Multisite Modeling of NH3 Adsorption and Desorption over Fe-ZSM5

Article pubs.acs.org/JPCC

Multisite Modeling of NH3 Adsorption and Desorption over Fe-ZSM5 Stavros A. Skarlis,*,†,‡ David Berthout,† André Nicolle,† Christophe Dujardin,‡ and Pascal Granger‡ †

IFP Energies Nouvelles, 1 et 4 avenue de Bois-Préau, 92852 Rueil-Malmaison Cedex, France Unité de Catalyse et de Chimie du Solide, UMR CNRS 8181, Université de Lille 1, 59655 Villeneuve d’Ascq, France

‡

ABSTRACT: The storage of NH3 over Fe-ZSM5 catalysts was investigated through a multisite approach where the reaction rate expressions are based on the mean-field approximation. A kinetic model, including five different surface sites was formulated from literature data and validated over a broad range of NH3 adsorption and desorption experiments, showing the interest of the selected approach. For simulation purposes, the values of the NH3 storage capacity used in our test cases were evaluated based on structural properties of zeolites with respect to Si/Al and Fe/Al ratios. Comparison between the predicted and the experimental values shows the limits of our evaluation.

1. INTRODUCTION Transportation is one of the most important sources of nitrogen oxide (NOx) emissions.1 NO and NO2 emissions from diesel engines are subjected to increasingly stringent restrictions2,3 since they contribute to environmental pollution4−6 and deterioration of public health.7,8 Among the numerous NOx abatement technologies available, the selective catalytic reduction by NH3 (NH3−SCR) is an efficient NOx abatement technique in lean conditions.9−11 A wide range of catalyst formulations have already been proposed including noble metals12 and/or transition metals.13,14 Iron and copper exchanged zeolites exhibit high performance over a wide range of temperatures (150−550 °C) as well increasing resistance to hydrothermal aging.15−19 The well-known “standard” NH3−SCR has been extensively investigated mostly over vanadia based catalysts and more recently on Fe-zeolites where the so-called “fast SCR” was found to play a key role for the enhancement of NOx conversion.20 Grossale et al. suggested a sequential SCR reaction mechanism involving the intermediate formation of ammonium nitrate and subsequent reaction with NO further producing nitrogen.21 The storage and release of NH3 is probably the most important chemical process for the dynamic behavior of SCR catalysts.22 The deNOx efficiency relies strongly on the availability and/or the inhibiting effect of NH3.23 Concerning Fe-zeolites, NH3 is believed to be physisorbed or weakly adsorbed, as well as chemisorbed over metallic and acidic sites. Three types of metallic sites were observed by Chmielarz et al.24 associated with (i) Fe3+ cations, mainly in tetrahedral coordination, (ii) FeOx species, and (iii) Fe2O3 bulky clusters. Moreover, these authors have proven, by means of FTIR spectroscopy, that Fe addition leads to the formation of Lewis acid sites, with acidic strength depending on the type of iron oxide. A detailed study by Capek et al. has shown that the © 2012 American Chemical Society

formation of metallic sites is strongly affected by the preparation method.25 Concerning the NH3 adsorption over acid sites, ammonia can be stored over Lewis sites which are substantiated through the identification of coordinated NH3 in the 3H form on the catalytic surface.26,27 Ammonia can also be chemisorbed over Brønsted acid sites forming NH4+ located in AlO4 tetrahedra.26,27 The kinetic modeling of the NH3 sorption over bare and Fezeolites has already been reported in numerous investigations. DFT calculations for the modeling of NH3 storage over HMordenite as well as semiglobal kinetic modeling were reported elsewhere.28−31 Detailed modeling of NH3 storage based on a multisite approach can be found in studies of Sjövall et al., concerning both Fe- and Cu-ZSM5 catalysts.32,33 Moreover, Klukowski et al., developed a kinetic model based on two adsorption sites, which account for Brønsted acidic sites and sites where ammonia can be adsorbed molecularly.34 This model was validated through experiments performed over a FeBeta zeolite. Finally, in a recent study, Chatterjee et al.,35 proposed a two-site model, involving an acidic site for the adsorption of NH3 and a redox site for the NH3−NO reaction, assuming also NH3 spillover from the acidic to the redox sites. To our knowledge there are no detailed kinetic models for Fe-ZSM5 catalysts accounting for interactions between ammonia and different surface species. In the present study, we aim to develop a detailed model that could accurately predict the adsorption and desorption of NH3 over Fe-ZSM5 in different operating conditions. Therefore a multisite approach has been adopted for the NH3 kinetic modeling and reaction rate expressions based on the mean-field approximation were employed. A flow model including mass transfer limitations in Received: October 21, 2011 Revised: March 23, 2012 Published: March 23, 2012 8437

dx.doi.org/10.1021/jp210115d | J. Phys. Chem. C 2012, 116, 8437−8448

The Journal of Physical Chemistry C

Article

both gas and solid phase is also taken into account. The calibration and validation of the model were performed by means of ammonia temperature-programmed desorption (NH3-TPD) on different bare H-ZSM5 and Fe-ZSM5 samples. In order to add value to existing models we have considered in our study weak and strong acidic sites contributing to NH3 storage, taking also into account dehydroxylation effects at high temperatures. Moreover, the values of NH3 storage capacity of each surface site, used in our test cases, were computed through a methodology taking into account particular structural properties of ZSM5 and then adjusted manually to fit the experimental results. The limits of this evaluation were examined. We have gained confidence in the multisite approach by validating it over a variety of operating conditions, ranging from low to high temperature applications (100−600 °C).

ηi =

∂t

+u

∂cg, i ∂z

= k m, is(cs, i − cg, i) +

∂ ⎛ ∂cg, i ⎞ ⎜Dax ⎟ ∂z ⎝ ∂z ⎠

ϕ=e

dt

= (cg, i in − cg, i out) + ϕfilm

Bim =

(1)

ρsCpsVs

ShDi dh

(7)

dTs = hsSgeom(Tg − Ts) + Φw ext + dt

∑ [−ΔhiMi ωi] i

Finally for the purposes of our simulation, we neglected gas accumulation in the gaseous film and a global external heat transfer within the solid phase in the axial direction as well as between the reactor and the ambient were not taken into consideration. 2.2. Nature of Surface Sites for NH3 Adsorption: Input for the Multisite Kinetic Model Development. 2.2.1. Setup of the Multisite Kinetic Model. The modeling of NH3 storage, employing a multisite approach was previously implemented by Sjövall et al. for a Cu-ZSM5 monolith type catalyst prepared by ion-exchange.32 These authors combined results from theoretical and experimental investigations in order to setup a kinetic model of NH3 storage over Cu-ZSM5 and considered coverage dependency of desorption activation energy barrier in order to fit correctly the broadness of the high temperature desorption peak associated with copper sites. Moreover in order to simplify their reactor model, internal diffusion process was neglected. In our present study the multisite kinetic model for the NH3 storage over Fe-ZSM5 is built mainly by extracting data from earlier TPD studies over bare and Fe exchanged ZSM5 samples. Various NH3-TPD tests combined with experiments employing optical techniques (UV−vis and FTIR) were analyzed in order to determine surface species which contribute to the storage of NH3 over Fe-ZSM5. Ammonia TPD provides information for the surface site identification but the TPD’s concentration signals should be analyzed carefully because they do not directly reflect the desorption kinetics. The occurrence of parallel readsorption as well as mass transfer processes can cause a significant shift of the maximum temperature of TPD peaks.48 Moreover, even though NH3-TPD is one of the most widely used methods to characterize the Brønsted sites density,49 it is also a controversial technique due to several limitations.50 The variation of particle sizes, in case of TPD

(2)

(3)

km,i is the mass transfer coefficient computed based on the Sherwood number and the ith gaseous species diffusivity (eq 4) k m, i =

ek m Di

(8)

ηi ω Vwash

(6)

Bulk temperature is deduced from the heat balance with the same assumptions as for species mass balance. The heat balance in the solid phase is given by eq 8

where ϕfilm is deduced from eq 341,42 k m, is(cg, i − cs, i) =

k(Ts) De

where De stands for the effective diffusion coefficient computed through the Bosanquet equation, taking into account the effect of tortuosity and porosity on the species diffusion coefficient.46 Finally the Biot number is another nondimensional number expressing the ratio between the convective and the diffusive mass transfer (external versus internal mass transfer resistance; eq 7)42,47

In our approach, the model of monolith was regarded as a series of continuously stirred tank reactors in order to simplify the axial dispersion term, as often done for fixed-bed catalyst systems. The number of tanks required for the macro-mixing of the gas in the axial direction was estimated based on the Péclet number.38 Each tank was then regarded as a component where the mass and heat balance for both the gas and solid phase were solved. Neglecting the mass transport limitations in the monolith could induce errors in the computation of adsorbed species.39 Therefore, the diffusion of species was taken into account both in the gas phase and the washcoat. A film model was used to compute the mass transfer between the surface and the gas phase whereas the diffusion within the washcoat was expressed using an effectiveness factor, based on a generalized Thiele modulus approach.40 The mass balance equation of gas species in each tank can be simplified as shown in eq 2 dcg, i

(5)

This effectiveness factor represents a ratio between the observed effective rate and the intrinsic reaction rate in the absence of mass transfer phenomena.45 For a first order reaction, the Thiele modulus is determined as42,44

2. MODELING 2.1. Reactor Model. The model was developed into the simulation environment LMS.Imagine.Lab AMESim.36 In the present work, a model of a monolith reactor was developed. Disregarding gas phase reactions, the 1D plug-flow gas species equations can be written as presented in the review of automotive catalyst modeling by Depcik et al.37 ∂cg, i

tanh(ϕ) ⎡ ϕ tanh(ϕ) ⎤ ϕ⎣⎢1 + Bi ⎦⎥ m

(4)

The Sherwood number used was computed through the expression developed by Groppi et al. for square channels.43 The effectiveness factor ηi for monolith reactors under isothermal conditions was computed using the Thiele modulus and the Biot number (eq 5)44 8438



Article

tests using fixed-bed reactors, NH3 readsorption as well as experimental conditions can significantly affect the TPD concentration signal, resulting in misleading results. Thus, in this study, we use a model that takes into account kinetic and mass transfer phenomena to simulate results of NH3 adsorption and desorption experiments. TPD signals enable us to identify families of surface sites associated with desorption activation energies at different temperature ranges. 2.2.2. NH3 Storage over Bare H-ZSM5. Many papers on this subject are available with miscellaneous results and interpretations. NH3-TPD combined with IR-spectroscopic experiments, employing ZSM5 samples (Si/Al = 11.8 and 18) were carried out by Topsøe et al.51 who identified three desorption peaks, observed within the temperature ranges of 60−100, 150−200, and 420−500 °C. These authors have assigned the desorption peak at ∼227 °C to weak acidic sites, probably terminal silanol groups on the external surface of the zeolite (3720−3740 cm−1) or possible nonzeolitic impurities. TPD peaks around 500 °C have been ascribed to strong acidic sites corresponding to an IR band of 3600 cm−1. The desorption peak at low temperatures (ca. 100 °C) corresponds to physisorbed NH3. According to Katada et al., these weak sites could be related to hydrogenbonded ammonia and are irrelevant to the acidic properties of the zeolite.52 Katada et al. reported the same observations over both MOR and ZSM5, highlighting also the role of diffusion phenomena, which cannot be excluded within this temperature regime.53 Other NH3-TPD experiments over bare ZSM5 zeolites characterized by different Si/Al ratios reported by Costa et al., revealed two desorption peaks centered at temperatures around 177 and 360 °C.49 The desorption peak released at 360 °C is attributed to ammonia adsorbed over weak acidic sites. According to Rodriguez-Gonzalez et al., this peak which appears within a temperature range of 370−440 °C is related to ammonia desorbed from Brønsted acid sites.54 Similar observations were also reported elsewhere.55,56 Moreover, Matsuura et al. showed that increasing the duration and consequently the maximum temperature of the TPD experiments, a third desorption peak is observed.57 According to the same source, this peak appears at around 500 °C and corresponds to NH3 bonded to very strong acidic sites probably extra-framework Al-structure. Based on the above-mentioned results, three different sites have been taken into account in our model to describe precisely the bare zeolite’s surface sites involved in NH3 storage. In order to capture the peak at low temperatures (L-T peak), a site for physisorption, called S1, has been added. Furthermore, in order to describe the desorption peaks at intermediate temperatures (I-T peak) corresponding to weak acid sites and at high temperatures (H-T peak) attributed to a mixture of Brønsted and Lewis acid sites, two acidic sites were added, called S2 and S3 respectively. 2.2.3. NH3 Storage over Fe-ZSM5. As far as the Feexchanged zeolites are concerned, results of TPD measurements over Fe-ZSM5 catalysts with different iron loadings are reported by Brandenberger et al. who observed an increase in intensity of L-T and I-T peaks, whereas the H-T peak attenuates.58 This latter change in intensity occurs due to the substitution of H protons by Fe cations taking place during the ion exchange process.59 The increase in the amount of NH3 desorbed at low and intermediate temperatures can be attributed to NH3 storage over metallic sites. This increase in the amount of desorbed NH3 is probably related to the creation of new Lewis acid sites.60 Anumiziata et al. have investigated

the adsorption of pyridine over Fe-ZSM5 by means of FTIR spectroscopy.61 The results show that pyridine interacts with Lewis metal sites such as Py-Fe2+ and Py-Fe3+. Katada et al. performed several NH3-TPD runs over ferrisilicates giving rise to desorption peaks at 277 and 427 °C.52 Therefore, we have added one metallic site, called S4 site, related to the enhancement of NH3 desorption at intermediate temperatures. Finally, analyzing the TPD curves reported in the same study of Brandenberger et al., another desorption peak was found at very high temperatures in the range of 650 °C. A small shoulder is also discernible over bare ZSM5 reported in the same article. Enhanced strong acid sites were reported by Mirodatos and Barthomeuf ascribed to an interaction between protonic [(Si,Al)On(OH)m] and Lewis sites.62 To our knowledge, neither theoretical studies nor experimental results have allowed to interpret the nature of this peak satisfactorily. Nevertheless, in order to deconvolute accurately TPD concentration signals, we have included an additional site, called S5, accounting for NH3 desorption at very high temperature. The latter possibly corresponds to an interaction between ammonia and water desorbed during dehydroxylation.63,64 2.3. Kinetic Model for NH3 Storage. For each of the above-mentioned surface sites, reaction rates of adsorption and desorption of NH3 were applied as presented in eqs 9−13. The adsorption rate constant is based on an Arrhenius expression, whereas that of desorption is described by a Temkin expression, the respective activation energy thus being surface coverage dependent (eq 13) RNH3_ads = kads(Ts)C NH3(1 − θj)

(9)

⎛ E ⎞ kads(Ts) = A ads exp⎜ − ads ⎟ ⎝ RTs ⎠

(10)

RNH3_des = kdes(Ts , θj)θj

(11)

⎛ E(θj)des ⎞ ⎟⎟ kdes(θj , Ts) = Ades exp⎜⎜ − RTs ⎠ ⎝

(12)

E(θj)des = Eo(1 − αθj)

(13)

Using the reaction rates of adsorption and desorption over each surface site, the chemical source term is computed through the eq 14: 5

ω NH3 = (∑ vjR j)mzeolite (14)

j=1

Finally, the surface coverage of each surface site is computed through the rates of NH3 adsorption and desorption and the total storage capacity according to the following equation: Nj

∂θj ∂t

5

=

∑ vjR j j=1

(15)

The present detailed kinetic model is based on the meanfield approximation. Thus we assume that all adsorbed species are distributed randomly over the surface.

3. RESULTS 3.1. Storage of Ammonia over H-ZSM5. The detailed model for the NH3 storage was initially calibrated and used for 8439



Article

Table 1. Kinetic Parameters for the Simulation of NH3 Adsorption and TPD Tests, Reported by Brandenberger et al.58 type of site S1 S2 S3 S4 S5

(weak non acidic sites) (weak acidic sites) (strong acidic sites) (metallic sites) (dehydroxylation effects)

Aads (m3/s kgzeolite)

Eads (kJ/mol)

Ades (mol/s kgzeolite)

Edes (kJ/mol)

α [−]

1200a 700a 950a 500a 800g

0b 0b 0b 0b 0b

1013c 1013c 1013c 1013c 1014g

95.0g 134.5d 195.0e 125.0g 255.0g

0.390f 0.155g 0.130g 0.100g 0.000g

a

Based on the parameters reported by Sjövall et al.32 bAssuming that the NH3 adsorption is a nonactivated process.65 cValues corresponding to adsorbed immobile molecules.66 dValue reported by Pinto et al.68 eValue reported by Matsuura et al.57 fValue reported by Wilken et al.71 gValues manually calibrated during simulation, so as to fit experimental curves.

Figure 1. Measured and computed NH3 outlet concentration during NH3 TPD test reported in ref 58.

9.2% very strong acidic sites (Edes = 160−250 kJ/mol). According to another work of Sahoo et al., four types of acid sites related to activation energies of adsorption around respectively 100 kJ/mol, 110 kJ/mol, 130 kJ/mol and the 140 kJ/mol were identified on various ZSM5 samples.69 The dependency of desorption energy on surface coverage is generally poorly discussed in literature, with regards to surface properties and only very few information was found. Mishin et al. have performed calorimetric measurements to estimate heats of adsorption of NH3 over different ZSM5 samples, proving the variation of the respective activation energy as a function of surface concentration.70 Wilken et al. obtained a value of α close to 0.38 during a NH3-TPD test over a Cu-Beta catalyst.71 This value was also used for the calibration of the multisite kinetic model of Sjövall et al.32 The above-mentioned values of activation energy were carefully evaluated in order to determine the set of parameters for the calibration of our model. Besides the existence of different surface sites there are other parameters that influence the shape of desorption peaks. As shown in eq 13 the surface coverage, including θ and α, has a direct impact on the desorption activation energy. Sjövall et al.32 has proposed a coverage dependent activation energy, for copper sites, so as to enable the simulation of broad desorption peaks at high temperature. Moreover, the effect of mass limitations on the temperature where each peak appears, cannot be excluded, as already discussed in section 2.1. Hence, kinetic parameters collected in Table 1 were used as a reference for the calibration of our model enabling us to simulate experimental data

the simulation of NH3 adsorption and TPD experiments over bare zeolites. For this purpose well-documented experiments reported in literature were simulated, presented hereafter. The calibration of kinetic parameters was performed in line with theoretical and experimental studies already published (Table 1). According to a recent work of Chen et al., the adsorption of NH3 on bare and/or Fe-exchanged zeolites is a nonactivated process.65 Consequently, the activation energy of adsorption is set to zero, for our model. It is worth noting that the previous authors found NH3 adsorption activated on acidic sites for the reactive systems NH3 + NO and NH3 + O2. In the present study, the pre-exponential factors of the adsorption were adjusted manually during calibration, so as to fit the experimental curves, taking into account respective values used for the model of Sjövall et al.32 The determination of the kinetic parameters of desorption was more complicated. First, the pre-exponential factor of desorption was set to 1013, according to the transition state theory for immobile adsorbed molecules.66 Similar values are also reported elsewhere.32 The activation energy of desorption was chosen based on several publications and it is discussed below. Recently, Metkar et al. recommended an average activation energy of desorption, around 115 kJ/mol.67 Pinto et al. have reported a broad distribution of acid strengths by performing deconvolution of NH3-TPD spectrum over a H-ZSM5 zeolite.68 This sample, characterized by an atomic Si/Al ratio of 15, was found to be composed of 47.7% weak adsorption sites (Edes = 50−65 kJ/ mol), 43.1% strong acid sites (Edes = 75−140 kJ/mol), and 8440



Article

mentioned published results we also assume in our methodology that the model’s acidic sites S2 and S3 exhibit a quite homogeneous acidic strength distribution. However since our model is based explicitly on data reported in literature we have not enough information about the ratio between Brønsted and Lewis sites contributing to each desorption peak of the NH3TPD signal and thus it is impossible to come to a safe conclusion. Further analysis on this topic is out of the scope of our study. In order to compute the NH3 surface coverage on each site, the NH3 storage capacity over each surface site (Nj) needs to be evaluated (see eq 15). Therefore, before starting the actual simulation we have tried to evaluate these values through a methodology presented below. Initially, the impact of the Si/Al ratio on the NH3 total storage capacity over bare H-ZSM5 zeolites was investigated. For this purpose results from TPD performed by Hidalgo et al. were used.75 Analyzing data with various Si/Al ratios, we have observed that for Si/Al of up to 28 the total NH3 storage capacity is reduced exponentially, when increasing the Si/Al ratio (Ntot ≈ e−0.11Si/Al). For higher ratios an almost constant behavior is found (Ctot ≈ 0.5), showing that if the Si content increases significantly, the total storage capacity is slightly affected by the sample’s composition. For the next step, the effect of the Si/Al ratio on the storage capacity over Brønsted acidic sites (Nacidic) of H-ZSM5 samples was determined based on experimental studies performed by Gonzalez et al.,54 Sazama et al.,76 and Kapustin et al..77 Fitting the experimental data, we have confirmed that a linear relation exists between Brønsted acidity and the aluminum content (Nacidic ≈ 11(Si/Al)). Rodriguez-Gonzalez et al. have obtained similar results, adding that when the zeolite’ s crystallinity is perfect and no undesirable effects due to high aluminum concentration occur, each aluminum atom corresponds to a Brønsted site.54 However, Costa et al. has proven experimentally that the aluminum content is not the only factor affecting the Brønsted acidity of ZSM5 zeolites.49 Finally, applying a molar balance, the storage capacity of weak sites (Nweak) for physisorption can be calculated by subtracting the storage capacity over acidic sites from the total storage capacity. The respective results, presented in Table 2, show that for the

obtained in wide operating conditions in order to verify the limit of our approach. 3.1.1. TPD of Ammonia Preadsorbed on H-ZSM5 at RT. The first experiment which was used for the model’s validation is a NH 3 TPD over H-ZSM5, earlier reported by Brandenberger et al.58 A H-ZSM5 sample was exposed to 1% vol. NH3 in nitrogen at room temperature, for 1 h. Then a TPD was performed from 30 to 800 °C applying a T-ramp of 10 °C/ min in flowing He. Simulated TPD curves are presented in Figure 1 and the respective kinetic parameters in Table 1. The existence of three desorption peaks, described in section 2.2, is confirmed. In this case the peaks are centered at three different temperatures: 140, 230, and 490 °C. Initially, we have attempted to simulate each desorption peak separately, before simulating the whole TPD signal. According to the analysis we have presented above, each of these peaks corresponds to NH3 desorbed from different surface sites. The first peak, appearing at low temperatures (140 °C), named as “L-T peak”, is attributed to physisorbed NH3. The activation energy for this desorption peak was manually calibrated in the order of magnitude of 100 kJ/mol so as to fit the experimental curves. The second peak, centered at 230 °C and named “intermediate T peak” is related to NH3 bonded to weak acidic sites. For this peak the activation energy of desorption is estimated to ca. 135 kJ/mol, in agreement with Pinto et al.,68 who reported similar values for weak acidic sites. Moreover Rodriguez-Gonzalez et al. used values between 129 and 142 kJ/mol in order to simulate a NH3 desorption peak over a H-ZSM5 zeolite (SiO2/Al2O3 = 30) appearing at ca. 350 °C.54,72 Finally the last peak found at high temperatures, named as “H-T peak”, is attributed to strongly bonded NH3 on acidic sites. For this peak the activation energy of desorption was set to 195 kJ/mol, in line with Matsuura et al., who assigned a similar value to a NH3 desorption peak at around 500 °C.57 For each peak, the “α” factor was adjusted to specific values, in order to center the separate desorption peaks at low, intermediate and high temperature respectively, as described above. Generally the adjusted “α” values reflect the extent of repulsive interactions between ammonia molecules that usually weaken their adsorption. A value of 0.35, combined with an activation energy of desorption of around 95 kJ/mol enables the correct simulation of the low temperature peak. These values are in agreement with the study of Wilken et al.,71 where a combination of α = 0.38 and Edes = 120 kJ/mol was proposed. For the simulation of the remaining peaks lower values of α were applied in order to fit the experimental curves. The chosen values, which apparently are of the order of magnitude of 0.15, indicate a limited weaker surface coverage dependency of the activation energy of desorption for S2 and S3 acidic sites. This trend seems to be in qualitative agreement with previous microcalorimetric measurements over two H-ZSM5 samples (Si/Al = 34), using NH3 and pyridine reported by Sharma et al.73 These authors observed an almost constant deferential enthalpy of adsorption (in the range of 150 kJ/mol) with increasing surface coverage. Thus they suggested a narrow distribution of acidic strength on the samples surface. More specific information is included in a review of A. Aroux, concerning the acidity characterization of different types of zeolites, including ZSM5, by means of microcalorimetry.74 The analysis shows that zeolites containing very few Lewis acidic sites show a very homogeneous Brønsted acid strength distribution, which apparently corresponds to a differential heat of adsorption around 150 kJ/mol. Based on the above-

Table 2. Comparison between the Estimated and the Actually Used NH3 Storage Capacity over Each Surface Site of the H-ZSM5 Samples Discussed in Section 3.1 (Si/Al = 13.5) evaluated storage capacity used values in TPD shown in Figure 1 used values in TPD shown in Figure 2

Nweak (mol/ kgzeolite)

Nacidic (mol/ kgzeolite)

Ntot (mol/ kgzeolite)

1.15 1.15

1.15 1.08

2.30 2.23

1.20

1.14

2.34

first test cases, where apparently a H-ZSM5 sample with Si/Al ratio of 13.5 is used, the calibrated values are quite in agreement with the evaluated ones. 3.1.2. TPD of Ammonia Preadsorbed on H-ZSM5 at 150 °C. The second experiment is an NH3 adsorption−desorption test over H-ZSM5 reported in the already mentioned publication of Sjövall et al.32 A total of 500 ppmv of NH3 were injected for 60 min at 150 °C followed by a period of Ar flush and a T-ramp of 10 °C/min. The results are presented in Figure 2. In this case, weak sites (S1) are able to accurately describe NH3 physisorption taking place during flushing. 8441



Article

Figure 2. Validation of the detailed kinetic model for NH3 storage through the NH3 adsorption and TPD test, reported in ref 32: (a) Evolution of the NH3 outlet concentration during an adsorption and TPD test and (b) Analysis of the TPD test.

Table 3. Modified Kinetic Parameters for the Simulation of the NH3 Adsorption and TPD Test Reported by Sjövall et al.32 type of site S1 (weak non acidic sites) S2 (weak acidic sites) S3 (strong acidic sites)

Aads (m3/s kgzeolite) 1200 700a 950a

Eads (kJ/mol)

a

b

0 0b 0b

Ades (mol/s kgzeolite) 13c

10 1013c 1013c

Edes (kJ/mol) d

92.7 132.0d 170.0d

α [−] 0.390e 0.155d 0.130d

a

Based on the parameters reported by Sjövall et al.32 bAssuming that the NH3 adsorption is a nonactivated process.65 cValues corresponding to adsorbed immobile molecules.66 dValues manually calibrated during simulation so as to fit experimental curves. eValue reported by Wilken et al.71

Therefore the first L-T peak does not appear during the TPD test. The I-T peak appears at ca. 260 °C, whereas the H-T peak appears at ca. 400 °C. The kinetic parameters used for the simulation of this experiment, shown in Table 3, are very similar to the ones reported in Table 1 (less than 10% deviation concerning the values of the activation energy). Notably, the activation energy of desorption for the S3 site is fixed to 170 kJ/mol instead of 195 kJ/mol, showing that in this sample less strong acidic sites

exist, which can be attributed to the zeolite structure (see also section 3.1.1). However an effect of the experimental conditions applied can not be excluded (e.g., heating rate). The values of the NH3 storage capacity over each site are presented in Table 2 and seem to be in quite accordance with the respective values, calculated through the algorithm presented in section 3.1.1. Our detailed kinetic model could reproduce quite sufficiently the adsorption and TPD test, taking the NH3 physisorption as 8442



Article

Figure 3. Validation of the detailed kinetic model through experimental results over Fe-ZSM5 zeolites with various Fe content, reported in ref 58.

of the simulation are reported in Figure 3a−f and the kinetic parameters used for simulation purposes in Table 1. Generally the shape of the TPD curves does not change dramatically compared to the curves presented in section 3.1. The proton exchange results in the deterioration of the storage capacity over the (S3) strong acidic sites and consequently in a decrease of the “H-T desorption peak”. This behavior, which can be captured by our model, emphasizes the fact that the “HT peak” is related to significantly strong acid sites. On the other hand, the intensities of “L-T and I-T peaks” increase, which can

well as the variation of acidity over the surface of the zeolite into account. 3.2. Storage of Ammonia over Fe-ZSM5. After evaluating the detailed model over bare ZSM5, we have tried to test its consistency over iron exchanged zeolites. For this reason a series of NH3 adsorption and TPD runs earlier reported58 were simulated. Each test includes the exposure of Fe-ZSM5 catalysts with different Fe loadings to 1 vol. % NH3 in nitrogen at room temperature, for 1 h followed by a TPD from 30 up to 800 °C in a He flow applying 10 °C/min. The results 8443



Article

Table 4. Comparison between the Estimated and the Actually Used NH3 Storage Capacity over Each Surface Site of the FeZSM5 Samples Discussed in Section 3.2 Nweak (mol/kgzeolite)

Nacidic_exch (mol/kgzeolite)

Nmetallic (mol/kgzeolite)

total capacity (mol/kgzeolite)

Fe/Al ratio

estimated

used

estimated

used

estimated

used

estimated

used

0.57 1.1 1.7 2 2.6 5

1.15 1.15 1.15 1.15 1.15 1.15

1.15a 1.15a 1.15a 1.15a 1.15a 1.03

0.80 0.68 0.55 0.48 0.35 0.40

1.03b 0.91b 0.86b 0.85b 0.84b 0.70b

0.14 0.25 0.34 0.45 0.46 0.21

0.20c 0.16c 0.01c 0.30c 0.15c 0.01c

2.09 2.08 2.04 2.08 1.96 1.76

2.38 2.22 2.02 2.30 2.14 1.74

a

Refers to the bare zeolite storage capacity over non acidic weak sites. bSum of the storage capacities of Lewis and Brønsted acidic sites. cSum of the storage capacities of Lewis-metallic sites and nonacidic weak metallic sites.

coverage dependency of the activation energy of desorption could be different depending on the type of iron surface species, with regards to Fe-isolated, -oligomeric or −polymeric ones. However, such aspects are quite unknown and should be definitely further investigated. Using our set of kinetic parameters, the desorption peak assigned to metal sites is centered in the same area as the I-T peak with a maximum at around 200 °C, coinciding with the temperature range where S2 sites appear. As discussed in section 2.2, the metallic sites are Lewis-type sites. However we have to point out that they should not be associated with Lewis sites existing in the bare zeolite. Metallic Lewis type sites are connected with bridging hydroxyl groups between the metal cation and Si4+,84 whereas the acidic Lewis sites of the nonexchanged zeolites are tricoordinated aluminum atoms.85 As was also described in section 2.2, a small peak appears at very high temperatures. The model captures this peak correctly through the (S5) site. The activation energy of this peak is manually calibrated to 255 kJ/mol in order to fit the experimental curves. Since this site is used to capture phenomena possibly associated with dehydroxylation effects (see also section 2.2.3), it is not considered as an actual surface site interacting with gaseous NH3 during storage. Thus a simple Langmuir expression is employed, leading to an α equal to zero. To calibrate the storage capacity over each site of the FeZSM5 samples, the impact of the iron loading was taken into account. Analyzing data from NH3 TPD test provided in the already mentioned experimental study of Brandenberger et al.,58 we have observed that the iron loading causes a linear deterioration of the storage capacity over acidic sites with an increase of the Fe/Al ratio (Nacidic_exch ≈ −1.5(Fe/Al)). Regarding the created metallic sites, the respective storage capacity (Nmetallic) was determined by the difference between the storage capacity of the acidic sites before and after the ionic exchange, taking into account the ion exchange degree. As it can be seen in Table 4, the above-mentioned estimation introduces quite important deviations in comparison with the respective calibrated values (up to 58% and 97% for the storage capacity of acidic and metallic sites respectively), showing that more parameters should be taken into account, namely catalyst preparation and pretreatment method. For example, in the experimental work of Forni et al.,86 the application of different durations of isothermal desorption resulted in the disappearance of the desorption peak at intermediate temperatures (occurring at nearly 300 °C). Similar results are reported in54 after flushing a H-ZSM5 sample with water-saturated He for 12 h. We have also simulated NH 3 adsorption and TPD experiments reported in a recent publication of Sjövall et al.33

be attributed to NH3 bonded to metal sites, as mentioned in section 2.2. The contribution of the (S4) metallic site can sufficiently account for the increase in NH3 storage capacity in the intermediate temperature range. A single deviation is observed in Figure 3c,e, where no increase appears. According to the authors, this behavior is mainly related to the experimental conditions. However the influence of the experimental parameters on the catalysts properties may also play a role. Sugawara et al. observed that the amount of NH3 desorbed at around 300 °C during NH3-TPD over Fe-ZSM5 samples with different iron loading is less than the amount of the catalysts’s iron content.78 These authors suggest that only specific metal species contribute to the NH3 storage, which notably is not encountered at low iron loads. Brandenberger et al. have examined the surface composition of the studied samples through UV−vis spectroscopy in ref 79. The results prove the abundance of isolated metal species at Fe/Al ratios up to 0.3 and metal clusters (dimeric and oligomeric species) as well as large metallic particles at higher ones. Thus we can conclude that mainly the metal clusters contribute to the NH3 adsorption rather than the isolated species, showing that the sample preparation has a crucial effect on the formation of surface sites responsible for the NH3 storage. The adsorption of ammonia over iron clusters has also been proven by DFT studies.80 Brandenberger et al.79 have observed a large number of metallic particles at high Fe concentrations, that could explain the deterioration of the overall NH3 storage capacity shown in Figure 3e. Regarding the increase in the amount of NH3 desorbed at around 110 °C, we have chosen on purpose not to capture it through a metal site, since we have no evidence about interaction of ammonia and metal species at this temperature regime. Ates has observed that the iron loading can almost duplicate the zeolite’s active surface area, due to the removal of impurities during the ion-exchange.81 Therefore we assume that the increase in the low temperature peak is attributed to additional physisorbed NH3, and thus it can be described by the (S1) site accounting for physisorption. The activation energy related to the (S4) metal sites was fixed manually to 125 kJ/mol with an α set to 0.1 so as to fit the experimental curves. Extracting data from the work of Wahba and Kemball,82 we found an activation energy of desorption of nearly 182 kJ/mol with α = 0.47. However, in other studies, the activation energy of NH3 desorption over iron crystals is set to 39.2 and 52.7 kJ/mol.83 As already discussed in section 3.1.1 the “α” factor of S4 metal sites is set to 0.1 (in order to manually fit the experimental curves) assuming that the metal sites heterogeneity is narrow. The Fe dispersion may also have some relevance to the determination of this factor. The surface 8444



Article

Figure 4. Validation of the detailed kinetic model through experimental results over a commercial Fe-ZSM5 zeolite, reported in ref 33.

Table 5. Modified Kinetic Parameters for the Simulation of NH3 Adsorption and TPD Tests Reported by Sjövall et al.33 type of site S1 S2 S3 S4 S5

(weak non acidic sites) (weak acidic sites) (strong acidic sites) (metallic sites) (dehydroxylation effects)

Aads (m3/s kgzeolite) a

1200 700a 950a 300a 800d

Eads (kJ/mol)

Ades (mol/s kgzeolite)

b

13c

0 0b 0b 0b 0b

10 1013c 1013c 1013c 1014d

Edes (kJ/mol) d

92.7 145.0d 165.0d 110.0d 255.0d

α [−] e

0.390 0.155d 0.130d 0.4100d 0.000d

Nj (mol/kgzeolite) 1.00d 0.09d 0.05d 0.20d 0.07d

a Based on the parameters reported by Sjövall et al.32 bAssuming that the NH3 adsorption is a nonactivated process.65 cValues corresponding to adsorbed immobile molecules.66 dValues manually calibrated during simulation so as to fit experimental curves. eValue reported in Wilken et al.71

acidic sites respectively, they imply a much more narrow distribution of the sample’s acidity strength compared to the Fe-ZSM5 samples already discussed in this study. We could attribute this behavior either to the catalyst’s composition and namely to the Si/Al ratio as well as to the existence of water in the feed. Considering the latter, Sjövall et al. has shown experimentally that H2O is competitively coadsorbed with NH3 on all surface sites of a Cu-ZSM5, leading to a deterioration of NH3 uptake at 150 °C.32 Considering the α factor of S4 metallic sites, which is set to 0.4 implies a quite broad heterogeneity of these sites. Anyhow, since the sample is a commercial one, an impact of additives or impurities can not be excluded. Finally considering the storage capacities of NH3 (Nj), they had to be manually adjusted so as to fit the experimental curves since no data about the structural properties of the sample were available so as to use our practical algorithm. The adsorption is performed at 159 and 210 °C; thus, the “LT peak” reasonably should not appear. Additionally, the very high temperature peak is not evidence since the TPD test is performed up to 450 °C. These results demonstrate that the 5site model as well as the methodology we have followed are

A commercial sample of Fe-ZSM5 was exposed to 510 ppmv NH3, 5 vol. % H2O, and 5 vol. % CO2 for 40 min at 159 and 210 °C followed by a 60 min period of 5 vol. % H2O in Ar and a TPD, applying 10 °C/min in a 5 vol. % H2O and Ar flow. The results are shown in Figure 4a−d. Considering the Figure 4d, a deviation is observed between experimental and simulated data, with respect to the trend of the two curves. The model, calibrated in the same way for both tests predict a maximum around 330 °C, (Figure 4b,d) whereas the experimental TPD signal exhibits a maximum peak at around 370 °C. We attribute this deviation to experimental conditions applied (e.g., a slight difference in the temperature ramp), since there is no justification why the maximum peak should shift to higher temperatures. In this case, the kinetic parameters had to be manually recalibrated in order to correctly capture the desorption peaks. The respective values presented in Table 5, show deviations in both activation energies of desorption (maximum deviation in activation energies of desorption lower than 13% with respect to activation energy of desorption) and in the α factor for the S4 metallic sites. Regarding the recalibrated activation energies of desorption, which are 145 and 165 kJ/mol for S2 and S3 8445



Article

Figure 5. Effect of the α factor on the determination of the NH3 desorption activation energy: ■ α = 0.35, ▲ α = 0.25, ● α = 0.15, × α = 0.05. The bold colored lines stand for the variation of the activation energy for the α factors used in the simulations presented in the article.

suitable for the accurate modeling and simulation of NH3 storage over various bare and iron-exchanged ZSM5 catalysts tested in various operating conditions.

the effect of experimental conditions and/or the state of the parent zeolite). In order to calibrate the NH3 storage capacity over each surface site, we have tried initially to evaluate the respective values by analyzing the zeolite properties. Comparing the estimated and the actual used values, we have found that even though our estimation was reliable in some cases, significant deviations were also observed. This implies that more parameters have to be taken into account, in order to calibrate the number of the different surface sites more accurately. It should be noted that we did not intend in this study to perform any fundamental analysis on the relation of the acidity with the Si/Al and Fe/Al ratios. Our goal was mostly the evaluation of the used parameters so as to investigate the limits of our approach. Finally, as it was explained in section 3.1, the calibration of the α factor of the activation energy for NH3 desorption was a quite crucial task. Since very few information was found in literature, we have assumed a narrow acidic strength distribution for S2 and S3 acidic sites. The same hypothesis was made for S4 metallic sites. A short analysis on the impact of the α factor on the determination of the activation energy of NH3 desorption is shown in Figure 5. The respective results, for every surface site used in the model, indicate reasonably that for α around 0.35 the activation energy of desorption varies significantly as a function of surface coverage, whereas for lower α values the activation energy is restricted in a narrow area. In our simulations we have tried to practically deconvolute the NH3-TPD signal over different samples by employing our multisite model. Therefore the used α factors are reasonably tuned to low values, so as to simulate narrow TPD curves, that are mainly affected by the activation energy. Moreover the way the parameters were calibrated enables the clear distinction of each desorption peak avoiding overlapping effects. More detailed analysis on the α factors is definitely required but it is out of the scope of the present study. Overall, our five-site kinetic model adds new features to already published models following the same approach, namely the variation of the surface acidity. Nevertheless, the calibration of the parameters required for simulation purposes can be quite difficult, increasing the complexity of the model application.

4. CONCLUSIONS AND DISCUSSION We have presented an extended kinetic model for depicting the adsorption and release of NH3 over Fe-ZSM5 catalysts. A multisite approach was implemented, taking into account the storage of NH3 over various surface sites. NH3 TPD profiles over both bare and iron-exchanged ZSM5 zeolites, reported in literature, underline the existence of at least three desorption peaks centered at low, intermediate and high temperatures, which are attributed to sites for weak physisorption, weak acidic sites, strong acidic sites as well as metal sites. Including these sites in our detailed kinetic model and adjusting the kinetics parameters in an acceptable range, we have obtained a good agreement with several previous experimental studies over both H-ZSM5 and Fe-ZSM5 catalysts. The model was able to describe surface acidity distribution as well as dehydroxylation effects at high temperature. Furthermore the single metallic site used was able to take into account interactions between NH3 and additional Lewis acidic sites created after the metal exchange process, associated with the migration of Fe cations in the zeolite framework. For the calibration of the kinetic parameters we have extensively reviewed literature, concluding a quite reliable range of kinetic parameters that could fit in different experimental conditions. The results prove that the selection of these values was correct and thus we successfully reproduced several NH3 adsorption and TPD experiments. It is important to recall, however, that in all tests catalysts with a Si/ Al ratio of 13.5 were used. Mishin et al. have shown that the Si/ Al ratio significantly affects the activation energy of NH3 desorption over various zeolites including ZSM5.70 Thus, it is likely that the obtained range of kinetic parameters would possibly not be applicable for the simulation of experiments involving samples with different Si/Al ratios. Moreover during simulations we have observed significant deviations between kinetic parameters used for different samples, characterized by the same Si/Al ratio, highlighting the fact that additional phenomena should be also taken into consideration (namely 8446



■

Article

AUTHOR INFORMATION

Sgeom: geometrical catalytic surface based on the open section ratio of the catalyst [m2] Sh: Sherwood number [−] Tg,Ts: temperatures of the gas and the solid phase respectively [K] u: gas velocity [m/s] Vs: volume of the solid phase [m3] Vwash: volume of the washcoat [m3]

Corresponding Author

*Tel.: +33 (0) 147525833. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS

Greek Symbols

Stavros Skarlis kindly acknowledges the EU FP7 Marie Curie Initial Training Network (ITN) VECOM.

■

NOMENCLATURE Aads.: pre-exponential factor of NH3 adsorption [m3/s kgzeolite] Ades.: pre-exponential factor of NH3 adsorption [mol/s kgzeolite] Bim: Biot number for mass transfer phenomena [−] cg,i_in,cg,i_out: inlet and outlet concentration flow rate of species [mol/(m3 s)] cg,i, cs,i: concentration of the i species in the gas and the solid phase respectively [mol/m3] CNH3: concentration of ammonia in the gas phase [mol/m3] Cps: monolith’s thermal capacity [W/(kg K)] Dax: axial diffusion coefficient [m2/s] De: effective diffusion coefficient, computed through the Bosanquet equation Di: diffusivity of gaseous species i [m2/s] dh: channel’s hydraulic diameter [m] e: characteristic length of the washcoat [m] Eads.: activation energy of NH3 adsorption [J/mol] Edes.: activation energy of NH3 desorption [J/mol] Eo: activation energy of NH3 desorption for surface coverage equal to zero [J/mol] hs: convection’s coefficient for the heat transfer (due to convection) between the gas and the solid phase [W/ (m2·K)] k(Ts): reaction rate constant [1/s] kads(Ts): rate constant for NH3 adsorption [m3/s kgzeolite] kdes(θj,Ts): rate constant for NH3 desorption [mol/s kgzeolite] km,i: mass transfer coefficient for the external diffusion of i species [m/s] Mi: molar weight of i species [kg/mol] mzeolite: mass of zeolite [kg] Nacidic: NH3 storage capacity over acidic sites of bare ZSM5 [mol/kgzeolite] Nacidic_exch: NH3 storage capacity over acidic sites of FeZSM5 [mol/kgzeolite] Nj: storage capacity over j surface site (j = S1, S2, S3, S4, and S5) [mol/kgzeolite] Nmetallic: NH3 storage capacity over iron−metallic sites of FeZSM5 [mol/kgzeolite] Ntot: NH3 total storage capacity over bare ZSM5 [mol/ kgzeolite] Nweak: NH3 storage capacity over weak sites of bare ZSM5 [mol/kgzeolite] R: global gas constant [8.314 J/mol K] Rg: gas mixture’s constant [kg/mol K] Rj: reaction rate of [J/mol] RNH3_ads,RNH3_des: reaction rate of the NH3 adsorption and desorption respectively [mol/s kgzeolite] s: surface area to volume ratio [1/m]

■

α: constant for the description of the dependence of the activation energy of desorption from the surface coverage [−] Δhi: specific enthalpy of ith species formation [W/kg] ηexch: percentage of the ion exchange degree [−] ηi: effectiveness factor for mass transport phenomena in the gas and the solid phase [−] θj: surface coverage of j surface site (j = S1, S2, S3, S4, and S5) [−] ρg: density of gas mixture [kg/m3] ρs: density of the solid phase [kg/m3] ϕ: Thiele modulus [−] ϕfilm: chemical source term due to ammonia’s adsorption and desorption, including mass transfer phenomena in both gas and solid phase [mol/m3 s] Φw_ext: global external heat transfers [W] vj: stoichiometric coefficient of NH3 adsorption and desorption over each site [−] ωi: chemical source term due to the chemical reactions for the species i [mol/s]

REFERENCES

(1) Klingstedt, F.; Arve, K.; Eränen, K.; Murzin, D. Y. Acc. Chem. Res. 2006, 39, 273−282. (2) U.S. Environmental Protection Agency. http://www.epa.gov/ otaq/hd-hwy.htm#regs (accessed Sept. 12, 2011). (3) European Union. http://europa.eu/legislation_summaries/ environment/air_pollution/l28186_en.htm. (accessed Sept. 12, 2011). (4) Badr, O.; Probert, S. D. Appl. Energy 1993, 46, 1−67. (5) Grassian, V. H. Environmental Catalysis. Fundamental Concepts in Molecular Simulation of NOx Catalysis; Grassian, V. H., Ed.; Taylor & Francis Group: London, 2005; Vol. 10, pp 233−234. (6) Kley, D.; Kleinmann, M.; Sanderman, H.; Krupa, S. Environ. Pollut. 1999, 100, 19−42. (7) Katsouyanni, K.; Karakatsani, A.; Messari, I.; Touloumi, G.; Hatzakis, A.; Kalandidi, A.; Trichopoulos, D. J. Epidemiol. Community Health 1990, 44, 321−324. (8) Pantazopoulou, A.; Katsouyanni, K.; Koureakremastinou, J.; Trichopoulos, D. Environ. Res. 1995, 69, 31−36. (9) Granger, P.; Parvulescu, V. I. Chem. Rev. 2011, 111, 3155−3207. (10) Koltsakis, G. C.; Haralampous, O. A.; Koutoufaris, I. Z. SAE Tech. Paper Ser. 2010, 2010−01−0893. (11) Rankovic, N.; Nicolle, A.; Berthout, D.; Da Costa, P. J. Phys. Chem. C 2010, 114, 7102−7111. (12) Baik, J. H.; Yim, S. D.; Nam, I.-S.; Lee, J.-H.; Cho, B. K.; Oh, S. H. Selective Catalytic Reduction of Diesel Engine NOx Emissions with NH3/Urea. 18th North American Catalysis Society Meeting, 2003. (13) Chatterjee, D.; Burkhardt, T.; Weibel, M.; Tronconi, E.; Nova, I.; Ciardelli, C. SAE Tech. Paper Ser. 2006, 2006−01−0468. (14) Tomasic, V. Catal. Today 2007, 119, 106−113. (15) Balle, P.; Geiger, B.; Kureti, S. Appl. Catal. B: Environ. 2009, 85, 109−119. (16) Ciardelli, C.; Nova, I.; Tronconi, E.; Chatterjee, D.; BandlKonrad, B.; Weibel, M.; Krutzsch, B. Appl.Catal. B: Environ. 2007, 70, 80−90. (17) Kamasamudram, K.; Currier, N.; Szailer, T.; Yezerets, A. SAE Int. J. Fuels Lubr. 2010, 3, 664−672. 8447



Article

(53) Katada, N.; Igi, H.; Kim, J. H. J. Phys. Chem. B 1997, 101, 5969− 5977. (54) Rodríguez-González, L.; Hermes, F.; Bertmer, M.; RodríguezCastellón, E.; Jiménez-López, A.; Simon, U. Appl. Catal. A: General 2007, 328, 174−182. (55) Akah, A.; Cundy, C.; Garforth, A. Appl. Catal. B: Environ. 2005, 59, 221−226. (56) Bagnasco, G. J. Catal. 1996, 159, 249−252. (57) Matsuura, H.; Katada, N.; Niwa, M. Microporous Mesoporous Mater. 2003, 66, 283−296. (58) Brandenberger, S.; Kröcher, O.; Wokaun, A.; Tissler, A.; Althoff, R. J. Catal. 2009, 268, 297−306. (59) Long, R. Q.; Yang, R. T. J. Catal. 2001, 198, 20−28. (60) Mhamdi, M.; Khaddar-Zine, S.; Ghorbel, A. Appl. Cata. A: General 2008, 337, 39−47. (61) Anunziata, O. A.; Beltramone, A. R.; Juric, Z.; Pierella, L. B.; Requejo, F. G. Appl. Catal. A: General 2004, 264, 93−101. (62) Mirodatos, C.; Barthomeuf, D. J. Chem. Soc., Chem. Commun. 1981, 39−40. (63) Borade, R.; Sayari, A.; Adnot, A.; Kaliaguine, S. J. Phys. Chem. 1990, 94, 5989−5994. (64) Chen, T.; Men, A.; Sun, P.; Zhou, J.; Yuan, Z.; Guo, Z.; Wang, J.; Ding, D.; Li, H. Catal. Today 1996, 30, 189−192. (65) Chen, X.; Li, W.; Schwank, J. W. Catal. Today 2011, 175, 2−11. (66) Chorkendorff, I.; Niemantsverdriet, J. W. Concepts of Modern Catalysis and Kinetics. Willey -VCH Verlag GmbH & Co. KGaA: Weinheim, Germany, 2003; pp 113−128. (67) Metkar, P. S.; Salazar, N.; Muncrief, R.; Balakotaiah, V.; Harold, M. P. Appl. Catal. B: Environ. 2011, 104, 110−126. (68) Pinto, R. R.; Borges, P.; Lemos, M. A. N. D.; Lemos, F.; Védrine, J. C.; Derouane, E. G.; Ribeiro, F. R. Appl. Catal. A: General 2005, 284, 39−46. (69) Sahoo, S. K.; Viswanadham, N.; Ray, N.; Gupta, J. K.; Singh, I. D. Appl. Catal. A: General 2001, 205, 1−10. (70) Mishin, I.; Brueva, T.; Kapustin, G. Adsorption 2005, 11, 415− 424. (71) Wilken, N.; Kamasamudram, K.; Currier, N. W.; Li, J.; Yezerets, A.; Olsson, L. Catal. Today 2010, 151, 237−243. (72) Rodríguez-González, L.; Rodríguez-Castellón, E.; JiménezLópez, A.; Simon, U. Solid State Ionics 2008, 179, 1968−1973. (73) Sharma, S. B.; Meyers, B. L.; Chen, D. T.; Miller, J.; Dumesic, J. A. Appl. Catal. A: General 1993, 102, 253−265. (74) Auroux, A. Top. Catal. 1997, 4, 71−89. (75) Hidalgo, C. V.; Itoh, H.; Hattori, T.; Niwa, M.; Murakami, Y. J. Catal. 1984, 85, 362−369. (76) Sazama, P.; Wichterlova, B.; Dedecek, J.; Tvaruzkova, Z.; Musilova, Z.; Palumbo, L.; Sklenak, S.; Gonsiorova, O. Microporous Mesoporous Mater. 2011, 143, 87−96. (77) Kapustin, G. I.; Brueva, T. R.; Klyachko, A. L.; Beran, S.; Wichterlova, B. Applied Catalysis 1988, 42, 239−246. (78) Sugawara, K.; Nobukawa, T.; Yoshida, M.; Sato, Y.; Okumura, K.; Tomishige, K.; Kunimori, K. Appl. Catal. B: Environ. 2007, 69, 154−163. (79) Brandenberger, S.; Kröcher, O.; Tissler, A.; Althoff, R. Appl. Catal. A: General 2010, 373, 168−175. (80) Lanzani, G.; Laasonen, K. Int. J. Hydrogen Energy 2010, 35, 6571−6577. (81) Ates, A. Appl. Catal. B: Environ. 2007, 76, 282−290. (82) Wahba, M.; Kemball, C. Trans. Faraday Soc. 1953, 49, 1351− 1360. (83) Dumesic, J. A.; Trevino, A. A. J. Catal. 1989, 116, 119−129. (84) Connell, G.; Dumesic, J. A. J. Catal. 1987, 105, 285−298. (85) John, W. J. Catal. 1967, 9, 225−236. (86) Forni, L.; Vatti, F. P.; Ortoleva, E. Microporous Mater. 1995, 3, 367−375.

(18) Oliver Kröcher New Challenges for urea-SCR systems: From vanadia-based to zeolite based SCR catalysts. IAV MinNOx Conference, Germany, 2007. (19) Rahkamaa-Tolonen, K.; Maunula, T.; Lomma, M.; Huuhtanen, M.; Keiski, R. L. Catal. Today 2005, 100, 217−222. (20) Brandenberger, S.; Kröcher, O.; Tissler, A.; Althoff, R. Catalysis Reviews 2008, 50, 492−531. (21) Grossale, A.; Nova, I.; Tronconi, E.; Chatterjee, D.; Weibel, M. J. Catal. 2008, 256, 312−322. (22) Malmberg, S.; Votsmeier, M.; Gieshoff, J.; Söger, N.; Mußmann, L.; Schuler, A.; Drochner, A. Top. Catal. 2007, 42−43, 33−36. (23) Grossale, A.; Nova, I.; Tronconi, E. J. Catal. 2009, 265, 141− 147. (24) Chmielarz, L.; Kustrowski, P.; Dziembaj, R.; Cool, P.; Vansant, E. F. Appl. Catal. B: Environ. 2006, 62, 369−380. (25) Č apek, L.; Kreibich, V.; Dĕdeček, J.; Grygar, T.; Wichterlová, B.; Sobalík, Z.; Martens, J. A.; Brosius, R.; Tokarová, V. Microporous Mesoporous Mater. 2005, 80, 279−289. (26) Long, R. Q.; Yang, R. T. J. Catal. 2000, 194, 80−90. (27) Devadas, M. Selective Catalytic Reduction (SCR) of Nitrogen Oxides with Ammonia over Fe-ZSM5; Ph.D Thesis, ETH Zurich Schwitzerland, 2006. (28) Bucko, T.; Benco, L.; Hafner, J.; Cejka, J. J. Chem. Phys. 2005, 120, 10263−10277. (29) Olsson, L.; Sjövall, H.; Blint, R. J. Appl. Catal. B: Environ. 2008, 81, 203−217. (30) Schuler, A.; Votsmeier, M.; Kiwic, P.; Gieshoff, J.; Hautpmann, W.; Drochner, A.; Vogel, H. Chem. Eng. J. 2009, 154, 333−340. (31) Vollmer, J. M.; Stefanovich, E. V.; Truong, T. N. J. Phys. Chem. B 1999, 103, 9415−9422. (32) Sjövall, H.; Blint, R. J.; Olsson, L. J. Phys. Chem. C 2009, 113, 1393−1405. (33) Sjövall, H.; Blint, R. J.; Gopinath, A.; Olsson, L. Ind. Eng. Chem. Res. 2009, 49, 39−52. (34) Klukowski, D.; Balle, P.; Geiger, B.; Wagloehner, S.; Kureti, S.; Kimmerle, B.; Baiker, A.; Grunwaldt, J. D. Appl. Catal. B: Environ. 2009, 93, 185−193. (35) Chatterjee, D.; Kocì, P.; Schmeißer, V.; Marek, M.; Weibel, M.; Krutzsch, B. Catal. Today 2010, 151, 395−409. (36) Mauviot, G.; Le Berr, F.; Raux, S.; Perretti, F.; Malbec, L.; Millet, C. Oil Gas Sci. Technol.−Rev. IFP 2009, 64, 285−307. (37) Depcik, C.; Assanis, D. Prog. Energy Combust. Sci. 2005, 31, 308−369. (38) Kumar, V.; Paraschivoiu, M.; Nigam, K. D. P. Chem. Eng. Sci. 2011, 66, 1329−1373. (39) Santos, H.; Costa, M. Int. J. Heat Mass Transfer 2008, 51, 1409− 1422. (40) Hayes, R. E.; Liu, B.; Moxom, R.; Votsmeier, M. Chem. Eng. Sci. 2004, 59, 3169−3181. (41) Froment, G. F. Appl. Catal. 1986, 22, 3−20. (42) J.Villermaux Génie de la réaction chimique. Conception et fonctionnement des réacteurs; Lavoiser Tec & Doc, 1993. (43) Groppi, G.; Belloli, A.; Tronconi, E.; Forzatti, P. AIChE J. 1995, 41, 2250−2260. (44) Carberry, J. J.; Kulkarni, A. A. J. Catal. 1973, 31, 41−50. (45) Mladenov, N.; Koop, J.; Tischer, S.; Deutschmann, O. Chem. Eng. Sci. 2010, 65, 812−826. (46) Zalc, J. M.; Reyes, S. C.; Iglesia, E. Chem. Eng. Sci. 2004, 59, 2947−2960. (47) Santos, H.; Costa, M. Int. J. Heat Mass Transfer 2008, 51, 1409− 1422. (48) Gorte, R. J. Catal. Today 1996, 28, 405−414. (49) Costa, C.; Dzikh, I. P.; Lopes, J. M.; Lemos, F.; Ribeiro, F. R. J. Mol. Catal. A: Chem. 2000, 154, 193−201. (50) Gorte, R. J. Catal. Lett. 1999, 62, 1−13. (51) Topsøe, N. Y.; Pedersen, K.; Derouane, E. G. J. Catal. 1981, 70, 41−52. (52) Katada, N.; Miyamoto, T.; Begum, H. A.; Naito, N.; Niwa, M.; Matsumoto, A.; Tsutsumi, K. J. Phys. Chem. B 2000, 104, 5511−5518. 8448


Multisite Modeling of NH3 Adsorption and Desorption over Fe-ZSM5

Recommend Documents