Alumina-supported platinum catalysts: Local atomic structure and catalytic activity for complete methane oxidation

Applied Catalysis A: General 486 (2014) 12–18

Contents lists available at ScienceDirect

Applied Catalysis A: General journal homepage: www.elsevier.com/locate/apcata

Alumina-supported platinum catalysts: Local atomic structure and catalytic activity for complete methane oxidation Vera P. Pakharukova a,b,∗ , Ilya Yu. Pakharukov a,b , Valerii I. Bukhtiyarov a,b , Valentin N. Parmon a,b a b

Boreskov Institute of Catalysis, pr. Lavrentieva 5, 630090 Novosibirsk, Russia Novosibirsk State University, Pirogova Street 2, 630090 Novosibirsk, Russia

a r t i c l e

i n f o

Article history: Received 4 June 2014 Received in revised form 23 July 2014 Accepted 15 August 2014 Available online 23 August 2014 Keywords: Alumina-supported platinum catalysts Local structure Radial distribution function of electron density Total methane oxidation

a b s t r a c t The structure of platinum species in a model set of monodisperse Pt/␥-Al2 O3 catalysts was studied using radial distribution function (RDF) of electron density. Catalyst preparation conditions were revealed to considerably affect the dispersion and structure of supported platinum. Cation vacancies on the ␥-Al2 O3 surface were essential to anchor electron deficient platinum atoms and clusters. Platinum segregation with forming finely dispersed metal particles was evident in some catalysts. The methane total oxidation over the Pt/␥-Al2 O3 catalysts was established to depend strongly on the platinum oxidation state. The catalysts containing oxidized platinum species and highly dispersed oxidizable Pt0 particles were more active. The metal–support interaction improved the performance of the catalysts indirectly, by stabilizing the platinum dispersion. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Alumina-supported platinum (Pt/␥-Al2 O3 ) catalysts appear to be one of the most attractive catalysts in a variety of practically important processes such as oxidation of hydrocarbons [1,2], gasoline reforming processes [3], CO oxidation [4,5], etc. Dispersion, structure, oxidation state of the active component as well as metal–support interaction are well known to impact significantly on catalytic performance. Thus, the total oxidation of methane over the Pt/␥-Al2 O3 catalysts was shown to be highly structure sensitive [6]. Hence, comprehensive studies on the atomic-scale structure and properties of catalytically active sites of the Pt/␥-Al2 O3 catalysts appear to be a challenge. Numerous studies were aimed at the state and structure of platinum in the Pt/␥-Al2 O3 systems. Both metallic Pt0 and oxide PtOx species were detected by high-resolution transmission electron microscopy (HRTEM) and X-ray photoelectron spectroscopy (XPS) [7,8]. X-ray absorption fine structure (XAFS) analysis revealed that three types of the Pt/␥-Al2 O3 catalysts with metal, oxide and metal-oxide platinum species were formed depending on a preparation procedure [6,9]. Belyi et al. used chemical adsorption for

∗ Corresponding author at: Boreskov Institute of Catalysis, pr. Lavrentieva 5, 630090 Novosibirsk, Russia. Tel.: +7 383 3269597; fax: +7 383 3308056. E-mail address: [email protected] (V.P. Pakharukova). http://dx.doi.org/10.1016/j.apcata.2014.08.014 0926-860X/© 2014 Elsevier B.V. All rights reserved.

quantitative determination of the metallic and ionic platinum species in the Pt/␥-Al2 O3 catalysts [10,11]. It was found that Pt ions, unlike Pt0 atoms, formed strong ␴-donor bonds with water [10]. Strong interaction with the support was suggested to cause a high thermostability of the ionic platinum species [11]. The data on characterization of the Pt0 particles are unambiguous, but this is not the case with the PtOx species. As a rule, the PtOx species are highly dispersed and disordered that makes it impossible to use traditional X-ray and electron diffraction methods for structural studies. Extended X-Ray Absorption Fine Structure (EXAFS) analysis was used to elucidate the structure of the PtOx species [6,9,12,13]. The EXAFS analysis detected 6-fold oxygen coordination of Pt4+ ions; that implied an octahedral geometry typical of ␣-PtO2 phase [6,9,12]. Considerable reduction and absence of longer Pt–Pt contributions indicated a lack of long-range atomic order in the PtOx species [9,12,13]. Detection of one Pt–O coordination shell did not allow the PtOx species to be unambiguously assigned to the ␣-PtO2 phase. The formation of highly dispersed disordered PtOx clusters on the support surface or coordination of the Pt ions with support oxygen ions seems more reliable. The interface features, metal–support interaction are not clearly understood. Several ideas were suggested, among which were binding the Pt0 atoms to the support oxygen ions [14,15] and incorporating the Pt ions into the alumina surface [16–22]. The commonly accepted opinion is that defects of the support surface are pinning sites for highly dispersed platinum species.

V.P. Pakharukova et al. / Applied Catalysis A: General 486 (2014) 12–18

EXAFS data reported by Koningsberger et al. [14,15] revealed the interaction between the alumina surface and Pt0 clusters as an important factor for the cluster anchoring. In spite of a catalyst prereduction, Pt–O interatomic distance was identified and assigned to coordination of the interfacial Pt0 atoms by the support oxygen ions. In terms of another model, the metal–support interaction consists of insertion of the Pt ions into cation vacancies of the ␥-Al2 O3 structure. A stabilization of the Pt ions in the vacant octahedral sites of the spinel-like alumina structure was suggested recently using RDF of electron density [16,17]. Different characterization techniques also indicated a platinum dissolution in the surface Al2 O3 layers. Some studies [18–20] showed a formation of platinum-alumina complexes, which, in contrast to the crystalline PtO2 particles, were soluble in acids and acetylacetone. The metal–support interaction was detected by temperature-programmed reduction (TPR). The platinum oxide species interacting with the support required high reduction temperature (∼220 ◦ C) and were identified as PtAl2 O4 species, whereas the crystalline platinum oxide particles were reduced at a lower temperature (≤100 ◦ C), which is close to the one for the bulk ␣-PtO2 phase [18,22]. The studies of model Pt/␥-Al2 O3 systems also showed a possibility of the Pt diffusion into the alumina. The Pt nanoparticles were deposited on ␥-Al2 O3 films [23,24]. Scanning tunneling microscopy (STM), low-energy electron diffraction (LEED) and Ion Scattering Spectroscopy (ISS) data indicated the Pt dissolution in the alumina after calcination at 300–530 ◦ C [23]. In summary, data on the structure of the platinum oxide species in the Pt/␥-Al2 O3 catalysts are still deficient; a nature of the metal–support interaction is not well understood. We report here a rigorous structural study of the Pt/␥-Al2 O3 catalysts with different performance to the CH4 oxidation. The local atomic structure of the supported platinum species was determined by RDF of electron density method. This method can reveal atomic arrangement in material in the range of 0.1–2 nm.

2. Experimental

13

2.2. Catalyst characterization High-resolution transmission electron microscopy was used to determine the average size of supported Pt particles. HRTEM studies were carried out using a JEOL JEM-2010 electron microscope with 1.4 A˚ lattice resolution at a 200 kV accelerating potential. The samples were ground, suspended in ethanol and mounted on a copper grid coated with a porous carbon film. Average linear (dl ) and volume-surface using vs ) particle diameters  (d  3were  calculated equations dl = ni di / ni and dvs = ni di / ni di2 , where ni is the number of particles with diameter di . A set of 200–700 particles was analyzed for each sample. The samples after 6 h-aging (430 o C; 1 vol % CH4 , 20.8 vol % O2 , rest N2 ) were studied by X-ray diffraction methods. Earlier, XAFS and HRTEM studies showed an absence of significant changes in the Pt oxidation state and Pt particle sizes after the aging [6,9]. Diffraction experiments were performed using a high-resolution diffractometer at the Siberian Synchrotron and Terahertz Radiation Centre (SSTRC, Budker Institute of Nuclear Physics, Novosibirsk, Russia). The measurements were carried out with step of 0.1◦ in the 2 range of 3–145◦ at a wavelength of 0.703 A˚ or in terms of magnitude of a scattering vector, h = 4 sin()/, in the range of 0.5–17 A˚ −1 . The structural features of the supported platinum species were determined by analyzing radial distribution functions of electron density 4r2 (r), where (r) is the electron density at distance r. These functions were obtained by the Fourier transformation of normalized scattering intensity in the X-ray diffraction [25–28]. The procedure used for calculating RDF (see elsewhere for details [16,26,29]) allows interatomic distances (r) and coordination numbers (CN) to be determined from position and area of peaks in the RDF curve. Accuracy of the r and CN values determination is 0.5% and 5%, respectively. Analysis of RDF provides information about the local atomic structure and makes it possible to identify phases with a crystallite size smaller than 3 nm. A comparison between the experimental RDF and RDF calculated from structural models is usually performed [16,29]. Data required for the model RDFs, such as interatomic distances and coordination numbers of compounds investigated here, were calculated from structural data of the ICSD database [30].

2.1. Catalyst preparation 2.3. Catalyst testing A series of monodisperse 0.75 wt% Pt/␥-Al2 O3 catalysts were prepared using the technique described earlier [6]. Commercial ␥-Al2 O3 (Puralox, Sasol Germany GmbH; 98% purity, SBET = 215 m2 g−1 , grain size 0.25–0.5 mm) was impregnated with aqueous platinum nitrate solutions with different Pt:HNO3 ratios. The freshly impregnated samples were dried at 120 ◦ C for 4 h and then calcined at 400 ◦ C for 4 h or at 600 ◦ C for 2 h. Details of the preparation procedure for all catalysts are shown in Table 1.

Table 1 Characteristics of the Pt/␥-Al2 O3 catalysts under study. Catalyst

Precursor (Pt:HNO3 ratio)

Calcination temperature (◦ C)

dl (nm)

dvs (nm)

DM (%)

Ia IIa IIIa IVb Va

1:0 1:5d 1:5c 1:1 1:5c

400 400 400 400 600

0.6 ± 0.3 1.1 ± 0.3 2.2 ± 0.4 1.3 ± 0.4 8.5 ± 1.8

1.0 1.3 2.7 1.6 9.2

100 84.8 42.2 70.6 12.2

a

Non-pretreated support. Acetic acid pretreated support. c The precursor solution was alkalinized with tetramethylammonium hydroxide up to pH 8.5. d Citric acid was added to the precursor solution. b

Catalysts for the complete methane oxidation were tested using a flow-circulation reactor BI-CATr (Boreskov Institute of Catalysis, Novosibirsk) at atmospheric pressure and a constant temperature of 430 ◦ C. The construction features of the BI-CATr setup were described elsewhere [31,32]. In brief, this set-up provides the gradientless mode and allows measuring the catalytic activity with a high accuracy due to an absence of temperature and concentration gradients in a catalyst bed [33]. The catalyst samples (weight (wcat ) was equal to 1 g) were placed in a steel reactor (i.d. = 25 mm). A multiplicity of reaction mixture circulation through the catalyst bed was varied from 17 to 280. It is sufficient to establish the gradientless mode. The composition of the outlet gas mixture (OGM) was identified in a stationary state, whichstability  was examined within an hour. The concentration of CH4

OGM CCH 4

was analyzed using an express

chromatograph with a thermocatalytic detector, the concentrations of CO2 , H2 O and O2 were analyzed by a gas chromatograph “Chromos GH1000” (“Chromosib”, Russia) equipped with a thermal conductivity detector and two columns (Porapak N, Zeolite Molecular Sieve 5A). The flow rate (U) of the initial gas mixture (IGM) was varied from 50 to 1100 ml/min with a constant composition (1 vol%

14

V.P. Pakharukova et al. / Applied Catalysis A: General 486 (2014) 12–18

Fig. 1. X-ray powder diffraction patterns of the Pt/␥-Al2 O3 catalysts and ␥-Al2 O3 support. Vertical lines indicate positions of reflections for Pt0 phase.

CH4 , 20.8 vol% O2 , rest N2 ). The specific stationary reaction rate R −1 −1 (mlCH4 gcat s ) was calculated for each IGM flow rate taking into account the OGM composition: R=

IGM − C OGM ) × U (CCH CH 4

4

wcat

(1)

The specific stationary reaction rate (R0.5 ) at the same temperature and the OGM composition (0.5 vol% CH4 , 0.5 vol% CO2 , 1 vol% H2 O, 19.8 vol% O2 ) was a criterion for comparing catalytic activities of the samples. The turnover frequency (TOF) value was determined using Eq. (2): TOF =

R0.5 M · DM 0.0075 × 22, 400

(2)

where M is the Pt molecular mass, 0.0075 is the platinum loading and 22,400 is the factor converting the ml of gas to the moles. A number of surface Pt atoms was calculated in assumption that the metal particles are spherical. Thus, the dispersion of platinum species DM can be calculated using Eq. (3) [34]: DM =

6 × Vm am × dvs

(3)

where Vm is the atomic volume (nm3 ) and am is the average surface area occupied by one metal atom (nm2 ) [6]. 3. Results and discussion The details of the preparation procedure as well as the average sizes of Pt particles for all catalysts are summarized in Table 1. Variations in the preparation procedure allow one to obtain catalysts containing the Pt particles with different sizes. Fig. 1 shows fragments of X-ray powder diffraction patterns of the Pt/␥-Al2 O3 catalysts and ␥-Al2 O3 support. The diffraction peaks of the ␥-Al2 O3 phase with a spinel-like structure (space group ¯ Fd3m, a = 0.793 nm) are predominant. The average crystallite size of ␥-Al2 O3 is 4.5 nm. The XRD patterns of the catalysts exhibit no peaks corresponding to the platinum phases. The diffraction peaks of the Pt0 phase are detected only for sample V containing well crystallized Pt0 particles (Fig. 1, Table 1). The absence of reflections attributable to the oxide or metal platinum phases implied a highly dispersed state of platinum species. HRTEM data (Table 1) were in agreement with this conclusion. Additional information on the state and local structure of platinum species in the catalysts was obtained using RDF of electron density analysis. Fig. 2 shows the RDF of the support in comparison with model ones for the ␥-Al2 O3 phase.

Fig. 2. Experimental RDF of the ␥-Al2 O3 support (1) in comparison with model ones for ␥-Al2 O3 structural formulas A∗1.75 A5.50 B∗2.00 B9.00 O32 (2) and A∗0.91 A6.90 B∗0.45 B13.05 O32 (3).

The ␥-Al2 O3 oxide is known to have a spinel structure (A8 B16 O32 ) [35]. Disorder and cation vacancies are intrinsic to the nanocrystalline alumina. Oxygen sublattice is fully occupied, while Al ions and vacancies are distributed over tetrahedral (Td —A) and octahedral (Oh —B) positions. Several authors considered alumina as oxyhydroxide and showed that in addition to ideal spinel sites, Al ions occupy also non-spinel (A* ) and (B* ) sites, a part of O atoms is substituted by OH groups. The structure was described by a general formula A∗x1 Ax2 B∗y1 By2 O32−z (OH)z [36,37]. The model RDFs were calculated for the formulas A∗1.75 A5.50 B∗2.00 B9.00 O22.75 (OH)9.25 [36] and A∗0.91 A6.90 B∗0.45 B13.05 O32 [37] with different cation distribution. The former model with a larger fraction of the non-spinel sites and lower filling of octahedral and tetrahedral sites is clearly preferred for describing the local structure of the alumina under study (see Fig. 2). Thus, the ␥-Al2 O3 support had defective structure with a great number of the cation vacancies. It is difficult to analyze RDF of electron density for composite catalysts, because it includes information about both the support and the supported components. Differential RDFs (d-RDFs) between RDFs of the catalysts and support were used for the analysis. As was shown before [16,38,39], this approach allows the local structure of supported nanoparticles to be determined. To identify the supported platinum species in the Pt/␥-Al2 O3 catalysts we considered an atomic arrangement in possible platinum phases. First, catalysts with the lowest and highest platinum dispersion were considered. Fig. 3 compares the experimental d-RDF of sample V and model RDFs calculated for the Pt0 (ICSD No. 77944) and

Fig. 3. Experimental d-RDF describing the Pt local arrangement in catalyst V compared with model RDFs for the Pt0 and ␣-PtO2 phases.

V.P. Pakharukova et al. / Applied Catalysis A: General 486 (2014) 12–18

15

Fig. 4. Experimental d-RDF describing the Pt local arrangement in catalyst I compared with model RDFs for the ␣-PtO2 and Pt0 phases.

Fig. 5. Experimental d-RDF describing the Pt local arrangement in catalyst I compared with calculated RDFs for models 1 and 2.

␣-PtO2 (ICSD No. 24922) phases. Positions of maxima correspond directly to the interatomic distances; peak areas are related to the coordination numbers. X-ray derived d-RDFs are dominated by the interatomic distances between heavy Pt atoms: the Pt–Pt contributions are the most intense; the O–O contributions are negligible. The analysis of d-RDF shows that the short-range atomic order is equivalent to that of the Pt0 phase. The Pt–Pt coordination peaks corresponding to the Pt0 structure are well resolved. The coordina˚ tion peaks typical of the PtO2 structure (r = 2.0, 3.1, 5.3, 6.9, 7.6 A) are not observed. The data showed clearly the metal state of platinum in V catalyst. This result is supported by XRD data (Fig. 1). It is important that the absence of any artifacts or additional contributions in the experimental d-RDF is an evidence of no changes in the support during the catalyst preparation. Fig. 4 shows the experimental d-RDF of catalyst I. The observed atomic correlations differ from the local atomic arrangement in the Pt0 or ␣-PtO2 phases. The absence of the Pt––Pt contributions ˚ indicates corresponding to the Pt0 phase (r = 3.9; 4.8; 5.5; 6.2 A) that the catalyst does not contain Pt0 particles or clusters. The first peak at 1.98 A˚ relates to short Pt-O interatomic distance. The calculated Pt–O coordination number (CN = 5.9) indicates the octahedral geometry as that in the ␣-PtO2 phase. The strong maxima corresponding to the Pt–Pt coordination shells in the ␣-PtO2 phase are not observed in the experimental d-RDF. Thus, analysis of d-RDF revealed that the platinum oxide species in the catalyst were not the ␣-PtO2 phase. Additional contributions in the experimental d-RDF can result from interaction of the platinum species with the support surface. Recently, some evidences of the metal–support interaction in the Pt/␥-Al2 O3 catalysts prepared in the same way were reported. H2 TPR and in situ XAFS studies detected retarded reduction of the PtOx species interacting with the support [40]. Small-angle X-ray scattering (SAXS) data also revealed the interaction between the platinum species and alumina [41]. The experimental d-RDF curve exhibits several peaks at dis˚ which tances r = 1.98, 2.81, 3.40, 4.45, 4.98, 5.98, 6.49, 7.18, 7.50 A, are close to the interatomic distances in the alumina structure. In fact, the spinel-like alumina structure is characterized by distances r = 2.81, 4.86, 5.61, 6.27, 7.43 A˚ (B–B distances), r = 3.43, 5.61, 6.58 A˚ (A–A distances), r = 3.29, 5.15, 6.50 A˚ (A–B distances), ˚ (B–O r = 1.93, 3.34, 3.46, 4.41, 4.50, 5.86, 5.93, 6.00, 6.53, 6.59 A) distances), r = 1.81, 3.31, 4.26, 5.08, 5.25, 5.89, 6.51, 7.05, 7.16, ˚ (A–O distances). As mentioned above, platinum dissolu7.57 A) tion in the surface ␥-Al2 O3 layers was suggested [16–23] with the stabilization of the Pt ions in the vacant octahedral sites of the spinel-like structure [16,17]. Defective structure of the alumina under study was shown. From the accepted structural formula A∗1.75 A5.50 B∗2.00 B9.00 O22.75 (OH)9.25 , 44% of the Oh cation sites are

vacant. Therefore, we studied the possibility of occupying the cation vacancies in the alumina surface by the Pt ions. In this case, the d-RDF should exhibit the strong Pt–Pt contributions reflecting distances between the Pt ions located in the Oh sites (B–B distances). The Pt–O (B–O distances) and Pt–Al (B–B, A–B distances) contributions arising from bonding with the alumina atoms are also expected. The nanocrystalline ␥-Al2 O3 oxides expose predominantly (1 1 0) faces [35]. Recently, it was shown that interaction regions exist as remote islands rather than as a continuous layer [11,42]. We considered models with random distribution of the Pt ions over the vacant Oh sites within regions 1 × 1 × 0.5 nm in size on the (110) alumina face. Model 1 and model 2 assumed filling 100% and 30% of the vacancies, respectively. Fig. 5 shows the experimental d-RDF in comparison with the calculated RDFs. The peak positions match very well. The calculated RDF for model 2 is characterized by a reduction of the Pt-Pt contributions due to decrease in a probability of Pt–Pt bonds. A better agreement between the experimental and model RDFs is provided with the assumption of partial occupation of the vacancies. Some discrepancies are nevertheless observed. The experimen˚ This tal d-RDF features a stronger coordination peak at 3.4 A. distance is characteristic of the atomic ordering around cations in the Td sites of the alumina structure (A–A, A–B distances). Therefore, the model assuming an insertion of the Pt ions into the vacant tetrahedral sites along with the octahedral ones on the alumina surface was also considered. According to model 3, the Pt cations occupy 30% of the vacant Oh and Td sites at the 4:1 ratio.

Fig. 6. Experimental d-RDF describing the Pt local arrangement in catalyst I in comparison with RDFs for models 2–4. Model 3 assumes occupying of 30% of the vacant Oh and Td sites at the 4:1 ratio. Model 4 is derived from Model 3 and takes into account the formation of ␣-PtO2 phase.

16

V.P. Pakharukova et al. / Applied Catalysis A: General 486 (2014) 12–18

Fig. 7. (A) Experimental d-RDF describing the Pt local arrangement in catalysts I (1), II (2), III (3), IV (4). (B) Comparison of d-RDFs of catalysts III (3) and IV (4) with calculated RDFs for PtOx -Al2 O3 species (6), Pt0 phase (7) and Pt0 + (PtOx -Al2 O3 ) composition (5).

As shown in Fig. 6, the model 3 is more appropriate for describing the observed atomic ordering. It should be noted that the presence of platinum oxide ␣-PtO2 in this sample cannot be excluded. The peak at r ∼ 5.36 A˚ may be assigned to the Pt–Pt bond in the ␣-PtO2 structure. The model RDF taking into account the ␣PtO2 formation is close to the experimental d-RDF (Fig. 6). However, a relative amount of the ␣-PtO2 phase does not exceed 10%. Thus, the RDF studies did not reveal forming a significant quantity of the ␣-PtO2 particles in catalyst I. The identified local structural configuration of supported platinum species was in good agreement with the incorporation of the Pt ions into the cation vacancies in the ␥-Al2 O3 surface. Platinum clusters and isolated ions seemed to be strongly bound with the alumina surface (PtOx Al2 O3 ). The situation was different for II, III and IV catalysts. The observed atomic correlations also suggested the incorporation of the Pt ions into the cation vacancies in the alumina surface. The experimental d-RDFs exhibit the coordination peaks at 1.98, 2.81, ˚ which are assigned to the interatomic distances of the 3.40 A, spinel-like phase (Fig. 7(A)). However, peaks at 5.0, 6.5 A˚ are reduced, while peaks at 2.8, 4.4, 6.0 A˚ are intensified. In addition, the coordination peaks at 4.4 and 6.0 A˚ are shifted to higher r values. These features may be assigned to formation of the Pt0 phase and are more pronounced with samples III and IV. The interatomic distances r = 2.76, 4.79, 6.19 A˚ are typical of the Pt0 phase. Model RDF was calculated, the presence of the metal Pt0 phase was taken into account. A good agreement between the experimental and calculated curves was observed for samples III and IV (Fig. 7(B)). The estimated amount of the Pt0 phase in these catalysts was ca. 30%. There was a lower amount of the Pt0 phase in sample II. This conclusion is consistent with the HRTEM data (Table 1). An increase in the amount of the Pt0 phase is associated with a decrease in the metal dispersion DM . The RDF data demonstrated clearly that there were two kinds of platinum in catalysts II, III and IV: the Pt0 particles and the highly dispersed clusters or Pt ions interacting with alumina. The data on the oxidation state of the platinum species in the catalysts under study are fully consistent with recent XAFS results [9,43]. XAFS studies also established formation of three types of Pt species (metal, oxide, oxide and metal) on the alumina surface and identified the octahedral oxygen coordination of the Pt4+ ions in the PtOx species. Reported amount of the Pt0 phase in catalyst IV (22%) is close to our estimation. The present data provided an insight into the metal–support interaction and structure of the PtOx species. The cation vacancies on the ␥-Al2 O3 surface behaved as pinning sites; the vacant octahedral sites were essential to anchor the Pt ions. This conclusion can be used to explain the origin of anomalous Pt–Pt distances observed

at direct imaging of individual Pt atoms on the (1 1 0) alumina faces [44,45]. Isolated platinum atoms, small Pt2 and Pt3 clusters on the ␥-Al2 O3 surface were resolved using Z-contrast imaging in scanning transmission electron microscopy (Z-STEM, where Z denotes nuclear charge). The triangular trimmers Pt3 exhibited atypical Pt–Pt distances: the short side was 2.8 ± 0.2 A˚ in length; the long ˚ The Pt–Pt distances in Pt2 dimmers also were sides were 3.3 ± 0.2 A. 2.8 ± 0.2 and 3.3 ± 0.2 A˚ in length. The observed distances [44,45] are close to typical distances between cations in the alumina structure and coincide with the experimental Pt–Pt distances observed here. The short distance is similar to the B–B interatomic distance, whereas long ones are consistent with the A-B distances in the alumina lattice. This is seen in Fig. 8, which shows a schematic top view of the alumina (1 1 0) face. It seems like the detected configuration of the supported Pt atoms was constrained to match the surface structure of the alumina. Thus, the Z-STEM data also indicate the absorption of the Pt ions into the first layer of the ␥-Al2 O3 lattice. It should be noted that first-principles density functional theory calculations revealed that H atoms in the alumina structure occupy the Td cation vacancies [46] and –OH capped Pt adatom structures are thermodynamically preferred over substitutional Pt structures in the Pt/␥-Al2 O3 system [45]. Unfortunately, X-ray diffraction techniques have a low sensitivity to precise location of light atoms; the Pt–Pt, Pt–Al contributions are the most intense in the RDFs. This limitation does not allow us to conclude unambiguously if there is an insertion of the Pt ions in the alumina surface layers or there is an ion-exchange with the surface –OH groups with a forming adatom Pt structures. However, our data clearly demonstrated that the Pt ions ordering resembled an arrangement of the cations in the alumina structure. The cation vacancies on the ␥-Al2 O3 surface were essential to anchor the Pt ions and clusters. Reported correlation between the concentrations of electron deficient platinum atoms and cation vacancies in the alumina confirms a key role of the vacancies in stabilization of the Pt ions on the alumina surface [17]. All the results showed that a variation in the preparation techniques caused the formation of a variety of platinum species on the ␥-Al2 O3 surface. The Pt ions could be pinned to the cation vacancies on the alumina surface. Nevertheless, a platinum segregation with formation of the Pt0 phase was evident in some systems. The state of supported platinum in the Pt/␥-Al2 O3 catalyst under study can be described as Pt0 + PtOx -Al2 O3 . The concentration of nitrate anions in the precursor solution was mentioned above [9] to be responsible for the growth of the Pt0 particles on the ␥-Al2 O3 surface (Table 1). Large platinum-containing aggregates are formed in the precursor solution due to the intensive hydrolysis process. The metal particles arise from these aggregates

V.P. Pakharukova et al. / Applied Catalysis A: General 486 (2014) 12–18

17

Fig. 8. Schematic top view of two (1 1 0) surfaces possible for the ␥-Al2 O3 oxide.

under the following temperature treatment. The addition of citric acid into the precursor solution results in suppression of hydrolysis and prevents the aggregate enlargement. Thus, catalyst II was characterized by a lower proportion of the metallic platinum and smaller metal dispersion. Sample IV contained the Pt0 phase in a considerable quantity, hence the acid pretreatment of the support also led to segregation of the metal phase. Increase in the acid strength of the support surface favors formation of the Pt0 phase [47]. Moreover, adsorbed acetic acid may serve as a reducing agent under the thermal treatment during the catalyst preparation. The calcination temperatures were only different at the preparation of samples III and V (400 and 600 ◦ C, respectively). While the Pt0 and PtOx -Al2 O3 species coexisted in catalyst III, the Pt0 phase was only detected in catalyst V. Obviously, the thermal treatment at 600 ◦ C led to elimination of the metal–support interaction and segregation of the Pt0 particles. The result is consistent with previous observations that treatment at temperatures ≥600 ◦ C led to decomposition of the PtOx -Al2 O3 species and formation of the Pt0 phase [12,18]. Notice that eliminating the interaction was accounted for by considerable sintering of Pt0 particles. Hence, it is reasonable to suppose that the Pt dispersing relates to the interaction with the support. The catalytic activities of the Pt/␥-Al2 O3 catalysts to the CH4 oxidation were studied. The steady-state catalytic activities are listed in Table 2. Appreciable differences in the catalytic activity of the catalysts indicate that the process under study is strongly structuresensitive. Considering the TOF values demonstrates clearly that the state of supported platinum is an important parameter for determining the activity of the alumina-supported platinum catalysts to the methane oxidation. Maximal catalytic activity was observed with the catalysts containing both the alumina stabilized PtOx species and highly dispersed Pt0 particles, while a lower TOF was observed with sample I containing the PtOx species only. The worst performance was characteristic of catalyst V with large Pt0 crystallites. Thus, the catalytic activity of platinum appears to depend

on a PtOx /Pt0 ratio. The results obtained agree with the models suggested earlier [6,48,49], which proposed that hydrocarbon combustion was catalyzed by oxidized platinum complexes interacting with alumina surface as well as by the Pt0 crystallites [48]. The latter species were more active. Hicks et al. [49] observed that the most active catalysts contained the crystalline Pt0 particles, whereas catalysts with a sufficient amount of dispersed PtO2 species were less active. The measured TOF values were 0.08 and 0.005 s−1 , respectively. In our case, a smaller TOF was also observed for sample I containing the oxidized PtOx -Al2 O3 species only. A part of the platinum captured by the support surface seemed to be inaccessible for reagent molecules. More recent studies clarified that partially oxidized platinum particles accounted for high catalytic activity rather than metal platinum [50,51]. Burch reported that the partially oxidized surface of Pt particles was preferred for the activation of C–H bonds in the CH4 molecules. The pure metal surface was considered as less active because of the absence of adsorbed oxygen for reaction with adsorbed CHx species. The completely oxidized Pt surface was less active because of slow dissociative adsorption of CH4 on the fully oxygen covered Pt surface [50]. Obviously, a degree of surface oxidation depends on the reaction conditions as well as on the metal dispersion. The correlation between oxidation extent and the Pt0 particle size was shown using TPR experiments [52]. XANES studies also revealed that large Pt0 particles were less oxidized than the small ones upon their treatment with flowing oxygen at 550 ◦ C [53]. In our study, the Pt0 crystallites with d ∼ 1–2 nm are expected to oxidize more easily than the Pt0 particles of 8.5 nm in size. Catalyst V containing large metal particles, as expected, exhibited the lowest catalytic activity. The highly dispersed Pt0 species were responsible for high activity to the methane oxidation. The sufficient activity of the catalysts containing platinum anchored to the alumina surface indicated the importance of the metal–support interaction. The anchorage improved indirectly the performance of the platinum catalysts by stabilizing Pt dispersion.

Table 2 Comparison of the catalyst performance for the CH4 oxidation. Catalyst

dl (nm)

DM (%)

−1 R0.5 (mlCH4 g−1 cat s )

TOF (s−1 )

Identified platinum species

I II III IV V

0.6 ± 0.3 1.1 ± 0.3 2.2 ± 0.4 1.3 ± 0.4 8.5 ± 1.8

100 84.8 42.2 70.6 12.2

0.0164 0.0252 0.0104 0.0187 0.0003

0.0190 0.0345 0.0286 0.0308 0.0029

PtOx -Al2 O3 Pt0 + PtOx -Al2 O3 Pt0 + PtOx -Al2 O3 Pt0 + PtOx -Al2 O3 Pt0

18

V.P. Pakharukova et al. / Applied Catalysis A: General 486 (2014) 12–18

In summary, the structural features of non-metallic PtOx species in the Pt/␥-Al2 O3 catalysts were identified using RDF of electron density. The obtained structural data are useful for understanding the structure–activity relationships of the aluminasupported platinum catalysts for a number of processes and for the knowledge-based design of catalytic materials. 4. Conclusions The local atomic structure of supported platinum in the Pt/␥Al2 O3 catalysts was studied using RDF of electron density. Both the metal Pt0 and oxide PtOx species were formed depending on the catalyst preparation conditions. The data provided an insight into the structure of the PtOx species. The results supported recent suggestions about interaction between platinum and alumina. In particular, the Pt ions were anchored to the cation vacancies on the ␥-Al2 O3 surface. The highly dispersed platinum clusters and electron deficient atoms were strongly bound to the alumina surface. Catalytic properties of the Pt/␥-Al2 O3 samples for the CH4 oxidation were inspected with respect to the revealed structural features. The high activity to the methane oxidation was accounted for by the coexistence of the alumina stabilized platinum oxide species and highly dispersed Pt0 particles. Potentialities of the method of RDF of electron density for determining the local atomic arrangement in supported nanoparticles and nanoparticle–support interface in catalytic materials were demonstrated. Acknowledgments The authors are grateful to V.I. Zaikovskii and I.E. Beck† for experimental contribution. The work was performed in the framework of the joint Research and Educational Center for Energoefficient Catalysis (Novosibirsk State University, Boreskov Institute of Catalysis). The structural studies were supported by the RSCF project No. 14-23-00037. Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.apcata.2014.08.014. References [1] P. Gelin, M. Primet, Appl. Catal. B: Environ. 39 (2002) 1–37. [2] T.V. Choudhary, S. Banerjee, V.R. Choudhary, Appl. Catal. A: Gen. 234 (2002) 1–23. [3] G.J. Antos, A.M. Aitani, J.M. Parera, Catalytic Naphtha Reforming Science and Technology, Marcel Dekker, New York, 1995. [4] M.J. Kahlich, H.A. Gasteiger, R.J. Behm, J. Catal. 171 (1997) 93–105. [5] E.D. Park, D. Lee, H.C. Lee, Catal. Today 139 (2009) 280–290. [6] I.E. Beck, V.I.I Bukhtiyarov, Yu. Pakharukov, V.I. Zaikovsky, V.V. Kriventsov, V.N. Parmon, J. Catal. 268 (2009) 60–67.

[7] A.S. Ivanova, E.M. Slavinskaya, R.V. Gulyaev, V.I. Zaikovskii, O.A. Stonkus, I.G. Danilova, L.M. Plyasova, I.A. Polukhina, A.I. Boronin, Appl. Catal. B: Environ. 97 (2010) 57–71. [8] M. Pan, J.M. Cowley, I.Y. Chan, J. Appl. Cryst. 20 (1987) 300–305. [9] I.E. Beck, V.V. Kriventsov, D.P. Ivanov, V.I. Zaikovsky, V.I. Bukhtiyarov, Nucl. Instrum. Methods A 603 (2009) 108–110. [10] A.S. Belyi, Kinet. Catal. 49 (2008) 562–567. [11] A.S. Belyi, React. Kinet. Catal. Lett. 57 (1996) 349–356. [12] P. Lagarde, T. Murata, G. Vlaic, E. Freund, H. Dexpert, J.P. Bournonville, J. Catal. 84 (1983) 333–343. [13] A. Boubnov, S. Dahl, E. Johnson, A.P. Molina, S.B. Simonsen, F.M. Cano, S. Helveg, L.J. Lemus-Yegres, J.-D. Grunwaldt, Appl. Catal. B: Environ. 126 (2012) 315–325. [14] D.C. Koningsberger, D.E. Sayers, Solid State Ionics 16 (1985) 23–27. [15] M.J. Vaarkamp, T. Miller, F.S. Modica, D.C. Koningsberger, J. Catal. 163 (1996) 294–305. [16] E.M. Moroz, V.P. Pakharukova, A.N. Shmakov, Nucl. Instrum. Methods A 603 (2009) 99–101. [17] E.M. Moroz, D.A.V Zyuzin, Yu. Tregubenko, I.E. Udras, A.S. Belyi, V.A. Likholobov, React. Kinet. Mech. Catal. 110 (2013) 459–470. [18] H.C. Yao, M. Sieg, H.K. Plummer Jr., J. Catal. 59 (1979) 365–374. [19] R.A. Herrmann, S.F. Adler, M.S. Goldstein, R.M. DeBaun, J. Phys. Chem. 65 (1961) 2189–2194. [20] M.F.L. Johnson, C.D. Keith, J. Phys. Chem. 67 (1963) 200–201. [21] G. Lietz, H. Lieske, H. Spindler, W. Hanke, J. Volter, J. Catal. 81 (1983) 17–25. [22] C.-P. Hwang, C.-T. Yeh, J. Mol. Catal. A: Chem. 112 (1996) 295–302. [23] Th. Bertrams, F. Winkelmann, Th. Uttich, H.-J. Freund, H. Neddermeyer, Surf. Sci. 331–333 (1995) 1515–1519. [24] L. Li, Z. Zhang, J.R. Croy, S. Mostafa, B.R. Cuenya, A.I. Frenkel, J.C. Yang, Microsc. Microanal. 16 (2010) 1192–1193. [25] B. Warren, E. X-ray Diffraction, Addison-Wesley, Reading, Massachusetts, 1969. [26] E.M. Moroz, Russ. Chem. Rev. 80 (2011) 293–312. [27] R.B. Neder, V.I. Korsunskiy, J. Phys.: Condens. Matter 17 (2005) 125–134. [28] T. Egami, in: S.J.L. Billinge, M.F. Thorpe (Eds.), Local Structure from Diffraction, Plenum, New York, 1998. [29] V.P. Pakharukova, E.M. Moroz, D.A. Zyuzin, J. Struct. Chem. 51 (2010) 319–323. [30] Inorganic Crystal Structure Database (ICSD-for-WWW), Fachinformationszentrum (FIZ), Karlsruhe, Germany, 2007. [31] Yu.I Pakharukov, N.N. Bobrov, V.N. Parmon, Catal. Ind. 1 (2009) 38–42. [32] Yu.I Pakharukov, N.N. Bobrov, V.N. Parmon, Catal. Ind. 1 (2009) 95–101. [33] N.N. Bobrov, V.N. Parmon, NATO Science Series II: Mathematics Physics and Chemistry, vol. 69, Kluwer, 2002. [34] J.R. Anderson, Structure of Metallic Catalysts, Academic Press, London, 1975. [35] S.V. Tsybulya, G.N. Kryukova, Phys. Rev. 77 (2008) 024112–024125. [36] V.A. Ushakov, E.M. Moroz, React. Kinet. Catal. Lett 24 (1984) 113–118. [37] L. Smrcok, V. Langer, J. Krestan, Acta Cryst. C 62 (2006) 83–84. [38] V.P. Pakharukova, E.M. Moroz, V.V. Kriventsov, T.V. Larina, A.I.L Boronin, Yu. Dolgikh, P.E. Strizhak, J. Phys. Chem. C 113 (2009) 21368–21375. [39] P.J. Chupas, K.W. Chapman, H. Chen, C.P. Grey, Catal. Today 145 (2009) 213–219. [40] A.A. Veligzhanin, Ya.V. Zubavichus, A.A. Chernyshov, A.L. Trigub, A.S. Khlebnikov, A.I. Nizovskii, A.K. Khudorozhkov, I.É. Beck, V.I. Bukhtiyarov, J. Struct. Chem. 51 (Suppl.) (2010) 20–27. [41] F.V. Tuzikov, I.É. Beck, N.A. Tuzikova, V.I. Bukhtiyarov, J. Struct. Chem. 51 (Suppl.) (2010) 1–10. [42] V.A. Ushakov, E.M. Moroz, P.A. Zhdan, A.I. Boronin, N.R. Bursian, S.B. Kogan, E.A. Levitskii, Kinet. Catal. 19 (1978) 744–748. [43] I.É. Beck, V.V. Kriventsov, D.P. Ivanov, E.P. Yakimchuk, B.N. Novgorodov, V.I. Zaikovskii, V.I. Bukhtiyarov, J. Struct. Chem. 51 (Suppl.) (2010) 11–19. [44] P.D. Nellist, S.J. Pennycook, Science 274 (1996) 413–415. [45] K. Sohlberg, S. Rashkeev, A.Y. Borisevich, S.J. Pennycook, S.T. Pantelides, Chem. Phys. Chem. 5 (2004) 1893–1897. [46] K. Sohlberg, S.J. Pennycook, S.T. Pantelides, J. Am. Chem. Soc. 121 (1999) 7493–7499. [47] Y. Yazawa, N. Takagi, H. Yoshida, S. Komai, A. Satsuma, T. Tanaka, S. Yoshida, T. Hattori, Appl. Catal. A: Gen. 233 (2002) 103–112. [48] J. Volter, G. Lietz, H. Spindler, H. Lieske, J. Catal. 104 (1987) 375–380. [49] R.F. Hicks, H. Qi, M.L. Young, R.G. Lee, J. Catal. 122 (1990) 280–294. [50] R. Burch, M.J. Hayes, J. Mol. Catal. A: Chem. 100 (1995) 13–33. [51] E. Becker, P.-A. Carlsson, H. Grönbeck, M. Skoglundh, J. Catal. 252 (2007) 11–17. [52] R.W. McCabe, C. Wong, H.S. Woo, J. Catal. 114 (1988) 354–367. [53] Y. Yazawa, H. Yoshida, T. Hattori, Appl. Catal. A: Gen. 237 (2002) 139–148.