Cerium nitrate which was the precursor of CeO_x will react with ammonium metavanadate precursor or ammonium tungstate precursor to produce cerium vanadate or cerium tungstate precipitation, affecting the uniformity of V₂O₅ and CeO₂ on the surface of TiO₂. Therefore, modified catalysts were prepared by multi-step ultrasonic impregnation method based on the common commercial SCR catalysts (containing 1%V₂O₅, 9%WO₃ and 90%TiO₂). TiO₂ was added into mixed solutions of ammonium metavanadate, ammonium metatungstate and oxalic acid. The slurry was exposed to an ultrasonic bath for 4 h, dried at 105°C for 4 h and calcined at 500°C for 4 h in air to obtain SCR catalyst. Ce(NO₃)₃ was dissolved in deionized water at different concentrations. The powder of SCR catalysts was impregnated in Ce(NO₃)₃ solution for 4 h, dried at 105°C for 4 h and calcined at 500°C for 4 h. Finally, CeO₂ doped catalyst V₂O₅–WO₃/TiO₂ was obtained and abbreviated as Ce_x/SCR. X represents the mass fraction of Ce in the sample, x = 0.5%, 1%, 3%, 5%. 80–100 mesh particles were selected by stainless steel sieve.

The schematic of the fixed-bed reactor system used for Hg⁰ oxidation was showed in Fig. 1. The reactor system consists of four parts: mercury generation, simulated flue gas mixing, temperature control and mercury analysis.The elemental mercury was generated by a PSA Cavkit 10.534 mercury generator. The water vapor was generated by an IAS Hovocal gas generator. The composition and concentration of simulated flue gas were controlled by mass flow controllers (MFCs). The fixed-bed reactor was made of a quartz tube heated in a vertical tubular furnace. The catalyst sample was placed between the quartz cotton and glass fiber filter membrane, which were loaded in the quartz reactor. The concentrations of Hg⁰ and Hg²⁺ in simulated flue gas were online monitored by Thermo CEMS. To avoid Hg contamination, all of the tubes, joints, and valves that Hg passed through were made of Teflon. To avoid condensation of water vapor, all pipelines with H₂O (g)-containing gas passing through were heated up to 120°C. The tail gas was treated by an activated carbon trap before being released to the air.

The inlet Hg⁰ concentration was set at 30 μg/Nm³ and the total flow rate of simulated flue gas was controlled at 2 L/min. The concentration of Hg⁰ and Hg²⁺ at the outlet was recorded after adsorption balance, which was defined as the fluctuation of concentration less than 3%. The time for the balance was more than 2 h. The total Hg⁰ oxidation rate (O_xi) within a certain time was defined as follows:(1) oxi=∑t0CHg2+tCHgTtt

where CHg2+t and CHgTt were the outlet Hg²⁺ and total mercury concentration at the reaction time, respectively.

XRD measurements were carried out on polycrystalline X-ray diffractometer (Smartlab, Rigaku, Japan) to examine the crystallinity and dispersivity of crystal pieces using Cu Ka radiation (9 KW) in the range of 3–90°. The specific surface area, pore volume and pore diameter of catalysts were tested on automatic nitrogen adsorption analyzer (Micrometritics ASAP2010). The surface area and pore size was calculated by Brunauer−Emmett−Teller (BET) method and Barrett−Joyner−Halenda (BJH) method, respectively. To investigate the elemental states, XPS was carried out by X-ray photoelectron spectrometer (PHI QuANTRO SXM, ULVAC-PHI, Japan). The observed spectra were calibrated with the C1s binding energy (BE) value of 284.6 eV.

Fig. 2 showed the Hg⁰ catalytic oxidation efficiencies over various catalysts with a space velocity of 60000 h⁻¹ in the temperature range of 150–400°C. The simulated flue gas consisted of 10 ppm HCl, 12% CO₂, 5% H₂, 300 ppm NO, 400 ppm SO₂ and N₂ as balance gas. The results indicated that Ce doping did not change the optimum Hg⁰ catalytic oxidation temperature over SCR catalyst, which was still 350°C. It was clearly found that the addition of CeO₂ resulted in enhancement of Hg⁰ oxidation activity in varying degrees. With the increase of Ce doping, the Hg⁰ oxidation efficiency over Ce_x/SCR catalyst first increased and then decreased at 350°C. Ce₁/SCR performed the best mercury oxidation and approximately 84.24% mercury oxidation efficiency was obtained. It was worth noting that the Hg⁰ oxidation rate of Ce₅/SCR decreased remarkably, even less than the Hg⁰ oxidation rate of SCR when the temperature exceeded 350°C. It was presumed that CeO₂–TiO₂ performed optimal Hg⁰ catalytic oxidation at about 150°C [15], and excessive Ce doping lead to poor Hg⁰ oxidation over Ce₅/SCR at high temperature.

Catalytic activity evaluation was carried out using a flow-through powder reactor system referred in Pang et al. [16]. NO concentration was tested for 30 min at each steady state. The NO conversion was calculated using the equation below:(2) Noxi=ΔNO¯NOin=NOin−NOout¯NOin

where NO_in represents the inlet NO concentration maintained at 500 ppm and NO_out represents the outlet average NO concentration.

The NO conversion over SCR and Ce₁/SCR catalyst at various temperature were showed in Fig. 6. The NO conversion over SCR catalyst was noticeably enhanced by Ce doping. SCR catalyst performed best NO conversion at 400°C while Ce₁/SCR catalyst performed best at 350°C. It was speculated that Ce had a good effect on catalytic oxidation at low temperature, which decreased the optimal denitrification temperature. At operating temperature (300–400°C) of catalyst in power plant, NO conversion over Ce₁/SCR catalyst was greater than 90%, which had a small difference from the optimal denitrification efficiency. Therefore, Ce₁/SCR can be well applied in power plant, which will have a better simultaneous removal performance of NO and Hg⁰ than SCR.

The XRD patterns of the raw and modified SCR catalysts were displayed in Fig. 7. TiO₂ phase other than rutile TiO₂ phase was detected in all samples. The diffraction line of all samples were narrow and sharp, which indicated the high crystallinity. The characteristic peaks of V₂O₅ and WO₃ were hardly detected, which were due to widely dispersion and poorer crystalline on the surface. CeO₂ was not observed when the loading of Ce was lower than 3% (wt.%). There were weak characteristic CeO₂ peaks at 28.38° when the loading of Ce was 5%, indicating a well-dispersed cluster with a small population of CeO₂ may exist.

The pore structure parameters of different samples were summarized in Table 1. As can be seen from the result, the BET surface area, BJH pore volume and average pore volume of Ce_x/SCR catalyst were lower than those of SCR. The BET surface area of Ce_x/SCR decreased with the increase of Ce doping amount. The pore volume and pore diameter of Ce_x/SCR catalyst did not show obvious relationship with Ce doping amount. The Nitrogen adsorption-desorption isotherms of Ce_x/SCR were the IV adsorption-desorption isotherm specified by IUPAC. H1 hysteresis loop indicated mesoporous materials, which might facilitate mass transfer in the catalytic reaction (Fig. 8a). As displayed in Fig. 8b, Ce doping had no significant effect on the pore size distribution of catalysts. There were much mesoporous of 2∼50 nm in Ce_x/SCR catalysts, indicating strong interactions between catalyst surface and adsorbate.

The effects of HCl and O₂ on mercury oxidation over Ce₁/SCR catalyst were shown in Fig. 9. Ce₁/SCR catalyst showed less than 5% mercury oxidation efficiency whether 5% O₂ was introduced into gas stream or not. Considering the measurement error, O₂ had little effect on Hg⁰ oxidation. That was probably because the level of reaction between O₂ and Hg⁰ was low, generating unstable Hg²⁺ which coule convert to Hg⁰ easily.(3) 2Hg+O2⇌2HgO

0.1 ppm HCl balanced in N₂ resulted in Hg⁰ oxidation efficiency of 13.02%, which is higher than the 2.73% Hg⁰ oxidation efficiency under pure N₂ condition. Hg⁰ oxidation efficiency of 64.68% was observed when HCl concentration further increased to 1 ppm. HCl exhibited a decisive effect on Hg⁰ oxidation over Ce₁/SCR catalyst. 0.1 ppm HCl resulted in Hg⁰ oxidation efficiency of 50.42% with the aid of 5% O₂, indicating more HCl can be oxidized to form active chlorine species in the presence of O₂. That might because O₂ replenished the consumed chemisorbed oxygen, regenerated the lattice oxygen and hence maintained the high surface oxygen concentration [15]. The reaction between HCl and Ce₁/SCR catalysts may occur through(4) 2HCl+O=2Cl+H2O

(5) Cl+Hg0= HgCl

(6) HgCl+Cl=HgCl2  

where Cl denoted an active chlorine species for oxidizing Hg⁰, and O represented chemisorbed or lattice oxygen on the surface of Ce₁/SCR catalysts.

The entire reaction can be written as follows:(7) Hg0+2HCl+12O=HgCl2+H2O 

To determine the oxidation states of the element in these catalysts and to further explain the mechanism of Hg⁰ oxidation over Ce₁/SCR catalyst, the catalysts were investigated by XPS technique. Ce₁/SCR catalyst was pretreated in the gas composed of 50 μg/m³, 10 ppm HCl and 5% O₂ at 350°C for 4 h to get sCe₁/SCR catalyst. Fig. 10 showed the O1s XPS spectra for SCR, Ce₁/SCR and sCe₁/SCR catalyst. The peak appeared at low binding energy (529–530.5 eV) could be ascribed to be the lattice oxygen (denoted as O_b) [17], while the binding energy of 531.0–532.9 eV was ascribed to the chemisorbed oxygen and weakly bonded oxygen species (denoted as O_a) [18]. O_a has been thought to be the most active oxygen and played an important role in oxidation reaction [18–19]. The surface atomic concentrations of catalysts are given in Table 2. It could be found that O_a relative concentration of Ce₁/SCR calculated by O_a/(O_a + O_b) was higher than that of SCR. The reason was that Ce doping created labile oxygen vacancies, which may conduce to the improvement of chemisorbed oxygen and was helpful for mercury oxidation. The Ce3d spectra of sCe₁/SCR and Ce₁/SCR catalysts were presented in Fig. 11. The bands labeled u1 and v1 represent the characteristic peaks of Ce³⁺ with 3d104f1 initial electronic state, whereas the peaks labeled u, u2, u3, v, v2, and v3 represent Ce⁴⁺ with 3d104f0 electronic state [20].

Table 2 showed a slight change in the Ce valence state ratios after the tests under a flue gas flow. This implied that the redox shift between Ce³⁺ and Ce⁴⁺ happened during the Hg⁰ catalytic oxidation process. In comparison with Ce₁/SCR, the relative concentration of Ce³⁺ species decreased in sCe₁/SCR, which might be due to sufficient O₂ in flue gas to oxidize Ce³⁺ to Ce⁴⁺.

The Hg4f XPS profile spectrum of sCe₁/SCR catalyst was presented in Fig. 12. The binding energies at 101.5 and 110.5 eV was corresponded to HgCl₂ and HgO, respectively. No Hg⁰ was observed on the surface of sCe₁/SCR catalyst, which probably because HCl accelerated Hg⁰ oxidation.

Previous studies have shown that the interconversion of V⁴⁺/V⁵⁺ participates in the reaction process of Hg⁰ catalytic oxidation over SCR. Based on the results above, a possible mechanism for Hg⁰ oxidation can be proposed as follows:(8) 2HCl(g)+V2O5=2Cl∗(ad)+V2O4+H2O

(9) 2Cl(ad)+Hg0(g)=HgCl2(g)  

(10) V2O4+2CeO2=V2O5+Ce2O3    

(11) 2Ce2O3+O2=4CeO2       

HCl was adsorbed onto the active sites of catalyst and reacted with chemisorbed oxygen from V₂O₅ to form active surface chlorine species, which then would react with gas-phase or weakly bonded Hg⁰ to produce HgCl₂. The redox process of V⁴⁺/V⁵⁺ was contributed by the shift of CeO₂/Ce₂O₃. The missing lattice oxygen of CeO₂ would be replaced by O₂. The redox cycle (V⁴⁺ + Ce⁴⁺ ↔ V⁵⁺ + Ce³⁺) played an important role in promoting Hg⁰ oxidation.