Continuous Electroformation of Gold Nanoparticles in Nanoliter Droplet Reactors

Abstract

Gold nanoparticles (AuNPs) are employed in numerous applications, including optics, biosensing and catalysis. Here, we demonstrate the stabilizer‐free electrochemical synthesis of AuNPs inside nanoliter‐sized reactors. Droplets encapsulating a gold precursor are formed on a microfluidic device and exposed to an electrical current by guiding them through a pair of electrodes. We exploit the naturally occurring recirculation flows inside confined droplets (moving in rectangular microchannels) to prevent the aggregation of nanoparticles after nucleation. Therefore, AuNPs with sizes in the range of 30 to 100 nm were produced without the need of additional capping agents. The average particle size is defined by the precursor concentration and droplet velocity, while the charge dose given by the electric field strength has a minor effect. This method opens the way to fine‐tune the electrochemical production of gold nanoparticles, and we believe it is a versatile method for the formation of other metal nanoparticles.

Droplets encapsulating a gold precursor are generated on a microfluidic device. The droplets are then guided through a pair of parallel electrodes, where they are exposed to an electrical current, and gold nanoparticles are created.

Gold nanoparticles (AuNPs) have unique size‐related optical, chemical, and electromagnetic properties that make them extraordinarily versatile nanomaterials for applications in optics, electronics, sensors, catalysis, and biomedicine.[ ¹ , ² , ³ , ⁴ ] The surface chemistry of AuNPs can be easily functionalized to bind ligands such as polymers, nucleic acids, peptides, proteins and antibodies.[ ⁵ , ⁶ , ⁷ ] Plenty of studies have focused on the synthesis of AuNPs by chemical and electrochemical reduction, plasma‐induced methods, spark ablation techniques, and seeding growth mechanisms.[ ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² ] Recently, new approaches have allowed for better control of their size, shape, chirality, stability and reproducibility, as well as for “greener” technologies.[ ¹³ , ¹⁴ , ¹⁵ , ¹⁶ ]

A main challenge in the synthesis of any colloidal nanoparticle is their tendency to aggregate, ^[17] where the aggregation not only results in a lower surface‐to‐volume ratio, detrimental for catalysis, but also in a drastic change on their physicochemical properties and interaction with other materials. For example, the intracellular uptake of spherical gold nanoparticles is governed by their size,[ ¹⁸ , ¹⁹ , ²⁰ ] and aggregation can lead to a potential cell toxicity. ^[20]

In order to limit or control aggregation, most studies report the use of capping or stabilizing agents. For example, the chemical reduction of gold (III) chloride by citrate remains one of the most employed methods for AuNPs synthesis, ^[8] because citrate reduces the gold precursor into colloidal gold and simultaneously binds to the nanoparticles, thereby stabilizing them. The main drawback of chemical‐based synthesis is that the reducing agents are highly reactive. ^[8] Thus, a precise control of the reaction time and several washing steps are usually needed. While other molecules (e.g., surfactants, polymers, and biomolecules) can be used to stabilize the surface of AuNPs, they modify their electron transfer mechanisms and the resulting electromagnetic properties. ^[21] The electrochemical method is an attractive alternative to chemical reduction, as it produces high yields with minimum unwanted side‐products, and the size of the synthetized nanoparticles can be adjusted by the precursor concentration and current density. ^[22] While the created nanoparticles are prone to aggregate at the surface of the electrodes, the ability to quickly stop the synthesis (by removing the electric potential) enables the use of new alternatives to control aggregation. For instance, a conductive, ultrasonic probe has been employed to simultaneously form AuNPs on its surface and quickly remove them to limit coagulation. ^[23] While interesting, most electrochemical‐based studies report experiments performed in bulk, which tend to produce batch‐to‐batch variations and require complex operations to control particle size.

In recent years, microfluidic technology has been successfully employed for the synthesis of metal nanoparticles.[ ²⁴ , ²⁵ , ²⁶ , ²⁷ ] Microscale features, including the use of laminar streams and a precise control of temperature and residence time, have resulted in improved size distributions of the nanoparticles, as well as in a better understanding of their nucleation and growth. In particular, droplet microfluidics has emerged as a powerful tool to form particles and crystals of nanometric sizes. ^[28] Water‐in‐oil droplets are formed at high frequencies with monodisperse volumes from pico‐ to nanoliters. Serving as individual reactors, droplets provide a reproducible environment, where the reaction conditions can be precisely tailored. This has been demonstrated for the synthesis of metal nanoparticles,[ ¹⁶ , ²⁹ , ³⁰ ] metal oxide particles, ^[31] metal–organic frameworks, ^[32] and quantum dots. ^[33]

Studies combining the advantages of electrochemical synthesis with nanoliter droplet reactors have focused on the deposition of nanostructures at the liquid‐liquid interface of non‐confined droplets.[ ³⁴ , ³⁵ ] In this contribution, we report the continuous electrochemical synthesis of gold nanoparticles inside droplets sandwiched between a pair of electrodes (confined). When flowing under confinement in a rectangular microchannel, droplets develop internal recirculation profiles that help to mitigate the aggregation of nanoparticles after nucleation. We demonstrate that nanoparticle size can be shifted towards smaller or larger particles on demand, by changing the precursor concentration, droplet velocity and electrical current magnitude.

A gold (III) chloride solution (1, 5 and 10 mM) was used as the gold precursor. Nanoliter‐sized droplets were prepared using a polydimethylsiloxane (PDMS) microfluidic device following a flow‐focusing approach (Figures 1A and D, Figure S1), where the precursor solution met an oil stream in a cross junction. ^[36] The symmetrical shearing force of the oil stream caused the precursor‐containing solution to detach and form droplets with an average volume of 1.5 nL. In order to minimize electrical current leakage through the oil phase, we employed a fluorinated oil with high electrical resistivity (Fluorinert FC‐40). The oil phase was supplemented with 0.1 %v/v of fluoro‐surfactant to form stable, monodispersed droplets (CV<3%). The ratio of the water‐to‐oil flow rates was set to one in all cases. A change in either the droplet size or shape results in different mixing conditions and shear rates, influencing the formation of gold nanoparticles. Therefore, to avoid any bias caused by different droplet shapes or sizes, the oil and aqueous flow rates were dynamically adjusted to maintain a consistent droplet volume throughout all experiments, while the channel length remained constant.

Figure 1.

A) Schematic illustration of the flow‐focusing approach for droplet generation. The recirculation flows inside the droplets are indicated with bold arrows. B) After generation, the particles are sandwiched by two carbon‐based membranes, which interface the droplets with the gallium electrodes. C) Photo of the microfluidic device. The power supply electrodes are connected to the patterned gallium electrodes via two platinum wires. Micrographs of the microchannels for D) droplet generation and E) for the flow channel in‐between the pair of membranes and parallel electrodes.

After generation, droplets moved through a region with two parallel gallium electrodes (Figures 1B and E), where an electrical current was induced inside the droplets by applying a potential difference across the electrodes. A key component for the functioning of the device was the integration of two thin (100 μm) carbon‐nanotube‐PDMS membranes between the droplets and electrodes (Figures 1B and E). ^[36] These membranes enabled the flow of electrons into the droplets, necessary for the electrochemical reduction of the gold ions, and were crucial to achieve stable (hydrolysis‐free) recirculation flows and a reproducible electrosynthesis. ^[36]

Since we were interested in the influence of the recirculation profiles on particle size, we performed experiments at three different droplet velocities: 5, 10 and 15 mm s⁻¹. The electrical current magnitude has been reported to have a small influence on particle size, compared to the duration of the electrical treatment. ^[23] Therefore, the residence time of each droplet (between the parallel electrodes) was kept constant at 10 s for all flow velocities, by changing the length of the parallel electrodes (5, 10 and 15 cm, respectively). The electrical current magnitude was adjusted so that the charge dose per droplet was consistent through all experiments. It is important to note that, in order to minimize any pressure‐driven artifact, only the length of the parallel electrodes was changed among experiments, while the overall channel length remained constant (15 cm, Figure 1C).

Previous studies with static electrochemical reactors required a complimentary procedure to prevent the formation of large aggregates. These procedures ranged from the addition of capping agents to the use of mechanical forces to detach particles from the electrodes surface.[ ⁸ , ¹⁷ , ²³ ] In our microfluidic setup, the effect of the hydrophobic carbon‐based membrane on the droplet boundary promotes an internal recirculation flow, composed of two main counter‐rotating circulations distributed along the axial centerline (Figure 1A).[ ³⁷ , ³⁸ ] Electrons reduce the gold ions into metallic gold at the membrane‐droplet interface, and are removed by the shear rate arising during droplet displacement, which limits their growth (Figure 1B).

We collected 30 μL aliquots of the nanoparticles‐containing droplets for all combinations of droplet velocity, charge dose, and precursor concentration (Figure S2). After droplet coalescence by a corona discharge process, each aliquot was analyzed by measuring the absorption spectra that originated from the surface plasmon resonance of the gold nanoparticles. ^[39] The peak of each absorption spectrum acquired through UV/Vis spectroscopy was correlated with the average nanoparticle size ^[39] (Figure 2A), and was verified using scanning electron microscopy (SEM) (Figures 2B–G). The average particle sizes estimated through UV/Vis absorption were consistent with those obtained by SEM. A maximum absorption peak at 553 nm relates to particles with a diameter around 80 nm (Figure 2A). A progressive decrease in the maximum absorption wavelength (from 540 to 526 nm in Figure 2A) corresponded to a decrease in particle size (from 60 to 30 nm in Figures 2B–G). In general, we obtained spherical nanoparticles with sizes ranging from 30 to 106 nm, depending on the conditions, and the SEM images revealed good mono‐dispersity (CV<16 %).

Figure 2.

A) UV/Vis absorbance spectra and B)–G) scanning electron microscopy images for nanoparticles grown under different experimental conditions. The maximum absorption correlates to the average nanoparticle size. Conditions: (spectrum 1/image B) 10 mM HAuCl₄, charge dose per droplet: 0.1 μC, droplet velocity: 15 mm s⁻¹, (2) 10 mM, 0.1 μC, 10 mm s⁻¹, (3) 5 mM, 3.0 nC, 15 mm s⁻¹, (4/C) 5 mM, 0.3 μC, 10 mm s⁻¹, (5/D) 1 mM, 2.5 μC, 15 mm s⁻¹, (6/E) 1 mM, 0.2 μC, 10 mm s⁻¹, (7/F) 1 mM, 0.3 μC, 5 mm s⁻¹, (8/G) 1 mM, 3.0 nC, 5 mm s⁻¹.

Particle size, as a function of droplet velocity and charge dose per droplet, is shown in Figure 3 for three concentrations of the metal precursor. In general, higher concentrations of gold (III) chloride yield larger particles, indicating that the gold reduction takes place preferably at the electrode surface (in the nuclei) rather than in the bulk of the droplet.[ ⁴⁰ , ⁴¹ , ⁴² ] Fluctuations in particle size were observed for charge doses below 1 μC per droplet, potentially indicating regions of higher growth along the electrode surface caused by the non‐uniform conductivity of the composite membranes. However, charge doses over 1 μC per droplet resulted in consistent particle sizes. Interestingly, particle size increased with droplet velocity, indicating that, as droplet velocity increased, the formed nuclei grew faster. This observation can be partially explained by two individual effects. First, the shear rate along the droplet boundary decreases with increasing droplet velocity for the studied conditions (8×10⁻⁴≤capillary number≤2×10⁻³), ^[37] becoming less likely for nuclei to detach from the electrode surface during droplet displacement. Second, faster droplet velocities result in faster recirculation flows inside the droplets, which improve the transport of unreduced gold (III) to the electrode surface, promoting nanoparticle growth. In this manner, AuNPs with sizes ranging from 30 to 106 nm could be synthetized by changing the precursor concentration and droplet velocity. Since these factors could be easily automated on the microfluidic setup, this study can lead to the fine‐tuning and screening of the optimal conditions for nanoparticle synthesis.

Figure 3.

Particle size as a function of charge dose per droplet and initial concentration of HauCl₄, for a droplet velocity of A) 5 mm s⁻¹, B) 10 mm s⁻¹ and C) 15 mm s⁻¹. Each point indicates the average particle size of >10 000 droplets. The shaded regions represent the 95 % confidence intervals for logarithmic models fitted to the individual points.

During electrochemical‐based synthesis, it is possible to introduce modifications in the spherical shape of the nanoparticles by changing the precursor concentration or adding pre‐defined seeds to the suspension.[ ²³ , ⁴³ , ⁴⁴ ] To demonstrate that the use of confined nanodroplet reactors can also be used to control such shape modifications, we added sodium hydroxide to the 1 mM HAuCl₄ solution (for a final pH of 7) and performed a new set of experiments. The hydroxide ions interact with the surface of the gold nanoparticles, modifying their synthesis reaction rate. ^[45] On the one hand, the resulting particle sizes were consistent with those obtained under similar conditions at a pH of 2. For example, at a droplet velocity of 5 mm s⁻¹, a decrease in size from about 60 to 30 nm was observed as the charge dose increased (Figure 4). However, instead of the spherical shapes observed at pH 2, mostly tetrahedral and prism‐like particles were observed at pH 7. It is important to remark that, in contrast to other studies, our system could produce tetrahedral‐shaped nanoparticles by changing the pH of the solution, without adding nanocrystals or other pre‐defined seeds. This observation highlights the use of confined droplet electrochemical reactors not only to control particle size based on their fluid dynamics, but also to control their shape by the addition of different ionic species.

Figure 4.

Scanning electron microscopy images of gold nanoparticles formed with a 1 mM HAuCl₄ precursor solution, titrated to pH 7 using sodium hydroxide. In contrast to the native pH (2), the addition of counter‐ions results in most particles having a non‐spherical shape (droplet velocity: 5 mm s⁻¹; charge dose per droplet: 3.0 nC (A), 1.5 μC (B) and 2.5 μC (C)).

In summary, we presented the electroformation of AuNP in nanoliter droplet reactors. The use of confined droplets (moving in rectangular microchannels) enables the continuous formation of nanoparticles, where their final size is mainly defined by the initial concentration of the metal precursor and droplet velocity. The morphology of the resulting nanoparticles can be modified by shifting the pH conditions. In this microfluidic setup, aggregation during particle formation is prevented by the internal recirculation flows and the electrostatic stability promoted by the low ionic strength of the solutions. ^[46] The size measurement using UV/Vis spectroscopy was generally performed within 24 hours after particle formation. The same samples were analyzed via SEM after 4 weeks, indicating a good stability after droplet collection. In conjunction with the proposed technique, droplet microfluidics offers a flexible framework for post‐processing the electroformed nanoparticles. For example, different chemical species can be pico‐injected to create multicore nanoparticles, and on‐chip, real‐time UV/Vis absorption spectroscopy ^[47] could be employed to screen individual droplets. Future efforts aim to interface cyclic voltammetry with the proposed microfluidic setup, to assess the reaction mechanism and growth dynamics. We believe that our approach is versatile and can be employed for the synthesis of other metal and semiconductor nanoparticles.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Saucedo-Espinosa M. A., Breitfeld M., Dittrich P. S., Angew. Chem. Int. Ed. 2023, 62, e202212459; Angew. Chem. 2023, 135, e202212459.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.