NANO LETTERS
Tug-of-War in Nanoparticles: Competitive Growth of Au on Au-Fe3O4 Nanoparticles
2009 Vol. 9, No. 12 4544-4547
Chao Wang,*,†,| Yujie Wei,*,‡ Hongyuan Jiang,§ and Shouheng Sun*,† Department of Chemistry, DiVision of Engineering, Brown UniVersity, ProVidence, Rhode Island 02912, and Department of Mechanical Engineering, the UniVersity of Alabama, Tuscaloosa, Alabama 35487 Received September 16, 2009
ABSTRACT Mechanical property of dumbbell-like Au-Fe3O4 nanoparticles (NPs) is investigated from a synthetic point of view by overgrowing Au2 on the Au1-Fe3O4 NPs. The competitive growth of Au2 on the preformed Au1-Fe3O4 NPs induced an interesting “tug-of-war” between Au2 and Fe3O4 in the formed Au2-Au1-Fe3O4 ternary nanostructure. An interpretation of the observed phenomena is proposed based on a mechanical analysis of the stress and strain distribution across the nanoparticle, which is further verified by control experiments with particle size tuned.
Synthesis of heterogeneous nanoparticles (NPs) with two or more particles in intimate contact is of great importance due to the unique electronic,1,2 magnetic,3-5 optical,1,6-9 and catalytic10,11 properties present in the structure. However, the strength of the contact between the two NPs, which is related to the mechanical property of the heterojunction, has rarely been investigated due likely to the lack of appropriate techniques to probe and manipulate the mechanical response of nanoscale objects.12,13 Here we report an insight into the mechanical property of dumbbell-like Au-Fe3O4 NP by overgrowing Au2 on the Au1-Fe3O4 NPs (the subscript 1 and 2 refers to the seed and the overgrown Au particles respectively). The competitive growth of Au2 on the preformed Au1-Fe3O4 NPs induced a “tug-of-war” between Au2 and Fe3O4 in the formed Au2-Au1-Fe3O4 ternary nanostructure. As a result, Au2 extracted Au1 out from the Au1-Fe3O4 conjugation, generating a new dumbbell-like Au2-Au1 and a dented Fe3O4 NP. The particle size-controlled growth and modeling analysis of the stress and strain distribution across the NPs indicated that this “tug-of-war” was due to the stress accumulated at the heterogeneous interface in Au1-Fe3O4. Our work would help understanding the structure stability at nanoscale and the rational design of composite nanostructures for multifunctional applications.14-17 * To whom correspondence should be addressed. E-mail: (C.W.)
[email protected]; (Y.W.)
[email protected]; (S.S.)
[email protected]. † Department of Chemistry, Brown University. ‡ University of Alabama. § Division of Engineering, Brown University. | Current address: Materials Science Division, Argonne National Laboratory, Argonne, IL 60439. 10.1021/nl903077t CCC: $40.75 Published on Web 10/20/2009
2009 American Chemical Society
The importance of the composite NPs with two NPs in intimate contact is evidenced by the current intensive research for applying NP methodology in a variety of advanced technologies. For example, tailoring the interaction between a semiconductor and a metal nanostructure has generated enormous interest in understanding the charge transfer across semiconductor-metal interface to design highly efficient light harvesting nanodevices.18-20 Nanoscale contact of Au NPs with metal oxide surface makes the “inert” Au highly active in catalyzing CO oxidation.21 To date, understanding the interparticle interactions has become an essential part in designing advanced composite nanomaterials for property optimization. The controlled synthesis of Au2-Au1-Fe3O4 NPs is illustrated in Figure 1a. Au1-Fe3O4 NPs were first synthesized by following the previously published procedure.1 Au2-Au1-Fe3O4 NPs were prepared by overgrowing Au2 onto Au1-Fe3O4 NPs in octadecene at 80 °C in the presence of HAuCl4 and oleylamine (see the Supporting Information). In the current reaction condition, the growth of Au2 did not enlarge Au1 in Au1-Fe3O4. Instead it led to a new nucleation and epitaxial growth of Au2 on Au1, forming a ternary structured NP. Figure 1b-f shows the TEM images of various composite NPs obtained after growth at 80 °C for 1, 3, and 6 h. The Au1-Fe3O4 seeds have average sizes of 5 nm for Au1 and 12 nm for Fe3O4. The size of Au2 was controlled by the reaction time with 7-9 nm Au2 formed after 1 h (Figure 1c) and g12 nm after 3 h (Figure 1d). A careful examination of the ternary Au2-Au1-Fe3O4 NPs after 3 h growth reveals that some NPs show cracks between Au1 and Fe3O4, which seems like that Au2 is trying to
Figure 1. Tug-of-war in Au2-Au1-Fe3O4 NPs. (a) Schematic illustration of the Au2 overgrowth on Au1 NP and Au1 NP detachment from the Fe3O4 NP, forming the new dumbbell-like Au1-Au2 and the dented Fe3O4 NP. (b-f) TEM images of the Au-Fe3O4 seeding NPs (b), and Au2-Au1-Fe3O4 NPs collected at 1 h (c), 3 h (e), and 6 h (f) in the synthesis. The gaps between Au1 and Fe3O4 in Au2-Au1-Fe3O4 NPs have been labeled with red arrows in (e).
nm, consistent with the size of Au1 in the ternary NPs. Figure 2e,f shows the HRTEM images of Au2-Au1 and the dented Fe3O4 NPs. It is seen that the overgrowth of Au2 on Au1 is epitaxial, and both particles show lattice fringes with interfringe distance measured to be 0.237 nm, corresponding to face-centered-cubic Au. The interfringe distance of Fe3O4 is measured to be 0.296 nm, close to that of (220) planes of magnetite with inverse spinel structure. The dent on the surface of Fe3O4 shows clear contrast difference with the rest of the NP when e-beam is focused close to the top surface of the particle (Supporting Information Figure S2). Growth of heterogeneous NPs has been well studied in the literature with mechanisms involving thermal-energy dynamics,22 surfactant-assisted growth,23,24 lattice mismatch,8,25 and nucleation control.1,11 For example, T. Mokari et al. studied the asymmetric growth of one-side Au tip on CdSe nanorod and ascribed the controlled tip growth to an Oswald ripening process that tends to reduce the whole surface energy of the nanocrystal.22 S. Habas et al. investigated the overgrowth of Pd or Au on Pt NPs of various shapes, and concluded that lattice mismatch played a key role in shaping the binary nanocrystals, that is, large lattice mismatch resulted in nonconformal growth to form asymmetric heterostructures.25 However, the Au2-Au1-Fe3O4 system shown here is different from those reported before. First, no lattice mismatch exists between Au2 and Au1 and the overgrowth is epitaxial (Figure 2e). In such a case, one would predict a conformal growth, in another word, size increase of Au1 instead of nucleation and growth of Au2. Second, the Au2-Au1 dimer is not an energetically favorable morphology compared with a single Au particle of larger size. The detachment of Au1 from Fe3O4 does not minimize the surface energy either, but rather increasing it by generation of new Au and Fe3O4 surfaces.26 On the basis of these analyses, some other factors have to be considered in order to understand the observed phenomena. We propose a mechanical interpretation of the tug-of-war process in Au2-Au1-Fe3O4 NPs by first understanding the mechanical size limit of Au1 in the Au1-Fe3O4 dumbbell system. The total free energy F of the dumbbell system can be written as
∫
1 σ:ε dV 2
∫ (γ
Figure 2. TEM images of an individual Au-Fe3O4 (a), Au2-Au1-Fe3O4 (b), Au2-Au1 (c), and dented Fe3O4 NP (d), and HRTEM images of a Au2-Au1-Fe3O4 (e) and a dented Fe3O4 NP (f).
F)
extract Au1 out of its conjugation with Fe3O4. At longer growth time, Au2 won over Fe3O4 on binding to Au1 and Au1 detached from Fe3O4, forming new Au2-Au1 dumbbells and dented Fe3O4 NPs, as shown in Figure 1f. It should be noted that the overgrowth on 5 nm Au monomer NPs produced only larger Au NPs, no Au1-Au2 dumbbells (Supporting Information Figure S1). The controlled growth is more visible by comparing individual NPs at higher resolution. Figure 2a-d shows TEM images of the individual Au1-Fe3O4, Au2-Au1-Fe3O4, Au2-Au1, and dented Fe3O4 NP, respectively. Particularly, the Fe3O4 particle shown in Figure 2d has a dent of about 5
where the first integration is in the volume of Fe3O4 NP (VF) and Au1 NP (VA), and σ and ε are the stress and strain tensors in the two NPs, respectively; the second integration is on the interface (S) of the two NPs, and γF and γA are respectively the surface energy of the Fe3O4 NP and Au1 NP, γ0 is the interfacial energy, and R is the radius of the Au1 NP. The increased strain energy can be further simplified and is controlled by the effective modulus E, the mismatch strain εm, and the equivalent volume RR3. Here R and β are geometrical factors on the order of unit. The size of the Au1 NP in the dimer system is governed by the requirement of F e 0, which leads to the critical size for Au1 NP
Nano Lett., Vol. 9, No. 12, 2009
VA+VF
S
F
+ γA - γ0)dS ≈ REεmR3 β(γF + γA - γ0)R2
4545
Re
R (γF + γA - γ0) β Eεm
Using typical parameters for Au and Fe3O4 (see the Supporting Information), we estimate the critical radius R e 4.4 nm, which qualitatively matches well with our experimental observations. During the growth of the second particle, which always starts from the far end of Au1 NP since lattice there is least distorted, the compressive stress by the surface energy of Au2 NP will decrease, and the strain energy is more localized at the interface. The reduced compressive stress in Au1 NP will result in possibly unbalanced stress across the interface, as illustrated in Figure 3a. The stress normal to the interface include several contributions, (a) the compressive stress 2γF/RF from the Fe3O4 NP with radius RF, (b) the compressive stress 2γA/RA from Au2 NP with radius RA, (c) the stress by the interfacial energy 2γ0/R, and (d) the cohesive stress due to the deformation of the interface. Note that there is a limit for the cohesive stress, the failure strength σy of the interface, beyond which the interface will break into two surfaces. Hence the equation below gives the critical size for Au2 NP 2γA 2γ0 2γF + < σy RF R RA
A reasonable estimate of the failure strength is on the order of several GPa.27-29 By taking σy ) 4 GPa, we obtain a maximum radius about 9.5 nm for the Au2 NP. Note in Figure 3b that failure may initiate at the surface junction of Au1-Fe3O4 NPs where stress concentrates (see the Supporting Information and Figure S3 for details). The above equation also suggests qualitatively that if the size of Au1 is small, Au2 has to be small. A big Au1 NP may grow big Au2 too. Figure 3c gives the strain-energy density contour in the Au2-Au1-Fe3O4 NPs. Concentration of strain energy in the interface of Au1-Fe3O4 is clearly seen. To further verify the mechanism that lattice distortioninduced nonhomogeneous strain-energy distribution and the failure criterion proposed above, we carried out control experiments by substituting the 5-12 nm Au1-Fe3O4 with Au-Fe3O4 NPs of different Au sizes. Figure 4 shows TEM images of NPs obtained by using 3-12 nm and 10-12 nm Au-Fe3O4 seeds after 6 h growth. We found that when the Au1 in the seeding NPs were small, for example, 3 nm, most NPs had no Au2 overgrowth, or only much smaller (
