Precisely Shaped, Uniformly Formed Gold Nanocubes with Ultrahigh

Aug 28, 2018 - Foundation of Korea (NRF) funded by the Ministry of Science and ICT (MSIT) (Grant No. 2016R1A2A1A05005430) and the. National Research ...
0 downloads 0 Views 11MB Size
Letter Cite This: Nano Lett. XXXX, XXX, XXX−XXX

pubs.acs.org/NanoLett

Precisely Shaped, Uniformly Formed Gold Nanocubes with Ultrahigh Reproducibility in Single-Particle Scattering and SurfaceEnhanced Raman Scattering Jeong-Eun Park, Yeonhee Lee, and Jwa-Min Nam* Department of Chemistry, Seoul National University, Seoul 08826, South Korea

Downloaded via UNIV OF NEW ENGLAND on October 9, 2018 at 05:02:20 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: Synthesizing plasmonic nanostructures in an ultraprecise manner is of paramount importance because the nanometer-scale structural details can significantly affect their plasmonic properties. Au nanocubes (AuNCs) have been a highly promising, heavily studied nanostructure with high potential in various fields, but an ultraprecise synthesis from 10 to 100 nm in size over a large number of AuNCs has not been well established. Precisely structured AuNC-based studies for a highly reproducible, quantitative plasmonic signal generation [e.g., quantitative surface-enhanced Raman scattering (SERS)] are needed for reliable use and exploration in the beneficial properties of AuNCs. Here, we developed a strategy for AuNC synthesis with the desired size and shape, ranging from 17 to 78 nm particularly with highly controlled corner sharpness, by precisely controlling the growth rate of different facets and AuNC-specific flocculation which enabled ultrahigh yields (∼98−99%). Importantly, the precisely shaped AuNCs can scatter light in a spectrally reproducible manner, and the SERS enhancement factors (EFs) for the AuNC dimers are very narrowly distributed (the EFs of 72 nm sharp-cornered cube dimers have a distribution within 1 order of magnitude). Our results pave the paths to ultrahigh yield synthesis of metal nanocubes with a precise size and shape and offer single-particle-level spectral controllability and reproducibility over a large number of particles. KEYWORDS: Gold nanocube, localized surface plasmon resonance, Rayleigh scattering, surface-enhanced Raman scattering

L

heavily because of their facile synthetic method and strong plasmonic properties, AuNCs are more beneficial in that Au nanoparticles offer good chemical stability, capabilities in various surface modifications, conjugation chemistries, biocompatibility, and tunable plasmonic properties.4,17 However, the synthesis of structurally controlled AuNCs in an ultrahigh yield is challenging, and the use of AuNCs has been limited due to the lack of reliable synthetic strategies with ultrahigh precision and yield. Although some of the previously reported papers claimed high yields,18−20 supporting data such as large area transmission electron microscopy (TEM) and scanning electron microscopy (SEM) images for a large number of particles and extensive spectroscopic/microscopic analysis have not been provided to convincingly support such claims, and it is widely believed that highly reliable synthesis of AuNCs with varying sizes and shapes has not been well established yet. The low yield could be partially resolved by universal gold nanoparticle seeds acquired through an iterative oxidative dissolution and regrowth reaction.21 However, the intricate and laborious seed-prepara-

ocalized surface plasmon resonance (LSPR) is a unique feature of plasmonic metal nanostructures, and this enables numerous applications including sensing,1,2 imaging,3 therapeutics,4 nonlinear optics,5 and catalysis.6−8 Since LSPR is primarily affected by the size and shape of plasmonic nanostructures, plentiful studies have been conducted on the creation of new structures and precise structural controlling strategies. Among many plasmonic metal nanostructures, metallic nanocubes (NCs) have been heavily studied and utilized because of their strong and tunable LSPR properties and the availability of various synthetic strategies to form metal NCs.9−16 The edges and corners of metallic NCs generate a significantly enhanced localized electromagnetic field on individual nanoparticles or between coupled nanoparticles, which can be employed in diverse areas including fundamental optics, plasmon-enhanced spectroscopy, and sensing applications.9−13,15 Furthermore, owing to its flat surfaces, it is often used as a building block for assembled structures such as face-to-face assembled cube dimers or as a template for further synthesis of more complicated structures.14,15 Thus, synthesizing plasmonic NCs with high reproducibility and controllability is a fundamental building block and of great importance in the field of materials science, chemistry, plasmonics, and nanotechnology. Among plasmonic NCs, although silver NCs (AgNCs) have been utilized more © XXXX American Chemical Society

Received: July 20, 2018 Revised: August 23, 2018 Published: August 28, 2018 A

DOI: 10.1021/acs.nanolett.8b02973 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 1. Synthesis and refinement of size and shape-controlled AuNCs. (a) Synthesis of corner-sharpness-controlled AuNCs with varied bromide densities followed by refinement by centrifugation-driven depletion-induced flocculation in surfactant micelle solutions. The proposed mechanism of modifying growth kinetics is based on the difference in growth rates of each facet due to the difference in adsorption of Br−. (b) Definition of edge length, edge radius, and sharpness index (SI). Representative TEM images (c) and SI value (d) of AuNCs obtained by adjusting the concentration of bromide from 0 to 200 mM at the fixed amounts of seed and gold precursor. (e) Bromide-concentration-dependent growth kinetics. The results were obtained at each maximum LSPR wavelength using a UV−vis spectrophotometer. TEM images (f) and SI values (n = 100) (g) of refined AuNCs with different amounts of seed and bromide; the insets show representative single-particle images to clearly visualize the sharpness difference. The numbers in the labels correspond to edge length, and R and S indicate round-cornered and sharp-cornered AuNCs, respectively. (h) Normalized UV−vis spectra for a series of AuNC solutions. Solid and dashed lines correspond to round-cornered and sharp-cornered AuNCs, respectively. The scale bars in parts c and f indicate 20 and 100 nm, respectively.

signals,11,22,23 but systematic and reliable experimental realiza-

tion process compromises practical synthesis of AuNCs. Further, the accurate control in structural features such as edge sharpness for NCs and the synthesis of small-sized AuNCs have not been achieved so far. The detailed structural features such as edges and corners of NCs can largely affect their optical

tion is limited by the lack of a proper synthetic method. This could be particularly impactful for reproducible and quantitative plasmonic signal generations for surface-enhanced spectrosB

DOI: 10.1021/acs.nanolett.8b02973 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

centrifugation, centrifugal force pushes particles close to each other, giving extra driving force to flocculation. Therefore, effective flocculation can occur with a relatively low depletion potential in a very short time (97−98%. As a proof of concept, first, we controlled the corner sharpness of AuNCs in a similar size (Figure 1c, Figure S1, and Figure S2). To demonstrate the effects of the bromide amount, the bromide concentration was varied from 0 to 200 mM, while the amounts of CTAC, seed, precursor, and ascorbic acid were kept constant (see the Methods section in the Supporting Information for details). To quantitatively characterize the structural features of each AuNC on the basis of the TEM images, we coined a term, sharpness index (SI), based on the edge length defined as the distance between the two closest faces of AuNCs and the edge radius defined as the radius of the circle that perfectly matches the corner curvature (Figure 1b). As shown in Figure 1d, the corner sharpness initially increased and then decreased with increasing bromide concentration. This could be explained by the growth rate changes through adsorbed bromides on each facet, as illustrated in Figure 1a. In the case of excessive bromide concentration, bromide ions can be additionally adsorbed on (111)/(110) surfaces as well as (100), decreasing the overall growth rate and rate difference between facets. Further, we explored the growth kinetics during shape evolution by UV−vis spectroscopy with three selected bromide concentrations (1, 40, and 200 mM). After adding Au precursors, we started to monitor the changes in extinction intensity with 10 s intervals at each maximum LSPR wavelength of the fully grown structures (Figure 1e and Figure S3). The slowest increase of extinction for 200 mM NaBr suggests that increasing the bromide concentration slows the reduction, supporting the proposed mechanism. More experimental studies on the possibility of reduction rate change due to ligand exchange of Au−halide complexes, the effect of other halide ions, and the advantage of using CTAC and Br− are separately shown in the Supporting Information (Figures S4−S6). To substantiate the above principles, we controlled the size and corner sharpness simultaneously by changing the amounts of seed and bromide. The seed amount was adjusted by controlling the volume of seed solution (300, 30, 9, 6, and 2 μL). We also varied the bromide concentration to control the corner sharpness at a fixed seed amount. Compared to the synthesis of sharp-cornered AuNCs, we applied a lower density of bromide to produce the round-cornered AuNCs. The Br− densities adsorbed on the round-cornered AuNC and the sharp-cornered AuNC, obtained from ion chromatography, confirm a higher number of Br− on the sharp-cornered AuNC surface (Table S2). After synthesis, the AuNCs were dispersed in a mixture of benzyldimethyldodecylammonium chloride (BDAC) and cetyltrimethylammonium bromide (CTAB) solution and centrifuged for 5−10 min at 500−1000 g. Since flocculation force has a positive correlation with the overlaid surface area between nanoparticles as well as the micelle concentration, the surfactant concentration required for flocculation decreases as AuNCs become larger (see the Methods section in the Supporting Information for details). On the basis of the results obtained with varied flocculation conditions, we determined the optimal surfactant concentration for each NC size (Figure S7 and note S1). Figure 1f shows different sizes of refined AuNCs, and the insets show representative images for clear visualization of representative structures; the numbers in the labels indicate the edge length, and R and S correspond to round-cornered and

copies such as surface-enhanced Raman scattering (SERS) and surface-enhanced fluorescence. Here, we report a synthetic strategy to form the AuNCs with ultraprecisely controlled size and shape, ranging from 17 to 78 nm in diameter, in an ultrahigh yield (98.3 ± 0.7%). The edge sharpness, edge, and size of AuNCs were precisely controlled via anisotropic growth facilitated by the quantitative modulation of facet-selective protection. Moreover, the synthetic yield of a targeted cube structure was improved by centrifugation-driven depletion-induced flocculation. Single-particle-level scattering spectrum results show that the synthesized AuNCs generate strong and tunable scattering signals in a highly reproducible and quantifiable manner. Importantly, bulk UV−vis spectral analysis for different batches of AuNCs proves that the optical signals from AuNCs are quite reproducible over a large number of particles. Average SERS EF values from AuNC dimers with a 1.1 ± 0.3 nm gap were higher than 107. Notably, the EFs of sharpcornered AuNC dimers were distributed within 1 order of magnitude. This is a big step forward in SERS because dimeric structures often generate a SERS EF distribution over several orders of magnitude due to the heterogeneity in dimeric interfaces and gaps,24−26 hindering the use of SERS probes for reliable and practical applications. The strategy and results in this study open up ultra-high-yield synthesis of metal NCs with precisely controlled size and shape and reproducible, quantitative, plasmonically enhanced signals including SERS from metal nanostructures. By a seed-mediated method, we performed the selective surface-protection-directed growth of AuNCs surrounded by (100) facets (Figure 1a, left). In order to induce an anisotropic growth, the growth rate along [100] needs to be reduced, with the rates along [110] and [111] being higher. We introduced bromide ions that are favorably adsorbed on the (100) facet, in the presence of cetyltrimethylammonium chloride (CTAC) to make a difference in accessibility of Au precursors to different facets and induce the growth-rate difference between (100) and the other facets.27,28 The bromide-concentration-controlled growth kinetics governs the corner sharpness of AuNCs. At low bromide densities, the number of adsorbed bromide ions is insufficient to block the (100) facet completely. Hence, the relative growth-rate difference between [100] and [111]/[110] is less significant, producing round-cornered AuNCs. When a sufficient amount of facet-directing agents is provided, effective and preferential binding to the (100) facet decreases the reduction rate, while the (111)/(110) facets are less affected by bromide ions, maximizing the growth-rate difference between (100) and the other two facets to be sharp-cornered AuNCs. We then employed the shape-selective refinement strategy to maximize the synthetic yields for AuNCs (Figure 1a, right). The method is based on centrifugation-driven depletion-induced flocculation, which consists of the selective flocculation of AuNCs using surfactant micelles and reversible redispersion of the sediment. The flocculation has been applied to select nanorods or nanobipyramids among mixtures of nanoparticles and typically takes more than 10 h to make nanoparticles settle down.29−31 Since the attractive force between two particles is proportional to the flat surface areas facing each other, precisely synthesized AuNCs with flat surfaces are advantageous in flocculation compared to those with curved surfaces, such as rods, spheres, and bipyramids. In this study, we dramatically shortened the time by adopting a brief centrifugation step, and the efficiency for flocculation here was much higher than the previously reported cases with noncube structures. During the C

DOI: 10.1021/acs.nanolett.8b02973 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 2. Precisely shaped AuNCs in a high yield. Low-magnification SEM images showing AuNCs in a high yield after refinement. The samples were prepared by drop casting AuNCs on a silicon wafer. The image was taken at the boundaries where the particles were densely packed. The scale bars indicate 1 μm.

determine an appropriate surfactant condition to make small particles, 17S and 18R AuNCs, flocculated owing to their low surface area resulting in low flocculation potential. 32S AuNCs have flocculation potentials similar to those of their byproducts, providing relatively lower yields (∼90%). The long-term stability of AuNC solution stored in an ambient condition was confirmed by the extinction spectra acquired with AuNC samples stored for ∼16 months (Figure S11). Next, we obtained the light scattering images from individual particles using dark-field (DF) microscopy. The scattering properties of 18R, 17S, 37R, and 32S AuNCs could not be clearly measured owing to the low signal intensity caused by their small volumes. The DF micrographs of six different larger AuNCs exhibit a highly uniform scattering intensity for each sample, and the scattering colors were identical (green), except for 78R and 72S AuNCs, which showed a yellow color of scattering (Figure 3a). Significantly, the highly consistent spectra from 25 different individual AuNCs for each case are reminiscent of the utterly narrow size and shape distribution of the synthesized AuNCs and indicate a unique and reproducible spectral feature from each particle (Figure 3b). The representative spectrum of each sample is marked with a dotted box in Figure 3c. As the corner becomes sharper between AuNCs of similar sizes and the size increases, the average peak wavelength is red-shifted, which is mainly attributable to the retardation effect,32 along with the increase in scattering intensity (Figure 3d and e). The 72S AuNCs showed a slightly

sharp-cornered NCs, respectively. With decreasing seed volume, the edge length increased from 17 to 78 nm (Figure S8 and Table S1). Larger-sized cubes can also be synthesized by this method (data not shown). Although synthesizing AuNCs smaller than 25 nm by a solution-based method was challenging and has not been achieved before because the self-diffusion distance of gold atoms is higher than those of other metals such as platinum,18 we successfully synthesized AuNCs with an edge length of 17−18 nm using our method. To our knowledge, this is the smallest size reported so far for synthetically controlled AuNCs. While the SI values of four large, sharp-cornered AuNCs are similar, that of 17S was lower than the others (Figure 1g). It is because their edge radius was similar to 32S, which can be mainly attributable to the surface energy that prohibits smaller edge radius. Normalized UV−vis spectra for a series of AuNC solutions show a gradual red-shift as the corner sharpens and the edge length increases, owing to the retardation effect (Figure 1h).32 After the refinement, the yield was improved to >97−99% for all of the AuNCs, measured with the images with a large number of particles over a large area, except for the smallest two NCs and 32S (n > 400, Figure 2 and Figure S9 for larger SEM images; yields are 98.0% for 37R, 90.0% for 32S, 99.5% for 41R, 98.4% for 41S, 98.0% for 54R, 97.2% for 53S, 99.0% for 78R, and 98.2% for 72S). The narrower spectral line width and removal of the shoulder peak at ∼800 nm after refinement imply highly monodisperse AuNCs (Figure S10). It was more difficult to D

DOI: 10.1021/acs.nanolett.8b02973 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 3. Single-particle scattering analysis on AuNCs and UV−vis spectra of different batches of AuNC solutions. Dark-field microscope images (a) and 25 single-particle scattering spectra (b) of individual AuNCs (the particles in part a do not correspond to the particles in part b). (c) The corresponding representative spectra of each sample marked with a dotted box in part b. Analyzed maximum peak position (d), scattering intensity (e), and spectral line width (f) of each AuNC obtained from the analysis of part b. (g) The solution extinction spectra from nine different batches of 53S AuNC solutions. The exposure times for acquisition of scattering images in part a are 200 ms for 41R and 41S AuNCs, 120 ms for 54R and 53S AuNCs, and 80 ms for 78R and 72S AuNCs. The images in part a were additionally adjusted to visualize the scattering signal while ensuring differences among each other. We acquired scattering spectra of 41R and 41S AuNCs with an exposure time of 20 s and those of 54R, 53S, 78R, and 72S AuNCs with an exposure time of 3 s in part b. The 20 s cases were normalized to the 3 s case under the assumption that the signals increase linearly with exposure time. The scale bars in part a indicate 2 μm.

lower scattering intensity compared to that of 78R AuNCs owing to the increased radiative decay for the larger NCs, as the light scattering scales with the square of the particle volume in general. Notably, although the maximum peak position of 78R was 562 nm as in the 53S case, an increased contribution from wavelengths greater than 600 nm results in a distinct color difference in the micrograph, as shown in Figure 3a. Analyzed spectral line widths also manifest the uniform scattering property (Figure 3f). When reference AuNCs synthesized according to the previously reported method18 were compared to our AuNCs of a similar scattering intensity, our AuNCs showed the line widths with clearly smaller deviations (Figure S12), implying the high homogeneity in particle structures synthesized with our method. Importantly, we also discovered the extraordinary reproducibility of the AuNCs by synthesizing nine different batches of 53S AuNC solutions and comparing their ensemble extinction spectra in solution (Figure 3g). The spectral information including the maximum peak position and line width suggest negligible batch-to-batch variations in the synthesized AuNCs, proving the reproducibility in structure and spectrum are very high in bulk scales. Formation of dimeric or multimeric structures is a typical way to create plasmonic couplings between particles to strongly enhance the signals from plasmonic nanostructures.4,33,34

Among those structures, cube dimers have gained attention as a plasmonically coupled nanostructure with a flat gap, showing a relatively uniform electric field inside the gap.35−39 However, experimentally obtaining reproducible and reliable signals from these structures with a largely enhanced plasmonic signal (typically, many orders of magnitude enhancements) is highly challenging. To investigate our dimeric AuNC structures with an ultrasmall nanogap (∼1 nm) on near-field enhancement, we observed the 1,4-benzenedithiol (BDT) SERS signal from face− face oriented AuNC dimers with two differently cornered AuNC structures at the single-particle level (Figure 4a and d). BDT was used as a linker molecule to assemble the NCs and to form a nanogap between NCs. Assembled NCs were loaded to a SiO2-supported TEM grid and mapped with TEM. Then, TEM images were correlated with photomultiplier tube images to distinguish face−face dimers from monomers and other types of assemblies including multimers and face−edge dimers in the Raman measurement setup (Figure S13). From the analysis of 22 different dimers for each case, the average gap size was obtained as 1.1 ± 0.3 nm (Figure S14). The Raman signal polar plot from the BDT with varying incident laser polarizations reveals that the maximum Raman enhancement occurs at the longitudinal mode of dimers (Figure S15). The assembled dimers generated highly symE

DOI: 10.1021/acs.nanolett.8b02973 Nano Lett. XXXX, XXX, XXX−XXX

Letter

Nano Letters

Figure 4. Single-particle SERS from AuNC dimers with face−face orientation. (a) Schematic illustration of face−face oriented AuNC dimers (78R and 72S AuNC dimers) with 1,4-benzenedithiol. (b) Cross-sectional images of simulated near-field enhancement for 78R and 72S dimers, respectively. The solid line depicts the boundary of AuNCs, and dotted boxes represent the flat interfacial area for each dimer. (c) 3D spatial plot images of simulated electric field enhancements in part b. (d) Representative Raman scattering spectra from individual AuNC dimers. (e, f) Single-particle SERS enhancement factor distributions for the Raman peak at 1069 cm−1 [log scale (e) and linear scale (f) obtained in the boxed regions in part e] from the measured Raman signals for 78R NC and 72S NC dimers, respectively (n = 22 for each case).

controllable hot spots and widely distributed EF values. Our NC-dimer-based SERS EF distribution results are reminiscent of the ultrahigh uniformity of AuNC structures, and can also be attributed to the uniformly distributed electromagnetic field in the flat gap of the assembled AuNCs, particularly for sharpcornered 72S cube dimers (Figure 4b and c). Although the maximum field enhancement is a bit higher for round-cornered AuNCs, the smaller electric field gradient for 72S cube dimers than 78R cube dimers can lead to a higher homogeneity in SERS signals, as 72S provides twice larger interfacial area and can form more uniform interfaces between particles. In this manner, the control of corner sharpness can be employed as one of the strategies for tuning the optical properties in assembled structures as well as in individual particles. In conclusion, we have developed a straightforward, generally applicable synthetic strategy to ultraprecisely control the cube size (17−78 nm in edge length) and shape, including the corner sharpness of AuNCs, in a very high yield. Quantitative control of bromide ions induces a growth rate difference between multiple facets, yielding sharp-cornered or round-cornered NCs in a highly selective manner, and the subsequent AuNC-specific flocculation process led to >98−99% yields for a targeted AuNC structure. The ultrahigh structural precision for size and sharpness and high synthetic yield enable the fine-tuning of plasmonic optical responses from AuNCs, allowing single-

metrical polar plot results, suggesting that the dimers were formed impeccably and symmetrically. The averaged EF of the 78R dimer is ∼8.0 × 107, which is 5 times higher than that of 72S, being ∼1.6 × 107, with a linearly polarized laser along with the longitudinal dimer axis. These values are believed to be strong enough for single-molecule detection experiments (Figure 4e and f).40 This trend is consistent with the electromagnetic-field simulation results, which show a higher maximum field enhancement for round-cornered AuNCs than sharp-cornered AuNCs (Figure 4b and c). Importantly, we found that a very narrow distribution of SERS EFs was acquired with both 78R and 72S cube dimers. The SERS EFs for the entire dimers that we analyzed ranged from 1.6 × 107 to 3.1 × 108 for 78R cube dimers and from 5.4 × 106 to 4.9 × 107 for 72S cube dimers (Figure 4e and f). It is remarkable that the SERS EFs are distributed within 1 order of magnitude for the 72S cube dimer case, and this is the narrowest distribution of SERS EFs with dimeric structures ever reported. Given that uniformly formed Ag nanosphere arrays41 generated widely distributed SERS EF values between 2.8 × 104 and 4.1 × 1010 and smaller SERS EF values from face-to-face AgNC dimers25 are broadly distributed from