Two-Dimensional Bipyramid Plasmonic Nanoparticle Liquid

Jan 5, 2016 - Two-Dimensional Bipyramid Plasmonic Nanoparticle Liquid Crystalline Superstructure with Four Distinct Orientational Packing Orders...
0 downloads 0 Views 8MB Size
Two-Dimensional Bipyramid Plasmonic Nanoparticle Liquid Crystalline Superstructure with Four Distinct Orientational Packing Orders Qianqian Shi,†,‡ Kae Jye Si,†,‡ Debabrata Sikdar,§ Lim Wei Yap,†,‡ Malin Premaratne,§ and Wenlong Cheng*,†,‡ †

Department of Chemical Engineering and §Advanced Computing and Simulation Laboratory (AχL), Department of Electrical and Computer Systems Engineering, Faculty of Engineering, Monash University, Clayton 3800, Victoria, Australia ‡ The Melbourne Centre for Nanofabrication, 151 Wellington Road, Clayton 3168, Victoria, Australia S Supporting Information *

ABSTRACT: Anisotropic plasmonic nanoparticles have been successfully used as constituent elements for growing ordered nanoparticle arrays. However, orientational control over their spatial ordering remains challenging. Here, we report on a self-assembled two-dimensional (2D) nanoparticle liquid crystalline superstructure (NLCS) from bipyramid gold nanoparticles (BNPs), which showed four distinct orientational packing orders, corresponding to horizontal alignment (H-NLCS), circular arrangement (C-NLCS), slanted alignment (S-NLCS), and vertical alignment (V-NLCS) of constituent particle building elements. These packing orders are characteristic of the unique shape of BNPs because all four packing modes were observed for particles with various sizes. Nevertheless, only H-NLCS and V-NLCS packing orders were observed for the free-standing ordered array nanosheets formed from a drying-mediated self-assembly at the air/water interface of a sessile droplet. This is due to strong surface tension and the absence of particle−substrate interaction. In addition, we found the collective plasmonic coupling properties mainly depend on the packing type, and characteristic coupling peak locations depend on particle sizes. Interestingly, surface-enhanced Raman scattering (SERS) enhancements were heavily dependent on the orientational packing ordering. In particular, V-NLCS showed the highest Raman enhancement factor, which was about 77-fold greater than the H-NLCS and about 19-fold greater than C-NLCS. The results presented here reveal the nature and significance of orientational ordering in controlling plasmonic coupling and SERS enhancements of ordered plasmonic nanoparticle arrays. KEYWORDS: self-assembly, bipyramid nanoparticles, plasmonic nanoparticle liquid crystalline superstructure, orientational packing, SERS he ability to precisely assemble “meta-atoms” in the artificial nanoparticle periodic table1 into ordered arrays has implications in photonic circuits, biosensing, and metamaterials.2−7 In particular, two-dimensional (2D) ordered plasmonic nanoparticle arrays can take full advantage of the shape-dependence, spatial arrangement, and directional proper-

ties of the nanoparticle.8−10 Furthermore, adjusting the structures in the hierarchical assembly of nanoparticles to couple

T

© XXXX American Chemical Society

Received: October 2, 2015 Accepted: January 5, 2016

A

DOI: 10.1021/acsnano.5b06206 ACS Nano XXXX, XXX, XXX−XXX

Article

www.acsnano.org

Article

ACS Nano

Figure 1. (a) Scheme of self-assembly fabrication process of BNP NLCSs. SEM characterization and 3D scheme of (b) H-, (c) C-, (d) S-, and (e) V-NLCSs made from mBNPs formed on a Si wafer. Insets in SEM images and top of 3D pictures show the cross-section of corresponding arrangements. The scale bars are 100 nm.

the external electromagnetic fields can lead to a strong confinement of light within nanoscale gaps,2 hence enabling their exciting applications in novel surface-enhanced Raman scattering (SERS) substrates.11−15 Sizes and shapes of constituent nanoparticle building blocks play a critical role in growing ordered arrays as well as in engineering their properties. To date, simple shapes, such as nanospheres, and complex anisotropic shapes such as nanorods, nanocubes, nanostars, and nanocuboids have been successfully assembled into ordered arrays using DNA-based16−18 or polymer-based8−12 strategies. These advances render the way forward for fabricating plasmonic ordered arrays at will simply using various artificial “meta-atoms” in the plasmonic periodic tables.1 However, the current progress has been muted to some extent owing to the difficulties resulting from the existence of complex nanoscale forces among plasmonic nanoparticles.19 We are far from the capability of manipulating nanoparticles as a chemist manipulates atoms. To date, due to the extreme difficulties in controlling nanoscale forces, only a limited number of nanoparticles have been successfully utilized in the fabrication of 2D ordered arrays.8,10,14,15,20−24 This is particularly true for anisotropic nanoparticles, in which the control over orientational order remains a challenge. The ability to control orientational ordering of anisotropic shapes can lead to the design of “one particle

shape, multiple superlattices”.14 Previously, two packing orders were observed for gold nanorods,8 semiconductor nanodisks,22 and silver nanoprisms;25 control over surface chemistry can lead to three kinds of ordered arrays from octahedral silver nanoparticles.14 Here, we report on a new type of plasmonic nanoparticle liquid crystalline superstructure (NLCS) using bipyramid gold nanoparticles (BNPs) as building blocks, which shows four types of packing orders, namely, horizontal alignment (H-NLCS), circular arrangement (C-NLCS), slanted alignment (S-NLCS), and vertical alignment (V-NLCS). We thoroughly investigated the effects of particle size, capping ligand length, evaporation rate, and substrates on these packing alignments. The orientational packing ordering directly influenced plasmonic coupling strength and modes and, hence, determined the SERS enhancements of plasmonic NLCSs. Among all the packing modes, the V-NLCS led to the highest Raman enhancement factor (EF), which could be further adjusted by tuning the particle size. These results were well-predicted by our numerical simulations.

RESULTS AND DISCUSSION BNPs were synthesized following the seed-mediated growth method reported earlier.26 The structure of the as-synthesized BNPs corresponds to twin pentagonal pyramids aligned along the length axis with rounded tips.27−29 Suspensions of purified B

DOI: 10.1021/acsnano.5b06206 ACS Nano XXXX, XXX, XXX−XXX

Article

ACS Nano

Figure 2. UV−vis spectroscopic data of four arrangements in NLCSs from (a) s-, (b) m-, and (c) lBNPs on ITO glass. Dashed lines in (a)−(c) represent the peak position of the BNP solution. The arrows show the positions of LSPRlong and LSPRtrans of BNPs in H-NLCSs.

Figure 3. TEM micrographs of mBNP NLCSs showing (a) horizontal and (c) vertical structure. Insets show the hexagonal close-packed arrangements in (a) and (c). (b and d) The corresponding UV−vis spectroscopic data from experiment and the simulation modes of (a) and (c). The rectangles depicted in (b) and (d) highlight the unit cell of the hexagonal lattice of H- and V-NLCS sheets, respectively.

changed from dark green to brown, then to orange-brown as the particle size increased (inset of Figure S1). We deliberately designed BNPs with three different sizes for later use in this work, namely, small BNPs (sBNPs, LSPRlong = 720 nm, LSPRtrans = 509 nm), medium BNPs (mBNPs, LSPRlong = 737 nm, LSPRtrans = 511 nm), and large BNPs (lBNPs, LSPRlong = 767 nm, LSPRtrans = 512 nm). Their respective lengths are 49.7 ± 2.8, 56.8 ± 2.6,

BNPs showed two dominant characteristic plasmonic peaks, corresponding to a longitudinal localized surface plasmon resonance (LSPRlong) mode in the low-energy region and a transverse localized surface plasmon resonance (LSPRtrans) mode in the high-energy region (Figure S1).30−33 By decreasing the amount of seeds, BNPs increased in size,27 leading to red-shifts of the two LSPR peaks. Typically, the color of a gold BNP solution C

DOI: 10.1021/acsnano.5b06206 ACS Nano XXXX, XXX, XXX−XXX

Article

ACS Nano and 65.1 ± 2.8 nm and respective widths are 17.4 ± 1.8, 18.7 ± 1.4, and 21.7 ± 1.2 nm (Figure S2). The purified BNPs were then used for growing NLCSs via a drying-mediated self-assembly process as illustrated in Figure 1a. First, BNPs were modified by thiolated polystyrene (PS) following our ligand exchange process reported earlier,10 rendering them hydrophobic and dispersible in chloroform. Then a droplet of chloroform solution was drop-casted onto a sessile water drop sitting on a solid substrate. Initial quick evaporation of chloroform followed by slow evaporation of water led to the formation of a two-dimensional NLCS. For all three different building block sizes (sBNPs, mBNPs, and lBNPs), we observed four distinct packing orders (Figures S3). The arrangements in mBNPs NLCS are demonstrated by top and cross-sectional views of scanning electron microscope (SEM) images, namely, H-NLCS (Figure 1b), C-NLCS (Figure 1c), S-NLCS (Figure 1d), and V-NLCS (Figure 1e). For HNLCS, all the neighboring BNPs lay down on the surfaces with nematic-like configurations; for C-NLCS, all the BNPs lay down on the surfaces with the nematic phase in the middle but parallel to each other at the perimeter of the circular patches; for SNLCS, the BNPs were slanted with one tip sitting on the substrate and the other one extruding at a certain angle with smectic-like phases; for V-NLCS, the particles were standing up and vertically aligned, forming crystalline phases. Additionally, the images of these four kinds of NLCSs were representative of large-scale ordering (Figure S4). The collective plasmonic properties of the NLCSs depended on both orientations and sizes of the BNPs. It can be seen from Figure 2 that the H-, C-, and S-NLCSs possessed two peaks, a weak shoulder peak and a dominant coupling peak, which correspond to LSPRtrans and LSPRlong of BNPs (see arrows in Figure 2).31 For V-NLCSs, only transverse peaks were found at 560, 564, and 570 nm for sBNPs, mBNPs, and lBNPs, because longitudinal plasmons are not excited on the vertically aligned nanoparticles. Both the longitudinal and transverse plasmonic peaks in the spectrum of H- and C-NLCSs exhibited red-shifts (50, 50, and 55 nm for s-, m-, and lBNPs of transverse peaks; 101, 119, and 140 nm for s-, m-, and lBNPs of longitudinal peaks, respectively) compared with the spectrum obtained from discrete BNPs in solution, which are the result of the interparticle coupling between the BNPs.10 Interestingly, the longitudinal extinction peaks of S-NLCSs are close to LSPRlong peaks of a BNP solution, indicating weak coupling of longitudinal plasmonic resonance modes. We further investigated the important experimental parameters to control orientational packing ordering. By increasing humidity during self-assembly, we could substantially reduce the growth kinetics of plasmonic NLCSs (Figure S5a). The humidity was controlled to ensure the full evaporation of water took 3 days to accomplish. Consequently, this slow-evaporation process led to the formation of dominant V-NLCSs (Figure S5b and c). According to three 2D NLCSs fabricated from different droplets, the rough fractions of V-NLCSs are over 90%. Note that VNLCS corresponds to a thermodynamically minimum energy state; however, kinetically trapped states often appear due to the stochastic evaporation process. By slowing down the rate of water evaporation, each BNP has enough time to find its preferred position with the lowest energy state. This finding is consistent with the previous self-assembly of nanorods.34−36 Furthermore, it was found that the determining parameter for fabrication of C-NLCSs was the concentration of BNP solutions. By reducing the BNP concentration while maintaining other

experimental parameters, we obtained an NLCS with nearly all circular arrangements (Figure S6). This can be attributed to the insufficient number of BNPs during the evaporation of chloroform and, hence, the inability to form a continuous film, but many small circular islands instead. In addition, substrate plays a key role in the self-assembly process. By using a holey copper grid following the same approach as depicted in Figure 1a, we could obtain free-standing plasmonic NLCS sheets. However, transmission electron microscope (TEM) characterization demonstrated that only H- and V-NLCSs were observed (Figure 3a and c) in freestanding sheets. This is due to strong surface tension and the absence of particle−substrate interaction. As expected, both LSPRlong and LSPRtrans peaks could be observed for free-standing H-NLCS sheets (Figure 3b), whereas only an LSPRtrans peak was found for V-NLCS sheets (Figure 3d). Similar results were obtained by our simulation (red lines in Figure 3b and d). Nevertheless, the simulated peak location and line width did not match exactly with experimental results. For instance, an additional shoulder peak was observed experimentally for VNLCS sheets but was not predicted by theoretical simulation. These variations can be attributed to the imperfection of H- and V-NLCS packing, which is evident in Figure 3c, showing a small fraction of BNPs adopting a laid down arrangement in the VNLCS sheet. Furthermore, the experimental UV spectra are broader than the simulation result, which arises from the difference in coupling modes and size distribution. Besides, there are instances (see Figure 3a) where the long axes of the adjacent nanoparticles are not parallel, in which case the increased interparticle interaction owing to reduced spacing contributed to a red-shift in the experimentally observed resonance peak. A similar trend of a minor red-shift is also observed in Figure 3d. Despite these minor mismatches, our simplifying assumptions can closely match the spectral positions of transverse and longitudinal peaks in Figure 3b and the transverse peak in Figure 3d. Notably, the packing types were found to be independent of particle sizes, as observed for the three different BNP NLCS sheets in Figure S7b−j. However, the nanoparticle sizes do affect the ordering parameter, which shows an improved ordering when particle size decreases in H-NLCS sheets (Figure S7a). Compared to previous nanorod-based systems, BNPs led to a better ordering likely due to high monodispersity of particle sizes. In addition, the length of polymeric ligands is also a key factor that directly affects the packing order and quality of the twodimensional BNP NLCSs. We investigated this effect by assembling BNP NLCSs with three different thiolated polystyrene ligands: molecular weights of 50 000, 20 000, and 12 000 g/mol. The assembled BNP NLCSs are shown in Figure S8. With a long PS ligand (Mn = 50 000 g/mol), the assembled structure exhibited a loose and disordered structure (Figure S8a), while shorter PS ligands (Mn = 12 000 g/mol) led to formation of a cracked, multistacked disordered structure (Figure S8c). In contrast, the medium-sized ligands led to highly ordered arrays under the same conditions (Figure S8b). These observations may be explained based on the interplay between the nanoscale interaction forces that were present during the self-assembly process. In particular, to form ordered arrays in our systems, the van der Waals (vdW) attraction forces need to be balanced with steric hindrance forces of the ligands. For longer ligands, steric forces are likely to dominate, whereas for shorter ligands, vdW forces outweigh the steric hindrance forces. In either cases, ordered lattices could not be obtained. D

DOI: 10.1021/acsnano.5b06206 ACS Nano XXXX, XXX, XXX−XXX

Article

ACS Nano

Figure 4. SERS spectra of 4-ATP from four arrangements of mBNP NLCSs with laser wavelengths of (a) 633 nm, (b) 782 nm, and (c) 830 nm. (d) Corresponding EFs calculated from C-, V-, and H-arrangements. (e, f) Simulated patterns of the normalized electric field (E) distribution in (e) V-NLCS and (f) H-NLCS showing the hot-spots generated for laser excitation at 633, 782, and 830 nm. The inner rectangles (in pink and gray) in the schematics represent the unit cell considered for simulation of V- and H-NLCS, respectively, whereas the outer rectangles (in orange) in both cases show the area of the sheet used for depicting the electric field patterns.

The observed four packing orders are a result of unique bipyramid shapes and one dimensionally confined self-assembly at the air/water interface. All the ordering processes could be explained by the volume restriction argument. As the solvent

evaporates, the concentration of BNPs increases. In theory, when the particle number density, n (i.e., the number of BNPs per unit volume), is greater than the critical value nc, a phase transition from disordered state to ordered state will occur. Additionally, E

DOI: 10.1021/acsnano.5b06206 ACS Nano XXXX, XXX, XXX−XXX

Article

ACS Nano V cS− ex = πrS2(L + 2h ic)sin θ

since BNPs have a similar elongated shape to nanorods, the ordering mechanism can be explained by Onsager’s model.37−39 According to Onsager’s model, rotational and translational entropy compete during the ordering transitions of hard rods. Rotational entropy is maximized when the rods have the possibility to point in any orientation, which decreases as alignment increases. Translational entropy has the highest value when the excluded volume is minimized (the accessible volume is maximized).19,40 When the translational and rotational freedom are restricted in a high-concentration solution, some of their orientational freedom was sacrificed to increase translational freedom by forming a parallel structure.8,19,41 These theories could qualitatively explain the formation of our H-, S-, and VNLCSs. Nevertheless, Onsager’s model considered only interactions among rigid rod or rod-like shapes. In practice, our BNP particles are capped with a fairly thick layer of polymeric ligands whose roles cannot be ignored for packing ordering. This layer of adaptive organic corona would be compressed by strong surface tension during the drying-mediated self-assembly process, but they could accommodate various packing by adjusting their configuration. The adaptive nature of the soft corona allowed us to assume polymer-coated BNP particles as soft nanocylinders that have an incompressible organic layer (pink dashed line in Supporting Figure S9a−c). According to the center-to-center distances that were measured between vertical, slanted, and horizontal BNPs (Figure S9g−i), the interparticle distances in VNLCS and S-NLCS were estimated to be ∼3.6 and ∼11.6 nm, while the side-to-side and edge-to-edge distances between two BNPs in H-NLCS were ∼14.4 and ∼7.8 nm, respectively. Therefore, we assume the thickness of the incompressible organic layer at the tips is half of the interparticle distance between vertical BNPs, which is ∼1.8 nm. Under this 1 assumption, the critical number density nc = 2

VcV− ex = πrv 2(L + 2h ic)

where VHc−ex, VSc−ex, and VVc−ex are the confined excluded volumes of horizontal, slanted, and vertical BNPs, rv is the center-tocenter distance between vertical BNPs (Figure S9d and g), rS is the center-to-center distance between slanted BNPs (Figure S9e and h), and Lss and Lee are the central distances between two BNPs in the side-by-side and end-to-end orientations in horizontal BNPs (Figure S9f and i), respectively. Hence, the values of VHc−ex, VSc−ex (where θ is assumed to be 45°), and VVc−ex were calculated to be 1.90 × 105, 1.23 × 105, and 0.94 × 105 nm3, respectively. Evidently, the value of VVc−ex is smaller than that of VSc−ex and VHc−ex. The smaller the excluded volume, the larger the translational entropy.40 Therefore, the translational entropy is maximized as more free volume is left to the system in a closepacked V-NLCS. It can be noted that the S-NLCS can be viewed as the transitional state from H- to V-NLCSs. The circled arrangements were caused by the microdroplet of chloroform at the low-concentration region during evaporation. Finally, these structures will “freeze” under different solvent layers after further evaporation of chloroform. It is worth mentioning that the final stage of chloroform evaporation could generate vortexes, leading to spiral-like assemblies of BNPs due to inward capillary forces (e.g., yellow circles in Figure S5c), similar to previous observations by others.42 Similar to our previous ordered plasmonic nanoparticle arrays,11−13 BNP-based NLCSs could also be used as SERS substrates. First, a BNP NLCS was subjected to plasma treatment for removal of surface ligands. Figure S11 reveals that all four types of BNP NLCSs maintained their structural integrity after plasma treatment. We used 4-aminothiophenol (4-ATP) as a model Raman analyte to investigate the effects of orientation order on SERS performance. Interestingly, we found that the EF was directly related to the orientational packing of BNPs. Figure 4a−c show the recorded SERS spectra of 4-ATP at laser wavelengths of 633, 782, and 830 nm, which were dominated by a set of peaks located at 1078, 1141, and 1578 cm−1.11 Among them, the strong fingerprint vibrational band at 1078 cm−1 was used to calculate the EF (see Supporting Information section III for sample EF calculation). It is known that the highest EF is usually obtained when the plasmonic resonance band matches closely the excitation laser wavelength.11 Specifically for our BNP NLCSs, the peak position of V-NLCS is located at 564 nm, which is close to the laser wavelength of 633 nm, while the LSPRlong peaks of H-NLCS and C-NLCS are located near the laser of 830 nm. As depicted in Figure 4d, the maximum EF calculated based on the 1078 cm−1 mode for vertical, horizontal, and circled arrangements were obtained at laser wavelengths of 633, 830, and 830 nm, which agrees well with the theoretical prediction. Due to the difficulty in calculating the exposure area of slanted BNPs, the EF of the S-NLCS was omitted here. With respect to all the laser wavelengths (633, 782, 830 nm) tested in our experiment, the VNLCS always gave the highest EFs among all packing orders. In particular, the EF of the V-NLCS (at a laser wavelength of 633 nm) was 77-fold and 19-fold greater than that of H- and CNLCSs, respectively. This may be explained by the lightning rod effect from the BNP tips and the reduced gaps, which resulted in stronger hot-spots in comparison with H- and C-NLCSs. To prove this, we compared their SERS enhancement with that of a V-NLCS fabricated from gold nanorods that had similar sizes to mBNPs (Figure S12). The results show that the BNP V-NLCS

π (L + 2h ic)(D + 2h ic)

(L and D are the length and width of BNPs, respectively; hic is the thickness of the incompressible organic layer at the tips)19 was calculated to be 1.1 × 104 μm−3. The calculated number densities were 4.7 × 104, 3.7 × 104, 3.5 × 104, 3.3 × 104 μm−3 for H-, C-, S-, and V-NLCSs, respectively. This indicated the occurrence of a phase transition during the assembly process, which led to the orientation of NBP particles from a disordered to an ordered state spontaneously. Considering soft nanocylinders confined within thin chloroform films, we can estimate the excluded volumes of BNP particles for various orientations before the final fully dried state (Figure S9d−f). Prior to full evaporation of chloroform, the solvent layer thickness may vary due to thermal fluctuation from region to region. The fluctuating chloroform film thickness before its full evaporation could be the reason for the three types of configurations observed: vertical, slanted, and horizontal alignments. The final V-, S-, and H-NLCSs also reflected the history of this confined assembly process of BNPs. Within the confined environment, it is expected that the excluded volumes differ from their corresponding scenarios for unconfined bulk solution. Considering the thickness of the confined volume within chloroform films matches the heights of standing, slanting, and lying BNPs, we could estimate the excluded volumes under confined environment (Vc−ex) by the following equations (see Supporting Information section II for further details): VcH− ex = 4LssLee(D + 2h ic) F

DOI: 10.1021/acsnano.5b06206 ACS Nano XXXX, XXX, XXX−XXX

Article

ACS Nano

Figure 5. (a) Schematic representation of the V-NLCS used in the simulation study, with the inner pink rectangle marking the unit cell and outer orange rectangle depicting the area of the sheet shown in the electric field patterns in (c); (b) EFs of 10 μM 4-ATP from small, medium, and large BNP V-NLCSs; (c) simulated patterns of the normalized electric field (E) distribution in three different sheets of small, medium, and large VNLCSs showing the hot-spots generated for laser excitation at 633 nm.

offers dramatically higher SERS enhancement than an ordered array of gold nanorods. It is also worthwhile to mention that the C-NLCS showed stronger SERS enhancement than the HNLCS due to the tip-to-tip structure that existed between the perimeters of two circular patches. To further explain the observed SERS enhancements, a numerical simulation was performed to calculate the electric-field distribution of four types of NLCSs at laser excitation wavelengths of 633, 782, and 830 nm. Simulated patterns of the normalized electric field distribution in a V-NLCS showed the strongest hot-spots at a laser excitation of 633 nm and gradually decreases in strength as the wavelength increases (Figure 4e). A similar study for H-NLCS showed an inverse trend where the strength of the hot-spots gradually increases when laser excitation wavelength gets longer, providing the strongest hot-spots at 830 nm (Figure 4f). Since the highest EF was achieved for V-NLCS under a laser wavelength of 633 nm, we further investigated the effect of particle size on Raman enhancement. The increase in particle size was observed to result in an increase of the enhancement factor (Figure 5b). This relationship was attributed to the fact that the transverse peak position of BNPs red-shifted (from 560 to 570 nm) closer to the laser wavelength (633 nm) as the width of the particle increases (Figure 2). As a result, the peak position of large BNPs is the closest to the excitation wavelength, in turn, strengthening the SERS signal. However, the size effect on EF is not as significant as that from packing type. In particular, comparison of the EF from V-NLCS shows that lBNPs are only 1.01 and 1.61 times stronger than mBNPs and sBNPs. This is

mainly caused by the fact that the differences between transverse peak positions of s-, m-, and lBNPs is much smaller than those between main peak positions of C-, H-, S-, and V-NLCSs. The normalized electric field distribution from simulated patterns of small, medium, and large V-NLCSs also illustrated that the strength of hot-spots was enhanced and more plasmonic coupling occurred as the particle size increases (Figure 5c).

CONCLUSION In conclusion, we report on a new type of 2D plasmonic nanostructures with four distinct orientational packing orders. We systematically investigated the formation of these packing orders under various conditions. The orientational control over bipyramid nanoparticles has a direct implication on plasmonic coupling and near-field electric fields (or Raman hot-spots). Both experimental and theoretical studies were undertaken to understand the effect of orientational packing order on plasmonic properties and SERS EF. In particular, the vertical arranged BNP NLCS exhibited the highest SERS EF of ∼77-fold higher than the horizontally arranged BNP NLCS. These results demonstrate the versatility of size/shape control in designing novel two-dimensional plasmonic nanostructures with a myriad of potential applications in electronics, photonics, metamaterials, sensing, and anticounterfeiting. METHODS Materials. Gold(III) chloride trihydrate (HAuCl4·3H2O, ≥99.9%), sodium borohydride (NaBH4), trisodium citrate, hexadecyltrimethylammonium bromide (CTAB), silver nitrate (AgNO3), L-ascorbic acid, G

DOI: 10.1021/acsnano.5b06206 ACS Nano XXXX, XXX, XXX−XXX

Article

ACS Nano

Characterization. The morphology and assembled structure of the BNPs and NLCSs were observed through TEM (FEI Tecnai G2 T20 TWIN LaB6 TEM operating at 200 kV) and SEM (FEI Helios Nanolab 600 FIB-SEM operating at 5 kV). An Agilent 8453 UV−vis spectrophotometer was used to measure the absorption spectra of the BNPs in water. For each size of BNP NLCSs, we prepared at least three samples: one on silicon wafer (used for the Raman test), one on ITO glass (used for the micro UV test), and one on a holey copper grid (for free-standing NLCS). For the substrate-supported sample, we took SEM images (from small scale to large scale) as maps to guide the position when performing the micro UV or Raman test (Figure S10). A J&M MSP210 microscope spectrometry system was used to obtain the absorption spectra of the ITO-supported BNP NLCSs, which were illuminated by high-intensity fiber light source under a 50× objective. The spectra acquisition in a local area was 7 × 7 μm2. On the basis of the “landmarks” in the SEM images, we located the light beam source to the exact areas of various packing orders to ensure their spectral correlation to respective ordering structures. Raman was characterized by a Renishaw RM 2000 Confocal micro-Raman system with excitation laser wavelengths of 633, 782, and 830 nm, a laser spot size of 1 μm, and a laser power of 0.1 mW. Numerical Simulations. The numerical simulations of the BNP ordered array (made of H- and V-NLCS arranged in a hexagonal lattice) were performed using CST Microwave Studio Suite. The frequencydomain FEM solver was used to obtain the optical response of these NLCSs under plane wave excitation by using periodic boundary conditions in lateral directions to model the behavior of these NLCSs. The simulation models are shown in Figure 3, where the rectangle in each case highlights the unit cell of the hexagonal lattice of the BNPs. The model parameters used in the simulations are taken from Supplementary Figures S2 and S3. The permittivity of gold in the BNPs was obtained from the literature,43−45 and the 2D NLCSs were assumed to be suspended in air. Open boundaries, emulating perfectly matched layers, were adopted in the transverse directions so that incident light can pass the boundaries with minimal reflection. Tetrahedral meshing with automatic mesh refinement was chosen to be fine enough for the frequency-domain simulations of the absorbance spectra over the wavelength range of interest. To estimate the SERS performance of each NLCS, we further studied the electric field distribution pattern at specific excitation laser wavelengths to compare the near-field confinement strength, which could roughly approximate the order of SERS intensity of these NLCSs. In the numerical calculations, the electric field vectors were monitored in threedimensional mesh points to generate the electric (E) field distribution maps. All the electric field patterns were obtained along the plane passing through the centers of the nanoparticles arranged in a hexagonal lattice.

cetyltrimethylammonium chloride solution (CTAC, 25 wt % in H2O), 4-aminothiophenol, indium tin oxide (ITO)-coated glass slides, and silicon wafers were purchased from Sigma-Aldrich. Hydrochloric acid (HCl, 32 wt %) was purchased from AJAX. Ammonium hydroxide (NH3·H2O, 25 wt %) was from Chem Supply. Hydrogen peroxide (H2O2, 30 wt %), tetrahydrofuran (THF), and chloroform were obtained from Merck KGaA. Thiol-functionalized polystyrene (Mn = 50 000, 20 000, and 12 000 g/mol) were purchased from Polymer Source Inc. All chemicals were used as-received unless otherwise indicated. Demineralized water was used in all aqueous solutions, which were further purified with a Milli-Q system (Millipore). All glassware used in the following procedures was cleaned in a bath of freshly prepared aqua regia and was rinsed thoroughly in H2O prior to use. Gilder extrafine bar grids (2000 mesh with 7 × 7 μm2 square holes) were purchased from Ted Pella. Synthesis of Bipyramid Gold Nanoparticles. BNPs were prepared by a typical seed-mediated method according to the literature.30 The first step involves the preparation of a gold seed solution by reducing HAuCl4 with NaBH4 in the presence of trisodium citrate. The orange-red solution was then aged at room temperature for at least 2 h before use. Then a growth solution was prepared by mixing HAuCl4 (2 mL, 25 mM) and CTAB (98 mL, 0.1 M), AgNO3 (1 mL, 10 mM), HCl (2 mL, 1.0 M), and L-ascorbic acid (0.8 mL, 0.1 M); the color of the solution turned from clear to violet-red after mixing. By decreasing the amount of seeds from 0.8 mL, to 0.7, and to 0.6 mL, three BNP samples of varying sizes were prepared. The final solution was allowed to age at 30 °C in a water bath and left undisturbed overnight. In order to get BNPs with high purity, a three-step purification method was involved after the preparation of the BNP solution.27 First, the as-grown Au BNP sample (100 mL) was centrifuged at 7830 rpm for 30 min. The precipitate was redispersed in a CTAC solution (80 mM, 75 mL), which was followed by subsequent addition and mixing of AgNO3 (0.01 M, 36 mL) and ascorbic acid (0.1 M, 18 mL). The resultant solution was kept in a water bath at 65 °C for 4 h to produce Au@Ag nanorods. The bimetallic Au@Ag nanorods were then centrifuged at 6000 rpm for 10 min. The precipitate was redispersed in CTAB (50 mM, 60 mL) and left undisturbed for 2 h at room temperature, during which the Au@Ag nanorods aggregated together and precipitated to the bottom of the container, while the spherical-like nanoparticles remained in the supernatant. The supernatant was discarded, and the precipitates were redispersed in CTAB (50 mM, 50 mL). The resulting solution was subsequently mixed gently with NH3·H2O (25 wt %, 1.0 mL) and H2O2 (30 wt %, 0.7 mL) and kept undisturbed for 2 h. During this process, the Ag segments were gradually etched away. A AgCl precipitate was seen to form at the bottom of the container. The clear supernatant was carefully taken out and centrifuged at 10 000 rpm for 10 min. The product was redispersed in a CTAB solution (50 mM, 25 mL) for further use. Synthesis of Gold Nanorod Particles. The gold nanorod particles are synthesized according to the seed-mediated growth method as in our previous report.8 Fabrication of BNP Sheets. The BNP NLCS sheet was fabricated by adopting the slightly modified, recently developed approach by our group.10 By using a two-step ligand-exchange procedure, the CTAB ligand was replaced by thiolated polystyrene (Mn = 50 000, 20 000, and 12 000 g/mol). The as-prepared CTAB-capped BNPs (12.5 mL) were centrifuged and redispersed in an excess thiol-functionalized polystyrene solution (8 mg/mL). After aging overnight at room temperature, the supernatant was discarded and the samples were purified by repeated centrifugation−precipitation cycles and redispersed in chloroform. Finally, one drop of chloroform solution of concentrated PS-capped BNPs was spread onto the surface of a convex-shaped water droplet on silicon wafer/ITO glass or a holey copper grid (with hole size of 7 μm × 7 μm), forming gold-colored reflective sheets. After the water droplet evaporated, a monolayer Au BNP ordered array was formed. Sample for Raman Test. For the Raman test, the Si wafersupported BNP NLCSs were first plasma-treated for 4 min and then submerged into 1 mL of 10 μM ATP−ethanol solution and left undisturbed overnight. The samples were rinsed with ethanol and dried before testing.

ASSOCIATED CONTENT S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsnano.5b06206. Detailed description of the experimental procedures and theoretical simulations employed for this work (PDF)

AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected]. Author Contributions

W.C. conceived the project, W.C. and Q.S. designed the experiments. Q.S., K.J.S., and L.W.Y. carried out experimental work on synthesis and characterization of BNP NLCSs; D.S. and M.P. developed the numerical models and D.S. carried out the numerical simulations; W.C., Q.S., K.J.S., and D.S. analyzed the experimental data and cowrote the paper. All authors have given approval to the final version of the manuscript. H

DOI: 10.1021/acsnano.5b06206 ACS Nano XXXX, XXX, XXX−XXX

Article

ACS Nano Notes

Programmable Nanoparticle Superlattices Using a Hollow ThreeDimensional Spacer Approach. Nat. Nanotechnol. 2012, 7, 24−28. (18) Cheng, W.; Campolongo, M. J.; Cha, J. J.; Tan, S. J.; Umbach, C. C.; Muller, D. A.; Luo, D. Free-Standing Nanoparticle Superlattice Sheets Controlled by DNA. Nat. Mater. 2009, 8, 519−525. (19) Bishop, K. J.; Wilmer, C. E.; Soh, S.; Grzybowski, B. A. Nanoscale Forces and Their Uses in Self-Assembly. Small 2009, 5, 1600−1630. (20) van der Stam, W.; Gantapara, A. P.; Akkerman, Q. A.; Soligno, G.; Meeldijk, J. D.; van Roij, R.; Dijkstra, M.; de Mello Donega, C. SelfAssembly of Colloidal Hexagonal Bipyramid- and Bifrustum-Shaped ZnS Nanocrystals into Two-Dimensional Superstructures. Nano Lett. 2014, 14, 1032−1037. (21) Saunders, A. E.; Ghezelbash, A.; Smilgies, D.-M.; Sigman, M. B.; Korgel, B. A. Columnar Self-Assembly of Colloidal Nanodisks. Nano Lett. 2006, 6, 2959−2963. (22) Hsu, S. W.; Ngo, C.; Tao, A. R. Tunable and Directional Plasmonic Coupling within Semiconductor Nanodisk Assemblies. Nano Lett. 2014, 14, 2372−2380. (23) Ye, X.; Chen, J.; Engel, M.; Millan, J. A.; Li, W.; Qi, L.; Xing, G.; Collins, J. E.; Kagan, C. R.; Li, J.; Glotzer, S. C.; Murray, C. B. Competition of Shape and Interaction Patchiness for Self-Assembling Nanoplates. Nat. Chem. 2013, 5, 466−473. (24) Jiang, Z.; He, J.; Deshmukh, S. A.; Kanjanaboos, P.; Kamath, G.; Wang, Y.; Sankaranarayanan, S. K.; Wang, J.; Jaeger, H. M.; Lin, X. M. Subnanometre Ligand-Shell Asymmetry Leads to Janus-Like Nanoparticle Membranes. Nat. Mater. 2015, 14, 912−917. (25) Bae, Y.; Kim, N. H.; Kim, M.; Lee, K. Y.; Han, S. W. Anisotropic Assembly of Ag Nanoprisms. J. Am. Chem. Soc. 2008, 130, 5432−5433. (26) Liu; Guyot-Sionnest, P. Mechanism of Silver(I)-Assisted Growth of Gold Nanorods and Bipyramids. J. Phys. Chem. B 2005, 109, 22192− 22200. (27) Li, Q.; Zhuo, X.; Li, S.; Ruan, Q.; Xu, Q. H.; Wang, J. Production of Monodisperse Gold Nanobipyramids with Number Percentages Approaching 100% and Evaluation of Their Plasmonic Properties. Adv. Opt. Mater. 2015, 3, 801−812. (28) Lee, J. H.; Gibson, K. J.; Chen, G.; Weizmann, Y. BipyramidTemplated Synthesis of Monodisperse Anisotropic Gold Nanocrystals. Nat. Commun. 2015, 6, 7571. (29) Zhou, G.; Yang, Y.; Han, S.; Chen, W.; Fu, Y.; Zou, C.; Zhang, L.; Huang, S. Growth of Nanobipyramid by Using Large Sized Au Decahedra as Seeds. ACS Appl. Mater. Interfaces 2013, 5, 13340−13352. (30) Guo, Z.; Wan, Y.; Wang, M.; Xu, L.; Lu, X.; Yang, G.; Fang, K.; Gu, N. High-Purity Gold Nanobipyramids Can be Obtained by an Electrolyte-Assisted and Functionalization-Free Separation Route. Colloids Surf., A 2012, 414, 492−497. (31) Malachosky, E. W.; Guyot-Sionnest, P. Gold Bipyramid Nanoparticle Dimers. J. Phys. Chem. C 2014, 118, 6405−6412. (32) Lee, S.; Mayer, K. M.; Hafner, J. H. Improved Localized Surface Plasmon Resonance Immunoassay with Gold Bipyramid Substrates. Anal. Chem. 2009, 81, 4450−4455. (33) Guffey, M. J.; Miller, R. L.; Gray, S. K.; Scherer, N. F. PlasmonDriven Selective Deposition of Au Bipyramidal Nanoparticles. Nano Lett. 2011, 11, 4058−4066. (34) Ryan, K. M.; Singh, S.; Liu, P.; Singh, A. Assembly of Binary, Ternary and Quaternary Compound Semiconductor Nanorods: from Local to Device Scale Ordering Influenced by Surface Charge. CrystEngComm 2014, 16, 9446−9454. (35) Lu, X.; Zhuang, Z.; Peng, Q.; Li, Y. Controlled Synthesis of Wurtzite CuInS2 Nanocrystals and Their Side-by-Side Nanorod Assemblies. CrystEngComm 2011, 13, 4039. (36) Singh, A.; Gunning, R. D.; Ahmed, S.; Barrett, C. A.; English, N. J.; Garate, J.-A.; Ryan, K. M. Controlled Semiconductor Nanorodassembly from Solution: Influence of Concentration, Charge and Solvent Nature. J. Mater. Chem. 2012, 22, 1562−1569. (37) Onsager, L. The Effects of Shape on the Interaction of Colloidal Particles. Ann. N. Y. Acad. Sci. 1949, 51, 627−659. (38) Dogic, Z.; Fraden, S. Ordered Phases of Filamentous Viruses. Curr. Opin. Colloid Interface Sci. 2006, 11, 47−55.

The authors declare no competing financial interest.

ACKNOWLEDGMENTS M.P. and W.C. acknowledge Discovery Grants DP110100713, DP140100883, DP120100170, and DP140100052. This work was performed in part at the Melbourne Centre for Nanofabrication (MCN) in the Victorian Node of the Australian National Fabrication Facility (ANFF). The authors also gratefully acknowledge the use of facilities at The Monash Center for Electron Micron Microscopy. The work of D.S. is supported by the DSDBI of the Victorian Government, through its Victoria India Doctoral Scholarship Program (managed by the Australia India Institute). REFERENCES (1) Tan, S. J.; Campolongo, M. J.; Luo, D.; Cheng, W. Building Plasmonic Nanostructures with DNA. Nat. Nanotechnol. 2011, 6, 268− 276. (2) Henzie, J.; Andrews, S. C.; Ling, X. Y.; Li, Z.; Yang, P. Oriented Assembly of Polyhedral Plasmonic Nanoparticle Clusters. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 6640−6645. (3) Tao, A.; Sinsermsuksakul, P.; Yang, P. Tunable Plasmonic Lattices of Silver Nanocrystals. Nat. Nanotechnol. 2007, 2, 435−440. (4) Hamon, C.; Liz-Marzan, L. M. Hierarchical Assembly of Plasmonic Nanoparticles. Chem. - Eur. J. 2015, 21, 9956−9963. (5) Jones, M. R.; Seeman, N. C.; Mirkin, C. A. Programmable Materials and the Nature of the DNA Bond. Science 2015, 347, 1260901. (6) Mazid, R. R.; Si, K. J.; Cheng, W. DNA Based Strategy to Nanoparticle Superlattices. Methods 2014, 67, 215−226. (7) Chen, Y.; Cheng, W. DNA-Based Plasmonic Nanoarchitectures: from Structural Design to Emerging Applications. Wiley Interdiscip. Rev.: Nanomed. nanobiotechnol. 2012, 4, 587−604. (8) Ng, K. C.; Udagedara, I. B.; Rukhlenko, I. D.; Chen, Y.; Tang, Y.; Premaratne, M.; Cheng, W. Free-Standing Plasmonic-Nanorod Superlattice Sheets. ACS Nano 2012, 6, 925−934. (9) Quan, Z.; Fang, J. Superlattices with Non-Spherical Building Blocks. Nano Today 2010, 5, 390−411. (10) Si, K. J.; Sikdar, D.; Chen, Y.; Eftekhari, F.; Xu, Z.; Tang, Y.; Xiong, W.; Guo, P.; Zhang, S.; Lu, Y.; Bao, Q.; Zhu, W.; Premaratne, M.; Cheng, W. Giant Plasmene Nanosheets, Nanoribbons, and Origami. ACS Nano 2014, 8, 11086−11093. (11) Chen, Y.; Si, K. J.; Sikdar, D.; Tang, Y.; Premaratne, M.; Cheng, W. Ultrathin Plasmene Nanosheets as Soft and Surface-Attachable SERS Substrates with High Signal Uniformity. Adv. Opt. Mater. 2015, 3, 919− 924. (12) Si, K. J.; Sikdar, D.; Yap, L. W.; Foo, J. K. K.; Guo, P.; Shi, Q.; Premaratne, M.; Cheng, W. Dual-Coded Plasmene Nanosheets as NextGeneration Anticounterfeit Security Labels. Adv. Opt. Mater. 2015, 3, 1710−1717. (13) Si, K. J.; Guo, P.; Shi, Q.; Cheng, W. Self-Assembled NanocubeBased Plasmene Nanosheets as Soft Surface-Enhanced Raman Scattering Substrates toward Direct Quantitative Drug Identification on Surfaces. Anal. Chem. 2015, 87, 5263−5269. (14) Lee, Y. H.; Shi, W.; Lee, H. K.; Jiang, R.; Phang, I. Y.; Cui, Y.; Isa, L.; Yang, Y.; Wang, J.; Li, S.; Ling, X. Y. Nanoscale Surface Chemistry Directs the Tunable Assembly of Silver Octahedra into Three TwoDimensional Plasmonic Superlattices. Nat. Commun. 2015, 6, 6990. (15) Lee, Y. H.; Lee, C. K.; Tan, B.; Rui Tan, J. M.; Phang, I. Y.; Ling, X. Y. Using the Langmuir-Schaefer Technique to Fabricate Large-Area Dense SERS-Active Au Nanoprism Monolayer Films. Nanoscale 2013, 5, 6404−6412. (16) Ku, J. C.; Ross, M. B.; Schatz, G. C.; Mirkin, C. A. Conformal, Macroscopic Crystalline Nanoparticle Sheets Assembled with DNA. Adv. Mater. 2015, 27, 3159−3163. (17) Auyeung, E.; Cutler, J. I.; Macfarlane, R. J.; Jones, M. R.; Wu, J.; Liu, G.; Zhang, K.; Osberg, K. D.; Mirkin, C. A. Synthetically I

DOI: 10.1021/acsnano.5b06206 ACS Nano XXXX, XXX, XXX−XXX

Article

ACS Nano (39) Gelbart, W. M. Molecular Theory of Nematic Liquid Crystals. J. Phys. Chem. 1982, 86, 4298−4307. (40) Frenkel, D. Order Through Entropy. Nat. Mater. 2015, 14, 9−12. (41) Zhang, H.; Liu, Y.; Yao, D.; Yang, B. Hybridization of Inorganic Nanoparticles and Polymers to Create Regular and Reversible SelfAssembly Architectures. Chem. Soc. Rev. 2012, 41, 6066−6088. (42) Kim, W. D.; Chae, W.-S.; Bae, W. K.; Lee, D. C. Controlled Vortex Formation and Facilitated Energy Transfer within Aggregates of Colloidal CdS Nanorods. Chem. Mater. 2015, 27, 2797−2802. (43) Sikdar, D.; Rukhlenko, I. D.; Cheng, W.; Premaratne, M. Tunable Broadband Optical Responses of Substrate-Supported Metal/Dielectric/Metal Nanospheres. Plasmonics 2014, 9, 659−672. (44) Sikdar, D.; Rukhlenko, I.; Cheng, W.; Premaratne, M. Optimized Gold Nanoshell Ensembles for Biomedical Applications. Nanoscale Res. Lett. 2013, 8, 1−5. (45) Johnson, P. B.; Christy, R. W. Optical Constants of the Noble Metals. Phys. Rev. B 1972, 6, 4370−4379.

J

DOI: 10.1021/acsnano.5b06206 ACS Nano XXXX, XXX, XXX−XXX