`
`www.elsevier.com/locate/jcat
`
`Hydrothermal stability of CuZSM5 catalyst in reducing NO by NH3
`for the urea selective catalytic reduction process
`Joo-Hyoung Park a, Hye Jun Park a, Joon Hyun Baik a, In-Sik Nam a,∗
`Jong-Hwan Lee c, Byong K. Cho c, Se H. Oh c
`
`, Chae-Ho Shin b,
`
`a Department of Chemical Engineering/School of Environmental Science and Engineering, Pohang University of Science and Technology (POSTECH), Pohang,
`Korea
`b Department of Chemical Engineering, Chungbuk National University, Chungbuk, Korea
`c General Motors R&D Planning Center, Warren, MI, USA
`Received 6 September 2005; revised 30 December 2005; accepted 6 March 2006
`
`Available online 5 April 2006
`
`Abstract
`
`To confirm the hydrothermal stability of CuZSM5 for urea selective catalytic reduction (SCR), NO removal activity over a series of the catalyst
`containing various amounts of copper, ranging from 1 to 5 wt%, was examined before and after hydrothermal treatment under a simulated feed
`◦
`C. The degree of catalyst aging varies with respect to the copper content of
`gas stream containing 10% water at a temperature range of 600–800
`the catalyst and the aging temperature. The optimal copper content seems to be about 4 wt% (nearly 125% in terms of ion-exchange level) from
`the standpoint of catalyst aging. The catalysts before and after aging were characterized by XRD, 27Al-MAS-NMR, BET, XAFS, and ESR to gain
`insight into the sintering mechanism of CuZSM5 for the urea SCR process. A slight alteration of the catalyst structure was observed by XRD and
`BET analysis on catalyst aging. The copper ions on the surface of CuZSM5 catalysts on aging remained as either isolated Cu2+
`and/or an oxide
`form of copper, as confirmed by EXAFS. However, ESR spectra of the catalysts clearly revealed four distinct local structures of Cu2+
`species
`in the framework of ZSM5 in the form of a square pyramidal site, a square planar site, and two unresolved distorted sites. Migration of Cu2+
`ions from the square pyramidal and/or the square planar sites of the isolated cupric ions to the two unresolved sites occurs during hydrothermal
`treatment. This causes an alteration of the population of isolated cupric ions on the square pyramidal and/or planar sites and is responsible for the
`hydrothermal stability of the CuZSM5 catalyst in the urea SCR process.
`© 2006 Elsevier Inc. All rights reserved.
`
`Keywords: Urea SCR; Deactivation; Hydrothermal stability; ESR; XAFS; CuZSM5
`
`1. Introduction
`
`The selective catalytic reduction (SCR) of NOx with urea is
`a recognized well-developed technology to remove NOx from
`light- and heavy-duty diesel engines [1,2]. For the actual diesel
`engine exhaust system, the H2O content generated from the
`combustion of diesel fuel with a high carbon number is dis-
`tinctive in the gas stream, and the development of a hot spot
`in the catalytic converter from the sudden burning of locally
`collected particulate can be expected. Consequently, achieving
`
`* Corresponding author. Fax: +82 54 279 8299.
`E-mail address: isnam@postech.ac.kr (I.-S. Nam).
`
`0021-9517/$ – see front matter © 2006 Elsevier Inc. All rights reserved.
`doi:10.1016/j.jcat.2006.03.001
`
`hydrothermal stability of the catalyst is a critical issue in the
`commercial application of urea SCR technology to the exhaust
`stream from diesel engines [3].
`CuZSM5 is one of the most promising catalysts for the SCR
`◦
`C,
`of NOx by urea at an exhaust temperature of around 150
`particularly for light-duty diesel engines [3,4]. For the decom-
`position of urea, 1 mol of urea is thermally decomposed and
`then easily hydrolyzed on the catalyst surface to produce 2 mol
`of ammonia and 1 mol of carbon dioxide, leading to the practi-
`cally equivalent overall reaction to NH3 SCR of NO [3–7].
`Numerous studies have been conducted regarding the na-
`ture of Cu on the surface of zeolite, which is generally recog-
`nized as an active reaction site for the present reaction system,
`
`BASF-2042.001
`
`
`
`48
`
`J.-H. Park et al. / Journal of Catalysis 240 (2006) 47–57
`
`and its reaction mechanism through redox chemistry [8–14].
`It has been commonly accepted that isolated Cu2+
`and Cu–
`O–Cu dimer species on the catalyst surface play key roles in
`the NH3 SCR reaction. Mizumoto et al. [12] prepared Cu-Y
`zeolite containing mainly isolated Cu2+
`to suggest a Langmuir–
`Hinshelwood-type reaction mechanism by the reaction between
`strongly adsorbed NH3 and weakly adsorbed NO on isolated
`copper ions over the catalyst surface. Choi et al. also confirmed
`a similar mechanism by temperature-programmed desorption
`experiments of NO and NH3 over Cu-exchanged Mordenite-
`type zeolite catalyst [13]. Komatsu et al. [14] studied NH3 SCR
`for a series of Cu zeolites with varying copper exchange lev-
`els and Si/Al ratios and concluded that the active reaction site
`of the CuZSM5 catalyst for NH3 SCR is copper dimer formed
`on the catalyst surface. The active reaction site on the surface
`of Cu-exchanged zeolite has been also investigated for the de-
`composition of NO and HC (hydrocarbon) SCR by lumines-
`cence, Fourier transform infrared spectroscopy, electron spin
`resonance (ESR), and X-ray absorption fine structure (XAFS)
`studies [15–23]. Iwamoto et al. proposed oxygen-bridged bin-
`uclear [Cu–O–Cu]2+
`in CuZSM5 for NO decomposition [15],
`and Sárkány et al. reported experimental evidence indicating
`that the complex forms all at once [16]. Ddeˇcek et al. examined
`the four distinctive sites of Cu ion coordination over various
`zeolite structures, including MFI, MOR, FER, BEA, and FAU,
`+
`emission
`using a multispectroscopic approach including Cu
`spectra, in situ infrared spectroscopy for Cu2+
`-adsorbing NO,
`and ESR of Cu2+
`[17].
`But little literature exists on the inactivation of the active re-
`action sites for NH3 SCR, although the hydrothermal stability
`of zeolite has long been considered a necessary issue to resolve
`for the commercial application of zeolite-type catalyst to auto-
`motive engines from the standpoint of alteration of the active
`reaction sites, including isolated Cu2+
`, and Cu2+
`dimer during
`the reaction. It has been commonly observed that the deacti-
`vation of SCR activity over CuZSM5 catalyst is due mainly
`to the degradation of zeolite support and/or the formation of
`Cu-aluminate by dealumination, the decrease in active reaction
`sites by the transformation of Cu2+
`to CuO, the redistribution
`of the reaction sites through the migration of Cu2+
`, or a com-
`bination of these mechanisms [19,20,24–28]. Furthermore, the
`optimal copper content on the catalyst surface has been little
`examined from the standpoint of catalyst aging for commercial
`application.
`The purpose of the present study was to optimize the copper
`content on the surface of ZSM5 catalyst from the standpoint
`of catalyst aging by evaluating the hydrothermal stability of a
`series of CuZSM5 catalysts. To accomplish this goal, various
`complementary spectroscopic techniques were used and their
`results compared. In particular, XAFS and ESR studies were
`carried out to investigate the nature of copper ions on the sur-
`face of CuZSM5 catalyst, such as the oxidation states of copper
`and the coordination structure of Cu2+
`in the framework of
`zeolite on the catalyst sintering with respect to aging temper-
`ature.
`
`Table 1
`Physicochemical properties of the catalysts employed in the present study
`
`Cu
`content
`(wt%)
`1.95
`
`Ion exchange
`level
`(%)
`61
`
`2.93
`
`92
`
`3.87
`
`124
`
`4.73
`
`150
`
`BET
`(m2/g)
`
`400
`
`372
`363
`
`387
`
`357
`321
`
`383
`355
`324
`
`357
`337
`294
`191
`
`Micropore
`surface area
`(m2/g)
`377
`
`331
`321
`
`362
`
`319
`303
`
`354
`328
`285
`
`323
`301
`260
`182
`
`Sample
`
`CuZSM5-61-fresh
`CuZSM5-61-600
`CuZSM5-61-700
`CuZSM5-61-800
`
`CuZSM5-92-fresh
`CuZSM5-92-600
`CuZSM5-92-700
`CuZSM5-92-800
`
`CuZSM5-124-fresh
`CuZSM5-124-600
`CuZSM5-124-700
`CuZSM5-124-800
`
`CuZSM5-150-fresh
`CuZSM5-150-600
`CuZSM5-150-700
`CuZSM5-150-800
`
`2. Experimental
`
`2.1. Catalyst preparation
`type ZSM5 (Si/Al ratio = 14), obtained from Tosoh
`+
`NH4
`Corp., was used as a parent catalyst for preparing the series of
`catalysts in the present study. The ZSM5 was exchanged with a
`0.01 M (CH3CO2)2Cu·H2O (Aldrich, 98%) solution to obtain
`a Cu-based ZSM5 at room temperature for 5 h [3,4]. The room
`temperature for exchanging copper ions into ZSM5 catalyst was
`used to avoid the formation of copper oxides on the catalyst
`surface during the course of the ion exchange [4]. This was fol-
`◦
`◦
`C for 12 h and calcining at 500
`C in
`lowed by drying at 110
`air for 5 h. The copper content was controlled by repeating the
`ion-exchange procedure.
`To investigate the hydrothermal stability of CuZSM5, the
`catalyst samples were sintered under a simulated feed gas
`stream containing 10% H2O in air balance with a flow rate
`◦
`C for 24 h. Cu-based
`of 500 cc/min at 600, 700, and 800
`ZSM5 catalysts containing a various copper loadings with re-
`spect to the aging temperatures were obtained and designated as
`CuZSM5-x-y, where x and y represent the percentage degree
`of Cu2+
`ion exchange, and the aging temperature, respectively.
`Table 1 lists the physicochemical properties of the catalysts pre-
`pared in this study.
`
`2.2. Reactor system and experimental procedure
`
`For urea SCR technology, urea is thermally decomposed into
`one mol of ammonia and one mol of isocyanic acid initially,
`and the isocyanic acid formed by the thermal decomposition
`of urea can readily undergo hydrolysis on the catalyst surface
`to produce another mol of ammonia [7]. Thus, the complete
`decomposition of 1 mol of urea produces 2 mol of ammonia
`and 1 mol of carbon dioxide [1,5–7],
`H2N–CO–NH2 + H2O → 2NH3 + CO2.
`
`(1)
`
`BASF-2042.002
`
`
`
`J.-H. Park et al. / Journal of Catalysis 240 (2006) 47–57
`
`49
`
`N2-filled ionization chambers. CuO, Cu2O, and Cu metal were
`used as reference compounds.
`The data analyses for EXAFS were carried out by a standard
`procedure. The inherent background in the data was removed
`by fitting a polynomial to the edge region and then extrapolat-
`ing through the entire spectrum from which it was subtracted.
`The resulting spectra, μ(E), were normalized to an edge jump
`of unity for comparing the X-ray absorption near-edge struc-
`tures (XANES) directly. The EXAFS function, χ (E), was ob-
`tained from χ (E) = {μ(E) − μ0(E)}/μ0(E) [29]. The result-
`ing EXAFS spectra were k2-weighted to compensate for the
`attenuation of EXAFS amplitude at high k and then Fourier
`−1 (cid:2) k (cid:2) 11.5 Å
`−1 with
`transformed in the range of 2.0 Å
`−1. To determine the
`a Kaiser–Bessel function of dk = 1 Å
`structural parameters, nonlinear least squares curve fitting was
`performed in the range of R (cid:2) ∼2 Å corresponding to the dis-
`tance to the Cu–O in the central-atom phase-corrected Fourier-
`transformed (FT) spectra, using the UWXAFS and IFEFFIT
`packages according to the following EXAFS formula [30]:
`(cid:4)
`(cid:4)
`χ (k) = −S2
`Ni
`Fi (k) exp
`kR2
`(cid:4)
`i
`2kRi + φi (k)
`× sin
`where the backscattering amplitude, Fi (k), the total phase shift,
`φi (k), and the photoelectron mean free path, λ(k), were theo-
`retically calculated for all scattering paths, including multiple
`ones by a curved wave ab initio EXAFS code FEFF8 [31].
`The ESR spectra were obtained at room temperature in the
`X-band using a JEOL model JES-TE300 spectrometer. The
`ESR signals of Cu2+
`were recorded in the field region from
`2000 to 4000 G with a sweep time of 10 min. A coaxial quartz
`cell was placed in the ESR cavity and connected with stainless
`steel capillaries of the flow system by Teflon ferrules. The sam-
`◦
`C in pure O2, evacuated, and sealed off
`ples were heated at 500
`in situ. To provide the maximum accuracy of the ESR signals,
`the packing height of the ampoule was constant in all cases,
`with the center of the sample positioned in the middle of the
`ESR cavity. The ESR signal from DPPH (g(cid:6) = 2.0036) was
`used as an internal standard. The Origin and Excel programs for
`Windows were used for the baseline correction, analysis of the
`fourfold hyperfine lines, and double-integration of the recorded
`ESR spectra.
`
`,
`
`(2)
`
`(cid:3)−2σ 2
`i k2
`
`(cid:3)−2Ri /λ(k)
`
`exp
`
`(cid:2) i
`
`(cid:3)
`
`o
`
`3. Results and discussion
`
`3.1. Effect of aging on NO removal activity
`
`The NO removal activity for a series of CuZSM5 catalysts
`◦
`C, where catalyst de-
`before and after aging, particularly at 700
`activation is specifically evident, is given in Fig. 1. The figure
`shows a bell-shaped activity profile with respect to the reac-
`tion temperature, typical for NH3 SCR catalyst maintaining a
`◦
`wide operating temperature window (200–400
`C) containing
`NO conversion >90%. In a reaction temperature range higher
`◦
`C, NO conversion gradually decreases mainly due to
`than 400
`
`This leads to the practically equivalent overall reaction of NO to
`NH3 SCR. The identical deNOx performance of the CuZSM5
`catalyst by urea SCR to that by NH3 SCR was observed under
`the specific reaction conditions [3]. Therefore, an NH3 SCR
`test to confirm the hydrothermal stability of the series of the
`CuZSM5 catalysts for urea SCR was performed for experimen-
`tal convenience.
`The activity of the CuZSM5 catalysts for NO reduction by
`NH3 was examined in a fixed-bed flow reactor, typically con-
`taining ca. 1 g of the catalyst was sieved to a mesh size of 20/30
`to minimize the mass transfer limitations of the catalyst. Be-
`fore the evaluation of activity, the catalyst was pretreated in situ
`with a total flow of 3300 cc/min containing 79% N2 and 21%
`◦
`C, and then cooled to room temperature. A reaction
`O2 at 500
`gas mixture consisting of 500 ppm NO, 500 ppm NH3, 5% O2,
`and 10% H2O in N2 balance was fed into the reactor system
`through Brooks mass flow controllers (model 5850E). A total
`flow rate of 3300 cc/min (corresponding to GHSV = 100,000
`−1) was mainly used for the catalyst activity testing. The NO
`h
`concentration was analyzed by an on-line chemiluminescence
`NO–NOx analyzer (Thermo Electron, model 42H). The details
`of the reactor system have been given elsewhere [3,7].
`
`2.3. Catalyst characterization
`
`BET surface areas were measured by Micromeritics ASAP
`2010 sorption analyzer with a static volumetric technique,
`based on the amount of N2 adsorbed at liquid N2 temperature.
`◦
`C in vacuum for 5 h before
`The samples were degassed at 200
`the adsorption measurements. BET surface area and micropore
`volume were calculated by the t-plot method. The micropore
`surface area was obtained by subtracting the external surface
`area from the total surface area.
`Powder XRD patterns for the catalysts were observed by us-
`ing an M18XHF X-ray diffractometer with Ni-filtered Cu-Kα
`radiation (λ ∼ 1.54184 Å). Data were collected in angles rang-
`◦
`◦
`◦
`to 90
`with a step size of 0.02
`and a step time of
`ing from 5
`10 s under continuous sample rotation during the scan.
`The 27Al magic-angle spinning nuclear magnetic resonance
`(MAS-NMR) spectra were obtained on a Bruker AVANCE 500
`spectrometer at an 27Al frequency of 130.325 MHz in 4-mm
`rotors at a spinning rate of 10.0 kHz. The spectra were obtained
`with the acquisition of ca. 10,000 pulse transients, which were
`reported with a π/4 rad pulse length of 5.00 µs and a recycle
`delay of 1.0 s.
`Extended X-ray absorption fine structure (EXAFS) spectro-
`scopic measurements were performed with the synchrotron ra-
`diation by using the EXAFS facility installed at beamline 3C1
`in the Pohang Accelerator Laboratory. The ring was operated
`at 2.5 GeV with 200 mA electron current and 1% coupling.
`The spectra were measured with an Si(111) channel-cut mono-
`chromator with an energy resolution of E/E = 2× 10
`−4 that
`remained constant at the Cu K-edge (8979 eV).
`The samples spread uniformly between adhesive tapes to ob-
`tain an optimal absorption jump. All of the data were recorded
`in a transmission mode at room temperature, and the intensities
`of the incident and transmitted beams were measured in 100%
`
`BASF-2042.003
`
`
`
`50
`
`J.-H. Park et al. / Journal of Catalysis 240 (2006) 47–57
`
`ation of the chemical state of copper on the catalyst surface
`[19,20,24–28].
`
`3.2. Confirming the structure of the catalyst support, ZSM5,
`by 27Al MAS-NMR and powder XRD
`
`27Al MAS-NMR spectra were examined for the CuZSM5-
`150 catalyst before and after hydrothermal treatment with the
`simulated aging gas stream containing 10% water, as shown in
`Fig. 4. No shoulder due to extra-framework AlO4 units was de-
`tected even for CuZSM5-150-800, except for that assigned to
`tetrahedral Al species. It may be direct evidence for the main-
`tenance of the catalyst structure without dealumination from
`the framework of zeolite due to hydrothermal treatment even
`◦
`at 800
`C. Identical NMR results were observed for all of the
`catalysts before and after aging. However, the intensity of the
`Al tetrahedral peak decreased, as also shown in Fig. 4, proba-
`bly due to the paramagnetic field of the Cu ions at their ion-
`exchange positions in the framework of zeolite structure. This
`characteristic of the Cu ions also produces an intensified and ex-
`tended side band of the NMR spectra, which may be originated
`from the anisotropic dipolar interaction between the electron
`spin of the Cu ions and the nuclei spin of the Al in the zeolite
`framework [33].
`Fig. 5 shows the typical powder XRD patterns for the series
`of the catalysts prepared before and after hydrothermal treat-
`ment. The relative intensities of X-ray peaks decrease due to
`the amount of Cu loadings on the catalyst surface and the aging
`temperature, particularly for underexchanged CuZSM5 cata-
`lysts. In addition, a fairly obvious weaker crystallinity for the
`overexchanged CuZSM5 compared with the underexchanged
`one is seen. It closely agrees with the decreasing trends of BET
`and micropore surface areas on the hydrothermal treatment of
`the catalysts as listed in Table 1. The irreversible decrease of the
`catalyst surface areas mainly causes a permanent loss of NO re-
`moval activity for overexchanged CuZSM5 catalysts aged at the
`high sintering temperature. It may be also attributed to the si-
`multaneous effects of the combination of high Cu loading and
`harsh aging conditions on the framework structure of ZSM5-
`type zeolite catalyst [34]. In addition, a broad XRD peak can be
`◦
`◦
`observed at 2θ of ca. 35.5
`and 39
`in the patterns of CuZSM5-
`150 catalysts, indicating formation of CuO on the catalyst sur-
`face, even for the fresh counterpart catalyst, but basically no
`CuO peak for the slightly underexchanged CuZSM5 catalyst,
`regardless of the aging temperatures. This may cause the de-
`crease of NO removal activity of overexchanged CuZSM5 cat-
`◦
`alyst aged at 800
`C, as mentioned earlier, but catalyst deactiva-
`tion for underexchanged CuZSM5 catalysts can barely be elu-
`cidated. Note that the additional XRD peaks for the formation
`of CuO and Cu2O on the catalyst surface of underexchanged
`CuZSM5 also cannot be identified, as shown in Fig. 5. Also
`note that formation of the Cu-aluminate related to the octahe-
`dral Al of the catalyst was hardly observed on the basis of the
`27Al MAS-NMR and XRD results, although we did not con-
`duct a TPR analysis to confirm its formation, as was done by
`Yan et al. [28].
`
`◦
`C
`Fig. 1. Activity of a series of CuZSM5 catalysts before and after aging at 700
`[(F) CuZSM5-61-fresh, (2) CuZSM5-92-fresh, (a) CuZSM5-124-fresh,
`(") CuZSM5-150-fresh,
`(E) CuZSM5-61-700,
`(1) CuZSM5-92-700,
`(e) CuZSM5-124-700, (!) CuZSM5-150-700] for the reduction of NO by
`NH3. Feed gas composition is 500 ppm NO, 500 ppm NH3, 5% O2, 10% H2O
`in N2 balance.
`
`(3)
`
`the oxidation of NH3 at the high reaction temperature with oxy-
`gen leading to a significant decrease in NH3 selectivity for NOx
`reduction as follows [32]:
`4NH3 + 3O2 → 2N2 + 6H2O.
`Among the catalysts before and after aging, CuZSM5-124 ex-
`hibited the best performance for NO removal activity main-
`◦
`C, which
`tenance, even at a reaction temperature below 200
`is critical for the application of the present urea SCR tech-
`nology to light-duty diesel engines. However, the activity for
`◦
`C decreases
`underexchanged CuZSM5-61 catalyst aged at 700
`significantly.
`Figs. 2 and 3 illustrate the aging effect of a series of
`CuZSM5 catalysts with respect to the copper content and
`the aging temperatures on NO removal activity in a reac-
`◦
`tion temperature range of 150–450
`C. For the catalysts con-
`taining <4% copper loading, NO removal activity increases
`as the Cu loading increases, regardless of the catalyst ag-
`ing. However, for a highly overexchanged catalyst such as
`CuZSM5-150, the activity decrease due to the catalyst sinter-
`ing is somewhat milder than that for slightly overexchanged
`catalyst, CuZSM5-124. For an underexchanged catalyst con-
`taining low copper content, such as CuZSM5-65, the worst
`catalyst deactivation can be observed, even for the catalyst
`◦
`aged at 600
`C. However, the cause of the sintering of Cu-
`based zeolite catalyst for SCR reaction has not yet been sys-
`tematically examined. It has been simply speculated, probably
`due to the structural destruction of thezeolite, and the alter-
`
`BASF-2042.004
`
`
`
`J.-H. Park et al. / Journal of Catalysis 240 (2006) 47–57
`
`51
`
`◦
`◦
`◦
`Fig. 2. Effect of Cu loading on the conversion of NO for CuZSM5 catalysts: (A) fresh, (B) aged at 600
`C, (C) aged at 700
`C, and (D) aged at 800
`C. Feed gas
`−1. [Reaction temperature: (") 150, (a) 175, (2) 200, (F) 300,
`composition is 500 ppm NO, 500 ppm NH3, 5% O2, 10% H2O in N2 balance, and SV is 100,000 h
`◦
`and (Q) 400
`C.]
`
`3.3. Local structure of copper ions by XAFS
`
`The shape and position of the Cu K-edge XANES provide
`information on the electronic structure and the local coordina-
`tion geometry of the absorbing Cu atom [22–24,35–42]. Fig. 6
`shows the XANES spectra for CuZSM5 catalysts, which are
`quite typical for the existence of the distorted octahedral Cu2+
`on the catalyst surface. The first derivatives of the XANES
`spectra for the catalyst samples on the hydrothermal treatment
`are given in Fig. 7; the general patterns are basically identical.
`In fact, the weak pre-edge peaks (I), corresponding to an elec-
`tric dipole-forbidden, but quadruply and vibronically allowed
`1s → 3d electronic transition and providing direct information
`on the oxidation state of copper, can be observed at ca. 8979 eV,
`
`species on the catalyst surface
`
`which is the fingerprint of Cu2+
`[23,36].
`The primary absorption peaks (II and III) for all of the cata-
`lysts appear at ca. 8986–8988 and 8995–8998 eV, respectively,
`which are attributed to the charge transfer from the metal ligand
`to the 3d orbital (ligand-to-metal charge transfer) correspond-
`ing to the degree of Cu(3d)–O(2p) bond covalence, and 1s →
`4p transition, respectively [23,36]. The shift of peak positions
`has been also observed at the lower absorption energy. The evo-
`lution of the absorption peaks (II and III) in Fig. 7 with respect
`to the Cu content and the aging condition may be due to the
`gradual polymerization of Cu2+
`ions with oxygen ion (O2−
`)
`on the catalyst surface. To identify and evaluate the peak posi-
`tion of the absorption edge (E0) defined as the maximum of the
`
`BASF-2042.005
`
`
`
`52
`
`J.-H. Park et al. / Journal of Catalysis 240 (2006) 47–57
`
`◦
`Fig. 3. Effect of aging temperatures on the conversion of NO for CuZSM5 catalysts at reaction temperatures of (A) 175, (B) 200, (C) 300, and (D) 400
`C. Feed
`−1 [(F) CuZSM5-61, (2) CuZSM5-92, (a) CuZSM5-124,
`gas composition is 500 ppm NO, 500 ppm NH3, 5% O2, 10% H2O in N2 balance, and SV is 100,000 h
`(") CuZSM5-150].
`
`first derivative of XANES spectra for CuZSM5 catalysts, the
`references including CuO, Cu2O, and Cu metal have been also
`analyzed and compared as listed in Table 2, where the apparent
`trend of the shift to the lower absorption energy can be seen.
`For the catalysts aged at the harshest sintering temperature
`◦
`of 800
`C except CuZSM5-61, an additional peak (IV) appears
`at ca. 8984 eV, which is due to the transition of copper ion into
`the bulk-type CuO [23]. It indicates that the copper on the sur-
`face of the catalyst can be transferred to a crystalline CuO by
`the hydrothermal treatment. These results were reconfirmed by
`EXAFS studies on the state of Cu on the catalyst surface.
`It has been generally recognized that EXAFS is an effec-
`tive tool for evaluating interatomic distances and coordination
`numbers to explore the geometric structure of reaction sites,
`due mainly to the interaction between an absorbing atom and
`a nearly surrounding atom [22,23,36,39,40]. Fig. 8 shows the
`radial distribution functions (RDFs) of Cu of CuZSM5 cata-
`lysts with CuO as a reference compound by Cu K-edge k2-
`
`Fig. 4. 27Al-MAS-NMR spectra of CuZSM5-150 before and after aging:
`◦
`C [∗ symbol indicating spinning
`(a) fresh, (b) aged at 600, (c) 700, and (d) 800
`sidebands].
`
`BASF-2042.006
`
`
`
`J.-H. Park et al. / Journal of Catalysis 240 (2006) 47–57
`
`53
`
`Fig. 5. XRD patterns of the under-exchanged (CuZSM5-92) and over-ex-
`changed (CuZSM5-150) before and after aging.
`(a) CuZSM5-92-fresh,
`(b) CuZSM5-92-700,
`(c) CuZSM5-92-800,
`(d) CuZSM5-150-fresh,
`(e) CuZSM5-150-700, and (f) CuZSM5-150-800. (Inset figure is a de-
`◦
`◦
`tailed XRD patterns for the identical catalysts ranged from 33
`to 43
`of the
`diffraction angle and the dashed lines are the characteristic peak positions of
`bulk CuO.)
`
`−1 (cid:2)
`weighted Fourier transforms (FTs) in the range of 2.0 Å
`−1 with a Kaiser–Bessel function. Dominant spectra
`k (cid:2) 11.5 Å
`with a Cu–O distance of ca 1.96 Å were commonly observed for
`the catalysts examined in this study [22,23]. As clearly shown in
`Fig. 9, the best fits of the EXAFS data for the CuZSM5 samples
`were attained; their parameters are summarized in Table 2. It is
`noteworthy that the coordination number of the Cu–O shell for
`CuZSM5 samples decreases with respect to the aging tempera-
`◦
`C, indicating that Cu2+
`ion may migrate
`tures from 600 to 800
`to the inaccessible sites or transform isolated Cu2+
`ions into
`small CuO-like phases.
`An additional weak Cu–(O)–Cu bond at ca. 2.95 Å [23]
`was seen for all but the underexchanged catalysts, including
`CuZSM5-61 and CuZSM5-92-fresh, due to the local structure
`of crystalline CuO formed on the catalyst surface. The am-
`plitude of the Cu–(O)–Cu peak for the CuZSM5 catalyst was
`much weaker than that for the bulk oxides, with coordination
`number <4 [23]. As the aging condition became more severe
`and the Cu loading content increased, the amplitude of the peak
`became evident. For highly overexchanged ZSM5 catalysts,
`a Cu–(O)–Cu peak was observed even for its fresh counterpart.
`It may be concluded that Cu zeolites prepared by the solution
`ion-exchange method can contain Cu–O–Cu dimeric oxoca-
`
`Fig. 6. Cu K-edge XANES spectra for CuZSM5 catalysts before and after
`aging: (A) CuZSM5-61, (B) CuZSM5-92, (C) CuZSM5-124, and (D) Cu-
`◦
`◦
`ZSM5-150 [(a) fresh, (b) aging at 600
`C, (c) aging at 700
`C, (d) aging at
`◦
`800
`C, and reference (e) CuO].
`
`tions that may be transferred to crystalline CuO as catalyst
`aging proceeds and/or the Cu loading content of the catalyst in-
`creases. Although the XAFS studies can examine the oxidation
`state of copper on the catalyst surface, the interatomic distance,
`and the coordination number of copper ion on the surface of
`CuZSM5 catalyst to characterize the geometric structure of the
`Cu atoms, it may not be sufficient to clearly assess the charac-
`teristics of Cu2+
`species, the amount of the isolated Cu2+
`, the
`degree of transformation of Cu2+
`into bulk-type CuO, and other
`features that may be critical to elucidate the aging mechanism
`of CuZSM5 catalyst to remove NO from automotive diesel en-
`gine by urea.
`
`3.4. Migration of copper among the reaction sites over the
`catalyst on aging
`To examine the alteration of the local characteristics of Cu2+
`ions by sintering, the CuZSM5 catalysts before and after hy-
`◦
`drothermal treatment were pretreated in situ at 500
`C for 2 h in
`pure O2 flow, evacuated, and sealed off for ESR study [22]. To
`
`BASF-2042.007
`
`
`
`54
`
`J.-H. Park et al. / Journal of Catalysis 240 (2006) 47–57
`
`Sample
`
`Sigmad R-factor
`
`CuZSM5-61-fresh
`CuZSM5-61-600
`CuZSM5-61-700
`CuZSM5-61-800
`
`CuZSM5-92-fresh
`CuZSM5-92-600
`CuZSM5-92-700
`CuZSM5-92-800
`
`CuZSM5-124-fresh
`CuZSM5-124-600
`CuZSM5-124-700
`CuZSM5-124-800
`
`5.3
`4.9
`5.1
`4.9
`
`5.6
`4.9
`5.1
`3.9
`
`5.4
`5.3
`5.0
`3.8
`
`0.0040
`0.0040
`0.0049
`0.0044
`
`0.0057
`0.0047
`0.0053
`0.0021
`
`0.0053
`0.0050
`0.0054
`0.0015
`
`0.0010
`0.0030
`0.0053
`0.0055
`
`0.0060
`0.0017
`0.0042
`0.0137
`
`0.0031
`0.0042
`0.0055
`0.0137
`
`Table 2
`Cu K-edge EXAFS least-square fitting results in CuZSM5 samples with respect
`to copper loadings and aging temperatures
`b Ec
`C.N.a Reff
`1.976 −6.6
`1.960 −7.7
`1.959 −8.3
`1.957 −7.8
`1.956 −7.5
`1.955 −6.4
`1.952 −7.2
`1.947 −7.4
`1.955 −7.1
`1.951 −7.3
`1.954 −6.8
`1.956 −5.2
`1.955 −7.1
`1.954 −7.6
`1.959 −6.6
`1.947 −6.4
`
`Edge
`energye
`8992.2
`8992.2
`8993.2
`8992.2
`
`8991.6
`8990.8
`8991.0
`8991.6
`
`8991.6
`8990.8
`8991.8
`8990.6
`
`8991.6
`8992.2
`8991.8
`8991.4
`
`0.0048
`0.0057
`0.0050
`0.0050
`
`0.0025
`0.0025
`0.0022
`0.0020
`
`5.1
`CuZSM5-150-fresh
`5.5
`CuZSM5-150-600
`5.3
`CuZSM5-150-700
`4.6
`CuZSM5-150-800
`a Coordination number.
`b Energy shift.
`c Interatomic distance.
`d Debye–Waller factor.
`e Determined from the position of the maximum of the derivative of XANES
`spectra for the samples.
`
`Fig. 7. First-derivative of XANES spectra for CuZSM5 catalysts before and
`after aging: (A) CuZSM5-61, (B) CuZSM5-92, (C) CuZSM5-124, and (D) Cu-
`◦
`◦
`ZSM5-150 [(a) fresh, (b) aging at 600
`C, (c) aging at 700
`C, (d) aging at
`◦
`800
`C, and reference (e) CuO].
`
`obtain better resolution of ESR spectra and investigate the ef-
`fect of measurement temperature on the state of Cu ions in the
`framework of the ZSM5 catalyst, the ESR studies were con-
`ducted at both room temperature and liquid nitrogen tempera-
`ture. No significant variation of the ESR spectra was observed.
`Fig. 10 shows the X-band ESR spectra for the CuZSM5-125
`catalysts, an axially symmetrical signal split into four hyper-
`fine lines that can be resolved as the parallel component of
`ESR spectra. This fourfold hyperfine splitting of the spectra
`was caused by the coupling between the 3d unpaired electron
`and the copper (I = 3/2) nuclear spin. For an orientationally
`disordered solid assuming axial symmetry, g anisotropy pro-
`duces a powder pattern in which the sharp features are referred
`to as parallel and perpendicular edges [20,22,24,41,42].
`The spectra of fresh samples exhibited two characteristic
`perpendicular g values: Cu2+
`in a square planar environment
`(by g(cid:6) = 2.27–2.30 and A(cid:6) = 150–170 G) and square pyra-
`midal (g(cid:6) = 2.33–2.35 and A(cid:6) = 135–145 G) [20,22,24,41,
`42]. For CuZSM-5 catalysts after aging, two additional Cu2+
`species, containing g(cid:6) = 2.30 and A(cid:6) = 165 G and g(cid:6) = 2.36
`
`and A(cid:6) = 130 G of ESR signals, have been found on the cata-
`lyst surface [20,24]. The exact geometry of the third and fourth
`Cu2+
`species is not completely understood, however.
`Based on the ESR parameters examined in the present study,
`the third species (g(cid:6) = 2.30 and A(cid:6) = 165 G) obtained over
`the aged catalyst can be coordinated to a distorted square sites
`anchored on the locations recessed from the main channels
`(i.e., in the side pockets of the zeolite structure), and the fourth
`species (g(cid:6) = 2.36 and A(cid:6) = 130 G) can be assigned to slightly
`distorted octahedral [20,43]. A recessed location would be ex-
`pected to alter the redox cycle of copper presumably during
`the SCR of NOx by impairing the accessibility of the reac-
`tants, hence decreasing the overall NO removal activity [18,42].
`Similar results were also observed for the CuZSM5 catalysts
`with differing copper contents. The g and hyperfine interaction
`constant values listed in Table 3 are nearly identical to those
`reported earlier for CuZSM5 [20,24].
`Double-integral (DI) values of ESR spectra for CuZSM5 cat-
`alysts were obtained for a comparative study to quantify the
`number of active Cu species on the catalyst surface, assum-
`ing the isolated Cu2+
`ions as the active reaction sites [22,42]
`for NO removal activity by NH3 SCR; these are given in Ta-
`ble 4. The altered number of reaction sites may be represented
`by the relative percentage of DI values of the ESR spectra for
`CuZSM5 catalysts, where the absorption area of ESR spectrum
`for CuZSM5-92-fresh is used as a basis for the calculation to
`confirm the migration of the copper ions among the reaction
`sites, an arbitrary criterion set for this comparative study. The
`relative percentages of DI values for fresh CuZSM5 catalyst
`
`BASF-2042.008
`
`
`
`J.-H. Park et al. / Journal of Catalysis 240 (2006) 47–57
`
`55
`
`Fig. 8. EXAFS RDF of Cu K-edge for CuZSM5 catalysts before and after
`aging: (A) CuZSM5-61, (B) CuZSM5-92, (C) CuZSM5-124, and (D) Cu-
`◦
`◦
`ZSM5-150 [(a) fresh, (b) aging at 600
`C, (c) aging at 700
`C, and (d) aging
`◦
`C, (e) CuO].
`at 800
`
`Fig. 9. Cu K-edge k2 weighted EXAFS data (!) and their best fits (–) for
`CuZSM5 catalysts before and after a