Low-Temperature Selective Catalytic Reduction of NO with NH3 over

Mar 27, 2008 - Sounak Roy,† B. Viswanath,‡ M. S. Hegde,† and Giridhar Madras*,§. Solid State ... The catalysts, Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe,...
0 downloads 0 Views 705KB Size
6002

J. Phys. Chem. C 2008, 112, 6002-6012

Low-Temperature Selective Catalytic Reduction of NO with NH3 over Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe, Co, Cu) Sounak Roy,† B. Viswanath,‡ M. S. Hegde,† and Giridhar Madras*,§ Solid State and Structural Chemistry Unit, Material Research Center, and Chemical Engineering Department, Indian Institute of Science, Bangalore 560012, India ReceiVed: December 13, 2007; In Final Form: February 7, 2008

The catalysts, Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe, Co, Cu), were synthesized in anatase phase by solution combustion. Selective catalytic reduction (SCR) of NO with NH3 was investigated over these catalysts. The reaction occurred at the lowest temperature over Ti0.9Mn0.1O2-δ, but the selectivity for N2 was highest over Ti0.9Fe0.1O2-δ. Therefore, both Mn and Fe were substituted in TiO2 (Ti0.9Mn0.05Fe0.05O2-δ). The reaction occurred at low temperature with a high selectivity over this catalyst. In order to understand the reaction mechanism and the nature of the active sites, temperature programmed desorption (TPD) of NH3 and hydrogen uptake studies were carried out. The relation between the Lewis acid sites and SCR window and the relation between Bronsted acid sites and low temperature was established. The order of the SCR reaction with respect to NO, NH3, and O2 was also investigated. It was also shown that the N2 selectivity of the SCR reaction has a strong inverse correlation with the oxidation of ammonia.

Introduction Selective catalytic reduction (SCR) of NOx by NH3 has been used extensively to treat stationary exhausts.1,2 The stoichiometric reaction of NO reduction by NH3 in presence of excess oxygen is

4NH3 + 4NO + O2 f 4N2 + 6H2O ∆G°298 ) -1627 kJ/mol (1) However, the other reactions that result in undesirable products are also thermodynamically favorable. These reactions are

4NH3 + 5O2 f 4NO + 6H2O ∆G°298 ) -960 kJ/mol (2) 4NH3 + 4NO + 3O2 f 4N2O + 6H2O ∆G°298 ) -196 kJ/mol (3) 2NH3 + 2O2 f N2O + 3H2O ∆G°298 ) -1103 kJ/mol (4) The common industrial catalysts for SCR of NO are TiO2 anatase support dispersed with 0.5-3 wt % of V2O5 and 5-10 wt % MoO3 or WO3. The overall surface areas of the catalysts are around 50-100 m2/g, and the reaction occurs around 300400 °C. Because of this high operating temperature, the catalyst bed must be located upstream of the desulfurizer and/or of the particulate control device to avoid reheating of the flue gas. This makes these catalysts susceptible to deactivation from high concentrations of sulfur dioxide and dust. In addition, retrofitting SCR devices into existing systems for flue gas cleaning is costly because space and access in many power plants are extremely limited. Therefore, there is a great interest in developing active SCR catalysts that work at low temperature. The second feature * Corresponding author. E-mail: [email protected]. Phone: +91-80-2293-2321. Fax: +91-80-2360-0683. † Solid State and Structural Chemistry Unit. ‡ Material Research Center. § Chemical Engineering Department.

of the reaction is the SCR window. To enhance N2 selectivity in SCR, NH3 to NO conversion (reaction 2) should be suppressed. N2O produced from reactions 3 and 4 is a harmful green-house gas and also needs to be suppressed. Therefore, three main issues namely the lowering of reaction temperature, enhancement of the SCR window and N2 selectivity need to be addressed by new catalysts. Al2O3 supported Mn oxides have been used for SCR reaction. In this catalyst, the oxidation state of the Mn, crystallinity and specific surface area are decisive for the performance of the low-temperature reaction.3,4 Slow conversion of Mn oxides in Al2O3 to MnAl2O4 or Mn2AlO4 may reduce the SCR activity. TiO2 is only weakly and reversibly sulfated in conditions approaching those of the SCR reaction and the stability of sulfates on the TiO2 surface is weaker than on other oxides such as Al2O3 and ZrO2. Therefore, anatase-TiO2 is considered to be the best support. TiO2-supported first row transition metal oxides (Mn, Cr, and Cu) have been identified as low-temperature SCR catalysts.5,6 However, the selectivities of these catalysts are poor but no clear reason has been provided for this observation. Despite extensive studies, active sites in the catalyst, the kinetics, and the reaction mechanism are a matter of debate. For the commercial V2O5/TiO2, the active sites are the vanadium sites; NH3 is adsorbed on the catalyst surface both in the form of molecularly coordinated NH3 and NH4+ ions. An increase in surface V-OH increases NO conversion.7 The reaction order with respect to NO is unity on vanadia-based catalysts.8-11 The reaction is zero order with respect to NH3.9,12 The reaction order with respect to oxygen is generally found in the range of 0-0.5. Many theoretical calculations have shown that the reaction occurs by H2NNO adduct formation, and NH2 + NO is the key reaction in SCR.13-16 The adduct H2NNO can produce selective products like N2 and H2O (reaction 1) and/or unselective products like N2O and H2 (reactions 3 and 4).17 However, the H2N-OO adduct is weakly bound and can produce NO and H2O (reaction 2).18

10.1021/jp7117086 CCC: $40.75 © 2008 American Chemical Society Published on Web 03/27/2008

Catalytic Reduction of NO with NH3 An alternative approach examined in this work is the substitution of first row transition metal ions as active sites in TiO2 matrix. A compound of the type Ti0.94+M0.13+O1.95 would have M3+ and Ti4+ as NH3 adsorption sites and the oxide ion vacancy can act as an adsorption site for NO that can dissociate. This study shows that the reaction can occur at low temperature and with high N2 selectivity over Ti0.9Mn0.05Fe0.05O2-δ. Further, this is the first study that shows a correlation between ammonia oxidation and selectivity in SCR. Experimental Section Synthesis and Characterization. Transition metal ions (Cr, Mn, Fe, Co, and Cu) substituted TiO2 catalysts were prepared by a novel, single step solution combustion method. It is a low temperature initiated process with a high transient temperature. The combustion process is fast (instantaneous), and due to production of huge amount of gases, the products are homogeneous nanocrystalline materials with desired composition and structure. The process utilizes the highly exothermic redox chemical reactions between metal nitrates and fuels, the metathetical (exchange) reaction between reactive compounds or reactions involving redox compounds/mixtures. Combustion synthesis is also known as “self-propagating high-temperature synthesis” (SHS). The combustion mixture for the preparation of 10 atom % M/TiO2 (M ) Cr, Mn, Fe, Co, and Cu) catalysts contained aqueous solution of stoichiometric amount of titanyl nitrate (TiO(NO3)2), respective metal nitrates (like Cr(NO3)3· 9H2O, Mn(NO3)2·4H2O, Fe(NO3)3·9H2O, Co(NO3)2·6H2O, and Cu(NO2)2·3H2O) and glycine (CH2NH2COOH). TiO(OH)2, which was prepared from hydrolysis of TiO(i-pr)4, was dissolved in HNO3 to produce TiO(NO3)2. The solution was introduced in to a muffle furnace maintained at 350 °C. The solution boiled with frothing and foaming with concomitant dehydration. At the point of its complete dehydration, the redox mixture ignited yielding a voluminous finely dispersed solid product. MnTiO3 and FeTiO3 were also prepared by solution combustion method. 5 wt % V2O5/TiO2 was made by conventional impregnation method.19 Here, ammonium vanadate (NH4VO3) was dissolved in 1 M oxalic acid and the blue complex (NH4)2[VO(C2O4)2] was obtained. A calculated amount of pre-synthesized TiO2 was added to this, and the slurry was dried and calcined at 500 °C for 6 h. The catalysts were characterized by X-ray diffraction, highresolution transmission electron microscopy, X-ray photoelectron spectroscopy, and surface analysis. Powder X-ray diffraction (XRD) patterns of catalysts were recorded on a Phillips X’Pert diffractometer using Cu KR radiation at a scan rate of 2θ ) 0.5° min-1. High-resolution transmission electron microscopy (HRTEM) studies were carried out using a TECNAI F30 electron microscope operated at 200 kV and the composition was confirmed by EDX analysis. X-ray photoelectron spectra (XPS) of the as prepared catalysts and used catalysts were recorded on an ESCA-3 Mark II VG scientific spectrometer using Al KR radiation (1486.6 eV). Binding energies reported are with respect to C (1s) at 285 eV. BET surface area was determined by nitrogen adsorption-desorption method at liquid nitrogen temperature using a Quantachrome NOVA 1000 surface area analyzer. The surface area and average pore diameter of the catalysts are shown in Table S1 (see the Supporting Information). H2 Uptake and NH3-TPD Studies. Hydrogen uptake experiments as a function of temperature were carried out with 50 mg of catalyst sample of 40-80 mesh size. The catalyst samples were plugged with ceramic wool in a continuous flow

J. Phys. Chem. C, Vol. 112, No. 15, 2008 6003 of 5% H2 in Ar at 30 cm3 min-1 flow rate. In order to minimize the contribution from adsorbed species, the catalysts are oxidized in situ in oxygen flow followed by degassing in Ar flow to experimental temperature. The catalysts were brought back to room temperature in oxygen environment, degassed with Ar flow and again the temperature was ramped up in H2 flow for the second cycle. The amount of H2 uptake is detected by a TCD detector, which is calibrated against the uptake of H2 with a known amount of CuO. For NH3-TPD studies all the catalysts were first calcined at 500 °C for 3 h. Then, 4.64 vol % NH3 in He was passed for 3 h at room temperature. Helium was passed over the catalyst at 100 cm3 min-1 for degassing, and the temperature was ramped from room temperature to 600 °C at 10 °C min-1. The product gases were analyzed with a quadrupole mass spectrometer. Catalytic Studies. The catalytic studies were carried out in a temperature programmed reaction system equipped with a quadrupole mass spectrometer SX200 (VG Scientific Ltd, England). The catalyst samples of 40-80 mesh were diluted with SiO2 to make catalyst bed volume 0.138 cm3. The catalysts were packed between two glass wool plugs in the center of a microreactor that consists of 4 mm ID, 15 cm length quartz tube inserted into a furnace. The reaction temperature was measured by chromel-alumel thermocouple dipped in the catalyst bed and regulated by a controller. The temperature was ramped at 10 °C-1. Experiments with other ramping rates (5 °C and 15 °C min-1) did not show a significant change in the profile. The gaseous products were sampled through a fine leak valve to an ultrahigh-vacuum (UHV) system housing the quadrupole mass spectrometer. The product gas passing through the mouth of a needle valve was leaked to a stainless steel (SS) chamber from 10-5 Torr to 0.02 Torr, which is pumped by a diffusion pump a 200 L min-1. The sampled gas from the SS chamber at 0.02 Torr was leaked to UHV chamber housing the mass spectrometer from 10-8 to 10-6 Torr. Thus, the dead volume of the sampling gas is replaced 8 to 10 times in 1 s by differential pumping method. The gases leaving the reaction zone at 100 sccm flow rate are detected within 1 s in the mass spectrometer. Gas flow rate was 100 sccm with GHSV of 43 000 h-1. Gases were from M/S Bhuruka Gases Ltd., India with 4.64 vol % NH3 in He, 4.74 vol % NO in He, and 99.9% O2. For SCR reactions, 150 mg of catalyst was taken and the feed gas ratio was NO:NH3:O2 ) 1 vol %:1 vol %:5 vol % (NO and NH3 in 10 000 ppm and O2 in 50 000 ppm) at a flow rate of 100 cm3 min-1. Unless otherwise mentioned, W/FNO was maintained at 201.6 × 103 g s mol-1. Results and Discussion Structural Studies. XRD patterns of Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe, Co, and Cu) are shown in Figure 1a. TiO2 crystallizes in anatase structure. The corresponding transition metal oxide X-ray lines were not found in the XRD pattern. The patterns show broad X-ray line width. Rietveld refinements of the catalysts were carried out taking the substituted noble metals in Ti4+ (4a) position and plotted in Figure S1a-e. The difference plot does not show the corresponding transition metal oxide X-ray lines except for small X-ray lines due to CuO in Ti0.9Cu0.1O2-δ. By careful XRD study we could show that up to 7 atom% Cu2+ ion can be substituted in TiO2. At higher Cu2+ concentration, CuO X-ray lines are observed. The cell parameters of the catalysts are given in Table 1. By substituting the first row transition metals in TiO2 lattice, the cell parameters and the cell volume decrease from Cr to Cu (Figure S2). The lattice parameter contraction is expected due to the decrease in

6004 J. Phys. Chem. C, Vol. 112, No. 15, 2008

Roy et al.

Figure 1. (a) XRD of Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe, Co, and Cu) and Ti0.9Mn0.05Fe0.05O2-δ. (b) Rietveld refinement of Ti0.9Mn0.05Fe0.05O2-δ.

the ionic radii of the transition metal ions. Because Mn and Fe substituted TiO2 were good SCR catalysts for NO, 5 atom% Mn + 5 atom% Fe were substituted for Ti in TiO2 and the Rietveld refinement (Figure 1b) shows complete solid solution formation. FeTiO3 and MnTiO3 synthesized by solution combustion method crystallize in illmenite structure (not shown). These compounds, where Fe or Mn in 2+ state and Ti in 4+ state without any oxide ion vacancies, were synthesized to compare the catalytic performance with Ti0.9M0.1O2-δ (M ) Mn and Fe). 5 wt % V2O5 crystallizes into anatase structure and no X-ray lines due to vanadium oxide was found (not shown) because of its noncrystalline nature.

Figure 2, panels a and b, represents the bright field and highresolution images of Ti0.9Mn0.1O2-δ respectively. The nanoparticles are well dispersed with size of 7-9 nm. TiO2 lattice fringes are observed in Figure 2b, which are of 3.5 Å distance, corresponding to (101) plane. No other lattice fringes corresponding to MnO2 or Mn2O3 were observed. In Figure 2, panels c and d, the bright field and high-resolution image of Ti0.9Fe0.1O2-δ are presented, respectively. The bright field image shows slightly smaller crystalline sizes of Ti0.9Fe0.1O2-δ compared to Ti0.9Mn0.1O2-δ whereas, in the HRTEM no lattice fringes due to Fe-oxides were observed other than TiO2 (101) plane. The average particle sizes of 8-9 nm were observed. Figure 3, panels a and b, shows the bright field and highresolution image of Ti0.9Mn0.05Fe0.05O2-δ, respectively. Fine dispersed particles of around 10 nm size showed only TiO2 (101) lattice fringes. The EDAX shown in Figure 3c confirms the composition which is atomic % Mn ) 4.97, Fe ) 5.06, Ti ) 89.97. Thus TEM and chemical analysis on lattice fringes confirm Ti0.9Mn0.05Fe0.05O2-δ solid solution formation. In Figure 4a, Mn(2p3/2,1/2) core level photoelectron spectra of Ti0.9Mn0.1O2-δ are presented. Mn(2p3/2,1/2) peaks at 641.5 and 652.9 eV indicate Mn in the 3+ state. Core level spectra of Fe(2p3/2) of Ti0.9Fe0.1O2-δ at 711.0 eV show Fe in the 3+ state (Figure 4b). XP spectra of Cr(2p), Co(2p), and Cu(2p) are provided in the supporting Figure S3. Co(2p) peak at 781.9 ev shows that it is in the 3+ state. Further, the Co(2p) region does not show a satellite due to Co2+ as in CoO. Co3+ in low spin does not show a satellite. The Co(2p) state is akin to Co3+ in low spin as in LiCoO2. The Cu(2p3/2) peak is observed at 933.9 eV, corresponding to Cu in the 2+ state. Cu is clearly in the 2+ state as seen from the satellite to main peak ratio of 0.6. The Cr(2p3/2) peak is observed at 579.6 eV, indicating Cr in the 3+ state. In addition to this, magnetic susceptibility of all of the transition metal ion substituted catalysts was measured. Accordingly, Cr, Mn, Fe, and Co are in the 3+ state and Cu is in the 2+ state. Ti(2p3/2) (Figure 4c) for both Ti0.9Mn0.1O2-δ and Ti0.9Fe0.1O2-δ is at same position; 459 eV, indicating that Ti is in the 4+ oxidation state. In other compounds also Ti is in the 4+ state. These characterizations clearly indicate that the transition metals are substituted in the TiO2 lattice in their higher oxidation state. Therefore, some oxide ion vacancies in the lattice need to exist for charge balance. H2 Uptake and NH3-TPD Studies. Figure 5a shows the temperature program reduction (TPR) profiles for unsubstituted TiO2 and Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe, Co, and Cu). Unsubstituted TiO2 shows a very small broad peak above 400 °C corresponding to the reduction of Ti4+ to Ti3+. All metal ion substituted catalysts except Ti0.9Co0.1O2-δ show two different peaks, one at low temperature and the second peak at higher temperature. The first peak is due to the reduction of the substituted metal ion, and the second one is due to conversion of Ti4+ to Ti3+. For Ti0.9Co0.1O2-δ, the reduction of Co is at higher temperature and merges with the peak for Ti4+ reduction. The H2/M molar ratios estimated from the area under each curve

TABLE 1: Rietveld Refined Parameters of Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe, Co, and Cu) and Ti0.9Mn0.05Fe0.05O2-δ catalysts

a/b (Å)

c (Å)

cell volume (Å3)

RBRAGG

RF

χ2

IR (Mn+)

TiO2 Ti0.9Cr0.1O2-d Ti0.9Mn0.1O2-d Ti0.9Fe0.1O2-d Ti0.9Co0.1O2-d Ti0.9Cu0.1O2-d Ti0.9Mn0.05Fe0.05O2-d

3.7904(5) 3.7863(3) 3.7812(1) 3.7800(1) 3.7768(1) 3.7742(1) 3.7813(4)

9.5067(1) 9.4867(4) 9.4853(3) 9.4826(1) 9.4735(4) 9.4711(1) 9.4354(3)

136.58 136.00 135.62 135.49 135.12 134.91 134.91

2.46 4.10 1.02 3.81 6.80 4.25 2.60

1.60 3.34 1.14 3.60 4.94 4.73 2.99

1.20 1.06 1.06 0.78 1.06 0.91 1.09

0.61 (Ti4+) 0.61 (Cr3+) 0.58 (Mn3+) 0.55 (Fe3+) 0.545 (Co3+) 0.57 (Cu2+)

Catalytic Reduction of NO with NH3

J. Phys. Chem. C, Vol. 112, No. 15, 2008 6005

Figure 2. (a) Bright field and (b) high-resolution transmission electron micrograph of Ti0.9Mn0.1O2-δ. (c) Bright field and (d) high-resolution transmission electron micrograph of Ti0.9Fe0.1O2-δ.

Figure 3. (a) Bright field and (b) high-resolution transmission electron micrograph and (c) EDAX of Ti0.9Mn0.05Fe0.05O2-δ.

are 0.5, 0.55, 0.3, 0.62, and 1.3, respectively, for Cr, Mn, Fe, Co, and Cu. If M3+ ions are reduced to M0, the ratio should be

1.5. However, the ratios are lower indicating only partial reduction of the metal ions. Cr in 3+ cannot be reduced to lower

6006 J. Phys. Chem. C, Vol. 112, No. 15, 2008

Roy et al.

Figure 5. (a). H2-TPR of TiO2 and Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe, Co, and Cu). (b). NH3-TPD of Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe, Co, and Cu).

Figure 4. XPS of (a) Mn(2p) of Ti0.9Mn0.1O2-δ, (b) Fe(2p) of Ti0.9Fe0.1O2-δ, and (c) Ti(2p) of Ti0.9Mn0.1O2-δ and Ti0.9Fe0.1O2-δ.

valent state, as it is proved by the low molar ratio of H2 to Cr ()0.5). For Mn and Fe the partial reduction can be from the 3+ to the 2+ state. On the other hand, for the reduction of Cu2+ to Cu0, the molar H2/M ratio should be unity and this confirms that Cu2+ ions in TiO2 reduce to the zerovalent state. From the TPR/H2 curve (Figure 5a), one can observe that the reduction of Ti0.9Mn0.05Fe0.05O2 starts at a slightly lower temperature than that of Ti0.9Mn0.1O2. The compound continued to

get reduced even up to 600 °C. The extent of reduction of Ti0.9Mn0.05Fe0.05O2 is higher than that of Ti0.9Mn0.1O2 and Ti0.9Fe0.1O2. The temperature program desorption profiles of NH3 over the catalysts are shown in Figure 5b. Except for Ti0.9Cr0.1O2-δ, all catalysts show two different NH3 desorption peaks, one at 40-150 °C, the other at 300 °C or above. Ammonia adsorbed on Bronsted acid sites desorbs at a low temperature than that of Lewis-coordinated ammonia during temperature programmed desorption.2,20,21 Thus, two peaks may be attributed to the Bronsted acid sites and Lewis acid sites, respectively. However, there is the possibility of oxidation of strongly chemisorbed NH3

Catalytic Reduction of NO with NH3

J. Phys. Chem. C, Vol. 112, No. 15, 2008 6007

Figure 6. (a) %NO conversion against temperature under SCR condition over Ti0.9M0.1O2-δ. (b) %NO conversion against temperature under SCR condition over Ti0.95Mn0.05O2-δ, Ti0.9Mn0.1O2-δ, Ti0.85Mn0.15O2-δ (c) %NO conversion against temperature under SCR condition over MnTiO3, FeTiO3 and V2O5/TiO2.

with the lattice oxygen and a small amount of N2O and NO were also desorbed, but the concentration was much lower than that of NH3 in our study. However, if we consider the two

different desorption peaks from Bronsted and Lewis acid site, then Ti0.9Cr0.1O2-δ shows only Bronsted acidity and no Lewis acidity, as reported in other studies.22 Bronsted acidity among

6008 J. Phys. Chem. C, Vol. 112, No. 15, 2008

Roy et al.

Figure 8. % Selectivity of N2, N2O and NO over Ti0.9Mn0.1O2-δ and Ti0.9Fe0.1O2-δ in SCR reaction.

Figure 7. Complete TPR profile of SCR reaction over Ti0.9Mn0.1O2-δ and Ti0.9Fe0.1O2-δ.

the catalysts follow the order Mn > Fe ∼ Cr > Co > Cu, whereas the Lewis acidity follows the order, Fe > Mn > Co > Cu . Cr. Catalytic Studies. SelectiVe Catalytic Reduction of NO with NH3 in the Presence of O2. NO reduction by NH3 in the presence of excess oxygen was carried out over 150 mg of Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe, Co, and Cu) with NO:NH3:O2 ) 1:1:5 at a flow rate of 100 cm3 min-1. Figure 6a shows the NO conversion in the presence of various catalysts. The NO conversion occurs at the lowest temperature over Ti0.9Mn0.1O2-δ, and at the highest temperature over Ti0.9Co0.1O2-δ while complete conversion does not take place over Ti0.9Cr0.1O2-δ. Based on T50 (50% NO conversion temperature), the catalytic activity follows the order Mn > Cu > Fe ∼ Cr > Co. Though NO conversion starts at a higher temperature over Ti0.9Fe0.1O2-δ compared to Ti0.9Mn0.1O2-δ, the SCR window is the largest. Similarly NO conversion starts at a low temperature over Ti0.9Cu0.1O2-δ, but the SCR window is narrow. The decrease in NO concentration at higher temperature is due to ammonia oxidation, as will be discussed subsequently. As Ti0.9Mn0.1O2-δ showed the best catalytic activity at low temperature, the Mn concentration was varied from 5 to 15% to determine the SCR activity. Figure 6b shows that the NO conversion in SCR condition over Ti0.95M0.05O2-δ, Ti0.9Mn0.1O2-δ,

and Ti0.85Mn0.15O2-δ. The conversions in the presence of Ti0.9Mn0.1O2-δ and Ti0.85Mn0.15O2-δ, are similar showing 10 %Mn concentration is the optimum concentration for NO conversion in this specified condition. Figure 6c shows the catalytic activity of FeTiO3 as well as MnTiO3 that indicates that NO conversion does not exceed 20%. Therefore, the active phase is the substituted anatase phase and not the illmenite phases. 5 wt % V2O5/TiO2 was also checked for the same reaction in comparatively mild reaction conditions. The reduction of NO with 5 wt % V2O5/TiO2 was investigated with a feed gas ratio of NO:NH3:O2 ) 0.5:0.5:3, and the flow rate was 50 cm3 min-1. The NO conversion is comparable but the SCR window is much narrower to that with Ti0.9Mn0.1O2-δ. Figure 7, panels a and b, show the complete reaction profile of SCR reaction with all of the reactants and products over Ti0.9Mn0.1O2-δ and Ti0.9Fe0.1O2-δ, respectively. The NO conversion starts below 100 °C and is complete at 180 °C over Ti0.9Mn0.1O2-δ. However, undesirable products such as NO and N2O are formed at high temperatures. Though the complete conversion of NO occurs only at 300 °C over Ti0.9Fe0.1O2-δ, the formation of NO and N2O is less compared to that of Ti0.9Mn0.1O2-δ. Figure 8 shows the product selectivity over Ti0.9Mn0.1O2-δ and Ti0.9Fe0.1O2-δ. This will be discussed in detail later. Ammonia Oxidation. The selectivity in a SCR reaction is mainly governed by ammonia oxidation (reactions 2 and 4). The oxidation of NH3 gives rise to NO and N2O, which are undesirable products in a desired SCR reaction. There are two possible mechanisms for NO formation in ammonia oxidation: one is the surface reaction between adsorbed atomic oxygen and nitrogen and the other is between oxygen and the NH fragment in which the desorption of NO is the rate-limiting step.23 N2 is produced mainly from dissociation of NO, which is produced by oxidation of adsorbed NH3.24 If the rate of ammonia oxidation is low, the selectivity for NO and N2O is the lowest (selectivity of N2 is highest). NH3 oxidation was carried out with NH3:O2 ) 1:5 over Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe, Co, and Cu). The feed O2 concentration was kept same as in SCR reaction. Figure 9 shows the NH3 conversion with

Catalytic Reduction of NO with NH3

Figure 9. Light off curve of NH3 oxidation over Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe, Co, and Cu) and Ti0.9Mn0.05Fe0.05O2-δ.

temperature over all the catalysts. The catalytic activity based on T50 follows the order: Cu > Co > Mn ∼ Cr > Fe. Cubased catalysts are generally known for NH3 to NO conversion.25,26 Figure 10a-c show the N2, NO, and N2O selectivity respectively over all of the catalysts. Ti0.9Cu0.1O2-δ shows the maximum selectivity for NO and N2O, whereas Ti0.9Fe0.1O2-δ shows the lowest selectivity for NO and N2O. Thus, the rate of NH3 oxidation over Ti0.9Cu0.1O2-δ is the fastest and selectivity the NO and N2O is highest. On the other hand, NH3 oxidation occurs at higher temperature over Ti0.9Fe0.1O2-δ, and NO and N2O selectivity is lowest. This observation correlates with the SCR reaction over these catalysts. Thus the NH3 oxidation reaction seems to be in good agreement with the alteration of the SCR performances at high temperature. In the SCR reaction, the NO and N2O formation is lowest over Ti0.9Fe0.1O2-δ compared to the other catalysts, whereas Ti0.9Cu0.1O2-δ has the narrowest SCR window. This indicates that, reactions 2-4 take place easily over Ti0.9Cu0.1O2-δ. In SCR condition in the presence of excess oxygen, NO forms from NH3 oxidation minimizing the SCR window. However, reaction 1 dominates over Ti0.9Fe0.1O2-δ. Therefore, to be a good catalyst for NOx abatement in SCR condition, the primary requirement of a catalyst is that it should not show high NH3 oxidation activity. This explains the catalytic behavior (i.e., NO conversion

J. Phys. Chem. C, Vol. 112, No. 15, 2008 6009 temperature as well as the selectivity) of different transition metals in TiO2 toward SCR reaction. Thus, among the catalysts for SCR reaction, Ti0.9Mn0.1O2-δ shows the highest catalytic activity at low temperature but with low selectivity. However, Ti0.9Fe0.1O2-δ shows lower catalytic activity but the best selectivity. Therefore, in search of a low temperature and higher N2 selective catalyst for SCR reaction, Ti0.9Mn0.05Fe0.05O2-δ was synthesized. NH3 oxidation was carried out over Ti0.9Mn0.05Fe0.05O2-δ, and it showed very poor catalytic activity compared to Ti0.9Mn0.1O2-δ and Ti0.9Fe0.1O2-δ (Figure 9). The product selectivity of the three catalysts for ammonia oxidation at 400 °C is presented in Figure 10a-c. Because the ammonia oxidation activity is poor over Ti0.9Mn0.05Fe0.05O2-δ, we can predict that Ti0.9Mn0.05Fe0.05O2-δ will be a better catalyst for SCR reaction. Accordingly, the SCR reaction was carried out over Ti0.9Mn0.1O2-δ, Ti0.9Fe0.1O2-δ, and Ti0.9Mn0.05Fe0.05O2-δ. The %NO conversion with temperature is plotted in Figure 11. Indeed, Ti0.9Mn0.05Fe0.05O2-δ shows higher catalytic activity than Ti0.9Fe0.1O2-δ in the SCR reaction. Figure 12 is the full reaction profile of the SCR reaction over Ti0.9Mn0.05Fe0.05O2-δ. Figure 8 compares the selectivity of products in the SCR reaction among Ti0.9Mn0.1O2-δ, Ti0.9Fe0.1O2-δ, and Ti0.9Mn0.05Fe0.05O2-δ. Thus, by incorporating Fe along with Mn in TiO2, a new catalyst that shows higher catalytic activity at low temperature as well as good N2 selectivity has been synthesized in this study. Kinetics of SCR. The rate of NO in SCR reaction over Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe, Co, and Cu) and Ti0.9Mn0.05Fe0.05O2-δ with variation in temperature has been plotted as an Arrhenius plot in Figure 13. The activation energies obtained from the slope of the Arrhenius plot are 58.0 kJ/mol for Ti0.9Cr0.1O2-δ, 39.8 kJ/mol for Ti0.9Mn0.1O2-δ, 27.4 kJ/mol for Ti0.9Fe0.1O2-δ, 75.1 kJ/mol for Ti0.9Co0.1O2-δ, 57.1 kJ/mol for Ti0.9Cu0.1O2-δ, and 26.0 kJ/mol for Ti0.9Mn0.05Fe0.05O2-δ. The activation energies for this reaction are comparable to that reported in the literature (40-90 kJ/mol) for zeolites based catalysts.27,28 The kinetic results correlate with the acidic nature of the catalysts obtained from TPD studies. The higher the acidity (Bronsted or Lewis) of the solid catalysts, the lower the activation energy of SCR. From the TPD study, Ti0.9Mn0.1O2-δ has the highest Bronsted acidity and Ti0.9Fe0.1O2-δ has the highest Lewis acidity, whereas Co, Cr, and Cu substituted TiO2 have less acidity. The same trend is observed in the activation energy of the reaction which follows the order Ti0.9Co0.1O2-δ > Ti0.9Cr0.1O2-δ ∼ Ti0.9Cu0.1O2-δ > Ti0.9Mn0.1O2-δ > Ti0.9Fe0.1O2-δ ∼ Ti0.9Mn0.05-

Figure 10. % Selectivity of (a) N2 (b) NO and (b) N2O in NH3 oxidation over Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe, Co, and Cu).

6010 J. Phys. Chem. C, Vol. 112, No. 15, 2008

Roy et al.

Figure 11. %NO conversion against temperature under SCR condition over Ti0.9Mn0.1O2-δ, Ti0.9Fe0.1O2-δ, and Ti0.9Mn0.05Fe0.05O2-δ.

Figure12. CompleteTPRprofileofSCRreactionoverTi0.9Mn0.05Fe0.05O2-δ.

Figure 14. Dependence of rate of NO conversion with (a) variation in NH3 concentration, (b) variation in NO concentration, and (c) variation in O2 concentration.

Figure 13. Rate of NO conversion with temperature as an Arrhenius plot.

Fe0.05O2-δ. Thus, by substituting Fe and Mn in TiO2, lowest activation energy is achieved in SCR.

Further studies on the order of the SCR reaction were carried out over Ti0.9Mn0.05Fe0.05O2-δ. Figure 14a shows the dependence of rate of NO at a fixed temperature (100 °C) with variation in concentration of NH3. For this experiment, the NO and O2 concentrations were kept constant at 1 and 5 vol %, respectively, whereas the NH3 concentration was varied between 0.5 and 4 vol %. The rate of NO conversion decreased linearly with an increase in NH3 partial pressure over Ti0.9Mn0.05Fe0.05O2-δ, and

Catalytic Reduction of NO with NH3 the slope of the curve is -0.7, showing some “inhibition effect”. This inhibition effect of NH3 can arise due to the reaction between NH3 and O2. NH3 inhibition effect over Mn2O3-W2O3/ Al2O3 has also been reported earlier.3 With an increase in NO concentration, the rate of NO conversion increased, and the dependence is linear with a slope of 0.66 at 100 °C which can be rounded up approximately as 0.7 (Figure 14b). NO conversion increases with increase in O2 concentration, with a slope of 0.7 (Figure 14c). The positive reaction orders with respect to NO and O2 and negative reaction order with respect to NH3 has been reported.3,8-11,28 Without oxygen, the SCR reaction takes place at a much higher temperature (not shown). This indicates that oxygen is essential for the reaction to occur. So, from the experimental finding, the rate of NO conversion over Ti0.9Mn0.05Fe0.05O2-δ can be expressed as a power law

rNO ) -ka

[NO]0.7[O2]0.7 [NH3]0.7

where ka is the apparent rate constant. ka was calculated at different temperatures (100-150 °C) and plotted in an Arrhenius plot (Figure S4). The activation energy obtained is 23.7 kJ/ mol, similar to that obtained from Figure 13. Summary and Conclusion The activity of Ti0.9M0.1O2-δ (M ) Cr, Mn, Fe, Co, and Cu), which crystallizes in the anatase phase, for SCR reaction of NO with NH3 was determined. MnTiO3 and FeTiO3, which crystallize in the illmenite phase, were also synthesized. Metal ion substituted anatase TiO2 showed higher SCR activity compared to MnTiO3 or FeTiO3. This indicates that the ionic state of the metals and oxide ion vacancies are the key species for higher catalytic activity. In SCR reactions, the catalysts need to work at a lower temperature, enhance the SCR window, and maintain a high N2 selectivity. While the NO conversion occurred at low temperature over Ti0.9Mn0.1O2-δ, the N2 selectivity was poor. However, the selectivity was highest over Ti0.9Fe0.1O2-δ, but the reaction occurred at a higher temperature. In order to optimize both criteria, Ti0.9Mn0.05Fe0.05O2-δ was synthesized for the first time. This catalyst had a high selectivity and also worked at a low temperature. H2 uptake studies indicated that Cu2+ easily reduced to Cu0, whereas the other metal ions were more difficult to reduce. This correlates with the high rates of NH3 oxidation to NO over Ti0.9Cu0.1O2-δ. It was proposed that the catalyst with the highest rate of NH3 oxidation has lowest N2 selectivity in SCR reaction. The total acidity (Lewis and Bronsted acidity) of the catalysts were determined by TPD experiments. NH3-TPD studies show that Ti0.9Mn0.1O2-δ has the highest Bronsted acidity and Ti0.9Fe0.1O2-δ has the highest Lewis acidity. Thus it can be concluded that the Bronsted acidity in Ti0.9Mn0.1O2-δ is responsible for the low temperature activity, whereas the Lewis acidity in Ti0.9Fe0.1O2-δ accounts for the wide SCR window. Though, it has been mentioned in the literature29 that neither type of acid sites are essential for SCR reaction, several other reports5-7,21 mention the importance of the specific acid sites. If we consider the total acidity (irrespective of Lewis or Bronsted acidity) of the catalyst, Mn and Fe substituted catalysts showed the highest acidity and better SCR activity. Therefore, NH3 can be adsorbed on the surface as NH4+ or as a molecular NH3 in the case of Bronsted acid sites and Lewis acid sites, respectively. NO can adsorb molecularly on the transition metal ion or it can dissociatively chemisorb in the oxide ion vacancy of the

J. Phys. Chem. C, Vol. 112, No. 15, 2008 6011 catalysts. The adsorbed NH3 or NH4+ can form an adduct with the adsorbed NO and the NH2NO adduct dissociates into N2 and H2O. This path of dissociation dominates in the case of Ti0.9Fe0.1O2-δ and Ti0.9Mn0.05Fe0.05O2-δ. However, NH2NO dissociating into N2O is more likely to be favored over Ti0.9Mn0.1O2-δ. So, a “dual site-redox” mechanism takes part on these catalysts surface. However, if the adsorbed NH3 reacts with the oxygen to form a NH2OO adduct, it produces NO and the SCR window becomes narrow. This seems to be the mechanism for Ti0.9Cu0.1O2-δ. Thus, we have showed that the SCR reaction takes place at the lowest temperature with the highest N2 selectivity over the catalysts that have Bronsted and Lewis acid sites for NH3 adsorption as well as oxide ion vacancies for NO dissociation. In this regard, a bi-metal ion substituted on anatase TiO2 (Ti0.9Mn0.05Fe0.05O2-δ) showed higher SCR activity and selectivity compared to V2O5/TiO2. Acknowledgment. The authors thank the Bangalore Institute of Technology for surface area analysis. Financial support from the Department of Science & Technology, Government of India is gratefully acknowledged. Supporting Information Available: Table S1 shows the surface area, pore diameter, and pore volume of the catalysts. Figure S1 shows the Rietveld refinement of (a) Ti0.9Cr0.1O2-δ, (b) Ti0.9Mn0.1O2-δ, (c) Ti0.9Fe0.1O2-δ, (d) Ti0.9Co0.1O2-δ, and (e) Ti0.9Cu0.1O2-δ. Figure S2 shows the change in cell volume due to transition metal ion substitution in TiO2 lattice. Figure S3 is the XPS of (a) Cr(2p) of Ti0.9Cr0.1O2-δ, (b) Co(2p) of Ti0.9Co0.1O2-δ, and (c) Cu(2p) of Ti0.9Cu0.1O2-δ. Figure S4 shows the Arrhenius plot to determine activation energy. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Busca, G.; Lietti, L.; Ramis, G.; Berti, F. Appl. Catal. B: EnVironmental 1998, 18, 1. (2) Busca, G.; Larrubia, M. A.; Arrighi, L.; Ramis, G. Catal. Today 2005, 107, 139. (3) Kapteijn, F.; Singoredjo, L.; Dekker, N. J. J.; Moulijn, J. A. Ind. Eng. Chem. Res. 1993, 32, 445. (4) Kapteijn, F.; Singoredjo, L.; Driel, M. V.; Andreini, A.; Moulijn, J. A.; Ramis, G.; Busca, G. J. Catal. 1994, 150, 105. (5) Smirniotis, P. G.; Pen˜a, D. A.; Uphade, B. S. Angew. Chem., Int. Ed. Engl. 2001, 40, 2479. (6) Pen˜a, D. A.; Uphade, B. S.; Reddy, E. P.; Smirniotis, P. G. J. Phys. Chem. B 2004, 108, 9927. (7) Topsoe, N. Y. Science 1994, 265, 1217. (8) Lietti, L.; Ramis, G.; Berti, F.; Toledo, G.; Robba, D.; Busca, G.; Forzatti, P. Catal. Today 1998, 42, 101. (9) Miyamoto, A.; Kobayashi, K.; Inomata, M.; Murakami, Y. J. Phys. Chem. 1982, 86, 2945. (10) Nam, I.; Eldridge, J. W.; Kittrell, I. R. Ind. Eng. Chem. Prod. Res. DeV. 1986, 25, 1186. (11) Beekman, J. W.; Hegedus, L. L.; Ind. Eng. Chem. Res. 1991, 30, 969. (12) Wong, W. C.; Nobe, K. Ind. Eng. Chem. Prod. Res. DeV. 1984, 23, 564. (13) Kulkarni, S. A.; Pundlik, S. S. Chem. Phys. Lett. 1995, 245, 143. (14) Wolf, M.; Yang, D. L.; Durant, J. L. J. Phys. Chem. A 1997, 101, 6243. (15) Anstrom, M.; Topsoe, N. Y.; Dumesic, J. A. J. Catal. 2003, 213, 115. (16) Soyer, S.; Uzun, A.; Senkan, S.; Onal, I.; Catal. Today 2006, 118, 268. (17) Sun, D.; Schneider, W. F.; Adams, J. B.; Sengupta, D.; J. Phys. Chem. A 2004, 108, 9365. (18) Gilardoni, F.; Weber, J.; Baiker, A. Int. J. Quantum Chem. 1997, 61, 683.

6012 J. Phys. Chem. C, Vol. 112, No. 15, 2008 (19) Choo, S. T.; Yim, S. D.; Nam, In-Sik Ham, S. W.; Lee, J. B. Appl. Catal. B: EnViron. 2003, 44, 237. (20) Amores, J. M. G.; Escribano, V. S.; Ramis, G.; Busca, G. Appl. Catal. B: EnViron. 1998, 18, 1. (21) Pena, D. A.; Uphade, B. S.; Smirniotis, P. G.; J. Catal. 2004, 221, 421. (22) Schneider, H.; Scharf, U.; Wokaun, A.; Baiker, A.; J. Catal. 1994, 147, 545. (23) Asscher, M.; Guthrie, W. L.; Lin, T.-H.; Somorjai, G. A. J. Phys. Chem. 1984, 88, 3233. (24) Bradley, J. M.; Hopkinson, A.; King, D. A. J. Phys. Chem. 1995, 99, 17032.

Roy et al. (25) Sazonova, N. N.; Simakov, A. V.; Nikoro, T. A.; Barannik, G. B.; Lyakhova, V. F.; Zheitvot, V. I.; Ismagilov, Z. R.; Veringa, H. React. Kinet. Catal. Lett. 1995, 57, 71. (26) Louis-Rosse, I.; Methivier, C.; Pradier, C. M.; Catal. Today 2003, 85, 267. (27) Capek, L; Vradman. L.; Sazama, P.; Herskowitz, M.; Wichterlova, B.; Zukerman, R.; Brosius, R.; Martens, J. A. Appl. Catal B 2007, 70, 53. (28) Huang, H. Y.; Long, R. Q.; Yang, R. T.; Appl. Catal B 2002, 235, 241. (29) Liu. Z.; Millington, J.; Bailie, J. E.; Rajaram, R. R.; Anderson, J. A; Microporous Mesoporous Mater. 2007, 104, 159.