STEM imaging of single Pd atoms in activated carbon fibers considered for hydrogen storage

CARBON

4 9 ( 20 1 1 ) 4 0 5 9–40 7 3

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/carbon

Letters to the Editor

STEM imaging of single Pd atoms in activated carbon fibers considered for hydrogen storage Klaus van Benthem a,*, Cecile S. Bonifacio a, Cristian I. Contescu b, Nidia C. Gallego b, Stephen J. Pennycook b a b

University of California at Davis, Dept. of Chemical Engineering and Materials Science, 1 Shields Ave, Davis, CA 95616, USA Oak Ridge National Laboratory, Materials Science and Technology Division, 1 Bethel Valley Road, Oak Ridge, TN 37831, USA

A R T I C L E I N F O

A B S T R A C T

Article history:

Aberration corrected scanning transmission electron microscopy was used to demonstrate

Received 6 December 2010

the feasibility of imaging individual Pd atoms that are highly dispersed throughout the vol-

Accepted 14 May 2011

ume of activated carbon fibers. Simultaneous acquisition of high-angle annular dark-field

Available online 18 May 2011

and bright-field images allows correlation of the location of single Pd atoms with micro˚ ngstro¨m imaging conditions revealed structural features of the carbon host material. Sub-A that 18 wt% of the total Pd content is dispersed as single Pd atoms in three re-occurring local structural arrangements. The identified structural configurations may represent effective storage sites for molecular hydrogen through Kubas complex formation as discussed in detail in the preceding article. Ó 2011 Elsevier Ltd. All rights reserved.

Nanostructured carbon materials are considered as solid state storage materials for hydrogen, and storage capabilities can significantly be enhanced by adding transition metals to their microstructure [1]. For instance, activated carbon fibers (ACFs) modified with Pd nanoparticles have shown a significant enhancement of H2 adsorption at room temperature [2]. Following the hypothesis of spillover [3], the source of H atoms in the Pd–ACF system is the H-rich phase of Pd-hydride (b-PdH0.66) that is readily formed in presence of hydrogen. Release of hydrogen atoms from b-PdH0.66 particles to the ACF microstructure is controlled by the interface between the particles and the carbon support, which destabilizes the hydride phase [4]. However, isolated transition metal atoms dispersed in the carbon microstructure may have the potential to contribute significantly to hydrogen storage via the formation of Kubas complexes, i.e. the reversible bonding of up to six H2 molecules to the individual metal atom [5,6]. In this letter, we report how atomic resolution scanning transmission

electron microscopy (STEM) can be used to obtain detailed information about the ACF microstructure in the presence of both nanoscale Pd catalyst particles and single Pd atoms dispersed throughout the volume of the ACFs. ACFs containing 2.55 wt.% Pd were synthesized from a mixture of isotropic pitch precursor and palladium acetylacetonate by melt-spinning, followed by stabilization and a combined carbonization and activation process [7,8]. Room temperature hydrogen sorption isotherms showed significantly larger hydrogen uptake than in the absence of Pd despite comparable surface areas [8]. Furthermore, the experimentally determined hydrogen uptake was 25–30% higher than expected for a case in which the entire Pd content is converted to b-PdH0.66 [9]. The objective of this letter is to report the feasibility of localization of individual heavy metal atoms and their bonding configuration with the surrounding carbon microstructure. A detailed quantitative analysis of the hydrogen absorption isotherms and possible storage

* Corresponding author: Fax: +1 530 752 1031. E-mail address: [email protected] (K. van Benthem). 0008-6223/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.carbon.2011.05.016

4060

CARBON

4 9 ( 2 0 1 1 ) 4 0 5 9 –4 0 7 3

mechanisms based on the microstructural characterization presented in this letter is provided in the preceding article in this issue [10]. For STEM investigations, fibers were ground in a methanol slurry and drop deposited onto holey carbon film mounted on TEM specimen grids. High-angle annular dark-field (HAADF) STEM images form by incoherent electron scattering and are sensitive to heavy elements with the scattering cross-section roughly proportional to the squared atomic number Z2 of the scattering element [11]. Bright-field (BF) images are sensitive to lighter elements and provide similar information as high

Fig. 1 – Pair of simultaneously acquired HAADF (a) and BF (b) STEM images. The bright image contrasts in (a) represent Pd particles. The BF image shows quasi-parallel graphene layers (arrowed). Circles mark the locations of two Pd nanoparticles with diameters well below 5 nm.

resolution TEM images. The simultaneous acquisition of HAADF and BF signals in the STEM however provides pixelto-pixel correlation between the two image signals, which makes this technique highly efficient for the investigation of carbon microstructures in the presence of catalytic Pd particles. Experiments were carried out using aberration-corrected dedicated STEM instrumentation operated at 300 keV with ˚ [12]. HAADF STEM imaging reelectron probe sizes below 1 A vealed a median Pd particle diameter of (9.66 ± 1.50) nm throughout the volume of the fibers with roughly 66% of the metal particles smaller than 5 nm. Fig. 1 shows a pair of simultaneously acquired HAADF and BF images exhibiting a Pd nanoparticle with a diameter of roughly 2 nm. The BF image (Fig. 2b) reveals lattice fringes for the ACF. In some areas, local short-range ordering with up to five quasi-parallel graphene sheets can be observed, while in other areas image contrasts indicate disordered or even amorphous carbon. When the projected specimen thickness is sufficiently small, HAADF STEM images revealed highly localized areas with

Fig. 2 – Pair of simultaneously acquired HAADF (a) and BF (b) images. (a) exhibits single Pd atoms embedded in the ACF microstructure. A section of the HAADF micrograph is color-coded to enhance contrast. (b) shows microstructural information for carbon. Pd atom locations extracted from (a) are marked but show no consistent image contrast.

CARBON

4 9 ( 20 1 1 ) 4 0 5 9–40 7 3

4061

Fig. 3 – Through-focal series of seven (a) HAADF and (b) BF images that were simultaneously acquired. The defocus is calibrated with respect to Gaussian focus. The calculated CTF for BF STEM imaging is shown in (c). The inverse lattice distance for graphene layers is marked to guide the eye. Pd atoms are visible within the thinnest areas displayed in the HAADF image for slightly positive defocus values. Surrounding graphitic layers are represented by bright contrasts in the corresponding BF images.

˚ (see Fig. 2a). A bright intensities and diameters below 1 A quantitative analysis of image intensities combined with through-focal series image acquisition unambiguously identified such image intensities to represent single Pd atoms dispersed in the carbon microstructure. The simultaneously acquired BF–STEM image (Fig. 2b) offers detailed information about the carbon microstructure that supports the isolated catalyst atoms. Image interpretation requires simulations to assess contrast reversal due to the partially coherent imaging conditions [12]. Through-focal series acquisition was used to deduce the association of dark image contrasts in the BF image with carbon atom locations. Seven pairs of ADF and BF images are displayed in Fig. 3. Images were extracted from a through-focal series, which contains a total of 61 image pairs recorded with defocus increments of 1 nm between two subsequent pairs. Image intensities are varying with changes in defocus (Fig. 3b) and eventually undergo contrast reversal in the Gaussian image plane, with respect to which the relative defocus df was calibrated (i.e. df = 0 nm). For the scattering angles contributing to the acquired bright field signal, image

intensities are considered mostly coherent. The contrast can thus be interpreted on the basis of the contrast transfer function (CTF), which was calculated for different values of defocus df for the 300 keV STEM used in this study (Fig. 3c), considering the size of the bright field detector corresponding to acceptance semi-angles of a = 10 mrad and typical electron optical parameters of Cs = 0.037 mm, C5 = 100 mm and an energy spread of 0.3 eV. Seven CTFs were calculated for defocus values between +15 nm and 15 nm for direct comparison to the BF images. A more positive defocus value refers to the electron probe focused closer to the source; positive values for the CTF refer to bright image intensities while negative values refer to dark intensities in the bright field image. At an inverse distance of 2.8 nm (marked in Fig. 3c), corresponding to the lattice spacing of 0.36 nm between two graphene sheets, the CTF reverses image contrast from bright to dark intensity at a defocus of approximately +0.5 nm. Hence, graphene layers appear with dark contrast in Fig. 2b. Most excitingly, graphene layers can even be observed in the ADF–STEM image in areas with very small specimen

4062

CARBON

4 9 ( 2 0 1 1 ) 4 0 5 9 –4 0 7 3

Fig. 4 – BF STEM image of ACFs recorded from an area that shows three re-occurring structural arrangements of the ACFs in the presence of individual Pd atoms (encircled). Regions of interest marked in (a) are reproduced in panels (b–d).

thickness (see Fig. 3a). For better visualization, part of the ADF intensities were color coded in Fig. 2a, and confirm the location of graphene layers. Due to the reduced depth of field after aberration correction [13], through-focal series acquisition of HAADF micrographs allows location of individual Pd atoms in the vertical direction of the sample to discriminate atoms embedded in the volume from those located on the sample surface. The three dimensional analysis of image intensities revealed that those Pd atoms embedded in the volume of the ACF remained stable during image acquisition while atoms that were exposed to the surface of the ACF flakes during TEM specimen preparation were highly mobile under the electron beam. Quantitative image analysis of three dimensional sample volume resulted in a concentration of roughly 0.05 Pd atoms per

nm3, i.e. about 18% of the palladium content is dispersed as single atoms throughout the volume of the ACF. Analysis of many areas on multiple TEM samples revealed three reoccurring structural arrangements of the carbon microstructure surrounding individual Pd atoms. Fig. 4a shows a BF image in which all three segments are observed. The micrograph was recorded at a positive defocus setting with graphene layers represented by bright image contrast. Fig. 4b shows a configuration with an isolated Pd atom located within a disordered region of the ACF that is either amorphous or contains graphitic flakes tilted far from the beam direction. Another Pd atom resides in an open volume between two diverging graphene sheets with a third graphene layer terminating at the Pd atom site. Fig. 4c shows a Pd atom surrounded by multiple almost concentric graphene layers,

CARBON

4 9 ( 20 1 1 ) 4 0 5 9–40 7 3

similar to onion-type configurations with a small pore in its center. A third configuration (Fig. 4d) places individual Pd atoms in between two quasi-parallel sheets with slightly increased interplanar spacing at the site of the Pd atom. The identified pore structures sketched in Fig. 4 have sizes between 0.6 and 1 nm and may therefore represent highly efficient building blocks for the physisorption of hydrogen [14]. In this communication we have reported that the simultaneous acquisition of HAADF and BF image signals with aberration-corrected STEM is an ideal tool for the quantitative microstructural characterization of heavy catalytic particles that are dispersed within activated carbons. The identification of single Pd atoms embedded in the volume of the ACFs indicates that Kubas binding of hydrogen to isolated metal atoms is a viable mechanism in addition to hydride formation and physisorption, while the role of hydrogen spill-over may be smaller than previously estimated. A detailed quantitative assessment of the contributions of each mechanism to the overall observed hydrogen adsorption is provided in a separate publication [10].

Acknowledgements Research sponsored by the Division of Materials Sciences and Engineering, Office of Basic Energy Sciences, U.S. Department of Energy. KvB and CSB acknowledge financial support through UC Davis start-up funds. KvB and SJP appreciate fruitful discussions about BF STEM contrast transfer functions with Dr. Mark P. Oxley.

R E F E R E N C E S

[1] Jorda-Beneyto M, Suarez-Garcia F, Lozano-Castello D, Cazorla-Amoros D, Linares-Solano A. Hydrogen storage on chemically activated carbons and carbon nanomaterials at high pressures. Carbon 2007;45(2):293–303. [2] Gallego NC, Baker FS, Contescu CI, Wu X, Speakman SA, Tekinalp HL, et al. Effect of Pd on the Hydrogen Adsorption Capacity of Activated Carbon Fibres. Carbon 2006; Aberdeen, Scotland 2007.

4063

[3] Lueking AD, Yang RT. Hydrogen spillover to enhance hydrogen storage - study of the effect of carbon physicochemical properties. Applied Catalysis a-General 2004;265(2):259–68. [4] Bhat VV, Contescu CI, Gallego NC. The role of destabilization of palladium hydride in the hydrogen uptake of Pdcontaining activated carbons. Nanotechnology 2009 May; 20(20):204011. [5] Dag S, Ozturk Y, Ciraci S, Yildirim T. Adsorption and dissociation of hydrogen molecules on bare and functionalized carbon nanotubes. Physical Review B. 2005 Oct;72(15):155404. [6] Kubas GJ, Ryan RR, Swanson BI, Vergamini PJ, Wasserman HJ. Characterization of the 1st Examples of Isolable Molecularhydrogen Complexes, MO(CO)3(PCY3)2(H2), W(CO)3(PCY3)2(H2),MO(CO)3 (PI-PR3)2(H2),W(CO)3(PIPR3)2(H2) - Evidence for a Side-on Bonded H2 ligand. Journal of the American Chemical Society 1984;106(2):451–2. [7] Tekinalp HL, Cervo E, Thies MC, Contescu CI, Gallego NC, Edie DD. The Effect of Pitch Composition on the Adsorption Behavior of Pd-doped ACF. Carbon 2007 Seattle, WA: The American Carbon Society 2007. [8] Wu X, Gallego NC, Contescu CI, Tekinalp H, Bhat VV, Baker FS, et al. The effect of processing conditions on microstructure of Pd-containing activated carbon fibers. Carbon 2008;46(1):54–61. [9] Contescu CI, Brown CM, Liu Y, Bhat VV, Gallego NC. Detection of Hydrogen Spillover in Palladium-Modified Activated Carbon Fibers during Hydrogen Adsorption. The Journal of Physical Chemistry C. 2009;113(14):5886–90. [10] Contescu CI, van Benthem K, Li S, Bonifacio CS, Pennycook SJ, Jena P, et al. Single Pd Atoms in Activated Carbon Fibers and their Contribution to Hydrogen Storage. Carbon. 2011;[CARBON-D-10-03226]. [11] Reimer L. Transmission Electron Microscopy. 2nd ed. Berlin, Heidelberg, New York: Springer; 1997. [12] Varela M, Lupini AR, van Benthem K, Borisevich AY, Chisholm MF, Shibata N, et al. Materials characterization in the aberration-corrected scanning transmission electron microscope. Ann Rev Mater Res. 2005;35:539–69. [13] van Benthem K, Lupini AR, Kim M, Baik HS, Doh S, Lee JH, et al. Three-dimensional imaging of individual hafnium atoms inside a semiconductor device. Applied Physics Letters 2005;87(3):3. [14] Aga RS, Fu CL, Krcmar M, Morris JR. Theoretical investigation of the effect of graphite interlayer spacing on hydrogen absorption. Physical Review B. 2007;76(16):165404.