One-pot hydrothermal synthesis of CuBi co-doped mesoporous zeolite Beta for the removal of NOx by selective catalytic reduction with ammonia

A series of CuBi co-doped mesoporous zeolite Beta (CuxBiy-mBeta) were prepared by a facile one-pot hydrothermal treatment approach and were characterized by XRD, N2 adsorption-desorption, TEM/SEM, XPS, H2-TPR, NH3-TPD and in situ DRIFTS. The catalysts CuxBiy-mBeta were applied to the removal of NOx by selective catalytic reduction with ammonia (NH3-SCR), especially the optimized Cu1Bi1-mBeta achieved the high efficiency for the removal of NOx and N2 selectivity, superior water and sulfur resistance as well as good durability. The excellent catalytic performance could be attributed to the acid sites of the support and the synergistic effect between copper and bismuth species. Moreover, in situ DRIFTS results showed that amides NH2 and NH4+ generated from NH3 adsorption could be responsible for the high selective catalytic reduction of NOx to N2. In addition, a possible catalytic reaction mechanism on Cu1Bi1-mBeta for the removal of NOx by NH3-SCR was proposed for explaining this catalytic process.

Scientific RepoRts | 6:30132 | DOI: 10.1038/srep30132 conventional zeolites owing to the crystalline framework and the hierarchically porous structure 7 . Thereinto, the mesoporous Beta zeolite (mBeta) with unique three-dimensional network of large pores (12MR) exhibits much high surface area and excellent hydrothermal stability, which is widely applied in fine chemistry 8 . In the past decades, a number of synthetic approaches of mBeta have been explored and some encouraging results have been obtained [8][9][10] . Such as, Xiao et al., synthesized highly mesoporous single-crystalline zeolite beta by using a commercial polymer, polydiallyldimethylammonium chloride (PDADMA) as both structure-directing agent and porogen, which showed better hydrothermal stability and higher catalytic activity than conventional zeolite Beta in large molecules involved acid-catalyzed reactions 8 . In addition, many metal oxides were reported to have good performance in the removal of NO x by NH 3 -SCR, e.g., Cu-loaded zeolite beta exhibited a good activity and hydrothermal stability in the NH 3 -SCR of NO x 11,12 . It is also reported that the introduction of Bi 2 O 3 can improve the SO 2 resistance 13 .
On the basis of our previous work 14 , a novel CuBi co-doped mesoporous zeolite Beta (Cu x Bi y -mBeta) has been synthesized by one-pot hydrothermal treatment approach, by which the copper and bismuth species can be well dispersed into the framework of mesoporous zeolite Beta. Therein, the optimized prepared catalyst Cu 1 Bi 1 -mBeta exhibits very high catalytic activity for the selective catalytic reduction of NO x with NH 3 . In addition, the N 2 selectivity, water vapor and sulfur resistance and durability of the Cu 1 Bi 1 -mBeta catalyst have been detailedly investigated. Finally, a possible catalytic mechanism of SCR of NO x with ammonia on this prepared catalyst CuBi-mBeta is proposed to clarify the catalytic process.

Results
Structure Characteristics. The powder XRD patterns of mBeta, Cu-mBeta, Bi-mBeta and Cu x Bi y -mBeta are shown in Fig. 1. It is found that all prepared samples keep the diffraction peak of typical zeolite beta structure, and no diffraction peaks corresponding to copper and bismuth species can be detected. Therefore, it is believed that the copper and bismuth species could be well-incorporated into the framework of zeolite as ions or highly dispersed into the mesoporous channels as metal oxides. It is noted that compared with the mBeta, the doping of Bi species can cause the inevitable destruction of zeolite framework to a certain extent, and the intensity of XRD peak of Cu x Bi y -mBeta decreases with the increase of Bi-loading content, as shown in Fig. 1.
The N 2 adsorption isotherms and pore size distribution curves for samples mBeta, Cu-mBeta, Bi-mBeta and Cu x Bi y -mBeta are shown in Fig. 2 and the corresponding pore structure parameters of all the samples are summarized in Table 1. All the samples exhibit typical type IV isotherms, confirming the presence of mesoporous structure. The reference sample mBeta shows well-defined mesopore of 3.8 nm, and the BET surface area and total pore volume are calculated to be 556 m 2 /g and 0.34 cm 3 /g, respectively, including the mesoporous surface area (166 m 2 /g) and mesoporous volume (0.16 cm 3 /g), respectively. Compared with the mBeta, the surface areas of Cu-mBeta, Bi-mBeta and Cu x Bi y -mBeta show obvious decrease after loading amount of copper and bismuth species, which is resulted from the generation of non-framework Cu or/and Bi ions (i.e., CuO and Bi 2 O 3 ) with the increase of Cu or Bi content, and inducing the collapse of pore structure of zeolite to some extent. Even so, the prepared Cu 1 Bi 1 -mBeta still keep high BET surface area and pore volume after doping with Cu and Bi species (Table 1), indicative of the unblocked mesoporous channels due to the high dispersity of Cu and Bi species. It is noted that the optimized Cu 1 Bi 1 -mBeta with the mesopore size of 3.6 nm shows high BET surface area (539 m 2 /g) and total pore volume (0.46 cm 3 /g).
The SEM image of as-prepared Cu 1 Bi 1 -mBeta, as shown in Fig. 3a, presents a rough surface morphology, demonstrating that the mesoporous structure has penetrated into the zeolite crystals by one-pot hydrothermal synthesis process. Additionally, no oxide aggregations can be found, as shown in Fig. 3a,b, indicating that the doped metal oxides are highly dispersed into the carrier mBeta. The clearly crystal lattices can be found in the high-magnification TEM image (Fig. 3c), confirming the zeolite crystallized structure. The element mapping in Fig. 3d-h further confirms that Cu and Bi species are highly dispersed into mesoporous zeolite Beta, which is consistent with the above results of XRD patterns. Spectroscopy Characteristics. The X-ray transmission spectroscopy (XPS) result of the Cu 1 Bi 1 -mBeta is show in Fig. 4a. The binding energy levels of Cu 2p 3/2 and Cu 2p 1/2 at around 934.5 eV and 954.2 eV, respectively (denoted with ★ ), accompanied by a shoulder peak of about 10 eV higher binding energy, are attributable to Cu(II). Additionally, the distinctive peaks at 936.6 eV and 956.4 eV, (denoted with ▼ ) can be ascribed  S meso was given by the difference between S total and S micro . c V meso was given by the difference between V total and V micro . to Cu(I), indicating that Cu species have variable valencies in the obtained sample Cu 1 Bi 1 -mBeta. In addition, the prepared Cu 1 Bi 1 -mBeta presents higher binding energy peak intensities of Cu(II) (934.5 eV) than that of Cu(I) (936.6 eV), as shown in Fig. 4a, suggesting that Cu 1 Bi 1 -mBeta contains much more amounts of Cu(II) than Cu(I) 15,16 . Figure 4b shows the Bi 4f XPS spectrum of Cu 1 Bi 1 -mBeta catalyst. Compared to pure Bi 2 O 3 at 164.2 eV and 158.9 eV, the Cu 1 Bi 1 -mBeta shows the Bi 4f 5/2 and Bi 4f 7/2 a little higher binding energies at 164.6 eV and 159.5 eV, respectively, which is attributed to the interaction between Bi and Cu or the support mBeta [17][18][19] .
The H 2 -TPR profiles of Cu-mBeta, Bi-mBeta and a series of Cu x Bi y -mbeta samples are shown in Fig. 5. It is found that the reference single loaded sample Cu-mBeta shows two weak reduction peaks at around 220 °C and 290 °C, indicating a two-step reduction of Cu 2+ ions, i.e., firstly to Cu + and then to Cu 20,21 . In addition, the reduction peak at 310 °C of reference Bi-mBeta can be ascribed to the reduction of Bi 2 O 3 . The H 2 -TPR profiles of co-loaded samples Cu x Bi y -mBeta show a distinctively different redox behavior from either Cu-mBeta or Bi-mBeta, and present enhanced reduction peak at 250-350 °C, attributed to the strong interaction between the active species CuO and Bi 2 O 3 . In addition, the reduction peak gradually shifts toward lower temperature range (from 350 to 310 °C) with the increase of Bi content ( Fig. 5 and Table 1), confirming that the addition of Bi 2 O 3 is beneficial to promote the catalytic redox reaction 22 . NH 3 -TPD experiments were carried out to obtain the acidity information of the prepared catalysts, as shown in Fig. 6. It is clear that a low-temperature peak at 150 °C and a high-temperature peak at 300 °C can be observed for the sample mBeta, which is assigned to weakly weak Lewis acid sites and strong Brønsted acid sites, respectively 23,24 . It is noted that all the samples Cu-mBeta, Bi-mBeta and Cu x Bi y -mBeta show weaker low-temperature peaks than that of the reference mBeta, owing to the part destruction of zeolite framework structure after introducing Cu and Bi species. However, it is interesting that only the Cu 1 Bi 1 -mBeta shows a similar desorption peak at high-temperature range (300-400 °C) to the mBeta, indicating that the Cu 1 Bi 1 -mBeta sample still keeps the strong acidity site of mBeta. The presence of rich acidic sites (Brønsted acid and Lewis acid) produced from the framework Al atoms and copper/bismuth species are helpful to the adsorption and activation of NH 3 , and thus producing many ammonia species, including NH 2 , coordinated NH 3 and ionic NH 4 + , which can greatly promote the selective catalytic reduction of NO x , as shown in Fig. 7.
The reaction of adsorbed NH 3 species towards NO + O 2 was evaluated by the in situ IR spectra at 250 °C and the results are shown in Fig. 7. When the catalyst are exposed to NH 3 for the 60 min and purged with N 2 , the peaks related to coordinated NH 3    on acidic sites could undergo the oxidative dehydrogenation to form NH 2 species (1560 cm −1 ), then produce intermediate specie NH 2 NO when NO and O 2 were added into reaction gas. It is noted that all the ammonia species, including NH 2 , coordinated NH 3 and ionic NH 4 + bound, disappeared after NO + O 2 purge, indicating that those ammonia species could participate in the reduction of NO x . Meanwhile, when NO and O 2 were added into reaction gas, the bands at 1235, 1367, 1542 and 1601 cm −1 could be detected in IR spectra. Thereinto, the bands at 1235 and 1367 cm −1 were assigned to monodentate nitrate, while the bands at 1542 and 1601 cm −1 are associated with bidentate nitrate and adsorbed NO 2 , respectively 15,25 . More interestingly, two peculiar peaks at 3335 and 3265 cm −1 related to the adsorbed NH 3 on acid sites became stronger with the increase of exposing time in the NO + O 2 , indicating that some acidic sites on the surface of the catalyst were released and then preferably adsorbed the NH 3 after NO and O 2 pass over the catalyst.
Catalytic performance. Figure 8 shows the NH 3 -SCR results of mBeta, Cu-mBeta, Bi-mBeta and a series of Cu x Bi y -mBeta catalysts. The NO x conversions over the prepared catalysts under high hourly space velocity of 64000 h −1 are shown in Fig. 8a. Compared with the references mBeta, Bi-mBeta and Cu-mBeta, the sample Cu x Bi y -mBeta show higher catalytic activity for the SCR of NO x . Especially, the optimized sample Cu 1 Bi 1 -mBeta with the 4.38 wt% Cu and 4.83 wt% Bi exhibits the highest catalytic performance, i.e., the NO x conversion efficient is above 90% within the wide operation temperature window of 170 °C to 400 °C. While, the excess doping of Cu and Bi species could induce the formation of oxides aggregates and thus decrease the active surface area (Table 1), e.g., the excess Bi would cover part of active center and results a lower catalytic activity. Therefore, the optimized sample Cu 1 Bi 1 -mBeta with the 1:1 ratio of Cu and Bi shows the highest catalytic performance. The results of N 2 selectivity of these catalysts were shown in Fig. 8b. For comparison, the reference Bi-mBeta shows a low N 2 selectivity, particular in 200 °C, owing to the part destruction of zeolite framwork with the addition of Bi species, as demonstrated by XRD results, which can affect the strong acidic sites (Fig. 6), thus decrease the adsoption ability of NH 3 on the reference Bi-mBeta. Furthermore, the Cu 1 Bi 1 -mBeta catalyst also presents as high as up to 100% N 2 selectivity in the whole temperature range investigated. It is believed that the high dispersity of active species  Cu and Bi on the support mesoporous zeolite, the richer acidic sites and the strong interaction between the active copper and bismuth species on the Cu 1 Bi 1 -mBeta could be main contributions to the catalytic activity.
It is well known that the real diesel exhaust presents large number of water vapor and trace S compounds, therefore, the effects of H 2 O and SO 2 on the SCR catalytic activity are also investigated on the samples Cu 1 Bi 1 -mBeta. It is evident that compared with the reference mBeta, Cu 1 Bi 1 -mBeta catalyst shows an excellent SO 2 resistance in the NH 3 -SCR above 200 °C (Fig. 8c). Even in the co-presence of H 2 O and SO 2 , the Cu 1 Bi 1 -mBeta catalyst still exhibits high activity of over 85% NO x conversion from 200 °C to 430 °C. The results of durability tested at 250 °C, as shown in Fig. 8d, indicate that the Cu 1 Bi 1 -mBeta is very stable in the presence of SO 2 (or H 2 O and SO 2 ). It is believed that the highly crystalline zeolite framework enables the catalyst to keep good stability against H 2 O, and the highly dispersity active speicies Bi can improve the SO 2 resistance to some extent 13 , which is very important for the NH 3 -SCR of NO x in the co-existence of H 2 O and SO 2 .

Discussion
Based on the above results and discussions, a possible catalytic reaction mechanism for the SCR of NO x was proposed, as illustrated in Fig. 9. Firstly, there are a large amount of oxygen vacancies ( ″ V 0 ) presented in the sample support mBeta due to the doping of hetero atoms Cu n+ , Bi 3+ and Al 3+ in the [SiO 4 ], leading to the generation of numerous surface activated oxygen (O * ) by adsorbing the O 2 , as shown in Step 1 (Fig. 9). Meanwhile, the existence of Bi 2 O 3 was reported 22,28 to be beneficial for the reduction of Cu 2+ to Cu + , and NO could be easily adsorbed and activated by the Cu n+ and generated the + ⁎ NO (Cu ) 2 and + ⁎ NO (Cu ) at higher temperatures or lower temperatures, respectively 29,30 . Afterwards, these activated + ⁎ NO (Cu ) 2 and + ⁎ NO (Cu ) could react with the surface activated oxygen (O * ) and thus produce large numbers of NO 2 , as shown in Step 1 (Fig. 9). When the concentration ratio of NO and NO 2 in reaction gas reaches to 1:1, the quick SCR reaction (NO + NO 2 + 2NH 3 = 2N 2 + 3H 2 O) occurs, during which the NO x conversion efficiency at low temperature can be greatly improved. Secondly, the presence of rich acidic sites (Brønsted acid and Lewis acid) produced from the framework Al atoms and copper/bismuth species is helpful to the adsorption and activation of NH 3 , i.e., the NH 3 adsorbed on strong Brønsted acid sites could be activated and generate NH 4 + , which was discovered from the in situ DRIFTs (Fig. 7), as shown in Step 2 (Fig. 9). Meanwhile, the NH 3 molecules could also be adsorbed on the weak Lewis acid sites to generate NH 3(ads) and react with the activated oxygen (O * ) to produce the amines NH/NH 2 , as confirmed by the in situ DRIFTs (Fig. 7). Finally, the produced amide NH/NH 2 and NH 4 + could directly react with NO x on the highly dispersed active sites and generate N 2 , as shown in Step 3 (Fig. 9). It is believed that the existence of highly dispersed varied valence Cu species and acidic sites can accelerate the adsorption and activation of NO and NH 3 in this reaction system, which greatly increases the NH 3 -SCR in the removal of NO x .
In conclusion, a series of Cu x Bi y -mBeta catalysts have been prepared by a facile one-pot hydrothermal treatment approach. The optimized Cu 1 Bi 1 -mBeta shows an excellent NH 3 -SCR activity and high N 2 selectivity (closely to 100%) toward NO x in a broad operation temperature window (170-400 °C). On the one hand, the mesopores structure is helpful for the homogeneous dispersion of active species, which can improve the diffusion and transport of reactants and products. Also, the highly crystalline zeolite framework enables the catalyst to keep good resistence toward water vapor. On the other hand, the highly dispersity of copper and bismuth active species, as well as the synergetic catalytic effect between copper and bismuth species and the richer acidic sites of the zeolite promote the reduction of CuO and generation of intermediate NO 2 . Meanwhile, the large number of amide NH/NH 2 and NH 4 + generated from NH 3 adsorption, as the key intermediates, greatly accelerate the selective catalytic reduction of NO x . More interesting, the prepared catalyst shows good durability and high resistance against H 2 O and SO 2 , which could also be ascribed to the crystalline zeolite framework and the highly dispersity active sites. The CuBi-mBeta catalyst with distinctive mirco-mesoporous structure demonstrates excellent NH 3 -SCR activity and high stability, which, as we believe, will present promising prospect in the practical application of catalytic purification of diesel exhausts.

Methods
Preparation of the mesoporous zeolite Beta (mBeta). Typically, 0.05 g NaCl and 0.15 g KCl were added into 2 mL distilled water and 14.4 g TEAOH solution. Afterwards, 3.9 g H 2 SiO 3 was dissolved into the above mentioned solution and stirred at 313 K for 6 h. Next, the solution containing 0.033 g NaOH, 0.17 g NaAlO 2 and 2 mL distilled water was slowly added into the resultant solution and further stirred at 313 K for 6 h. Finally, 0.5 g CTAB solution was added into the obtained solution and further stirred at 353 K for 8 h. The obtained mixed solution was hydrothermally treated for 48 h at 423 K. Subsequently, the products were washed with distilled water and dried at 383 K for 12 h. The final product mBeta was obtained after calcinations at 823 K for 6 h to remove any organics.
Preparation of Cu 1 Bi 1 -mBeta. Typically, 0.05 g NaCl and 0.15 g KCl were added into 2 mL distilled water and 14.4 g TEAOH solution. Afterwards, 3.9 g H 2 SiO 3 was dissolved into the above mentioned solution and stirred at 313 K for 6 h. Then, the solution containing 0.033 g NaOH, 0.17 g NaAlO 2 and 2 mL distilled water was slowly added into the resultant solution and stirring. Next, 1 mmol Cu(NO 3 ) 2 • 3H 2 O, 1 mmol Bi(NO 3 ) 3 • 5H 2 O and 2 mL distilled water was slowly added into the above precursor solution and further stirred at 313 K for 6 h. Then, 0.5 g CTAB solution was added into the obtained solution and further stirred at 353 K for 8 h. Next, the obtained mixed solution was hydrothermally treated for 48 h at 423 K. Subsequently, the products were washed with distilled water and dried at 383 K for 12 h. The final product, Cu 1 Bi 1 -mBeta, was obtained after calcinations at 823 K for 6 h to remove any organics.
For comparison, the Cu-mBeta, Bi-mBeta and Cu x Bi y -mBeta were synthesized by a facile one-pot hydrothermal treatment approach, similarly to the above process of Cu 1 Bi 1 -mBeta. Herein, the x and y represent the millimole amount of Cu and Bi in the initial precursor solution, respectively. In addition, the final pH value of the synthetic gel including copper and bismuth species is about 11. Sample characterization. The XRD patterns were recorded on a Rigaku D/Max-2200PC X-ray diffractometer using Cu target at 40 kV and 40 mA. The N 2 adsorption and desorption measurements were performed using Micromeritics Tristar 3000 at 77 K. The total surface area and pore volume were calculated using the BET and BJH method. Field emission scanning electron microscopy (SEM) analysis was performed on a JEOL JSM6700F electron microscope. Field emission transmission electron microscopy (TEM) analysis was conducted with a JEOL 200CX electron microscope operated at 200 keV. X-ray photoelectron spectroscopy (XPS) signals were collected on a Thermo Scientific ESCALAB 250 instrument using monochromated Al X-ray resource at 1486.6 eV operated at 15 kW. The temperature-programed reduction with hydrogen (H 2 -TPR) and temperature programmed desorption of ammonia (NH 3 -TPD) were performed on Micromeritics Chemisorb 2750 instrument attached with ChemiSoft TPx software. TPR was carried out from room temperature to 900 °C under 5% H 2 in Ar at a flow rate of 25 mL/min. The H 2 signal was detected by a thermal conductivity detector (TCD). The NH 3 -TPD of the samples was carried out from room temperature to 800 °C at a flow rate of 25 mL/min. The amount of NH 3 desorbed was measured using a thermal conductivity detector (TCD). In situ diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was performd on a Bruker spectrometer equipped with an MCT detector. Prior to each experiment, the sample was pretreated at 400 °C for 1 h in a flow of N 2 and then cooled down to 200 °C. The background spectrum was collected in flowing N 2 and automatically subtracted from the sample spectrum. The reaction conditions were controlled as follows: 100 mL min −1 total flow rate, 500 ppm NH 3 , 500 ppm NO, 5% O 2 and N 2 as the balance. All spectra were recorded by accumulating 100 scans with a resolution of 4 cm −1 . The contents of Cu and Bi were measured by using Inductively Coupled Plasma Atomic Emission Spectroscopy (ICP-AES) analyzer on a Vista AX.
Catalytic activity test. The catalysis measurements were carried out in a fixed-bed quartz reactor using 0.2 g catalyst of 40-60 meshes. Before catalytic test, the catalysts were dried at 423 K for 16 h. The feed gas mixture contained 500 ppm NO, 500 ppm NH 3 , 0 or 3% H 2 O, 0 or 50 ppm SO 2 , 5% O 2 and Ar as the balance gas. The total flow rate of the feed gas was 300 mL/min, corresponding to a GHSV of 64,000 h −1 . The composition of the product gas was analyzed by a chemiluminescence NO/NO 2 analyzer (Thermal Scientific, model 42i-HL) and gas chromatograph (Shimadzu GC 2014 equipped with Porapak Q and Molecular sieve 5A columns). The activity data were collected when the catalytic reaction practically reached steady-state condition at each temperature.
The NO x (X NOx ) and NH 3 (X NH3 ) conversions and N 2 selectivity (S N2 ) were calculated as where NO x includes NO and NO 2 , C i presents the concentration of the "i" species, and the "in" and "out" present the gas concentration of inlet and the exit of the reactor, respectively.