Article pubs.acs.org/Langmuir
Comparative Study of Ligand Binding during the Postsynthetic Stabilization of Metal Oxide Nanoparticles C. Grote,† T. A. Cheema,† and G. Garnweitner* Institute for Particle Technology, Technische Universität Braunschweig, Volkmaroder Strasse 5, D-38104 Braunschweig, Germany S Supporting Information *
ABSTRACT: In the absence of stabilizers in the reaction medium, the nonaqueous synthesis of metal oxide nanoparticles usually results in agglomerated products. Stabilization is however often possible in a postsynthetic treatment, involving the addition of organic ligands that coordinate to the nanoparticle surface. The ligands are commonly expected to chemisorb via functional groups; however, we have recently shown that also weakly and unspecifically interacting ligands can lead to stabilization. Here, we present detailed investigations on the stabilization, comparing the binding of weakly coordinating ligands to a system with strongly and selectively binding stabilizers and additionally exploring the effect of ligand chain length. Although in all cases stabilization and disintegration of agglomerates to the primary particle level are achieved, strong differences are observed with respect to the processes at the particle surface. Moreover, these processes are shown to be more complex than simple ligand adsorption and need to be understood for proper design and choice of stabilizers.
■
INTRODUCTION Metal oxide nanoparticles have become a principal material in nanotechnology, as they can be employed in very diverse fields and in different forms, e.g., dispersed in aqueous media for cancer diagnosis and treatment in biomedicine1−5 or embedded in a polymer matrix material to form nanocomposites with enhanced mechanical or optical properties.6−8 For such applications, it has been proven that the surface chemistry has a tremendous effect on the properties and performance, with for example different cytotoxicity and mechanisms of endocytosis occurring for differently coated nanoparticles9,10 or different mechanical properties of nanocomposites being observed for nanoparticles functionalized with different organic ligands.11−14 For most applications, the nanoparticles are processed in the form of liquid dispersions, where they can be safely handled and applied. Here, the surface chemistry plays an important role in the stabilization of nanoparticles against agglomeration, which needs to be accomplished in order to keep the nanoparticles in dispersion and obtain products with high homogeneity and e.g. optimum optical properties.15 The addition of surfactants or other small-molecule organic ligands to the synthesis medium can be utilized to achieve instant stabilization and may offer additional benefits such as shape control16,17 but can also adversely affect the particle formation and lead to greatly reduced crystallinity.18,19 Hence, in the most cases, the stabilization in a separate step following the synthesis by means of a surface modification appears to be the better choice in order to allow for a simple synthesis system and a stabilization or functionalization tailored to a desired medium and way of processing.20−22 In contrast to the classical colloidal methods of electrostatic stabilization by generation of a potential on the particle surface and steric stabilization by © 2012 American Chemical Society
adsorption of polymers, the stabilization of nanoparticles can generally be achieved also through adsorption of small organic ligands such as surfactants.23 Whereas the stabilization of metal nanoclusters and semiconductor nanocrystals has been studied in more detail,24−27 the stabilization of metal oxide nanoparticles is still hardly understood and predominantly achieved empirically for various systems and applications. The achievement of a fundamental understanding of the particle interactions for small nanoparticles would thus facilitate the development of more efficient strategies for stabilization, also enabling the controlled self-assembly of such nanoscale building blocks to more complex structures. The application of the colloidal theories of stabilization to small nanoparticles below 20 nm in size, although feasible in principle, has however shown to be problematic in practice for judging the stability of nanoparticle dispersions, especially for organic dispersion media. In particular, often mixed electrostatic and steric effects play a role for the stability of nanoparticles in polar organic systems.28,29 Recently, Peukert and co-workers showed that the colloidal stability of ZnO nanoparticles in ethanol can be explained by dimensionless numbers, based on geometry as well as attraction and repulsion parameters, better than considering only the energy barrier as calculated by the DLVO model.30 One aspect that is crucial for a deep understanding of the colloidal stability of nanoparticles is the precise knowledge of their surface chemistry. Metal oxide nanoparticles have shown to be particularly difficult in this context, as their interactions Received: May 3, 2012 Revised: September 2, 2012 Published: September 6, 2012 14395
dx.doi.org/10.1021/la301822r | Langmuir 2012, 28, 14395−14404
Langmuir
Article
ITO system. To retrieve the stabilized nanoparticles, the dispersion was added into ethyl acetate with a volume ratio of 5:1 (ethyl acetate:dispersion) to induce precipitation, shaken for 5 min, and then centrifuged. The obtained precipitate was dried at room temperature under vacuum. For 13C NMR spectroscopy, 60 mg of the thusobtained nanoparticle powder was redispersed in 800 μL of CDCl3 and then analyzed employing a Bruker AV II-600 instrument. To determine the ITO or ZrO2 content in dispersion, gravimetry was performed in ceramic crucibles using 200 μL of dispersion and heating to 600 °C in air. The volumetric particle size distribution of the dispersions was determined (after dilution to 40 mg mL−1 for concentrated dispersions) using dynamic light scattering (DLS) on a Malvern Zetasizer Nano ZS device, performing a 173° backscatter measurement with three repeated runs. Thermogravimetric analysis (TGA) was performed using dried powder samples. All TGA measurements were carried out on a Mettler Toledo TGA/SDTA 851 under O2 in the range of 25−950 at 10 °C min−1.
with organic ligands in liquid media are rather complex; however, it is known that surface modification is possible by the anchoring of various chemical functionalities such as carboxylates and phosphonates.22 Additionally, it is known that the specific synthesis method of metal oxides has a high influence on their surface chemistry,31 but precise and yet generally applicable methods for describing the surface state of nanoparticles and allowing a comparison have not been established yet. Although the exchange of ligands on the surface and their effect on particle stability for example have been described by Boal et al.,23 general knowledge as to the affinity of different ligands to the nanoparticle surface and the rate of ligand exchange has not been achieved so far. We have recently reported that, in contrast to the widespread notion of ligands anchoring to the nanoparticle surface being a prerequisite for colloidal stability, the stabilization of indium tin oxide (ITO) nanoparticles can also be achieved via weakly coordinating ligands, with the initial reaction medium still playing an important role despite its unsuitability for stabilization.32 In this contribution, we present further insights into the binding of ligands to the surface of metal oxide nanoparticles, showing that weakly binding ligands can lead to long-term stability. The ITO system is compared to the stabilization of ZrO2 nanoparticles by addition of fatty acids, which results from the selective anchorage of the carboxyl groups to Zr centers on the particle surface.33 Thereby, two different model systems with weak and strong ligand binding are compared, showing that the differences in the nature of ligand binding result in different mechanisms of stabilization. From these insights, important conclusions toward an optimization of the nanoparticle stabilization can be drawn, which might represent a first step in the development of general concepts, beyond the classical theories of stabilization, for a rational stabilization of metal oxide nanoparticles in organic hydrophobic media.
■
■
RESULTS AND DISCUSSION ITO and ZrO2 nanoparticles were synthesized via the nonaqueous approach in benzyl alcohol as described earlier.33,34 Because of the absence of stabilizers in the reaction medium, the nanoparticles are obtained in agglomerated form, with agglomerate sizes in the micrometer range as evident from their fast sedimentation. However, they can be stabilized in nonpolar organic media through a postsynthetic stabilization treatment, involving the washing of the precipitate in a volatile organic solvent followed by addition of a solution of the stabilizer in the dispersion medium (Figure 1). TEM images of the stabilized
EXPERIMENTAL SECTION
ITO nanoparticles with 15 wt % SnO2 content were prepared by reaction of In(III) acetylacetonate and Sn(IV) tert-butoxide in benzyl alcohol (BnOH) as described by Ba et al.34 The nanoparticles were isolated from the reaction mixture by centrifugation and then subjected to washing (resuspension and centrifugation cycles) with CHCl3. The yield of one synthesis was about 165 mg of ITO nanoparticles (0.638 mmol). For the stabilization experiments, a solution of a defined amount of an n-alkylamine as stabilizer (denoted NCn in the following, whereby n refers to the alkyl chain length) in 5 mL of CHCl3 was prepared and was added to the nanoparticles under stirring at room temperature, immediately after the washing procedure. The unstabilized particle fraction was separated from the dispersion by centrifugation at 6940g (8500 rpm) for 15 min. To isolate the stabilized nanoparticles from the dispersion for characterization, methanol was added to induce precipitation in a volume ratio of 1:1, followed by centrifugation at 6120g (7500 rpm) for 10 min and drying at ambient temperature under vacuum. The ZrO2 nanoparticles were synthesized by nonaqueous synthesis as described earlier33,35 with slight modifications. 80 mL of Zr(IV) npropoxide in 1-propanol (70 wt %, Aldrich) was mixed with 500 mL of BnOH and reacted for 4 days at 220 °C in a steel reactor fitted with an additional glass cup (Parr Instruments). From the obtained suspension, an aliquot of 5 mL corresponding to 275 mg of ZrO2 nanoparticles (2.23 mmol) was used for a stabilization experiment. The nanoparticles were retrieved by centrifugation, washed 2 times with EtOH, and then mixed with a solution of n-carboxylic acid (denoted Cn in the following, n referring to the chain length; 0.023− 0.244 mmol in 5 mL of CHCl3) and left under stirring for 24 h. Any present agglomerates were removed by centrifugation in analogy to the
Figure 1. Photographs of (a) ITO and (b) ZrO2 nanoparticle dispersions before (left) and after (right) the stabilization treatment in CHCl3 by addition of (a) dodecylamine and (b) n-decanoic acid.
nanoparticles after drying of the dispersions show rather uniform and nonagglomerated nanoparticles despite high concentrations used for the measurements (Figures S1 and S2). Whereas the ITO nanoparticles possess cubic morphology with an average core size of 8 ± 2 nm, the ZrO2 nanoparticles are edgy and smaller with 6 ± 2 nm. We have reported recently that the ITO nanoparticles can be stabilized in chloroform by addition of n-alkylamines of different chain lengths that 14396
dx.doi.org/10.1021/la301822r | Langmuir 2012, 28, 14395−14404
Langmuir
Article
Figure 2. 13C NMR spectra of (a) modified ZrO2 nanoparticles (ZrO2-C10-0.1) and (b) free decanoic acid.
undergo only a weak interaction with the particle surface.32 In contrast, ZrO2 nanoparticles prepared by the same synthetic procedure cannot be transformed to stable dispersions by addition of n-alkylamine ligands but are colloidally stabilized by fatty acids via a selective interaction of the carboxylic acid group, as can be deduced from 13C NMR spectroscopy. As a representative example, Figure 2 shows the spectrum of ZrO2 nanoparticles after stabilization with n-decanoid acid in a molar ratio of 1:0.1 (ZrO2:stabilizer, thus denoted ZrO2-C10-0.1). To remove any weakly adsorbed components from the system, the particles were precipitated, dried, and redispersed in CDCl3 for the measurement (in contrast, the ITO nanoparticles are not redispersible after complete drying). By comparison of the spectrum of the redispersed nanoparticles to a spectrum of the ligand, it is clearly visible that aside from the solvent and a small amount of EtOH stemming from the washing of the particles (labeled ∗), only the signals of the ligand are visible, suggesting a strongly preferred binding of the ligand over all other organics in the system, such as in particular BnOH stemming from the initial synthesis of the nanoparticles, ethanol as the washing solvent, and ethyl acetate as precipitation agent. From the corresponding 1H NMR spectrum (Figure S4), the amount of BnOH (or any derivative aromatic species stemming from the synthesis, such as benzoates) remaining on the nanoparticles is below 0.005 (expressed as molar ratio BnOH:decanoic acid), illustrating the high selectivity of ligand binding. Similar to our earlier observations,33 a strong binding of the ligand to the nanoparticle surface via the carboxylic group is evidenced by the signals of the C1, C2, and C3 atoms. Although the chemical shift of the carbonyl atom C1 is difficult to determine due to strong peak broadening at ∼181 ppm, a considerable downfield shift of the vicinal carbon atom (C2) from 34.4 to 37.3 ppm as well as significant peak broadening clearly points to binding of the ligand to the ZrO2 nanoparticles via the carboxyl group.36 Also, the C3 signal shows peak broadening and a shift from 24.8 to 25.6 ppm, whereas the terminal signals C8−C10 of the ligand are clearly visible without substantial broadening or shift. For comparison, a 13C NMR spectrum of ITO nanoparticles stabilized with dodecylamine is presented in Figure S3: no particular carbon signal shifts or peak broadening can be detected compared to a spectrum of the pure ligand, illustrating this weak and unspecific interaction. By comparing these two model systems with strong and weak ligand binding, we would like to shed further light on the stabilization mechanism of metal oxide nanoparticles. Additionally, for each system the effect of a short-chain ligand (C5− C6) is compared to a long-chain ligand (C10−C12). Initially, the following questions arose: (i) whether a reliable (i.e., long-
term) colloidal stabilization of nanoparticles can be achieved by both strongly and weakly coordinating ligands; (ii) whether there are fundamental differences in the stabilization for these two systems; (iii) whether the chain length of the utilized ligand additionally exerts a crucial influence; and (iv) whether the adsorption of the ligand in all cases directly initiates the deagglomeration of the nanoparticles or whether other processes, and possibly also other species bound to the particle surface, are important. To elucidate these issues, a certain amount of nanoparticles obtained directly from the synthesis was used for stabilization experiments, while varying the nature and concentration of added stabilizer. For the ITO nanoparticle system with weakly binding ligands, not the full quantity of nanoparticles could be stabilized below a critical stabilizer concentration. In these cases, a certain fraction of micrometer-sized agglomerates remained in the dispersion which was removed by centrifugation. Figure 3a shows the obtained solid contents for two
Figure 3. Solid content of (a) ITO and (b) ZrO2 nanoparticle dispersions obtained by addition of stabilizers in different molar ratios (stabilizer:metal oxide) after removal of agglomerates by centrifugation, also showing the theoretical maximum solid content for each system (dashed lines).
representative stabilizers, dodecylamine (denoted NC12) and amylamine (NC5). It is clearly visible that for increasing molar ratios of stabilizer:ITO the content in stabilized nanoparticles rises strongly. Additionally, there is a strong dependence on the chain length of the stabilizer: very small amounts of dodecylamine already result in the stabilization of significant quantities of the nanoparticles, with a molar ratio of 0.04:1 (NC12:ITO) being sufficient for the stabilization of almost the full fraction of nanoparticles (within measurement error). In contrast, the addition of substantially larger amounts of amylamine is required to achieve stabilization, with no stabilized nanoparticles being observed for molar ratios below 0.14:1. Even for larger amounts of NC5 (molar ratio of 0.5:1) 14397
dx.doi.org/10.1021/la301822r | Langmuir 2012, 28, 14395−14404
Langmuir
Article
Figure 4. Thermogravimetric analysis of (a) ITO nanoparticles before and after a different number of washing cycles, (b) washed as well as stabilized and unstabilized (marked by ∗) ITO nanoparticles after addition of amylamine, and washed and stabilized ITO nanoparticles for different amounts of (c) amylamine and (d) dodecylamine added.
on the particle surface stemming from the synthesis were investigated considering the number of washing cycles. Figure 4a shows the weight loss for unwashed ITO nanoparticles directly obtained from the synthesis as well as the weight loss for nanoparticles that were washed twice and ten times in CHCl3, respectively. The plot of the unwashed particles shows a two-step weight loss. Comparing this curve with the weight loss curve of pure BnOH (see Figure S5) identifies the first step in a temperature range
