Influence of the Preparation Method of Ag-K/CeO 2 ZrO 2 -Al 2 O 3 Catalysts on Their Structure and Activity for the Simultaneous Removal of Soot and NOx

: Ag/CeO 2 -ZrO 2 -Al 2 O 3 , a known catalyst for the simultaneous removal of NOx and soot, was modified by the addition of K, and was prepared using various techniques: wet impregnation, incipient wetness, and chemical vapor impregnation at different temperatures. The effect of the preparation method on catalyst activity was studied. It was found that catalysts prepared via wet impregnation, incipient wetness, and chemical vapor impregnation at 80 °C were able to utilize in situ formed N 2 O at low temperatures, to simultaneously remove NOx and soot. The difference in preparation method affected the catalyst’s ability to produce and use N 2 O as an oxidant for soot. The temperature at which chemical vapor impregnation was performed greatly influenced the catalyst’s ability to oxidize soot. The introduction of K to the Ag/CeO 2 -ZrO 2 -Al 2 O 3 vastly improved the soot oxidation activity, particularly for the catalyst prepared via wet impregnation. However, the incorporation of K had an adverse effect on the reduction of NOx.


Introduction
It is well known that emissions from vehicles are hazardous to both the environment and human health [1][2][3], hence the automotive industry must meet regulations in order to reduce these emissions. The four key emissions that are restricted by the EURO limits are carbon monoxide, unburnt hydrocarbons (HCs), nitrous oxides (NOx), and particulate matter (PM) [4]. The most recent EURO limit (EURO 6) saw a 67% decrease in the permitted NOx concentration emitted from diesel passenger vehicles, as compared to the previous limit [5]. These limits are now being rigorously enforced through on-road emission measurements as well as in laboratory rolling-road tests. Many countries are phasing out gasoline and diesel-powered vehicles in favor of electric ones, however, it will still be many years until there are no longer diesel and gasoline vehicles on the roads.
Gasoline vehicles operate under stoichiometric conditions, allowing the exhaust gas to be treated by a single catalytic converter, however, this is not the case for diesel vehicles [6]. Diesel vehicles have several exhaust aftertreatment components which remove CO, HCs, NOx, and PM (or soot). A single aftertreatment system would increase vehicle performance due to a lower weight and back pressure, in addition to faster warm up compared to the combination of separate systems. Two of the more complex systems to combine are the regeneration of the diesel particulate filter (DPF), used to remove PM, and the selective catalytic reduction (SCR) of NOx. The regeneration of the DPF occurs via the catalytic combustion of fuel at high temperatures, resulting in the complete oxidation of the trapped soot (Eq. 1). However, the SCR of NOx to N2 using NH3 as the reductant (which is formed in situ from aqueous urea), operates under normal exhaust temperatures, typically 100-360 °C for passenger vehicles [7] (Eq. 2 and 3).
Recently, Ag/CeO2 catalysts have been shown to be active for soot combustion [8][9][10][11][12][13][14]. Ceria is a widely used support for soot oxidation catalysts, due to its ability to promote oxygen storage as well as its redox behavior [15][16][17][18][19][20][21]. The incorporation of Ag into the surface of ceria has been shown to greatly enhance soot oxidation, reducing the temperature required for complete soot combustion by >250 °C [12,22]. Hence, it has the potential to remove the need for fuel injection into the exhaust to regenerate the DPF in passenger vehicles. The removal of the fuel injection protocol would improve the fuel economy of vehicles and lower CO2 emissions.
Davies et al. [23] found that an Ag/CeO2-ZrO2-Al2O3 (Ag/CZA) catalyst was active for the simultaneous removal of NOx and soot. It was shown that the Ag catalyst could reduce NOx to N2 whilst oxidizing soot at low temperatures, typical of a diesel exhaust, even though the catalyst was a poor SCR catalyst in the absence of soot. This was a consequence of the production of N2O from the non-selective SCR (Eq. 4) reaction, initiating soot oxidation at low temperatures in the presence of the Ag catalyst (Eq. 5 and 6). At higher temperatures, N2O is no longer formed as the SCR reaction becomes more selective and hence NOx reduction by NH3 is dominant and soot oxidation occurs utilizing O2 and NO2 as oxidants.

Non-Selective Catalytic Reduction
Soot Oxidation by N2O Building on this previous work, we have been investigating the effect of the catalyst preparation method to investigate the performance for the simultaneous removal of NOx and soot. Potassium, a well-established promoter of soot oxidation [24,25], has also been incorporated into the catalyst. There were significant differences in catalytic activity between the various preparation methods, and these are presented and discussed in this paper.

Catalyst activity
Catalytic testing was carried out using a simulated exhaust gas, which deliberately excluded the major combustion products (CO2 and H2O) as these gases could mask small changes in the composition of the gas stream. In particular, this allowed the desorption of CO2 from the catalyst (in the absence of soot) and its formation during the oxidation of soot to be effectively monitored. Figure 1. Variation of reactants and products as a function of temperature for NOx reduction and soot combustion where the simulated exhaust gas consisted of 500 ppm NO, 500 ppm NH3, 8% O2, with a balance of N2. The temperature was increased in 25 °C intervals from 125 to 550 °C. A total of 0.25 g of catalyst was mixed in a 10:1 ratio by mass with soot. The catalysts were prepared by the following methods: (a) Chemical Vapor Impregnation (CVI) at 60 °C, (b) CVI at 70 °C, and (c) CVI at 90 °C.

Chemical Vapor Impregnation (CVI) Catalysts
The temperature at which the CVI preparation method was carried out influenced the activity towards soot oxidation and NOx reduction (see Figure 1). The catalyst prepared at a CVI temperature of 60 °C (CVI60) proved to be a poor catalyst for NOx reduction, with the NO concentration remaining at 500 ppm for most of the experiment until 475 °C, where the concentration increased. This increase in concentration corresponds to the decrease in concentration of NH3, suggesting that the high temperature oxidation of NH3 produces NO. CO2 was first observed at 275 °C, the CO2 trace is very similar to that observed for this catalyst in the absence of soot (see Supplementary Information, Figure S1). This implies that the presence of the CO2 is mainly due to decomposition of residual carbonate (which had survived the calcination treatment) or to the dissociation of surface carbonate or hydrogencarbonate species (which had formed by adsorption of CO2 and H2O from ambient air prior to catalyst testing).
CVI70 does show a reduction in the concentration of NOx and NH3, but this does not occur simultaneously, this is partially due to NH3 storage by the support, probably on Lewis acid sites (oxygen vacancies) on the CZA. The adsorbed NH3 species can subsequently take part in the SCR reaction via a Langmuir-Hinshelwood mechanism [26]. N2O begins to form at 275 °C but is not used Concentartion / ppm ( CO 2 x 10) to oxidize soot, as the same quantity of N2O was formed in the absence of soot (see supplementary information, Figure S1). Similarly, CO2 was observed from 125 °C, but again cannot be attributed to soot oxidation, as CO2 was observed in the absence of soot, indicating the presence of residual or adsorbed carbonate or hydrogencarbonate species. At temperatures above 400 °C, greater amounts of CO2 were observed due to uncatalyzed soot oxidation at the higher temperatures as seen in the reaction without soot (Supplementary Information Figure S1). CVI80 showed more characteristics of the required catalytic activity. Up to 400 °C, NOx reduction occurred, as shown by the decrease in concentration of NO and NH3 (see Figure 2). In the absence of soot, between 250 °C and 400 °C, the N2O trace is almost a mirror image of the inverted NO peak. This is consistent with the non-selective SCR (Eq 4) reaction, which shows that for every mole of NO consumed there is 1 mole of N2O formed. Between 250 and 500 °C, in the presence of soot, N2O was observed with a maximum (approx. 150 ppm) concentration at 375 °C. This maximum is less than that observed in the absence of soot, suggesting that the N2O is being utilized to oxidize soot. This is supported by the other reaction data, as with the decrease in the N2O peak there is a simultaneous increase in CO2. In the presence of soot at 250 °C, a concentration of 70 ppm of CO2 was measured, compared to the case in the absence of soot, which only produced 30 ppm of CO2 at 250 °C. This implies that low temperature oxidation (<300°C) of soot by N2O is possible using this catalyst. However, this does not account for the full decrease in the concentration of N2O, this is due to some of the N2O being stored by soot until higher temperatures [23]. Over this catalyst, a relatively high concentration of CO2 was observed, particularly at high temperatures, again showing the ability of the catalyst to directly catalyze the oxidation of soot.
The CVI90 catalyst demonstrated exceptionally good activity for soot oxidation, especially at high temperatures, with CO2 production onset at 375 °C and peaking at 500 °C with a concentration >9500 ppm. However, the NO concentration did not decrease at all during the reaction; indeed, the concentration increased at temperatures above 400 °C when NH3 oxidation is both thermodynamically and kinetically favored. Hence, CVI90 is not a suitable catalyst for the simultaneous removal of NOx and soot but it is an effective catalyst for soot oxidation by NO2 and O2. The 2 % Ag-20 % K/CZA catalyst made via incipient wetness displayed interesting properties for soot oxidation. The catalyst used low-temperature N2O (225-400 °C) to oxidize soot, as shown by the relative difference in CO2 and N2O concentrations between the reaction with and without the presence of soot ( Figure 3). From Figure 3, it can be determined that, for this catalyst, N2O is formed as a result of NH3 oxidation, even when the NOx-reduction reaction is suppressed by the presence of soot. The catalyst also uses NO2 at high temperatures to oxidize the soot. Despite being an active catalyst for soot oxidation at low temperatures, the concentration of CO2 observed is significantly lower than that from other preparation methods at higher temperatures. In the absence of soot, the catalyst performs the desired NOx-reduction reaction at 125-300 °C, however, at >300 °C, the NO concentration increases as the direct oxidation of NH3 begins to dominate. In the presence of soot, the NOx-reduction reaction does not take place; instead, the data are consistent with the NH3 being converted through partial oxidation (to N2O) and deep oxidation (to NO and NO2).

Wet Impregnation Catalyst
In the absence of soot, the wet impregnation catalyst showed good NOx-reduction performance, with both NH3 and NO reaching 0 ppm ( Figure 4). Between 175 and 300 °C, a peak of N2O was observed in the absence of soot. This peak was not observed when soot was introduced to the system. The absence of N2O is consistent with the facilitating low temperature oxidation of soot. However, in the presence of soot, N2O was observed at higher temperatures (>300 °C), probably because its formation now occurs over a temperature range where soot oxidation by NO2 and O2 can predominate. In the absence of soot, a large spike of CO2 was observed at 325 °C, which corresponds closely to the decomposition temperature for Ce(CO3)2 [27,28], suggesting that the calcination treatments of the support and the fully-formulated catalyst had not converted all the CZA-carbonate to CZA-oxide. This CO2 has a diluting effect on the gas stream when it desorbs, explaining the fluctuations in the NH3, NO, and N2O traces. When soot was introduced into the system, NOx reduction was suppressed, but N2O was still formed (by partial oxidation of NH3) albeit at a higher temperature. Very high concentrations of CO2 were detected in the presence of soot, when a stepwise oxidation of soot was observed, as described by Davies et al. for K-free Ag/CZA [23]. However, the catalyst of Davies et al. had the ability to simultaneously reduce NOx and utilize in-situ generated N2O to oxidize soot, whilst this catalyst is less effective for NOx reduction in the presence of soot. This suggests that K2CO3 is suppressing this reaction. Overall, the performance data shows that, in the absence of soot, this catalyst is active for NOx reduction (mainly through SCR), whilst in the Concentration / ppm Temperature / °C presence of soot it has much lower NOx-reduction activity but is capable of oxidizing the soot over a wide range of temperatures.

Catalyst Characterization
Raman Spectroscopy, X-ray Diffraction (XRD), BET surface area, X-Ray Photoelectron Spectroscopy (XPS), and Scanning Electron Microscopy (SEM) analysis were carried out on each catalyst with the key information shown in Table 1. The catalysts prepared via CVI were black in color and hence were diluted with KBr to allow a Raman spectrum to be obtained. Ceria-based soot oxidation catalysts have been postulated to have high activity due to the reversible Ce 4+ to Ce 3+ transition as well as the formation of surface and bulk oxygen vacancies [29]. It has been shown that, despite the reduction in surface area, with the incorporation of Ag into the ceria surface, an increase in catalytic activity is observed. Raman spectroscopy studies, identifying the relationship between activity and defect concentration, have shown that the increase in soot oxidation is due to an increase in oxygen vacancies on the ceria surface [30]. The preparation method effects the catalytic activity, as the greater the defect concentration in the ceria lattice, the higher the Concentration / ppm catalytic activity. Ag has also been shown to play an important role in activating molecular oxygen at low temperatures [31].
The Raman analysis (see Supplementary Information, Figure S2) showed that all catalysts had a band between 141 and 158 cm −1 , which is attributed to Ag lattice vibrational modes (i.e., phonons) [32,33]. The CZA support displayed a band centered at 451-469 cm −1 , from the F2g Raman mode of ceria. The peaks present between 500 and 600 cm −1 are indicative of the defect-induced mode of ceria's fluorite phase [13]. The F2g peak distortion ratio was calculated for CVI70, CVI80, CVI90, and Incipient Wetness (IW) prepared catalysts using the F2g peak and the distortion peak (D). It was not possible to calculate the distortion ratio for CVI60 and Wet Impregnation (IMP) due to the absence of the ceria distortion peak. The calculated defect ratios (ID/IF2g) are shown in Table 1.The differences in the ratios show that the formation of bulk oxygen vacancies (Vo-b) are influenced by the different preparation techniques in this study, in agreement with previous work [11]. Additionally, the Ag species supported on ceria can lead to oxygen reverse spillover [9,10], creating Vo-s, which are filled by the bulk oxygen of ceria via diffusion [11]. Furthermore, a band was observed at 1052-1064 cm −1 for all catalysts, excluding CVI80, which is due to residual carbonate from the precipitation process using Na2CO3 or following the adsorption of CO2. From the XPS data ( Table 2) it can be seen that Na is present on the surface of all of the catalysts.  Figure 5 shows the powder X-ray diffraction patterns for the catalysts studied. All catalysts displayed typical diffraction peaks indexed to a cubic fluoride ceria phase, which agrees with the Raman data. The shift in CeO2 peaks, compared to pure ceria [34], is due to the incorporation of ZrO2 and Al2O3 into the ceria lattice. The peaks at 2θ = 38° and 44° shows the formation of metallic silver crystallites. From the XRD pattern, Ag is not present in the form of AgO, which agrees with the work of others investigating similar catalysts [8,10,11,35,36]. The three catalysts that have been shown to utilize N2O to oxidize soot (IW, IMP and CVI80) display significantly lower intensities of the main Ag peak, hence suggesting that the Ag is less crystalline in these catalysts. From the XRD patterns, the Ag crystallite size for each catalyst was calculated using the Scherer equation (Table 1). The catalysts, which were both active soot oxidation catalysts and had the ability to utilize in situ generated N2O to oxidize soot at low temperatures (IMP and CVI80), had the smallest Ag crystallite sizes of 320 and 400 Å, respectively. However, IW, which is also an active catalyst for the simultaneous removal of NOx and soot, but overall a poor soot oxidation catalyst, had a much larger Ag crystallite size (>1000 Å). There were small K2CO3 peaks visible for all catalysts, showing that under ambient conditions the K was present as the carbonate.
The BET analysis showed that all the catalysts had relatively low surface areas, in the range 3.8-26.4 m 2 g −1 (Table 1). From the BET analysis, the IMP catalyst has the greatest surface area, however, there does not appear to be a direct correlation between surface area and the catalysts activity towards soot oxidation.  Figure 6 shows SEM images of the six catalysts. It is clear from the images that the preparation method has a significant impact on the morphology of the catalyst. Each catalyst, although in varying degrees, shows needle-like structures distributed over the surface of the catalyst, and, in order to determine the composition of the needle-like structures, the CZA and K/CZA supports were imaged and analyzed. Figure 7 shows SEM images for the CZA (a) and K/CZA (b) supports. The key difference between the two images is that, for K/CZA, needle-like structures are visible, which are not present for CZA. Elemental analysis from EDX showed that no K was present on the CZA support where there are no needles, but for the K/CZA support, K was clearly present as were the needle structures (see Supplementary Information, Figure S3 and S4). This suggests that the needles are rich in K. The preparation method and temperature has an effect on the morphology of the K2CO3 needles.
From the images of the catalysts it can be observed that the phase separation of the CZA support has occurred as shown in Figure 6. This phase separation is observable from the pure CZA support as well as the K/CZA, hence suggesting that the preparation method is not the key factor in the phase separation. Na may also distort the measured surface concentrations of the cerium.  Table 2. As for EDX analysis (see Supplementary Information, Figure S5), XPS reveals the presence of significant amounts of fluorine for the CVI preparations, which was possibly unexpected and is attributed to the use of the C13H13AgF6O2 ((1,5-Cyclooctadiene)(hexafluoroacetylacetonato)silver(I)) precursor used to deposit Ag. The CVI preparation temperature influences the surface F concentration in the following way: as preparation temperature increases, the relative concentration of F decreases, as does Ag. XPS also confirmed the presence of high amounts of carbonate, although the concentration does vary significantly, evidenced by a C(1s) peak at ca. 289.5 eV in all spectra. As the temperature of the CVI preparation increased, there was a decrease in the relative surface concentration of Ag, and this is most likely associated with the Ag dispersion decreasing, which is consistent with the XRD results. It is worth noting that the most active CVI80 catalyst had a low Ag surface concentration and large Ag crystallites, and the other active catalysts prepared by incipient wetness and impregnation showed the same characteristics. Hence, these catalyst features are important to utilize in situ formed N2O at low temperatures to simultaneously remove NOx and soot.
The oxidation state determination of Ag by XPS can be complicated due to negligible binding energy shifts between oxidation states [37]. The use of the modified Auger parameter allows for a more robust chemical assignment. Here, for the CVI60, CVI70, and CVI80 preparations, the calculated Auger parameter was 720.1 (±0.2) eV, characteristic of metallic Ag. However, for the CVI90 catalyst, the Auger parameter was difficult to determine due to uncertainty in the absolute peak maximum, however, it is reasonable to assume that the Ag was in a similar state to the other CVI catalysts. The IW and IMP catalysts, as for the CVI90 catalysts, exhibited a low Ag concentration and a weak Auger parameter, suggesting that the Ag is present as large particles, which is confirmed by the XRD analysis.

Materials and Methods
The experimental study prepared 2 % Ag-20 % K/CZA by a variety of different methods (wet impregnation, incipient wetness and chemical vapour impregnation at different temperatures) and testing them for their ability to simultaneously reduce NOx and oxidise carbon black in a simulated fuel-lean exhaust gas. Ag/CZA is a known catalyst for soot combustion as well as having the ability for simultaneously remove NOx and soot. K is a well-known promoter of soot oxidation. The method of catalyst preparation was studied in order to establish the ideal preparation technique and to develop a deeper understanding of the properties that influence the catalyst's ability to simultaneously remove NOx and soot.

Ceria zirconia alumina (CZA) Support
The CZA support was prepared by co-precipitation. Additionally, in the present study, 0. 25

Wet Impregnation (IMP)
A total of 2 wt% Ag-20 wt% K/CZA was prepared by dissolving AgNO3 (99.9999%, Sigma-Aldrich, Dorset UK) and K2CO3 (99%, Fisher Scientific, Loughborough, UK) in de-ionized water (5 cm 3 ). The solution was heated to 90 °C with constant stirring before the CZA support (0.78 g) was added. The solution was maintained at 90 °C and was stirred until all the excess water had been evaporated or absorbed to leave a grey paste. The solid was then dried (16 h, 110 °C) and then calcined under flowing air (5 h, 10 °C min −1 , 500 °C). This catalyst is referred to as IMP.

Incipient Wetness (IW)
A weight loading of 2 wt% Ag-20 wt% K/CZA was prepared by dissolving AgNO3 and K2CO3 in minimum amounts of distilled water before the CZA support was added. The catalyst was dried (16 h, 110 °C) and calcined under flowing air (5 h, 10 °C min −1 , 500 °C). This catalyst is referred to as IW.

Characterization
The specific surface area was obtained by applying the BET method to a 20-point N2 adsorption isotherm measured over the range 0.05-0.35 p/p0 at −196 °C. The data were obtained using a Quantachrome Quadrabsorb (Bracknel, UK). The catalysts were degassed under vacuum for 16 h at 120 °C before surface area analysis was carried out.
Powder X-ray diffraction was carried out using a Panalytical X'Pert diffractometer (Malvern, UK) equipped with a Cu X-ray source operating at 40 kV and 40 mA. The ICDD standard database was used for phase identification. Crystallite size was estimated by using the Scherrer equation. Raman spectroscopy was performed using a Renishaw inVia confocal Raman microscope (Miskin, Wales, UK) equipped with an argon ion visible green laser. The laser wavelength used was 514 nm and spectra were collected in reflective mode using a highly sensitive charge couple device detector. CVI catalysts were dark in colour and were therefore difficult to obtain a Raman spectrum from the pure catalysts, hence they were mixed with KBr to enable suitable spectra to be obtained.
X-ray Photoelectron Spectroscopy (XPS) was carried out on the catalysts using a Kratos Axis Ultra DLD photoelectron spectrometer (Manchester, UK) utilizing monochromatic AlKα radiation operating at an energy of 120 W (10 x 12 kV). The data were analyzed using CasaXPS (version 2.3.23, Teignmouth, UK) and were modified using Wagner sensitivity factors, as supplied by the instrument manufacturer after the subtraction of a Shirly background. All spectra were calibrated to the C(1s) line taken to be 284.8 eV.
Scanning Electron Microscopy (SEM) was performed on a Tescan MAIA3 field emission gun scanning electron microscope (Cambridge, UK) fitted with an Oxford Instruments X-ray MaxN 80 detector. The catalysts were loaded onto a carbon tape. To help prevent the charging of the samples, the catalysts were sputter coated with Au/Pd.

Catalyst Performance Testing
Previous work has established that carbon black (Cabot Black Pearls 2000, Cabot Corporation, Latvia) was a reliable diesel soot mimic [23], samples taken from a single batch were used throughout the catalytic tests in this work.
The activity of the catalysts for the selective catalytic reduction of NO by NH3 was analyzed using a micro-reactor connected to an Fourier Transform Infra-Red (FTIR) gas analyzer (Gasmet). A reactant gas mixture comprising of 500 ppm NO, 500 ppm NH3, 8% O2, and N2 as the balance gas was passed over either 0.25 g of catalyst or over 0.25 g of catalyst and 0.025 g carbon black at a flow rate of 200 ml min −1 . Based on the catalyst volume, the GHSV for the experiments was 40,000 h -1 . The carbon black-catalyst mixture was prepared by mixing together with a spatula until homogenously achieving loose contact, which is comparable to the contact between a catalyst washcoat and trapped soot in a DPF [38]. For these sets of experiments, water and CO2 were not introduced into the system to allow their formation to be detected to monitor the oxidation of carbon black. The furnace temperature was periodically increased in 25 °C increments from 125 up to 550 °C. The results from the experiments were recorded and analyzed using Origin17 (OriginLab Corporation, Northampton, Massachusetts, USA)

Conclusions
In this study, the effect of the preparation method for 2% Ag-20% K/CZA for the simultaneous removal of NOx and soot from diesel exhaust by the in situ catalytic generation and utilization of N2O was investigated. A supported Ag catalyst, known to be active from previous studies, was modified by the addition of K and prepared by wet impregnation, incipient wetness, and chemical vapor impregnation at four different temperatures. It was found that only the CVI80, impregnation, and incipient wetness catalysts utilized the in situ generated N2O to oxidize soot at low temperatures. However, despite the incipient wetness catalyst utilizing in situ generated N2O to oxidize soot at low temperatures, overall, it was shown to be a poor soot oxidation catalyst. The addition of K vastly improves the catalytic oxidation of soot but has a negative impact on the reduction of NOx compared to previous work [23]. The results indicate that a 20 wt % loading of potassium is excessive, by overly hindering NOx reduction, though still allowing N2O to be formed directly from NH3. Hence, a loading study is required to identify the optimum surface concentration of K for both N2O formation by non-selective NOx reduction and soot oxidation.

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