Self-Assembly of a Micrometers-Long One-Dimensional Network of Cemented Au Nanoparticles

One-dimensional Au nanoparticles connected by ionic linkers self-assemble into a micrometer-scale network. Reactions between linker ions reinforce the assembly and lead to the formation of a continuous network. This continuous network demonstrates roomtemperature Coulomb-blockade characteristics that are indicative of electron transport along the 1D pathways. 1 Figure 1. Absorption spectra of the assembly at each stage of the transformation. For clarity, the PSS–CdS–Au spectrum has been shifted vertically by 0.5 units. The change in the color of the solution, from wine red to violet blue and then muddy yellow, is visible in the photograph. 2 Maheshwari, Kane, & saraf in AdvAnced MAteriAls 20 (2008) To guide the bridged cluster into a long-range 1D assembly, 200 μL of PSS solution (1 mg mL–1) was added to 5 mL of Cd–Au solution. The PSS sulfonate groups preferred to attach to the free Cd ions[10] on the nanoparticle surface that were not involved in particle linkages. In contrast to the Au– Cd suspension stability (less than 2 weeks), the PSS–Cd–Au solution was remarkably stable and showed no change in color or in the position of the SPR band (indicative of agglomeration) over a time period of more than 6 months. This increase in stability was attributed to the steric and electrostatic hindrance that arises from the anionic PSS. The only significant change was the appearance of the PSS absorbance band at ~265 nm (Figure 1). The 1D necklace of Au nanoparticles that was deposited on the PSS chain was clearly visible in the TEM image (Figure 2a). The arrangement of the particles was fairly discrete, with an interparticle distance of 1–2 nm. The addition of Na2S to the solution, at a ratio 0.7 times the stoichiometric ratio with respect to Cd, changed the absorbance spectrum significantly (PSS–CdS–Au curve in Figure 1). The SPR band of Au was evident from the weak peak at 540 nm that extended to 800 nm. The shift in the SPR peak from ~620 nm back to ~540 nm for the Au particles (Au-10-nm curve in Figure 1) indicated weaker plasmon coupling resulting from the higher dielectric constant of CdS compared to that of the ionic bridge. A shoulder at 415 nm that was highly blue shifted compared to bulk CdS (510 nm) was attributed to formation of interparticle CdS (cement) (PSS–CdS–Au curve in Figure 1). The formation of CdS cement was more evident in the photoluminescence (PL) spectrum (Figure 3). Figure 3 shows a single emission peak at ~455 nm for the Au particle necklace on PSS both before (inset in Figure 3) and after cementing (i.e., for Au–Cd–PSS and Au–CdS–PSS, respectively). This 455 nm peak arose from radiative recombination of electrons and holes in the interband transition for the Au nanoparticles.[13,15] However, the spectrum of the cemented necklace (Au–CdS–PSS) had an emission band centered at ~610 nm, in addition to the 455 nm peak. This new 610 nm band was associated with the formation of CdS nanoscale cement between Au particles. The larger red shift in photoluminescence for CdS cement at 610 nm than for pure CdS at 530 nm[16] was attributed to the change in surface states of CdS due to the interface with Au. The exothermic reaction of Na2S and the affinity of the gold surface for sulfur led to the formation of a “metallurgical” junction and a small amount of diffusion of Au into the CdS cement. The assignment of the 610 nm band to an impurity energy level from Au in the bandgap of CdS (i.e., CdS–Au) is consistent with reports in the literature.[17] The cementing of the Au nanoparticles into the necklace, due to addition of Na2S, was also observed by TEM (Figure 2b). In Figure 2b, we see that the discreet particle necklace (Figure 2a) was transformed into a continuous wire where silhouettes of the nanoparticles appear as nodules in the wire. The width of the wire was ~10 nm, which indicated that it was composed of a 1D row of particles. The morphology of the necklace, characterized by branches and loops, appeared to remain intact. Also, the length of the clusters remained approximately the same, which indicated that the necklace was fairly robust; it did not fracture during processing, for example when making the TEM samples. The fact that topology of the necklace did not change significantly on cementing suggested that the necklace was a 3D network of 1D necklaces. In some regions, the wire appeared to have agglomerations with widths of ~20 nm or larger that seem to be remnants of entanglements that were observed in the precemented necklace seen in Figure 2a. Elemental analysis by energy dispersive Xray (EDX) with a ~60 nm beam spot indicated the presence of CdS in the necklace (Figure S7). On cementing, the neckFigure 2. TEM and electrical analysis of the 1D branched assemblies. a) Discreet Au nanoparticle assembly resulting from the PSS–Cd–Au interaction. The inset scale bars are 20 nm. b) On addition of Na2S, the discreet nanoparticle assembly is transformed to a continuous structure. c) Current (I) in response to applied bias (V) for the necklace network between Au electrodes spaced 50 μm apart as a function of temperature. d) Differential conductivity corresponding to the response in (c). self-asseMbly of μM-long 1D networK of ceMenteD au nanoparticles 3 lace became conducting with nonlinear current–voltage (I– V) characteristics and a significant threshold bias VT required for current flow. (Figure 2c). This VT was indicative of a notable coulomb blockade effect at 296 K that increased at lower temperatures (Figure 2c–d). This observation was consistent with the 1D electron transport that is expected from the necklace morphology.[18] Bias, V > 40 V led to unstable currents and eventually to a loss of conductivity. The cause of this irreversible collapse was local resistive heating of the structure. Therefore, the bias was limited to a maximum value of 25 V during current–voltage measurements because this meant the structure was stable for several days. The behavior shown in Figure 2c–d indicates an increase in VT from ~1.8 V to ~15 V as the temperature is reduced from room temperature (RT) to 50 K. The increase in VT is not consistent with classical coulomb blockade behavior, which predicts no significant variation in VT with temperature [19] except for small effects such as those from the narrowing of the Fermi–Dirac distribution that affects electron tunneling.[20] Assuming a dielectric constant of ε = 3 for the medium surrounding the nanoparticle, for a device with a single isolated particle of diameter d = 10 nm, the estimated bias VT = e/2C = e/(4πεε0d) is ~50 mV corresponding to a barrier energy of 50 meV (i.e., 2kT at RT). Here, C is the capacitance of the particle, ε is the charge of an electron, and ε0 is the elec tric permittivity in vacuum. This increase in VT and the thermal behavior can both be explained by modeling the necklace as a percolating structure punctuated with nanoparticle islands made of one (singlet), two (doublet), or more particles (Figure 4). These islands can be thought of as defects resulting from CdS cement breaking off or as gaps, left by chance, that were too difficult for the cement to fill during the assembly process. The isolated particles are coupled to the percolating strands by a tunnel junction that is similar to a classical coulomb blockade setup. Thus, the large percolating clusters act as wires and the islands act as devices. At RT only the “single” particle clusters (i.e., singlets) will contribute to the blockade voltage. As the isolated particles are in series the net capacitance CT is