Surfactant Layers on Gold Nanorods

Conspectus

Gold nanorods (Au NRs) are an exceptionally promising tool in nanotechnology due to three key factors: (i) their strong interaction with electromagnetic radiation, stemming from their plasmonic nature, (ii) the ease with which the resonance frequency of their longitudinal plasmon mode can be tuned from the visible to the near-infrared region of the electromagnetic spectrum based on their aspect ratio, and (iii) their simple and cost-effective preparation through seed-mediated chemical growth. In this synthetic method, surfactants play a critical role in controlling the size, shape, and colloidal stability of Au NRs. For example, surfactants can stabilize specific crystallographic facets during the formation of Au NRs, leading to the formation of NRs with specific morphologies.

The process of surfactant adsorption onto the NR surface may result in various assemblies of surfactant molecules, such as spherical micelles, elongated micelles, or bilayers. Again, the assembly mode is critical toward determining the further availability of the Au NR surface to the surrounding medium. Despite its importance and a great deal of research effort, the interaction between Au NPs and surfactants remains insufficiently understood, because the assembly process is influenced by numerous factors, including the chemical nature of the surfactant, the surface morphology of Au NPs, and solution parameters. Therefore, gaining a more comprehensive understanding of these interactions is essential to unlock the full potential of the seed-mediated growth method and the applications of plasmonic NPs. A plethora of characterization techniques have been applied to reach such an understanding, but many open questions remain.

In this Account, we review the current knowledge on the interactions between surfactants and Au NRs. We briefly introduce the state-of-the-art methods for synthesizing Au NRs and highlight the crucial role of cationic surfactants during this process. The self-assembly and organization of surfactants on the Au NR surface is then discussed to better understand their role in seed-mediated growth. Subsequently, we provide examples and elucidate how chemical additives can be used to modulate micellar assemblies, in turn allowing for a finer control over the growth of Au NRs, including chiral NRs. Next, we review the main experimental characterization and computational modeling techniques that have been applied to shed light on the arrangement of surfactants on Au NRs and summarize the advantages and disadvantages for each technique. The Account ends with a “Conclusions and Outlook” section, outlining promising future research directions and developments that we consider are still required, mostly related to the application of electron microscopy in liquid and in 3D. Finally, we remark on the potential of exploiting machine learning techniques to predict synthetic routes for NPs with predefined structures and properties.

Key References

• González-Rubio G.; Mosquera J.; Kumar V.; Pedrazo-Tardajos A.; Llombart P.; Solís D. M.; Lobato I.; Noya E. G.; Guerrero-Martínez A.; Taboada J. M.; Obelleiro F.; MacDowell L. G.; Bals S.; Liz-Marzán L. M.. Micelle-Directed Chiral Seeded Growth on Anisotropic Gold Nanocrystals. Science 2020, 368. 1472–1477.¹ Chiral cosurfactants were used to induce the formation of Au NRs containing wrinkles with a helical arrangement.

• González-Rubio G.; Kumar V.; Llombart P.; Díaz-Núñez P.; Bladt E.; Altantzis T.; Bals S.; Peña-Rodríguez O.; Noya E. G.; MacDowell L. G.; Guerrero-Martínez A.; Liz-Marzán L. M.. Disconnecting Symmetry Breaking from Seeded Growth for the Reproducible Synthesis of High Quality Gold Nanorods. ACS Nano 2019, 13, 4424–4435.² Through the combination of CTAB and the cosurfactant 1-decanol, small Au NRs were obtained and subsequently used as anisotropic seeds to obtain homogeneous Au NRs with tunable aspect ratio and dimensions.

• Pérez-Juste J.; Pastoriza-Santos I.; Liz-Marzán L. M.; Mulvaney P.. Gold Nanorods: Synthesis, Characterization and Applications. Coord. Chem. Rev. 2005, 249, 1870–1901.³ A comprehensive review providing a detailed examination of the synthesis and properties of Au NRs, with a focus on the crucial role of surfactants to direct anisotropic growth.

1. Introduction

Several decades of intense research have resulted in a rather precise understanding of the behavior of gold at the nanoscale. For example, gold nanoparticles (Au NPs) display fascinating optical properties related to the tunability of localized surface plasmon resonances (LSPRs), as a function of Au NP size and shape.⁴ LSPRs feature exceptionally high absorption and scattering of a specific incident electromagnetic radiation and induce the enhancement of electric fields, local heating, and the efficient production of hot electrons, all of which can be also tailored through the morphology of the NPs. Among a multitude of available shapes, single-crystal gold nanorods (Au NRs) have been arguably the most widely studied variety.⁵ For this class of nanomaterials, a simple morphological parameter (length-to-diameter aspect ratio) can be adjusted to tune the (longitudinal) plasmon wavelength.³ Indeed, it has been reported that the longitudinal LSPR wavelength increases in a quasi-linear fashion, through the visible and into the infrared range, as the aspect ratio is increased. This tunability, together with reliable synthetic methods available, render Au NRs ideal candidates for a wide range of applications, including solar harvesting, sensing, and photothermal therapy, among others.⁶

Even though seedless Au NR growth has been reported, the colloidal synthesis of Au NRs essentially relies on so-called seed-mediated growth, which was pioneered by Murphy and El-Sayed.⁷ Subsequently, this method was optimized by many other groups, including us,⁸ to achieve narrow size and shape distributions, as well as high yield, for a wide range of sizes and aspect ratios. A key concept of seed-mediated growth is the spatial and temporal separation between nanocrystal nucleation and growth, which leads to greater reaction control and, in turn, more homogeneous NPs. From the wealth of acquired information, we conclude that, for the growth of colloidal anisotropic NRs to occur, molecular templates are required to drive anisotropic growth while maintaining the high-symmetry face-centered cubic gold lattice.

Although some flexibility may be allowed in terms of the colloidal stabilizer for seed preparation, the growth solution invariably contains gold (complex) ions, a mild reducing agent, such as ascorbic acid, and typically a quaternary ammonium surfactant, most frequently cetyltrimethylammonium bromide (CTAB). By using a weak reducing agent, Au³⁺ is reduced to Au⁺ in the growth solution, but further reduction into Au⁰ should only occur on the seed surface, to avoid unwanted secondary nucleation.

A crucial event herein is the symmetry breaking from quasi-spherical seeds into anisotropic single-crystal Au NRs. This step occurs during seeded growth and requires the presence of Ag⁺ ions, which additionally determine the nanorod aspect ratio. A variety of mechanisms have been proposed to explain symmetry breaking and anisotropic development, because seeded growth takes place in a complex mixture of metal salts, anions, and surfactants. Currently, the most widely accepted mechanisms are as follows:

• (i)Facet-specific capping by surfactants. Computational simulations indicate that CTA⁺ covers (100) and (110) facets more effectively, compared to (111). Therefore, the easier diffusion of gold atoms toward (111) facets would result in the preferential growth of Au NRs in this direction.⁹
• (ii)Stabilization of specific facets via underpotential deposition of Ag⁺ ions. The use of Ag⁺ as an essential additive suggests that the underpotential deposition of silver atoms takes place on distinct Au crystal facets. This phenomenon leads to symmetry breaking and creation of rod-shaped structures.¹⁰
• (iii)Facet-specific capping by halides. Halides have a strong binding affinity for metal surfaces. Various studies have demonstrated the efficacy of halides to direct growth into specific nanocrystal shapes. Halogens not only serve as face-specific capping agents, but can also modulate the redox potentials of metal ions and even affect the extent of Ag deposition on the surface of nanocrystals.¹⁰

2. The Surfactant Shell

There is general agreement regarding the crucial influence of the distribution and packing of surfactants on the growth and behavior of metallic (and other) NPs. Also known as “amphiphiles”, surfactant molecules comprise, in general, a hydrophobic tail and a hydrophilic head. The hydrophobic part usually consists of an aliphatic chain, whereas the polar head can vary according to its charge: nonionic, cationic, anionic, or zwitterionic. Because of their amphiphilic nature, surfactants readily adsorb at air–water or oil–water interfaces, stabilizing them and reducing the interfacial tension. When the surfactant concentration in aqueous solution reaches the so-called critical micelle concentration (CMC), amphiphilic molecules commence self-assembly into supramolecular aggregates, termed micelles. The hydrophobicity of the surfactant and various characteristics of the solutions (e.g., ionic strength) are parameters that determine the CMC. Additionally, the size and shape of micelles depend on a surfactant geometric factor referred to as the critical packing parameter (CPP). Importantly, the CPP can be affected by changes in solution conditions, including temperature, pH, and ionic strength, because the effective headgroup area of a specific surfactant can vary under different conditions.¹¹ As a result, the same surfactant can self-assemble into aggregates with various morphologies.¹² The most common surfactant aggregates in water are spherical micelles, cylindrical or worm-like micelles, bilayer vesicles, and lamellar phases (Figure 1).¹³

Figure 1.

Phase diagram for cetyltrimethylammonium bromide as a function of concentration and temperature in aqueous solution. Adapted with permission from ref (13). Copyright 1996 American Chemical Society.

As mentioned above, cationic surfactants are essential components during the preparation of Au NPs, because they can seemingly control aggregation and functionalization. The most common surfactants used with Au NRs are CTAB and cetyltrimethylammonium chloride (CTAC, a CTAB analogue with Cl^– counterions). Notwithstanding, more exotic surfactants have also been applied, such as cationic gemini surfactants.¹⁴ Furthermore, the binding energies of anions to Au are correlated to the anion polarizability, with Br^– exhibiting a higher affinity than Cl^–.¹⁵ This affinity difference enhances the adsorption strength of cationic surfactant micelles and the surfactant density on the Au surface in the presence of Br^–.¹⁶

Numerous studies have been performed to elucidate the organization of surfactant aggregates on the Au NR surface. This is however an intricate question that has not been fully clarified so far. Two kinds of surfactant aggregates have been postulated on Au NP surfaces, micelles and bilayers, the main difference being that a bilayer is a lamellar-type arrangement in which two molecular layers form a continuous, flat structure, whereas micelles are heterogeneous mixtures of surfactant assemblies spaced by water channels.^9,17 In the case of CTAB, the bilayer model has been widely reported, and a thickness around 32 Å has been experimentally determined (vide infra).¹⁸ Conversely, for CTAC discrete globular micellar aggregates have been proposed. However, despite these observations, it can be expected that the interpretation of the interaction between surfactants and Au NPs is not unequivocal. For example, for dissolved micellar aggregates, the solution conditions (temperature, concentration, etc.) also affect the organization of surfactant aggregates on Au NPs, eventually inducing unusual micellar morphologies.

The interaction between surfactants and Au NPs is also influenced by the surface morphology of the Au rods. For example, CTAB binds more strongly onto (100) and (110) than (111) facets, the latter being reportedly present on NR tips.⁹ Additionally, it is commonly accepted (mainly from favored tip-to-tip self-assembly observations), that the formation of bilayers is less favorable on highly curved surfaces, such as those found at tips with several facet junctions, thus leading to more exposed surfaces.¹⁹ These effects may thus be critical for the diffusion of gold ions toward the NP surface and ultimately for Au NR growth.

3. Influence of Additives

Chemical additives have been extensively studied as modulators of the structure of surfactant aggregates, in turn influencing properties such as flexibility or thermodynamic stability. Additives may affect surfactant aggregation properties by external interactions with the aggregates or being embedded therein. The most common examples of external interaction are the counterions of charged surfactants, which interact with the surfactant polar heads, thereby reducing electrostatic repulsions between them. For example, CTAC and CTAB only differ in the counterion, Cl^– vs Br^–, both anions having the same charge, but they have different CMC (0.96 mM for CTAB; 1.3 mM for CTAC), and therefore their organization on Au NPs may be significantly different.

Regarding the embedded mechanism, a surfactant can be combined with a second one, giving rise to binary mixtures, the less abundant surfactant being considered as an embedded additive that modifies the micellar aggregates. The same mechanism can be observed by adding molecules with a weaker amphiphilic nature, usually termed cosurfactants, which feature short to medium hydrophobic chains and small hydrophilic headgroups. The main difference between surfactants and cosurfactants is that the latter cannot form aggregates by themselves.

In the context of this Account, a milestone was the report by Murray and colleagues, who employed the embedded mechanism by using mixtures of CTAB and sodium oleate to improve the synthesis of monodisperse Au NRs (Figure 2A).²⁰ By using this binary mixture in the seeded growth method, Au NRs were obtained with essentially no shape impurities (less than 0.5% of the total number of NPs) and a striking dimensional tunability, with diameters ranging between 15 and 50 nm and LSPR wavelengths from 650 to 1150 nm. Importantly, unusually thick Au NRs (diameters larger than 30 nm), could also be synthesized in the presence of this surfactant mixture. The presence of sodium oleate induces the following effects that may be responsible for these results: (i) sodium oleate reduces the micelle charge; (ii) sodium oleate enhances the tendency toward elongated micelles;²¹ and (iii) the oleate double bond acts at as a soft reducing agent that transforms Au³⁺ into Au⁺. Subsequently, the same group showed that Au NRs can be efficiently prepared by using a mixture of CTAC (instead of CTAB) and sodium oleate, effectively reducing bromide concentration to a minimum.²²

Figure 2.

Three representative examples of the application of additives for the preparation of Au NRs. (A, B) Chemical structure of additives used in combination with CTAB, TEM images of the obtained Au NRs, and normalized extinction spectra showing the tunability of longitudinal LSPR frequency (scale bars: 100 nm). Colored lines in panel A, from purple to brown, represent increasing NR aspect ratio from 3.8 to 7.5. Similarly, colored lines in panel B indicate increasing aspect ratio, from 3.6 to 6.3. (C) Chemical structure of the chiral additive used in combination with CTAC, a HAADF-STEM image of the chiral Au NRs (scale bar: 200 nm), a selected orthoslice obtained from a tomography reconstruction (scale bar: 50 nm), and an anisotropy factor spectrum for chiral Au nanorods with increasing particle size: 165 × 73 nm (red), 210 × 112 nm (blue), and 270 × 175 nm (magenta). Panel A reproduced with permission from ref (20). Copyright 2013 American Chemical Society. Panel B adapted with permission from ref (2). Copyright 2019 American Chemical Society. Panel C reproduced with permission from ref (1). Copyright 2020 AAAS.

In a later development, our groups reported a significant improvement in the reproducibility of high quality Au NRs, by using CTAB in the presence of the cosurfactant 1-decanol (Figure 2B).² This combined surfactant/cosurfactant system allowed a separation of the symmetry breaking step from seeded growth, so that both processes could be optimized separately. Upon initial symmetry breaking into small Au NRs (21 nm long; 7.5 nm wide), these could be subsequently used as anisotropic seeds to obtain Au NRs of different aspect ratios and dimensions. Equally important for the performance of this synthetic method was the application of 1-decanol as cosurfactant for seeded growth. It has been reported that 1-decanol/CTAB mixtures in 0.20–0.30 ratio form stable bilayers in solution. Both experimental evidence (2D NMR) and computational studies indicated that the former mixture formed a highly ordered bilayer on the (250) lateral facets of Au NRs, inducing anisotropic growth on the Au NR seeds.²

Inspired by the former results, which clearly illustrate the dramatic impact of cosurfactants on Au NR synthesis, we subsequently proposed the use of chiral cosurfactants to induce the formation of dissymmetric plasmonic NRs. We explored the application of cosurfactants with axial chirality, 1,1′-bi(2-naphthol) (known as BINOL), to grow chiral gold layers over the surface of preformed Au NRs.¹ BINOL has been experimentally shown to induce the formation of chiral worm-like micelles in combination with CTAB/CTAC, in agreement with computational models (vide infra). However, BINOL-based micelles proved unable to induce chiral growth, likely due to insufficiently strong binding of the micelles onto the Au NR surface. On the contrary, when using 1,1′-binaphthyl-2,2′-diamine (BINAMINE, an analogue of BINOL in which hydroxyls are replaced by amines), sharp (2 to 3 nm wide) wrinkles with a helical arrangement over the seed NR surface were achieved, as confirmed by 3D high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) (Figure 2C). This successful result is likely due to the higher affinity between the BINAMINE amino groups in the micellar aggregates and the gold surface, which stabilize the highly energetic surface formed during chiral shell growth. Circular dichroism (CD) analysis of the chiral NPs revealed anisotropy factors up to 0.25, which are among the highest values reported for colloidal NPs. Furthermore, the position of the CD bands can be shifted from the visible to the near-infrared spectral region by increasing the thickness of the chiral shell.

4. Characterization of Surfactant Layers

Based on the above considerations, it is obvious that an accurate characterization of the structure and distribution of surfactant layers on Au NRs is crucial. We present in this section an overview thereof, highlighting recent progress in such characterization techniques. Based on their historical development, we propose a classification into two main categories: bulk and single-particle characterization techniques.

4.1. Bulk Characterization Techniques

Bulk methods include scattering and spectroscopy techniques, which can accurately determine the average thickness and structure of ligands surrounding the Au NRs in an ensemble manner. El-Sayed’s group provided the first hints regarding the adsorption of CTAB on Au NRs, by combining thermogravimetric analysis (TGA) and Fourier-transform infrared spectroscopy (FTIR).¹⁷ The authors concluded that CTAB is packed into a hexagonal arrangement, forming a bilayer in which the inner layer is bound to the gold surface via the surfactant ammonium headgroups. It should be noted, however, that these experiments may underestimate the surface ligand density due to loss of surfactant molecules during purification, hence leading to inaccurate quantification. In addition, a direct characterization of morphological characteristics such as thickness, density, patterning, or interdigitation degree of the surface ligands could not be provided by either TGA or FTIR. Similarly, mass spectrometry (MS) and zeta potential measurements have also been shown to yield information on the composition and surface charge of ligand layers. These techniques can e.g. be used to monitor ligand exchange of CTAB for thiolated PEG.²³

In a different direction, some of us used a combination of surface-enhanced Raman scattering (SERS) and X-ray photoelectron spectroscopy (XPS) to demonstrate that the replacement of bromide for iodide ions on the surface of Au NRs was key toward the formation of dumbbell-like NRs.²⁴

The above-mentioned techniques are commonly used for dry samples, in which ligand layers are collapsed and dehydrated. Capillary forces acting during sample preparation inevitably alter the thickness and surface distribution of ligands, thus affecting the quantification of surface coverage and ligand density. To avoid such drying effects, small-angle scattering of X-rays and neutrons (SAXS and SANS), as well as solution nuclear magnetic resonance (NMR), have been applied to study the structure of ligand layers (e.g., thickness) on Au NRs in their native environment.^18,25,26 The measured surfactant layer thickness (32 Å) was found to agree with a partially interdigitated bilayer model (on the basis of reported measurements on flat surfaces).²⁷ The same techniques were applied to study the difference between CTAB and CTAC assemblies on Au NRs, concluding that CTAC organizes into a patchy shell, whereas CTAB forms a closed/collapsed bilayer.¹⁶ However, the actual internal stacking of the ligands could not be unraveled due to the overall weak scattering signal arising from the low concentration of the Au NR dispersion used in the study.^18,25

Compared to other bulk techniques, NMR does enable the determination of ligand conformation at the molecular level.²⁸ In a recent study, Murphy’s group investigated the conformation of ligands grafted on Au NRs with different aspect ratios and Au spheres with variable diameters. It should be noted, however, that most other NMR-based studies of Au NR surface chemistry have focused on spherical NPs with diameters below 6 nm because line broadening in larger NPs hinders reliable peak assignments and integration during NMR analysis. Moreover, the low sensitivity of NMR requires highly concentrated samples, which may not always be possible.²⁹

In summary, we stress that the ensemble nature of bulk characterization techniques leads to average information on the structure and composition of surfactant layers. To decipher structural differences, such as ligand distribution on tips versus lateral facets of Au NRs, a more direct single-particle level characterization toolbox should complement bulk characterization.

4.2. Single-Particle Characterization Techniques

One of the first local measurement techniques, applied to decipher the topography of molecular assemblies–of suitable height and conductivity–was scanning tunneling microscopy (STM). By collecting the tunneling current between a sharp metallic tip and the sample, mixed-ligand coatings could be distinguished on both Au nanospheres and Au NRs.^30,31 Stellacci’s group investigated Au NRs stabilized with a binary mixture of ligands, exhibiting stripe patterns orthogonal to the rod axis, in agreement with computer simulations.³² Yet, a sufficiently detailed analysis of standard surfactant ligands by STM remains challenging. Successful measurements require clean samples, free of isolated ligands, which is not always trivial to manage. Signal disturbances induced by sample movement may also compromise spatial resolution, potentially leading to biased structural analysis. Similarly, atomic force microscopy (AFM) measurements in peak force tapping mode, in combination with scanning electron microscopy (SEM), have enabled direct visualization of a patchy ligand structure with cysteamine on the tips and mPEG-SH on the side of the Au NRs.³³

On the other hand, (scanning) transmission electron microscopy ((S)TEM) enables one to investigate nanostructures down to the atomic scale. In contrast to bulk characterization techniques, (S)TEM can potentially visualize the morphology, thickness, and distribution of ligand layers at the single-particle level. However, in practice this is highly challenging due to the low contrast and sensitivity of organic molecules toward the electron beam.³⁴ To overcome the lack of contrast in conventional imaging modes, one can use TEM spectroscopic techniques such as electron energy loss spectroscopy (EELS) and energy filtered TEM (EFTEM).^35,36 Huang and co-workers investigated the distribution of CTAB and (16-mercaptohexadecyl)trimethylammonium bromide (MTAB) layers on Au NRs, using STEM-EELS.³⁵ By quantification of the carbon signal, the MTAB distribution was found to be homogeneous around the nanorods, whereas a lower density was observed for CTAB at the tips compared to the sides. A shell thickness of approximately 24 Å was extracted for CTAB, which is smaller than the values reported by ensemble techniques.¹⁸ An asymmetric CTAB distribution surrounding Au nanorods was also reported by Kotov and Norris, using EFTEM together with secondary electron imaging.³⁷ A noncentrosymmetric distribution of CTAB was determined in this case for NR tips, leading to an electrostatic potential gradient at the surface.

A significant drawback related to TEM spectroscopic techniques is the need for relatively long acquisition times and high electron dose. Especially during imaging of beam-sensitive ligands, the measurements might affect the ligand structure. To improve the resistance of the ligands to the electron beam, Huang and colleagues used graphene as the support for TEM.³⁵ In addition to graphene’s high thermal and electrical conductivity, these supports provide minimal background, leading to improved contrast for ligands. Graphene grids therefore also allow direct visualization of ligands at the surface of Au NPs, albeit in a dried state.³⁴

5. Computational Modeling

Several attempts have been made to use theoretical methods to analyze the behavior of surfactant molecules adsorbed onto the surface of Au NRs. Such studies aimed to help understand the effects of different surfactant molecules, their concentrations, and other chemical moieties, such as halide counterions or Ag⁺ ions, on the growth mechanisms of the nanorods as well as the packing of the resulting surfactant layers. The methods used in these studies have been based principally on either molecular dynamics (MD) simulations or density functional theory (DFT) calculations.

In MD simulations, the equations of motion for each atom or molecule are solved numerically, so that the positions and velocities of the atoms or molecules can be determined as a function of time. In the case of surfactants interacting with Au NRs, MD simulations can provide insight into thermodynamic, kinetic, and structural aspects. In an early study, Meena and Sulpizi used MD simulations to study the role of halide ions (chloride and bromide) in controlling the shape of the NPs and the mechanism of anisotropic growth.⁹ The study showed that the presence of halide ions leads to the formation of elongated surfactant micelles on Au NRs, leaving water channels containing bromide ions that provide a path for the diffusion of gold reactants toward the NR surface (Figure 3A). This concept was against the common understanding that CTAB forms a regular bilayer on Au NRs, but already a higher packing density was observed for (110) and (100) compared to (111) surfaces, potentially explaining anisotropic growth. In a subsequent contribution, the same authors investigated the effect of surfactant concentration and chain length on the adsorption behavior.³⁸ The study shows that, under certain conditions, the adsorption of cationic surfactants on Au NRs may indeed lead to the formation of a stable monolayer on the nanorod surface. The main conclusion coming out of this study was the preferential adsorption of surfactant at different crystalline facets, which was proposed to explain the symmetry breaking stage in seeded growth and the stabilization of higher index facets, compared to (111), which would be less protected and thus more reactive.

Figure 3.

(A) Snapshots from MD simulations of CTA⁺ (blue sticks) at and above the CMC on a Au(100) surface. Bromide ions and water molecules are represented by green spheres and red sticks, respectively. (B) MD simulation of (R)-BINOL/CTAC adsorbed onto a gold nanorod. (C) DFT-based simulation of the incorporation and growth of Au onto a pre-existing structure for different FCC surfaces. Step 1: surface adsorption of CTAB–AgBr complex. Step 2: release of Br. Step 3: addition of Au atoms to the lattice. Surface Au atoms are shown in yellow, bulk Au in orange, Ag in gray, Br in brown, N in blue, CTA⁺ as sticks. Panel A reproduced with permission from ref (38). Copyright 2016 Wiley-VCH. B) Reproduced with permission from ref (1). Copyright 2020 American Association for the Advancement of Science. C) Reproduced with permission from ref (40). Copyright 2014 American Chemical Society.

A similar conclusion was reached by some of us, also using MD simulations to understand the different effects of femtosecond laser irradiation on Au NRs, at different CTAB concentrations.³⁹ Around the CMC, femtosecond laser irradiation could induce focusing of the Au NR aspect ratio because heat could be efficiently released to the solvent through water channels, whereas polydispersity would be enhanced at higher concentrations because of unwanted fast reshaping due to shielding by continuous CTAB bilayers. In a later work, our group demonstrated the effect of cosurfactants in a binary (surfactant/cosurfactant) packing, on the uniform symmetry breaking and seeded growth of Au NRs. Molecular dynamics simulations revealed that the presence of 1-decanol allowed dense packing on lateral (250) facets, thus forming bilayers and hindering growth on those facets.² On the contrary, the bilayer would get destabilized on (111) facets present on Au NR tips. Therefore, this disposition of decanol/CTAB aggregates provokes Au precursors from solution to readily reach the Au NR tips, inducing the anisotropic growth of Au seeds. The idea of cosurfactants affecting the distribution of ligands on Au NRs was more recently employed to explain the formation of quasihelical wrinkles on preformed Au NRs, in the presence of CTAC and chiral cosurfactants such as BINOL or BINAMINE (Figure 3B).¹ MD simulations demonstrated in this case the formation of quasihelical micelles that could wrap the Au NRs and serve as templates for chiral overgrowth.

On the other hand, DFT calculations are based on quantum mechanics and can be applied to predict the adsorption energies between ligand molecules and metal surfaces. Of particular note here is the work by López and colleagues, who used binding energies to understand the role of different halide counterions and to reproduce the energetic variations during various steps of the seeded growth of Au NRs, in the presence of Ag⁺ (Figure 3C).⁴⁰ This study relied on previous work by Spalla’s group,⁴¹ who claimed that it is a complex between Ag⁺ ions and CTAB, which is actually deposited on the Au surface, playing a crucial role in the anisotropic growth.

6. Conclusions and Outlook

Motivated by the high scientific and technological relevance of gold nanorods, we have focused this Account on the issue of understanding (and manipulating) surfactant layers on these NPs. From the above discussion, it is apparent that we do not know yet whether CTAB/CTAC molecules organize into a bilayer structure or rather form micelles on the Au NR surface. Nevertheless, it is generally accepted that the formation of micellar structures is favored at low concentrations of surfactant and when Cl^– is utilized as the counterion.^16,39 Therefore, we still need a technique that allows an unambiguous visualization of surface ligands with sufficient resolution and without alterations from their natural environment. We propose that the major recent progress in electron microscopy techniques will eventually enable achieving this goal. However, we stress again that most TEM studies of surface ligands have been carried out on previously dried samples and under conventional TEM conditions (i.e., in high vacuum), thus unavoidably resulting in structural changes of the ligand layers due to capillary forces. To study the structure of ligands in their native environment (in a colloidal dispersion), it is essential to perform TEM measurements in the liquid phase. Promising results were recently reported by Alivisatos and co-workers,⁴³ in which the distribution of ligands was indirectly assessed by monitoring the reshaping of Au NRs during oxidative etching inside a graphene liquid cell (GLC)⁴² by encapsulating Au NRs dispersion and oxidative solution between graphene monolayers.⁴³ However, we are not aware of any demonstration of directly deciphering the structure of ligand layers in solution, using TEM.

Another limitation that must be overcome is related to TEM data sets conventionally corresponding to 2-dimensional (2D) projections of 3-dimensional (3D) objects, which in some cases (such as the determination of ligand distribution around an entire NP) can be misleading. Electron tomography then becomes the technique of choice, because it allows characterizing the structure of NPs in 3D, even with atomic resolution.⁴⁴ Visualizing the interface between an Au NP and the surfactant layer by electron tomography is, however, extremely challenging because it requires the acquisition of multiple images along different viewing directions, thereby demanding a high electron dose. As indicated above, the ligand structure may be altered during the measurements because they are highly sensitive to electron beams.

To retrieve the original 3D ligand structure, further progress is clearly demanded. A promising route involves combining the concept of fast electron tomography,⁴⁵ by which the image acquisition time can be significantly reduced, with liquid phase TEM. However, commercial liquid phase TEM holders with Si₃N₄ chips suffer from a limited tilting range and a low signal-to-noise ratio (SNR), thus hampering reliable 3D structure analysis of ligands in a liquid environment. Such a challenge can be tackled by using GLCs, which allow a sufficient tilting range and improved SNR because of the much thinner layer of liquid in the cell, compared to commercial holders. The risk of potential beam damage can be further minimized by exploiting advanced direct detectors and techniques that enable low-dose imaging.⁴⁶

Another challenge would be the direct observation of the dynamics of the adsorption and desorption of ligands (e.g., during ligand exchange) on Au NRs, at the single particle level and in 3D. Given the fast diffusion of ligands (or micelles) in a liquid, the use of advanced direct electron detectors or even ultrafast electron microscopy techniques should be explored in the future.^46,47 Advances in data treatment during electron tomography experiments should also enable further progress. For example, advanced reconstruction methods are needed to process images with low SNR. We envisage that machine learning (ML)-guided reconstruction methodologies, currently used for medical computed tomography, can be adapted to the development of algorithms for TEM.⁴⁸ Through the implementation of these improvements, liquid-phase fast electron tomography reconstructions will enable the next breakthrough in the characterization of surface ligands.

Although Au NRs can currently be produced in a controlled manner, synthesis optimization has been traditionally based on a trial-and-error approach. A broad range of synthesis parameters must be tested and screened during a large number of experiments before reliable recipes can be established. With the rapid development of characterization tools, we foresee great opportunities for ML techniques to enable more efficient and target-oriented synthetic strategies, not only for Au NRs but also for a broad range of NPs. Despite its increasing popularity, ML-guided design of inorganic NPs has just emerged and only a handful of publications are available on ML-mediated Au NP synthesis.^49,50 We foresee that improved knowledge from a complete characterization of the NPs and their ligand shells (in 3D and in solution) can be used to train ML routines, which in turn will predict synthetic routes for NPs with the desired composition, surface chemistry, and properties.

In summary, we have described a wide range of characterization techniques that have been used to investigate the structure and distribution of ligands at the surface of Au NRs, each having both advantages and drawbacks. Therefore, we propose that a combination of complementary bulk and local (single-particle level and below) characterization techniques would be a suitable way forward. For instance, we envision that one could derive the overall thickness and local structural details of ligand layers in their native environment, through a combination of SANS and TEM measurements (both in liquid). Moreover, we remark that computational modeling at the atomic and single-molecule scales will become indispensable to validate experimental observations when the structure of ligands cannot be unambiguously deciphered by experiments.

Biographies

Jesús Mosquera received his Ph.D. in Chemistry from University of Santiago de Compostela in 2014. He is now working as an Assistant Professor at the Universidade da Coruña.

Da Wang received his Ph.D. in Physics from Utrecht University in 2018. From 2018 to 2022, he worked as a postdoctoral researcher in the Electron Microscopy for Materials Science (EMAT) at the University of Antwerp. His research is focused on colloidal self-assembly and advanced electron microscopy techniques. He is now working as a tenure track group leader at the South China Normal University.

Sara Bals received her Ph.D. from the University of Antwerp (Belgium) in 2003. She is a Full Professor and the spokesperson of the Electron Microscopy for Materials Science (EMAT) research facility at the University of Antwerp.

Luis M. Liz-Marzán has a Ph.D. from the University of Santiago de Compostela (1992). He is currently an Ikerbasque Research Professor and Scientific Director of CIC biomaGUNE, and Physical Chemistry Professor at Cinbio, Universidade de Vigo.

Author Contributions

^∇ J.M. and D.W. contributed equally. CRediT: Jesús Mosquera funding acquisition (equal), methodology (equal), writing-original draft (equal), writing-review & editing (equal); Da Wang methodology (equal), writing-original draft (equal), writing-review & editing (equal); Sara Bals conceptualization (equal), funding acquisition (equal), methodology (equal), supervision (supporting), writing-original draft (equal), writing-review & editing (equal); Luis M. Liz-Marzán conceptualization (lead), funding acquisition (equal), methodology (equal), supervision (lead), writing-original draft (equal), writing-review & editing (equal).

The authors declare no competing financial interest.

Special Issue

Published as part of the Accounts of Chemical Research special issue “Ligand and Surface Chemistry of Nanoparticles”.