Simple and Rapid High-Yield Synthesis and Size Sorting of Multibranched Hollow Gold Nanoparticles with Highly Tunable NIR Plasmon Resonances.

full papers www.MaterialsViews.com

Gold Nanoparticles

Simple and Rapid High-Yield Synthesis and Size Sorting of Multibranched Hollow Gold Nanoparticles with Highly Tunable NIR Plasmon Resonances Adam J. Blanch, Markus Döblinger, and Jessica Rodríguez-Fernández*

Branched gold nanoparticles with sharp tips are considered excellent candidates for sensing and field enhancement applications. Here, a rapid and simple synthesis strategy is presented that generates highly branched gold nanoparticles with hollow cores and a ca.100% yield through a simple one-pot seedless reaction at room temperature in the presence of Triton X-100. It is shown that multibranched hollow gold nanoparticles of tunable dimensions, branch density and branch length can be obtained by adjusting the concentrations of the reactants. Insights into the formation mechanism point toward an aggregative type of growth involving hollow core formation first, and branching thereafter. The pronounced near-infrared (NIR) plasmon band of the nanoparticles is due to the combined contribution from hollowness and branching, and can be tuned over a wide range (≈700–2000 nm). It is also demonstrated that the high environmental sensitivity of colloidal dispersions based on multibranched hollow gold nanoparticles can be boosted even further by separating the nanoparticles into fractions of given sizes and improved monodispersity by means of a glycerol density gradient. The possibility to obtain highly monodisperse multibranched hollow gold nanoparticles with predictable dimensions (50–300 nm) and branching and, therefore, tailored NIR plasmonic properties, highlights their potential for theranostic applications.

1. Introduction The synthesis of gold nanoparticles with well-defined shapes has been the focus of much attention in recent times due to their attractive physical and chemical properties. These

Dr. A. J. Blanch, Dr. J. Rodríguez-Fernández Photonics and Optoelectronics Group Department of Physics and CeNS Ludwig-Maximilians-Universität München Amalienstr. 54, 80799 Munich, Germany E-mail: [email protected] Dr. A. J. Blanch, Dr. M. Döblinger, Dr. J. Rodríguez-Fernández Nanosystems Initiative Munich (NIM) Schellingstr. 4, 80799 Munich, Germany Dr. M. Döblinger Department of Chemistry and CeNS Ludwig-Maximilians-Universität München Butenandstr. 11, 81377 Munich, Germany DOI: 10.1002/smll.201500095

4550 www.small-journal.com

include highly tunable plasmonic resonances, catalytic activity, relative ease of surface modification and enhancement of surface enhanced Raman scattering (SERS) signals which affords the potential for use in sensing devices.[1–4] The low reactivity of gold is also an advantage in biocompatibility, where gold nanostructures have been used for various applications including photothermal therapy.[5] The properties of gold nanoparticles are strongly dependent on their size and shape, and many different morphologies have been produced thus far, from spheres and rods to cubes and polyhedral plates.[4,6] Particular interest has been paid to anisotropic shapes, most notably nanorods, as well as branched structures.[3,6] The latter may be referred to as nanostars,[3,7–10] nanourchins,[11–13] spiky[14] or multibranched nanoparticles,[15–17] depending on the particular study, however they possess the common elements of multiple arms radiating from a central core. Such structures are of interest due to the anisotropic distribution of the electromagnetic field near the branch tips, where the radius of curvature is highest. In addition, branched nanostructures generally have

© 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

small 2015, 11, No. 35, 4550–4559

www.MaterialsViews.com

plasmon resonances in the near-infrared (NIR) region, which is useful for biological applications.[13,16,18] Colloidal synthesis of branched gold nanostructures is generally achieved via one of two main pathways; namely, seeded growth and homogenous nucleation approaches.[1,3] For seeded growth, small precursor nanoparticles are generated using a strong reducing agent, aged, then applied to a secondary growth solution containing additional metal salt. Further reducing agent is added in the presence of shape directing agents (usually surfactants and ionic species such as Ag+ or Br−)[19] and the initial nanoparticles are enlarged by metal deposition to form the desired nanoparticles. For nonseeded growth, the nanoparticles are produced in a one-pot synthesis, where a single reduction step is performed. As in this instance nuclei are formed in situ, generally larger nanoparticles and broader size distributions result.[3] Despite being a popular nonionic surfactant, Triton X-100 (TX-100) is not commonly used for gold nanoparticle synthesis. Gold colloids have been previously generated through photo-irradiation[20–22] or microwave[23] induced nucleation of nanoparticles from a solution of gold precursor in Triton X-100, however these studies have been limited to spherical nanoparticles. More recently, branched nanoparticles with well-defined tips have been reported to be synthesized in this surfactant using both seeded[8] and nonseeded[7] approaches. In the seeded method, star-shaped nanoparticles with regular branches can be produced under certain conditions. However, the yield of nanostars is often very low, with multipodal and unbranched spheroids constituting a large percentage of the generated nanoparticles.[8] For the nonseeded approach, a very high concentration of the surfactant was applied in order to operate in a liquid crystal regime. This allowed production of gold nanoparticles with long thorns but highly irregular structures.[7] Here, we show that under similar conditions used to grow nanostars in the seeded approach, highly branched gold nanoparticles with hollow cores and NIR plasmon resonances can be grown in a simple and rapid one pot seedless method at room temperature with a ca. 100% yield. We demonstrate control over the extent of branching and nanoparticle dimensions by modifying the reaction conditions, and gain insights into their hollow nature and growth mechanism through electron microscopy and optical spectroscopy. Finally, we examine the sensitivity of the nanoparticles to the surrounding medium before and after separation using density gradients which are able to provide well-refined fractions of narrowed size range. The ease of production and high density of sharp tips possessed by these multibranched hollow gold nanoparticles promotes them as candidates for use in sensing, while their highly tunable NIR plasmon resonance allows for their potential use in biological and photothermal applications.

2. Results and Discussion 2.1. Synthesis and Mechanism of Growth Multiply-branched gold nanoparticles were obtained in a simple and rapid one-pot seedless process upon addition of small 2015, 11, No. 35, 4550–4559

AgNO3, HAuCl4 and ascorbic acid to an aqueous solution of TX-100 (see the Experimental Section for details). After addition of the reducing agent the solution quickly acquires a faint coloring that strongly darkens within 3–5 min. After 1.5–2 h of undisturbed growth, thiolated polyethylene glycol (PEG) is typically added to stabilize the nanoparticles for centrifugation and subsequent redispersion in water. When the above synthesis is carried out in the presence of TX100-stabilized Au seeds, it results in the formation of a mixture of gold nanostars with five or less branches along with a considerable fraction of non-branched nanoparticles, as reported by Pallavicini et al.[8] However, we have found that in the absence of seeds colloidal dispersions solely consisting of multiply-branched gold nanoparticles can be produced. Figure 1A shows a representative transmission electron microscopy (TEM) image of the nanoparticles obtained when concentrations of 124 × 10−3 m TX-100, 0.096 × 10−3 m AgNO3, 0.49 × 10−3 m HAuCl4, and 1.34 × 10−3 m ascorbic acid (hereafter referred as the standard set of conditions) are combined in the final volume of the growth solution. All nanoparticles are three dimensional (see scanning electron microscopy (SEM) image in Figure 1B), and are characterized by a high number of tips (see high-resolution TEM image (HRTEM) in Figure 1C). Their optical response is characterized by a broad localized surface plasmon resonance in the NIR (see Figure 1D) which is due to the hollow character of their cores and multiple branches, as we demonstrate further below. The branches are relatively uniform in width, but highly variable in length and may be up to 100 nm long, while the cores are highly polydisperse in size (see Figure 1A) and are often asymmetrical. We have found that some slight differences in the size distribution and branching

Figure 1. A) TEM and B) SEM images of multiply-branched gold nanoparticles produced via seedless growth in TX-100 at the standard conditions. C) HRTEM image of multiple narrow branches on a single particle. D) UV–vis-NIR spectrum of the same dispersion, recorded in a 2 mm path length quartz cell to reduce the absorbance contribution from water and allow the region above 1400 nm to be examined. The discontinuities from ≈1420–1490 nm and 1870–2090 nm are due to noise from water absorption which cannot be subtracted from the baseline.


www.small-journal.com

4551


extent of the nanoparticles can give rise to some slight variations in the plasmon band position and width from batch to batch (see Figure S1, Supporting Information). Nevertheless, in all cases the sample is comprised of essentially 100% multiply-branched nanoparticles such as those depicted in Figure 1 and whose size polydispersity (and therefore, optical response) can be effectively improved by centrifugal sorting in a glycerol gradient, as we demonstrate further below. It is also worth noting here that when the nanoparticles are left in the growth solution, there is a significant shift in their plasmon band likely due to surface reshaping, as reported elsewhere.[24] However, capping with PEG-thiol and subsequent washing prevents any sort of reshaping for at least two weeks (see Figure S2, Supporting Information). The extent of the branching can be controlled by adjusting the concentration of ascorbic acid. This can be qualitatively assessed from the TEM images shown in Figure 2A–H. A quantitative assessment of nanoparticle branching was obtained by measuring the branch density through analysis of TEM images of at least ten individual nanoparticles at

each ascorbic acid concentration (see the gallery of the nanoparticles analyzed in Figure S4, Supporting Information). The TEM images were converted to binary images in the software package ImageJ and values for both the perimeter and area of each nanoparticle were extracted using the inbuilt “analyze particles” algorithm. The ratio of these two parameters gives an indication of the branch density. Figure 2I depicts the evolution of the perimeter-to-area ratio (P/A) as a function of the ascorbic acid concentration. It clearly shows that nanoparticle branching increases steadily when increasing the ascorbic acid concentration from 0.77 × 10−3 to 2.63 × 10−3 m, while above this concentration the P/A value is saturated. The UV–vis-NIR spectra of each dispersion are shown in Figure 2K. At the lowest ascorbic acid concentration examined (0.77 × 10−3 m), where little to no branching occurs and the nanoparticles possess only small protrusions, the plasmon band is the most blue-shifted and centered at ≈850 nm. The plasmon band gradually red-shifts and broadens as the ascorbic acid concentration is increased up to 1.34 × 10−3 m.

Figure 2. A–H) TEM images showing the morphology and branching evolution of Au nanoparticles formed in the presence of (A) 0.77, (B) 0.92, (C) 1.07, (D) 1.34, (E) 1.97, (F) 2.63, (G) 3.49, and (H) 4.68 × 10−3 M ascorbic acid while keeping the concentrations of TX-100, HAuCl4, and AgNO3 at standard conditions. Scale bars are 200 nm (main) and 70 nm (inset). I) Evolution of the average 2D perimeter/area ratio per nanoparticle (P/A) as a function of the ascorbic acid concentration. The mean area and 2D perimeter/area per nanoparticle values are averaged from ≥10 individual nanoparticles (see TEM images of the nanoparticles used in Figure S4, Supporting Information). J) Evolution of the mean nanoparticle area as a function of the ascorbic acid concentration used to synthesize each corresponding batch. K) UV–vis-NIR spectra for each of the colloidal dispersions shown in (A–H).



small 2015, 11, No. 35, 4550–4559


Figure 3. A–C) HRTEM images of individual multibranched gold nanoparticles obtained at an ascorbic acid concentration of (A) 0.77 × 10−3, (B) 1.34 × 10−3, and (C) 2.63 × 10−3 M. D–F) STEM images of single nanoparticles from the very same batches: (D) 0.77 × 10−3, (E) 1.34 × 10−3, and (F) 2.63 × 10−3 M. G) Representative EDX spectrum from a single nanoparticle formed at 1.34 × 10−3 M ascorbic acid concentration. Inset: immersion mode SEM images of a nanoparticle from the same sample before (left) and after (right) milling with a focused ion beam. Note that in these images the branches appear more rounded than in (B) and (C) due to the visibility of the PEG shell in immersion mode SEM.

This is due to the gradual increase in branching in this concentration range (see Figure 2I). At 1.34 × 10−3 m the plasmon band reaches its maximum red-shift (≈1250 nm) and breadth. At concentrations higher than 1.34 × 10−3 m the plasmon band shift no longer follows this trend despite the clear increase in branching still occurring up to 2.63 × 10−3 m ascorbic acid, as shown in Figure 2I. This may be due to the formation of slightly smaller nanoparticles for concentrations >1.34 × 10−3 m (see plot of mean nanoparticle area vs. ascorbic acid concentration in Figure 2J along with size distributions in Figure S10, Supporting Information), and it is most likely associated with the increased rate of reaction at higher concentrations of the reducing agent. Differences in the size distribution and length/width ratio of the branches between the different samples may also be a factor contributing to the plasmon band evolution observed here.[25] It is worth noting that in the literature the effect of ascorbic acid concentration on the branching behavior of gold nanoparticles has been reported to have dissimilar effects. For instance, in cetyltrimethylammonium bromide (CTAB) based syntheses, increasing the ascorbic acid concentration usually leads to a reduction in branch length and density.[15,26] The same has also been reported in a surfactantless synthesis,[9] although core size has been reported to be reduced,[15] increased[27] or unaltered[26] with increasing ascorbic acid depending on the concentrations of other reactants. Alternatively, in CTAB with bis(p-sulfonatophenyl)phenyl phosphine dehydrate dipotassium salt coated seeds[27] or with sodium dodecyl sulfate[28] as a shape directing surfactant, increasing ascorbic acid concentration may increase the amount of branches and/or their length. From these mixed reports it is clear that the effect of the reducing agent concentration on the extent of nanoparticle branching is very sensitive to subtle differences in the synthesis conditions. Nevertheless, for the seedless TX-100 based synthesis that we report here, higher concentrations of ascorbic acid undoubtedly promote increased branching, in agreement with the work from Pallavicini et al.,[8] with a maximum branch density obtained for an ascorbic acid concentration of 2.63 × 10−3 m. Above this concentration hyperbranching, understood as branching from the branches rather than from the core, becomes far more apparent, indicating that the core’s branch small 2015, 11, No. 35, 4550–4559

density limit has been reached at this point. The saturation of branching density at high ascorbic acid concentration is also apparent for a series of nanoparticles obtained at a lower Ag+ concentration, as shown in Figure S5, Supporting Information. One interesting observation from the TEM images depicted in Figure 2 is that at the highest ascorbic acid concentration used (4.68 × 10−3 m), the multiply-branched gold nanoparticles begin to appear hollow in the center (see Figure 2H). To investigate this apparent hollowness we carried out HRTEM and STEM (scanning transmission electron microscopy) imaging at 300 kV. As can be seen in Figure 3A–F, even multiply-branched gold nanoparticles obtained at low ascorbic acid concentrations (0.77 × 10−3 and 1.34 × 10−3 m) possess nonsolid centers. This is further confirmed after milling sections from individual nanoparticles with a focused ion beam (FIB). The nanoparticle cores are mostly hollow, or at least contain large void spaces (see SEM inset in Figure 3G and additional images in Figure S6, Supporting Information). Even though some slight restructuring and tip melting does occur upon exposure to the focused ion beam, it is still possible to clearly observe the void within the nanoparticle interior. All in all, these results confirm the hollow (cagelike) nature of the multiply-branched gold nanoparticles obtained via our one-pot seedless process. It is worth noting here that hollow gold nanoparticles with a similar cage-like morphology are typically synthesized by performing galvanic replacement reactions on pre-formed silver or cobalt nanoparticles, resulting in hollow nanoparticles with either a spiky[12,29–31] or a smooth surface.[32–34] Only a few examples can be found in the literature reporting on the one-pot seedless formation of hollow gold nanoparticles, one being the nanoboxes that He et al. have synthesized through a surfactantless co-reduction of HAuCl4 and AgNO3, which resulted in a Ag–Au alloy.[35] In order to investigate the composition of our multibranched hollow nanoparticles we have performed energy-dispersive X-ray spectroscopy (EDX) analysis on several samples. An EDX spectrum for a single nanoparticle produced at our standard conditions (where the Au3+:Ag+ ratio is approximately 5:1) is shown in Figure 3G. Further results of elemental quantification for this and several other nanoparticles are given in the Supporting



4553


Figure 4. A) UV–vis–NIR spectra recorded from aliquots of an Au nanoparticle growth solution (at standard conditions) for which growth was arrested at the indicated reaction times. B) Corresponding TEM images taken from each aliquot. Scale bars are 50 nm. Additional TEM images and spectra are provided in Figure S13, Supporting Information.

Information (Figure S7). The EDX results indicate that the major component of our nanoparticles (approximately 93%) is Au, though a small percentage of Ag is present, partly as a thin sheath across the apex of the branch tips (see Figure S8, Supporting Information). This Au:Ag composition is similar to a reported composition of 91% gold: 9% silver for nanoparticles produced via reduction of HAuCl4 in the presence of AgNO3 using ascorbic acid at a ratio of 10:1 Au3+:Ag+ without any surfactant.[14] However, the low percentage of Ag in our nanoparticles strongly differs from the typically high silver content of hollow nanoparticles prepared by galvanic replacement of preformed Ag nanoparticles (reported as 13%–28%,[30] 35%–85%,[33] ≈55%,[32] and 57%[36]). This small percentage of Ag does not appear to strongly affect the plasmon band of our nanoparticles. Their plasmon resonance seems to be dominated by the major Au component, with their hollow nature and multiple branches being responsible for the NIR location of the band (see discussion below). Further investigations reveal that the amount of Ag+ added to the growth solution is critical to the formation of branched structures, and only a low concentration of Ag+ (90 s; while for times 0.13 m (standard conditions, see Figure S15, Supporting Information). Some of those nanoparticles tend to agglomerate with each other giving rise to cage-like gold nanostructures whose completeness seems to be hindered by their slow diffusion in such viscous media. We have found neither spectral evidence (no plasmon band in ≈520–560 nm range, see Figures S12, S13, and S16, Supporting Information) nor visual evidence (see TEM images in Figure 4, Figures S12 and S13, Supporting Information) for the formation of small solid gold nanoparticles at the 0.13 m TX-100 concentration we have used for the synthesis of our multibranched hollow gold nanoparticles. However, the comparatively much lower viscosity of the 0.13 m TX-100 growth solutions vs. that of solutions with [TX-100] >0.13 m (Figure S15, Supporting Information) could enable the rapid diffusion and agglomeration of small gold nanoparticles forming in the medium during the early stages of reaction. This rapid agglomeration could account for the lack of spectral and visual evidence for their formation. Taken altogether, our results suggest that a galvanic replacement of a preformed template is unlikely in our case, and that the growth pathway for our multibranched hollow gold nanoparticles may occur via deposition of Au0 onto small Au nanoparticles nucleating and rapidly agglomerating in situ, as reported previously for both flower-like[40] and urchin-like[11,41] gold nanostructures. This is also similar to the expected mechanism for the formation of Ag–Au nanoboxes as described by He et al.,[35] where Au nuclei are formed initially followed by simultaneously reduction of Ag

and Au to further nanoparticle growth. The Ag within the nanoparticle alloy is subject to an electrochemical driving force to reduce Au3+ ions via the galvanic displacement reaction, leading to hollow nanoboxes, analogous to our results. However, in contrast to the system of He et al., we find that Ag+ is reduced by ascorbic acid directly. Therefore it may also be possible that Ag0 could form in the reaction medium and participate in the formation of Au nuclei, as reported by Sau et al.[15] A more thorough investigation would still be needed in order to undoubtedly confirm the growth mechanism of our nanoparticles.

2.2. Size Separation and Plasmonic Sensitivity We now shift our focus toward the investigation of the plasmonic performance of these nanoparticles. Their inherent hollowness, multiple branches, and NIR plasmon resonances make them excellent candidates for plasmonic sensing. Therefore, we investigated their sensitivity to the dielectric properties of the surrounding medium. For this purpose, we synthesized four batches of approximately the same average size at 0.187 × 10−3 m AgNO3, but possessing different branch densities, which was controlled by varying the ascorbic acid concentration between 0.83 × 10−3 and 2.57 × 10−3 m. Representative TEM images of the as-synthesized nanoparticles are shown in Figure 5C. The spectra obtained after redispersion of the colloids in glycerol/water mixtures are shown in Figure 5A. In all cases, the plasmon band red-shifts as the refractive index of the glycerol/water mixtures increases along with the glycerol content. It is worth highlighting here the high dielectric sensitivity (190 nm/RI) of the non-branched, and therefore only hollow, gold nanoparticles (0.83 × 10−3 m, see Figure 5B). From our results it is also obvious that higher branch densities clearly lead to a larger plasmon shift per refractive index unit (Δλ/RI, see Figure 5B), with the sensitivity increasing up to 319 nm/RI for the nanoparticles with

Figure 5. A) UV–vis-NIR spectra (normalized at 400 nm) for nanoparticles grown at 0.186 × 10−3 M AgNO3 using different concentrations of ascorbic acid (as indicated) and resuspended in aqueous solutions containing an increasing weight percentage of glycerol. B) Evolution of the plasmon peak position as a function of solvent refractive index for each sample. Values on the left side are the corresponding slope values, in terms of shift per refractive index unit (Δλ/RI), for each linear regression. C) Representative TEM images for nanoparticles from each dispersion. D) Evolution of the plasmon sensitivity in terms of shift per refractive index unit (Δλ/RI) as a function of the 2D perimeter to area ratio (determined from 10 individual nanoparticles in each ensemble). small 2015, 11, No. 35, 4550–4559



4555


Figure 6. A) Photograph illustrating the separation of multibranched hollow gold nanoparticles grown at 0.187 × 10−3 M AgNO3 with 1.31 × 10−3 M ascorbic acid in a glycerol density gradient (2 mL steps) at 3220 × g for 50 min. B) Size distributions for each sample as determined from analysis of the TEM images. Bottom panel: Color-coded TEM images for the original unsorted dispersion, upper, middle, and lower fractions from left to right. Scale bars are 100 nm. C) UV–vis-NIR spectra for the original dispersion and sorted fractions.

the highest branching (2.57 × 10−3 m). This higher sensitivity may be due to a higher number of sharp tips per unit area, as the evolution of Δλ/RI correlates well with the average 2D perimeter-area ratio (Figure 5D), which quantifies the extent of branching. However, it is also known that sensitivity to the refractive index of the medium increases as the plasmon resonance maximum is red-shifted,[42] and therefore we cannot exclude that part of this improvement may be related to this effect. The environmental sensitivity of our gold nanoparticles is quite high due to their inherent hollowness and branching. Even though nanoparticles grown at a AgNO3 concentration of 0.187 × 10−3 m such as those shown in Figure 5 are smaller and have a narrower size distribution than those grown at standard conditions (0.096 × 10−3 m AgNO3), they are still quite polydisperse ( = 66 ± 11 nm, see Figure 6C). Polydispersity of particle size is a common drawback to all the seedless syntheses reported here and elsewhere[3] and it may have a detrimental effect on the plasmonic sensitivity. With the aim of reducing the size distribution of our multibranched hollow nanoparticles, and thus improving their environmental sensitivity, we have carried out their separation in glycerol density gradients under mild centrifugation conditions, as reported by Bonaccorso et al.[43] Figure 6 shows the results for the separation of a sample analogous to the one shown in Figure 5 (1.31 × 10−3 m ascorbic acid) using a glycerol step gradient. Separation of colored bands is easily visible in the centrifuge tube within 20 min of centrifugation (see photograph in Figure 6A). Allowing the nanoparticles to migrate into the higher density bands for longer times (50 min in this case) allows for better separation. This improvement is due to the more distinct boundary between steps in the higher density fractions which cause the bands to separate, although some nanoparticles precipitate as a pellet (and thus are lost) at longer centrifugation times. The distinct nanoparticle fractions obtained, referred to as “lower,” “middle,” and “upper,” may be collected via pipette. Representative TEM images from the original (unsorted) sample and from each isolated


fraction are shown in Figure 6. The corresponding size distributions are provided in Figure 6B. From these results the efficiency of the separation process is obvious, yielding fractions of multibranched hollow nanoparticles with significantly narrower size distributions (namely 56 ± 5 nm, upper fraction; 67 ± 5 nm, middle fraction; and 81 ± 6 nm, lower fraction) as compared to the original sample (66 ± 11 nm). This monodispersity improvement is clearly reflected in the UV–vis-NIR spectra shown in Figure 6C, with the plasmon band of each fraction being significantly narrower and more symmetric than that of the original dispersion. The plasmon bands red-shift and gain in intensity with increasing average nanoparticle size from the upper (663 nm) to the middle (687 nm), and to the lower fraction (725 nm). It is worth noting here that by carrying out this glycerol-assisted separation we can easily assess the effect of size on the plasmon band of hollow gold nanoparticles with the same or different branch densities, as demonstrated in Figure 6C and Figure S15, Supporting Information, respectively. Furthermore, it is a highly versatile and reproducible process, since it works equally well for nanoparticles having greater branching densities (see Figures S18–S21, Supporting Information) or core sizes (Figure S22, Supporting Information), and can be efficiently scaled in volume (from 15 to 50 mL) and concentration (10×), as shown in Figure S23, Supporting Information. Having shown the success of the glycerol-assisted separation, we evaluated the plasmonic sensitivity of the more monodisperse fractions obtained vs. that of the corresponding original, unsorted, dispersion. Specifically, we examined the environmental sensitivity of three fractions of different mean size (49 ± 7, upper; 67 ± 10, middle; and 80 ± 13 nm, lower) obtained after performing an up-scaled separation of a dispersion consisting of multibranched hollow gold nanoparticles analogous to the ones shown in Figure 5 (1.85 × 10−3 m ascorbic acid). Each fraction was redispersed in different solvents with increasing refractive index from water (1.33) to pyridine (1.51), as shown in Figure 7A. We note here that the PEG-capped nanoparticles were highly stable in all solvents,


small 2015, 11, No. 35, 4550–4559


Figure 7. A) UV–vis-NIR spectra (normalized to the peak maxima) for density gradient separated particle fractions (1.85 × 10−3 M ascorbic acid growth, separated in a 50 mL centrifuge tube—see Figure S23, Supporting Information, for details. The average size for each distribution is given beneath its associated spectra after redispersion in different solvents. B) Peak shift as a function of the solvent refractive index. The values on the left side are slope (Δλ/RI) values for each linear regression.

and in all cases their plasmon band red-shifts with increasing refractive index. A comparison of the plasmon sensitivity of these separated fractions (222 nm/RI, upper; 328 nm/RI, middle; and 385 nm/RI, lower, see Figure 7B) vs. that of the unsorted sample (293 nm/RI, see Figure 5B) clearly shows that it is possible to selectively isolate fractions whose sensitivity is larger than that of the original dispersion, and commensurate with that of an unsorted dispersion of approximately the same average size but with higher branching (319 nm/RI, see Figure 5B). The results in Figure 7B also show that the larger nanoparticles have a greater sensitivity to the surrounding medium. This may be explained by larger core sizes being able to accommodate a greater number of branches, which would likely increase sensitivity as established previously. Nevertheless, we cannot exclude here any influence stemming from differences in hollowness between the three different fractions, since this could also play a role in their different sensitivities. The sensitivities reported here for the larger separated nanoparticle fractions shown in Figure 7 are higher than those obtained for gold nanorods within the same size domain and with plasmon resonances centered at similar wavelengths (55 × 16 nm, λmax 728 nm, Δλ/RI of 224; 74 × 17 nm, λmax 846 nm, Δλ/RI of 288).[44] This may be due to the hollow nature of our nanoparticles and the increased number of branches, and therefore sharp tips, that they possess. Furthermore, such small, highly branched and hollow nanoparticles with a plasmon resonance compatible with common laser wavelengths should be useful for SERS applications, which have been recently reported to benefit from the high tip density provided by smaller branched nanoparticles.[12] Conversely, larger multibranched hollow nanoparticles with strong NIR plasmonic absorption are desirable for photothermal and photodynamic therapies,[16] where their hollow nature could allow for the possible encapsulation of therapeutic agents. All in all, the possibility to sizeselectively separate our multibranched hollow nanoparticles into fractions with on-demand dimensions and branch density opens the door for their exploitation as highly versatile theranostic agents.

the reaction of HAuCl4, AgNO3 and ascorbic acid in a TX-100 solution at room temperature. The extent of branching can be controlled by adjusting the ascorbic concentration, with the nanoparticles remaining hollow; while the presence of Ag+ ions during growth is found to be key for branch formation while also influencing branch aspect ratio and overall core size. The nanoparticles mainly consist of Au and contain a small amount of Ag, likely due to a thin silver coating on the branch surface. We found no evidence of a templated galvanic replacement-driven formation. Instead, our findings indicate that nanoparticle hollowness likely stems from the fast agglomeration of small Au nanoparticles nucleated in situ during the first ca. 0.5 min of reaction followed by further deposition into a cage-like geometry, while branch formation is the predominant type of growth thereafter. The hollow nature and multiple branches of the nanoparticles determine the NIR character of their plasmon resonance, which can be spectrally tuned over a wide range (≈700–2000 nm) as a function of the nanoparticle dimensions, hollowness, branch density and branch length. We have also demonstrated that the already high dielectric sensitivity of the multibranched hollow gold nanoparticles can be improved even further by separating the colloidal dispersions into fractions of different sizes and narrower size distributions through a highly versatile centrifugal sorting in a glycerol gradient. In summary, we have presented a straightforward synthesis and size sorting approach to obtain narrow size distributions of multibranched gold nanoparticles with hollow cores, predictable dimensions (50–300 nm) and branching, and therefore with tailored plasmonic properties in the NIR. Nanoparticle hollowness offers a competitive advantage for the conversion of absorbed photons into heat, while the spikes can serve as spots for electric field amplification. The low affinity of TX-100 for the gold surface facilitates functionalization with suitable capping ligands, offering versatility. Moreover, the possibility to obtain nanoparticles with tailor-made absorption, scattering, and field enhancing properties in the NIR highlights the potential of these nanoparticles for theranostic applications.

3. Conclusion

4. Experimental Section

We have shown that multiply-branched hollow gold nanoparticles with NIR plasmon resonances can be synthesized with high yield via a simple and rapid seedless process involving

Chemicals: Polyethylene glycol tert-octylphenyl ether (Triton X-100, ≈99%, Fluka), L-ascorbic acid (AA, ≥99.0%, Sigma), silver nitrate (AgNO3, 99.9999%, Aldrich), hydrogen tetrachloroaurate(III)

small 2015, 11, No. 35, 4550–4559



4557


trihydrate (HAuCl4·3H2O, ≥49% Au basis, Sigma-Aldrich), α-methoxyω-mercapto polyethylene glycol (CH3O-PEG-SH, 5000 Dalton, Rapp Polymere), glycerol (≥99.5%, Sigma-Aldrich), ethanol (≥99.8%, Sigma-Aldrich), 1-butanol (≥99%, Sigma), pyridine (≥99%, Aldrich), N,N-dimethylformamide (DMF, ≥99.9%, Aldrich), and dimethyl sulfoxide (DMSO, ≥99.9%, Sigma-Aldrich) were purchased from the retailer and used as received. All water used in this study was from a Millipore Milli-Q ultrapure water purification system having a resistivity of 18.2 MΩ cm and total organic carbon of 2 ppb. Synthesis of Multibranched Hollow Gold Nanoparticles: 18.75 g of TX-100 was dissolved in 100 mL H2O to give a stock solution of approximately 0.26 M TX-100, which was diluted for synthesis. In general, synthesis was performed in 20 mL glass vials, with appropriate volumes of 4 × 10−3 M AgNO3 (10–350 µL; standard conditions = 125 µL), 0.102 M HAuCl4 (25 µL), and 0.08 M AA (50–320 µL; standard conditions = 87.5 µL) added in this order to 5 mL of 0.13 M TX-100. Both AgNO3 and AA solutions are freshly prepared before each synthesis. The vial was mixed via manual agitation after addition of AgNO3 and again after addition of HAuCl4. AA is pipetted directly into the solution which is subsequently mixed by immediate but gentle shaking for 1–2 s afterward. It should be noted that this synthesis may be sensitive to the mixing protocol; the manual shaking should be performed in the same manner to obtain reproducibility, while reactants are recommended to be injected into the solution bulk. Addition of all reagents is completed in less than 30 s. At standard conditions (0.096 × 10−3 M AgNO3, 0.49 × 10−3 M HAuCl4, and 1.34 × 10−3 M AA) the solution changes color from light yellow to transparent after AA addition, subsequently developing a faint gray-green color within 30 to 40 s which darkens over 3–5 min. The solution is then allowed to stand for 1.5 to 2 h of undisturbed growth at room temperature. After this time, the nanoparticles were capped by addition of 5 µL of 5 × 10−3 M PEG-thiol per mL of sample aliquot with subsequent gentle magnetic stirring for at least 30 min. Note that the nanoparticles exhibit poor stability if coated only with TX-100 and aggregate on centrifugation due to weak interaction with the surfactant. Addition of PEG-SH induces facile displacement of TX-100 and allows for repeated washing steps.[8] After PEG capping, the dispersion was diluted with water in a 1:4 dilution and washed by three centrifugation-water redispersion steps (10 krpm for 15 min to ensure collection of smaller nanoparticles) to remove the excess surfactant and capping agent. All glassware was precleaned with aqua regia (a mixture of HCl:HNO3 in a 3:1 volume ratio) and rinsed thoroughly with deionized water beforehand, including brief sonication. Arrested Growth Studies: Synthesis was initiated as stated in the previous section, using 3× volumes to increase the available material in earlier fractions. 2 mL of the growth solution was extracted at the given time intervals and immediately injected into premade quenching solutions of 48.5 mL H2O containing 0.05 × 10−3 M PEG and kept on ice. These were mixed manually for 2–3 min before centrifugation for 15 min at 9 krpm. UV–vis–NIR and TEM analysis of the particles was performed after a second washing step at the same centrifugation conditions and redispersion in 400 µL of H2O. Separation in Density Gradients: Density gradients were prepared by careful layering of aqueous solutions containing 10%–90% glycerol by weight into 15 or 50 mL polypropylene centrifuge tubes. Steps of 10% were used with step volumes of 1–2 or 3 mL for the


different tubes, respectively. 200 to 600 µL of a concentrated dispersion of PEG-capped multibranched gold nanoparticles in water was then layered on top of the gradient before processing in an Eppendorf 5810 R centrifuge with an A-4–62 swinging bucket rotor at 4 krpm. The samples were spun for 5 or 10 min iterations with photographs recorded at each interval to monitor the progression of any bands. At completion, bands were removed via pipette and diluted with water. Samples were subsequently pelleted and resuspended in water three times before analysis. UV–vis–NIR Spectroscopy: UV–vis–NIR spectra were recorded in 10 mm path length SUPRASIL quartz cuvettes on a Cary Varian 5G spectrophotometer at 909 nm min−1 using 0.13 M TX-100 (or another appropriate solution) as a baseline. For spectra over the region 1400–2300 nm, 2 mm path length cells were used, allowing the partial subtraction of strong water absorbance in this range. Transmission Electron Microscopy: For TEM analysis 2–5 µL aliquots of samples were drop-dried onto carbon film coated copper grids. Low accelerating voltage TEM images were recorded on a JEOL JEM 1101 TEM operating at 80 kV. An FEI Titan 80–300 equipped with an EDAX EDX detector was used at 300 kV for TEM imaging, EDX spectroscopy, and scanning transmission electron microscopy in high-angle annular dark field (STEM-HAADF) mode. Size measurements were performed manually using ImageJ. Values for perimeter and area were extracted from TEM images of single nanoparticles by converting to a binary image and using the inbuilt “analyze particles” algorithm. Scanning Electron Microscopy/Focused Ion Beam: 3 µL of sample was typically deposited onto a Si substrate for SEM analysis and images were recorded using a Gemini Ultra Plus field emission SEM (Zeiss) instrument at 2 kV and a working distance of 2 mm. For both SEM imaging and focused ion-beam processing 20 µL of nanoparticle dispersion was deposited on a glass coverslip and blown dry with N2. The coverslip was carbon coated before imaging in a FEI Strata 400S at 5 kV. A section of each individual particle was removed by scanning a 30 kV Ga Ion beam at 1.5 pA over a rectangular area, which was complete in 3 to 5 s.

Supporting Information Supporting Information is available from the Wiley Online Library or from the author.

Acknowledgements The authors acknowledge funding from the European Union’s Seventh Framework Programme for research, technological development, and demonstration under Grant Agreement No. 310250 (project UNION). The authors also acknowledge the Karlsruhe Nano Micro Facility (KNMF, www.kit.edu/knmf) of the Forschungszentrum Karlsruhe for provision of access to instruments at their laboratories and the authors would like to thank Dr. Torsten Scherer for assistance in using the Electron Microscopy and Spectroscopy Laboratory (EMSL) at the Institute of Nanotechnology (INT) at the Karlsruhe Institute of Technology (KIT).


small 2015, 11, No. 35, 4550–4559

www.MaterialsViews.com [1] L. Zhang, W. Niu, G. Xu, Nano Today 2012, 7, 586. [2] S. E. Lohse, C. J. Murphy, Chem. Mater. 2013, 25, 1250. [3] A. Guerrero-Martínez, S. Barbosa, I. Pastoriza-Santos, L. M. Liz-Marzán, Curr. Opin. Colloid Interface Sci. 2011, 16, 118. [4] J. Xiao, L. Qi, Nanoscale 2011, 3, 1383. [5] R. A. Sperling, P. R. Gil, F. Zhang, M. Zanella, W. J. Parak, Chem. Soc. Rev. 2008, 37, 1896. [6] T. K. Sau, A. L. Rogach, Adv. Mater. 2010, 22, 1781. [7] S. Umadevi, H. C. Lee, V. Ganesh, X. Feng, T. Hegmann, Liq. Cryst. 2013, 0, 1. [8] P. Pallavicini, A. Donà, A. Casu, G. Chirico, M. Collini, G. Dacarro, A. Falqui, C. Milanese, L. Sironi, A. Taglietti, Chem. Commun. 2013, 49, 6265. [9] H. Yuan, C. G. Khoury, H. Hwang, C. M. Wilson, G. A. Grant, T. Vo-Dinh, Nanotechnology 2012, 23, 075102. [10] A. Casu, E. Cabrini, A. Donà, A. Falqui, Y. Diaz-Fernandez, C. Milanese, A. Taglietti, P. Pallavicini, Chem. - Eur. J. 2012, 18, 9381. [11] X. Wang, D.-P. Yang, P. Huang, M. Li, C. Li, D. Chen, D. Cui, Nanoscale 2012, 4, 7766. [12] Z. Liu, Z. Yang, B. Peng, C. Cao, C. Zhang, H. You, Q. Xiong, Z. Li, J. Fang, Adv. Mater. 2014, 26, 2431. [13] Z. Liu, L. Cheng, L. Zhang, Z. Yang, Z. Liu, J. Fang, Biomaterials 2014, 35, 4099. [14] L.-C. Cheng, J.-H. Huang, H. M. Chen, T.-C. Lai, K.-Y. Yang, R.-S. Liu, M. Hsiao, C.-H. Chen, L.-J. Her, D. P. Tsai, J. Mater. Chem. 2012, 22, 2244. [15] T. K. Sau, A. L. Rogach, M. Döblinger, J. Feldmann, Small 2011, 7, 2188. [16] P. Vijayaraghavan, C.-H. Liu, R. Vankayala, C.-S. Chiang, K. C. Hwang, Adv. Mater. 2014, 26, 6689. [17] G. Kawamura, Y. Yang, K. Fukuda, M. Nogami, Mater. Chem. Phys. 2009, 115, 229. [18] L. Sironi, S. Freddi, M. Caccia, P. Pozzi, L. Rossetti, P. Pallavicini, A. Donà, E. Cabrini, M. Gualtieri, I. Rivolta, A. Panariti, L. D’Alfonso, M. Collini, G. Chirico, J. Phys. Chem. C 2012, 116, 18407. [19] M. R. Langille, M. L. Personick, J. Zhang, C. A. Mirkin, J. Am. Chem. Soc. 2012, 134, 14542. [20] K. Mallick, Z. L. Wang, T. Pal, J. Photochem. Photobiol. Chem. 2001, 140, 75. [21] A. Pal, Talanta 1998, 46, 583. [22] T. K. Sau, A. Pal, N. R. Jana, Z. L. Wang, T. Pal, J. Nanoparticle Res. 2001, 3, 257.

small 2015, 11, No. 35, 4550–4559

[23] B. Aswathy, S. Suji, G. S. Avadhani, R. Aswathy, S. Suganthi, G. Sony, J. Mol. Liq. 2011, 162, 155. [24] L. Rodríguez-Lorenzo, J. M. Romo-Herrera, J. Pérez-Juste, R. A. Alvarez-Puebla, L. M. Liz-Marzán, J. Mater. Chem. 2011, 21, 11544. [25] W. Jia, J. Li, L. Jiang, ACS Appl. Mater. Interfaces 2013, 5, 6886. [26] W. Ahmed, E. S. Kooij, A. Silfhout van, B. Poelsema, Nanotechnology 2010, 21, 125605. [27] H. A. Day, D. Bartczak, N. Fairbairn, E. Mcguire, M. Ardakani, A. E. Porter, A. G. Kanaras, CrystEngComm 2010, 12, 4312. [28] C.-H. Kuo, M. H. Huang, Langmuir 2005, 21, 2012. [29] Y. Jin, S. Dong, J. Phys. Chem. B 2003, 107, 12902. [30] S. Pedireddy, A. Li, M. Bosman, I. Y. Phang, S. Li, X. Y. Ling, J. Phys. Chem. C 2013, 117, 16640. [31] L. Zhao, K. Ding, X. Ji, J. Li, H. Wang, W. Yang, Colloids Surf. Physicochem. Eng. Asp. 2011, 386, 172. [32] A. M. Goodman, Y. Cao, C. Urban, O. Neumann, C. Ayala-Orozco, M. W. Knight, A. Joshi, P. Nordlander, N. J. Halas, ACS Nano 2014, 8, 3222. [33] Y. Choi, S. Hong, L. Liu, S. K. Kim, S. Park, Langmuir 2012, 28, 6670. [34] A. M. Schwartzberg, T. Y. Olson, C. E. Talley, J. Z. Zhang, J. Phys. Chem. B 2006, 110, 19935. [35] W. He, X. Wu, J. Liu, X. Hu, K. Zhang, S. Hou, W. Zhou, S. Xie, Chem. Mater. 2010, 22, 2988. [36] J. G. Smith, Q. Yang, P. K. Jain, Angew. Chem. Int. Ed. 2014, 53, 2867. [37] S. R. Jackson, J. R. McBride, S. J. Rosenthal, D. W. Wright, J. Am. Chem. Soc. 2014, 136, 5261. [38] P. N. Njoki, I.-I. S. Lim, D. Mott, H.-Y. Park, B. Khan, S. Mishra, R. Sujakumar, J. Luo, C.-J. Zhong, J. Phys. Chem. C 2007, 111, 14664. [39] J. Rodríguez-Fernández, J. Pérez-Juste, F. J. García de Abajo, L. M. Liz-Marzán, Langmuir 2006, 22, 7007. [40] J. Xie, Q. Zhang, J. Y. Lee, D. I. C. Wang, ACS Nano 2008, 2, 2473. [41] Z. Liu, F. Zhang, Z. Yang, H. You, C. Tian, Z. Li, J. Fang, J. Mater. Chem. C 2013, 1, 5567. [42] K. M. Mayer, J. H. Hafner, Chem. Rev. 2011, 111, 3828. [43] F. Bonaccorso, M. Zerbetto, A. C. Ferrari, V. Amendola, J. Phys. Chem. C 2013, 117, 13217. [44] H. Chen, X. Kou, Z. Yang, W. Ni, J. Wang, Langmuir 2008, 24, 5233.


Received: January 12, 2015 Revised: March 18, 2015 Published online: June 10, 2015


4559

One-step synthesis of hollow porous gold nanoparticles with tunable particle size for the reduction of 4-nitrophenol.

Stepwise synthesis of cubic Au-AgCdS core-shell nanostructures with tunable plasmon resonances and fluorescence.

Activation of oxygen on gold and silver nanoparticles assisted by surface plasmon resonances.

Synthesis of hollow mesoporous silica nanoparticles with tunable shell thickness and pore size using amphiphilic block copolymers as core templates.

Tunable Dipole Surface Plasmon Resonances of Silver Nanoparticles by Cladding Dielectric Layers.

Multipole plasmon resonances in self-assembled metal hollow-nanospheres.

1064 nm SERS of NIR active hollow gold nanotags.

Plasmon hybridization in composite nanostructures with tunable resonances and vertex truncation analysis.

A tunable high-pass filter for simple and inexpensive size-segregation of sub-10-nm nanoparticles.

One-step synthesis of zero-dimensional hollow nanoporous gold nanoparticles with enhanced methanol electrooxidation performance.

Synthesis of Gold Nanoparticles with Buffer-Dependent Variations of Size and Morphology in Biological Buffers.

Trapping Iron Oxide into Hollow Gold Nanoparticles.

Highly sensitive SERS-based immunoassay of aflatoxin B1 using silica-encapsulated hollow gold nanoparticles.

Gold Nanobipyramid-Directed Growth of Length-Variable Silver Nanorods with Multipolar Plasmon Resonances.

Luminescent Gold Nanoparticles with Size-Independent Emission.

Designing Plasmon Resonances.

Templated fabrication of hollow nanospheres with 'windows' of accurate size and tunable number.

Hollow spherical gold nanoparticle superstructures with tunable diameters and visible to near-infrared extinction.

Ag2S-hollow Fe2O3 nanocomposites with NIR photoluminescence.

Tunable surface plasmon resonance frequencies of monodisperse indium tin oxide nanoparticles by controlling composition, size, and morphology.

Size tunable Au@Ag core-shell nanoparticles: synthesis and surface-enhanced Raman scattering properties.

Controllable synthesis of concave cubic gold core-shell nanoparticles for plasmon-enhanced photon harvesting.

Facile synthesis of gold nanoworms with a tunable length and aspect ratio through oriented attachment of nanoparticles.

A microwave-facilitated rapid synthesis of gold nanoclusters with tunable optical properties for sensing ions and fluorescent ink.