† Corresponding author. E-mail:
Project supported by the National Natural Science Foundation of China (Grant No. 51406111), the Major Program of the National Natural Science Foundation of China (Grant No. 51590902), the Natural Science Foundation of Shanghai, China (Grant No. 14ZR1417000), the Scientific Innovation Project of Shanghai Education Committee, China (Grant No. 15ZZ100), and Young Eastern Scholar of Shanghai, China (Grant No. QD2015052).
We simulated the asymmetrical growth of cube-shaped nanoparticles by applying the Monte Carlo method. The influence of the specific mechanisms on the crystal growth of nanoparticles has been phenomenologically described by efficient growth possibilities along different directions (or crystal faces). The roles of the thermodynamic and kinetic factors have been evaluated in three phenomenological models. The simulation results would benefit the understanding about the cause and manner of the asymmetrical growth of nanoparticles.
Shape control of metal nanostructures is a promising strategy of tailoring their physical, chemical, and biocompatible properties for various applications, such as catalysis, sensing, bioimaging, photothermal therapy, plasmonics, and spectroscopy.[1–4] Most metals crystallize in the same cubic close-packed structure: a face-centered cubic (fcc) lattice. It is known that the external morphology of a crystal is decided by its internal point group symmetry. During the past decade, a wide variety of polyhedral fcc metal (gold, silver, etc.) nanostructures with Oh symmetry, including cubes, truncated cubes, cuboctahedra, truncated octahedra, and octahedra, have been synthesized.[2,4,5] These cases are consistent with the point group symmetry of the crystals. In the meantime, highly anisotropic metal nanostructures which do not conserve the point group symmetry were also widely reported in the literature, for example, one dimensional (1D) rods and wires.[6,7] This raises the critical issue here: why and how asymmetrical growths could take place in the fcc metals? As we know, there are several mechanisms that could lead to asymmetrical growth of the fcc crystals. Commonly, twinning has been proven as a cause of asymmetrical growth of metals and produces multiple twinned particles.[2] Shen et al. demonstrated an alternative mechanism for the formation of 1D silver nanostructures, in which continuous formation of (111) twins, local 4H phase, and other stacking faults along one direction induces the asymmetrical growth of the crystal of fcc silver.[8] In the case of two-dimensional nanoplate,[9] twinning also plays an important role, though the detail has not yet been revealed. Moreover, oriented attachment is another important mechanism of the formation of 1D nanostructure that involves a spontaneous self-organization of adjacent particles.[10,11] Differing from the above mentioned mechanisms, the evolution of a single-crystal nanoparticle, e.g. a nanocube, to a 1D nanostructure is mysterious and unexplored. Obviously, this evolution is a kinetic-controlled growth. A kinetic-controlled growth here refers to a growth leading to some shapes which deviate from the thermodynamic equilibrium one with minimal Gibbs free energy. In fcc materials, the surface energies γ of low-index crystallographic facets have the energetic sequence of γ{111} < γ{100} < γ{110}.[12] As a result, the equilibrium shape is the truncated octahedron (or Wulff polyhedron) for the single-crystal fcc nanoparticles. However, the preferential binding of additives to certain crystal facets can alter the sequence of the surface energies, and thus other shapes, like the cube enclosed by six {100} facets, may become the equilibrium one. As an example, bromide binds most strongly to the {100} facets of Ag, Pd, and Pt. This benefits the formation of nanocubes.[2,4,13–15]
A cube has a smaller surface area than a cuboid of the same volume, which means a lower Gibbs free energy. However, we find exceptions in almost every case, where considerable numbers of cuboid nanoparticles exist among the cubic particle clusters, especially when the average size of the particles is smaller than 10 nm. The aspect ratio r between the longest side lmax and the shortest side lmin of these cuboid nanoparticles is usually smaller than 1.5. This phenomenon appears in metals,[13–16] alloys,[17,18] and even metallic oxides.[19,20] Besides the microcubes assembled by Pd nanocubes, some cuboid microstructures can also be obtained. All these experimental results indicate that this asymmetrical growth may not arise from some specific kinetic mechanisms, but from thermodynamics alone. For a nanocube, the three directions are equivalent under the thermodynamics-controlled growth. Therefore, the formation of cuboid nanoparticles must be governed by extra mechanisms, which means that the growth is preferable along one direction rather than equally along three directions. So far, several kinetic mechanisms, such as surfactant-assisted growth,[4,21] electric-field-directed growth,[22] and localized-oxidative-etching-promoted growth,[23] have been proposed to account for the formation of 1D nanostructures whose aspect ratios are usually larger than 2. Although some asymmetrical growth mechanisms have been proposed.[24] for cases like the experimental results, there are very limited theoretical works that can be used to quantitatively understand the observations. To explore the asymmetrical growth mechanism of cube-shaped nanoparticles, we establish models to study the role of thermodynamic and kinetic factors in the anisotropic growth of cube-shaped nanoparticles using Monte Carlo (MC) simulation.
The growth of a nanoparticle can be described by the growth rate along specific faces or directions. In consideration that the synthesis of the cube-shaped nanoparticles often takes place in the solutions, the growth of the particles is not limited by the substrate. When the particles are crowded, the side facing a neighbor particle will have reduced chance to grow. However, it is acceptable to neglect this neighborhood competition when the concentration of the particles is low.
We focus on why and how the cuboids appear among cubes. We can simplify the cube formation simulation using a simple cubic (sc) model instead of the fcc model. In a homopolar crystal with a simple cubic lattice, each atom can be treated as a cube. Because bonding may occur at each face of the cube, an atom may have up to six first-nearest neighbors. The interactions of the particle with its second and third neighbors, which share an edge and corners with the particle, respectively, also exist. However, the interactions with the second and the third neighbors are much weaker than that with the first-nearest neighbors and therefore can be neglected reasonably. In the beginning of the crystallization, hundreds of atoms assemble under the nucleation mechanism. To simplify, this atom assembly is assumed to be a cube. A new atom attaches on one of the six faces of the cube with equal possibility. Then another atom tends to attach on the already formed assembly at the site where most chemical bonding can be formed, which means that the attachment process favors releasing the maximum energy. Therefore, if an atom attaches on one face, a whole layer including that atom is formed soon after being adhered to that face. Another layer is then formed, and so on. This is the well-known layer growth model (Kossel–Stranski model).[25–27] Our simulation is based on the layer growth model. The schematic figure of the growth process is shown in Fig.
First, we consider the growth process of the cuboid nanoparticle dominated only by the thermodynamic factor, which is isotropic. This model represents approximately the growth condition without anisotropic perturbations, like surfactant, nonuniform electric field, localized-oxidative-etching, and so on. After the nucleation process, many atoms move freely and perform random Brownian motion in the solution. The possibility that one atom attaches on a face is proportional to the area of that face, which means the growth possibility Pi ∝ Si (i = x,y,z). The thermodynamics term
Since the kinetic factor is excluded, the kinetic term
Then we consider the case with both thermodynamic and kinetic factors. Although several kinetic mechanisms, such as surfactant-assisted growth, electric-field-directed growth, and localized-oxidative-etching-promoted growth, have been proposed for certain types of inhomogeneous nanoparticle synthesis procedures respectively, we will focus on the common factor of these mechanisms to universally describe the effect of the kinetic factor. At the same time, the thermodynamic factor always exists and cannot be neglected.
One type of kinetic factor depends on the shape of the nanopaticle. Juste et al. attributed the inhomogeneous growth of gold nanorods to the electric field.[22] The nonuniform electric field is much larger at the tip of the nanorods than at the sides, and it gradually increases with the growth of the particles. This type of kinetic factor is a positive feedback process, which accelerates the inhomogeneous growth and lengthens the particle persistently along one direction once the asymmetrical growth takes place along this direction. To describe this type of kinetic factor, we consider that growth possibility Pi (i = x,y,z) increases with li, which means increased growth possibility along the long side. The expression of the kinetic term in Eq. (
The thermodynamic term is the same as that in the previous model. Equation (
In the previous subsection, the shape-dependent model is presented. Now, we discuss another type of growth mechanism, which is independent of the shape of the nanoparticles. The simplest case is a constant kinetic factor along the preferred growth direction, say x. With this assumption, we obtain the kinetic term as
It is known that, at the initial stage of the growth, since Px > Py, the inhomogeneous growth along the x direction takes place and therefore the nanoparticles gradually elongate along the x-direction and deviate from cubes. Since the symmetry is broken along the x-axis, it is predictable that lx > ly ∼ lz. Defining that r0 ≡ lx/ly ∼ lx/lz, the following equations are obtained from Eq. (
It is known that li ∝ vi ∝ Pi, where vi is the growth rate along the i-direction. Therefore, the equation lx/ly = Px/Py = r0 holds. Then by substituting Eq. (
This means that when lx is large enough, the growth reaches a stable stage, when the ratio between Px and Py keeps a constant. What should be emphasized is that equation (
Following the models established in Section 2, we perform a Monte Carlo simulation for the growth of the cuboid nanoparticles. In the simulation, we carry out a growth process from a 5 × 5 × 5 size cube. In each step, a random number ranged from 0 to 1 is sampled by Eqs. (
Furthermore, we can also define
To give a clear impression of the size of the obtained cube-shaped nanoparticles, we calculate the atom number of the platinum (Pt) cubes with the side length l. The results are shown in Table
We further study the shape distribution of the cuboid nanoparticles when only the thermodynamic factor is considered and the results are shown in Fig.
Schematic images are presented for comparing our simulation to the experiments. In the experiments, one usually takes SEM or TEM images after the formation of the cuboid nanoparticles, which always lay on the base. It always looks like a batch of squares and rectangles spread in a plane. 100 simulated nanoparticles are picked randomly from our simulation, and are laid on a random face to form the schematic images shown in Fig.
We now turn to study the influence of the kinetic together with the thermodynamic factor on the shape of the cuboid nanoparticles. First, the growth process governed by both thermodynamic factor and shape-dependent kinetic factor (see Eq. (
It is interesting that the growth trends for γ = 0.5 and 1 are opposite, which reflects the different properties of the thermodynamic and kinetic factors. In order to address the two different trends more clearly, ⟨r⟩ is introduced, which stands for the average value of aspect ratio r. ⟨r⟩ can be calculated based on the distribution of r. The result is shown in Fig.
To help us understand the numerical result, the schematic images are generated for γ = 1 at different N̅ in Fig.
We then discuss the growth of the cuboid nanoparticles following the model with both thermodynamic and shape-independent kinetic factors. Two cases for small and large α are considered and the simulation results are shown in Fig.
The aspect ratios of all the cuboid nanoparticles are asymptotic to r0 if the average size is large enough. This feature has been explained analytically in Subsection 2.3. In the early stage of the growth, due to the kinetic term along the x direction, Px is larger than Py and Pz and thus growth along the x direction is the fastest. When the growth reaches the stable growth stage, the nanoparticle continues to elongate and at the same time it thickens, while r keeps a constant. Therefore, the distributions of D and r are packets. Figures
Then the average values of r for different α values based on the simulation results are calculated and shown in Fig.
To present our simulation results more clearly, we generate schematic images based on the simulation data in Fig.
The effects of the thermodynamic and kinetic factors on the asymmetrical growth of the cube-shaped nanoparticles have been investigated by the Monte Carlo method. Based on different types of growth processes in experiments, we simplified the universal model into three cases with specified physical concepts. They are a) thermodynamic factor only model; b) thermodynamic with shape-dependent kinetic factor model; and c) thermodynamic with shape-independent kinetic factor model. For the first growth model, the possibility that the nanoparticles deviate from the cube-shape is large (when N̅ ∼ 103, p1.2 ∼ 0.3) at the early stage of the growth, when the size of the nanoparticles is small. However, when the size of the nanoparticles is large, due to the homogeneous property of the thermodynamic factor, the possibility of deviating from cubes is fairly low. This result can explain why there is a considerable portion of cuboid nanoparticles accompanying the cube-shaped nanoparticles obtained in the experiments when the size of the nanoparticles is small, no matter what the materials and experimental conditions are. For the model with both thermodynamic and shape-dependent kinetic factors, the thermodynamic factor is dominant for a wide range (0 < γ < 0.93), for which case the nanoparticle tends to be a cube when its size is large enough. When 0.96 < γ < 1, the kinetic factor plays a leading role and makes the nanoparticle elongated. The final products have different appearances, including the cube-shaped, cuboid, and rod-shaped nanoparticles. This result explains the experiments in which nonuniform shaped products are obtained. Finally, a model with both thermodynamic and shape independent kinetic factors were analyzed. The kinetic factor accelerates the growth along one direction until the growth process reaches a stable stage where r is a constant. This result explains the type of experiments in which uniform nanorods are obtained. Our simulation based on three simplified phenomenological models successfully represents three major types of known nanoparticle products in experimental works. This theoretical work benefits the understanding of the cause and manner of the asymmetrical growth of nanoparticles and could be useful for guiding the shape control of metallic nanostructures.
1 | |
2 | |
3 | |
4 | |
5 | |
6 | |
7 | |
8 | |
9 | |
10 | |
11 | |
12 | |
13 | |
14 | |
15 | |
16 | |
17 | |
18 | |
19 | |
20 | |
21 | |
22 | |
23 | |
24 | |
25 | |
26 | |
27 |