Applied Catalysis B: Environmental 20 (1999) 67±73
A comparison between Cu-ZSM-5, Cu±S-1 and Cu±mesoporoussilica±alumina as catalysts for NO decomposition Giuliano Morettia,*, Carlo Dossib, Achille Fusib, Sandro Recchiab, Rinaldo Psarob a
Centro CNR ``SACSO'', Dipartimento di Chimica, UniversitaÁ ``La Sapienza'', Piazzale A. Moro 5, 00185 Rome, Italy Centro CNR ``CSSSCMTBSO'', Dipartimento di Chimica Inorganica, Metallorganica e Analitica, UniversitaÁ di Milano, Via Venezian 21, 20133 Milan, Italy
b
Received 7 June 1998; received in revised form 6 September 1998; accepted 27 September 1998
Abstract We prepared H-ZSM-5 (Si/Al80), amorphous mesoporous silica±alumina (MSA, Si/Al90), and Silicalite-1 (S-1) according to the methods described by Bellussi and co-workers, and compared their ion-exchange capacity for copper ions. The Cu-ZSM-5, Cu±MSA and Cu±S-1 samples thus obtained have been investigated as catalysts for the NO decomposition reaction at 773 K. We found that using copper acetate solutions with concentrations in the range 0.008 M[Cu2]0.1 M, both at room temperature and at 323 K, it is very easy to prepare over-exchanged Cu-ZSM-5, Cu±S-1 and Cu±MSA catalysts. XRD and Vis±UV DRS techniques show that after thermal treatments of the fresh samples in air at 823 K for 4 h no segregation of CuO phase occurs, suggesting the presence of low nuclearity [CunOx(OH)y]q species (q2(nÿx)ÿy0). These results were con®rmed by TPR studies. At 773 K only over-exchanged Cu-ZSM-5 catalysts showed NO decomposition activity (NO 1% in He, W/F0.1 gs/cm3) with a turnover frequency essentially determined by the Si/Al ratio, in agreement with previous literature data. Instead, the activity of Cu±S-1 and Cu±MSA catalysts was not measurable under our experimental conditions. The present results con®rm that the framework topology and the presence of framework AlOÿ 4 species are fundamental to develop active copper species for NO decomposition. It is demonstrated that the active sites in Cu-ZSM-5 consist of copper species strongly anchored to framework AlOÿ 4 species. The most active sites, as they occur only on ZSM-5 support with the lower Si/Al atomic ratios, might consist of dimeric Cu species (Cu Cu2 Oÿ) strongly anchored to next-nearestneighbour framework AlO4ÿ species. # 1999 Elsevier Science B.V. All rights reserved. Keywords: Nitric oxide decomposition; Cu-ZSM-5; Cu±silicalite; Cu±mesoporous-silica±alumina
1. Introduction Since the discovery of the remarkable NO decomposition activity of Cu-ZSM-5 catalysts by Iwamoto *Corresponding author: Tel.: +39-06-49913759; fax: +39-06490324; e-mail:
[email protected]
and co-workers in 1986 [1] a lot of effort has been devoted to understand the reasons of the peculiarity of this system in comparison to other copper containing systems or transition metal ions exchanged in ZSM-5 [1±3]. Iwamoto et al. [1,2] reported that overexchanged Cu-ZSM-5 catalysts (Cu/Al atomic ratio >0.5) were the most active and stable catalysts for the
0926-3373/99/$ ± see front matter # 1999 Elsevier Science B.V. All rights reserved. PII: S0926-3373(98)00096-4
68
G. Moretti et al. / Applied Catalysis B: Environmental 20 (1999) 67±73
NO decomposition. These results were con®rmed by Li and Hall [3] which also investigated the effects of NO partial pressure and the presence of O2 in the feed on the kinetics of the reaction. The effects of the Si/Al atomic ratio of the parent ZSM-5 used to prepare the Cu-ZSM-5 model catalysts were investigated by our group [4±8]. It was demonstrated that the most active catalysts are those with the lowest Si/Al atomic ratios and Cu exchange levels in the range 90±150%. These results have also lead to the development of a structural model for the copper active sites in Cu-ZSM-5 catalysts [4]. The most active sites should be those containing two close copper ions linked to next-nearest-neighbours framework AlOÿ 4 species. This can explain the higher turnover frequencies observed for the catalysts with the higher Al framework content. Recently, we discovered by adsorption measurements and diffuse re¯ectance FT-IR techniques that sites of this kind are able to chemisorb dinitrogen at low temperature [9]. Our model is also able to describe the behaviour of active sites consisting of low nuclearity or monomeric copper species (Cu2Oÿ) linked to the isolated framework AlOÿ 4 species (the principal active species on catalysts with low copper loading and/or high Si/Al ratio). These species have much lower activity for NO decomposition into N2 and O2 with respect to dimeric Cu species (Cu Cu2 Oÿ) strongly anchored to next-nearest-neighbour framework AlOÿ 4 species. For catalysts at constant copper loading, with Cu/Al 0.5, the model shows that the activity should decrease increasing the Al content in the ZSM-5 framework, as observed experimentally [4]. It is clear that being able to demonstrate the structure of the active sites, and learning the limitations of the Cu-ZSM-5, could lead to develop new more active and stable catalysts which could ®nd practical application. In fact, it has been reported that the increase of 1 order of magnitude in the turnover frequency could lead to a practical catalyst [10,11]. As remarked by Armor [11], NO decomposition still offers a very attractive approach to NOx removal. However, since any combustion process is going to produce almost 16% water vapour, one must focus on a catalyst that is stable for long times in such wet environments. In this paper, we report a comparison between Cu-ZSM-5 (Si/Al80), Cu±MSA (amorphous mesoporous silica±alumina, Si/Al90) and
Cu±S-1 model catalysts, prepared by the same ion exchange procedure and with similar Cu loadings. The comparison between zeolite and silica±alumina catalysts with high Si/Al ratios is more relevant for practical applications in wet environments because, as reported by Sano and co-workers [12], these catalysts are the most stable under hydrothermal conditions. 2. Experimental 2.1. Catalysts preparation The zeolite H-ZSM-5 (Si/Al80) was prepared according to the procedure reported by Bellussi et al. [13]. A solution of 1 g of Al(NO3)39H2O (C. ErbaRPE-ACS) in 20 g of ethyl alcohol was added to 45 g of tetraethylsilicate (TES) (Fluka-purum). This solution was poured in a 400 ml pyrex glass vessel containing 100 g of a 20% aqueous solution of tetrapropylammonium hydroxide (TPA-OH) (Flukapurum). The resulting mixture was kept at 333 K for 3 h and then heated under autogenous pressure in a 350 ml stainless steel autoclave in an oven at 448 K, without stirring, for 24 h. The crystalline product was separated from the liquid by centrifugation (4500 rpm 30 min), washed several times with distilled water, dried 2 h at 383 K and ®nally calcined at 823 K for 5 h. Silicalite-1 (S-1), a pure-silica ZSM-5, with the same framework topology of ZSM-5 (MFI), was prepared according to the same procedure used for HZSM-5. The amorphous mesoporous silica±alumina (MSA) with Si/Al90 was prepared according to [14]. A clear sol with a pH around 12 was obtained by adding TES to a TPA-OH solution containing aluminium isopropoxide (Fluka-puriss.). The molar ratios of the components were the following: 180:1:18:3600SiO2:Al2O3:TPA-OH:H2O. The solution was charged in a stainless steel autoclave and kept at 453 K, without stirring, under autogenous pressure for 30 min. The clear sol was then transformed in a clear dense gel by partial evaporation of the solvent. The microcrystalline powder was recovered and washed by centrifugation. The sample was dried at 363 K for 24 h in air and ®nally calcined at 823 K in air for 10 h.
G. Moretti et al. / Applied Catalysis B: Environmental 20 (1999) 67±73
69
The copper containing catalysts were prepared by ion exchange using cupric acetate solutions. In the preparations 2 g of support (H-ZSM-5, MSA or S-1) was treated with 250 ml of solution of given copper concentration for 2 h under stirring at room temperature or at 323 K. The samples after ®ltration were washed several times with distilled water, dried at 383 K for 2 h, and stored over a saturated solution of NH4Cl at room temperature (relative humidity79%). Chemical analyses of the catalysts were performed by atomic absorption (Varian SpectrAA-30). Details about the prepared catalysts are reported in Table 1. Cu±MSA and Cu-ZSM-5 samples will be identi®ed by their Cu% of exchange while for Cu±S-1 samples the Cu wt% values will be quoted.
pumped capillary inlet. Temperature programmed reduction (TPR) studies were done in ¯owing H2 (8%)/Ar mixture on a modi®ed computer-controlled Micromeritics Pulse Chemisorb 2700 apparatus. The temperature was increased at 8 K minÿ1 from 200 to 773 K. Before the TPR run the fresh samples were treated in 2% O2 in He from room temperature to 773 K at 3 K minÿ1 and then cooled in Ar. BET surface area, pore size distribution and total pore volume of MSA were determined from N2 adsorption±desorption isotherms at 77 K using a manual volumetric technique. The samples were preheated 4 h at 473 K and 10ÿ2 Pa.
2.2. Catalysts characterisation
The activity measurements were made at atmospheric pressure using 1% NO in He in a continuous-¯ow microreactor interfaced to a mass spectrometer. The NO decomposition into N2 and O2 was studied mainly at 773 K. Before each experiment the catalysts were treated in He for several hours at 773 K. The turnover frequencies, expressed as mol NO converted to N2 per mol Cu per second, were calculated for conversion 0.5). Similar results were reported by Parrillo et al. [15] for HZSM-5 with Si/Al35 and for S-1 samples synthesised using an organic template. The low pH values (
