Poisoning E ﬀ ects of Phosphorus, Potassium and Lead on V 2 O 5 -WO 3 / TiO 2 Catalysts for Selective Catalytic Reduction with NH 3

: The poisoning e ﬀ ect of single elements on commercial V 2 O 5 -WO 3 / TiO 2 catalysts has been studied in the past decades. In this study, the combined e ﬀ ects of two multi-element systems (phosphorus-potassium and phosphorus-lead) on V 2 O 5 -WO 3 / TiO 2 catalysts were studied by diverse characterizations. The results show that potassium and lead can result in the deactivation of catalysts to di ﬀ erent degrees by reacting with active acid sites and reducing the amount of V 5 + . However, phosphorus displays slight negative inﬂuence on the NO x conversion of the catalyst due to the comprehensive e ﬀ ect of reducing V 5 + amount and generating new acid sites. The samples poisoned by phosphorus–potassium and phosphorus–lead have higher NO x conversion than that by potassium or lead, because doped potassium or lead atoms may react with new acid sites generated by phosphate, which liberates more V–OH on the surface of catalysts and reduces the poisoning e ﬀ ects of potassium or lead on vanadium species and active oxygen species. Pb. The atomic ratio of V 5 + / V 4 + decreases from 1.21 to 0.43 and 0.51 after doping with K and Pb. Compared with D-K and D-Pb, the atomic ratio of V 5 + / V 4 + in D-P-K and D-P-Pb samples increases largely. The result indicates that the treatment with (NH 4 ) 2 HPO 4 can inhibit the transformation from V 5 + into V 4 + resulting from the adverse inﬂuence of Pb and K.


Introduction
The selective catalytic reduction (SCR) of nitrogen oxides (NO x ) with ammonia is one of the most effective technologies to reduce NO x emission generated by combustion process [1][2][3], such as coal-fired power plants and municipal solid waste incinerators [4]. The catalyst mostly used for this purpose is V 2 O 5 -WO 3 /TiO 2 . V 2 O 5 is considered as the active phase and WO 3 is considered as a promoter that inhibits the transformation of anatase to rutile, favors the spreading of the vanadia on the catalyst surface, and increases the acidity of the catalyst [5,6].
Though V 2 O 5 -WO 3 /TiO 2 catalysts show excellent activity in NH 3 -SCR process, many compounds including alkali/alkali earth metal elements, Pb, Zn, P and As are found in exhaust flue gas streams, which can lead to the deactivation of SCR catalysts in their lifetime. The deactivation mechanism of single element was studied in the past decades [1,[7][8][9][10][11][12][13]. Recently, the combined effects of different elements in the flue gas aroused the attention of scholars. Yu et al. [14,15] suggested that SO 2 in the flue gas promoted the activity of K-and Pb-poisoned catalysts. Kong et al. [16] discovered that new acid sites generated on the SCR catalyst by loading HgCl 2 . Our recent work indicated SO 2 and HCl can promote the activity of Pb-poisoned SCR catalyst [4]. However, few studies were about the combined effects of P and other poisoning elements. Schobing et al. [1] found the simultaneous presence of Ca and Zn on the catalyst increased the deactivation, but the additivity of individual effects was not observed and the presence of phosphates anions reduced the poisoning effects of Ca and Zn. Klimczak et al. [17] found an alleviated poisoning of Ca in the presence of phosphates and sulfates due to a regeneration of the ammonia adsorption capacity of the catalyst. It seems that phosphorus (P) could neutralize the negative effects of some poisoning elements, even though P has relatively small negative influence on V 2 O 5 -WO 3 /TiO 2 catalysts in consideration of the large amount of phosphorus addition. However, Chen et al. [18] discovered that alkali metal had the strongest poisoning effect on the V-based catalyst, and the poisoning effect of Pb was between K 2 O and Na 2 O (K 2 O > PbO > Na 2 O). And K is a major component of alkali metals in the coal-fired flue gas. Tokarz et al. [19] measured a large amount of Pb contained in the flue gas of municipal solid waste incinerators. It is; therefore, desirable to study the combined poisoning effects of P and K or Pb on commercial V 2 O 5 -WO 3 /TiO 2 catalysts.
In this work, the commercial V 2 O 5 -WO 3 /TiO 2 catalyst was deactivated by P, K, Pb and the combination of P and K or Pb. The combined poisoning effects of P and K or Pb were studied by a series of characterizations. The influence of surface O, V, P and Ti species was investigated by X-ray photoelectron spectroscopy (XPS) and the amount and types of acid sites present in catalysts were further studied by temperature-programmed desorption of NH 3 (NH 3 -TPD) and in situ diffuse reflectance infrared Fourier-transform spectroscopy (in situ DRIFTS). Figure 1a shows the SCR activity of the catalyst samples in the temperature range of 250 to 380 • C under a GHSV of 60,000 h −1 . The experiments at each specific temperature were maintained for 1 h. The NO x conversion of fresh sample keeps around 90% in the temperature range of 250 to 380 • C. Compared with fresh sample, the NO x conversion of D-P (poisoned by P) decreases slightly (~4%). The poisoning effects of K and Pb on the V 2 O 5 -WO 3 /TiO 2 catalyst obviously decrease the NO x conversion of D-K (poisoned by K) and D-Pb (poisoned by Pb) sample. However, the NO x conversion rates of D-P-K (poisoned by P and K) and D-P-Pb (poisoned by P and Pb) pretreated with (NH 4 ) 2 HPO 4 are higher than that of D-K and D-Pb, respectively. It implies that the treatment with (NH 4 ) 2 HPO 4 counteracts the poisoning effects of K and Pb on the catalyst to some degree. Furthermore, Figure  S1a shows the N 2 selectivity of poisoned catalyst increases at increasing temperature in the range 250-280 • C. Then N 2 selectivity of poisoned catalyst decreases with increasing temperature above 300 • C. Figure 1b shows the NO x conversion of the fresh catalyst and poisoned samples at 350 • C for 3500 min. The fresh catalyst shows high stability. Its NO x conversion reaches the stable stage in 10 min and keeps around 95% until the end of the test. The NO x conversion rates of D-K and D-Pb samples also slightly decrease with time. The NO x conversion rates of the catalysts treated by (NH 4 ) 2 HPO 4 (D-P, D-P-K and D-P-Pb) decrease by 2%, 3% and 3%, respectively, in the first 1500 min, and then are relatively stable during the test. The N 2 selectivity of fresh sample keeps stable, and that of the poisoned catalyst decreases a little during the test ( Figure S1b).

Structure and Surface Acid Sites
XRD patterns of fresh catalyst and poisoned samples are displayed in Figure 2 Figure S2).   Table 1 summarizes the element loading amount, specific surface areas (S BET ), total pore volume (V p ) and average pore diameter (D A ) for all catalysts. The N 2 adsorption-desorption isotherms of all samples and pore size distribution are displayed in Figures S3 and S4. According to the International Union of Pure and Applied Chemistry (IUPAC) classification, the isotherms are type-IV nitrogen isotherm corresponding to typical mesoporous materials. The S BET and V p of the D-Pb and D-K samples are almost similar to the fresh catalyst, suggesting that Pb and K have little influence on the pore structure of the catalyst. The treatment with (NH 4 ) 2 HPO 4 decreases the S BET and increases the D A of the samples, The SEM images of fresh catalysts and poisoned samples are exhibited in Figure S5. A little agglomeration is observed on the samples containing P, which may block the pores and change S BET and V p of the samples. Table 1. Element loading amount, specific surface areas, total pore volume and average pore diameter of samples.

Sample
Loading Amount (wt %) The results of structure analysis imply that significant deactivation effect of Pb and K may well result from the chemical reaction between poisoning elements and active sites over catalysts. Moreover, the NO x conversion of samples containing P and another poisoning element is not affected by the reduced S BET . There may be other reasons for the activity change of these samples. Therefore, further characterizations were conducted.
It has been recognized that the surface acid sites over V 2 O 5 -WO 3 /TiO 2 catalyst play a crucial role in the selective catalytic reduction process [4,20]. Therefore, the NH 3 -TPD was performed to investigate the combined effects of poison elements on acid sites of the catalyst. The NH 3 -TPD profile are exhibited in Figure 3. It shows that all curves recorded from 100-600 • C are divided into two parts. The part below 280 • C is attributed to NH 3 desorption on the weak acid sites, and the other part above 280 • C is ascribed to NH 3 desorption on the strong acid sites. The total amounts of NH 3 adsorption on each sample are calculated from the integrated areas of the corresponding curves and normalized by the value of Fresh sample. They are sequenced as follows: It shows that the amount of desorbed NH 3 on the D-P sample increases by 27% compared with fresh sample especially the amount of NH 3 adsorbed on the weak acid sites. However, the Mass Spectrum (MS) signal of desorbed NH 3 decreases obviously with doping K (by 77%) or Pb (by 36%), which suggests the doping of K or Pb reduces the acidity of catalyst significantly. The treatment with (NH 4 ) 2 HPO 4 increases the amount of NH 3 adsorption of K and Pb poisoning samples by 161% and 6%, respectively. It may be the reason that D-P-K and D-P-Pb samples have higher DeNO x efficiency than D-K and D-Pb samples.
Besides the surface concentration of acid sites, the types of acid sites are also important for the SCR catalysts [21]. Therefore, a set of in situ DRIFTS experiments was conducted to obtain a thorough understanding of the interaction between NH 3 and the surface acid sites.
The in situ DRIFTS results of samples exposed to 500 ppm NH 3 in N 2 at 350 and 200 • C for 30 min are displayed in Figure 4. Figure 4a shows that three bands at 1605, 1401 and 1238 cm −1 are detected in the Fresh catalyst at 350 • C. It is recognized that the bands located at 1605 and 1238 cm −1 are ascribed to NH 3 adsorbed on Lewis acid sites [21][22][23][24], and the band at 1401 cm −1 is associated with asymmetric deformation modes of ammonium ions on Brønsted acid sites [25,26]. The observed bands at 1610, 1426, 1409 and 1659 cm −1 on D-P sample are stronger obviously than that on fresh sample at 350 • C. The band located at 1659, 1426 and 1409 cm −1 were proved to be NH 3 adsorbed on the Brønsted acid sites [23,27,28]. It implies that the treatment with (NH 4 ) 2 HPO 4 generates new Brønsted and Lewis acid sites and improves the acidity of V 2 O 5 -WO 3 /TiO 2 catalyst [29][30][31]. For the in situ DRIFTS spectrum of D-K and D-Pb, only very weak bands are detected. It indicates that K and Pb can result in huge damage to Lewis acid sites and Brønsted acid sites. However, when K or Pb is introduced to the sample pretreated with (NH 4 ) 2 HPO 4 , the bands attributed to NH 3 adsorbed on Lewis acid sites and Brønsted acid sites can be detected with enhanced intensity. It suggests the treatment with (NH 4 ) 2 HPO 4 can weaken the poisoning effects of K and Pb on the V 2 O 5 -WO 3 /TiO 2 catalyst.  The stability of the adsorbed ammonia on the catalysts was analyzed by comparing the spectra of samples collected at 200 and 350 • C, respectively. The intensity of all peaks due to the adsorbed ammonia decreased with increasing temperature except the peaks detected on D-P. As shown in Figure 4b, the bands located at 1662, 1608, 1445, 1298 and 1238 cm −1 are observed on the fresh catalysts. The bands at 1662 and 1445 cm −1 are assigned to NH 4 + chemisorbed on Brønsted acid site. The bands at 1608, 1298 and 1238 cm −1 are attributed to NH 3 adsorbed on the Lewis acid sites [21,32,33]. When the temperature reaches 350 • C, the bands on the fresh catalysts become to weaken and the band at 1298 cm −1 vanishes. However, the bands on the D-P samples still keep strong intensity especially the bands at 1426 and 1409 cm −1 . It implies that the acid sites generated by P species have higher thermo stability. This is maybe the reason the D-P has good NO x conversion at high temperature. The Brønsted acid sites on the D-P-K and D-P-Pb decrease with increasing temperature clearly. The doped K and Pb atoms tend to react with acid sites generated by P species. The results suggest that the treatment with (NH 4 ) 2 HPO 4 can weaken the poisoning effect of Pb and K on fresh catalyst because of the formation of new Lewis acid sites and Brønsted acid sites. A total of 500 ppm NO and 5% O 2 were introduced to the D-P sample and fresh sample pretreated with NH 3 and purged with N 2 for 10 min. The reaction spectra between NO + O 2 and adsorbed NH 3 species are presented in Figure 5. For the spectrum of fresh catalyst, the band at 1238 cm −1 gradually weakens with time and vanishes after NO + O 2 is introduced for 30 min, and the bands at 1605 and 1401 cm −1 weaken during the test. In addition, new bands are detected at 1626 and 1599 cm −1 , which could be assigned to adsorbed NO 2 and bridged nitrate [34,35]. The results indicate that adsorbed NH 3 species on fresh sample are consumed in the NO + O 2 stream. For the spectrum of D-P sample, the bands at 1659, 1607, 1426 and 1409 cm −1 weaken during test time and the band at 1659 cm −1 almost vanishes for 30 min. A new band located at 1595 cm −1 assigned to adsorbed bridged nitrate is detected. The results indicate that adsorbed NH 3 species on Lewis acid sites and Brønsted acid sites, partly generated by the treatment of (NH 4 ) 2 HPO 4 , can be consumed in the NO + O 2 stream as well. The above experiments suggest that the reaction mechanism on the samples pretreated with (NH 4 ) 2 HPO 4 still follows the Eley-Rideal mechanism: Adsorbed ammonia species react with gas-phase or weakly-adsorbed NO [36].

Surface Chemical State
XPS was carried out to illustrate the change of oxidation state of the active species after poisoning. Figure 6 shows the XPS results of O 1s, V 2p 3/2 , P 2p and Ti 2p. The spectra of O 1s were fitted into three peaks which are the lattice oxygen in the metal oxides (O γ ), the surface oxygen by hydroxyl species (O β ) and the absorbed oxygen or/and weakly bonded oxygen species (O α ) [21,37,38]. The positions and concentrations of oxygen species are displayed in the Table 2. Surface chemisorbed oxygen has been reported to be the most active oxygen and plays an important role in oxidation reactions [20,39]. It can be found that O β concentration of D-P sample is higher than that of fresh sample, indicating that there are more hydroxyl species (-OH) after doping which can form P-OH on the surface of catalysts [31]. However, the D-P sample has lower NO x conversion than fresh catalyst, indicating P-OH has lower activity than that of V-OH. The O β concentration of D-K and D-Pb samples is lower than that of fresh sample, suggesting K and Pb can react with surface hydroxyl species and form -O-K or -O-Pb-O-, in accordance with other researches [20,40]. For D-P-K and D-P-Pb samples, it can be found that O β concentration increases largely, suggesting more -OH species still exist on the catalyst surface. The binding energy value of O γ shifts from 529.9 to 529.7 and 529.8 eV after the Pb and K doping. This may be because Pb and K donates electrons to lattice oxygen. However, in the D-P-K and D-P-Pb samples, the binding energy value of O γ keeps similar to the fresh catalyst. It suggests that the treatment with (NH 4 ) 2 HPO 4 may obstruct the electron interaction of Pb and K and the catalyst.  Based on previous reports [41,42], the SCR activity of catalyst is positively correlated with the surface V 5+ /V 4+ ratio. The spectra of V 2p 3/2 are fitted into two peaks related to V 5+ (~517.1 eV) and V 4+ (~516.2 eV) [43]. The atomic ratio of V 5+ /V 4+ can be determined according to the peaks area ratio of V 5+ /V 4+ . Table 3 summarizes the peak position and atomic ratio of V 5+ /V 4+ in different samples. The atomic ratio of V 5+ /V 4+ decreases obviously after doping poisoning elements, indicating P, Pb and K can result in the transformation from V 5+ into V 4+ in V 2 O 5 -WO 3 /TiO 2 catalysts. However, the adverse influence of P on surface V 5+ /V 4+ ratio was much weaker than K and Pb. The atomic ratio of V 5+ /V 4+ decreases from 1.21 to 0.43 and 0.51 after doping with K and Pb. Compared with D-K and D-Pb, the atomic ratio of V 5+ /V 4+ in D-P-K and D-P-Pb samples increases largely. The result indicates that the treatment with (NH 4 ) 2 HPO 4 can inhibit the transformation from V 5+ into V 4+ resulting from the adverse influence of Pb and K.
The P 2p binding energy of samples containing P is detected at 133.8 eV, suggesting that P in the samples exists in a pentavalent-oxidation state (P 5+ ) [44], and no peak is observed at 128.6 eV that is the characteristic binding energy of P 2p in Ti-P, indicating the absence of Ti-P bonds in the samples [29,30,45]. Table 3. XPS results of V 2p 3/2 and surface atomic ratio of V 5+ /V 4+ for each sample.

Samples
Binding Energy (eV)   For the spectra of Ti 2p, two peaks centered at 458.6-459 eV and 464.4-464.9 eV can be ascribed to Ti 2p 3/2 and Ti 2p 1/2 , respectively, suggesting the presence of Ti 4+ in all samples [25,46]. When the catalyst treated with (NH 4 ) 2 HPO 4 , the Ti 2p 3/2 and Ti 2p 1/2 peaks shift to higher energies, from 458.7 to 458.9 eV and 464.6 to 464.8 eV, respectively. It means the titanium species in D-P samples have lower density of electron cloud density than that in fresh samples [45,47]. It may because the electrons around the titanium species transfer to oxygen species, which interact with P dopants. It also proves that P 5+ can replace part of Ti 4+ in TiO 2 lattice to form a Ti-O-P linkage. The similar location of the Ti 2p peaks for fresh, D-K and D-Pb samples, indicates that K and Pb may not react with TiO 2 [4,48]. In addition, D-P-K and D-P-Pb have a similar Ti 2p peak center with D-P, which implies that K or Pb does not affect the reaction between P species and TiO 2 . The XPS spectra of W 4f were displayed in the Figure S6 and the information was supplied in the Supplementary Materials.

Poisoning Effect of Phosphorus, Potassium and Lead on V 2 O 5 -WO 3 /TiO 2 Catalysts
Based on the NH 3 -TPD, DRIFT, XPS spectra analysis and previous studies [29,49,50], the proposed combined poisoning effect of phosphorus, potassium or lead on V 2 O 5 -WO 3 /TiO 2 can be concluded in Figure 7. For fresh catalyst, based on Eley-Rideal mechanism, the adsorbed NH 4 + species on -OH can be activated by V 5+ active sites and react with NO in the flue gas [51,52]. The poisoning effects of K and Pb on the catalyst can be assigned to the neutralization of the acid sites, the decrease of surface chemisorption oxygen and the reduction of V 5+ species [10,40]. This results in weakening the NH 3 adsorption capacity and reducibility of the catalyst. After the treatment with (NH 4 ) 2 HPO 4 , the doped P species occupy the acid sites. However, new Brønsted acid sites (P-OH) form [31], and the acid sites also have the ability to adsorb NH 3 . The NH 4 + species adsorbed on the Brønsted acid sites (P-OH) could be activated by surface V active sites due to the excellent redox property of nearby V 5+ species. Hence, the slightly weakened NH 3 -SCR performance of D-P samples can be assigned to the reduction of V 5+ species. For D-P-K and D-P-Pb, the in situ DRIFTS analysis results suggest doped Pb and K atoms may react with acid sites generated by phosphate, and make them inactive in the adsorption of NH 3 . Each K atom consumes one acid site, and each Pb atom reacts with two acid sites. Hence, it seems that more K and Pb around the P atom will liberate more V-OH on the surface of catalysts, so the treatment with (NH 4 ) 2 HPO 4 alleviates the poisoning effect of potassium or lead on the catalysts [53].

Catalysts Preparation
The commercial V 2 O 5 -WO 3 /TiO 2 catalyst (named as Fresh, with 1.16 wt.% V and 2.42 wt.% W) was ground and sieved to 60-80 mesh. All the chemicals used in the catalyst preparation process were analytical grade and purchased from Sinopharm Chemical Reagent Co., Ltd (Shanghai, China). The P poisoned samples were prepared by impregnating the commercial catalyst particles into aqueous (NH 4 ) 2 HPO 4 (0.25 mol/L) solution for 6 h at room temperature. Then the samples were dried at 105 • C for 12 h followed by calcination at 500 • C for 4 h in a muffle furnace. The P poisoned catalyst was denoted as D-P. Fresh catalysts and D-P sample were impregnated into aqueous KNO 3 (0.13 mol/L) or Pb(NO 3 ) 2 (0.25 mol/L) solution for 6 h at room temperature. Then the samples were dried at 105 • C for 12 h followed by calcination at 500 • C for 4 h in a muffle furnace. The samples were denoted as D-K, D-Pb, D-P-K and D-P-Pb. The element loading amount detected by SEM-EDS is summarized in Table 1.

Activity Test
The catalytic activities of the samples were evaluated in a simulated fixed-bed reactor made of quartz tube with the dimensions of Φ 6.2 × 580 mm in the laboratory. The simulated flue gas consisted of 500 ppm NO, 500 ppm NH 3 , 5% O 2 and N 2 balance. In each experimental run, the catalysts loaded in the reactor were 1.2 cm 3 and the gas hourly space velocity (GHSV) was kept at 60,000 h −1 . For this system, the concentrations of NO, NO 2 and NH 3 were measured by an Antaris IGS analyzer (Thermo Fisher Company, Waltham, M.A., USA) and the concentration of O 2 was measured by a flue gas analyzer (T-350, Testo Company, Hampshire, United Kingdom).
The conversion of NO x was defined as: where C NO x ,in and C NO x ,out are the NO x concentration (NO + NO 2 ) of the simulated gas stream in the inlet and outlet of the reactor, C NH 3 ,in and C NH 3 ,out are the inlet and outlet concentrations of NH 3 , and C N 2 O,out is the outlet concentration of N 2 O.
The temperature-programmed desorption of ammonia (NH 3 -TPD) on samples was carried out on a chemisorption analyzer (ChemBET-3000TPR-TPD, Quantachrome, Boynton Beach, F.L., USA)-mass spectrum (DYCOR LC-D200, Ametek, Kent, O.H., USA). A total of 100 mg of the sample was treated in helium at 500 • C for 1 h and cooled to room temperature. Then the sample was exposed to 5% NH 3 for 40 min and subsequently purged with helium at 100 • C for 30 min to remove weakly-adsorbed NH 3 from the surface. Finally, the sample was heated to 600 • C at a rate of 10 • C/min in He and the signal of NH 3 was recorded by mass spectrum.
In situ diffuse reflectance infrared Fourier-transform spectra (DRIFTS) were collected using a Bruker Vertex 70 infrared spectrometer (Billerica, M.A., USA) with KBr optics and a mercury-cadmium-telluride (MCT) detector. The sample was treated at 500 • C for 1 h and then cooled to the desired temperature in N 2 atmosphere. The spectra were collected with 500 ppm NH 3 in N 2 at the set time. All the spectra were collected at a resolution of 4 cm −1 by accumulating 64 scans. The spectra obtained in this study were transformed into absorption spectra by using Kubelka-Munk function.
X-ray photoelectron spectra (XPS) were measured with a Thermo ESCALAB 250 (Waltham, M.A., USA) using monochromated Al Kα X-rays (h = 1486.6 eV) as a radiation source at 150 W. Sample charging effects were eliminated by correcting the observed spectra with the C 1s binding energy (BE) value of 284.6 eV. The analysis of XPS spectra of samples was conducted by "Avantage" software supplied by Thermo Fisher Scientific Corporation, and uses "Scofield factor" as sensitive factor.

Conclusions
In this study, P, K and Pb decrease the DeNO x efficiency of commercial V 2 O 5 -WO 3 /TiO 2 catalysts to different degrees. Furthermore, the doping of P, K and Pb does not change crystalline phase and textural structure of commercial catalysts. However, the treatment with (NH 4 ) 2 HPO 4 improves the NO x conversion of Pb-and K-poisoned catalysts. That may be because the samples pretreated with (NH 4 ) 2 HPO 4 have higher content of acid sites with higher thermal stability. Moreover, they have increased V 5+ /V 4+ ratio and active oxygen species, including V-OH and P-OH, on the surface of catalysts. It is believed that the treatment with (NH 4 ) 2 HPO 4 can decrease the poisoning effects of K and Pb on commercial V 2 O 5 -WO 3 /TiO 2 catalysts. This work may offer the reference to further research in the complex poisoning effects of multiple elements on commercial V 2 O 5 -WO 3 /TiO 2 catalysts.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4344/10/3/345/s1, Figure S1: (a) The N 2 selectivity of fresh and poisoned samples; (b) the N 2 selectivity of fresh and poisoned samples at 350 • C with time, Figure S2: The SEM-EDS images of fresh and poisoned catalyst, Figure S3: N 2 adsorption-desorption isotherm of fresh and poisoned samples, Figure S4: The pore size distribution of fresh and poisoned samples, Figure S5: The SEM images of fresh catalysts and poisoned samples, Figure S6: XPS spectra of the W 4f of fresh and poisoned samples.