Alumina-Supported Manganese Catalysts for Soot Combustion Prepared by Thermal Decomposition of KMnO 4

Alumina-supported manganese catalysts with cryptomelane and/or birnessite structure have been prepared using a simple method based on the thermal decomposition of potassium permanganate. The samples have been characterized by XRD, FTIR, TGA, DSC, N2 adsorption at −196 °C, SEM, H2-TPR and XPS, and their catalytic activity for soot combustion has been tested and compared to that of a reference Pt/alumina catalyst. The thermal decomposition of alumina-supported KMnO4 yields a mixture of supported birnessite and potassium manganate which is the most effective, among those prepared, to lower the soot combustion temperature. However, this material is not useful for soot combustion because the accelerating effect is not based on a catalytic process but on the oxidation of soot by potassium manganate. A suitable soot combustion catalyst is obtained after potassium manganate is removed by water washing, yielding only the birnessite phase OPEN ACCESS Catalysts 2012, 2 353 on the γ-Al2O3 support. This birnessite phase can be transformed into cryptomelane by calcination at 600 °C. These two samples, γ-Al2O3-supported birnessite and cryptomelane are suitable catalysts for soot combustion in NOx/O2 mixtures, as their catalytic activity is based on the NO2-assited mechanism, that is, both catalysts accelerate the oxidation of NO to NO2 and NO2 promotes soot oxidation. The soot combustion temperatures obtained with these birnessite/cryptomelane alumina-supported catalysts are similar to that obtained with the reference Pt/alumina catalyst.


Introduction
One of the main problems related with the development of the XXI century society is environmental pollution; the emission of nano/micro particles of carbon (soot) by diesel-engines being one of the unsolved issues to overcome.
In a previous study [5], manganese catalysts with birnessite and cryptomelane structure were prepared, characterized, and tested as catalysts for soot combustion with NO x /O 2 .The activity of these catalysts was compared to that of some other commercially available pure manganese oxides (MnO, MnO 2 , Mn 2 O 3 , Mn 3 O 4 , and natural MnO 2 ) and it was concluded that both birnessite and cryptomelane are more active for soot combustion than the pure oxides tested.In that study, cryptomelane was prepared by the reflux method [14,15] and birnessite was prepared following the method of Ching et al. [16].In a further study [6], manganese oxides with cryptomelane and/or birnessite structure were successfully prepared by a simple method based on the thermal decomposition of potassium permanganate.Their catalytic activity to accelerate soot combustion was confirmed, and it was also concluded that the catalytic combustion of soot occurs by the NO 2 -assisted mechanism, which is the most convenient to operate under real conditions.
For the practical use of soot combustion catalysts, the active phases have to be loaded in a proper substrate.A typical DOC (Diesel Oxidation Catalyst) used in a CRT (Continuously Regenerating Trap) system for CO and hydrocarbons oxidation, and also to oxidize NO to NO 2 , which promotes the combustion of soot collected in a filter placed downstream (NO 2 -assisted soot combustion), consists of an inert substrate coated with an alumina layer where different active components (typically containing noble metals) are loaded.
The goal of the current study is to evaluate if alumina-supported manganese catalysts with cryptomelane and/or birnessite structure can be prepared by a simple method based on the thermal decomposition of potassium permanganate [6].The soot combustion activity of alumina-supported manganese catalysts has been tested and compared with that of a reference Pt/alumina catalyst.

Characterization Results
The X-ray diffraction patterns of the manganese-alumina samples prepared are included in Figure 1.In addition, in order to analyze potential changes occurring on the alumina support during the preparation of the supported catalysts, the XRD patterns of the γ-Al 2 O 3 support calcined at 400 °C and 600 °C have also been included.The diffraction peaks of γ-Al 2 O 3 appear at 2θ = 19.51°,32.7°, 37.5°, 39.24°, 45.5° and 67.1°, being consistent with the diffractograms reported for this material (JCPDS: 00-046-1215) [17,18].The XRD patterns of γ-Al 2 O 3 are equal after calcinations at 400 °C and 600 °C, evidencing the thermal stability of this support in the range of temperatures analyzed.
The diffraction patterns of MnAlT400 and MnAlT400W present peaks assigned to the γ-Al 2 O 3 support along with peaks at 2θ = 12.54° and 25.10°, which can be attributed to birnessite [19].The expected peaks of potassium manganate were not detected by XRD in MnAlT400, probably due to its high dispersion on the support.However, the presence of potassium manganate was evidenced in a previous study performed with unsupported manganese samples (without alumina) after calcinations of potassium permanganate at 400 °C [6].The XRD pattern changes significantly for MnAlW600, showing characteristics peaks of criptomelane (JCPDS 44-0141) [20].
Evidence of the presence of potassium manganate in MnAlT400 are the greenish color of the water used in the washing step performed to MnAlT400 and the increase of the intensity of the birnessite peaks of MnAlT400W with regard to those of MnAlT400.In addition, the K/Mn molar ratio determined from Atomic Absorption Spectrometer (AAS) analysis values (data in Table 1), is higher in MnAlT400 than in MnAlT400W, that is, decreases after the washing step due to potassium manganate removal.The SEM images included in Figure 2 also support the formation of potassium manganates upon calcination of γ-Al 2 O 3 -supported KMnO 4 .γ-Al 2 O 3 (Figure 2a) consists of quite spherical particles of 100-200 µm, the difference in contrast between the particles in this figure is assigned to "charging effect" on the surface of particles probably because this material is dielectric and the electrons are accumulated on it.These spherical particles are covered by potassium manganate needles of 15-40 µm length (Figure 2b).These needles are removed by water washing, and therefore, are neither observed in MnAlT400W nor in MnAlW600.Figure 3 shows the FTIR spectra of the manganese-alumina samples together with those of the γ-Al 2 O 3 support calcined at 400 °C and 600 °C.The bands between 3800-3680 cm −1 and 2000-1500 cm −1 are assigned to the O-H vibration of adsorbed water and/or to surface hydroxyl groups [21], and those between 980-550 cm −1 are attributed to Al-O bond vibrations [22].The spectra of γ-Al 2 O 3 after calcinations at 400 °C and 600 °C are quite similar to each other, supporting that the γ-Al 2 O 3 support is thermally stable in the range of temperatures studied, which is consistent with the XRD conclusions.As can be observed in Figure 3, the band at 3745 cm −1 , assigned to adsorbed water and/or hydroxyl groups on γ-Al 2 O 3 , disappear upon KMnO 4 impregnation, and the position of those at 3460 and 1643 cm −1 shift with respect to their position on bare alumina.The bands at 1400, 918 and 760 cm −1 are attributed to KMnO 4 , and the shift in position of these bands with regard to the typical positions reported for pure KMnO 4 [6,23] evidences a certain interaction between KMnO 4 and γ-Al 2 O 3 .
In sample MnAlT400, the IR bands below 700 cm −1 are assigned to vibrational modes of Mn-O bonds [24] and the bands at 509 y 422 cm −1 are characteristics of birnessite [25].The band at around 1019 cm −1 , which can be ascribed to Mn=O vibrations, suggests the presence of manganate-type oxides (Mn(VI)Ox) [26].This band disappears after washing the MnAlT400 sample (see IR spectra of MnAlT400W) due to the removal of this soluble specie.After calcinations at 600 °C of the washed sample (see IR spectra of MnAlTW600) a new band appears at 1022 cm −1 , which has been reported for the vibration mode of Mn 3+ -O [27], and finally some studies attribute the bands between 450 and 750 cm −1 to the vibration of Mn-O bonds in cryptomelane [28].
Figures 4 and 5 show the TGA and DSC results, respectively, and Table 2 compiles the weight losses determined from TGA profiles for different temperature ranges.The weight losses observed below 250 °C can be partially attributed to water release and the weight losses occurring above 500 °C are associated with structural changes that imply O 2 delivery [6].For instance, in γ-Al 2 O 3 -supported KMnO 4 , the endothermic DSC peaks at temperatures below 200 °C are due to the release of physisorbed water, and the exothermic peaks at 288 °C and 537 °C are associated to phase transitions occurring with oxygen release.The weight loss suffered by γ-Al 2 O 3 -supported KMnO 4 between 250 and 500 °C is related to the decomposition of KMnO 4 , considering that the γ-Al 2 O 3 support is stable in the range of temperatures studied, as deduced from XRD and FTIR.This KMnO 4 decomposition is in accordance with previously reported DSC results [6].
The quantitative weight loss values included in Table 2 show that the weight loss suffered until 250 °C by the washed sample (MnAlT400W) is lower than that suffered by the counterpart non-washed sample (MnAlT400), which is attributed to the hygroscopic character of manganates, as it was reported by Kappenstein et al. [29].The weight loss of MnAlW600 below 250 °C is due to the release of water adsorbed in the porous structure of cryptomelane.The structural changes related with the formation of manganese oxide phases, as determined by XRD and FTIR, are in accordance with the DSC profiles (Figure 5).
In conclusion, the TGA results discussed are consistent with those previously obtained with pure KMnO 4 (without alumina) [6].Birnessite and potassium manganate are formed after KMnO 4 calcination at 400 °C, and transformation to cryptomelane occurs after potassium manganate removal by water washing and calcination at 600 °C.
N 2 adsorption-desorption isotherms are shown in Figure 6, and different parameters obtained from these isotherms are compiled in Table 3.All samples present type IV isotherms, according to IUPAC classification [30], with considerable N 2 adsorption at low relative pressures, which is indicative of microporosity [31], and hysteresis cycles.These profiles indicate that the materials present also a mesoporous structure and them shown capillary condensation above P/Po 0.3.According to the De Boer [30][31][32] classification, the hysteresis cycles present contributions of slit-shaped pores (B) and interconnected pores with bottleneck shape (E).The mesopores size distributions determined by the BJH method (inset in Figure 6) show peaks centered between 35.98 Å and 37.86 Å, without significant change among samples.The micropore size distribution, obtained by the HK method, (Figure 7) is monomodal for MnAlT400, while bimodal for bare γ-Al 2 O 3 , MnAlT400W and MnAlW600, suggesting the presence of primary and secondary micropores [33].These results evidence that potassium manganate on MnAlT400 blocks part of the γ-Al 2 O 3 microporosity, in agreement with results reported by other authors [34], and the closed micropores becomes again open after removal of the soluble salts by water washing.The BET surface areas included in Table 3 are consistent with this conclusion.The BET surface area of γ-Al 2 O 3 decreases after KMnO 4 impregnation, and increases again after the washing step (sample MnAlT400W).The H 2 reduction profile obtained with MnAlW600 shows two main reduction peaks at 400 °C and 434 °C, which could indicate that the reduction proceeds following the sequence of steps KMn 8 O 16 (cryptomelane)→Mn 3 O 4 →MnO.A similar H 2 reduction profile is obtained with MnAlT400W, but the temperatures of the main peaks are shifted to lower temperatures.The intensity of the second peak is always higher than that of the first one in both samples.On the contrary, the H 2 reduction profile of MnAlT400 (non-washed sample) differs from those of the washed samples (MnAlW600 and MnAlT400W) due to the presence of salts with manganese cations in oxidation states higher than +4.In conclusion, the H 2 reduction experiments suggest the presence of Mn cations with oxidation state near +4, as expected for birnessite and cryptomelane.
XPS characterization was used in order to obtain additional information about the oxidation state of manganese on the different samples by using a method reported elsewhere [6]. Figure 9 plots the formal oxidation state of manganese for three different commercial manganese oxides (MnO, Mn 3 O 4 and MnO 2 , by Aldrich) with regard to the difference between the binding energy of the Mn 2p 3/2 and Mn 2p 1/2 peaks measured experimentally by XPS with each oxide (+2, +2.7 and +4, respectively).These reference values allow us to drawn a calibration line, and thus, the formal oxidation state of manganese in MnAlT400, MnAlT400W and MnAlW600 can be estimated by introducing in this trend line the XPS values calculated by difference between the corresponding binding energies of the Mn 2p 3/2 and Mn 2p 1/2 transitions.The formal oxidation states estimated are +4.3,+3.5 and +3.7 for MnAlT400, MnAlT400W and MnAlW600, respectively.The value obtained with the unwashed sample (MnAlT400; +4.3) is slightly above +4, which is consistent with the presence of manganate salts (in addition to birnessite) where the oxidation state of manganese is +6.The values obtained with the washed samples are between +3 and +4 because the manganates have been removed and only the birnessite and/or cryptomelane phases remain on the samples.As expected, the formal oxidation state obtained with the sample calcined at 600 °C (MnAlW600; +3.7) is slightly higher to that obtained with the sample calcined at 400 °C (MnAlW600; +3.5).These XPS results are in agreement with the 200 300 400 500 600 700 800

MnAlT400 MnAlW600
MnAlT400W H 2 -TPR conclusions and confirm the presence of manganese cations with the oxidation states expected for birnessite and cryptomelane phases, which are the phases identified by XRD and FTIR.

Figure 9.
Formal oxidation state of manganese as a function of the difference between the binding energy of the Mn 2p 3/2 and Mn 2p 1/2 XPS peaks.

Catalytic Tests
The catalytic combustion of soot was evaluated in O 2 /N 2 and NO x /O 2 /N 2 .The soot conversion profiles obtained in O 2 /N 2 , and included in Figure 10, show that only the sample MnAlT400 is able to lower the soot combustion temperature with regard to the uncatalyzed reaction in these reactions conditions.This is attributed to the presence of potassium manganates, that is, this is not a catalytic effect but a redox reaction where potassium manganate oxidises soot.
In experiments performed in NO x /O 2 /N 2 (Figure 11), the lowest soot combustion temperature is also achieved with MnAlT400.However, in the presence of NO x , the washed samples (MnAlW600 and MnAlT400W) are also able to lower the soot combustion temperature with respect to the uncatalyzed reaction.These results are consistent with those previously obtained with unsupported manganese samples of similar composition.In the previous study [6], it was concluded that the combustion of soot assisted by birnessite and cryptomelane catalysts (once manganese salts have been removed) occurs by the NO 2 mechanism, that is, acceleration of the NO oxidation to NO 2 , which is much more oxidizing than NO and O 2 .For comparison, the soot combustion profile obtained with a Pt/Al 2 O 3 reference catalyst has been included in Figure 11.Note, that the onset of the soot combustion temperatures obtained with the manganese samples are not very different to that obtained with Pt.Therefore, these oxides are able to accelerate soot combustion and would be a valuable alternative for catalytic combustion of soot, since they are cheaper than Pt catalysts and easy to prepare.In order to analyze in more detail the reaction mechanism by which these alumina-supported manganese catalysts (MnAlW600 and MnAlT400W) accelerate soot combustion in a NO x /O 2 gas mixtures, blank experiments (only with the catalysts and without soot) have been performed with the NO x /O 2 /N 2 gas mixture.The NO 2 percentage (produced by NO oxidation) in total NO x basis is plotted in Figure 12 as a function of temperature.These results confirm that the manganese catalysts prepared in this study are able to oxidize NO to NO 2 , and the amount of NO 2 yielded is not very different to that produced by the Pt/Al 2 O 3 reference catalyst (the amounts of NO 2 evolved are reported in Figure 12 in mmol/gr).These blank experiments confirm that the catalyzed combustion of soot takes place with the participation of the NO 2 -assisted mechanism.

Preparation of Sample
Alumina-supported manganese samples were synthesized by means of the wet impregnation method [38].γ-Al 2 O 3 (Carlo Erba) was impregnated with Commercial KMnO 4 (J.T. Baker, analytical grade) in solution and the solvent was removed at 90 °C.The nominal MnO 4 − loading was 4.5 wt%.
Three alumina-supported manganese samples, referred to as MnAlT400, MnAlT400W and MnAlW600, were prepared following different procedures: -MnAlT400: After KMnO 4 impregnation on γ-Al 2 O 3 , the sample is calcined in air at 400 °C for 6 h (heating rate 10 °C/min).-MnAlT400W: The previously prepared sample MnAlT400 is washed with distilled deionized to remove soluble salts (potassium manganate), and after washing, the sample is dried at 60 °C for 24 h.-MnAlW600: The previously prepared sample MnAlT400W is calcined at 600 °C for 6 h (heating rate 10 °C/min).
In addition, a 1% Pt/-Al 2 O 3 catalyst was prepared by excess-volume impregnation of -Al 2 O 3 using Pt(NH 3 ) 4 (NO 3 ) 2 as precursor.After impregnation, the catalyst is heated in static air at 110 °C for 12 h and calcined at 700 °C for 5 h.

Characterization of Samples
The samples were characterised using the same techniques and the technical specifications reported by Becerra et al. [6] in brief: (i) Powder X-ray diffraction patterns were obtained in a Rigaku MiniFlex II diffractometer; (ii) infrared measurements were carried out using a NICOLET 380 spectrometer with a DTGS detector; (iii) thermogravimetric analysis (TGA) was performed in a thermo balance, by TA Instruments, model TGA Q500 and (iv) differential scanning calorimetric measurements were performed with a TA Instruments equipment, model DSC Q100.
Elemental analysis was performed in an Atomic Absorption Spectrometer (AAS) Perkin Elmer, model 3110.For this purpose, the samples were dissolved (even the alumina support) by a fusion process with lithium metaborate [39].50 mg of sample are mixed with 300 mg of lithium metaborate and calcined at 1000 °C for 15 min.The mixture is cooled down to room temperature, and 20 mL of distilled and deionized water and 5 mL of 37% HCl are added.This solution is heated until boiling under continuous stirring, and once the solids are dissolved, 1 mL of 15% lanthanum chloride is added.Finally, the solution is diluted with water for elemental analysis.The Mn and K contents were determined at wavelengths of 279.5 nm and 766.5 nm, respectively.
Scanning Electron Microscopy (SEM) images were recorded with a Jeol JSM 5910LV microscope, operated at 15 kV and 19 kV in high vacuum mode, with Back-Scattered Electrons (BES) and secondary electron imaging (SEI) for image generation.
XPS analysis was carried out in a VG-Microtech Multilab electron spectrometer using Mg-Kα (1253.6 eV) radiation source.H 2 -TPR experiments were performed in a Micromeritics device; model Pulse ChemiSorb 2705, with a TCD detector.

Catalytic Tests
Catalytic tests were performed in a fixed-bed reactor at atmospheric pressure under a gas flow (500 mL/min) containing 5% O 2 balanced with N 2 or 500 ppm NO x + 5% O 2 balanced with N 2 .The model soot used was a carbon black from Degussa S.A. (Printex-U).The experiments consisted of heating the soot-catalyst mixtures from 25 to 750 °C at 10 °C/min.The soot-catalyst mixtures contained 80 mg of catalyst + 20 mg of soot + 300 mg SiC, and were prepared with a spatula following the so-called "loose contact" procedure [40].Blank experiments were performed only with the catalysts (without soot) under similar experimental conditions.The gas composition was monitored by specific NDIR-UV (Non Dispersive Infrared Ultra-Violet) gas analyzers for NO, NO 2 , CO, CO 2 and O 2 and the soot conversion was determined from CO and CO 2 evolved.

Conclusions
From the results presented and discussed in this paper, it can be concluded that different alumina-supported manganese oxide catalysts have been obtained by impregnation of potassium permanganate on γ-Al 2 O 3 and subsequent thermal decomposition.Some of the prepared materials can be considered for the regeneration of DPF filters since they are able to catalytically accelerate the soot combustion process.
Thermal decomposition of alumina-supported potassium permanganate yields a mixture of supported birnessite and potassium manganate.This material is the most effective, among those prepared to lower the soot combustion temperature, but has no practical relevance for soot combustion because the accelerating effect is based on a redox process thus consuming the raw samples.A suitable soot combustion catalyst is obtained after potassium manganate removal by water washing, only leaving the birnessite phase on the γ-Al 2 O 3 support.The birnessite phase can then be transformed into cryptomelane by calcinations at 600 °C.
The γ-Al 2 O 3 -supported birnessite and cryptomelane samples are suitable catalysts for soot combustion in NO x /O 2 mixtures, and their catalytic activity is based on the NO 2 -assisted mechanism, that is, both catalysts accelerate the oxidation of NO to NO 2 which promotes soot oxidation.

Figure 10 .
Figure 10.Soot conversion profiles in catalytic tests performed in O 2 /N 2 .

Figure 11 .
Figure 11.Soot conversion profiles in catalytic tests performed in NO x /O 2 .

3 Figure 12 .
Figure 12.NO 2 percentage in blank experiments (without soot) performed in NO x /O 2 .The amounts of NO 2 evolved are reported between parentheses in mmol/gr).

Table 2 .
Weight loss in TGA experiments in N 2 atmosphere.