Divalent Metal Nanoparticles

See allHide authors and affiliations

Science  19 Jan 2007:
Vol. 315, Issue 5810, pp. 358-361
DOI: 10.1126/science.1133162


Nanoparticles can be used as the building blocks for materials such as supracrystals or ionic liquids. However, they lack the ability to bond along specific directions as atoms and molecules do. We report a simple method to place target molecules specifically at two diametrically opposed positions in the molecular coating of metal nanoparticles. The approach is based on the functionalization of the polar singularities that must form when a curved surface is coated with ordered monolayers, such as a phase-separated mixture of ligands. The molecules placed at these polar defects have been used as chemical handles to form nanoparticle chains that in turn can generate self-standing films.

Nanoparticles that consist of crystals of tens to thousands of atoms have been used as “artificial atoms” to form supracrystals that mimic ionic solids (1) or to form the nanoscale equivalent of ionic liquids (2). Breaking the interaction symmetry in these isotropic and almost spherical materials is a major challenge. It is increasingly evident that nanoparticles would become a much more powerful research tool if it were possible to place a known small number of molecules in their ligand shell to enable directional assembly, for example. This would enlarge the scope of potential applications, because anisotropic assemblies have distinctive properties that cannot be found or produced in isotropic assemblies (3, 4). Indeed, a large research effort to direct the assembly of nanoparticles (NPs), based primarily on biomolecules (59) or other templating agents (10), has been hindered by a lack of control over the number of receptors that interact with the templating agent. Approaches based solely on stoichiometry tend to require dilute aqueous solutions, resulting in limited throughput (11, 12). Recently, creative methods to introduce single valency in NPs have been developed, mostly through solid-state reactions (1316). We present an approach to functionalize monolayer protected metal NPs at two diametrically opposed points that exist as a consequence of the topological nature (17, 18) of the particles. Specifically, we have made NPs that act as divalent building blocks (“artificial monomers”), and can be reacted with complementary divalent molecules to form chains that can then produce self-standing films.

Monolayer protected metal NPs are supramolecular assemblies consisting of a metallic core coated with a self-assembled monolayer (SAM) composed of one or more types of thiol-terminated molecules (ligands). It is known that molecules in SAMs on flat gold surfaces form a two-dimensional (2D) crystal in which every molecule conforms to the same tilt angle and direction relative to the surface normal (19, 20) in order to maximize the van der Waals interactions with its nearest neighbors. Landman and co-workers (21) addressed the question of the morphology of the ligand shell SAM on the faceted surface of a gold NP. They found that ligand molecules conform to one single tilt angle relative to a common particle diameter rather than assuming their equilibrium tilt angle on each crystallographic facet, which would generate a large number of line defects along facet edges. That is, the vectorial projection of the tilted ligand molecules propagates around the particle. This needs to be reconciled with the fact that on a topological sphere a 2D crystal cannot exist unless two separate point defects are present (22, 23). This is commonly known as the “hairy ball theorem” that states that it is not possible to “align hairs” onto a sphere without generating two singularities (such as the whirl on the back of our heads). Recently, we have shown (18, 24) that mixtures of thiolated molecules, which on flat gold surfaces separate into randomly distributed domains (25), form ordered alternating phases (ripples) when assembled on surfaces with a positive Gaussian curvature, such as the core of an NP (Fig. 1A). These types of domains will profoundly demarcate the two diametrically opposed singularities at the particle poles, where the rings collapse into points (Fig. 1B). We conjecture that, in the case of a self-assembled ligand shell, the polar singularities manifest themselves as defect points, that is, sites at which the ligands must assume a nonequilibrium tilt angle. Ligands at the poles, being not optimally stabilized by intermolecular interactions with their neighbors, should be the first molecules to be replaced in place-exchange reactions (SOM Text S1).

Fig. 1.

From rippled particles to NP chains. Idealized drawing of (A) a side view and (B) a top view of a rippled particle showing the two polar defects that must exist to allow the alternation of concentric rings. (C) Schematic depiction of the chain formation reaction.

Gold NPs coated with a binary mixture of 1-nonanethiol (NT) and 4-methylbenzenethiol (MBT) were synthesized and characterized by scanning tunneling microscopy (STM). Ordered rings similar in nature and spacing to the ones observed previously (18) were found. These molecules were chosen for multiple reasons: They have a strong driving force for phase separation and a large STM contrast, and are terminated with unreactive methyl groups. Transmission electron microscopy (TEM) was used to determine the size distribution and the molecular weight of the NPs. To place-exchange at the polar defects, the particles were dissolved in a solution containing 40 molar equivalents (relative to the moles of particles) of 11-mercaptoundecanoic acid (MUA) activated by N-hydroxysuccinimide. After stirring for 30 min, the reaction was rapidly quenched by filtration over a Sephadex column or by inducing precipitation with deionized water. A two-phase “polymerization” reaction inspired by the well-known procedure to synthesize nylon was then performed by combining an organic (toluene) phase containing the MUA functionalized particles with a water phase containing divalent 1,6-diaminohexane (DAH) (Fig. 1C).

Within a few minutes, a precipitate begins to form at the water-toluene interface (fig. S1); after a few hours, the reaction reaches equilibrium. The precipitate can be re-dissolved in tetrahydrofuran (THF) and dropped onto a TEM grid or, conversely, collected directly on a TEM grid and rinsed with THF. TEM images in both cases show a large population of linear chains of NPs, ranging from 3 to 20 NPs in length (Fig. 2). The low fraction of branched chains and the absence of 3D aggregates in the images strongly supports the idea of selective functionalization at the two opposed polar defects and suggests that polar singularities react even faster than other defects in the ligand shell.

Fig. 2.

TEM images of chains that compose the precipitate obtained when MUA pole-functionalized rippled NPs are reacted with DAH in a two-phase reaction. Scale bars 200 nm, inset 50 nm.

Many control experiments were performed to observe the formation (or lack thereof) of a precipitate and the presence or absence of chains in TEM images of both the precipitate and the toluene phase (fig. S2). Mixed-ligand rippled NPs showed precipitate only when pole-functionalized with MUA, either activated or not activated. NPs containing carboxylic acid groups everywhere in the ligand shell formed, as expected, only large 3D aggregates resulting from nondirectional interparticle bonding. In general, precipitate was not observed when DAH was not present in the water phase or when the particles lacked carboxylic acid terminated ligands, proving that the precipitate is a product of an amide-coupling reaction leading to an amide bond (in the case of activation) or to a salt. A statistical analysis was performed on chains and controls resulting from one-phase syntheses (26). After analyzing more than 40 samples with a total population exceeding 50,000 particles, it was found that the average fraction of particles in chains was 20% (SD = 8%), whereas in all control experiments only 3 to 5% of the NPs were found in chainlike structures. The one-phase synthesis had a lower yield than the interfacial two-phase reaction.

To further prove that the chains observed in our TEM images were due to molecular linking of our particles, the synthesis was performed with one of two divalent linking molecules of different lengths: DAH and O,O′-Bis(2 aminoethyl)octadecaethylene glycol (EGDA). We then measured the average interparticle separation along the chain (defined as the distance between the two nearest points or facets). Assuming an all-trans conformation for the molecular linkers (i.e., two MUA molecules covalently bonded to either DAH or EGDA), the expected interparticle distance would be 3.6 and 9.6 nm, respectively. Analysis of TEM images showed an average interparticle separation of 2.2 nm (SD = 0.4 nm) for DAH and of 4.2 nm (SD = 0.9 nm) for EGDA (Fig. 3A), proving that the observed chains are kept together by the molecular linkers. The larger spread in the EGDA distribution was expected because of its greater conformational freedom (Fig. 3B). Evidence that chains are present in solution (as opposed to being caused by solvent drying) was also obtained through light-scattering experiments. We found an increase in scattering intensity for THF solutions of chains as compared with solutions of identical optical density containing only the starting particles (fig. S3), proving the presence of aggregates in solution. Moreover, NPs of two different sizes (average diameter 10 and 20 nm, respectively) were pole-functionalized with 16-mercapto-1-hexadecylamine and reacted with sebacoyl chloride (a divalent molecule) to form chains of random composition (fig. S4). When the same particles were cast on a TEM grid without previous pole functionalization or without reacting with sebacoyl chloride, primarily isolated particles were obtained. To prove the dynamic nature of the chains, we exposed them to a large excess of NT for 3 days and observed only isolated particles in TEM images (fig. S5). Poles provide a distinctive way to place NPs on a surface with their ripples parallel to the substrate plane. STM images of samples prepared in this way lack the striations present when the sample is prepared such that the ripples are perpendicular to the substrate plane (fig. S6).

Fig. 3.

Variation in interparticle distance with linker molecule. (A) Distribution of interparticle distances in chains with DAH linkers (blue) and EGDA linkers (red). The average interparticle distance for DAH was 2.2 nm ± 0.4 nm; for EGDA, it was 4.2 nm ± 0.9 nm. Note that the measured distributions barely overlap. Insets show TEM images of representative chains of each type. Scale bars, 20 nm. (B) Schematics illustrating (left) the potential geometry of NPs that could result in the measured interparticle distances being smaller than the length of the linker and (right) the conformational freedom of EGDA that could result in the observed wide distribution of interparticle distances for this type of chain.

Place-exchange reactions have been thoroughly studied by Murray and co-workers (2729). They found that molecules exchange first at defects in the ligand shell or at corners and edges of the core crystal. Using nuclear magnetic resonance, they calculated the initial rate of the ligand-exchange reaction: For 1-octanethiol–coated NPs they found a second-order reaction rate constant of 3 10–2 M–1 s–1 (27). In our system, the place-exchange reaction reaches equilibrium (at least for pole functionalization) after 10 min (30). Given that the number of NPs that contain at least one MUA molecule in their ligand shells can be experimentally evaluated (by making the conservative assumption that every MUA-functionalized particle will react to form an insoluble chain), the second-order rate constant was found to be 1.67 M–1 s–1, about two orders of magnitude larger than that observed for homoligand NPs (27). Applying the fastest second-order rate constant published (27), the number of MUA molecules that would have reacted at defects other than the poles under our reaction conditions is only 1 MUA molecule per 100 NPs. It should be pointed out that (i) we assume that every chain precipitates but observe few dimers in TEM images of the precipitate, hence they must still be soluble; and (ii) we assume that only one MUA molecule is located at each pole. Faster rates of reaction would be obtained if any of these assumptions were removed. These kinetic experiments show that defects at the molecularly defined polar singularities of mixed ligand NPs are thermodynamically distinct from those at crystallographically defined vertices of the core crystal. Most experiments were performed with particles with an average diameter of ∼4 nm; as the size of the particles changes, it is reasonable to expect that the rate of polar reactions will vary. The polar singularities will exist only in a certain size range: On surface hemispheres with a radius larger than 20 nm, ripples do not form (18), and a similar behavior is expected on NPs. Investigations are under way to determine the lower size limit; we have observed ripples and chains in NPs as small as 3 nm in diameter.

Enough NP chains at the water-toluene interface have been produced to form a continuous film as large as 1 cm2 (fig. S7). Our preliminary conclusion is that these films (whose thickness can reach 60 μm) must be composed of multiple interwoven chains. Whereas the chains described here are soluble in dichloromethane (DCM), the films when placed in DCM quickly lose any unreacted particles but maintain their structural integrity and then take weeks to dissolve. The van der Waals interactions between the ligand shells of the particles together with the interchain morphology provide enough mechanical strength to make these purely NP films self-standing. Control experiments show that, in situations where chains are not formed, one of two extreme cases happen. When the entire ligand shell can react with diamine (e.g., there is carboxylic acid in the ligand shell), coarse powders that are insoluble in any solvent are formed. When the ligand shell cannot react with diamines, semicontinuous films form, but only after the toluene has completely evaporated; these materials can at times be self-standing but always readily dissolve in organic solvents.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S7


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article