Use of pillared clays in the preparation of washcoated clay honeycomb monoliths as support of manganese catalysts for the total oxidation of VOCs

Catalysis Today xxx (xxxx) xxx–xxx

Contents lists available at ScienceDirect

Catalysis Today journal homepage: www.elsevier.com/locate/cattod

Use of pillared clays in the preparation of washcoated clay honeycomb monoliths as support of manganese catalysts for the total oxidation of VOCs ⁎

José M. Gaticaa, , Jorge Castiglionib, Carolina de los Santosb, M. Pilar Yestea, Gustavo Cifredoa, Martín Torresb, Hilario Vidala a b

Departamento C.M. I.M. y Química Inorgánica, Universidad de Cádiz, Puerto Real 11510, Spain Laboratorio de Fisicoquímica de Superficies, DETEMA, Facultad de Química, Universidad de la República, 11800 Montevideo, Uruguay

A R T I C L E I N F O

A B S T R A C T

Keywords: Environmental catalysis Honeycomb monoliths Manganese Pillared clays VOCs oxidation

Manganese catalysts supported by impregnation onto honeycomb monoliths extruded from commercial clay, previously coated with aluminium-pillared clay, were prepared and tested in the total combustion of propane and acetone. Samples characterization included chemical analysis, nitrogen physisorption, electron microscopies (SEM-EDS, HAADF and EELS), XRD, TPR and O2-TPD experiments. The active phase (around 5 wt%), consisting of MnO2 particles as majority phase, with homogeneous size and shape, that tends to concentrate in some regions of the surface of the clay support, exhibited high efficiency to oxidize the two model VOCs investigated and stability in severe reaction conditions. Light-off temperatures as low as 225 and 330 °C were found for the oxidation of acetone and propane respectively. The better performance observed in comparison to the monoliths without pillared clay was attributed to the higher active phase loading. Differences found as function of the VOC's nature and concentration were related to the different oxidation mechanisms proposed in literature, either just Mars-van Krevelen or this with also Eley-Rideal contribution, for acetone and propane respectively. These results combined with the intrinsic advantages of the honeycomb monolithic design open up new possibilities for using pillared clays as catalytic support in VOCs oxidation under more affordable conditions.

1. Introduction Literature related to the use of manganese-based oxides as alternative to noble metals for the abatement of VOCs is so far extensive, dealing with both pure manganese oxides [1–4] or Mn-containing mixed oxides [5] and MnOx deposited onto various supports [6–8], not only in the form of powders but also onto honeycomb monoliths, either using cordierite, washcoated with high surface alumina [9], or metallic substrates [10–12]. On the contrary, it is surprising that references which employ Mn-based catalysts supported onto clays for this application are much scarcer [13], in spite of the fact that clays are cheap and abundant materials, and that many of them possess ideal rheological properties for preparing structured supports by extrusion. Pillared clays with large surface area can be also used as supports for metal catalysts to achieve homogeneous dispersion, increased reactant adsorption area, and shape selectivity [14]. The utilization of clays pillared with transition metal oxides has been extended to reactions such as the dehydrogenation of cyclohexane to benzene [15], the Fischer–Tropsch synthesis [16], the selective catalytic oxidation of H2S [17], the selective catalytic reaction of NO [18], the

⁎

hydrodesulfurization of thiophenes [19], and selective organic transformations [20], among others. In these examples, the activity is intrinsically associated to the metal oxides acting as pillars, but in other occasions the active phase is further incorporated by impregnation on the pillared clays surface area, which has been also done to prepare supported Mn oxides [21]. The traditional methods for preparing pillared clays are carried out in very diluted systems, so the contact between the clay in suspension and the intercalating solution allows diffusion of cations, used as pillared precursors, into the clay sheets, generating a homogeneous pillars structure [22,23]. However, their scale up for their industrial production remains a challenge (time-consuming processing, high volumes handling and lack of reproducibility) [24,25]. Therefore it is of great interest the development of optimized standard procedures or experimental techniques for the use of such materials. Recently the use of concentrated dispersions and microwave or ultrasound irradiation during the intercalation step have been proposed [26–28]. Another alternative might be to deposit pillared clays onto honeycomb monoliths as a way to minimize the amount of pillared clay needed, maximizing the contact with active phases, while taking profit of the

Corresponding author. E-mail address: [email protected] (J.M. Gatica).

http://dx.doi.org/10.1016/j.cattod.2017.04.025 Received 12 January 2017; Received in revised form 28 March 2017; Accepted 10 April 2017 0920-5861/ © 2017 Elsevier B.V. All rights reserved.

Please cite this article as: Jose Manuel, G.M., Catalysis Today (2017), http://dx.doi.org/10.1016/j.cattod.2017.04.025

Catalysis Today xxx (xxxx) xxx–xxx

J.M. Gatica et al.

methodology to that employed by other authors for depositing clays onto metallic monoliths [43]. Briefly, the synthesized powder pillared clay was used to prepare a stable aqueous suspension (1.7 wt%). The adequate amount of solid was dispersed in deionised water and the mixture was sonicated for 3.5 h. After this time, a polyvinyl alcohol (PVA) solution was added to the suspension to improve its stability in clay/PVA ratio of 1:1 by weight. The mixture was kept under vigorous stirring for 24 h and then, the ClayM monolith was immersed in it during 1 h under ultrasounds (110 W). After careful withdrawal, the excess of colloidal solution was removed by blowing. Then the monolith was dried overnight at 60 °C and finally calcined at 450 °C for 2 h at a heating rate of 1 °C/min. The above resulting PILC/ClayM monoliths were impregnated by immersion with a 1 M solution of Mn(NO3)2·4H2O (from Sigma Aldrich, 98.5% pure) with continuous stirring for 30 min. These active phaseloaded monoliths were further dried by using microwaves (500 W, 1 min), and finally submitted to calcination at 450 °C for 2 h using a heating rate of 2 °C/min. According to previous TemperatureProgrammed Oxidation (TPO) experiments [40], this treatment ensures metal precursor decomposition, while keeping structural stability of the clays supports [29]. Hereafter the final monoliths will be referred to as Mn/PILC/ClayM. A similar procedure, skipping the intermediate washcoating step or the last impregnation, was followed to prepare Mn/ ClayM [40] and PILC/ClayM monoliths respectively, both used as reference samples.

extra advantages of structured supports with respect to packed bed reactors [29,30]. In this sense, our group has demonstrated the possibility of preparing clay-based honeycomb monoliths with great potential for several environmental applications as a cheaper alternative to commercial structured supports such as those based on commercial high-tech cordierites [31–37]. On the other hand, in previous studies we also prepared aluminium-pillared clays and used them as catalytic support of manganese oxides for the total oxidation of acetone and propane [38,39]. With the above precedents, the novel aim of this research was to prepare and characterize manganese catalysts supported onto honeycomb monoliths extruded from commercial clay, and to test them in the total combustion of propane and acetone, two model VOCs which are often present in end-of-pipe emissions. Moreover, before the incorporation of the manganese phase by using a simple impregnation technique, the clay honeycomb monoliths were also coated with a pillared clay by a washcoating procedure in order to further improve the physicochemical properties of the catalyst and consequently its performance. On this regard, the bare clay honeycomb support [34], this impregnated with the manganese phase [40] and the one just washcoated with the pillared clay were also studied as references of the ternary system. 2. Experimental 2.1. Materials and catalyst preparation

2.2. Characterization

The clay employed in this work to prepare clay honeycomb monoliths, named as ARGI, was provided by VICAR S.A., and came from deposits located at the east of Spain. It was received as a powder with a grain size of 100 μm and a nominal composition (induced couple plasma, ICP, analysis) as follows: 57% SiO2, 28.4% Al2O3, 1% Fe2O3, 1.5% TiO2, 0.5% CaO, 2.5% K2O, 0.5% MgO, 0.3% Na2O and 7.8% of undetermined ashes (mass contents for the dry sample). Extrusion of this clay was achieved without needing other additives except water (0.3–0.4 ml/g of paste) and following the methodology previously reported [41]. The resulting green monoliths were dried overnight at 90 °C and finally calcined at 440 °C for 4 h to enhance their mechanical resistance without altering the clay structure [34]. The monoliths finally obtained (Fig. 1), presented a honeycomb-type circular section with a diameter of 1.4 cm, a density of approx. 50 cells/cm2, 0.33 mm of wall thickness and a 72% open frontal area. The resulting monoliths (from now on ClayM) were subsequently washcoated by an aluminium pillared clay, named as PILC. This was previously prepared from a calcium-rich montmorillonite with low sodium and potassium content from Uruguay as reported elsewhere [42] leading to a material with the following textural properties: SBET = 235.2 m2/g; pore volume = 0.167 cm3/g; micropore volume = 0.096 cm3/g; and average pore diameter = 10.7 nm. Deposition of this PILC over the ClayM monoliths was conducted using a similar

The samples prepared in this investigation were characterized by means of different techniques such as chemical and textural analyses, Scanning Electron Microscopy with Energy Dispersive Spectroscopy (SEM-EDS), X-ray diffraction (XRD), High-angle annular dark-field (HAADF) electron microscopy, Electron Energy Loss Spectroscopy (EELS), Temperature-Programmed Reduction (TPR) and O2 Temperature-Programmed Desorption (O2-TPD). The metal content of the supported catalysts was estimated by inductively coupled plasma spectroscopy (ICP) analysis, using a NexION (Perkin-Elmer) mass spectrometer and an IRIS Intrepid Thermo Elemental instrument, for Mn/PILC/ARGI and Mn/ARGI monoliths respectively. Textural characterization was performed by means of N2 physisorption at −196 °C using a Micromeritics ASAP2020 instrument. For this analysis the monoliths were pre-evacuated at 200 °C for 2 h. Specific surface area was measured by the BET method. Total pore volume (Vp) was calculated from the amount of nitrogen adsorbed at relative pressures around 0.99. Pore size distribution and pore mean size diameter were determined by the BJH method from the desorption branch of the isotherms. SEM images and EDS compositional data corresponding to small pieces of the monoliths were obtained using a QUANTA-200 scanning electron microscope equipped with a Phoenix microanalysis system using a nominal resolution of 3 nm. In addition, EDS maps corresponding to the distribution of Mn and Si were recorded in a Field Emission Gun (FEG) scanning electron microscope (Nova NanoSEM 450) equipped with an EDAX system. XRD studies were carried out in a Bruker diffractometer, D8 Advance 500 model, to study ClayM and Mn/ClayM monolithic samples, previously crushed and sieved. In this case diffractograms were recorded using Cu Kα radiation and a graphite monochromator. The 2θ angle ranged from 2° to 130°, with a step of 0.03° and a counting time per step of 45 s. In the case of PILC/ClayM and Mn/PILC/ClayM samples, similar equipment was employed but using Mo Kα radiation. Measurements were performed from 0.9° to 49.4° during 7 h. The tube operated at 50 kV and 50 mA, and the sample turned at 10 rpm. The XRD data were properly processed to allow comparison between diagrams recorded using different radiation. Complementary analysis

Fig. 1. Image of a typical clay honeycomb monolith extruded in this work.

2

Catalysis Today xxx (xxxx) xxx–xxx

J.M. Gatica et al.

was performed by using the Fullprof Rietveld programme [44] to simulate the XRD diagrams, assuring in this way the correct identification of the phases. A JEOL 2010F transmission electron microscope, equipped with a FEG, operating at 200 kV and with a structural resolution of 0.19 nm, was used to record HAADF images and EELS data. Both images and spectra were processed with Digital Micrograph software. For these studies we used powdered samples obtained from crushing small pieces of the monoliths. Qualitative TPR experiments were carried out using a Thermostar QMS 200 (Pfeiffer) mass spectrometer (MS), employing a flow of 60 ml/ min of H2(5%)/Ar, over crushed monolith pieces (100 mg) and with a heating rate of 10 °C/min. Before the experiments the samples were cleaned by heating at 200 °C for 1 h under O2(5%)/He flowing at 60 cm3/min. In addition, quantitative hydrogen consumption data were obtained in a Micromeritics AutoChem II 2920 using a TCD detector. Finally, O2-TPD experiments were performed in an experimental device coupled to a Pfeiffer, model Thermostar GSD301T1, quadrupole MS, employing a flow of 60 ml/min of He, over crushed monolith pieces (200 mg) and with a heating rate of 10 °C/min. The pre-treatment of the samples consisted of a heating at 450 °C for 1 h under O2(5%)/He flowing at 60 cm3/min, with further cooling down to 150 °C under the flow of diluted oxygen and finally to 25 °C in He.

Table 1 Summary of main results obtained in the compositional and textural characterization of the honeycomb monolithic samples investigated. Sample

Mn (wt%)a

SBET (m2/g)

Vpb (cm3/g)

Vmicroc (cm3/g)

Dpd (nm)

ClayM PILCe PILC/ClayM Mn/ClayM Mn/PILC/ClayM

– – – 3.07 ± 0.01 5.18 ± 0.05

27.6 72.3 32.5 28.9 20.1

0.089 0.085 0.081 0.064 0.069

0.001 0.066 0.006 0.002 0.005

14.8 6.5 14.6 11.5 21.9

a

As measured by ICP analysis. Estimated from the amount of nitrogen adsorbed at P/P0 = 0.99. Estimated by means of t-plot (Harkins-Jura) analysis from N2 physisorption data. d Determined from the adsorption branch of the N2 physisorption isotherm following the BJH method. e In the form of powder and calcined at the same temperature as the monoliths. b c

Fig. 3 and results collected in Table 1. First notice how the specific surface area of the ClayM monolith increases upon the PILC washcoating, while mesopores below 10 nm develop. This result is consistent with the mesoporous character of the PILC as illustrated by Fig. 4. The isotherm of the PILC calcined at the same temperature as the monoliths is of Type IV with H4 hysteresis loop, associated with narrow slip-like pores. Moreover, considering the BET surface area of the PILC, the amount of pillared clay incorporated to the ClayM substrate by the washcoating procedure could be estimated around 11 wt%. Second, it is remarkable the different effect of the subsequent metal impregnation on the textural properties as function of the presence of PILC. While direct coating of the clay honeycomb monoliths with the manganese phase harder affects neither the specific surface area nor the porosity of the support (compare ClayM and Mn/ClayM values), previous modification with the pillared clay induces a significant change of the texture (notice the SBET reduction, the mean pore diameter increase and the disappearance of low size mesopores when passing from PILC/ ClayM to Mn/PILC/ClayM). This could be related to a more significant deposition of manganese with preferential blockage of small pores.

2.3. Activity testing Catalytic tests were run over entire monolith pieces (2 cm long), placed inside a Pyrex glass reactor. Before the reaction, each monolith was subjected to a pretreatment at 450 °C for 30 min under Ar stream. In the case of propane oxidation, the activity was measured operating at atmospheric pressure between 200 and 450 °C, with a feedstock of 79.7% Ar, 19.8% O2 and 0.5% C3H8 (167 ml/min). To evaluate the influence of the VOC's concentration some extra experiments were performed with a mixture of 59.5% Ar, 39.5% O2 and 1.0% C3H8, and same total flow. For both conditions propane conversion and reaction products were analyzed by gas chromatography (GC14B Shimadzu). Regarding acetone oxidation, the activity was also measured at atmospheric pressure between 150 and 450 °C, using in general a reactant mixture containing 1200 ppm of C3H6O in a gas stream, 97.2% Ar and 2.8% O2, total flow being 124 ml/min. As above, complementary experiments were performed with double amount of VOC. Acetone conversion and reaction products were analyzed by gas chromatography (GC2014 Shimadzu). Stability tests in the case of manganese-containing monoliths were also performed for both reaction at 450 °C for 24 h operating with VOCs concentration of 1.0 vol.% and 1200 ppm for propane and acetone, respectively.

3.2. Characterization by means of SEM-EDS In general, images obtained by SEM for PILC/ClayM sample (Fig. 5A) were similar to those previously observed for the clay honeycomb monolith support [34]. No matter the piece of monolith analyzed, agglomerates of particles of heterogeneous size and irregular shape as well as inter-particle/agglomerate void spaces, which may contribute to the porosity stated above, were observed. However, a perusal analysis through EDS of PILC/ClayM revealed interesting findings. First of all, although relatively homogeneous, chemical composition resulted to be intermediate between that of pure ClayM [34] and PILC [42], confirming the successful deposition of the pillared clay. Furthermore, a lower relative content of potassium and titanium was detected in bright contrast zones which could be therefore PILCrich, considering the absence of these elements in such clay [42], in comparison with darker areas, rather attributable to bare ClayM. Concerning the active phase dispersed, Fig. 5 also shows the overlapping of the EDS maps obtained for Mn and Si in order to visualize the general surface distribution of the former and its relative position respect to the most representative element of the two clay supports. It should be noticed that the concentration of the minority elements that might be the most appropriate to distinguish between the two clays was not high enough as to allow discerning the exact location of the pillared clay and its likely interaction with the manganese. In any case, two valuable remarks can be made from the mapping study. First, notice the clearly higher amount of manganese present in the pillared clay-containing sample which is in agreement with the previously commented ICP analysis. And second, the active phase distribution in this sample (Fig. 5B4) seems to be less homogeneous than in Mn/ClayM

3. Results and discussion 3.1. Compositional and textural characterization of the catalysts Results obtained by ICP analysis of the investigated catalysts are shown in Table 1. It can be observed that the amount of manganese introduced over the PILC/ClayM honeycomb monoliths is higher than that over the ClayM monoliths. This increase of the metal content can be considered a first indication of the positive effect of washcoating the starting clay support with the pillared clay. The above results are consistent with those derived from the N2 physisorption experiments. For all the honeycomb monolithic samples prepared isotherms of Type II indicative of macroporous adsorbents, including a characteristic hysteresis loop (type H3) associated with capillary condensation in mesopores of samples with aggregates of plate-like particles [45], were recorded (Fig. 2). Processing of these isotherms allowed obtaining the pore size distribution curves shown in 3

Catalysis Today xxx (xxxx) xxx–xxx

J.M. Gatica et al.

Fig. 2. Nitrogen adsorption/desorption isotherms of the monolithic samples investigated. Both adsorption (o) and desorption (•) branches are depicted.

was previously found in the Mn/ClayM sample [40] in which high contrast particles associable to manganese were detected but in much lower content. In fact, the average manganese content on the surface of the Mn/PILC/ClayM resulting from the SEM-EDS analysis of many different regions in scan mode was estimated as 22.9 ± 2.1 wt%. This result is clearly higher than that measured on Mn/ClayM (4.87 ± 1.07 wt%) [40], and suggests again that the washcoating PILC favours the incorporation of more manganese on the monolithic support.

3.3. Structural characterization Fig. 6 shows the X-ray diffractograms obtained for the Mn/PILC/ ClayM monoliths prepared as well as for the reference samples employed in this study. The first observation worthy of comment is that this technique not only also allows detecting the manganese phase in the ternary system but also its structural identification. In this way, the peaks at 37.3° and 56.7°, which are absent both in ClayM and PILC/ ClayM, can be assigned to MnO2 pyrolusite (PDF file 24-735). This situation contrasts with that of Mn/ClayM which only showed peaks characteristic of the clay support [34]. This difference is consistent with the chemical analyses (both at surface and massive levels) above commented that pointed to a higher manganese content in the pillared clay coated monoliths, and it is also in good agreement with results obtained by other authors [46] who studying TiO2-supported manganese catalysts only detected peaks related to the metal phase for loadings higher than 11% by weight. On the other hand, distinguishing the PILC from the ClayM by means of XRD is a much harder task because the most characteristic reflections generated by its structure are positioned at low angles where the contribution of the baseline is high. Nevertheless, comparison of the diagrams of PILC/ClayM and ClayM represented versus the d spacing parameter instead of 2θ (Fig. 7) allowed finding (in the former) a not well defined band around 18 Å that might correspond to the pillared clay. This effect was more difficult to observe in the Mn/PILC/ClayM. A reasonable explanation could be a progressive distortion of the pillars during the washcoating process and subsequent impregnation with the metal precursor solution. A more detailed analysis of the XRD diagrams using Rietveld allowed discarding the existence of other manganese-containing phases

Fig. 3. Pore size distribution curves as obtained by means of BJH analysis of the desorption branch of the N2 isotherms (−196 °C) for the monolithic samples.

(Fig. 5C). In any case, it should be considered that this description corresponds to very wide regions (400 × 370 microns) of the samples surface. When a look is had a much higher magnification inside the zones where manganese seems to concentrate in the Mn/PILC/ClayM monolith (Fig. 5B1 and B2), very fine rounded particles (average size around 150 nm) homogeneously spread out all over the smaller explored monolith surface areas (25 × 20 microns, Fig. 5B1) can be clearly observed. On this regard, the use of EDS technique on selected areas in spot mode analysis (Fig. 5B3) allowed confirming that those particles correspond to the Mn-containing phase. A similar observation 4

Catalysis Today xxx (xxxx) xxx–xxx

J.M. Gatica et al.

Fig. 4. Nitrogen adsorption/desorption isotherms and pore size distribution curves as obtained by means of BJH analysis of the desorption branch of the N2 isotherms (−196 °C) corresponding to the PILC sample used for the preparation of washcoated clay honeycomb monoliths. The figure shows results obtained for both the (A) fresh sample and (B) that calcined at the same temperature than the coated monoliths.

obtained in the analysis of the Mn/PILC/ClayM but with a lower signal/ noise ratio according to the higher thickness of the manganese aggregates in this sample.

other than MnO2 in the Mn/PILC/ClayM sample. In addition, this refinement revealed that the MnO2 phase was present in the form of a bimodal distribution, as highly anisotropic crystallites with average sizes of 6 and 28 nm for 91 wt% and 9 wt%, respectively. In this sense, it should be noticed that particles observed by SEM (Fig. 5B1 and B2) must correspond to aggregates of the crystals detected by XRD. These aggregates were also observed by means of HAADF technique (Fig. 8), which can distinguish Mn from the lighter elements of the clay taking profit of the different contrasts that can be generated in the recorded images as a function of the atomic number. In this sense, it is remarkable the much smaller size of the manganese aggregates in the Mn/ClayM sample. This result might explain why the manganese phase was neither observed in its SEM images nor detected in its XRD diagram. Confirmation of the chemical nature of the aggregates observed by HAADF was possible by recording the Manganese L2,3edge EELS spectra (also included in Fig. 8). In our case the more conventional study focusing in the Oxygen K-edge spectra was not possible considering the majority presence of this element in the clay [47]. Fig. 8 shows two manganese L2,3-edge spectra after background subtraction, corresponding to different areas of one Mn/ClayM aggregate, detected by HAADF. They consist of two white lines L3 and L2 due to the transitions from 2p3/2 and 2p1/2 core states to 3d unoccupied states localized on the excited manganese ions. The L3 peak position tends to shift to higher energy as the formal valence of manganese ions is increased [48]. According to [49], peaks maxima centred at 642 and 652 eV in A spectrum can be related to MnO2 while the doublet [48] with maxima at 639 and 641 eV and the peak at 651 eV [49] can be reasonably attributed to a more reduced manganese phase, Mn3O4. Its poor crystallinity and/or high dispersion might explain its no detection by XRD. Also important, similar spectra to the just described were

3.4. Active phase redox characterization As well known, the interpretation of reduction profiles in supported catalysts obliges to separate those processes related with the dispersed phase from those corresponding to the support. In this sense, although the clay materials employed do not belong to the category of reducible oxides, their heating treatment in this kind of experiments can produce water from their dehydroxylation which is not associated to hydrogen consumption. Fig. 9 shows the results of the TPR study using mass spectrometry for the evolved gasses analysis. The analysis of the m/e 18 signal reveals slight differences between profiles of PILC/ClayM and ClayM. In particular, the shoulder appearing around 300 °C in the former can be attributed to dehydroxylation of the pillared clay, being another evidence of its incorporation over the ClayM monolith surface as consequence of the washcoating. Besides this the intense desorption peak around 500 °C must be associated with the massive clay support dehydroxylation in both samples. Focusing on the m/e 2 signal which provides information of the H2 consumption, the reduction profiles obtained for the Mn-containing samples are similar, consisting of two well defined and narrow peaks, the one at lower temperature with higher relative intensity. The assignment of these peaks to different oxidation states of manganese is certainly a hard task [50]. In any case, our results are in good agreement with those obtained in [50–52] among others, in which the XRD and XPS characterization of the samples investigated allowed tentatively relating this TPR profile to MnO2. Moreover, the two peaks 5

Catalysis Today xxx (xxxx) xxx–xxx

J.M. Gatica et al.

Fig. 5. Scanning electron micrographs obtained for small pieces of external walls of the PILC/ClayM (A) and Mn/PILC/ClayM monoliths at two different magnifications (B1 and B2), and typical EDS spectrum obtained for the latter by spot mode analysis over bright fine particles of the selected area indicated (B3). Besides Mn, the other elements which appear (except Au used to metallize the samples) proceed from the clays used as support. Mn (yellow) and Si (purple) distributions as obtained by EDS mapping for the Mn/PILC/ClayM (B4) and Mn/ClayM (C) monoliths are also included. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

0.62 mmol H2/g and 0.78 mmol H2/g for the Mn/ClayM and Mn/ PILC/ClayM samples, leading to manganese average oxidation numbers of 3.2 and 3.6, respectively. Considering the above commented EELS data that showed the coexistence of MnO2 and Mn3O4, relative contents of 47% and 53% in the Mn/ClayM, and 73% and 27% in the Mn/PILC/ Clay were estimated for these phases, respectively. According to these results, the manganese would be slightly more reduced in the not containing pillared clay sample. Furthermore, the maxima of the peaks in the TPR profile of Mn/PILC/ClayM occur at approximately 40 °C above the respective ones in Mn/ClayM. This difference can be indicative of a slightly higher dispersion of the manganese phase in the latter [51,54]. All these results are consistent with data discussed in the structural characterization section.

would be the result of a two steps reduction: from MnO2 to Mn3O4 (via Mn2O3) at lower temperature, and the reduction of Mn3O4 to MnO at higher temperature. Other authors also proposed the formation of this phase on cordierite honeycombs after impregnation with manganese nitrate followed by drying at 120 °C during 4 h and final calcination at 500 °C for 2 h [9]. Nevertheless, taking into account other studies, some contribution of other manganese oxides than MnO2 to the TPR profiles recorded by mass spectrometry cannot be completely discarded. According to [53] the intensity ratio between the two signals here recorded allow excluding the presence of Mn2O3 but not that of Mn3O4. With the aim of elucidating this issue, complementary TPR studies but using a quantitative TCD detector were performed (results not shown). The estimate of the total hydrogen consumption resulted to be 6

Catalysis Today xxx (xxxx) xxx–xxx

J.M. Gatica et al.

3.5. Catalytic activity First of all it should be mentioned that only total combustion products (CO2 and H2O) were detected in both investigated reactions. The study of the propane oxidation reaction (Fig. 11A) indicated that the activity of the bare ClayM honeycomb monolith is negligible under our experimental conditions while PILC/ClayM only exhibits some activity above 350 °C, keeping below 20% of conversion at the highest temperature studied. On the contrary, monoliths containing manganese showed activity even from 200 °C, reaching a conversion of ca. 92% at 450 °C. These results are competitive if compared to other catalysts employed in this reaction, for example to high surface area ceria for which a 90% of conversion at 450 °C was reported [56]. The activity found in this work for the oxidation of propane is also in the order of that observed for co-precipitated manganese/aluminium Mn3O4/Al2O3 mixed oxide catalysts [6]. In addition, and also noticeable, the results obtained in this study are even better than those previously measured for the same system but unsupported. In particular, temperatures of 400 °C and 500 °C were obtained for T50 and T100 values respectively for a powdered MnOx (6 wt%)/PILC in reasonably comparable experimental conditions [39]. It is also remarkable that among the two manganese-containing samples here studied, Mn/ClayM was (though narrowly) the most active, in spite of the fact that the one including the pillared clay had a higher loading of manganese. This result can be tentatively explained on the basis of the possible mechanism for this reaction. According to literature [57] combustion of saturated and unsaturated hydrocarbons (n-hexane, propane, toluene, etc.) is carried out through an Eley-Rideal mechanism (related with the available active phase surface) or a combination of this with a Mars-van Krevelen one (which involves the active phase mass). In such scenario, the higher dispersion of manganese in the Mn/ClayM monolith might compensate for its lower content, making it exhibit better catalytic performance. In any case, the partial contribution of a Mars-van Krevelen mechanism in this reaction should not be completely excluded. Previous studies [50] have suggested that the Mn3O4 phase, as the one detected in our catalysts by EELS, is particularly active for VOCs oxidation reactions operating through that type of mechanism. Let's recall that, combining EELS results with quantitative TPR data, we estimated a slightly higher content of Mn3O4 in the Mn/ClayM sample. Therefore this could also enhance its catalytic activity. Some authors have used changes in the partial pressure of the reagents to gain insight into the kinetic aspects of VOCs oxidation processes [58]. In this sense, Fig. 11A also shows the results of running the propane combustion over the metal-loaded monoliths under double concentration of VOC. As can be observed, in both cases there is a clear reduction of the conversion which suggests a saturation of the monolith surface under 1% of propane. However, this is less pronounced in the Mn/PILC/ClayM sample in which the Mars-van Krevelen component could be favoured under these conditions. In the case of the acetone oxidation (Fig. 12A) the bare ClayM and just washcoated with PILC showed almost equal performance, and although their activity was clearly worse than that of the manganesecontaining samples, conversion reached 90% at 450 °C. This result compared to those observed in the case of propane is reasonably attributable to the well-known higher reactivity of the acetone molecule under same experimental conditions. In any case, the most interesting result was that observed for the Mn/PILC/ClayM monolith with a light-off temperature as low as 225 °C, and full conversion at 350 °C. This performance was much better than that obtained with the Mn/ClayM monolith, which again can be related to mechanistic aspects. According to literature [57] combustion of oxygenated VOC molecules (alcohols, aldehydes, ketones, etc.) proceeds via a Mars-van Krevelen mechanism. Thus the catalyst with higher manganese loading, Mn/PILC/ClayM, should be more active. Recall that according to O2TPD experiments the mobility of lattice oxygen was similar in both Mncontaining catalysts; however its relative content was significantly

Fig. 6. X-ray diffractograms obtained for all the honeycomb monoliths studied.

Fig. 7. X-ray diffractograms of ClayM and PILC/ClayM as function of d parameter.

In order to obtain complementary information about the availability of oxygen in the manganese-containing samples, additional O2-TPD experiments were carried out (Fig. 10). In both cases very similar in shape profiles were obtained. These were characterized by two main peaks, the most intense one around 540 °C, and the second one, less intense at upper temperature, centred approx. at 870 °C. The interpretation of the origin of these peaks should be reasonably made in the same terms as the TPR profiles, although in this case corresponding to oxygen desorption under inert atmosphere, that according to the temperature at which takes place can be ascribed to lattice oxygen in agreement with other authors [50,55]. However, the shoulder of the first peak at lower temperature can be associated with desorption of surface oxygen species (O− and O2−) and/or labile oxygen species (β species), with different Mn–O bond strengths [50]. In any case, two significant differences can be noticed in relation to the most intense peak when the two samples studied are compared. First, its intensity is clearly higher in Mn/PILC/ClayM which is in good agreement with the higher loading of manganese oxide in this sample. And second, its position is slightly shifted to lower temperature in Mn/ClayM, which resembles the effect observed by TPR and that was taken as a sign of higher dispersion of the manganese phase. 7

Catalysis Today xxx (xxxx) xxx–xxx

J.M. Gatica et al.

Fig. 8. HAADF images of the manganese-containing samples and EELS spectra of the Mn/ClayM sample, recorded for the labelled zones.

under 1200 ppm of acetone and GHSV equal to 14,000 h−1, the T50 and T90 values were 46 °C and 99 °C higher than those reported for a cordierite honeycomb washcoated with Cu0.15Mn0.3O0.55 operating at 10,000 h−1 [62]. On the contrary, our results were very similar to those found for a commercial monolithic Pd/Al2O3 [62]. Finally, in order to check the potential of the prepared catalysts in a more realistic context, we have performed stability tests by monitoring the conversion versus time-on-stream at the highest temperature studied in this work, 450 °C. In the case of acetone (Fig. 12B) loss of conversion lower than 3% for both catalysts was recorded after 24 h. In the case of propane (Fig. 11B) no significant change was observed for Mn/ClayM while the conversion decreased about 10% for the pillaredcontaining monolith. In a previous study we had already reported that in a similar study performed over the Mn/ClayM sample but at less severe conditions (300 °C and 0.5% of propane) the catalytic activity remained constant [40]. Therefore, the data here reported widen the operative window and demonstrate that the use of clays for manufacturing honeycomb supports allows obtaining catalysts which are adequate in relation with their stability under long-term and severe reaction conditions.

higher in the one with pillared clay. In fact, the results obtained for this catalyst are even better than those reported for the complete oxidation of acetone over manganese oxide catalysts supported on alumina- and zirconia-pillared clays [7]. It should be also noticed that this catalyst reasonably keeps the performance when compared to the unsupported MnOx(6 wt%)/PILC (T50 = 200 °C, T100 = 350 °C) [38]. For this reaction the effect of doubling the VOC's concentration over the metal-loaded monoliths was also studied (Fig. 12A). Unlike that observed with propane, no significant changes were observed in the 200–450 °C range upon increasing the concentration of acetone from 1200 to 2400 ppm in the Mn/PILC/ClayM sample. This suggests that under our experimental conditions, the saturation of active sites is not reached. By contrast, being the operating Mars-van Krevelen mechanism less sensitive to the active phase dispersion, the Mn/ClayM monolith (with lower active phase content) suffered a slight deactivation between 300 and 400 °C. To better evaluate the performance of the catalysts proposed in this work, Table 2 summarizes the best results found in literature [59–62] for the total oxidation of propane and acetone, using both fix bed powdered catalysts and honeycomb monoliths. Although making comparisons is a complex task due to the different experimental conditions employed in each case, two general remarks can be stated: (i) in the case of propane oxidation, although powdered references behave better, our catalysts are more active than those reported of monolithic type. Notice that for the Mn/ClayM monolith, operating with 0.5% propane and GHSV of 19,000 h−1, we obtained T50 and T90 values that are 85 °C and 23 °C lower than those reported in [60]; (ii) for the acetone oxidation our catalysts also exhibit an intermediate performance. As an example, for the Mn/PILC/Clay sample, working

4. Conclusions Clay honeycomb monoliths previously washcoated with pillared clay were used to prepare supported manganese catalysts by means of a simple impregnation technique for the oxidation of propane and acetone, and compared with the same system without pillared clay. Textural and chemical characterization of the catalysts demonstrated that the pillared clay helps to optimize the active phase deposited onto 8

Catalysis Today xxx (xxxx) xxx–xxx

J.M. Gatica et al.

Fig. 9. Diagrams showing both H2 consumption (m/e 2) and H2O production (m/e 18) during the TPR-MS experiments.

oxidize the two model VOCs investigated, with promising stability according to tests under severe conditions (450 °C, 24 h). This is illustrated by conversion values around 90% at only 300 °C and 450 °C for the oxidation of acetone and propane respectively. In any case, the most remarkable result obtained was a light-off temperature as low as 225 °C for the acetone oxidation on the pillared clay-containing catalyst, which is ca. 80 °C lower than that observed for the manganese catalyst supported onto the bare clay monolith. Moreover, the performance of this catalyst was similar or even better to that of the unsupported Mn/pillared clay in the acetone and propane oxidation, respectively. This proves the value of minimizing the amount of pillared clay through deposition onto a structured support, main target of this work. The effect of the VOC's concentration was also studied for both oxidation reactions, finding interesting differences between the one of acetone and that of propane, which could be reasonably related to their different mechanisms, either Mars-van Krevelen (involving lattice oxygen as confirmed by O2-TPD experiments) or Eley-Rideal, respectively. All these data, along with the intrinsic advantages of the honeycomb monolithic design, show the potential of the proposed formulation including clay (as received and especially further coated with a pillared clay) as a competitive material in the field of VOCs oxidation. In addition, they represent a step towards a more rational use of pillared clays, minimizing the necessary amount required as catalytic support, and under experimental conditions which may be closer to those of real applications.

Fig. 10. Diagrams obtained in the O2-TPD experiments.

the surface of the honeycomb monolith surface by increasing its amount (from 3 to 5 wt%, and from 5 to 30 wt%, for massive and surface contents, respectively) while keeping a relatively homogeneous distribution, at least in macroscopic areas of approx. 0.15 mm2 (EDS results). XRD with Rietveld analysis and HAADF images pointed out significant differences in the size of MnO2 crystals aggregates, this being higher in the pillared clay containing sample. Moreover, combination of TPR and EELS analyses revealed the presence of Mn3O4 along with MnO2 as active phases. The two types of Mn-coated clay honeycomb monoliths prepared in this study (with and without pillared clay) exhibited high efficiency to

Acknowledgements The authors thank the Ministry of Economy and Competitiveness of Spain (Project MINECO/FEDER Ref: MAT2013-40823-R), the Junta de Andalucía (FQM-110 and FQM-334 groups), Comisión Sectorial de Investigación Científica (CSIC-Udelar), and PEDECIBA-Química (Uruguay) for their financial support. They also acknowledge the SCCyT of Cadiz University (UCA) and the SCAI of Malaga University (UMA) for using their XRD, chemical analysis and electron microscopy 9

Catalysis Today xxx (xxxx) xxx–xxx

J.M. Gatica et al.

Fig. 11. Catalytic activity of the bare, pillared coated and manganese impregnated clay honeycomb monoliths in the propane oxidation (A) and conversion as function of time-on-stream for experiments run over the Mn-containing monoliths with 1 vol.% of propane at 450 °C (B).

Fig. 12. Catalytic activity of the bare, pillared coated and manganese impregnated clay honeycomb monoliths in the acetone oxidation (A) and conversion as function of time-on-stream for experiments run over the Mn-containing monoliths with 1200 ppm of acetone at 450 °C (B). Table 2 Summary of data reported in literature dealing with propane and acetone oxidation showing the best results in T50 and T90 values. VOC

Concentration (ppm)

GHSV (/h)

Total flow rate (ml/min)

Catalyst

Catalyst mass (g)

Design

T50 (°C)

T90 (°C)

Ref.

Propane Propane

5000 10,000

n.r. 2300

100 15

0.1 1

Powder Monolith

147 420

154 473

[59] [60]

Acetone Acetone

500 n.r.

n.r. 10,000

50 n.r.

0.3 n.r.

Powder Monolith

65 185

105 196

[61] [62]

Acetone

n.r.

10,000

n.r.

Ru-Re/γ-Al2O3 Clay honeycomb monolith Mesoporous chromia Cu0.15Mn0.3Ce0.55/ Cordierite honeycomb Commercial monolithic Pd/Al2O3

n.r.

Monolith

230

n.r.

[62]

n.r. Not reported. [5] [6] [7] [8] [9]

M. Tomatis, H. Xu, J. He, X. Zhang, J. Chem. (2016) Article ID 8324826. E. Finocchio, G. Busca, Catal. Today 70 (2001) 213–225. L.M. Gandía, M.A. Vicente, A. Gil, Appl. Catal. B 38 (2002) 295–307. A. Gil, M.A. Vicente, S.A. Korili, Catal. Today 112 (2006) 117–120. S. Azalim, R. Brahmi, M. Agunaou, A. Beaurin, J.-M. Giraudon, J.-F. Lamonier, Chem. Eng. J. 223 (2013) 536–546. [10] F.N. Aguero, B.P. Barbero, O. Sanz, F.J. Echave Lozano, M. Montes, L.E. Cadús, Ind. Eng. Chem. Res. 49 (2010) 1663–1668. [11] F.G. Durán, B.P. Barbero, L.E. Cadús, Chem. Eng. Technol. 37 (2) (2014) 310–316. [12] M.H. Castaño, R. Molina, S. Moreno, Rev. Acad. Colomb. Cienc. Ex. Fis. Nat. 39 (150) (2015) 26–35.

division facilities. References [1] L.M. Gandía, A. Gil, S.A. Korili, Appl. Catal. B 33 (2001) 1–8. [2] M. Baldi, E. Finocchio, F. Milella, G. Busca, Appl. Catal. B 16 (1998) 43–51. [3] M. Baldi, V. Sanchez Escribano, J.M. Gallardo Amores, F. Milella, G. Busca, Appl. Catal. B 17 (1998) L175–L182. [4] B. Puértolas, A. Smith, I. Vázquez, A. Dejoz, A. Moragues, T. García, B. Solsona, Chem. Eng. J. 229 (2013) 547–558.

10

Catalysis Today xxx (xxxx) xxx–xxx

J.M. Gatica et al.

[37] Z. Tian, T. Chafik, M. Assebban, S. Harti, H. Vidal, J.M. Gatica, G.A. Cifredo, N. Bahlawane, P.M. Kouotou, K. Kohse, Appl. Energy 108 (2013) 528–529. [38] C. de Los Santos, M. Torres, A. Quevedo, J. Castiglioni, Actas del VIII Congreso de Catálisis del Mercosur, Bahía Blanca, Argentina, 2015. [39] C. de Los Santos, M. Torres, A. Quevedo, R. Faccio, M. Sergio, J. Castiglioni, Actas del XXIV Congreso Iberoamericano de catálisis, Medellin, Colombia, 2014. [40] J.M. Gatica, J. Castiglioni, C. de los Santos, M.P. Yeste, G.A. Cifredo, M. Torres, H. Vidal, Int. J. Chem. Environ. Eng. 6 (4) (2015) 230–235. [41] J.M. Gatica, J.M. Rodríguez-Izquierdo, D. Sánchez, C. Ania, J.B. Parra, H. Vidal, Carbon 42 (2004) 3252–3254. [42] A. de León, C. De los Santos, L. Latrónica, A.M. Cesio, C. Volzone, J. Castiglioni, M. Sergio, Chem. Eng. J. 241 (2014) 336–343. [43] L.M. Martínez, M.I. Domínguez, N. Sanabria, W.Y. Hernández, S. Moreno, R. Molina, J.A. Odriozola, M.A. Centeno, Appl. Catal. A 364 (2009) 166–173. [44] J.J. Rodríguez-Carvajal, Phys. B 192 (1993) 55–69. [45] K.S.W. Sing, D.H. Everett, R.A.W. Haul, L. Moscou, R.A. Pierotti, J. Rouquérol, T. Siemieniewska, Pure Appl. Chem. 57 (4) (1985) 603–619. [46] P.R. Ettireddy, N. Ettireddy, S. Mamedov, P. Boolchand, P.G. Smimiotis, Appl. Catal. B 76 (2007) 123–134. [47] J.A. Pérez-Omil, J.J. Delgado, W. Ouahbi, A.B. Hungría, N. Browning, M.A. Cauqui, J.M. Rodríguez-Izquierdo, J.J. Calvino, J. Phys. Chem. C 114 (2010) 8981–8991. [48] H. Kurata, C. Colliex, Phys. Rev. B 48 (4) (1993) 2102–2108. [49] H.K. Schmid, W. Mader, Micron 37 (2006) 426–432. [50] V.P. Santos, M.F.R. Pereira, J.J.M. Orfao, J.L. Figueiredo, Appl. Catal. B 99 (2010) 353–363. [51] F. Arena, T. Torre, C. Raimondo, A. Parmaliana, Phys. Chem. Chem. Phys. 3 (2001) 1911–1917. [52] M. Sun, B. Lan, L. Yu, F. Ye, W. Song, J. He, G. Diao, Y. Zheng, Mater. Lett. 86 (2012) 18–20. [53] L. Christel, A. Pierre, D.A.R. Abel, Thermochim. Acta 306 (1997) 51–59. [54] C. Reed, Y. Lee, S.T. Oyama, J. Phys. Chem. B 110 (2006) 4207–4216. [55] H.C. Genuino, S. Dharmarathna, E.C. Njagi, M.C. Mei, S.L. Suib, J. Phys. Chem. C 116 (2012) 12066–12078. [56] C. de los Santos, D. Gómez, H. Vidal, J.M. Gatica, M. Sergio, J. Castiglioni, Libro de Actas SECAT 2013 Catalizadores y reactores estructurados, Seville, Spain P62 (2013), pp. 335–336. [57] F.N. Aguero, B.P. Barbero, L. Gambaro, L.E. Cadús, Appl. Catal. B 91 (2009) 108–112. [58] G. Mazloom, S.M. Alavi, Reac. Kinet. Mech. Cat. 110 (2013) 387–403. [59] K. Baranowska, J. Okal, Appl. Catal. A 499 (2015) 158–167. [60] M. Assebban, A. El Kasmi, S. Harti, T. Chafik, J. Hazard. Mater. 300 (2015) 590–597. [61] Y.S. Xia, H.X. Dai, L. Zhang, J.G. Deng, H. He, C.T. Au, Appl. Catal. B 100 (2010) 229–237. [62] H. Lu, Y. Zhou, H. Huang, B. Zhang, Y. Chen, J. Rare Earths 29 (2011) 855–860.

[13] J.E. Colman-Lerner, M.A. Peluso, J.E. Sambeth, H.J. Thomas, Reac. Kinet. Mech. Cat. 108 (2013) 443–458. [14] S. Cheng, Catal. Today 49 (1999) 303–312. [15] T.J. Pinnavia, M.S. Tzou, S.D. Landau, J. Am. Chem. Soc. 107 (1985) 4783–4785. [16] E.P. Giannelis, E.G. Rightor, T.J. Pinnavia, J. Am. Chem. Soc. 110 (1988) 3880–3885. [17] K.V. Bineesh, D.K. Kim, H.J. Cho, D.W. Park, J. Ind. Eng. Chem. 16 (2010) 593–597. [18] R.T. Yang, J.P. Chen, E.S. Kikkinides, L.S. Cheng, Ind. Ing. Chem. Res. 31 (6) (1992) 1440–1445. [19] J.T. Kloprogge, W.J.J. Welters, E. Booy, V.H.J. de Beer, R.A. Van Santen, J.W. Geus, J.B.H. Jansen, Appl. Catal. A 97 (1) (1993) 77–85. [20] B.M. Choudary, A. Durga Prasad, V. Bhuma, V. Swapna, J. Org. Chem. 57 (1992) 5841–5844. [21] L.M. Gandía, M.A. Vicente, A. Gil, Appl. Catal. A 196 (2000) 281–292. [22] G. Fetter, G. Heredia, L.A. Velázquez, A.M. Maubert, P. Bosch, Appl. Catal. A 162 (1997) 41–45. [23] L. Fang, L. Wang, T. Zhou, L. Liu, J. Zhou, M. Li, Appl. Clay Sci. 141 (2017) 88–94. [24] C.B. Molina, J.A. Casas, A.H. Pizarro, J.J. Rodríguez, Pillared clays as green chemistry catalysts: application to wastewater treatment, in: J.P. Humphrey, D.E. Boyd (Eds.), Clay: Types, Properties and Uses, Nova Science Publishers Inc., New York, 2011, pp. 435–474. [25] K.V. Bineesh, D. Park, Structural modification of montmorillonite clay by pillaring process: its characterization and applications, in: J.P. Humphrey, D.E. Boyd (Eds.), Clay: Types, Properties and Uses, Nova Science Publishers Inc., New York, 2011, pp. 371–390. [26] M.A. Vicente, A. Gil, F. Bergaya, Pillared clays and clay minerals, in: F. Bergaya, G. Lagaly (Eds.), Developments in Clay Science, vol. 5A, Elsevier, Amsterdam, 2013, pp. 523–557. [27] A. Pérez, M.A. Centeno, J.A. Odriozola, R. Molina, S. Moreno, Catal. Today 133–135 (2008) 526–529. [28] V. Singh, V. Sapehiyia, G.L. Kad, J. Mol. Catal. A 210 (2004) 119–124. [29] D.M. Gómez, J.M. Gatica, J.C. Hernández-Garrido, G.A. Cifredo, M. Montes, O. Sanz, J.M. Rebled, H. Vidal, Appl. Catal. B 144 (2014) 425–434. [30] A. Cybulski, J.A. Moulijn, Structured Catalysis and Reactors, Marcel Dekker, Inc., New York, 1998. [31] J.M. Gatica, H. Vidal, Use of clays to manufacture honeycomb monoliths for pollution control applications, in: J.P. Humphrey, D.E. Boyd (Eds.), Clay: Types, Properties and Uses, Nova Science Publishers Inc., New York, 2011, pp. 253–274. [32] S. Harti, G. Cifredo, J.M. Gatica, H. Vidal, T. Chafik, Appl. Clay Sci. 36 (2007) 287–296. [33] T. Chafik, S. Harti, G. Cifredo, J.M. Gatica, H. Vidal, J. Hazard. Mater. 170 (2009) 87–95. [34] G. Cifredo, S. Harti, J.M. Gatica, H. Vidal, Appl. Clay Sci. 47 (2010) 392–399. [35] J.M. Gatica, H. Vidal, J. Hazard. Mater. 181 (2010) 9–18. [36] J.M. Gatica, D.M. Gómez, S. Harti, H. Vidal, Appl. Surf. Sci. 277 (2013) 242–248.

11