Introduction

The physical properties of metal nanoparticles strongly depend on their assembled states. It is known that an assembly of gold nanoparticles has different light-absorption properties to that of single particles.1, 2 Based on this phenomenon, a wide variety of biomedical applications has been accomplished.3, 4 In addition, by manipulating the hybridization process involved in anchoring DNA to the surface of particles, cluster formation and morphology can be precisely controlled.5, 6, 7 These manipulation methods for the nanoparticle assembly are versatile and can be used to construct biosensors with good response to stimuli.8, 9 Furthermore, additional functions have been derived from the specific distribution of nanoparticles in an assembly. For instance, a class of unique optical functional materials involving metamaterials has been manufactured with a well-ordered assembly of gold nanoparticles.10, 11, 12, 13 Therefore, precisely controlling the morphology and relative positions of nanoparticles in an assembly is critically important to obtain superior or unexpected functions.14, 15, 16, 17, 18, 19, 20

Structural manipulation with light has various advantages. A rapid, time-sensitive, and site-sensitive response can be achieved without additives. In previous reports, the photoisomerizations of azobenzene21, 22, 23, 24, 25, 26 and stilbene derivatives27, 28, 29 tethered to the surface of nanoparticles have been investigated. Based on the structural changes triggered by the light irradiation, the dispersion states were controlled. The aggregation/dispersion of gold nanoparticles can also be modulated reversibly by the photo-triggered dimerization of thymine and the splitting of the thymine dimer.30 Although the formation of aggregates can be controlled with various methods, as mentioned here, modulating the morphology of the aggregates is less well established. In particular, few examples have controlled the interparticle distances in the nanoparticle aggregates. We have previously demonstrated that imidazolium-presenting gold nanoparticles can form stable and well-dispersed aggregates in aqueous solutions.31 The hydrophobicity of the surface was modulated via the anion exchange of the imidazolium cation at the surface of the nanoparticles. Finally, the interparticle distances of the gold nanoparticles in the aggregates could be regulated at the nanometer scale. Accordingly, the light-absorption properties in the visible region originating from the surface plasmon resonance of the gold nanoparticles could be tuned under mild conditions. The next challenge is to construct a modulation system for the distances between the gold nanoparticles in the aggregate by external stimuli, such as light irradiation.

Here we report the modulation of the interparticle distances in aggregates of gold nanoparticles via light irradiation. Modified nanoparticles covered with an imidazolium salt were prepared and mixed with a carboxylate-containing linker that has a photo-cleavable o-nitrobenzyl ester. In the first step, the linker molecules gathered the synthesized nanoparticles to form a stable and well-dispersed aggregate under aqueous conditions. Triggered by ultraviolet (UV) irradiation, the o-nitrobenzyl ester in the linker was cleaved, resulting in the generation of another carboxylate group. Finally, the nanoparticles were tightly bound by the dicarboxylate linkers. These changes can be monitored by microscopic observations. In addition, significant changes were induced in the optical properties of the gold nanoparticles. To the best of our knowledge, this is the first example of a light-driven modulation system for the distribution of nanoparticles in an aggregate.

Experimental Procedure

General

All of the reactions were conducted under a nitrogen atmosphere unless otherwise stated. Chromatographic purifications were performed using Wako gel C-200 (Wako Pure Chemical Industries, Ltd., Osaka, Japan). An ultrafilter unit (USY-5, molecular weight cutoff=50 000) was purchased from ADVANTEC Co., Ltd (Tokyo, Japan). 1H and 13C nuclear magnetic resonance (NMR) spectra were obtained with a JEOL EX-400 spectrometer (400 MHz; JEOL Ltd, Tokyo, Japan) using chloroform-d1, deuterium oxide and dimethylsulfoxide-d6 as solvents and either tetramethylsilane or trimethylsilylpropionic acid sodium salt as an internal reference. UV–visible spectra were measured on a SHIMADZU UV-3600 spectrophotometer using quartz cuvettes with a 1-cm optical path length (SHIMADZU CORPORATION, Kyoto, Japan). Transmission electron microscopy (TEM) was performed using a JEOL JEM-1025 operated at a 100-kV electron beam accelerating voltage (JEOL Ltd). One drop of sample was deposited onto a carbon grid, the excess liquid was blotted off with a Kimwipe and the grid was dried under ambient conditions. Dynamic light scattering was measured on a FPAR-1000 (Otsuka Electronics Co., Ltd, Osaka, Japan) at 20.0 °C, using the Marquardt method to solve the autocorrelation functions. The samples were irradiated with UV light with a spiral-shaped low-pressure mercury lamp. The samples were added to a quartz-glass tube and were placed at the center of the spiral-shaped lamp. Carboxylate 132 and imidazolium-presenting gold nanoparticles NP1–331 with various degrees of hydrophobicity were prepared according to previous reports.

Preparation of the gold nanoparticle aggregates

To the aqueous dispersion of nanoparticles (5 ml, 4 × 10−4 mol of imidazolium), 1 ml of 0.04 m aqueous solution of 1 was added and left for 1 h, followed by washing with 1 ml of water five times using an ultrafilter unit (ADVANTEC USY-5, molecular weight cutoff=50 000). Then the nanoparticles were dispersed in 6 ml of distilled water and measured.

Photoreactions

A typical procedure involved the following steps: the sample dispersion (3 ml) was added to a quartz-glass test tube, which was located at the center of the spiral-shaped mercury lamp. Then the UV irradiation was performed at room temperature.

Results and Discussion

The synthesis of the gold nanoparticle aggregates and the schematic model of the regulation of the interparticle distances are illustrated in Scheme 1. To form a water-dispersive aggregate of gold nanoparticles, the carboxylate anion 1, which has a photo-cleavable functional group, was mixed into the imidazolium-presenting gold nanoparticles by employing ionic liquid structures33, 34 via anion exchange with chloride anions. Under aqueous conditions, the gold nanoparticles should be aggregated because of the enhanced hydrophobicity due to 1. Using light irradiation, another carboxyl group could be generated, followed by an interaction with the neighbor imidazolium cation. Following this procedure, the nanoparticles are expected to be tightly bound. Compound 1 was synthesized according to a previous study.32 The sodium form of 1 was obtained as a pale yellow solid with a cation exchange resin. Modified gold nanoparticles with the chloride form of imidazolium were prepared according to our previous report.31 To examine the influence of surface hydrophobicity on the behavior of the aggregates, three types of nanoparticles were prepared with various substituent structures in imidazolium. NP1 has the most hydrophilic surface of the three, and NP3 has the most hydrophobic and bulky imidazolium moiety substituent. The anion exchange from chloride anions to carboxylate anions was examined using 1H NMR measurements (Supplementary Figure S1). To D2O solutions of imidazolium, 10 eq. of 1 was added and incubated for 60 min at room temperature. Then 1H NMR measurements were performed. The peak position of the NMR signal derived from the proton at the 2-position in the imidazolium ring was shifted from 8.55 to 9.74 p.p.m. by adding 1. This result suggests that anion exchange occurred.

Initially, the aggregation behaviors of the nanoparticles were investigated. Figure 1 shows representative TEM images of NP2 before and after adding 1. Well-dispersed states were obtained from all of the dispersions of the synthesized nanoparticles before adding 1.31 By contrast, aggregate states were clearly induced in the dispersions, especially those containing NP2 and NP3, which possess relatively hydrophobic alkyl groups in the imidazolium moiety in the presence of 1. From the sample containing NP1, which has the most hydrophilic modification of the three, aggregates were partially formed. The interparticle distances were estimated as an averaged value of the distance between the centers of the particles from the TEM observations. Table 1 presents the averaged interparticle distances between the nanoparticles in the aggregates. It is likely that the nanoparticles with longer alkyl chains can aggregate because of their strong hydrophobicities.

Figure 1
figure 1

TEM images of NP2 (a) before and (b) after addition of 1.

Table 1 Summary of absorption maxima and the interparticle distances of the aggregates of the modified gold nanoparticlesa

It is known that the light-absorption ability of gold nanoparticles can indicate the dispersion state of the nanoparticles.31 By forming aggregates, the absorption band can be shifted to a longer-wavelength region. Correspondingly, as a representative result, an absorption band was observed with a peak at 512 nm before adding 1 to the dispersion of NP2 (Figure 2). By contrast, a red-shifted absorption band was observed with a peak at 522 nm. With the NP3 nanoparticles, a peak shift to the longer-wavelength region was also induced by adding 1. These data clearly indicate that carboxylate 1 can enhance the formation of aggregates. Furthermore, from the dynamic light scattering measurements, an increase of the hydrodynamic radii from 5.4±1.4 to 11.3±2.0 nm was observed by adding 1 to the dispersion of NP2 nanoparticles. These data also support the formation of well-dispersed synthesized nanoparticle aggregates.

Figure 2
figure 2

UV absorption spectra of the nanoparticles in aqueous dispersions at room temperature. The pictures show the appearances of the dispersions before and after UV irradiation for 10 min.

Before photoreactions with the synthesized nanoparticles, we examined the reactivity of 1 and determined the products after light irradiation. Because of the absorption maximum of 1 at 266 nm, a low-pressure mercury lamp (>300 nm) was used to induce a photo-cleavage reaction. To 5 ml of a D2O solution of 1 (0.01 M), UV light was irradiated through quartz glass, and the reactions were monitored using 1H NMR (Figure 3). After photo-irradiating the solution of 1, the signal intensity derived from the aromatic ring in 2 decreased with the photo-irradiation time, whereas the signal intensity around δ 2.48 p.p.m., derived from monosodium succinate, increased. These data indicate the progress of the photo-cleavage leading to the generation of the carboxyl group. By evaluating the conversion, 80% of 1 was determined to be consumed within 30 min. The spectra showed that less by-product was generated by UV irradiation. By contrast, by increasing the irradiation time, the consumption yield of 1 was improved. However, unexpected side-reactions were observed. The photo-induced cleavage of 1 was also monitored with UV–vis measurements (Figure 4). UV light from a low-pressure mercury lamp irradiated a 3.3 × 10−5 m aqueous solution of 1 through quartz glass, and changes in absorptions were recorded using a UV–vis spectrometer. Photo-irradiation to 1 caused the decrease of the absorption band at 266 nm. These data indicate the generation of o-nitrosobenzaldehyde.35, 36 In other words, the data support the production of a carboxyl group.

Figure 3
figure 3

(a) NMR spectra of sodium succinate and 1 (0.01 m) before and after low-pressure mercury lamp irradiation (in D2O, 400 MHz, with trimethylsilylpropionic acid sodium salt used as an internal standard). (b) Plot of conversion vs irradiation time. The conversions were calculated from the ratio of the NMR signal intensities observed at 2.41 and 5.54 p.p.m.

Figure 4
figure 4

(a) Time course of the NP2 spectra in aqueous dispersion during UV irradiation. (b, c) Magnified spectra. (d) Time course of the absorption change at 266 nm during UV irradiation.

The photoreaction was carried out with the same light source at room temperature for 10 min. From the TEM observations, the interparticle distances were evaluated (Figure 5). The UV irradiation decreased the interparticle distances. The averaged diameter of NP2 decreased from 2.2±0.7 to 1.6±0.5 nm. From the dynamic light scattering measurements, a drastic increase in the hydrodynamic radii, from 11.3±2.0 to 618±403 nm, was observed. These data clearly indicate that UV irradiation can induce tight binding in the aggregates. Even from the dispersion of NP1, which was too well dispersed to determine the interparticle distances in the presence of 1, aggregates with an interparticle distance of 1.6±0.5 nm were observed. Thus the data show that the product of 1 after UV irradiation has a superior nanoparticle aggregation property.

Figure 5
figure 5

TEM images of (left) NP1, (middle) NP2, (right) NP3 after (upper) addition of 1 and (bottom) UV irradiation for 10 min. The scale bars represent 200 nm.

The optical properties were greatly influenced by the UV irradiation (Figure 2 and Supplementary Figure S2). After the UV irradiation of the dispersion, the absorption band with the 266-nm peak decreased. It was hypothesized that the photo-cleavable group would be removed, leading to the generation of carboxylates. Significantly, UV irradiation caused a red-shift in the surface plasmon absorption. Peak shifts by +37 nm (NP1), +30 nm (NP2) and +5 nm (NP3) after UV irradiation were observed in the dispersions. The relatively smaller degree of red-shift in NP3 after UV irradiation could be derived from the intrinsic tight binding among the nanoparticles. Because of the strong hydrophobicity of the alkyl chains in NP3, aggregation could be induced only by adding compound 1. The conversion after 10 min of irradiation was calculated to be 69%. After longer irradiations, the consumption rate increased to 80% after a 20-min irradiation and finally 100% after a 30-min irradiation.

In the absence of 1, the absorption property of the well-dispersed NP2 was not strongly influenced by the light irradiation (Supplementary Figure S3). In addition, to prove the necessity of a photo-responsive anion, we conducted the same experiment with sodium benzoate instead of 1 (Supplementary Figure S4). Similar to 1, sodium benzoate also induced the aggregation of NP2, and the red-shifted absorption with the 520 nm peak was observed. However, significantly less red-shift was observed after the UV irradiation of the sample. These data indicate that the photo-moiety in 1 is essential to operate this system.

Conclusion

Our results demonstrate two significant points. First, a simple tool for manipulating the distribution of nanoparticles after the formation of aggregates was developed. In addition, these changes can be detected as a significant alteration in the optical properties. Based on this method, advanced optical materials can be produced with unique characteristics originating from precisely controlled nanostructures. Second, we can control the distribution of the nano-sized components in the system with small molecules and light irradiation. Our materials can significantly absorb light in the near infrared region. Our materials could potentially improve the conversion efficiency of solar cells by enhancing their light absorption efficiency in the near infrared region. In addition, because photosynthesis in several types of algae can be enhanced or suppressed by light irradiation in the specific wavelength range between the deep red and near infrared regions, materials that filter incident sunlight can be applied to plant growth.37 Our materials can be used as filtering materials for modulating plant growth. Thus our concept presented here is feasible not only for constructing nano-structured materials but also for dynamically regulating nano-structures.

scheme 1

Schematic illustration of the modulation of interparticle distances with the photo-responsive linker 1. A full color version of this scheme is available at the Polymer Journal online.