Selective catalytic reduction (SCR) has proved to be the most effective technology in the control of NO_x emissions from both stationary and mobile sources,such as coal-fired power plants and heavy duty diesel engines ^{[1, 2, 3, 4, 5, 6]}. The most widely used SCR catalyst is a V₂O₅-based catalyst,which is very active when operating at 300-400 °C and highly tolerant to sulfur ^{[7, 8]}. However,the V₂O₅-based catalyst still suffers from a low N₂ selectivity,narrow temperature window,and sublimation of V₂O₅ at high temperatures even after modification with molybdenum and tungsten species ^{[9, 10, 11, 12, 13, 14, 15]}. The prices of tungsten and molybdenum resources have also increased recently. Therefore,improving the V₂O₅-based catalyst with a low cost composition is still of great importance in industrial applications ^{[16, 17, 18]}.

In recent years,many non-noble metals (Cu,Fe,Mn,Co,Ce,Zr,Nb,Sn,La,etc.) have been explored to promote the selectivity,operational temperature window,and stability of the V₂O₅-based catalyst ^{[7, 19, 20, 21, 22, 23]}. Normally,the introduced metals are in the form of oxides or vanadates,which have many effects on the active V species ^{[13, 17, 24, 25]}. The redox properties and dispersion of the vanadium species,the surface acid sites,and the inhibition of phase transfer from anatase to rutile are those most commonly targeted for improvement by introducing the metals. For example,the introduction of Fe₂O₃ improves the adsorption and activation of NH₃ ^[26],or the oxidation of NO to NO₂ ^[19]. Liu and co-workers ^[27] studied the effect of Mn on V₂O₅/TiO₂ for NH₃-SCR. They found that adding Mn significantly enhanced the activity,redox cycle,and the reactive intermediates,thus promoting the NH₃-SCR performance. Recently,some studies have shown that introducing the metals as vanadates can significantly improve the activity and N₂ selectivity,and especially the stability ^{[28, 29, 30]}. He and co-workers ^[29] have reported iron vanadates on TiO₂,which showed high activity,N₂ selectivity,and H₂O/SO₂ durability in NH₃-SCR reactions,which was due to the electronic inductive effect between the Fe³⁺ and V⁵⁺ species. Therefore,modifying the V₂O₅-based catalysts with low cost metals is possible and could be of great importance in industrial applications.

In this work,we introduced abundant,low cost transition metals (Cu,Fe,Mn,Co) into V₂O₅/TiO₂,forming highly dispersed M-V/TiO₂ catalysts for SCR reactions. The activity,N₂ selectivity,and SO₂/H₂O resistance of the different M-V/TiO₂ catalysts were studied. Both Cu-V/TiO₂ and Fe-V/TiO₂ performed excellently. The crystal structures and morphologies were characterised by X-ray diffraction (XRD),Raman spectroscopy,transmission electron microscopy (TEM),high- resolution TEM (HRTEM),and energy-dispersive X-ray analysis and element mapping (EDX mapping). Redox properties,surface active oxygen species and acid sites were studied by hydrogen temperature-programmed reduction (H₂-TPR),X-ray photoelectron spectroscopy (XPS),ammonia temperature- programmed desorption (NH₃-TPD),and in-situ diffuse reflectance infrared Fourier transform spectroscopy (in-situ DRIFTS). The relationship between the catalytic performance and the physical and chemical properties of the catalysts was studied.

Titanium (IV) oxide (99.8%,anatase) with a particle size of about 25 nm was purchased from Aladdin. All other chemicals were purchased from Sinopharm Chemical Regent Company and all were used without further purification.

The M-V/TiO₂ catalysts were prepared by a co-impregnation method. In a typical process,the desired amounts of the nitrates of the different metals (Cu,Fe,Mn,Co) were first dissolved in deionised water. The amounts of Cu(NO₃)₂·3H₂O,Fe(NO₃)₃·9H₂O,Mn(NO₃)₂ (50 wt% aq.),and Co(NO₃)₂·6H₂O were 0.23 g,0.05 g,0.28 g,and 0.23 g,respectively. The desired amount of ammonium metavanadate (NH₄VO₃) was dissolved in an aqueous solution of HNO₃. These amounts were 0.16 g,0.25 g,0.18 g,and 0.18 g,respectively. The total amount of all the M-V oxides was kept at 10% (mass fraction) of TiO₂ to achieve approximately monolayer coverage ^[29]. These two solutions were then mixed by magnetic stirring. TiO₂ (2.0 g) was then added to the mixed solution and dispersed with ultrasound and stirred for 1 h before being dried in a rotary evaporator at 48 °C with the help of a vacuum pump. These mixtures gave molar ratios of the metal to vanadium of 1:1.5,1:17,1:2,and 1:2,respectively,which were designed to favour the formation of vanadates,namely CuV₂O₆/Cu₂V₂O₇,Fe_0.12V₂O₅,MnV₂O₆,and CoV₂O₆. Finally,the so obtained samples were dried at 80 °C for 12 h and calcined at 400 °C for 6 h in air. For comparison,the V₂O₅/TiO₂ catalyst was also prepared by the same impregnation method without the addition of a nitrate.

For the comparative studies on the structure and the physical and chemical properties,the pure metal (Cu,Fe,Mn,Co) vanadates and corresponding metal (Cu,Fe,Mn,Co) oxides,namely 10% MO_x/TiO₂,were also prepared by the same impregnation method. The TiO₂ support and the ammonium metavanadate (NH₄VO₃) was absent from the preparation of these catalysts.

XRD was performed on a Rigaku D/MAX-RB X-ray diffractometer using Cu K_α (40 kV,40 mA) radiation and a secondary beam graphite monochromator. Nitrogen adsorption- desorption isotherms of the samples were measured at −196 °C using a Micromeritics ASAP 2010 instrument,and the corresponding pore size distribution curves were calculated from desorption branches by the BJH method. The specific surface areas of the samples were calculated by the Brunauer-Emmett-Teller (BET) method. TEM and HRTEM were recorded on a JEM-200CX and JEM-2100F microscope,respectively. EDX mapping was obtained using an Inca Energy 200 TEM system from Oxford Instruments. To prepare the (HR)TEM and EDX-mapping samples,the M-V/TiO₂ samples were first dispersed in ethanol by ultrasonication for 30 min and a Cu grid was dipped into the solution (a Mo grid was used instead of a Cu grid to avoid disturbing the copper in the Cu-V/TiO₂ catalyst) and dried under the irradiation of infrared lamp before analysis. XPS of all the elements was recorded on a Perkin-Elmer PHI 5000C ESCA system equipped with a dual X-ray source,using a Mg-K_α (1253.6 eV) anode and a hemispherical energy analyser. The background pressure during data acquisition was kept below 10⁻⁶ Pa. All the binding energies were calibrated using contaminant carbon (C 1s = 284.6 eV) as a reference. The visible Raman spectra were recorded on an inVia-reflex Renishaw spectrometer equipped with a holographic notch filter,a CCD detector,and irradiated with a laser at 514 nm. The UV Raman spectra were recorded on a home-assembled UV Raman system equ ipped with a Jobin-Yvon T64000 triple-stage spectrograph. A He-Cd laser at 325 nm was used as the excitation source.

H₂-TPR and NH₃-TPD were conducted on a TianjinXQ TP-5080 auto-adsorption apparatus,and NH₃ and H₂ were monitored by a TCD. Each sample (80 mg) was pre-treated with high-purity N₂ (99.999%,30 mL/min) at 300 °C for 30 min before either test. For H₂-TPR,after cooling to room temperature,the flowing gas was switched to 10% H₂-90% N₂ and the reactor temperature was raised to 800 or 900 °C at a rate of 10 °C/min. For NH₃-TPD,after cooling to 100 °C,the samples were exposed to a flow of NH₃ at the same temperature for 1 h. Finally,the temperature was raised to 500 °C at a heating rate of 10 °C/min.

In-situ DRIFTS experiments were conducted from 1100 to 1900 cm⁻¹ at a spectral resolution of 4 cm⁻¹(number of scans,64) on an FTIR spectrometer (Nicolet 6700) equipped with a Harrick DRIFTS cell and an MCT/A detector. The catalysts were pre-treated at 400 °C in a flow of N₂ (50 mL/min) for 0.5 h to remove physisorbed water and other impurities,then cooled to the target temperature under a N₂ flow before each experiment. Background spectra were recorded in the N₂ flow and automatically subtracted from the corresponding spectra.

The SCR reactions were carried out in a fixed-bed quartz micro-reactor (i.d. 8 mm) operating in steady state flow mode. The reaction conditions were as follows: 0.4 g catalysts,550 ppm NO,550 ppm NH₃,5 vol% O₂,8 vol% H₂O (when used),200 ppm SO₂ (when used),and the balance N₂. The total flow rate was 250 mL/min,and thus a gas hourly space velocity (GHSV) of 26000 h⁻¹. The temperature was increased from 150 to 400 °C. At each temperature step the concentrations of NO,NH₃,and N₂O were recorded. The concentrations of NO in the inlet and outlet gases were measured by a 4000VM NO_x analyser. The concentrations of N₂O and NH₃ were measured by a transmission IR N₂O analyser and IQ 350 ammonia analyser. NO conversion and N₂ selectivity were calculated according to the following expressions:

NO conversion = ([NO]_in - [NO]_out)/[NO]_in × 100%

N₂ selectivity = (1 - 2[N₂O]/([NO_x]_in + [NH₃]_in - [NO_x]_out - [NH₃]_out)) × 100%

The [NO]_in,[NO]_out,[NO_x]_in,[NO_x]_out,[NH₃]_in,[NH₃]_out,and [N₂O]_out correspond to the inlet and outlet concentration of a steady-state,respectively. NO_x means the total concentration of NO and NO₂.

Figure 1(a) shows the NO conversion over the different M-V/TiO₂ and V₂O₅/TiO₂ catalysts as the temperature increased from 150 to 400 °C. All the samples had an operational temperature window (the temperature range at which the conversion of NO is higher than 90%) within this range. Of these,Cu-V/TiO₂ showed the widest temperature window (225-375 °C),followed by Fe-V/TiO₂ (225-350 °C) and Mn-V/TiO₂ (275-375 °C). No obvious temperature window existed for Co-V/TiO₂. The V₂O₅/TiO₂ sample also showed a temperature window of 200-325 °C,which is a similar width to Cu-V/TiO₂ and Fe-V/TiO₂,but at lower temperatures. It should be noticed that while the loading of M-V and V were the same,the actual amount of vanadium species for V₂O₅/TiO₂ (10% V) was higher than for Cu-V/TiO₂ (6% V). It is widely accepted that the vanadium species is one of the most active components in the SCR reaction. Therefore,the high performance of V₂O₅/TiO₂ was simply due to the increased amount of V compared with Cu-V/TiO₂ and Fe-V/TiO₂. When the amount of the vanadium pentoxide was decreased to 6% (denoted as 6% V₂O₅/TiO₂),giving an equal molar amount of vanadium species as in Cu-V/TiO₂,the temperature window moved to 225-300 °C. This range is significantly more narrow than Cu-V/TiO₂.

Fig. 1. (a) NO conversion and (b) N2 selectivity for NH3-SCR over M-V/TiO2 as a function of temperature at a total feed gas rate of 250 mL/min. Reaction conditions: [NO] = [NH3] = 550 ppm,[O2] = 5%,GHSV = 26000 h−1.

The N₂ selectivity over the samples for the NH₃-SCR reactions was also studied,as shown in Fig. 1(b). It was found that the selectivity decreased for all samples as the reaction temperature increased. Among them,Cu-V/TiO₂ showed the highest selectivity,which remained above 94.9% for the temperature range of 150-400 °C. The next better ones were Fe-V/TiO₂ and Mn-V/TiO₂,where the selectivity for both remained above 89.6% over the same temperature range. In contrast,both Co-V/TiO₂ and V₂O₅/TiO₂ showed unstable selectivity,which rapidly decreased above 350 °C. The selectivity for V₂O₅/TiO₂ was below 90.0% for the entire temperature window of 200-325 °C and decreased to below 40% at 400 °C. Furthermore,the N₂ selectivity of 6% V₂O₅/TiO₂ decreased rapidly above 300 °C and was below 50% at 400 °C.

N₂O can be generated by the nonselective oxidation of NH₃,which is usually characterised by N₂ selectivity and has a significant influence on the NH₃-SCR performance ^[31]. For example,the N₂ selectivity of both the Co-V/TiO₂ and V₂O₅/TiO₂ catalysts decreased rapidly above 350 °C,and the NO conversion of the two catalysts decreased simultaneously,indicating that the decrease in activity was probably due to the nonselective oxidation of NH₃. The Cu-V/TiO₂ catalyst showed the highest N₂ selectivity over the temperature region of interest,indicating that the Cu-V/TiO₂ catalyst was efficient in reducing the unselective oxidation of NH₃.

State-of-the-art V-W-TiO₂ catalysts are (1-2)% V₂O₅/TiO₂ and (8-10)% WO₃/TiO₂. Therefore,we decided to study the influence of Cu and Fe on V₂O₅/TiO₂ catalysts with 1% V₂O₅. Fig. 2(a) shows that both Cu-1%V/TiO₂ and 1%V-10%W/TiO₂ have similar operational temperature windows of about 300-475 °C,which is considerably higher than Cu-V/TiO₂ and 6% V₂O₅/TiO₂. Fe-1%V/TiO₂ showed a more narrow temperature window of 375-475 °C. Fig. 2(b) shows that both Cu-1%V/TiO₂ and Fe-1%V/TiO₂ had much better selectivity than 1%V-10%W/TiO₂. Specifically,the selectivity of 1%V-10%W/TiO₂ decreased to below 90.0% between 400-475 °C,which was still within the operational temperature window.

Fig. 2. (a) NO conversion and (b) N2 selectivity in NH3-SCR reaction as a function of temperature in the feed gas of 250 mL/min total rate over the Cu-1%V/TiO2,Fe-1%V/TiO2,and 1%V-10%W/TiO2. Reaction conditions: [NO] = [NH3] = 550 ppm,[O2] = 5%,GHSV = 26000 h−1.

From these results,the introduction of Cu and Fe simultaneously improved both the NO conversion and N₂ selectivity over V₂O₅/TiO₂,whereas Mn and Co did not. The addition of Cu achieved similar NO conversion and better N₂ selectivity compared with the state-of-the-art V-W-TiO₂ catalysts,implying that a copper-modified V₂O₅-based catalyst might be a potential NH₃-SCR catalyst.

As both Cu-V/TiO₂ and Fe-V/TiO₂ catalysts showed high NO conversion and N₂ selectivity,they were chosen for further study into their stability and SO₂/H₂O durability. First,the stabilities of the Cu-V/TiO₂ and Fe-V/TiO₂ catalysts were tested at 175 and 275 °C. The feed gases were the same as in the activity measurements. Fig. 3(a) shows that both Cu-V/TiO₂ and Fe-V/TiO₂ were stable for 20 h operation in the SCR reaction.

Fig. 3.(a) Stability test at 175 and 275 °C for 20 h and (b) SO2/H2O durability test of the Cu-V/TiO2 and Fe-V/TiO2 catalysts. Reaction conditions: [NO] = [NH3] = 550 ppm,[O2] = 5%,[H2O] = 8 vol%,[SO2] = 200 ppm,GHSV = 26000 h−1.

2-15 vol% H₂O and 30-2000 ppm SO₂ are usually present in exhaust flows and have a strong effect on the performance of the catalysts. The SO₂/H₂O durability then was further investigated by introducing 8 vol% H₂O and 200 ppm SO₂ in the feed stream at 175 and 275 °C (Fig. 3(b)). The NO conversion over Cu-V/TiO₂ at 175 °C showed a small decrease from 46.4% to 44.1% after the introduction of H₂O and SO₂,but returned to 45.0% after the contaminants were turned off. The introduction of H₂O and SO₂ did not cause obvious effect on the performance of the catalyst. With the Fe-V/TiO₂ catalyst,the NO conversion decreased from 49.4% to 44.5%. When the temperature was increased to 275 °C,the combined impact of H₂O and SO₂ induced a 2.0% decrease in the NO conversion over Cu-V/TiO₂,but the NO conversion was recovered when the supply of H₂O and SO₂ was stopped. The NO conversion over the Fe-V/TiO₂ catalyst was not affected by the introduction of H₂O and SO₂ at this temperature,which agrees with other reports ^{[28, 29]}. In summation,both the Cu-V/TiO₂ and Fe-V/TiO₂ catalysts were highly sulfur-tolerant and water-resistant.

Table 1 shows the textural parameters of M-V/TiO₂ catalysts. The BET surface area of the TiO₂ support was about 89.2 m²/g. However,this decreased to 62.6-65.8 m²/g after the M-V loading,which was probably due to the impregnated species blocking the pores of the TiO₂ support or by aggregation of the TiO₂ support. The V₂O₅/TiO₂ catalyst showed the lowest surface area at 56.3 m²/g,indicating that the TiO₂ particles agglomerate more easily after loading with V₂O₅. Although the BET surface areas of the catalysts showed variation,there was no direct relationship with the activity of catalysts. For example,the NO conversion over the Cu-V/TiO₂ catalyst was the highest,but its BET surface area was relatively small (62.6 m²/g). In addition to the BET surface area,the textural parameters of the M-V/TiO₂ catalysts after the reaction test were also studied. The surface areas of some of the used catalysts were slightly increased,but this was within the error of the instrument. The decrease in activity at higher temperature was not due to the decrease of surface area,but more likely because of the nonselective oxidation of NH₃ as illustrated in Fig. 1(b). Based on these results,it appears that the textural parameters do not play an important role in the SCR reaction.

Table 1 Textural parameters of the M-V/TiO2 catalysts,obtained from N2 sorption results. Sample SBET a (m2/g) VBJH b (cm3/g) DBJH c (nm) TiO2 89.2 0.36 15.4 Cu-V/TiO2 62.6 0.26 16.4 Fe-V/TiO2 64.7 0.27 14.8 Mn-V/TiO2 65.8 0.25 17.0 Co-V/TiO2 63.9 0.28 15.6 V2O5/TiO2 56.3 0.24 18.9 Cu-V/TiO2 (used) d 67.3 0.32 15.6 Fe-V/TiO2 (used) d 62.6 0.28 15.6 Mn-V/TiO2 (used) d 63.1 0.29 15.5 Co-V/TiO2 (used) d 65.8 0.30 15.6 V2O5/TiO2 (used) d 57.9 0.25 15.8 a BET surface area. b BJH desorption pore volume. c Average pore diameter. d The catalysts after the activity test.

Table 1 Textural parameters of the M-V/TiO2 catalysts,obtained from N2 sorption results.

XRD patterns of the catalysts were recorded to study the crystal structure of M-V/TiO₂ catalysts,after being calcined at 400 °C for 6 h. As shown in Fig. 4,only the diffraction peaks of anatase (JCPDS Card No. 21-1272) were observed in all the samples. No other peaks belonging to the metal (Cu,Fe,Mn,Co),vanadates,metal oxides,or vanadium species were detected. This indicates that the vanadium species and the introduced transition metals were amorphous or in a highly dispersed crystallite state on the TiO₂ surface.

Fig. 4. XRD patterns of the M-V/TiO2 catalysts.

We further prepared the M-V oxides by the same process,but leaving out the TiO₂. The corresponding XRD patterns are shown in Fig. 5. Interestingly,vanadates were found in the M-V oxides. For Cu-V,the diffraction peaks could be assigned to the Cu₂V₂O₇,CuV₂O₆,and CuO. For Fe-V,the diffraction peaks of Fe_0.12V₂O₅ and Fe₂O₃ were observed. For Mn-V,the diffraction peaks of MnV₂O₆ and V₂O₅ were observed. For Co-V,the diffraction peaks of CoV₂O₆ and V₂O₅ were observed. This indicates that the metal vanadates might be formed over the M-V/TiO₂ obtained through similar processes.

Fig. 5. XRD patterns of the M-V oxides without TiO2 support. (a) Cu-V; (b) Fe-V; (c) Mn-V; (d) Co-V.

Raman spectroscopy was used to study the structure of these surface metal oxide species. Visible Raman spectra (Fig. 6(a)) indicated that the M-V/TiO₂ catalysts only displayed the typical Raman shifts of anatase TiO₂ (E_g at 145 and 636 cm⁻¹,B_1g at 396 cm⁻¹ and A_1g at 517 cm⁻¹) ^[29]. No other Raman shifts that could be ascribed to metal,vanadates,or metal oxides were observed.

Fig. 6. (a) Vis-Raman and (b) UV-Raman spectra of the M-V/TiO2 catalysts.

It has been well demonstrated that UV-Raman spectroscopy is more surface-sensitive for TiO₂-containing samples because TiO₂ has a strong absorption in the UV region ^{[32, 33]}. UV-Raman spectroscopy was therefore recorded with 325 nm excitation. Fig. 6(b) shows that new bands were observed in the 700-1100 cm⁻¹ region from species other than anatase TiO₂. In this region,the band at 915-950 cm⁻¹ was assigned to a polymerised V-O-V stretching mode. In particular,the bands within 930-960 cm⁻¹ were assigned to the bridging V-O-V stretching modes of a polymeric surface vanadate species,such as VO₄ units ^{[14, 34]}. The band at ~990 cm⁻¹ corresponds to terminal V=O groups in bulk V₂O₅ and indicates crystalline V₂O₅ is present on the surface of TiO₂ ^{[34, 35, 36]}. The V₂O₅/TiO₂ sample had two bands at 917 and 990 cm⁻¹,indicating that the vanadium species had agglomerated. For the M-V/TiO₂ samples,the 917 and 990 cm⁻¹ bands were absent (except in Fe-V/TiO₂),indicating that the introduced metals increased the dispersion of the vanadium species. Furthermore,the appearance of bands at around 940 cm⁻¹ showed that vanadates are present on the TiO₂. The presence of these bands in Fe-V/TiO₂ was probably due to the small amount of iron in the iron vanadates (Fe_0.12V₂O₅) as seen in the XRD results.

Combining the XRD and Raman results,we proposed that the dispersal of the vanadium species on the surface of the TiO₂ was improved by the introduced transition metals. Furthermore,vanadates were formed and might be beneficial for the SCR reaction as has been demonstrated previously ^{[28, 29, 37]}.

The morphology of Cu-V/TiO₂ was investigated by TEM as shown in Fig. 7(a). The catalyst particles were about 30-80 nm in diameter,which is significantly larger than the purchased TiO₂ (~25 nm). This might be caused by the aggregation of the TiO₂ support after loading the metal vanadates. This may also have caused the decrease in the BET surface area.

Fig. 7. (a) TEM image,(b) HRTEM image,and (c-f) EDX mapping images of the Cu-V/TiO2 catalyst.

The morphology of the Cu-V/TiO₂ catalyst was further studied by HRTEM. Fig. 7(b) indicates that the TiO₂ support was mainly anatase,as indicated by the (101) lattice fringes at 0.346 nm. No vanadium-containing phases were observed,indicating that the Cu-V was highly dispersed. The distributions of different elements were further studied by EDX mapping as shown in Fig. 7(c-f). The distributions of V and Ti were quite similar,indicating that the V species were uniformly dispersed on the TiO₂ support. The distributions of O and Cu were similar to each other but there was evidence they extended beyond the presence of V and Ti,which was probably due to the formation of CuO that was dispersed on the TiO₂ during sample preparation. The presence of CuO was observed in the XRD patterns in Fig. 5.

The redox properties of the M-V/TiO₂ catalysts were investigated by H₂-TPR. For comparison,the corresponding transition metal oxides over TiO₂ (MO_x/TiO₂) and V₂O₅/TiO₂ were also studied as shown in Fig. 8. The assignment of the peak positions to the corresponding chemical processes of MO_x/TiO₂ and V₂O₅/TiO₂ are summarised in Table 2 ^{[1, 38, 39, 40, 41, 42, 43, 44, 45, 46]}. It was widely accepted that the particle size has a significant influence on the reduction temperatures. Normally,the oxides with smaller particles are more easily reduced. For example,the isolated V species on a support are reduced at about 550 °C,while polymeric species are reduced above 700 °C ^{[47, 48, 49]}. However,the interaction between different oxides can cause co-reduction,which means that one component becomes easier to reduce when the nearby components are already reduced. For example,the copper species in copper vanadates were reduced at higher temperatures when compared with copper oxide,whereas the vanadium species in copper vanadates were reduced at lower temperature when compared with vanadium oxides ^[47]. Similar phenomena have been reported for different vanadates and were observed in this work ^{[50, 51]}.

Fig. 8. H2-TPR profiles of the catalysts. (a) MOx/TiO2; (b) M-V/TiO2.

Table 2 Temperature dependent reduction results for the MOx catalysts. MOx Peak position (°C) Chemical process Ref. V2O5 500-760 V2O5 + H2 → V2O4 + H2O [40, 41] CuO 200-220 2CuO + H2 → Cu2O + H2O [38, 39] 250-300 Cu2O + H2 → Cu + H2O Fe2O3 400-520 3Fe2O3 + H2 → 2Fe3O4 + H2O [42, 43] 450-620 Fe3O4 + H2 → 3FeO + H2O >700 FeO + H2 → Fe + H2O 320-350 2MnO2 + H2 → Mn2O3 + H2O [1, 44] MnO2 350-420 3Mn2O3 + H2 → 2Mn3O4 + H2O 460-530 Mn3O4 + H2 → 3MnO + H2O Co3O4 300-380 Co3O4 + H2 → 3CoO + H2O [45, 46] 400-480 CoO + H2 → Co + H2O

Table 2 Temperature dependent reduction results for the MOx catalysts.

Fig. 8(a) shows that the V₂O₅/TiO₂ catalyst had a reduction band at 534 °C,which was assigned to the reduction of an isolated V species from V⁵⁺ to V^{4+ [40, 41]}. The small peak at 680 °C was attributed to a V⁴⁺ to V³⁺ reduction ^{[40, 41]}. The profile of CuO/TiO₂ has two peaks at 184 and 274 °C,which were attributed to a Cu²⁺ to Cu⁺ and a Cu⁺ to Cu⁰ reduction,respectively ^{[38, 39]}. For the Cu-V/TiO₂ catalyst (Fig. 8(b)),the main reduction band was at 212 °C. According to the literature and the H₂-TPR profiles of CuO/TiO₂ and V₂O₅/TiO₂,this peak could be assigned to the reduction of a copper species that has been affected by the vanadium species. The small band at 372 °C was probably due to the reduction of a highly dispersed vanadium species. Therefore,the present H₂-TPR results suggest the presence of copper vanadates on Cu-V/TiO₂ ^[29]. A small band at about 862 °C was due to the reduction of a polymeric vanadium species ^{[47, 48, 49]}.

Fig. 8(b) also shows the H₂-TPR profiles of the Fe-V/TiO₂,Mn-V/TiO₂,and Co-V/TiO₂ catalysts. There are main reduction peaks at 451,522,and 438 °C for the Fe-V/TiO₂,Mn-V/TiO₂,and Co-V/TiO₂ catalysts,respectively. Combined with the reported reduction processes of different metal oxides from Table 2 and the H₂-TPR results of the transition metal oxides in Fig. 8(a),these reduction bands could be attributed to the co-reduction of the transition metal and an isolated vanadium species due to the formation of vanadates. The small bands at above 700 °C were attributed to the reduction of vanadium species.

The main outcome of these experiments is that there is only one main reduction band for the M-V/TiO₂ catalysts. The formation of vanadates resulted in the reduction of vanadium species at lower temperatures whereas the reduction of the transition metals moved to higher temperatures.

XPS was used to investigate the surface compositions and chemical states of the elements. The surface atomic concentrations,as estimated from the survey scan of the XPS data,are summarised in Table 3. It was clear that the ratios of transition metals to vanadium were close to the ratios of the precursor materials except for Fe-V/TiO₂,indicating that the transition metals and vanadium species were highly dispersed on the TiO₂ supports.

Table 3 Binding energies,active oxygen species,and the ratios of M to V in the V2O5/TiO2 and M-V/TiO2 catalysts. Catalyst Binding energy (eV) Oα /(Oα+Oβ) M/V Oα Oβ V2O5/TiO2 531.25 529.74 56.09% — Cu-V/TiO2 531.60 530.20 65.97% 1/2.7 Fe-V/TiO2 531.60 530.20 59.32% 1/3.8 Mn-V/TiO2 531.60 530.20 58.43% 1/1.7 Co-V/TiO2 531.60 530.20 55.05% 1/2.3

Table 3 Binding energies,active oxygen species,and the ratios of M to V in the V2O5/TiO2 and M-V/TiO2 catalysts.

The O 1s bands of the different catalysts in Fig. 9(a) were deconvoluted using a curve-fitting procedure. The sub-bands at lower binding energy (530.2-530.5 eV) belong to the lattice oxygen O²⁻ (denoted O_β),and the sub-bands at higher binding energy (531.6-532.1 eV) correspond to surface adsorbed oxygens (denoted O_α),such as O₂²⁻ (or O⁻) from defect oxides or hydroxyl-like groups ^{[19, 52, 53, 54, 55, 56]}. The corresponding peak- fitting results of O 1s and the relative concentration ratios of O_α/(O_α+O_β) for these catalysts are summarised in Table 3. The O_α/(O_α+O_β) ratio for Cu-V/TiO₂ was much higher than for other M-V/TiO₂ and V₂O₅/TiO₂. Surface oxygen O_α has been shown to have higher mobility than the lattice oxygen O_β,and O_α is therefore more active in the oxidation of NO to NO₂. The presence of NO₂ is beneficial to the so called “fast SCR” reaction and is helpful in the NH₃-SCR process ^{[57, 58, 59, 60, 61, 62]}. Thus,the improved SCR performance of Cu-V/TiO₂ was partially due to the presence of more surface oxygen O_α. For V₂O₅/TiO₂ and the other M-V/TiO₂ samples,the relative concentration ratios of O_α/(O_α+O_β) were quite similar,implying that the different activities were because of other factors,such as the surface acids sites and the redox properties.

Fig. 9. XPS spectra for (a) O 1s,(b) V 2p,(c) M (Cu,Fe,Mn,Co),and (d) Ti 2p of the catalysts.

The XPS results from the different transition metals are shown in Fig. 9(c). The metals were present in more than one valence state,indicating that they had complicated coordination spheres due to the formation of metal oxides and vanadates,as demonstrated in the XRD and Raman results,and the formation of new surface states. Cu 2p_3/2 and the corresponding satellite peak confirmed that the copper species were present mainly as Cu²⁺ in CuV₂O₆,Cu₂V₂O₇,or CuO. The Fe species were present as Fe³⁺,which was doped into the crystal lattice of V₂O₅ or as amorphous Fe₂O₃. For both the Mn and Co species,the more highly oxidised Mn⁴⁺ and Co³⁺ species were observed as well as Mn³⁺ and Co²⁺,indicating that other metal oxides were formed as well as the MnV₂O₆ and CoV₂O₇ vanadates. The XPS,XRD,and Raman results indicated that the M-V oxides on TiO₂ were a mixture of vanadates,metal oxides,and/or V₂O₅.

The V 2p and Ti 2p XPS results for V₂O₅/TiO₂ and M-V/TiO₂ are shown in Fig. 9(b) and (d),respectively. The binding energies of V 2p_3/2 and Ti 2p_3/2 are at 517.2 and 458.9 eV,respectively,indicating that the vanadium was mainly present as V⁵⁺ and the titanium as Ti⁴⁺. Fig. 9(b) indicates that both the V 2p of Cu-V/TiO₂ and Mn-V/TiO₂ had shifted to higher binding energies. Fig. 9(d) also indicates that the Ti 2p_3/2 peaks had shifted to higher binding energies,which was probably due to the electric effect between the vanadates and the TiO₂ support.

NH₃-TPD was performed to study the adsorption and activation of NH₃ on the active sites of the catalyst surface,as this is a crucial process in the NH₃-SCR reaction. Normally,the NH₃ desorption peaks at 120-240 °C corresponded to partially ionic NH₄⁺ bound to weak Brönsted acid sites. The peaks at 250-450 °C were assigned to the desorption of NH₄⁺ bound to strong Brönsted acid sites or coordinated NH₃ bounded to Lewis acid sites.

Fig. 10(a) shows that all the NH₃-TPD profiles of the M-V/TiO₂ catalysts can be deconvoluted into three peaks. According to the above assignments,the peaks below 240 °C (198 °C for Cu-V/TiO₂,190 °C for Fe-V/TiO₂,185 °C for Mn-V/TiO₂,184 and 238 °C for Co-V/TiO₂,176 and 230 °C for V₂O₅/TiO₂) belong to the desorption of partially ionic NH₄⁺ bound to weak Brönsted acid sites ^{[63, 64, 65]}. The peaks from 250 to 450 °C (266 and 353 °C for Cu-V/TiO₂,250 and 340 °C for Fe-V/TiO₂,252 and 335 °C for Mn-V/TiO₂,332 °C for Co-V/TiO₂,305 °C for V₂O₅/TiO₂) can then be assigned to the desorption of ionic NH₄⁺ bound to strong Brönsted acid sites or a coordinated NH₃ bound to Lewis acid sites ^{[20, 66, 67]}.

Fig. 10. (a) NH3-TPD profiles,(b) the ratios of strong acid sites to weak acid sites of the catalysts,and (c) the normalised peak intensity of the total acid sites (strong acid sites + weak acid sites) for M-V/TiO2 and V2O5/TiO2.

Because the strong Brönsted acid sites and Lewis acid sites located at 250-450 °C are difficult to identify based on the NH₃-TPD profiles,both acid sites were assigned as strong acid sites as distinct from the weak acid sites at 120-240 °C. The ratios of the strong acid sites to the weak acid sites are summarised in Fig. 10(b),whereas the total amount of surface acids sites is summarised in Fig. 10(c). Both results indicate that the introduction of transition metals significantly increased the total amount of acid sites and the strong acid sites,especially for Cu-V/TiO₂.

NH₃ adsorption in-situ DRIFTS was used to investigate the surface acidity of these catalysts,and the corresponding results are presented in Fig. 11. Normally,the bands at 1430-1440 and 1650-1680 cm⁻¹ are assigned to the asymmetric bending vibration of N-H bonds in NH₄⁺ chemisorbed on Brönsted acid sites ^{[58, 68, 69, 70, 71, 72]}. The bands at 1195-1200,1220-1225,and 1595-1610 cm⁻¹ were assigned to the asymmetric and symmetric bending vibrations of N-H bonds in NH₃ coordinated to Lewis acid sites ^{[69, 72, 73, 74]}. For V₂O₅/TiO₂ (Fig. 11(a)); the intensity of bands associated with the Brönsted acid sites was much stronger than that with the Lewis acid sites. With the introduction of transition metals,the band intensities due to the Lewis acid sites were significantly increased (except for Fe-V/TiO₂). For Cu-V/TiO₂ (Fig. 11(b)),the intensities of the Brönstedacid sites became very weak and could only be observed by zooming in on the region of interest. For Mn-V/TiO₂ and Co-V/TiO₂ (Fig. 11(d) and (e)),both Lewis acid sites and Brönsted acid sites were observed. The only exception was Fe-V/TiO₂,which had Lewis acid sites and Brönsted acid sites similar to V₂O₅/TiO₂. This was probably due to the small amount of iron in the vanadates.

Fig. 11. In-situ DRIFTS of NH3 adsorption on the catalysts. (a) V2O5/TiO2; (b) Cu-V/TiO2; (c) Fe-V/TiO2; (d) Mn-V/TiO2; (e) Co-V/TiO2; (f) The ratios of IL-acid/IB-acid for different catalysts. The inset in (b) zooms in on the 1400-1500 cm−1 region.

Fig. 11(f) further summarises the ratios of I_L-acid/I_B-acid. Here,I_L-acid and I_B-acid corresponded to the band intensities of Lewis acid sites at 1595-1610 cm⁻¹ and Brönsted acid sites at 1430-1440 cm⁻¹,respectively. The introduction of metals increased the amount of Lewis acid sites,especially for the Cu-V/TiO₂ catalyst. Combining the in-situ DRIFTS and NH₃-TPD results,it appears that the strong acid sites on Cu-V/TiO₂ are mainly Lewis acid sites,while the strong acid sites on V₂O₅/TiO₂ and Fe-V/TiO₂ are mainly Brönsted acid sites. It has been widely reported that both Lewis acid and Brönsted acid improve the SCR reaction but they followed different mechanisms ^{[75, 76, 77]}. The SCR reaction mechanism at Brönsted acid sites is as follows:

At Lewis acid sites,the coordinated NH₃ and NO react another way:

Therefore,the amount and the strength of the acid sites are important for the NO conversion and N₂ selectivity.

Fig. 12 shows the normalised peak intensities due to the Brönsted acid sites and Lewis acid sites for each catalyst as the temperature increases. Fig. 12(a) shows that the intensities due to the Brönsted acid sites for M-V/TiO₂ (M = Mn,Co) and V₂O₅/TiO₂ decreased with increasing temperature and finally tended to zero at about 250 °C. For Cu-V/TiO₂,the amount of Brönsted acid sites decreased rapidly although this was from a small starting amount. In contrast,the Brönsted acid sites on Fe-V/TiO₂ were much more stable than the other M-V/TiO₂ and V₂O₅/TiO₂ and were stable to 350 °C,due to the increased prevalence of stronger Brönsted acid sites,as seen in the NH₃-TPD profiles.

Fig. 12. The normalised peak intensity of the (a) Brönsted acid sites and (b) Lewis acid sites on different catalysts.

Fig. 12(b) indicated that the intensity due to the Lewis acid sites also decreased as the temperature increased. All the Lewis acid sites on the M-V/TiO₂ had not disappeared by 400 °C,indicating that they were more stable than the Brönsted acid sites. Remarkably,the Lewis acid sites on V₂O₅/TiO₂ were not nearly as stable and completely disappeared by 350 °C,which means that NH₃ will be effectively not adsorbed on V₂O₅/TiO₂ above 350 °C. It has been reported that the strong adsorption of NH₃ on a surface could determine the upper temperature limit for NO_x-reduction,and might account for the oxidation of NH₃ at higher temperatures ^[78]. Coincidentally,an obvious decrease in NO_x conversion and N₂ selectivity also occurred at 350 °C. We then proposed that the strength of the acids sites is very important in the N₂ selectivity at higher temperature ^{[79, 80, 81]}.

Different transition metals (Cu,Fe,Mn,Co) were successfully introduced to modify V₂O₅-based catalysts (M-V/TiO₂) for the SCR of NO with NH₃. The introduced metals induced the formation of vanadates,which helped disperse the V species as seen in the XRD,Raman,and (HR)TEM results. The introduced metals also significantly increased the amount of surface acid sites and the strength of these acid sites. The strong acid sites might be responsible for the high N₂ selectivity at higher temperatures. Among these catalysts,Cu-V/TiO₂ showed a much higher activity and selectivity than the other M-V/TiO₂ and V₂O₅/TiO₂. XPS,NH₃-TPD,and in-situ DRIFTS results suggested that the improved performance was probably due to more active surface oxygen species and more strong surface acid sites. The Cu-V/TiO₂ might be as a potential SCR catalyst for stationary flue gas.