Effect of support synthesis methods on structure and ...

Catalysis Communications 84 (2016) 171–174

Contents lists available at ScienceDirect

Catalysis Communications journal homepage: www.elsevier.com/locate/catcom

Short communication

Effect of support synthesis methods on structure and performance of VOx/CeO2 catalysts in low-temperature NH3-SCR of NO Thanh Huyen Vuong a,b, Jörg Radnik a, Matthias Schneider a, Hanan Atia a, Udo Armbruster a, Angelika Brückner a,⁎ a b

Leibniz Institute for Catalysis at the University of Rostock, Albert-Einstein-Str. 29a, D-18059 Rostock, Germany School of Chemical Engineering, Hanoi University of Science and Technology, 1 Dai Co Viet, Hanoi 10000, Vietnam

a r t i c l e

i n f o

Article history: Received 20 May 2016 Received in revised form 28 June 2016 Accepted 2 July 2016 Available online 04 July 2016 Keywords: VOx/CeO2 catalysts Low temperature NH3-SCR Citrate sol-gel Preparation methods Operando-EPR Pseudo-in-situ-XPS

a b s t r a c t CeO2 supports were prepared by a citrate (C) or a precipitation method (P) before deposition of vanadia by wet impregnation to obtain supported V/CeO2 catalysts. The V/CeO2-P catalyst is more active, reaching ≈100% NO conversion and N2 selectivity already below 225 °C at a space velocity of GHSV = 70,000 h−1. XRD, UV-vis-DRS, Raman, pseudo-in-situ-XPS and operando-EPR spectroscopy revealed that this is due to higher surface area and a more effective incorporation of V sites into the support surface, which keeps them in their active valence states +5 and +4 and prevents reduction to inactive V3+ as observed on V/CeO2-C. © 2016 Elsevier B.V. All rights reserved.

1. Introduction V2O5-WO3/TiO2 catalysts, used for selective catalytic reduction (SCR) of NOx in power plants, operate only at 300–500 °C which is too high for, e.g., diesel or lean-burn gasoline engines. Therefore, development of new catalysts [1,2] or improvement of vanadium-based catalysts [3,4] for low-temperature NH3-SCR of NOx is needed, whereby vanadia supported on (modified) ceria appeared to be particularly effective [3,5–7]. With VOx/CexZr1 − xO2, we recently reached almost full NO conversion and N2 selectivity around 200 °C at a space velocity of GHSV = 70,000 h−1 [8]. This was attributed to the formation of special \\O\\Ce–O–V(_O)–O–Zr–O– moieties supposed to support oxygen transport. In these moieties, the V sites shuttle reversibly between V5+ and V4+ while Ce and Zr species remained essentially tetravalent. Interestingly, we found that the same amount of vanadia deposited on differently prepared CeO2 supports leads to catalysts of rather different performance. A sensitive dependence of the performance of ceria-based catalysts on the synthesis procedure was also observed in other reactions [9–11], for which ceria was prepared by hydrothermal [12], precipitation [9,13], citrate sol-gel [10,14], solution combustion [15], or microemulsion [11] protocols. It is supposed that, depending on the synthesis protocol, oxygen vacancies and structural defects are formed ⁎ Corresponding author. E-mail address: [email protected] (A. Brückner).

http://dx.doi.org/10.1016/j.catcom.2016.07.002 1566-7367/© 2016 Elsevier B.V. All rights reserved.

that in turn influence the Ce4+/Ce3+ redox behavior [16,17]. However, at least in NH3-SCR, a detailed analysis of these relations has not yet been performed. This is the aim of the present study. To this end, we compare structure-reactivity relationships of two VOx/CeO2 catalysts in which 5 wt.% of V2O5 were deposited by the same procedure on CeO2 supports prepared by precipitation (V/CeO2-P) or a citrate solgel method (V/CeO2-C). Synthesis, catalytic performance and selected characterization results of the latter catalyst have already been described in our recent paper [8] and are cited here only for comparison. 2. Experimental CeO2 supports were prepared by a sol-gel method from citric acid and cerium nitrate (CeO2-C) [8,14], or by precipitation from cerium nitrate in aqueous ammonia (CeO2-P) [9]. The V/CeO2 catalysts with a nominal content of 5 wt.% V2O5 were prepared by wet impregnation of the calcined CeO2 powders with a solution of ammonium metavanadate in oxalic acid. Catalysts were characterized by X-ray diffraction (XRD) for their crystallinity, by inductively coupled plasma optical emission spectroscopy (ICP-OES) for chemical composition, by nitrogen adsorption for BET surface area and pore properties, by X-ray photoelectron spectroscopy (XPS) for their surface properties as well as by Raman, UV-vis diffuse reflectance and electron paramagnetic resonance (EPR) spectroscopy.

172

T.H. Vuong et al. / Catalysis Communications 84 (2016) 171–174

Fig. 1. NO (solid symbols and lines), NH3 conversion (open symbols and dashed lines) over: (A) V/CeO2-C (squares) and V/CeO2-P (circles); (B) V-free CeO2-C (squares) and CeO2-P (circles); and N2O concentration (solid squares and dotted lines) over V-free CeO2-C as a function of temperature. Feed composition: 0.1% NO, 0.1% NH3, 5% O2/He, GHSV = 70,000 h−1

Catalytic performance in NH3-SCR of NO was measured between 100 and 300 °C with 100 mg catalysts and a feed composition of 0.1% NO, 0.1% NH3, 5 vol.% O2/He at a GHSV of 70,000 h−1. For more experimental details see Supporting information SI-A. 3. Results and discussion 3.1. Catalytic behavior From Fig. 1B it is evident that already the pure supports show some activity in the catalytic reaction. While NO and NH3 conversion is low over the whole temperature range on support CeO2-P, sample CeO2-C shows almost similar conversion around 200 °C as the corresponding catalyst V/CeO2-C. However, above 200 °C undesired NH3 combustion and N2O formation reduce NO conversion and N2 selectivity. Thus, both pure supports show detrimental catalytic behavior and are therefore not further considered here. Catalyst V/CeO2-C is much less active than V/CeO2-P, reaching a maximum NO conversion of 90% at only 275 °C which even drops at higher temperature due to undesired oxidation of NH3 (Fig. 1A). In contrast, V/CeO2-P reaches full conversion already at 225 °C and remains stable. In due course, we present the results of a comprehensive characterization study to find out reasons for this different behavior. Above 200 °C, NO conversion declines markedly and a rather high concentration of undesired N2O is formed (Fig. 1B). In the following section, we present the results of a comprehensive characterization study of the catalysts to find out reasons for their different behavior. 3.2. Catalyst characterization XRD patterns of both CeO2-C and CeO2-P supports show the characteristic peaks of the cubic fluorite structure (Fig. SI-1). They are sharper and more intense for CeO2-C due to its larger mean crystallite size and a smaller strain derived by the Williamson-Hall equation [18] (Table 1, SIA). The strain in CeO2-P is about twice as high as in CeO2-C which might

promote oxygen mobility inside the support. Deposition of vanadia does not lead to any changes, indicating that highly dispersed and/or amorphous VOx surface species dominate in both samples. However, the Raman spectra of both catalysts show some weak bands of V2O5 nanocrystals too small to be detectable by XRD [19] (Fig. SI-2). The bulk V/Ce atomic ratios are similar and close to the nominal value corresponding to 5% of V2O5 (Table 1). The BET surface area and pore volume of CeO2-P and V/CeO2-P are significantly higher than those of CeO2-C and V/CeO2-C. This might be due to the fact that the P-samples are mesoporous while the C-samples are non-porous or macroporous [20] (Fig. SI-3). The UV-vis-DR spectra of the pure supports show bands below 250 nm, around 280 and 350 nm. The first is usually assigned to O2− → Ce3 + charge-transfer (CT) transitions while the latter two arise from O2 − → Ce4 + CT and interband transitions (Fig. 2) [18,21]. For CeO2-P, the intensity below 250 nm is higher suggesting a higher content of Ce3+ which, for maintaining electroneutrality, must create oxygen vacancies. This is supported by the fact that the absorption edge of CeO2-P (Eg = 2.72 eV) is red-shifted compared to CeO2-C for which Eg = 3.00 eV is close to the bulk value of CeO2 (3,15 eV) [22], as well as by Raman data discussed below. Such red shift is caused by a drop in lattice symmetry due to increased disorder, which may result from small nanoparticles with abundant surface defects and/or replacement of Ce by other metal ions with different diameter [18,21]. It agrees with the smaller crystallite size and higher strain of CeO2-P (Table 1). Deposition of vanadia on these supports leads to an increase of absorbance below 400 nm, characteristic for O2− → V5+ CT bands of VOx single sites and small VxOy clusters [23]. The weak shoulder extending from 400 to 500 nm in sample V/CeO2-C might arise from few V2O5 nanocrystals [24], which are evident, too, from the Raman spectra. Interestingly, a marked red shift of the absorption edge is observed for V/ CeO2-P. One could suppose that this is due to a higher V site agglomeration in this sample, since it has been shown previously that the absorption edge energy of VxOy-containing oxides decreases with the number of V-O-V bridges [25]. However, the Raman spectra in Fig. SI-2 do not

Table 1 Bulk V/Ce ratio[a], chemical composition crystallite size, specific surface area and pore volume, band gap energy of supports and catalysts. Sample CeO2-C CeO2-P V/CeO2-C V/CeO2-P a b c

V/Ce ratio[a]

Mean crystallite size[b] (nm)

Strain[b]

Surface area (m2 g−1)

Pore volume (cm3 g−1)

Band gap energy (eV)[c]

24.3 14.1

0.00138 0.00265

12.7 61.2 11.5 45.4

0.026 0.060 0.043 0.053

3.00 2.72 2.58 2.25

0.100 0.092

From ICP. From XRD using the Williamson-Hall equation (see SI-A). From UV-vis-DRS.

T.H. Vuong et al. / Catalysis Communications 84 (2016) 171–174

173

Table 2 Rate constants of V reduction and reoxidation (k/min−1) Sample

V/CeO2-C[a]

V/CeO2-P

kred1 kred2 kreox1 kreox2

0.104 0.271 0.003 0.181

0.0185 0.0185 0.023 1.096

a

Fig. 2. UV-vis spectra of pure CeO2 supports and 5% V2O5/CeO2 catalysts.

show more intense V2O5 bands for sample V/CeO2-P and the EPR spectra in Fig. 3 (discussed below) suggest a higher V site agglomeration in V/ CeO2-C rather than in V/CeO2-P. Based on these results, we assign the red shift of the absorption edge in sample V/CeO2 to an incorporation of V in the surface of CeO2 and, thus, a more effective attachment of the VOx species to the support than in V/CeO2-C. A similar red shift was also observed upon doping of ceria with other elements such as Nd3+ [26]. 3.3. Spectroscopic in-situ-studies In-situ-UV-vis spectra of V/CeO2-P recorded at 200 °C after 60 min pretreatment in air flow at 275 °C and 45 min exposure to the NH3SCR feed stream at 200 °C indicate a loss of intensity in the region of CT bands of V5+ between 400 and 500 nm, while the band of d-d transitions of reduced V5-x species above 600 nm rises (Fig. SI-4). This change is reversible in NO/O2 flow after removing NH3 from the feed. A similar behavior has been observed, too, for V/CeO2-C of the V/ CexZr1 − xO2 catalyst series [8]. For this series, the reduction/reoxidation rates of surface VOx species (derived by kinetic evaluation of the timedependent UV-vis absorbance at 700 nm) scaled with catalytic activity, the least active V/CeO2-C catalyst showing the lowest rates. For comparison, the same time dependent UV-vis experiment was performed with V/CeO2-P. The rate constants of V reduction in a flow of 0.1% NH3/He and reoxidation in 5% O2/He flow at 200 °C (Table 2) were derived from fits of the experimental data in Fig. SI-5 by a first-order rate law (Eqs. (1) and (2) in SI-E) The fits revealed two V reduction and reoxidation processes, a fast and a slow one (Fig. SI-5). We assume that the fast processes (kred2

From [8].

and kreox2) might be related to V sites on the top surface. These are the ones in immediate contact with reactants and most probably govern the catalytic activity. The slow processes (kred1 and kreox1) may reflect V sites in subsurface layers, which might play, if at all, only a minor role for catalysis, since particularly their reoxidation is hindered. Interestingly, reduction and reoxidation of top surface V species proceeds with similar rates in the less active catalyst V/CeO2, while in V/CeO2-P reduction is by 2 orders of magnitude slower, yet reoxidation is by a factor of about 10 faster. For NH3-SCR of NO on FeOx [27] and MnOx containing catalysts [28] it was found that NO is not directly reduced by NH3 but first oxidized to adsorbed nitrite and/or nitrate species by Fe3+ or Mn4+ according to a Mars-van Krevelen mechanism – a process which is hardly catalyzed by metal ions in non-reducible valence states such as Fe2+ or Mn2+. Anticipating that a Vn+/V(n − 1)+ redox shuttle is working in NH3-SCR of NO over V/CeO2, too, fast reoxidation of reduced V species would be beneficial to ensure high catalytic activity. Thus, the high reoxidation rate of V/CeO2-P could be one reason for its higher activity compared to V/CeO2-C. A more specific analysis of reaction-dependent changes on the surface was performed by pseudo-in-situ-XPS after 60 min pretreatment in air flow at 300 °C and after 30 min exposure to SCR feed at 200 °C without subsequent contact to ambient atmosphere (Table 3, spectra plotted in Fig. SI-6). Generally, six features are observed in the Ce 3d region which are labelled with v, v″, v‴ (3d5/2) and u, u″, u‴ (3d3/2) (Fig. SI-6). For clarity, only EB values of the v peak are listed in Table 3 The surface V/Ce ratio of the more active catalyst V/CeO2-P is lower than that of V/CeO2-C, probably since its higher surface area leads, at a similar V dispersion, to a lower V surface concentration. After air pretreatment, the surface of this catalyst contains exclusively V5 + (reflected by EB = 517.0 eV [29,30]) while that of V/CeO2-C represents a mixture of V5+ and V4+ (EB = 518.0 and 516.5 eV [8]). After treatment in SCR feed, the surface V5 + sites on V/CeO2-P are partially reduced to V4+ while on V/CeO2-C a considerable part of V3+ is formed. Under the same conditions, Ce remains essentially tetravalent in V/ CeO2-C while it is partially reduced to Ce3+ in V/CeO2-P. This may be explained by the removal of O and localization of the residual electrons at the neighboring Ce, leaving behind\\Ce3+\\□\\V5+(_O)\\O\\Ce3+\ and \\Ce3+\\□\\V4+(_O)\\O\\Ce4+\\ species. This has been confirmed by DFT calculations of VOx/CeO2 catalysts [31]. Preferential electron trapping by Ce4+ could keep a considerable part of the V sites in V/CeO2-P in their most active valence state +5 under reaction conditions, which would agree very well with the observed higher activity

Table 3 Pseudo-in-situ-XPS binding energies (EB/eV) and atomic ratios of V/CeO2 after pretreatment in air and SCR feed at 200 °C

Fig. 3. In-situ-EPR spectra of V/CeO2-C and V/CeO2-P recorded at 200 °C in Ar after pretreatment in O2 at 300 °C (C1 and P1), after 30 min treatment in 0.2% NH3/He (C2 and P2) and after 30 min exposure to total SCR feed flow (C3 and P3, spectrum in 0.2% NO, 5% O2/He subtracted). The dashed lines show spectra fitted with spin Hamiltonian parameters in Table SI-1

Sample

EB (V 2p3/2)

EB (Ce 3d3/2)

V/Ce

V/CeO2-C air

516.5 (V4+ + V5+) 518.0 (V5+) 516.6 (V4+ + V5+) 515.0 (V3+) 517.0 (V5+) 516.2 (V4+)

885.5 (Ce4+)

0.73

885.5 (Ce4+)

0.67

V/CeO2-C after SCR V/CeO2-P air V/CeO2-P after SCR

4+

885.5 (Ce ) 885.5 (Ce4+) 882.5 (Ce3+)

0.52 0.44

174

T.H. Vuong et al. / Catalysis Communications 84 (2016) 171–174

In-situ- and operando-EPR spectra (recorded together with on-line mass spectrometry, Fig. SI-7) widely supports the conclusions drawn from UV-vis and XPS data. After pre-oxidation, the EPR spectra of both catalysts show only very small signals around g = 2.00 from oxygen de4+ signal is detectfects such as O•− and/or O•− 2 (Fig. 3, C1 and P1). No V ed, which is surprising at least for V/CeO2-C as XPS points to some surface V4 +. Possibly, this is due to the local environment of these sites on the surface of V/CeO2-C, leading to short relaxation times. Exposure to NH3/He flow at 200 °C reduces the V5 + species in sample V/ CeO2-P to VO2+, evident from the characteristic EPR signal with hyperfine structure (hfs). Switching to the total SCR feed containing also NO and O2 raises the intensity of this signal. This points to oxidation of EPR-silent V3 + (formed during the preceding treatment in NH3) to EPR-active VO2 + and agrees very well with the in-situ-XPS result (Table 3). The fact that no VO2 + signal is observed in V/CeO2-C after NH3-treatment (Fig. 3, C2) suggests that reduction to V3 + may have been very pronounced which is, however, at least partly reversible under SCR feed (Fig. 3, C3). From the g and A tensor parameters derived by spectra simulation and explained in more detail in SI-D (Fig. 3, Table SI-1), it is evident that the hfs spectrum of catalyst V/CeO2-C could be fitted with one single VO2 + site while two different VO2 + species A and B were required for V/CeO2-P. The parameters of the single VO2+ sites in V-CeO2-C and of VO2+ species A in V-CeO2-P are rather similar but differ significantly from VO2 + species B in V-CeO2-P, which are less distorted and less ionic. Similar species have been found in V/ CexZr1-xO2 catalysts and assigned to V incorporated in \\O\\Ce\\O\\V(_O)\\O\\Zr\\O\\ surface moieties [8]. Accordingly, we assign the VO2+ species B in V/CeO2-P to similar sites, tightly incorporated in the catalyst surface

Acknowledgements The authors thank Dr. J. Rabeah, Dr. J. Engeldinger, and Mr. R. Eckelt for experimental support. Thanh Huyen Vuong acknowledges a grant of Project 911 of the Vietnamese Ministry of Education and Training. Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.catcom.2016.07.002. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20]

4. Conclusions In contrast to the citrate sol-gel protocol, the precipitation method creates a mesoporous CeO2-P support with markedly higher surface area, higher concentration of Ce3+ bulk species and oxygen vacancies, which favor oxygen transport through the lattice. This is beneficial for the corresponding V/CeO2-P (assuming a Mars-van Krevelen redox cycle) which is much more active than the respective V/CeO2-C material and approaches full NO conversion and N2 selectivity already below 225 °C. While surface V sites are partly reduced under SCR conditions to (probably inactive) V3+ on V/CeO2-C, this deep reduction does not occur on V/CeO2-P on which, instead, Ce4+ is preferentially reduced to Ce3+. Moreover, EPR results suggest that V sites in V/CeO2-P might be in a more tight junction with the CeO2 surface, probably forming active \\Ce3+\\□\\V5+/4 +(_O)\\O\\Ce3 +/4 +\\ which are anticipated to boost catalytic activity.

[21] [22] [23] [24] [25] [26] [27] [28] [29] [30]

[31]

Z. Liu, Y. Yi, S. Zhang, T. Zhu, J. Zhu, J. Wang, Catal. Today 216 (2013) 76–81. J. Li, H. Chang, L. Ma, J. Hao, R.T. Yang, Catal. Today 175 (2011) 147–156. Z. Liu, S. Zhang, J. Li, J. Zhu, L. Ma, Appl. Catal. B 158–159 (2014) 11–19. T. Boningari, R. Koirala, P.G. Smirniotis, Appl. Catal. B 140–141 (2013) 289–298. C. Li, Q. Li, P. Lu, H. Cui, G. Zeng, Front. Environ. Sci. Eng. 6 (2012) 156–161. Z. Lian, F. Liu, H. He, Catal. Sci. Technol. 5 (2015) 389–396. Z. Lian, F. Liu, H. He, K. Liu, RSC Adv. 5 (2015) 37675–37681. T.H. Vuong, J. Radnik, E. Kondratenko, M. Schneider, U. Armbruster, A. Brückner, Appl. Catal. B (2016), http://dx.doi.org/10.1016/j.apcatb.2016.03.063. M.M. Natile, G. Boccaletti, A. Glisenti, Chem. Mater. 17 (2005) 6272–6286. J. Liu, Y. Li, J. Zhang, D. He, Appl. Catal. A (2016) (Ahead of Print). J. Rebellato, M.M. Natile, A. Glisenti, Appl. Catal. A 339 (2008) 108–120. M. Hirano, E. Kato, J. Mater. Sci. Lett. 15 (1996) 1249–1250. B. Djuričić, S. Pickering, J. Eur. Ceram. Soc. 19 (1999) 1925–1934. P.T.M. Pham, M.T. Le, T.T. Nguyen, I.V. Driessche, Materials (2014) 7379–7397. A. Bumajdad, J. Eastoe, A. Mathew, Adv. Colloid Interf. Sci. 147–148 (2009) 56–66. L. Vivier, D. Duprez, ChemSusChem 3 (2010) 654–678. L. Li, Y. Zhan, Q. Zheng, Y. Zheng, C. Chen, Y. She, X. Lin, K. Wei, Catal. Lett. 130 (2009) 532–540. B. Choudhury, A. Choudhury, Mater. Chem. Phys. 131 (2012) 666–671. X.J. Wang, H.D. Li, Y.J. Fei, X. Wang, Y.Y. Xiong, Y.X. Nie, K.A. Feng, Appl. Surf. Sci. 177 (2001) 8–14. K.S.W. Sing, D.H. Everett, R.A.W. Haul, L. Moscou, R.A. Pierotti, J. Rouquerol, T. Siemieniewska, Pure Appl. Chem. 57 (1985) 17. B.M. Reddy, P. Bharali, P. Saikia, S.-E. Park, M.W.E. van den Berg, M. Muhler, W. Grunert, J. Phys. Chem. C 112 (2008) 11729–11737. Z. Wang, Z. Quan, J. Lin, Inorg. Chem. 46 (2007) 5237–5242. A. Brückner, Appl. Catal. A Gen. 200 (2000) (287-197). P. Rybarczyk, H. Berndt, J. Radnik, M.M. Pohl, O. Buyevskaya, M. Baerns, A. Brückner, J. Catal. 202 (2001) 45–58. X. Gao, I.E. Wachs, J. Phys. Chem. B 104 (2000) 1261–1268. B. Choudhury, A. Choudhury, Curr. Appl. Phys. 13 (2013) 217–223. R. Pérez Vélez, I. Ellmers, H. Huang, U. Bentrup, V. Schünemann, W. Grünert, A. Brückner, J. Catal. 316 (2014) 103–111. U. Bentrup, A. Brückner, M. Richter, R. Fricke, Appl. Catal., B 32 (2001) 229–241. M.C. Biesinger, L.W.M. Lau, A.R. Gerson, R.S.C. Smart, Appl. Surf. Sci. 257 (2010) 887–898. J.F. Moulder, W.F. Stickle, P.E. Sobol, K.D. Bomben, Handbook of x-ray photoelectron spectroscopy: a reference book of standard spectra for identification and interpretation of XPS data, Eden Prairie, Minn. Physical Electronics, c1995, 1995. M.V. Ganduglia-Pirovano, C. Popa, J. Sauer, H. Abbott, A. Uhl, M. Baron, D. Stacchiola, O. Bondarchuk, S. Shaikhutdinov, H.-J. Freund, J. Am. Chem. Soc. 132 (2010) 2345–2349.