The main goal of this work is to elucidate the role of capping agents in regulating the size of palladium nanoparticles by combining in situ small angle x-ray scattering (SAXS) and ligand-based kinetic modeling.
The size, size distribution and stability of colloidal nanoparticles are greatly affected by the presence of capping ligands. Despite the key contribution of capping ligands during the synthesis reaction, their role in regulating the nucleation and growth rates of colloidal nanoparticles is not well understood. In this work, we demonstrate a mechanistic investigation of the role of trioctylphosphine (TOP) in Pd nanoparticles in different solvents (toluene and pyridine) using in situ SAXS and ligand-based kinetic modeling. Our results under different synthetic conditions reveal the overlap of nucleation and growth of Pd nanoparticles during the reaction, which contradicts the LaMer-type nucleation and growth model. The model accounts for the kinetics of Pd-TOP binding for both, the precursor and the particle surface, which is essential to capture the size evolution as well as the concentration of particles in situ. In addition, we illustrate the predictive power of our ligand-based model through designing the synthetic conditions to obtain nanoparticles with desired sizes. The proposed methodology can be applied to other synthesis systems and therefore serves as an effective strategy for predictive synthesis of colloidal nanoparticles.
Controlled synthesis of metallic nanoparticles is of great importance due to the large applications of nanostructured materials in catalysis, photovoltaic, photonics, sensors, and drug delivery1,2,3,4,5. To synthesize the nanoparticles with specific sizes and size distribution, it is vital to understand the underlying mechanism for the particle nucleation and growth. Nevertheless, obtaining nanoparticles with such criteria has challenged the nano-synthesis community due to the slow progress in understanding the synthesis mechanisms and the lack of robust kinetic models available in the literature. In 1950s, LaMer proposed a model for the nucleation and growth of sulfur sols, where there is a burst of nucleation followed by a diffusion-controlled growth of nuclei6,7. In this proposed model, it is postulated that the monomer concentration increases (due to the reduction or decomposition of the precursor) and once the level is above the critical supersaturation, the energy barrier for particle nucleation can be overcome, resulting in a burst nucleation (homogeneous nucleation). Owing to the proposed burst nucleation, the monomer concentration drops and when it falls below the critical supersaturation level, the nucleation stops. Next, the formed nuclei are postulated to grow via the diffusion of monomers towards the nanoparticles surface, while no additional nucleation events occur. This results in effectively separating the nucleation and growth in time and controlling the size distribution during the growth process8. This model was used to describe the formation of different nanoparticles including Ag9, Au10, CdSe11, and Fe3O412. However, several studies illustrated that the classical nucleation theory (CNT) cannot describe the formation of colloidal nanoparticles, in particular for metallic nanoparticles where the overlap of the nucleation and growth is observed1,13,14,15,16,17. In one of those studies, Watzky and Finke established a two-step mechanism for the formation of iridium nanoparticles13, in which a slow continuous nucleation overlaps with a fast nanoparticle surface growth (where growth is autocatalytic). The slow nucleation and fast autocatalytic growth were also observed for different types of metal nanoparticles, such as Pd14,15,18, Pt19,20, and Rh21,22. Despite recent advances in developing nucleation and growth models1,23,24,25, the role of the ligands is often ignored in the proposed models. Nevertheless, ligands are shown to affect the nanoparticles size14,15,26 and morphology19,27 as well as the catalytic activity and selectivity28,29. For example, Yang et al.30 controlled the Pd nanoparticle size ranging from 9.5 and 15 nm by varying the concentration of trioctylphosphine (TOP). In the synthesis of magnetic nanoparticles (Fe3O4), the size noticeably decreased from 11 to 5 nm when the ligand (octadecylamine) to metal precursor ratio increased from 1 to 60. Interestingly, the size of Pt nanoparticles was shown to be sensitive to the chain length of amine ligands (e.g., n-hexylamine and octadecylamine), where smaller nanoparticle size could be obtained using longer chain (i.e., octadecylamine)31.
The size alteration caused by different concentration and different types of the ligands is a clear evidence for the contribution of ligands in the nucleation and growth kinetics. Unfortunately, few studies accounted for the role of ligands, and in these studies, several assumptions were often made for the sake of simplicity, which in turn make these models applicable only for specific conditions32,33. More specifically, Rempel and co-workers developed a kinetic model to describe the formation of quantum dots (CdSe) in the presence of capping ligands. However, in their study, the binding of the ligand with nanoparticle surface is assumed to be at equilibrium at any given time32. This assumption might hold true when the ligands are in large excess. Our group recently developed a new ligand-based model14 which accounted for the binding of capping ligands with both the precursor (metal complex) and the surface of nanoparticle as reversible reactions14. In addition, our ligand-based model could potentially be used in other metal nanoparticle systems, where the synthesis kinetics seem to be affected by the presence of the ligands.
In the current study, we use our newly developed ligand-based model to predict the formation and growth of Pd nanoparticles in different solvents including toluene and pyridine. For our model input, in situ SAXS was utilized to obtain the concentration of nanoparticles and size distribution during the synthesis. Measuring both the size and concentration of particles, complemented by kinetic modeling, allows us to extract more precise information on the nucleation and growth rates. We further demonstrate that our ligand-based model, which explicitly accounts for the ligand-metal binding, is highly predictive and can be used to design the synthesis procedures to obtain nanoparticles with desired sizes.
1. Pd Acetate Recrystallization
CAUTION: This protocol involves hands-on operations with high temperature glassware and solution. Use personal protective equipment including goggles and heat-resistant gloves. All the operations involving solution handling should be conducted in a fume hood and avoid other heating sources nearby due to the corrosive and flammable properties of anhydrous acetic acid.
2. Preparation for Pd Acetate – TOP Synthesis Solution14
3. Colloidal Pd Nanoparticle Synthesis14
4. Pd Nanoparticle Characterization- Ex situ Small-angle X-ray Scattering (SAXS)34
5. Obtaining Kinetic Data from in situ SAXS on Colloidal Pd Nanoparticle Synthesis at Synchrotron
6. Modeling Approach and Simulation Procedures for Nucleation and Growth of Palladium (Pd) Metal Nanoparticles
7. Obtaining Nucleation and Growth Rates from Both the Experimental Data and Model
To systematically examine whether the capping ligands alter the kinetics of nucleation and growth, we took the two following approaches: (i) the binding of the ligand with the metal was not considered in the kinetic model similar to previous studies (i.e., the nucleation and autocatalytic growth) (ii) the reversible binding of capping ligand with the precursor and surface of the nanoparticle was taken into account in the model (i.e., ligand-based model described in Protocol). Regarding the Pd synthesis in toluene, as shown in Figure 1, without accounting for the ligand-metal binding, the model failed to capture the time evolution of the nanoparticles concentration () and concentration of Pd atoms (). As an alternative, we implemented our newly developed kinetic model (Figure 2) and as depicted in Figure 3, the model accurately predicts our in situ data (both and during reaction).This further indicates that the capping ligands indeed affect the nucleation and growth kinetics of Pd nanoparticles.
Estimating the rate constants (Table 1) from the model further enables us to obtain useful information on the kinetics of the nanoparticle formation. In this regard, Figure 4A shows the comparison between the nucleation and growth rates (as estimated from the model) and the results clearly reveal that nucleation is slow while the growth is fast, which agrees well with previous studies1,14. Both modeling and experimental results demonstrate that the metal precursor/monomer does not undergo burst nucleation. This is illustrated by the in situ SAXS and modeling results where the nucleation continues till the end of synthesis (Figure 3B and Figure 4A). The continuous formation of nuclei, therefore, contradicts the LaMer burst nucleation and growth model but supports the continuous nucleation reaction in the Finke-Watzky two step mechanism. In addition, the nucleation can be fitted by pseudo-first order; however, we cannot exclude the possibility that the nucleation could be higher in order. Herein, as shown in Figure 4B, the ligand plays a central role in the continuity of nucleation by further binding to the nanoparticle surface and reducing the concentration of active sites (i.e., [B]). This drastically decreases the particle growth rate and expands the time window for the nucleation throughout the synthesis. In addition, our current results presented in this work in combination with our previous study14 (where the synthesis was conducted under different experimental conditions) indicate that the ligand and precursor concentrations do not have a significant effect on the rate and equilibrium constants, which shows the chemical fidelity between the model and the real system.
Next, we probed the applicability of our ligand-based model to a different solvent system, where pyridine was used as a solvent instead of toluene. We can see that despite the significant difference observed for the nucleation and growth kinetics in pyridine compared to toluene (Figure 5 and Table 1), the model accurately captures the in situ data, and , and allows for more accurate estimation of rate constants (Table 1). One of the important features that makes a kinetic model robust is that it should be able to predict synthetic conditions for achieving nanoparticles with desired sizes. Therefore, we implemented our ligand-based model (using the same rate constants reported in Table 1) to predict the size under different concentrations of metal precursor, Pd(OAc)2, in pyridine. Figure 6 shows that the model can provide a very accurate estimation of the nanoparticle size under different concentrations of the metal precursor. The modeling as well as the experimental results demonstrate that the nanoparticles become larger in size at higher precursor concentration. This is because the growth is second order kinetics while the nucleation is first order which makes the growth faster at higher precursor concentration14.
Figure 1. Experimental and two-step modeling results for the synthesis of Pd nanoparticles in toluene: (A) concentration of Pd atoms and (B) concentration of nanoparticles. The rate constants are = s-1 and = L.mol-1.s-1. Experimental condition: [Pd(OAc)2]=25 mM , TOP:Pd molar ratio= 2, and T (°C) = 100. Please click here to view a larger version of this figure.
Figure 2. The schematic of ligand-mediated nucleation and growth model. In this proposed model, the capping ligands can associate and dissociate from both the metal precursor and nanoparticle surface, thereby, affecting the nucleation and growth kinetics (through altering the concentration of kinetically active precursor and the number of free surface sites, respectively). Please click here to view a larger version of this figure.
Figure 3. Experimental and ligand-based modeling results for the synthesis of Pd nanoparticles in toluene: (A) concentration of Pd atoms and (B) concentration of nanoparticles. The rate constants are summarized in Table 1. Experimental condition: [Pd(OAc)2]=25 mM , TOP:Pd molar ratio= 2, and T (°C) = 100. Please click here to view a larger version of this figure.
Figure 4. (A) The rates of the nucleation and growth extracted from the ligand-based model for the synthesis of Pd nanoparticles in toluene and (B) ratio. Experimental condition: [Pd(OAc)2]=25 mM , TOP:Pd molar ratio= 2, and T (°C) = 100. Please click here to view a larger version of this figure.
Figure 5. Experimental and ligand-based modeling results for the synthesis of Pd nanoparticles in pyridine: (A) concentration of Pd atoms and (B) concentration of nanoparticles. The rate constants are summarized in Table 1. Experimental condition: [Pd(OAc)2]=2.5 mM , TOP:Pd molar ratio= 2, and T (°C) = 100. Please click here to view a larger version of this figure.
Figure 6. Model prediction of final nanoparticle size as a function of precursor concentration in pyridine solution (experimental data from Mozaffari et al.14). The error bars represent the standard deviation of the particle size distribution. Experimental condition: TOP:Pd molar ratio= 2, and T (°C) = 100. Please click here to view a larger version of this figure.
k1-nuc | k2-growth | k3-f (A+L) | k4-f (B+L) | K5-eq (A+L) | K6-eq (B+L) | |
Units | s-1 | L.mol-1.s-1 | L.mol-1.s-1 | L.mol-1.s-1 | L.mol-1 | L.mol-1 |
25 mM Pd in Toluene | 1.8×10-5 | 10×10-1 | 4.7×10-3 | 3×10-1 | 1.5×101 | 1×103 |
2.5 mM Pd in Pyridine | 1.74×10-5 | 2.34×101 | 1.7×10-1 | 2.13×10-2 | 3.54×102 | 1.44×102 |
Table 1. The extracted rate constants for Pd nanoparticle synthesis in different solvents (toluene and pyridine). Experimental condition: TOP:Pd molar ratio= 2, and T (°C) = 100.
In this study, we presented a powerful methodology to examine the effect of capping ligands on the nucleation and growth of metal nanoparticles. We synthesized Pd nanoparticles in different solvents (toluene and pyridine) using Pd acetate as the metal precursor and TOP as the ligand.We used in situ SAXS to extract the concentration of reduced atoms (nucleation and growth events) as well as the concentration of nanoparticles (nucleation event), where both experimental observables were used as the model inputs. In addition, by considering the slope of the concentration of the nanoparticles and concentration of the atoms at the early reaction time, our methodology (the use of in situ SAXS and kinetic modeling), allowed us to estimate the upper and lower bonds for the nucleation and growth rate constants (more details can be found in ref. 14, which was the first study to decouple the contributions of nucleation and growth to the total metal reduction).
There are three critical steps in systematically examining the effects of ligand-metal binding on the nucleation and growth of colloidal nanoparticles: (i) measuring the evolution of size as well as the concentration of nanoparticles (steps 4.1-4.3). This is an important step as it can provide more detailed information on both the nucleation and growth events, (ii) developing a robust kinetic model, which explicitly accounts for the reactions of capping ligands with the metal and also includes the most relevant reactions during the formation and growth of nanoparticles (step 6.4), and (iii) constructing an appropriate link between the experimental observables and those extracted from the model (e.g., size measured experimentally versus size extracted from the model).
It is important to note that due to the small size of the particles (< 10 nm in diameter), and the fast nucleation and growth rates in the beginning of the reaction, a high energy and high flux X-ray beam is needed for obtaining in situ data, which can be only realized at the synchrotron. Even with synchrotron beams, it is difficult to capture any size below 0.5 nm unless the concentration of the particle is high enough. A rule of thumb principle is that SAXS intensity reduces with 6th power of the particle size but it is only linearly proportional to the concentration of the nanoparticles. In addition, for smaller nanoparticles, data acquisition up to much higher wave vector q (wider angle) is required, where the background scattering from solvents become more significantly detrimental to signal to noise ratio. This limits the size and concentration of small nanoparticles that can be detected in the early stages of the reaction, especially when the nucleation is slow and continuous as shown in this work. However, while the high energy/flux allows the acquisition of in situ data, the beam can also cause damage to the sample (agglomeration of nanoparticles and/or deposition on the cell walls). Therefore, in step 5.1, the beam energy and X-ray exposure time need to be tested and adjusted to the level that provides the best data quality (signal to noise ratio) for the detection of small nanoparticles in the early stages of the reaction without causing damage to the sample. The troubleshooting has to be done at the synchrotron during the in situ SAXS measurement, i.e., to monitor the SAXS spectra and ensure that no agglomeration/precipitation occurs during the synthesis. Through a few tests, the beam energy was finally set at 18 keV with an appropriate exposure time (0.1 s) to capture enough signal, and hence, the small Pd nanoparticle size in the early stage of reaction. We also note that while the current kinetic model does not account for agglomeration, if such growth mechanism is dominant, the model can be modified to include agglomeration steps (for example, B + B → C and B + C → 1.5C, where B and C represent the small and larger nanoparticles, respectively)1. However, agglomeration as well as other modes of growth (i.e., Ostwald and digestive ripening)40 would be best described by population based models24,25,32,33.
As already discussed in the manuscript, the underlying mechanism governing the nanoparticle nucleation and growth is poorly understood, particularly in the presence of coordinating ligands. For example, recent studies showed that TOP-Pd binding lowers the nucleation and growth rate of Pd nanoparticles14,15,16,30. Therefore, we accounted explicitly for the ligand-metal binding in our kinetic model. What distinguishes our method from other relevant studies is that our ligand-based model considers the ligand binding with both the precursor and surface of metal nanoparticle as reversible reactions and no priori assumptions are made on whether the ligands are in equilibrium with either of them. In addition, unlike previous studies where only one experimental observable (either size33 or concentration of atoms23, etc.) was used for model verification, our ligand-based model uses both the particle size and concentration of nanoparticles as model inputs. Therefore, it allows us to obtain more accurate estimates for the reaction rate and equilibrium constants.
Using our proposed methodology, we demonstrated the predictive power of our ligand-based model. In this regard, we showed that the model can predict the synthesis conditions to obtain nanoparticles with various sizes, which as a result minimizes the need for trial and error. Furthermore, with this simple "heat-up" synthesis method, the nanoparticle size can be tuned by changing the type of solvent or the metal concentration. These different sized Pd nanoparticles can have potential applications in catalysis, drug delivery, and sensors15,41. The presented synthesis strategy along with the kinetic modeling can be potentially used to provide insights on the role of capping ligands in the nucleation and growth of different types of nanoparticles to guide their controlled synthesis.
For future work, we direct our research toward developing kinetic models with the ability of predicting the size distribution during the synthesis. In addition, we will further investigate the validity of our ligand-based model under different experimental conditions, including different temperature ranges and different types of ligands and metals.
The authors have nothing to disclose.
The work was primarily funded by the National Science Foundation (NSF), Chemistry Division (award number CHE-1507370) is acknowledged. Ayman M. Karim and Wenhui Li acknowledge partial financial support by 3M Non-Tenured Faculty Award. This research used resources of the Advanced Photon Source (beamline 12-ID-C, user proposal GUP-45774), a U.S. Department of Energy (DOE) Office of Science User Facility operated for the DOE Office of Science by Argonne National Laboratory under Contract No. DE-AC02-06CH11357. The authors would like to thank Yubing Lu, a Ph.D. candidate in the Chemical Engineering Department at Virginia Tech for his kind help with the SAXS measurements. The presented work was partly executed at the Center for Integrated Nanotechnologies, an Office of Science User Facility operated for the U.S. Department of Energy (DOE) Office of Science. Los Alamos National Laboratory, an affirmative action equal opportunity employer, is operated by Los Alamos National Security, LLC, for the National Nuclear Security Administration of the U.S. Department of Energy under contract DE-AC52-06NA25396.
palladium acetate (Pd(OAc)2) | ALDRICH | 520764 | |
anhydrous acetic acid | SIAL | 338826 | |
trioctylphosphine | ALDRICH | 718165 | |
pyridine | MilliporeSigma | PX2012-7 | |
toluene | SIAL | 244511 | |
1-hexanol | SIAL | 471402 | |
N8 Horizon SAXS | Bruker | A32-X1 | |
glovebox | Vaccum Atmospheres Co. | 109035 | |
MR HEI-TEC 115V Hotplate | Heidolph | 5053000000 | |
hotplate Monoblock insert | Heidolph | 5058000800 | |
heat-On 25-ml insert | Heidolph | 5058006200 | |
7 mL vials | SUPELCO | 27518 | |
micro stir bar PTFE | VWR | 58948-353 | |
egg-Shaped Bars | Fisherbrand™ | 14-512-121 | |
25 mL round bottom flasks | ALDRICH | Z167495 | |
quartz capillary | Hampton Research | HR6-148 | |
MATLAB R2016b | MathWorks | ||
Bruker SAXS 1.0v | Bruker | ||
Diffrac Measurement Center 4.0v | Bruker |