A highly dispersed Pd–Mg-SiO2 catalyst active for methanation of CO2

ct 6-50 greg sion m in he P epa con e M syn an abu inter dehydr liar Sa ), DG of ap a catalyst to achieve acceptable rates and selectivities. Extensive studies have been conducted on several metal-based catalytic sys- tems using Ni [4–7], Ru [8–11], and Rh [12] as well as metal oxides [13,14]. A unique activity has been observed using bimetallic and bifunctional catalyst concepts. A Ni(II)-Ferrite catalyst with a spi- nel structure has been shown to be active for CO2 methanation with a high selectivity at 300 �C [15]; however, the ultra-fine par- cursor is soluble. The droplets are dispersed in a continuous oil phase. By the addition of precursors from several species, the aggregate solids formed within the micelles can be used to directly contact multiple species and form uniformly dispersed bimetallic or mixed oxide nanoparticles [21,22]. He et al. prepared CeO2/ ZnO mixed oxide catalysts using reverse microemulsions for an oxidative coupling of CH4 with CO2 which showed significantly higher activity than CeO2/ZnO catalysts prepared by conventional impregnation methods [22]. Other catalysts synthesized using re- verse microemulsions have shown improved activities compared * Corresponding author. Journal of Catalysis xxx (2009) xxx–xxx Contents lists availab f w.e ARTICLE IN PRESS E-mail address: mcfar@engineering.ucsb.edu (E.W. McFarland). production of synthetic natural gas [3] and the removal of trace amounts of CO2 in hydrogen feeds for ammonia synthesis. There have been proposed applications for space travel to obtain drinking water by combining the on-board hydrogen rocket fuel with carbon dioxide produced by the astronauts. In all applications, increasing the rate and selectivity of the reduction pathway is highly desirable. The methanation reaction is exothermic; however, an 8-elec- tron process is required to reduce the fully oxidized carbon to methane and there are significant kinetic limitations which require the metal oxide to produce water and reduce the metal species leaving behind oxygen vacancies which are the sites for CO2 reduction. For efficient utilization of bifunctional pathways intimate mix- ing of the heterospecies is required and phase separation is mini- mized. Recently, reverse microemulsion methods have been utilized to achieve highly dispersed catalysts [19,20]. In a reverse microemulsion synthesis, the reverse micelles form nanometer- size droplets with an aqueous phase core in which the catalyst pre- 1. Introduction The potential utilization of CO2 as chemical feedstock is of widespread has been used as an oxidant for the zene to styrene [2], and in a fami hydrogenation of CO2, CO2(g) + 4H2(g)! CH4(g) + 2H2O(g The methanation of CO2 has a range 0021-9517/$ - see front matter Published by Elsevier doi:10.1016/j.jcat.2009.05.018 Please cite this article in press as: J.-N. Park, E doi:10.1016/j.jcat.2009.05.018 ndant and inexpensive est [1]. Carbon dioxide ogenation of ethylben- batier reaction for the 0298K = �27 kcal mol�1. plications including the ticulate catalysts were unstable and rapidly sintered. Sehested et al. have demonstrated that a Ni–Fe alloy catalyst is more selec- tive for the conversion of CO2 to CH4 than a pure nickel catalyst at 330 �C in excess hydrogen [16]. de Leitenburg et al. have investi- gated why Ru or Rh supported on CeO2 produces only CH4 as the reaction product whereas Pt, Pd, and Ir give both CO and CH4 over a range of temperatures up to 630 �C showing the maximum yield of CH4 at approximately 450 �C [14]. A bifunctional mechanism has been invoked to explain the behavior of iron- and ceria-containing catalysts [17,18]. The hydrogen is thought to react with oxygen in A highly dispersed Pd–Mg/SiO2 catalyst a Jung-Nam Park a,b, Eric W. McFarland a,* aDepartment of Chemical Engineering, University of California, Santa Barbara, CA 9310 bDepartment of Chemistry, Sungkyunkwan University, Suwon 440-746, South Korea a r t i c l e i n f o Article history: Received 10 February 2009 Revised 15 May 2009 Accepted 22 May 2009 Available online xxxx Keywords: CO2 activation Methanation Pd Mg Reverse microemulsion a b s t r a c t A catalyst formed as an ag using a reverse microemul taining particles 5 to 10 n activity and selectivity of t to several other catalyst pr to CH4 at a carbon dioxide CO2 reduction to CO, and th tive. The results support a Journal o journal homepage: ww Inc. .W. McFarland, A highly dispe ive for methanation of CO2 80, USA ate of highly dispersed palladium and magnesium in silica was prepared synthesis. After calcination, the catalyst consists predominately of Pd con- size distributed within an amorphous oxide of Mg and Si. The catalytic d–Mg/SiO2 catalyst for CO2 methanation were characterized and compared rations. At 450 �C, the Pd–Mg/SiO2 catalyst had greater than 95% selectivity version of 59%, while similar catalysts without Mg have activity only for g- and Si-containing oxide (without a transition metal) are relatively inac- ergistic effect between the Pd and the Mg/Si oxide. Published by Elsevier Inc. le at ScienceDirect Catalysis lsevier .com/locate / jcat rsed Pd–Mg/SiO2 catalyst active for methanation of CO2, J. Catal. (2009), SiO2 catalysts synthesis was from mixed solutions of the Pd(NO3)2�H2O (Aldrich, 7 mL, 0.028 M) with Ni(NO3)2�H2O (Al- diffraction, XRD, was performed on selected samples over the scat- tering range of 30 6 2h 6 75 with steps of 0.017� using CuKa radi- ation (Philips X’Pert Pro X-ray diffractometer). The surface area was estimated using BET analysis of the N2 adsorption isotherms (Micromeritics, Model ASAP 2400). 2.3. Catalytic activity Methanation of CO2 was performed with 0.1 g of catalyst (40 to 60 mesh) in a 6 mm diameter fixed bed reactor. The reactions were performed at atmospheric pressure over a temperature range of 25 to 450 �C using a mixture of carbon dioxide and hydrogen at a ratio 3 �1 The catalyst loadings and BET surface areas are listed out in Ta- nal of Catalysis xxx (2009) xxx–xxx ARTICLE IN PRESS drich), Fe(NO3)2�H2O (Aldrich), or Li(NO3)2�H2O (Aldrich, 3 mL, 0.169 M), respectively. The silica impregnated with Pd catalyst, Pd/SiO2 (IMP), was synthesized on silica spheres produced from a reverse microemulsion (ME) identical to that described above for the Pd/SiO2 (ME), with the Pd precursor replaced with NH4OH solu- tion and without a reducing agent [30]. SiO2 spheres (0.35 g) were added to a solution of Pd(NO3)2�H2O (0.05 g) in 40 ml of water and were stirred for 4 h at 80 �C then dried at 110 �C and calcined at 550 �C in air for 6 h; the heating rate was 1 �C min�1. The Pd load- ing was 6.2 wt% and the loading of the second metal was 3.6 wt%. 2.2. Catalyst characterization The catalyst morphology was investigated using transmission electron microscopy, TEM (JEN2100F, JEOL) and the elemental dis- tribution was monitored with energy dispersive X-ray spectros- copy, EDS, with a system integrated to the TEM (Oxford INCA). to those synthesized by conventional impregnation methods for applications in hydrogenation [23] and oxidation [24]. We are investigating Pd–Mg/SiO2 as a bifunctional catalyst for CO2 methanation motivated by the properties of Pd to dissociate molecular hydrogen [25,26] and make available hydrogen atoms for the subsequent transfer and reaction with activated surface car- bonate species formed by the reaction of CO2 on a Mg-containing oxide [27–29]. The conceptual approach is to provide an alterna- tive pathway to potentially minimize the CO byproduct by using metal oxides that inhibit CO desorption. To create intermixed Pd and Mg sites, the reverse microemulsion synthesis route is used. In this report, we address the following questions: (1) Is a highly dispersed and intermixed Pd–Mg/SiO2 catalyst active for CO2 methanation? (2) What is the sensitivity of the catalyst perfor- mance to the synthesis method? (3) How does the activity and selectivity compare to the individual Pd and Mg oxide components and other methanation catalysts? 2. Experimental 2.1. Catalyst synthesis The Pd–Mg/SiO2 catalyst was synthesized from a reverse micro- emulsion (ME) by adding a mixture of Pd(NO3)2�H2O (Aldrich, 7 ml, 0.028 M) and a Mg(NO3)2�H2O (Aldrich, 3 mL, 0.169 M) to a vigor- ously stirred solution containing a non-ionic surfactant (40 mL, Igepal CO-520, Aldrich) and cyclohexane (100 mL, Fisher Scien- tific). While stirring, Hydrazine (64 lL, Aldrich 98%) was added and the mixture was stirred for another hour. A NH4OH solution (EMD 28%) was then added to adjust the pH to 11. After another hour of stirring a mixture of TEOS (tetraethyl orthosilicate, 0.995 mL, Acros organic) and C18 TMS (n-octadecyl trimethoxysi- lane, 0.357 mL, Gelest Inc.) was added dropwise. Hydrolysis and condensation of the silica precursors were allowed to proceed in the stirred mixture for 3 days at 20 �C. After condensation, the Pd–Mg/SiO2 was precipitated and washed with ethanol. The etha- nol wash followed by centrifugation was repeated thrice. The Pd– Mg/SiO2 was dried at 100 �C and was calcined at 550 �C in air for 6 h; the heating rate was 1 �C min�1. The Pd/SiO2 (ME) [30], Mg/ SiO2, and Ni/SiO2 catalysts were synthesized using nitrate precur- sors; Pd(NO3)2�H2O (Aldrich, 10 mL, 0.02 M), Mg(NO3)2�H2O (Al- drich, 10 mL, 0.02 M), and a Ni(NO3)2�H2O (Aldrich, 10 mL, 0.02 M), respectively. For the Pd–Ni/SiO2, Pd–Fe/SiO2, and Pd–Li/ 2 J.-N. Park, E.W. McFarland / Jour To determine the particle size distribution and average particle size, 100 particles were measured from the TEM images. Errors were taken as the standard deviation of the particle sizes. X-ray Please cite this article in press as: J.-N. Park, E.W. McFarland, A highly dispe doi:10.1016/j.jcat.2009.05.018 ble 1. The surface areas of the Pd/SiO2 and Pd(Imp)/SiO2 were found to be 185 and 134 m2/g, respectively. The surface areas of catalysts synthesized from reverse microemulsions were in the range 110 to 190 m2 g�1. The morphology of the Pd–Mg/SiO2 catalysts characterized by TEM is shown in Fig. 1. The micrographs show that the as-synthe- sized Pd–Mg/SiO2 catalyst after calcination at 550 �C in air for 6 h contains well-dispersed electron-dense particles identified as con- taining Pd by EDS within a matrix of less dense solid that does not appear crystalline, Fig. 1a. The particles appear uniform in shape with a diameter of approximately 5.5 ± 2.0 nm. Features within the image are consistent with the remnants of silica shells sur- rounding the electron-dense centers, however, aggregation has oc- curred. After reaction with CO2 and hydrogen for 10 h, the particles remain well dispersed and though the presence of larger particles is observed, a large fraction of the Pd-containing nanoparticles Table 1 Catalyst loadings and BET surface areas. Catalyst Pd (wt%) Second metal (wt%) BET surface area (m2 g�1) Pd/SiO2 Pd: 6.2 – 185 Pd(Imp)/SiO2a Pd: 6.2 – 134 Mg/SiO2 Mg: 6.2 – 123 Pd–Mg/SiO2 Pd: 6.2 Mg: 3.6 189 Mg(Imp)/Pd/SiO2b Pd: 6.2 Mg: 3.6 80 Pd–Fe/SiO2 Pd: 6.2 Fe: 3.6 188 Pd–Ni/SiO2 Pd: 6.2 Ni: 3.6 155 Ni/SiO2 Ni: 6.2 – 169 Pd–Li/SiO2 Pd: 6.2 Li: 3.6 109 of CO2: H2 = 1:4 (flow rate = 10.2 cm min ) mixed with 2 cm3 min�1 Ar. The space time was 1.1 s. The catalyst samples were pretreated at 450 �C in H2 prior to reaction. The product gas stream was analyzed by parallel gas chromatography (GC, SRI, Inc.) with a 5 Å molecular sieve column, and mass spectrometry (SRS, Inc.). To determine conversion and selectivity, the products were collected after one hour of steady-state operation at each temperature. The conversions of carbon dioxide and hydrogen were determined from the ratios of the difference in the molar flow rates in the feed minus the product to the feed molar flow rate. The molar flow rates were assumed proportional to their concentration in the gas streams. Selectivities were calculated as moles of indi- vidual products produced per mole of total products (methane plus carbon monoxide) produced. The yields were determined as the product of the conversion and the selectivity. For temperature-pro- gramed methanation, the heating rate was 7.5 �C min�1. 3. Results and discussion 3.1. Catalyst characterization a Palladium was impregnated on SiO2 spheres. b Magnesium was impregnated on Pd/SiO2 from a reverse microemulsion. rsed Pd–Mg/SiO2 catalyst active for methanation of CO2, J. Catal. (2009), nal o ARTICLE IN PRESS 30 35 40 45 50 55 60 65 70 75 In te ns ity [a . u .] 2 θ Pd-Li/SiO2 Ni/SiO2 Pd-Ni/SiO2 Pd-Fe/SiO2 Mg(Imp)/Pd/SiO2 Pd-Mg/SiO2 Mg/SiO2 Pd(Imp)/SiO2 Pd/SiO2***** *: PdO(a) # : Pd * Pd-Mg/SiO2 (1) as synthesized (2) H pretreated : PdO(b) J.-N. Park, E.W. McFarland / Jour maintain a size of approximately 5.5 nm, Fig. 1b. We speculate that there is a relatively small population of Pd particles that were not fully encapsulated by the Mg/Si oxide that were able to sinter; however, there remains a significant population of Pd-containing nanoparticles that remain stable. EDS performed within localized �20 nm regions of the image, Fig. 1c and d, indicate the presence of Pd, Mg, and Si. The TEM images are consistent with a stable well-dispersed aggregate of Pd-containing nanoparticles in an amorphous oxide matrix. The XRD data for the as-synthesized catalysts after calcination at 550 �C in air for 6 h are shown in Fig. 2a. Broad reflections at approximately 25� (not shown) were consistent with an amor- phous metal oxide phase or very small nanocrystallites. When present, reflections for the Pd phase were consistent with those for palladium oxide (PdO (101) at 33.9� 2h and Pd (110) at 41.9� 2h). We have previously shown the presence of the same PdO phase in as-synthesized Pd/SiO2 and Pd(Imp)/SiO2 catalysts [30]. The palladium phase in Pd–X/SiO2 (X = Li, Mg, Fe, Ni) was also observed to be PdO and the absence of reflections for the sec- ond metal (X) phase is consistent with an amorphous or nano- crystalline phase. There was no evidence for a crystalline phase of Ni or Mg in the Ni/SiO2 or Mg/SiO2 catalysts calcined at 550 �C. An estimate of the size of the Pd particles was made using the Scherrer equation and found to be in reasonable agreement 30 35 40 45 50 55 60 65 70 75 ## **** (3) (2) 2 θ In te ns ity [a .u] 2 (3) after reaction (1) * # Fig. 1. (a) XRD data obtained from all the catalysts studied after initial synthesis and calcination, (b) XRD data obtained from the Pd–Mg/SiO2 catalyst as synthesized (and calcined), following H2 pretreated, and after reaction for approximately 10 h. Please cite this article in press as: J.-N. Park, E.W. McFarland, A highly dispe doi:10.1016/j.jcat.2009.05.018 with the TEM’s at approximately 7 to 10 nm. Fig. 2b shows the XRD data from the Pd–Mg/SiO2 catalysts as synthesized, after H2 pretreated, and after the methanation reaction. Not surprisingly, the PdO is reduced to Pd metal after H2 pretreatment and after 10 h under methanation reaction conditions the metallic Pd phase remains (Pd (111) at 40.1� 2h and (200) at 46.7� 2h). XPS per- formed ex situ before and after reaction (data not shown) was consistent with a metallic Pd and oxide containing Mg and Si. The particle sizes of the Pd aggregates in Pd–Mg/SiO2 after reac- tion for more than 10 hours as calculated by the Scherrer equation from the XRD data indicate that there is an increase in the Pd phase size from �10 nm in the as-synthesized catalyst to �19 nm after reaction. This may represent the superposition of the sharper reflections from the larger sintered particles on the broader reflection from the smaller, stable aggregates observed by TEM. 3.2. Catalyst activity Table 2 shows the data for the conversions of CO2 and H2 and the selectivities and yields to CH4 and CO over various catalysts at 450 �C. At 450 �C, Pd/SiO2 is active for CO2 reduction to CO in the presence of H2 and Mg/SiO2 (without Pd) is relatively inactive; however, the Pd–Mg/SiO2 catalyst had greater than 95% selectivity to CH4 at a carbon dioxide conversion of 59%. The CO2 and H2 con- version on the Pd–Mg/SiO2 catalyst synthesized from the reverse microemulsion was significantly higher than that observed on the Pd/SiO2, Mg/SiO2, or the other catalysts. The compositionally identical Mg/Pd/SiO2 catalyst synthesized by Mg impregnation of the Pd/SiO2 catalyst (made by reverse microemulsion) was less ac- tive and less selective than when both Mg and Pd were introduced simultaneously in the reverse microemulsion. The Pd–Mg/SiO2 cat- alyst synthesized from the reverse microemulsion has significantly greater contact between Pd-containing nanoparticles and Mg-con- taining oxide. The catalyst made from the reverse microemulsion also showed larger CO2 conversion and methane selectivity com- pared to the same catalyst prepared by impregnation [19,20]. The Pd–Ni/SiO2 catalyst showed higher conversion and selectivity to CH4 compared to Ni/SiO2. On the other hand, Pd–Fe/SiO2 showed the highest selectivity to CO (97%) at high CO2 conversion (45%). The overall conversion of CO2 for the several catalysts followed the general ordering Pd–Mg/SiO2 > Pd–Ni/SiO2 > Pd–Fe/SiO2 > Pd– Li/SiO2 > Pd/SiO2 > Ni/SiO2�Mg/SiO2, while the selectivity to CH4 decreased in the following order Pd–Mg/SiO2 > Pd–Ni/SiO2 > Pd– Li/SiO2 > Mg(Imp)/Pd/SiO2 > Ni/SiO2 > Pd–Li/SiO2 > Pd/SiO2 = Mg/ SiO2. The data in Table 2 are consistent with the reaction stoichiom- etries and the requisite increase in H2 conversion is coincident with an increase in selectivity for methane production. The reac- tion proposed by Choe et al. [31] in Scheme 1suggests that the alternative product, CO, is likely produced as an intermediate dur- ing methanation. The stability of the CO moiety on the catalyst will determine whether or not the CO will desorb or progress for fur- ther reduction. The CO dissociation is thought to be rate determin- ing for the remaining reduction steps. It was observed that with the carbonate forming catalyst combinations, Pd–X/SiO2 (X = Mg, Li) there was a high selectivity to CH4 (95%, 88%) and a correspond- ing higher H2 conversion (27%, 22%) as the intermediate carbonate XO2–CO is stabilized and CO desorption is inhibited. The Pd–Fe/ SiO2 catalyst is active for CO2 activation at high conversion; how- ever, unlike the Pd–Mg/SiO2 catalyst, there is no means for stabiliz- ing the carbonate on the Fe/Si oxide. The hydrogen will react with the hematite surface oxygen to produce water and form oxygen f Catalysis xxx (2009) xxx–xxx 3 vacancies which will activate additional CO2 to fill the vacancy and produce CO. The pathway to further dissociate CO does not compete significantly with CO desorption. Utilization of H2 is low rsed Pd–Mg/SiO2 catalyst active for methanation of CO2, J. Catal. (2009), nal o ARTICLE IN PRESS 4 J.-N. Park, E.W. McFarland / Jour since the formation of CO requires only one H2 while that of methanation requires four. The activities of the individual components of the Pd–Mg/SiO2 catalyst are also listed out in Table 2. The Pd/SiO2 catalyst was rea- sonably active for CO2 reduction to CO; however, an intermediate carbonate is not anticipated and minimal methane production was observed. Carbon dioxide will only weakly adsorb molecularly Fig. 2. Images obtained by TEM of the Pd–Mg/SiO2 catalyst (a) before, and (b) after reac particles. EDS results from the regions indicated in the TEM’s are shown below with the Table 2 Conversion, selectivity, and yield for the reaction of CO2 and H2 over 0.1 g of the studied catalysts at 450 �C. Catalyst CO2 conversion (%) H2 conversion (%) Selectivity (%) Yield (%) CH4 CO CH4 CO Pd/SiO2 40.8 11.4 10.4 89.6 4.3 36.5 Pd(Imp)/SiO2 40.6 9.6 6.5 93.5 2.6 38.0 Mg/SiO2 0.8 6.7 10.3 89.7 0.1 0.7 Pd–Mg/SiO2 59.2 26.9 95.3 4.7 56.4 2.8 Mg(Imp)/Pd/SiO2 40.0 15.3 76.2 23.8 30.4 9.5 Pd–Fe/SiO2 44.7 5.7 2.8 97.2 1.3 43.4 Pd–Ni/SiO2 50.5 23.4 89.0 11.0 44.9 5.6 Ni/SiO2 36.8 14.8 81.8 18.2 30.1 7.0 Pd–Li