Tomographic Heating Holder for In Situ TEM: Study of Pt/C and PtPd/Al2O3 Catalysts as a Function of Temperature

1 TomographicheatingholderforinsituTEM: StudyofPt/CandPtPd/Al2O3catalysts asafunctionoftemperature Lionel C. Gontard,1,* Rafal E. Dunin-Borkowski,2 Asunción Fernández,1 Dogan Ozkaya,3 and Takeshi Kasama4 1 Instituto de Ciencia de Materiales de Sevilla (CSIC), 41092, Sevilla, Spain 2 Ernst Ruska-Centre for Microscopy and Spectroscopy with Electrons (ER-C) and Peter Grünberg Institute (PGI), Forschungszentrum Jülich, D-52425 Jülich, Germany 3 Johnson Matthey Technology Centre, Blount’s Court, Sonning Common, Reading RG4 9NH, UK 4 Center for Electron Nanoscopy, Technical University of Denmark, DK-2800 Kgs. Lyngby, Denmark Abstract: A tomographic heating holder for transmission electron microscopy that can be used to study supported catalysts at temperatures of up to ~1500 °C is described. The specimen is placed in direct thermal contact with a tungsten filament that is oriented perpendicular to the axis of the holder without using a support film, allowing tomographic image acquisition at high specimen tilt angles with minimum optical 2 shadowing. We use the holder to illustrate the evolution of the active phases of Pt nanoparticles on carbon black and PtPd nanoparticles on γ-alumina with temperature. Particle size distributions and changes in active surface area are quantified from tilt series of images acquired after subjecting the specimens to increasing temperatures. The porosity of the alumina support and the sintering mechanisms of the catalysts are shown to depend on distance from the heating filament. Keywords: in situ TEM, TEM specimen holder, particle growth, heterogeneous catalysis, platinum, carbon black, platinum-palladium, alumina * email: [email protected] . Phone: +34 954489576 1. Introduction When studying heterogeneous catalysts, which frequently comprise metal nanoparticles on carbon black or an oxide support, an important goal is to observe the evolution of the active phase with temperature in the presence of reactant gases. Catalytic treatment can lead to changes in the surface area of the supported metal particles and hence catalytic activity due to processes such as sintering. In order to understand and control catalyst production, it is necessary to study how the dispersion and morphology of both the nanoparticles and the support depend on reaction conditions. Although phase diagrams of small metal particles have been proposed theoretically based on thermodynamics (Ajayan & Marks, 1988), they are incapable of predicting nanoparticle shapes that result from purely kinetic considerations. Recently, Barnard et al. (2009) proposed a quantitative phase map for gold nanoparticles based on ab initio calculations. However, a further complication is that such thermodynamic models do not take into account the 3 role of the support, which determines the strength of attachment and thereby particle mobility and also the posibility of phase transformations (Bett et al., 1974; Tauster, 1987; Sepúlveda et al., 1998; Gónzalez et al., 2009). In situ transmission electron microscopy (TEM) refers to techniques that allow direct observation of dynamic properties at the nanoscale through imaging, spectroscopy or diffraction. Such techniques can be used to understand the kinetics of phase transitions in heterogeneous catalysts and the importance of sintering mechanisms (Benavidez et al., 2012; Hansen et al, 2013). Possible approaches involve injecting gas directly into an environmental TEM (ETEM) (Hansen et al., 2002) and/ or using dedicated sample holders that contain hot stages, gas or liquid cells, electrical contacts, fiber optics and lasers for illuminating or heating the specimen while images are acquired with atomic spatial resolution (Kamino et al., 2005; Creemer et al., 2010; Cavalca et al., 2012; Merhareen et al., 2013). In situ experiments that have made use of such instrumentation are nowadays common. Most recent results have included studies of the role of promoters in Ru-based heterogeneous catalysts (Hansen et al., 2002), the oxidation of In, the reoxidation of Si and the formation of graphene on MgO-supported Pt nanoparticles (Kamino et al., 2005; Peng et al., 2012), as well as photodegradation of Cu2O and photodeposition of Pt on a GaN:ZnO photocatalyst (Cavalca et al., 2012). Here, we describe a TEM specimen holder that provides direct contact between a specimen and a filament that has a small thermal mass. The filament reaches a temperature that depends on the current flowing through it. There is then no need to use a support film, which can prevent effective thermal transfer and degrade image resolution. Kamino et al. (2005) developed a similar specimen holder, which was 4 equipped with a longitudinal tungsten wire aligned parallel to the rotation axis of the holder. In contrast, the filament in our holder is oriented perpendicular to the holder axis and has a narrow design, permitting acquisition of images at high specimen tilt angles (± 80°) with minimal optical shadowing. This capability allows electron tomography to be used to analyse heterogenous catalysts in three dimensions in combination with high temperature heating experiments. We demonstrate the functionality of the holder by imaging the growth of Pt catalyst nanoparticles on carbon black in situ in the TEM as a function of temperature and by measuring the resulting changes in particle size distribution (PSD) and surface area of the active phase. We also study the formation of bimetallic PtPd particles on γ-Al2O3 and use high-angle annular dark-field (HAADF) scanning TEM (STEM) (Gontard et al., 2008, 2009) to obtain three-dimensional information about the porosity of the support after successive heat treatments. 2. Materials and Methods 2.1. Tomographic heating holder The tomographic specimen holder that we describe here contains a pre-wound tungsten filament taken from a light bulb (Fig. 1). We mount the filament in a removable cartridge, which is clamped onto two electrical contact pads in a specimen holder from E.A. Fischione Instruments, Inc. (Export, PA). We glue the filament onto two pieces of a copper grid using silver epoxy and place a thin electrically-insulating sheet of mica between the copper and the clamp. 5 Samples can be deposited directly onto the wire, either in the form of a dry powder or from solution. The ends of the wire, which has an electrical resistance of approximately 5 Ω and can be replaced between experiments if required, can be accessed electrically from outside the microscope. The wire temperature, which can be calibrated using an optical pyrometer before insertion into the TEM column, proved to be highly reproducible between different filaments. For example, a temperature of ~1000 °C was achieved by passing a current of ~90 mA through the wire. When operated in high vacuum, the filaments were typically found to fail at currents of ~220 mA, corresponding to a temperature of just over 1500 °C. Tungsten wire has high thermomechanical stiffness and therefore does not sag significantly at high temperature, as well as a high melting point of 3422 °C and a low vapour pressure. Experiments were performed with filaments operated for up to 12 hours at currents of up to 150 mA (corresponding to a temperature of ~1300 °C) with no significant signs of tungsten evaporation. 2.2. Pt supported on carbon black Pt particles with high dispersions (> 75 m2 g-1), which were supported on mesoporous high-surface-area carbon blacks (Vulcan XC-72R) Pt (19.1 wt %) and had been reduced in a N2 atmosphere at 900 °C, were provided by the Johnson Matthey Technology Centre in the form of a dry powder (Ralph & Hogarth, 2002; Ozkaya et al., 2003). The samples are used in proton exchange membrane fuel cells, which require high loadings of electrocatalysts. A small amount of powder that had been stored in air was placed directly onto the filament of the heating holder, inserted into a Philips CM300 TEM 6 operated at 300 keV and reheated in situ by passing a current through the filament. The temperature was gradually increased from room temperature to 975 °C in steps of either 25 or 50 °C, allowing the specimen to stabilise between each step before acquiring a bright-field (BF) TEM image. Measurements of PSDs from the recorded images were obtained using dedicated image processing software based on adaptive segmentation (Gontard et al., 2011). 2.3. PtPd supported on alumina PtPd nanoparticle formation was studied in situ in the TEM, starting from a dry powder of the precursor (Pt (HN3)4)2 Pd2(EDTA) 2.5 wt. % on γ-Al2O3 (SFC-140). γ-Al2O3 is a metastable polytype of Al2O3, which is used extensively as a catalytic support material because of its high porosity and large surface area. In the present specimen, the formation of PtPd particles was expected to occur at temperatures of between 600 and 850 °C. The experimental procedure was as follows: (1) The powder was placed onto the filament of the heating holder and inserted into a Tecnai F20 TEM. A region of the specimen that could be tilted to high angles without shadowing was identified. A tilt series of HAADF STEM images of the region of interest was acquired at low magnification over a specimen tilt range of -52° to +42° using a 4° tilt increment; (2) In situ heating was performed in a Philips CM300 TEM, with the temperature of the filament increased from 500 to 850 °C, allowing the specimen to equilibrate between steps. BF TEM images were acquired from the same area at high magnification at each temperature; (3) The specimen was allowed to cool to room temperature, the holder was reinserted into the Tecnai F20 TEM and a new tilt series of HAADF STEM images was acquired from the same area. Alignment and tomographic reconstruction using the 7 simultaneous iterative reconstruction technique (SIRT) were carried out in Inspect3D software (15-20 iterations). Voltex visualization of the final reconstructions was performed using Amira V2.3 software. PSDs were measured using image processing software based on adaptive segmentation (Gontard et al., 2011). 3. Results 3.1. Sintering of Pt particles on C Figure 2 shows a series of BF TEM images, which illustrate the sintering of Pt particles supported on carbon black during in situ heating at increasing temperature. Even though the specimen had already been reduced in an inert atmosphere at high temperature, Fig. 2 shows that re-sintering occurs during heating in high vacuum, beginning at ~550 °C, accelerating between 650 and 850 °C and appearing to stop at 850 °C. Faceting of the particles can then be seen, particularly for several larger particles. The shape of the C support does not change appreciably during sintering, although it rotates and translates within the field-of-view. In Figs. 3A and B, circles and arrows are used to highlight differences between the same particles viewed at 500 and 900 °C. Figure 4 shows PSDs of the nanoparticles measured from selected images in Fig. 2. Several statistical parameters are plotted, including mean size, standard deviation and maximum size, for five selected temperatures (500, 650, 725, 875 and 950 °C). The mean particle size was used to estimate the active surface area . This is an important quantity in catalysis, as it describes the surface area of the active phase that is exposed to gases and can be used to understand catalyst deactivation. We made two 8 approximations: (1) That the contact surface of the nanoparticles with the support is exposed to the gas; (2) that the particles have spherical shapes with radii . The surface area is then simply given by the expression      4 ∙4 3   3 ∙   where  is the number of particles,  and  are the surface area and mass of particle  and  is the density of Pt (i.e., 21.450 kg m ⁄). Figure 5 shows measurements and the evolution of  with temperature. The rate of loss of active area calculated from the slope of a linear fit, i.e.,   ⁄910mkg ⁄C. 3.2. Formation of Pt-Pd particles Figure 6A shows TEM images of the precursor (Pt (HN3)4)2 Pd2(EDTA) 2.5 wt. % on γAl2O3 (SFC-140) recorded at room temperature. Figure 6B shows an image of the same region of the specimen recorded after heating at 850 °C. The images on the left side of the figure are HAADF STEM images, in which the local intensity is sensitive to atomic number (brighter regions correspond to particles). The BF TEM images shown in the right column of Fig. 6 were acquired from the regions marked using rectangles in the left column. The images show that the PtPd particles grow at high temperature and that the γ-alumina support increasse in porosity. Figure 6C shows the measured PSDs of the PtPd nanoparticles. Figure 7 shows individual HAADF STEM images recorded at different specimen tilt angles from the same area of the alumina sample before and after heating, taken from a tilt series of images used for electron tomography. Figure 8A shows a voltex 9 visualisation of the three-dimensional reconstruction of this region of the specimen after heating, while Fig. 8B shows an orthoslice generated from the reconstructed volume. Although the tilt step of 4° used for acquisition of the tilt series is relatively large and reduces the spatial resolution of the reconstructed volume, this value was chosen to minimise the total electron dose on the specimen. For example, it has been observed on samples of Pd supported on alumina that under electron irradiation Al2O3 can be reduced and a metastable intermetallic compound of Al2Pd can form (Kamino et al., 1992). In the present experiments, the tilt range used was smaller than the maximum range that the holder allows (± 80°), which also affects the quality of the final reconstruction, in particular in the z direction. 4. Discussion The results shown in Figs. 1 to 8 can be used to discuss the kinetic route that catalyst nanoparticles follow when they grow on a specific support and to determine the relative importance of different sintering mechanisms (Benavidez et al., 2012; Hansen et al., 2013). Three possible sintering mechanisms are: (1) Coalescence, involving motion of the particles over the support and fusion (Ruckenstein & Lee, 1984). In general, at temperatures of 0.5 × the melting point, metallic particles are expected to be in a quasimelted state (continuously fluctuating between different structures), which favours coalescence (Ajayan & Marks, 1988; Barnard et al., 2009). (2) Ostwald ripening, involving the migration of individual atoms from the particles onto the support, across the support and onto other particles (Simonsen et al., 2010). Ostwald ripening is strongly dependent on atomic diffusivity (Linderoth et al., 1997; Horch et al., 1999) and can also happen in faceted particles in the form of inhibited nucleation, which arises 16 Gontard, L.C., Dunin-Borkowski, R.E., Gass, M.H., Bleloch, A.L. & Ozkaya, D. (2009). Three-dimensional shapes and structures of lamellar-twinned fcc nanoparticles using ADF STEM. J Electron Microsc 58, 167-174. Gontard, L.C., Ozkaya, D. & Dunin-Borkowski, R.E. (2011). A simple algorithm for measuring particle size distributions on an uneven background from TEM images. Ultramicroscopy 111, 101-106. González, J.C., Hernández, J.C., López-Haro, M., del Río, E., Delgado, J.J. , Hungría, A.B., Trasobares, S., Bernal, S., Midgley, P.A. & Calvino, J.J. (2009). 3D characterization of gold nanoparticles supported on heavy metal oxide catalysts by haadf-stem electron tomography. Angew Chem Int Ed 48, 5313-5315. Guisbiers, G., Abudukelimu, G. & Hourlier, D. (2011). Size-dependent catalytic and melting properties of platinum palladium nanoparticles. Nanoscale Res Lett 6, 1-5. Hansen, T.W., Wagner, T.W., Hansen, P.L., Dahl, S., Topsoe, H. & Jacobsen, C.J.H. (2002). Atomic-resolution in-situ transmission electron microscopy of a promoter of a heterogeneous catalyst. Science 294, 1508-1509. Hansen, T.W., DeLaRiva, A.T., Challa, S.R. & Datye, A.K. (2013).S intering of catalytic nanoparticles: particle migration or ostwald ripening?. Acc Chem Res 46(8), 1720-1730. Harris, P.J.F. (1986). The sintering of platinum particles in an alumina-supported catalyst: further transmission electron microscopy studies. J Catal 97, 527-542. 17 Harris, P.J.F. (1995). Growth and structure of supported metal catalyst particles. Int Mater Rev 40(3), 97-115. Horch, S., Lorensen, F., Helveg, S., Lægsgaard, E. , Stensgaard, I., Jacobsen, K.W., Nørskov, J.K. & Besenbacher, F. (1999). Enhancement of surface self-diffusion of platinum atoms by adsorbed hydrogen. Nature 398, 134-136. José-Yacamán, M., Gutierrez-Wing, C., Miki, M., Yang, D.-Q., Piyakis, K.N. & Sacher, E. (2005). Surface Diffusion and Coalescence of Mobile Metal Nanoparticles. J Phys Chem B 109, 9703-9711. Kamino, T., Kuroda, K. & Saka, H. (1992) In situ HREM/microanalysis study of reduction of Al2O3 with palladium. Ultramicroscopy 41, 245-248. Kamino, T., Yaguchi, T., Konno, M., Watabe, A., Marukawa, T., Mima, T., Kuroda, K., Saka, H. , Arai, S., Makino, H., Suzuki, Y. & Kishita, K. (2005). Development of a gas injection/specimen heating holder for use with transmission electron microscope. J Electron Microsc 54(6), 497-503. Kamino, T., Yaguchi, T., Sato, T. & Hashimoto, T. (2005). Development of a technique for high resolution electron microscopic observation of nano-materials at elevated temperatures. J Electron Microsc 54(6), 505-508. 18 Linderoth, T.R, Horch, S., Lægsgaard, E., Stensgaard, I. & Besenbacher, F. (1997). Surface diffusion of Pt on Pt(110): Arrhenius behavior of long jumps. Phys Rev Lett 78(26), 4978-4981. Mehraeen, S., McKeown, J.T., Deshmukh, P.V., Evans, J.E., Abellan, P., Xu, P., Reed, B.W., Taheri, M.L., Fischione, P.E. & Browning, N.D. (2013). A (S)TEM gas cell holder with localized laser heating for in-situ experiments. Microsc Microanal 19, 470-478. Mishra, Y.K., Mohapatra, S., Avasthi, D.K., Lalla, N.P. & Gupta, A. (2010). Tailoring the size of gold nanoparticles by electron beam inside transmission electron microscope. Adv. Mat. Lett. 1(2), 151-155. Ozkaya, D., Thompsett, D., Goodlet, G., Spratt, S., Ash, P. & Boyd, D. (2003). Characterisation of C supported Pt nano-particles using HREM. Inst Phys Conf Ser 179, 127. Peng, Z., Somodi, F., Helveg, S., Kisielowski, C., Specht, P. & Bell, A.T. (2012). High-resolution in-situ and ex situ TEM studies on graphene formation and growth on Pt nanoparticles. J Catal 286, 22-29. Ralph, B.T.R. & Hogarth, M.P. (2002). Catalysis for low-temperature fuel cells. Part I: The cathode challenges. Plat Met Rev 46, 3-14. 19 Ruckenstein, E. & Lee, S.H. (1984). Redispersion and migration of Ni supported on alumina. J Catal 86, 457. Sepúlveda-Escribano, A., Coloma, F. & Rodíguez-Reinoso, F. (1998). Platinum catalysts supported on carbon blacks with different surface chemical properties. Appl Catal A 173, 247-257. Simonsen, S.B., Chorkendorff, I., Dahl, S., Skoglundh, M., Sehested, J. & Helveg, S. (2010). Direct observations of oxygen-induced platinum nanoparticle ripening studied by in-situ tem. J Am Chem Soc 132, 7968-7975. Sankaranarayanan, S.K.R.S., Bhethanabotla, V.R. & Joseph, B. (2005). Molecular dynamics simulation study of the melting of Pd-Pt nanoclusters. Phys. Rev. B 71, 195415. Tauster, S.J (1987), Strong metal-support interactions. J Am Chem Soc 4, 170-175. Wynblatt, P. & Gjostein, N.A. (1976). Particle growth in model supported metal catalysts-I. Theory. Acta Mettal 24, 1165-1174. Wynblatt, P. (1976). Particle growth in model supported metal catalysts-II. comparison of experiment with theory. Acta Mettal 24, 1175-1182.  20 Figure 1. (A) Modified TEM specimen holder with tungsten heating filament oriented perpendicular to the holder axis. The numbers correspond to (1) the clamp; (2) an insulating mica sheet; (3) Cu electric contacts; (4) silver epoxy resin. (B-D) Three magnified images of the filament. The large arrow in (D) indicates an aggregate of the deposited specimen. 21 Figure 2. Series of BF TEM images showing sintering of Pt nanoparticles on a carbon black support as a function of temperature between 500 and 975 °C. The specimen was deposited directly onto the tungsten filament in the modified side-entry TEM specimen holder shown in Fig. 1. The dark contrast at the bottom left corner of the images acquired at 925 and 950 °C arises from the closest part of the filament to the region of interest. 22 Figure 3. (A) and (B) Detail extracted from the BF TEM images shown in Fig. 2, illustrating the coalescence of Pt nanoparticles between indicated specimen temperatures of 500 and 975 °C. Red circles in both images show the initial and final configurations of selected Pt particles that have coalesced. White arrows indicate particles that remained strongly attached to the support and experienced smaller changes in their size. (C) Detail of the attachement of a faceted Pt nanoparticle to the support. (D) Model showing crystalline facets and possible Pt(111) contact plane of a Pt nanoparticle such as that shown in (E). (E) Experimental high-resolution TEM image of one Pt particle on graphitised carbon (recorded from a similar specimen) showing atomic defects present at the surfaces (arrows 1 and 2) (Gontard et al., 2007). 23 Figure 4. Top: Particle size distributions (PSDs) of the Pt particles shown in Figs. 2 and 3 measured at five selected temperatures between 650 and 950 °C fitted with lognormal curves. Bottom: Dependence of several statistical parameters (mean size, standard deviation and maximum size) on temperature. The values increase with temperature because of particle sintering. 24 Figure 5. Active surface area of the Pt particles described in Figs. 2-4, measured at five selected temperatures between 650 and 950 °C (see text for details). Althought the evolution of surface area is not linear we used the slope of a linear approximation as an estimate of the rate of loss of active surface area. 25 Figure 6. (A, B) ADF STEM images (left) and BF TEM images (right) of (Pt(HN3)4)2Pd2(EDTA) 2.5 wt% on γ-Al2O3 acquired (A) at room temperature (using the Tecnai F20) and (B) after increasing the temperature to 850 °C (using the Philips CM300) in high vacuum conditions. The PtPd particle size and the porosity of the alumina differ between the marked zones. (C) PSDs measured from zones 2 and 3 and from zone 3 alone, illustrating the formation of two families of particles with different average sizes.