Controlled Synthesis of Bimetallic Gold‐Silver Nanostars: Atomic Insights and Predictive Formation Model

Abstract

The nucleation and growth of bimetallic gold‐silver nanostars (GNSs) are investigated to elucidate their atomic‐scale formation mechanism. Motivated by the increasing demand for nanomaterials with enhanced optical and catalytic properties, particularly for applications in biosensing, bioimaging, and photothermal therapy, this work focuses on understanding the factors governing GNSs formation. GNSs are synthesized by reducing HAuCl₄ with ascorbic acid in the presence of AgNO₃, exploring the influence of temperature, delay time in AgNO₃ introduction, and AgNO₃ concentration. High‐resolution electron microscopy, energy‐dispersive X‐ray spectroscopy, high‐resolution X‐ray photoelectron spectroscopy, and synchrotron‐based powder X‐ray diffraction are used to characterize their morphology, size, composition, and stability. These findings reveal that AgNO₃ promotes anisotropic growth through the formation of metallic Ag and AgCl on GNSs surfaces, leading to thorn‐like structures. A detailed analysis of kinetics, particle concentration, and nucleation barriers enables the development of a theoretical model to predict optimal synthesis conditions. This work provides new insights into controlling GNSs morphology and properties, which are critical for optimizing their performance in catalysis, sensing, and biomedical applications. The novelty lies in the discovery of the role of AgCl in directing GNSs growth and the formulation of a predictive model for synthesis optimization.

A systematic study of gold‐silver nanostars (GNSs) formation reveals the key role of silver and AgCl in directing GNSs growth. Using a rapid and green synthesis approach, temperature and AgNO₃ addition influence the process. This enables control over GNSs morphology, size, and concentration, allowing their properties to be tuned for various fields such as catalytic and sensing applications.

1. Introduction

Metallic nanoparticles (NPs) synthesized from noble metals (e.g. Au, Cu, Ag, Pt, Hg) are widely used nanomaterials, and, owing to a high surface‐to‐volume ratio and unique properties that can be designed and tailored, they have found extensive use in multiple fields, such as medicine, materials engineering, chemistry, and physics. Gold nanoparticles (GNPs) are renowned for their chemical and physical stability, reproducible production, and ease of surface modification. Their synthesis through various methods allows for the creation of a wide range of morphologies, porosities, and functionalities.^{[ 1} , ² , ^{3 ]}

Bimetallic nanoparticles (BMNPs) represent an advanced class of nanomaterials with enhanced technological properties, which vary based on their size, composition, and structure.^{[ 4} , ^{5 ]} In particular, gold and silver‐based BMNPs are highly valued for their superior sensitivity as surface‐enhanced Raman scattering (SERS) substrates, benefiting from the synergistic enhancement of the optical and catalytic properties of both elements.^{[ 6} , ⁷ , ^{8 ]} Additionally, silver‐silver chloride (Ag‐AgCl) nanostructures have attracted attention as promising photocatalysts due to the improved stability of AgCl under UV light.^{[ 9} , ^{10 ]} The combination of gold with silver and silver chloride has demonstrated superior absorbance and localized surface plasmon resonance (LSPR) properties compared to Ag‐AgCl alone, along with increased stability and synthesis reproducibility, making them ideal for use in catalysis and sensing applications.^{[ 4} , ¹¹ , ¹² , ¹³ , ^{14 ]}

The well‐known Turkevich method for the synthesis of spherical GNPs is the basis for gold‐silver BMNPs.^{[ 15 ]} This method involves the reduction of [AuCl₄]⁻ anions in aqueous solution by sodium citrate, which also serves as a stabilizing agent through electrostatic repulsion. In many works, AgNO₃ has been added to the reaction system to produce various morphologies and, usually, it has been also assumed that gold and silver are distributed equally in the NPs. Recently, Blommaerts et al. showed a time‐resolved composition evolution of spherical NPs during the synthesis and proved that although there is a single absorption peak in the UV–vis spectrum, these NPs do not exhibit a homogenous atomic distribution and that they have a gold‐core and silver‐enriched shell nanostructure.^{[ 16 ]} The presence of a single absorption peak, instead of two, is explained by plasmon band broadening.

Among the diverse GNPs, gold nanostars (GNSs) are extensively studied for their applications in biosensing, bioimaging,^{[ 17 ]} and photothermal therapy,^{[ 18} , ^{19 ]} due to their enhanced local optical and electromagnetic properties, particularly in their star‐like protrusions.^{[ 20} , ²¹ , ²² , ^{23 ]} Venttisette et al. successfully 3D‐printed hydrogel scaffolds incorporated with GNSs, using light‐based stereolithography (SLA), and showed SERS‐sensing properties of the scaffolds.^{[ 24 ]} Researchers have also proposed the synthesis of virus‐like particles.^{[ 25 ]} Moreover, highly curved surfaces of the nanostars expose more low‐coordinated atoms, making them a promising candidate for enhanced catalytic activity. Ullah et al.^{[ 26 ]} demonstrated that GNSs coated with varying Ag thicknesses exhibited significantly higher catalytic activity compared to pure GNSs in the reduction process of 4‐nitrophenol to 4‐aminophenol and the degradation of methyl orange—both stable, yet toxic and mutagenic compounds. Increased Ag thickness further boosted the catalytic performance.

A simple, rapid, and efficient method to produce GNSs is a one‐pot, seedless, bottom‐up synthesis. This method involves the reduction of HAuCl₄ and AgNO₃ by ascorbic acid (AA) under acidic conditions, avoiding the use of environmentally unfriendly reagents and solvents, such as cetyltrimethylammonium bromide surfactant or dimethylformamide, making it a green synthesis.^{[ 3} , ^{23 ]} It has been suggested that in this type of synthesis, gold ions are reduced by AA, forming gold clusters that aggregate into polycrystalline NPs.^{[ 27 ]} Afterward, AgNO₃ promotes anisotropic growth of the GNPs, because of (111) plane surface blocking, which causes thorn growth at the non‐blocked sites.^{[ 28 ]} Others proposed that reduced silver on the surface of the GNPs serves as active sites or seeds for further precipitation and growth of the gold thorns.^{[ 27 ]} A third possible mechanism is that in the presence of chlorine anions, AgCl precipitates on the gold seeds, decreases the gold growth rate, causing anisotropic growth and the formation of GNSs. Although silver was detected on the GNS surface using X‐Ray photoelectron spectroscopy (XPS) measurements, no diffraction peak of AgCl was found in the X‐Ray diffraction (XRD) pattern.^{[ 29 ]} Intriguingly, the final morphology of GNSs depends on several factors, including the gold‐to‐silver ratio, reaction time, pH, and AA relative concentration.^{[ 17} , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ^{32 ]} However, the mechanism of GNS formation at the atomic level when using HAuCl₄, AgNO₃, and AA as precursors is not fully understood yet. It is also hypothesized that gold and silver are distributed evenly throughout the particle, forming an alloy structure. This argument is based on high‐angle annular dark field scanning transmission electron microscopy (HAADF‐STEM) and energy‐dispersive X‐ray spectroscopy (EDS) micrographs, and on the presence of a single plasmon band instead of two,^{[ 30 ]} as supposed to be found in core‐shell structures.^{[ 33 ]}

A few mechanisms were suggested for the synthesis of GNPs via Turkevich approach, such as the nucleation‐diffusional growth mechanism proposed by Turkevich,^{[ 15 ]} referred to as the “organizer” model for nucleation which hypothesized burst nucleation; the Finke‐Watzky mechanism^{[ 34} , ^{35 ]} which includes slow and continuous nucleation followed by an autocatalytic surface growth; the four‐step seed‐mediated growth mechanism presented by Polte et al.^{[ 36} , ^{37 ]} However, all these mechanisms do not allow us to describe quantitatively the kinetics and temperature dependence of the GNSs formation and evolution over time and, therefore, do not allow us to choose the optimal conditions for the synthesis of uniform GNSs, which might be critical for various applications. For instance, Kim et al.^{[ 38 ]} showed how GNSs that were synthesized using AgNO₃ found to be an effective probe for tip‐enhanced Raman scattering and tip‐enhanced fluorescence, but due to the dispersion of the GNS shape, also the nano‐optical properties showed variation.

Aiming to shed light into the mechanism of the formation of GNSs in the presence of silver, in this study, we investigate the complex process of nucleation and growth of bimetallic gold‐silver nanostars (referred here as GNSs). We focused on the relationship between the reaction temperature and the timing of AgNO₃ introduction and the morphology, size, particle concentration, stability, and optical properties of the acquired GNS. Imaging was conducted using HAADF‐STEM in a Thermo‐Fisher high‐resolution aberration corrected scanning transmission electron microscopy (HR‐STEM) equipped with a Bruker Dual‐X EDS detector, providing detailed insights into the synthesized GNSs. Additionally, synchrotron high‐resolution powder X‐ray diffraction (HR‐PXRD), and other techniques such as dynamic light scattering (DLS) and nanoparticle tracking analysis (NTA), together with the calculation of the energy barriers for nucleation and growth, and the development of a theoretical model that rationalizes the process, enables us to propose a mechanism for the synthesis of this type of widely used nanomaterials.

2. Results

2.1. Effect of temperature and delay of AgNO₃ addition on GNS synthesis

We delved into the impact of temperature on the growth of GNSs. We kept the AgNO₃ concentration constant (15mM) and examined two delay periods for the addition of AgNO₃ to the system (2 and 5 s). The thiol‐poly(ethylene glycol)‐carboxylic acid (HS‐PEG‐COOH) molecules were used to counteract the GNSs growth and secure their physical stability by binding strongly to the NP surface via gold‐sulfur covalent bond and PEGylating their surface. The terminal carboxylic acid group is a reactive moiety for potential modifications such as the conjugation of peptides and proteins.^{[ 39} , ^{40 ]}

Figure 1 presents the high‐resolution scanning electron microscopy (HR‐SEM) micrographs of the synthesized GNSs morphologies under different conditions, and the polydispersity index (PDI) values obtained from DLS measurements. At 3 °C and 2 s delay time (Figure 1A), we observe highly uniform GNSs. Once the temperature is increased up to 20 °C or the delay time is longer (Figure 1B–H), the uniformity decreases, with a notable increase in the presence of plates instead of thorns on the GNS surface. Above 25 °C and 5 s delay (Figure 1J,L), spherical NPs are formed to some extent, while at 35 °C, spherical particles are present even after a 2 s delay (Figure 1M). At 40 °C, nearly all the particles are spherical (Figure 1O,P). In addition, the higher the temperature of the synthesis, the greater the PDI of the particles.

Figure 1.

HR‐SEM micrographs of GNSs, synthesized by varying the delay time for the addition of 10 µL of 15 mM AgNO₃ to the reaction solution (2 or 5 s) and the solution temperature (3, 10, 15, 20, 25, 30, 35, and 40 °C). The HS‐PEG‐COOH was added to all the samples 60 s post‐reaction initiation. On the right bottom part are presented PDI values measured via DLS, which allows assesment of the monodispersity as a function of morphology.

The DLS technique measures the size of particles by analyzing the intensity fluctuations of the scattered light of the particle in suspension. GNSs with thorns and plates tend to scatter more light than spherical particles with smooth surfaces. Therefore, this type of morphology causes the hydrodynamic diameter (D_h) to be overestimated compared to the actual size, which is a limitation of the system. Moreover, molecules conjugated to the GNSs surface, especially hydrophilic molecules which absorb water, such as HS‐PEG‐COOH, can further increase the measured diameter value. The PDI value is important to estimate the broadness of the NP size distribution in the system, and it ranges between 0 and 1. PDI values smaller than 0.1 indicate a highly monodispersed particle distribution. For non‐spherical particles, values smaller than 0.2 are considered suitable. A value of 1 indicates that the particles are highly polydisperse with several size populations.^{[ 28} , ^{41 ]} Figure 1 enables a comparison of the PDI values and the GNS morphology (GNS size is presented in Figure 2 below). We expect a slower GNS formation at lower temperatures, therefore, a more controlled synthesis, which might be the reason for the PDI values lower than 0.2 observed at temperatures below 25 °C.

Figure 2.

Characterization of the GNSs synthesized at different temperatures and delay times before adding AgNO₃ (2 and 5 s). A) Hydrodynamic diameter (D_h) measured using DLS. B) Manual size analysis using ImageJ software (n = 90). C) UV–Vis spectra of GNSs synthesized with 2 s delay time before adding AgNO₃. D) UV–vis spectra of GNPs synthesized with 5 s delay time before adding AgNO₃. E) Hydrodynamic diameter (D_h) measured immediately after synthesis (Fresh) and after 7 months for 2 s delay. F) Hydrodynamic diamter (D_h) measured immediately after synthesis and after 7 months for 5‐s delay. The average and standard variation values are presented (n = 3, 13 run each).

Figure 2 presents further characterization of those particles. Figure 2A,B shows the size distribution of the GNSs, measured via DLS and HR‐SEM image analysis using ImageJ software, respectively. Both methods indicate a similar trend of average size increasing at a temperature up to 20 °C, both in 2 and 5 s delay time before AgNO₃ addition, followed by a gradual size decrease. GNSs with a delay time of 5 s were found to be smaller, on average, than those with a 2 s delay, corresponding to the appearance of the spherical particles visualized in HR‐SEM (Figure 1). The average size of the GNSs measured by these two methods differs, which is attributed to the fact that in the DLS, the D_h includes the hydrated HS‐PEG‐COOH layer and the adsorbed water molecules, resulting in a larger measured particle size. Conversely, HR‐SEM measures the size of dry GNSs, which excludes the surrounding hydration layer. The size obtained by manual ImageJ analysis represents a diameter of tip‐to‐tip distance.

Figure S1, Supporting Information presents raw DLS data for GNS synthesized under the same conditions as in Figure 1A, but on different occasions. This demonstrates the reproducibility of the synthesis, providing evidence that the results are consistent across multiple trials.

The GNSs UV–Vis absorption spectra, which are presented in Figure 2C,D, support the morphology evolution that occurs due to the increase in the temperature. In general, GNPs cause a distinct optical feature due to the LSPR, depending on the size and shape of the particle. The current results are in very good agreement with previous studies.^{[ 30} , ⁴² , ^{43 ]} At lower temperatures, where the nanostar morphology is mostly observed, the maximum absorbance is between 750–800 nm, and the absorption peak is wide. Reduction in GNSs size and transformation into spherical GNPs causes a blueshift, a decrease in absorbance peak wavelength and peak width. At 40 °C and 5 s delay, a strong absorbance peak was found at ≈545 nm, which is usually attributed to spherical GNPs with a diameter of 50–60 nm.^{[ 44 ]} Moreover, at ≈400 nm, we can observe an increased absorbance, which might suggest the formation of a second plasmon band due to the core‐shell structure of the GNSs.^{[ 33} , ^{45 ]} Ngo et al.^{[ 45 ]} demonstrated that pure silver nanostars (SNSs) exhibit an absorbance peak between 350 and 400 nm. When SNSs were coated with a gold shell by using galvanic replacement, they showed broadening of the characyeristic SNSs peak and an increase in absorption at 700–800 nm, corresponding to the increased gold presence on the SNSs surface. Our results, which show the opposite trend, suggest that the GNSs are coated with silver, a possibility we will discuss later in the paper.

The D_h of GNSs stored for 7 months at 4 °C was measured again by DLS to assure their physical stability and lack of aggregation. The results shown in Figure 2E,F confirm that most of the GNSs remain unchanged and are very stable, except the NPs prepared at 10 °C with a 5 s delay, and at present, no clear explanation can be provided for this exception.

2.2. Effect of AgNO₃ Concentration on GNS Synthesis

To investigate the effect of AgNO₃ concentration on GNSs formation, we set the delay period to 2 s and conducted the experiments at two different temperatures (3 or 20 °C). Figure 3 presents the GNSs imaged by HR‐SEM, highlighting a clear variance in morphology and size, particularly at 3 °C. At this lower temperature, the GNSs exhibit plate‐like morphology (Figure 3A,C), while increasing AgNO₃ concentration reduces plate formation on the surface and decreases particle size (Figure 3E,G,I). At 20 °C, the morphological change is less prominent, yet still noticeable. Figure 3B also shows plate‐like morphology, and as the AgNO₃ concentration increases, the particle's surface becomes more spherical (Figure 3D,F,H,J). Despite these variations, the PDI values indicate that all synthesized GNSs are sufficiently monodispersed.

Figure 3.

HR‐SEM micrographs of GNSs, synthesized by varying the AgNO₃ concentration (2, 5, 15, 50, or 100 mM) and the solution temperature (3 or 20 °C). The AgNO₃ was added 2 s after the synthesis start, and HS‐PEG‐COOH was added to all the samples 60 s post‐reaction initiation. On the right bottom part are presented PDI values measured via DLS, which allows monodispersity assessment compared to morphology.

Figure 4 presents additional characterization of the GNSs. Figure 4A,B illustrates the visible change in the GNSs solutions. At 3 °C (Figure 4A), both the hue of the solution and its intensity change once the AgNO₃ concentration increases, which corresponds to the change in the absorbance behavior presented in Figure 4C. As the solutions become darker, their absorbance increases. Additionally, the maximum absorbance peak changes between the samples, influencing the observed color shade. We can see the most significant change between the 5 and 15 mM samples. At 20 °C, once the AgNO₃ concentration increases, the hue of the solution does not change but exhibits higher intensity, supported by the increase in absorbance with no change in the postion of the peak (Figure 4B,D). Figure 4E presents the GNSs hydrodynamic size, which gradually decreases once the AgNO₃ concentration increases, while in parallel, the particle concentration measured by NTA increases (Figure 4F). Using 2 and 5 mM AgNO₃ results in the low GNSs concentration, especially at 3 °C, and once reaching 15 mM, the GNSs concentrations at 3 and 20 °C are similar.

Figure 4.

GNS solutions that were synthesized at A) 3 °C and B) 20 °C with increasing AgNO₃ concentrations (from left to right: 2, 5, 15, 50, or 100 mM). C) UV–VV is spectra of GNSs synthesized at 3 °C. D) UV–vs spectra of GNSs at 20 °C. E) Hydrodynamic diameter (D_h) measured using DLS. The average and standard variation values are presented (n = 3, 13 runs each). F) Particle concentrations measured using NTA.

2.3. GNSs Chemical Analysis

We performed chemical analysis of the most uniform GNSs, produced at 3 °C and a 2 s delay for AgNO₃ addition (15 mM). The GNSs were imaged in HAADF‐STEM mode, and EDS maps were acquired both from the entire particle and a single thorn at higher modification, as presented in Figure 5A,B, respectively. The EDS spectrum of the thorn is presented in Figure S2A, Supporting Information. Silver was found mainly on the surface of the GNSs, and it is not equally distributed within the particle volume. Figure 5B shows the segregation of silver on the surface of the gold thorns, forming a layer with a thickness of 1.7 ± 0.4 nm (average and S.D). GNPs that were synthesized under the same conditions but without AgNO₃ addition, showed no Ag signal in the EDS spectrum (Figure S2B, Supporting Information). The particles remain stable and show no mixing of gold and silver after storage at 4 °C for 5 months (Figure S2C, Supporting Information). Figure S3, Supporting Information, presents EDS results of all the GNSs featured in Figure 1. The atomic fraction of gold atoms in the NPs is ≈10 to 17 times greater than the one of silver (the initial HAuCl₄:AgNO₃ ratio was 17), with no specific pattern observed across the different temperatures and no significant change between the different delay times prior to adding silver.

Figure 5.

Chemical analysis of GNSs synthesized at 3 °C and 2‐s delay before adding AgNO₃. EDS elemental map, measured in HR‐STEM system (HAADF‐STEM mode) of a A) whole GNS and B) a GNS single thorn (HAADF‐STEM image, Au, Ag, Cl and overlay of the elements). C) The atomic fraction profile measured from the thorn's edge, marked in a white line on the image of the thorn's overlay. D) HR‐TEM image of the gold lattice structure, from a GNS thorn, showing the presence of a twofold twin in the center of the thorn. E) Fast Fourier transform (FFT) pattern of the yellow square. F) FFT pattern of the green square.

Additionally, chlorine atoms were also discovered on the GNSs surface, partially overlapping with the silver layer. A profile of the atomic fraction of gold, silver and chlorine is presented in Figure 5C, which emphasizes the greater silver and chlorine concentration on the thorn's surface, and the presence of gold mainly in the core of the particle. For pure GNPs, chlorine segregation on the surface was not found (Figure S2B, Supporting Information).

Several thorns in the GNSs, and specifically the one presented in Figure 5B, show a distinct corrugated morphology. A previous study by Yuan et al,^{[ 29 ]} where GNSs were synthesized in a similar manner, proved that the corrugated thorns are twinned crystals elongating along the thorn's axis direction [211]. This shape was mainly found in the longer thorns, as was indeed found in our sample as well. It has been suggested that this shape results from the physical adsorption of AgCl onto the GNSs surface and the lack of surfactants such as CTAB.^{[ 29 ]} Our synthesized GNSs show a similar corrugated shape in some of the thorns (Figure 5B) that also correlates with the presence of AgCl.

Figure 5D presents the gold lattice structure of the GNSs thorns. The thorn contains a two‐fold twinned crystal structure with a [011]||[011] zone axis (Figure 5E,F). These findings align with the results described by Yuan et al.^{[ 29 ]} Moreover, an undefined layer was detected on the thorn's surface. We believe that it is the Ag‐AgCl layer as identified using EDS. The absence of a clear lattice structure of the layer is probably because it is a very thin layer and the misalignment with the zone axis. Notably, achieving alignment of the zone axis for gold was challenging, making it even more difficult for the layer.

To gain better insight into the composition of the GNSs surface, we have performed HR‐XPS measurements. Spectra were recorded at three different time points (0 s, 30 s, 60 s), showing a reduction in C1s and O1s signals over time, which correlates to HS‐PEG‐COOH removal from the surface, alongside an increase in Au4f signal (Figures S4A–C, Supporting Information). The best measurement was obtained at 30 s, and the results for Ag and Cl are presented in Figure S4D,E,F, Supporting Information, respectively, for this time point. Ag appears to be present at binding energy (B.E.) of Ag3d 5/2 of 368.0 eV. For reference, B.E. for metallic Ag is 368.2 eV, and for AgCl is 368.1 eV,^{[ 46 ]} making the differentiation between these two species very challenging. The kinetic energy (K.E.) of AgMN1 M4NN was measured at 357.7 eV, resulting in an Auger parameter of 725.7 eV, which is between the metallic Ag (726.1 eV) and the AgCl (723.5 eV).^{[ 46 ]} For Cl2s (Figure S4F, Supporting Information), no distinct peaks were detected. However, the sensitivity (S) ratio of S(Cl2s):S(Ag3d 5/2) is 0.16, and if even all of the Ag was in AgCl form, we would expect a peak height of ≈400cps of Cl2s signal. The noise level in that region is in the same order as the potential peak magnitude, meaning that in this system, it is challenging to observe Cl because it is under the detection limit. To conclude, HR‐XPS reassures the presence of an Ag metallic layer on the GNS surface, while suggesting that AgCl exists in small quantities close to the detection limit. While in HR‐TEM we measure a specific area, the HR‐XPS captures a much bigger area.

In the next step, we synthesized GNSs at different temperatures and varying delay periods before introducing AgNO₃ to the reaction solution. We then examined the influence of these parameters on particle concentration, average size and morphology compared to the cases where AgNO₃ is absent. HS‐PEG‐COOH was added after allowing the GNSs to grow for 60 s. Figure S5, Supporting Information presents the GNSs morphology visualized by HR‐SEM and the average size measured via NTA. Once again, various alterations in morphology and size can be observed while varying the temperature and the delay time. The results reveal an interesting effect on particle concentration, as seen in Figure 6A. For delay times ranging from 2 to 10 s, we observed a concentration increase of about two orders of magnitude compared to the GNPs that were synthesized without AgNO₃. Moreover, for a 0 s delay, where both HAuCl₄ and AgNO₃ were already present in the solution before adding AA, we observed a decreased GNSs concentration compared to conditions without AgNO₃. Additionally, the results show a gradual increase in concentration as the temperature rises.

Figure 6.

A) Concentration of GNSs synthesized at different temperatures (3, 10, 15, or 35 °C) and various delay periods of adding AgNO₃ (0, 2, 5, 8, or 10 s) and GNPs synthesized without AgNO₃, as measured by NTA. (B) HR‐PXRD patterns of GNSs prepared at 3 or 40 °C, and without AgNO₃ at 40 °C (Binning 0002). The Miller indices represent gold reflections, according to ICDD card number 00‐004‐0784. The black rectangle is the chosen enlarged area that is presented in (C). С) HR‐PXRD patterns that emphasize the presence of AgCl (111) and (200) peaks (marked with ♥) and NaCl contamination (200) peak (marked with ♦).

We employed synchrotron HR‐PXRD to study the structural features of the GNSs. We measured two GNSs samples synthesized at 3 or 40 °C, with a 2 s delay prior to adding AgNO₃. As shown previously, the GNSs formed at 3 °C are the most uniform, while at 40 °C are mostly spherical NPs (Figure 1). For comparison, GNPs were synthesized at 40 °C without AgNO₃. The HR‐PXRD patterns are presented in Figure 6B, showing clearly the characteristic diffraction peaks of crystalline Au and in Figure 6C the (111) and (200) reflections of AgCl (according to ICDD card number 00‐006‐0480). The intensities of these peaks are very low, but we could detect them due to the advantages and uniqueness of the HR‐PXRD technique. To our knowledge, it is the first time that AgCl diffraction peaks were detected using an X‐ray method, confirming the presence of AgCl in the GNSs synthesized by this approach. The latter finding may shed light on the nucleation and growth mechanisms of the GNSs.

We used Rietveld refinement to extract the gold lattice parameter, coherence length, and microstrains accurately. Additionally, we have calculated the lattice distortion according to Equation 1:

$$\frac{\mathrm{\Delta }a}{a_{AuICDD}}=\frac{a_f-a_{AuICDD}}{a_{AuICDD}}\ast 100$$ (1)

where a_f is the fitted lattice parameter and a_{Au ICDD} is the known gold lattice parameter (according to ICDD card number 00‐004‐0784). The extracted values are summarized in Table 1 . We observed an increase in the coherence length and microstrains with the increase of the temperature and in the absence of AgNO₃.

Table 1.

Rietveld refinement results of the GNSs.

Sample	Fitted lattice parameter (S.D.) [Å]	∆a/a(Au ICDD) [%]	Coherence length ± S.D. [nm]	Microstrains ± S.D.	Au phase fraction ± S.D. [%]
3 °C	4.07878(8)	0.004	18.3 ± 0.2	9589 ± 173	99.2 ± 0.2
40 °C	4.07966(8)	0.026	23.5 ± 0.3	12 352 ± 187	99.2 ± 0.2
40 °C without AgNO3	4.08072(7)	0.052	31.3 ± 0.5	14 000 ± 161	100

2.4. Effect of Temperature on Pure GNPs Synthesis

Our experimental approach involved the same synthesis method that was used for GNSs, except for the use of AgNO₃. Shortly, we prepared the deionized water (DIW) solutions with HCl and AA, stabilized the solutions at different temperatures, and then added a fixed amount of HAuCl₄ and let the GNPs grow over different periods. Here, we achieved pure GNPs, with no silver mixed in the structure. The achieved morphologies are presented in Figure 7 . One can observe surface pattern alterations depending on the temperature. At higher temperatures of 35 and 45 °C, the GNPs exhibit mostly a perfect spherical shape (Figure 7, two bottom rows), while lower temperatures (3, 10, or 15 °C) induce a surface pattern, which shows a gradual smoothening of the surface over time (Figure 7, three upper rows). Moreover, the results show an inverse correlation between the average size of the GNPs and the temperature, previously confirmed using the DLS and NTA measurements, which are presented in Figure 8A,B, respectively. The difference in the GNPs sizes measured by the two methods arises from their methodological disparities. As mentioned above, DLS captures the fluctuations in light scattering, which translates to hydrodynamic size, while NTA tracks the Brownian motion of individual particles, captured via laser beam. Therefore, it is reasonable to expect slight changes between the results obtained by the two methods. Figure 8C shows a significant increase in the particle concentration with the increase in temperature. Figure S6, Supporting Information presents the measured concentration values on a regular scale. For temperatures 15, 35 and 45 °C, the average size and concentration remain almost constant throughout all time intervals, which means the whole process of nucleation and growth finishes in less than 5 s. For 3 °C, the concentration increases with time while the average size decreases, indicating the appearance of new particles of smaller size. A similar effect, but to a lesser extent, is observed at 10 °C.

Figure 7.

HR‐SEM images of GNPs synthesized while varying the temperature conditions (3, 10, 15, 35, or 40 °C) and the delay time before adding HS‐PEG‐COOH (5, 10, 60, 180, or 600 s). AgNO₃ was not used.

Figure 8.

GNPs' average size and concentration at different synthesis temperatures and process longitudes are presented. A) Hydrodynamic diameter of GNPs measured using DLS (n = 3, 13 runs each).B) GNPs size measured using NTA. (n = 3) C) GNPs concentration in solution (Ln scale), measured using NTA. The average and standard deviation values are presented.

3. Discussion

In this work, we investigated the formation of GNSs through a simple and fast method that excludes toxic materials. In this reaction, AA is used to reduce [AuCl₄]⁻ ions, and AgNO₃ is added after a few seconds, promoting the anisotropic growth of the gold in a thorn‐like morphology. We have examined the influence of temperature conditions during the synthesis and AgNO₃ concentration on the GNSs's morphology, average size, uniformity, absorbance, stability, and chemical composition. In addition, delay time prior to adding AgNO₃ was inspected and found to be one of the main factors that control the final properties of the GNSs. The results show how subtle alterations in the synthesis conditions can greatly affect the final product. Furthermore, it is the first time we have revealed that there is an Ag layer on the GNSs surface, followed by both Cl and Ag atoms, as measured by HR‐TEM. The AgCl phase was confirmed in tiny quantities by HR‐PXRD, which supports the presence of an AgCl outer layer on the GNS. Although there is a possibility that these X‐Ray signals are due to AgCl NPs that might form as a byproduct, we believe it resulted from the outer layer of the GNSs due to the following reasons: (1) the presence of small amounts of crystalline AgCl NPs would lead to more profound peaks in HR‐PXRD method; (2) no additional morphology or size distribution that might correlate to AgCl NPs was found using the HR‐SEM, DLS and NTA methods; (3) the AgCl synthesis is slower (minutes and not few seconds as in the case of GNPs) and sometimes includes stabilizing agents and templates in the synthesis approach.^{[ 47} , ⁴⁸ , ⁴⁹ , ^{50 ]}

To comprehend the mechanism of GNSs synthesis, we suggest examining the chemical reactions. The overall chemical reaction between HAuCl₄ and AA, which is responsible for the nucleation and growth of GNPs is described as^{[ 51 ]}

$$2HAuC{l}_4+3C_6{H}_8{O}_6\to 2Au+3C_6{H}_6{O}_6+8HCl$$ (2)

The distribution of gold and AA precursors in aqueous solution is highly influenced by pH. Previous studies have confirmed the effect of pH on the final GNP morphology output.^{[ 30 ]} By adding HCl and AA to 10 mL DIW, the pH is ≈3. Therefore, the most stable precursors in this condition are [AuCl ₄]⁻ and H ₂ AA (L‐ascorbic acid). HAA ⁻ (ascorbate anion) is another possible precursor of AA, which is present in much smaller concentrations. AA is well‐known for its ability to react with various metals and form complexes.^{[ 52 ]} The rate of the reduction reactions of Au³⁺ to Au¹⁺ ([AuCl ₄]⁻ to [AuCl ₂]⁻, respectively) by AA is usually attributed to the GNPs nucleation rate.^{[ 51} , ^{53 ]} The reactions that are happening in parallel are^{[ 51 ]}

$${\left[AuC{l}_4\right]}^{-}+H_{2}AA\to {\left[AuC{l}_2\right]}^{-}+DHA+2HCl$$ (3a)

$${\left[AuC{l}_4\right]}^{-}+HA{A}^{-}\to {\left[AuC{l}_2\right]}^{-}+DHA+2HCl$$ (3b)

where DHA is dehydroascorbic acid, which is the oxidized form of L‐ascorbic acid.

However, the formation of the initial Au‐nuclei from [AuCl₂]⁻ ions still needs to be thoroughly understood. A similar GNPs synthesis method was studied by T. Yao et al.^{[ 54 ]} where Au nanoclusters were synthesized through the reduction of HAuCl₄ by sodium citrate in an aqueous surfactant solution (polyvinyl pyrrolidone), at 70 °C for 260 min. Using X‐ray absorption fine structure spectroscopy, they determined that [AuCl₄]⁻ ions were partially reduced to [AuCl₃]⁻, and several ions were combined to form Au_nCl_n+x complex clusters with elongated Au‐Au bonds. They supposed that two dispersed [AuCl₃]⁻ ions were first connected via the Au‐Au pair with the bond length significantly larger than that of the bulk Au, yielding Cl₃‐Au‐Au‐Cl₃‐dimers which could then react with an additional [AuCl₃]⁻ ion to form trimers. Analogously, the higher‐order Au chains and small Au clusters could be formed in which the Au atoms are mostly coordinated with other Au atoms. At a certain size, these clusters transform into GNPs surrounded by some Cl⁻ ions. We assume that in the case of [AuCl₂]⁻ ions, a very similar mechanism of GNP formation is realized: it starts from the formation of Cl₂‐Au‐Au‐Cl₂ dimers and continues via formation of trimers and then larger chains surrounded by some Cl⁻ ions, as schematized in Figure 9 . These clusters can be considered as stable pre‐nucleation clusters,^{[ 55 ]} which upon attaining a critical size undergo phase transformation in stable GNPs. The latter process can be described in the frame of classical nucleation theory (see the next Model section).

Figure 9.

Schematic representation of the formation of pre‐nucleated Au nanoclusters via formation of Cl₂‐Au‐Au‐Cl₂‐ dimers, trimers and larger chains. Created in BioRender. Pokroy Lab (2025) https://BioRender.com/r73s821.

Our proposed mechanism resembles closely the four‐step seed‐mediated growth process described by Polte et al.^{[ 36 ]} for GNPs formation using the Turkevich method. In that approach, sodium citrate acts both as reducing agent and a stabilizer of the GNPs, while in our synthesis, AA serves as the reducing agent with HS‐PEG‐COOH as a stabilizer. This distinction is crucial because it leads to faster kinetics in our approach; all four steps happened in less than 2 min, while in the sodium citrate‐based process, only the nucleation phase might take a few minutes. The temperature conditions are also notably different. With AA, there is no need for more than 45 °C, compared to sodium citrate, which requires 75–85 °C for the reaction. Another key difference is the particle size. The Turkevich method produces relatively small GNPs, while our approach allows the synthesis of larger GNPs by varying the temperature.

The change of the gold amount precipitated in the gold particles can be measured via the concentration of gold ions in the solution, as has been done in the work of Luty‐Błocho et al.^{[ 51 ]} They described the change of [AuCl ₄]⁻ concentration ($C_{{[AuC{l}_4]}^{-}}$) under conditions of large excess of reductant in comparison with Au³⁺ ions concentration, by the first order reaction

$$\frac{d{C}_{{\left[\mathrm{AuC}{l}_4\right]}^{-}}}{\mathrm{dt}}=-k_{\mathrm{obs}}{C}_{{\left[\mathrm{AuC}{l}_4\right]}^{-}}$$ (4)

where $k_{obs}=k\ast C_{H_{2}AA}$, k is the rate constant of the reaction, $C_{H_{2}AA}$ is concentration of H₂AA in solution at time t. In accordance with Equation (4), concentration of the Au³⁺ complex ions $,C_{A{u}^{3+}},$ in their reduction to Au¹⁺ ions, decreases exponentially:^{[ 51 ]}

$$C_{A{u}^{3+}}=C_{0,A{u}^{3+}}\exp \left(-k_{obs}t\right)$$ (5)

where $C_{0,A{u}^{3+}}$ is an initial concentration of Au³⁺ ions.

The temperature dependence of the coefficient k_obs has been experimentally measured in the range of temperatures from 15 to 35 °C, and the exponential dependence $k_{obs}=k_{0,obs}\exp (-\frac{E_{LB}}{k_{b}T})$ was found with the activation energy E_LB = (41.0 – 44.5) kJ mole⁻¹ depending on pH (k_b is Boltzmann's constant); it has been interpreted as the energy barrier to transfer electrons during corresponding chemical reaction, where the gold ions removed from the solution are attached to the gold particles. The change of the total particles’ volume, V _Au, tot is proportional to the change of the gold ions concentration in the solution

$$\frac{d{V}_{Au,tot}}{dt}=-\beta k_{obs}{C}_{A{u}^{3+}}\exp \left(-k_{obs}t\right)$$ (6)

where β is temperature independent proportionality coefficient.

However, this expression cannot explain the temperature dependencies of the particle concentration and the GNP average size and predict the behavior of the GNPs system during the reduction reaction. Therefore, to determine optimal conditions for synthesizing GNSs, a simple model that considers the nucleation process is presented below, based on the results shown in Figure 7 and Figure 8.

3.1. The model

Nucleation and growth of GNPs can be described by the following simplified model. Let us assume that the nucleation rate, I, is approximately constant during the initial period, Δt₀, after which it turns out to zero. The volume fraction, ƒ, of the nucleated GNPs after the time Δt₀

$$f=I\underset{0}{\overset{\mathrm{\Delta }{t}_0}{\int }}V\left(\mathrm{\Delta }{t}_0-\tau \right)d\tau$$ (7)

where V(Δt ₀ − τ) is the volume of particles nucleated at the moment τ. Assuming a linear growth of the spherical particle radius with the rate ν

$$V\left(\mathrm{\Delta }{t}_0-\tau \right)=\frac{4\pi }{3}{\nu }^3{\left(\mathrm{\Delta }{t}_0-\tau \right)}^3$$ (8)

after integration of Equation (7) one can find

$$f=\frac{\pi }{3}{v}^{3}I\mathrm{\Delta }{t}_0^4$$ (9)

which is followed to JMAK equation for assumed simplifications.^{[ 56 ]} Cessation of nucleation occurs due to a decrease of gold supersaturation in the solution when the volume fraction of gold particles reaches a certain critical value, f₀ . The corresponding time when this critical fraction is achieved

$$\mathrm{\Delta }{t}_0={\left(\frac{3f_0}{\pi {\nu }^{3}I}\right)}^{\frac{1}{4}}$$ (10)

The concentration of the GNPs (C_Au ) achieved during this time is the following

$$C_{Au}=I\mathrm{\Delta }{t}_0={\left(\frac{3f_0}{\pi }\right)}^{\frac{1}{4}}{\left(\frac{I}{\nu }\right)}^{\frac{3}{4}}$$ (11)

and the average radius ($\overline{R}$) of the particles

$$\overline{R}={\left(\frac{3f_0}{\pi }\right)}^{\frac{1}{4}}{\left(\frac{\nu }{I}\right)}^{\frac{1}{4}}$$ (12)

The GNP radius is assumed to be half of the average size measured in the NTA method (Figure 8B).

Assuming exponential temperature dependencies for the nucleation rate, I, and the rate of the average particle size increase

$$I=I_0\exp \left(-\frac{W^{\ast }}{k_{B}T}\right)$$ (13)

$$\nu ={\nu }_0\exp \left(-\frac{E^{\ast }}{k_{B}T}\right)$$ (14)

where W* is the energy barrier for particle nucleation, E* is the activation energy of particle growth, I₀ and ν₀ are the corresponding pre‐exponential factors, one can write:

$$C_{Au}=C_0\exp \left(-\frac{3E_A}{4k_{B}T}\right)$$ (15)

$$\overline{R}=\overline{R_0}\exp \left(\frac{E_A}{4k_{B}T}\right)$$ (16)

where E_A = W*  − E*, $C_0={(\frac{3f_0}{\pi })}^{\frac{1}{4}}{(\frac{I_0}{{\nu }_0})}^{\frac{3}{4}}$ and $\overline{R_0}=\frac{C_0{\nu }_0}{I_0}$. That means that the activation energies for concentration and the average radius differ by three times and are opposite in sign.

It should be noted that after nucleation is stopped, the growth of the particles may continue since supersaturation still exists while it is not sufficient for nucleation. In this case, the average radius continues to grow with a new rate, ν₁:

$${\nu }_1={\nu }_{01}\exp \left(-\frac{E^{\ast }}{k_{B}T}\right)$$ (17)

where ν₀₁ is pre‐exponential constant. During some additional time, Δt ₁, the final radius will be the sum:

$$\overline{R}=\nu \mathrm{\Delta }{t}_0+{\nu }_1\mathrm{\Delta }{t}_1={\left(\frac{3f_0}{\pi }\right)}^{\frac{1}{4}}{\left(\frac{\nu }{I}\right)}^{\frac{1}{4}}+{\nu }_1\mathrm{\Delta }{t}_1$$ (18)

Since the second term in Equation (18) increases with temperature (unlike the first term), the effective value of activation energy measured using Equation (16) can be smaller than the third part of the activation energy measured from Equation (15).

As can be seen from experimental dependencies of particle concentration on time of growth at different temperatures (Figure 8C), the concentrations reach a saturated value at 10 °C and above, while at 3 °C it continues to grow. At the same time, the amount of precipitated gold can be estimated by the value C*R³ (Figure 10A). These amounts reach a constant value of about 2*10¹⁶ at 10 °C and above, while it is about twice smaller at 3 °C even after 600s. Therefore, only the data for 10 °C and above can be used for estimation of activation energies using Equations (15) and (16).

Figure 10.

A) The precipitated amount of gold as a function of experimental time at different temperatures. B) Activation energy estimation from the temperature dependence of saturated concentrations reached during 60 to 600 s at temperatures 10–45 °C. The approximation is given by the equation lnC = ‐A(1/T) + B, with A = 6164 ± 52 which corresponds to activation energy E_A = 51.25 ± 0.4 kJ mole⁻¹. C) Dependence of GNPs average size on temperature, after 5 or 10 s of reaction, giving the activation energies −15.1 and −12.4 kJ mole⁻¹, correspondingly; D) Dependence of GNPs average size on temperature, after 60, 180 or 600 s of reaction, giving the average activation energy ≈−8.5 kJ mole⁻¹. The graphs are based on NTA measurements of GNPs average size and concentration.

Using the experimental data obtained, the temperature dependence of the saturated concentration, $\stackrel{\sim }{C}$, can be described as following:

$$\stackrel{\sim }{C}=9.3\ast {10}^{16}\exp \left(-\frac{6164.4}{T}\right)=9.3\ast {10}^{16}\exp \left(-\frac{51.25kJ}{R_{g}T}\right)$$ (19)

where R_g is the gas constant.

Activation energy estimated from the temperature dependence of saturated concentrations (Figure 10B) varies depending on experimental time. However, after 60 s, it varies only slightly: it gives ≈51.6 kJ mole⁻¹ after 60 s, ≈51.3 kJ mole⁻¹ after 180 s, and after 600 s, it is ≈50.8 kJ mole⁻¹.

Activation energy found from the temperature dependence of the average radius (Figure 7C) varies between −(12 – 15) kJ mole⁻¹ after 5 s and 10 s (Figure 10C) and between −(8–9) kJ mole⁻¹ after 60–600 s (Figure 10D).

It should be noted that for temperatures ≥10 °C, the average radius reached a certain saturated value characteristic to a temperature already after 10 s, while at 3 °C, after initial growth, it decreases during all measured time of 600 s. This decrease in the average radius can be explained by the permanent nucleation of new particles during this time, as confirmed by an increase of concentration during 600 s at 3 °C (Figure 8C). The same effect is observed at 10 °C and, to a lesser extent, at 15 °C for the first 10 s. Therefore, it is reasonable to use the time 10 s for estimation of corresponding activation energy, which gives −12 kJ mole⁻¹. Therefore, the equation describing the temperature dependence of the saturated radius can be written as follows:

$$R=0.73\exp \left(\frac{1490.4}{T}\right)\approx 0.73\exp \left(\frac{12.4kJ}{R_{g}T}\right)$$ (20)

As mentioned above, the temperature dependence of the saturated concentration gives the activation energy ≈51 kJ mole⁻¹, and therefore, the expected activation energy for the average particle radius, according to the proposed model, should be about −17 kJ mole⁻¹. However, the measured activation energy equals ≈−12 kJ mole⁻¹, a little bit smaller. As discussed above, it can be explained by continuation of particle growth after stopping the nucleation process (see Equation (18)) Therefore, the developed model is generally validated, with $E_A=W^{\ast }-E^{\ast }\approx \frac{4}{3}\ast 51.2=68.3\mathrm{kJ}\mathrm{mol}{\mathrm{e}}^{-1}$.

The duration of nucleation process can be now estimated from the Equation (10):

$$\mathrm{\Delta }{t}_0={\left(\frac{3f_0}{\pi {\nu }^{3}I}\right)}^{\frac{1}{4}}={\left(\frac{3f_0}{\pi {\nu }_0^3{I}_0}\right)}^{\frac{1}{4}}\exp \left(\frac{W^{\ast }+3E^{\ast }}{4k_{B}T}\right)$$ (21)

Assuming E* =  αW*, the activation energy E_A = W* (1 − α) and $W^{\ast }=\frac{E_A}{1-\alpha }$, $E^{\ast }=\frac{\alpha E_A}{1-\alpha }$ and then, from Equation (21):

$$\mathrm{\Delta }{t}_0=\mathrm{\Delta }{t}_{00}\exp \left[\frac{E_A\left(1+3\alpha \right)}{4k_{B}T\left(1-\alpha \right)}\right]$$ (21a)

Figure 11 presents the temperature dependence of this duration for the experimental value of E_A = 68.2 kJ mole⁻¹ and different values of α. The value of Δt ₀₀ was chosen assuming the nucleation process was stopped after Δt ₀ = 10 s at 15 °C.

Figure 11.

Nucleation time for different activation energies of particle growth, where E* = αW*, and α variation.

As can be seen, for temperatures below 15 °C, the nucleation process exceeds 10 s, while for higher temperatures, the nucleation stops after a period Δt < 10 s. This result should be considered in determining the optimal conditions for obtaining GNSs in this synthesis method.

The activation energy for the particle growth, E*, can be estimated using the results of Luty‐Błocho et al.^{[ 51 ]} mentioned above (Equation (6)). The change of gold amount in the solution had been measured at temperatures 15–35 °C during ≈10 s, that is, under the conditions when the GNPs nucleation process had already been stopped, and the gold ions were attached to the existing particles.

Assuming the total particles’ volume is proportional to the cube of average particle radius, after integration of Equation (6), one can write:

$$\frac{{\overline{R}}^3\left(t\right)-{{\overline{R}}_0}^3}{{{\overline{R}}_{\max }}^3-{{\overline{R}}_0}^3}=1-\exp \left(-k_{\mathrm{obs}}t\right)$$ (22)

where ${\overline{R}}_0$ and ${\overline{R}}_{max}$ are the initial and the maximum average radius of GNPs. Initial linear time dependence of average radius assumed above can be realized for k _0,obs ≪ 1 and ${\overline{R}}_0=0$, when ${\overline{R}}^3(t)={(vt)}^3\approx {{\overline{R}}_{max}}^3{k}_{obs}t$, and then:

$$v\approx {\overline{R}}_{max}{\left(\frac{k_{obs}}{t^2}\right)}^{\frac{1}{3}}$$ (23)

Using Equation (9) for the temperature dependence of the rate v, and Equation (23), one can find $E^{\ast }\approx \frac{E_{LB}}{3}$, therefore, E* ≈ (13.7 − 14.8) kJ mole⁻¹. Using these values, and E_A = W*  − E* ≈ 68.3 kJ mole⁻¹, one can obtain W* ≈ 82.5 ± 0.5 kJ mole⁻¹, and E* ≈ 0.18W*, that is, α ≈ 0.18.

Let us check if the obtained energy barrier for nucleation of GNP, W*, from a supersaturated liquid solution is reasonable. From the classical theory of homogeneous nucleation, the energy barrier for nucleation of spherical particles can be evaluated as follows:^{[ 57 ]}

$$W^{\ast }=\frac{16\pi {\gamma }_{Au-sol}^3}{3\mathrm{\Delta }{F}^2}$$ (24)

where γ _{Au − sol} is the surface energy of the GNPs in the solution, ΔF is the free energy gain per unit volume when gold atoms are transferred from the solution to the particle, $\mathrm{\Delta }F=\frac{\mathrm{\Delta }{E}_1}{\mathrm{\Omega }}$, where ΔE ₁ is the energy change caused by transferring one gold atom from the solution to the GNP, Ω is the atomic volume of GNP. Using the reasonable values for the surface energy ${\gamma }_{Au-sol}=(0.82÷1.2)\frac{J}{m^2}=(0.05÷0.075)\frac{eV}{A^2},$ and for $\mathrm{\Delta }{E}_1=-(1÷2)(\frac{eV}{atom})$(the sublimation energy for gold is 3.82 eV atom⁻¹), Ω  =  12.5 A ³, one can obtain: W* =  (0.65÷1.46)/(1÷4) eV. The experimental value estimated above, $W^{\ast }\approx 82\frac{kJ}{mole}$ can be easily obtained from this estimation. For example, for ${\gamma }_{Au-sol}=0.9\frac{J}{m^2}$, $\mathrm{\Delta }{E}_1=-1.0\frac{eV}{atom}$, it yields $W^{\ast }\approx 0.82eV=79.4\frac{kJ}{mole}$. Therefore, the nucleation energy barrier may indeed provide the observed activation energy. Table 2 presents the summary of all estimated energies.

Table 2.

Activation energies estimated from the temperature dependencies of GNPs concentration and average radius (kJ mole⁻¹).

EA ± S.D. [kJ mole−1]	W* ± S.D. [kJ mole−1]	E* ± S.D. [kJ mole−1]
68.3 ± 0.5	82.5 ± 0.5	14.3 ± 0.5

This model supports the experimental approach to GNSs synthesis and provides a strong explanation for the experimental results. As discussed, the nucleation rate of gold is slower at lower temperatures, meaning more gold ions remain in the solution for the later step of the process and can be used for the thorns' growth after adding the AgNO₃. Based on both our experimental results and the model, we can assume that only at 3 and 10 °C, there are sufficient gold ions left in the solution during the first 10 s of the process. Therefore, using a temperature higher than 10 °C and waiting longer than 10 s before adding AgNO₃ will result in the formation of spherical GNSs's morphology, rather than promoting thorn growth or achieving nanostars. A similar conclusion was suggested by Phiri et al. as well.^{[ 58 ]} Furthermore, as can be seen from Figure 10A, there is no significant change in the total amount of gold participating after 60 s, meaning terminating the synthesis reaction of GNS after 60 s is reasonable.

In our experiments, the synthesis of GNSs involves introduction of AgNO₃ to the GNPs solution after a reasonable period of 2 or 5 s, causing an anisotropic growth of the gold thorns. However, as mentioned before, the full mechanism of this synthesis is still lacking understanding due to the multiple reactions that might occur simultaneously after introducing AgNO₃.

The chemical reaction between the silver ions and AA is described by:^{[ 59 ]}

$$2A{g}^{+}+H_{2}AA\to 2A{g}^0+DHA+2H^{+}$$ (25)

The HAA⁻ precursor plays a crucial role in the reduction of Ag⁺ and increases the reduction rate; However, this precursor is present in relatively low amounts in the solution with pH = 3.^{[ 59} , ^{60 ]} Moreover, in the presence of Cl⁻ anions, Ag⁺ cations will precipitate in the following manner:^{[ 29 ]}

$$A{g}^{+}+C{l}^{-}\to AgCl$$ (26)

Additionally, after the formation of Ag⁰ (Equation (25)), a galvanic reaction might occur between Ag⁰ and [AuCl₄]⁻ as following:^{[ 61 ]}

$$3A{\mathrm{g}}^0+{\left[AuC{l}_4\right]}^{-}\to 3A\mathrm{gCl}+A{u}^0+C{l}^{-}$$ (27)

Another possible reaction is between Ag⁰ and [AuCl₂]⁻, which requires only one silver atom instead of three as in Equation (27):^{[ 61 ]}

$$A{\mathrm{g}}^0+{\left[AuC{l}_2\right]}^{-}\to A\mathrm{gCl}+A{u}^0+C{l}^{-}$$ (28)

Equations ((26), (27), (28)) describe the formation of AgCl precipitates on the GNSs surface, leading to anisotropic growth until all the gold ions are precipitated. It is well known that AgCl is insoluble in water and in many other common solvents, as well as in gold.^{[ 62 ]} Thus, this property might help stabilize the GNS for extended periods with the support of the HS‐PEG‐COOH layer and prevent silver diffusion inside the gold core.

Surprisingly, in our experiments, after adding AgNO₃ to the synthesis solution and with a delay time of 2 to 10 s, the particle concentration of synthesized GNSs increased dramatically, by about two orders of magnitude, compared to the GNPs synthesized without silver (Figure 6A).

To explain this significant increase in the GNS concentration, one can assume three possibilities: 1) easier and faster formation of Au‐nanocluster, in the presence of AgNO₃ consuming chlorine anions surrounding Au‐nanoclusters; 2) decrease of the energy barrier for GNP nucleation, W*, due to decrease of interface energy γ _{Au − sol} , or/and increase of the free energy gain ΔF due to transferring gold atoms from the solution to the particle; 3) decrease of the GNP growth rate (as presented in Equation (14)), that can be probably due to increase of the activation energy of particle growth, E*, which, in turn, may be caused by formation of insoluble AgCl on the surface of growing GNPs. The two latter factors should result in a decrease of activation energy E_A , controlling the saturated value of GNPs concentration, as shown in Equation (15).

The temperature dependencies of GNSs concentrations obtained with AgNO₃ introduced to the reaction solution after varying delay periods are given in Figure S7, Supporting Information. Figure S7, Supporting Information demonstrated a substantial decrease of the activation energy, E_A , as compared to the case of pure gold (from ≈0.55 to ≈0.3 eV, and even to 0.05–0.08 eV). We suppose that this decrease is primarily due to an increase of effective energy barrier E* for attaching new gold atoms to the existing GNPs. This is validated by the change of a spherical shape of the GNPs in the case of pure gold to a thorn‐like GNS, when AgNO₃ was added to the solution. The latter is dictated by partial coating of GNP surface with Ag and the insoluble AgCl clusters. The presence of chlorine on the surface of the gold thorn's edges was also confirmed by EDS analysis in the HAADF‐STEM mode (Figure 3) and HR‐PXRD measurements. Therefore, the observed increase in the GNSs concentration in the presence of AgNO₃ most probably is connected with the increase in the duration of the nucleation process, Equation (21). At the same time, the total amount of gold in the GNPs almost does not change, although their number, shape, and size undergo significant changes (see further explanation in Figure S8, Supporting Information).

Furthermore, for a delay period of 0 s, we observe a reduction in the GNSs concentration compared to GNPs without AgNO₃ use, with a particularly noticeable change at 10 and 15 °C. Under these conditions, nucleation is inhibited to some extent. This can be explained by a competition between [AuCl ₄]⁻ ions and immediately present Ag⁺ ions for reaction with AA precursors, resulting in a reduced availability of AA for reaction with [AuCl₄]⁻ to form [AuCl₂]⁻ ions (reactions (3a,b)) and initiate the GNP nucleation.

4. Conclusions

In this study, a range of gold‐based NPs was successfully synthesized, including pure GNPs and bimetallic gold‐silver nanostars, under varying synthesis conditions. Our findings highlight the significant impact of temperature on the final outcomes, particularly in terms of morphology, size, and particle concentration. Additionally, the concentration and timing of AgNO₃ addition during synthesis are crucial, as they greatly affect the characteristics of the resulting NPs. The temperature dependencies of GNSs concentration and their average size were described in the frame of the developed analytical model, which enabled us to predict the optimal conditions for GNSs synthesis. For the synthesis of highly uniform GNSs, it is essential to maintain lower temperatures and introduce AgNO₃ before the complete precipitation of [AuCl₄]^⁻ ions. Delaying the addition of AgNO₃ diminishes the effect of silver on the synthesis, preventing the desired morphological alterations. Our study also demonstrates that silver coats the gold‐core NPs surface, followed by accumulation of silver and chloride atoms as AgCl on the GNSs thin outer layer, slowing the growth rate of the GNPs and extending the nucleation process for new particles. The insights gained in this work contribute to a deeper understanding of the synthesis and nanostructure formation of bimetallic GNSs, laying the groundwork for more controlled, precise, and scalable fabrication techniques in nanomaterials science.

5. Experimental Section

Materials

HAuCl₄ was supplied by Alfa Aesar (Ward Hill, MA, USA). Silver nitrate (AgNO₃) was purchased from Wieland Edelmetalle GmbH (Pforzheim, Germany). L‐ascorbic acid was purchased from Fisher Scientific (Loughborough, UK). Heterobifunctional thiol‐poly(ethylene glycol)‐carboxylic acid (HS‐PEG‐COOH, molar weight of 2000 g mL⁻¹) was supplied by Laysan Bio Inc. (Arab, AL, USA). Concentrated HCl was purchased from Bio‐Lab Ltd. (Jerusalem, Israel). Si (100) wafers were purchased from Sil'tronix Silicon Technologies (Archamps, France).

Synthesis of the GNSs

The GNSs seedless synthesis was adopted from the previous work of Phiri et al.^{[ 58 ]} and Cheng et al.^{[ 27 ]} with several modifications. Briefly, 10 µL of 1 M HCl and 50 µL of a 100 mM AA (reducing agent) stock aqueous solution were added to 10 ml of DIW to obtain a solution of pH 3, in a 20 mL glass bottle. The temperature was stabilized by using refrigerated bath circulator (WBL‐100, mrc – Laboratory Instruments, Holon, Israel) to 3, 10, 15, 20, 25, 30, 35, or 40 °C. Under gentle magnetic stirring (300 rpm, magnetic bar size of 3 mm x 8 mm), 50 µL of 50 mM HAuCl₄ was added to the mixture and after 2 or 5 s, 10 µL of 15mM AgNO₃ in water was incorporated into the system, resulting in a rapid color transition from transparent to blue‐green color. For the final step, after 60 s since the reaction started, 250 µL of 2.5 mM of HS‐PEG‐COOH in water was added to the system. HS‐PEG‐COOH forms a stable covalent bond via the Au‐S interaction, which effectively stops the GNS growth. GNS dispersions were stirred (30 rpm) for 2 h at room temperature and then stored at 4 °C. Moreover, GNS were synthesized in varying AgNO₃ concentrations to 2, 5, 15, 50, and 100 mM. After stabilizing the temperature to 3 or 20 °C, the synthesis started as described previously, 10 µL of AgNO₃ was added after 2 s, followed by 250 µL of 2.5 mM of HS‐PEG‐COOH after 60 s. To investigate the influence of the delayed incorporation of AgNO₃ to the reaction mixture, GNSs were synthesized at temperatures of 3, 10, 15, and 35 °C, and 10 µL of 15 mM of AgNO₃ was added after 0, 2, 5, 8 or 10 s.

For developing an analytical model for the nucleation and growth and determining the corresponding activation energies, GNPs were synthesized using the same method described above, with a few modifications. In this case, the solutions were stabilized at 3, 10, 15, 35 and 45 °C, AgNO₃ was not used in the synthesis, and only HS‐PEG‐COOH was added after 5, 10, 60, 180 or 600 s.

Characterization of the GNSs—HR‐SEM

Samples were prepared by spraying GNS and GNP dispersions using N₂ on polished (001) Si substrate with native oxide after a thermal oxidation process at 1100 °C. The particles morphology was visualized by Zeiss Ultra Plus FEG‐SEM (Zeiss, Oberkochen, Germany) with an accelerating voltage of 4 keV and 3–4 mm working distance, using InLens signal.

Characterization of the GNSs—EDS

The atomic composition of the GNSs was performed using the EDS detector in the HR‐SEM system with an accelerating voltage of 8 KeV and a working distance of 8.5 mm by INCA (Oxford Instruments, High Wycombe, UK).

Characterization of the GNSs—DLS

The D_h and the PDI (an estimation of the nanoparticle size distribution) were measured by DLS at a scattering angle of 173°, utilizing a Zetasizer Nano‐ZS (Malvern Instruments, Malvern, UK). Each sample was measured three times, with 13 runs each. Results are expressed as average ± S.D. All samples were measured by Intensity.

Characterization of the GNSs—NTA

The size (D), size distribution, and concentration (expressed as particles mL⁻¹) of the GNSs were measured using NTA in a NanoSight NS300 system with a red laser of 638 nm (Malvern Instruments). The samples of Figure 3 measured in different NTA system (NanoSight NS500‐Zeta HSB, red laser of 638 nm, Malvern Instruments); Therefore, we expect a slight discrepancy in the obtained values between the systems.

Characterization of the GNSs—UV–vis spectrophotometry

The light absorption of the GNSs was measured by UV–vis spectrophotometry between the wavelengths of 300–800 nm, with a scan rate of 600 nm min⁻¹, on an Agilent Cary 100 UV–vis Spectrophotometer (Agilent Technologies, Santa Clara, CA, USA).

Characterization of the GNSs—HR‐TEM

2 mL of GNS or GNP suspension was washed three times with DIW by centrifugation (30 min, 8000 rpm) and redispersion in 2 mL DIW. Then, 20 µL of a GNS dispersion (synthesized at 3 °C, with 2 s delay before adding AgNO₃) or GNP dispersion (synthesized at 3 °C, without using AgNO_3, growth for 60 s), was dropped on a holy carbon coated 200 or 300 mesh Cu (SPI Supplies, West Chester, PA, USA) and was dried in a vacuum oven for 1 h, at room temperature. GNSs micrographs and EDS maps were acquired using probe‐corrected FEI/ThermoFisher Titan Cubed Themis G2 (60‐300) (ThermoFisher Scientific, Waltham, MA, USA) operated at an acceleration voltage of 200 KeV, 0.05 nA, using HAADF‐STEM mode. The microscope is equipped with a Bruker Dual‐X EDS detector (Bruker). Lattice imaging was acquired using HR‐TEM mode. The data was processed using Velox software (Thermo‐Fisher).

Characterization of the GNSs—HR‐PXRD

The measurements were conducted at the ID22 beamline of the European Synchrotron Radiation Facility (ESRF), in Grenoble, France. The sample preparation process involved synthesizing 40 mL of GNSs (synthesizing every 10 mL separately and then summing it all up). GNSs were synthesized at 3 and 40 °C and AgNO₃ addition after 2 s. A sample without AgNO₃ at 40 °C was used as a control. To ensure that the GNSs remain dispersed and to prevent agglomeration on the plastic walls during centrifugation, the GNSs suspension was divided into aliquots of 2 mL in Eppendorf tubes and washed three times with DIW (2 mL) by centrifugation (8000 rpm, 30 min). Subsequently, the washed GNSs suspension was collected in a 50‐mL Falcon tube, frozen at −80 °C and freeze‐dried (Labconco Free Zone 4.5 plus L Benchtop Freeze Dry System, Labconco, MO, USA), resulting in a dark‐blue fine powder. A borosilicate capillary was coated with vacuum grease (which does not diffract) and the dry GNSs were affixed to the grease. Then, the capillary was exposed to a beam in a continued rotating mode, at room temperature at a wavelength of 0.35426587 Å. Rietveld refinement^{[ 63 ]} was applied to accurately calculate lattice parameters, coherence length (estimation for the grain size) and microstrains, by using GSAS‐II software.^{[ 64 ]} The weighted R‐factor (Rw) in the Rietveld refinement is a value that quantifies how well the calculated model fits the experimental data. The lower Rw, the better the fit. The acquired Rw values are less than 10 for all refinements, which generally suggests a good fit.

Characterization of the GNSs—HR‐XPS

The GNSs sample (3 °C, 15 mM AgNO₃, 2 s delay before adding AgNO₃, HS‐PEG‐COOH after 60 s) was prepared as described for HR‐SEM samples, on 1cm*1cm Si wafer. HR‐XPS measurements were performed using ESCALAB QXi (Thermo Scientific, USA). Samples were irradiated with a focused X‐ray Al‐Kα monochromatic X‐ray source, beam size 900 µm. Cluster sputtering was performed, in order to reduce the organic material on the surface by using 500Ar⁺ cluster, 4keV, 2 mm raster. The sample was measured at three times: 0, 30, 60 s. The detection limit of the system is 0.05%. HR‐XPS survey spectra were recorded for the elements: C1s, Ag3d, AgMN1, Au4f, O1s, Cl2s.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Dobrynin D., Zlotver I., Polishchuk I., Kauffmann Y., Suharenko S., Koifman R., Kuhrts L., Katsman A., Sosnik A., Pokroy B., Controlled Synthesis of Bimetallic Gold‐Silver Nanostars: Atomic Insights and Predictive Formation Model. Small 2025, 21, 2410152. 10.1002/smll.202410152

Data Availability Statement

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