SECOND CYCLE, 30 CREDITS STOCKHOLM, SWEDEN 2020
Colloidal self-assembly of
anisotropic gold nanoparticles
Master Thesis Report
Samuel Emilsson
KTH ROYAL INSTITUTE OF TECHNOLOGY
Samuel Emilsson (samemi@kth.com) Molecular Science and Engineering KTH Royal Institute of Technology
Place for Project
Lyon, France
Examiner
Professor Patrick Norman
Division of Theoretical Chemistry and Biology KTH Royal Institute of Technology
Supervisor
Dr. Anthony Désert Professor Stéphane Parola
Functional Materials and Photonics
ENS de Lyon
Abstract
The colloidal self-assembly of plasmonic gold nanoparticles (AuNPs) is of interest to utilize the plasmonic coupling effects that arise between nanoparticles. The enhanced properties of anisotropic AuNPs make them particularly attractive in self-assemblies. Herein, a literature study into the different strategies used to obtain self-assemblies of AuNPs using molecular linkers is presented.
The use of nanospheres (AuNS) and nanorods (AuNRs) were mainly reviewed. Thereafter, two different nanobipyramids (AuBPs) were investigated for use in self-assemblies. The concentration of cetyltrimethylammonium bromide (CTAB), which coats the AuNP surface, was manipulated to study the stability of the AuNPs. A stable, meta-stable and non-stable region were identified for the nanoparticles.
At low CTAB levels, the AuNPs preferentially assemble end-to-end. The addition of L-cysteine to stable AuNP dispersion induced end-to-end assembly, showing promise as a molecular linker for AuBPs. The addition of excess CTAB stabilized the assemblies over time. The kinetic behaviour of the two AuBPs differed, suggesting the effect of the AuNP shape on the self-assembly kinetics. This study provides a starting point for the development of a robust self-assembly strategy for anisotropic AuNPs by using L-cysteine as a molecular linker.
Keywords
Self-assembly, Gold nanoparticle, Anisotropic, Nanobipyramid, Surface plasmon resonance
Sammanfattning
Den kolloidala självsammansättningen av ytplasmoniska guld nanopartiklar (AuNPs) är av intresse för att utnyttja de plasmoniska kopplingseffekterna som uppstår mellan nanopartiklar. De fördelaktiga egenskaperna hos anisotropa AuNP gör dem särskilt intressanta för självsammansättningar. En litteraturstudie har gjorts på de olika strategier som används för att erhålla självsammansättningar av AuNPs med hjälp av molekylära länkar. Användningen av nanosfärer (AuNS) och nanostavar (AuNRs) i självsammansättningar undesöktes huvudsakligen. Därefter undersöktes två olika nanobipyramider (AuBPs) för användning i självsammansättningar. Koncentrationen av cetyltrimetylammonium bromid (CTAB), som täcker AuNP-ytan, manipulerades för att undersöka AuNPs stabilitet. En stabil, meta- stabil och instabil region identifierades för nanopartiklarna. Vid låga CTAB-nivåer sammansätts AuNPs ände-mot-ände. Tillsatsen av L-cystein till stabila AuNP dispersioner inducerade sammansättningar ände-mot-ände, vilket visar L-cysteins potential som en molekylär länk för AuBPs. Tillsatsen av en stor mängd CTAB stabiliserade självsammansättningarna för en längre tid. Det kinetiska beteendet hos de två AuBPs skilde sig, vilket tyder på effekten av AuNP-formen på den självsammansättningskinetiken.
Denna studie erbjuder en startpunkt för utvecklingen av en robust självsammansättningstrategi för anisotropa AuNPs genom att använda L-cystein som en molekylär länk.
Nyckelord
Självsammansättningar, Guldnanopartiklar, Anisotrop, Nanobipyramider, Ytplasmonresonans
Acknowledgements
I want to thank Anthony Désert for the supervision, advise, and valuable feedback provided throughout
the project. I also want to thank Antonio Carone whom I have collaborated closely with. Thank you
for all the help and guidance in the lab. Further, I want to thank Stéphane Parola for welcoming me to
the lab at ENS de Lyon and for also advising me throughout my work. Thank you also to the rest of the
staff and researchers as ENS de Lyon who have made me feel welcome. At KTH, I want to thank Patrick
Norman and Mikael Gullstrand for the feedback provided.
Abbreviations
SPR Surface plasmon resonance SPB Surface plasmon band
L-SPR Longitudinal surface plasmon resonance T-SPR Transverse surface plasmon resonance AuNPs Gold nanoparticles
AuNS Gold nanospheres AuNRs Gold nanorods AuBPs Gold bipyramids AR Aspect ratio
SERS Surface enhanced Raman scattering CTAB Cetyltrimethylammonium bromide CD Circular dichroism
TEM Transmission electron microscopy
Contents
1 Introduction 1
2 Theoretical Background 2
2.1 Nanotechnology . . . . 2
2.2 Plasmonics . . . . 3
2.3 Anisotropic gold nanoparticles . . . . 4
2.4 Synthesis of gold nanoparticles . . . . 5
2.5 Self-assembly of nanoparticles . . . . 7
3 Literature Report: Self-Assembly of gold nanoparticles 9 3.1 Property enhancements upon assembly . . . . 9
3.1.1 SPR band shift . . . . 9
3.1.2 Hot spot formation . . . . 10
3.2 Self-assembly of AuNPs . . . . 13
3.2.1 Amino acids as molecular linkers . . . . 13
3.3 Assembly of gold nanospheres . . . . 14
3.3.1 Stability of AuNS . . . . 14
3.3.2 Controlling assembly . . . . 16
3.4 Assembly of gold nanorods . . . . 19
3.4.1 Stability of AuNRs . . . . 19
3.4.2 Anisotropic coupling . . . . 21
3.4.3 Controlling assembly . . . . 23
3.4.4 Assembly growth kinetics . . . . 25
3.4.5 Stabilizing assemblies . . . . 27
3.5 Assembly of Nanobipyramids . . . . 29
3.5.1 Amino acid-assisted assembly . . . . 29
3.5.2 Alternative assembly methods . . . . 30
3.5.3 Previous work in research group . . . . 31
3.6 Future Outlook . . . . 32
3.6.1 Purpose of the study . . . . 32
4 Method and methodology 33
4.1 Materials . . . . 33
4.2 Synthesis of Gold nanobipyramids . . . . 33
4.2.1 Seed synthesis . . . . 33
4.2.2 Overgrown seeds . . . . 34
4.2.3 Modified synthesis of overgrown seeds . . . . 35
4.2.4 High concentration synthesis of AuBPs . . . . 35
4.3 CTAB stability test . . . . 35
4.4 Cysteine-mediated self-assembly . . . . 36
4.5 UV-vis-NIR spectroscopy . . . . 37
5 Results & discussion 38 5.1 Preparation of Gold nanobipyramids . . . . 38
5.1.1 Characterization . . . . 38
5.1.2 Concentration determination . . . . 40
5.2 Arginine-mediated self-assembly . . . . 41
5.3 Stability testing . . . . 42
5.3.1 Overgrown seeds . . . . 42
5.3.2 Gold nanobipyramids . . . . 43
5.3.3 System comparison . . . . 44
5.3.4 Low-CTAB mediated self-assembly . . . . 44
5.4 Cysteine-mediated self-assembly . . . . 45
5.4.1 Overgrown seeds . . . . 45
5.4.2 Gold nanobipyramids . . . . 47
6 Conclusions 51 6.1 Future Outlook . . . . 51
References 52
Introduction
Nanotechnology has emerged as one of the most interesting new research fields of the past decades.
The strive for engineering new structures and products with atomic precision has been a major motivation. However, it has also led to the discovery the unique properties that exist at the nanoscale.
Several different materials and objects have thereafter been developed to enhance these properties.
Nanotechnologies have also been developed for a variety of applications including medicine, energy, catalysis, biotechnology, and electronics. [1]
A nanoobject which has been heavily researched are gold nanoparticles. Gold nanoparticles (AuNPs) have a number of interesting features which can be used in a variety of applications. Firstly, metal nanoparticles such as gold possess a unique optical property known as the surface plasmon resonance (SPR). This allows the metal particles to absorb and scatter light at distinct bands in the Vis-NIR range. Further, by changing the shape and size of the nanoparticle, this band can be fine-tuned.
AuNPs have the advantage of being stable and highly tunable, making them attractive for plasmonic applications. Thereby, several different AuNPs have been developed, including nanospheres (AuNS), nanorods (AuNRs) and nanobipyramids (AuBPs). AuNRs and AuBPs are particularly interesting thanks to their anisotropy which leads to two surface plasmon bands which are also more tunable. Further investigation of AuNPs has also shown that when near to each other, plasmon coupling effects arise.
Therefore, efforts to control self-assembly of AuNPs has been investigated to further enhance these effects.
In this report, the colloidal self-assembly of AuNPs will be investigated. This is done in two parts. First,
an extensive literature study will review the different strategies that exist for using molecular linkers
to create self-assemblies of different AuNP shapes. The second part involves the experimental efforts
of obtaining self-assembly of two different AuBP shapes. This will be done by first mapping out the
dependence of the CTAB concentration on the particle stability. The use of L-cysteine as a molecular
linker will then be investigated by changing the L-cysteine concentration and AuNP. The assembly
process will be monitored through changes in a Vis-NIR spectrum over time. Due to the COVID-19
pandemic, the experimental study was limited.
Theoretical Background
2.1 Nanotechnology
Over the past decades, huge technological leaps have been made in fields ranging from physics and electronics to biotechnology and chemistry. Much of this development has been achieved thanks to nanotechnology [2]. An overarching motivation behind nanotechnology is controlling or enhancing systems at the nanoscale to promote wanted properties at larger scales. [1] One definition of nanotechnology is “engineering with atomic precision”. This assumes a level of understanding and control over the nanoscale, which usually considered to include the range from 1-100 nm.
However, nanotechnology was around long before understanding of the nanoscale was achieved. An early account of nanotechnology is the “Lycurgus Cup” crafted in Ancient Rome, containing gold nanoparticles which could change color based on the illumination source. It was not until it was analyzed in modern times that it was understood that nanoparticles were the reason for the phenomena.
Indeed, the development of analysis methods have therefore been an important step in the expansion of nanotechnology. [2]
Nanoobjects are objects which are limited in one or more external dimensions to the nanoscale and their
development has received a lot of attention. For instance, 2D materials such as graphene have huge
potential and unique characteristics. Another such nanoobject are nanoparticles. [1] Nanoparticles
are interesting due to the unique properties that arises when particles are confined to the nanometer
range. A central property is “quantum confinement” which implies that nanoparticles have spectral
properties which lie between that of individual atoms and the bulk material. Another property unique
for metal nanoparticles is the surface plasmon resonance, which this report will focus on. Since these
unique properties are highly size- and shape-dependent, controlling size and shape during synthesis has
become a major goal in nanoparticle development. [2]
2.2 Plasmonics
Within the field of nanoscience, a rapidly growing interest concerns plasmonics. The use of plasmonic nanostructures has shown great potential in a number of fields such as medicine [3], biosensing, photocatalysis [4] and energy storage.
The surface plasmon resonance band is a strong absorption band observed in the Vis-NIR-region for metallic nanoparticles. It arises from the collective oscillation of conduction band electrons induced by an external electric field, known as the surface plasmon resonance (SPR), as seen in 2.1b. For nanoparticles smaller than the incident wavelength of light, the electrons move coherently with the external electric field, creating a SPR that is highly localized. Due to the confinement of the surface plasmons, the localized electric field is more intense than the external electric field and decays rapidly as seen in 2.1a. [5] The confinement of this intense electric field is used in several spectroscopic methods to enhance its signal. [6] For example, enhancements of surface-enhanced Raman spectroscopy (SERS) with several orders of magnitude has been achieved with plasmonic nanoparticles [7].
(a) (b)
Figure 2.1: (a) Schematic of the localized surface plasmon resonance [8] (b) Schematic of electronic cloud displacements in a nanoparticle [5]
Fine-tuning this SPR band through the controlled synthesis of metal nanoparticles has been a major challenge in the development of SPR-active materials. This has led to the development of several strategies for the synthesis of metal nanoparticles of different shapes. Through this, greater understanding of the dependence of the composition, size and geometry of the nanoparticles on the SPR has been achieved [9]. The goal in the development of nanoparticles for this function is a highly pure, monodisperse nanoparticle with a well-defined SPR in terms of wavelength range, intensity and bandwidth. [8], [10]
The noble metals silver and gold have been used within plasmonics due to their intense plasmonic signal
[10]. Moreover, both gold and silver nanoparticles have also been synthesized with low polydispersity,
high chemical inertness, bio-compatibilty and easy functionalization. [11] Gold, however, shows
the highest stability. In this report, focus will be put on the properties and synthesis of gold
nanoparticles.
2.3 Anisotropic gold nanoparticles
A variety of different AuNPs have been successfully synthesized, showing a broad range of SPR bands (as shown in 2.2a). Amongst these, gold nanospheres (AuNS) and nanorods (AuNR) are the most researched. Although the size of symmetrical AuNS can be tuned, the size-dependent SPR shift is limited. Gold nanorods, however, are significantly more tunable with regards to SPR shifts due to their anisotropic nature. Since AuNRs have the shape of elongated cylinders, they contain two SPR modes:
the longitudinal and transverse bands (see 3.10). The transverse SPR band (T-SPR) is relatively low in intensity and only shifts slightly with size. The longitudinal SPR band (L-SPR) is by contrast intense and highly dependent on the length-to-width ratio of the nanorod. [8], [12]
This effect can be generalized for all anisotropic gold nanoparticles. The L-SPR is dependent on the aspect ratio of the particle. The aspect ratio is defined as the length-to-width ratio of the nanoparticle (see 3.10). Thereby, manipulating the geometric length of the nanoparticles leads to changes of the plasmonic band. For instance, the SPR peak can be tuned from 600-1300 nm for AuNRs compared to just 520-650 nm for AuNS [13].
Another more recently discovered anisotropic AuNP are gold nanobipyramids (AuBPs). AuBPs were first found as a byproduct to nanorods but containing sharp end tips. These tips showed a significant localized enhancement of the electric field and a narrower peak than conventional nanorods. [10], [14] It has also been synthesized highly monodisperse, which is a current limitation of gold nanorods.
Moreover, the SPR band has been successfully shifted to the NIR region. Since this region is biologically transparent, it is attractive for use in biological systems. [10] AuBPs have shown enhanced performances in sensing [7], [15] and medicine [3].
(a) (b)
Figure 2.2: a) Examples of gold metal nanoparticle shapes seen through transmission electron microscopy. (a) Nanospheres, (b) nanohexahedrons (i.e. cubes), (c) nanotriangles, (d) nanorods, (e) nanostars, (f) UV-visible spectra showing the optical properties of the different nanoparticle shapes [8];
b) Left: Schematic of anisotropic AuNP and its aspect ratio; Right: Schematic spectra of anisotropic
AuNP with transverse and longitudinal SPR bands.
2.4 Synthesis of gold nanoparticles
Two main strategies exist to produce nanoparticles: a top-down or bottom-up approach. The top-down approach involves taking the bulk material and fragmenting it by ie. crushing or grinding. Crushing and grinding are universal and cheap top-down methods. However, they usually evoke little control and leads to defects and polydispersity of the final material. [16] Development of highly precise top-down methods such as lithography have greatly improved the control of the final material. However, one must usually sacrifice either precision, price or time with these methods [17]. The bottom-up approach relies on going from an atomically dispersed phase to a solid condensed nanoparticle. This requires careful balance of nucleation and growth to obtain uniform particles at a given size. In this report, bottom-up approaches will be the main focus. [16]
Gold nanoparticles have been synthesized for 2000 years, although until recent decades only with limited control. [18] To increase the control, strategies such as photochemical synthesis [19] and electrochemical synthesis have been explored. [20] Successful shape-control has also been managed through post-synthetic etching, by using precise laser pulses [21]. The most explored synthetic method is however wet chemical synthesis of AuNPs. This process generally involves using a reducing agent in the presence of a capping agent (such as a surfactant) to reduce Au(III) (often HAuCl4) to grow AuNPs. As with many other synthetic methods, wet chemical synthesis was initially limited in control. [22]
A breakthrough in controlled synthesis of nanoparticles was achieved with the discovery of the seed- mediated growth method which provided access to a unique variety of shapes by building on the chemical synthesis protocol. [20] The principle behind seed-mediated growth is the synthesis of a small gold nanoparticle seeds followed by dispersion in a growth solution, containing Au(III), a weaker reducing agent and a templating capping agent. In this way, the nucleation step and growth are separated, allowing for more control of the critical nucleation step. The size and shape of the nanoparticle can then be controlled based on the concentration and size of the seeds, the concentration of gold salt and capping agents. [8], [18]
Figure 2.3: Morphology dependence of gold nanoparticles grown from either single crystal (d) or multiply twinned (e) seeds, in the presence (a–c) and absence (f–h) of silver nitrate [18]
The seed-mediated growth method allowed for development of protocols for controlled growth of
anisotropic AuNPs such as nanorods. It was established that using silver nitrate (AgN O
3) as an additive
in the growth step greatly improved the yield of anisotropic shapes as well as their aspect ratio. [23],
[24]. In 2005, Liu and Guyot-Sionnest discovered that the crystallinity of seeds could be controlled and in turn was crucial for the final shape of the anisotropic AuNP [25]. Seeds which were prepared with the surfactant, CTAB, were monocrystalline (figure 2.3d). When these were grown in the presence of Ag(I), high yields of gold nanorods were obtained. In contrast, seeds prepared with sodium citrate were pentatwinned (figure 2.3e) and growing them in the presence of Ag(I) yielded AuBPs with a pentatwinned base (figure 2.3c). In summary, it was now possible to control the shape of the AuNP by either changing the concentration of Ag(I) or the crystallinity of the seeds.
With this discovery, development of AuBPs was made possible. As mentioned previously, the bipyramids possess several attractive properties for use in plasmonics. To improve the purity of the AuBPs, changes in the capping agent and post-purification have been explored [10]. A successful improvement was made by thermally aging the seeds, which increased the yield of pentatwinned seeds and thereby bipyramids. [14], [26] However, the synthesis was still limited to relatively low concentrations of AuBPs and a limited monodispersity.
These issues were addressed in the protocol developed by Chateau et al. [27] Firstly, an overgrowth step was introduced after the thermal aging of the pentatwinned seeds. This led to the synthesis of very small truncated bipyramids. These could then be purified and concentrated through centrifugation, leading to a higher concentration and yield. Secondly, instead of adding the seed solution to a growth solution in one step, the growth solution was added to the overgrown seeds in a multistep process. A schematic of the process is presented in figure 2.4a. This allowed for concentrations of AuBPs up to 15 mM, compared to 0.5 mM previously reported in literature. By varying the seed concentration, the SPR of the bipyramids could also be readily fine-tuned, as seen in figure 2.4b. Furthermore, changes in the composition of surfactants and the concentrations of Ag(I) and overgrown seeds allowed for synthesis of a wide variety of shapes and sizes of the nanoparticles.
(a) (b)
Figure 2.4: a) Schematic of novel procedure presented by Chateau et al. Source: [27]; b) To the left:
Spectrum of of different synthesized bipyramids as a function of seed concentration; To the right: TEM images of AuBPs from the spectrum.Source: [27]
In this report, the synthesis of overgrown seeds and AuBPs were done according to the procedure of
Chateau et al. [27] It is these overgrown seeds and AuBPs that were then used for self-assembly in the
present study.
2.5 Self-assembly of nanoparticles
Naturally, self-assemblies of smaller objects into more complex systems exist on all scales, including the nanoscale. There exist multiple examples of this in both inanimate and living structures. One example of this is gemstone opals, which in some cases are naturally assembled colloidal crystals. Furthermore, complex colloidal dispersions are essential parts of living cells. An excellent example is DNA, which is assembled through the precise interaction between two complimentary single strands of DNA.
With nature as an inspiration, it has been the subject of extensive research to explore the possibilities of controlled self-assembly. [28][29][30] However, that requires understanding the factors which govern these assemblies. As Scarabelli puts it, there are three important characteristics to consider for self- assemblies: the balance of attractive and repulsive interactions, the rational design of building blocks, and the generation of enhanced or novel properties [8]. Research has been produced for all three of these domains. Since the design of building blocks has already been covered extensively when discussing AuNPs and the potential enhancement of properties will be discussed later, it is most interesting to first discuss the balance of interactions during self-assembly. Several theories have been established for understanding and in turn balancing the forces that control the interaction between the “building blocks”. For nanostructures such as AuNPs in liquid dispersion, as the ones synthesized in this report, it is important to consider colloidal interactions.
Colloids are objects ranging in size from a few nanometers to microns and range from macromolecules to the AuNPs in this report. A colloidal dispersion is a multiphase system consisting of colloids equally distributed in a liquid medium. The dispersion is governed by two major forces: electromagnetic and gravitational forces. These play important roles in stabilizing and destabilizing the colloidal dispersion.
Sedimentation is an example of a gravitational force which destabilizes the dispersions. [28]
Figure 2.5: Schematic of the forces governing the DLVO approximation. Source: [31]
The electromagnetic forces in colloidal dispersions can be explained using the DLVO approximation,
which combines the repulsive electrostatic double layer interaction and attractive Van der Waals
interaction [28], [30], [31]. This can be seen in figure 2.5 Electrostatic repulsion occurs when two
similarly charged surfaces or objects are brought close to each other. This force is highly dependent on
ions in the solutions, which can screen this effect. Therefore, the range of this force is highly dependent
on the ionic strength in the dispersion. In contrast, the Van der Waals interaction is an attractive force
governed by molecular interactions at very short range. To obtain a stable suspension, the repulsive
forces need to be greater than the attractive forces.
It is also important to consider the non-negligible entropic contribution in colloidal dispersion. For hydrophobic particles in aqueous solutions, the hydrophobic effect can act as an important destabilizing mechanism. By aggregating, hydrophobic particles are breaking the highly ordered network of water molecules surrounding them, leading to a release of entropy. This has been suggested to play a major role in the assembly of nanoparticles. [32]
The most common strategies for stabilizing colloids against aggregation is through charge-stabilization or steric forces. These can be seen in figure 2.6. This is done by either chemically binding or physically adsorbing the surface of the colloid. [28] For charge-stabilization, charged molecules such as surfactants are used. In this way, the repulsive electrostatic force is enhanced. This is the case for the AuNPs used in this report, where cationic CTAB has been used to stabilize the AuNPs. Studies have shown that CTAB forms a bilayer on the gold surface, leading to a charge reversal to positively charged AuNPs [33].
Stabilization through steric forces is done by using bulky molecules. These bulky molecules sterically hinder AuNPs from coming close to each other. Polymers are commonly used for this purpose. Often, both charge- and steric stabilization are present simultaneously. [28]
Figure 2.6: Schematic of electrostatic (left) and steric (right) stabilization of nanoparticles.
In summary, controlling self-assembly in a colloidal dispersion is to a large extent about controlling the
attractive and repulsive forces affecting the colloids. Manipulating these forces can be done in several
ways, such as changing the pH, temperature, concentration of already present species or by introducing
new species.
Literature Report: Self-Assembly of gold nanoparticles
To investigate the self-assembly of plasmonic gold nanoparticles, an understanding of previous attempts to create self-assemblies is important. Therefore, a more extensive review of the literature on this subject is needed. In this literature study, different methods of self-assembly will be presented. The focus will be on assemblies in dispersion using molecular linkers. Further, previous assemblies on gold nanospheres, nanorods and bipyramids will be discussed and important aspects controlling assembly will be mapped out. However, before discussing self-assembly in more detail, it is important to establish why it is investigated.
3.1 Property enhancements upon assembly
3.1.1 SPR band shift
The motivation for investigating self-assembly of plasmonic nanoparticles stems from several possible enhanced and novel properties. Firstly, bringing the electron clouds of nanoparticles close to each other allows for plasmon coupling effects and another step in tuning the SPR response. Self-assemblies of gold nanoparticles lead to a shift of the SPR band. This can be manipulated in such a way that a red-shift (increasing wavelength) or blue-shift (decreasing wavelength) occurs, enabling shifting the SPR band within a larger window.[9]
The directionality of assemblies is of importance for which coupling effects are seen. In the same way
as the shape and size of individual AuNPs affects the SPR formation, the shape and size of assemblies
do too. The simplest coupling effects can be seen between two AuNPs, which can be modelled as two
dipoles. Constructive interference along the dimer axis (longitudinal) leads to a sharper red shift which
increases with decreasing interparticle distance. A destructive interference occurs along the transversal
axis, leading to a slight blue shift and peak broadening. Another way of looking at this interaction is
through plasmon hybridization, where a bonding and antibonding plasmon dimer is formed at lower
and higher energies, respectively, compared to individual particles (see figure 3.1). [34], [35]
Figure 3.1: Schematic of plasmon hybridization of AuNP dimers
This coupling can be extended beyond two particles with a similar pattern. It has been found that for 1D assemblies, also known as AuNP chains, increasing the length of chain leads to a larger red shift.
Therefore, it follows a behavior similar to increasing the aspect ratio of anisotropic AuNPs. This is advantageous since extended red shifts can be made possible, while maintaining intense and narrow peaks. However, due to the short scale of the near-field enhancement, this effect has a limit around 10 AuNP (depending on the size and shape of the AuNP). Increasing the length beyond this limit doesn’t lead to further shifts. [34]
2D and 3D assemblies form when AuNPs arrange in clusters that extend in more than one direction.
This generally leads to a highly broad red shifted peak, unlike the sharp peak observed for chains. In 1D, the dipoles are only constructive. However, in 2D and 3D there is a mix of constructive and destructive interactions between the rows which reduces the overall dipole. In solution, 3D clusters are generally easier to form than 1D clusters. However, these clusters also serve their own function, as will be seen in further discussions. [35] In this report, mostly 1D clusters will be portrayed and attempted.
The SPR band shift which occurs upon assembly/aggregation of nanoparticles can be used for sensing.
In this case, the shift can act as evidence of the existence of certain molecules or ions which may induce assembly. This principle has been used in several studies for the detection of amino acids inducing aggregations of AuNS [36], [37]. It has also been used to selectively detect low concentrations of P b
2+. In a report by Cai et al., cysteine functionalized AuNRs were side-by-side assembled upon addition of P b
2+, which could be monitored on a UV-Vis-NIR spectrum [38].
SPR band shifting that is dependent on assembly has also enable to the development of active plasmonics [9]. Active plasmonics implies that the SPR band can be actively and reversibly tuned. Several examples of reversible assemblies exist in literature. For instance, Sun et al. were able to to reversibly shift the SPR band by changing the pH [39]. Liu et al. managed to create a reversible, temperature-dependent assembly by using DNA [40]. In similar fashion, Moaseri et al. created AuNS aggregates by quick exposure to low pH which slowly disassembled at higher temperature, making them useful for medical applications [41].
3.1.2 Hot spot formation
Another enhanced property which arises when bringing the electron clouds of the AuNPs together
is the formation of regions with very strong electric fields, called extrinsic hot spots. One of the
key characteristics of plasmonics is the locally enhanced electric field on the surface of plasmonic nanoparticles. However, this electric field can be further enhanced. Studies have shown that sharp edges and tips on anisotropic AuNPs form intrinsic hot spots. Furthermore, these hot spots can be further enhanced by bringing together the tips/edges of two adjacent AuNPs within 10 nm of each other [11].
These hot spots have created significant enhancements of a variety of surface enhanced spectroscopy applications. These include surface-enhanced Raman scattering (SERS), surface-enhanced fluorescence (SEF), and surface-enhanced infrared absorption (SEIRA) [42]. All spectroscopy methods have in common that hot spots formed during plasmon coupling significantly enhance the signal. The signal enhancement is dependent on the shape and size of the AuNPs and their assemblies, as well as the interparticle distance. SERS and SEIRA are mechanistically similar, and apart from the electromagnetic enhancement also experience a chemical enhancement. This occurs when molecules adsorb to the AuNPs at certain sites on the surface, leading to coupling effects. For SEF, studies have shown that the quantum yield and excitation rate can be significantly enhanced by modifying the distance between the nanoparticle and fluorophore molecule.
Figure 3.2: SERS data collected from individual 4-ATP modified gold nanorods and bipyramids (blue and pink curves, respectively) and end-to-end assembled gold nanorods and bipyramids (black and red curves, respectively). Source: [7]
A study by Pardehkhorram et al., was able to show the impact of both the shape and assembly of anisotropic AuNPs on the electric field enhancements [7]. This was done by looking at the degree of enhancement the different setups had on the signal in SERS. Normally, signals from Raman scattering of molecules are low. The enhancement factor (EF) can be measured by comparing the SERS signal to a regular Raman signal. In the study, EF measurements of individual AuNRs and AuBPs were first obtained at ∼ 10
4and ∼ 10
5respectively. The EF values of assembled AuNRs and AuBPs were then obtained at ∼ 10
6and ∼ 10
7respectively. The highly enhanced SERS signals can be seen in figure 3.2.
Such EF allow for ultra-low detection of molecules using SERS. Simulations also showed this highly enhanced electric field was located in-between end-to-end assemblies of AuNRs and AuBPs (see figure 3.3).
Another feature of hot spot formation is the enhancement of chiroptical activity of chiral structures.
Plasmonic circular dichroism has shown promise towards enantioselective catalysis, chiral separation
and detection [43]. Circular dichroism (CD) is a technique which is used to detect chirality of molecules
and structures since chiral molecules exhibit a difference in absorbance for left and right circularly
Figure 3.3: FEM simulation of near-field distribution for (a) single gold bipyramids and (b) nanorods, (c) longitudinally assembled gold bipyramids and (d) nanorods at an incident radiation of 785 nm.
Enhancement efficiency map distribution around (e) single bipyramids and (g) nanorod tips, and (f) end-to-end assembled bipyramids and (h) nanorods separated by 2 nm from the tips. Source: [7]
polarized excitation. However, the CD signal intensity of chiral molecules is often weak. This signal is greatly enhanced for chiral molecules exposed to plasmonic hot spots. This has been successfully achieved when using individual AuNPs. Furthermore, chiral molecular linkers have been used to create AuNP assemblies. In these assemblies, the CD response of the linkers were further enhanced. In a study by Zhu, et al. assemblies of AuNRs were used for sensitive chirality detection of cysteine by creating assemblies with chiral cysteine and measuring the CD response [44]. The use of chiral linkers has also induced chirality into the assemblies of AuNRs themselves [43].
The formation of extrinsic hot spots upon self-assembly of AuNPs can also be applied to other interesting applications. “Hot electrons” which are formed at hot spots during irradiation have been investigated in plasmon catalysis applications. One such application is the use of “hot electrons” to drive chemical reactions on the surface of the AuNPs. Unlike the use of non-selective heat generation for catalysis, plasmon catalysis can potentially more selectively activate chemical bonds through the transfer of ”hot electrons” [45]. This could be used in the conversion of solar energy to chemical energy. Although hot electrons can transfer directly to adsorbed molecules, the process can also be assisted by transferring plasmon-generated hot electrons to a semi-conductor material. This prolongs the lifetime of hot electrons enabling reactions such as water splitting and CO2 reduction [46].
Although many studies on plasmonic catalysis have focused on intrinsic hot spots, studies have also taken advantage of the enhanced extrinsic hot spots formed during assembly. The photochemical conversion of 4-ATP into DMAB was performed at the hot spots of end-to-end assembled AuNRs.
This resulted in a significant enhancement of the plasmon catalysis compared to when using individual AuNRs [6]. Plasmon catalysis has also been used for initiating radical polymerization. Highly localized polymer growth on the tips of AuBPs, an intrinsic hot spot, was achieved. The near field enhancement was also improved by selective polymerization in-between two AuNS, exploiting extrinsic hot spots [47].
In summary, for several of the above-mentioned catalysis applications the formations of extrinsic hot spots via assembly of AuNPs could further enhance performance.
The motivation for investigating assemblies of gold nanoparticles stems largely from the enhancement of
plasmonic features such as hot spots. Furthermore, many applications show that the use of anisotropic
AuNPs (especially AuBPs) show additional enhancement.
3.2 Self-assembly of AuNPs
To achieve these enhanced properties, one must obtain controlled self-assembly. This requires controlling the directionality and size of the assembled cluster so that the wanted properties are obtained. Another important factor is controlling the interparticle distance since both the plasmon shift and hot spots depend on it. This can be achieved with many different strategies and techniques [48] as for example: top-down approaches through templates, direct assemblies by applying an external electric field to align the AuNPs, evaporation-induced self-assembly technique using of capillary forces during the drying process to control assembly, etc.
In this report, self-assembly is achieved in liquid dispersions. They are advantageous compared to top- down methods since they are cheaper and potentially can be produced at higher volumes. Self-assembly in liquid dispersions relies on controlling the colloidal forces acting on the surface of the particles. A widely used technique for doing this is through surface modification by the binding of ligands [33].
Several different ligands have been used for the self-assembly of AuNPs. A ligand which has been readily used is DNA-oligomers. This has been achieved by first binding ssDNA strands to the gold surface and then adding half complimentary DNA strands to induce assembly [49]. Another approach has been to use smaller biomolecules, such as peptides and amino acids. There are several advantages with using small biomolecules as linkers in self-assembly of AuNPs. Firstly, they are naturally occurring and thereby readily available and compatible for biomedical applications. Amino acids also have selective chirality, making them useful for building chiral assemblies. Further, their small size allows for short interparticle spacing during assembly ( 1 nm) [32], [50]. This is highly advantageous for both hot spot generation and for achieving large shifts in the SPR band.
3.2.1 Amino acids as molecular linkers
Using amino acids as linkers in self-assembly largely relies on a two-step mechanism. The amino acids must both bind to the AuNPs and bind to each other. Therefore, an important factor in evaluating the effectiveness of amino acids as binders is their binding affinity to gold surfaces. Hoefling et al. obtained adsorption free energies of amino acids on Au(111) which were obtained through molecular dynamics simulations [51]. From these simulations, a general trend could be seen for the different side groups of amino acids: aromatic < sulfur < positive < polar < aliphatic ∼ negative. Since this trend refers to the adsorption free energy, the binding affinity is the highest for aromatic amino acids and lowest for negatively charged amino acids. It is well established that thiols (containing sulfur) bind readily to gold surfaces, which was also confirmed from this study.
Although binding studies to gold surfaces are useful in finding amino acid linker candidates, it has limitations. The interaction between AuNPs and amino acids depends on both the type of amino acid and AuNP. Firstly, depending on the synthesis method, different capping agents are used to stabilize the AuNPs in dispersion. The charge and binding affinity of these capping agents influence which amino acids can be used, as well as the shape and size of the AuNP.
In this report, assemblies of several different AuNPs will be reviewed. They will be presented with
varying degrees of relevance to the assembly of AuBPs which is investigated in our study. However,
all these studies contribute in different ways to the understanding of self-assemblies of AuNPs in dispersions. Most studies have at least one common variable with our system. The variables which are changed include: capping agent, AuNP structure and binder. These studies are summarized in Table B.
3.3 Assembly of gold nanospheres
Being so heavily researched, a natural start when approaching assemblies using amino acids is by looking at AuNS. The size can vary from a few nanometers to several hundreds. However, in this context, most studies use AuNS with a diameter of around 10 nm [36], [52], [53]. At this size, the SPR band of individual nanospheres is about 520 nm. Considering the size is an important prerequisite for self- assemblies using amino acids. A modeling study by Shao et al. showed that the binding strength of different amino acids changed with varying AuNS diameters. This has to do with the changes in surface curvature of the AuNS and the density of the solvation shell created by surrounding water molecules.
[54]
3.3.1 Stability of AuNS
Most synthesized AuNS are not bare gold surfaces but instead capped with citrate molecules as stabilizing agents. Citrate molecules are bulky and contain three carboxyl groups which, when charged, create a negatively charged shell around the AuNS. The effect of this negatively charged shell has been confirmed in several studies where the zeta potential was measured. A highly negative zeta potential was measured for stable citrate capped AuNS, while ligand exchanges which decreased the zeta potential lead to aggregation [55]. Since assembly relies on destabilizing a colloidal dispersion, one strategy is decreasing the zeta potential by replacing or neutralizing citrate molecules (see figure 3.4).
An extensive study by Zakaria et al. on the aggregation of AuNS looked closer at citrate’s role by changing the pH while also adding different molecular binders [53]. They found that there was a critical zeta potential value for stability, which for them was at around -25 mV (note that this value can change with particle size and shape). Furthermore, they found that the stability of the AuNS was largely dependent on the charge state of citrate. For non-modified AuNS, the particles were stable until the pH went down to 3. This corresponds to being below the final pKa-value of citrate, meaning that it is in its fully protonated state, decreasing its ability to electrostatically stabilize the AuNS. This led to aggregation of AuNS, as seen in figure 3.4. The same trend could be seen when also adding different small molecular ligands.
When comparing aggregations at different pH-values, there was a sharp increase in aggregate sizes after passing one of the pKa-values of citrate, which corresponds to the protonation of one of the carboxyl groups.
For ligand-induced aggregation to take place, sufficient ligand exchange must occur. Therefore, the
affinity of different amino acids to AuNS have been investigated, along with their ability to induce
aggregation. Several studies show that the trend follows what was first suggested by simulations: side
groups with thiols, such as cysteine, bind most readily to AuNS, leading to rapid aggregation. Positive
side groups, such as amines are also effective at binding to AuNS and inducing aggregation. This is the
case for Arginine and Lysine. Histidine, however, only obtained aggregations at high concentrations or at very low pH (high degree of protonation). Neutral side groups like alanine showed no effect on aggregation. This confirms its poor ability to bind to gold. Finally, glutamate, which contains a carboxyl group side chain, managed to further decrease the zeta potential. This suggests that it binds and stabilizes the AuNS better than citrate. This was also confirmed at pH 3, where the addition of glutamate prevented aggregation of AuNS. [53], [56]
Figure 3.4: Schematic of AuNS aggregation pathways through a) lowering the pH below 3 or b) ligand exchange as described by Zakaria et al. [53] These pathways lead to a zeta potential closer to zero, which causes aggregation of particles.
Interestingly, the stabilizing ability of charged carboxyl groups was not limited to the side groups of amino acids. The aggregation of AuNS using cysteine and lysine were compared to cysteamine and cadaverine respectively. These molecules have the same structure as the amino acids, but without the carboxyl group. Aggregation was significantly faster for cysteamine and cadaverine, meaning that the carboxyl group had an inhibitory role. [53] This is inhibitor could be useful when trying to achieve self- assembly, since rapid aggregation is difficult to control. Further, this effect was used to follow enzymatic decarboxylation reactions of lysine and histidine. The decarboxylation products induced aggregation while their amino acid counterpart didn’t [57].
Figure 3.5: Schematic of proposed binding mechanism of cysteine-mediated aggregation of citrate- capped AuNS
The reduction of negatively charged carboxyl groups are not the only factor leading to aggregation.
Addition of cysteine displayed rapid aggregation even at high pH when the zeta potential was below -25 mV [53]. This suggests that bridging occurs between AuNS through hydrogen bonding or electrostatic interactions between the amino acid backbones of cysteine. A suggested mechanism is therefore that the thiol side group binds to the gold surface, allowing the amino acid backbone to interact with other cysteine molecules (see figure 3.5. A similar mechanism has been suggested for amine-based amino acids, like lysine and arginine. [53], [56]
3.3.2 Controlling assembly
To achieve self-assembly is not necessarily the same as obtaining aggregation. Aggregations can be useful in sensing when one differentiates between individual particles and aggregates. However, this is usually achieved with little control over interparticle distance, directionality and cluster size. Self- assemblies of AuNS using amino acids have been achieved for both 3D clusters and linear chains.
The purpose for these assemblies differs, but they both require balance of the repulsive and attractive forces explained above. This entails controlling factors like concentration, temperature, pH and ionic strength.
Figure 3.6: Effect of surface ligands on interparticle spacing between AuNPs and how this spacing influences the NIR extinction and the degree of dissociation of nanoclusters to individual particles.
Source: [58]
Since AuNS are isotropic, they do not have any preferential directionality upon assembly. Therefore, the aggregates which are obtained when adding excess cysteine or lysine at low pH can be considered uncontrolled 3D clusters. Controlling them requires limiting the cluster size and interparticle distance.
In a study by Moaseri et al. this was achieved through addition of cysteine and rapid changes in pH.
[58] The study focused on achieving two things: a large NIR plasmonic shift, and reversibility between clusters and individual particles. This was accomplished by first partially replacing citrate molecules on the surface with cysteine. These AuNS were then exposed to pH 3 leading to quick aggregation due to charge neutralization. However, the aggregation was quenched after just 1 second by quickly increasing the pH to 7. In this way, the size of the clusters was limited. Using this method, clusters were obtained with interparticle spacing below 1 nm and a broad absorption peak stretching to the NIR.
Additionally, the clusters were slowly reverted to individual particles when exposed to pH 5 and 37
◦
C (which represents the conditions of a cancerous cell). This shows promise for use in biomedical
imaging and treatment. It was also concluded that the ligand exchange had to be balanced to achieve
both an NIR shift and reversibility (as seen in figure 3.6). A more recent publication from the same group
also showed further optimization of this technique with glutathione, showing that the cluster sizes were highly tunable with respect to pH and that meta-stable clusters could be achieved for extended time periods [41].
Linear assemblies
For intense and narrow plasmonic red shifting, linear assemblies are preferable. Linear assemblies require that a dipole is formed which controls the preferential directionality. For this reason, Wang and Zhang developed a model for interactions between AuNS [59]. They identified that although the interaction potential between particles is govern largely by long range electrostatic repulsion and short ranged van der Waals forces (see figure 2.5), an anisotropic dipolar interaction can be induced, leading to dimer formation. When they form, the electrostatic double layer around the particles rearrange in such a way that the electrostatic repulsion is lower at the ends than the side, inducing further growth into longer AuNS chains. This showed that if a dipole is induced in AuNS, the electrostatic interactions, originally isotropic, can enforce anisotropic assembly. The study also showed that if the electrostatic repulsion along the sides of the chain became lower than the Van der Waals attraction, 2D and 3D growth occurred.
Different strategies have been used to enhance dipole formation. One strategy involves exchanging the solvent from water to a solvent with a lower dielectric constant (lower polarity). This enhances weak dipolar interaction between AuNS and induces chain growth [59]. The combined effect of replacing the solvent and adding salt has been studied and led to successful linear assemblies [60]. Another strategy involves evoking a dipole on the surface of AuNS by partial ligand exchange. As discussed previously, replacing citrate with amino acids can lead to both reduction of surface charge and creation of active binding sites. If these amino acids replace citrate in non-uniform patches, a charge separation can be created which induces a dipole. In a recent study, this was observed for AuNS which were exposed to very low levels of amino acids. In the study, AuNS were first centrifugated to remove a large portion of the citrate molecules on the surface and dispersed in ethanol. Thereafter, a low level of amino acids, insufficient to cover the whole surface of the AuNS, was added. This led to AuNS chains, suggesting that non-uniform surface modification could produce 1D assemblies. [36] This electric dipole-driven mechanism was also suggested by Li et al. when using MEA (a small molecule similar to amino acids) [61].
Figure 3.7: Mechanism of Arginine-mediated assembly of AuNP as proposed by Sethi et al. [52]
Sethi et al. have used this strategy with cysteine and arginine. This was examined by varying the concentration of added amino acid, temperature, ionic strength, and solvent of the dispersion containing citrate capped AuNS. For both amino acids, there were certain concentration windows which induced certain behavior. At low concentrations, slow assembly kinetics were observed, but usually with higher degrees of linear assemblies and few larger clusters (see figure 3.7. As the concentration increased, faster kinetics were observed. These often led to large aggregates and precipitates. Finally, at very high amino acid levels, stabilization of individual AuNS was observed again. This behavior suggests that at low concentrations, the amino acid only partially covers the AuNS, inducing a dipole. [56] A similar behavior was observed in an extensive study by Abbas et al., where a sigmoid-type function was observed for the assembly rate when increasing concentration. In this study, this behavior was connected to the zeta potential, which reached an unstable range (low absolute value) at low concentrations of linker molecule. [62]
Due to cysteine superior binding affinity to gold, these windows were observed at significantly lower ( 1/10) concentrations than arginine. This made it harder to find a window at which controlled assembly was observed for cysteine. Furthermore, the strong bonds formed with cysteine are thought to be more static, limiting its movement along the gold surface to create a charge segregation and dipole, which is thought to occur with more loosely bound arginine. Therefore, uncontrolled clusters were obtained in a greater degree for cysteine compared to linear assemblies for arginine. [52], [56]
Another dependency which was tested was the temperature. For arginine-containing systems, increasing the temperature increased the assembly kinetics. However, for cysteine elevated temperature disrupted the electrostatic interactions, limiting the assembly. The solvent was also partially replaced with ethanol for the arginine induced assembly, in order to decrease the solvent dielectric constant.
Here, the behavior followed what was observed by Wang and Zhang. Decreasing the solvent dielectric enhanced linear assemblies. As expected, the assembly increased upon the addition of NaCl as well, which decreases the range of charges. Broader peaks were observed in this case, likely due to the shortening of the range of electrostatic repulsion. [52], [56]
Figure 3.8: Schematic model representing the interaction of the different parameters involved in the self-assembly process, including the concentration of AuNPs (C
N P), linker concentration (C
C L), zeta potential (ζ), and linker ionization (I). Ar represents the assembly rate and Ad represents the assembly degree. Source: [62]
The studies by Sethi et al. showed that AuNS dimers and chains could be formed through inducing
charge separation with Arginine. Further, it showed that the forces guiding linear assemblies can be controlled through fine-tuning of temperature, solvent, and concentration. These dependencies were further explicated by Abbas et al. when using the biomolecule p-ATP [62]. The study found that properties of self-assembly (assembly rate and degree) could be connected to the initial conditions of both the nanoparticle solution (concentration, pH, ζ-potential) and the linker reagent (pH, pKa, concentration). This is summarized in figure 3.8. The molar ratio of linker to nanoparticle (C
C L/C
N P) and ionization degree of the linker were found to be the most important when achieving assemblies.
Zhang et al. also used these dependencies to build complex assemblies such as hexagonal micro flakes and ultralong microwires with slow and fast growth kinetics respectively by exposing cysteine functionalized AuNPs to different pH and solvent conditions. [63]
Despite these successes in predicting and controlling 1D assemblies in dispersions using isotropic AuNS, it still has limitations. The isotropic nature makes 1D assemblies sensitive to changes in the dispersion.
3.4 Assembly of gold nanorods
With the turn of the century, significant advancements in the synthesis of AuNRs led to the first look at assemblies of AuNRs by Nikoobakht et al. They were able to show that manipulations of nanoparticle concentration, ionic strength and surfactant concentration led to assemblies of AuNRs in different directions [64]. A few years later, Catherine Murphy’s team were able to show that preferential end- to-end assemblies of AuNRs in dispersion could be achieved by site-specific adsorption of Biotin- Streptavidin connectors [65]. Thus, showing that controlled assemblies could be obtained. Since then, the research on self-assemblies of AuNRs in dispersion has expanded greatly. A major reason for this is the anisotropic shape of AuNRs. As will be made evident, using anisotropic AuNRs in assembly adds a level of complexity. With this, new enhanced properties arise, along with challenges.
3.4.1 Stability of AuNRs
AuNRs differ from AuNS both in terms of shape and capping agent, in turn effecting their stability. Most AuNRs are synthesized using the surfactant, cetyltrimethylammonium bromide (CTAB), as a capping agent. CTAB contains a positively charged quaternary ammonium head group and a hydrophobic tail. The ammonium head binds to the gold surface and the hydrophobic tail faces outwards. Due to their hydrophobicity, a bilayer of CTAB forms in water with the hydrophobic tails facing inwards, as seen in figure 3.9a. Thereby, the surface of CTAB-capped AuNRs exhibit a positive charge which electrostatically stabilize the particles. This differs from AuNS which are negatively charged [12].
The degree of surface coverage of CTAB molecules on AuNRs play a vital role for their stability and in-
turn ability to self-assemble. Studies have shown that the surface coverage varies both between particles
and along the same particle. A study by Merrill et al. examined the effect of CTAB concentration on
the stability of the AuNRs during multiple purification steps. The study found that above a certain
threshold concentration, the particles remain stable. However, below this threshold, AuNRs started to
aggregate. Surprisingly, two populations of AuNRs could be observed, one which remained stable and
one which aggregated. This suggested that the surface coverage of CTAB differed between AuNRs in solution [66].
Further, it was suggested already in 2003 that the CTAB surface coverage at the ends of AuNRs was lower than at the sides, allowing the ends to be functionalized more easily [65]. This has been observed in several reports since then and the lower surface coverage has been attributed to factors like the higher surface curvature at the ends and the facet-dependent binding densities [12], [33]. In a recent report, the surface coverage of CTAB on AuNRs was quantified using STEM-EELS [67]. The relative binding density as a function of position along the AuNR surface is shown in figure 3.9b. It confirmed that the density of CTAB molecules at the AuNR tips was around 30 % lower than the sides. However, it also found that there were large variations in the surface coverage between different particles. This confirms the observation by Merrill et al. above. Furthermore, it shows that the CTAB coverage is a decisive factor affecting assembly of AuNRs.
(a) (b)
Figure 3.9: a) Representation of CTAB-capped gold nanorod. CTAB molecules are represented by their positively charged head group and hydrophobic tail. b) Plot of the relative binding density as a function of position along the boundary for CTAB-coated AuNRs. Black line represents the average relative binding density. Source: [67]
One of the consequences of the lower surface coverage at the ends of AuNRs is the preference to assemble in an end-to-end manner, forming a 1D AuNR chain. This type of alignment has impressively been obtained without the addition of any molecular linker. Jain et al. obtained 1D assemblies when diluting AuNRs in low CTAB concentrations. At the lowest AuNR concentration (0.12 nM) and CTAB concentration (1 µM), 1D assemblies were observed [68].
The age of the CTAB-capped AuNRs have also been reported to play a role in their stability, and in
turn assembly kinetics. Several studies have noted that stock solutions with AuNRs which are left for
several days form 1D assemblies more readily than freshly prepared AuNRs [50], [69], [70]. Haidar
et al. showed that AuNRs which had been aged for 5-20 days assembled in less than 10 minutes upon
addition of cysteine, compared to hours for freshly prepared AuNRs [50]. This has been attributed to
the aligning of CTAB molecules along the gold surface, suggesting that the relative binding density of
CTAB decreases at the tips over time [69].
3.4.2 Anisotropic coupling
Interestingly, it is not only the ability to aggregate or assemble that changes with the shape anisotropy of AuNRs. The coupling effects of AuNR assemblies are more complex. Due to their anisotropic shape, AuNRs exhibit both a transversal and L-SPR band. Upon coupling of AuNR dimers, both these bands shift. Furthermore, these coupling effects are highly dependent on the orientation of the AuNRs with respect to each other.
(a) (b)
Figure 3.10: a) Schematic of Vis-NIR spectrum of gold nanorods (AuNRs) upon dimer formation. Red curve: end-to-end assembly, Blue curve: side-to-side assembly; Black: Reference curve of individual AuNRs b) Plasmon hybridization scheme for rod dimers in different geometric arrangements. Note that transversal – transversal hybridizations are left out. The dark modes are represented with red crosses.
Source: [71]
The most used alignments are side-to-side (SS) and end-to-end (EE). The spectral response of their coupling effects has been extensively studied [71]. EE dimer formations lead to attractive coupling of the longitudinal modes and a large red shift of the L-SPR. The transverse modes couple repulsively, leading to a slight blue shift of the T-SPR. Upon EE chain formation, the L-SPR is further red-shifted. For SS dimers, the transverse modes are attractively coupled, leading to a slight lowering of energy (red shift).
However, the longitudinal modes are coupled repulsively. Since they are significantly more intense, the averaged spectrum shows a slight broadened blue shift and the T-SPR is washed away. Upon SS chain formation, the combined band is further blue shifted. The spectral response of SS and EE assemblies are presented in figure 3.10a.
It is important to note that for both interactions there are also possible attractive and repulsive coupling which are dark modes and therefore not seen in a spectrum. Furthermore, these two alignments lead to coupling of the longitudinal modes, exhibiting the largest coupling effects, and thus making them the most interesting. Other geometries are however also possible, exhibiting their own coupling effects. T- and L-shaped geometries lead to coupling between the transversal and longitudinal modes, exhibiting a lower shift of energy. A hybridization scheme of coupled dimers with different geometries can be seen in figure 3.10b.
The anisotropic shape of AuNRs clearly make them more tunable with regards to their SPR. However, the
added sensitivity to the coupling geometry also makes it more difficult to follow assemblies using spectral responses. The coupling hybridization scheme shows that many different couplings give red and blue shift, respectively. Therefore, a red shift does not necessarily act as proof for EE assembly, for example.
This could also be a combination of coupling modes in an uncontrolled aggregation. Furthermore, as previously reported, the shifts are highly dependent on interparticle distance. Therefore, the control of plasmon coupling requires the control of numerous factors such as AuNR size and shape, assembly geometry, and interparticle distance.
Enhancement of properties by assembly type
