Impact of the Crosslinker’s Molecular Structure on the Aggregation of Gold Nanoparticles

2.1 Materials and Instrumentation

Biphenyl-4,4′-dithiol (95%, “BI”), p-terphenyl-4,4″-dithiol (96%, “TER”), nonane-1,9-dithiol (95%, “NON”), oleylamine (technical grade, “OA”) and toluene (p.a.) were obtained from Sigma-Aldrich. α-Methoxypoly(ethylene glycol)-ω-(11-mercaptoundecanoate) (~2000 g/mol)(PEGMUA) was synthesized as described previously [27]. Ultrapure water (18.2 MΩcm, Millipore) (MQ) was used for all procedures. Biphenyl-4,4′-dithiol, p-terphenyl-4,4″-dithiol, nonane-1,9-dithiol were dissolved in toluene (c=10 mM). To obtain solutions with lower concentrations, these solutions were further diluted. The gold nanoparticles used in this work were synthesised as described in Schulz et al. [23], the obtained particles were stabilised by citrate in water and had a diameter of 12.0±0.7 nm. UV/Vis spectra were recorded using a Varian Cary 50 conc, a Varian Cary 50 bio or a Perkin Elmer Lambda 25. DLS measurements were performed with a Zetasizer Nano ZS system (Malvern Instruments). The instrument uses a He-Ne-Laser (4.0 mW, 633 nm). The number of runs per measurement was set to 30, three measurements where always performed to ensure the reproducibility. A JEOL JEM-1011 transmission electron microscope operated at 100 kV was used for the TEM measurements.

2.2 Preparation of gold clusters in water

To induce the aggregation of the AuNP in water, 10 mL of citrate-stabilised AuNP (5 nM) were mixed with 5 mL of BI or TER in toluene (1 mM). The aqueous phase containing the AuNP was red and clear. Above, a turbid phase was formed by the linker solutions. The solutions were kept in the shaker for around 10 h.

In order to clean the samples from unbound molecules, the organic phase was removed. Afterwards, 5 mL of toluene was added and the solutions were placed in the ultrasonic bath for a few minutes. This process was repeated four times. After washing with toluene, the solutions were centrifuged for 20 min at 20,000 g. The clear and colourless supernatants were removed and 1 mL ultrapure water was added. This process was repeated three times. Samples of the purified solutions were analysed by UV/Vis spectroscopy and TEM. Samples for TEM were prepared by drop casting ~8 µL of the samples on a TEM grid.

The ligand exchange in water was tested with different amounts and concentrations of the educts. The effect of the centrifugation time on the aggregation was studied by repeating the experiments: 5 mL of AuNP in MQ were mixed with 0.5 mL toluene and the same amount of a 1 mM BI or TER solution. Additionally, one sample was prepared as a reference with 5 mL of AuNP in MQ and 1 mL toluene. Afterwards, UV/Vis spectra were recorded and 2 mL of the samples were transferred into Eppendorf-tubes and centrifuged 5 times (20,000 g, 15 min). After each centrifugation step the pellets were resuspended by ultrasonication and UV/Vis spectra were recorded. To test the stability of the obtained aggregates, small amounts of PEGMUA solution (1 mM) were added to some of the prepared aggregates.

2.3 Phase transfer from aqueous to organic phase

For the phase transfer into the organic phase, a solution of 200 mg (7.5·10⁻⁴ mol) oleylamine in 4 mL toluene was prepared. 0.5 mL of the solution was mixed with 10 mL toluene, then 10 mL of aqueous AuNP-solution (5 nM) was added. The aqueous phase blurred, the organic phase stayed clear and became reddish. After adding 500 µL (1 M) sodium chloride solution, the phases were well separated and clear. The aqueous phase was removed, the oleylamine-stabilised particles (OA-AuNP) in toluene were stored in a glass bottle.

This procedure was repeated several times with different reaction parameters such as different amounts of solvents, sodium chloride solution or AuNP solution without any notable influence on the reaction. In all reactions oleylamine was used in excess. To prove the absence of aggregation, DLS measurements were carried out.

2.4 Preparation of gold clusters in the organic phase

Solutions containing BI, TER or NON were added in different amounts (usually ranging from 10 µL to 1500 µL, 1 mM) to solutions containing OA-AuNP (3 mL, 3 nM). The aggregation was monitored using UV/Vis spectroscopy and DLS, the obtained samples were investigated using TEM. When the aggregation was monitored using UV/Vis or DLS, the samples were not shaken during the aggregation except directly after the addition of the linker solution.

3.1 Gold clusters in the aqueous phase

The aggregation of the AuNP in MQ after adding solutions containing BI or TER could be followed by the strong red shift of the UV/Vis spectra due to the plasmon coupling between the AuNP (Figure 2). The spectra are showing a slight red-shift due to the binding of the ligands as well as plasmon coupling [26] due to aggregation. Depending on the number of centrifugation cycles, the plasmon coupling becomes more and more pronounced. While the initial aggregation of particles crosslinked with BI is faster, the development of the plasmon coupling is more pronounced for particles crosslinked with TER. TEM measurements indicate a dendritic aggregation (see SI, Figure S1).

Fig. 2:

UV/Vis spectra (left) and TEM micrographs (right) of AuNP aggregates. In the upper row, the results for AuNP reacted with with biphenyl-4,4′-dithiol (BI) are shown, lower row shows the results for AuNP with p-terphenyl-4,4″-dithiol (TER). UV/Vis spectra are shown for different numbers of centrifugation cycles, with apparent increase of aggregation for each cycle.

During our experiments, the aggregation in water has shown to be difficult to control and uniform and size controlled aggregates in high yield could not be obtained. Also, TEM measurements of the aggregates are difficult to interpret because citrate-stabilised AuNP tend to also form aggregates during drying on the TEM grid. This makes it hard to distinguish aggregates formed in solution from those resulting from drying effects. Adding PEGMUA to aggregates in solution seem to reverse the aggregation to a certain degree. The plasmon coupling in the UV/Vis becomes less pronounced and TEM measurements reveal a lot of isolated particles (the images differ from those of particles directly functionalised with PEGMUA, though). Additional information is provided as supporting information, Figure S2.

A comparison of the centrifugation-induced aggregation of citrate-stabilised AuNP and TER-functionalised AuNP is provided in the SI, Figure S3.

3.2 Phase transfer from aqueous to organic phase

To induce the aggregation in toluene we first used a simple and straightforward protocol to transfer the citrate-stabilised AuNP from water into toluene using oleylamine as a ligand. This protocol is based on previous work by others [28], [29], [30]. We found the so-obtained oleylamine-stabilised AuNP to be long-term stable, the narrow size-distribution of the particles was preserved and no aggregation was observed in DLS measurements. Citrate-AuNP can be added repeatedly to increase the AuNP concentration in the organic phase by at least a factor of 10. This makes this approach interesting for experiments where high particle concentrations are needed, e.g. diffraction experiments. In contrast to citrate-stabilised AuNP, the oleylamine-stabilised particles formed closed packed films on TEM-grids when drop casted (Figure 3).

Fig. 3:

UV/Vis spectra, DLS data (left) and TEM measurements (right) of oleylamine-functionalised AuNP. A spectrum of citrate-stabilised AuNP is also included. The particles form two-dimensional, hexagonal packed layers on the TEM grid.

Since amine-bond ligands on gold are easily replaced by thiols [1], our OA-AuNP are a platform to study ligand-exchange induced aggregation in detail. The aggregates should be clearly differentiable from the OA-AuNP in TEM measurements, because these self-assemble into monolayers. The spectra of the AuNP are slightly affected by the bonding of the amine. This can be attributed to the changed refraction index after phase-transfer into toluene. DLS measurements confirm the absence of any aggregates.

3.3 Gold clusters in the organic phase

The UV/Vis spectra (Figure 4) for BI/TER [here, we focus on samples which have been prepared by adding 15 µL or 150 µL 1 mM ligand solution to the OA-AuNP solution (3 mL, 3 nM)] show a fast, initial increase of the absorbance accompanied by a small shift of the plasmon resonance by about 10 nm after the addition of the ligand solution. On a larger timescale (several hours), the AuNP start to aggregate and plasmon coupling becomes obvious in the UV/Vis spectra. Complementary DLS measurements do not indicate any aggregation directly after the addition of the ligand solution, but after several hours the particles were aggregated to clusters with sizes of roughly 100 nm. Aggregates prepared by the addition of NON are showing a different behaviour: According to the UV/Vis spectra and the DLS measurements, the particles start to aggregate directly after the addition of NON in toluene. We attribute the signals at large hydrodynamic diameters in the DLS data for BI and TER after 5 min to precipitates of the ligands – at least in part. BI and TER are not easily soluble in toluene, also indicated by the concentration-dependent turbidity of the according solutions. The precipitates could be removed by syringe filtration. In DLS measurements of purified samples just the main peak remained.

Fig. 4:

UV/Vis spectra (a), distributions of hydrodynamic diameters weighted by intensity as obtained by DLS measurements (b) and TEM measurements for oleylamine-stabilised AuNP (3 mL, 3 nM) functionalised with 150 µL of BI, TER or NON in toluene (c). Measurements are shown for particles before (label ‘AuNP’), directly after the addition of the ligand solution (‘0 min’), 5 min after the addition (‘5 min’) and 18 h (‘18 h’) after the addition of the linker solution. The TEM measurements were performed for samples prepared from the final (after 18 h) samples. Scale bars correspond to 100 nm.

As can be seen from the TEM measurements reacting OA-AuNP with BI or TER results in spherical aggregates. Aggregates with BI appear to be more dense, which is in accordance with the stronger plasmon coupling observable in the UV/Vis spectra and can probably be attributed to the molecules’ length. TER, about 5 Å longer, yields less-dense aggregates, which is in agreement with the weaker plasmon coupling observable in the UV/Vis spectra (Figure 4). For NON, the DLS data indicate aggregates of about 100 nm after the addition of the linker solution. After 18 h, much bigger aggregates are found (~1 µm). Based on TEM analysis we assume that the initially formed aggregates grow together to form superclusters, which seem to be made up from spherical cluster of about 100 nm. Similar structures were observed before with smaller AuNP [17], [18].

Regarding the aggregation mechanism, we assume a monomer-aggregate growth mechanism, where in a first step a dithiol monolayer forms on the particles’ surface, which causes the fast inital shift of the plasmon resonance. We note that this shift in toluene is in full accordance with predictions based on Mie-theory (see SI, Figure S4) because the refractive indices of BI and TER are significantly larger than those of toluene, oleylamine or NON. Afterwards and much slower, the particles start to aggregate and monomers (which here means functionalised thus “sticky” particles) start to grow onto big aggregates. TEM measurements taken during the aggregation process for NON-functionalised AuNP reveal big aggregates as well as unbound single AuNP, whereas after the aggregation, no monomers were observed. Similar growth mechanisms were discussed by Ahonen et al. [31] and Haute et al. [32].

The distinct differences between the aggregation of NON and the aromatic molecules BI/TER could be linked to two aspects: First, NON might more easily penetrate the oleylamine-layer of the particles because of its structural similarity. Second, there is a chance that the NON linker does not crosslink different particles, but binds with both thiol groups on one particles (a process termed backbonding or loop formation [33]), thus strongly destabilising the particles and inducing aggregation. The aromatic molecules, in contrast, are even claimed to stand perpendicular on the particles surface [33], [34].

A discriminative response of AuNP reacted with an aromatic and an aliphatic dithiol in ethanol/water mixtures was also reported by Rajendra et al. supporting the idea of fundamental structural differences in the according ligand layers [35]. This interpretation is also in accordance with structural models of dithiol-self assembled monolayers (SAMs) on planar gold substrates [33], [34].

By varying the amount of added linker solution, the speed of the aggregation reactions can be controlled (see Figure 5). From the recorded kinetics, the aggregation induced by BI seems to be strongly hindered. We attribute this to the length of the molecule: After the binding of one thiol group on the particles’ surface the other thiol group is not completely exposed, thus further aggregation is sterically hindered.

Fig. 5:

Aggregation kinetics obtained for the addition of 15 µL or 150 µL of ligand solution to 3 mL of AuNP (3 nM) stabilised with oleylamine. The aggregation was measured by calculating the aggregation parameter from the UV/Vis spectra (see SI).

Another interesting aspect concerning the aggregation kinetics can be seen when the oleylamine-stabilised AuNP are cleaned from unbound ligands prior to aggregation. This was done by centrifuging the solution with 20,000 g, removing the supernatant and adding clean toluene. This process was repeated two or four times. When the aggregation was repeated with exactly the same reaction parameters, the aggregation was significantly faster (see Figure 6 and SI, Figures S5 and S6). The aggregation kinetics could again be controlled by varying the amount of linker molecules, thus supporting our results discussed above.

Fig. 6:

Aggregation kinetics and UV/Vis spectra obtained for the addition of 150 µL of BI in touluene to 3 mL of AuNP (3 nM) stabilised with oleylamine, which have been cleaned by two centrifugation cycles. UV/Vis spectra (right) were taken every 2 min and are shown from the beginning (blue) until 2 h (yellow). The aggregation parameter obtained for uncleaned oleylamine-stabilised particles is also shown.