Synergistic effect of VOx and MnOx surface species for improved performance of V2O5/Ce0.5Ti0.5−xMnxO2−δ catalysts in low-temperature NH3-SCR of NO

Thanh Huyen Vuong ab, Stephan Bartling a, Ursula Bentrup a, Henrik Lund a, Jabor Rabeah a, Hanan Atia a, Udo Armbruster a and Angelika Brückner *a
aLeibniz Institute for Catalysis at the University of Rostock, Albert-Einstein-Str. 29a, D-18059 Rostock, Germany. E-mail:; Fax: (+49) 381 1281 51 244
bSchool of Chemical Engineering, Hanoi University of Science and Technology, 1 Dai Co Viet, 10000 Hanoi, Vietnam

Received 23rd October 2018 , Accepted 26th October 2018

First published on 26th October 2018

Supported V2O5/Ce0.5Ti0.5−xMnxO2−δ (x = 0, 0.01, 0.05, 0.1, and 0.2) catalysts and bare supports were investigated in the selective catalytic reduction (SCR) of NO by NH3 between 100 and 300 °C. Incorporation of Mn into Ce0.5Ti0.5O2 increased the NO removal efficiency, reaching both NO and NH3 conversions above 85% between 175 to 300 °C at remarkably high N2 selectivities of ≥95%. Moreover, Mn-modified supports were also much more stable against deactivation in the presence of water. Deposition of vanadia on the Ce0.5Ti0.5−xMnxO2−δ supports enhanced the N2 selectivity, reaching ≥98% in the whole temperature range for Mn percentages of ≤0.1. Characterization results revealed that incorporation of Mn leads to supports with higher lattice disorder and smaller crystallite size and enhances the oxygen mobility and the reducibility of the supports. Operando EPR studies show that the synergistic effect of VOx and MnOx surface species in moieties containing Mn3+/Mn2+ and VO3+/VO2+ in close vicinity contributes to the prevention of undesired NH3 oxidation, thus improving the N2 selectivity of these catalysts.

1. Introduction

Nitrogen oxides are considered as main pollutants contributing to a series of environmental issues such as acid rain, photochemical smog, ozone depletion, fine particulate pollution and even accelerated global warming.1–4 Selective catalytic reduction of NOx with NH3 (NH3-SCR) is one of the most popular techniques to control NOx emission from combustion processes. This technology is widely applied in stationary sources and industrial processes, in which supported catalysts based on V2O5–WO3(MoO3)/TiO2 or zeolites are commonly used.4–7 However, these materials operate properly only in a rather high and narrow temperature range of 300–500 °C, which limits their practical application for the abatement of NOx from exhaust gases of other sources such as diesel or lean-burn gasoline engines, the temperature of which is much lower. Therefore, development and optimization of catalysts being active and selective for NH3-SCR of NOx already at temperatures well below 300 °C is highly desirable. Among a variety of catalysts tested for this purpose, those containing manganese belong to the most active ones.8–11 Unfortunately, most of these materials have major drawbacks. They form N2O as a harmful greenhouse gas, especially with rising reaction temperature, and they are poorly resistant to other exhaust gas components such as SO2 and H2O.12 To date, significant efforts have been made to solve these limitations. One promising approach is to mix or dope MnOx with other oxides. Several binary oxides including Ce–Mn,13,14 Mn–Zr,15 and Mn–Ni,16 and ternary compositions such as Mn–Ce–Ti,17 Sn–Mn–Ce,18 Ce–Fe–Mn,19 Zr–Ce–V,20 and Cu–Ce–Zr (ref. 21) have been investigated to enhance the N2 selectivity and prevent deactivation by H2O and SO2. Remarkably high performance has been obtained with Zr-doped CeVO4 showing an NO conversion of 95% above 175 °C at a gas-hourly space velocity (GHSV) of 26[thin space (1/6-em)]000 h−1 and still 90% between 200 and 350 °C at 400[thin space (1/6-em)]000 h−1, though fast deactivation in the presence of SO2 and SO2/H2O was observed at 190 °C.20 Ceria-modified FeMnOx was even more active reaching almost full NO conversion already around 110 °C at GHSV = 30[thin space (1/6-em)]000 h−1 which did not drop below 90% in the presence of SO2 and SO2/H2O, yet no information about N2 selectivity was given.19

Recently, we have developed a series of VOx/Ce1−xTixO2 catalysts in which the best binary oxide support Ce0.5Ti0.5O2 already achieved high NO and NH3 conversions of ≈80% at 200 °C in the absence of vanadium. Deposition of vanadia on the surface of this support boosted the NH3-SCR activity tremendously, reaching 100% NO conversion and 100% N2 selectivity even well below 200 °C at GHSV = 70[thin space (1/6-em)]000 h−11.22 Remarkably, the performance of this catalyst was still appreciable at a tenfold higher space velocity of 750[thin space (1/6-em)]000 h−1. This catalyst is among the best materials tested so far for LT-NH3-SCR of NO, considering that many other catalysts in the literature have been tested at much lower space velocities and frequently without providing N2 selectivity.

Bearing in mind the beneficial activity effect of manganese evident from literature data,8,12 the aim of this work was to improve the catalytic performance of the binary oxide Ce0.5Ti0.5O2 even more by combining it with MnOx. It is generally accepted that lattice oxygen participates in the NH3-SCR reaction.23 This means that a Mars–van Krevelen mechanism is operative in which NOx is first oxidized by lattice oxygen, and the resulting O vacancies are replenished by uptake of gas-phase O2.24–27 Mn ions are able to switch easily between different valence states and have markedly higher redox potentials compared to Ti ions (Mn3+/Mn2+ = 1.542 V, Mn4+/Mn2+ = 1.224 V, Ti4+/Ti3+ = −0.055 V).28 Thus, we anticipated that partial replacement of Ti in Ce0.5Ti0.5O2 by Mn might boost its activity in NH3-SCR of NOx by facilitating the initial oxidation of NO and ideally avoiding the formation of N2O. To this end, Ce0.5Ti0.5−xMnxO2−δ supports with different Mn contents and the corresponding supported vanadia catalysts have been tested in NH3-SCR of NO and their structure–reactivity relationships were elucidated by a combination of standard characterization methods as well as in situ and operando spectroscopy techniques including electron paramagnetic resonance (EPR) and diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS).

2. Experimental section

2.1. Catalyst preparation

The preparation of different Ce0.5Ti0.5−xMnxO2−δ supports was carried out by co-precipitation in the presence of a surfactant followed by a hydrothermal treatment. Hexadecyltrimethylammonium bromide (CTAB, Sigma Aldrich, ≥98%) was chosen as a surfactant since it is known to facilitate the generation of high surface areas and pore volumes.29–31 Appropriate amounts of Mn(NO3)2·4H2O (Sigma Aldrich, 99%) and Ce(NO3)3·6H2O (Acros, 99,5%) were dissolved in deionized water at room temperature to obtain 0.125 M mixed solution A. Meanwhile, a suitable amount of CTAB was dissolved in deionized water at 30 °C and aqueous ammonia was added to this solution under vigorous stirring at room temperature (solution B). For co-precipitation, suitable amounts of 0.125 M (NH4)2TiO(C2O4)2·H2O (Acros, 98%) and solution A were added dropwise to solution B under vigorous stirring to keep the pH at about 9. A molar ratio of (Ce + Ti + Mn)/CTAB = 1 was adjusted. After stirring at room temperature for 1 h, the obtained suspension was aged in a glass-sealed autoclave at 120 °C for 48 h. The resulting precipitate was filtered and subsequently washed with deionized water and ethanol. The obtained powder was dried at room temperature, then at 100 °C for 12 h, and subsequently calcined in air at 550 °C for 5 h.

Deposition of 5 wt% V2O5 was performed by wet impregnation. This value was chosen since samples loaded with 5 wt% V2O5 showed the highest performance in our previously studied series of VOx/Ce1−xZrxO2 and VOx/Ce1−xTixO2 catalysts, in which vanadia loadings lower and higher than 5 wt% were also tested, in comparison with a commercial 2% V2O5/8% WO3–TiO2 catalyst, which was less active than any of the ceria-based catalysts.22,32 The required amount of ammonium metavanadate (NH4VO3, Sigma Aldrich, 99%) was added to 0.2 M oxalic acid (C2H2O4·2H2O, Acros, 99.5%). The required amount of the calcined support powder was suspended in this solution. After stirring at room temperature for 2 h, the excess water was evaporated using a water bath, and the solid residue was dried at 120 °C for 12 h and subsequently calcined in air at 400 °C for 5 h.

2.2. Catalyst characterization

XRD powder patterns were recorded on a PANalytical X'Pert diffractometer equipped with an Xcelerator detector using automatic divergence slits and Cu Kα1/α2 radiation. The samples were mounted on silicon zero background holders. Peak intensities were converted from automatic to fixed divergence slits (0.25°) for further analysis. The peak positions and profile were fitted with a pseudo-Voigt function using the HighScore Plus software package (PANalytical). Phase identification was conducted by using the PDF-2 database of the International Center of Diffraction Data (ICDD). The crystallite size was calculated by applying the Scherrer equation using integral breadth under the assumption of spherically shaped crystallites for each peak separately (20° ≤ 2θ ≤ 75°). K is set to 1.0747.33 The mean value over all peaks is presented here.

Nitrogen adsorption isotherms were measured at −196 °C on an ASAP 2010 apparatus (Micromeritics GmbH, Aachen, Germany). Before analysis, the samples were degassed at 200 °C in a vacuum at 0.01 mbar for 3 h. The specific surface areas were calculated by the BET method, and the pore size distributions were determined by the BJH method. The elementary analysis was performed using a Varian 715-ES ICP-emission spectrometer and the ICP Expert software.

X-ray photoelectron spectra were obtained on a Thermo ESCALAB 220 iXL spectrometer (ThermoFisher) at room temperature using monochromatic AlKα radiation. Binding energies were corrected with reference to the C 1s value of 284.8 eV. Signal intensities were normalized using the sensitivity factors of Scofield34 and the transmission function of the spectrometer.

Raman measurements were recorded under ambient conditions using a Renishaw inVia Raman microscope utilizing a 633 nm excitation source with a power of 0.17 mW.

EPR spectra were recorded on an X-band cw-spectrometer ELEXSYS 500-10/12 (Bruker) using a microwave power of 6.3 mW and a modulation frequency and amplitude of 100 kHz and 5 G, respectively. In situ and operando EPR experiments were performed in a home-made quartz plug-flow reactor connected to a gas-dosing device with mass flow controllers (Bronkhorst) at the inlet and a quadrupole mass spectrometer (Omnistar, Pfeiffer Vacuum GmbH) at the outlet for on-line product analysis. This reactor was filled with 110 mg of catalyst particles (250–350 μm). The impact of oxidation, reduction and SCR reaction conditions was investigated separately. For these experiments, the samples were 1) treated in an O2 flow (30 ml min−1) at 300 °C for 1 h and then exposed subsequently at 200 °C for 30 min to a flow (50 ml min−1) of 2) 0.2% NH3/Ar, 3) 0.2% NO and 5% O2/Ar, and 4) 0.2% NH3, 0.2% NO, and 5% O2/Ar (SCR feed). Between subsequent steps, samples were flushed with argon, cooled to 150 °C and 20 °C for recording of the EPR spectra and reheated in argon again to 200 °C before switching to the next gas flow. During these experiments, NO and NH3 concentrations in the effluent were measured by on-line mass spectrometry.

In situ-DRIFTS measurements were performed on a Nicolet 6700 FTIR spectrometer equipped with a liquid-nitrogen cooled MCT detector. A high-temperature reaction cell (Harrick) with a temperature programmer (Eurotherm) connected to a gas-dosing device with mass flow controllers (Bronkhorst) at the inlet was filled with 145 mg of catalyst particles (250–350 μm). The catalysts were pretreated at 300 °C for 1 h in air followed by cooling to 175 °C and subsequently exposed to a flow of the desired gases (30 ml min−1). Background spectra were collected in flowing He and subtracted from the sample spectra for each measurement at the respective temperature.

H2-TPR measurements were performed in a U-shaped quartz reactor on a Micromeritics Autochem II 2920 instrument. For each experiment, 30 mg of support or 200 mg of V-containing catalyst particles (250–350 μm) were preheated to 400 °C in 5% O2/He (30 ml min−1) for 30 min before cooling to room temperature under flushing with Ar. TPR runs were carried out from room temperature to 800 °C under a 5% H2/Ar gas flow (50 ml min−1) at a heating rate of 10 °C min−1. Hydrogen consumption was monitored using a TCD detector.

2.3. NH3-SCR activity test

NH3-SCR activity measurements were carried out in a continuous-flow fixed-bed quartz reactor (length 200 mm, internal diameter 6 mm) using 100 mg catalyst particles (250–350 μm). The total flow rate of the NH3-SCR feed-gas mixture was 100 ml min−1, corresponding to a gas hourly space velocity (GHSV) of 70[thin space (1/6-em)]000 h−1. The feed contained 1000 ppm NO, 1000 ppm NH3, 5 vol% O2, 100 ppm SO2 (when used), 8 vol% H2O (when used), and balance He. The composition of the product gas was analyzed on a multigas sensor (Limas 11HW, ABB, Germany) including a catalytic converter delivering NO, NO2, and NH3 concentrations, and with an on-line gas chromatograph (HP 6890) using a molecular sieve 5A column for analysis of N2, N2O, and O2. Due to the inlet temperature and chemical equilibrium, NO was partly converted into NO2 before entering the reactor, and this was considered when calculating the conversion and selectivity. The reactant and product contents in the effluent gas were monitored during 1 h of continuous reaction in the steady state at each temperature from 100 to 300 °C in a step of 25 °C.

NOx and NH3 conversions and N2 selectivity were obtained by the following equations:35

image file: c8cy02193g-t1.tif(1)
image file: c8cy02193g-t2.tif(2)
image file: c8cy02193g-t3.tif(3)

3. Results and discussion

3.1. Catalytic behavior

Fig. 1 illustrates the NH3-SCR performance of all Ce0.5Ti0.5−xMnxO2−δ supports (x = 0–0.2) and their respective V/Ce0.5Ti0.5−xMnxO2−δ catalysts. The NOx and NH3 conversions of the bare supports increase with increasing Mn content already well below 250 °C. Above this temperature, the NOx conversion drops gradually with increasing Mn content. This is still negligible for x = 0.05 and 0.1 but visible for x = 0.2. This is due to the oxidation of NH3 which is then no longer available for selective reduction of NO. Consequently, the N2 selectivity drops as well with increasing Mn content (Fig. 1C and E). For Mn contents of 0.01 < x < 0.2, the catalytic performance was markedly higher than that for Mn-free Ce0.5Ti0.5O2−δ, and both NOx and NH3 conversions could be maintained above 85% from 175 to 300 °C, while the N2 selectivity remained above 95% in the whole temperature range.
image file: c8cy02193g-f1.tif
Fig. 1 NOx conversion (A and B), NH3 conversion (C and D), and N2 selectivity (E) over pure Ce0.5Ti0.5−xMnxO2−δ supports (open symbols) and their respective V/Ce0.5Ti0.5−xMnxO2−δ catalysts (filled symbols) as a function of temperature. Feed composition: 0.1% NO, 0.1% NH3, 5% O2/He, GHSV = 70[thin space (1/6-em)]000 h−1.

The deposition of 5 wt% vanadia improves the NH3 and NO conversions below 200 °C for the Mn-free V/Ce0.5Ti0.5O2 catalyst compared to the bare Ce0.5Ti0.5O2 support, however, this effect is negligible for all the other Mn-containing catalysts (Fig. 1B, D and E), suggesting that the major impact on activity is brought about by Mn incorporation but not by V deposition. However, surface vanadia species are beneficial for N2 selectivity which remains above 94% in the whole temperature range for the most active catalyst V/Ce0.5Ti0.3Mn0.2O2−δ while it drops drastically above 200 °C for the respective bare support (Fig. 1E), due to combustion of NH3 which is then no more available for reduction of NO (compare Fig. 1A and C).

A major problem for catalysts in low-temperature NH3-SCR is stability since the presence of H2O and SO2 in the exhaust gas stream is more problematic than at higher temperature.18,36,37 To check the supports and catalysts for potential deactivation, catalytic tests in the presence of H2O and SO2 have been performed at 200 °C. After adding 8 vol% H2O into the feed gas, the NOx conversion decreases significantly over the Mn-free support Ce0.5Ti0.5O2−δ while only a slight and almost equal decrease is observed for Mn contents of x = 0.1 and 0.2 (Fig. 2A). This indicates that incorporation of Mn into the supports slows down deactivation by water. Moreover, this activity loss is almost completely reversible within 4 h when H2O is removed from the feed stream. A much more severe and irreversible activity decline is caused by the simultaneous presence of water and SO2 for all supports, whereby this deactivation is most severe for the highest Mn content of x = 0.2.

image file: c8cy02193g-f2.tif
Fig. 2 Influence of H2O and SO2 on NOx conversion (A) and N2 selectivity (B) over Ce0.5Ti0.5−xMnxO2−δ supports (dashed lines and open symbols) and V/Ce0.5Ti0.5−xMnxO2−δ catalysts (solid lines and filled symbols) at 200 °C. Feed composition: 0.1% NO, 0.1% NH3, 5% O2/He, 8% H2O, 0.01% SO2, GHSV = 70[thin space (1/6-em)]000 h−1.

A similar and reversible decline of activity is also observed for the V-containing catalysts when only water is present in the feed. However, in the simultaneous presence of water and SO2 the activity decline is much smaller with the V-containing catalysts and, remarkably, the N2 selectivity remains stable at virtually 100% for all the catalysts. This suggests that deposition of vanadia on the support surface stabilizes the catalysts against deactivation (Fig. 2A). A reason may be that MnOx species facilitate the oxidation of SO2 to SO3 due to their higher reduction potential in comparison with V2O5 (MnO2/Mn2+ = 1.224 V, Mn2O3/Mn2+ = 1.485 V, V2O5/VO2+ = 0.957 V).25 Thus, active Mn3+/4+Ox in the bare supports might be converted into poorly active MnSO4 and/or Mn(HSO4)2 surface species which reduce NO and NH3 conversions but improves N2 selectivity.8,38 Partial coverage of MnOx by VOx species may hinder the formation of such species and, thus, be a reason for the observed higher stability of the V-containing catalysts. The opposite trend is observed for the N2 selectivity (Fig. 1B). It increases in the presence of both H2O and H2O/SO2, reaching quickly 100% for all the catalysts except the support with the highest Mn content of x = 0.2, for which this selectivity improvement takes a bit longer.

To gain more information about structure–reactivity relationships and the different roles of V and Mn in V/Ce0.5Ti0.5−xMnxO2−δ catalysts, comprehensive characterization studies have been performed, the results of which are described in the following sections.

3.2. Bulk and surface properties of bare supports and catalysts

The Mn-free Ce0.5Ti0.5O2−δ support crystallizes in the cubic fluorite structure of CeO2 without any peaks of TiO2, indicating the replacement of Ce4+ (97 pm) by smaller Ti4+ ions (74 pm) to form a homogeneous solid solution22 (see the XRD patterns in Fig. S1A). The additional presence of broad peaks may indicate either a subnanometer-sized material or monoclinic CeTi2O6 which is usually formed at Ce/Ti-ratios between 0.4 and 1.38,39 Incorporation of increasing amounts of Mn broadens all reflections which is due to the decrease of the mean crystallite size (Table 1). This might be due to replacement of Ce4+ and/or Ti4+ in bulk lattice positions mainly by smaller Mn4+ ions (d = 53 pm).28 Incorporation of lower valent Mn3+ or even Mn2+ seems to play a minor role, since in this case O vacancies should be created for which, however, no evidence is found in the Raman spectra (vide infra). Moreover, some minor additional peaks of Mn2O3 (2θ = 38.25 and 65.74)40 were found in the sample with the highest Mn content, too. The deposition of vanadia on the support surface did not lead to additional reflections of V2O5, confirming the highly dispersed and/or amorphous nature of these species.
Table 1 Mean crystallite size, specific surface area, pore volume and pore size of the supports and catalysts
Sample Mean crystallite size Surface area (m2 g−1) Pore volume (cm3 g−1) Average pore size (nm)
Ce0.5Ti0.5O2−δ 5.9 169.9 0.45 7.58
Ce0.5Ti0.49Mn0.01O2−δ 6.8 148.1 0.36 8.50
Ce0.5Ti0.45Mn0.05O2−δ 5.2 140.7 0.43 9.05
Ce0.5Ti0.4Mn0.1O2−δ 5.3 182.8 0.46 10.71
Ce0.5Ti0.3Mn0.2O2−δ 3.6 147.9 0.51 11.10
V/Ce0.5Ti0.5O2−δ 5.4 143.6 0.35 7.18
V/Ce0.5Ti0.49Mn0.01O2−δ 6.7 146.8 0.33 8.00
V/Ce0.5Ti0.45Mn0.05O2−δ 5.4 126.6 0.38 8.99
V/Ce0.5Ti0.4Mn0.1O2−δ 4.1 168.9 0.43 10.45
V/Ce0.5Ti0.3Mn0.2O2−δ 3.2 143.8 0.48 10.05

As shown in Table 1, the supports do not differ much in their specific surface area and pore volume. Just a small increase of the pore size is observed with increasing Mn content. Moreover, the N2 adsorption–desorption isotherms of all the supports point to mesoporosity (Fig. S2).41 Deposition of vanadia on the supports had a negligible impact on the BET surface area and pore properties (Table 1).

The Raman spectra of all the samples (Fig. 3) show an intense band around 462 cm−1 due to the triply degenerate F2g mode of the cubic fluorite structure of ceria.42,43 Besides, a weak band at 282 cm−1 and a shoulder at 402 cm−1 are related to the longitudinal and transversal stretching modes of Ce–O bonds on the (111) surface of ceria nanocrystals.44 It has been shown that partial reduction of Ce4+ to Ce3+, i.e. creation of oxygen vacancies in the ceria lattice, leads to decreasing intensity of this band while the intensity of a band at 570 cm−1 (assigned to oxygen vacancies) increases.44 The spectra in Fig. 3 do not show any defect feature around 570 cm−1 and the intensity of the band at 282 cm−1 does not drop. Moreover, no red shift of the F2g (seen after partial reduction of ceria44) occurred upon incorporation of Mn. This suggests that the reduction of Ce4+ as well as replacement of Ce4+ and/or Ti4+ by lower valent Mnn+ ions (n < 4) and, consequently, creation of O vacancies is negligible in Ce0.5Ti0.5−xMnxO2−δ supports. Instead, Ce4+ and/or Ti4+ might be replaced in their lattice positions by isovalent Mn4+ which does not cause charge mismatch as a reason for O vacancy formation. This is in agreement with the XPS results discussed below. However, the band at 282 cm−1 shifts to lower wavenumbers with increasing Mn content, being particularly obvious for sample Ce0.5Ti0.3Mn0.2O2−δ. This shift is due to the increase of lattice distortion45 since the size of Mn4+ (53 pm) is significantly smaller than those of Ce4+ (d = 94 pm) and Ti4+ (0.61 pm).28 Moreover, a band around 704 cm−1 is seen, which can be assigned to Ti atoms with sevenfold coordination in a fluorite structure, being in good agreement with the observed formation of a solid Ce0.5Ti0.5O2 solution.46,47 The absence of this band in the sample Ce0.5Ti0.3Mn0.2O2−δ might be due to its comparably low Ti content (Table 2). Although Mn2O3 is found in the XRD powder pattern of the Ce0.5Ti0.3Mn0.2O2−δ support with the highest Mn content (Fig. S1A), no respective Raman bands are observed, probably since reduced Mn oxides are weak Raman scatterers.17,48

image file: c8cy02193g-f3.tif
Fig. 3 Raman spectra of the supports (A) and V catalysts (B).
Table 2 Bulk and surface compositions of supports and catalysts
Sample Bulk atomic ratioa Surface atomic ratiob

image file: c8cy02193g-t4.tif

image file: c8cy02193g-t5.tif

image file: c8cy02193g-t6.tif

image file: c8cy02193g-t7.tif

image file: c8cy02193g-t8.tif

image file: c8cy02193g-t9.tif

a Results from ICP. b Results from XPS.
Ce0.5Ti0.5O2 0.986 1.630
Ce0.5Ti0.4Mn0.1O2−δ 0.712 0.167 0.711 0.280
Ce0.5Ti0.3Mn0.2O2−δ 0.514 0.331 0.841 0.227
V/Ce0.5Ti0.5O2 0.879 0.075 2.235 0.167
V/Ce0.5Ti0.4Mn0.1O2−δ 0.678 0.163 0.072 1.595 0.400 0.224
V/Ce0.5Ti0.3Mn0.2O2−δ 0.568 0.301 0.077 0.927 0.294 0.225

After impregnation with vanadia, a weak band at 282 cm−1 reflecting surface Ce–O bonds is observed only in the sample V/Ce0.5Ti0.5O2 (Fig. 3B). The absence of this band in the Mn-containing catalysts might be due to the higher V dispersion on their surfaces, evident from the surface V/(Ce + Ti + Mn) ratios in Table 2. Bands of crystalline V2O5 were not detected on all the samples, only a weak band of the symmetric V[double bond, length as m-dash]O stretching vibration around 1022 cm−1 was observed. In most supported V catalysts such a band is only properly observed after removing adsorbed water by thermal treatment in situ. Since the spectra in Fig. 3B have been recorded from samples stored in ambient atmosphere, this band is very weak Fig. 3B. Taking into account that bands at 1008, 1015, 1022–1030 and 1044 cm−1 are assigned to monomeric, dimeric, trimeric, and polymeric VOx, respectively,49,50 the band around 1022 cm−1 arises most probably from oligomeric surface VOx species. The missing V[double bond, length as m-dash]O bands for V/Ce0.5Ti0.5O2 and V/Ce0.5Ti0.3Mn0.2O2−δ might be due to the high dispersion of the surface vanadia species.

XPS measurements were performed to identify the surface characteristics and the valence states of the catalysts (Table 2, Section SI-C in the ESI). For all the samples, the O 1s peak splits into two features at binding energies (EB) of 531.5–531.8 eV and 529.7–530.0 eV (Fig. S3). The former can be assigned to weakly adsorbed surface oxygen while the latter results from lattice oxygen.51 The Mn 2p region (Fig. S4) contains two main peaks (2p1/2 and 2p3/2) and a satellite feature around 647 eV. The large line width suggests a superposition of Mn3+ and Mn4+ reflected by contributions at 641.9±0.3 and 643.4±0.3 eV, respectively, while the peak at 640.9 ± 0.3 eV and a weak satellite feature at 647 eV correspond to Mn2+.52,53

All peaks in the Ce 3d spectra fall in the range typical for 3d5/2 (ν, ν′′ ν′′′) and 3d3/2 (υ, υ′′, υ′′′) core electrons of Ce4+ (Fig. S5). In particular the υ′′′ peak at 917.7 eV is a fingerprint for Ce4+.54,55 No typical signal of Ce3+ was observed. Usually the two 3d peaks of Ce3+ split into two features each. Thus, the 3d5/2 peak consists of two contributions from the Ce 3d94f1 O 2p6 final state (ν′ around 885 eV) and the Ce 3d94f2 O 2p5 state (ν0 around 880–881 eV).54,56 In the samples containing both Ce4+ and Ce3+, the ν′ peak of Ce3+ is frequently difficult to separate from the ν and ν′′ features of Ce4+. It is therefore on its own not sufficient for doubtless identification of Ce3+. A clear indication of the presence of Ce3+ is only when the ν0 peak of Ce3+ is clearly discernible from the superimposed ν peak of Ce4+ below 881 eV. In the spectra in Fig. S5 no feature below 881 eV is visible. Therefore, the presence of Ce3+ can be safely excluded in all the samples. This agrees well with the fact that no Raman bands of oxygen vacancies at 570 cm−1 were observed, since these would have had to be formed upon replacement of Ce4+ by Ce3+. The Ti 2p EB values of all the samples (Fig. S6) correspond to Ti4+ states (458.5–459.2 eV (ref. 57)). This seems plausible, given that the reduction potential of the Ti4+/Ti3+ pair is −0.055 V the lowest in the mixed oxide.28 Possibly, partial reduction of Mn4+ keeps Ti4+ in its highest oxidation state. The V 2p3/2 peaks of all the supported vanadia catalysts are located around EB = 517.5 eV being characteristic of V5+ (Fig. S7).55,57 While the Mn/Ce bulk atomic ratios of the bare supports increase with increasing Mn content, as expected, the Mn/Ce ratio on the surface of Ce0.5Ti0.3Mn0.2O2−δ is lower than that in the bulk (Table 2). This may be due to the minor formation of Mn2O3 clusters in this support suggested, too, by the XRD (Fig. S1) and TPR results (vide infra). The Ti/Ce atomic ratios drop with decreasing Ti content in both the bulk and the surface. However, the surface Ti/Ce ratios are for all supports higher than the corresponding bulk ratios. This indicates that Ti is enriched on the surface of all supports. The surface V/(Ce + Ti + Mn) ratios are generally much higher than the respective bulk values determined by ICP-OES (Table 2), which is expected for catalysts prepared by impregnation procedures. Moreover, while the surface V/(Ce + Ti + Mn) ratios do not differ much for catalysts with no or low Mn content, it is higher for the most active catalyst V/Ce0.5Ti0.3Mn0.2O2−δ, which may be one reason for its superior activity (cf.Fig. 1).

The EPR spectra of the Mn-free support Ce0.5Ti0.5O2 and catalyst V/Ce0.5Ti0.5O2 show only weak signals of paramagnetic oxygen defects such as O˙ and/or O2˙ in the range of g = 2.058–1.997 (Fig. 4A).58,59 Additionally, a rhombic signal with gx = 1.976, gy = 1.963 and gz = 1.942 can be assigned to Ti3+ in the support lattice. This line lost its intensity and changed to axial symmetry (g = 1.963, g = 1.942)60 after depositing vanadia. The EPR spectra of V/Ce0.5Ti0.5O2 contained an additional weak isotropic signal at g = 1.993 that may be related to Ti3+ on the surface of this sample.61 No VO2+ signal is seen, suggesting that EPR-silent pentavalent V5+ is dominant on the surface of this catalyst, in agreement with the XPS results (Fig. S7).

image file: c8cy02193g-f4.tif
Fig. 4 The EPR spectra of Ce0.5Ti0.5O2−δ and V/Ce0.5Ti0.5O2−δ recorded at 20 °C (A) and Mn-containing supports (solid lines) and V-loaded catalysts (dashed lines) recorded at 200 °C (B).

The Mn-containing samples show a sextet of narrow lines with g = 2.004 and a hyperfine structure splitting (hfs) constant of A = 90 G, arising from Mn2+ single sites in an octahedral environment which might be located in the cationic positions of the cubic crystal lattice.62 This sextet is superimposed on a broad background signal around g = 2, the intensity of which increases with increasing Mn content (Fig. 4B). This line is due to magnetically interacting Mn2+ species.63 In the V-containing catalysts, an EPR signal of paramagnetic VO2+ species has not been detected. A reason might be that EPR-silent VO3+ is dominant on the catalyst surface, in agreement with the XPS data (Fig. S7). On the other hand, the total EPR signal intensity of the V-containing catalysts is lower in comparison with the V-free supports. This may suggest that some paramagnetic V4+ could be in tight contact with the MnOx species, leading to magnetic interactions which broaden the EPR line and reduce the signal amplitude. Since XPS does not point to such reduced V species, they may be located in the subsurface on the interface to the MnOx species.

Catalytic activity in NH3-SCR of NOx is generally influenced by the reducibility of the catalysts since the reaction involves a Mars–van Krevelen redox cycle in which NOx species are first oxidized by lattice oxygen.64 Therefore, selected catalysts and their corresponding bare supports were also analyzed by H2-TPR (Fig. 5). The Mn-free support Ce0.5Ti0.5O2−δ showed a signal with a maximum at about 445 °C. This value is lower than the reduction temperatures of surface and lattice oxygen in pure CeO2 (500 °C and above 600 °C, respectively).65 This can be explained by the formation of Ce–O–Ti moieties in the Ce0.5Ti0.5O2−δ solid solution that improves its reducibility.22,66 Incorporation of Mn enhances the reducibility of the Ce0.5Ti0.5−xMnxO2−δ supports even more. This is reflected by the shift of the TPR peak maximum to lower temperatures and the increase of the H2 consumption with increasing Mn content (Fig. 5, Table S1). This is in nice agreement with the catalytic test results showing, too, an activity increase with increasing Mn content, i.e. with increasing reducibility created by these Mn species. The TPR signals of the bare supports are broad and reflect the reduction of different MnOx species.48 A precise discrimination is not possible; however, the shoulder at the lowest temperature (293 °C) may be assigned to the reduction of Mn4+ to Mn3+. This agrees with the XPS results which show the presence of different Mnn+ species in these supports. The deposition of vanadia on the surface of these supports shifts the reduction peak maximum to higher temperatures (505–540 °C) and increases the H2 consumption. This results most probably from the reduction of surface V5+ species.67,68 Moreover, a weak peak at 250–275 °C in the sample V/Ce0.5Ti0.3Mn0.2O2−δ with the highest Mn content can be assigned to the reduction of MnO2 to Mn2O3. The catalytic performance of this catalyst is slightly higher compared to that of the corresponding V-free Ce0.5Ti0.3Mn0.2O2−δ support. It is also significantly higher than those of the other catalysts which do not differ much (Fig. 1). Apart from minor amounts of MnOx clusters, this catalyst has the highest surface ratios of V/(Ce + Ti + Mn) and Mn/Ce (Table 2). Possibly, the effective redox interaction of MnOx and V surface species, being most abundant on this catalyst, may be responsible for the superior activity in comparison with V-containing catalysts with lower Mn content. Thus, the TPR findings seem to be somehow in line with the observed trend in catalytic activity.

image file: c8cy02193g-f5.tif
Fig. 5 H2-TPR profile of pure supports (a) Ce0.5Ti0.5O2−δ, (b) Ce0.5Ti0.45Mn0.05O2−δ, (c) Ce0.5Ti0.4Mn0.1O2−δ, and (d) Ce0.5Ti0.3Mn0.2O2−δ and vanadium-containing catalysts (e) V/Ce0.5Ti0.5O2−δ, (f) V/Ce0.5Ti0.45Mn0.05O2−δ, (g) V/Ce0.5Ti0.4Mn0.1O2−δ, and (h) V/Ce0.5Ti0.3Mn0.2O2−δ.

3.3. Spectroscopic in situ and operando investigation of active sites

3.3.1. In situ DRIFTS investigations. At first, the interaction of the feed components with the different catalysts was studied by in situ DRIFTS experiments. The spectra obtained after treatment of the samples with 0.1% NH3/He are displayed in Fig. 6A. For the V-containing catalysts bands at 1678/1427 cm−1 are seen which belong to the symmetric and asymmetric modes of NH4+ indicating Brønsted sites, while the bands at 1606/1218 cm−1 are assigned to the asymmetric and symmetric N–H bending vibrations of NH3 coordinated to Lewis acid sites.69 In the region around 2040 cm−1 a decreasing intensity of the 2ν overtone band of V5+[double bond, length as m-dash]O moieties and a shift to lower wavenumbers are observed during exposure to NH3 (Fig. S8), indicating the interaction of these moieties with NH3. After flushing with helium, this band increases and shifts to higher wavenumbers again, yet its initial intensity is not completely restored. This finding suggests that the changes of the 2ν overtone V5+[double bond, length as m-dash]O band upon NH3 exposure are on the one hand due to an adsorbate effect,70 but on the other hand also due to reduction of VO3+ to VO2+ as evidenced, too, by in situ EPR spectroscopy described below. As expected, no Brønsted sites exist on the supports, only bands at 1606 and 1184 cm−1 are seen, resulting from NH3 coordinated on Lewis sites. Small features at 1545, 1427 and 1370 cm−1 in the V-free Ce0.5Ti0.5O2−δ support stem from formate species71,72 formed by reaction of a surface carbonate impurity with NH3 and surface hydroxyl groups.
image file: c8cy02193g-f6.tif
Fig. 6 In situ DRIFT spectra of bare supports and V-containing catalysts recorded after 45 min exposure to 0.1% NH3/He flow at 175 °C followed by purging with He for 30 min (A) and after exposure to 0.1% NO/He as well as 0.1% NO and 5% O2/He flow at 175 °C followed by purging with He (B).

The spectra obtained after exposure of the samples to 0.1% NO/He as well as to 0.1% NO and 5% O2/He are shown in Fig. 6B. The bands which appear between 1620–1530 cm−1 and 1240–1210 cm−1 are characteristic of adsorbed nitrate species. Depending on the band positions, one can distinguish monodentate (1530–1480 cm−1 and 1290–1250 cm−1) as well as chelating and bridging bidentate (1500–1650 cm−1 and 1200–1300 cm−1) nitrate species.73 The formation of nitrate species is favoured on the supports, in particular when Mn is present, while on the V-containing catalysts only minor nitrate formation is observed. Hence, the spectra of V/Ce0.5Ti0.4Mn0.1O2−δ are exemplarily depicted in Fig. 6B for comparison. Comparing the spectra obtained after exposure to NO/He and NO/O2/He, respectively, it is obvious that nitrate formation proceeds already without additional gaseous oxygen. This suggests the participation of lattice oxygen supplied by the support. The Mn-containing supports exhibit an essentially higher oxygen mobility and oxidation activity because the formation of bidentate nitrates species (bands at 1570/1595/1619 cm−1) is facilitated, while on Ce0.5Ti0.5O2δ only monodentate nitrate species are formed (bands at 1531/1214 cm−1). This finding is in agreement with DFT calculations of García Pintos et al.74 for Mn-doped CeO2, which showed that Mn promotes the release of surface oxygen and consequently the formation of surface oxygen vacancies. As expected, these vacancies are quickly replenished when O2 is present in the feed, which leads to increased nitrate formation and consequently much more pronounced surface nitrate bands as reflected by Fig. 6B. Due to coverage of the strongly oxidizing MnOx by VOx species the oxidation activity of the V-containing catalysts is lower. Instead, on the V-containing catalysts the adsorption of NH3 is facilitated because additional Brønsted sites are created by the VOx species (cf.Fig. 6A).

To gain more information about the activity of the adsorbate species generated by adsorption of NH3 or NO/O2, their subsequent reaction with NO/O2 or NH3 was studied. It should be mentioned here that for the different supports and V-containing catalysts in general similar trends concerning adsorbate formation and reactivity could be observed. To demonstrate the different behaviours of both types of catalysts, therefore, only the experiments with Ce0.5Ti0.4Mn0.1O2−δ and V/Ce0.5Ti0.4Mn0.1O2−δ will be exemplarily discussed.

The adsorbate spectra of Ce0.5Ti0.4Mn0.1O2−δ and V/Ce0.5Ti0.4Mn0.1O2−δ obtained after 45 min exposure to a NH3/He flow at 175 °C followed by purging with He and subsequent switch to NO/O2/He are shown in Fig. 7A. Upon exposure to NO/O2/He, the bands of pre-adsorbed NH3 and NH4+ vanish gradually on Ce0.5Ti0.4Mn0.1O2−δ (Fig. 7A), while new bands appear at 1618/1595/1570/1536 cm−1 as well as at 1241/1212 cm−1 resulting from monodentate and bidentate nitrate species (cf.Fig. 6B). The V/Ce0.5Ti0.4Mn0.1O2−δ catalyst shows a similar behaviour, however, the extent of nitrate formation is extremely low (note the different unit ranges on the ordinate in Fig. 7A and B). Interestingly, after complete vanishing of the NH3 adsorbate bands after 5 min exposure of the V/Ce0.5Ti0.4Mn0.1O2−δ catalyst to NO/O2/He, a single band at 1610 cm−1 appears which is related to adsorbed NO2.75 The formation of nitrates needs longer exposure times to the NO/O2/He feed and is seen only after 10 min.

image file: c8cy02193g-f7.tif
Fig. 7 In situ DRIFT spectra of Ce0.5Ti0.4Mn0.1O2−δ and V/Ce0.5Ti0.4Mn0.1O2−δ recorded after 45 min exposure to 0.1% NH3/He flow at 175 °C followed by purging with He for 30 min and subsequent switch to 0.1% NO and 5% O2/He flow (A) and after pre-adsorption of 0.1% NO and 5% O2/He for 45 min at 175 °C followed by He flushing and subsequent exposure to 0.1% NH3/He flow (B).

The adsorbate spectra of Ce0.5Ti0.4Mn0.1O2−δ and V/Ce0.5Ti0.4Mn0.1O2−δ obtained after pre-adsorption of NO/O2/He and subsequent exposure to NH3 at 175 °C are depicted in Fig. 7B. When pre-adsorbed nitrates on Ce0.5Ti0.4Mn0.1O2−δ become in contact with NH3, the typical nitrate bands lose their intensity, while new bands grow at 1607 and 1178 cm−1 stemming from coordinatively adsorbed NH3. It is noticeable that particularly the bands around 1570 cm−1 resulting from bidentate nitrate species vanish, while the bands of the monodentate nitrate species (1543/1512/1229 cm−1) do not completely disappear. This points to a higher reactivity of bidentate nitrate species as also observed by Hu et al.51,76 over Mn-containing Ce0.75Zr0.25O2 and Co3O4/TiO2 catalysts. The weak nitrate bands on V/Ce0.5Ti0.4Mn0.1O2−δ vanish quickly, and the surface is dominated by adsorbed NH3 as reflected by the typical bands of NH4+ at 1678/1427 cm−1 and of NH3 coordinated to Lewis acid sites at 1606/1218 cm−1 (cf.Fig. 6A). In the spectral region of the 2ν overtone of V5+[double bond, length as m-dash]O moieties (Fig. S9), no changes of the band at 2039 cm−1 were observed upon exposure to NO/O2/He and He flushing indicating that the adsorbed nitrates are located at the support. Only upon subsequent exposure to NH3 that a drastic decrease and a shift to lower wavenumbers are observed caused by the interaction of V5+[double bond, length as m-dash]O moieties with NH3, as already described above.

The in situ DRIFTS results reflect interesting differences in the roles of vanadium and manganese sites in NH3-SCR. NH3 can be adsorbed and activated on all surfaces while oxidation of NO to adsorbed NOx species proceeds only on Ce–O–M (M = Ce, Mn, Ti) surface sites but not on vanadium species. Previously we have found that NH3-SCR proceeds on V- and Mn-free Ce1−xTixO2 supports after a Langmuir–Hinshelwood (L–H) mechanism in which both NO and NH3 are first adsorbed to give activated surface species that further react with N2 and H2O, while an Eley–Rideal (E–R) mechanism is identified on the corresponding supported VOx/Ce1−xTixO2 catalysts in which gaseous NO reacts with adsorbed NHx.22 The results of the present study suggest that NH3-SCR on Mn-containing bare supports proceeds after a L–H mechanism, too.

3.3.2. Operando EPR investigations. Operando EPR spectra after subsequent treatment in different gas flows for the Mn-free Ce0.5Ti0.5O2−δ support and the related V/Ce0.5Ti0.5O2−δ catalyst are depicted in Fig. 8 (for results of the simultaneous mass spectrometric analysis of the effluent gas stream see Fig. S10). After oxidative pre-treatment, the EPR spectra of Ce0.5Ti0.5O2 show weak signals in the range of g = 2.058–2.011, which can be assigned to oxygen species such as O˙ and/or O2˙ nearby Ce4+ and Ti4+.59 These signals do not change depending on the feed composition, suggesting that they might be located in the bulk and do not participate in the reaction. Besides these oxygen species, a rhombic signal with g1 = 1.983, g2 = 1.965, and g3 = 1.945 was also observed, which can be ascribed to Ti3+ in the lattice positions of the CeO2 matrix.60 Under an NH3/He flow this signal increases slightly, suggesting that some Ti4+ species exposed at the surface are reduced in the presence of NH3. Moreover, new weak signals appeared at g = 2.006 and g = 1.993. The former can be assigned to electrons trapped in surface oxygen vacancies while the latter may be due to the perpendicular component of a Ti3+ site with a slightly different environment, the parallel component of which is hidden by the other signals at higher field.61 When the catalyst is subsequently exposed to a flow of 0.1% NO and 5% O2/He, the weak Ti3+ signal at g = 1.993 vanishes, the intensity of the rhombic Ti3+ EPR signal drops again due to reoxidation and its components shift slightly to g2 = 1.967 and g3 = 1.942. This may be due to slight changes in the coordination environment, possibly by filling of oxygen defects. In any case, the changes indicate that these Ti species are exposed on the surface and accessible by reactants.
image file: c8cy02193g-f8.tif
Fig. 8 In situ-EPR spectra of samples Ce0.5Ti0.5O2−δ and V/Ce0.5Ti0.5O2−δ recorded at 20 °C after (A) 1 h pretreatment in an O2 flow at 300 °C, (B) 30 min exposure to 0.1% NH3/Ar, (C) 30 min exposure to 0.1% NO and 5% O2/Ar and (D) 30 min exposure to the total SCR feed flow.

Upon subsequent switch to the total NH3-SCR feed flow, the initial shape of the rhombic Ti3+ signal is restored, and its intensity increases again. This indicates that it might be these Ti species that undergo a reversible redox Ti4+/Ti3+ cycle during NH3-SCR, thus, accounting for the significant catalytic activity of this bare support. However, it must be stressed that the observed changes in the Ti3+ signal are only small, suggesting that the major part of Ti3+ might be hidden in the bulk of the support and not accessible by reactants. Ti3+ species were not observed on the Ce0.5Ti0.5O2 support of our previous investigation, in which no CTAB surfactant and no hydrothermal treatment were used during catalyst synthesis and in which the surface area was significantly lower (113 m2 g−1).22 This indicates that the catalyst preparation route has a significant impact on both the catalytic performance and redox properties.

Despite the slight differences of the present and the previous Ce0.5Ti0.5O2−δ support,22 the catalytic performance and the behaviour of the respective V/Ce0.5Ti0.5O2−δ catalysts in the presence of different feed components are almost identical. As likewise observed previously,22 it is only the vanadium species for which a reaction-dependent redox cycle can be observed by operando EPR (Fig. 8). After pre-treatment in an O2 flow at 300 °C, only weak signals tentatively attributed to paramagnetic oxygen species are seen. Upon exposure to 0.1% NH3/Ar at 200 °C, the typical signal of reduced VO2+ appeared, which vanished again in a NO2/O2 flow due to reoxidation of these sites and reappeared again under the total SCR feed. In analogy to our previous study,22 we conclude that these redox-active vanadyl sites in close vicinity to Ce and Ti might govern the catalytic activity.

In the case of Mn-containing samples Ce0.5Ti0.3Mn0.2O2−δ and V/Ce0.5Ti0.3Mn0.2O2−δ, the hfs sextet of Mn2+ single sites (g = 2.004 and A = 90 G) remained unchanged in the EPR spectra under the different gas flows, suggesting that these species might be hidden in the bulk and do not participate in the NH3-SCR reaction (Fig. 9A and B). In contrast, the broad background EPR signal of interacting Mn2+ species in Ce0.5Ti0.3Mn0.2O2−δ “feels” the presence of the feed components. It increases slightly upon switching from Ar to the NH3/He flow. This is seen more clearly from the normalized intensity I/IAr plotted in Fig. 9C and suggests the formation of additional Mn2+ species, probably due to the reduction of EPR-silent Mn3+ present as Mn2O3 in the as-prepared sample (evident from the XRD pattern of Ce0.5Ti0.3Mn0.2O2−δ, Fig. SA). Upon switching to the NO/O2 flow, the intensity of the Mn2+ cluster signal dropped again, confirming the reversible reoxidation of Mn2+ to Mn3+ in these clusters (Fig. 9A and C). When the sample was exposed afterwards to the total SCR feed flow, the Mn2+ cluster signal became slightly broader and increased again, indicating that Mn3+ ions were again partly reduced. However, it did not reach the intensity observed under more reducing conditions in the NH3/Ar flow. This confirms clearly that a transient Mn2+/Mn3+ equilibrium is established under SCR conditions, due to the reversible redox behaviour of Mn ions within MnxOy agglomerates. This also suggests that it might be these clusters (and not Mn single sites) which boost the catalytic activity of supports with higher Mn content (Fig. 1). Note that the sample Ce0.5Ti0.45Mn0.05O2−δ, in which Mn single sites are dominant, is only marginally more active than the Mn-free support.

image file: c8cy02193g-f9.tif
Fig. 9 In situ-EPR spectra at 150 °C of the support Ce0.5Ti0.3Mn0.2O2−δ (A) and the catalyst V/Ce0.5Ti0.3Mn0.2O2−δ (B) after 1 h pretreatment in an O2 flow at 300 °C and cooling to 150 °C in an Ar flow, and subsequent exposure for 30 min each to 0.1% NH3/Ar, 0.1% NO, 5% O2/Ar and the total SCR feed flow (flushing with Ar between subsequent steps); (C) relative total EPR intensity (double integrals, normalized on the intensity after pretreatment in O2 and cooling to 150 °C in an Ar flow) at 150 °C.

A similar behaviour was observed for the V-containing V/Ce0.5Ti0.3Mn0.2O2−δ catalyst (Fig. 9B and C). However, compared to the V-free support, the spectra of this catalyst differ more significantly in intensity during subsequent treatment in NH3/Ar, NO/O2, and the SCR feed. This might be due to the additional participation of VOx species in the catalytic redox cycle, in which they shuttle between EPR-silent VO3+ and EPR-active VO2+, as observed accordingly for the Mn-free V/Ce0.5Ti0.5O2−δ catalyst (Fig. 8). However, in contrast to the latter, no signal of VO2+ single sites with a hyperfine structure is observed in Fig. 9B, though the formation of reduced VO2+ species in the NH3-containing feed is indicated, too, in the DRIFT spectra by the decreased intensity of the ν(V[double bond, length as m-dash]O) band at 2044 cm−1 (Fig. S8). The reason might be the magnetic interaction between these VO2+ species and the MnOx moieties, which prevents resolution of hfs. Thus, the missing hfs of the VO2+ signal in the Mn-containing catalysts is a clear indication of the close vicinity of MnOx and VOx species, which might be a reason for the higher activity of these samples compared to that of Mn-free V/Ce0.5Ti0.5O2−δ (cf.Fig. 1). Moreover, the lower total EPR intensity of the V-containing catalysts in comparison with that of the V-free supports (Fig. 9), observed already in the ex situ EPR spectra of the as-prepared catalysts (cf.Fig. 4B), is another indication of the effective magnetic interaction and, thus, the close vicinity of MnOx and VOx. In contrast to the V-free Ce0.5Ti0.3Mn0.2O2−δ support, the EPR signal of the V/Ce0.5Ti0.3Mn0.2O2−δ catalyst grows when switching from NH3/Ar to the total SCR feed, suggesting that under the more reducing NH3/Ar flow, some V3+ species may have been formed that are reoxidized to VO2+ in the SCR feed and contribute to the operando EPR signal (Fig. 9B and C). In general, the behaviour of the supports and catalysts with lower Mn content was similar to that of Ce0.5Ti0.3Mn0.2O2−δ and V/Ce0.5Ti0.3Mn0.2O2−δ, respectively. Therefore, the spectra of these catalysts are only shown in the ESI (Fig. S11 and S12), but their relative total intensities (double integrals) are plotted in Fig. 9C as well.

Deriving a direct relationship of the behaviour of the total relative EPR intensity (Fig. 9C) with the catalytic activity (Fig. 1) is difficult since the former reflects the simultaneous redox behaviour of both Mn2+ and VO2+, between which a discrimination is not possible. Also, the magnetic interaction between neighbouring Mn2+ and V4+ may reduce the total relative EPR intensity. Thus, for example, it cannot be decided without doubt, whether the lower total relative EPR intensity of the catalyst V/Ce0.5Ti0.3Mn0.2O2−δ compared to that of V/Ce0.5Ti0.4Mn0.1O2−δ under SCR conditions is due to more pronounced magnetic interactions between Mn2+ and V4+ or to a higher percentage of Mn3+/Mn4+ and V5+. However, one can make an educated guess: while the Mn/Ce surface ratio is the same for both catalysts, the V/(Ce + Ti + Mn) surface ratio is lower for V/Ce0.5Ti0.4Mn0.1O2−δ than for V/Ce0.5Ti0.3Mn0.2O2−δ (Table 2), suggesting that more VOx is dispersed on the latter surface in contact with MnOx. Taking the reduction potentials of Mn3+/Mn2+ (1.542 V) and VO3+/VO2+ (0.957 V) into account,28 a higher percentage of V5+ may be present under SCR conditions in the most active catalyst V/Ce0.5Ti0.3Mn0.2O2−δ as a consequence of electron transfer between adjacent V4+ and Mn3+. In any case, EPR spectroscopy reveals, without doubt, a tight interaction between surface VOx and MnOx species.

3.4. Synergistic effect of V and Mn

From the catalytic tests it is evident that the introduction of Mn into Ce0.5Ti0.5O2−δ increases the activity but at the same time reduces the N2 selectivity (Fig. 1). Furthermore, increasing amounts of Mn lead to severe and irreversible deactivation in the presence of H2O and SO2 (Fig. 2). When vanadia is deposited on the surface of the Ce0.5Ti0.5−xMnxO2−δ supports, there is only a marginal increase in activity, but a remarkable improvement of N2 selectivity and, particularly, of stability. This beneficial effect might arise from the synergistic interaction between Mn and V surface species.

XPS measurements have shown that the V/(Ce + Mn + Ti) surface ratio is highest for the most active catalyst with the highest Mn content, suggesting that the highly active but less selective and stable Mn surface sites might be partially covered by V. The tight interaction between Mn and V is also evident from the TPR measurements showing a higher reducibility of the V/Ce0.5Ti0.5−xMnxO2−δ catalysts, as well as from the EPR spectra in which effective magnetic interaction of Mn2+ with VO2+ in its vicinity lowers the signal intensity of the former in comparison with the V-free support (Fig. 4B).

Since the SCR reaction requires both oxidation and reduction steps, an ideal catalyst should contain components with optimally aligned redox potential. The redox potential of the metal ions in the V2O5/Ce0.5Ti0.5−xMnxO2−δ catalysts decreases in the order Ce4+/Ce3+(1.72 V) > Mn3+/Mn2+ (1.54 V) > Mn4+/Mn2+ (1.22 V) > VO3+/VO2+ (0.96 V) > Ti4+/Ti3+ (−0.06 V).28 This means that Mn has a higher tendency than V to obey a lower oxidation state and, thus, to take up electrons. Given that in V2O5/Ce0.5Ti0.5−xMnxO2−δ catalysts VOx is enriched on the surface covering partly the highly active but less selective MnOx surface species (evidenced by XPS), it is probable that the initial activation of NH3 to adsorbed NH2 converts V5+ to V4+ (Scheme 1A → B). Due to the higher redox potential of Mn, the fast electron transfer from V4+ to neighbouring Mn3+ (Scheme 1B → C) might form Mn2+ (visible by EPR, Fig. 9), followed by further NH3 activation and V4+ formation (visible by FTIR, Fig. S8). Reoxidation by O2 restores the initial state of the catalyst surface making it ready for the next catalytic cycle (Scheme 1C → D). As we have discussed previously, the Mars–van Krevelen mechanism of NH3-SCR implies that active metal ions reduced by NH3 are reoxidized by lattice oxygen while the formed oxygen vacancies are subsequently refilled by oxygen from the gas phase.22 It is probable that the oxygen vacancy created by reduction in the immediate vicinity of V and Mn is first filled by a nearby lattice oxide, leading to another vacancy at a certain distance which is then replenished by gas phase oxygen and so on. This requires that O vacancies and oxide ions can effectively travel through the solid (Scheme 1D). This is known as a unique property of ceria, which makes it an outstanding support for NH3-SCR catalysts.

image file: c8cy02193g-s1.tif
Scheme 1 Synergistic redox effect of V and Mn in NH3-SCR of NO.

When uncovered by vanadia, MnOx surface species with their higher redox potential tend to oxidize NH3 to N2O and SO2 to SO42− and/or HSO4 with reduced N2 selectivity and stability. The synergistic effect between V and Mn (Scheme 1) obviously modulates the total surface redox potential which suppresses undesired overoxidation while retaining high activity.

4. Conclusions

Catalytic tests revealed that the performance of bare Ce0.5Ti0.5−xMnxO2−δ supports in NH3-SCR of NO increased with increasing Mn content. Appreciable activity was already achieved with medium Mn contents (x = 0.05–0.10), reaching both NO and NH3 conversions above 85% between 175 to 300 °C at remarkably high N2 selectivities of ≥95%. Deposition of vanadia on Mn-containing supports does not lead to higher NH3 and NO conversions, as has been observed for Mn-free Ce0.5Ti0.5O2. However, it improves N2 selectivity which remained above 98% in the whole temperature range for Mn percentages of ≤0.1 and above 95% for catalyst V/Ce0.5Ti0.3Mn0.2O2−δ with the highest Mn content. This suggests that high activity is brought about by Mn incorporation while undesired NH3 combustion to N2O, i.e. low N2 selectivity, is suppressed by vanadia. As far as stability against feed gas pollutants is concerned, incorporation of Mn into the supports slows down deactivation by water which, furthermore, is completely reversible. A much stronger deactivation is caused by the simultaneous presence of H2O and SO2 for all supports (probably due to formation of MnSO4 and/or Mn(HSO4)2 surface species), which is partly suppressed by coverage of the MnOx species with vanadia.

XRD and Raman data suggest that the lattice disorder and the number of oxygen vacancies increase with increasing Mn content in the solid Ce0.5Ti0.5−xMnxO2−δ solutions. It is supposed that this improves oxygen mobility and in turn catalytic activity since the initial step in NO reduction is its temporal oxidation by a Mars–van Krevelen mechanism. This is also supported by TPR results, which point to the increase in reducibility of the catalysts and bare supports with increasing Mn content, as well as by in situ DRIFTS results, which evidence the formation of surface nitrates, yet only in Mn-containing catalysts but not in Mn-free V/Ce0.5Ti0.5O2.

In summary, it appears that Mn is the major factor for increasing the low-temperature activity of VOx/Ce0.5Ti0.5−xMnxO2−δ catalysts beyond that of the corresponding Mn-free VOx/Ce0.5Ti0.5O2 material, while the role of VOx species is first and foremost NH3 adsorption and activation, demonstrated by in situ DRIFTS in accordance with our previous study on Mn-free VOx/Ce1−xTixO2 catalysts.22 Moreover, the tight contact between MnOx and VOx surface species in the VOx/Ce0.5Ti0.5−xMnxO2−δ catalysts (evidenced by EPR) is assumed to modulate the total redox potential of the surface, which improves N2 selectivity and stability since overoxidation to N2O and SO42−/HSO4 is suppressed.

Conflicts of interest

There are no conflicts of interest to declare.


Dr. Thanh Huyen Vuong acknowledges a grant from the Vietnamese Ministry of Education and Training. We thank Dr. Giovanni Agostini and Dr. Matthias Schneider for experimental help in the XPS and XRD measurements.


  1. C. J. Pereira and M. D. Amiridis, ACS Symp. Ser., 1995, 587, 1–13 CrossRef CAS.
  2. V. P. Aneja, P. A. Roelle, G. C. Murray, J. Southerland, J. W. Erisman, D. Fowler, W. A. H. Asman and N. Patni, Atmos. Environ., 2001, 35, 1903–1911 CrossRef CAS.
  3. H. Bosch and F. Janssen, Catalytic Reduction of Nitrogen Oxides: A Review on the Fundamentals and Technology, Elsevier, 1998 Search PubMed.
  4. K. Skalska, J. S. Miller and S. Ledakowicz, Sci. Total Environ., 2010, 408, 3976–3989 CrossRef CAS PubMed.
  5. F. Nakajima and I. Hamada, Catal. Today, 1996, 29, 109–115 CrossRef CAS.
  6. M. Koebel, M. Elsener and G. Madia, Ind. Eng. Chem. Res., 2001, 40, 52–59 CrossRef CAS.
  7. P. Forzatti and L. Lietti, Chim. Ind., 1996, 78, 685–691 CAS.
  8. S. Zhang, B. Zhang, B. Liu and S. Sun, RSC Adv., 2017, 7, 26226–26242 RSC.
  9. J. Li, H. Chang, L. Ma, J. Hao and R. T. Yang, Catal. Today, 2011, 175, 147–156 CrossRef CAS.
  10. P. G. Smirniotis, D. A. Peña and B. S. Uphade, Angew. Chem., Int. Ed., 2001, 40, 2479–2482 CrossRef CAS PubMed.
  11. T. Boningari, P. R. Ettireddy, A. Somogyvari, Y. Liu, A. Vorontsov, C. A. McDonald and P. G. Smirniotis, J. Catal., 2015, 325, 145–155 CrossRef CAS.
  12. C. Liu, J. W. Shi, C. Gao and C. Niu, Appl. Catal., A, 2016, 522, 54–69 CrossRef CAS.
  13. C. Liu, G. Gao, J.-W. Shi, C. He, G. Li, N. Bai and C. Niu, Catal. Commun., 2016, 86, 36–40 CrossRef CAS.
  14. S. Yang, Y. Liao, S. Xiong, F. Qi, H. Dang, X. Xiao and J. Li, J. Phys. Chem. C, 2014, 118, 21500–21508 CrossRef CAS.
  15. J. Zuo, Z. Chen, F. Wang, Y. Yu, L. Wang and X. Li, Ind. Eng. Chem. Res., 2014, 53, 2647–2655 CrossRef CAS.
  16. P. Zhang, Y. Sun, W. Su, Y. Wei and J. Liu, RSC Adv., 2016, 6, 107270–107277 RSC.
  17. Z. Liu, J. Zhu, J. Li, L. Ma and S. I. Woo, ACS Appl. Mater. Interfaces, 2014, 6, 14500–14508 CrossRef CAS PubMed.
  18. H. Chang, X. Chen, J. Li, L. Ma, C. Wang, C. Liu, J. W. Schwank and J. Hao, Environ. Sci. Technol., 2013, 47, 5294–5301 CrossRef CAS PubMed.
  19. L. J. France, Q. Yang, W. Li, Z. Chen, J. Guang, D. Guo, L. Wang and X. Li, Appl. Catal., B, 2017, 206, 203–215 CrossRef CAS.
  20. X. Zhao, L. Huang, H. Li, H. Hu, X. Hu, L. Shi and D. Zhang, Appl. Catal., B, 2016, 183, 269–281 CrossRef CAS.
  21. S. Ali, L. Chen, F. Yuan, R. Li, T. Zhang, S. u. H. Bakhtiar, X. Leng, X. Niu and Y. Zhu, Appl. Catal., B, 2017, 210, 223–234 CrossRef CAS.
  22. T. H. Vuong, J. Radnik, J. Rabeah, U. Bentrup, M. Schneider, H. Atia, U. Armbruster, W. Grünert and A. Brückner, ACS Catal., 2017, 7, 1693–1705 CrossRef CAS.
  23. P. R. Ettireddy, N. Ettireddy, T. Boningari, R. Pardemann and P. G. Smirniotis, J. Catal., 2012, 292, 53–63 CrossRef CAS.
  24. B. Beck, M. Harth, N. G. Hamilton, C. Carrero, J. J. Uhlrich, A. Trunschke, S. Shaikhutdinov, H. Schubert, H.-J. Freund, R. Schlögl, J. Sauer and R. Schomäcker, J. Catal., 2012, 296, 120–131 CrossRef CAS.
  25. E. Tronconi, I. Nova, C. Ciardelli, D. Chatterjee and M. Weibel, J. Catal., 2007, 245, 1–10 CrossRef CAS.
  26. G. Busca, L. Lietti, G. Ramis and F. Berti, Appl. Catal., B, 1998, 18, 1–36 CrossRef CAS.
  27. I. E. Wachs, Dalton Trans., 2013, 42, 11762–11769 RSC.
  28. CRC Handbook of Chemistry and Physics, ed. D. R. Lide, CRC Press, Boca Raton, FL, 2005, pp. 12–16 Search PubMed.
  29. A. Bumajdad, J. Eastoe and A. Mathew, Adv. Colloid Interface Sci., 2009, 147–148, 56–66 CrossRef CAS PubMed.
  30. T. Sukonket, A. Khan, B. Saha, H. Ibrahim, S. Tantayanon, P. Kumar and R. Idem, Energy Fuels, 2011, 25, 864–877 CrossRef CAS.
  31. S. Wu, X. Yao, L. Zhang, Y. Cao, W. Zou, L. Li, K. Ma, C. Tang, F. Gao and L. Dong, Chem. Commun., 2015, 51, 3470–3473 RSC.
  32. T. H. Vuong, J. Radnik, E. Kondratenko, M. Schneider, U. Armbruster and A. Brückner, Appl. Catal., B, 2016, 197, 159–167 CrossRef CAS.
  33. J. I. Langford and A. J. C. Wilson, J. Appl. Crystallogr., 1978, 11, 102–113 CrossRef CAS.
  34. J. H. Scofield, J. Electron Spectrosc. Relat. Phenom., 1976, 8, 129–137 CrossRef CAS.
  35. W. Li, R.-t. Guo, S.-x. Wang, W.-g. Pan, Q.-l. Chen, M.-y. Li, P. Sun and S.-m. Liu, RSC Adv., 2016, 6, 82707–82715 RSC.
  36. E. Garcia-Bordeje, J. L. Pinilla, M. J. Lazaro and R. Moliner, Appl. Catal., B, 2006, 66, 281–287 CrossRef CAS.
  37. Y. Jangjou, D. Wang, A. Kumar, J. Li and W. S. Epling, ACS Catal., 2016, 6, 6612–6622 CrossRef CAS.
  38. W. Yang, J. Zhang, Q. Ma, Y. Zhao, Y. Liu and H. He, Sci. Rep., 2017, 7, 1–14 CrossRef PubMed.
  39. M. Luo, J. Chen, L. Chen, J. Lu, Z. Feng and C. Li, Chem. Mater., 2001, 13, 197–202 CrossRef CAS.
  40. Z. Wang, G. Shen, J. Li, H. Liu, Q. Wang and Y. Chen, Appl. Catal., B, 2013, 138–139, 253–259 CrossRef CAS.
  41. K. S. W. Sing, D. H. Everett, R. A. W. Haul, L. Moscou, R. A. Pierotti, J. Rouquerol and T. Siemieniewska, Pure Appl. Chem., 1985, 57, 603–619 CAS.
  42. B. M. Reddy, A. Khan, Y. Yamada, T. Kobayashi, S. Loridant and J.-C. Volta, J. Phys. Chem. B, 2003, 107, 5162–5167 CrossRef CAS.
  43. B. M. Reddy, P. Lakshmanan and A. Khan, J. Phys. Chem. B, 2004, 108, 16855–16863 CrossRef CAS.
  44. C. Schilling and C. Hess, J. Phys. Chem. C, 2018, 122, 2909–2917 CrossRef CAS.
  45. L. Liu, Z. Yao, B. Liu and L. Dong, J. Catal., 2010, 275, 45–60 CrossRef CAS.
  46. K. J. Moreno, A. F. Fuentes, M. Maczka, J. Hanuza and U. Amador, J. Solid State Chem., 2006, 179, 3805–3813 CrossRef CAS.
  47. L. S. M. Traqueia, T. Pagnier and F. M. B. Marques, J. Eur. Ceram. Soc., 1997, 17, 1019–1026 CrossRef CAS.
  48. P. R. Ettireddy, N. Ettireddy, S. Mamedov, P. Boolchand and P. G. Smirniotis, Appl. Catal., B, 2007, 76, 123–134 CrossRef CAS.
  49. Z. Wu, M. Li and S. H. Overbury, ChemCatChem, 2012, 4, 1653–1661 CrossRef CAS.
  50. Z. Wu, A. J. Rondinone, I. N. Ivanov and S. H. Overbury, J. Phys. Chem. C, 2011, 115, 25368–25378 CrossRef CAS.
  51. H. Hu, S. Cai, H. Li, L. Huang, L. Shi and D. Zhang, ACS Catal., 2015, 5, 6069–6077 CrossRef CAS.
  52. E. S. Ilton, J. E. Post, P. J. Heaney, F. T. Ling and S. N. Kerisit, Appl. Surf. Sci., 2016, 366, 475–485 CrossRef CAS.
  53. M. C. Biesinger, B. P. Payne, A. P. Grosvenor, L. W. M. Lau, A. R. Gerson and R. S. C. Smart, Appl. Surf. Sci., 2011, 257, 2717–2730 CrossRef CAS.
  54. E. Bêche, P. Charvin, D. Perarnau, S. Abanades and G. Flamant, Surf. Interface Anal., 2008, 40, 264–267 CrossRef.
  55. J. F. Moulder, W. F. Stickle, P. E. Sobol and K. D. Bomben, Handbook of x-ray photoelectron spectroscopy : a reference book of standard spectra for identification and interpretation of XPS data, Physical Electronics, Eden Prairie, Minn., c1995., 1995 Search PubMed.
  56. S. Watanabe, X. Ma and C. Song, J. Phys. Chem. C, 2009, 113, 14249–14257 CrossRef CAS.
  57. M. C. Biesinger, L. W. M. Lau, A. R. Gerson and R. S. C. Smart, Appl. Surf. Sci., 2010, 257, 887–898 CrossRef CAS.
  58. C. Naccache, P. Meriaudeau, M. Che and A. J. Tench, Trans. Faraday Soc., 1971, 67, 506–512 RSC.
  59. M. Che and A. J. Tench, in Adv. Catal., ed. H. P. D. D. Eley and B. W. Paul, Academic Press, 1983, vol. 32, pp. 1–148 Search PubMed.
  60. J. M. Coronado, A. Javier Maira, A. Martinez-Arias, J. C. Conesa and J. Soria, J. Photochem. Photobiol., A, 2002, 150, 213–221 CrossRef CAS.
  61. S. Yamazoe, K. Teramura, Y. Hitomi, T. Shishido and T. Tanaka, J. Phys. Chem. C, 2007, 111, 14189–14197 CrossRef CAS.
  62. B. Murugan, A. V. Ramaswamy, D. Srinivas, C. S. Gopinath and V. Ramaswamy, Chem. Mater., 2005, 17, 3983–3993 CrossRef CAS.
  63. H. Lettenmayr, W. Jantsch and L. Palmetshofer, Solid State Commun., 1987, 64, 1253–1255 CrossRef CAS.
  64. E. Tronconi, I. Nova, C. Ciardelli, D. Chatterjee and M. Weibel, J. Catal., 2006, 245, 1–10 CrossRef.
  65. X. Yao, R. Zhao, L. Chen, J. Du, C. Tao, F. Yang and L. Dong, Appl. Catal., B, 2017, 208, 82–93 CrossRef CAS.
  66. P. Li, Y. Xin, Q. Li, Z. Wang, Z. Zhang and L. Zheng, Environ. Sci. Technol., 2012, 46, 9600–9605 CrossRef CAS PubMed.
  67. H. Bosch, B. J. Kip, J. G. van Ommen and P. J. Gellings, J. Chem. Soc., Faraday Trans. 1, 1984, 80, 2479–2488 RSC.
  68. Y. Peng, C. Wang and J. Li, Appl. Catal., B, 2014, 144, 538–546 CrossRef CAS.
  69. A. A. Davydov, Molecular spectroscopy of oxide catalyst surfaces, John Wiley & Son Ltd, The Atrium, Southern Gate, Chichester, England, 2003 Search PubMed.
  70. L. J. Burcham, G. Deo, X. Gao and I. E. Wachs, Top. Catal., 2000, 11, 85–100 CrossRef.
  71. Y. Denkwitz, A. Karpenko, V. Plzak, R. Leppelt, B. Schumacher and R. J. Behm, J. Catal., 2007, 246, 74–90 CrossRef CAS.
  72. G. N. Vayssilov, M. Mihaylov, P. S. Petkov, K. I. Hadjiivanov and K. M. Neyman, J. Phys. Chem. C, 2011, 115, 23435–23454 CrossRef CAS.
  73. K. I. Hadjiivanov, Catal. Rev.: Sci. Eng., 2000, 42, 71–144 CrossRef CAS.
  74. D. García Pintos, A. Juan and B. Irigoyen, J. Phys. Chem. C, 2013, 117, 18063–18073 CrossRef.
  75. M. Y. Mihaylov, E. Z. Ivanova, H. A. Aleksandrov, P. S. Petkov, G. N. Vayssilov and K. I. Hadjiivanov, Appl. Catal., B, 2015, 176–177, 107–119 CrossRef CAS.
  76. H. Hu, K. Zha, H. Li, L. Shi and D. Zhang, Appl. Surf. Sci., 2016, 387, 921–928 CrossRef CAS.


Electronic supplementary information (ESI) available. See DOI: 10.1039/c8cy02193g

This journal is © The Royal Society of Chemistry 2018