The effects of Bi2O3 on the selective catalytic reduction of NO by propylene over Co3O4 nanoplates

Bi2O3/Co3O4 catalysts prepared by the impregnation method were investigated for the selective catalytic reduction of NO by C3H6 (C3H6-SCR) in the presence of O2. Their physicochemical properties were analyzed with SEM, XRD, H2-TPR, XPS, PL and IR measurements. It was found that the deposition of Bi2O3 on Co3O4 nanoplates enhanced the catalytic activity, especially at low reaction temperature. The SO2 tolerance of Co3O4 in C3H6-SCR activity was also improved with the addition of Bi2O3. Among all catalysts tested, 10.0 wt% Bi2O3/Co3O4 achieved a 90% NO conversion at 200 °C with the total flow rate of 200 mL min−1 (GHSV 30 000 h−1). No loss in its C3H6-SCR activity was observed at different temperatures after the addition of 100 ppm of SO2 to the reaction mixture. These enhanced catalytic behaviors may be associated with the improved oxidizing characteristics of 10.0 wt% Bi2O3/Co3O2. XRD results showed that Bi2O3 entered the lattice of Co3O4, resulting in the formation of lattice distortion and structural defects. H2-TPR results showed that the reduction of Co3O4 was promoted and the diffusion of oxygen was accelerated with the addition of Bi2O3. XPS measurements implied that more Co3+ formed on the 10.0% Bi2O3/Co3O2 catalysts. The improved oxidizing characteristics of the catalyst with the addition of Bi2O3 due to the synergistic effect of the nanostructure hybrid, thus enhanced the C3H6-SCR reaction and hindered the oxidization of SO2. Therefore, the 10.0% Bi2O3/Co3O4 catalyst exhibited the highest NO conversion and strongest SO2 tolerance ability.


Introduction
Lean burn engines, which are generally used in gasoline and diesel powered vehicles, are more fuel-efficient than the stoichiometric gasoline engines. 1 They also effectively reduce unburned hydrocarbons, CO 2 and CO in exhausts. 2 However, lean burn engines operate with a large excess of air, leading to a signicant concentration of oxygen in the exhausts, where the noble-metal three-way catalysts cannot work well to reduce nitrogen oxides (NO x ). 3 A large amount of NO x produced by lean burn engines leads to serious air pollution and public health problems.
In order to control NO x emission under the lean burn conditions, selective catalytic reduction of NO by hydrocarbons (e.g. propylene) has been undertaken and reported in the literatures as one potential application (HC-SCR). Many classes of catalysts, including supported noble metals (e.g. Pt, 4,5 Au 6,7 ), metal oxides (e.g. Ag 2 O, 8,9 CuO, 10,11 SnO 2 , 12-14 CoO x (ref. [15][16][17][18]) and zeolite types (ZSM-5, 19 MCM-41 (ref. 20)) have been investigated. In general, the noble metals are active and stable even at lower temperature, but the formation of N 2 O is undesirable by using such precious metals, particularly, platinum-based catalysts. The zeolite-based catalysts were low thermal stability. Among metal oxides catalysts, cobalt oxides (e.g. Co 3 O 4 ) are considered as one promising catalyst for HC-SCR due to its high catalytic activity. 21 When combined with other oxides, such as CeO 2 , 16 ZrO 2 , 15 Al 2 O 3 , 18 sulphated ZrO 2 , 22 the catalytic performances of the cobalt oxide catalyst could be improved as reported. These results implied that the chemical environment around cobalt oxide plays a crucial role in controlling the overall activity of cobalt containing catalysts in SCR reactions. Bi 2 O 3 , a common oxide semiconductor, is widely used in the elds of chemical engineering and electronics such as NO detection 23 and the oxidation or ammoxidation of propylene. 24,25 In the oxidation/ammoxidation of propene over bismuth/ molybdate catalyst, bismuth was thought to involve the ratedetermining hydrogen abstraction from propylene, 26 exhibiting its mild oxidizing characteristics. This property might be also benecial for the partial oxidation of propylene in C 3 H 6 -SCR for NO reduction. Therefore, it is of considerable interest to explore the application of Bi 2 O 3 in the reduction of NO with propylene.
In the present study, Co 3 O 4 nanoplates and Bi 2 O 3 /Co 3 O 4 were prepared with the solvothermal and impregnation method respectively. Their catalytic performances in the NO reduction by C 3 H 6 in the presence of O 2 were investigated. The catalysts were characterized with X-ray diffraction (XRD), temperature programmed reduction with hydrogen (H 2 -TPR), and X-ray photoelectron spectra (XPS). The effects of Bi 2 O 3 on the selective catalytic reduction of NO by propylene over Co 3 O 4 nanoplates were expected to be elucidated. The Co 3 O 4 support was synthesized via the solvent-thermal method, 50 mmol CoCl 2 solution (200 mmol L À1 , Sinopharm Chemical Reagent Co. China) and 25 mL of NaOH (2 mol L À1 , Sinopharm Chemical Reagent Co. China) were added to a roundbottom ask, ultrasonicated for about 20 minutes to obtain a light brown uniform suspension. And then the suspension was transferred into a stainless steel autoclave with Teon liner. The autoclave was sealed and maintained at 120 C for 12 h. The obtained product was collected aer washing with deionized water for several times, nally calcined at 500 C for 5 h in air (S1

Catalytic activity tests
C 3 H 6 -SCR over the catalysts was carried out at atmospheric pressure in a xed-bed quartz reactor (diameter ¼ 10 mm). 0.1 g catalyst was used in each run with a reaction mixture composed of 200 ppm NO, 200 ppm C 3 H 6 , 100 ppm SO 2 (when needed) and 10 vol% O 2 in balance gas N 2 . The total ow rate was 200 mL min À1 , corresponding to a GHSV 30 000 h À1 . Reaction temperature ranges from 100 to 500 C. The concentration of NO was continuously measured by a NO analyzer (Thermo Environmental Instruments Inc., model 42c), which monitors NO, NO 2 , and NO x (NO x represents NO + NO 2 ). The removal efficiency of NO was calculated as NO removal (%) ¼ (1 À C/C 0 ) Â 100%, where C and C 0 are concentrations of NO in the outlet and inlet, respectively.

Catalysts characterization
Scanning electron microscopy (SEM) images were taken on a Hitachi S4800 scanning electron microscope operating at 5.0 kV. X-ray powder diffraction (XRD) was carried out on Brukeraxs D8 Discover (Cu Ka ¼ 1.5406Å). The scanning rate is 1 min À1 in the 2q range from 20 to 80 degree. The reducibility of Bi 2 O 3 /Co 3 O 4 catalysts was estimated by temperature programmed reduction with hydrogen analysis (H 2 -TPR). The experiments were carried out with a Micromeritics 2910 apparatus using H 2 /Ar (3/97, v/v) gas with a total ow rate of 15 mL min À1 . In each run, 0.030 g of the catalyst was previously activated at 500 C for 30 min under air, and then cooled to RT. TPR started with the introduction of the mixture of H 2 and Ar. The catalyst was heated from room temperature (RT) to 1000 C (10 C min À1 ). H 2 consumption was continuously monitored with the thermal conductivity detector. X-ray photoelectron spectra (XPS) of the catalysts were measured in a VG Multilab 2000 spectrometer by using Al Ka (1486.6 eV) radiation as the X-ray source. Photoluminescence (PL) measurement was carried out on a Shimadzu RF-5301 PC uorescence spectrophotometer. Raman spectra were recorded using a Horiba Jobin-Yvon Lab Ram HR800 Raman microspectrometer, with an excitation laser at 514 nm.

Scanning electron microscope (SEM) observation and XRD analysis
Co 3 O 4 nanoplates with different dimensions were observed on the support Co 3 O 4 (S1), shown in Fig. 1(a). Fig. 1 Fig. 3. For the reduction of Bi 2 O 3 , a sharp peak was observed at ca. 490 C, implying the reduction of Bi 3+ in a narrow temperature range. A broad reduction peak from 330 to 460 C with a large shoulder at the lower reduction temperature (ca. 380 C) appeared on Co 3 O 4 support (S1). Many researchers reported that the reduction of Co 3 O 4 was a two-step reduction process involving the intermediate reduction of CoO. 16,21,27 Two main clear reduction peaks respectively located around 186 C and 310-480 C were shown in the TPR spectra. The low temperature TPR peak was associated with the reduction of Co 3+ to Co 2+ , and the peak at high temperature was the subsequent reduction of CoO to metallic cobalt. In the TPR spectrum of Co 3 O 4 synthesized in the present work, there are no obvious two peaks probably due to an abroad particle size distribution as shown in SEM observation. The large shoulder at 376 C (peak I) in Fig. 3 should be attributed to the reduction of Co 3+ to Co 2+ , and main reduction peak (peak II) is ascribed to the reduction of CoO to metallic cobalt.
The reduction process of Bi 2 O 3 /Co 3 O 4 catalysts (S2-S4) became complicated with the introduction of Bi 2 O 3 . The reduction peak at 487 C became wider and shied towards to the lower temperature, especially on the Bi 2 O 3 /Co 3 O 4 catalyst with the highest Bi loading amount (S4). In addition, two new peaks around 342 C (peak I) and 402 C (peak II) respectively ascribed to the reduction of Co 3+ to Co 2+ , Co 2+ to metallic Co appeared on Bi 2 O 3 /Co 3 O 4 catalysts (S2-S4). 28 Compared with the bulk Co 3 O 4 (S1), both the reduction peak of Co 3+ and that of Co 2+ shied to lower temperature aer the deposition of Bi 2 O 3 , implying the promoted reduction of Co 3 O 4 . Moreover, the larger reduction peak I than peak II on S2-S4 samples indicated that the ratio of Co 3+ /Co 2+ was higher on S2-S4 samples than that on the Co 3 O 4 (S1). It revealed that the deposition of Bi 2 O 3 on Co 3 O 4 affected the oxidized state of cobalt in the synthesized Co 3 O 4 , more Co 3+ were present on the supported samples (S2-S4) than the pure Co 3 O 4 .

X-ray photoelectron spectroscopy (XPS)
XPS measurements were carried out on Co 3 Fig. 4. In the Co 2p ( Fig. 7(a)), the main peaks located at 779.6-781.3 eV and 794.8-796.5 eV are ascribed to Co 2p 1/2 and Co 2p 3/2 spin-orbital peaks, respectively. 29 Based on the restriction that Co 2p 3/2 binding energies of Co 2+ and Co 3+ components are 781.3 eV and 779.6 eV, respectively, the spin-orbit doublet splitting is 15.2 eV with a xed ratio of 2/1 for the 2p 3/2 -to-2p 1/2 peak area, 29 the Co 2p spectra of S1 and S3 can be tted to the Co 2+ (peak II and IV) and Co 3+ (peak I and III) components. 30,31 The satellite peak of Co 2+ and that of Co 3+ in Co 3 O 4 were also respectively observed at 787.0 eV and 791.0 eV. 32,33 Fig. 4 shows that the peak areas of peak I and III increased with the addition of Bi 2 O 3 , implying that the surface Co 3+ and the surface content ratio of Co 3+ /Co 2+ increased with the addition of Bi 2 O 3 . More Co 3+ was present on 10.0% Bi 2 O 3 /Co 3 O 4 (S3) than the bulk Co 3 O 4 support (S1), as consistent with the TPR results.
The O 1s XPS spectra of Co 3 O 4 and Bi 2 O 3 /Co 3 O 4 catalysts are shown in Fig. 4(b). For the Co 3 O 4 sample, there are two peaks (I and II). The peak I located at $530.1 eV is attributed to the surface lattice of Co 3 O 4 , and the peak II at $531.5 eV is associated with OH À groups. 34 In the case of 10.0% Bi 2 O 3 /Co 3 O 4 , besides the peak I and II, one new peak at $532.8 eV appeared, which should be related to the contribution of the oxygen from Bi 2 O 3 .

Photoluminescence (PL) and Raman spectra
PL emission spectra originating from the recombination of free charge carriers are useful to reveal the migration, transfer and separation of photogenerated charge carriers. Fig. 5 shows photoluminescence emission spectra of different catalysts at room temperature. All samples show one luminescence peak center at about 358 nm, which can be attributed to the radiative recombination of charge carriers. The pure Co 3 O 4 has the strongest PL emission peak. This charge recombination process of Co 3 O 4 can be greatly inhibited by the deposition of Bi 2 O 3 on    support (S1) rstly increased with reaction temperature, reached the maximum conversion (ca. 60%) at ca. 300 C and then decreased at higher temperature. The NO conversion was further increased with the addition of Bi 2 O 3 into Co 3 O 4 with the activity order: Co 3 O 4 (S1) < 5.0% Bi 2 O 3 /Co 3 O 4 (S2) < 15.0% Bi 2 O 3 /Co 3 O 4 (S4) < 10.0% Bi 2 O 3 /Co 3 O 4 (S3). Among all catalysts tested, 10.0% Bi 2 O 3 /Co 3 O 4 (S3) possessed the highest activity for NO conversion in the reaction temperature window, reaching ca. 90% NO conversion at 200 C. NO conversion under the lower reaction temperature (100-250 C) over S3 also reached the highest among the catalysts tested. In contrast, the mixture of 10% Bi 2 O 3 nanoparticles and Co 3 O 4 support (S5) showed lower activity than S1 and S3. It was indicated that the interaction between Bi 2 O 3 and Co 3 O 4 in S3 is not the simply physical mixture like S5. The chemical interaction between them took place in S3 and should contribute the admirable catalytic performance of S3 in the C 3 H 6 -SCR reaction. SO 2 usually exists in the diesel engine exhaust. So it is necessary to investigate the SO 2 tolerance of the catalyst in C 3 H 6 -SCR. Fig. 8 exhibited the effects of 100 ppm SO 2 co-fed in the reaction gas on the NO conversions over the catalysts at the different reaction temperatures. NO conversion over the S3   (10% Bi 2 O 3 /Co 3 O 4 ) catalyst clearly did not change in the wide reaction window. The steady-state NO conversion reached 90.3% on S3 at 250 C in the presence and absence of SO 2 . In contrast, NO conversion decreased from 65.8% to 35.7% at 300 C on the Co 3 O 4 support (S1) when 100 ppm SO 2 was contained in the feed gas. NO conversions at other reaction temperatures also reduced in the presence of SO 2 . These results obviously suggested that 10% Bi 2 O 3 /Co 3 O 4 exhibited good resistibility against SO 2 that coexists with NO and C 3 H 6 in the reaction mixture. Fig. 9 shows the SO 2 durability of 10% Bi 2 O 3 /Co 3 O 4 catalyst with the reaction time at the optimum reaction temperature 200 C in the C 3 H 6 -SCR of NO. When NO conversion reached to the maximum (89.3%), 100 ppm SO 2 was added in the reaction system, NO conversion immediately decreased. It was probably due to the competitive adsorption of NO and SO 2 on the active site. 20 min later, NO conversion reduced to 63.6%. Aer that, NO conversion recovered to 85.7%, and maintained at ca. 88% through the whole reaction period of 90 min. This result further illustrates the outstanding SO 2 resistibility of 10% Bi 2 O 3 /Co 3 O 4 in the long time-reaction.

Discussions
In this study, the deposition of Bi 2 O 3 with the proper loading amount on Co 3 O 4 nanoplates enhanced NO conversion over Co 3 O 4 , especially at low reaction temperature (<200 C). 10.0% Bi 2 O 3 /Co 3 O 4 catalyst also showed the strong resistibility against SO 2 in the feed gas. XRD results showed Bi 2 O 3 could enter the lattice of Co 3 O 4 , and promote the formation of the lattice distortion and structural defect as demonstrated by PL spectra and IR spectra. The H 2 -TPR and XPS results showed that more Co 3+ appeared with the deposition of Bi 2 O 3 . These changes were probably related to the promotive effects of Bi 2 O 3 in the C 3 H 6 -SCR reaction.
Scheme 1 illustrated the mechanistic investigations for the HC-SCR reactions in the previous literatures. 17,36,37 According to these ndings, the reactants (C 3 H 6 , NO and NO 2 ) are supposed to be rst adsorbed on the active sites over the catalyst surface. Subsequently, the adsorbed nitrates formed via NO oxidation by O 2 . C 3 H 6 was also activated to form C x H y O z species such as formate, acetate and so on. As these C x H y O z species become available, nitrates subsequently reacted with them to yield nitrogen-containing organic species, such as NCO species. The nal step would be the interaction of NOC intermediates with NO x (NO, NO x ), decomposing into N 2 , CO x and H 2 O as nal products. This proposed reaction process reveals the crucial role of O 2 in the feed gas and the importance of oxidizing characteristics of the catalyst surface.
In our work, PL and IR results showed that more oxygen vacancies were produced on the Co 3 O 4 aer the doping of Bi 2 O 3 . The vacancy could accelerate the adsorption, activation and diffusion of oxygen, which was suggested to be available for the oxidation reactions involved in HC-SCR. The doped Bi 2 O 3 also increased the Co 3+ concentration on the surface. The richness of Co 3+ could promote the adsorption and activation of NO and (or) C 3 H 6 . What is more is the mild oxidizing characteristics of   This journal is © The Royal Society of Chemistry 2019 bismuth oxide in the selective oxidation and ammoxidation of propene to acrolein and acrylonitrile. It will accelerate the formation of C x H y O z species. In short, the addition of Bi 2 O 3 into the Co 3 O 4 in the present study probably inuenced the oxidation process in the C 3 H 6 -SCR reaction, favored the activation of C 3 H 6 and NO, and then enhanced the following NCO intermediate formation and its decomposition with reaction with NO x to N 2 .
About poisoning HC-SCR catalyst with SO 2 , the previous studies reported that the suppression effect of SO 2 on the SCR catalyst could attributed to the formation of sulphate on the catalyst. 38

Conflicts of interest
There are no conict to declare.