Preparation and Characterization of Supported Bimetallic PdIV–CoIII

Article pubs.acs.org/Langmuir

Preparation and Characterization of Supported Bimetallic PdIV−CoIII Model Catalyst from Organometallic Single Source Precursor for Aerobic Oxidation of Alcohols Elizabeth Erasmus,*,† J. W. (Hans) Niemantsverdriet,‡ and Jannie C. Swarts† †

Chemistry Department, University of the Free State, P.O. Box 319, 9300, Bloemfontein, South Africa Schuit Institute of Catalysis, Eindhoven University of Technology, P.O. Box 513, 5600 MB Eindhoven, Netherlands

‡

ABSTRACT: The bimetallic paddlewheel catalyst precursor, [PdIICoII(μ-OOCCH3)4] H2O·2CH3COOH (1), prepared from [Pd3(μ-OOCCH3)6] and [Co(OOCCH3)2], was used as a single source precursor to prepare, after binding to a surface-hydroxylated silicon wafer and oxidation, the bimetallic oxides of PdIVCoIII/SiO2 catalyst supported on a model planar (i.e., two-dimensional) silicon wafer. This catalyst catalyzes the aerobic oxidation of alcohols to its corresponding carbonyl compounds. The bimetallic tetracarboxylato catalyst precursor was bonded to the surface-hydroxylated silicon wafer by spincoating and also by grafting. X-ray photoelectron spectroscopy (XPS) revealed that one of the four μ-acetato bridging ligands was substituted by Si−O fragments in a covalent bond formation process during grafting of 1 onto the wafer. In contrast, during the spin-coating process, all four acetato ligands remained intact during fixation on the silicon surface. Upon oxidation and workup, the grafted sample’s Pd:Co ratio remained unchanged (1.0:1.3), whereas the spin-coated sample’s Pd content decreased with respect to Co content. XPS determined binding energies were interpreted to imply that after oxidation in an oxygen/argon mixture of the grafted sample both PdII and CoII were oxidized to produce PdO2 (337.5 eV) and CoIII2O3 (781.1 eV) which most probably interacts with the silicon surface via PdIV−O−Si and CoIII−O−Si bonds. Solvent free aerobic oxidation of octadecanol to its corresponding carbonyl compound was achieved on this oxidized PdIVCoIII/SiO2 model catalyst using molecular oxygen as oxidant under solvent-free conditions. The use of the single source catalyst precursor, 1, resulted in a PdIVCoIII/SiO2 catalyst with superior catalytic activity toward the oxidation of octadecanol over a catalyst prepared from a physical mixture of the separate reactant compounds tripalladium(II) hexaacetate and cobalt(II) diacetate.

1. INTRODUCTION Oxidation of an alcohol to its corresponding carbonyl compounds, either an aldehyde or a carboxylic acid, is an essential and extensively used reaction on laboratory scale and in large scale chemical industries.1 Normally oxidation of alcohols is carried out utilizing either stoichiometric amounts of inorganic oxidants such as chromate or permanganate,2,3 which leads to large amounts of byproducts being liberated. Alternatively, expensive homogeneous catalysts are employed, but these are difficult to separate from the reaction mixture.4,5 These experimental disadvantages of traditional oxidation routes create the opportunity to develop new heterogeneous catalysts that allow the use of molecular oxygen or hydrogen peroxide as oxidant,6−9 since they are more environmentally friendly, easier to work, with and more economic than the traditional inorganic oxidants. Toward this end, we recently showed how silver supported on a silicon wafer can be used to oxidize alcohols to either an aldehyde or carboxylic acid depending on the reaction conditions.10 Platinum and palladium heterogeneous catalysts have widely been investigated, either unpromoted or promoted by Pb0 or PbII,11 or Bi0 or BiIII.11−13 A variety of other metals can also be employed as promoters such as CoII or CoIII,14,15 SnIV,16 and Ru.17 Normally the non-noble metal promoters are inactive when used in isolation as oxidation catalysts.8 However, © 2012 American Chemical Society

addition of non-noble promoters activates the surrounding noble metal catalyst sites which lead to increased oxidation rates,18 and selectivity,16 for the oxidation of hydrocarbons. It is important to note that this non-noble metal promoter is not an active catalyst site itself. The generation of a homogeneous dispersion of the two different metals of a bimetallic heterogeneous catalyst on a solid support surface is very difficult, because metals with similar physical and chemical properties tend to separate and aggregate from other metals. The relative distribution of the different metal types of a bimetallic catalyst on any solid support as well as the catalytic activity of such a heterogeneous catalytic system strongly depends on the catalyst preparation method. It has been found that, for a Pd−Sn heterogeneous bimetallic catalyst supported on alumina (PdSn/Al2O3), addition of SnIV as Sn(C4H9)4 to the Pd/Al2O3 has a positive influence on catalytic hydrogenation as compared to addition of SnCl4.19 Promoting of palladium acetate on carbon with bismuth-oxoacetate to form a bimetallic heterogeneous catalyst has a positive effect on oxidative catalytic performance compared to the activity of the monometallic palladium supported on carbon.20 Received: August 14, 2012 Revised: October 16, 2012 Published: October 23, 2012 16477

dx.doi.org/10.1021/la3032978 | Langmuir 2012, 28, 16477−16484

Langmuir

Article

1610 (vas CO), 1508 (vas CO), 1419 (vs CO), 1377 (vs CO), 1325 (vs CO). The crystal structure of this compound is reported elsewhere; see ref 26. Anal. Calcd for 2: C, 26.71; H, 4.11. Found: C, 26.88; H, 4.42. 2.4. Preparation of Silicon Wafer Having a Hydroxylated (Si−OH) Surface. Silicon wafers with (100) surface orientation were oxidized in air at 750 °C for 24 h to form an amorphous SiO2 overlayer and then cleaned by allowing it to stand in a mixture 1/1 (v/ v) of H2O2 (25% solution) and NH4OH (35% solution) at room temperature. Afterward, the Si-wafers were hydroxylated to form Si− OH bonds on the surface of the wafer, by boiling the wafers in double distilled water for 1 h and dried at room temperature under nitrogen. 2.5. Spin-Coating of 1, [Pd I I 3 (μ-OOCCH 3 ) 6 ], and [CoII(OOCCH3)2] onto a Surface Hydroxylated Silicon Wafer. The hydroxylated SiO2-wafer was washed three times with ethanol. The surface of the hydroxylated SiO2-wafer was spin-coated under dry nitrogen at 2880 rpm with the catalyst-precursor (5 mM solution of 1, [PdII3(μ-OOCCH3)6] or [CoII(OOCCH3)2] in acetone). A catalyst loading on the silicon surface of 7−8 1 molecules nm−2 or 1.30 (±0.1) × 10−5 mol m−2 was calculated as described elsewhere27,28 from eq 1.

The question of whether use of a bimetal organometallic compound as a single source precursor to form a bimetallic heterogeneous catalyst rather than a simple physical mixture of two or more different metal salts would enhance small-particle metal dispersion and at the same time have a positive effect on the catalytic performance was addressed by three studies utilizing organometallic precursors Mo2Co2Sn(Cp)2(CO),21 Pt2Au4(CCtBu)8,22 and [Bi2Pd2(O2CCF3)10(HO2CCF3)2].23 In all three cases, a more homogeneous metal distribution and higher catalytic activity was observed over catalysts that were prepared by physically mixing two or three separate monometallic precursors. This paper reports the use of [PdIICoII(μ-OOCCH3)4]H2O·2CH3COOH (1) as a single source precursor for preparation of a PdIVCoIII/SiO2 bimetallic heterogeneous catalyst suitable for oxidation of an alcohol, here octadecanol, to the corresponding carbonyl compound. A flat two-dimensional silicon-based support (Si-wafer, SiOH/SiO2/Si(100)) was used to model conventional porous three-dimensional supports. Using a flat-model support has many advantages including the resolving of the z-coordinate and less sample charging problems during X-ray photoelectron spectroscopy (XPS) measurements. The resolved z-coordinate implies no diffusion limitations interfering with the reactions and the ability to detect all particles on the surface with surface sensitive techniques such as XPS.10 The insight gained from the method of binding, oxidation states of the metals, and reactivity toward oxidation of alcohols from the two-dimensional flat model catalyst may contribute to a better understanding of the conventional porous three-dimensional supported catalyst.

m = 1.35 × C0

n ρω2tevp

(1)

In eq 1, m is the number of moles of catalyst deposited onto the hydroxylated SiO2-wafer, C0 is the bulk concentration of the catalyst precursor in solution, n (= 0.0012 kg m−1 s−1) is the viscosity of the solvent (here acetone), ρ is the density (= 789 kg m−3) of acetone, ω is the rotation speed (2880 rpm), and tevp (here 4−5 s) is the evaporation time. 2.6. Grafting of 1, [PdII3(μ-OOCCH3)6], and [CoII(OOCCH3)2] onto a Hydroxylated Silicon Oxide on Silicon Wafer. Hydroxylated SiO2-wafer was submerged in a gently stirred 5 mM solution of 1 (or [PdII3(μ-OOCCH3)6] and/or [CoII(OOCCH3)2]) in acetone for 24 h. The hydroxylated SiO2-wafer was then removed from the solution and washed with ethanol, and the excess ethanol was spun off at 2880 rpm under N2. 2.7. Oxidation of 1, [Pd I I 3 (μ-OOCCH 3 ) 6 ], and/or [CoII(OOCCH3)2] on SiO2-Wafer. Freshly prepared silicon wafers with 1 or [PdII3(μ-OOCCH3)6] and/or [CoII(OOCCH3)2]) deposited on them (prepared by spin-coating as well as grafting methods) were placed in an oven filled with 20% O2/Ar. It was then oxidized under 20% O2/Ar flow of 250 mL min−1. The temperature of the oven was increased from 25 to 200 °C at a rate of 10 °C min−1. It was then kept at the same temperature for 30 min before the temperature was further increased at a rate of 10 °C min−1 to 500 °C. The samples were kept at this temperature for 4 h before they were allowed to cool spontaneously overnight to room temperature. 2.8. Flat Model Catalyst Test Reaction. Octadecanol (0.1 g) was heated to 105 °C (upon which the solid melted) in an open-to-air 50 mL glass beaker. The oxidized PdIV,CoIII-coated silicon wafer (approximate dimensions 4 mm × 4 mm) was placed face-down in the liquid in order for the active sites to be in contact with the liquid. The temperature was maintained at 105 °C and samples for ATRFTIR analyses were removed after 24 h.

2. EXPERIMENTAL SECTION 2.1. Chemicals and Instruments. Ethanol, palladium(II) chloride, cobalt(II) acetate tetrahydrate, octadecanol, and octadecanoic acid (Aldrich) were used without any further purification. Palladium black was prepared according to published procedures.24 Attenuated total reflectance (ATR) FTIR spectra were recorded on a Nexus FTIR equipped with a Nicolet Smart Golden Gate. Atomic force microscopy was preformed on a Shimadzu SPM - 9600 instrument with a 125 um scanner. XPS data were recorded on a Kratos AXIS Ultra spectrometer, equipped with a monochromatic Al Kα X-ray source and a delay-line detector (DLD). Spectra were obtained using an aluminum anode (Al Kα = 1486.6 eV) operating at 150 W, with survey scans at constant pass energy of 160 eV and region scans at a constant pass energy of 40 eV. The background pressure was 2 × 10−9 mbar. XPS data obtained was fitted with CasaXPS. All spectra were corrected for charging typically limited to 0.2 eV by using the Si 2p peak of SiO2 which was 103.4 eV. 2.2. Tripalladium(II) Hexaacetate [(Pd3(OOCCH3)6], 2. A modification of a published25 procedure was used. Palladium black (141 mg, 1.33 mmol) was boiled under gentle reflux in a solution of 5 mL of acetic acid and 0.06 mL of HNO3, until evolution of brown NO2 fumes ceased (±1 h). The warm solution was filtered and the filtrate was allowed to cool. The orange-brown crystals that formed were filtered, washed excessively with acetic acid (30 mL) and water (50 mL), and air-dried overnight. Yield 43 mg (0.064 mmol, 15%) of 2. IR (cm−1): 2916 (CH stretch), 1595 (vas CO), 1416 (vs CO), 1348 (CH bending). Anal. Calcd for 2: C, 21.39; H, 2.69. Found: C, 21.4; H, 2.7. 2.3. Palladium II -cobalt II Tetracarboxylate[Pd II Co II (μOOCCH3)4]·H2O·2CH3COOH, 1. A mixture of 2 (340 mg; 0.5 mmol) and [Co(OOCCH3)2]·4H2O (362 mg, 1.45 mmol) in 10 mL of acetic acid was refluxed for 2 h. The mixture was allowed to cool slowly, overnight. The brown precipitate which was formed was filtered off and washed with cold benzene and cold hexane and then dried. Yield 710 mg (1.43 mmol, 98%). IR (cm−1): 1699 (vas CO),

3. RESULTS AND DISCUSSION The use of a flat two-dimensional model silicon-wafer (SiOH/ SiO2/Si(100)) to support 1, a single source heterobimetallic catalyst precursor, was chosen as a proof of concept not only to demonstrate that using a well-defined single source heterobimetallic catalyst precursor for preparation of a bimetallic catalyst are more effective for the oxidation of alcohols than using a physical mixture of two different catalyst precursors to obtain a heterobimetallic catalyst but also to demonstrate that a two-dimensional catalyst support systems are useful in the study of catalyzed organic reactions. To try to address the difficulty of attaining a homogeneous dispersion of the metal-based components of a bimetallic 16478

dx.doi.org/10.1021/la3032978 | Langmuir 2012, 28, 16477−16484

Langmuir

Article

Scheme 1. Synthesis of 1a

a

The structure of this compound is reported in ref 26. The additional acetate and water ligands are shown at the positions they were crystallographically found.

Scheme 2. Illustration of Spin-Coated (top) and Chemically Grafted (bottom) Binding of 1 onto a Hydroxylated Silicon-Wafera

a

XPS results show that the Si−O−Co bond as indicated only represents 60% of the binding modes (i.e., XSi−O−Co = 0.6). The remaining Co−O bond fraction, 0.4, most likely binds CH3COOH and H2O in a similar fashion as the starting material.

dropping one drop of a 5 mM solution of the precatalyst onto the hydroxylate silicon wafer, which is spinning at 2880 rpm until all the solvent has evaporated (4−5 s) under a steady flow of nitrogen. By utilizing eq 1 (section 2.5), the loading may then be calculated as 7−8 molecules/nm2.10,27,28 The complex adheres to the surface probably by means of hydrogen bonding between the water axial ligand of 1 and the hydroxyl groups of the silicon wafer (Scheme 2). XPS measurements (see discussion below) showed that during this process the two axial acetic acid molecules are removed. The grafting method involves a chemical reaction between the hydroxylated silicon surface and 1. XPS measurements show this procedure results in one or more of the μ-OOCCH3 ligands of 1 to exchange with the hydroxyl groups on the surface of the SiO2-wafers in a ligand exchange reaction (see Scheme 2) to form a direct covalent bond between the PdII, CoII centers and the O-atoms protruding from the surface of the silicon-wafers. Especially the graft-process should lead to improved catalyst−support interaction and good PdII/CoII dispersion. To confirm the oxidation states of the palladium and cobalt before and after oxidation as well as to determine the influence

heterogeneous catalyst at atomic level on the surface of a SiO2modified support, the bimetallic precursor 1 was used as a single-source of both active metals rather than a mixture of two components, one containing only PdII and the other only CoII. The synthesis of 1 involves the reaction between tripalladium(II) hexaacetate [Pd3(μ-OOCCH3)6] (2) and cobalt(II) diacetate [Co(OOCCH3)2]·4H2O (3) in glacial acetic acid to yield the desired product in almost quantitative yield (98%), Scheme 1. Crystallographically it was shown that the product 1 contains an axial H2O and two acetic acid ligands associated to the Co via a hydrogen interaction.26 These axial ligands come from solvents used during the preparation. This bimetallic coordination complex, 1, was then coated on top of a flat-model hydroxylated SiO2-wafer by spin-coating (the two-dimensional equivalent of impregnation) or 24 h chemical grafting by means of ligand exchange. Spin-coating is the two-dimensional version of wetimpregnation and involves deposition of 1 onto the hydroxylated surface of a SiO2-wafer by spin-coating for a specific time (until Newton rings appear, showing that all the excess solvent has evaporated), here 4−5 s. This is achieved by 16479

dx.doi.org/10.1021/la3032978 | Langmuir 2012, 28, 16477−16484

Langmuir

Article

Figure 1. XPS spectra of cobalt (left), palladium (middle), and carbon (right) on hydroxylated silicon surfaces, before (bottom) and after (top) oxidation. The peaks observed in the oxidized carbon (top right) are due to C−O and C−C bonds normally present on the surface. X1 represents the carboxylato C peak, OOCCH3, while X2 represents the side chain C peak, OOCCH3, of the carboxylato ligand.

Table 1. Metal Ratios and Binding Energies from XPS Data for 1−3 Incorporated onto a Si-Wafer Having a Hydroxylated Surface before and After Oxidationa Si

Pd

Co

OOCCH3

Co 2p3/2 (eV)

Co 2p1/2 (eV)

Pd 3d5/2 (eV)

Pd 3d3/2 (eV)

spin-coated

“fresh” “oxidized”

5.3(0.2) 8.7(0.2)

1.0(0.2) 1.0(0.2)

1.1(0.2) 1.7(0.2)

3.8(0.2)

781.8 781.7

797.4 797.5

338.2 337.4

343.5 342.7

grafted

“fresh” “oxidized”

2.2(0.2) 3.5(0.2)

1.0(0.2) 1.0(0.2)

1.2(0.2) 1.3(0.2)

2.6(0.2)

781.9 781.1

797.5 796.6

338.3 337.5

343.7 342.8

Pd (2)b

“fresh” “oxidized”

5.3(0.2) 8.7(0.2)

1.0(0.2) 1.0(0.2)

337.9 337.1

343.2 342.5

Co (3)b

“fresh” “oxidized”

2.2(0.2) 3.5(0.2)

Pd (2) + Co (3)b

“fresh” “oxidized”

2.2(0.2) 3.5(0.2)

337.7 337.2

343.0 342.6

1.0(0.2) 1.0(0.2)

0.9(0.2)

1.0(0.2) 1.0(0.2)

1.0(0.2)

781.5 781.1

797.5 796.7

1.1(0.2) 1.4(0.2)

0.9(0.2)

781.7 781.2

797.7 796.8

a

The value in brackets is the standard deviation from a set of three experiments. bThe data is for the metal species 2−3 grafted onto the hydroxylated Si-wafer surface.

of the second metal on the binding energy of the first metal, XPS analysis was performed on all the prepared samples (mono- and bimetallic, spin-coated and grafted). Charging problems normally obtained during XPS analysis of aluminabased three-dimensional supported catalysts were not observed in the present flat two-dimensional silicon support model system due to the inherent conductivity of the silicon support. Also, catalyst testing could be performed on the same catalystwafer as the XPS analysis was done. It is known that there is always some deposited carbon on the surface of any sample.29,30 This interferes with XPS to quantify the C-content associated with catalyst deposition although the contaminant is not expected to have acetate groups in its structure. To quantify the intrinsic amount of carbon present on the hydroxylated Si-wafer, a “blank” surface was analyzed. This “blank” hydroxylated Si-wafer was prepared in exactly the same way as the hydroxylated Si-wafer with 1 deposited on the surface (either by impregnation or by grafting), except no metal complex was dissolved in the grafting or impregnation solutions. The carbon content on the “blank” wafer was determined by XPS. The signal from the “blank” was then subtracted from the carbon spectrum found for the spincoated and grafted hydroxylated Si-wafer with the PdIICoII on the surface. The remaining carbon measured on the surface of the two hydroxylated Si-wafer samples containing the PdIICoII

was assigned to originate from acetate (μ-OOCCH3) ligands bridging the PdII and CoII; see Figure 1, right. XPS binding energies obtained for the acetate ligands OOCCH3 (289.3 eV) and OOCCH3 (286.8 eV) were found within experimental error at the same position reported for acid carboxylic functions (289.4 and 287.0 eV).31 The XPS data of the two differently prepared samples (compound deposition by means of spin-coating and chemical grafting) revealed that the ratios silicon/palladium/cobalt/ carbon (carbon from the bridging carboxylate ligands) were different (see Table 1). The average Si:Pd:Co:OOCCH3 ratio for samples that were prepared by spin-coating was found to be 5.3:1.0:1.1:3.8. This carbon content is within experimental error that is expected if no μ-OOCCH3 ligands are displaced from 1, where only the two CH3COOH ligands associated with axial H2O and are replaced by hydrogen bonding between the axial H2O and the hydroxyl groups from the Si-surface; see Scheme 1, top. However, for samples that were obtained from grafting, the average Si:Pd:Co:OOCCH3 ratio was 2.2:1:1.2:2.6. This implies that at least one of the μ-OOCCH3 ligands was substituted by a Si−O−Co and/or Si−O−Pd coordination bond, and that during the grafting process at least half of the axial H2O·2CH3COOH ligands may also have been replaced by a Si−O−Co bond; see Scheme 1, bottom. 16480

dx.doi.org/10.1021/la3032978 | Langmuir 2012, 28, 16477−16484

Langmuir

Article

concentration. Also, the metal ratio of palladium to cobalt did not change significantly after oxidation for the grafted sample (Table 1). From this it is possible to conclude that even after oxidation (removing of the organic bridges between the metals), the metals are still bonded through the hydroxyl derived oxygen on the Si-wafer (i.e., metal ester bond, M−O− Si), preventing reduction in metal concentration or aggregation to form a range of larger particles of different sizes which would show lower intensities. The sample prepared by spin-coating showed a change in metal ratio, from Pd:Co = 1:1.1 to 1:1.7. Thus, the palladium content “apparently” decreased after oxidation. The reason could be twofold. First, it may be that palladium aggregates to form larger particles, or second cobalt might segregate to the outside of the PdIVCoIII mixed oxide particle and thereby attenuate the signal from palladium. Whichever process is taking place, it caused a only minor change in the Pd:Co ratio of the impregnated sample. Comparison of the binding energies of Pd 3d5/2 and Co 2p3/2 from the catalyst obtained from the single source bimetallic catalyst precursor 1 with the binding energies of the monometallic catalyst precursors, PdII3(μ-OOCCH3)6] (Pd 3d5/2 = 337.9 eV) and [CoII(OOCCH3)2] (Co 2p3/2 = 781.5 eV), revealed that the binding energy for both Pd 3d5/2 and Co 2p3/2 from 1 is ca. 0.4 eV larger. This demonstrates the influence of the two metals on each other when they are within the same molecule (Table 1). The catalyst obtained by the physically mixing PdII3(μ-OOCCH3)6] and [CoII(OOCCH3)2] had binding energies that were between the monometallic and bimetallic source, demonstrating that the second metal does influence the binding energy, but not as much as when the two metals originate from within the same molecule. After oxidizing the catalyst, the binding energy of Co 2p3/2 (ca. 781.1 eV) for the catalysts prepared by monometallic, physical mixture of monometallic and grafting of the bimetallic catalyst precursors were within experimental error the same, thus showing that the cobalt oxide that formed are the same for all three. The binding energy (Pd 3d5/2 = ca. 337.1 eV) of the PdO2 obtained by oxidizing the monometallic palladium and physical mixture of monometallic catalyst precursors were within experimental error the same, thus showing that the metal oxide obtained are the same. However, the oxidation of palladium of the sample obtained by grafting and spin-coating of the bimetallic catalyst precursor were ca. 0.4 eV higher. This shows that the PdO2 obtained from 1 is still under the influence of the cobalt (now CoxOy), which could be interpreted that even after oxidation of 1 on the two-dimensional support, the palladium and cobalt oxides are within close proximity of each other. The oxidized PdIVCoIII catalysts were investigated in the solvent free oxidation of octadecanol using molecular oxygen present in air to prove that catalyst supported on a twodimensional surface, here a modified silicon wafer, can be used in the study of simple organic reactions. The reaction is shown in Scheme 3. This reaction was monitored by ATR FTIR (see Figure 2), following the appearance of the carbonyl peaks at 1718 and 1734 cm−1 and the disappearance of the hydroxyl peak (at 3200 cm−1) of the alcohol. Comparison of the position of the carbonyl (CO) stretching frequency of the product obtained by oxidation (maximum at 1734 cm−1) with the wavelength of carbonyl stretching frequency of the carboxylic acid (1710 cm−1)1b and the aldehyde (1730 cm−1)1 clearly showed that the aldehyde is formed as the major product.

For the purpose of this publication, one has to distinguish between 1 which is anchored on a hydroxylated SiO2 surface but not further treated with 20% O2/Ar; these samples are referred to as “fresh” and samples which were oxidized in 20% O2/Ar. The latter set of samples is referred to as “oxidized”. Table 1 summarizes Si:Pd:Co:C2 ratios for “fresh” and “oxidized” samples. The XPS binding energies for palladium and cobalt (Table 1) of the “fresh” catalyst for the two different preparation methods (spin-coated and grafting) are within experimental error the same. It can therefore be assumed that the chemical environment of each metal after deposition on the hydroxylated Si-wafer is almost the same. This is to be expected as the (C− O)3−PdII−(O−Si) and (C−O)3−CoII−(O−Si) chemical environments around the Pd and Co centers on the surface of the Si-wafer should be much the same as the PdII−(O−C)4 and CoII−(O−C)4 chemical environments around the metals in the free precursor 1. In the cobalt region of the XPS spectra of the “fresh” samples (Figure 1, left), the expected shakeup feature was observed at 786.4 eV, which is indicative of an ionic Co species present. The Co 2p3/2 peak was found at a binding energy of ca. 781.8 eV, which is within the reported values of 781.2−782.0 eV reported for cobalt acetate on SiO2.32 The Pd 3d5/2 peak which appears just above 338 eV (see Figure 1, middle) for the “fresh” catalyst corresponds well with the known Pd(OCCH3)2 Pd 3d5/2 peak at 337.75 eV.33 The next step in the preparation of the catalyst is the oxidation of 1 on the Si(OH)/SiO2/Si(100). In this process, the metals become oxidized and the organic fragment is “burned” off. From the XPS spectra (see Figure 1, right), it can clearly be seen that none of the peaks belonging to the organic material, the acetate bridges marked X1 (OOCCH3) and X2 (OOCCH3) in the “fresh” catalyst’s spectra, remained in those of the oxidized sample. The new binding energies after oxidation for palladium and cobalt in both impregnated and grafted samples are given in Table 1. Noteworthy is that, for both preparations, the binding energy of palladium (for Pd 3d5/2) decreased upon oxidation by about 1 eV (from ca. 338 eV to ca. 337 eV). The Pd 3d5/2 peak at ca. 337.4 eV is between the reported binding energies for PdO (336.6 eV) and PdO2 (338 eV).34 However, it corresponds well with reported binding energy value of 337.3 eV for the oxidized product of K2Pd(NO2)4 on alumina.35 The binding energy for the impregnated cobalt sample’s peaks after oxidation remained within experimental error unchanged after oxidation. The grafting approach led to a decrease of ca. 0.8 eV in the Co 2p3/2 peak binding energy. The position of the binding energy of the Co 2p3/2 peak is between Con+ in cobalt oxides (780 eV) and Co2+ bound to a zeolite (782 eV).36,37 These results demonstrate the difficulty of identifying Co redox states by XPS. Both cobalt(II) and cobalt(III) oxides show Co 2p3/2 peaks at ca. 780 eV.38 Experimentally we observe that, after oxidation, the Co 2p3/2 peak changed position to ca. 781.1 and 781.7 eV for the impregnated and grafted sample respectively. We attribute this to a CoIII species (including Co2O3, which is reported to be have a binding energy of 781.3 eV,39 and O−CoIII−O−Si species), which is within the range of 779.3−781.8 eV reported for calcined Co/SiO2.32 The sample that was prepared by grafting showed similar intensities in counts per second before and after oxidation, which shows that there was no loss or change in metal 16481

dx.doi.org/10.1021/la3032978 | Langmuir 2012, 28, 16477−16484

Langmuir

Article

sublimed during course of the catalytic reaction, an estimation of the “quasi”-turnover frequency (TOF) was determined, which would give an indication of how active the model catalyst is. In the estimation of the quasi-TOF number, 0.1 g of octadecanol (3.69 × 10−3 mol, 2.22 × 1020 molecules) was fully converted to the carbonyl in 60 h by the catalyst prepared by spin-coating having 7−8 active sites/nm2. This translates to a total of 1.12−1.28 × 1014 active sites for a support with a total surface area of 4 × 4 mm2; see Experimental Section. The quasi-TOF was determined by eq 2. The quasi-TOF of the catalyst obtained by spin-coating is 8−9 molecules s−1. Since the intensity of the counts per second and the percent area of the peaks (from the measured XPS spectra, which is an indication of concentration) for the metals are similar for catalyst prepared by spin-coating and grafting, we can assume that the same amount of active sites is present on the catalyst prepared by grafting. Thus, using the data from the catalyst prepared by spin-coating to determine the quasi-TOF of the catalyst prepared by grafting and allowing for 40 h of conversion time, the quasi-TOF of the catalyst obtained by grafting is 12−14 molecules s−1. This shows that the grafted sample is a more active catalyst for the oxidation of alcohols than the spin-coated catalyst. This two-dimensional model catalyst prepared by spin-coating and grafting compares very well with the bimetallic three-dimensionally supported catalyst Au60Pd40/activated carbon having a TOF = 18 000 h−1 (5 s−1) for the oxidation of benzaldehyde,40 and the monometallic palladium-grafted hydroxyapatite supported catalyst having a TOF = 9800 h−1 (3 s−1) for the oxidation of 1-phenylethanol under solvent free conditions.41 The two-dimensional model catalyst performed even slightly better than the conventional three-dimensional catalyst. This clearly shows that the use of a single source bimetallic catalyst precursor, here [PdIICoII(μOOCCH3)4]H2O·2CH3COOH (1), rather than the two separate salts of each metal, results in a more active catalyst.

Scheme 3. Aerobic Oxidation of Octadecanol to Its Corresponding Carbonyl Compound to the Aldehyde Using a Flat PdIVCoIII/Si(OH)/SiO2/Si(100) Model Catalysta

a

The major carbonyl product is the aldehyde.

Figure 2. ATR FTIR spectra of octadecanol (bottom), octadecanol after 24 h oxidation in the presence of PdIVCoIII/Si(OH)/SiO2/ Si(100) prepared by spin-coating (middle), and after 24 h oxidation in the presence of PdIVCoIII/Si(OH)/SiO2/Si(100) prepared by grafting (top).

⎡ ⎢ TOF = ⎢ ⎢⎣

Comparison of the oxidation of octadecanol using the two different catalysts (one obtained from spin-coated precursors and the other by grafting) under identical reaction conditions revealed that after 24 h the catalyst obtained by oxidation of the grafted precursor, 1, produced about four times more carbonyl product than the catalyst obtained by oxidation of the spincoated precursor. This conclusion stems from comparison of the intensities of the carbonyl peak (see Figure 2). Quantitatively, after 40 h reaction time for the catalyst obtained via the grafted route, all the octadecanol was converted to the carbonyl product since the OH peak at 3340 cm−1 was no longer visible. In contrast, alcohol oxidation on the catalyst obtained via the spin-coating route was only complete after 60 h. A detail kinetic study was not conducted, since the goal of this study was to demonstrate how a single source catalyst precursor such as 1 could be synthesized and anchored onto a two-dimensional model SiO2-support surface, to show the ease of characterization of the supported catalyst, and to demonstrate its operational effectiveness as a catalyst with the simple catalytic reaction where octadecanol is aerobically oxidized. However, even though a detailed kinetic study was not conducted, and minute amount the alcohol, octadecanol,

of molecules converted ⎤ ( amountamount )⎥ of active sites

time in s

⎥ ⎥⎦

(2)

To quantify any enhancement the grafted PdIVCoIII/SiO2 catalyst might have over the catalytic activity of the grafted monometallic oxidized PdIV/SiO2 or CoIII/SiO2 catalyst, the latter two single metallic catalysts were tested under identical conditions. After 24 h, ATR FTIR was recorded for each octadecanol oxidation. Figure 3 shows the graph of percent of reflectance increase of the carbonyl peak. The starting material, octadecanol was taken as zero % carbonyl content, and the PdIVCoIII/SiO2 (grafted sample) catalyst’s results described above as 100%. As Figure 3 shows, relative to the PdIVCoIII/ SiO2 (grafted sample), the monometallic PdIV/SiO2 catalyst generated only 22% carbonyl from octadecanol compared to PdIVCoIII/SiO2 in the allowed 24h reaction time. The monometallic CoIII/SiO2 catalyst only generates 15% carbonyl content under identical conditions. Theoretically a combination of PdIV/SiO2 and CoIII/SiO2 would thus generate 22 + 15 = 37% carbonyl compounds during 24 h. There is thus a factor 2.8 increase in catalytic activity if PdIVCoIII/SiO2 is prepared from 1 over what is expected from the combination of PdIV/ SiO2 and CoIII/SiO2. When the two monometallic acetate precursors, 2 and 3, were dissolved together in the same solution to have the same Pd and Co content and then grafted onto the hydroxylated Si-wafer, a factor of 1.8 reflectance 16482

dx.doi.org/10.1021/la3032978 | Langmuir 2012, 28, 16477−16484

Langmuir

Article

CH3COO− ligands during the deposition process. This result is consistent with PdII and CoII bound to the Si(OH)/SiO2/ Si(100) wafer through oxygen bridge atoms such as M−O−Si with M = PdII and CoII prior to oxidation. Upon oxidation in oxygen, the organic ligand system was burned off and XPS revealed that the grafted sample’s Pd/Co ratio remained ca. 1.0:1.3. In the spin-coated sample, the Pd content decreased by 35%, to 1.0:1.7 (or more appropriately 0.65:1.1). This result is attributed to segregation and aggregation of the like metals into larger monometal ion clusters. Solvent free oxidation of octadecanol was achieved more successfully on the grafted and oxidized PdIVCoIII/SiO2 catalytic surface using molecular oxygen as oxidant; this bimetallic catalyst showed a twofold gain in catalytic reactivity over that expected when measuring the activity of a catalyst obtained from mixing monometal ion catalytic systems of Pd and Co. This observed synergistic effect is attributed to more control in the preparation of smaller and more homogeneously distributed mixed metal PdIVCoIII-oxide particles. The paper illustrates how multicomponent catalysts can be prepared with a homogeneous composition by nature from binuclear organometallics.

Figure 3. Comparative graph showing the percent increase in carbonyl peak, taking starting material as zero and PdIVCoIII/SiO2 (24 h chemical grafting sample) as 100%. Pd−Co = PdIVCoIII/SiO2 (1, grafted sample), Pd = PdIV/SiO2, Co = CoIII/SiO2, and Pd + Co = PdIVCoIII/SiO2 (prepared from a solution containing separate components 2 and 3 mixed in such a way that Pd and Co content are equal). All model catalysts contain the same amount of metal in total.

■

AUTHOR INFORMATION

Corresponding Author

increase of the carbonyl peak was obtained over the 37% yield theoretically calculated above; that is, 66% carbonyls were obtained. This is 34% lower than the 100% reflectance increase of the carbonyl peak that found for the sample prepared by the single source bimetallic 1, but higher than the sum of the separate monometallic catalysts. This implies that there exists a more effective synergistic effect between the catalyst composing of PdIV and CoIII when prepared by the single source bimetallic precursor 1 than otherwise. The reason for this observation is most likely a more homogeneous particle constitution having a Pd:Co ratio of 1:1.3. Synergism may then be enhanced by the smaller Pd/Co catalyst particle size. Use of the bimetallic precursor appears to stabilize the smaller particle size during particle formation better. Synergistic effects in bimetallic catalyst have been reported before.42,43 Seminario et al. showed by using computational chemistry when Pt is mixed with Co, it leads to better O2 dissociation than on pure Pt.44 By assuming PdIV reacts the same as Pt, the PdIV−CoIII combination would lead to better O2 dissociation, and this in combination with palladium’s alcohol dehydrogenation ability45,46 leads to better catalytic activity. The combination PdIV and CoIII in the same molecule to create a bimetallic single source catalyst precursor, 1, showed enhanced synergistic effects over the catalyst prepared from a physical mixture of the separate reactant compounds tripalladium(II) hexaacetate and cobalt(II) diacetate.

*Telephone: ++27-(0)51-401-9656. Fax: ++27-(0)51-4446384. E-mail: [email protected]. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors acknowledge financial support from SASOL, Eindhoven University of Technology, and the University of the Free State during the course of this study. Tiny Verhoeven and Gilbere Mannie are acknowledged for recording the XPS data.

■

REFERENCES

(1) Parshall, G. W.; Ittel, S. D. Homogeneous Catalysis, Second Edition; Wiley Interscience: New York, 1992. (2) Ley, S. V.; Madin, A. Comprehensive Organic Synthesis; Pergamon: Oxford, 1991. (3) Sheldon, R. A.; Kochi, J. K. Metal-Catalyzed Oxidations of Organic Compounds; Academic Press: New York, 1981. (4) Stahl, S. S.; Thorman, J. L.; Nelson, R. C.; Kozee, M. A. Oxygenation of Nitrogen-Coordinated Palladium(0): Synthetic, Structural, and Mechanistic Studies and Implications for Aerobic Oxidation Catalysis. J. Am. Chem. Soc. 2001, 123, 7188−7189. (5) Muldoon, J.; Brown, S. N. Practical Os/Cu-Cocatalyzed Air Oxidation of Allyl and Benzyl Alcohols at Room Temperature and Atmospheric Pressure. Org. Lett. 2002, 4, 1043−1045. (6) Mertens, P. G. N.; Corthals, S. L. F.; Ye, X.; Poelman, H.; Jacobs, P. A.; Sels, B. F.; Vankelecom, I. F. J.; De Vos, D. E. Selective alcohol oxidation to aldehydes and ketones over base-promoted gold− palladium clusters as recyclable quasihomogeneous and heterogeneous metal catalysts. J. Mol. Catal. A: Chem. 2009, 313, 14−21. (7) Wang, F.; Ueda, W. Selective oxidation of alcohols using novel crystalline Mo-V-O oxide as heterogeneous catalyst in liquid phase with molecular oxygen. Selective oxidation of alcohols using novel crystalline Mo-V-O oxide as heterogeneous catalyst in liquid phase with molecular oxygen. Catal. Today 2009, 144, 358−361.

4. CONCLUSION The current study shows that a planar model PdIV−CoIII/ Si(OH)/SiO2/Si(100) catalyst can be prepared from a single source bimetallic precursor, [Pd II Co II (μ-OOCCH 3 ) 4 ]H2O·2CH3COOH (1). The latter was deposited onto a hydroxylated SiO2-wafer via two different routes, spin-coating and 24 h chemical grafting. Characterization with XPS after surface coating but before oxidation revealed that incorporation of 1 via spin-coating proceeded without acetate ligand loss and implies hydrogen binding between the water axial ligand of the PdIICoII tetraacetato precursor to the hydroxyl groups of the support. In contrast, the grafted sample lost on average ca. 1.5 16483

dx.doi.org/10.1021/la3032978 | Langmuir 2012, 28, 16477−16484

Langmuir

Article

(8) Mallot, T.; Baiker, A. Oxidation of Alcohols with Molecular Oxygen on Solid Catalysts. Chem. Rev. 2004, 104, 3037−3058. (9) Zhan, B. Z.; Thompson, A. Recent developments in the aerobic oxidation of alcohols. Tetrahedron 2004, 60, 2917−2935. (10) Erasmus, E.; Niemantsverdriet, J. W.; Swarts, J. C. A new approach to silver-catalysed aerobic oxidation of octadecanol: Probing catalysts utilising a flat, two-dimensional silicon-based model support system. Catal. Commun. 2012, 27, 193−199. (11) Alardin, F.; Delmon, B.; Ruiz, P.; Devillers, M. Stability of bimetallic Bi−Pd and Pb−Pd carbon-supported catalysts during their use in glyoxal oxidation. Catal. Today 2000, 61, 255−262. (12) Wenkin, M.; Ruiz, P.; Delmon, B.; Devillers, M. The role of bismuth as promoter in Pd−Bi catalysts for the selective oxidation of glucose to gluconate. J. Mol. Catal. A: Chem. 2002, 180, 141−159. (13) Radnik, J.; Benhmid, A.; Kalevaru, V. N.; Pohl, M. M.; Martin, A.; Lucke, B.; Dingerdissen, U. Deactivation of Pd Acetoxylation Catalysts: Direct Observations by XPS Investigations. Angew. Chem., Int. Ed. 2005, 44, 6771−6774. (14) Hronec, M.; Cvengrosova, Z.; Tuleja, J.; Ilavsky, J. Liquid-Phase Oxidation of Hydrocarbons and Alcohols Catalyzed by Heterogeneous Palladium and Platinum Catalysts. Stud. Surf. Sci. Catal. 1990, 55, 169−176. (15) Skoglundh, M.; Johansson, H.; Lowendahl, L.; Jansson, K.; Dahl, L.; Hirschauer, B. Cobalt-promoted palladium as a three-way catalyst. Appl. Catal., B 1996, 7, 299−319. (16) Pinxt, H.; Kuster, B. F. M.; Marin, G. B. Promoter effects in the Pt-catalysed oxidation of propylene glycol. Appl. Catal., A 2000, 191, 45−54. (17) Deffernez, A.; Hermans, S.; Devillers, M. Bimetallic Bi−Pt, Ru− Pt and Ru−Pd and trimetallic catalysts for the selective oxidation of glyoxal into glyoxalic acid in aqueous phase. Appl. Catal., A 2005, 282, 303−313. (18) Lefranc, H. European Patent EP 667,331, 1999. (19) Aduriz, H. R.; Bodnariuk, P.; Coq, B.; Figueras, F. Aluminasupported bimetallics of palladium alloyed with germanium, tin, lead, or antimony from organometallic precursors: I. Preparation and characterization. J. Catal. 1989, 119, 97−107. (20) Wenkin, M.; Renard, C.; Ruiz, P.; Delmon, B.; Devillers, M. On the role of bismuth-based alloys in carbon-supported bimetallic Bi-Pd catalysts for the selective oxidation of gluconic acid. Stud. Surf. Sci. Catal. 1997, 108, 391−398. (21) Song, C.; Parfitt, D. S.; Schobert, H. H. Bimetallic dispersed sulfide catalysts from organometallic clusters for coal liquefaction. Catal. Lett. 1993, 21, 27−34. (22) Chandler, B. D.; Schabel, A. B.; Blanford, C. F.; Pignolet, L. H. Preparation and Characterization of Supported Bimetallic Pt−Au Particle Catalysts from Molecular Cluster and Chloride Salt Precursors. J. Catal. 1999, 187, 367−384. (23) Li, B.; Zhang, H.; Huynh, L.; Diverchy, C.; Hermans, S.; Devillers, M.; Dikarev, E. Bismuth−Palladium Heterometallic Carboxylate as a Single-Source Precursor for the Carbon-Supported Pd−Bi/C Catalysts. Inorg. Chem. 2009, 48, 6152−6158. (24) Wieland, H. Hydrogenation and Dehydrogenation. Chem. Ber. 1912, 45, 484−493. (25) Stephenson, T. A.; Morehouse, S. M.; Powell, A. R.; Heffer, J. P.; Wilkinson, G. Carboxylates of palladium, platinum, and rhodium, and their adducts. J. Chem. Soc. 1965, 3632−3640. (26) Akhmadullina, N. S.; Cherkashina, N. V.; Kozitsyna, N. Y.; Stolarov, I. P.; Perova, E. V.; Gekhman, A. E.; Nefedov, S. E.; Vargaftik, M. N.; Moiseev, I. I. Synthesis of palladium(II) 3d-metal(II) paddlewheel acetate-bridged heterodimetallic complexes: Unexpected catalysis by water molecules. Inorg. Chim. Acta 2009, 362, 1943−1951. (27) Niemantsverdriet, J. W.; Engelen, A. F. P.; de Jong, A. M.; Wieldraaijer, W.; Kramer, G. J. Realistic surface science models of industrial catalysts. Appl. Surf. Sci. 1999, 144, 366−374. (28) Gunter, P. L. J.; Niemantsverdriet, J. W.; Ribeiro, F. H.; Somorjai, G. A. Surface science approach to modeling supported catalysts. Catal. Rev. Sci. Eng. 1997, 39, 77−168.

(29) Castaneda, C. M.; Ashbaugh, L.; Wakabayashi, P. Use of proton backscattering to determine the carbon and oxygen content in fine particle samples deposited on PTFE((CF2)n) membrane disk filters. J. Aerosol Sci. 2010, 41, 99−107. (30) Kyotani, T.; Nagai, T.; Inoue, S.; Tomita, A. , Formation of New Type of Porous Carbon by Carbonization in Zeolite Nanochannels. Chem. Mater. 1997, 9, 609−615. (31) Dablemont, C.; Lang, P.; Mangeney, C.; Piquemal, J. Y.; Petkov, V.; Herbst, F.; Viau, G. FTIR and XPS Study of Pt Nanoparticle Functionalization and Interaction with Alumina. Langmuir 2008, 24, 5832−5841. (32) Okamoto, Y.; Nagata, K.; Adachi, T.; Imanaka, T.; Inamura, K.; Takyu, T. Preparation and characterization of highly dispersed cobalt oxide and sulfide catalysts supported on SiO2. J. Phys. Chem. 1991, 95, 310−319. (33) Mathew, J. P.; Srinivasan, M. Photoelectron spectroscopy (XPS) studies on some palladium catalysts. Eur. Polym. J. 1995, 31, 835−839. (34) Katzer, J. R.; Smith, G. C. A.; van Hoof, J. H. C. Paramagnetic platinum and oxygen species on supported platinum. J. Catal. 1979, 59, 278−292. (35) Benkhaled, M.; Morin, S.; Pichon, C.; Thomazeau, C.; Verdon, C.; Uzio, D. Synthesis of highly dispersed palladium alumina supported particles: Influence of the particle surface density on physico-chemical properties. Appl. Catal., A 2006, 312, 1−11. (36) Zsoldos, Z.; Vass, G.; Lu, G.; Guczi, L. XPS study on the effects of treatments on Pt2+ and Co2+ exchanged into NaY zeolite. Appl. Surf. Sci. 1994, 78, 467−475. (37) Boix, A.; Fierro, J. L. G. X-ray photoelectron spectroscopy analysis of platinum- and/or cobalt-loaded zeolites relevant for selective catalytic reduction of Nox. Surf. Interface Anal. 1999, 27, 1107−1113. (38) Dillard, J. G.; Koppelman, M. H. X-ray photoelectron spectroscopic (xps) surface characterization of cobalt on the surface of kaolinite. J. Colloid Interface Sci. 1982, 87, 46−55. (39) Tan, B. J.; Klabunde, K. J.; Sherwood, P. M. A. XPS studies of solvated metal atom dispersed (SMAD) catalysts. Evidence for layered cobalt-manganese particles on alumina and silica. J. Am. Chem. Soc. 1991, 113, 855−861. (40) Villa, A.; Janjic, N.; Spontoni, P.; Wang, D.; Su, D. S.; Prati, L. Au−Pd/AC as catalysts for alcohol oxidation: Effect of reaction parameters on catalytic activity and selectivity. App. Catal., A 2009, 364, 221−228. (41) Mori, K.; Hara, T.; Mizugaki, T.; Ebitani, K.; Kaneda, K. Hydroxyapatite-Supported Palladium Nanoclusters: A Highly Active Heterogeneous Catalyst for Selective Oxidation of Alcohols by Use of Molecular Oxygen. J. Am. Chem. Soc. 2004, 126, 10657−10666. (42) Bianchini, C.; Shen, P. K. Palladium-Based Electrocatalysts for Alcohol Oxidation in Half Cells and in Direct Alcohol Fuel Cells. Chem. Rev. 2009, 109, 4183−4206. (43) Dimitratos, N.; Villa, A.; Wang, D.; Porta, F.; Su, D.; Prati, L. Pd and Pt catalysts modified by alloying with Au in the selective oxidation of alcohols. J. Catal. 2006, 244, 113−121. (44) Balbuena, P. B.; Altomare, D.; Agapito, L.; Seminario, J. M. Theoretical Analysis of Oxygen Adsorption on Pt-Based Clusters Alloyed with Co, Ni, or Cr Embedded in a Pt Matrix. J. Phys. Chem. B 2003, 107, 13671−13680. (45) Keresszegi, C.; Burgi, T.; Mallat, T.; Baiker, A. On the Role of Oxygen in the Liquid-Phase Aerobic Oxidation of Alcohols on Palladium. J. Catal. 2002, 211, 244−251. (46) Keresszegi, C.; Mallat, T.; Grunwaldt, J. D.; Baiker, A. A simple discrimination of the promoter effect in alcohol oxidation and dehydrogenation over platinum and palladium. J. Catal. 2004, 225, 138−146.

16484

dx.doi.org/10.1021/la3032978 | Langmuir 2012, 28, 16477−16484