Water–gas shift activity of pt catalysts prepared by different methods

: Platinum supported on ceria and zirconia was prepared through di ﬀ erent preparation methods: Coprecipitation (CP), spray drying (SD), and ﬂame spray pyrolysis (FSP). The catalysts were characterized by XRD, TPR, N 2 adsorption, and H 2 chemisorption, and the water–gas shift activity in the range 190–310 ◦ C and initial stability at 300–310 ◦ C were tested. Although the spray-dried Pt / CeO 2 / ZrO 2 catalyst shows the highest initial activity, it deactivates rapidly at 300 ◦ C and levels out at similar activity as the coprecipitated Pt / CeO 2 and Pt / CeO 2 / ZrO 2 within a few hours. Flame spray pyrolysis appears to be a promising preparation method concerning the stability of catalysts, although the initial activity is rather poor. High activity is related to high Pt dispersion, low reduction temperature, and small support particles. The support particle size is also much a ﬀ ected by the preparation method.


Introduction
The water-gas shift (WGS) reaction has been an important step in industrial hydrogen production from hydrocarbons since the early 1940s: Recently, there is a renewed interest in the reaction because of its importance in the production of hydrogen feed with a low CO level for fuel cell applications [1]. New technologies require improved catalysts, since the commercial Cu/ZnO/Al 2 O 3 catalyst has limitations that will cause problems. One disadvantage is that this catalyst is designed to operate at steady-state conditions for a long time without interruption, whereas the catalysts for fuel processing have to deal with many start-up/shutdown cycles. In addition, operation at too high temperatures must be avoided, since the sintering of copper in these catalysts starts at about 330 • C [2]. It is also sensitive to oxygen and poisoned by sulfur [3]. The use of precious metals has been widely investigated. Precious metal catalysts are stable at higher temperatures; they are nonpyrophoric and also more tolerant of catalyst poisons such as sulfur. A recent review focuses on the use of Pt-based catalysts for WGS [4]. Ceria provides interesting support for precious metals, which are not easily oxidized by water and thus have, in general, poor WGS activity. By using ceria as support, CO adsorbed on the metal can react with oxygen from ceria [5]. 2 of 10 Ceria based catalysts are, however, prone to deactivation under reaction conditions [6,7], and hence promoters and stabilizers are used to improve the performance of the catalyst. The application of cerium-zirconium mixed oxides has become quite prominent, and the incorporation of zirconium into the ceria fluorite-type structure is expected to improve the activity and stability of the mixed oxide [1,[8][9][10]. The preparation method highly influences the properties of the ceria supported catalysts as well [11][12][13][14]. Flame spray pyrolysis and spray drying are methods that are expected to give high surface area catalysts with finely dispersed active metal. The short residence time at high-temperature results in the formation of small particles, and the homogeneous precursor solution yields homogeneous precipitate particles [15][16][17][18][19][20]. Such properties are considered beneficial for the water-gas shift reaction [12,[21][22][23][24][25].
In this study, Pt catalysts supported on ceria and/or zirconia were synthesized by three different preparation methods: Coprecipitation/impregnation, spray drying, and flame spray pyrolysis. The effect of the preparation methods on the characteristics and WGS activity for the catalysts was investigated. Figure 1 shows the XRD patterns of the catalysts. Crystalline CeO 2 and ZrO 2 supports are detectable for most of the catalysts. An exception is Pt/ZrO 2 _SD, where ZrO 2 seems to be crystalline when it is prepared by coprecipitation or flame spray pyrolysis, but extremely broad peaks are observed when it is prepared by spray drying. Despite a calcination temperature of 650 • C, zirconia seems to have an amorphous structure in Pt/ZrO 2 _SD. Due to the low Brunauer-Emmet-Teller (BET) surface area, it is not very probable that it consists of very small particles. The ceria particle sizes are very close to those of the coprecipitated samples, indicating that also the spray-dried samples are more agglomerated than the samples prepared by flame spray pyrolysis.

Characterization
Catalysts 2020, 10, x FOR PEER REVIEW 2 of 10 with oxygen from ceria [5]. Ceria based catalysts are, however, prone to deactivation under reaction conditions [6,7], and hence promoters and stabilizers are used to improve the performance of the catalyst. The application of cerium-zirconium mixed oxides has become quite prominent, and the incorporation of zirconium into the ceria fluorite-type structure is expected to improve the activity and stability of the mixed oxide [1,[8][9][10]. The preparation method highly influences the properties of the ceria supported catalysts as well [11][12][13][14]. Flame spray pyrolysis and spray drying are methods that are expected to give high surface area catalysts with finely dispersed active metal. The short residence time at high-temperature results in the formation of small particles, and the homogeneous precursor solution yields homogeneous precipitate particles [15][16][17][18][19][20]. Such properties are considered beneficial for the water-gas shift reaction [12,[21][22][23][24][25].
In this study, Pt catalysts supported on ceria and/or zirconia were synthesized by three different preparation methods: Coprecipitation/impregnation, spray drying, and flame spray pyrolysis. The effect of the preparation methods on the characteristics and WGS activity for the catalysts was investigated. Figure 1 shows the XRD patterns of the catalysts. Crystalline CeO2 and ZrO2 supports are detectable for most of the catalysts. An exception is Pt/ZrO2_SD, where ZrO2 seems to be crystalline when it is prepared by coprecipitation or flame spray pyrolysis, but extremely broad peaks are observed when it is prepared by spray drying. Despite a calcination temperature of 650 °C, zirconia seems to have an amorphous structure in Pt/ZrO2_SD. Due to the low Brunauer-Emmet-Teller (BET) surface area, it is not very probable that it consists of very small particles. The ceria particle sizes are very close to those of the coprecipitated samples, indicating that also the spray-dried samples are more agglomerated than the samples prepared by flame spray pyrolysis. Previous studies show the temperature at which zirconia is expected to become crystalline varies. Liang et al. [26] reported crystalline zirconia after treatment at 400 °C, while Ksapabutr et al. [27] and Rezaei et al. [28] increased the calcination temperature to 500 and 600 °C, respectively. Devassy and Halligudi [29] detected crystalline ZrO2 after calcination at 350 °C, but found also that Previous studies show the temperature at which zirconia is expected to become crystalline varies. Liang et al. [26] reported crystalline zirconia after treatment at 400 • C, while Ksapabutr et al. [27] and Rezaei et al. [28] increased the calcination temperature to 500 and 600 • C, respectively. Devassy and Halligudi [29] detected crystalline ZrO 2 after calcination at 350 • C, but found also that the temperature had to be increased to 450 • C to obtain a crystalline phase after adding the active metal. The findings of Devassy and Halligudi partly support the results in this study, since a calcination temperature of 350 • C was enough to crystallize the support in Pt/ZrO 2 _CP before the Pt impregnation, while the Pt/ZrO 2 _SD may need a higher temperature. Pt/ZrO 2 _FSP is exposed to higher temperatures in the flame and should therefore be expected to be crystalline. In Pt/CeO 2 /ZrO 2 _CP and Pt/CeO 2 /ZrO 2 _SD, zirconia is not found. The shift of the ceria peaks to higher 2θ values when adding zirconia to the support indicates that some of the Ce 4+ are substituted by Zr 4+ in the lattice [30], but also here it is possible that zirconia is present as an amorphous phase. Pt phases are only visible in the coprecipitated Pt/CeO 2 and Pt/CeO 2 /ZrO 2 . This is probably because the platinum particle size in the spray-dried and flame-synthesized catalysts are below the XRD detection limit.

Characterization
The surface area and Pt dispersion of the catalysts are listed in Table 1. The catalysts prepared by flame spray pyrolysis have the highest surface area for all compositions and highest Pt dispersion for Pt/CeO 2 and Pt/ZrO 2 . This is not surprising, since flame spray pyrolysis is known for giving high surface area materials with a good metal dispersion. Spray drying produced Pt/CeO 2 and Pt/ZrO 2 with the lowest surface area and dispersion, which is more unexpected. However, one obvious difference between FSP and SD is the lower temperature used for spray drying, which could significantly affect the catalyst properties. In addition, spray drying often results in a droplet-particle ratio of 1:1, whereas flame spray pyrolysis gives numerous nanoparticles formed from each droplet [31]. The spray-dried Pt/CeO 2 /ZrO 2 , on the other hand, obtained better platinum dispersion than the other Pt/CeO 2 /ZrO 2 catalysts, and the surface area is also quite high. Hydrogen spillover from Pt to the support and partial reduction of the support may affect the measured hydrogen consumption and hence the measured platinum dispersion. This may not affect catalysts equally, since the reducibility of ceria and zirconia may vary between different supports and different preparation methods.  Table 1 shows that the support particle sizes of the coprecipitated and the spray-dried catalysts are similar, while flame spray pyrolysis produced slightly larger support particles. This somewhat contradicts the surface area results, since a large surface area often correlates with small particles. However, this may indicate a larger extent of agglomeration of the primary crystallites in the coprecipitated and spray-dried samples. The Pt particles are generally the smallest for the catalysts prepared by flame spray pyrolysis, but this is in principle the same as the Pt dispersion results. It was not possible to calculate the Pt particle size for Pt/ZrO 2 _SD, since no hydrogen uptake was observed during the chemisorption experiment.
It is also observed that for coprecipitation and flame spray pyrolysis both the surface area and the Pt dispersion decrease with increasing zirconia content. Concerning the spray-dried catalysts, Pt/CeO 2 /ZrO 2 has the highest surface area and platinum dispersion. Results reported by others show the varying influence of the ceria content, and different preparation procedures to obtain different compositions have been used in some cases [10,[32][33][34][35][36][37][38]. The preparation method is a factor that affects the structural properties, but definite conclusions on the correlation of different preparation methods and characterization results are difficult to find.
The TPR profiles are shown in Figure 2. Comparison of the Pt/CeO 2 catalysts shows that the reduction temperature peaks of the coprecipitated and spray-dried Pt/CeO 2 are in the same range, 100-200 • C. There are also traces of reduction of Pt/CeO 2 _FSP in this temperature interval, but the main reduction lies in a much higher temperature range (350-550 • C). The amount of hydrogen consumed is highest for Pt/CeO 2 _CP and quite similar for the other two. Similar trends are observed for the Pt/CeO 2 /ZrO 2 samples, but the Pt/ZrO 2 samples are different. Pt/ZrO 2 _CP is almost not reduced at all and Pt/ZrO 2 _SD has the highest hydrogen consumption, with significantly higher reduction temperatures. To compare the coprecipitated samples, Pt/CeO 2 _CP and Pt/CeO 2 /ZrO 2 _CP are quite similar, while Pt/ZrO 2 _CP is reduced at higher temperatures (~617 • C) with substantially lower hydrogen consumption. This is also reported by others [33,39], but the most common result in the literature appears to be higher hydrogen consumption for Pt/CeO 2 /ZrO 2 relative to the other two [37,40]. The reduction temperature is higher for Pt/ZrO 2 _SD also for the spray-dried samples, but a lot more hydrogen is consumed in this case. The high hydrogen consumption is possibly explained by the reduction of nitrate residues, since Pt/ZrO 2 _SD did not crystallize during calcination. It could also be that the oxygen present in the amorphous structure is, due to weaker bonds, more easily reacted. The spray-dried samples are all reduced in two steps. The catalysts prepared by flame spray pyrolysis are quite similar to each other in terms of reduction temperature (400-450 • C) and hydrogen consumption. Pt/CeO 2 _FSP and Pt/CeO 2 /ZrO 2 _FSP show in addition minor peaks before 200 • C, which is absent in Pt/ZrO 2 . A possible explanation as to why the spray-dried and flame-synthesized catalysts have more reduction peaks than the coprecipitated catalysts could be differences in Pt distribution and hence the availability of Pt. The platinum particles might be more distributed inside the crystal lattice for the first two methods, as opposed to the last one where Pt is impregnated on the surface. The Pt particles at the surface will probably be reduced first, and then the Pt particles in the bulk. There could also be differences in the oxidation state of Pt, and the number of reduction peaks might depend on how the support is reduced. Figure 3a shows the CO conversion as a function of reaction temperature and Figure 3b shows the tendencies towards deactivation for all catalysts as obtained at 300/310 • C. Pt/CeO 2 /ZrO 2 _SD shows the highest initial CO conversion, but deactivates significantly at 300 • C. Pt/CeO 2 _CP and Pt/CeO 2 /ZrO 2 _CP are also slightly better than the other catalysts, and they are quite similar concerning both activity and initial stability. The catalysts prepared by flame spray pyrolysis have in general the lowest activity. Based on the S BET and Pt dispersion results, this was unexpected but consistent with the higher reduction temperature and a small amount of reduced Pt obtained in TPR. The stability is relatively good, however, indicating that FSP is a promising preparation method. Catalysts prepared by FSP are exposed to very high temperatures during preparation and are thereby expected to be more resistant to sintering than coprecipitated and spray-dried catalysts. This may support its good stability, since deactivation possibly occurs due to loss of metal surface area [7]. Disregarding the FSP catalysts, the ranking based on catalyst activity is closely related to the specific surface area. A correlation between high platinum dispersion and high catalytic activity when comparing the samples within the preparation methods is also present. Considering all the samples, there is no clear trend and hence the preparation method is an important factor.  Figure 2. TPR profiles for all catalysts. Reduction in 7% H2/Ar using a heating rate of 10 °C/min. Figure 3a shows the CO conversion as a function of reaction temperature and Figure 3b shows the tendencies towards deactivation for all catalysts as obtained at 300/310 °C. Pt/CeO2/ZrO2_SD shows the highest initial CO conversion, but deactivates significantly at 300 °C. Pt/CeO2_CP and Pt/CeO2/ZrO2_CP are also slightly better than the other catalysts, and they are quite similar concerning both activity and initial stability. The catalysts prepared by flame spray pyrolysis have in general the lowest activity. Based on the SBET and Pt dispersion results, this was unexpected but consistent with the higher reduction temperature and a small amount of reduced Pt obtained in TPR. The stability is relatively good, however, indicating that FSP is a promising preparation method. Catalysts prepared by FSP are exposed to very high temperatures during preparation and are thereby expected to be more resistant to sintering than coprecipitated and spray-dried catalysts. This may support its good stability, since deactivation possibly occurs due to loss of metal surface area [7]. Disregarding the FSP catalysts, the ranking based on catalyst activity is closely related to the specific surface area. A correlation between high platinum dispersion and high catalytic activity when comparing the samples within the preparation methods is also present. Considering all the samples, there is no clear trend and hence the preparation method is an important factor. The crystallite size of the support also seems to be essential. With the exception of Pt/CeO 2 _FSP, the smaller the support particles, the higher the activity. The specific surface area and Pt dispersion are in general expected to reflect the support particle size, but as discussed in Section 2.1, the Pt dispersion results are possibly affected by hydrogen spillover. By using the Weisz-Prater criterion it is verified that the reaction is not diffusion limited. Pt/ZrO 2 _SD shows the lowest CO conversion. This was expected due to the low surface area and Pt dispersion, and the high reduction temperature. According to Tabakova et al. [41], the catalytic activity may also depend on the crystallinity of the catalyst, and the fact that Pt/ZrO 2 _SD shows an amorphous phase could thereby be another contribution to the low activity. Pt/ZrO 2 is in general the less active catalyst for all preparation methods. This is related to the fact that the abovementioned properties also are quite poor for Pt/ZrO 2 _CP and Pt/ZrO 2 _FSP.

Activity Measurements
According to Tabakova et al. [41], the catalytic activity may also depend on the crystallinity of the catalyst, and the fact that Pt/ZrO2_SD shows an amorphous phase could thereby be another contribution to the low activity. Pt/ZrO2 is in general the less active catalyst for all preparation methods. This is related to the fact that the abovementioned properties also are quite poor for Pt/ZrO2_CP and Pt/ZrO2_FSP.
In coprecipitation, a total amount of 0.115 mol precursors (cerium (III) nitrate hexahydrate, zirconyl (IV) nitrate hydrate (≥99.5%, Acros Organics, Geel, Belgium) were dissolved in 450 mL of a 40% ethylene glycol solution (≥99.5%, Fluka, Steinheim; Germany) in deionized water. Then, 0.99 mol urea (≥99.5%, Merck, Darmstadt, Germany) was added under stirring. The solution was placed in an oil bath (100 • C) for 16 h and then rapidly cooled in cold water. The suspension was filtered and washed 4 times with 300 mL deionized water (45 • C), and the resulting precipitate was dried (100 • C, 16 h) and calcined in a muffle furnace for 30 min at 350 • C (heating up: 2 • C/min; cooling down: ca. 2 • C/min). The calcined supports were impregnated with Pt (tetraammineplatinum (II) nitrate (99.995%, Aldrich, Steinheim, Germany)) by incipient wetness. The impregnated samples were dried and calcined in the same way as described above. Further details on the preparation method have been published elsewhere [42,43].
For spray drying, nitrate precursors (cerium (III) nitrate hexahydrate, zirconyl (IV) nitrate hydrate (≥99.5%, Acros Organics, Geel, Belgium), tetraammineplatinum (II) nitrate (99.995% (Aldrich, Steinheim, Germany) were dissolved in deionized water to give a total concentration of 10 wt %. Using a Lab-Plant SD-06 Laboratory-Scale Spray Dryer, the nitrate solution was pumped through a jet nozzle (d = 0.5 mm) and sprayed into a heated chamber. Compressed air induces droplet formation, and an outer air stream with a variable temperature separates the droplets from the nozzle and dries the material. The product flows in a continuous spiral through a cyclone and into a collection bottle. The pump rate was~120 mL/h, the inlet temperature 165 • C, and the air flow 4.3 m/s. The samples were calcined in a muffle furnace for 30 min at 350 • C (heating up: 2 • C/min; cooling down: ca. 2 • C/min). The Pt/ZrO 2 was calcined at 650 • C for 30 min (heating up: 2 • C/min; cooling down: ca. 2 • C/min). The principle of spray drying together with a picture of the laboratory unit is shown on Figures S4 and  S5 in Supplementary.
Flame spray pyrolysis was performed in a laboratory-scale flame spray unit. Appropriate amounts of precursors (cerium (III) 2-ethylhexanoate (49% in 2-ethylhexanoic acid), platinum (II) acetylacetonate (97%), and zirconium (IV) n-propoxide (70% w/w in n-propanol, (Alfa Aesar, Karlsruhe, Germany) were dissolved in toluene (>99.7%, Riedel-de Haën, Seelze, Germany) and the solution was sprayed into a hydrogen flame at 100 mL/h by a syringe pump (IVAC P700, Cardinal Health, 1180 Rolle, Switzerland). Hydrogen was supplied at 3.1 L/min. The liquid was dispersed by 9 L/min oxygen. The product particles were collected by a glass fiber filter (Advantec GC50, Cole-Parmer, Vernon Hills, IL, USA) with the aid of a vacuum pump (PIAB M50, Hongham, MA 02043, USA). Further details on the setup are reported elsewhere [44,45] and a picture of the setup and the principle of particle formation is shown in Figures S2 and S3 in Supplementary.

Characterization
X-ray diffraction analysis of the catalysts was performed on a Bruker AXS D8 Focus diffractometer using CuK α radiation (λ = 1.54Å) (76187 Karlsruhe, Germany). The XRD patterns were compared with standards in a database to identify the phases and the crystallite sizes were calculated by using the Scherrer equation [46].
The specific surface area (S BET ) of the catalysts was determined by N 2 adsorption at −196 • C using a COULTER SA 3100 instrument. The specific surface area was calculated by the Brunauer-Emmet-Teller (BET) equation in the relative pressure interval of 0.05-0.20. Prior to the measurements, the samples were outgassed at 150 • C for 2 h.
The Pt dispersion was determined by H 2 chemisorption, using a Micromeritics ASAP 2010 (Norcross, GA, USA) instrument. An amount of 0.2 g catalyst was loaded into a quartz reactor and held in place by quartz wool. Prior to the chemisorption measurement, the samples were reduced in H 2 flow while heating at 10 • C/min up to 300 • C. The samples were then evacuated in He flow at 300 • C for 1 h before cooling down to 35 • C, at which the analysis was performed. In order to distinguish between chemisorbed and physisorbed hydrogen, two adsorption isotherms were measured.
Temperature-programmed reduction (TPR) experiments were performed using a Quantachrome CHEMBET 3000 (Boynton Beach, FL, USA) instrument, where the effluent gas is analyzed by a thermal conductivity detector (TCD). An amount of 0.15 g catalyst was loaded into a quartz reactor and quartz wool was used to keep it in place. The samples were outgassed in Ar flow at 300 • C for 1 h prior to the measurements, and after cooling down to ambient temperature, the samples were heated in 7% H 2 /Ar (60 mL/min) at 10 • C/min up to 800 • C.

Activity Measurements
WGS activity measurements were carried out in a fixed-bed reactor using 0.10 g catalyst diluted with SiC (50/50 vol%). The catalysts were prereduced at 220 • C for three hours in 20% hydrogen in nitrogen (5% H 2 in N 2 for the FSP catalysts) at a total gas flow of 100 N mL/min. A 1:1:1 mixture of N 2 , CO, and steam was fed at 150 N mL/min total flow during reaction experiments, and the pressure was close to ambient. Measurements were performed in the temperature interval 190-300 • C, and the initial catalyst deactivation was studied at 300 • C (310 • C for the FSP catalysts) by doing repeated measurements for 15-20 h. The dry product gas (H 2 , CO, CO 2 , N 2 ) was analyzed using an Agilent 3000 Micro GC (Santa Clara, CA, USA). A flowsheet of the experimental setup is given as Figure S1 in Supplementary.

Conclusions
It is shown that the preparation method is of importance concerning the properties of the Pt-based catalysts, and thereby the water-gas shift activity. The spray-dried Pt/CeO 2 /ZrO 2 catalyst shows the highest initial activity, but deactivates rapidly at 300 • C, and levels out at similar activity as the coprecipitated Pt/CeO 2 and Pt/CeO 2 /ZrO 2 within a few hours. The more stable Pt/CeO 2 _CP could have become the most active if the stability were tested for a longer period. Flame spray pyrolysis appears to be a promising preparation method concerning the stability of the catalysts. However, the initial activity is rather poor, although this could be attributed to incomplete reduction. High activity is related to high Pt dispersion, low reduction temperature, and small support particles. The Pt dispersion is an important factor, but the measurements are complicated due to spillover when Pt is supported on reducible supports such as ceria and zirconia. A comparison of the dispersion results and the activity measurements indicate that the reducibility of the support could vary between the different preparation methods. Additionally, the support particle size is much affected by the preparation method.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4344/10/10/1132/s1, Figure S1: Schematic presentation of the experimental setup used for water-gas activity testing. Figure S2: The principle of particle formation in a flame. Figure S3: The setup for flame spray pyrolysis employed at DTU. Figure  S4: The principle of spray drying. Figure  Funding: Financial support was provided by the Research Council of Norway through the projects 153967/420-Scientific Design and Preparation of New Catalysts and Supports and 158516/S10 (NANOMAT) and through a travel grant from the Strategic Area Materials at NTNU.

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