Letter pubs.acs.org/NanoLett
Atomic Observation of Catalysis-Induced Nanopore Coarsening of Nanoporous Gold Takeshi Fujita,*,†,‡ Tomoharu Tokunaga,§ Ling Zhang,† Dongwei Li,∥ Luyang Chen,† Shigeo Arai,§ Yuta Yamamoto,§ Akihiko Hirata,† Nobuo Tanaka,§ Yi Ding,*,∥ and Mingwei Chen*,†,⊥,¶ †
Advanced Institute for Materials Research, Tohoku University, Sendai 980-8577, Japan PRESTO, JST, 4-1-8 Honcho Kawaguchi, Saitama 332-0012, Japan § Ecotopia Science Institute, Nagoya University, Nagoya 464-8603, Japan ∥ Center for Advanced Energy Materials and Technology Research (AEMT), School of Chemistry and Chemical Engineering, Shandong University, Jinan 250100, People’s Republic of China ⊥ State Key Laboratory of Metal Matrix Composites, School of Materials Science and Engineering, Shanghai Jiao Tong University, Shanghai 200030, People’s Republic of China ¶ CREST, JST, 4-1-8 Honcho Kawaguchi, Saitama 332-0012, Japan ‡
S Supporting Information *
ABSTRACT: Dealloyed nanoporous metals have attracted much attention because of their excellent catalytic activities toward various chemical reactions. Nevertheless, coarsening mechanisms in these catalysts have not been experimentally studied. Here, we report in situ atomic-scale observations of the structural evolution of nanoporous gold during catalytic CO oxidation. The catalysis-induced nanopore coarsening is associated with the rapid diffusion of gold atoms at chemically active surface steps and the surface segregation of residual Ag atoms, both of which are stimulated by the chemical reaction. Our observations provide the first direct evidence that planar defects hinder nanopore coarsening, suggesting a new strategy for developing structurally stable and highly active heterogeneous catalysts. KEYWORDS: CO oxidation, heterogeneous catalyst, environmental transmission electron microscopy, nanoporous metal, In situ TEM
T
no additional supports (see also Figures S1−S4 in the Supporting Information).14 Previously, we reported on the atomic origins of catalysis in NPG,15 but the degradation of NPG by the aforementioned coarsening mechanisms is not fully understood. Here, we provide the first direct atomic-scale observations of the coarsening process in the porous catalyst that show completely different degradation mechanisms as compared to conventional nanoparticulate catalysts. More importantly, the atomic observation provides compelling evidence that planar defects such as twins can effectively prevent structure coarsening, suggesting a new strategy for developing chemically active and structurally sound catalysts. The effects of planar defects on catalysis have been highlighted recently,16,17 and our observation offers the first direct experimental evidence of this important phenomenon in catalysis. Figure 1 shows the representative catalytic performance of NPG in CO oxidation. The conversion rate remained at ∼91% until 200 min and then gradually decreased to 70% of the initial conversion rate after 550 min. The corresponding scanning
he lifetime of a catalyst is a crucial factor in industrial applications, particularly for heterogeneous catalysts such as nanoparticles, because the sintering (i.e., coalescence and Ostwald ripening) of particulate catalysts is the main reason for the shortening of the catalyst lifetime.1−4 Similar to nanoparticulate catalysts, dealloyed nanoporous metals such as nanoporous gold (NPG) have recently been demonstrated to possess excellent catalytic activities toward various chemical/ electrochemical reactions.5−7 Interestingly, the catalytic performance of nanoporous catalysts is dependent on the size of the nanopores and metal ligaments.8,9 Catalytic reactions can cause the coarsening of the nanoporous structure and thereby the degeneration of catalytic activities. Although nanoparticle sintering has been extensively studied to date, the mechanisms of catalysis-induced coarsening of nanoporous catalysts remain unclear and demand further investigation. Environmental highresolution transmission electron microscopy (E-HRTEM) possesses atomic resolution for in situ characterization of catalytic reactions10−13 but can only offer limited potential for the observation of nanoparticle/oxide-support catalysts because electron beam irradiation changes the shape and crystalline orientation of the particulate catalysts.13 In contrast, NPG is structurally more stable against electron beam irradiation and thermal coarsening as compared to nanoparticles and requires © 2014 American Chemical Society
Received: October 17, 2013 Revised: January 28, 2014 Published: February 7, 2014 1172
dx.doi.org/10.1021/nl403895s | Nano Lett. 2014, 14, 1172−1177
Nano Letters
Letter
Figure 1. Catalytic performance of NPG at 30 °C. Insets are SEM images of NPG samples after catalytic reactions at initial conversion rate C0, and 90, 80, and 70% of the initial conversion rate, as indicated by arrows.
electron microscopy (SEM) images of the NPG samples at the initial conversion rate C0 and at 90, 80, and 70% of the initial conversion rate are shown as the insets of Figure 1. The characteristic length scale of the nanostructure (i.e., gold ligament and nanopore dimensions) significantly increased with increasing reaction time, and the coarsening of the nanostructure led to the degeneration of the overall catalytic performance. At 70% of the initial conversion rate, some areas showed that the sizes of several ligaments increased significantly. Neglecting those regions, the average characteristic length scale was almost the same as at 80%. We then used scanning transmission electron microscopy (STEM) coupled with energy dispersive X-ray spectroscopy (EDS) to investigate the local Au and Ag content. Figure 2a−c displays the STEM images and the X-ray mappings of the Ag-Lα and Au-Lα lines and the corresponding mixed-color images from the marked areas for the NPG catalyst at the C0, 90, and 70% conversion rates, respectively. Initially, the Au and Ag are uniformly distributed in the ligament (Figure 2a). As the nanoporous structure gradually coarsened, faint heterogeneous regions of Ag were detected, as indicated by the arrow in Figure 2b. Interestingly, the heterogeneous feature is more prominent at a conversion of 70%. Ag-segregated domains 5−20 nm in size were detected (Figure 2c). These domains were frequently observed in other areas at 70%, as shown in Supporting Information Figure S5. It should be noted that the catalytic performance of NPG in the current experiment (Figure 1) differs slightly from that of our previous studies.8,9 This is a result of the variation in the reaction conditions and sample structures (see Supporting Information Table S1 and Figure S6). However, the coarsening behavior is quite universal regardless of the speed of the reactions.18 We focused the E-HRTEM on an ∼10 nm nanopore, viewed along the [011] direction prior to the reaction, and then exposed it to a CO/air gas mixture at pressures of up to 30 Pa in order to observe how the nanopore coarsened during the reaction (Figure 3a−f). The nanopore was located across two twin interfaces and 20 nm away from another large nanopore. The twin interfaces were typical Σ3 coincident-site lattice
Figure 2. High-angle annular dark-field STEM images of NPG after catalytic reactions and corresponding EDS elemental mapping. Au-Lα (in red), Ag-Lα (in green), and corresponding mixed-color images from selected areas for (a) initial stage, and at conversion rates of (b) 90% and (c) 70%. Arrows in (b,c) show segregated Ag atoms on NPG surface.
boundaries and may have been induced by deformation during dealloying.19 The shape of the nanopore exposure to the reaction gas mixture was nearly round, and the nanopore had a diameter of ∼8.5 nm and a high density of surface atomic steps, as shown in Figure 3a. Immediately after the nanopore was exposed to the CO/air gas mixture, the surface became faceted (Supporting Information Movie S1). After the nanopore was exposed to the reaction gas for about 50 s, the faceting surface dynamics became more vivid without any obvious additional nanopore coarsening (Figure 3b and Supporting Information Movie S2). Each local surface plane of the nanopore shown in Figure 3b is indexed in Figure 3c. The topmost surface planes of the reconstructed nanopore are either {111} or {100}, regardless of the geometry. The round nanopore became geometrically regular with flat and low-index surfaces. Significant coarsening of the nanopore could be observed when the reaction time was extended, during which the surface planes of the nanopore still tended to be the low-index {111} and {100} planes (Figure 3d). The selection of either the {111} or {100} surface solely depends on whether the geometry of the {111} or {100} plane is kinetically and thermodynamically favored during pore coarsening. As shown in Figure 3 and Supporting Information Movie S3, the nanopore coarsened by the diffusion of gold atoms at the chemically active surface steps on the topmost 1173
dx.doi.org/10.1021/nl403895s | Nano Lett. 2014, 14, 1172−1177
Nano Letters
Letter
Figure 3. Sequence of nanopore coarsening during catalytic reaction: (a) Initial state. (b) Faceted nanopore just after surface was exposed to CO/air gas mixture for ∼50 s. (c) Crystallography of nanopore shown in (b); dotted lines represent twin planes. (d) Preferential growth direction normal to (100) surface plane during catalytic reaction; see also Supporting Information Movie S3. (e) Necking of gold ligament before rupture; dotted selected area is enlarged to show deformation twins. (f) Coalescence of small and large nanopores after rupture.
surface. The wide steps on the flat surface of the geometrically regular pore accelerated the layer-by-layer atomic motion and nanopore growth. Before the pore coalesced with a neighboring pore, the gold ligament between them became increasingly thinner, and this thinning process was accompanied by nanopore growth. Deformation twinning could be observed at the necking region before the rupture of the ligament (Figure 3e), indicating that large local stresses and strains were produced during nanopore coarsening. Eventually, the nanopore merged with the neighboring nanopore (Figure 3f). This process is very similar to the computational simulation of nanopore coarsening driven by high temperatures.20 We systematically measured the growth rates along the different facets of the nanopore. The coarsening rate of the {100} surface was much higher than that of the {111} surface (Figure 3d and Supporting Information Movie S3). This is in addition to the fact that the {111} plane had a lower surface energy than the {100} plane.21 The rapid growth of the {100} surfaces actually provided wide surface steps at the corners in connection with the {111} surfaces, which assisted the coarsening along the {111} surfaces. Figure 4 shows the effect of exposure time on the height of and width between the {111} planes, defined in the inset, during steady coarsening of the nanopore. The growth rate along the width direction, which was perpendicular to the twin interfaces, was obviously higher than that along the height direction, which was parallel to the twin interfaces (Figure 4; the sequence is shown in Supporting Information Movie S3). Interestingly, the difference in the growth rates was caused by the twin boundaries, which worked as pinning sites at the propagating front of the coarsening nanopore (Figure 5; the sequence is shown in Supporting Information Movie S4). The atomic motion along the {111} surfaces usually stopped at the twin boundary during nanopore coarsening along the height direction. After the topmost atoms of the twin boundary moved away (the kink and the step of the twin boundary are circled in Figure 5a,d, respectively), a layer of gold atoms quickly diffused
Figure 4. Height of and width between nanopores plotted as functions of exposure time. Height (red arrow) and width (green arrow) between {111} planes are defined in inset.
along the (11̅1) plane (Figure 5b,c and e,f). Although the residual silver atoms in the NPG may have stabilized the atomic steps in catalytic environments,15 it is unlikely that the residual silver atoms segregated at the twin boundaries stabilized the atomic edges because the amount of residual silver was only ∼1 atom % in the NPG and the total reaction time (
