Potential of Ceria-Zirconia-Based Materials in Carbon Soot Oxidation for Gasoline Particulate Filters

: ZrO 2 and Ce 0.8 Zr 0.2 O 2 mixed oxides were prepared and tested in the oxidation of carbon soot at di ﬀ erent oxygen partial pressures and degrees of catalyst / soot contact to investigate their activity under typical gasoline direct injection (GDI) operating conditions. Under reductive atmospheres, generation of oxygen vacancies occurs in Ce 0.8 Zr 0.2 O 2 , while no reduction is observed on ZrO 2 . Both materials can oxidize carbon under high oxygen partial pressures; however, at low oxygen partial pressures, the presence of carbon can contribute to the reduction of the catalyst and formation of oxygen vacancies, which can then be used for soot oxidation, increasing the overall performance. This mechanism is more e ﬃ cient in Ce 0.8 Zr 0.2 O 2 than ZrO 2 , and depends heavily on the interaction and the degree of contact between soot and catalyst. Thus, the ability to form oxygen vacancies at lower temperatures is particularly helpful to oxidize soot at low oxygen partial pressures, and with higher CO 2 selectivity under conditions typically found in GDI engine exhaust gases.


Introduction
The gasoline engine market, based on direct injection (DI) technology is growing rapidly, especially in relation to their better efficiency and lower CO 2 emissions [1]. In addition, for 2020 the European Union has imposed a limit on CO 2 emissions for cars of 95 g/km; the GDI engines are able to increase fuel economy, and consequently reduce these emissions [2]. On the other hand, however, these engines produce a higher amount of soot than conventional gasoline engines. In particular, attention is focused not only on particulate emissions (PM), but also on the particulate number (PN), and consequently it is necessary to implement new emission control strategies [3]. As required for diesel engines, it is possible to introduce a particulate filter to reduce emissions of soot: the so-called gasoline particulate filter (GPF) [4][5][6]. However, due to the different operating conditions of a diesel engine and a GDI engine [3], it is not possible to simply transfer the technologies optimized and developed in the last twenty years for diesel particulate filters (DPF), but it is necessary to investigate in detail the mechanisms of soot accumulation and oxidation under typical GDI operating conditions [7]. Exactly as it happens for DPFs, the accumulation phase must necessarily be followed by a regeneration phase, in order to oxidize the accumulated soot and prevent an increased pressure drop [8]. Since particulate oxidizes at around 600 • C, in order to save fuel and to avoid filter damage, a catalyst impregnated on the filter, which lowers the combustion temperature, can be used. To ensure regeneration, however, issues arising from the different atmosphere of the exhaust gas of a GDI and a diesel engine must be taken into consideration [3]. In the diesel engine, in fact, regeneration is favored by the high oxygen content and by the presence of nitrogen oxides, NO x , which participate in the oxidation reaction. Consequently, the catalyst formulations developed for the DPFs cannot be directly transferred to the GDI engines, real systems. Tight contact has been obtained by grinding the mixture in an agate mortar for ten minutes. This is less representative of real conditions, but results are generally more reproducible, and a much stronger interaction is observed [42]. A third type of mixing has only been employed in the case of ceria-zirconia, and it is realized by milling powders for 8 h in a high-energy mill to obtained a supertight contact [40]. This is not representative of the real application, but it is a good reference for a high degree of redox interaction that can be obtained between carbon and catalyst, and is useful in the understanding of the mechanism of reaction and on the intrinsic activity of reducible materials. This contact is characterized by a different morphology, where a layer of soot forms a thin shell covering ceria-zirconia crystallite cores promoting interactions at a nanoscale at the ceria-zirconia/carbon interface [40,41]. The different contact modes are represented in Scheme 1.
Catalysts 2020, 10, x FOR PEER REVIEW 3 of 13 Two types of mixing conditions have been employed to put into contact carbon soot and catalysts to simulate different degrees of interactions. For loose contact, soot and catalyst have been mixed in a vial for two minutes, which is representative of the weak degree of interaction that is found in real systems. Tight contact has been obtained by grinding the mixture in an agate mortar for ten minutes. This is less representative of real conditions, but results are generally more reproducible, and a much stronger interaction is observed [42]. A third type of mixing has only been employed in the case of ceria-zirconia, and it is realized by milling powders for 8 h in a high-energy mill to obtained a supertight contact [40]. This is not representative of the real application, but it is a good reference for a high degree of redox interaction that can be obtained between carbon and catalyst, and is useful in the understanding of the mechanism of reaction and on the intrinsic activity of reducible materials. This contact is characterized by a different morphology, where a layer of soot forms a thin shell covering ceria-zirconia crystallite cores promoting interactions at a nanoscale at the ceria-zirconia/carbon interface [40,41]. The different contact modes are represented in Scheme 1. Scheme 1. Different kinds of contact obtained by mixing soot (black) and catalyst (yellow) using a spatula, an agate mortar and a high-energy mill.
The composition and BET surface area of investigated materials are reported in Table 1. Ceriazirconia and zirconia show a surface area in the range of 60-80 m 2 /g. When soot is mixed with the catalyst in a loose and tight contact mode, the surface area is not affected, while the surface area of ceria-zirconia decreases by effect of milling from 79 to 29 m 2 /g. This is typically observed upon the milling of high surface area powders, and is probably due to the adhesion of fine crystallites on the surface of larger particles [40,43]. X-ray diffraction (XRD) profiles are shown in Figures 1A,B. ZrO2 shows the simultaneous presence of tetragonal (space group P42/nmc; main reflection at 30.2°) and monoclinic ZrO2 (space group P121/c1; main reflection at 28.2°) and milling with soot does not modify the XRD profile. Ceria-Scheme 1. Different kinds of contact obtained by mixing soot (black) and catalyst (yellow) using a spatula, an agate mortar and a high-energy mill.
The composition and BET surface area of investigated materials are reported in Table 1. Ceria-zirconia and zirconia show a surface area in the range of 60-80 m 2 /g. When soot is mixed with the catalyst in a loose and tight contact mode, the surface area is not affected, while the surface area of ceria-zirconia decreases by effect of milling from 79 to 29 m 2 /g. This is typically observed upon the milling of high surface area powders, and is probably due to the adhesion of fine crystallites on the surface of larger particles [40,43]. X-ray diffraction (XRD) profiles are shown in Figure 1A,B. ZrO 2 shows the simultaneous presence of tetragonal (space group P42/nmc; main reflection at 30.2 • ) and monoclinic ZrO 2 (space group P121/c1; main reflection at 28.2 • ) and milling with soot does not modify the XRD profile. Ceria-zirconia (CZ) catalysts ( Figure 1B) exhibit reflections characteristic of a pure fluorite phase. The soot/CZ mixture in loose and tight contact do not evidence any difference with parent CZ. The XRD profiles do not show significant modification of the ceria-zirconia structure, and all samples are indexed in a fluorite cubic cell. The CZST sample shows a sharpening of XRD signals due to an increase of the particle size. In addition, a small signal appears at ca. 30.5 degrees, due to the main reflection of the (011) plane of ZrO 2 , originating from abrasion during milling. Rietveld analysis also reveals a small decrease of the lattice cell parameter due to insertion of ZrO 2 into CZ, with a slight change in stoichiometry of solid solution that goes from Ce 0.8 Zr 0.2 O 2 to Ce 0.78 Zr 0.22 O 2 in agreement with previous observations [40]. No signals of reaction between the oxide and carbon are revealed by XRD.
Catalysts 2020, 10, x FOR PEER REVIEW 4 of 13 zirconia (CZ) catalysts ( Figure 1B) exhibit reflections characteristic of a pure fluorite phase. The soot/CZ mixture in loose and tight contact do not evidence any difference with parent CZ. The XRD profiles do not show significant modification of the ceria-zirconia structure, and all samples are indexed in a fluorite cubic cell. The CZST sample shows a sharpening of XRD signals due to an increase of the particle size. In addition, a small signal appears at ca. 30.5 degrees, due to the main reflection of the (011) plane of ZrO2, originating from abrasion during milling. Rietveld analysis also reveals a small decrease of the lattice cell parameter due to insertion of ZrO2 into CZ, with a slight change in stoichiometry of solid solution that goes from Ce0.8Zr0.2O2 to Ce0.78Zr0.22O2 in agreement with previous observations [40]. No signals of reaction between the oxide and carbon are revealed by XRD. In order to characterize the reduction behavior of the materials and understand the role of carbon in the removal of lattice oxygen, we have further investigated the redox and oxygen storage properties of pure catalyst and catalyst soot mixtures under different contact conditions by H2-TPR and OSC measurements; the results are summarized in Figures 2 and 3, respectively. For zirconia, a temperature-programmed reduction (TPR) feature of a typical non-reducible support was found, while CZ shows one broad hydrogen uptake peak centered at around 558 °C, due to the reduction of surface and bulk Ce 4+ cations [44]. In carbon-containing samples, the H2 consumption peak shifts to progressively lower temperatures as the contact become more intimate. The reduction peak for bare CZ centered at 558 °C progressively shifts to 530, 506 and 420 °C, respectively for loose, tight and supertight contact, with an overall reduction degree, similar for all the investigated samples, of 45%. No evidence for reduction is found in Zr samples. In bare CZ, the reduction peak is correlated to consumption of H2 with the formation of H2O. With CZ/soot mixtures, the outlet gas composition has been followed by MS/FTIR, indicating the production of H2O (see inset of Figure 2A), with no formation of CH4, thus excluding the reduction of carbon by hydrogen. Figures 2C,D show CO formation profiles in TPR, as followed by Fourier-transform infrared spectroscopy (FTIR). Pure CZ shows the development of a very low amount of CO at around 600 °C, likely related to the desorption of residual carbonate species adsorbed on the surface. When soot/catalyst mixtures are investigated, the amount of CO formed during the reaction increases by increasing the contact with soot. No evidence for CO formation is observed in Zr samples, regardless of the contact conditions. Overall, these results indicate that deposited carbon promotes the reduction of ceria-zirconia as a function of the degree of interaction with the surface; the more robust the carbon/ceria contact, the lower the reduction temperature. This is particularly evident in CZST, where the morphology of carbon distributed as a thin envelope on the surface of the catalyst acts as a booster for the reduction reaction, with a reduction profile centered 140 °C lower than bare CZ. One possible mechanism is that carbon at the interface forms C-O* adsorbed species that act as a driving force for the reduction of the In order to characterize the reduction behavior of the materials and understand the role of carbon in the removal of lattice oxygen, we have further investigated the redox and oxygen storage properties of pure catalyst and catalyst soot mixtures under different contact conditions by H 2 -TPR and OSC measurements; the results are summarized in Figures 2 and 3, respectively. For zirconia, a temperature-programmed reduction (TPR) feature of a typical non-reducible support was found, while CZ shows one broad hydrogen uptake peak centered at around 558 • C, due to the reduction of surface and bulk Ce 4+ cations [44]. In carbon-containing samples, the H 2 consumption peak shifts to progressively lower temperatures as the contact become more intimate. The reduction peak for bare CZ centered at 558 • C progressively shifts to 530, 506 and 420 • C, respectively for loose, tight and supertight contact, with an overall reduction degree, similar for all the investigated samples, of 45%. No evidence for reduction is found in Zr samples. In bare CZ, the reduction peak is correlated to consumption of H 2 with the formation of H 2 O. With CZ/soot mixtures, the outlet gas composition has been followed by MS/FTIR, indicating the production of H 2 O (see inset of Figure 2A), with no formation of CH 4 , thus excluding the reduction of carbon by hydrogen. Figure 2C,D show CO formation profiles in TPR, as followed by Fourier-transform infrared spectroscopy (FTIR). Pure CZ shows the development of a very low amount of CO at around 600 • C, likely related to the desorption of residual carbonate species adsorbed on the surface. When soot/catalyst mixtures are investigated, the amount of CO formed during the reaction increases by increasing the contact with soot. No evidence for CO formation is observed in Zr samples, regardless of the contact conditions. Overall, these results indicate that deposited carbon promotes the reduction of ceria-zirconia as a function of the degree of interaction with the surface; the more robust the carbon/ceria contact, the lower the reduction temperature. This is particularly evident in CZST, where the morphology of carbon distributed as a thin envelope on the surface of the catalyst acts as a booster for the reduction reaction, with a reduction profile centered 140 • C lower than bare CZ. One possible mechanism is that carbon at the interface forms C-O* adsorbed species that act as a driving force for the reduction of the material. This adsorbed oxygen then reacts more easily with hydrogen, forming H 2 O. In other words, carbon, when finely dispersed over ceria-zirconia, can act as a catalyst for ceria-zirconia reduction reaction. Traces of CO are also formed by direct oxidation of carbon with lattice oxygen; again, these are more evident in CZST, where the carbon is finely dispersed over ceria-zirconia.
Catalysts 2020, 10, x FOR PEER REVIEW 5 of 13 material. This adsorbed oxygen then reacts more easily with hydrogen, forming H2O. In other words, carbon, when finely dispersed over ceria-zirconia, can act as a catalyst for ceria-zirconia reduction reaction. Traces of CO are also formed by direct oxidation of carbon with lattice oxygen; again, these are more evident in CZST, where the carbon is finely dispersed over ceria-zirconia. Similar results are confirmed by oxygen storage capacity analysis carried out on pure zirconia and ceria-zirconia, and on soot-catalyst mixtures at different temperatures, as seen in Figure 3. The data were collected according to the method described in the experimental part; zirconia, as expected, does not show any measurable oxygen storage capacity, while OSC of ceria-zirconia increases from 1796 µg O2/g at 300 °C, to 2922 and 8762 µg O2/g, respectively, at 400 °C and 500 °C. For bare CZ and CZL, removal of oxygen is similar in the temperature range investigated, while for tight and supertight contact, a higher OSC has been found, confirming the enhanced "reducibility" of the material in the presence of soot. reaction. Traces of CO are also formed by direct oxidation of carbon with lattice oxygen; again, these are more evident in CZST, where the carbon is finely dispersed over ceria-zirconia. Similar results are confirmed by oxygen storage capacity analysis carried out on pure zirconia and ceria-zirconia, and on soot-catalyst mixtures at different temperatures, as seen in Figure 3. The data were collected according to the method described in the experimental part; zirconia, as expected, does not show any measurable oxygen storage capacity, while OSC of ceria-zirconia increases from 1796 µg O2/g at 300 °C, to 2922 and 8762 µg O2/g, respectively, at 400 °C and 500 °C. For bare CZ and CZL, removal of oxygen is similar in the temperature range investigated, while for tight and supertight contact, a higher OSC has been found, confirming the enhanced "reducibility" of the material in the presence of soot. Similar results are confirmed by oxygen storage capacity analysis carried out on pure zirconia and ceria-zirconia, and on soot-catalyst mixtures at different temperatures, as seen in Figure 3. The data were collected according to the method described in the experimental part; zirconia, as expected, does not show any measurable oxygen storage capacity, while OSC of ceria-zirconia increases from 1796 µg O 2 /g at 300 • C, to 2922 and 8762 µg O 2 /g, respectively, at 400 • C and 500 • C. For bare CZ and CZL, removal of oxygen is similar in the temperature range investigated, while for tight and supertight contact, a higher OSC has been found, confirming the enhanced "reducibility" of the material in the presence of soot.
Carbon particles can act as an "activator" of the oxygen, moving it towards the surface and making it more available for the hydrogen present in the gas phase, as depicted in Scheme 2. The higher the contact between carbon and the catalyst, the higher the amount of oxygen that can be transferred at a lower temperature. Carbon particles can act as an "activator" of the oxygen, moving it towards the surface and making it more available for the hydrogen present in the gas phase, as depicted in Scheme 2. The higher the contact between carbon and the catalyst, the higher the amount of oxygen that can be transferred at a lower temperature.

Scheme 2.
Mechanism of reaction for oxygen transfer over soot/CZ mixtures.

Catalytic Activity
The catalytic activity has been investigated by means of thermogravimetric experiments and temperature programmed oxidation under different oxygen concentration, with the objective to understand whether a reducible CZ and a non-reducible Zr material can be used under oxygen conditions, simulating those of gasoline exhaust gases, in which the amount of oxygen is typically low. Representative weight loss profiles obtained for bare soot and CZ in loose contact under different atmospheres (air, 1% O2/N2 and N2) are shown in Figure 4. The non-catalytic reaction ( Figure 4A) is strongly dependent on the amount of oxygen in the gas phase. Under inert atmosphere a negligible weight loss is observed, while in the presence of O2 a complete oxidation is reached with a T50 of 705 °C (1O2/N2) and 630 °C (air). When soot is mixed with CZ in loose contact mode ( Figure 4B), a remarkable increase in the activity is achieved independently of the O2 concentration. The oxidation of carbon in CZL in the absence of oxygen is not quantitative, and is associated with the oxidation of the soot by the surface/bulk oxygens of ceria-zirconia. The weight loss (2.7%) is due to the formation of CO/CO2 evolving from the system, and it also accounts

Catalytic Activity
The catalytic activity has been investigated by means of thermogravimetric experiments and temperature programmed oxidation under different oxygen concentration, with the objective to understand whether a reducible CZ and a non-reducible Zr material can be used under oxygen conditions, simulating those of gasoline exhaust gases, in which the amount of oxygen is typically low. Representative weight loss profiles obtained for bare soot and CZ in loose contact under different atmospheres (air, 1% O 2 /N 2 and N 2 ) are shown in Figure 4. Carbon particles can act as an "activator" of the oxygen, moving it towards the surface and making it more available for the hydrogen present in the gas phase, as depicted in Scheme 2. The higher the contact between carbon and the catalyst, the higher the amount of oxygen that can be transferred at a lower temperature.

Scheme 2.
Mechanism of reaction for oxygen transfer over soot/CZ mixtures.

Catalytic Activity
The catalytic activity has been investigated by means of thermogravimetric experiments and temperature programmed oxidation under different oxygen concentration, with the objective to understand whether a reducible CZ and a non-reducible Zr material can be used under oxygen conditions, simulating those of gasoline exhaust gases, in which the amount of oxygen is typically low. Representative weight loss profiles obtained for bare soot and CZ in loose contact under different atmospheres (air, 1% O2/N2 and N2) are shown in Figure 4. The non-catalytic reaction ( Figure 4A) is strongly dependent on the amount of oxygen in the gas phase. Under inert atmosphere a negligible weight loss is observed, while in the presence of O2 a complete oxidation is reached with a T50 of 705 °C (1O2/N2) and 630 °C (air). When soot is mixed with CZ in loose contact mode ( Figure 4B), a remarkable increase in the activity is achieved independently of the O2 concentration. The oxidation of carbon in CZL in the absence of oxygen is not quantitative, and is associated with the oxidation of the soot by the surface/bulk oxygens of ceria-zirconia. The weight loss (2.7%) is due to the formation of CO/CO2 evolving from the system, and it also accounts The non-catalytic reaction ( Figure 4A) is strongly dependent on the amount of oxygen in the gas phase. Under inert atmosphere a negligible weight loss is observed, while in the presence of O 2 a complete oxidation is reached with a T50 of 705 • C (1O 2 /N 2 ) and 630 • C (air). When soot is mixed with CZ in loose contact mode ( Figure 4B), a remarkable increase in the activity is achieved independently of the O 2 concentration. The oxidation of carbon in CZL in the absence of oxygen is not quantitative, and is associated with the oxidation of the soot by the surface/bulk oxygens of ceria-zirconia. The weight loss (2.7%) is due to the formation of CO/CO 2 evolving from the system, and it also accounts for the lattice oxygen of CZ used for oxidation during the reaction. Weight loss in 1% O 2 /N 2 and in air shows very similar profiles, but when a lower amount of oxygen is used, a shift of the oxidation curve profile to higher temperature is observed; in particular, T50 shifts from 515 • C for reaction under air to 545 • C for reaction in 1% O 2 /N 2 . Figure 5 shows the T50 for all the investigated catalysts as a function of the reaction atmosphere. Under an inert atmosphere, weight loss is negligible in bare soot and in zirconia/soot mixtures regardless of the contact type; this is consistent with the lack of lattice oxygen available for promotion of redox reaction in zirconia and the lack of volatile matter in soot composition, which can be released during thermal treatment. No oxidation is therefore observed in these samples under an inert atmosphere. With the introduction of ceria in CZ, a higher activity can be observed already under inert atmosphere, with a T50 for the loose contact of 775 • C, which progressively decreases to 546 and 420 • C for tight and supertight contact, respectively. This is consistent with the fact that ceria-zirconia-based systems can use the surface/bulk oxygen to oxidize soot even in the absence of oxygen in the gas phase, through the oxidation of carbon at the carbon/CZ interface with the formation of an oxygen vacancy [24,45]. The oxidation reaction proceeds to consume available oxygen, and the activity is closely related to the amount of reducible oxygen/OSC, and is progressively inhibited due to the lack of reoxidation of the material when the inert gas phase is used [46,47]. The introduction of oxygen facilitates oxidation in all formulations. If we focus on CZ, we can observe that the catalytic activity in 1% O 2 /N 2 slightly differs from that in air (∆T50 close to 25/35 • C), while for Zr materials the T50 in 1% O 2 /N 2 is 70/80 • C, which is higher compared to oxidation in air. In the case of zirconia, the change of oxygen concentration heavily affects T50, as carbon reacts directly with O from the gas phase, while in CZ materials the dependence on the amount of oxygen in the gas phase is less significant, and mediated by the available oxygen from the CZ surface. Therefore, even when the oxygen concentration is very low, ceria-based materials are able to release their surface/bulk oxygen to oxidize the soot with formation of vacancies that are refilled through oxygen of the gas phase, promoting oxidation. It is therefore the oxygen from ceria that sustains the catalytic activity through a mechanism that exploits both the available surface oxygen and the bulk oxygen [47]. for the lattice oxygen of CZ used for oxidation during the reaction. Weight loss in 1% O2/N2 and in air shows very similar profiles, but when a lower amount of oxygen is used, a shift of the oxidation curve profile to higher temperature is observed; in particular, T50 shifts from 515 °C for reaction under air to 545 °C for reaction in 1% O2/N2. Figure 5 shows the T50 for all the investigated catalysts as a function of the reaction atmosphere. Under an inert atmosphere, weight loss is negligible in bare soot and in zirconia/soot mixtures regardless of the contact type; this is consistent with the lack of lattice oxygen available for promotion of redox reaction in zirconia and the lack of volatile matter in soot composition, which can be released during thermal treatment. No oxidation is therefore observed in these samples under an inert atmosphere. With the introduction of ceria in CZ, a higher activity can be observed already under inert atmosphere, with a T50 for the loose contact of 775 °C, which progressively decreases to 546 and 420 °C for tight and supertight contact, respectively. This is consistent with the fact that ceriazirconia-based systems can use the surface/bulk oxygen to oxidize soot even in the absence of oxygen in the gas phase, through the oxidation of carbon at the carbon/CZ interface with the formation of an oxygen vacancy [24,45]. The oxidation reaction proceeds to consume available oxygen, and the activity is closely related to the amount of reducible oxygen/OSC, and is progressively inhibited due to the lack of reoxidation of the material when the inert gas phase is used [46,47]. The introduction of oxygen facilitates oxidation in all formulations. If we focus on CZ, we can observe that the catalytic activity in 1% O2/N2 slightly differs from that in air (ΔT50 close to 25/35 °C), while for Zr materials the T50 in 1% O2/N2 is 70/80 °C, which is higher compared to oxidation in air. In the case of zirconia, the change of oxygen concentration heavily affects T50, as carbon reacts directly with O from the gas phase, while in CZ materials the dependence on the amount of oxygen in the gas phase is less significant, and mediated by the available oxygen from the CZ surface. Therefore, even when the oxygen concentration is very low, ceria-based materials are able to release their surface/bulk oxygen to oxidize the soot with formation of vacancies that are refilled through oxygen of the gas phase, promoting oxidation. It is therefore the oxygen from ceria that sustains the catalytic activity through a mechanism that exploits both the available surface oxygen and the bulk oxygen [47]. The catalytic activity for the combustion of soot was also determined from peak-top temperature (Tm) during temperature programmed oxidation (TPO) of catalyst/soot mixtures under 10% O2/N2 or 1% O2/N2, mainly to verify gas composition after combustion and evaluate the selectivity to CO2. Figure 6 compares CO and CO2 profiles of all samples.
The results are in agreement with those obtained from TGA, with ceria-zirconia displaying a lower oxidation temperature than zirconia for any contact conditions, and regardless of oxygen content. Selectivity to CO2 is always higher than 97%@1% O2, compared to a value of 60-80% obtained The catalytic activity for the combustion of soot was also determined from peak-top temperature (Tm) during temperature programmed oxidation (TPO) of catalyst/soot mixtures under Catalysts 2020, 10, 768 8 of 13 10% O 2 /N 2 or 1% O 2 /N 2 , mainly to verify gas composition after combustion and evaluate the selectivity to CO 2 . Figure 6 compares CO and CO 2 profiles of all samples.
Catalysts 2020, 10, x FOR PEER REVIEW 8 of 13 oxygen is comparable to selectivity in bare soot. Selectivity in ceria-zirconia systems is only negligibly affected by the different oxygen concentration and contact conditions, while in bare soot and Zr-based samples, the CO/CO2 ratio is strongly influenced by atmosphere and carbon catalyst contact. In particular, zirconia at low oxygen concentration in loose conditions has negligible catalytic activity and selectivity (Tm = 680 °C, Sel = 63%), which is increased under tight conditions, as well as 10% oxygen, to reach 97% with a Tm of 535 °C.  The specific reaction rate for soot oxidation was also calculated to evaluate in more detail the ability of ceria-based materials to help the transfer of lattice oxygen to carbon in oxygen-poor atmospheres. The values at different temperatures are reported in Figure 7. While no oxidation is observed in uncatalyzed reaction at 500 °C (S@500 in Figure 7), for zirconia, both in tight and loose contact, in the 400-500 °C temperature range, the specific reaction rate is quite low, and does not exceed 300 µgsoot/gsoot*s*m 2 . For CZ, higher reaction rate values are observed regardless of the contact type, and the rates increase as the contact become more intimate (CZST > CZT > CZL). The results are in agreement with those obtained from TGA, with ceria-zirconia displaying a lower oxidation temperature than zirconia for any contact conditions, and regardless of oxygen content. Selectivity to CO 2 is always higher than 97%@1% O 2 , compared to a value of 60-80% obtained with Zr. The data are summarized in Table 2; it can be observed that selectivity to CO 2 in ZrL at 1% oxygen is comparable to selectivity in bare soot. Selectivity in ceria-zirconia systems is only negligibly affected by the different oxygen concentration and contact conditions, while in bare soot and Zr-based samples, the CO/CO 2 ratio is strongly influenced by atmosphere and carbon catalyst contact. In particular, zirconia at low oxygen concentration in loose conditions has negligible catalytic activity and selectivity (Tm = 680 • C, Sel = 63%), which is increased under tight conditions, as well as 10% oxygen, to reach 97% with a Tm of 535 • C. The specific reaction rate for soot oxidation was also calculated to evaluate in more detail the ability of ceria-based materials to help the transfer of lattice oxygen to carbon in oxygen-poor atmospheres. The values at different temperatures are reported in Figure 7. While no oxidation is observed in uncatalyzed reaction at 500 • C (S@500 in Figure 7), for zirconia, both in tight and loose contact, in the 400-500 • C temperature range, the specific reaction rate is quite low, and does not exceed 300 µg soot /g soot *s*m 2 . For CZ, higher reaction rate values are observed regardless of the contact type, and the rates increase as the contact become more intimate (CZST > CZT > CZL).
Catalysts 2020, 10, x FOR PEER REVIEW 9 of 13 In an oxygen-poor atmosphere, ceria-zirconia has a specific reaction rate from four (CZL) to seven (CZT) times higher than the corresponding zirconia-based systems, confirming the importance of the redox behavior for oxidation under low oxygen pressure. We have previously reported that soot oxidation over ceria-based catalysts is driven by two different coexisting phenomena, one influenced by the amount of surface active oxygens that predominates in oxygen rich atmosphere, and another related to the OSC (bulk oxygen) of the material that prevails when gas-phase oxygen is absent [47]. In particular, (i) oxygen withdrawn from ceria-zirconia oxidizes carbon at the soot catalyst interface; and (ii) the resulting vacancy or other surface defects can be the center for the activation of gas phase oxygen and the formation of active oxygen species that contributes to oxidize soot on a parallel route. In the absence of gas phase oxygen, route (i) is the only available way to oxidize carbon, while in the presence of oxygen, the two mechanisms may coexist in proportions that depend on oxygen partial pressure and the degree of carbon catalyst contact. The concurrence of the two mechanisms could therefore be important in GPF applications, where the oxygen concentration is lower than in DPF.
Overall, the results obtained in this study highlight the characteristics of materials for use in the catalytic oxidation of particulate matter under oxygen-poor conditions such as those found in GDI engines. It is shown that the activation and transfer of lattice/surface oxygen are fundamental in order to develop catalytic materials suitable for GPF. Ceria-zirconia compositions overtake ZrO2 in both activity and selectivity under these conditions; several known strategies could then be accomplished to increase the transfer of oxygen in these materials, like the engineering of the shape and size of ceria particles (nanocubes, nanorods, etc.) or tailoring the composition of the materials, by lattice/surface doping with rare earth, noble or transition metals elements.

Catalyst Preparation
The ceria-zirconia (Ce0.8Zr0.2O2, CZ80) sample was prepared by the co-precipitation of an acidic solution of cerium and zirconium nitrate (Treibacher Industrie AG, Althofen, Austria) with NH4OH in the presence of H2O2. The precipitate was dried overnight at 100 °C, and calcined in air at 500 °C for 3 h. Zirconia was synthesized by the calcination of zirconium hydroxides (Mel Chemicals, Manchester, UK) at 500 °C for 3 h. Conventional catalyst/soot mixtures were obtained in loose (L) and tight (T) contact modes by mixing the appropriate amount of catalyst (CZ80 or ZrO2) with soot (Printex U by Degussa, Esse, Germany, surface area of 100 m 2 /g), respectively, in a vial for 2 min or in an agate mortar for 10 min. The catalyst/soot weight ratio of 20:1 was adopted in this study. For CZ80, improved contact (supertight, ST) was achieved in a high-energy Spex mill equipped with a In an oxygen-poor atmosphere, ceria-zirconia has a specific reaction rate from four (CZL) to seven (CZT) times higher than the corresponding zirconia-based systems, confirming the importance of the redox behavior for oxidation under low oxygen pressure.
We have previously reported that soot oxidation over ceria-based catalysts is driven by two different coexisting phenomena, one influenced by the amount of surface active oxygens that predominates in oxygen rich atmosphere, and another related to the OSC (bulk oxygen) of the material that prevails when gas-phase oxygen is absent [47]. In particular, (i) oxygen withdrawn from ceria-zirconia oxidizes carbon at the soot catalyst interface; and (ii) the resulting vacancy or other surface defects can be the center for the activation of gas phase oxygen and the formation of active oxygen species that contributes to oxidize soot on a parallel route. In the absence of gas phase oxygen, route (i) is the only available way to oxidize carbon, while in the presence of oxygen, the two mechanisms may coexist in proportions that depend on oxygen partial pressure and the degree of carbon catalyst contact. The concurrence of the two mechanisms could therefore be important in GPF applications, where the oxygen concentration is lower than in DPF.
Overall, the results obtained in this study highlight the characteristics of materials for use in the catalytic oxidation of particulate matter under oxygen-poor conditions such as those found in GDI engines. It is shown that the activation and transfer of lattice/surface oxygen are fundamental in order to develop catalytic materials suitable for GPF. Ceria-zirconia compositions overtake ZrO 2 in both activity and selectivity under these conditions; several known strategies could then be accomplished to increase the transfer of oxygen in these materials, like the engineering of the shape and size of ceria particles (nanocubes, nanorods, etc.) or tailoring the composition of the materials, by lattice/surface doping with rare earth, noble or transition metals elements.

Catalyst Preparation
The ceria-zirconia (Ce 0.8 Zr 0.2 O 2 , CZ80) sample was prepared by the co-precipitation of an acidic solution of cerium and zirconium nitrate (Treibacher Industrie AG, Althofen, Austria) with NH 4 OH in the presence of H 2 O 2 . The precipitate was dried overnight at 100 • C, and calcined in air at 500 • C for 3 h. Zirconia was synthesized by the calcination of zirconium hydroxides (Mel Chemicals, Manchester, UK) at 500 • C for 3 h. Conventional catalyst/soot mixtures were obtained in loose (L) and tight (T) contact modes by mixing the appropriate amount of catalyst (CZ80 or ZrO 2 ) with soot (Printex U by Degussa, Esse, Germany, surface area of 100 m 2 /g), respectively, in a vial for 2 min or in an agate mortar for 10 min. The catalyst/soot weight ratio of 20:1 was adopted in this study. For CZ80, improved contact (supertight, ST) was achieved in a high-energy Spex mill equipped with a zirconia jar and balls [40]. A reference sample, to obtain the non-catalytic combustion profile, was prepared diluting soot with powder quartz (Carlo Erba Reagents S.r.l., Milan, Italy) in a weight ratio of 1:20.

Catalyst Characterization
Surface area of the materials has been measured by means of a Tristar 3000 gas adsorption analyzer (Micromeritics, Norcross, GA, USA), according to the B.E.T. method by nitrogen adsorption at 77 K. Structural features of the catalysts were investigated by XRD. Diffractograms were recorded on a Philips X'Pert diffractometer (Ni-filtered Cu-Kα radiation) in the range 20 • -145 • using a step size of 0.02 • and a counting time of 40 s per angular abscissa (PANalytical B.V., Almelo, the Netherlands). Phase identification has been performed by means of Philips X'Pert HighScore software (Version 1.0b, PANalytical B.V., Almelo, the Netherlands).
In TPR experiments, the catalysts (50 mg) were heated at a constant rate (10 • C/min) in a U-shaped quartz reactor from room temperature to 900 • C under a flowing hydrogen/nitrogen mixture (35 mL/min, 4.5% H 2 in N 2 ). The hydrogen consumption was monitored using a thermal conductivity detector (TCD). For ZrO 2 and Ce 0.8 Zr 0.2 O 2 , pretreatment under air at 500 • C for 1 h was carried out. The outlet gas composition was also followed by an online quadrupole mass-spectrometer (Omnistar, Balzers Instruments, Balzers, Liechtenstein). In addition, H 2 -TPR experiments under 1.8% H 2 in N 2 (total flow 500 mL/min) in a tubular quartz reactor followed by FTIR gas analyzers (MultiGas 2030, MKS Instrumentrs, Inc., Andover, MA, USA) have been performed.
The OSC of samples was investigated by carrying out TGA experiments (Q500, TA Instruments, New Castle, DE) in 4.5% H 2 in N 2 mixture flow (100 mL/min). Each sample was treated in N 2 atmosphere for 1 h at 150 • C, followed by heating at a constant rate (10 • C/min) up to a fixed temperature (300, 400 and 500 • C), and finally the N 2 /H 2 mixture was introduced while keeping that temperature for 60 min. The observed weight loss is due to oxygen removal by H 2 to form water, and it can be associated with total oxygen storage capacity at that temperature.

Catalytic Activity
The soot oxidation activity was measured by running thermogravimetric analysis from r.t. to 800 • C (heating rate 10 • C/min) under a different atmosphere (N 2 , 1% O 2 /N 2 and air) [48,49]. Before the catalytic tests, the samples (soot + catalyst, ca. 20 mg) were subjected to a 1 h pre-treatment at 150 • C under inert atmosphere (60 mL/min), in order to eliminate the adsorbed water. The activity was measured by means of T50, the temperature at which 50% of the weight loss is observed. Catalytic activity measurements have also been carried out by TPO experiments heating at a constant rate (10 • C/min) 20 mg of sample under O 2 gas flow (1% or 10% O 2 (v/v) balance N 2 ; total flow 0.5 L/min). The catalyst temperature was measured by a chromel-alumel thermocouple, located on the catalyst bed. The outlet gas composition (i.e., CO, and CO 2 ) was measured by FTIR gas analyzers (MultiGas 2030, MKS Instrumentrs, Inc., Andover, MA, USA). As a matter of fact, reproducibility of results was verified by running several TGA/TPO experiments on similar samples, and the results, in terms of T50/Tp, were always within ± 3 • C.
Reaction rate measurements were performed by TGA isothermal experiments after pretreatment of ca. 20 mg of a catalyst-soot mixture under nitrogen atmosphere for 1 h at 150 • C. Then the temperature has been increased at a constant rate (10 • C/min) up to the reaction temperature (300, 400 or 500 • C), followed by switching to air. The reaction was followed for 1 h. A specific reaction rate has been determined, according to Van Setten et al. [50], and normalized to the soot initially present in the reactor and to the catalyst surface area (µg soot /(g sootinitial ·s·m 2 ) [51]. The reaction rate was calculated at 3% of conversion. A reaction rate in an inert atmosphere has also been performed.

Conclusions
Ceria-based materials has been widely studied as a catalyst in diesel particulate filters, but their role and applicability in gasoline particulate filters, under oxygen-poor conditions, is still at the initial stage, and this preliminary work investigates their potential in soot oxidation at low oxygen partial pressure. The comparison of the behavior of ceria-zirconia and zirconia in soot oxidation shows that the former is less influenced by the variation of the oxygen atmosphere, and is capable of carbon oxidation at a lower temperature with almost full selectivity to CO 2 ; in contrast, activity in ZrO 2 is highly influenced by the oxygen atmosphere, and lower CO 2 selectivities are obtained. Lattice and surface oxygen in ceria-zirconia can help with oxidation at low oxygen partial pressure. The mechanism works through the interface between carbon and catalysts, and is therefore highly dependent on the way carbon and catalysts are put into contact. The presence of soot over ceria-zirconia particles acts as a driving force for the reduction of the material enhancing the oxygen removal rate, and its use in oxidation reaction in the substitution of gas phase oxygen. Funding: This research was funded in part by Interreg V Italy-Austria project COAT4CATA project number ITAT1019.

Conflicts of Interest:
The authors declare no conflict of interest.