Characterization of Palladium (Pd) on Alumina Catalysts Prepared

Jun 21, 2008 - Jaehoon Kim, M. Jason Kelly, H. Henry Lamb, George W. Roberts* and Douglas J. Kiserow. Department of Chemical and Biomolecular ...
0 downloads 0 Views 766KB Size
Download PDF
10446

J. Phys. Chem. C 2008, 112, 10446–10452

Characterization of Palladium (Pd) on Alumina Catalysts Prepared Using Liquid Carbon Dioxide Jaehoon Kim,†,‡ M. Jason Kelly,† H. Henry Lamb,† George W. Roberts,*,† and Douglas J. Kiserow†,§ Department of Chemical and Biomolecular Engineering, North Carolina State UniVersity, Raleigh, North Carolina 27695-7905, U.S. Army Research Office, Chemical Science DiVision, PO Box 12211, Research Triangle Park, North Carolina 27709-2211, and Energy and EnVironment Research DiVision, Korea Institute of Science and Technology (KIST), 39-1 Hawolgok-dong, Seongbuk-gu, Seoul, 136-791, Korea ReceiVed: December 6, 2007; ReVised Manuscript ReceiVed: April 25, 2008

Palladium (II) hexafluoroacetylacetonate (Pd(hfac)2) dissolved in liquid carbon dioxide (L-CO2) was used to deposit Pd nanoparticles onto low-surface-area R-alumina (13 m2/g) and high-surface-area γ-alumina (207 m2/g). These nanoparticles were prepared by contacting Pd(hfac)2 dissolved in L-CO2 with the alumina at 6.9 MPa and 28.5 °C and then slowly venting gaseous CO2 until L-CO2 was completely evaporated. After depressurization to remove the CO2 and unabsorbed Pd(hfac)2, the impregnated Pd(hfac)2 was reduced in hydrogen at a relatively low temperature of 75 °C. The adsorption isotherm of Pd(hfac)2 on γ-alumina suggests a weak interaction between the organometallic compound and the support. The average Pd particle size on the low-surface-area R-alumina, measured by scanning electron microscopy, increased from 13.1 ( 3.5 to 59.9 ( 11.3 nm, and the metal dispersion, measured by pulsed CO chemisorption, decreased from 11 to 3%, as the Pd loading on the alumina was increased from 0.15 to 1.54 wt %. With the high-surface-area γ-alumina, Pd particle size, measured by scanning transmission electron microscopy, increased from 3.1 ( 1.9 to 7.0 ( 5.9 nm, and the metal dispersion decreased from 56 to 5%, as Pd loading was increased from 0.58 to 3.94 wt %. 1. Introduction Supported metal catalysts, composed of nanosized metal particles attached to a high-surface-area support, have been extensively used in many industrial applications, including petroleum refining, chemical synthesis, environmental cleanup, and fuel cells. Typical metals include platinum (Pt), palladium (Pd), rhodium (Rh), ruthenium (Ru), and nickel (Ni), alone or in combination. The most common supports are alumina, silica, and various kinds of high-surface-area carbon. The chemical and physical properties of supported metal catalysts, and their performance in catalytic reactions, depends on the method of preparation.1,2 Preparation of supported metal catalysts using supercritical carbon dioxide (scCO2) has received considerable attention recently.3–15 The unusual physical properties of scCO2, including zero surface tension, low viscosity, and tunable density, make it an interesting alternative medium for metal deposition, with the potential to overcome many of the disadvantages associated with traditional, aqueous catalyst preparation techniques. These disadvantages include slow diffusion of the metal salt into the pores of the support due to the high viscosity of water, changes in the structure of the support during drying due to the high surface tension of water, poor metal dispersion, and redistribution or inhomogeneous distribution of the active compound.1,16–18 Catalysts with very small metal particle sizes and high activity have been prepared from scCO2. For example, Saquing et al. * Corresponding author e-mail: [email protected]; phone: 1-919515-7328. † North Carolina State University. ‡ Korea Institute of Science and Technology. § U.S. Army Research Office.

reported that 1.6-3.1 nm diameter Pt particles were synthesized on a high-surface-area carbon aerogel (629-741 m2/g) by impregnating dimethyl(1,5-cyclooctadiene) platinum(II) (PtMe2(cod)) using scCO2 at 27.6 MPa and 80 °C and subsequently reducing the organometallic compound to metallic Pt by heat treatment between 300-800 °C.12 High Pt dispersions, in the range of 37-70% as measured by hydrogen chemisorption, were obtained. Zhang et al. showed that uniformly distributed Pt particles, with diameters in the range of 1.2-6.4 nm, could be prepared on various supports, including carbon aerogel (690 m2/g), carbon black (290 m2/g), silica aerogel (800 m2/g), silica (120 m2/g), and γ-alumina (260 m2/g) via impregnation from scCO2.4 Zhang et al. also reported that homogeneously distributed Ru particles with diameters in the range of 1.7-3.8 nm were produced on carbon aerogel (629-741 m2/g) by impregnating bis(2,2,6,6,-tetramethyl-3,5-heptanedionato)(1,5cyclooctadiene) ruthenium(III) (Ru(cod)(TMHD)2) from scCO2 at 27.6 MPa and 80 °C and then reducing to metallic Ru at 300-1000 °C.3 Ye et al. showed that nanoparticles of Pd (5-10 nm), Rh (3-5 nm), and Ru (∼1 nm) were deposited on multiwall carbon nanotubes (MWCNT) by dissolving metal-β-diketonate precursors [palladium(II) hexafluoroacetylacetonate (Pd(hfac)2), rhodium(II) acetylacetonate (Rh(acac)2), ruthenium(III) acetylacetonate (Ru(acac)3)] in scCO2 and then reducing them at 80-150 °C using H2.11 Lin et al. reported that highly active Pt/MWCNT electrocatalysts with Pt particle diameters in the 5-10 nm range were prepared by depositing Pt(acac)2 from a scCO2/methanol mixture at 8 MPa and 200 °C, followed by reduction to metallic Pt at 200 °C using H2.7 We reported the first use of liquid carbon dioxide (L-CO2) in the preparation of supported metal catalysts.19 As in the case of scCO2, L-CO2 also has attractive physical properties for supported catalyst preparation. In the previous communication,

10.1021/jp711495n CCC: $40.75  2008 American Chemical Society Published on Web 06/21/2008

Characterization of Pd on Alumina Catalysts

J. Phys. Chem. C, Vol. 112, No. 28, 2008 10447

TABLE 1: Pore Properties of Alumina Supports surface area (m2/g)

average pore diameter (nm)

pore volume (cm3/g)

support

BET

BJH

BET

BJH

BET

BJH

R-alumina γ-alumina

13.3 207.0

13.0 250.3

32.7 11.6

32.8 9.3

0.11 0.60

0.11 0.58

a

microporea volume (cm3/g) 5.8 × 10-4 6.2 × 10-4

micropore area (m2/g) 1.3 4.4

Micropores are defined as pores with diameters less than 0.60 nm.

we showed that Pd nanoparticles were deposited on low-surfacearea R-alumina (13.3 m2/g) and high-surface-area γ-alumina (207 m2/g) by impregnating Pd(hfac)2 dissolved in L-CO2 at mild conditions (6.9 MPa and 28.5 °C), followed by reduction to metallic Pd with H2 at a low temperature of 75 °C.19 In this paper, we will describe the effects of Pd(hfac)2 concentration in L-CO2 on Pd loadings, particle sizes, and dispersions, as determined by inductively coupled plasma-emission spectrometry (ICP-ES), scanning electron microscopy (SEM), scanning transmission electron microscopy (STEM), and pulsed CO chemisorption. The adsorption equilibrium of Pd(hfac)2 on the supports and the partitioning of Pd(hfac)2 between the supports and fluid phase also will be discussed. 2. Experimental Section Materials. Pd(II) hexafluoroacetylacetonate (Pd(hfac)2) was purchased from Sigma-Aldrich (St. Louis, MO) and used as received. Low-surface-area R-alumina and high-surface-area γ-alumina were purchased from Alfa-Aesar (Ward Hill, MA). Table 1 shows some properties of these aluminas. The aluminas were in the form of cylindrical particles, with a 3.25 mm diameter and a 5.5 mm height. The CO2 source was Coleman (purity of 99.99%), and the H2 source was ultra high purity (purity of 99.999%) obtained from National Welders (Charlotte, NC). Apparatus. A schematic of the system used for catalyst preparation is shown in Figure 1. The apparatus consists of a high-pressure vessel (VS), a pneumatic flow control valve (V2), an ISCO pump to feed CO2 (P1), isolation valves (V1), a computer feedback control system (F), a temperature control unit for the high-pressure vessel (TC), a heating tape for the high-pressure vessel (HT), and a magnetic stirrer (M). The highpressure vessel (VS) is cylindrical in shape with an inside diameter of 1.75 cm and an inside height of 7.61 cm, giving a volume of 18.3 mL. The vessel is equipped with a sapphire window and a window screw cap for visual inspection of the Pd(hfac)2 in L-CO2 solution. A stainless steel support was placed at the bottom of the vessel to hold a metal mesh and to provide space for a magnetic stirring bar. The high-pressure vessel and the screw caps were manufactured by the North Carolina State University Instrumentation Shop. The sapphire windows were purchased from Rubicon Technology (Franklin Park, IL). The temperature of the high-pressure vessel was controlled by housing the vessel, plus the inlet and outlet tubing, in a model 4EM mechanical convection incubator (TC) manufactured by Precision Scientific (Winchester, VA). The inlet H2 temperature was controlled by flowing H2 through a 15 m long tube connected to the top port of the vessel and located inside the convection incubator. Temperature changes were monitored using type K thermocouples (T) from Omega Engineering, Inc. (Stamford, CT). The temperature of the high-pressure vessel was maintained within (0.1 °C during impregnation. The temperature during reduction was controlled by a heating tape wrapped around the high-pressure vessel and connected to a temperature controller [Digi-Sense Advanced Temperature Controller; Cole-Parmer (Vernon Hills, IL)].

Figure 1. Schematic of catalyst preparation using the L-CO2 apparatus. VS, high-pressure vessel; V1, isolation valves; V2, pneumatic flow control valve; S1, CO2 source tank; P1, ISCO syringe pump at the CO2 source; S2, H2 source tank; F, feedback control loop; P, pressure transducer; T, thermocouple; TC, high pressure vessel temperature control unit; HT, heating tape; M, magnetic stirrer.

The vent rate of CO2 was controlled through a computerregulated proportional integral derivative (PID) feedback control loop (F) by combining a type 766 research control valve (model 1001GCN36SVCPP11ST) (V2) from Badger Meter, Inc. (Milwaukee, WI) with a high-accuracy pressure transducer ((0.1 psi) (model PX01C0-7.5KG5T) (P) from Omega Engineering, Inc. Liquid CO2 was supplied using a Model 500D syringe pump from ISCO, Inc. (Lincoln, NE) (P1). Adsorption and Catalyst Preparation Procedure. The procedure for synthesizing supported Pd catalysts using L-CO2 impregnation was described in detail in the previous communication.19 For an absorption experiment, the supports first were pretreated by placing the support particles in a metal mesh and heating at 300 °C for a minimum of 3 h in air to remove absorbed water and any volatile substances. Known amounts of Pd(hfac)2 and the pretreated support were then placed in the high-pressure vessel. The vessel was closed and L-CO2 was slowly fed into it. After the desired amount of CO2 was introduced and the desired pressure (6.9 MPa) and temperature (28.5 °C) were reached, the system was isolated for 12-24 h to ensure that Pd(hfac)2 was completely dissolved and that

10448 J. Phys. Chem. C, Vol. 112, No. 28, 2008 adsorption equilibrium was reached. The homogeneity of the solution of Pd(hfac)2 in L-CO2 was visually inspected using the view port in the high-pressure vessel. In early experiments, the amount of the adsorbed Pd(hfac)2 after 12 h was found to be the same as after 24 h. Thus, the subsequent adsorption period was set to 12 h. At the end of this period, CO2 was vented at a constant rate of 0.3 mL/min. After the pressure of the vessel reached ambient, the impregnated support was taken from the vessel and was weighed using a high-accuracy analytical balance ((0.1 mg) [Explore Pro, model: EP214DC; Ohaus Corporation (Pine Brook, NJ)]. The amount of Pd(hfac)2 deposited was determined from the change of mass before and after adsorption. Characterization. Surface area, pore size, and pore volume of the two alumina supports were measured using a TriStar 3000 analyzer (Micromeritics, GA). The results are given in Table 1. Prior to the measurements, the alumina samples were outgassed at 200 °C for 4 h. The surface area, pore size, and pore volume were calculated using the BET (Brunauer, Emmet, and Teller) and BJH (Barrett, Joyner, and Halenda) methods. Micropore volume and micropore area were obtained using the t-plot method. The metal loadings on the supports were measured using ICPES. A Perkin-Elmer Optima 2000DV ICP-ES was operated at an RF power of 1500 W, with an argon flow rate of 18 mL/ min and a nebulizer flow rate of 0.62 L/min. Each sample was placed in a Teflon bottle and was dissolved in a solution of 2 parts of HNO3 and 6 parts of HCl by refluxing the samples on a hot plate with a sand bath for 30 min. Standards were made from a NIST-traceable 1000 µg Pd/L stock solution. Analyses were performed at wavelengths of 340.46, 324.27, and 248.89 nm. The standard curves for each of the analytical wavelengths were linear. Each wavelength yielded the same concentration of element in the acid digestate. The morphology of the catalysts was characterized using scanning electron microscopy (SEM, Hitachi 4800, Japan) and scanning transmission electron microscopy (STEM, Hitachi HD2000, Japan). Samples were prepared by crushing the catalyst with a mortar and pestle and then dispersing the crushed catalyst in ethanol. Three drops of the dispersed samples were deposited on a copper mesh grid coated with a carbon film and a Formvar film manufactured by Ted Pella, Inc. (Redding, CA). After complete evaporation of ethanol, the samples were examined using STEM. The size of Pd particles was estimated by analyzing the SEM and STEM images with Canvas by ACD systems (Miami, Florida). The average diameter of particles was defined as the arithmetic average diameter of each hemispherically shaped particle in the images. The average diameter of irregularly shaped particles was estimated by assuming that their geometry was hemispherical. The dispersion of Pd on the catalysts was measured using a custom-built pulsed carbon monoxide (CO) chemisorption apparatus. The apparatus consists of a glass reactor, a furnace, a temperature controller, mass flow rate controllers, a vacuum pump, and a residual gas analyzer. Prior to the chemisorption experiment, 0.5 g of the catalyst was placed in the glass reactor and the sample was purged overnight at room temperature with 5 sccm He. The sample was then purged with 5% H2 in He at 20 sccm until the H2 signal in the residual gas analyzer was stable. Using the furnace, the temperature of the reactor then was ramped to 250 at 5 °C/min with the same flow rate of 5% H2 in He. The temperature was held at 250 °C for 30 min to reduce any oxidized Pd. The sample was then purged using 40 sccm He at 250 °C for 30 min to remove H2. The purge gas was switched to 20 sccm He and the reactor was cooled to 25

Kim et al.

Figure 2. Adsorption isotherms for Pd(hfac)2 (a) on low-surface-area (13 m2/g) R-alumina and (b) on high-surface-area (207 m2/g) γ-alumina. The absorption conditions were 6.9 MPa and 28.5 °C. The dashed lines show the calculated values of q1.

°C. The system was stabilized at 25 °C with a flow of He at 20 sccm for 30 min before injecting CO. For the CO chemisorption experiments, 1 mL of 5% CO in He was pulsed through the reactor with 20 sccm of He carrier gas. The He carrier gas was allowed to flow through a 1 mL sample loop for 10 s to inject CO in the sample loop into the reactor. The sample loop then was refilled with a 4 sccm flow of 5% CO in He gas for 3 min. The number of pulses was in the range of 10-20, depending on the sample. Carbon monoxide concentrations were recorded by a residual gas analyzer (Model RGA100) using software from Stanford Research Systems, Inc. (Sunnyvale, CA). A typical pulsed CO signal is shown in Figure S1 in the Supporting Information. Helium and 5% H2 in He were purified using All-Pure gas purifiers manufactured by Alltech Associates, Inc. (Columbia, MD) before introduction into the reactor. Pd dispersion was calculated assuming the stoichiometry for CO on Pd is 0.5. Results and Discussion Adsorption of Pd(hfac)2 on Alumina Supports. Figure 2 shows adsorption isotherms of Pd(hfac)2 on the low-surfacearea R-alumina and on the high-surface-area γ-alumina. In both cases, the concentration of Pd(hfac)2 adsorbed on the support, expressed as mmol/g of support, increases monotonically as the concentration of Pd(hfac)2 in the solution (expressed as mmol/g L-CO2) increases. The dashed lines in Figure 2 show the calculated amounts of Pd(hfac)2 that would be adsorbed if the

Characterization of Pd on Alumina Catalysts

J. Phys. Chem. C, Vol. 112, No. 28, 2008 10449

TABLE 2: Comparison of Langmuir Adsorption Isotherm Parameters

adsorbate

pressure (MPa)

temperature (C)

K (g CO2/mmol adsorbate)

q∞ (mmol adsorbate/g adsorbent)

Pd(hfac)2 toluene DDT m-xylene c furfural ethyl acetate naphthalene d ketoprofen e PtMe2(cod) f Ru(cod)(tmhd)2

6.9 11.2 10.4 15 29 20 18 18 27.6 27.6

28.5 45 40 50 40 40 40 40 80 80

25.7 952 3915 1922 3019 230 16 925 268 2313

0.48 1.1 2.17 3.89 3.44 13.25 0.82 1.86 3.61 0.31

adsorbent 2

γ-alumina (220 m /g) activated carbon activated carbon (900-1100 m2/g) activated carbon (1300 m2/g) activated carbon (963 m2/g) activated carbon (963 m2/g) silica aerogel silica aerogel carbon aerogel (741.2 m2/g) carbon aerogel (741.2 m2/g) e

b

q1 (mmol adsorbate/g adsorbent)

q∞/q1

ref

0.44 NAa 3.52 8.17 9.03 21.19 NA NA 6.27 1.57

1.09 NA 0.62 0.48 0.38 0.63 NA NA 0.58 0.20

this study 1 2 3 4 4 5 5 6, 7 8

a Not available. b 1,1-Bis(4-chlorophenyl)-2,2,2,-trichloroethane. c Furan-2-carboxaldehyde. d (RS)-2-(3-benzoylphenyl) propionic acid. Dimethyl(1,5-cyclooctadiene) platinum(II). f Bis(2,2,6,6,-tetramethyl-3,5-heptanedionato)(1,5-cyclooctadiene) ruthenium(III).

support surfaces were completely covered by a monolayer of Pd(hfac)2. This amount, which is designated q1, was calculated from eq 1,

q1 )

SA πr 2NA

(1)

where NA is Avogadro’s number, SA is the specific surface area of the support (area/weight), and r is the radius of a Pd(hfac)2 molecule. The radius of Pd(hfac)2 was estimated to be 0.5 nm using a space filling model for the equilibrium structure of the compound (Chem 3D Ultra, CambridgeSoft, Cambridge, MA). With R-alumina, the uptake of Pd(hfac)2 continued to increase with solution concentration, even after complete monolayer coverage was reached (Figure 2a). This indicates that multilayer adsorption of Pd(hfac)2 took place on R-alumina. The slope of the isotherm prior to reaching complete monolayer coverage is larger than the slope after a complete monolayer had been adsorbed. The adsorption behavior of Pd(hfac)2 on the γ-alumina is much different. As shown in Figure 2b, the concentration adsorbed reached a constant value at high concentrations of Pd(hfac)2. Moreover, this constant value corresponded very closely to the calculated value for complete monolayer coverage (q1). The kind of adsorption behavior shown in Figure 2b can be characterized using the Langmuir isotherm (eq 2),

q)

q∞KC 1 + KC

(2)

where q is the concentration of absorbate (e.g., Pd(hfac)2) absorbed per unit mass of absorbent (e.g., alumina), K is the equilibrium constant for adsorption, C is the concentration of adsorbate in solution, and q∞ is the maximum concentration, that is, the value of q when C is very large. The equilibrium constant K describes partitioning of absorbate between the support and the fluid phase. A high value of K indicates that the adsorbate has a high affinity for the surface of the support. For monolayer adsorption, the value of q∞ cannot exceed q1, but it can be less that q1 if the density of sites on the support with which the adsorbate can complex is low or if portions of the support are not physically accessible to the adsorbate. Table 2 lists the Langmuir model parameters for adsorption of Pd(hfac)2 from L-CO2 onto γ-alumina and compares these parameters with those for other hydrocarbons and organometallic compounds adsorbed onto various supports from scCO2. The maximum loading (q∞) for select hydrocarbons—toluene, DDT, m-xylene, furfural, and ethyl acetate—from scCO2 on activated carbon is very high.20–23 This is probably due primarily to the high surface area of activated carbon (∼1000 m2/g).

The maximum loading of naphthalene on hydrophilic silica aerogel was significantly smaller than that of ketoprofen on the same support.24 This suggests that the hydrophilic ketoprofen may be able to complex with more sites on the hydrophilic silica aerogel surface than the hydrophobic naphthalene. When PtMe2(cod), a hydrocarbon-based organometallic compound, was adsorbed onto a carbon aerogel, the maximum loading was comparable to those of the hydrocarbons on the activated carbons.6,12 However, when Ru(cod)(tmhd)2 was absorbed onto the same carbon aerogel, the maximum loading was an order of magnitude smaller than that of PtMe2(cod).3 The low maximum loading of Ru(cod)(tmhd)2 was attributed to the larger size of this molecule (∼1 nm) compared to PtMe2(cod) (∼0.5 nm). Because many of the pores in the carbon aerogel were too small to admit Ru(cod)(tmhd)2, only a fraction of the carbon aerogel surface was accessible to this compound. This is reflected in the ratio q∞/q1, which was estimated to be 58% for PtMe2(cod) and only 19% for Ru(cod)(tmhd)2.3 The maximum concentration of Pd(hfac)2 on γ-alumina is small compared with q∞ for the hydrocarbons on activated carbon and for PtMe2(cod) on the carbon aerogel. This results from the higher surface area per unit weight of the carbon supports compared to γ-alumina. However, the ratio q∞/q1 for Pd(hfac)2 on γ-alumina was 1.09, which indicates that the surface of the γ-alumina can be completely covered with adsorbed Pd(hfac)2 and suggests that the entire surface of the γ-alumina was accessible to Pd(hfac)2. In fact, Table 1 shows that only about 2% of the total area of the γ-alumina is in pores with diameters of 0.6 nm or less. As shown in Figure 2b, a concentration of about 0.5 mmol Pd(hfac)2/g L-CO2 was required to obtain complete coverage of Pd(hfac)2 on the γ-alumina surface. This is reflected in the low value of the equilibrium constant K for this system, as shown in Table 2. Values of K for the other systems in this table are 1-2 orders of magnitude higher. Pd Catalysts on r-Alumina. Figure 3 shows representative SEM images of Pd particles on R-alumina at two different metal loadings. At 0.24 wt % Pd, uniformly distributed particles with an average diameter of 13.1 ( 3.5 nm were deposited on the support (Figure 3a). At a higher metal loading of 0.65 wt %, the average particle diameter was 36.1 ( 8.6 nm (Figure 3b). Figure 4 shows the dependence of metal loading, Pd dispersion, and Pd particle size on the concentration of Pd(hfac)2 in L-CO2. Clearly, metal loading and particle size can be controlled by adjusting the Pd(hfac)2 concentration. As this concentration increased from 0.6 to 10.5 wt %, the metal loading increased from 0.15 to 1.54 wt %, and the Pd dispersion decreased from about 11 to about 2.5% (Figure 4a). Figure 4b shows a

10450 J. Phys. Chem. C, Vol. 112, No. 28, 2008

Kim et al.

Figure 3. SEM images of Pd deposited on R-alumina at metal loadings of (a) 0.24 wt % and (b) 0.65 wt %.

comparison of the average particle size estimated from the SEM images and from the CO chemisorption measurements. The particle diameters (d) were calculated from the dispersions using eq 3,25

d)

112 D

(3)

where D is the dispersion. This calculation is based on the assumption that the particle shape is hemispherical. The particle sizes from SEM agree reasonably well with those from dispersion (CO chemisorption) in the low-to-medium Pd(hfac)2 concentration range (0.6-5.9 wt %). However, at the higher concentrations (8.2-10.5 wt %), the particle sizes from SEM are larger than those from CO chemisorption. This is probably because the actual morphology of the Pd particles is not hemispherical and because there are small Pd particles that do not appear in the SEM images. As shown in Figure S2a of the Supporting Information, particles with diameters in the range of about 5 nm are observed in the STEM images of 1.15 wt % Pd on R-alumina prepared at 8.2 wt % Pd(hfac)2 concentration. These smaller Pd particles, not observed in the SEM images, provide additional CO sorption sites. In addition, the shape of the Pd particles is star-like (Figure S2b in the Supporting Information) and the larger Pd particles are comprised of smaller Pd crystallites, resulting in a porous structure (S2c in the Supporting Information). Therefore, CO can access a larger area of Pd relative to particles with a nonporous, hemispherical shape. As a result, the value of the dispersion is higher, and the particle size from CO chemisorption is smaller compared with the particle size observed from SEM.

Figure 4. The effect of Pd(hfac)2 concentration in L-CO2: (a) dependence of Pd metal loadings and metal dispersion on Pd(hfac)2 concentration; (b) dependence of the Pd particle size on the Pd(hfac)2 concentration. The support is R-alumina.

The particle size measured via SEM increased from 14.5 ( 4.1 to 59.9 ( 11.3 nm as the Pd(hfac)2 concentration increased from 0.6 to 10.5 wt %. The increase in particle size results from the increased amount of Pd(hfac)2 that adsorbs on R-alumina as the Pd(hfac)2 concentration in L-CO2 is increased. Multilayer adsorption of Pd(hfac)2 occurs above about 1.2 wt % Pd(hfac)2 in L-CO2, so that the highest four concentrations in Figure 4 are in the range of multilayer adsorption. Because the interaction between Pd(hfac)2 and alumina is weak, the surface mobility of absorbed Pd(hfac)2 should be high, and the growth of Pd particles may accelerate when a large number of neighboring Pd(hfac)2 molecules are present near the initial Pd nuclei. Pd Catalysts on γ-Alumina. Figure 5 shows STEM images and particle size distributions of Pd particles deposited on γ-alumina. At the lowest metal loading of 0.58 wt %, most Pd particle diameters are in the range of 1-4 nm (Figure 5a). When the metal loading was increased to 2.45 wt %, the Pd particle diameter increased and the size distribution broadened (Figure 5b). At the highest metal loading of 3.94 wt %, the distribution of particle diameters was notably broader compared to those at the lower metal loadings; Pd particles ranging from 1 to 24 nm in diameter were observed (Figure 5c). The formation of larger Pd particles and the broader particle size distribution at higher Pd(hfac)2 concentrations probably results from the weak interaction between Pd(hfac)2 and the γ-alumina support, as previously discussed. The weak interactions allow higher surface mobility of Pd(hfac)2 molecules, which increases the possibility for Pd nuclei to grow via incorporation of Pd(hfac)2 molecules.

Characterization of Pd on Alumina Catalysts

J. Phys. Chem. C, Vol. 112, No. 28, 2008 10451

Figure 5. STEM images and particle size distributions of Pd deposited on γ-alumina at metal loadings of (a) 0.58, (b) 2.45, and (c) 3.94 wt %.

The dependence of metal loading, dispersion, and particle size on Pd(hfac)2 concentration in L-CO2 is shown in Figure 6 for the γ-alumina support. As with Pd on R-alumina, the Pd loading on γ-alumina can be controlled by adjusting the Pd(hfac)2 concentration; the metal loading increased from 0.58 to 3.94 wt % as the concentration increased from 1.2 to 27.8 wt % (Figure 6a). At the lower loadings of 0.58-1.77 wt %, high Pd dispersions of 51-56% were obtained. However, Pd dispersion decreased from 35 to 5% when the metal loading was increased from 2.01 to 3.94 wt %. The particle sizes of Pd on γ-alumina estimated from the STEM images are compared with those from the chemisorption measurements in Figure 6b. The Pd particle size from STEM increased from 3.2 ( 1.9 to

7.0 ( 5.9 nm when the Pd(hfac)2 concentration increased from 1.2 to 27.8 wt %. The particle sizes from STEM agree well with those from dispersion in the lower metal loading regime. This suggests that the use of a stoichiometric factor of 0.5 for CO adsorption on Pd and the hemispherical particle shape assumption of eq 3 are reasonably valid. The good agreement also indicates that most of the Pd surface is accessible to CO. In the higher-metal-loading regime, however, the particle size from STEM (7.0 ( 5.9 nm) is significantly smaller than that from dispersion (23 nm). This large discrepancy may be due to underestimation of the particle size when scanning a small area of the 3.94 wt % Pd on γ-alumina. A significant number of Pd particles with 20-30 nm diameters were observed when

10452 J. Phys. Chem. C, Vol. 112, No. 28, 2008

Kim et al. R-alumina increased from 14.5 ( 4.1 to 59.9 ( 11.3 nm and Pd loading increased from 0.15 to 1.54 wt % when the Pd(hfac)2 concentration increased from 0.6 to 10.5%. The discrepancy between particle sizes from SEM and those from pulsed CO chemisorption at high metal loadings of 1.15-1.54 wt % is probably the result of smaller Pd particles, plus nonhemispherical and porous Pd particles. Pd particle size on γ-alumina increased from 3.2 ( 1.9 to ∼23 nm and Pd loading increased from 0.58 to 3.94 wt % when the Pd(hfac)2 concentration increased from 1.2 to 27.8 wt %. At low metal loadings of 0.58-1.77 wt %, high Pd dispersions of 51-56% were obtained. Acknowledgment. This research was performed while Dr. Jaehoon Kim held a National Research Council Research Associateship Award with the U.S. Army Research Office. The authors acknowledge the U.S. Army Research Office, Agreement No. DAAG55-98-D-0003 for financial support and the National Science Foundation (NSF) Science and Technology Center (STC) Program, Agreement No. CHE-9876674, for additional support. The authors would also like to thank Dr. Wayne P. Robarge in the Department of Soil Science at the North Carolina State University for ICP-ES measurements. Supporting Information Available: Typical CO pulse chemisorption data; STEM images of 1.15 wt % Pd on R-alumina; SEM images of 3.94 wt % Pd on γ-alumina and of 0.58 wt % Pd on γ-alumina. This material is available free of charge via the Internet at http://pubs.acs.org.

Figure 6. The effect of Pd(hfac)2 concentration in L-CO2: (a) dependence of metal loading and metal dispersion on Pd(hfac)2 concentration; (b) dependence of the Pd particle size on Pd(hfac)2 concentration. The support is γ-alumina.

scanning a larger area of the 3.94 wt % sample (Figure S3a in the Supporting Information). Large Pd particles were not observed when scanning a large area of the lower-metal-loading samples (Figure S3b in the Supporting Information). Details on Pd crystallite morphology investigated using extended X-ray absorption fine structure (EXAFS) spectroscopy and diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy will be the subject of a future publication. It was not an objective of this research to optimize the procedure for catalyst synthesis. With the present techniques, the amount of unadsorbed Pd(hfac)2 ranged from 14 to 76%, depending of the initial Pd(hfac)2 concentration, when R-alumina was used. With γ-Al2O3, the amount of unadsorbed Pd(hfac)2 ranged from 8 to 54%, again depending on the initial Pd(hfac)2 concentration. Most of the undeposited Pd(hfac)2 remained at the bottom of the high-pressure vessel after venting of CO2. It should be possible to recover and recycle this material. Conclusions Supported Pd catalysts were prepared on low-surface-area R-alumina and high-surface-area γ-alumina by impregnating Pd(hfac)2 from L-CO2 at mild conditions, followed by reduction at a relatively low temperature of 75 °C. The adsorption isotherm of Pd(hfac)2 on R-alumina exhibits multilayer type behavior, whereas the adsorption isotherm on γ-alumina exhibits monolayer, Langmuirian behavior. The values of the parameters in the Langmuir adsorption isotherm suggest that Pd(hfac)2 has low affinity for the surface of the γ-alumina. The Pd particle sizes and metal loadings can be controlled by adjusting the Pd(hfac)2 solution concentration in L-CO2. Pd particle size on

References and Notes (1) Schwarz, J. A.; Contescu, C.; Contescu, A. Chem. ReV. 1995, 95, 477. (2) Farrauto, R. J.; Bartholomew, C. H., Fundamentals of Industrial Catalytic Processes, 2nd Ed.; John Wiley & Sons: Hoboken, New Jersey, 2003; p 9. (3) Zhang, Y.; Kang, D. F.; Aindow, M.; Erkey, C. J. Phys. Chem. B 2005, 109, 2617. (4) Zhang, Y.; Kang, D. F.; Saquing, C.; Aindow, M.; Erkey, C. Ind. Eng. Chem. Res. 2005, 44, 4161. (5) Zhang, Y.; Erkey, C. Ind. Eng. Chem. Res. 2005, 44, 5312. (6) Saquing, C. D.; Kang, D.; Aindow, M.; Erkey, C. Microporous Mesoporous Mater. 2005, 80, 11. (7) Lin, Y. H.; Cui, X. L.; Yen, C.; Wai, C. M. J. Phys. Chem. B 2005, 109, 14410. (8) Lin, Y. H.; Cui, X. L.; Ye, X. R. Electrochem. Comm. 2005, 7, 267. (9) Jiang, R. C.; Zhang, Y.; Swier, S.; Wei, X. Z.; Erkey, C.; Kunz, H. R.; Fenton, J. M. Electrochem. Solid State Lett. 2005, 8, A611. (10) Haji, S.; Zhang, Y.; Kang, D. F.; Aindow, M.; Erkey, C. Catal. Today 2005, 99, 365. (11) Ye, X. R.; Lin, Y. H.; Wang, C. M.; Engelhard, M. H.; Wang, Y.; Wai, C. M. J. Mater. Chem. 2004, 14, 908. (12) Saquing, C. D.; Cheng, T. T.; Aindow, M.; Erkey, C. J. Phys. Chem. B 2004, 108, 7716. (13) Ohde, H.; Ohde, M.; Wai, C. M. Chem. Comm. 2004, 930. (14) Morley, K. S.; Licence, P.; Marr, P. C.; Hyde, J. R.; Brown, P. D.; Mokaya, R.; Xia, Y. D.; Howdle, S. M. J. Mater. Chem. 2004, 14, 1212. (15) Ye, X. R.; Lin, Y. H.; Wai, C. M. Chem. Comm. 2003, 642. (16) Neimark, A. V.; Kheifets, L. I.; Fenelonov, V. B. Ind. Eng. Chem. Prod. Res. DeV. 1981, 20, 39. (17) Neimark, A. V.; Fenelonov, V. B.; Kheifets, L. I. React. Kinet. Catal. Lett. 1976, 5, 67. (18) Cervello, J.; Garcia de la Banda, J. F.; Hermana, E.; Jimenez, J. F. Chem.-Ing.-Tech. 1976, 48, 520. (19) Kim, J.; Roberts, G. W.; Kiserow, D. J. Chem. Mater. 2006, 18, 4710. (20) Tan, C. S.; Liou, D. C. Ind. Eng. Chem. Res. 1990, 29, 1412. (21) Macnaughton, S. J.; Foster, N. R. Ind. Eng. Chem. Res. 1995, 34, 275. (22) Benkhedda, J.; Jaubert, J. N.; Barth, D.; Zetzl, C.; Brunner, G. Sep. Sci. Technol. 2001, 36, 2197. (23) Lucas, S.; Cocero, M. J.; Zetzl, C.; Brunner, G. Fluid Phase Equilib. 2004, 219, 171. (24) Smirnova, I.; Mamic, J.; Arlt, W. Langmuir 2003, 19, 8521. (25) Xu, X. P.; Szanyi, J.; Xu, Q.; Goodman, D. W. Catal. Today 1994, 21, 57.

JP711495N