Nanostructured Equimolar Ceria-Praseodymia for Total Oxidations in Low-O 2 Conditions

: A Gasoline Particulate Filter (GPF) can be an e ﬀ ective solution to abate the particulate matter produced in modern direct injection gasoline engines. The regeneration of this system is critical, since it occurs in oxygen deﬁciency, but it can be promoted by placing an appropriate catalyst on the ﬁlter walls. In this paper, a nanostructured equimolar ceria-praseodymia catalyst, obtained via hydrothermal synthesis, was characterized with complementary techniques (XRD, N 2 -physisorption, FESEM, XPS, Temperature Programmed Reduction, etc.) and its catalytic performances were investigated in low oxygen availability. Pr-doping signiﬁcantly a ﬀ ected ceria structure and morphology, and the weakening of the cerium–oxygen bond associated to Pr insertion resulted in a high reducibility. The catalytic activity was explored considering di ﬀ erent reactions, namely CO oxidation, ethylene and propylene total oxidation, and soot combustion. Thanks to its capability of releasing active oxygen species, ceria-praseodymia exhibited a remarkable activity and CO 2 -selectivity at low oxygen concentrations, proving to be a promising catalyst for coated GPFs.


Introduction
In the recent years, several strategies have been implemented by automotive manufacturers in order to comply with the CO 2 emission targets imposed by many nations (e.g., 95 g/km in the European Union by 2021) [1]. Diesel engines provide a good efficiency, but they require a complex after-treatment apparatus [2] and they have encountered a decline in popularity in recent years; hence, many efforts have been devoted to improving the overall fuel economy of gasoline engines. Among the different solutions developed, the Gasoline Direct Injection (GDI) technology presents several advantages with respect to Port Fuel Injection (PFI), since a GDI engine can work with higher efficiency and lower fuel consumption [3]. However, the reduced degree of mixing in the combustion chamber entails the production of a greater range and quantity of noxious combustion by-products in the fuel-rich regions of the flame, e.g., unburned hydrocarbons and carbonaceous particulate matter (soot) [3][4][5][6]. Since the actual regulations strictly limit not only the mass but also the number of solid particles that can be emitted, GDI vehicles have been equipped with a Gasoline Particulate Filter (GPF) in order to keep the emissions within the limits [1,5,7]. The structure and operating principle of a GPF are similar to that of a Diesel Particulate Filter (DPF), but the regeneration phase is more critical due to the lack of

Catalyst Characterization
The synthesized material was investigated using different complementary techniques, in order to unveil its physico-chemical and textural properties. Figure 1 displays the powder XRD diffractogram of the Ce50Pr50 catalyst, compared with that of pure ceria prepared using the same synthesis procedure. The mixed oxide exhibits the typical pattern of the fluorite crystalline structure of ceria, marked by the presence of the (111), (200), (220), (311) and (222) planes [12,13,[25][26][27], as can be seen in Figure 1. This similarity indicates that Pr ions are well incorporated into the ceria lattice, which is confirmed by the absence of additional peaks, attributable to segregated oxides. However, the peaks in the XRD pattern of the Ce50Pr50 catalyst are slightly shifted towards lower angles with respect to those of pure ceria, and this displacement can be the result of lattice expansion. Actually, cerium and praseodymium cations have very similar ionic radii, but the latter element is more reducible; therefore, the observed shift is probably induced by the presence of Pr 3+ , whose ionic radius is bigger than that of Ce 4+ [28]. growth during the synthesis, inducing an increase in the relative abundancy of the (110) planes. Furthermore, praseodymium addition was found to also influence the final dimension of the particles. Indeed, the average crystallite size was estimated using Scherrer's equation, and the value obtained for the Ce50Pr50 catalyst (42 nm) was significantly lower than that observed for pure ceria nanoparticles synthesized with the same procedure (139 nm). The specific surface area of the Ce50Pr50 sample was calculated via nitrogen physisorption, using the BET method. A value of 13 m 2 /g was found, slightly higher than that of pure CeO2 (7 m 2 /g), in agreement with the smaller size of the ceria-praseodymia crystallites pointed out by the XRD results.   The (200)/(111) and (220)/(111) peak intensity ratios were calculated, and the values obtained were 0.32 and 0.63, respectively. Both these parameters are higher than those observed on pure ceria samples prepared in analogous conditions (being 0.30 and 0.51, respectively [12]), and especially the (220)/(111) ratio shows a remarkable increase. This result suggests that Pr addition can affect crystal growth during the synthesis, inducing an increase in the relative abundancy of the (110) planes. Furthermore, praseodymium addition was found to also influence the final dimension of the particles. Indeed, the average crystallite size was estimated using Scherrer's equation, and the value obtained for the Ce50Pr50 catalyst (42 nm) was significantly lower than that observed for pure ceria nanoparticles synthesized with the same procedure (139 nm).
The specific surface area of the Ce50Pr50 sample was calculated via nitrogen physisorption, using the BET method. A value of 13 m 2 /g was found, slightly higher than that of pure CeO 2 (7 m 2 /g), in agreement with the smaller size of the ceria-praseodymia crystallites pointed out by the XRD results. Figure 2 reports some FESEM micrographs of the Ce50Pr50 catalyst, in which two different types of nanostructures can be noticed. Both nanopolyhedra and nanorods are present in the sample, instead of the nanocubes previously observed in pure ceria prepared using analogous synthesis conditions [12]. This confirms that Pr addition directly affects the ripening process, promoting the growth of (110) crystal planes and thus favoring edge truncation and elongation of the particles. Energy Dispersive X-ray analysis (EDX) was performed focusing the spot on different points of the sample. The average composition of the mixed oxide resulted to be very close to the nominal one, with 49.4% of cerium and 50.6% of praseodymium. Furthermore, EDX mapping, whose results are reported in Figure 3, showed that the two elements are evenly distributed in the space, confirming the good homogeneity of the Ce50Pr50 sample.
The oxygen species on the surface of the sample were investigated analysing the O 1s XP spectrum shown in Figure 4, in which two distinct peaks can be observed. The intense peak centered at 528.1 eV can be ascribed to lattice oxygen (the so-called "O β ") [29,30] and it is slightly shifted towards lower binding energy as a consequence of Pr addition [25]. The peak at higher binding energy (530.9 eV) is instead assigned to the various "O α " capping oxygen species chemisorbed on the surface, e.g., hydroxyl (OH − ), carbonate (CO 3 2− ), peroxide (O 2 2− ) or superoxide (O 2− ) ions [31,32]. The relative abundancy of the O α species, calculated from the area of the deconvoluted peaks and reported in Table 1, proved to be significant. A high quantity of chemisorbed oxygen can foster oxygen mobility and spillover at the catalyst surface, with beneficial effects during the oxidation reactions [13]; in particular, reactive oxygen species play a key role in the soot oxidation, in which a direct interaction between the two solid phases is limited by the few contact points. The specific surface area of the Ce50Pr50 sample was calculated via nitrogen physisorption, using the BET method. A value of 13 m 2 /g was found, slightly higher than that of pure CeO2 (7 m 2 /g), in agreement with the smaller size of the ceria-praseodymia crystallites pointed out by the XRD results.    conditions [12]. This confirms that Pr addition directly affects the ripening process, promoting the growth of (110) crystal planes and thus favoring edge truncation and elongation of the particles. Energy Dispersive X-ray analysis (EDX) was performed focusing the spot on different points of the sample. The average composition of the mixed oxide resulted to be very close to the nominal one, with 49.4% of cerium and 50.6% of praseodymium. Furthermore, EDX mapping, whose results are reported in Figure 3, showed that the two elements are evenly distributed in the space, confirming the good homogeneity of the Ce50Pr50 sample. The oxygen species on the surface of the sample were investigated analysing the O 1s XP spectrum shown in Figure 4, in which two distinct peaks can be observed. The intense peak centered at 528.1 eV can be ascribed to lattice oxygen (the so-called "Oβ") [29,30] and it is slightly shifted towards lower binding energy as a consequence of Pr addition [25]. The peak at higher binding energy (530.9 eV) is instead assigned to the various "Oα" capping oxygen species chemisorbed on the surface, e.g., hydroxyl (OH − ), carbonate (CO3 2− ), peroxide (O2 2− ) or superoxide (O 2− ) ions [31,32]. The relative abundancy of the Oα species, calculated from the area of the deconvoluted peaks and reported in Table 1, proved to be significant. A high quantity of chemisorbed oxygen can foster oxygen mobility and spillover at the catalyst surface, with beneficial effects during the oxidation reactions [13]; in particular, reactive oxygen species play a key role in the soot oxidation, in which a direct interaction between the two solid phases is limited by the few contact points.  conditions [12]. This confirms that Pr addition directly affects the ripening process, promoting the growth of (110) crystal planes and thus favoring edge truncation and elongation of the particles. Energy Dispersive X-ray analysis (EDX) was performed focusing the spot on different points of the sample. The average composition of the mixed oxide resulted to be very close to the nominal one, with 49.4% of cerium and 50.6% of praseodymium. Furthermore, EDX mapping, whose results are reported in Figure 3, showed that the two elements are evenly distributed in the space, confirming the good homogeneity of the Ce50Pr50 sample. The oxygen species on the surface of the sample were investigated analysing the O 1s XP spectrum shown in Figure 4, in which two distinct peaks can be observed. The intense peak centered at 528.1 eV can be ascribed to lattice oxygen (the so-called "Oβ") [29,30] and it is slightly shifted towards lower binding energy as a consequence of Pr addition [25]. The peak at higher binding energy (530.9 eV) is instead assigned to the various "Oα" capping oxygen species chemisorbed on the surface, e.g., hydroxyl (OH − ), carbonate (CO3 2− ), peroxide (O2 2− ) or superoxide (O 2− ) ions [31,32]. The relative abundancy of the Oα species, calculated from the area of the deconvoluted peaks and reported in Table 1, proved to be significant. A high quantity of chemisorbed oxygen can foster oxygen mobility and spillover at the catalyst surface, with beneficial effects during the oxidation reactions [13]; in particular, reactive oxygen species play a key role in the soot oxidation, in which a direct interaction between the two solid phases is limited by the few contact points.    Figure 5a depicts the Ce 3d core level region of the XP spectrum. Five "u" and five "v" peaks were identified, related to 3d 5/2 and 3d 3/2 states, respectively. The v • , v', u • and u' peaks were ascribed to Ce 3+ ions, while the other six peaks were assigned to the 4+ oxidation state [33,34]. A similar deconvolution was performed on the Pr 3d XP spectrum, in the 968-923 eV region (Figure 5b), ascribing four peaks (herein denoted as a • , a', b • and b') to Pr 3+ ions and other six peaks to the 4+ oxidation state [35]. The relative abundancy of the Ce and Pr cations was estimated after the deconvolution, and the calculated values are resumed in Table 1. Although Pr addition was proved to increase the Ce 3+ quantity [25,36], cerium maintains 4+ as primary valence. Instead, most of the Pr cations are in the lower oxidation states, confirming the higher reducibility of this element and suggesting a strong presence of oxygen vacancies at the catalyst surface.
Catalysts 2020, 10, x FOR PEER REVIEW 5 of 16 Figure 5a depicts the Ce 3d core level region of the XP spectrum. Five "u" and five "v" peaks were identified, related to 3d5/2 and 3d3/2 states, respectively. The v°, v', u° and u' peaks were ascribed to Ce 3+ ions, while the other six peaks were assigned to the 4+ oxidation state [33,34]. A similar deconvolution was performed on the Pr 3d XP spectrum, in the 968-923 eV region ( Figure  5b), ascribing four peaks (herein denoted as a°, a', b° and b') to Pr 3+ ions and other six peaks to the 4+ oxidation state [35]. The relative abundancy of the Ce and Pr cations was estimated after the deconvolution, and the calculated values are resumed in Table 1. Although Pr addition was proved to increase the Ce 3+ quantity [25,36], cerium maintains 4+ as primary valence. Instead, most of the Pr cations are in the lower oxidation states, confirming the higher reducibility of this element and suggesting a strong presence of oxygen vacancies at the catalyst surface.  In order to better investigate the reducibility of the catalyst, different types of Temperature Programmed Reduction (TPR) were performed. When H2 is used, the TPR profile of pure ceria is usually characterized by the presence of two peaks, since the reduction occurs in two steps. A lower In order to better investigate the reducibility of the catalyst, different types of Temperature Programmed Reduction (TPR) were performed. When H 2 is used, the TPR profile of pure ceria is Catalysts 2020, 10, 165 6 of 16 usually characterized by the presence of two peaks, since the reduction occurs in two steps. A lower temperature peak centered at 500-600 • C is associated to the surface reduction, while a band located above 700 • C is related to the slow release of bulk β-oxygen at very high temperatures [37,38]. For the Ce50Pr50 sample, a single intense peak was observed during the H 2 -TPR, as reported in Figure 6a, pointing out that Ce and Pr ions are simultaneously reduced from a 4+ to a 3+ valence state. The position of this asymmetric component, centered around 460 • C, suggests that Pr addition significantly increases the ceria reducibility; in fact, the incorporation of Pr ions can weaken the Ce-O bond, thus fostering the oxygen release from both the surface and the bulk.
Catalysts 2020, 10, x FOR PEER REVIEW 6 of 16 temperature peak centered at 500-600 °C is associated to the surface reduction, while a band located above 700 °C is related to the slow release of bulk β-oxygen at very high temperatures [37,38]. For the Ce50Pr50 sample, a single intense peak was observed during the H2-TPR, as reported in Figure  6a, pointing out that Ce and Pr ions are simultaneously reduced from a 4+ to a 3+ valence state. The position of this asymmetric component, centered around 460 °C, suggests that Pr addition significantly increases the ceria reducibility; in fact, the incorporation of Pr ions can weaken the Ce-O bond, thus fostering the oxygen release from both the surface and the bulk. The results of the CO-TPR are reported in Figure 6b. In this case, the CO2 concentration at the outlet was monitored, as this species is the result of the reaction between the CO and the oxygen released from the catalyst lattice. While reduction peaks were barely visible for pure ceria nanocubes [36], here, a peak around 550 °C can be clearly observed for the Ce50Pr50 catalyst, as a result of its higher reducibility and OSC (the operating conditions adopted in this work differ from the ones of the CO-TPR reported in [23], and therefore the peak locates at a different temperature). Furthermore, a soot-TPR analysis was performed, heating a bed of catalyst and soot in tight contact (i.e., mixed in a ball mill) in absence of gas-phase oxygen. The concentration profiles of the CO and CO2 produced when the soot is oxidized by the catalyst are displayed in Figure 6c. Observing the plot, two reaction steps can be identified: at low temperatures (200-600 °C), when the surface oxygen species are still abundant, the catalyst reduction produces mainly CO2 and almost no CO, pointing out the good selectivity of Ce50Pr50; however, at high temperatures (above 600 °C) more CO starts to be generated, since most of the available oxygen of the catalyst has already been consumed.

Catalytic Activity
The catalytic activity of nanostructured Ce50Pr50 was investigated by performing different oxidation reactions over the catalyst powder. The CO conversion, soot combustion and VOCs' total oxidation were tested in low oxygen concentration (1%), in order to simulate the conditions of a Gasoline Particulate Filter.
The oxidizing capability of the Ce50Pr50 catalyst was firstly evaluated through CO oxidation tests and compared to that of pure ceria. The results obtained, in terms of CO converted to CO2 as a function of the temperature, are reported in Figure 7. In the absence of a catalyst, the thermal conversion of CO starts to be significant only above 500 °C, and it is still lower than 10% at 600 °C. However, this reaction can be activated at much lower temperatures by pure CeO2, and the Ce50Pr50-mixed oxide shows even better catalytic performances. For a quantitative evaluation, the T50% can be considered, namely the temperature at which 50% of the incoming CO is oxidized. The Ce50Pr50 sample was able to reach such a conversion at only 359 °C in high oxygen excess, exhibiting a specific reaction rate of the CO oxidation of 47.16 μmolCO/h/m 2 at this temperature, while the T50% of pure CeO2 was 413 °C. Nanostructured ceria-praseodymia is therefore more The results of the CO-TPR are reported in Figure 6b. In this case, the CO 2 concentration at the outlet was monitored, as this species is the result of the reaction between the CO and the oxygen released from the catalyst lattice. While reduction peaks were barely visible for pure ceria nanocubes [36], here, a peak around 550 • C can be clearly observed for the Ce50Pr50 catalyst, as a result of its higher reducibility and OSC (the operating conditions adopted in this work differ from the ones of the CO-TPR reported in [23], and therefore the peak locates at a different temperature).
Furthermore, a soot-TPR analysis was performed, heating a bed of catalyst and soot in tight contact (i.e., mixed in a ball mill) in absence of gas-phase oxygen. The concentration profiles of the CO and CO 2 produced when the soot is oxidized by the catalyst are displayed in Figure 6c. Observing the plot, two reaction steps can be identified: at low temperatures (200-600 • C), when the surface oxygen species are still abundant, the catalyst reduction produces mainly CO 2 and almost no CO, pointing out the good selectivity of Ce50Pr50; however, at high temperatures (above 600 • C) more CO starts to be generated, since most of the available oxygen of the catalyst has already been consumed.

Catalytic Activity
The catalytic activity of nanostructured Ce50Pr50 was investigated by performing different oxidation reactions over the catalyst powder. The CO conversion, soot combustion and VOCs' total oxidation were tested in low oxygen concentration (1%), in order to simulate the conditions of a Gasoline Particulate Filter.
The oxidizing capability of the Ce50Pr50 catalyst was firstly evaluated through CO oxidation tests and compared to that of pure ceria. The results obtained, in terms of CO converted to CO 2 as a function of the temperature, are reported in Figure 7. In the absence of a catalyst, the thermal conversion of CO starts to be significant only above 500 • C, and it is still lower than 10% at 600 • C. However, this reaction can be activated at much lower temperatures by pure CeO 2 , and the Ce50Pr50-mixed oxide shows even better catalytic performances. For a quantitative evaluation, the T 50% can be considered, namely the temperature at which 50% of the incoming CO is oxidized. The Ce50Pr50 sample was able to reach such a conversion at only 359 • C in high oxygen excess, exhibiting a specific reaction rate of the CO oxidation of 47.16 µmol CO /h/m 2 at this temperature, while the T 50% of pure CeO 2 was 413 • C. Nanostructured ceria-praseodymia is therefore more efficient towards CO abatement with respect to pure ceria [18,21,25] and to Pr-doped ceria synthesized with different techniques [27,39].
Catalysts 2020, 10, x FOR PEER REVIEW 7 of 16 efficient towards CO abatement with respect to pure ceria [18,21,25] and to Pr-doped ceria synthesized with different techniques [27,39]. The great activity of the Ce50Pr50 catalyst seems to be preserved even when the oxygen availability in the gas phase is limited: indeed, the conversion curve obtained in low-oxygen conditions is quite similar to that measured by feeding the reactor with 10% of oxygen, and the T50% is just 20 °C higher (see Table 2). This interesting performance can be ascribed to the redox properties and high reducibility previously discussed. In fact, since CO oxidation occurs via a MvK-type mechanism over ceria-based catalysts [11], the ability to easily release surface oxygen species accessible to CO molecules plays a key role in defining the catalytic activity. Moreover, also the decrease in the particle size induced by Pr addition has a beneficial effect on the reaction, since the quantity of active sites available to the gaseous reagents is directly related to the specific surface area. Pure ceria also exhibited a limited shift in the CO conversion curve towards higher temperatures at low-oxygen conditions, presenting a T50% of 437 °C. In order to investigate the oxidizing activity of pure and Pr-doped ceria towards VOCs, ethylene and propylene were chosen as probe molecules. In fact, these two species are among the unburned hydrocarbons which can be found in the exhaust gas emitted by gasoline cars, since they can partially slip away from the three-way catalyst at certain operative conditions [41]. The The great activity of the Ce50Pr50 catalyst seems to be preserved even when the oxygen availability in the gas phase is limited: indeed, the conversion curve obtained in low-oxygen conditions is quite similar to that measured by feeding the reactor with 10% of oxygen, and the T 50% is just 20 • C higher (see Table 2). This interesting performance can be ascribed to the redox properties and high reducibility previously discussed. In fact, since CO oxidation occurs via a MvK-type mechanism over ceria-based catalysts [11], the ability to easily release surface oxygen species accessible to CO molecules plays a key role in defining the catalytic activity. Moreover, also the decrease in the particle size induced by Pr addition has a beneficial effect on the reaction, since the quantity of active sites available to the gaseous reagents is directly related to the specific surface area. Pure ceria also exhibited a limited shift in the CO conversion curve towards higher temperatures at low-oxygen conditions, presenting a T 50% of 437 • C. Table 2. Comparison of the parameters obtained from the different catalytic tests performed in low-oxygen conditions (1% O 2 ) over the Ce50Pr50 catalyst and pure CeO 2 . The temperatures at which 10%, 50% and 90% of conversion were reached (T 10% , T 50% and T 90% , respectively) are displayed.

Reaction
Catalyst In order to investigate the oxidizing activity of pure and Pr-doped ceria towards VOCs, ethylene and propylene were chosen as probe molecules. In fact, these two species are among the unburned hydrocarbons which can be found in the exhaust gas emitted by gasoline cars, since they can partially slip away from the three-way catalyst at certain operative conditions [41]. The abatement of these compounds is of high importance due to the fact that some of them are toxic for human health [42][43][44] and they can also foster the photochemical generation of secondary pollutants, such as ozone [45]. Since ceria-based oxides can be used as catalysts [46][47][48] or noble metals supports [29,49,50] for VOCs' oxidation, CeO 2 and Ce50Pr50 are expected to affect these kinds of reactions.
The catalytic tests were carried out at two different concentrations of oxygen, in order to evaluate the effect of this parameter. The conversion curves resulting from the ethylene and propylene oxidation tests are reported in Figure 8. Both CeO 2 and Ce50Pr50 were able to catalyze the two oxidation reactions, since the temperatures required for the conversion were considerably lower in the presence of these catalysts with respect to the uncatalyzed tests. However, the mixed oxide showed a remarkably higher activity, as evidenced by the catalytic parameters resumed in Table 2.
Catalysts 2020, 10, x FOR PEER REVIEW 8 of 16 abatement of these compounds is of high importance due to the fact that some of them are toxic for human health [42][43][44] and they can also foster the photochemical generation of secondary pollutants, such as ozone [45]. Since ceria-based oxides can be used as catalysts [46][47][48] or noble metals supports [29,49,50] for VOCs' oxidation, CeO2 and Ce50Pr50 are expected to affect these kinds of reactions. The catalytic tests were carried out at two different concentrations of oxygen, in order to evaluate the effect of this parameter. The conversion curves resulting from the ethylene and propylene oxidation tests are reported in Figure 8. Both CeO2 and Ce50Pr50 were able to catalyze the two oxidation reactions, since the temperatures required for the conversion were considerably lower in the presence of these catalysts with respect to the uncatalyzed tests. However, the mixed oxide showed a remarkably higher activity, as evidenced by the catalytic parameters resumed in Table 2.
In the case of ethylene, pure ceria was more sensitive to the O2 quantity, showing a lower conversion in oxygen shortage, as displayed in Figure 8a. Instead, the oxygen concentration induced only minimal changes on the conversion curve of Ce50Pr50: in fact, the light-off temperature was slightly lower when oxygen was in high excess, as could be expected, but then 50% of conversion was reached at about 215 °C regardless of the O2 abundancy. At high temperatures, competitive adsorption between ethylene and oxygen might possibly explain the higher catalytic activity observed at a low oxygen concentration (1%): indeed, the higher conversion of ethylene could be due to its higher degree of adsorption on the catalyst with respect to the conditions at a high oxygen content (10%).
Conversely, oxygen availability seems to remarkably affect Ce50Pr50 performances in the case of propylene oxidation, as can be seen by observing Figure 8b. In detail, when 10% of oxygen was fed to the reactor, the Ce50Pr50 catalyst exhibited considerable activity, with a T50% of 205 °C; under 1% of O2, instead, 50% of conversion was reached at 327 °C, indicating that Ce50Pr50 is less active with a shortage of oxygen, although it is still able to significantly promote the propylene oxidation. On the contrary, the role of oxygen seems to be much less relevant for pure CeO2, whose T50% only slightly increased from 462 °C to 475 °C when reducing the oxygen concentration.
Furthermore, the Ce50Pr50 catalyst was also capable of affecting the selectivity to CO2 during the VOCs' oxidation. The uncatalyzed reaction produces 25% of CO in the case of ethylene and 17% In the case of ethylene, pure ceria was more sensitive to the O 2 quantity, showing a lower conversion in oxygen shortage, as displayed in Figure 8a. Instead, the oxygen concentration induced only minimal changes on the conversion curve of Ce50Pr50: in fact, the light-off temperature was slightly lower when oxygen was in high excess, as could be expected, but then 50% of conversion was reached at about 215 • C regardless of the O 2 abundancy. At high temperatures, competitive adsorption between ethylene and oxygen might possibly explain the higher catalytic activity observed at a low oxygen concentration (1%): indeed, the higher conversion of ethylene could be due to its higher degree of adsorption on the catalyst with respect to the conditions at a high oxygen content (10%).
Conversely, oxygen availability seems to remarkably affect Ce50Pr50 performances in the case of propylene oxidation, as can be seen by observing Figure 8b. In detail, when 10% of oxygen was fed to the reactor, the Ce50Pr50 catalyst exhibited considerable activity, with a T 50% of 205 • C; under 1% of O 2 , instead, 50% of conversion was reached at 327 • C, indicating that Ce50Pr50 is less active with a shortage of oxygen, although it is still able to significantly promote the propylene oxidation. On the contrary, the role of oxygen seems to be much less relevant for pure CeO 2 , whose T 50% only slightly increased from 462 • C to 475 • C when reducing the oxygen concentration.
Furthermore, the Ce50Pr50 catalyst was also capable of affecting the selectivity to CO 2 during the VOCs' oxidation. The uncatalyzed reaction produces 25% of CO in the case of ethylene and 17% in the case of propylene oxidation (average values measured in the 300-500 • C range). When Ce50Pr50 is employed, instead, the CO formation is totally negligible in the case of ethylene (CO 2 selectivity is always almost 100%), and it is marginal also when propylene is used, with a CO 2 selectivity ranging from 97% at 250 • C to values over 99% above 350 • C.
Afterwards, the activity of Ce50Pr50 towards the oxidation of particulate matter was evaluated. This heterogeneous reaction involves two solid phases, and is therefore strongly dependent on the quality of the contact at the interface, i.e., the force with which the two solids are brought together and the compatibility of their surface features [12,[51][52][53][54]. For this reason, the catalytic activity was probed in two different conditions: loose contact (LC), which is more representative of the real situation in a GPF, was achieved by gently mixing the two powders with a spatula, while tight contact (TC), which can provide further information about the intrinsic activity, was reached via ball milling. Synthetic Printex-U carbon black was used as carbonaceous particulate, since it represents a suitable substitute for GDI soot [55], and 1% of O 2 was fed to the reactor.
The results of the soot oxidation tests are reported in Figure 9. As expected, better performances were observed at the higher degree of contact, but in both cases the catalyst was able to significantly reduce the temperature required for the soot conversion. Indeed, while 50% of the soot initially present was oxidized only at 652 • C without a catalyst, the same conversion was reached more than 100 • C earlier with Ce50Pr50: in detail, the T 50% was 510 • C in LC and 465 • C in TC. These values are comparable to those obtained at higher oxygen concentrations [25]. Instead, non-doped ceria exhibited a lower activity in oxygen shortage, since a higher T 50% of 485 • C was obtained when soot oxidation was performed in TC on this catalyst [40] (see Table 2).
Catalysts 2020, 10, x FOR PEER REVIEW 9 of 16 Afterwards, the activity of Ce50Pr50 towards the oxidation of particulate matter was evaluated. This heterogeneous reaction involves two solid phases, and is therefore strongly dependent on the quality of the contact at the interface, i.e., the force with which the two solids are brought together and the compatibility of their surface features [12,[51][52][53][54]. For this reason, the catalytic activity was probed in two different conditions: loose contact (LC), which is more representative of the real situation in a GPF, was achieved by gently mixing the two powders with a spatula, while tight contact (TC), which can provide further information about the intrinsic activity, was reached via ball milling. Synthetic Printex-U carbon black was used as carbonaceous particulate, since it represents a suitable substitute for GDI soot [55], and 1% of O2 was fed to the reactor. The results of the soot oxidation tests are reported in Figure 9. As expected, better performances were observed at the higher degree of contact, but in both cases the catalyst was able to significantly reduce the temperature required for the soot conversion. Indeed, while 50% of the soot initially present was oxidized only at 652 °C without a catalyst, the same conversion was reached more than 100 °C earlier with Ce50Pr50: in detail, the T50% was 510 °C in LC and 465 °C in TC. These values are comparable to those obtained at higher oxygen concentrations [25]. Instead, non-doped ceria exhibited a lower activity in oxygen shortage, since a higher T50% of 485 °C was obtained when soot oxidation was performed in TC on this catalyst [40] (see Table 2). The Ce50Pr50 catalyst seems therefore able to interact effectively with the gas-phase O2, even when the concentration of the latter is as low as 1%. As a result, this material was capable of preserving the selectivity towards CO2 despite the low oxygen availability: indeed, when the soot combustion was performed in the absence of a catalyst, the carbon was converted mainly to CO, as the CO to CO2 reaction step was hindered by the lack of gas-phase oxygen; instead, in the presence of Ce50Pr50, the CO2 selectivity exceeded 96.5% in TC and reached 98% in LC.
Finally, kinetic studies were carried out in order to further analyze how the oxygen concentration affects the soot oxidation reaction when it is promoted by Ce50Pr50. The oxygen partial pressure in the gaseous mixture can be expected to affect the reaction rate, and will therefore feature in the rate equation with a certain reaction order. Kastrinaki et al. [56] described the isothermal oxidation of soot on ceria-based catalysts using a multi-population, first-order model: different oxidation rate constants were employed for different soot populations, which were Figure 9. Soot conversion (a), CO 2 (b) and CO (c) concentration at the outlet as a function of the temperature during the soot oxidation tests performed over the Ce50Pr50 catalyst in low-O 2 conditions. The Ce50Pr50 catalyst seems therefore able to interact effectively with the gas-phase O 2 , even when the concentration of the latter is as low as 1%. As a result, this material was capable of preserving the selectivity towards CO 2 despite the low oxygen availability: indeed, when the soot combustion was performed in the absence of a catalyst, the carbon was converted mainly to CO, as the CO to CO 2 reaction step was hindered by the lack of gas-phase oxygen; instead, in the presence of Ce50Pr50, the CO 2 selectivity exceeded 96.5% in TC and reached 98% in LC.
Finally, kinetic studies were carried out in order to further analyze how the oxygen concentration affects the soot oxidation reaction when it is promoted by Ce50Pr50. The oxygen partial pressure in the gaseous mixture can be expected to affect the reaction rate, and will therefore feature in the rate equation with a certain reaction order. Kastrinaki et al. [56] described the isothermal oxidation of soot on ceria-based catalysts using a multi-population, first-order model: different oxidation rate constants were employed for different soot populations, which were distinguished by their contact conditions with the catalyst. Hence, the macroscopic oxidation process results from several parallel microscopic processes and can be modelled as follows with solutions of the type where m is the total soot mass, m i is the soot mass of the i-th population, k i is the rate constant for the oxidation in the contact conditions of the i-th population, β i is the reaction order of oxygen for the i-th population and the subscript 0 indicates the initial soot mass. In tight contact, the contact conditions between the soot and the catalyst can be assumed to be homogeneous throughout the sample: the sum could therefore be reduced to a single term, and the macroscopic oxidation process can be effectively described by a monoexponential-type model with a solution of the type The reaction order of oxygen can then be evaluated by carrying out isothermal oxidations at different oxygen partial pressures and by extracting the characteristic reaction times τ from monoexponential fittings of the soot conversion curves. In fact, extracting the logarithm from both members of equation (3) provides the following equation: The obtained τ values can be plotted against the respective values of the oxygen fraction in the gaseous mixture in a double logarithmic plot. A linear regression of the experimental points will provide a line whose slope is the reaction order of oxygen and whose intercept is the rate constant for tight contact conditions. This kind of kinetic evaluation was carried out by performing isothermal soot oxidation at 450 • C at three different O2 concentrations, namely 1%, 5% and 10%; the results of these tests are presented in Figure 10. The soot oxidation curves in Figure 10a point out the detrimental effect of the lack of oxygen, since a reduction in its abundancy results in a lower reaction rate, as could be expected. After fitting these profiles with a monoexponential function in order to obtain the τ values, the plot in Figure 10b was realized, following Equation (6), and a linear regression of the points was performed. The resulting reaction order for oxygen (β) is very close to 1, in agreement with the results of Neeft et al. [57] and Darcy et al. [58] regarding the non-catalytic oxidation of Printex-U soot. Nevertheless, despite this direct dependence of the reaction rate on the oxygen concentration, the Ce50Pr50 catalyst showed promising performances towards soot combustion even in oxygen shortage conditions, thanks to its high intrinsic activity.
obtain the τ values, the plot in Figure 10b was realized, following Equation (6), and a linear regression of the points was performed. The resulting reaction order for oxygen (β) is very close to 1, in agreement with the results of Neeft et al. [57] and Darcy et al. [58] regarding the non-catalytic oxidation of Printex-U soot. Nevertheless, despite this direct dependence of the reaction rate on the oxygen concentration, the Ce50Pr50 catalyst showed promising performances towards soot combustion even in oxygen shortage conditions, thanks to its high intrinsic activity.

Material Synthesis
The Ce50Pr50 catalyst was prepared via hydrothermal synthesis. Equimolar amounts of Ce(NO 3 ) 3 ·6H 2 O and Pr(NO 3 )3·6H 2 O (10 mmol overall, Sigma-Aldrich, Darmstadt, Germany) were dissolved in 10 mL of deionized water. A second solution was prepared by mixing 48 g of NaOH pellets (Sigma-Aldrich) and 70 mL of deionized water. The solution of lanthanides was then slowly added dropwise into the NaOH solution, under continuous stirring, and a light brown slurry formed. Afterwards, the slurry was stirred at room temperature for 1 h and eventually transferred into the Teflon liner of a stainless-steel autoclave, which was then closed and heated at 180 • C for 24 h inside a furnace. The content of the autoclave was centrifuged and the precipitate was carefully washed by consecutive cycles of suspension and centrifugation in deionized water, ethanol and deionized water again. The washed precipitate was then dried at 60 • C overnight. Finally, the dried powder was gently ground in a mortar and calcined in air at 650 • C for 4 h. A pure CeO 2 sample was prepared for comparison using the same procedure, starting from 10 mmol of Ce(NO 3 ) 3 ·6H 2 O.

Catalyst Characterization
Powder X-ray diffractograms were collected in a Philips X'Pert PW3040 diffractometer using Cu Kα radiation (λ = 1.5418 × 10 −10 m), with values of 2θ ranging between 20 • and 70 • , step size of 0.013 • and an acquisition time of 0.2 s/step. The diffraction peaks were assigned referring to the Powder Diffraction Files by International Centre of Diffraction Data (ICDD). The crystallite size was estimated using Scherrer's equation; the instrumental peak broadening was corrected using lanthanum hexaboride (LaB 6 ) as a calibration standard.
Nitrogen physisorption was carried out in a Micromeritics Tristar II 3020, in order to determine the surface area and pore volume of the prepared catalyst. Before the analysis, an outgassing pretreatment was performed by heating the sample at 200 • C for 2 h under nitrogen flow. The specific surface area was calculated with the Brunauer-Emmett-Teller (BET) method.
The morphology of the catalyst was investigated via field emission scanning electron microscopy (FESEM), using a Zeiss Merlin equipped with a Gemini-II column. An Oxford x-act X-ray detector allowed to estimate the sample composition via energy-dispersive X-ray spectroscopy (EDX).
X-ray photoelectron spectroscopy (XPS) was performed in a PHI Versa Probe instrument, using a 187.85 eV band-pass energy, a 45 • take-off angle and an X-ray spot size of 100.0 µm.
The catalyst reducibility was investigated via different types of temperature programmed reduction (TPR) analyses. H 2 -TPR was performed in a Thermo Scientific 1100 TPDRO equipped with a TCD detector. Some powder (50 mg) was pre-treated by heating it up to 550 • C with a 10 • C/min ramp in He flow. During the analysis, the sample was heated from room temperature up to 800 • C with a 10 • C/min ramp, under a 20 mL/min flow of 5% H 2 in He.
The CO-TPR analysis was carried out in a quartz U-tube reactor (4 mm internal diameter) containing a bed of 150 mg of silica and 45 mg of catalyst powder, placed inside a PID-controlled furnace. Before the analysis, the catalytic bed was pre-treated at 150 • C, flowing 100 mL/min of air for 1 h followed by flushing with N 2 for 30 min. The reactor was then heated from room temperature to 700 • C with a rate of 10 • C/min under a 100 mL/min flow of 2000 ppm of CO in N 2 . The concentrations of CO and CO 2 in the outlet were continuously monitored with an ABB Uras 14 non-dispersive infrared (NDIR) analyser.
The soot-TPR was carried out with the same equipment used for CO-TPR. This time, the bed was made of 150 mg of silica, 45 mg of catalyst powder and 5 mg of Printex-U soot, mixed together in a ball mill at 290 rpm for 15 min. Before the analysis, the catalytic bed was pre-treated at 100 • C, flowing 100 mL/min of pure nitrogen for 1 h. The reactor was then heated from room temperature to 800 • C at a rate of 10 • C/min under a 100 mL/min flow of pure N 2 . CO and CO 2 in the outlet were monitored with the NDIR analyser.

Catalytic Activity Tests
The general setup used for the temperature-programmed oxidation (TPO) tests consisted of a PID-controlled furnace, a quartz U-tube reactor with a fixed catalytic bed and a K-type thermocouple with its tip placed as closed as possible to the catalytic bed. A by-pass line was used to let the feed gas concentration stabilize before starting the tests. The CO and CO 2 concentration in the reactor outlet was monitored with an ABB Uras 14 continuous NDIR gas analyser; furthermore, an Emerson XStream X2GP paramagnetic analyser was employed for measuring the concentration of O 2 . In the present work, four different TPOs were performed.

CO Oxidation
In this test, the catalytic bed was prepared with 100 mg of catalyst. The bed was pre-treated at 100 • C for 30 min under a 50 mL/min flow of air. The reactor was then cooled to 50 • C, while a mixture of N 2 containing 1000 ppm of CO and 1% or 10% of O 2 was sent to the by-pass line with a flow rate of 50 mL/min. After the concentrations of CO, CO 2 and O 2 had stabilized, the flow was sent to the reactor. Upon reaching the steady state, the gas composition at the outlet was recorded and the oven temperature was increased. The procedure was then repeated in order to sample a sequence of steady states at different temperatures, until the total CO conversion was reached. A blank test was performed in the same way, except preparing the reactor bed with 100 mg of SiO 2 .

VOCs Oxidation
Two different tests were carried out to investigate the catalytic activity towards ethylene and propylene complete oxidation. The catalytic bed was prepared with 100 mg of catalyst, and it was pre-treated at 100 • C for 30 min under a 50 mL/min flow of air. The reactor was then cooled to 50 • C, while a mixture of N 2 containing 500 ppm of the VOC (ethylene or propylene) and 1% or 10% of O 2 was sent to the bypass line with a flow rate of 50 mL/min. After the concentrations of CO, CO 2 and O 2 had stabilized, the flow was sent to the reactor. Upon reaching the steady state, the gas composition at the outlet was recorded and the oven temperature was increased. The procedure was then repeated in order to sample a sequence of steady states at different temperatures, until the total VOC conversion was reached. Blank tests were performed in the same way, except the reactor bed was prepared with 100 mg of SiO 2 .

Soot Oxidation
The soot oxidation tests were performed on catalytic beds containing 45 mg of catalyst, 150 mg of silica (Sigma-Aldrich) and 5 mg of Printex-U soot (Degussa). Two different contact conditions were investigated: loose contact (LC) conditions were obtained by gently mixing the solid mixture with a spatula for three minutes before inserting it in the reactor, while tight contact (TC) conditions were obtained by mixing the powders in a ball mill at 290 rpm for 15 minutes. The catalytic bed was pre-treated at 100 • C under a 100 mL/min flow of air for 30 min. Afterwards, a mixture of 1% of O 2 in N 2 was sent with a flow rate of 100 mL/min. The furnace temperature was then raised continuously with a 5 • C/min ramp up to 750 • C.
Isothermal oxidation tests were also performed, preparing a tight contact bed as described above. After the pretreatment, the reactor was flushed with pure nitrogen for some minutes and then sealed. The oven temperature was then raised to 450 • C with a rate of 10 • C/min while a flow of N 2 containing 1%, 5% or 10% O 2 was sent to the bypass line with a flow rate of 100 mL/min. When the goal temperature was reached, the flow was redirected to the reactor and the analysis was carried out until the outlet concentrations of CO and CO 2 were again equal to their baseline values.

Conclusions
In this work, a nanostructured equimolar ceria-praseodymia catalyst has been investigated in order to study its behaviour and catalytic activity in low-oxygen conditions. Although Pr ions were well incorporated into the ceria lattice, forming a single phase, the doping caused important structural and morphological changes. Indeed, praseodymium addition affected the crystal growth during the synthesis, resulting in the formation of two different kinds of nanoparticles. Furthermore, Pr weakened the cerium-oxygen bond, allowing to obtain a highly reducible material. Thanks to these properties, the catalyst exhibited a promising activity towards different oxidation reactions, especially in the case of soot combustion. Moreover, its performances only slightly worsened when the oxygen concentration in the inlet gaseous mixture was abated from 10% to 1%. The CO 2 selectivity was very high both in VOCs and soot oxidation, and it was fully preserved in oxygen shortage. These results point out the beneficial effects of coupling the Pr 4+ -Pr 3+ redox cycle to that of cerium: indeed, the thus-obtained material is capable of fruitfully delivering active oxygen species to particulate matter and gaseous pollutants. Therefore, equimolar ceria-praseodymia could be suitably employed in catalyzed GPFs, in which a high catalytic activity is required to comply with the emission abatement targets in low oxygen availability.