Reverse Micelle Strategy for the Synthesis of MnO_x–TiO₂ Active Catalysts for NH₃-Selective Catalytic Reduction of NO_x at Both Low Temperature and Low Mn Content

Abstract

MnO_x–TiO₂ catalysts (0, 1, 5, and 10 wt % Mn nominal content) for NH₃-SCR (selective catalytic reduction) of NO_x have been synthesized by the reverse micelle-assisted sol–gel procedure, with the aim of improving the dispersion of the active phase, usually poor when obtained by other synthesis methods (e.g., impregnation) and thereby lowering its amount. For comparison, a sample at nominal 10 wt % Mn was obtained by impregnation of the (undoped) TiO₂ sample. The catalysts were characterized by using an integrated multitechnique approach, encompassing X-ray diffraction followed by Rietveld refinement, micro-Raman spectroscopy, N₂ isotherm measurement at −196 °C, energy-dispersive X-ray analysis, diffuse reflectance UV–vis spectroscopy, temperature-programmed reduction technique, and X-ray photoelectron spectroscopy. The obtained results prove that the reverse micelle sol–gel approach allowed for enhancing the catalytic activity, in that the catalysts were active in a broad temperature range at a substantially low Mn loading, as compared to the impregnated catalyst. Particularly, the 5 wt % Mn catalyst showed the best NH₃-SCR activity in terms of both NO_x conversion (ca. 90%) and the amount of produced N₂O (ca. 50 ppm) in the 200–250 °C temperature range.

1. Introduction

Nitrogen oxides (NO_x) are still the major air pollutants that are discharged by both combustion engines, as the primary mobile source, and by power plants and nitric acid production plants, as stationary ones.^1,2 To reduce NO_x emissions, multiple denitrification (deNO_x) technologies have emerged; among them, the selective catalytic reduction (SCR) with ammonia (NH₃-SCR) is one of the most reliable and common methods due to NO_x reduction efficiency, durable performance, and reasonable costs.³⁻⁵

TiO₂-supported V₂O₅ catalysts, promoted by either WO₃ or MoO₃, have been recognized as the most important commercial system in NH₃-SCR, in the name of their high activity and resistance to SO₂ poisoning.^4,6−8 However, the disadvantages of the narrow range of operating temperature (300–400 °C) and of the fast and irreversible deactivation above 450 °C lead to a pressing need for SCR catalysts characterized by both low cost and low T efficiency.⁶⁻⁹

Various transition metal oxides have also been reported to efficiently catalyze SCR reactions at low T, where the doping metal content reaches up to 30 wt %.¹⁰ Manganese oxides (MnO_x) have shown excellent performance, mostly ascribed to various types of labile oxygen species and Mn oxidation states (Mn²⁺, Mn³⁺, and Mn⁴⁺), which can have a significant role in accomplishing the catalytic cycle enhancing the NO oxidation, the nitrate formation, and the NH₃ activation steps.¹¹⁻¹⁵ Furthermore, Mn is less toxic than other transition metals like V, Co, and Ni and is highly abundant in soil.

The NH₃-SCR cycle typically involves the adsorption and activation of both NO and NH₃ molecules on the catalyst active sites.^10,16 Optimizing the interaction between the two activated molecules is crucial to avoid undesired parallel reactions, most notably NH₃ overoxidation and N₂O production.^17,18

The need of satisfactory catalytic activity can be fulfilled by tailoring the textural features of the support and the redox properties of the active species. The main principle is that the catalyst should be dispersed on a suitable support to obtain stable active species. In addition, it is important that there is adequate metal-support interaction and optimum performance at low metal loading, which consequently reduces the cost of the catalyst. Impregnation and coprecipitation methods are applied to produce a wide range of materials but usually provide a poor control on the formed species and their dispersion and on their interaction with the support. Among the different oxides used as a support for the SCR catalyst, TiO₂ offers excellent dispersion of the active components, proving to be an ideal candidate support for Mn-based low-temperature SCR catalysts.¹⁵

Ettireddy et al. explored the effect of Mn loading by wet impregnation of different TiO₂ phases: anatase was identified as the best support with an optimal Mn content of about 17 wt %, corresponding to monolayer coverage.¹⁹ On the other hand, plentiful papers in the literature report that significant NO_x conversion and N₂ selectivity are obtained by using catalysts having Mn loading larger than 15 wt %.^20,21

Bimetallic systems are often used to improve the catalytic performance. Putluru et al. prepared a series of Mn/TiO₂ and Fe-promoted Mn/TiO₂ catalysts by both precipitation and impregnation methods, and the results were discussed by directly comparing their activity with that of commercial V₂O₅–WO₃/TiO₂ (3 wt % V₂O₅, 7 wt % WO₃).²² The catalysts prepared by precipitation exhibited superior low temperature activity with respect to the impregnated ones, reaching 96% NO conversion with an Fe-promoted Mn/TiO₂ catalyst. Nevertheless, the required total active metal content was 25 wt %, whereas NO conversion values dropped to 65% for a metal content below 10 wt %.

In this scenario, the sol–gel method is an outstanding route providing a potential control of the doping process. Sol–gel is a wet chemical process where a liquid “sol” transforms into a “gel” phase by competition, at the molecular level, between the reactions responsible for the formation and the growth of colloidal particles.

Its versatility and mild conditions allow for tailoring the textural and surface properties through the synthesis parameters, making the sol–gel processing a powerful tool for the synthesis of materials.²³⁻³⁰

A superior dispersion with no detectable sintering or formation of MnO_x crystalline phases up to a Mn/Ti molar ratio equal to 0.4 was obtained by traditional sol–gel preparation. Nevertheless, the initial expectations of an improved catalytic activity at Mn loading below 20 wt % were not met.³¹ The same authors also explored the effect of transition metals in addition to Mn/TiO₂-based catalysts for low temperature SCR. Activity was enhanced by adding Fe [Fe(0.1)–Mn(0.4)/TiO₂] via the one pot sol–gel procedure, and the NO conversion was more than 90% at about 100 °C.¹³

The work by Kim et al.³² aimed at investigating the effect of the preparation method on the deNO_x activity of the Mn/TiO₂-based catalyst. The sol–gel approach allowed for reaching a better deNO_x performance as compared to an impregnated Mn/TiO₂ catalyst in the 150–350 °C temperature range. At 250 °C, 90% NO conversion was achieved with the sol–gel catalyst at the highest Mn content (30 wt %). The observed enhancement of the deNO_x activity due to the increase in the Mn content was ascribed to the remarkable dispersion of MnO₂ on the catalyst surface by the efficient incorporation of Mn into the TiO₂ matrix during the sol–gel preparation.³²

Although sol–gel catalysts are more efficient with respect to those obtained by conventional methods (e.g., impregnation), several shortcomings of the reported systems still need to be addressed, including the narrow temperature range, the substantial Mn content, and to some extent, the low N₂ selectivity.

Among the different types of chemical processes that may be considered under the heading of “sol–gel”, the reverse micelle approach provides a favorable microenvironment to control the nucleation and growth of nanoparticles (NPs), simultaneously enhancing the metal dispersion. Reverse micelles are nanometric droplets of the aqueous phase stabilized by surfactants in an organic phase, and the rate of hydrolysis and condensation are driven by the frequency of intermicellar exchanges and by the water content in the micelle core, in addition to the structure-directing role of the surfactant. The supramolecular template sol–gel technique overcomes the obstacles encountered in more traditional procedures, where the high hydrolysis rate makes the handling of transition metal alkoxides not straightforward. In addition, transition metals often exhibit several (stable) coordination states, and when coordinatively unsaturated, they are able to expand their coordination sphere by solvation or alkoxy bridging.³³ The vigorous reaction of Ti alkoxides with water leads to (undesired) formation of Ti oxo/hydroxo precipitates. The most common inhibitors are complexing ligands, β-diketones, polyhydroxy acids, or carboxylic acids, which reduce the reactivity of Ti alkoxides by increasing its coordination number and decreasing the number of easily hydrolyzable groups.³³

In this work, a one-pot reverse micelle strategy has been adopted to prepare TiO₂-supported MnO_x species at low metal content (namely, 1, 5, and 10 wt % Mn). A thorough physicochemical characterization was carried out by X-ray diffraction (XRD) followed by Rietveld refinement, microRaman spectroscopy, N₂ isotherm measurement at −196 °C, energy-dispersive X-ray (EDX) analysis, DR (diffuse reflectance) UV–vis spectroscopy, temperature-programmed reduction (TPR) technique, and X-ray photoelectron spectroscopy (XPS) to gain insights into both structural and surface properties of the MnO_x–TiO₂ catalysts. The activity in the SCR of NO_x with NH₃ was evaluated by directly comparing the sol–gel MnO_x–TiO₂ catalysts with a catalyst prepared by impregnating with the same Mn precursor the (undoped) TiO₂ prepared by the same reverse micelle sol–gel method. The results were discussed in relation with the structural and surface features, in order to explain the enhanced performance of the sol–gel-derived catalysts, in terms of both low T activity and low Mn content, obtained during the NH₃ SCR catalytic tests.

2. Results and Discussion

2.1. Structural and Textural Properties of the Catalysts

The diffraction patterns of the various catalysts, displayed in Figure 1, reveal the occurring phases. At a first glance, all the studied samples are multiphase, with anatase (A) as the main phase along with the presence of minor phases, which are labeled B (for brookite), R (for rutile), and P (for pyrolusite, i.e., β-MnO₂); furthermore, the pronounced broadness of the XRD peaks suggests a low degree of crystallinity, that is, a nanometric crystal size of the various phases.

Figure 1.

XRPD patterns of the prepared samples; the different crystalline phases are labeled as follows: (A) anatase, (B) brookite, (R) rutile, and (P) Pyrolusite.

The amount of each phase, the anatase cell parameters, and the crystal size were determined by Rietveld refinement (data shown in Table 1). Due to their low content, for the minor phases, the only refined parameter was their amount, whereas the other parameters were kept constant or equal to anatase, when possible. The plots of the refinements (data not shown) confirm the very good quality of all the refinements as confirmed by the values of sigma and R_wp (weighted profile R-factor) reported in Table 1.

Table 1. Determination of Anatase Cell Parameters, Different Phase Contents, and Crystal Size as Obtained by Rietveld Refinement.

	TiO2	Mn-1	Mn-5	Mn-10	Mn-10-IMP
a (Å) anatase	3.78898	3.789194	3.787906	3.78774	3.787105
b (Å) anatase	9.48788	9.483329	9.468461	9.463282	9.484697
cell volume (Å3)	136.21	136.16	135.86	135.77	136.03
anatase (wt %)	91	90.49	91.23	86.19	91.42
brookite (wt %)	9	9.51	8.77	10.7
rutile (wt %)				3.11
pyrolusite (wt %)					8.57
crystal size anatase (Å3)	163	179	153	162	190
sigma	1.98	1.63	1.71	1.57	1.89
Rwp	4.27	4.34	4.23	3.65	4.17

According to the data given in Table 1, the phase composition of the sol–gel samples is almost the same up to 5 wt % Mn, whereas with the Mn-10 catalyst, the appearance of rutile induces a non-negligible reduction of the anatase content. This finding is not surprising, considering that the Mn dopant addition is reported to promote the anatase to rutile (ATR) phase transformation.³⁴ With all these catalysts, no trace of Mn-containing crystalline phases was detected. The anatase lattice parameter suggests that at least a certain amount of Mn⁴⁺ ions has been dissolved in the anatase matrix by substituting Ti⁴⁺ cations. Figure 2 shows that TiO₂ and Mn-1 have a very similar unit cell, in agreement with the low Mn content of the latter sample, and then, a progressive shrinking of the unit cell occurs, perfectly compatible with a partial substitution of Ti⁴⁺ ions (octahedral Ti⁴⁺ has r = 0.605 ų⁵) by Mn⁴⁺ ions (octahedral Mn⁴⁺ has r = 0.53 ų⁵). Therefore, by increasing the amount of Mn in the sol–gel process, the metal tends to dissolve within the anatase phase. The impregnated sample has a completely different behavior; indeed, TiO₂ is present only as an anatase phase, likely due to the calcination step following the impregnation, whereas Mn occurs in the pyrolusite (β-MnO₂) phase. The amount of pyrolusite calculated by Rietveld refinement (almost 9 wt %) is very close to the nominal one (10 wt %), indicating that only a small (if any) fraction of Mn was dissolved in the TiO₂ matrix, in agreement with the adopted (impregnation) procedure.

Figure 2.

Anatase unit cell volume of the sol–gel prepared samples.

As a further confirmation, the anatase cell volume of Mn-10-IMP is similar to that of the TiO₂ support, likely indicating that in the impregnated sample, the two oxides crystallized independently and no solid solution of MnO₂ in TiO₂ formed. Therefore, the insertion on Mn inside the anatase lattice is possible only if the system is mixed at an atomic level (here, during the reverse micelle-assisted sol–gel synthesis) and not when mixing occurs at a large scale, as during impregnation.

The reported XRD results strongly evidence the effectiveness of the reverse micelle sol–gel approach where the nanometer-size aqueous systems promote the intimated contact between the precursors. The surfactant, indeed, provides a cagelike environment that favors an excellent control of the final powder stoichiometry with the possibility of obtaining homogeneity and mixing at the atomic scale, ultimately favoring the inclusion of Mn in the studied samples.

The Raman spectra of the prepared catalysts are shown in Figure 3a,b 100–275 and 275–1200 cm^–1 range: the typical anatase TiO₂ Raman scattering peaks (black line, Figure 3a,b) are seen at 144 (E_g), 198 (E_g), 320 (one of the weak two-phonon bands), 397 (B_1g), 516 (A_1g + B_1g), 639 (E_g), and 795 (B_1g) cm^-1.36 Some weak signals, due to the low content of brookite TiO₂ and superimposed to the anatase spectrum, are discernible at 247 (A_1g), 322 (B_1g), 366 (B_2g) cm^–1, as well. In addition, the known³⁶ strongest modes of brookite at 128 (A_1g) and 636 (A_1g) cm^–1 are completely hidden by the nearly coincident modes of anatase.

Figure 3.

Raman spectra of the following samples: TiO₂ (black curve); Mn-1 (red curve), Mn-5 (green curve), Mn-10 (blue curve), and Mn-10-IMP (orange curve). The spectra are reported in two (complementary) sections, (a) (100–275) and (b) (275–1200 cm^–1) in order to allow the reader to appreciate scattering bands with very different intensities.

For all the TiO₂ phases, the modes below 400 cm^–1 are ascribable to O–Ti–O bending vibrations and the ones above 400 cm^–1 are due to Ti–O stretching vibrations. Thus, both types of modes could be affected by the possible incorporation of Mn. In particular, the most sensitive one is the E_g peak at 144 cm^–1, as it is much responsive to local oxygen coordination surrounding the metal ion. Actually, as shown in Figure 3a, the blue shift (i.e., toward higher wavenumbers) of the 144 cm^–1 peak, which also becomes asymmetric and larger, increases with the Mn content for the sol–gel catalysts, owing to the disruption of the topological order in the TiO₂ matrix, related to the probable insertion of Mn ions. Moreover, such Mn species could generate point defects in the TiO₂ lattice and its contraction, in agreement with the evolution of such bands.³⁷ Moreover, by comparing the Mn-10 spectrum (blue line) with Mn-10-IMP (orange line), a shift of the 144 cm^–1 peak occurs (Figure 3a), similar to the one observed for the Mn-5 sample, but the peak does not lose its symmetry. For this reason, the broad anatase peak centered at 198 cm^–1 becomes more intense and, finally, Mn incorporation cannot be inferred by Raman spectroscopy.

Because the modes of the vibrations due to MnO₂ and Mn₂O₃ species are overlapped to those of TiO₂ in the same spectral region, a subtraction (Figure S1) of the undoped TiO₂ spectrum to each of the Mn-doped TiO₂ spectra has been performed, in order to single out any Mn-related bands. In Figure 3b, the 530 and 651 cm^–1 peaks (related to amorphous MnO₂ and Mn₂O₃, respectively) are visible with the Mn-1, Mn-5, and Mn-10 samples.³⁷ Their intensities increase with the Mn nominal content. With Mn-10, the 318 and 382 cm^–1 modes, assigned to Mn₂O₃, are observable,³⁸ and for Mn-5 and Mn-10, a broad band centered at 1050 cm^–1 related to amorphous MnO₂ appears. The Mn-10-IMP spectrum is quite superimposable to the undoped TiO₂ spectrum (no peak shift nor width enlargement are observable), and it does not show MnO_x-related modes, except for the very broad signal at 1050 cm^–1. Moreover, no trace of pyrolusite, instead detected by XRD, could be singled out (the reference RRUFF Project Raman pyrolusite spectrum being not reported), thus suggesting that the formed Mn oxides (detected by both XRD and DR-UV–vis spectroscopy, vide infra) are not homogenously dispersed on the surface of TiO₂ and/or that their amount is below the sensitivity of the Raman technique in the examined spots. This result confirms that impregnation does not allow a fine dispersion of Mn on the sample, whereas the sol–gel method does. Indeed, the EDX analysis (Table 2) shows Mn/Ti atomic ratios close to nominal values for the sol–gel catalysts, whereas the value determined with the impregnated sample is very high, likely due to the aforementioned heterogeneity induced by the impregnation technique.

Table 2. Textural Properties of the Studied Samples as Determined by N₂ Isotherms at −196 °C, FESEM and EDX Analyses.

sample	BET SSA (m2 g–1)	total pore volume (cm3 g–1)	micropore volumea (cm3 g–1)	EDX (nominal) Mn/Ti atomic ratio
TiO2	70	0.090	0.0002
Mn-1	57	0.096	0.0017	0.0164(0.0147)
Mn-5	76	0.109	0.0017	0.0656(0.0765)
Mn-10	68	0.102	0.0004	0.106(0.161)
Mn-10-IMP	40	0.070		0.604 (0.161)

^aAs obtained according to the t-plot method.

Table 2 also reports the textural properties of the catalysts as obtained by N₂ physisorption at −196 °C. The samples showed type IV isotherms (Figure S2) with a H2 hysteresis loop, characteristic of mesoporous solids with interparticle porosity. As expected, the impregnated sample shows a drop in both the surface area and pore volume, whereas the other samples show Brunauer–Emmett–Teller (BET) specific surface area (SSA) values in the 57–76 m² g^–1 range and total pore volume values in the 0.09–0.102 cm³ g^–1 range, indicating a limited effect of Mn doping on the sample porosity.

The field-emission scanning electron microscopy (FESEM) analysis (Figure S3) showed the occurrence of spherical NPs with uniform shape and size, forming large agglomerates/aggregates, likely due to NP coalescence after calcination. The Mn-10-IMP sample showed instead larger NPs but still with a quasispherical shape and forming large agglomerates/aggregates.

2.2. Characterization of Bulk and Surface Mn Species

The DR UV–vis spectra of the studied catalysts are reported in Figure 4. As expected, TiO₂ shows a typical UV–vis spectrum due to the CT (charge-transfer) transitions from O^2– to Ti⁴⁺: a band gap (E_g) of 3.01 eV has been calculated by applying the Tauc’s plot method and assuming an indirect semiconductor behavior (F(R)*(hυ))^1/2 in agreement with the fact that it mainly contains anatase (indirect semiconductor), with a minor amount of brookite (direct semiconductor). The DR UV–vis spectrum of Mn-10-IMP, instead, shows a broad absorption in the whole investigated range and resembles the literature spectra usually reported for MnO₂.³⁹ This result is in agreement with XRD patterns (vide supra), showing the occurrence of pyrolusite in the sample.

Figure 4.

DR UV–vis spectra of the following samples: TiO₂ (black curve); Mn-1 (red curve), Mn-5 (green curve), Mn-10 (blue curve), and Mn-10-IMP (orange curve).

Considering the Mn-doped catalysts obtained by sol–gel, the Mn-1 spectrum shows two broad and intense absorptions at about 420 and 560 nm, that is, in the region of Mn ion d–d transitions (360–800 nm). At higher Mn contents (i.e., Mn-5 and Mn-10 spectra), the two bands become more intense, another component is observed at ca. 390 nm, and simultaneously, the O^2– to Ti⁴⁺ CT band (at 350 nm in TiO₂) suffers a slight red shift (i.e., toward longer wavelength), likely due to Mn doping by inclusion in the TiO₂ lattice, as shown by XRD. The component at 390 nm is assigned in the literature to d–d transitions of octahedral Mn⁴⁺ ions,⁴⁰ whereas the other two bands are ascribed to different MnO_x species, likely amorphous (as confirmed by both XRD and Raman techniques) and located at the NP surface, in agreement with XPS results (vide infra).⁴¹

The catalysts prepared by sol–gel synthesis show a progressive decrease of the E_g value to 2.06 (Mn-1), 1.64 (Mn-5), and 2.00 eV, (Mn-10) confirming that, along with the formation of surface MnO_x species, bulk Mn doping occurs by using the adopted synthesis technique, in agreement with XRD results (vide supra). Moreover, this result is in agreement with both experimental⁴² and theoretical studies,⁴³ the latter showing that substitution of Ti⁴⁺ ions by Mn ions leads to a remarkable red shift of the optical absorption edge both by overall reduction of the band gap and introduction of curvy intermediate bands into the forbidden gap. The (unexpected) increase of the band gap energy with the Mn-10 sample may be due to an extensive doping, which can lead to the well-known Moss-Burnstein effect occurring at a high dopant content. Indeed, with Mo-doped TiO₂ samples with the similar metal content obtained by the same synthesis procedure,⁴⁴ the same effect was observed; this result is another proof of the effectiveness of the sol–gel synthesis in the doping of TiO₂ by heteroatom ions (here, Mn).

The surface chemical composition of the Mn-doped TiO₂ catalysts has been investigated by means of XPS, allowing for both a qualitative and quantitative analysis of the surface chemical composition. Table 3 reports the surface elemental composition (at. %) along with the Mn/Ti atomic ratios, as derived by the quantification of the O 1s, Mn 2p, and Ti 2p spectral regions.

Table 3. XPS Surface Elemental Composition of Mn-Doped TiO₂ Catalysts. The Values are Averaged out of at Least Three Replicates and are Reported as Mean Value ± 1S.

	XPS surface elemental composition (at. %)
Sample	Mn	Ti	O	Mn/Ti atomic ratio
Mn-1	1.1 ± 0.2	26.5 ± 1.6	72.4 ± 1.6	0.042 ± 0.004
Mn-5	2.6 ± 0.3	26.3 ± 1.1	71.1 ± 1.4	0.10 ± 0.01
Mn-10	3.5 ± 0.5	24 ± 2	73 ± 2	0.15 ± 0.02
Mn-10-IMP	10.5 ± 0.5	17.5 ± 1.0	72.3 ± 1.3	0.58 ± 0.04

The surface amount of Mn was found to increase with the nominal Mn content, as shown by both the absolute Mn at. % and the Mn/Ti atomic ratio values reported in Table 3, in agreement with a previous work concerning Mo-doped TiO₂ samples obtained by means of the same synthetic procedure.^44,45 Also, the impregnation procedure resulted in a higher surface Mn availability, as compared to the sol–gel process, as seen by comparing the Mn/Ti ratio of the Mn-10-IMP and the Mn-10 samples.

In order to gather information on the surface chemical speciation and Mn oxidation states, a curve-fitting procedure was applied to the Ti 2p, Mn 2p, and O 1s high-resolution XP spectra. The Ti 2p spectral region showed the typical spin–orbit splitting doublet of Ti 2p_3/2 and Ti 2p_1/2 species with binding energy (BE) and separation values consistent with Ti⁴⁺ in the TiO₂ lattice (Figure S4). The Mn 2p and O 1s XP regions (Figures 5 and 6, respectively) provided interesting results on the possible oxidation states of Mn and oxygen surface vacancies.

Figure 5.

Mn 2p XP spectra for all Mn-doped catalysts. Curve fitting has been applied only to the 2p_3/2 peak component.

Figure 6.

O 1s XP spectra and corresponding curve fittings as obtained with the samples (from top to bottom): Mn-1, Mn-5, Mn-10, and Mn-10-IMP.

Mn 2p XP spectra can be very complex due to the coupling of angular momenta associated with partially filled core and valence shells containing unpaired electrons. This can create a number of final states, which will manifest in the XP spectrum in the form of multiplet splitting. In particular, out of the six stable manganese oxidation states (0, +2, +3, +4, +5, and +7), three have significant multiplet splitting (Mn²⁺, Mn³⁺, and Mn⁴⁺), and thus, the qualitative and quantitative XPS investigation of Mn-containing samples is particularly challenging.^11,46−49 The separation of the 2p_3/2 and 2p_1/2 spin–orbit splitting components of Mn is large enough to consider only the (more intense) 2p_3/2 signal and its associated structure.

The curve fitting procedure applied to the Mn 2p_3/2 XP spectra of Mn-1 to Mn-10-IMP samples is reported in Figure 5. Due to the complexity of this signal, a first parameter to focus on could be the main peak shape because it presents some detailed features depending on the presence/prevalence of the specific oxidation state, as described by Biesinger et al. who reported Mn 2p spectra for various standard samples and discussed in detail the corresponding fitting parameters.⁴⁶ Comparing the Mn 2p spectra of the present work with those by Biesinger et al., it is possible to observe the occurrence of a shake-up satellite around 646.0 ± 0.2 eV, suggestive of surface Mn(II) species, whose relative abundance decreases as the total Mn % increases (see Table S1 in the Supporting Information file). It is close to 9% in the Mn-5 sample and goes down to 3% in the Mn-10-IMP sample. Also, a prominent shoulder at lower BE values (640.0 ± 0.2 eV), typical of Mn(III), could be identified in all the samples with a comparable relative abundance. With the Mn-10-IMP sample, it is also possible to identify a peak shape and a multiplet splitting strongly resembling those of Mn(III) species. The Mn 2p_3/2 XP spectrum of Mn-1 was not curve fitted, due to the very low signal-to-noise ratio, but the general peak shape is close to those of both Mn-5 and Mn-10. A further spectral feature is the 2p spin–orbit splitting value; it is shown to increase as the Mn oxidation state increases.¹⁷ The value for all our catalysts is always 11.6 ± 0.1 eV, compatible with the presence of both Mn(II) and Mn(III). All those findings indicated that a mix of Mn(II) and Mn(III) is generally distinguishable at the surface of the Mn-doped catalysts: Mn(II) species could be hypothesized to be more abundant in the case of the sol–gel samples and their relative amount decreasing at a higher Mn content; conversely, Mn(III) species are predominant at the surface of the impregnated catalyst. Moreover, due to complexity of Mn 2p multiplet splitting, the occurrence of small amounts of Mn(IV) species could not be excluded, especially in the Mn-10-IMP sample on the base of the main peak shape. All these experimental evidences are confirmed by TPR analysis (vide infra).

The curve-fitted O 1s XP spectra are reported in Figure 6; the curve fittings shown from three (with the Mn-1 and Mn-5 samples) to four peak components (with the Mn-10 and Mn-10-IMP samples). The lowest BE peak is due to lattice oxygen species bonded to Ti and/or Mn (529.7 ± 0.2 eV), whereas the BE peak at around 531.4 ± 0.3 eV is attributed to nonlattice oxygen/OH species. Finally, the component at 530.5 ± 0.1 eV is due to oxygen in oxygen-deficient regions, whereas the highest BE peak (at 533.2 ± 0.2 eV) is ascribed to adsorbed/structural water.^50,51 It is important to notice how the amount of oxygen vacancies increases with the dopant concentration, as can be derived by the relative abundance of the peak component at 530.5 ± 0.1 eV, which is around 10–11% in Mn-5 and increases up to 16% in Mn-10 and Mn-10-IMP. As a whole, XPS analysis showed the occurrence of surface Mn species at the variable oxidation state [Mn(II) species being mainly found on the samples prepared by sol–gel]. Correspondingly, oxygen vacancies are observed due to the charge balance. Their occurrence in the Mn-10-IMP sample shows that the surface composition of pyrolusite differs from that of its bulk because only Mn²⁺ and Mn³⁺ species are observed by XPS.

Moreover, comparison of the Mn/Ti atomic ratio at the surface (XPS) and in the bulk (EDX) shows the occurrence of a surface enrichment in Mn, as previously observed with another set of Mo-doped TiO₂ samples with the similar heteroatom content that were obtained by the same synthesis procedure.⁴⁴

The occurrence of different redox species and their dispersion at the catalyst surface is notoriously one of the main factors controlling the catalytic activity in the studied reaction. The redox ability of the catalysts was investigated by H₂-TPR, and the corresponding results are shown in Figure 7. The amount of H₂ consumption, reported in the following paragraphs, has been calculated in the range between room temperature and 500 °C.

Figure 7.

H₂-TPR profiles reported for the following samples: Mn-1 (red curve), Mn-5 (green curve), Mn-10 (blue curve), and Mn-10-IMP (orange curve). The left Y axis refers to the Mn-1, Mn-5, and Mn-10 samples; and the right Y axis refers to the Mn-10-IMP sample.

According to the literature, the reduction process of manganese oxides covers a broad temperature range where (bulk) MnO₂ is successively reduced to lower oxidation states in the order MnO₂ → Mn₂O₃ → Mn₃O₄ → MnO. The following T ranges are usually reported for the reduction of manganese oxides: 290–350 °C for MnO₂ reduction (but for pyrolusite, β-MnO₂, the T range increases up to 430 °C), 350–420 °C for Mn₂O₃ reduction to Mn₃O₄, and 450–500 °C for Mn₃O₄ reduction to MnO, which is not further reducible at the partial pressure of H₂ adopted during TPR analysis.^19,22,52,53 Reducibility is significantly affected by several other variables, like oxide dispersion and crystallite size, interaction with the support, synthesis method, and so forth.¹⁹

The undoped TiO₂ (not reported) showed some H₂ consumption, in agreement with the literature reporting that the support is not inert toward reduction.^54,55 In the TPR profile of the Mn-10-IMP catalyst (Figure 7), two sharp and well-defined reduction peaks were observed at ca. 325 and 435 °C, respectively. At higher T, no further peaks, ascribable to TiO₂ reduction, were identified. The TPR spectrum of Mn-10-IMP likely corresponds to the reduction of MnO₂ to MnO, suggesting the following two-step reduction process: MnO₂ → Mn₂O₃/Mn₃O₄ and Mn₂O₃/Mn₃O₄ → MnO.⁵⁶ The related H₂ consumption was 91.8 μmol g^–1, which corresponded to a Mn average oxidation state of 3.9, in good agreement with XRD results, where crystalline β-MnO₂ was detected. The Mn-10 sample also showed two reduction peaks in the 250–550 °C range, indicating a two-step reduction process; however, the peaks were shifted to higher T as compared to the Mn-10-IMP sample, indicating an overall lower average Mn oxidation state.⁵⁷ This hypothesis is confirmed by the H₂ consumption that for the Mn-10 sample dropped to 16.3 μmol g^–1, from which an average Mn oxidation state of 2.5 was calculated. Moreover, the shape of the overall TPR profile is very broad and structured as compared to the Mn-10-IMP sample, indicating the likely occurrence of several nonstoichiometric MnO_x species, in agreement with Raman spectroscopy.⁵⁷

Concerning the H₂ consumption above 500 °C, the peak at ca. 600 °C is due to the formation of Ti³⁺ species, forming oxygen deficient TiO_x species, whereas the minor H₂ consumption at c.a. 750 °C could be assigned to the reduction of the brookite phase, although in the same T range, other authors pointed out the occurrence of the ATR transition, which hampers a straightforward assignment in this case.⁵⁸⁻⁶⁰

The Mn-5 sample showed a H₂ consumption of 9.5 μmol g^–1 corresponding to a calculated Mn average oxidation state of 2.3. The two reduction peaks observed for Mn-5 are even more shifted to higher T compared to Mn-10, in agreement with a slightly lower average oxidation state.

The TPR spectrum of the Mn-1 sample showed very low intensity and the amount of consumed H₂ was 1.8 μmol g^–1, which corresponded to a Mn average oxidation state of 2.3.

As expected, the introduction of Mn by sol–gel synthesis induced a remarkable change in the TPR curve shape and in the H₂ consumption, with respect to the impregnation procedure, being responsible for the formation of a mixture of oxidation states.

The different reducibility of the MnO_x species, driven also by the strength of the interaction with the support, is responsible of the observed broadened TPR profile. However, the sol–gel method favors a strong Mn–TiO₂ interaction, likely enhancing the reduction of the MnO_x species with respect to unsupported Mn₂O₃ and Mn₃O₄ with a consequent shift of the reduction onset to lower T values.^10,27,61 This modulation of Mn reducibility could (positively) affect the NO_x catalytic conversion, thus impacting the SCR activity of the Mn-doped TiO₂ samples.

Another physico-chemical property, which may affect the NH₃-SCR catalytic activity, is surface acidity; Figure 8 shows the sample NH₃-TPD profiles in the 100–550 °C range. With respect to the samples prepared by the reverse micelles sol–gel method, the Mn-10-IMP sample showed a smaller amount of desorbed NH₃ (d_NH3) per g of the catalyst (i.e., 75 μmol g^–1), in agreement with its smaller SSA value, and a sharper TPD curve in the 125–350 °C range, indicating the occurrence of fewer and weaker acidic sites. The Mn-1 sample (d_NH3 = 242 μmol g^–1) showed a different shape in a broader T range (125–450 °C), indicating the presence of abundant and heterogeneous acidic sites.²⁰ The Mn-5 (d_NH3 = 244 μmol g^–1) and Mn-10 (d_NH3 = 244 μmol g^–1) samples showed very similar features not only in terms of the amount of acidic sites, but also in terms of the broad TPD curve profile extending in a wide T range, although Mn-5 showed an onset at slightly lower T. As a whole, the Mn-1, Mn-5, and Mn-10 TPD profiles indicated the occurrence of different types of NH₃ adsorbed species with different thermal stabilities at the catalyst surface.⁶² Concerning the nature of those acidic sites, that is, Brønsted or Lewis sites, NH₄⁺ ions decompose at a lower T with respect to NH₃-Lewis acid site adducts; in agreement with the literature, the desorption at low T (below 200 °C) may be assigned to NH₃ molecules obtained by the decomposition of NH₄⁺ ions bound to Brønsted acid sites and the desorption at high T is associated with NH₃ molecules coordinated to Lewis acid sites. The strength of such Lewis acid sites is weak to medium by considering the desorption T range.⁶³ Brønsted acid sites are also important for the SCR catalytic activity, and they seem to be more abundant with the Mn-5 sample, showing a desorption onset at a slightly lower T and a component at ca. 170 °C.

Figure 8.

NH₃-TPD curves of Mn-1 (red curve), Mn-5 (green curve), Mn-10 (blue curve), and Mn-10-IMP (orange curve) as measured in the 100–550 °C range.

2.3. Catalytic Activity—NH₃-SCR

The results of the catalytic activity, in terms of NO_x conversion and N₂ selectivity, of all the studied catalysts operating under identical experimental conditions are illustrated in Figure 9a,b. All the conversion curves presented a similar trend as a function of T, with a maximum of conversion followed by a more or less pronounced decline, depicting a “bell” or Gaussian shape rather than a sigmoidal one. This behavior is clearly attributed to the competition of the standard SCR reaction with at least two side reactions, namely, the catalytic oxidation of NH₃ to give either N₂ or N₂O.⁶⁴ In fact, beyond the maximum T, the NH₃ oxidation becomes kinetically prevalent, so that the reacting NH₃ is consumed in side reactions, with a consequent decrease in NO_x conversion.

Figure 9.

NO_x conversion (a), N₂ selectivity (b), and N₂O production (c) over the sol gel synthesized and impregnated samples.

The Mn-5 catalyst showed the best SCR activity, exhibiting both the highest NO_x conversion and the lowest amount of N₂O produced in the 200–250 °C range; the conversion was almost 90%; and the N₂O concentration was ca. 50 ppm. The Mn-10 sample had the same NO_x conversion as Mn-5; however, the selectivity was much lower (more than five times with respect to the Mn-5 catalyst at the maximum conversion) and high N₂O production was observed, while a lower Mn loading (Mn-1) resulted in lower conversion. With the Mn-10 sample, a rapid increase in the N₂O produced was observed suggesting the prominence of the (parasitic) oxidation of NH₃ that, being oxidized, did not participate in the SCR reaction and this is likely promoted by the high content of the reducible Mn species present in the Mn-10 catalyst, meaning an increased oxidative behavior. It has been widely demonstrated that the high activity of the Mn-based SCR catalysts is the result of the high NO oxidation and its subsequent adsorption in the form of NO_2(ads) over the manganese active sites. Activated ammonia also adsorbs on the same acid sites, and it recombines with nitrites consequently decomposing into N₂ and H₂O.⁶⁵ However, the rate of formation of such intermediates (NH₄NO₂ or NH₂NO) depends on the NH₃-SCR reaction rate, that is, favored at low temperatures, in comparison with the nonselective NH₃ oxidation rate, which prevails at high temperatures. N₂O is produced by either overoxidation of NH₃ [reaction 1] at higher T (>250 °C) or formation and decomposition of ammonium nitrate and/or nitrite intermediates at low R [reactions 2 and 3],^17,18,52,66 as shown in the following set of reactions

1

2

3

4

where eq 1 represents the ammonia oxidation, eq 2 represents the nitrate formation, eq 3 represents the ammonium nitrate formation, eq 4 represents the ammonium nitrate decomposition, and □ depicts an active site.

The N₂ selectivity and N₂O production (Figure 9b,c) was highly dependent on the synthesis method and the Mn loading likely due to the competition between the abovementioned reactions and the SCR reaction. One of the widely accepted theories for the TiO₂-based SCR reactions is a bifunctional mechanism whereby NH₃ is activated over the acid sites predominantly located on the TiO₂ support with participation of the −OH terminations of the metal sites and the subsequent reaction with the nitrates.^65,67 An increase of the metal loading shifts the NH₃–NO_x activation predominantly over Mn, and above a certain amount, the mechanism changes and the ammonium nitrate formation and decomposition [reactions 3 and 4] become the primary SCR pathway, favoring the N₂O production. This hypothesis can explain the drastic change in the N₂ selectivity for the sol–gel catalysts with 5 and 10 wt % Mn loading (Mn-5 and Mn-10, respectively), which still showed similar overall NO_x conversion. In fact, the similar amount and nature of the acid sites, as demonstrated by NH₃-TPD reported in Figure 8, justified the very close behavior concerning NO_x conversion, correlated to NH₃ adsorption, while the lower amount and better dispersion of manganese in the Mn-5 catalyst hindered the mechanism of N₂O production previously described and improved the N₂ selectivity. Finally, above 250 °C, the N₂O production and N₂ selectivity decreased because the NH₃ overoxidation produced NO instead of N₂O.

The impregnated sample had much lower NO_x conversion and selectivity (i.e., high N₂O produced per NO converted) that could be attributed to the surface occlusion of the TiO₂ acid sites that typically adsorbs NH₃ and activates it for the further SCR reaction, in agreement with the overall NH₃ desorbed amount determined as by NH₃-TPD (vide supra). As evidenced in the previous sections, the contact and interaction between Mn and Ti superficial active sites were highly improved due to the one-pot synthesis and the resulting high dispersion of Mn that was not possible to achieve by conventional impregnation. Indeed, the XRD analysis (Figure 1), the Raman spectroscopy (Figure 3) and the DR UV–vis spectroscopy (Figure 4) demonstrated the high Mn dispersion and that after calcination, the MnO₂ and Mn₂O₃ phases remained amorphous in the sol–gel samples with respect to the impregnated sample, where crystalline pyrolusite was detected. The remarkable low-temperature NO_x SCR activity can be likely attributed to such increased dispersion and to the enhanced contact between the multiple TiO₂ and MnO_x active sites, where NO and NH₃ adsorption occurs, therefore interaction and rearrangement toward NH₄NO₂ and NH₂NO intermediates can take place more easily, mainly on the amorphous phases, improving the NO_x conversion at low temperatures.⁶⁸⁻⁷⁰

3. Conclusions

The physico-chemical characterization results showed that the adopted reverse micelle sol–gel synthesis has an impact on the Mn surface and bulk dispersion and on the strength of Mn/TiO₂ interaction. Specifically, with respect to the traditional impregnation method, the reverse micelle method allowed for a fair dispersion of 1, 5, and 10 wt % Mn both in the bulk and at the surface of TiO₂ NPs, with surface MnO_x species resulting mainly amorphous and characterized by a different redox behavior as compared to those obtained by impregnation (i.e., crystalline pyrolusite).

As compared to the literature, the NH₃-SCR catalytic tests showed that the studied catalysts have enhanced deNO_x activity in NH₃-SCR, notwithstanding the low Mn content; both 5 at 10 wt % Mn catalysts were active at low T, the best catalytic activity being obtained at 5 wt % nominal Mn content.

The ensemble of the characterization techniques and the catalytic tests allowed for figuring out in the set of studied catalysts and “optimal” Mn content (i.e., nominal 5 wt %), corresponding to a material characterized by abundant and probably weaker (Brønsted) acidic sites and well-dispersed Mn-species. The same surface species were stabilized as the strong interaction with the support, as obtained by the reverse micelle sol–gel synthesis, which also favored the formation of lower oxidation states of Mn, that is, Mn(II) and Mn(III), in the surface amorphous MnO_x phases.

As a whole, the reported synthesis procedure could be utilized for those types of application (her, NH₃-SCR), requiring the fair dispersion of metal species and their stabilization by the support.

4. Experimental Section

4.1. Synthesis Method

All the chemicals were provided by Sigma-Aldrich (Italy).

Mn-doped and undoped TiO₂ NPs were prepared by the reverse micelle microemulsion sol–gel method⁷¹ using polyoxyethylene (20) oleyl ether (Brij O20) as a surfactant and cyclohexane as an oil phase, and titanium(IV)butoxide 97% (Ti-(BuO)₄) and manganese(II) nitrate tetrahydrate [Mn(NO₃)₂·4H₂O, purum p.a. ≥97%] as precursors.

In a typical synthesis, a proper amount of Brij-O20 is dispersed in cyclohexane by stirring at 50 °C. Meanwhile, Mn(NO₃)₂·4H₂O is dissolved in distilled water and the aqueous solution is added to the oil/surfactant solution and stirred for 45 min, with the formation of a water in-oil (w/o) microemulsion of surfactant nanoreactors. Afterward, Ti-(BuO)₄ is dropwise added to the emulsion.

The mixture is stirred for 2 h at the constant temperature of 50 °C, and finally, the emulsion is broken by the addition of 2-propanol, followed by sonication. The solid phase is then collected by centrifugation and dried at 100 °C for 24 h, followed by calcination in air at 500 °C for 2 h with a temperature ramp of 2.5 °C min^–1. Proper precursor amounts were used in order to obtain Mn wt % nominal contents of 0 wt % (TiO₂), 1 wt % (Mn-1), 5 wt % (Mn-5), and 10 wt % (Mn-10). Undoped titania was prepared by the same procedure.

For comparison, a sample with a 10 wt % nominal Mn content (Mn-10-IMP) was prepared by the incipient wetness impregnation method by using the same Mn precursor; the support (i.e., the TiO₂ sample prepared by the reverse micelle sol–gel method) was impregnated with few drops of the Mn precursor aqueous solution under stirring at 80 °C. The resulting powder was then calcined by following the same procedure adopted for the other sol–gel prepared samples.

4.2. Methods

X-ray powder diffraction (XRPD) patterns were obtained on an X’Pert Phillips diffractometer operating with Cu Kα radiation (1.541874 Å) and equipped with a PIXcel 1D detector (step: 0.026°2θ; time per step: 2 s). Lattice parameters, crystallite size, and quantitative phase analysis were determined by Rietveld analysis, as implemented in MAUD software.⁷² The instrumental broadening was characterized by means of the standard NIST 660a (LaB₆).

Raman spectra were collected by means of a Renishaw InVia Reflex microRaman spectrometer (Renishaw plc, Wotton-under-Edge, UK) equipped with a cooled charge-coupled device camera. The Raman source was a diode-pumped solid-state laser (λ_ex = 514.5 nm), and the sample inspection was performed through a microscope objective (10×) in backscattering light collection mode. 5 mW laser power, 5 s of exposure time, and four accumulations were employed to collect each spectrum of the pelletized (to facilitate the surface focusing) powders. 0.1 mW laser power was selected for the pelletized (undoped) TiO₂ powder only to prevent the scattering signal saturation of the detector. For each sample, five spectra have been recorded on different regions of the surface. Then, each data set was averaged and the resulting averaged Mn-doped TiO₂ spectra were normalized to the averaged (undoped) TiO₂ spectrum, with respect to the intensity of the 144 cm^–1 peak.

N₂ adsorption/desorption isotherms at −196 °C were acquired on samples previously outgassed at 200 °C for 2 h (Micromeritics Tristar II) to remove water and other atmospheric contaminants. The sample SSA was calculated according to the BET method; the pore total volume was measured at P/P⁰ = 0.99; and the pore size distribution was calculated by applying the BJH (Barrett–Joiner–Hallenda) method to isotherm desorption branch.

FESEM micrographs were taken on a ZEISS Supra 40 FESEM instrument (Carl-Zeiss AG, Oberkochen, Germany) equipped with an EDX probe that was used for semiquantitative determination of the sample chemical composition; typically, on each samples, five spots (ca. 5 μm³ volume) were analyzed and the average Ti, O, and Mn contents were calculated.

DR UV–vis spectra of the powders were recorded on a UV–vis Varian Cary 5000 spectrophotometer equipped with a DR integration sphere in the 200–1000 nm range.

XPS analyses were performed on a Versa Probe II Scanning XPS Microprobe spectrometer (Physical Electronics GmbH). The measurements were carried out with a monochromatized Al Kα source (X-ray spot 100 μm) at a power of 24.4 W. Wide scans and high-resolution spectra were acquired in fixed analyzer transmission mode with a pass energy of 117.40 and 29.35 eV, respectively. An electron gun was used for charge compensation (1.0 V 20.0 μA). All the BEs were referenced to the C 1s line at 284.8 ± 0.1 eV for adventitious carbon. Data processing was performed by using the MultiPak software (v. 9.8.0.19).

H₂-TPR analysis was carried out on a TPD/R/O (temperature-programmed desorption/reduction/oxidation) 1100 ThermoQuest instrument on ca. 50 mg of the sample. In a typical experiment, the powder was kept under constant flow (20 mL min^–1) of 5% H₂ in the Ar mixture and heated from room temperature to 800 °C with a ramp of 10 °C min^–1. The measured H₂ uptake was then normalized to the sample weight, in order to allow comparison.

NH₃-TPD (temperature-programmed desorption) analysis was carried out in a quartz bed-fixed reactor connected to a NH₃ ND-IR (nondispersive infrared) ABB Uras 14 gas analyzer. In each experiment, ca. 200 mg of the sample was pretreated under N₂ flow (100 mL min^–1) for 60 min at 500 °C (temperature ramp = 10 °C min^–1). Then, a mixture of 2000 ppm NH₃ in He (400 mL min^–1) was fed into the reactor at 100 °C for the adsorption run, which lasted until the NH₃ concentration value reached the initial one. The desorption run was performed under N₂ flow (400 mL min^–1) first at r.t. to remove the loosely physisorbed fraction and subsequently by increasing the temperature up to 500 °C (temperature ramp = 5 °C min^–1).

4.3. Catalytic Activity Measurements

SCR of NO_x was carried out by placing 500 mg of catalyst powder in a fixed bed tubular quartz reactor (inner diameter = 10 mm) under a total feed gas flow rate of 300 N mL min^–1, which translates to gas hourly space velocity of about 50,000 h^–1, as values above 30,000 h^–1 are typically used industrially. The gas mixture composition was as follows: 500 ppm NO, 500 ppm NH₃, 3% O₂, and balance N₂.

Catalytic tests were performed under isothermal conditions by increasing T from 50 to 350 °C with steps of 32 °C and by waiting to reach stable T and concentrations of the reactive species before each T increase.

The NO_x concentration was analyzed by means of a NO/NO₂ UV Limas gas analyzer (ABB SpA) and the N₂O and NH₃ concentrations by a ND-IR Uras gas analyzer (ABB SpA).

The NO_x and NH₃ conversions and the N₂ selectivity were calculated from gas concentration according to the following equations

5

6

7

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.1c03153.

• Raman spectra after subtraction of the undoped TiO₂ spectrum; N₂ adsorption/desorption isotherms at −196 °C; FE-SEM micrographs; and Ti 2p XP spectra (PDF)

The authors declare no competing financial interest.