Regeneration of Commercial SCR Catalysts: Probing the Existing

Jul 17, 2015 - To investigate the poisoning and regeneration of SCR catalysts, fresh and arsenic-poisoned commercial V2O5–WO3/TiO2 catalysts are ...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/est

Regeneration of Commercial SCR Catalysts: Probing the Existing Forms of Arsenic Oxide Xiang Li, Junhua Li,* Yue Peng, Wenzhe Si, Xu He, and Jiming Hao State Key Joint Laboratory of Environment Simulation and Pollution Control, School of Environment, Tsinghua University, Beijing 100084, P. R. China S Supporting Information *

ABSTRACT: To investigate the poisoning and regeneration of SCR catalysts, fresh and arsenic-poisoned commercial V2O5−WO3/TiO2 catalysts are researched in the context of deactivation mechanisms and regeneration technology. The results indicate that the forms of arsenic oxide on the poisoned catalyst are related to the proportion of arsenic (As) on the catalyst. When the surface coverage of (V+W+As) is lower than 1, the trivalent arsenic species (AsIII) is the major component, and this species prefers to permeate into the bulkphase channels. However, at high As concentrations, pentavalent arsenic species (AsIV) cover the surface of the catalyst. Although both arsenic species lower the NOx conversion, they affect the formation of N2O differently. In particular, N2O production is limited when trivalent arsenic species predominate, which may be related to As2O3 clogging the pores of the catalyst. In contrast, the pentavalent arsenic oxide species (As2O5) possess several As−OH groups. These As−OH groups could not only enhance the ability of the catalyst to become reduced, but also provide several Brønsted acid sites with weak thermal stability that promote the formation of N2O. Finally, although our novel Ca(NO3)2-based regeneration method cannot completely remove As2O3 from the micropores of the catalyst, this approach can effectively wipe off surface arsenic oxides without a significant loss of the catalyst’s active components. with vanadium oxide.11−14 During these studies, the arsenic oxide poisoning process was simulated through As 2 O 3 sublimation in a stainless steel reactor or tube furnace under simulated flue gas conditions. It was concluded that deviations may exist between their estimates and the actual results. Therefore, a thorough study of the predominant forms of arsenic oxide and its ability to poison commercial catalysts is urgent to clarify the role of arsenic in the deactivation process. The regeneration of deactivated SCR catalysts is a viable and environmentally friendly option for the denitration market because of its low cost, recyclable purpose and ability to prolong the lifetime of catalysts. Hot-water washing and electrophoresis treatments were found to be effective methods to remove alkali metals on SCR catalysts, and we previously demonstrated that H2O2 solutions are an effective way to reduce the arsenic levels.15,16 While some of the vanadium and sulfates on the surface of the catalysts are simultaneously removed under an H2O2 environment, which reduces the catalytic activity. Therefore, it is crucial to develop a method that can remove the arsenic and maintain the active constituents of the catalysts.

1. INTRODUCTION Selective catalytic reduction (SCR) with NH3, an effective method for controlling NOx emissions, has been widely used in recent years. V2O5−WO3/TiO2 catalysts are at the core of SCR technology on stationary source applications, particularly for coal-fired power plants.1−5 However, alkali metals (e.g., K, Na), heavy metals (e.g., As, Pb) and SO2, originating from the complex flue gas components, often result in deactivating the catalyst and collapse its working lifetime.6,7 Therefore, many researchers have focused on the deactivation mechanisms for the poisons of commercial SCR catalysts.8−10 Nevertheless, their research objects were usually alkali metals, alkaline earth metal, and acidic gas, and paid little attention to heavy metals because of their physiological toxicity and variable valence states. Heavy metal arsenic, which exists as vapor-phase arsenic compounds in flue gas, is another effective poison that deactivates SCR catalysts. As a rule, the fly ash is reinjected into the boiler in the former, which leads to high concentrations of gaseous arsenic at the SCR inlet, so their toxicity is more common in wet bottom furnaces than in dry-bottom furnaces. Presently, there is no agreement regarding the role of arsenic in the deactivation process. Some have argued that gaseous As2O3 or its dimer As4O6 could diffuse into the catalyst and block its micropores to prevent NO from reaching the active sites; others have suggested that arsenic in its 5+ oxidation state could react © 2015 American Chemical Society

Received: Revised: Accepted: Published: 9971

April 27, 2015 July 16, 2015 July 17, 2015 July 17, 2015 DOI: 10.1021/acs.est.5b02257 Environ. Sci. Technol. 2015, 49, 9971−9978

Article

Environmental Science & Technology In this study, poisoned commercial V 2 O 5−WO 3 /TiO 2 catalysts obtained from coal-fired power plants were studied and compared with regenerated and fresh catalysts. The arsenic content, valency, surface acidity and reducibility were investigated using N2 adsorption, XRD, Raman, XPS, H2-TPR, NH3TPD, DRIFTS spectroscopies. Density functional theory (DFT) calculations were also used to explore the probable structures of As on the catalyst surface. Finally, different regeneration methods were used and compared based on both the As removal ratios and active constituent (i.e., V and W) residual ratios.

loading of each sample was below 1 wt %, and the As3# sample was chosen for the subsequent regeneration studies. All of the catalysts were crushed and sieved with >60 meshes for chemical characterization and within 40−60 meshes for the activity measurements. The deactivated catalysts were regenerated by removing As with different regenerating solutions containing H2O, acetic acid, citric acid, Na2CO3, H2SO4, NH3·H2O, NaOH, H2O2, and Ca(NO3)2, followed by washing with deionized water, drying and calcination. Typically, the poisoned catalysts (5 g) were introduced to the specific regenerating solution (50 mL) and impregnated for 3 h after one 30 min ultrasonic treatment. The catalysts were then cleaned several times with deionized water and filtered. The precipitates were collected and dried at 110 °C overnight to remove any residual water. The final samples were calcined at 500 °C for 3 h in air with a heating rate of 2 °C·min−1. For the regeneration procedure with Ca(NO3)2, after impregnated and regenerated by Ca(NO3)2·4H2O aqueous solutions, the poisoned catalysts were cleaned with a dilute HNO3 solution (pH = 2) to remove excess Ca cations and then washed several times with DI water. Activity Measurements. SCR activity tests were performed in a fixed-bed quartz flow reactor (i.d.: = 6 mm) using 100 mg of the catalyst. The feed gas mixture contained 500 ppm of NO, 500 ppm of NH3, and 3% O2 and N2 as the balance gas. The total flow rate was 200 mL·min−1, corresponding to a gas hourly space velocity (GHSV) of 120 000 mL·g−1·h−1. The concentrations of the inlet and outlet gases (NO, NH3, NO2, and N2O) were monitored by an FTIR spectrometer (Gasmet FTIR DX-4000). The NOx conversion and N2 selectivity were calculated as follows:

2. EXPERIMENTS AND METHODS Catalyst Preparation and Regeneration. Fresh industrial SCR catalysts and three types of As-poisoned industrial SCR catalysts were obtained from a coal-fired power plant in the Inner Mongolia province of China. These catalysts were defined as fresh, As1#, As2#, and A3# based on the different runtimes, and their physical parameters are were listed in Table 1. The V2O5 Table 1. Physical, Chemical and Surface Properties of Fresh and Poisoned Catalysts sample

fresh

As1#

As2#

As3#

As/Ti (at %) As/Tib (at %) As2O3 contentc (wt %) SBET (m2/g) θV+W+Asc (%) Vp (cc/g) rp (nm) SdBET (m2/g) NH3 desorption (μmol/m2cat) H2 consumption (μmol/m2cat)

0 0 0 52.2 52.5 0.33 19.4 53.3 1.8 5.3

0.61 0.42 0.47 50.5 97.9 0.26 19.1 53.2 1.8 6.0

1.8 0.89 0.94 50 143.7 0.25 18.9 53.9 1.4 6.1

7.1 2.7 2.9 49.9 340.4 0.25 19.7 57.3 1.1 7.8

a

a

b

NOx conversion =

c

By XPS. By ICP. By XRF.

[NOx ]inlet − [NOx ]outlet × 100% [NOx ]inlet

⎛ ⎞ 2[N2O]outlet N2selectivity = ⎜1 − ⎟ × 100% [NOx ]inlet + [NH3]inlet − [NOx ]outlet − [NH3]outlet ⎠ ⎝

Where NOx included NO and NO2. Catalyst Characterization. Powder XRD patterns were recorded on a powder X-ray diffractometer (Rigaku, D/max2200/PC, Japan) between 20° and 90° at a stepsize of 5°·min−1 using a Cu Kα radiation source (λ = 0.15405 nm, 40 mV and 200 mA). The elemental composition of the catalysts was performed by ICP with an IRIS Intrepid II XSP apparatus from Thermo Fisher Scientific Inc. and XRF with an XRF DEX-LE from Shimadzu. The specific surface areas of the catalysts were calculated by the N2 adsorption method at 77 K using a Micromeritics ASAP 2020 instrument in the static mode, and the pore volumes and average pore diameters were determined by the Barrett−Joyner−Halenda (BJH) method from the desorption branches of the isotherms. X-ray photoelectron spectroscopy (XPS) was performed using an ESCALab220i-XL electron spectrometer from VG Scientific using 300 W Al Kα radiation. All of the binding energies were referenced to the C 1s peak (B.E. = 284.8 eV). The temperature-programmed reduction (TPR) of H2 was performed on a chemisorption analyzer (Micromeritics, ChemiSorb 2720 TPx). One hundred milligrams of each sample was treated under Ar at 300 °C for 60 min. The samples were

(1)

(2)

then reduced under 10% H2/Ar (50 mL/min) at a rate of 10 °C· min−1 from 50 to 950 °C. Temperature-programmed desorption (TPD) of NH3 was performed in a fixed-bed quartz reactor using an FTIR spectrometer detector (MKS, MultiGas 2030HS). Prior to the measurements, 100 mg samples were pretreated at 350 °C for 60 min and then purged with NH3 (500 ppm) at 100 °C until adsorption equilibrium was reached, followed by sweeping with N2 for 60 min. Subsequently, the treated catalysts were heated from 100 to 600 °C at a rate of 10 °C·min−1 using a flow rate of 200 mL·min−1. In situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFT) experiments were performed on an FTIR spectrometer (Nicolet NEXUS 6700) equipped with a Harrick IR cell and MCT/A detector cooled by liquid nitrogen. Prior to each experiment, the catalyst was preheated at 400 °C for 1 h with N2 at a flow of 100 mL·min−1. All of the DRIFTS spectra were recorded by accumulating 32 scans with a resolution of 4 cm−1. DFT Calculations. All calculations were based on DFT and were performed using the Material Studio 5.5 modeling Dmol3 software package. In the electronic structure calculations, double numerical basis sets plus polarization functions (DNP) and a 9972

DOI: 10.1021/acs.est.5b02257 Environ. Sci. Technol. 2015, 49, 9971−9978

Article

Environmental Science & Technology

Forms of Arsenic Oxide on the Catalysts. The effect of As loading on the morphology and structural properties of the deactivated catalysts was investigated. The details of these investigations are listed in Table 1. Additionally, the XRD patterns of fresh and poisoned samples are shown in Figure 2.

relativistic semicore pseudo potential (DSPP) were chosen. The exchange-correlation interaction was treated within the generalized gradient approximation (GGA) using the Perdew-BurkeErnzerh (PBE) functional. Self-consistent field (SCF) calculations were performed with a convergence criterion of 10−6 Hartree on the total energy, and the real-space cutoff radius was set to 5.2 Å. The V6O20H10 cluster with an exposed V2O5 (010) plane and TiO2 (001) slab were constructed as matrices for further study. Additional details of this model could be found in our previous work.15,17 To explore the most favorable structures that may exist in the catalyst, trivalent and pentavalent arsenious oxide clusters were doped on the V2O5 cluster and TiO2 (001) slabe, respectively. The total energies of the systems were also calculated and compared after the incorporation of O2.

3. RESULTS AND DISCUSSION Activity Test. Figure 1 shows the NOx conversion and N2O production over the As1#, As2#, and As3# samples with fresh

Figure 2. XRD patterns of fresh (a), As1# (b), As2# (c), and As3# (d) catalysts.

For all samples, the anatase phase of TiO2 (JCPDS, PDF 211272) is predominant in each XRD pattern and no diffraction lines attributed to crystalline V2O5 or WO3 can be detected, indicating that vanadium and tungsten oxides are present in a noncrystalline state and are highly dispersed on the surface of TiO2. However, a small peak appears in the XRD patterns for all As poisoned samples, which is attributed to the reflection of clauditite As2O3 (2θ = 27.5°). Its normalized diffraction intensity remains nearly unchanged as the As content increases. Because no other peaks can be associated with As2O5, which has been reported to be the primary form of As on deactivated SCR catalysts,11,12,18 pentavalent arsenic may exist as a bound moiety to the catalyst or as an amorphous constituent of the catalyst. The crystallite size (d) was estimated from the (101) reflection of the anatase phase using the Scherrer equation. The specific surface area of spherical anatase (SdBET) can be estimated by SdBET = 6/(ρanatase·d(101)), where ρanatase is the density of pure anatase.19,20 The BET surface areas of the four types of catalysts are approximately 50 m2/g, and there is a slight inverse relationship between the BET surface area (SBET) and As content. However, the difference between the SdBET and SBET values becomes larger with the change in the As content, particularly for As3#, where the As content increases from 0 to 2.9 wt %. These results suggest that the support surface is rough with TiO2 particles and is covered with amorphous arsenic species. The pore volumes (Vp) of the catalysts are clearly reduced with the loss of surface area resulting from the introduction of arsenic oxide. The reduced pore volume of As1# is notable, particularly because its mass percent of As is only 0.47%, which demonstrates that arsenic oxide at low As content preferential clogs the catalyst pores. However, the mean pore radius (rp) is not significantly affected by the addition of As, as evidenced by its relatively constant values of 18−20 nm. These results indicate that the surface of the catalyst covered by arsenic oxide is obvious for samples with high As contents but that the pores of the catalysts become plugged for samples with low As contents. Parameters for the theoretical capacity of a monolayer in the literature can be used to further evaluate the theoretical surface

Figure 1. NOx conversion (a) and N2O production (b) of fresh and poisoned catalysts. Reaction conditions: [NO] = [NH3] = 500 ppm, [O2] = 3%, total flow rate =200 mL/min, GHSV = 120 000 mL/(g·h).

catalysts for the SCR reaction at 250−500 °C under a GHSV of approximately 120 000 mL·g−1·h−1. The ratio of NOx conversion decreases sharply with increasing As content, which suggests that the toxic effect of arsenic oxide on the V2O5−WO3/TiO2 catalyst is rather serious. However, considering that the N2O production originates primarily from unselective NH3 oxidation5 for the three types of poisoned catalysts, the As3# sample exhibits poor N2 selectivity and yields more N2O than the fresh catalyst at 250−450 °C. This result is consistent with previous findings.17 In contrast, the formation of N2O is lower for the As1# and As2# samples than the fresh catalysts at temperatures between 250 and 500 °C. Thus, catalysts affected by arsenic poisoning show both an enhanced (high As content) and inhibitory (low As content) effect for N2O formation despite a clearly reduced SCR activity. Moreover, the results suggest that the level of As poisoning affects the deactivated catalysts differently. 9973

DOI: 10.1021/acs.est.5b02257 Environ. Sci. Technol. 2015, 49, 9971−9978

Article

Environmental Science & Technology coverage (θ) of the poisoned SCR catalysts; specifically, a monolayer capacity of 0.145% w/w for V2O5/m2,21 a monolayer capacity of 7 μmol for WO3/m2,22 and a monolayer capacity of 153.6 μg for As3+/m223 has been assumed. The surface coverage of (V+W) has also been calculated by simply summing the surface coverage of V and W. As reported in Table 1, the coverage can be differentiated with the As content reported for the poisoned catalysts. The surface coverage of (V+W+As) approaches 100% for As1# and exceeds 100% for As2# and As3#, whereas it is dilute for the fresh V2O5−WO3/TiO2 catalysts. Combined with the XRD patterns, As2O3 species (in the claudetite phase) cannot be the major components of the As3# sample, and other arsenic oxide forms must be present, particularly for catalysts poisoned with high As contents. The Raman spectra of the four types of samples were recorded to study the interaction between arsenic oxide and the active ingredients (i.e., vanadium oxide and tungsten oxide) under ambient conditions. As shown in Figure 3, all of the spectra

Figure 4. As 3d (a) and O 1s (b) from XPS spectra of fresh and poisoned catalysts.

that oxidation of As3+ occurs on the surface of TiO2. Thus, as described previously,11,18 the As species are anchored onto the catalyst’s surface as pentavalent arsenate (V) at high As content. Correspondingly, As2O3 species are the main forms that exist for poisoned catalysts at low As concentrations, particularly for those with a theoretical V+W+As monolayer coverage below 1. The XPS spectra of O 1s for the fresh and poisoned catalysts are shown in Figure 4(b). The spectra are deconvolved into three groups of sub-bands centered at approximately 530.0, 531.6, and 533.0 eV; these peaks can be assigned to lattice oxygen species (O2−) and surface chemisorbed oxygen species containing O− and O2−, respectively. The relative proportions of each O species for each sample are calculated and shown in Figure 4(b). For the As3# sample, the amount of surface adsorbed oxygen first decreases and then increases to 29.8%. In general, chemisorbed oxygen species are more reactive in oxidation reactions than the lattice oxygen species because of their higher mobility.28,29 Good correlation is observed between the relative proportions of chemisorbed oxygen species and the amount of N2O produced, as shown in Figure 1(b). Therefore, surface adsorbed oxygen can be directly related to the nonselective catalytic reduction (NSCR) or NH3 oxidation side reactions that contribute to the formation of nitrous oxide. Redox Properties and Surface Acidities. Reducibility and surface acidity are considered the most two important properties of SCR catalysts. Figure 5(a) illustrates the H2-TPR profiles of the fresh and poisoned catalysts. Two peaks can be observed for the fresh samples: the main reduction band centered at 526 °C with a shoulder at 482 °C, which are assigned to the reduction of interacting surface tungsten oxide (W6+ to W4+) and to the reduction of V5+ to V3+, respectively, and the high-temperature peak (at 831 °C) corresponding to the reduction of W4+ to W.30−32 Both peaks shift to lower temperatures, 476 and 805 °C, respectively, and the area of the low-temperature peak and the total H2 consumption (Table 1) increase significantly following the addition of high-content arsenic oxide. These results suggest that the reduction of As species occurs concomitantly with the reduction of W6+ to W4+ and V5+ to V3+ at 400−560 °C.

Figure 3. Raman spectra of fresh and poisoned catalysts.

exhibit the second-order feature of TiO2 at approximately 795 cm−1 and a broad band in the 930−990 cm−1 range assigned to the symmetrical VO or WO stretching modes of the twodimensional surface species.24−26 This broad band becomes less pronounced and shifts toward lower frequencies as the As content increases. Since the Raman cross-section of the vanadium oxide species is approximately 4 times larger than that of the tungsten oxide species, these shifts may be caused by the longer bond distance of VO, which originates from V−O− As following the introduction of arsenic oxide. The XPS technique was used to explore the contents and chemical environments of As and O on the surface of fresh and poisoned catalysts. The surface As/Ti molar ratio is larger than the bulk molar ratio calculated by ICP. These results indicate an enrichment of As species on the catalyst surface with As content increase. Combined with XRD results, blocked pores in the catalyst may make the claudetite phase of arsenic oxide the main form of the oxide at low As content. The XPS spectra of As 3d are shown in Figure 4(a), and two peaks can be obtained by fitting the curves. As reported previously, the peak centered at 44.1− 45.1 eV can be assigned to As3+ and the peak centered at 45.9 eV can be assigned to As5+.27 As shown in Figure 4(a), only the As3+ signals exist in the As1# catalyst, and the presence of the As5+ species increases gradually and becomes the predominant form of arsenic with an increase in the As content. Furthermore, the As 3d binding energy moves to higher energy, which clearly suggests 9974

DOI: 10.1021/acs.est.5b02257 Environ. Sci. Technol. 2015, 49, 9971−9978

Article

Environmental Science & Technology

the reducibility of the catalyst and that the reduction is related to higher N2O production. The NH3-TPD profiles of fresh and poisoned catalysts are shown in Figure S1 of the Supporting Information (SI) and their total acidities are shown in Table 1. Two NH3 desorption peaks at 223 and 330 °C can be identified on the TPD curve for the fresh sample. Relatively strong (330 °C) acidities disappear gradually when the As content of poisoned catalysts is increased as the amount of NH3 desorption (Table 1) decreases, which is consistent with their SCR activities. At the same time, weak (223 °C) acidities diminish because the NH3 desorption peak shifts toward lower temperatures. Consequently, the NH3-TPD results imply that the addition of arsenic oxide decreases the NH3 adsorption and thus lowers the catalytic performance. In situ DRIFT spectra of adsorbed NH3 over fresh and poisoned catalysts were used to investigate the surface acid sites and structural changes. Figure 6(a) shows the FTIR spectra of NH3 adsorption over the four types of catalysts at 200 °C. The negative band at approximately 3663 cm−1, ascribed to an overlap between the V−OH and Ti−OH stretching frequencies, weakens with an increase in the As content. The peak at 3614 cm−1, which is attributed to the As−OH stretching frequency, gradually strengthens.11,14,31 In the N−H vibrational stretching region, predominantly coordinated NH3 bands at 3100−3400 cm−1 and NH4+ bands at 2800−3020 cm−1 can be found. The bands centered at 1603, 1146, and 1232 cm−1 can be ascribed to asymmetric and symmetric bending vibrations of NH3 linked to Lewis acid sites. The bands at 1440 and 1674 cm−1 are assigned to coordinated NH4+ species chemisorbed onto Brønsted acid sites.33−35 The negative peak at approximately 1990 cm−1, which is ascribed to the overtones of the VO or WO vibrational modes for the fresh sample, disappears after poisoning. This result indicates that the VO or WO modes are perturbed by As species and that chemical bonding occurs in the form of V− O−As or W−O−As.2,36 To further explore the thermal stability of Brønsted and Lewis acid sites for the four catalysts, the integrated peak areas for the different types of acid sites from 100 to 350 °C were evaluated and are presented in Figure 6(b). The corresponding DRIFT spectra of NH3 adsorption as a function of temperature are shown in SI Figure S2. Following the addition of As, both the Brønsted and Lewis acidity of the poisoned catalysts decrease more rapidly than the fresh sample, which is consistent with the NH3-TPD results shown above. At low temperature, the arsenic poisoned samples show fewer Lewis acid sites, whereas

Figure 5. H2-TPR (upper) and H2 consumption rate profiles (bottom) of fresh (a), As1# (b), As2# (c), and As3# (d) catalysts.

Additionally, we calculated the initial H2 consumption rate for the first reduction band of each sample to better explore the reducibility influenced by the addition of As. The results of these experiments are shown in Figure 5(b). The initial H 2 consumption rate decreases in the sequence As3# > As2# > As1# > fresh sample, in accordance with the total H 2 consumption. These results suggest that arsenic oxide enhances

Figure 6. In situ FTIR spectra (a), the contributions of the Lewis and Brønsted acid sites (b) of fresh and poisoned catalysts, and the NH3 desorption of the As3# catalyst in the temperature range of 50−350 °C (c). 9975

DOI: 10.1021/acs.est.5b02257 Environ. Sci. Technol. 2015, 49, 9971−9978

Article

Environmental Science & Technology

coordination (i.e., three As−O bonds and an AsO bond) for the pentavalent arsenic species with two adjacent arsenic sites on the optimized V6O20H10 cluster and TiO2 (001) plane. From the configurations of the poisoned models, both the As2O3 and As2O5 clusters can bond in a stable manner to the V2O5 (010) plane (As/Ti and As/V/Ti) and TiO2 (001) plane (As/Ti) via long V−O and Ti−O bonds. To determine the most favorable forms of these structures, the total energies of the relative systems were compared following incorporation of O2. The results show that the systems with the As2O5 cluster on top of the V6O20H10 cluster or TiO2 (001) plane have a considerably lower energy than the system with the As2O3 cluster plus the incorporation of O2. However, when arsenic oxide and vanadia with a 4-fold coordination bond together on top of the TiO2 (001) plane, the system with the trivalent arsenic species is more stable (−6928.17 Ha.) than the system with the pentavalent arsenic species (−6928.11 Ha.). Therefore, the As2O5 species prefers to cover the V2O5 active sites and bond with surface Ti−OH group, whereas the trivalent arsenic species prefers to remain around the tetrahedrally coordinated VO4 species consisting of a single (mono-oxo) terminal VO bond and three bridging V−O bonds. To further investigate the influence of As species on the V2O5 cluster, the projected density of states (PDOS) of the V orbitals and the density of states (DOS) have been calculated to compare the energy difference between the bottom of the conduction band and the Fermi surface before and after poisoning. As shown in SI Figure S3, The energy difference of V in As(III)V/Ti is nearly unchanged compared with that in V/ Ti, whereas the bottom of the conduction band for the V orbitals in As(V)V/Ti clearly shifts to a lower-energy region (at 1.2 eV). This demonstrates that the addition of As2O5 improves the ability of the V2O5/TiO2 catalyst to be oxidized, which is consistent with the N2O production and TPR results. Comparison of Different Regeneration Methods. Although hydrogen dioxide solutions could effectively remove As2O5 from poisoned catalysts, the active constituents of the catalysts, such as V2O5 and WO3, are lost because of their strong solubility.15 Therefore, further efforts should focus on regenerating poisoned catalysts while maintaining as many of the original active ingredients as possible to lengthen the lifetime of the catalyst. The As3# samples were regenerated using different types of cleaning solutions, and the As removal ratios and active constituent residual ratios were calculated by XRF to compare the various regeneration methods (Table 2 and SI Table S2). As

weak Brønsted acid sites increase slightly for As3#. Thus, the V O bond, which is related to Lewis acid sites, is disturbed in the presence of arsenic species, and abundant As5+−OH moieties can provide additional Brønsted acid sites, which have a lower thermal stability. The As2# sample possesses the fewest Brønsted and Lewis acid sites, which may be correlative with the influence of As3+ species and surface coverage. Moreover, as shown in Figure 6(c) and SI Figure S2, a weak peak at 1538 cm−1, which we attribute to the δ(NH2−) mode, is present and becomes stronger with increasing temperature for the As3# catalyst between 200 and 350 °C.37,38 This existence of this species indicates the higher oxidative ability of the catalysts and results in superfluous N2O production because many pentavalent arsenic species cover the catalyst surface. In the O−H vibrational stretching region, the negative band corresponding to surface O−H stretching frequencies of TiO2 essentially vanish above 200 °C, and the only remaining peak is ascribed to the As−OH stretching frequency at 3613 cm−1. These results suggest that the Brønsted acid sites of As3# are provided by the As−OH groups and that these sites are more conducive to N2O production than N2 production at high temperatures. Optimized Structures by Theoretical Calculations. As noted above, the form of the arsenic oxide on the poisoned catalyst is related to the As proportion of the catalyst’s composition. At low As concentrations, trivalent arsenic species are the major components and are inclined to permeate into bulk-phase channels. At high As concentrations, pentavalent arsenic species, which possess a higher oxidation capacity, cover the surface of the catalyst. Finally, at intermediate As concentrations, both As3+ and As5+ are found on the catalyst surface. To explore some likely forms of As that exist following its interacting with V2O5 and TiO2, the optimized structures and corresponding total energies of As2O3 and As2O5 clusters on the V2O5 cluster and the TiO2 (001) plane have been determined. These results are displayed in Figure 7, and the complete calculation results are listed in SI Table S3. The arsenic oxide dimers are modeled as having a 3-fold coordination (i.e., three As−O bonds) for the trivalent arsenic species and a 4-fold

Table 2. Comparison of Relevant Parameters on Major Regeneration Methods regeneration method

H2O2

NaOH

concentration (%) pH As removal ratio (%) V residual ratio (%) W residual ratio (%)

4 1.8 66.2 74.9 95.7

1 13 77.9 82.7 77.3

Ca(NO3)2 1 63.3 96 96.2

2 11 72.4 94.1 94

4 74.2 92.2 93.1

shown in Table 2, arsenic oxide is removed more effectively from poisoned catalysts in alkaline conditions than in acidic conditions. A NaOH solution (1 wt %) demonstrates the highest As removal efficiency (77.9%), nevertheless this approach is subjected to a higher loss of active components (V-17.3%; W-22.7%). Additionally, secondary poisoning may occur following the introduction of the alkali metal (Na), which is considered an important poison for SCR catalysts used at coal-

Figure 7. Optimized structures of trivalent (left) and pentavalent (right) arsenic species on the V2O5 (010) plane and TiO2 (001) plane. Vanadium atoms are green, oxygen atoms are red, arsenic atoms are purple, titanium atoms are gray and hydrogen atoms are pale yellow. 9976

DOI: 10.1021/acs.est.5b02257 Environ. Sci. Technol. 2015, 49, 9971−9978

Environmental Science & Technology



fired plants. Moreover, Ca(NO3)2 cleaning solutions in alkaline conditions remove arsenic oxide effectively with a lower loss of active components, which may be related to the larger pKsp value of Ca3(AsO4)2 (i.e., 18.17). Figure 8 shows the NOx conversion

Article

AUTHOR INFORMATION

Corresponding Author

*Phone: +86 10 62771093; e-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was financially supported by the National Natural Science Foundation of China (21325731, 21407088, and 51478241), the National High-Tech Research and Development (863) Program of China (2013AA065401 & 2013AA065304), and the China Postdoctoral Science Foundation (2013M530643).



(1) Cristiani, C.; Bellotto, M.; Forzatti, P.; Bregani, F. On the morphological properties of tungsta-titania de-NOxing catalysts. J. Mater. Res. 1993, 8 (08), 2019−2025. (2) Topsoe, N. Y. Mechanism of the selective catalytic reduction of nitric-oxide by ammonia elucidated by in-situ online Fourier-transform infrared-spectroscopy. Science 1994, 265 (5176), 1217−1219. (3) Martens, J. A.; Cauvel, A.; Francis, A.; Hermans, C.; Jayat, F.; Remy, M.; Keung, M.; Lievens, J.; Jacobs, P. A. NOx abatement in exhaust from lean-burn combustion engines by reduction of NO2 over silvercontaining zeolite catalysts. Angew. Chem., Int. Ed. 1998, 37 (13−14), 1901−1903. (4) Burch, R.; Scire, S. Selective catalytic reduction of nitric-oxide with ethane and methane on some metal exchanged ZSM-5 zeolites. Appl. Catal., B 1994, 3 (4), 295−318. (5) Busca, G.; Lietti, L.; Ramis, G.; Berti, F. Chemical and mechanistic aspects of the selective catalytic reduction of NOx by ammonia over oxide catalysts: A review. Appl. Catal., B 1998, 18 (1−2), 1−36. (6) Nicosia, D.; Czekaj, I.; Kröcher, O. Chemical deactivation of V2O5/ WO3−TiO2 SCR catalysts by additives and impurities from fuels, lubrication oils and urea solution: Part II. Characterization study of the effect of alkali and alkaline earth metals. Appl. Catal., B 2008, 77 (3−4), 228−236. (7) Castellino, F.; Jensen, A. D.; Johnsson, J. E.; Fehrmann, R. Influence of reaction products of K-getter fuel additives on commercial vanadia-based SCR catalysts: Part II. Simultaneous addition of KCl, Ca(OH)2, H3PO4 and H2SO4 in a hot flue gas at a SCR pilot-scale setup. Appl. Catal., B 2009, 86 (3−4), 206−215. (8) Chen, L.; Li, J.; Ge, M. The poisoning effect of alkali metals doping over nano V2O5−WO3/TiO2 catalysts on selective catalytic reduction of NOx by NH3. Chem. Eng. J. 2011, 170 (2−3), 531−537. (9) Wu, Z.; Jin, R.; Wang, H.; Liu, Y. Effect of ceria doping on SO2 resistance of Mn/TiO2 for selective catalytic reduction of NO with NH3 at low temperature. Catal. Commun. 2009, 10 (6), 935−939. (10) Tang, F.; Xu, B.; Shi, H.; Qiu, J.; Fan, Y. The poisoning effect of Na+ and Ca2+ ions doped on the V2O5/TiO2 catalysts for selective catalytic reduction of NO by NH3. Appl. Catal., B 2010, 94 (1−2), 71− 76. (11) Hilbrig, F.; Göbel, H. E.; Knözinger, H.; Schmelz, H.; Lengeler, B. Interaction of arsenious oxide with DeNox-catalysts: An X-ray absorption and diffuse reflectance infrared spectroscopy study. J. Catal. 1991, 129 (1), 168−176. (12) Hums, E. A catalytically highly-active, arsenic oxide resistant VMo-O phase  results of studying intermediates of the deactivation process of V2O5-MoO3-TiO2 (anatase) DeNOx catalysts. Res. Chem. Intermed. 1993, 19 (5), 419−441. (13) Senior, C. L.; Lignell, D. O.; Sarofim, A. F.; Mehta, A. Modeling arsenic partitioning in coal-fired power plants. Combust. Flame 2006, 147 (3), 209−221. (14) Lange, F. C.; Schmelz, H.; Knözinger, H. Infrared-spectroscopic investigations of selective catalytic reduction catalysts poisoned with arsenic oxide. Appl. Catal., B 1996, 8 (2), 245−265.

Figure 8. NOx conversion (a) and N2O production (b) of fresh, poisoned and regenerated catalysts. Reaction conditions: [NO] = [NH3] = 500 ppm, [O2] = 3%, total flow rate =200 mL/min, GHSV = 120 000 mL/(g·h).

and N2O production over the fresh catalyst, poisoned catalyst and regenerated catalysts with different concentrations of Ca(NO3)2 solutions. The results show that the catalytic activity is recovered above 80% and that N2O production declines sharply and reaches the level of fresh catalyst at 350−500 °C when the mass concentration of Ca(NO3)2 in the regenerated liquid is higher than 2%. This result suggests that most of the arsenic oxides can be removed by treatment with Ca(NO3)2 and a low concentration of acid following solutions without excessive losses of V2O5 and WO3. To explore the residual As species that may exist on the catalyst, surface and bulk phase As/Ti molar ratios of poisoned and regenerated catalysts using the Ca(NO3)2 regeneration method were also estimated by XPS and ICP, respectively (SI Table S4). In contrast to the poisoning process, the surface As/Ti molar ratios are lower than the bulk molar ratios after regeneration of the catalyst, regardless of the Ca(NO3)2 concentration. These results suggest that surfaceexposed As2O5 species can be removed more easily than the As2O3 species that accumulate in channels of the catalyst. The small XRD peak attributed to the clauditite phase of As2O3 remains for the regenerated catalysts (not shown). Furthermore, according to the normalized peak area ratios between the (101) and (040) planes in the XRD patterns listed in SI Table S3, the ratio maintains a high value (of 0.51) when the concentration of Ca(NO3)2 in the regenerated liquid is 2%, although its value decreases after regeneration. Therefore, another preferable regeneration method is still required to remove As2O3 that resides in the micropores of the catalyst.



REFERENCES

ASSOCIATED CONTENT

S Supporting Information *

Associated figures and tables are provided. The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.est.5b02257. 9977

DOI: 10.1021/acs.est.5b02257 Environ. Sci. Technol. 2015, 49, 9971−9978

Article

Environmental Science & Technology (15) Peng, Y.; Li, J.; Shi, W.; Xu, J.; Hao, J. Design strategies for development of SCR catalyst: Improvement of alkali poisoning resistance and novel regeneration method. Environ. Sci. Technol. 2012, 46 (22), 12623−12629. (16) Peng, Y.; Li, J.; Si, W.; Luo, J.; Wang, Y.; Fu, J.; Li, X.; Crittenden, J.; Hao, J. Deactivation and regeneration of a commercial SCR catalyst: Comparison with alkali metals and arsenic. Appl. Catal., B 2015, 168− 169 (0), 195−202. (17) Peng, Y.; Li, J.; Si, W.; Luo, J.; Dai, Q.; Luo, X.; Liu, X.; Hao, J. Insight into deactivation of commercial SCR catalyst by arsenic: an experiment and DFT study. Environ. Sci. Technol. 2014, 48 (23), 13895− 13900. (18) Hums, E. Is advanced SCR technology at a standstill? A provocation for the academic community and catalyst manufacturers. Catal. Today 1998, 42 (1−2), 25−35. (19) Kompio, P. G. W. A.; Brückner, A.; Hipler, F.; Auer, G.; Löffler, E.; Grünert, W. A new view on the relations between tungsten and vanadium in V2O5WO3/TiO2 catalysts for the selective reduction of NO with NH3. J. Catal. 2012, 286 (0), 237−247. (20) Wang, C.; Yang, S.; Chang, H.; Peng, Y.; Li, J. Dispersion of tungsten oxide on SCR performance of V2O5WO3/TiO2: Acidity, surface species and catalytic activity. Chem. Eng. J. 2013, 225 (0), 520− 527. (21) Lietti, L.; Nova, I.; Ramis, G.; Dall’Acqua, L.; Busca, G.; Giamello, E.; Forzatti, P.; Bregani, F. Characterization and Reactivity of V2O5− MoO3/TiO2 De-NOx SCR Catalysts. J. Catal. 1999, 187 (2), 419−435. (22) Alemany, L. J.; Lietti, L.; Ferlazzo, N.; Forzatti, P.; Busca, G.; Giamello, E.; Bregani, F. Reactivity and Physicochemical Characterization of V2O5-WO3/TiO2 De-NOx Catalysts. J. Catal. 1995, 155 (1), 117−130. (23) Jegadeesan, G.; Al-Abed, S. R.; Sundaram, V.; Choi, H.; Scheckel, K. G.; Dionysiou, D. D. Arsenic sorption on TiO2 nanoparticles: Size and crystallinity effects. Water Res. 2010, 44 (3), 965−973. (24) Vuurman, M. A.; Wachs, I. E.; Hirt, A. M. Structural determination of supported vanadium pentoxide-tungsten trioxidetitania catalysts by in situ Raman spectroscopy and x-ray photoelectron spectroscopy. J. Phys. Chem. 1991, 95 (24), 9928−9937. (25) Madia, G.; Elsener, M.; Koebel, M.; Raimondi, F.; Wokaun, A. Thermal stability of vanadia-tungsta-titania catalysts in the SCR process. Appl. Catal., B 2002, 39 (2), 181−190. (26) Bond, G. C.; Tahir, S. F. Vanadium oxide monolayer catalysts Preparation, characterization and catalytic activity. Appl. Catal. 1991, 71 (1), 1−31. (27) Shan, W.; Liu, F.; He, H.; Shi, X.; Zhang, C. A superior Ce-W-Ti mixed oxide catalyst for the selective catalytic reduction of NOx with NH3. Appl. Catal., B 2012, 115−116 (0), 100−106. (28) Qu, R.; Gao, X.; Cen, K.; Li, J. Relationship between structure and performance of a novel cerium-niobium binary oxide catalyst for selective catalytic reduction of NO with NH3. Appl. Catal., B 2013, 142, 290−297. (29) Lange, F.; Schmelz, H.; Knözinger, H. An X-ray photoelectron spectroscopy study of oxides of arsenic supported on TiO2. J. Electron Spectrosc. Relat. Phenom. 1991, 57 (3−4), 307−315. (30) Reiche, M. A.; Maciejewski, M.; Baiker, A. Characterization by temperature programmed reduction. Catal. Today 2000, 56 (4), 347− 355. (31) Reiche, M. A.; Bürgi, T.; Baiker, A.; Scholz, A.; Schnyder, B.; Wokaun, A. Vanadia and tungsta grafted on TiO2: influence of the grafting sequence on structural and chemical properties. Appl. Catal., A 2000, 198 (1−2), 155−169. (32) Engweiler, J.; Harf, J.; Baiker, A. WOx/TiO2 catalysts prepared by grafting of tungsten alkoxides: Morphological properties and catalytic behavior in the selective reduction of NO by NH3. J. Catal. 1996, 159 (2), 259−269. (33) Topsoe, N. Y.; Dumesic, J. A.; Topsoe, H. Vanadia-titania catalysts for selective catalytic reduction of nitric-oxide by ammonia 0.2. studies of active-sites and formulation of catalytic cycles. J. Catal. 1995, 151 (1), 241−252.

(34) Topsoe, N. Y.; Topsoe, H.; Dumesic, J. A. Vanadia-titania catalysts for selective catalytic reduction (SCR) of nitric-oxide by ammonia 0.1. combined temperature-programmed in-situ FTIR and online mass-spectroscopy studies. J. Catal. 1995, 151 (1), 226−240. (35) Wu, Z.; Jiang, B.; Liu, Y.; Wang, H.; Jin, R. DRIFT study of manganese/titania-based catalysts for low-temperature selective catalytic reduction of NO with NH3. Environ. Sci. Technol. 2007, 41 (16), 5812−5817. (36) Ramis, G.; Busca, G.; Bregani, F.; Forzatti, P. Fourier transforminfrared study of the adsorption and coadsorption of nitric-oxide, nitrogen-dioxide and ammonia on vanadia titania and mechanism of selective catalytic reduction. Appl. Catal. 1990, 64 (1−2), 259−278. (37) Chang, H.; Li, J.; Chen, X.; Ma, L.; Yang, S.; Schwank, J. W.; Hao, J. Effect of Sn on MnOx-CeO2 catalyst for SCR of NOx by ammonia: Enhancement of activity and remarkable resistance to SO2. Catal. Commun. 2012, 27, 54−57. (38) Thirupathi, B.; Smirniotis, P. G. Nickel-doped Mn/TiO2 as an efficient catalyst for the low-temperature SCR of NO with NH3: Catalytic evaluation and characterizations. J. Catal. 2012, 288, 74−83.

9978

DOI: 10.1021/acs.est.5b02257 Environ. Sci. Technol. 2015, 49, 9971−9978