Pt Catalyst over SiO 2 and Al 2 O 3 Supports Synthesized by Aerosol Method for HC-SCR DeNOx Application

Silica-supported platinum (Pt/SiO2) and alumina-supported platinum (Pt/Al2O3) catalysts have been prepared by an aerosol spray pyrolysis method. Systematic characterization of each catalyst using TEM, XRD, and XPS revealed that crystalline and metallic Pt nanoparticles were well dispersed on the surface of silica and alumina supports. The sintering effect on Pt particles over Al2O3 at high temperature (~250°C) is more prominent than those over SiO2; this suggests that there is stronger interaction between Pt particles and SiO2 support, when compared to Pt over Al2O3 support, resulting in Pt particles size below 3 nm. Moreover, steady-state catalytic experiments for selective reduction of nitrogen monoxide by propene have demonstrated that NOx conversions to N2O and N2 in Pt/SiO2 and Pt/Al2O3 catalysts are 29.8% and 55.8% at 250°C, respectively.


INTRODUCTION
Emission of NO x gases from both mobile and stationary sources causes serious environmental problems such as acid rain and photochemical smog (Pitchon and Fritz, 1999;Zhang et al., 2011;Wanga et al., 2012).Thus, NO x treatment technologies have drawn much attention in view of catalysis, especially for treating various exhaust gases produced by state-of-the-art automobile engines (Haj et al., 2002).Hydrocarbon (HC) species are often emitted from automobiles under lean-burn operating conditions, in the forms of fragments (propane, propene, etc.) and/or more primitive polycyclic aromatic hydrocarbons (PAHs) which are known as mutagenic and carcinogenic compounds causing cataracts, kidney, and liver damage (Lu et al., 2011;Dong and Lee, 2009).Therefore, the European Union implemented increasingly stringent regulations on emissions, limiting HC and NO x concentrations to lower than 0.17 g km -1 for passenger cars using diesel engines (The European Parliament and The Council of The European Union 2007).
The unburned HCs existing in the exhaust line can be used as a reducing agent for NO x treatment (Pitchon and Fritz, 1999;Haj et al., 2002) prior to tail-pipe emission, which is known as hydrocarbon-enhanced selective catalytic reduction (HC-SCR) of NO x .Using the HCs for NO x treatment enables more efficient fuel uses, while reducing HC and NO x emissions under lean-burn conditions (Burch, 2004;Wang et al., 2011).In this prospect, developing various catalysts for HC-SCR has been an intensive research area, which was initiated by Iwamoto et al. in 1990 (Iwamoto, 1990;Jayat et al., 1999).Then, Burch (Burch and Millington, 1995;Burch and Millington, 1996;Burch and Watling, 1997;Burch and Watling, 1998;Burch et al., 1998;Burch et al., 2002;Burch, 2004) performed a series of pioneering investigations for revealing catalytic performance and reaction kinetics of HC-SCR catalysts.Among various formulations of catalysts including Ag, Pd, Co, Ni, Au, Rh, and Cu, Pt-based catalysts showed excellent performance for NO x reduction at relatively low temperatures (Burch andMillington, 1995, 1996;Burch and Watling, 1997;Jayat et al., 1999;Burch et al., 2002).However, platinum suffers from three serious drawbacks: a narrow temperature window of operation, an objectionably high selectivity toward N 2 O (Burch and Watling, 1997) rather than a major species of NO or NO 2 , and a poor thermal stability (Asoro et al., 2010).In an attempt to resolve the issue of thermal stability, namely that Pt particles are easily sintered into large sizes, thereby significantly changing their behavior and catalytic performance (Suzuki et al., 2010), supporting materials such as silica and alumina have been introduced (Gonzalez et al., 1997;Nagai et al., 2006).As an example, Burch et al. (1996) reported that the maximum conversion of NO for Pt/SiO 2 and Pt/Al 2 O 3 catalysts reached 75% and 60%, respectively.Furthermore, such metal oxide supports are known to lower the working temperatures of Pt catalysts and minimize the undesirable corrosion effect of water vapor and sulfur oxides (SO x ) (Burch and Watling, 1997;Gonzalez et al., 1997;Männikkö et al., 2012).
Recent studies suggested that specific catalytic activities are sensitive to not only the reducing gases and supporting materials, but also the dispersion, size, crystallographic nature, and microstructure of the Pt catalysts (Seker and Gulari, 2000;García-Cortés et al., 2003;Vaccaro et al., 2003;Després et al., 2004;Radic et al., 2004;Mulla et al., 2006).Given a limited amount of precious Pt in a sample, well-dispersed Pt particles with a moderate size (3-7 nm) are ideal for efficient conversion of NO x (Jayat et al., 1999;Vaccaro et al., 2003;García-Cortés et al., 2003;Després et al., 2004).The existence of very large particles (> 10 nm) can cause a significant deterioration in particle dispersion, and can lead to less catalytic activity.Very small particles (< 2 nm) result in far better dispersion, but easy oxidation, which causes deactivation of the catalytic function (García-Cortés et al., 2003).In addition, crystallographic orientation of Pt determined by the ratio of Pt (100) peak to Pt (111) peak from X-ray diffraction may affect N 2 selectivity because NO molecules can be dissociatively adsorbed on Pt (100), while NO is mostly adsorbed molecularly on Pt (110) and Pt (111) (García-Cortés et al., 2003).Regarding the chemical oxidation state of Pt, several researchers (Després et al., 2004;Mulla et al., 2006) reported that formation of PtO and PtO 2 can degrade catalytic activity and eventually lead to deactivation of the catalyst.Another important factor determining catalytic activity is the catalyst preparation method; some materials prepared by the sol-gel method show improved activity and N 2 selectivity, even in the presence of high water vapor concentrations (2.6-10%) (Seker and Gulari, 2000;Burch et al., 2002).
The Pt catalysts on metal oxide supports produced by the incipient wetness impregnation and ion-exchange methods (Burch andMillington, 1995, 1996;Burch and Watling, 1997;Burch et al., 1998;Männikkö et al., 2012) have been used to reveal that the aforementioned thermal stability problem is due to weak interactions with Pt and the supporting material.It has been established that the thermal stability issue with Pt is less severe in catalysts synthesized by the sol-gel method (Gonzalez et al., 1997;Cho et al., 1998;Romero-Pascual et al., 2002;Hu et al., 2006).Alternatively, flame spray pyrolysis (FSP) has been used for producing aluminasupported platinum catalysts (Strobel et al., 2003(Strobel et al., , 2006)).More recently, the aerosol-based process has been widely investigated and used for making catalysts for deNOx (Boningari et al., 2012), deVOCs (Lin et al., 2015), deCO (Byeon et al., 2014) applications.While the aerosol methods have been recognized to be well developed, the methods have had limited success in control of agglomeration of catalysts.The uncontrolled agglomeration might reduce exposed surface areas of catalysts, leading to degradation of catalytic activity and performance.
Thus, we focused on the agglomeration/microstructural control and thermal stability of catalysts.Combining the sol-gel method with the aerosol spray-pyrolysis method (called the aerosol-gel method hereafter), we may take the unique benefits of each method (Lee et al., 2007;Firmansyah et al., 2009Firmansyah et al., , 2012)), together with the advantage of costeffective production of microstructure-controlled particles (Lee et al., 1997;Santiago et al., 2011).Better understanding the process, one can extend this method to the synthesis of binary hetero-structured catalysts.
In this study, we have applied the aerosol method for the synthesis of well-dispersed Pt particles on SiO 2 and Al 2 O 3 supports The two types of catalysts exhibited comparable NO conversion with respect to the previous reports.The effect of sintering on particle size distribution and relevant possible mechanisms were explored.

Preparation of Catalysts
A precursor solution for Pt/SiO 2 catalysts was prepared as follows: 56 mg platinum (II) acetylacetonate (Pt(acac) 2 , Aldrich, 97%) well dissolved in 33.6 mL ethanol (Aldrich, 99.5%), was mixed with 3.2 mL TEOS (Aldrich, 98%) dissolved in 17.6 mL ethanol (Aldrich, 99.5%), and the mixture solution was stirred at room temperature for 30 min.Then, 12.5 mL deionized (DI) water and 1.2 mL HCl (0.1 M) were introduced to the mixture, and then the solution was stirred at room temperature for 2 hours in order to initiate the sol-gel reaction of TEOS with water.Herein only SiO 2 sols are formed, maintaining the Pt(acac) 2 dissolved in the solution at low temperature.
A precursor solution for Pt/Al 2 O 3 catalysts was prepared likewise: 40 mg Pt(acac) 2 was first dissolved in 40 mL ethanol and stirred at 50°C for 30 minutes.A second solution was made by dissolving 800 mg aluminum nitrate nonahydrate (ANN, Aldrich, 98%) into 40 ml ethanol.The two solutions were mixed together, and stirred at 50°C for 30 minutes.Then, 20 mL of water and 0.1 mL of HCl (0.1 M) were introduced into the solution and stirred again under 50°C for two days.
As-prepared two different solutions were aerosolized to 1-5 µm droplets in an atomizer running with compressed air at a pressure of 25 psi and a gas flow rate of 2.2 L min -1 .The droplets were therefore fed into a diffusion dryer, filled with silica gel to remove the solvent.They were then transferred to a quartz tube (1/4" in diameter), and heated by tube furnace, in which Pt(acac) 2 is pyrolyzed to form tiny Pt nanoparticles.It should be noted that the droplets contain not only SiO 2 sols and Pt aggregates but also remaining TEOS solution and Pt(acac) 2 , in the case of Pt/SiO 2 catalysts.In the case of Pt/Al 2 O 3 , however, the droplets contain liquidphase reactants of Pt(acac) 2 and Al(NO 3 ) 3 until the reactant temperatures reach an adequate level in the furnace; this occurs as aluminum nitrate is never sol-gel reacted, but rather pyrolyzed at high temperature.It is noted that organic species in TEOS and Pt(acac) 2 appear to leave carbon behind after the decomposition reaction when Ar was used instead of air.This is the reason why air has been used as a carrier gas.
The formation positions of solid particles (i.e., Pt with respect to its metal oxide support) can vary depending on the differences in their reaction kinetics.The disparity in formation timing of Pt and metal oxides enables control of the morphology in such a way that Pt dots are formed on the outer surface of the SiO 2 or Al 2 O 3 particles.For reference (control) materials, pure SiO 2 or Al 2 O 3 nanoparticles were also prepared in the same way as described above (without Pt source), and then examined for their performance as catalytic promoters.The residence time of particles in the furnace is estimated at approximately 0.19 s.The reacting temperature of the furnace was set to either 600°C for Pt/SiO 2 catalysts, or 700°C for Pt/Al 2 O 3 catalysts.Assynthesized catalysts nanoparticles were collected by a glass membrane filter with 1 µm pores (Milipore).More details about experimental setup are available elsewhere (Lee et al., 2007;Firmansyah et al., 2009Firmansyah et al., , 2012)).

Characterization of Catalysts
As-prepared catalysts particles were dispersed in ethanol with an ultrasonicator.A few milliliters of the suspension were dropped onto a carbon-coated TEM grid and then dried in air.The size and morphology of the catalysts were observed by a high-resolution transmission electron microscope (HR-TEM; JEM 2010F & 2011, JEOL, 200 kV) equipped with an energy dispersive spectrometer (EDS).Crystalline phase of the samples was characterized in view of mole composition and crystallite size by using X-ray photoelectron spectroscopy (XPS; ESCALAB 250 XPS spectrometer) with Al K Xray (1486.6 eV) with a resolution of 0.45 eV and X-ray diffraction (Bruker, AXS) with Cu K radiation (λ = 0.154 nm), respectively.Pt loading was measured by an inductive coupled plasma optical emission spectrometer (ICP-OES; JY ULTIMA 2 C HR, Horiba).

Test for Catalytic Activity of NO Reduction
Catalytic performance of each sample was tested as follows: each powder sample weighing 34 mg was placed inside a 1/8" diameter tube with 1 cm length, where both tube ends were blocked by a layer of glass wool.The tube was connected to 1/4" stainless steel tubes in a way that the sample holder was located at the center of a tube furnace, and provided, a reactant gas mixture composed of 236 ppm NO, 440 ppm C 3 H 6 , and 5% O 2 in He.Before testing, the powder samples were heated to temperatures up to 700°C for 4 minutes, with sufficient air flow to flush out any remaining liquid solvent from the sample.The samples were then cooled under Ar flow to 150°C.The temperature of the sample holder was controlled to a designated temperature in a range of 150-400°C.After the desired temperature was reached, the reactant gases began to flow through the catalyst bed in the sample holder at a total flow rate of 66 ccm.The mole composition of the reactant gases before and after passing through the catalyst bed was measured with Fourier transform infrared spectroscopy (FTIR; Nicolet 380, Thermo Electron) equipped with a 10-m optical-path gas cell.
The FTIR measurement was performed over a range of 800-2800 cm -1 with a resolution of 4 cm -1 .A calibration experiment was performed to obtain a correlation between the known composition and the resultant peak intensity for gases including NO, C 3 H 6 , CO 2 , N 2 O, and NO 2 at steady state.The resultant conversion efficiency (C Z ) of gas species Z, e.g., NO or C 3 H 6 gases, was calculated as follows: where [Z] I and [Z] T are the concentrations of the gas, Z, which were measured before and after the sample holder, respectively.There was no measured NO 2 species in the inlet for all temperatures (before sample holder), however, NO 2 species was generated immediately upon the contact of NO gas with Pt catalysts at temperatures higher than 250°C, causing full propene conversion (García-Cortés et al., 2003).The relative yield, X Z , and selectivity, S Z , of a product Z are defined as follows (Joubert et al., 2006): where n is the number of nitrogen atoms in the gas Z.The N-containing products (Z) of the reaction are N 2 , N 2 O, and NO 2 .N 2 cannot be detected by FTIR; therefore the product yield of N 2 is given by (Pitchon and Fritz 1999):

RESULTS AND DISCUSSION
Fig. 1(a) shows a typical TEM image of the as-prepared Pt/SiO 2 catalysts produced by the aerosol method.During the time for sol-gel reaction in the atomizer (2 hours in this study), small silica sols and aggregates were suspended in a colloidal solution, in which Pt(acac) 2 and a part of TEOS remain dissolved.The solution was aerosolized, subsequently heated up, and finally turned into spherical nanoparticles.Through HR-TEM characterization, it was revealed that the many tiny dark dots covering the large gray spheres were in-fact metallic Pt dots on the surface of amorphous silica spheres.Unlike the case of Pt/SiO 2 , with Pt/Al 2 O 3 , two precursors of Pt(acac) 2 and Al(NO 3 ) 3 remain dissolved in the atomized droplets, until they are decomposed to solid phase at high temperatures.Nevertheless, Fig. 1 measurement of the catalysts.As a result, the Pt dots are distributed from 1.3 to 3.5 nm in diameter, with mode diameters of 1.9 and 2.3 nm for Pt/SiO 2 and Pt/Al 2 O 3 , respectively.The larger size of Pt dots when supported by Al 2 O 3 is likely attributed to higher furnace temperature, which might facilitate to sinter Pt particles into larger dots.Additionally, it can be postulated that, in the absence of sol-gel reactions which can form a matrix structure to anchor the Pt particles (Firmansyah et al., 2012), coproduced Pt dots may have a better chance to move and sinter on the surface of alumina spheres.
Of particular interest is understanding the formation mechanism of Pt dots on the surface of the metal oxide spheres, in view of size and hetero-structure of Pt particles; when a droplet evaporates during the spray-pyrolysis process, the dissolved solutes are locally concentrated near the droplet surface, and diffuse toward the center of the droplet in response to the concentration gradient, while the solvent diffuses toward the outermost region, and evaporates.It is generally accepted that the overall morphology of single-component particles is determined by the competition of solute diffusion and solvent evaporation rates (Jayanthi et al., 1993;Lenggoro et al., 2000;Lee et al., 2007).In the present study, there are two diffusing solutes in a solvent droplet: for Pt/SiO 2 there are non-ionic molecules of Pt(acac) 2 and TEOS, and for Pt/Al 2 O 3 , there are Pt(acac) 2 molecules and Al 3+ ions (from dissolved ANN).A disparity of relative diffusion speed between the solutes likely determines local abundances of the solutes, in a way that a fast diffusing solute moves toward the droplet center more quickly than a slowly diffusing solute.Noting that product particles come out of the abundant regions, the existence of Pt dots on the surface seems to indicate that the solute of Pt(acac) 2 is the least diffusing species.In order to confirm that the diffusion coefficients of the three solute species, as a measure of the diffusion rate, are estimated to compare as follows.
The diffusion coefficient of Al 3+ ion in a dilute solution at 298 K was determined according to Nernst-Haskell equation: , where R g is the gas constant, F is the Faraday constant, ν + and ν - are respectively the valences of the cation and anion, and λ 0 is the limiting ionic conductance (Reid et al., 1987;Firmansyah et al., 2009Firmansyah et al., , 2012)).In this study, the D s value for Al 3+ ions is estimated to be 1.15 × 10 -5 cm 2 s -1 via ν Al3+ = 3, ν NO3-= 1, λ 0 Al3+ = 59.0 Ω -1 cm 2 mol -1 and λ 0 NO3-= 71.4 Ω -1 cm 2 mol -1 .For nonionic solutes of TEOS and Pt(acac) 2 , the corresponding values of D s are calculated by Wilke and Chang's equation (Reid et al., 1987;Firmansyah et al., 2009Firmansyah et al., , 2012) ) : where M B is the solvent molecular weight in g mol -1 , η B is the solvent viscosity in cP, V A is the molar volume of solute A at its normal boiling temperature in cm 3 mol -1 , and ϕ is the association factor.As a result, the diffusion coefficient of Pt(acac) 2 in ethanol is estimated to be 4.1 × 10 -6 cm 2 s -1 and the diffusion coefficient of TEOS in water is estimated at 8.0 × 10 -6 cm 2 s -1 .It is quite clear that the lighter species of TEOS and Al 3+ diffuse to the droplet center at higher speeds, leaving behind the heaviest Pt(acac) 2 near the surface; this may explain why Pt(acac) 2 decomposes to Pt dots mainly on the surface of the silica and alumina spheres.Another route to explain the presence of Pt on the surface of the metal oxide supports is from a disparity of evaporation rates between water and ethanol.It is noted that Pt(acac) 2 dissolves mainly in ethanol which is 40% more volatile than water (Reid et al., 1987), while TEOS and Al 3+ dissolves in water.Hence, faster evaporation/consumption of ethanol near a droplet surface initiates the precipitation of Pt(acac) 2 whereas TEOS and Al(NO 3 ) 3 both retreat to the center along with shrinking water.This is likely how a pair of solutes is segregated.
Figs. 2(a) and 2(b) show XRD patterns of Pt/SiO 2 and Pt/Al 2 O 3 catalysts in comparison with those of SiO 2 and Al 2 O 3 supports only, respectively.Fig. 2(a) clearly shows four major peaks of ( 111), ( 200), (220), and (331) crystalline facets of Pt as well as a very broad peak at ~23° in 2θ units.The broad peak also appears in the pure SiO 2 sample, indicating that the silica is almost amorphous.Likewise, Fig. 2(b) shows similar diffraction patterns of crystalline Pt on a mixture of amorphous and γ-phase alumina.Scherrer equation (Wang et al., 2006) is applied to the largest peak of Pt (111) so that the crystallite size for Pt/SiO 2 and Pt/Al 2 O 3 are estimated as 2.8 nm and 3.1 nm, respectively, which is consistent with TEM image analysis as shown in Fig. 1(c).
The oxidation states of Pt on silica and alumina supports are examined by XPS.Fig. 3(a) shows that the Pt 4f corelevel XPS spectrum of Pt/SiO 2 catalysts is mainly composed of three pairs of doublets.The highest doublet (at 71.2 and 74.7 eV) corresponds to metallic Pt.The second highest doublet (at 72.3 and 75.5 eV), 1 eV blue-shifted from the metallic peaks of Pt, is assigned to Pt [II] states as in PtO or Pt(OH) 2 .The weakest third doublet at 73.6 and 76.7 eV suggests that a minor level of Pt [IV] as in PtO 2 exists in the present catalysts (Kang et al., 2005;Choi et al., 2010).For Pt/Al 2 O 3 catalysts, however, the Pt 4f core-level XPS spectrum is considerably different from that of Pt/SiO 2 , mainly because the Pt 4f lines are overlapped with the Al 2p lines.Because of this interference, it is almost impossible to measure the 4f 5/2 peak intensity of Pt on alumina especially when the Pt content is below 5%.Hence, the Pt 4d 5/2 line is measured as an alternative (Shyu and Otto, 1988).Fig. 3(b) shows the Pt 4d 5/2 core-level XPS spectrum of Pt/Al 2 O 3 catalysts deconvoluted to three subpeaks.In Fig. 3(b), the highest peak at 314 eV is the signal from metallic Pt (Shyu and Otto 1988).The second peak at 315.3 eV is assigned to Pt [II] state as in PtO or Pt(OH) 2 (Shyu and Otto, 1988).The third peak at 316.9 eV corresponds to Pt [IV], indicating the existence of PtO 2 (Shyu and Otto, 1988).The surfaceoxygenated species of PtO and PtO 2 in the as-prepared Pt catalysts, which are presumably produced during the decomposition reaction at elevated temperatures of 600°C (for Pt/SiO 2 ) and 700°C (for Pt/Al 2 O 3 ) in air (Hauff et al., 2012), are believed to hinder the NOx dissociation on the surface of metallic Pt (Romero-Pascual et al., 2002;Hu et al., 2006).Considering that the relative abundance of Pt, PtO, and PtO 2 is often approximated by the corresponding XPS peak area, one may notice from Figs. 3(a) and 3(b) that metallic Pt, as a major species, dominates other minor species of PtO and PtO 2 .Furthermore, relative abundance of each species approximated from the surface area of deconvoluted peaks in Figs.3(a) and 3(b), and shown in Table 1, indicates that metallic Pt 0 is dominant for both catalysts.The value of metallic Pt 0 reported here (73%-78%) are in good agreement with previous results for the Pt samples heat treated by flowing H 2 /N 2 at 450°C (Despres et al., 2004) or N 2 at 1500°C (Kang et al., 2005).
The overall HC-SCR reaction for the produced catalysts of Pt/SiO 2 or Pt/Al 2 O 3 is described as: where the coefficients of a, b, c, d, e, f, g, and h are dependent on the types of catalysts, reactant concentration, and temperature.To investigate the reaction mechanism, a compositional analysis for the mixture of the reactants and the products was performed using gas-cell FTIR.Figs.4(a)-4(d) show FTIR spectra obtained before and after passing the reactant gases in Eq. ( 7) through SiO 2 , Al 2 O 3 , Pt/SiO 2 and Pt/Al 2 O 3 catalysts tested at different temperatures, respectively.FTIR spectrum at the bottom of each figure indicates a little but distinct peak of CO 2 even before contacting the catalysts.The trace level of CO 2 is believed to come from indoor atmosphere by gaseous diffusion to the gas cell of the FTIR machine.Two small double peaks at 1820-1960 cm -1 are assigned to NO gas (Smith et al., 1988) and marked with a dotted box.Other peaks at 912 cm -1 , 1598 cm -1 and 2210 cm -1 are assigned to C 3 H 6 (Yoon et al., 2002), NO 2 (Li and Guan, 2009) and N 2 O (Sobczak et al., 2008), respectively.Though the peak of NO 2 gas is overlapped with that of H 2 O (Tshilongo et al., 2009) at ~1600 cm -1 , it is quite clear at which temperature the NO 2 gas peak begins to be prominent in the presence of Pt (see Figs. 4(c) and 4(d)).
Figs. 4(a) and 4(b) show the spectrum of evolved gases from the catalytic support of SiO 2 and Al 2 O 3 without Pt.The profiles of spectrum for 150°C and 200°C are almost the same as the inlet, implying that there is no catalytic activity of the supports in this temperature range.At temperatures of 250°C and 300°C, we observe propene combustion catalyzed by metal oxide support in which the SiO 2 support appears more active than that of Al 2 O 3 ; this is due to the fact that NOx and O 2 cannot form an adsorbed species on the inert support of SiO 2 .As a result, O 2 (g) will react directly with hydrocarbon in the gas phase to generate CO 2 and H 2 O when the temperature is higher than 250°C.On the other hand, NO x and O 2 can form an adsorbed species (NO ads ) (Burch et al., 1998) on Al 2 O 3 which has higher surface acidity (Hirashima et al., 1998).In this case, hydrocarbon loses chance to react with O 2 (g) in the gas phase, rather reacts with NO ads at higher temperatures, and eventually generate CO 2 and H 2 O.For this reason, alumina might be apparently less active than silica for hydrocarbon oxidation.At temperatures of 350°C and 400°C, NO reduction of about 20% is observed with Al 2 O 3 catalysts, while NO reduction with SiO 2 is not observed at any temperatures tested.These facts suggest that Al 2 O 3 should be more favorable as a catalyst support for the NO treatment than SiO 2 , and may contribute to better activity when it is combined with Pt.For Pt/SiO 2 catalyst at 150°C, the spectrum is almost identical to that of original reactant gas mixtures (noted as 'inlet' in Fig. 4(c)).As temperature is increased to 200°C, the peaks of NO and C 3 H 6 exhibit a slight decrease, while the peaks of N 2 O and CO 2 species show distinctive increases.This suggests that NO gas is partly reduced to N 2 O and N 2 , and that the O from the catalyst surface might be consumed to oxidize the incoming C 3 H 6 gas to generate CO 2 .This is exactly how HC-SCR catalysts are supposed to function via a redox process (Burch and Millington, 1996).A dramatic change is observed in every gas species when the temperature is increased further to 250°C: a sudden disappearance of C 3 H 6 , associated with a huge increase in CO 2 peak, a relatively large decrease in NO peak height, a noticeable increase of NO 2 peak, and a sudden stop to change in N 2 O peak.Unlike the first three observations, the fourth and fifth observations indicate that NO is not properly reduced to N 2 O or N 2 , but rather oxidized to NO 2 even at the expense of C 3 H 6 .The preferential oxidation of NO suggests that there is more abundant chemisorbed oxygen than chemisorbed nitrogen on Pt, making the oxidation more probable than reduction.A larger peak of NO 2 and a small peak of N 2 O at 275°C and at 300°C were observed; the conversion of NO to NO 2 for Pt/SiO 2 at these temperatures is similar with the catalytic result as reported by Jayat et al. (1999).Moreover, the role of C 3 H 6 in consuming the chemisorbed oxygen at the Pt surface becomes insignificant as the catalyst surface is almost completely covered by chemisorbed oxygen.As the temperature is increased to 350°C, the NO 2 peak and NO peaks slightly decrease, starting at this temperature, generation of gas phase NO 2 from the catalyst is limited by the NO 2 equilibrium concentration as indicated by previous authors (Seker and Gulari, 2000;García-Cortés et al., 2003).Eventually, when the temperature is increased to 400°C, C 3 H 6 and N 2 O almost completely disappear, suggesting that Pt oxidation is predominantly arising, which eventually leads to the degradation process of catalytic activity.
Similar to the Pt/SiO 2 catalyst, the spectrum for the Pt/Al 2 O 3 catalyst, at 150°C, is also almost identical to that of the original reactant gas mixtures (noted as 'inlet' in Fig. 4(d)).As temperature is increased to 200°C and 250°C, the NO peak decreases, the C 3 H 6 peak disappears, and the N 2 O peak experiences a marked increase.Thus, at this temperature, NO gas is partly reduced to N 2 O and N 2 , and oxygen likely binds to the catalyst (Pt) surface.We observe that NO 2 gas (which indicates an oxidation process on the Pt surface) does not appear at this temperature, though C 3 H 6 (which indicates a reduction process on the Pt surface) is no longer available.Thus, the Pt/Al 2 O 3 catalyst appears to perform better in NO x conversion than the Pt/SiO 2 catalyst.Moreover, Pt/Al 2 O 3 exhibits clearly higher peak of N 2 O gas, suggesting that most of the Pt sites are likely in a reduced state promoting more NO decomposition to N 2 and N 2 O.A continuous decrease of N 2 O species is observed at 275°C and 300°C while NO 2 concentration is observed to increase to maximum yield of about 26% at these temperatures.In addition, the role of C 3 H 6 in consuming the chemisorbed oxygen at the Pt surface over the alumina support becomes insignificant at that temperature, suggesting that the catalyst surface begins to be covered by more chemisorbed atomic oxygen.Finally, in a process similar to that of the Pt/SiO 2 catalyst, as temperature is increased further to 350°C and 400°C, the presence of C 3 H 6 and N 2 O is minute; suggesting that the oxidation process in Pt is again predominantly arising, which leads to the degradation of catalytic activity.And again, the decrease of NO 2 concentration at such high temperature is affected by the equilibrium concentration (Seker and Gulari, 2000;García-Cortés et al., 2003).
As explained in section 2, the degree of conversion X R (refer to Eqs. ( 1)-( 2)) for NO and C 3 H 6 were measured on the basis of FTIR measurement of gas mixture before and after the gases passed through the catalysts.In addition, the as-obtained conversions of NO to NO 2 and N 2 O were used to estimate the conversion of NO to N 2 on the basis of mass balance.Every corresponding conversion of each species was plotted as a function of temperature in Figs.5(a  Before catalytic testing, we performed an extra heat pretreatment in an attempt to further enhance catalytic performance as Després et al. (2004) and Kang et al. (2005) did.We observe that non-pretreated samples show less NO conversion, 39% and 45% for Pt/SiO 2 and Pt/Al 2 O 3 at 250°C, respectively.In order to investigate the effect of the pre-treatment on the Pt catalysts particles, we analyzed pretreated samples that had been heated up to 700°C under air with a heating rate of 10 °C min -1 .The benefit of using air as a carrier gas is that it allows for reducing the PtO monolayer (Després et al., 2004), and increasing the size of Pt particles (García-Cortés et al., 2003;Santiago et al., 2011) to obtain the optimal size for catalytic activity.solution (see section 2).These reactions thereby reduce the sintering effect, and thus the dispersion of Pt particles over SiO 2 is superior to that of the dispersion of Pt particles over Al 2 O 3 after sintering.Well dispersed Pt particles, with sizes larger than 2 nm may show a better NO x conversion according to previous works (García-Cortés et al., 2003;Vaccaro et al., 2003).In addition, a different Pt particle size may exhibit various reactivity to Pt in which a higher rate of surface reaction may be obtained with large Pt crystallites, as compared to small Pt crystallites (Radic et al., 2004).However, Pt particles that are too large (> 10 nm) are less active due to weaker O bond formation (García-Cortés et al., 2003), while Pt particles that are too small are easily oxidized to PtO which is less active for NOx conversion (García-Cortés et al., 2003).Thus, a moderate size (3-8 nm) of Pt particles is recommended for efficient conversion of NO x (Jayat et al., 1999;García-Cortés et al., 2003;Vaccaro et al., 2003;Després et al., 2004), therefore, changing the average size of Pt particles over SiO 2 and Al 2 O 3 supports from ~2 nm to larger sizes of 3.8 and 8.4 nm may contribute to the NO x conversion observed in Figs.5(a) and 5(b).

CONCLUSIONS
Catalyst of well dispersed Pt particles over SiO 2 and Al 2 O 3 supports were fabricated successfully via an aerosol method combined with pyrolysis.According to calculations for aerosol particle formation, the solute of Pt particles concentrates locally near the droplet surfaces while TEOS and Al 3+ diffuse toward the center of the droplets, implying that Pt particles disperse on the metal oxide supports.The results of HR-TEM studies corroborate our calculations showing that the Pt/SiO 2 and Pt/Al 2 O 3 catalyst particles were uniformly dispersed at the surface of each supporting material in the respective samples.The XRD patterns confirm that Pt/Al 2 O 3 particle size is slightly larger than that of the Pt/SiO 2 particles.The results of XPS and peak deconvolution show that the surface of Pt/SiO 2 and Pt/Al 2 O 3 are mostly covered by metallic Pt.Moreover, the maximum NO x reduction obtained for Pt/SiO 2 and Pt/Al 2 O 3 catalysts are 29.8% and 55.8%, at 250°C, respectively.
The Pt particles over SiO 2 and Al 2 O 3 supports from pristine catalysts before heat pretreatment are homogeneously distributed with similar particle sizes of ~2 nm.However, the Pt particles sizes are completely different at high temperature after the heat pretreatment, where the sintering effect on Pt over Al 2 O 3 is more severe than that of Pt over SiO 2 , suggesting that there is a stronger interaction between Pt and the SiO 2 support.

Fig. 1 .
Fig.1(a) shows a typical TEM image of the as-prepared Pt/SiO 2 catalysts produced by the aerosol method.During the time for sol-gel reaction in the atomizer (2 hours in this study), small silica sols and aggregates were suspended in a colloidal solution, in which Pt(acac) 2 and a part of TEOS remain dissolved.The solution was aerosolized, subsequently heated up, and finally turned into spherical nanoparticles.Through HR-TEM characterization, it was revealed that the many tiny dark dots covering the large gray spheres were in-fact metallic Pt dots on the surface of amorphous silica spheres.Unlike the case of Pt/SiO 2 , with Pt/Al 2 O 3 , two precursors of Pt(acac) 2 and Al(NO 3 ) 3 remain dissolved in the atomized droplets, until they are decomposed to solid phase at high temperatures.Nevertheless, Fig.1(b)shows a very similar HR-TEM image of Pt/Al 2 O 3 catalysts: Pt dots on large Al 2 O 3 spheres.The sizes of Pt in both cases were measured and their size distributions from TEM image analysis are displayed in Fig.1(c): more than 350 Pt nanoparticles in each case were considered and the same set of Pt particles were measured twice to yield an average diameter in an attempt to reduce the uncertainty for size

Fig. 6 .
Fig. 6. a) TEM micrograph of the pre-treated sample of Pt/SiO 2 , b) TEM micrograph of the pre-treated sample of Pt/Al 2 O 3 , c) Histogram plot of particle size distribution for the pre-treated samples of Pt/SiO 2 and Pt/Al 2 O 3 .

Table 1 .
The relative abundance of different Pt species determined from Peak Deconvolution of XPS measurements.
250°C for Pt/SiO 2 and at 200°C for Pt/Al 2 O 3 .For NO conversion, the profile exhibits a maximum peak at 55% and 66% conversion at 300°C for Pt/SiO 2 and Pt/Al 2 O 3 , respectively.However, the higher NO x conversion of Pt/Al 2 O 3 consequently produces more N 2 O gas, and therefore reduces the N 2 yield (production) and N 2 selectivity.The maximum conversion of NO to N 2 for both catalysts occurs at 300°C, which corresponds to 20% and 17.5% conversion for Pt/SiO 2 and Pt/Al 2 O 3 , respectively.As a result, the N 2 selectivity for both catalysts abruptly increases to 50% and 31% for Pt/SiO 2 and Pt/Al 2 O 3 at 250°C, respectively.In addition, the maximum yield of NO 2 is produced at the same level for both catalysts, and thus the maximum NO conversions (reduction) to N 2 + N 2 O gases for both catalysts reach about 29.8% at 250°C for Pt/SiO 2 , and 55.8% at 200~250°C for Pt/Al 2 O 3 .These facts imply that the catalytic performance of Pt/Al 2 O 3 catalyst is likely better than Pt/SiO 2 catalyst in this test.