Element-specific anisotropic growth of shaped platinum alloy nanocrystals

See allHide authors and affiliations

Science  19 Dec 2014:
Vol. 346, Issue 6216, pp. 1502-1506
DOI: 10.1126/science.1261212


Morphological shape in chemistry and biology owes its existence to anisotropic growth and is closely coupled to distinct functionality. Although much is known about the principal growth mechanisms of monometallic shaped nanocrystals, the anisotropic growth of shaped alloy nanocrystals is still poorly understood. Using aberration-corrected scanning transmission electron microscopy, we reveal an element-specific anisotropic growth mechanism of platinum (Pt) bimetallic nano-octahedra where compositional anisotropy couples to geometric anisotropy. A Pt-rich phase evolves into precursor nanohexapods, followed by a slower step-induced deposition of an M-rich (M = Ni, Co, etc.) phase at the concave hexapod surface forming the octahedral facets. Our finding explains earlier reports on unusual compositional segregations and chemical degradation pathways of bimetallic polyhedral catalysts and may aid rational synthesis of shaped alloy catalysts with desired compositional patterns and properties.

Nanoparticle growth starts at the edges

The high activity of precious metals such as platinum for reactions that occur in fuel cells can be enhanced by alloying with metals such as nickel and cobalt to form shaped nanoparticles, where platinum is concentrated at the corner and edge sites. Gan et al. used a combination of high-resolution imaging and modeling to follow the formation of octadedral nanoparticles of these alloys with increasing growth times. A platinum-rich phase with an extended morphology forms initially and becomes the edges and corners for the particles, and the alloying metals deposit to fill in the facets.

Science, this issue p. 1502

Shape control can be an effective approach for tuning the physical and chemical properties of inorganic nanocrystals (NCs) (1, 2). For metal alloy NCs, shape control offers a flexible means of simultaneously tailoring surface structure and surface composition so as to fine-tune their catalytic properties. For example, owing to the exceptionally high catalytic reactivity of {111}-oriented surfaces, octahedral bimetallic Pt-Ni NCs with their all-{111} orientations have been considered the ultimate “dream electrocatalysts” for the technologically important oxygen reduction reaction (ORR) in hydrogen fuel cells (36). Although a variety of shaped bimetallic Pt alloy NCs such as cubes (7, 8), octahedra (914), icosahedra (15), and concave structures (1618) have been prepared by solution-phase co-reduction of metal precursors, their detailed elemental surface compositions have been largely overlooked and thus remained poorly understood. Contrary to the widely held notion of essentially homogeneous elemental distribution and also contrary to previously theoretically predicted Pt-rich surface composition (1921), recent studies uncovered an unusual compositional segregation in shaped Pt alloy NCs (22, 23)—for example, Pt-rich frames and Ni-rich facets in Pt-Ni nano-octahedra (22). The elemental distribution resulted in complex structurally corrosive degradation patterns of the alloy octahedra during the ORR electrocatalysis; these patterns proved key to understanding their catalytic activity and stability trajectories.

We address some critical unresolved questions concerning the atomic origins of the complex compositional distribution and segregations in shaped bimetallic NCs. Specifically, how do atoms of different elements self-assemble into the compositionally anisotropic structure during the solution-phase co-reduction? Is the process thermodynamically or kinetically controlled? Although there have been numerous NC growth studies, they tend to focus on monometallic NCs or generally highlight the morphological evolution, not the compositional evolution, during NC growth through the use of transmission electron microscopy (TEM), either ex situ (24, 25) or in situ within a liquid TEM cell (2629). Typical growth models for shaped monometallic NCs involve the initial nucleation of seeds in the shape of cuboctahedra (the thermodynamically most stable shape); this is followed by a kinetically controlled anisotropic growth, which is induced either through blocking of desired facet orientations by suitable capping ligands or through selective oxidative etching of undesired facets (30). Although these models shed light on the growth of shaped alloy NCs, the compositional evolution at the atomic scale has remained unaddressed, and it is generally assumed that all alloying components follow the same anisotropic growth pathway.

We present an element-by-element study of the growth mechanism of Pt-Ni nano-octahedra using aberration-corrected scanning TEM (STEM) coupled with elemental mapping by electron energy loss spectroscopy (EELS). By tracking both structural evolution and intraparticle composition evolution during the solution-phase growth, we reveal a novel element-specific anisotropic growth mechanism of shaped alloy NCs where different alloying components follow vastly different anisotropic growth pathways. The Pt-rich phase grows first into hexapod-like concave NCs via a ligand-controlled kinetic process. Deposition of a Ni-rich phase into the concave surfaces follows in a thermodynamically controlled step, and finally, Pt-Ni octahedra form with Pt-rich frames and Ni-rich {111} facets. We track the elemental composition trajectory of other transition metal alloys as well, and conclude that the element-specific growth we observe appears to be responsible for a wide range of shaped Pt alloy NCs. Our finding provides the origin for previously reported unusual compositional segregation and chemical degradations of shaped Pt alloy NCs and enables comprehensive understanding of their entire “life cycle.”

Octahedral PtNi1.5 NCs were synthesized by solvothermal reduction of Pt(acac)2 (4 mmol/liter; acac = acetylacetonate) and Ni(acac)2 (28 mmol/liter) in 100 ml of dimethylformamide (DMF) at 120°C, where DMF acted as both reducing agent and solvent (12, 22). Unlike other methods requiring capping agents such as oleylamine (9, 10) and polyvinylpyrrolidone (13), no dedicated surfactants were needed to induce the shape-selective growth, leaving clean particle surfaces for catalytic applications. The low-temperature solvothermal synthesis is also slow [up to 42 hours, in contrast to other synthetic reactions that can complete within several minutes (7, 9)] and allowed us to capture intermediate structures at different growth stages (after 4, 8, 16, and finally 42 hours). The collected NCs were then characterized using a spherical aberration–corrected FEI Titan TEM operated at 300 kV for atomic imaging and an FEI Titan STEM (“PICO”) operated at 80 kV for high-angle annular dark field (HAADF) imaging and EELS elemental mapping.

Figure 1 presents the TEM analysis of the Pt-Ni NCs after an initial growth period of 4 hours. Energy-dispersive x-ray spectroscopy (EDX) showed an average composition of Pt75Ni25, suggesting a faster reduction of Pt ions relative to Ni ions, which in turn can be related to a more positive reduction potential of the former. The Pt-rich NCs exhibited two distinct morphologies. One is the near-spherical cuboctahedra around 2 to 8 nm in diameter; the other features a branched structure at larger sizes between 6 and 12 nm (Fig. 1B). As a typical Wulff polyhedron, a cuboctahedron represents a thermodynamically stable structure at the nucleation stage because of its lowest surface energy. The formation of larger branched NCs, however, indicates a subsequent, kinetically controlled anisotropic growth. High-resolution TEM (HRTEM) images of the branched NCs along two representative orientations (Fig. 1, C and D) suggest an early-stage hexapod structure with asymmetrical arms growing along 〈100〉 directions. This asymmetry likely arises from the competition between the anisotropic growth and an opposing surface diffusion/smoothening effect for lowering surface energy, which leads to a dynamic equilibrium between branched and spherical shapes. Once a critical particle size is exceeded (e.g., 6 to 8 nm in Fig. 1B), surface diffusion slows to a point where the anisotropic growth dominates.

Fig. 1 Characterization of Pt-Ni NCs after 4 hours of reaction time.

(A) Typical low-magnification TEM image and (B) particle size distribution, showing smaller near-spherical NCs (~4 nm) and larger branched NCs (~9 nm). They exhibited a Pt-rich average composition of Pt75Ni25. The inset of (B) shows a representative HRTEM image and atomic structure model of the near-spherical cuboctahedral NC. (C and D) HRTEM images and structural models of the branched NCs along the [001] and [101] directions, respectively. The insets show the fast Fourier transform (FFT) patterns to determine the crystallographic directions.

Indeed, when extending the reaction to 8 hours, most NCs evolved into the proposed hexapod structure (Fig. 2A), with average size around 11 nm (Fig. 2B). The average composition (Pt73Ni27) remained nearly unchanged throughout this growth stage, which implies that the growth of a Pt-rich phase along the multipod arms continued to dominate the process. Geometrically, an ideal hexapod NC with six identical arms is most stable when situating on three of the arms on a flat substrate (i.e., along the [111] zone axis), thus showing a three-fold symmetry (18). However, the asymmetrical arms described above led to other projections, such as a crisscross along the [001] zone axis (Fig. 2C). To better understand the 3D morphology, we acquired a series of images at different rotations of selected hexapod NCs (Fig. 2, D and E), with which a hexapod structural model was iteratively compared and refined. Our detailed analysis reveals a striking one-to-one correspondence between the TEM images and the projections of the 3D hexapod model at each rotation angle, which strongly supports our hypothesized growth process of the hexapod NCs.

Fig. 2 Characterization of Pt-Ni NCs after 8 hours of reaction time.

(A) Typical low-magnification TEM image and (B) size distribution, showing that most of the NCs evolved into branched structures with an average composition of Pt73Ni27. (C) HRTEM image and the corresponding FFT pattern of a branched NC along the [001] zone axis. Inset shows FFT as in Fig. 1, C and D. (D and E) A series of TEM images of two selected hexapods under different tilting conditions, showing one-to-one correspondence with the hexapod structural model. (F and G) HAADF images (red) that mainly represent Pt and EELS spectrum imaging of Ni (green), showing a slight Ni segregation at the concave surface of the Pt-Ni hexapod NCs.

The anisotropic growth of cuboctahedra into nanohexapods after 8 hours indicates much faster growth along 〈100〉 directions relative to other directions such as 〈111〉. We performed additional controlled syntheses to elucidate the likely mechanistic origin of this extremely anisotropic growth. First, in the absence of any Ni precursor while keeping all other synthetic conditions constant, the obtained pure Pt NCs showed exclusively near-spherical shape (fig. S1A). Thus, the Ni precursor must affect the formation of the hexapod NCs. Second, replacing Ni(acac)2 by Ni(acetate)2 resulted in spherical NCs as well (fig. S1B); in contrast, replacing Ni(acac)2 by K(acac) again induced an anisotropic growth into branched Pt NCs (fig. S1C). We conclude that the DMF-solvated acetylacetonate ligand plays a critical role in inducing the kinetically controlled anisotropic growth. Specifically, the DMF-solvated acetylacetonate ligand has stronger adsorption on the {111} surfaces of preformed cuboctahedra, leading to a faster growth along 〈100〉 directions and thus forming the hexapods. The presence of adsorbed organic ligands on the hexapod NCs was further evidenced by Fourier transform infrared (FTIR) spectra (fig. S2) and complemented by thermogravimetric analysis (fig. S3).

To gain element-specific insight into the growth mechanism of the hexapod NCs, we characterized their intraparticle composition using aberration-corrected STEM and EELS elemental mapping at 80 KV. As shown in Fig. 2, F and G, for the NCs oriented along two different directions, the HAADF images (red) that mainly represent the distribution of Pt and the EELS mapping of Ni (green) demonstrated slight segregation of Ni at the concave surfaces of the hexapods, whereas Pt was distributed more homogeneously. This result implies a deposition of Ni-rich phase at the concave surfaces of the hexapod NCs (see below).

When the synthetic reaction was extended to 16 hours, a transformation of the hexapod NCs to concave octahedra (Fig. 3A) was observed. Meanwhile, the Ni content of the NCs increased substantially to Pt62Ni38 while the NCs maintained their average size (Fig. 3B). These results suggest an accelerated deposition of Ni-rich phase at the concave surfaces; by contrast, the growth of the Pt-rich phase along the hexapod arms (which determined the ultimate particle size) slowed down because of the depletion of the Pt precursor. Figure 3C presents an atomic-scale HRTEM image of a resulting concave octahedral NC along the [110] zone axis. The concave {111} surfaces are clearly visible with a high density of atomic steps. The EELS elemental mappings of NCs close to two representative zone axes, [110] (Fig. 3D) and [100] (Fig. 3E), and additional EDX mapping results (fig. S4) give direct evidence of the substantial Ni enrichment at the concave {111} facets and the Pt enrichment at the corners.

Fig. 3 Characterization of the concave Pt-Ni NCs obtained after 16 hours of growth.

(A) Typical low-magnification TEM image (magnified image in inset) showing the concave octahedral shape. (B) Particle size distribution of the dominant concave octahedral NCs with an average composition of Pt62Ni38. (C) HRTEM image of a concave octahedral NC along the [110] direction, showing the atomic surface steps at the concave {111} facets. (D and E) HAADF images (red) that mainly represent Pt and EELS spectrum imaging of Ni (green) of the concave octahedral Pt-Ni NC along the [110] and [100] directions, respectively. (F) A sketch of the delayed anisotropic growth of Ni-rich phase through a step-induced, layer-by-layer deposition of Ni-rich atoms at the concave {111} surfaces of the preformed Pt-rich hexapods. Upper right: Planar view of the concave {111} surface. Ni-rich adatoms at the step sites are more energetically favorable than those at the terrace sites because of a higher coordination number (CN).

Although the growth of Pt-rich hexapods after 8 hours was ascribed to the selective capping of DMF-solvated acetylacetonate ligand on the {111} surfaces, we doubt that the subsequent deposition of Ni-rich phase at the concave {111} surfaces can be explained by another ligand-blocking effect on {100} facets. Instead, the atomic steps associated with the concave {111} surfaces appear to enable the slow yet steady deposition of the Ni-rich phase through a step-induced, layer-by-layer growth mechanism. In fact, surface defects such as atomic steps, dislocations, and stacking faults have been reported to play important roles in seed-mediated growth of bimetallic and trimetallic core-shell NCs (31, 32). As illustrated in Fig. 3F, the preformed Pt-rich hexapods after 8 hours featured intrinsic {111} surface steps at the concave surfaces and {100}/{110} steps at the sidewalls of the arms. Relative to the terrace sites, Ni adatoms energetically prefer the step sites because of a higher coordination number (Fig. 3F, upper right); thus, the process is thermodynamic in nature. Deposition of a Ni atom at the step results in the advance of the atomic step, thereby leading to a continuous, step-induced deposition of primarily Ni atoms. In this context, the sidewalls play an important role in providing the initial step sites to induce a layer-by-layer growth of Ni-rich {111} facets, as sketched in Fig. 3F. Unlike the previously reported two-pot, two-step, seed-mediated growth of bimetallic core-shell NCs (31, 32), the present anisotropic growth of Pt-rich hexapods was followed by a self-organized slower deposition of a Ni-rich phase on top of the concave {111} surfaces. The entire particle formation proceeded spontaneously in a one-pot co-reduction that remained geometrically and compositionally anisotropic. Thus, the current synthesis also enables us to control the extent of concavity by simply controlling the reaction time.

Ultimately, a complete transformation to octahedral Pt-Ni NCs with smooth {111} surfaces occurred after a reaction time of 42 hours (Fig. 4A and fig. S5). The Ni average composition increased to Pt40Ni60, whereas the particle size still increased slightly (Fig. 4B). This change is consistent with a continued, selective anisotropic deposition of an Ni-rich phase at the concave {111} surfaces. As soon as the filling of the Ni-rich phase reaches the top of the hexapod arms and flat {111} surfaces form, there are no more step edges available; metal deposition therefore self-terminates and the octahedron is completed. This also accounts for the atomic Ni ratio in the final Pt40Ni60 octahedra being far below that of the initial metal precursors (Pt:Ni = 1:7). Elemental STEM-EELS mapping of the octahedral NCs (Fig. 4, C and D) unambiguously shows Ni enrichment at the {111} facets versus Pt enrichment at the corners and edges, consistent with our previous finding (22). The Pt enrichment at both the edges and the corners also indicates a refined anisotropic growth trajectory of the Pt-rich phase; that is, the rapid growth of Pt-rich hexapods along 〈100〉 directions was followed by a slower growth along 〈110〉 directions. In contrast, as mentioned above, the growth of the Pt-rich phase along 〈111〉 directions was inhibited because of the selective adsorption of DMF-solvated acetylacetonate ligand on the {111} facets.

Fig. 4 Characterization of the final Pt-Ni nano-octahedra after 42 hours of reaction time, and overall pathway.

(A) A typical low-magnification HAADF-STEM image. (B) Particle size distribution shows dominance of octahedral NCs around 13 nm, with an average composition of Pt40Ni60. (C and D), HAADF images (red) that mainly represent Pt and EELS spectrum imaging of Ni (green) of the octahedral PtNi1.5 NCs along the [110] and [100] directions, respectively, showing Pt-rich frames along the edges/corners and Ni-enriched facets. (E) Atomic structural models of octahedral Pt bimetallic alloy NCs (Pt-M; M = Ni, Co, etc.) during the solution-phase co-reduction and during the acidic ORR electrocatalysis. (i) Initially formed Pt-rich cuboctahedra. (ii) Rapid growth along the 〈100〉 directions, resulting in Pt-rich hexapods. (iii) A delayed, step-induced deposition of M-rich phase at the concave {111} surfaces. (iv) The complete formation of Pt-M octahedra with Pt-rich corners/edges and M-rich facets. (v) Selective etching of the M-rich {111} facets during acidic ORR electrocatalysis, leading to Pt-rich concave octahedra. (vi) Degradation of octahedra into Pt-rich hexapods after long-term electrode potential cycling.

Our conclusion as to the formation and growth of a Pt-richer phase and a Ni-richer phase on two vastly distinct time scales is fully corroborated by time-resolved XRD patterns (fig. S6). All NCs at different growth stages showed Bragg reflections assigned to a face-centered cubic (fcc) lattice with peaks shifting to higher 2θ values with increasing reaction time, indicating the increase of bulk Ni content in the NCs. Whereas the Bragg reflections of the Pt-Ni NCs growing less than 16 hours showed symmetric Gaussian peaks of a Pt-rich fcc phase, the final Pt-Ni octahedra after 42 hours of growth exhibited asymmetric diffraction peaks, implying the coexistence of a Pt-richer phase and a Ni-richer phase.

To demonstrate the general importance of the element-specific anisotropic growth mechanism, we further synthesized Pt-Co NCs by replacing Ni(acac)2 with Co(acac)2 while keeping all other conditions unchanged. After 42 hours of reaction time, octahedral PtCo1.5 NCs were successfully prepared. By following their morphological and compositional evolution after different growth times (fig. S7), we observed a growth trajectory of the PtCo1.5 octahedra similar to that of the PtNi1.5 octahedra. The growth of Pt-rich hexapods/concave octahedra before the final formation of Co-rich octahedra again suggests a delayed anisotropic deposition of a Co-rich phase at the concave {111} surfaces. Thus, the element-specific anisotropic growth appears to be an important mechanism for the formation of a variety of shaped Pt alloy NCs in solution-phase co-reduction.

Figure 4E presents a comprehensive, atomic-scale “life-cycle” model of our bimetallic nano-octahedra, including their unusual anisotropic growth pathway and their previously reported degradation pathway during acidic ORR electrocatalysis (22). Our results reveal a previously overlooked element-specific, compositionally anisotropic growth mechanism of shaped Pt alloy NCs, where rapid growth of Pt-rich hexapods/concave octahedra along 〈100〉 directions precedes delayed deposition of Ni-rich phase at the concave {111} sites. Whereas the growth of Pt-rich hexapods is a ligand-controlled kinetic process, the step-induced deposition of the Ni-rich phase at the concave surface resembles a thermodynamically controlled process accomplished in a much longer time. The element-specific anisotropic growth provides the origin of our previously reported compositional segregation (Ni-rich facets and Pt-rich corners/edges) and chemical degradation pathway of the Pt-Ni octahedra (22), which underwent a selective etching of the Ni-rich {111} facets and thus activity instability during the ORR electrocatalysis in acidic electrolyte (Fig. 4E and fig. S8). While forming a catalytically active Pt-rich shell, the selective etching of the Ni-rich {111} facets resulted in concave octahedra with the exposure of less active facets such as {100} and {110}. Extended potential cycling further resulted in the re-emergence of Pt-rich hexapods and almost none of the catalytically active {111} surfaces survived, leading to substantial activity degradation. Evidently, the fate of the shaped Pt bimetallic NCs during long-term ORR electrocatalysis is substantially determined by the early stages of their element-specific anisotropic growth during synthesis.

Our results highlight the importance of understanding the element-by-element growth mechanism of shaped alloy NCs. The possibility of controlling the element-specific anisotropic growth modes of such NCs may enable the rational synthesis of Pt alloy nano-octahedra ORR electrocatalysts with desired surface composition (e.g., Pt-richer {111} facets) and sustained high activity.

Supplementary Materials

Materials and Methods

Figs. S1 to S8

References (33, 34)

References and Notes

  1. Acknowledgments: Supported by U.S. Department of Energy EERE award DE-EE0000458 via subcontract through General Motors; the Ernst Ruska Center for Microscopy and Spectroscopy with Electrons, Forschungszentrum Jülich GmbH, Germany; and Deutsche Forschungsgemeinschaft (DFG) grants STR 596/4-1 (“Pt stability”) and STR 596/5-1 (“Shaped Pt bimetallics”). We thank the Zentraleinrichtung für Elektronenmikroskopie (Zelmi), Technical University Berlin, for its support of TEM and EDX measurements; N. Erini for carrying out FTIR measurements; and R. Schomack and M. Gleich for analysis of thermal gravimetric data.

Stay Connected to Science

Navigate This Article