Improved dispersion of transition metals in mesoporous materials through a polymer-assisted melt infiltration method

C. Ciotonea abd, B. Dragoi a, A. Ungureanu *a, C. Catrinescu a, S. Petit b, H. Alamdari c, E. Marceau d, E. Dumitriu a and S. Royer *bcd
a“Gheorghe Asachi” Technical University of Iasi, Faculty of Chemical Engineering and Environmental Protection, 73 D. Mangeron Bvd, 700050 Iasi, Romania. E-mail: aungureanu@tuiasi.ro
bUniversité de Poitiers, CNRS UMR 7285, Institut de Chimie des Milieux et Matériaux de Poitiers (IC2MP), Bâtiment B35, 6 Rue Michel Brunet - TSA 51106, 86073 Poitiers Cedex 9, France
cDepartment of Mining, Metallurgical and Materials Engineering, University Laval, Québec, Québec G1V 0A6, Canada
dUniv. Lille, CNRS, ENSCL, Centrale Lille, Univ. Artois, UMR 8181 – UCCS – Unité de Catalyse et de Chimie du Solide, F-59000 Lille, France. E-mail: sebastien.royer@univ-lille1.fr

Received 15th May 2017 , Accepted 13th September 2017

First published on 14th September 2017


Melt infiltration (MI) has been described as an effective way to disperse transition metals (TM) in ordered mesoporous supports. The alternative method described here is based on the infiltration of molten precursors into the support pores in the presence of a support polymer porogen. The resulting materials contain metal oxide particles (metallic particles after reduction) smaller than 2 nm, dispersed within the support micropores, which prove to be stable upon reduction up to 900 °C and to exhibit improved catalytic performances for the hydrogenation of cinnamaldehyde due to the high fraction of the active metal exposed.


Introduction

Mesoporous materials such as silica SBA-15, synthesized for the first time by Zhao et al.,1 are of great interest for catalytic applications due to their high surface area, tunable pore size, adjustable framework compositions and surface properties.1–3 Noble or transition metals in an oxidic or metallic form can be dispersed inside the pores of mesoporous materials to increase catalytic efficiency via an increase of the number of active sites.4,5 Impregnation is the most employed technique to achieve this goal. It is based on the filling of the pores with an aqueous solution containing metal precursors such as nitrates, chlorides or acetates, followed by a drying step and thermal treatments. Calcination results in the decomposition of the deposited precursors and in the stabilization of oxide nanoparticles over the support surface.6,7 This is a simple, scalable, low-cost process, but with some disadvantages. The metal loading is limited by the precursor solubility and by the support pore volume, and agglomeration and subsequent sintering of the metal-containing phases may occur upon drying and thermal treatments.8

Melt infiltration (MI) has been recently described as another simple way to disperse metals while overcoming the drawbacks of impregnation. This method involves two simple steps: blending the support and the precursor in their solid form, and heating the resulting mixture at low temperature.9 During the heating step, the precursor melts and migrates into the pores of the support. As temperature increases, the infiltrated precursor decomposes, leaving dispersed oxide particles inside the pores. Accordingly, the MI technique has also been termed in the literature as “solvent-free”,10,11 “solid-state grinding”12,13 or “capillary infiltration”.14 The amount of precursor in the pores is much higher than that attained using an impregnation solution, allowing one to achieve higher loadings of the active element in the catalyst. In addition, the drying step, which is necessary to eliminate the solvent in a classical impregnation process, is irrelevant here. The diffusion of the precursor out of the pores upon solvent evaporation, and its associated agglomeration on the external surface of the support or in the pore's mouth, are thus avoided.15

A wide range of precursors can be used for MI. Hydrated metal nitrates are usually preferred due to their low cost, to their low melting point and decomposition temperature, and to the fact that they yield pure oxide phases after calcination. Furthermore, the water present in these compounds enhances the wetting of the support by the molten phase.9–11,16–19

Ordered mesoporous silica and alumina are widely used as support materials for MI due to their high surface area and to the composition of their surface, which is rich with hydroxyl groups, that ensures the wettability of the support.19 SBA-15 supports are usually prepared using a copolymer surfactant as a structure directing agent (SDA), i.e. P123 (PEO20–PPO70–PEO20). The SDA is later eliminated by calcination to produce a solid with an open mesoporous structure.1 The calcined support is generally used for MI, but Jiang et al. have performed infiltration before removing the SDA, leading to improvements in the dispersion and morphology of the supported nanoparticles.18 Temperature, humidity and pressure have also been reported to influence the degree of dispersion of the active element into the pores.20

Although SBA-15 exhibits both meso- and microporosity, most reported results show that the supported elements introduced by conventional MI are primarily located in mesopores.12,17,18 The driving forces for infiltration are related to the interfacial tensions, acting at the solid/liquid/gas interfaces. But kinetically speaking, reaching the equilibrium state depends on how fast mass transport occurs within pores of different sizes: the infiltration of the molten phase is easier and faster into large mesopores than into micropores. As a consequence, infiltration time should not be overlooked as an important parameter to control.21

The aim of the present work is to investigate the parameters on which the infiltration process of Ni-based molten phases in pores of SBA-15 is based, for a higher dispersion and better control of the active element localization into the micropores of the support. With this objective in mind, a series of experiments were designed in order to evaluate the conditions under which the meso- and micropores of SBA-15 can be effectively and selectively infiltrated, and to assess if the dispersed active metal can be stabilized into the micropores while maintaining or improving the catalyst performances.

Experimental details

Material synthesis

A SBA-15 support was synthesized according to the procedure proposed by Zhao et al.1 4 g of Pluronic P123 (poly(ethylene oxide)-block-poly(propylene oxide)-block-poly(ethylene oxide)-block, PEO20–PPO70–PEO20, with MW = 5800 g mol−1) was dissolved in a 1.6 M solution of HCl at 40 °C. 8.5 g of a silica precursor, tetraethyl orthosilicate (TEOS), was added dropwise to the solution, followed by magnetic stirring for 24 h. The resulting gel was subjected to hydrothermal treatment for 48 h at 100 °C. After recovering by filtration, washing, and drying, the material was calcined at 550 °C for 6 h in a muffle furnace, using a heating ramp of 1.5 °C min−1. A reference material, denoted C-RI, was prepared according to the incipient wetness impregnation route using an aqueous solution of Ni(NO3)2·6H2O, and characterized for the sake of comparison with MI-derived materials. The detailed procedure of material preparation can be found elsewhere.22

Ni(NO3)2·6H2O was also used as the precursor for MI. Following the method reported by Tian et al.,23 the solid precursor was mixed with SBA-15 powder, and the mixture was ground at room temperature. Both calcined and uncalcined supports were used for infiltration, the latter retaining the P123 polymer within the pores. After grinding, the mixture was heat-treated in a Teflon-lined autoclave at 57 °C for different periods of time, varying from 0.5 to 4 days. Once the heat-treatment was completed, the solid was calcined under air flow at 500 °C for 5 h, using a heating rate of 1.5 °C min−1.

Nomenclature. The samples prepared with the calcined support were identified as C-t, C referring to “calcined” and t to the heat-treatment time in days. The samples prepared with an uncalcined support were identified as UC-t, UC referring to “uncalcined” and t to the heat-treatment time.

For comparison purposes, reference MI samples were obtained by direct calcination of the mixture, without heat-treatment in the autoclave, as described in the literature.9 These samples were identified as C-0 and UC-0.

Material characterization

Ni-Containing materials were characterized in the calcined state using ICP-OES, nitrogen physisorption, TEM-EDX, XRD at low and high angles and TPR.

The chemical composition of the samples was determined by ICP analyses using a sequential scanning inductively coupled plasma optical emission spectrometer (Perkin Elmer Optima 2000 DV). A known amount of the sample was dissolved in a diluted HF–HCl solution and heated under microwave until complete dissolution. The results showed that the final Ni loading was in agreement with the expected 10 wt% value ± 0.3 wt%.

N2-physisorption isotherms were recorded at −196 °C on an Autosorb 1-MP automated gas sorption system (Quantachrome Instruments). Before analysis, the samples were outgassed under dynamic vacuum at 350 °C for 3 h. The textural properties were determined from the adsorption/desorption isotherms by using the Autosorb 1 software. The BET surface area was determined using the multipoint BET algorithm in the P/P0 range of 0.10–0.25. The t-plot method was applied to quantify the micropore volume and surface area (de Boer statistical thickness of 3.8–6.5 Å). The mesopore size distribution was determined using the NLDFT equilibrium algorithm for cylindrical pores.

The samples were analysed by transmission electron microscopy (TEM, JEOL 2100 UHR, operated at 200 kV with a LaB6 source and equipped with a Gatan Ultra scan camera). They were embedded in a polymeric resin (Spurr), before being sliced into 50 nm-thick sections using an ultramicrotome. The samples were deposited on a carbon grid. The pore structure and NiO particle distribution in the samples were then determined.

X-ray diffraction was performed using the powder samples, using a PANalytical Empyrean X-ray diffractometer in the Bragg–Brentano configuration, with CuKα radiation (λ = 1.54184 Å). For small-angle analysis, the data were collected in a 2θ range from 0.5 to 5° with a step size of 0.01°. Wide-angle analyses were performed using a 2θ range between 10 and 80° with a step size of 0.05°. The crystal mean size was calculated using the Scherrer equation: Dhkl = /b[thin space (1/6-em)]cos[thin space (1/6-em)]θ, where K is a structure constant (0.9 for spherical crystals); λ is the incident ray wavelength; b is the peak width at half height after correction for instrumental broadening; and θ is the Bragg angle.

Temperature-programmed reduction under H2 (H2-TPR) was conducted on an Autochem chemisorption analyser (Micromeritics) equipped with a TCD. Before H2-TPR run, the solid resulting from MI was activated up to 500 °C (heating ramp of 5 °C min−1, plateau of 1 h, under simulated air, total flow rate of 30 mL min−1). After cooling down to 50 °C, the H2-containing flow was stabilized (30 mL min−1, 3.0 vol% H2 in Ar) and the temperature-programmed reduction was performed (from 50 to 900 °C, with a temperature ramp of 5 °C min−1).

Catalytic performances in the high-pressure hydrogenation of cinnamaldehyde

SBA-15-supported NiO nanoparticles obtained after calcination of the materials were reduced to their metallic form at 500 °C for 10 h (heating ramp of 6 °C min−1) under H2 flow (1 L h−1). The reaction was performed in a Parr reactor under the following conditions: 1 mL cinnamaldehyde, 40 mL isopropanol, 0.265 g catalyst, 10 bar H2 and reaction temperature of 60 °C. Preliminary tests performed with different granulometric fractions, loadings of catalyst, and stirring rates, disclosed no diffusional limitations under the selected reaction conditions provided that a grain size lower than 0.126 mm was used. Verification of the absence of diffusional resistance during reaction was performed over the most active catalyst (UC-4). In addition, the experiment was conducted under N2 pressure (instead of H2). No hydrogenation product was detected after 6 h of reaction. Aliquots of the reaction mixture were withdrawn periodically and analysed by gas chromatography (HP 5890, equipped with a DB-5 capillary column and a FID detector). The identification of the reaction products was achieved from the retention times of pure compounds and from additional analysis by GC-MS (Agilent 6890 N, equipped with an Agilent 5973 MSD detector and a DB-5-ms column). For the recycling tests, the UC-2 catalyst was recovered from the end reaction mixture (after 60 min of reaction) by centrifugation, washed with fresh 2-propanol and ethanol, dried at room temperature overnight, calcined at 500 °C and reactivated under H2 at 500 °C before being tested under standard conditions (1 mL cinnamaldehyde, 40 mL isopropanol, 10 bar H2, reaction temperature of 60 °C). The catalytic performance of UC-2 for each run was expressed as CNA conversion after 5 min of reaction, relative to the conversion registered for the previous run at the same reaction time (i.e., relative activity, %).

Results and discussion

The TEM images of several C-t samples (infiltrated on calcined SBA-15 during time t and subsequently calcined at 500 °C) are shown in Fig. 1. In the C-0 solid prepared without a plateau at 57 °C, a large amount of the nickel precursor recrystallized out of the support pores and only a small fraction of precursor migrated into the support mesopores (Fig. 1a and b). The nickel dispersion in the material is inhomogeneous, with highly loaded zones (as presented in Fig. 1a and b), and large nickel free zones (not presented). The size of the NiO particles outside the pores was quite large, around 35 nm. In contrast, the small amount of infiltrated precursor led to ∼7–8 nm NiO nanoparticles (NPs) accommodated inside the mesopores of the support, at a size similar to the mesopore diameter. The average particle size for each population is listed in Table 1, and the histogram is presented in Fig. S1.
image file: c7cy00963a-f1.tif
Fig. 1 TEM images of the samples prepared over the calcined support, with an infiltration time of “t” days (C-t) after decomposition of the precursor; a and b) without a plateau at 57 °C (no infiltration time, sample C-0); c and d) t = 1 day of infiltration at 57 °C (sample C-1); e and f) t = 2 days of infiltration at 57 °C (sample C-2).
Table 1 Selected textural and structural properties of the synthesized materials: loss of surface area and pore volume compared to SBA-15, size of the NiO NPs
Sample Loss of SBETa/% Loss of Sμb/% Loss of Vporec/% Loss of Vμd/% D pore /nm NiO particle size/nm
XRD TEM
a Specific surface area, m2 gSBA−1. b Microporous surface area, m2 gSBA−1. c Total pore volume, cm3 gSBA−1. d Micropore volume, cm3 gSBA−1. e Mean pore size. f n.d.: not detectable. Characteristics of the SBA-15 materials (two batches have been used, denoted + and ‡): SBET = 820 m2 g−1, Sμ = 207/255+ m2 g−1, Vpore = 1.25 cm3 g−1 and Vμ = 0.10/0.12+ cm3 g−1.
C-0+ 0 −13 0 −25 8.1, 8.4 30.7 7.9, 35.0
C-1 −23 −65 −17 −69 6.0, 8.4 9.0 7.1, 29.4
C-2 −29 −68 −24 −72 6.0, 8.4 9.0 7.6
UC-0+ 0 −1 0 −8 8.4 25 5.7, 115.1
UC-0.5+ −24 −49 −19 −50 5.0, 8.4 15.3 Not determined
UC-1 −18 −77 −7 −82 8.1, 8.4 n.d.f 2.6, 8.7
UC-2 −25 −73 −14 −79 8.4 n.d.f 2.5
UC-4+ −30 −54 −24 −55 8.4 n.d.f 1.4
C-RI+ −25 −48 −21 −50 6.0, 8.2 9.0 8.0, 16.5


1 day heat-treatment at 57 °C resulted in better infiltration of the precursor. As shown in Fig. 1c and d, only a very small amount of large particles was observed out of the support pores, and nickel redispersion occurs throughout all silica grains. Increasing the heat-treatment time from 1 to 2 days (Fig. 1e and f) resulted in the full infiltration of the precursor into the mesopores. No large NiO crystals were observed at the external surface after calcination. The high magnification images confirm that NiO NPs were accommodated inside the mesopores, with a final particle size of ∼7–8 nm (Fig. 1f). The efficiency of the MI method is thus very sensitive to the infiltration time.

As the NP size is seen to be determined by the primary mesopore diameter, a way to further reduce their size would be to infiltrate the molten precursor inside the intra-wall pores (hereafter denoted micropores) connecting or not the primary mesopores.24–26 This could be achieved by preparing samples on uncalcined SBA-15 supports (UC-t series): the presence of the SDA P123 was supposed to reduce the available space inside the mesopores, forcing the nickel precursor to diffuse along the silica pore walls and migrate into the micropores. The TEM images of the UC-t samples after calcination are shown in Fig. 2. Again, the absence of a plateau at 57 °C resulted in very low penetration of the precursor into the mesopores (UC-0), with external particle formation (Fig. 2 and S1). As in the case of C-0, the UC-0 material presented (i) highly loaded nickel zones (large external particles of NiO), and (ii) large nickel free zones (pure SBA-15 silica). With heat-treatment of 1 day at 57 °C, the precursor fully infiltrated the support (Fig. 2c and d), with NiO NPs visible both in the pores (NP size >3 nm) and inside the silica walls (NP size <3 nm).


image file: c7cy00963a-f2.tif
Fig. 2 TEM images of the samples prepared over the uncalcined support, with an infiltration time of “t” days (UC-t), after decomposition of the precursor; a and b) melt impregnation on the uncalcined support without a plateau at 57 °C (sample UC-0); c and d) t = 1 day at 57 °C (sample UC-1).

Fig. 3 shows the high-magnification TEM images of the UC-t samples prepared with longer infiltration times. After two days of infiltration, very few NPs larger than 3 nm were observed. Instead, more NPs smaller than 3 nm were seen in the silica walls. This phenomenon is better observed after 4 days of infiltration, after which NiO NPs smaller than 2 nm are visible inside the walls, with no particles in the mesopores (Fig. 3d). The side view of the samples (Fig. S2) confirms that the transition metals are mostly located inside the walls.


image file: c7cy00963a-f3.tif
Fig. 3 TEM images of the samples prepared over the uncalcined support (UC-t) heat-treated at 57 °C for longer times and calcined at 500 °C: a and b) t = 2 days at 57 °C (sample UC-2); c) t = 4 days at 57 °C (sample UC-4); d) t = 4 days at 57 °C (sample UC-4), with 3 min exposure of the sample to an electron beam to determine the presence of transition metal phases.

To reveal the presence of NiO NPs at high magnification in the case of UC-4 (Fig. 3d), the sample is exposed to an electron beam for 3 min before acquiring the image. Under electron beam exposure, the transition metal phase is more easily observed, but the silica pore structure is significantly modified (increase in the wall width and decrease in the pore diameter, due to the partial collapse of the structure).

This visible evolution of decreasing particle size with heat-treatment time is unambiguously confirmed by the NiO size distribution assessed from the TEM images, showing the displacement of the mean particle size from a broad population (from 1 nm to >10 nm) for the UC-1 material (Fig. 4a) to a narrow distribution of particle size (<2 nm) for the UC-4 material (Fig. 4c). The improved MI process on the uncalcined support thus results not only in a much better distribution of the metal oxide throughout the silica grain, but also in highly dispersed metal oxide NPs, that is not attained on calcined SBA-15 for a similar time of heat treatment.


image file: c7cy00963a-f4.tif
Fig. 4 Evolution of the particle size distribution in the samples prepared over the uncalcined support, with an infiltration time of “t” days (UC-t): a) sample UC-1; b) sample UC-2; c) sample UC-4. All materials were calcined at 500 °C prior to TEM analysis.

X-ray diffraction analyses performed on the C-t and UC-t samples confirm this evolution. Fig. 5 shows the diffraction patterns of the C-t and UC-t samples, respectively. The X-ray diffraction pattern of the sample obtained by incipient wetness impregnation is also included for comparison (sample C-RI). In all the samples, the detected crystalline phase is NiO. The mean crystallite size of NiO particles in sample C-0 was calculated to be 30.7 nm (Table 1), as the population is dominated by large particles grown at the external surface of the silica grains. It was verified that a heat-treatment of 12 h resulted in roughly the same diffraction pattern (C-0.5 sample, not shown). Significant peak broadening is observed for 1 and 2 day heat-treatment (samples C-1 and C-2), with the mean crystallite size evaluated to be 9 nm (Table 1), a value in agreement with that assessed from TEM measurements (most or all particles are confined in the mesopores, ∼7–8 nm, Fig. 1f). A similar situation arose for the material prepared by incipient wetness impregnation (sample C-RI, Fig. 5A), for which the particle size also fits the pore diameter as also shown for cobalt based nanoparticles also prepared by IWI.22


image file: c7cy00963a-f5.tif
Fig. 5 Evolution of wide angle X-ray diffractograms with heat-treatment time: A) for the samples prepared over the calcined support, with an infiltration time of “t” days (C-t) and the reference sample prepared over the calcined support, using the IWI-MD method (C-RI) and B) for the sample prepared over the uncalcined support, with an infiltration time of “t” days (UC-t); bottom NiO JCPDS file no. 47-1049.

The effect of heat-treatment time on the evolution of the diffraction patterns of the UC samples is quite different, as shown in Fig. 5B. Again, the sample UC-0 contains NiO NPs with a mean crystallite size of about 25 nm, explained by the very large particles that crystallized outside the support. But on uncalcined supports, unlike C-0.5, the NiO peaks broaden after only 12 h of heat-treatment, suggesting that the precursor easily migrates into the pores and yields very small and well dispersed NiO NPs. In line with this evolution, the NiO main peak, still visible for C-1 and C-2, has almost disappeared after 4 days of heat-treatment. Due to the poor intensity of the visible diffraction peaks, it was not possible to calculate precisely the mean crystallite size, but this is in line with the very small size revealed by TEM analysis (<3 nm).

Table 1 summarizes the textural characteristics of SBA-15 and the C and UC samples, analysed in terms of loss of surface area and pore volume compared to the initial calcined support, while the N2 physisorption isotherms are presented in Fig. 6. They are all of type IV according to the IUPAC classification, and confirm that the mesostructure is maintained upon MI and calcination. These observations are quite similar to those observations for the C-RI material (Fig. 6 and Table 1). In line with the TEM results, except for C-0 and UC-0, the pore volumes and specific surface areas decrease as the heat-treatment time increases and more NPs are present in the SBA-15 pores. It is however difficult to draw general tendencies on the textural property evolutions with final localization and size of NPs. Some differences are observed in terms of loss of micropore surface area and volume, but they seem to be rather linked to the different micropore volumes initially present in the two SBA-15 materials used for MI. In addition, the calcination of the material, with the active phase precursor infiltrating into the intra-wall pores, can affect the final material textural properties, including B.E.T. surface area and pore volume. All of the materials show an open mesoporosity and high surface area after thermal stabilization, confirming that the infiltration process have a limited effect on the mesopore architecture.


image file: c7cy00963a-f6.tif
Fig. 6 N2 physisorption isotherms recorded for the MI-derived materials. A) for samples prepared over the calcined support, with an infiltration time of “t” days (C-t) and the reference sample prepared over calcined support, using the IWI-MD method (C-RI) and B) for the sample prepared over the uncalcined support, with an infiltration time of “t” days (UC-t).

Hysteresis loops present a delay in the closure of the desorption branch at P/P0 = 0.5–0.65 when NPs are located in the mesopores (C-1, C-2, RI, UC-0.5), evidencing partial blockage and resulting in two maxima in the pore size distribution (Table 1).15 Such maxima in the pore size distribution can originate from the formed particles not fully plugging the mesopores. Even if such particles are not observed by TEM, their formation cannot be excluded on the basis of the N2-physisorption results. This perturbation disappears for UC-2 and UC-4, confirming that an open mesoporosity is recovered following the migration of the precursor and formation of NPs in the micropores.

The duration of the heat-treatment step and the presence of the SDA thus account for the complete infiltration of the liquid precursor into the support pores, preventing the formation of large particles on the external surface. Although the influence of heat-treatment time can explain the efficient infiltration of the precursor into the support mesopores, the presence of the polymer during infiltration seems to be the key factor to explain the enhanced migration of the precursor into the micropores, since in the case of the C-t samples, even longer heat-treatment times did not result in micropore infiltration. The favoured migration of the precursor into the micropores is highlighted by the evolution of the cumulative secondary pore volume evolution with the infiltration time (Fig. S3). Compared to cumulative pore volume at low size recorded for calcined SBA-15 or UC-0, lower values are obtained for UC-4.

Some authors have attempted to connect this phenomenon to the physico-chemical properties of the polymeric surfactant and its interaction with the molten precursor and support during the MI process. The P123 SDA is composed of hydrophobic PPO central chains and two hydrophilic PEO end chains,27,28 the former being located in the mesopores of the support and the latter extending into the silica walls. When the molten precursor is in contact with the calcined support, i.e. in the absence of polymers (C-t samples), it easily migrates along the mesopores, but the surface tension is possibly too high to allow the precursor to completely wet the silica walls and infiltrate into the micropores. In the presence of the polymer and applying an intermediate step for precursor diffusion (UC-t samples), the extension of the hydrophilic PEO branches into the SBA-15 walls may favour the wetting of the walls, as shown by Park et al. using PEO-functionalized SBA-15,29 and provide an adequate chemical environment for the migration of the precursor into the intra-wall micropores. One may also hypothesize that because of the hydrophobic nature of the PPO core of the micelles located inside the main mesopores, the system tends to minimize the precursor/PPO interface and maximize the precursor/PEO interface, favouring the migration of the precursor along the silica walls up to the silanol-rich intra-wall pore surface.

The reducibility of the metallic phase was characterized by TPR (Fig. 7). Two main reduction peaks are observed in most profiles. They are associated, below 500 °C, to larger NiO NPs behaving like bulk NiO, and above 500 °C, to smaller NiO NPs or stabilized Ni2+ surface ions.30–33 Comparison with the TEM results shows that attributing the two peaks to external vs. intra-porosity NPs would be an oversimplification, as two peaks are visible in the TPR profile of C-2, though external NPs are scarce. In the pores, interactions between NiO NPs and the silica walls probably play a role in modifying their reactivity toward hydrogen.


image file: c7cy00963a-f7.tif
Fig. 7 H2 temperature programmed reduction profiles recorded for the MI-derived materials. A) for the samples prepared over the calcined support, with an infiltration time of “t” days (C-t) and the reference sample prepared over the calcined support, the using IWI-MD method (C-RI) and B) for the sample prepared over the uncalcined support, with an infiltration time of “t” days (UC-t).

Sample C-0 exhibits the main reduction peak at lower temperature, in line with a high number of external, larger NiO NPs. As heat-treatment time increases in the C-t series, the intensity of the second peak increases at the expense of the first one, basically because a larger number of smaller NiO crystals are embedded in the mesopores. The TPR profile of sample C-RI is very similar to those of C-1 and C-2, which demonstrates that in terms of NiO size distribution and localization, simply increasing the heat-treatment time brings limited benefits when compared with preparation by classical impregnation.

In contrast, the TPR profiles of the samples infiltrated in the presence of the SDA (UC-t samples) are noticeably different (Fig. 7B). The sample prepared on the uncalcined support without a plateau at 57 °C (UC-0) exhibits only one peak at 401 °C. Compared to C-0, this shows that in the presence of a polymer and in the absence of plateau, the Ni precursor cannot enter the pores at all and decomposes outside SBA-15, as would be the case with a salt that does not melt like nickel acetate.32 But in the UC series, a major change takes place after only 12 h of heat treatment. The intensity of the first peak decreases drastically with increasing heat-treatment time, while the reduction peak at 580 °C associated with smaller NiO NPs increases. In contrast with the C series, the first peak totally disappears when the heat-treatment time reaches 4 days. Based on the TEM analyses, in this series the second broad peak can be attributed to the reduction of the homogeneous population of small NiO particles present in the micropores and formed with the help of the SDA. The presence of a shoulder above 700 °C on some profiles may also indicate the existence of a small amount of nickel silicate.

The catalytic performance of selected C-t and UC-t samples was assessed in the liquid-phase hydrogenation of cinnamaldehyde (CNA) (reaction scheme shown in Scheme S1). Prior to reaction, the catalysts were activated under pure H2 flow for 10 h at 500 °C to ensure reduction of NiO. A selectivity in hydrocinnamaldehyde, HCNA, of 90% was measured between 10 and 90% conversion for all tested materials, in line with the efficiency of nickel catalysts for C[double bond, length as m-dash]C bond hydrogenation (Fig. S4).5,34 These similarities led us to use the reaction to probe the nickel active surface area, by comparing the evolution of CNA conversion with reaction time for the different catalysts.

The evolutions of the conversion with reaction time, obtained for C-0, C-2 and C-RI catalysts, are presented in Fig. 8A. The reference C-0 sample, which exhibits large NiO particles outside the pores, presents the lowest activity, reaching only 78% CNA conversion after 360 min of reaction, with a CNA hydrogenation rate of 6.9 mmolCNA conv. gcata.−1 h−1, as calculated based on the initial slope of the conversion curve (Table 2). For the two close materials C-RI and C-2 (homogeneous particle size of ∼8 nm, NPs mostly in the mesopores of the support, similar TPR profiles), an almost complete CNA conversion is achieved in 360 min, and close CNA hydrogenation rates of 14.4–18.5 mmolCNA conv. gcata.−1 h−1 are measured (Table 2).


image file: c7cy00963a-f8.tif
Fig. 8 Evaluation of the catalytic properties of selected MI-derived materials for the cinnamaldehyde hydrogenation: A) for the samples prepared over the calcined support, with an infiltration time of “t” days (C-t) and the reference sample prepared over the calcined support, using the IWI-MD method (C-RI) and B) for the sample prepared over the uncalcined support, with an infiltration time of “t” days (UC-t).
Table 2 CNA hydrogenation rate and fraction of Ni atoms exposed at the NP surface
Sample Hydrogenation ratea/mmolCNA conv. gcata.−1 h−1 % atoms exposed on the NP surfaceb
a Calculated based on the initial slope of the conversion curve, at XCNA < 40%. b % of atoms exposed on the surface, calculated using the particle sizes determined by TEM statistical analysis, considering that dNi = 0.84 × dNiO based on the differences of molar mass and density, and that for each particle, Ni dispersion (%) = 97.1/dNi (nm).34
C-0 6.9 11.8
C-2 18.5 15.5
C-RI 14.4 11.3
UC-0 37.7 9.5
UC-1 63.3 24.9
UC-2 112.5 47.3
UC-4 131.1 84.5


The catalytic properties of the UC-t samples are shown in Fig. 8B. A CNA conversion of 97% is achieved after 360 min over UC-0. But for samples UC-1 and UC-2, full CNA conversion is obtained in less than 60 min (Fig. 8B), and after 40 min only for the most active sample, UC-4. Table 2 shows that with the increase in infiltration time, which results in an improvement in nickel dispersion, the CNA hydrogenation rate also increases, up to 131.1 mmolCNA conv. gcata.−1 min−1 for UC-4, which is 2 times higher than for UC-1, and ∼9 times higher than for C-2 and C-RI (solids with mesopore-confined Ni NPs). The catalytic activity demonstrated by the UC-t samples is outstanding, considering the mild reaction conditions, i.e. temperature of 60 °C and pressure of 10 bar.

The hydrogenation rates are plotted in Fig. S5 as a function of the fraction of nickel atoms exposed at the NP surface deduced from TEM measurements (Table 2). Due to the specific localization of NPs (confined in mesopores or micropores), all atoms of the NPs are not expected to be available for reaction, some being in close contact with the silica surface. For the UC-t series, the calculated rate is roughly linked to the fraction of nickel atoms exposed, the other catalysts displaying similar nickel dispersion (arising mostly from the fraction of smaller particles, as large particles outside the pores contribute to a very low extent to the metallic surface area) and low catalytic activity. This reflects the positive effect of the infiltration time on the Ni NP dispersion and surface active site accessibility. It also shows that CNA can access the surface active sites without encountering significant internal diffusional limitations. Consequently, confining the NPs in micropores both ensures a high number of accessible active sites due to a high dispersion of the metal NPs and maintains the mesopores fully open, for an effective diffusion of the reactants.

Catalyst stability was evaluated during 3 successive runs with the selected Ni/SBA-15 catalyst (UC-2 sample, Fig. S6). The results showed that the catalytic performances are not essentially changed over three catalytic runs (a conversion decrease of <5% was registered between the runs), confirming the good stability and reusability of the Ni-based catalysts prepared by our MI approach.

It has been finally confirmed by TEM analysis performed using UC-4 that Ni particles do not sinter during reduction and remain exclusively in the micropores, even after reduction under H2 at 900 °C. Indeed, Fig. 9 shows the presence of very small metallic particles (<3 nm), with only a very few large metallic particles in the mesopores. This stresses the importance of micropore confinement on the stabilization of the NPs, as it has been formerly demonstrated that Ni0 particles smaller than 3 nm located in the mesopores would sinter up to a size close to the support pore diameter during reduction.35


image file: c7cy00963a-f9.tif
Fig. 9 TEM images recorded over the samples prepared over the uncalcined support, with an infiltration time of 4 days (UC-4), after reduction at 900 °C.

Conclusion

Two ways to enhance the dispersion of an active metal in a SBA-15 support by melt infiltration have been explored: the optimization of the diffusion heat-treatment at a temperature close to the melting point of the precursor, and the presence of the SBA-15 structure directing agent during melt infiltration. While the diffusion of the molten precursor into the support mesopores is facilitated by a longer heat-treatment, preventing the crystallization of the precursor outside the support, a spectacular improvement in dispersion is achieved when the infiltration is performed on the uncalcined support containing the P123 copolymer, allowing the infiltration of the precursor into the micropores of the support within a few days. This enhancement is attributed to the interactions with the hydrophilic branches of the polymer extending into the walls of the support. A heat-treatment of 4 days in the presence of the structure directing agent results in a final NiO particle size of 1–2 nm, a size that is retained even after reduction at a temperature as high as 900 °C. The catalytic activity of Ni/SBA-15 materials in the liquid-phase hydrogenation reaction of cinnamaldehyde is shown to significantly increase when the very small nickel particles are located in the silica intra-wall micropores, both because the metal phase dispersion is high and because the diffusion of the reactant is facilitated in the mesopores of the support that are not plugged by confined nickel particles. Furthermore, the catalysts were reusable, with no essential changes in catalytic performance after three catalytic runs.

Conflicts of interest

There are no conflicts of interest to declare.

Acknowledgements

This work was partially supported by two grants of the Romanian National Authority for Scientific Research, CNCS-UEFISCDI (project numbers PN-II-RU-TE-2012-3-0403 and PN-II-ID-PCE-2011-3-0868). C Ciotonea (784016L) acknowledges the Ministère des Affaires étrangères et du Développement International for Eiffel Excellence fellowships. C. Ciotonea also thanks the Fondation de l'Université de Poitiers for financial support. S. Royer, E. Marceau and C. Ciotonea acknowledge the Chevreul Institute (FR 2638), Ministère de l'Enseignement Supérieur et de la Recherche, Région Nord – Pas de Calais and FEDER for funding.

References

  1. D. Zhao, J. Feng, N. Huo, G. Melosh, H. Fredrickson, B. F. Chmelka and G. D. Stucky, Science, 1998, 279, 548 CrossRef CAS PubMed.
  2. U. Ciesla and F. Schuth, Microporous Mesoporous Mater., 1999, 27, 131 CrossRef CAS.
  3. A. Taguchi and F. Schuth, Microporous Mesoporous Mater., 2005, 77, 1 CrossRef CAS.
  4. P. Mäki-Arvela, L. P. Tiainen, M. K. Lindblad, D. N. Kumar, R. Sjöholm, T. Ollonqvist, J. Väyrynen, T. Salmi and D. Y. Murzin, Appl. Catal., A, 2003, 241, 271 CrossRef.
  5. B. M. Reddy, G. M. Kumar, I. Ganesh and A. Khan, J. Mol. Catal. A: Chem., 2006, 247, 80 CrossRef CAS.
  6. F. Schuth, A. Wingen and J. Sauer, Microporous Mesoporous Mater., 2001, 44–45, 465 CrossRef CAS.
  7. J. R. A. Sietsma, J. D. Meeldijk, M. Versluijs-Helder, A. Broersma, A. J. van Dillen, P. E. de Jongh and K. P. de Jong, Chem. Mater., 2008, 20, 2921 CrossRef CAS.
  8. H. Friedrich, J. R. A. Sietsma, P. E. de Jongh, A. J. Verkleij and K. P. de Jong, J. Am. Chem. Soc., 2007, 129, 10249 CrossRef CAS PubMed.
  9. Y. M. Wang, Z. Y. Wu and J. H. Zhu, J. Solid State Chem., 2004, 177, 3815 CrossRef CAS.
  10. C. F. Zhou, Y. M. Wang, Y. Cao, T. T. Zhuang, W. Huang, Y. A. Chun and J. H. Zhu, J. Mater. Chem., 2006, 16, 1520 RSC.
  11. J. K. Shon, S. S. Kong, Y. S. Kim, J. H. Lee, W. K. Park, S. C. Park and J. M. Kim, Microporous Mesoporous Mater., 2009, 120, 441 CrossRef CAS.
  12. Y. M. Wang, Z. Y. Wu, H. J. Wang and J. H. Zhu, Adv. Funct. Mater., 2006, 16, 2374 CrossRef CAS.
  13. M. B. Zheng, J. Cao, S. T. Liao, J. S. Liu, H. Q. Chen, Y. Zhao, W. J. Dai, G. B. Ji, J. M. Cao and J. Tao, J. Phys. Chem. C, 2009, 113, 3887 CAS.
  14. K. Jian, T. C. Truong, W. P. Hoffman and R. H. Hurt, Microporous Mesoporous Mater., 2008, 108(1–3), 143 CrossRef CAS PubMed.
  15. J. R. A. Sietsma, H. Friedrich, A. Broersma, M. Versluijs-Helder, A. J. van Dillen, P. E. de Jongh and K. P. de Jong, J. Catal., 2008, 260, 227 CrossRef CAS.
  16. A. Boulaoued, I. Fechete, B. Donnio, M. Bernard, P. Turek and F. Garin, Microporous Mesoporous Mater., 2012, 155, 131 CrossRef CAS.
  17. W. Yue and W. Zhou, Chem. Mater., 2007, 19, 2359 CrossRef CAS.
  18. Q. Jiang, Z. Y. Wu, Y. M. Wang, Y. Cao, C. F. Zhou and J. H. Zhu, J. Mater. Chem., 2006, 16, 1536 RSC.
  19. T. M. Eggenhuisen, J. P. den Breejen, D. Verdoes, P. E. de Jongh and K. P. de Jong, J. Am. Chem. Soc., 2010, 132, 18318 CrossRef CAS PubMed.
  20. P. E. de Jongh and T. M. Eggenhuisen, Adv. Mater., 2013, 25, 6672 CrossRef CAS PubMed.
  21. C. Ciotonea, I. Mazilu, B. Dragoi, C. Catrinescu, E. Dumitriu, A. Ungureanu, H. Alamdari, S. Petit and S. Royer, ChemNanoMat, 2017, 3, 233 CrossRef CAS.
  22. C. Ciotonea, B. Dragoi, A. Ungureanu, A. Chirieac, S. Petit, S. Royer and E. Dumitriu, Chem. Commun., 2013, 49, 7665 RSC.
  23. W. H. Tian, L. B. Sun, X. L. Song, X. Q. Liu and Y. Yon, Langmuir, 2010, 26, 17398 CrossRef CAS PubMed.
  24. A. Galarneau, H. Cambon, F. Di Renzo, R. Ryoo, M. Choi and F. Fajula, New J. Chem., 2003, 27, 73 RSC.
  25. C. J. Rasmussen, A. Vishnyakov, M. Thommes, B. M. Smarsly, F. Kleitz and A. V. Neimark, Langmuir, 2010, 26, 10147 CrossRef CAS PubMed.
  26. F. Kleitz, F. Bérubé, R. Guillet-Nicolas, C. M. Yang and M. Thommes, J. Phys. Chem. C, 2010, 114, 9344 CAS.
  27. P. Van der Voort, P. I. Ravikovitck, K. P. De Jong, M. Benjelloun, E. Van Bavel, A. H. Jansen, A. V. Neimark, B. M. Weckhuysen and E. F. Vansant, J. Phys. Chem., 2002, 106, 5873 CrossRef CAS.
  28. A. Galarneau, H. Cambon, F. Di Renzo, R. Ryoo, M. Choi and F. Fajula, New J. Chem., 2003, 27, 73 RSC.
  29. Y. Park, T. Kang, P. Kim and J. Yi, J. Colloid Interface Sci., 2006, 295, 464 CrossRef CAS PubMed.
  30. C. Louis, Z. X. Cheng and M. Che, J. Phys. Chem., 1993, 97, 5703 CrossRef CAS.
  31. A. Ungureanu, B. Dragoi, A. Chirieac, S. Royer, D. Duprez and E. Dumitriu, J. Mater. Chem., 2011, 21, 12529 RSC.
  32. E. Marceau, M. Che, J. Čejka and A. Zukal, ChemCatChem, 2010, 2, 413 CrossRef CAS.
  33. F. Negrier, E. Marceau and M. Che, Chem. Commun., 2002, 1194 RSC.
  34. E. Marceau, A. Löfberg, J. M. Giraudon, F. Négrier, M. Che and L. Leclercq, Appl. Catal., A, 2009, 362, 34 CrossRef CAS.
  35. J. P. Dacquin, D. Sellam, C. Batiot-Dupeyrat, A. Tougerti, D. Duprez and S. Royer, ChemSusChem, 2014, 7, 631 CrossRef CAS PubMed.

Footnote

Electronic supplementary information (ESI) available: Supplementary characterization results. See DOI: 10.1039/c7cy00963a

This journal is © The Royal Society of Chemistry 2017