pubs.acs.org/JPCL
Solvent-Mediated End-to-End Assembly of Gold Nanorods Yiliang Wang, A. Eugene DePrince, III, Stephen K. Gray, Xiao-Min Lin,* and Matthew Pelton* Center for Nanoscale Materials, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, Illinois 60439
ABSTRACT We demonstrate a new method for the bottom-up assembly of anisotropic nanoparticles, showing that alkanethiol molecules can induce controlled end-to-end assembly of gold nanorods in mixed water/acetonitrile solutions. The assembly is driven by solvent-mediated interactions among hydrophobic alkanethiol ligands selectively bound to the ends of the nanorods and among hydrophilic cetyltrimethylammonium bromide (CTAB) surfactants on the sides of the rods. It occurs only when the gold-nanorod samples have been aged for approximately two weeks. We compare the kinetics of solvent-mediated assembly using undecanethiol ligands to assembly processes driven by covalent bonding using R,ω-undecanedithiol ligands and processes driven by hydrogen bonding using 11-mercaptoundecanoic acid ligands. Our experiments demonstrate the different assembly mechanisms involved as well as the conditions needed to obtain selective end-to-end assembly. SECTION Nanoparticles and Nanostructures
T
he bottom-up production of functional nanomaterials requires the ability not only to synthesize nanoparticles with controlled size, shape, and composition but also to arrange the particles into arbitrary configurations.1,2 The anisotropic assembly of nanoparticles into low-symmetry arrangements presents a particular challenge. Gold nanorods (GNRs) are ideal model systems for studying this problem because their optical extinction spectrum exhibits a strong peak due to the longitudinal plasmon resonance;3,4 when the nanorods are assembled, these plasmon resonances couple, forming new, hybridized plasmon modes that depend on the arrangements of the rods.5,6 This means that the assembly process can be monitored in situ by tracking changes in the optical spectrum of a GNR solution. Commonly used methods for GNR growth produce rods that are stabilized in aqueous solution by a bilayer of cetyltrimethylammonium bromide (CTAB) surfactant.7-9 The local density of CTAB is lower at the ends of the rods than on the sides, allowing thiol-terminated molecules at low concentrations to selectively displace CTAB and bind to the gold surfaces at the ends of the rods. If these molecules have an appropriate functional group opposite the thiol group, they can bind to one another, inducing end-to-end assembly of the nanorods. This was demonstrated using biomolecular recognition linkers,10-14 small functional molecules capable of forming intermolecular hydrogen bonds,15-19 and dithiol ligands that form covalent bonds to the gold surfaces.20,21 Side-by-side and end-to-end assembly was also demonstrated using electrostatic interactions among lyotropic chromonic molecules.22 Functionalization of the GNR ends with organic molecules often requires the use of a mixed aqueous-organic solvent. GNRs functionalized on their ends with hydrophobic organic molecules will have complex interactions with the mixed solvents since their sides continue to be covered with hydrophilic
r 2010 American Chemical Society
CTAB surfactants. It was found that nanorods that have been functionalized on their ends with large mercaptopolystyrene molecules assemble into end-to-end, side-by-side, or more complex configurations, depending on the solvent composition and the molecular weight of the polystyrene.23-25 On the other hand, hydrophobic alkanethiol molecules were previously found not to induce GNR assembly over an extended period of time,15,16,20 except for random aggregation in the case of extremely short alkane chains (1-propylmercptan).20 In this work, we show that long-chain alkanethiol ligands are in fact able to induce controlled end-to-end assembly of GNRs in a mixture of water and acetonitrile. By measuring the assembly kinetics for different solvent compositions and different concentrations of undecanethiol (C11SH), 11-mercaptoundecanoic acid (MUA), and R,ω-undecanedithiol (C11DT) ligands, we elucidate the mechanisms responsible for assembly as well as the conditions necessary in order to obtain a desired final product. We synthesize GNRs using a seed-mediated procedure.26 The GNRs in our experiments have an average aspect ratio of 4.3 ( 0.7, which results in a longitudinal plasmon resonance centered at a wavelength of 835 nm. Figure 1a shows the change of optical extinction over time after C11SH ligands are added to GNRs in a water-acetonitrile mixture such that the final ligand concentration is 1 μM. The initial conversion of GNR monomers into end-to-end dimers results in the decrease of the peak at 835 nm and the rise of a new peak at longer wavelengths, with a corresponding isosbestic point in the extinction spectrum.15 The formation of longer chains
Received Date: July 22, 2010 Accepted Date: August 25, 2010 Published on Web Date: September 01, 2010
2692
DOI: 10.1021/jz1010048 |J. Phys. Chem. Lett. 2010, 1, 2692–2698
pubs.acs.org/JPCL
assembly process confirm the existence of a significant amount of end-to-end assembly, as illustrated in Figure 1c and in the Supporting Information (Figure S10). As a control experiment, we monitor GNRs in the same solvent mixture without adding C11SH. Negligible changes in the extinction spectrum are seen over a period of 5 h, indicating that the addition of C11SH is necessary for the GNR assembly (see the Supporting Information, Figure S1). Our interpretation of the time-dependent experimental spectra is verified using rigorous electrodynamics simulations. Figure 1b shows the calculated extinction spectra for isolated GNRs and for end-to-end chains of up to four GNRs. The calculation results show a large initial shift between the spectrum of an isolated GNR and the spectrum of an end-toend dimer, with smaller shifts as additional GNRs are added to form a trimer and a tetramer. Our experiments thus clearly indicate that alkanethiol ligands can induce end-to-end assembly of GNRs in mixed solvents. This is contrary to the results of previous studies, where alkanethiol ligands with chain lengths between five and nine carbon atoms were not found to induce assembly.16,20 Although we use slightly longer carbon chains and we synthesize our GNRs using a seed-mediated rather than a photochemical method, we believe the most likely reason for the differences between our results and those of the previous studies is related to the aging of the samples. The data shown in Figure 1 are collected using a sample that has been precipitated after synthesis by centrifugation, redispersed in the same volume of deionized water, and then aged for approximately 2 weeks. As shown in Figure 2a, newly synthesized GNR solutions do not exhibit assembly when C11SH molecules are added. The freshly made samples are well passivated by a CTAB bilayer, inhibiting functionalization by C11SH. Assembly begins to occur for samples that have been aged for approximately 1 week, becoming more rapid as the samples are aged longer and stabilizing after 12 days (see Figure 2 and Supporting Information Figure S2). All other experiments reported here are therefore performed on samples that have been aged between 12 and 15 days. We speculate that this aging effect is due to reduction of CTAB coverage at the ends of the rods over time. We did not observe any significant changes in the ζ potential of the rods as they aged, suggesting that the CTAB coverage of the entire nanorod does not change significantly. On the other hand, the CTAB coverage at the ends of the rods is likely lower and less robust due to the higher curvature of the end surfaces and could be changing. This, in turn, could lead to significant increases in ligand exchange rates at the ends of the rods, resulting in faster self-assembly. Insight into the assembly mechanism can be obtained by quantitative analysis of the assembly kinetics. Our kinetics study is based upon monitoring the change of the single nanorod (monomer) concentration over time after the ligand was added. The extinction of the GNR solution at the singleparticle longitudinal plasmon resonance, At, is directly proportional to the nanorod monomer concentration through the Beer-Lambert law, At = εl[Ct], where ε is the molar extinction coefficient, l is the optical path length, and [Ct] is the nanorod monomer concentration at time t. Any assembly process
Figure 1. (a) Measured extinction spectra of gold nanorods in a solution of 10% water and 90% acetonitrile, following the addition of undecanethiol ligands such that the final ligand concentration is 1 μM. Curves i-xiv correspond to reaction times of 0-260 min, with a time interval of 20 min between measurements. (b) Calculated extinction spectra, normalized by the number of rods, for a single gold nanorod and end-to-end chains of two, three, and four nanorods. (c) Transmission electron micrograph of gold nanorods assembled in a solution of 10% water and 90% acetonitrile, 36 min after the addition of undecanethiol ligands, such that the final ligand concentration is 5 μM. The scale bar is 200 nm.
follows, indicated by a gradual bathochromic shift of the coupled plasmon peak. Transmission electron microscopy (TEM) studies of nanorods deposited from solution after the
r 2010 American Chemical Society
2693
DOI: 10.1021/jz1010048 |J. Phys. Chem. Lett. 2010, 1, 2692–2698
pubs.acs.org/JPCL
Figure 2. Measured extinction spectra of gold nanorods in a solution of 10% water and 90% acetonitrile, following the addition of undecanethiol ligands such that the final ligand concentration is 5 μM. For (a), the nanorod sample was aged for 1 day; curves i-iii correspond to reaction times of 0, 60, and 120 min. For (b), the nanorod sample was aged for 3 days; curves i-v correspond to reaction times of 0, 30, 60, 105, and 135 min. For (c), the nanorod sample was aged for 7 days; curves i-ix correspond to reaction times of 0-120 min, with a time interval of 15 min. For (d), the nanorodsample was aged for 11 days; curves i-x correspond to reaction times of 0-45 min, with a time interval of 5 min.
involving the formation of longer chains of GNRs is not directly considered in our kinetic analysis since these events cause spectral changes at longer wavelengths. If, for a certain portion of the experiment, the kinetics are dominated by a secondorder process of GNR dimerization, then the rate for this process will be proportional to k[Ct]2, where k is the kinetic constant for dimerization. The second-order rate equation thus becomes kt = εl(A0 - At)/(A0At), where A0 is the extinction at the beginning of the experiment.20 Normalizing the extinction with respect to A0 and plotting ε(1 - At)/At versus t therefore gives a straight line whose slope is proportional to k. Figure 3 shows a series of such plots for GNRs in a solution of 10% H2O and 90% CH3CN after different amounts of C11SH are added. The plots show clear straight-line portions, reflecting a second-order process of GNR assembly into endto-end dimers; linear fits to these portions give the dimerization kinetic constants, k. Before the dimerization process, there is an incubation period that corresponds to the time required for the ligands to displace CTAB from the ends of the rods and bind to the gold surface.20 The intercept of the straight-line fit with the time axis can be taken as an estimate of the ligand exchange time. Following dimerization, the kinetic curves deviate from straight lines as longer GNR chains begin to form.20 This general trend is observed for most experimental conditions, although in some cases, when the dimerization occurs much more readily, the ligand ex-
r 2010 American Chemical Society
Figure 3. Kinetics of GNR assembly in a solution of 10% water and 90% acetonitrile, following the addition of undecanethiol ligands such that the final ligand concentrations are 1 (squares), 2 (circles), and 5 μM (triangles). Solid lines are fits to a second-order reaction model.
change region becomes less obvious. The fact that we observe second-order kinetics indicates that the dimerization process dominates at early stages of assembly. Only later, when the
2694
DOI: 10.1021/jz1010048 |J. Phys. Chem. Lett. 2010, 1, 2692–2698
pubs.acs.org/JPCL
Table 1. Kinetic Constant for Dimerization, k, and Ligand Exchange Time, τexh, for the End-to-End Assembly of Gold Nanorods in a Mixture of 10% Water and 90% Acetonitrile, Following the Addition of Different Ligand Molecules R,ω-undecanedithiol
undecanethiol ligand concentration
k (106 M-1 s-1)
τexh (s)
k (106 M-1 s-1)
11-mercaptoundecanoic acid
τexh (s)
k (106 M-1 s-1)
τexh (s)
1 μM
3
4010
23
490
21
49
2 μM
11
2000
38
140
42