Effect of Residual Na + on the Combustion of Methane over Co 3 O 4 Bulk Catalysts Prepared by Precipitation

The effect of the presence of residual sodium (0.4 %wt) over a Co3O4 bulk catalyst for methane combustion was studied. Two samples, with and without residual sodium, were synthesized by precipitation and thoroughly characterised by X-ray diffraction (XRD), N2 physisorption, Wavelength Dispersive X-ray Fluorescence (WDXRF), temperature-programmed reduction with hydrogen followed by temperature-programmed reduction with oxygen (H2-TPR/O2-TPO), temperature-programmed reaction with methane (CH4-TPRe), ultraviolet–visible–near-infrared diffuse reflectance spectroscopy (UV-vis-NIR DRS), Raman spectroscopy and X-ray photoelectron spectroscopy (XPS). It was found that during calcination, a fraction of the sodium atoms initially deposited on the surface diffused and migrated into the spinel lattice, inducing a distortion that improved its textural and structural properties. However, surface sodium had an overall negative impact on the catalytic activity. It led to a reduction of surface Co3+ ions in favour of Co2+, thus ultimately decreasing the Co3+/Co2+ molar ratio (from 1.96 to 1.20) and decreasing the amount and mobility of active lattice oxygen species. As a result, the catalyst with residual sodium (T90 = 545 ◦C) was notably less active than its clean counterpart (T90 = 500 ◦C). All of this outlined the significance of a proper washing when synthesizing Co3O4 catalyst using a sodium salt as the precipitating agent.


Introduction
Methane is a powerful greenhouse effect gas (25 times higher than that of CO 2 ) that appears as a residue or flue gas in many different applications.These include thermal plants, refineries or natural gas engines.To protect the environment from its noxious effect, this pollutant needs to be properly removed [1].Usually, the off-gas treatment technology must be efficient at low temperatures when operating with large gas flows with small amounts of methane.Under these conditions, the most suitable technique is catalytic oxidation.
The traditional catalysts for oxidation of methane have been based on noble metals, such as platinum or palladium.In the recent years, extensive research has been made with the aim of finding other alternatives.This is mainly due to their high price, low availability and tendency to deactivate and age [2,3].The most significant part of this research is focused on both single and mixed transition metal oxides.Materials such as perovskites, hexaaluminates or spinels composed of different metals have been studied and proven active for the oxidation of methane [4][5][6][7].Spinel oxides, specifically, are mixed oxides with general formula AB 2 O 4, where A is a metal with oxidation state +2 and tetrahedral coordination and B is a metal with oxidation state +3 and octahedral coordination [8].The activity of these types of materials for oxidation reactions lies on the easiness that the metal cations For this reason, in the present paper, two Co 3 O 4 catalysts were prepared by a basic precipitation method, intentionally leaving some residual sodium in one of them to study its effect over the catalytic activity for the complete oxidation of methane.The obtained catalysts were thoroughly characterised and the role of the surface cobalt and oxygen species on the oxidation process was ascertained.

Synthesis of Co 3 O 4 Precursors
When preparing the two bulk Co 3 O 4 catalysts by precipitation with sodium carbonate, the catalytic precursors were filtrated and washed with varying amounts of deionized water between 1-8 L. The filtrates extracted from the filtration step were recovered and their pH was measured to estimate the extent of the removal of residual sodium from the precursor.The evolution of pH value as a function of the volume of water used for washing is shown in Figure 1.
Catalysts 2018, 8, x FOR PEER REVIEW 3 of 15 For this reason, in the present paper, two Co3O4 catalysts were prepared by a basic precipitation method, intentionally leaving some residual sodium in one of them to study its effect over the catalytic activity for the complete oxidation of methane.The obtained catalysts were thoroughly characterised and the role of the surface cobalt and oxygen species on the oxidation process was ascertained.

Synthesis of Co3O4 Precursors
When preparing the two bulk Co3O4 catalysts by precipitation with sodium carbonate, the catalytic precursors were filtrated and washed with varying amounts of deionized water between 1-8 L. The filtrates extracted from the filtration step were recovered and their pH was measured to estimate the extent of the removal of residual sodium from the precursor.The evolution of pH value as a function of the volume of water used for washing is shown in Figure 1.It was observed that the pH of the filtrates remained almost constant (pH = 7.8) after washing the precipitate with four litres of water, which would suggest that all residual sodium from the synthesis process was removed with this amount of water.This residual pH was somewhat above 7 due to the basicity and slight solubility of the precipitate.

Physico-Chemical Characterisation
The structural characterisation of the catalysts was performed by XRD.The diffractograms in Figure 2a only revealed signals attributable to a pure phase of spinel-type Co3O4 (JCPDS 042-1467).No signals attributable to Na2O, or any other Co-Na mixed oxides, were detected in the catalyst with residual sodium (CC-Na).However, the diffraction peaks displayed by the CC-Na sample were markedly wider than those displayed by the Na-free CC sample and were shifted towards lower diffraction angles, as can be seen in Figure 2b.This revealed that the presence of trace amounts of sodium (0.4 wt% as determined by WDXRF) induced a distortion of the spinel lattice of Co3O4.It was observed that the pH of the filtrates remained almost constant (pH = 7.8) after washing the precipitate with four litres of water, which would suggest that all residual sodium from the synthesis process was removed with this amount of water.This residual pH was somewhat above 7 due to the basicity and slight solubility of the precipitate.

Physico-Chemical Characterisation
The structural characterisation of the catalysts was performed by XRD.The diffractograms in Figure 2a only revealed signals attributable to a pure phase of spinel-type Co 3 O 4 (JCPDS 042-1467).No signals attributable to Na 2 O, or any other Co-Na mixed oxides, were detected in the catalyst with residual sodium (CC-Na).However, the diffraction peaks displayed by the CC-Na sample were markedly wider than those displayed by the Na-free CC sample and were shifted towards lower diffraction angles, as can be seen in Figure 2b.This revealed that the presence of trace amounts of sodium (0.The crystallite size of Co3O4 was estimated from the full width half maximum (FWHM) of the characteristic peak located at 37.1°, assignable to the (311) crystalline plane, by applying the Scherrer equation.Besides, the cell size could be calculated by performing a profile matching of the diffractograms.It was found that the crystallite size of the CC-Na catalyst was almost half of that of the CC catalyst, while its cell size was notably larger.This was in agreement with the aforementioned distortion caused by the inclusion of sodium into the spinel lattice.This distortion can be explained in terms of the difference in ionic radius between Co 2+ and Co 3+ ions (72 and 69 pm, respectively) and Na + ions (113 pm).
The results from the textural characterisation, also displayed in Table 1, showed that the catalyst with residual sodium exhibited almost twice as much surface area as the Na-free sample, and a pore volume a 25% larger.This suggested that the distortion caused in the spinel lattice also caused an enhancement of the textural properties of the Co3O4.The pore size distributions, shown in Figure 3, also evidenced a shrinking effect over the mean pore diameter when cobalt oxide is contaminated with residual sodium.These effects have already been reported in other studies regarding the doping of Co3O4 catalysts with other metallic elements [27,28].It must be noticed that all those variations were originated by a relatively low bulk concentration of sodium (0.4 wt%).The crystallite size of Co 3 O 4 was estimated from the full width half maximum (FWHM) of the characteristic peak located at 37.1 • , assignable to the (311) crystalline plane, by applying the Scherrer equation.Besides, the cell size could be calculated by performing a profile matching of the diffractograms.It was found that the crystallite size of the CC-Na catalyst was almost half of that of the CC catalyst, while its cell size was notably larger.This was in agreement with the aforementioned distortion caused by the inclusion of sodium into the spinel lattice.This distortion can be explained in terms of the difference in ionic radius between Co 2+ and Co 3+ ions (72 and 69 pm, respectively) and Na + ions (113 pm).
The results from the textural characterisation, also displayed in Table 1, showed that the catalyst with residual sodium exhibited almost twice as much surface area as the Na-free sample, and a pore volume a 25% larger.This suggested that the distortion caused in the spinel lattice also caused an enhancement of the textural properties of the Co 3 O 4 .The pore size distributions, shown in Figure 3, also evidenced a shrinking effect over the mean pore diameter when cobalt oxide is contaminated with residual sodium.These effects have already been reported in other studies regarding the doping of Co 3 O 4 catalysts with other metallic elements [27,28].It must be noticed that all those variations were originated by a relatively low bulk concentration of sodium (0.4 wt%).The structure of the samples were additionally studied by UV-vis-NIR DRS, Raman spectroscopy and XPS. Figure 4a shows the UV-vis-NIR diffuse reflectance spectra.Both samples displayed absorption bands at 630, 1210, 1300 and 1510 nm, associated with the presence of tetrahedral Co 2+ species.Since the Co3O4 oxide also contains octahedral Co 3+ species, their corresponding absorption bands (weak shoulders at 400 and 720 nm) were also expected to be visible [29].However, the former was not detectable, while the latter was somewhat visible, only for the CC sample.This could suggest that the amount of Co 3+ present in this catalyst was larger with respect to the Na-containing oxide.In addition, a band at 1370 nm was only detected for the CC sample.This band was associated with the 2ν harmonics band from residual hydroxyl groups [30].On the other hand, both samples displayed absorption bands at 550, 850 and 1190 nm, related to Co 2+ ions with octahedral coordination as in CoO.Interestingly, these bands were more noticeable for The structure of the samples were additionally studied by UV-vis-NIR DRS, Raman spectroscopy and XPS. Figure 4a shows the UV-vis-NIR diffuse reflectance spectra.Both samples displayed absorption bands at 630, 1210, 1300 and 1510 nm, associated with the presence of tetrahedral Co 2+ species.Since the Co 3 O 4 oxide also contains octahedral Co 3+ species, their corresponding absorption bands (weak shoulders at 400 and 720 nm) were also expected to be visible [29].However, the former was not detectable, while the latter was somewhat visible, only for the CC sample.This could suggest that the amount of Co 3+ present in this catalyst was larger with respect to the Na-containing oxide.In addition, a band at 1370 nm was only detected for the CC sample.This band was associated with the 2ν harmonics band from residual hydroxyl groups [30].The structure of the samples were additionally studied by UV-vis-NIR DRS, Raman spectroscopy and XPS. Figure 4a shows the UV-vis-NIR diffuse reflectance spectra.Both samples displayed absorption bands at 630, 1210, 1300 and 1510 nm, associated with the presence of tetrahedral Co 2+ species.Since the Co3O4 oxide also contains octahedral Co 3+ species, their corresponding absorption bands (weak shoulders at 400 and 720 nm) were also expected to be visible [29].However, the former was not detectable, while the latter was somewhat visible, only for the CC sample.This could suggest that the amount of Co 3+ present in this catalyst was larger with respect to the Na-containing oxide.In addition, a band at 1370 nm was only detected for the CC sample.This band was associated with the 2ν harmonics band from residual hydroxyl groups [30].On the other hand, both samples displayed absorption bands at 550, 850 and 1190 nm, related to Co 2+ ions with octahedral coordination as in CoO.Interestingly, these bands were more noticeable for On the other hand, both samples displayed absorption bands at 550, 850 and 1190 nm, related to Co 2+ ions with octahedral coordination as in CoO.Interestingly, these bands were more noticeable for the CC-Na sample.This suggested that the presence of this species was larger with respect to the CC sample.An additional feature could also be seen when examining the spectra in detail, as depicted in Figure 4b.When zooming into the 1150-1250 nm region, where the well-defined absorption signals at around 1190 and 1210 nm were located, it could be noted that both signals for the CC-Na sample were shifted with respect to the CC sample.Specifically, the former was redshifted 12 nm from its original position, while the latter was blueshifted 10 nm from its original position.These findings could further confirm the lattice distortion caused by the residual sodium.
Raman spectroscopy results, shown in Figure 5, also pointed out the aforementioned lattice distortion.The resulting spectra for both samples showed five Raman bands.The bands located at 194, 519 and 617 cm −1 could be attributed to the F 2g mode.The bands located at 479 and 687 cm −1 could be assigned to the E g and A 1g modes, respectively [16].However, the CC-Na sample showed these bands slightly shifted and became wider with respect to those of the CC sample.Additionally, the band attributed to the A 1g mode also showed a strong asymmetry.This is a clear sign of lattice distortion according to the bibliography [31,32].
the CC-Na sample.This suggested that the presence of this species was larger with respect to the CC sample.An additional feature could also be seen when examining the spectra in detail, as depicted in Figure 4b.When zooming into the 1150-1250 nm region, where the well-defined absorption signals at around 1190 and 1210 nm were located, it could be noted that both signals for the CC-Na sample were shifted with respect to the CC sample.Specifically, the former was redshifted 12 nm from its original position, while the latter was blueshifted 10 nm from its original position.These findings could further confirm the lattice distortion caused by the residual sodium.
Raman spectroscopy results, shown in Figure 5, also pointed out the aforementioned lattice distortion.The resulting spectra for both samples showed five Raman bands.The bands located at 194, 519 and 617 cm −1 could be attributed to the F2g mode.The bands located at 479 and 687 cm −1 could be assigned to the Eg and A1g modes, respectively [16].However, the CC-Na sample showed these bands slightly shifted and became wider with respect to those of the CC sample.Additionally, the band attributed to the A1g mode also showed a strong asymmetry.This is a clear sign of lattice distortion according to the bibliography [31,32].The two features with the lowest binding energies, located at 779.5 and 780.7 eV, were assigned to Co 3+ and Co 2+ ions, respectively.The signal centered at 782.2 eV was attributed to the presence of CoO in the surface of the samples [33].The abundance of this species was higher for the CC-Na sample (14%) than for the CC sample (8%), as also suggested by the results from UV-vis-NIR DRS.The two signals with the highest binding energies (785.7 and 789.3 eV) were identified as the satellite signals from Co 3+ and Co 2+ ions, respectively.The comparable intensity of both satellite signals pointed out that the surface of the catalysts was mainly composed of Co3O4 [34].On the other hand, the O 1s spectra exhibited two signals for both samples, located at 529.9 and 531.3 eV, respectively.The low binding energy signal was attributed to lattice oxygen species (Olatt).The high binding energy signal was assigned to weakly adsorbed oxygen species (Oads) [35].Finally, for the CC-Na catalyst an extra peak at a binding energy of 1071.8 eV was attributed to the presence of Na + ions.
The deconvolution and integration of the XPS spectra allowed for a quantitative analysis of the composition of the surface, as shown in Table 2.It must be noticed that the concentration of sodium on the surface of the CC-Na sample was several times higher (5.1 wt%) than in the bulk (0.4 wt%).This evidenced that the majority of the sodium atoms present in that sample were located on its surface.A certain amount had likely migrated and diffused into the spinel lattice during the  The signal centered at 782.2 eV was attributed to the presence of CoO in the surface of the samples [33].The abundance of this species was higher for the CC-Na sample (14%) than for the CC sample (8%), as also suggested by the results from UV-vis-NIR DRS.The two signals with the highest binding energies (785.7 and 789.3 eV) were identified as the satellite signals from Co 3+ and Co 2+ ions, respectively.The comparable intensity of both satellite signals pointed out that the surface of the catalysts was mainly composed of Co 3 O 4 [34].On the other hand, the O 1s spectra exhibited two signals for both samples, located at 529.9 and 531.3 eV, respectively.The low binding energy signal was attributed to lattice oxygen species (O latt ).The high binding energy signal was assigned to weakly adsorbed oxygen species (O ads ) [35].Finally, for the CC-Na catalyst an extra peak at a binding energy of 1071.8 eV was attributed to the presence of Na + ions.
The deconvolution and integration of the XPS spectra allowed for a quantitative analysis of the composition of the surface, as shown in Table 2.It must be noticed that the concentration of sodium on the surface of the CC-Na sample was several times higher (5.1 wt%) than in the bulk (0.4 wt%).This evidenced that the majority of the sodium atoms present in that sample were located on its surface.A certain amount had likely migrated and diffused into the spinel lattice during the calcination process.In addition to that, the Co 3+ /Co 2+ molar ratio of the catalyst with sodium was notably lower than for its clean counterpart, while the O ads /O latt molar ratio was higher.This was explained as a consequence of sodium atoms donating electrons to oxygen ions within the spinel lattice.The resulting increase in electronic density transformed Na sites into strong Lewis sites that are able to donate electrons to cobalt cations with a concomitant effective decrease in their overall oxidation state [22,23].This effect was further confirmed by the slight shift (0.4 eV) of the O 1s spectra of the CC-Na towards lower binding energy values.This indicated an increase in the electron density of oxygen anions and also caused a weakening of Co-O bonds, which is coherent with observed increase in the O ads /O latt molar ratio.
calcination process.In addition to that, the Co 3+ /Co 2+ molar ratio of the catalyst with sodium was notably lower than for its clean counterpart, while the Oads/Olatt molar ratio was higher.This was explained as a consequence of sodium atoms donating electrons to oxygen ions within the spinel lattice.The resulting increase in electronic density transformed Na sites into strong Lewis sites that are able to donate electrons to cobalt cations with a concomitant effective decrease in their overall oxidation state [22,23].This effect was further confirmed by the slight shift (0.4 eV) of the O 1s spectra of the CC-Na towards lower binding energy values.This indicated an increase in the electron density of oxygen anions and also caused a weakening of Co-O bonds, which is coherent with observed increase in the Oads/Olatt molar ratio.

Redox Characterisation
Figure 7a shows the H2-TPR profiles of the cobalt oxide samples.As described by other authors, the TPR profile of a Co3O4 catalyst, namely the CC sample, consisted of two different contributions to the H2 uptake [36,37].A first contribution, located at 310 °C, assignable to the reduction of Co 3+ ions into Co 2+ ; and a second contribution, centered at 350-400 °C, attributable to the subsequent reduction of Co 2+ into metallic cobalt.The TPR profile of the sample with residual sodium also exhibited these two contributions, but their shape was notably different.This was probably a consequence of the distortion of the spinel lattice caused by sodium.Furthermore, the second H2 uptake did not take the shape of a single peak, but instead was formed of at least two different contributions.This suggested the presence of Co 2+ species with different reducibilities, namely free CoO that was detected by both DRS and XPS analysis along with Co 2+ from the spinel lattice.
Immediately after the H2-TPR analysis, an O2-TPO run was carried out to check the capacity of reoxidation of the catalysts after their complete reduction to metallic cobalt.The TPO profiles, shown in Figure 7b, clearly evidenced the influence of sodium on the reoxidability of the Co3O4 catalysts.The CC sample showed a high intensity oxidation peak centered at 330 °C with a tail starting around 200 °C.The sample with sodium showed a much lower intensity peak located at about 360 °C, with a tail starting around 250 °C.Furthermore, the total O2 uptake of the Na-containing sample was only

Redox Characterisation
Figure 7a shows the H 2 -TPR profiles of the cobalt oxide samples.As described by other authors, the TPR profile of a Co 3 O 4 catalyst, namely the CC sample, consisted of two different contributions to the H 2 uptake [36,37].A first contribution, located at 310 • C, assignable to the reduction of Co 3+ ions into Co 2+ ; and a second contribution, centered at 350-400 • C, attributable to the subsequent reduction of Co 2+ into metallic cobalt.The TPR profile of the sample with residual sodium also exhibited these two contributions, but their shape was notably different.This was probably a consequence of the distortion of the spinel lattice caused by sodium.Furthermore, the second H 2 uptake did not take the shape of a single peak, but instead was formed of at least two different contributions.This suggested the presence of Co 2+ species with different reducibilities, namely free CoO that was detected by both DRS and XPS analysis along with Co 2+ from the spinel lattice.
Immediately after the H 2 -TPR analysis, an O 2 -TPO run was carried out to check the capacity of reoxidation of the catalysts after their complete reduction to metallic cobalt.The TPO profiles, shown in Figure 7b, clearly evidenced the influence of sodium on the reoxidability of the Co 3 O 4 catalysts.The CC sample showed a high intensity oxidation peak centered at 330 • C with a tail starting around 200 • C. The sample with sodium showed a much lower intensity peak located at about 360 • C, with a tail starting around 250 • C. Furthermore, the total O 2 uptake of the Na-containing sample was only 51% than that of the Na-free oxide.These results revealed that not only the sample with sodium was more difficult to reoxidize, but also that it could not be completely reoxidized, after complete reduction, below 600 • C.This was likely due to the reducing effect of sodium ions over cobalt species and the subsequent weakening of Co-O bonds, which in turn made more difficult the reincorporation of oxygen into the spinel lattice.
Catalysts 2018, 8, x FOR PEER REVIEW 8 of 15 51% than that of the Na-free oxide.These results revealed that not only the sample with sodium was more difficult to reoxidize, but also that it could not be completely reoxidized, after complete reduction, below 600 °C.This was likely due to the reducing effect of sodium ions over cobalt species and the subsequent weakening of Co-O bonds, which in turn made more difficult the reincorporation of oxygen into the spinel lattice.Nevertheless, in spite of the differences in reducibility between the two catalysts, their reduction started at the same temperature (about 200 °C) as can be seen in Figure 7a.This could be attributable to the high reducing power of hydrogen, that may hide subtle differences in reducibility.For this reason, a CH4-TPRe experiment in the absence of oxygen was conducted over both oxides, since the ability of methane for reducing is lower than that of hydrogen.The results are shown in Figure 8.The CH4-TPRe profiles of the Co3O4 catalysts consisted of two reaction peaks, in a similar fashion to the H2-TPR profiles.However, these peaks did not exhibit the same shape as the TPR peaks, nor did they appear as the same temperatures.The CC sample was characterised by a low temperature reduction peak, assigned to the reaction of methane with O 2− ions associated with Co 3+ ions, centered at 469 °C.The corresponding peak of the CC-Na sample was centered at 544 °C.On the other hand, the second reaction peak, attributed to the reaction of methane with O 2− ions associated with Co 2+ Nevertheless, in spite of the differences in reducibility between the two catalysts, their reduction started at the same temperature (about 200 • C) as can be seen in Figure 7a.This could be attributable to the high reducing power of hydrogen, that may hide subtle differences in reducibility.For this reason, a CH 4 -TPRe experiment in the absence of oxygen was conducted over both oxides, since the ability of methane for reducing is lower than that of hydrogen.The results are shown in Figure 8.
51% than that of the Na-free oxide.These results revealed that not only the sample with sodium was more difficult to reoxidize, but also that it could not be completely reoxidized, after complete reduction, below 600 °C.This was likely due to the reducing effect of sodium ions over cobalt species and the subsequent weakening of Co-O bonds, which in turn made more difficult the reincorporation of oxygen into the spinel lattice.Nevertheless, in spite of the differences in reducibility between the two catalysts, their reduction started at the same temperature (about 200 °C) as can be seen in Figure 7a.This could be attributable to the high reducing power of hydrogen, that may hide subtle differences in reducibility.For this reason, a CH4-TPRe experiment in the absence of oxygen was conducted over both oxides, since the ability of methane for reducing is lower than that of hydrogen.The results are shown in Figure 8.The CH4-TPRe profiles of the Co3O4 catalysts consisted of two reaction peaks, in a similar fashion to the H2-TPR profiles.However, these peaks did not exhibit the same shape as the TPR peaks, nor did they appear as the same temperatures.The CC sample was characterised by a low temperature reduction peak, assigned to the reaction of methane with O 2− ions associated with Co 3+ ions, centered at 469 °C.The corresponding peak of the CC-Na sample was centered at 544 °C.On the other hand, the second reaction peak, attributed to the reaction of methane with O 2− ions associated with Co 2+ The CH 4 -TPRe profiles of the Co 3 O 4 catalysts consisted of two reaction peaks, in a similar fashion to the H 2 -TPR profiles.However, these peaks did not exhibit the same shape as the TPR peaks, nor did they appear as the same temperatures.The CC sample was characterised by a low temperature reduction peak, assigned to the reaction of methane with O 2− ions associated with Co 3+ ions, centered at 469 • C. The corresponding peak of the CC-Na sample was centered at 544 • C. On the other hand, the second reaction peak, attributed to the reaction of methane with O 2− ions associated with Co 2+ ions, appeared at 627 • C for the CC sample.For the sample with sodium, it did not appear until the isothermal step at 850 • C. On the basis of these results, it was evident that the presence of sodium in the Co 3 O 4 catalyst remarkably decreased the mobility of the oxygen species of the spinel lattice, as well as the reducibility of the cobalt ions themselves.However, this was only clearly visible when the reduction of the samples was carried out with methane.
It must be considered that according to the results from the CH 4 -TPRe tests, these Co 3 O 4 catalysts were not completely reduced during the reaction process.This is because the temperature levels needed for their complete reduction with methane were higher than 600 • C, and their reoxidation occurred at temperatures around 300 • C. For this reason, the results from the O 2 -TPO analysis may not be too relevant, but they could serve as an additional indication of the detrimental effect of sodium on the redox properties of Co 3 O 4 .

Catalytic Performance
The activity tests were carried out three times consecutively for each catalyst.In both cases, the second and third cycles were characterised by an identical light-off curve.The light-off curves for the first and second/third cycles are shown in Figure 9.It was clear that the sample without sodium was far more active than the sample with sodium, thus evidencing the inhibiting effect caused by this metal.Considering that the primary mechanism of methane oxidation over metallic oxides is the Mars-van Krevelen mechanism, the negative effect of sodium on the activity of the CC-Na catalyst could be explained as a consequence of the reduced oxygen mobility.This was demonstrated by CH 4 -TPRe, and the decrease on the amount of Co 3+ on the surface and the subsequent decrease in the active lattice oxygen species, as shown by the XPS results.However, it must be noticed that both samples were able to achieve a 100% methane conversion at 600 • C.This pointed out that the aforementioned inhibition effect was only apparent at low temperatures, as can be observed from the marked differences between the T 10 and T 50 values (Table 3).
Catalysts 2018, 8, x FOR PEER REVIEW 9 of 15 ions, appeared at 627 °C for the CC sample.For the sample with sodium, it did not appear until the isothermal step at 850 °C.On the basis of these results, it was evident that the presence of sodium in the Co3O4 catalyst remarkably decreased the mobility of the oxygen species of the spinel lattice, as well as the reducibility of the cobalt ions themselves.However, this was only clearly visible when the reduction of the samples was carried out with methane.It must be considered that according to the results from the CH4-TPRe tests, these Co3O4 catalysts were not completely reduced during the reaction process.This is because the temperature levels needed for their complete reduction with methane were higher than 600 °C, and their reoxidation occurred at temperatures around 300 °C.For this reason, the results from the O2-TPO analysis may not be too relevant, but they could serve as an additional indication of the detrimental effect of sodium on the redox properties of Co3O4.

Catalytic Performance
The activity tests were carried out three times consecutively for each catalyst.In both cases, the second and third cycles were characterised by an identical light-off curve.The light-off curves for the first and second/third cycles are shown in Figure 9.It was clear that the sample without sodium was far more active than the sample with sodium, thus evidencing the inhibiting effect caused by this metal.Considering that the primary mechanism of methane oxidation over metallic oxides is the Mars-van Krevelen mechanism, the negative effect of sodium on the activity of the CC-Na catalyst could be explained as a consequence of the reduced oxygen mobility.This was demonstrated by CH4-TPRe, and the decrease on the amount of Co 3+ on the surface and the subsequent decrease in the active lattice oxygen species, as shown by the XPS results.However, it must be noticed that both samples were able to achieve a 100% methane conversion at 600 °C.This pointed out that the aforementioned inhibition effect was only apparent at low temperatures, as can be observed from the marked differences between the T10 and T50 values (Table 3).A differential kinetic analysis was performed for conversion values below 10%, assuming a first order with respect to methane and zeroth with respect to oxygen, which is a suitable model for the Mars-van Krevelen mechanism in the presence of excess oxygen [38,39].The results shown in Table 3 further confirmed the negative influence of residual sodium over the CC-Na catalyst.The activation energy of this sample was significantly higher with respect to the CC sample.In addition, the preexponential factor of the CC-Na sample was about four times higher than that of the CC sample, A differential kinetic analysis was performed for conversion values below 10%, assuming a first order with respect to methane and zeroth with respect to oxygen, which is a suitable model for the Mars-van Krevelen mechanism in the presence of excess oxygen [38,39].The results shown in Table 3 further confirmed the negative influence of residual sodium over the CC-Na catalyst.The activation energy of this sample was significantly higher with respect to the CC sample.In addition, the pre-exponential factor of the CC-Na sample was about four times higher than that of the CC sample, which could be a consequence of the beneficial effect of residual sodium over the structural properties of this catalyst.The fact that the first reaction cycle was characterised by a different light-off curve pointed out that it worked as 'conditioning' cycle, after which the tested catalysts became stable under the employed reaction conditions.For this reason, the tested catalysts were characterised after the reaction tests to look for differences with respect to fresh counterparts that could be caused by this conditioning process.The results from the textural/structural characterisation, shown in Table 4, revealed a loss of specific surface area and pore volume for both catalysts.The loss of specific surface area was comparable between the two catalysts (around 33% in both cases).The loss of pore volume was more severe for the Na-free catalyst.Additionally, a general enlargement of the average pore size was observed for both catalysts.Interestingly, the loss of surface area and pore volume of the CC catalyst was not accompanied by a marked increase in the crystallite size, as it would be expected if the effect were caused by sintering.Moreover, the shrinkage of the cell size was noticed for both samples (from 8.052 to 8.046 Å over the CC sample and from 8.064 to 8.048 Å over the CC-Na sample).All this suggested that the conditioning cycle caused a moderate surface restructuring of the catalysts that made them slightly less active, but more thermally stable.
The effect of such surface restructuration could be seen through the XPS analysis of the samples after the light-off runs.A decrease of the Co 3+ /Co 2+ molar ratio was found for both samples, from 1.96 to 1.64 for the CC sample and from 1.20 to 1.02 for the CC-Na sample; which was accompanied by an increase in the O ads /O latt molar ratio from 0.40 to 0.52 for the Na-free oxide and from 0.57 to 0.70 for the sample with sodium.In addition to that, the CC-Na sample suffered an increase in the Na surface concentration with respect to its fresh counterpart, achieving a molar concentration of 11.3% (9.0 wt%) of sodium.

Stability over a Prolonged Time
Finally, the stability of the catalysts over a prolonged period of time (150 h) was tested at 450 • C.This was to check if the presence of residual sodium could also affect the performance of the catalysts.It must be pointed out that a fresh sample of both oxide catalysts was used for this run.Both catalysts were thermally stable, although the conversion level they could achieve was remarkably different (Figure 10).The catalyst without sodium achieved a constant methane conversion of around 75%.The CC-Na catalyst could only achieve around 40% conversion for the entire time span of the test.This evidenced that the presence of residual sodium did not affect the thermal stability of the catalysts, but it did influence their activity.
The characterisation of the post-run catalysts revealed the effect of prolonged periods of operation over their properties.As can be seen in Table 4, the loss of properties for the sample without sodium was less marked than for the catalyst with sodium.However, interestingly, none of the catalysts exhibited any crystallite size growth after the stability test, which pointed out that they were structurally stable at temperatures up to 450 • C. catalysts exhibited any crystallite size growth after the stability test, which pointed out that they were structurally stable at temperatures up to 450 °C.Comparing the impact caused by the light-off cycles with the loss of properties caused by the prolonged operation at constant temperature (450 °C), it became clear that operating at higher temperatures (600 °C) for short periods of time is more critical than operating at lower temperatures for extended periods of time.Indeed, if we compare the conversion levels obtained at 450 °C for the two aforementioned cases, it can be seen that the difference in conversion between the first and the third light-off cycle (around 20%) was much larger than the difference in conversion between the first light-off cycle and the start of the stability tests (around 5%).This evidenced that the loss of properties in the light-off cycles took place during the time the catalysts were operated at high temperatures (above 450 °C).For the stability tests it took place during the first hour of the test.

Catalyst Preparation
Two bulk Co3O4 catalysts were prepared following a simple precipitation route, described elsewhere [18].This route consisted of an aqueous basic precipitation of a cobalt precursor, resulting in the formation of cobalt hydroxycarbonate.A solution of Na2CO3 (Sigma-Aldrich, St. Louis, MO, USA) 1.2 M was added drop-by-drop to 100 mL of Co(NO3)2•6H2O (ACS reagent, Alfa Aesar, Haverhill, MA, USA) 0.5 M, while the temperature was kept constant at 80 °C, until pH 8.5 was attained.After that, the obtained precipitates were filtrated and washed with different amounts of deionised water between one and eight litres.In this way, the so-called CC-Na precursor was washed with two litres of water to allow some residual sodium to remain within the catalytic precursor.On the contrary, the CC precursor was washed with up to eight litres of water with the aim of removing all the remaining sodium from the precipitation process.The catalyst precursors were dried in static air at 110 °C for 16 h and subjected to calcination in static air at 600 °C to produce the final catalysts.Comparing the impact caused by the light-off cycles with the loss of properties caused by the prolonged operation at constant temperature (450 • C), it became clear that operating at higher temperatures (600 • C) for short periods of time is more critical than operating at lower temperatures for extended periods of time.Indeed, if we compare the conversion levels obtained at 450 • C for the two aforementioned cases, it can be seen that the difference in conversion between the first and the third light-off cycle (around 20%) was much larger than the difference in conversion between the first light-off cycle and the start of the stability tests (around 5%).This evidenced that the loss of properties in the light-off cycles took place during the time the catalysts were operated at high temperatures (above 450 • C).For the stability tests it took place during the first hour of the test.

Catalyst Preparation
Two bulk Co 3 O 4 catalysts were prepared following a simple precipitation route, described elsewhere [18].This route consisted of an aqueous basic precipitation of a cobalt precursor, resulting in the formation of cobalt hydroxycarbonate.A solution of Na 2 CO 3 (Sigma-Aldrich, St. Louis, MO, USA) 1.2 M was added drop-by-drop to 100 mL of Co(NO 3 ) 2 •6H 2 O (ACS reagent, Alfa Aesar, Haverhill, MA, USA) 0.5 M, while the temperature was kept constant at 80 • C, until pH 8.5 was attained.After that, the obtained precipitates were filtrated and washed with different amounts of deionised water between one and eight litres.In this way, the so-called CC-Na precursor was washed with two litres of water to allow some residual sodium to remain within the catalytic precursor.On the contrary, the CC precursor was washed with up to eight litres of water with the aim of removing all the remaining sodium from the precipitation process.The catalyst precursors were dried in static air at 110 • C for 16 h and subjected to calcination in static air at 600 • C to produce the final catalysts.
Catalytic activity was measured from 200 to 600 • C. Conversion measurements were taken each 25 • C. Additionally, stability tests were carried out for a total time on stream of 150 h at 450 • C. Methane conversion was calculated by the difference between inlet and outlet CH 4 concentrations.Inlet and outlet streams were analysed using an on-line Agilent Technologies 7890N gas chromatograph (Agilent Technologies, Santa Clara, CA, USA) equipped with a thermal conductivity detector (TCD) and two columns: a PLOT 5A molecular sieve column (Agilent Technologies, Santa Clara, CA, USA) for the analysis of CH 4 , O 2 , N 2 and CO; and a PLOT U column (Agilent Technologies, Santa Clara, CA, USA) for CO 2 analysis.To ensure that the obtained catalytic results were not affected by both mass and heat transfer limitations, the criteria for intra-granular and extra-granular mass diffusion, energy diffusion and temperature gradients were checked, according to the Eurokin procedure (Eurokin, Delft, The Netherlands) [40].

Conclusions
Two bulk Co 3 O 4 catalysts were prepared by a simple precipitation method using sodium carbonate as the precipitating agent.One of the catalysts was properly washed after the precipitation process while the other kept some residual sodium ions.
The characterization of the two catalysts evidenced that the presence of the residual sodium ions had a great impact on the properties of the catalyst.Firstly, the insertion of the sodium atoms into the spinel lattice induced a distortion that improved the structural properties of the Co 3 O 4 , inhibiting crystallisation and increasing the specific surface area.However, this distortion was also partially responsible for a decrease in the reducibility of the catalyst.This was in a subtle way when the reducing agent was hydrogen, but severely when the samples were reduced with methane.On the other hand, the concentration of sodium ions on the surface of the catalyst induced a reduction of the Co 3+ ions into Co 2+ , due to their high Lewis acidic properties, thus increasing the electron density within the oxygen ions of the lattice and weakening the Co-O bonds, making it more difficult for the catalyst to reoxidize.This subsequently provoked a decrease in the abundance of lattice oxygen species that were active for the methane oxidation reaction.The overall effect of the presence of residual sodium in the Co 3 O 4 was a notable decrease in the catalytic activity, especially at low temperatures.These results pointed out the importance of properly washing the catalytic precursors after the synthesis process, especially when the precipitating agent was a sodium base.In this sense, the use of ammonia-based precipitants or synthesis routes such as sol-gel complexation may be considered as suitable alternatives for preparing active Co 3 O 4 catalysts for methane combustion.

Figure 1 .
Figure 1.pH of the filtrate from the washing step of the catalytic precursors (CC and CC-Na).

Figure 1 .
Figure 1.pH of the filtrate from the washing step of the catalytic precursors (CC and CC-Na).

Figure 2 .
Figure 2. (a) XRD diffractograms of the catalysts; (b) detail of the most intense diffraction signal.

Figure 2 .
Figure 2. (a) XRD diffractograms of the catalysts; (b) detail of the most intense diffraction signal.

Figure 3 .
Figure 3. Pore size distributions of the catalysts.

Figure 4 .
Figure 4. (a) UV-vis-NIR DRS spectra of the samples; (b) detail of the region around 1200 nm.

Figure 3 .
Figure 3. Pore size distributions of the catalysts.

Figure 3 .
Figure 3. Pore size distributions of the catalysts.

Figure 4 .
Figure 4. (a) UV-vis-NIR DRS spectra of the samples; (b) detail of the region around 1200 nm.

Figure 4 .
Figure 4. (a) UV-vis-NIR DRS spectra of the samples; (b) detail of the region around 1200 nm.

Figure
Figure6a,b show the Co 2p and O 1s XPS spectra of the samples, respectively.The Co 2p3/2 spectra of both oxides was composed of five different signals.The two features with the lowest binding energies, located at 779.5 and 780.7 eV, were assigned to Co 3+ and Co 2+ ions, respectively.The signal centered at 782.2 eV was attributed to the presence of CoO in the surface of the samples[33].The abundance of this species was higher for the CC-Na sample (14%) than for the CC sample (8%), as also suggested by the results from UV-vis-NIR DRS.The two signals with the highest binding energies (785.7 and 789.3 eV) were identified as the satellite signals from Co 3+ and Co 2+ ions, respectively.The comparable intensity of both satellite signals pointed out that the surface of the catalysts was mainly composed of Co3O4[34].On the other hand, the O 1s spectra exhibited two signals for both samples, located at 529.9 and 531.3 eV, respectively.The low binding energy signal was attributed to lattice oxygen species (Olatt).The high binding energy signal was assigned to weakly adsorbed oxygen species (Oads)[35].Finally, for the CC-Na catalyst an extra peak at a binding energy of 1071.8 eV was attributed to the presence of Na + ions.The deconvolution and integration of the XPS spectra allowed for a quantitative analysis of the composition of the surface, as shown in Table2.It must be noticed that the concentration of sodium on the surface of the CC-Na sample was several times higher (5.1 wt%) than in the bulk (0.4 wt%).This evidenced that the majority of the sodium atoms present in that sample were located on its surface.A certain amount had likely migrated and diffused into the spinel lattice during the

Figure
Figure 6a,b show the Co 2p and O 1s XPS spectra of the samples, respectively.The Co 2p 3/2 spectra of both oxides was composed of five different signals.The two features with the lowest binding energies, located at 779.5 and 780.7 eV, were assigned to Co 3+ and Co 2+ ions, respectively.The signal centered at 782.2 eV was attributed to the presence of CoO in the surface of the samples[33].The abundance of this species was higher for the CC-Na sample (14%) than for the CC sample (8%), as also suggested by the results from UV-vis-NIR DRS.The two signals with the highest binding energies (785.7 and 789.3 eV) were identified as the satellite signals from Co 3+ and Co 2+ ions, respectively.The comparable intensity of both satellite signals pointed out that the surface of the catalysts was mainly composed of Co 3 O 4[34].On the other hand, the O 1s spectra exhibited two signals for both samples, located at 529.9 and 531.3 eV, respectively.The low binding energy signal was attributed to lattice oxygen species (O latt ).The high binding energy signal was assigned to weakly adsorbed oxygen species (O ads )[35].Finally, for the CC-Na catalyst an extra peak at a binding energy of 1071.8 eV was attributed to the presence of Na + ions.The deconvolution and integration of the XPS spectra allowed for a quantitative analysis of the composition of the surface, as shown in Table2.It must be noticed that the concentration of sodium on the surface of the CC-Na sample was several times higher (5.1 wt%) than in the bulk (0.4 wt%).This evidenced that the majority of the sodium atoms present in that sample were located on its surface.A certain amount had likely migrated and diffused into the spinel lattice during the calcination process.In addition to that, the Co 3+ /Co 2+ molar ratio of the catalyst with sodium was notably lower than for its clean counterpart, while the O ads /O latt molar ratio was higher.This was explained as a consequence of sodium atoms donating electrons to oxygen ions within the spinel lattice.The resulting increase in

Figure 6 .
Figure 6.(a) Co 2p XPS spectra of the samples; (b) O 1s XPS spectra of the samples.

Figure 6 .
Figure 6.(a) Co 2p XPS spectra of the samples; (b) O 1s XPS spectra of the samples.

Figure 9 .
Figure 9. Light-off curves of the catalysts.

Figure 9 .
Figure 9. Light-off curves of the catalysts.

Figure 10 .
Figure 10.Stability of the catalysts over time.

Figure 10 .
Figure 10.Stability of the catalysts over time.

Table 2 .
Surface composition results from XPS.

Table 2 .
Surface composition results from XPS.

Table 3 .
Results from the kinetic analysis of activity results.

Table 4 .
Comparison of textural/structural properties between fresh and used catalysts.