffraction (XRD) pattern measurements were carried out with nickel-filtered CuK 1 radiation (λ = 1.5418 Å) collected on a Rigaku diffractometer (Rigaku, Tokyo, Japan) in the 2θ range of 10 • -80 • . Temperature programmed reduction (TPR-H2) profiles were obtained on Micromeritics TPR/TPD2900 equipment (Micromeritic, Norcross, GA, USA) provided with a thermal conductivity detector in 5% H 2 /Ar flow at a rate of 40 mL·min −1 from room temperature to 800 • C and a 5 • C·min −1 heating rate. The surface charges of materials were determined by zeta potential in a Zetasizer NanoZS Marvern instrument (Malvern, Worcestershire, UK) with a provided electrode tray. Magnetic measurements were carried out at room temperature in a vibrating sample magnetometer (VSM) Lakeshore 7400 series (Lakeshore, New Orleans, LA, USA) with a maximum applied field of 20 kOe. The VSM had been previously calibrated with a pure nickel sphere.

Catalytic Activity
The catalytic activity evaluation of cinnamaldehyde was performed in a batch compact Parr ® 5513 model reactor using 30 mg of catalyst weight and mechanical agitation up to 700 rpm. The reaction was performed at 100 • C, 2000 kPa of H 2 , with a substrate/catalyst mole ratio equal to 600 in a volume of 30 mL of cyclohexane. The reaction was monitored by taking non-invasive samples periodically from the reaction mixture at different reaction times until total cinnamaldehyde conversion. The product quantifications were carried out using a gas chromatograph (Hewlett Packard HP-4890, Palo Alto, CA, USA) equipped with a capillary column (HP-5) and a flame ionization detector (FID).

Material Synthesis
The different steps carried out to produce well-defined were coated with TEOS to produce the starting core-SiO 2 structure. The first layer of silica between the iron species in the core and the further deposition of the metal MeO 2 (Me = Ti, Sn, Ce) oxide was designed to: (1) provide a regular, homogeneous and inert surface of SiO 2 for metal oxide deposition, (2) inhibit the effect of the metal oxides changing the magnetic behavior of the iron species in the core and (3) evaluate the effect of the metal oxide in the catalytic performance of Pt without the effect of the iron species present in the core. Furthermore, the surface of the Fe 2 O 3 -SiO 2 -MeO 2 (Me = Ti, Sn, Ce) structures were functionalized with APTMS before the impregnation of Pt, to enhance the metallic dispersion and to carefully control the Pt particle size with successive impregnation-reduction cycles. were coated with TEOS to produce the starting core-SiO2 structure. The first layer of silica between the iron species in the core and the further deposition of the metal MeO2 (Me = Ti, Sn, Ce) oxide was designed to: (1) provide a regular, homogeneous and inert surface of SiO2 for metal oxide deposition, (2) inhibit the effect of the metal oxides changing the magnetic behavior of the iron species in the core and (3) evaluate the effect of the metal oxide in the catalytic performance of Pt without the effect of the iron species present in the core. Furthermore, the surface of the Fe2O3-SiO2-MeO2 (Me = Ti, Sn, Ce) structures were functionalized with APTMS before the impregnation of Pt, to enhance the metallic dispersion and to carefully control the Pt particle size with successive impregnationreduction cycles.

HR-TEM
The morphology and composition of the core-shell structures during the synthesis was followed by HR-TEM and the corresponding EDX analysis. The Fe3O4-NPs ( Figure S1) used as cores corresponded to micro-spheres with regular shapes formed by nanosize Fe3O4 particles joined together in intimate contact, forming aggregates of Fe3O4-NPs with a mean diameter of ~230 nm. Single Fe3O4-NPs have been reported to have regular square shapes of 20-50 nm [35] as well as 180nm colloidal nanocrystal clusters of Fe3O4, each one composed of many single magnetite crystallites of 10 nm [36]. The partial reduction process of the Fe(OH)x species to Fe3O4 is likely carried out by a dehydration process promoted by the presence of ethylene glycol (reducing agent) at 200 °C during the solvothermal synthesis [29]. These Fe3O4 nanoparticles are then coated by a first SiO2 shell by means of Stober's method, generating a structure Fe3O4-SiO2 core-shell ( Figure S2). This procedure allowed for the isolation of the magnetic core, thus providing a uniform surface to coat a second metal oxide shell. The mean diameter of ~320 nm for the Fe3O4-SiO2 structures allows for an average estimated thickness of ~50 nm silica.
The HR-TEM micrographs of Fe2O3-SiO2-MeO2 (Me = Ti, Sn and Ce) structures are shown in Figure 1a-c. As it was said before, during the calcination step to produce the -MeO2 oxides (Me = Ti, Sn and Ce), the Fe3O4 core was completely oxidized to γ-Fe2O3, in all of the samples, with almost no changes in the morphology. The Fe2O3-SiO2-TiO2 and Fe2O3-SiO2-CeO2 structures showed regular spherical shapes, which were indicative of a uniform metal oxide deposition on the surface of -SiO2. The Fe2O3-SiO2-SnO2 displayed an irregular surface attributable to the presence of larger SnO2 crystals. This trend is in agreement with the tin-based core-shell reported by Pang et al. [37], where the materials showed a non-defined coating after the deposition of SnO2 on the titanium dioxide core surface.
In Figure 1d-f the corresponding STEM-EDS is shown, where the analysis was carried out in the zone marked in Figure 1a

HR-TEM
The morphology and composition of the core-shell structures during the synthesis was followed by HR-TEM and the corresponding EDX analysis. The Fe 3 O 4 -NPs ( Figure S1) used as cores corresponded to micro-spheres with regular shapes formed by nanosize Fe 3 O 4 particles joined together in intimate contact, forming aggregates of Fe 3 O 4 -NPs with a mean diameter of~230 nm. Single Fe 3 O 4 -NPs have been reported to have regular square shapes of 20-50 nm [35] as well as 180-nm colloidal nanocrystal clusters of Fe 3 O 4 , each one composed of many single magnetite crystallites of 10 nm [36]. The partial reduction process of the Fe(OH) x species to Fe 3 O 4 is likely carried out by a dehydration process promoted by the presence of ethylene glycol (reducing agent) at 200 • C during the solvothermal synthesis [29]. These Fe 3 O 4 nanoparticles are then coated by a first SiO 2 shell by means of Stober's method, generating a structure Fe 3 O 4 -SiO 2 core-shell ( Figure S2). This procedure allowed for the isolation of the magnetic core, thus providing a uniform surface to coat a second metal oxide shell. The mean diameter of~320 nm for the Fe 3 O 4 -SiO 2 structures allows for an average estimated thickness of~50 nm silica.
The HR-TEM micrographs of Fe 2 O 3 -SiO 2 -MeO 2 (Me = Ti, Sn and Ce) structures are shown in Figure 1a-c. As it was said before, during the calcination step to produce the -MeO 2 oxides (Me = Ti, Sn and Ce), the Fe 3 O 4 core was completely oxidized to γ-Fe 2 O 3 , in all of the samples, with almost no changes in the morphology. The Fe 2 O 3 -SiO 2 -TiO 2 and Fe 2 O 3 -SiO 2 -CeO 2 structures showed regular spherical shapes, which were indicative of a uniform metal oxide deposition on the surface of -SiO 2 . The Fe 2 O 3 -SiO 2 -SnO 2 displayed an irregular surface attributable to the presence of larger SnO 2 crystals. This trend is in agreement with the tin-based core-shell reported by Pang et al. [37], where the materials showed a non-defined coating after the deposition of SnO 2 on the titanium dioxide core surface.
In Figure 1d-f the corresponding STEM-EDS is shown, where the analysis was carried out in the zone marked in Figure 1a  SnO 2 and~15 nm for CeO 2 . The similar thicknesses for TiO 2 and SnO 2 materials were a consequence of the reported modified-Stöber method [33,38] that favored the interaction between the precursor alkoxide with silica and the smallest thickness of the CeO 2 coating materials added with the solvothermal method.
Materials 2018, 11, x FOR PEER REVIEW 6 of 16 more than 500 particles. The results obtained indicated a thickness of an average of ~20 nm for TiO2, ~30 nm for SnO2 and ~15 nm for CeO2. The similar thicknesses for TiO2 and SnO2 materials were a consequence of the reported modified-Stöber method [33,38] that favored the interaction between the precursor alkoxide with silica and the smallest thickness of the CeO2 coating materials added with the solvothermal method. The HR-TEM micrographs of the Fe2O3-SiO2-MeO2-Pt (Me = Ti, Sn, Ce) catalysts are shown in Figure 2. All of the catalysts showed narrow particle size distributions, and their diameter increased in the order of Fe2O3-SiO2-CeO2-Pt < Fe2O3-SiO2-SnO2-Pt < Fe2O3-SiO2-TiO2-Pt. This effect is mainly  attributable to the nature of the metal oxide during the Pt deposition. In the catalyst syntheses, the Fe2O3-SiO2-MeO2 (Me = Ti, Sn, Ce) were contacted with the metal precursor (K2PtCl6) dissolved in aqueous solution to deposit the active phase on the shells (SiO2, TiO2, SnO2 or CeO2 modified with APTMS). The core-shell materials can accumulate PtCl6 2− by a ligand-exchange process from the -NH2 surface group and/or adsorb the metal precursor on the surface of the metal oxide by electrostatic interaction. In Table 1, the zeta potential (ZP) measurements for the Fe2O3-SiO2-MeO2 (Me = Ti, Sn, Ce) aqueous dispersion are displayed. For all of the solids, ZP is a positive value, ascribable mainly to the presence of protonated amine groups from APTMS on the core-shell surface (-NH3 + ), because the pKa of APTMS is 10.6 [39].    [40], IEP SnO2 = 6.4-7.3 [41] and IEP CeO2 = 6.8-8.3 [40]. The IEP is a reference for predicting the charge-dependent behavior of oxide minerals and their suspensions, in which IEP is a zero-point charge arising from the interaction of H + and OH − with the solid and water [42]. In our case, the ZP at the pH of impregnation is positive, which enhances the metal precursors' electrical attraction from the dissolution to the stationary layer of fluid attached to the dispersed particle. During the reduction process with NaBH 4 , the nucleation of PtCl 6 2− to produce Pt NPs was enhanced in Fe 2 O 3 -SiO 2 -CeO 2 , mainly by the positive surface charge of -CeO 2 at pH = 6.5. In the case of Fe 2 O 3 -SiO 2 -SnO 2 , the IEP of -SnO 2 was close to the pH of the materials' dispersion (see Table 1) which is in line with an increase in the Pt NPs after the reduction process in comparison with the -CeO 2 -based core-shell. For the Fe 2 O 3 -SiO 2 -TiO 2 , the IEP was lower than for the other solids, which provided a negative surface charge on -TiO 2 , which promoted an increase of Pt clusters during the reduction of Pt 4+ to metallic Pt. Therefore, the large and homogeneous Pt metallic particle sizes can be attributed to the successive impregnation-reduction deposition. For comparison, we included HR-TEM micrograph for the Fe 2 O 3 -SiO 2 -TiO 2 -Pt, at 5 wt % Pt, prepared by the metal deposition in one step and a sequence of Pt depositions in multi-cycles for the Fe 2 O 3 -SiO 2 -TiO 2 -Pt catalyst (see Materials and methods section). The Fe 2 O 3 -SiO 2 -TiO 2 -Pt prepared in one step showed a mean Pt crystal size of 13.4 nm with a wide particle size distribution, which confirmed the irregular nucleation when this metal was loaded on the TiO 2 shell surface during the Pt deposition. Moreover, the multi-step metal precursor addition provided for, in the first cycle, the nucleation of smaller Pt NPs, which grew uniformly in every cycle as shown in the Supplementary Material ( Figure S3).

ICP
The elemental composition of the Fe 2 O 3 -SiO 2 -MeO 2 -Pt (Me = Ti, Sn, Ce) catalysts was determined by ICP spectroscopy and the results are in Table 1. A similar Fe content can be seen with almost no differences related to the metal oxide, which indicated a thin coverage of the double shell, as determined by EDX-Scan analysis. The Pt content approaches the nominal amount of 5 wt %, which was indicative of successful consecutive Pt impregnation-reduction cycles.

Specific Area
The adsorption-desorption nitrogen isotherms of the materials presented a type II IUPAC classification ( Figure S4), which was indicative of non-porous structures. This result implied a non-textural modification during the sol-gel coating process not observed in the first shell of silica (see Table 1). It should be pointed out that, even though the materials had low surface areas, large platinum dispersions were observed in 5 wt % of Pt content catalysts, which supported the successful functionalization with APTMS and further Pt surface deposition impregnation-reduction cycles.

XRD
The XRD profiles of the Pt and non-Pt content structures are shown in Figure 3. Diffraction peaks can be seen, which indicated only the presence of maghemite γ-Fe 2 O 3 (JCPDS 39-1356) as crystalline phases with an absence of Fe 3 O 4 (JCPDS 65-3107) completely oxidized to Fe(III) ( Figure S5). For the Fe 2 O 3 -SiO 2 -MeO 2 (Me = Ti, Sn, Ce) structures, it can be seen that the coating process with the double shell of the metal oxide almost did not change the chemical composition of the crystalline Fe 2 O 3 cores. There was only the apparition of segregated phases for the SnO 2 structures, and the lack of detection of crystalline structures for TiO 2 and CeO 2 indicated that they were present with a high degree of dispersion (the crystal size of the oxides was lower than 5.0 nm). It has been reported for -TiO 2 [4] that the nanocrystal size is controlled by the thickness of the TiO 2 coating process. On the other hand, for -SnO 2 , the appearance of segregated phases of SnO 2 rutile with a tetragonal crystal system [43] has also been reported. Therefore, the appearance of only SnO 2 diffraction peaks is in agreement with the HR-TEM results, regarding a more heterogeneous second coating process attributable to the behavior of the tin precursor to form polycrystalline SnO 2 nanoparticles [44].
There was only the apparition of segregated phases for the SnO2 structures, and the lack of detection of crystalline structures for TiO2 and CeO2 indicated that they were present with a high degree of dispersion (the crystal size of the oxides was lower than 5.0 nm). It has been reported for -TiO2 [4] that the nanocrystal size is controlled by the thickness of the TiO2 coating process. On the other hand, for -SnO2, the appearance of segregated phases of SnO2 rutile with a tetragonal crystal system [43] has also been reported. Therefore, the appearance of only SnO2 diffraction peaks is in agreement with the HR-TEM results, regarding a more heterogeneous second coating process attributable to the behavior of the tin precursor to form polycrystalline SnO2 nanoparticles [44]. With regard to the Pt-content catalysts, Fe2O3-SiO2-MeO2-Pt (Me = Ti, Sn, Ce) showed the same XRD profile as that of the pristine core-shell materials. This confirmed that the NaBH4 used to perform the reduction of the metal precursor had no effect on the structure of the Fe2O3-SiO2-MeO2 (Me = Ti, Sn, Ce). However, the diffraction line at 39.6°, which corresponded to the metallic Pt(111) with a face-centred cubic structure, was too weak to be used to determine particle size in the case of Fe2O3-SiO2-MeO2-Pt (Me = Ti, Sn, Ce). Indeed, it is well recognized that the XRD technique is limited by the size of the particles. In this case, the size of the particle is less than 5 nm, which was in agreement with the HR-TEM characterization (see Table 1).

TPR-H2
The TPR-H2 profiles shown in Figure 4, corresponded to values normalized to per gram Fe to compare the intensity of the TCD signals independent of the more easily reducible FeOx species and large iron content. In order to better see the reduction profiles of the Fe2O3-SiO2-MeO2-Pt (Me = Ti, Sn, Ce) catalysts in Figure 4, the profiles of Fe2O3-NPs is also shown. The additional experiment to produce Fe2O3-NPs was carried out on non-coated Fe3O4-NPs submitted to an oxidation treatment. The Fe2O3-NPs profiles indicated a first reduction step at 370 °C from hematite Fe2O3 to Fe3O4, followed by a broad reduction process started above 400 °C that can be deconvoluted into two peaks attributable to the Fe3O4→FeO and FeO→Fe steps [45]. The similarity of the TPR profiles for the Fe2O3-SiO2-MeO2-Pt; (Me = Ti, Sn, Ce) catalysts to the Fe2O3-NPs profile was a consequence of the larger content of Fe2O3 compared to the TiO2, SnO2 and CeO2 oxides; the reduction process of TiO2, SnO2 and CeO2 was masked by the reduction process of Fe2O3. Unfortunately, the broadening of the peaks of Figure 4 (that could not be deconvoluted into different reduction steps) are likely unassignable to a particular reduction process. The absence of a reduction step at low temperature indicates that the deposition of the platinum nanoparticles reduced with NaBH4 at room temperature is the metallic state. To support this result, after the TPR of the -CeO2-Pt system, an oxidation process in air at 300 °C for 1 h was carried out before a second TPR profile ( Figure S6). The reduction step at 100 °C in the second TPR confirms the presence of metallic Pt in the -CeO2-Pt catalyst. With regard to the Pt-content catalysts, Fe 2 O 3 -SiO 2 -MeO 2 -Pt (Me = Ti, Sn, Ce) showed the same XRD profile as that of the pristine core-shell materials. This confirmed that the NaBH 4 used to perform the reduction of the metal precursor had no effect on the structure of the Fe 2 O 3 -SiO 2 -MeO 2 (Me = Ti, Sn, Ce). However, the diffraction line at 39.6 • , which corresponded to the metallic Pt(111) with a face-centred cubic structure, was too weak to be used to determine particle size in the case of Fe 2 O 3 -SiO 2 -MeO 2 -Pt (Me = Ti, Sn, Ce). Indeed, it is well recognized that the XRD technique is limited by the size of the particles. In this case, the size of the particle is less than 5 nm, which was in agreement with the HR-TEM characterization (see Table 1).

TPR-H 2
The TPR-H 2 profiles shown in Figure 4, corresponded to values normalized to per gram Fe to compare the intensity of the TCD signals independent of the more easily reducible FeO x species and large iron content. In order to better see the reduction profiles of the Fe 2 O 3 -SiO 2 -MeO 2 -Pt (Me = Ti, Sn, Ce) catalysts in Figure 4, the profiles of Fe 2 O 3 -NPs is also shown. The additional experiment to produce Fe 2 O 3 -NPs was carried out on non-coated Fe 3 O 4 -NPs submitted to an oxidation treatment. The Fe 2 O 3 -NPs profiles indicated a first reduction step at 370 • C from hematite Fe 2 O 3 to Fe 3 O 4 , followed by a broad reduction process started above 400 • C that can be deconvoluted into two peaks attributable to the Fe 3 O 4 →FeO and FeO→Fe steps [45]. The similarity of the TPR profiles for the Fe 2 O 3 -SiO 2 -MeO 2 -Pt; (Me = Ti, Sn, Ce) catalysts to the Fe 2 O 3 -NPs profile was a consequence of the larger content of Fe 2 O 3 compared to the TiO 2 , SnO 2 and CeO 2 oxides; the reduction process of TiO 2 , SnO 2 and CeO 2 was masked by the reduction process of Fe 2 O 3 . Unfortunately, the broadening of the peaks of Figure 4 (that could not be deconvoluted into different reduction steps) are likely unassignable to a particular reduction process. The absence of a reduction step at low temperature indicates that the deposition of the platinum nanoparticles reduced with NaBH 4 at room temperature is the metallic state. To support this result, after the TPR of the -CeO 2 -Pt system, an oxidation process in air at 300 • C for 1 h was carried out before a second TPR profile ( Figure S6). The reduction step at 100 • C in the second TPR confirms the presence of metallic Pt in the -CeO 2 -Pt catalyst.

Magnetic Measurements
The magnetic hysteresis loops at room temperature of the materials scaled to g −1 of material are shown in Figure 5. The Fe2O3-SiO2-MeO2-Pt (Me = Ti, Sn, Ce) structures display a ferromagnetic behavior with a magnetic saturation of ~40 emu·µg −1 . The remanence and coercivity properties of the hysteresis loop (inset of Figure 5) was attributable to the grain size of the core and limited for the silica coating process [2]. An increase in the coercivity of nanostructures has been reported due to an increase in both magnetocrystalline and shape anisotropy, which exert influence on the nanostructures' magnetic properties [46]. Due to the metal oxide content, structures did not show a change in the shape of the hysteresis loop, which confirmed that the magnetic properties of the material were confined to the core structure. The observed remanence and coercivity values allowed for the easy separation of the prepared catalysts from the reaction medium using external magnetic fields [32]. This phenomenon is appreciated because it is possible to separate the solid from the middle of the reaction in 30 s after putting it in contact with a magnet ( Figure 5).

Magnetic Measurements
The magnetic hysteresis loops at room temperature of the materials scaled to g −1 of material are shown in Figure 5. The Fe 2 O 3 -SiO 2 -MeO 2 -Pt (Me = Ti, Sn, Ce) structures display a ferromagnetic behavior with a magnetic saturation of~40 emu·µg −1 . The remanence and coercivity properties of the hysteresis loop (inset of Figure 5) was attributable to the grain size of the core and limited for the silica coating process [2]. An increase in the coercivity of nanostructures has been reported due to an increase in both magnetocrystalline and shape anisotropy, which exert influence on the nanostructures' magnetic properties [46]. Due to the metal oxide content, structures did not show a change in the shape of the hysteresis loop, which confirmed that the magnetic properties of the material were confined to the core structure. The observed remanence and coercivity values allowed for the easy separation of the prepared catalysts from the reaction medium using external magnetic fields [32]. This phenomenon is appreciated because it is possible to separate the solid from the middle of the reaction in 30 s after putting it in contact with a magnet ( Figure 5).

Magnetic Measurements
The magnetic hysteresis loops at room temperature of the materials scaled to g −1 of material are shown in Figure 5. The Fe2O3-SiO2-MeO2-Pt (Me = Ti, Sn, Ce) structures display a ferromagnetic behavior with a magnetic saturation of ~40 emu·µg −1 . The remanence and coercivity properties of the hysteresis loop (inset of Figure 5) was attributable to the grain size of the core and limited for the silica coating process [2]. An increase in the coercivity of nanostructures has been reported due to an increase in both magnetocrystalline and shape anisotropy, which exert influence on the nanostructures' magnetic properties [46]. Due to the metal oxide content, structures did not show a change in the shape of the hysteresis loop, which confirmed that the magnetic properties of the material were confined to the core structure. The observed remanence and coercivity values allowed for the easy separation of the prepared catalysts from the reaction medium using external magnetic fields [32]. This phenomenon is appreciated because it is possible to separate the solid from the middle of the reaction in 30 s after putting it in contact with a magnet ( Figure 5).

Catalytic Performance
The selective catalytic hydrogenation of CAL to COL is an ambitious challenge and aspects such as the nanometric size (1-10 nm) of the active sites, surface structure, metal-support interactions and electronic effects become more important in achieving it [47], and these factors should be considered for describing a catalyst's performance. To increase the selectivity towards the high-value-added COL, the nature of the shell as support for Pt NPs was studied as part of this work. The CAL conversion curves with reaction time for the Fe 2 O 3 -SiO 2 -MeO 2 -Pt (Me = Ti, Sn, Ce) catalysts are shown in Figure 6a, and to better illustrate the effect of the metal oxide type on the conversion levels, Figure 6b shows the successful fit of the experimental data with a pseudo 1st order reaction.

Catalytic Performance
The selective catalytic hydrogenation of CAL to COL is an ambitious challenge and aspects such as the nanometric size (1-10 nm) of the active sites, surface structure, metal-support interactions and electronic effects become more important in achieving it [47], and these factors should be considered for describing a catalyst's performance. To increase the selectivity towards the high-value-added COL, the nature of the shell as support for Pt NPs was studied as part of this work. The CAL conversion curves with reaction time for the Fe2O3-SiO2-MeO2-Pt (Me = Ti, Sn, Ce) catalysts are shown in Figure 6a, and to better illustrate the effect of the metal oxide type on the conversion levels, Figure 6b shows the successful fit of the experimental data with a pseudo 1st order reaction. The calculated pseudo-1st-order reaction rate constants, the initial reaction rates, maximum conversions at 500 min. and selectivity are shown in Table 2. More than 99% conversion and 67% COL production selectivity were obtained at 300 min. when Fe2O3−SiO2−CeO2-Pt was used as catalyst, while the Fe2O3−SiO2−TiO2-Pt system only reached 78% conversion and 65% selectivity for COL during the same time of reaction. In contrast, the conversion decreased to 40% and the selectivity value lowered to 50% for the Fe2O3−SiO2−SnO2-Pt system. For the selective hydrogenation of CAL, it is well-documented that the particle size of active metal species can exert vital impacts on their catalytic performance [47]. Generally, the decrease in the metal particle size leads to both the increased activity and the decreased COL selectivity. This is due to more active sites on small-sized metal particles that have less-coordinated surface sites and the higher adsorption strength of C=C bonds versus C=O bonds. In the other way, the use of -TiO2, -SnO2 and -CeO2 as shells was designed with the aim to increases the selectivity towards COL product, due to they have been reported as non-inert supports which could provide SMSI effect even at room temperature [23].
In the present work, although the sizes of Pt NPs were different (see Table 1), Fe2O3-SiO2-CeO2-Pt had the best catalytic performance. Due to the different sizes of Pt particles, there was a size The calculated pseudo-1st-order reaction rate constants, the initial reaction rates, maximum conversions at 500 min. and selectivity are shown in Table 2. More than 99% conversion and 67% COL production selectivity were obtained at 300 min. when Fe 2 O 3 −SiO 2 −CeO 2 -Pt was used as catalyst, while the Fe 2 O 3 −SiO 2 −TiO 2 -Pt system only reached 78% conversion and 65% selectivity for COL during the same time of reaction. In contrast, the conversion decreased to 40% and the selectivity value lowered to 50% for the Fe 2 O 3 −SiO 2 −SnO 2 -Pt system. For the selective hydrogenation of CAL, it is well-documented that the particle size of active metal species can exert vital impacts on their catalytic performance [47]. Generally, the decrease in the metal particle size leads to both the increased activity and the decreased COL selectivity. This is due to more active sites on small-sized metal particles that have less-coordinated surface sites and the higher adsorption strength of C=C bonds versus C=O bonds. In the other way, the use of -TiO 2 , -SnO 2 and -CeO 2 as shells was designed with the aim to increases the selectivity towards COL product, due to they have been reported as non-inert supports which could provide SMSI effect even at room temperature [23].
In the present work, although the sizes of Pt NPs were different (see Table 1), Fe 2 O 3 -SiO 2 -CeO 2 -Pt had the best catalytic performance. Due to the different sizes of Pt particles, there was a size dependence of hydrogenation activity among these catalysts. However, the pseudo-1st-order reaction rate constant follows the trend k -CeO2 > k -TiO2 >> k -SnO2 , which is different from the trend for the Pt mean particle size, where -CeO 2 -Pt > -SnO 2 -Pt > -TiO 2 -Pt. This behavior can be explained by the nature of the -SnO 2 shell in the Fe 2 O 3 -SiO 2 -SnO 2 -Pt. The -SnO 2 shell possess an irregular surface with evident formations of SnO 2 clusters with particle sizes larger than the Pt NPs, as was demonstrated by HR-TEM and XRD characterization. Despite the production of Pt NPs with a mean size of 3.5 nm (see Table 1), the crystals were almost randomly dispersed in the -SnO 2 interstices as hetero-aggregates, which could block the active sites, thereby decreasing the conversion levels. In relation to the results for k -CeO2 > k -TiO2 , the ratio of the pseudo-1 st -order reaction rate constant was k -CeO2 :k -TiO2 = 2.2:1, which is in agreement with the differences in the Pt particle sizes, where their size ratio in -TiO 2 -Pt:-CeO 2 -Pt was 1.9:1.
The selectivity was the most interesting result; Fe 2 O 3 -SiO 2 -TiO 2 -Pt catalysts showed a high selectivity to the desired product during the entire reaction course. For a better understanding, Figure 7 shows the product distribution as a function of time for all of the catalysts until 500 min. of reaction. The decrease of the catalyst selectivity in the Fe 2 O 3 -SiO 2 -CeO 2 -Pt system is in line with work reported by Wei et al in the selective hydrogenation of CAL over a Pt/CeO 2 catalyst [19]. The Fe 2 O 3 -SiO 2 -CeO 2 -Pt catalyst displayed a sharp decrease in the CAL concentration in the initial reaction period. The CAL concentration reached near total consumption after 240 min. Meanwhile, the COL concentration increased gradually in the total reaction period of 120 min. However, the production of HCOL also increased rapidly after 120 min. of reaction, reflecting a thermodynamically favorable hydrogenation of the C=C bond. Despite of the presence of CeO 2 which could promote the SMSI effect at the reaction temperature, as was observed at lower conversion level, the low mean Pt crystal size enhance the over-hydrogenation of both COL and HCAL producing mainly HCOL at high conversion level.
In the selective hydrogenation of CAL, it has been reported that the adsorption of the C=O bond towards the formation of the unsaturated alcohol is favored on larger metallic particle sizes [47][48][49][50]. However, for the Fe 2 O 3 -SiO 2 -TiO 2 -Pt catalyst, a continuous increase in the COL formation can be seen with no further second hydrogenation step. We propose that the chemical properties of the TiO 2 loading shell could generate an active site favoring the adsorption through the C=O, which inhibited the coplanar adsorption of CAL by both Pt crystal size and SMSI effect. Moreover, in the Fe 2 O 3 -SiO 2 -TiO 2 -Pt catalyst, the total hydrogenation product (HCOL) results from the HCAL hydrogenation with non-consumption of COL at 180 min which are in agreement with the preferred C=O adsorption on the catalyst surface.
Finally, for the Fe 2 O 3 -SiO 2 -SnO 2 -Pt catalyst, the competitive adsorption of the substrate by the SnO 2 crystallites and the homogeneous and narrow Pt particle sizes produced HCAL and COL in almost the same proportion. Moreover, although CAL hydrogenation does not require acid sites, in the smallest Pt nanoparticles, the Lewis acid properties of SnO 2 influenced the reaction rate and the selectivity of the process by modifying the adsorptive properties of Pt because of surface electron density [50]. Therefore, the observed selectivity results can be attributed to surface modification and metal-support interactions as well as a decrease in the Pt particle size [47].
CAL concentration reached near total consumption after 240 min. Meanwhile, the COL concentration increased gradually in the total reaction period of 120 min. However, the production of HCOL also increased rapidly after 120 min. of reaction, reflecting a thermodynamically favorable hydrogenation of the C=C bond. Despite of the presence of CeO2 which could promote the SMSI effect at the reaction temperature, as was observed at lower conversion level, the low mean Pt crystal size enhance the over-hydrogenation of both COL and HCAL producing mainly HCOL at high conversion level.

Conclusions
Fe 2 O 3 -SiO 2 -MeO 2 -Pt (Me = Ti, Sn, Ce) magnetic core-shell catalysts were successfully prepared. The use of -SiO 2 shells provided a protected magnetic core, and its structure ensured the deposition of a second metal oxide shell, which preserved the shape of the Fe 2 O 3 -SiO 2 structure. The Pt metal nanoparticles deposited in continuous cycles provided support to the catalyst with a homogeneous metal dispersion on the -MeO 2 surface (Me = Ce, Sn, Ti). In addition, the nature of the metal oxide affected the Pt mean particle diameters in the order of -TiO 2 -Pt > -SnO 2 -Pt > -CeO 2 -Pt. This trend was attributed to the IEP of the solids, which affected the interaction of metal precursors with the surfaces of the metal oxide shells. Consequently, the catalytic activity showed the trend of -CeO 2 -Pt > -TiO 2 -Pt > -SnO 2 -Pt, which is line with the physicochemical properties of the core-shell catalysts. In the case of -CeO 2 -Pt, an excellent activity and a continuous decrease in the production of COL was attributed to the Pt mean particle size (2.6 nm), which enhanced the consumption of both CAL and COL to produce of HCOL at a higher CAL conversion. In the opposite direction, -SnO 2 -Pt showed a minor activity in comparison with the other catalysts, mainly attributed to the irregular SnO 2 shell surface, which allowed for the formation of Pt nanoparticles inside the SnO 2 crystals, thereby blocking the active phase, which decreased the catalytic performance. The large stability towards the production of the unsaturated cinnamic alcohol of the -TiO 2 -Pt catalyst is attributed to particle size (4.9 nm) and the chemical properties of the TiO 2 loading shell favoring the adsorption of CAL by the C=O bond, inhibiting the coplanar adsorption of CAL to produce HCAL and/or HCOL. This study suggests both Pt nanoparticles size and SMSI effects have resulted beneficial in selective hydrogenation of CAL to produce COL as the main product.