New Insights into the Side-Face Structure, Growth Aspects, and

Article pubs.acs.org/Langmuir

New Insights into the Side-Face Structure, Growth Aspects, and Reactivity of Agn Nanoprisms

Aurélie Le Beulze,† Etienne Duguet,† Stéphane Mornet,† Jérôme Majimel,† Mona Tréguer-Delapierre,*,† Serge Ravaine,‡ Ileana Florea,§ and Ovidiu Ersen§ †

CNRS, Univ. Bordeaux, ICMCB, UPR 9048, F-33600 Pessac, France CNRS, Univ. Bordeaux, CRPP, UPR 8641, F-33600 Pessac, France § IPCMS, UMR 7504 CNRS Uds, 23 rue du Lœss, 67087 Strasbourg, France ‡

S Supporting Information *

ABSTRACT: We report an improved synthesis of colloidal Agn nanoprisms using carboxyl compounds (citrate or succinate) and long chain macromolecules (polyvinylpyrrolidone (PVP)). The side-facet structure of the triangular nanostructure was determined in detail using electron tomography in scanning transmission mode (3D STEM) and HRTEM. It has been found that they are built up by {100} facets with a single parallel twin plane. The best conditions for producing uniform Ag nanoprisms with tunable sizes and high yields in the presence of carboxyl compounds additive system are described, and a growth mechanism is proposed. This approach provides also a route to synthesize Ag nanodisks and Au−Ag alloyed nanoprisms.

■

INTRODUCTION Among metal nanostructures, silver nanoprisms are highly interesting because of their well-defined anisotropy.1,2 This anisotropy together with sharp corners and edges greatly influences various properties, among which the optical extinction can be singled out.3,4 The nanoprisms display localized surface plasmon resonances (SPR) that can be tightly controlled through changes of their aspect ratio.2,3,5,6 The most intense optical modes have been associated with dipolar and quadrupolar excitations, in analogy to spherical nanoparticles.7 Such modes correspond to field enhancements at the corners (dipolar mode) and at the edges (quadrupolar modes). The dipolar modes have extremely high extinction cross sections at relatively long wavelengths and can be widely tuned by altering the length of the nanoprisms, giving rise to a number of interesting applications in fields as diverse as photonics, ultrasensitive chemical and biological sensors, labeling, and optoelectronics.2,3,5,8−11 This has fostered the development of variety of physical and chemical methods to produce Ag nanoprims with controlled dimensions.1,2,16 Among them, the seed-mediated growth in aqueous medium assisted by carboxylate-containing compounds and long chains of polyvinylpyrrolidone (PVP) has been shown to be successful for the growth of Ag nanoprisms with a relative control over size/ shape uniformity and a spectral tunability of the localized surface plasmon resonances. It yields nanoprisms covered by large {111} facets at both top and bottom surfaces.9,21−24 Yet, the exact nature of the side-facets geometry has not been fully established. They would be built up with {100}, {110}, or {111} faces and one or more stacking faults extending across © 2014 American Chemical Society

the entire nanostructure. By adjusting some experimental parameters such as the reagents/capping ligands ratio or the temperature, the relative control over prism length (45 nm−5 μm) or edge (5−200 nm) has been demonstrated.17 Efforts have been carried out to elucidate the seeded-growth mechanism leading to speculation that preferential binding of the capping ligands occurs onto certain facets. The citrate ions and the PVP macromolecules would bind to the {111} and {100} seed facets, respectively, promoting two-dimensional growth.25,26 However, the anisotropic growth is not guaranteed by simply introducing selective binding ligands into the reaction medium. The introduction of seeds with internal defect structures would be another key parameter.27 Their stacking faults would provide the driving force for twodimensional growth. Despite these advances in silver nanoprisms synthesis, it is still a challenging task to generate uniform Ag nanoprisms with controllable size in bulk quantities using the existing protocols. Furthermore, there are still many questions concerning the conditions under which the growth of Agn nanoprisms takes place. Every now and then we read new reports which did not really fit with the face-blocking and crystal-twinning models mentioned above.24,28−32 It is still not clear which factors play the most important role. When defects and capping agents are both involved, we are unsure which of them will control the growth of a nanocrystal seed. Received: October 28, 2013 Revised: January 17, 2014 Published: January 21, 2014 1424

dx.doi.org/10.1021/la4039705 | Langmuir 2014, 30, 1424−1434

Langmuir

Article

Figure 1. (top) Extinction spectrum of the PSS/citrate-stabilized seeds and triangular nanoparticles solutions synthesized with citrate and PVP at pH = 5.6. TEM image of the triangular nanoparticles. (bottom) High-resolution images of the crystalline lattice of the side and top facets. Electron diffraction pattern with the assignment of the reflections taken from an individual triangular nanoplate ([111] zone axis). Preparation Procedure for Silver Seeds. Typically, 49 mL of an aqueous solution containing AgNO3 (0.1 mM) and trisodium citrate (0.3 mM) was prepared in a 100 mL two-necked round-bottom flask and immersed in an ice bath. The solution was bubbled with argon (Ar) under constant stirring for 30 min to remove oxygen. Then the Ar flow was kept constant at the surface of the solution throughout the reaction. Under vigorous stirring, 0.5 mL of aqueous NaBH4 (cold and freshly prepared, 50 mM) was quickly added into the ice-cold solution. The reaction was allowed to proceed for 15 min, and during this time, 3−5 drops of NaBH4 solution were added every 2 min to the solution to ensure complete reduction of the Ag+ ions. Finally, 0.5 mL of aqueous PSS (5 mg/mL) and 0.25 mL of NaBH4 (50 mM) were added to the solution in a dropwise fashion over a 5 min time period. After 1 h of reaction time, the flask was removed from the ice bath. Stirring and Ar flow were maintained for 4 h at room temperature to allow the excess of borohydride to be decomposed by water. The average particle diameter was 2.5 ± 0.6 nm as determined by statistical analysis of TEM images (Figure S1, Supporting Information). The colloidal suspension absorbs at 378 nm and possesses an exceptionally narrow band (Figure 1) (full width at half-maximum, fwhm = 50 nm). The seed dispersion was used directly for all further experiments. In order to study the role of the seed size on the silver nanoprisms synthesis, seeds of larger sizes were also prepared by playing with the composition of capping agent mixture. In the absence of PSS, the average size of the citrate-stabilized seeds was larger (10.0 ± 0.2 nm) with many imperfections (Figure S1). By substituting PSS with PVP, seeds of 10.0 ± 2.0 nm coexisting with smaller ones (5.0 ± 1.0 nm) were obtained. In the presence of both PVP and citrate, the preparation produced stable seeds with an average size of 4.9 ± 0.9 nm. Preparation Procedure of Silver Nanoprisms. Typically, 100 mL of ultrapure water was mixed with 100 mL of aqueous L-ascorbic acid (0.5 mM), 3.5 mL of as-prepared 4 h-aged seeds, 38 mL of aqueous PVP (0.7 mM in terms of monomer units), and 38 mL of aqueous trisodium citrate or succinate (30 mM) in a 500 mL round

In this paper, we report an improved synthesis of colloidal silver nanoprisms through seed-mediated growth and their detailed morphological and crystallographic analysis via electron tomography (3D TEM). What are widely accepted as “essential reactants” for silver nanoprims synthesis, i.e., carboxylated compounds, PVP, Ag seeds and ascorbic acid, were kept. We studied the influence of the control of the rate of supply of the silver precursors to the seeds on the silver nanoprism features and yields, and we derived a growth mechanism. Furthermore, by exploiting the nanoprisms’ affinity for some common inorganic (metal halides and gold salt) or organic species such as 11-mercaptoundecanoic acid (MUA), we investigated the possibility to use Ag triangular nanoprisms as precursor for the synthesis of stable Ag nanodisks and Ag− Au alloyed nanoprims.

■

EXPERIMENTAL SECTION

Chemicals. All solvents and chemicals were purchased from SigmaAldrich and used without further purification: silver nitrate (AgNO3, 99.99%); sodium borohydride (NaBH4, 99.99%); trisodium citrate (Na 3 C 6 H5 O7 , ≥99%); poly(sodium 4-styrenesulfonate) (PSS, (C8H7O3S)n, average Mw ∼70 000 g.mol−1) 30 wt % in H2O); disodium succinate (Na2C4H4O4, ≥98%); polyvinylpyrrolidone (PVP, (C6H9NO)n, average Mw ∼55 000 g.mol−1); 11-mercaptoundecanoic acid (MUA, 98%); gold(III) chloride trihydrate (HAuCl4·3H2O, 99.99%) and potassium gold(III) chloride (KAuCl4, 99.99%); HCl ; Lmalic acid disodium salt (C4H6O5, ≥99%); ethylenediaminetetraacetic acid (EDTA, C10H12O8N2, 99.99%); and L-ascorbic acid (AA, C6H8O6, ≥99%). Ultrapure water (18.2 MΩ·cm−1) was used directly from a LaboStar 7 UV TWF system (Odémi, Grisy, France). Any glassware was cleaned in a bath of freshly prepared aqua regia (HCl/HNO3, 3:1), then rinsed thoroughly with H2O, and dried at 90 °C in an oven before use. 1425

dx.doi.org/10.1021/la4039705 | Langmuir 2014, 30, 1424−1434

Langmuir

Article

flask. Then, under magnetic stirring, 12.5 mL of aqueous AgNO3 (5 mM) was added to the solution in a dropwise fashion with a rate of addition fixed at 10 mL/h. The solution progressively changed color via yellow, orange, pink, purple, and finally blue, and then it intensified as the addition of silver precursor continued. Throughout the reaction, the mixture was maintained under ambient atmosphere and at room temperature (≈20 °C). Moreover, the silver precursor solution and the round flask containing the growth solution were protected from light exposure. The blue silver sol could also be obtained from AgClO4 under similar condition. For studying the effect of the pH on the size and shape of the as-produced nanoparticles, a Mettler Toledo Titrator T70 equipped with a DGi115-SC electrode was used. It allows for the regulation of the pH during the growth step while pouring the aqueous solution of silver nitrate into the stirred growth medium, which contained the seeds, the organic stabilizers(s), and the reducing agent. By substituting the calomel electrode by a perfectION combination silver/sulfide electrode, the pAg (pAg = −log [Ag+]) of the solution could be also regulated, and the rate at which the free silver ions of an aqueous solution of silver nitrate were consumed could be measured. Details on the experimental conditions for calibration have been reported in the Supporting Information. Standard aqueous solutions of silver ion (Fluka; 1, 100, and 1000 ± 2 mg.L−1) were used for calibration. To form silver disks, 20 μL of HCl (0.1 M) was added in one shot into 1 mL of as-prepared silver nanoprisms (1.3 × 1014 part/L) under stirring. To prepare the gold triangular particles from silver nanoprisms, 825 μL of ascorbic acid (10 mM) was added to 13.9 mL of the silver nanoprisms solution under vigorous stirring. Then 10 mL of an aqueous HAuCl4 solution (10 mL, 0.5 mM) was dropwise added at a rate of 1 mL/h via a syringe pump into the reaction solution. The modified silver particles were then directly characterized without additional washing. Characterization Techniques. UV−vis absorption spectra (300− 1300 nm) were recorded using an UV-3600 Shimadzu UV−vis−NIR spectrophotometer. 10 mm optical path length plastic or quartz cells were used. Transmission electron microscopy (TEM) characterization was performed on a Philips CM20 microscope operating at 75 kV and high-resolution transmission electron microscopy (HRTEM) with a JEOL 2200FS microscope operating at 200 kV. One drop of the colloidal solution was deposited on carbon-coated copper grids. The experimental data for tomography were acquired by means of a spherical aberration (Cs) probe corrected JEOL 2100F transmission electron microscope with a field emission gun operating at 200 kV. The acquisition of STEM-HAADF tilt series was performed in the scanning mode, using the HAADF annular detector and a 10 cm camera length corresponding to inner and outer semiangles for the HAADF detector of 60 and 160 mrad, respectively. Details of the experimental conditions for optimized acquisitions have been reported elsewhere.33 Once the acquisition of the tilt series was completed, the images were first roughly aligned using a cross-correlation algorithm. A refinement of this initial alignment and a precise determination of the tilt axis direction were then obtained using the IMOD software.34 The volume of the analyzed object was then calculated using algebraic reconstruction techniques (ARTs)35 implemented in the TomoJ software36 with the number of iterations not exceeding 20. Visualization and quantitative analysis of the final volume were carried out using Slicer37 and ImageJ software.38 To obtain the 3D models corresponding to the nanoparticle surfaces, a data segmentation procedure based on a simple selection of the voxels as a function of their intensities was used.

in terms of shape dispersity. Figure 1 shows the normalized UV−vis spectrum and electron microscopy images of Ag nanoprims generated from seed-mediated growth described in the Experimental Section. The method involved silver-seedcatalyzed reduction of AgNO3 by L-ascorbic acid (AA) in the presence of citrate and PVP at pH = 5.6. Four localized surface plasmon resonance peaks are observed, indicating the presence of thin and flat nanostructures once the seeded growth was initiated. The dominant extinction peak at 700 nm is the inplane dipole resonance mode, whereas the three weak bands near 332, 450, and 500 nm are the out-of-plane quadrupole, out-of-plane dipole, and in-plane quadrupole modes, respectively. These observations are consistent with the characteristic resonance peaks for triangular Ag nanostructures of 50 nm length.6 TEM analysis confirmed that the prevailing shapes were triangular nanostructures (95%) coexisting with hexagonal and circular ones (5%). As observed when they were stacked perpendicular to the substrate, they were thin and flat in agreement with the optical observations. Their thickness was ∼5 nm, i.e., slightly higher than the dimension of the seeds (2.5 nm). Their crystalline structure, in particular, their side-face structure, was finely characterized in order to disclose the mechanism involved in the nanoparticle growth using HRTEM and electron tomography (3D-STEM). HRTEM characterization revealed that each particle was a single crystal. The corresponding fast Fourier transform (FFT) of a triangle aligned along a ⟨111⟩ zone axis is shown in Figure 1. Two sets of spots were identified. The outer set one (triangles) was indexed as {220} planes whereas the inner one (circles) was assigned to the 1/3 {422} forbidden Bragg reflection21 already observed for both silver and gold nanoplates.5 The corresponding reticular spacings were measured on the HRTEM micrograph associated with the previous FFT (Figure 1). The existence of this forbidden reflection reveals the presence of a single {111} stacking fault lying parallel to the flat top and bottom {111} surfaces and extending across the entire nanoplate. The existence of that kind of single stacking fault led to a triangle side constituted by two inclined lateral planes as described by Bögels.39,40 HRTEM as well as electron tomography experiments allowed to identify those lateral planes as {100} ones (Figure 2). This kind of particle shape is deriving from a cuboctahedron composed by six {100} square faces and height {111} triangular ones. It is generated by truncating the vertices of a cube or of an octahedron at 1/2 edge length. The remaining challenge is now to gain understanding of the mechanisms leading to such a geometrical feature. Role of the Different Reagents. A key issue in determining the reaction mechanism is to establish the specific roles of each reagent. We conducted a set of syntheses by excluding one of the reagents (e.g. seeds, metal precursor (AgNO3), reducing agent L-ascorbic acid, and organic stabilizers) while all the other parameters were kept the same. The UV−vis spectra and TEM images of the final colloids produced in some of those different experimental conditions are displayed in the following figures and in the Supporting Information. The rate of reduction of silver ions was followed by monitoring their consumption via pAg measurements. Role of the Seeds. With no premade seeds, a typical Ag nanocrystal synthesis produces irregularly shaped particles with size ranging from 50 to 100 nm (Figure 3). The reaction requires many hours for completion. The standard redox

■

RESULTS AND DISCUSSION Structural Characteristics. Seed-mediated growth in the presence of citrate ions and PVP in an aqueous medium has been continuously improved since its first demonstration by Sun et al.15 It was observed by us and others that the quality or monodispersity of the as-synthesized silver nanoprisms is far from ideal. We found that controlling the rate of supply of the silver precursor allowed us to significantly improve the method 1426

dx.doi.org/10.1021/la4039705 | Langmuir 2014, 30, 1424−1434

Langmuir

Article

into the aqueous solution of silver ions, indicating that about 95% of the silver ions are still free in solution (eq 2). C6H 7O6 + Ag + → C6H6O6 + Ag 0 + H+

Ag n + x Ag + ↔ Ag n + x x +

(1)

(complexation equilibrium) (2)

Ag n + xC6H 7O6 ↔ Ag n

x−

+ xC6H6O6 + x H

+

(3)

Ag n + Ag + + xC6H 7O6 → Ag n + 1 + xC6H6O6 + x H+ (Ag n x −: a seed particle that has already picked up x e−) (4)

The introduction of ascorbic acid led to a sudden drop in the number of free silver ions. The abrupt decrease is attributed to the reduction of Ag+ onto the silver seeds. They possess a large surface area which acts as a substrate for electron-transfer reaction. It clearly helps to efficiently catalyze the decomposition of ascorbic acid41,42 (eq 3). Analysis of the shape of the Ag+ consumption curve showed that it could not be fitted to a first-order or second-order rate law. It must therefore be concluded that several processes contribute to the decay of the Ag+ concentration (eq 4). Kinetically, transfer of an electron to the seeds forms Agn−, followed by reaction Agn− with Ag+ to yield Agn+1. Another kinetic path could consist on the addition of Ag+ to Agn and electron transfer from the ascorbic acid to Agn+1+. The rate of the reduction is controlled both by the rate of the electron pick-up from ascorbic acid by the particle and the migration of silver ion to the seeds. Both reactions efficiently compete with eq 1, even when few seeds are present. We observed that the catalytic reaction even occurred when the catalyst concentration was 4 or 5 orders of magnitude less than the concentration of one of the reagents. At longer times (>100 s), a decrease in the reduction rate is systematically observed due to a continuous decrease of the pH of the reactive medium. Changes in pH as dramatic as 3 or 4 units are observed after introducing ascorbic acid in the absence of stabilizer. The reduction of Ag+ with HAA− produces H+ ions in solution. This induces a deceleration of the reduction rate of silver ion which is consistent with the pH-dependent reduction activity of ascorbate. Control experiments showed the importance of the seed size on the catalytic growth process. Large preformed seeds (10

Figure 2. Top left: ADF-STEM projection at 2.5° tilt angle of a nanoprism (scale bar = 20 nm). 5 nm gold beads were previously deposited on the grid as fiducial markers for alignment. Top right: high-resolution lateral view of the Ag nanoprism laying on its side. Bottom: two outside views of the 3D modeling of the nanoprism. Schematic illustration of a section of one side of the triangle with the identification of surface planes constituting the edges of side faces.

potential of free Ag+/Ag0 (−1.75 VNHE)41 is so negative that the spontaneous formation of Agn clusters is unfavorable under these experimental conditions explaining the slow growth process (eq 1). When ultrafine seeds (2.5 nm) are added to the growth solution, the reaction produces within a few minutes toward plate-like nanoparticles with a narrow size distribution (Figure 3). As explained below, the morphology of the platelike nanoparticles strongly depends on the chemical environment. The change of reaction rate induced by the presence of the seeds was monitored by recording the decrease of redox potential vs silver electrode of an aqueous solution of silver nitrate as a function of time after the addition of ascorbic acid (Figure 3). Note that, before the addition of the reducing agent, no strong change of potential occurred upon addition of seeds

Figure 3. (left) Time evolution of the potential of a silver electrode (vs Ag/AgCl) upon addition of ascorbic acid (0.17 mM) in an aqueous solution of Ag+ (0.2 mM; initial pH = 6.5; after 400 s = 3.8 ; seeds: 1.2 vol %). No stabilizer was added. (right) Morphology dependence of silver nanoparticles grown with or without no-premade seeds (2.5 nm). In the absence of seeds, irregularly shaped nanoparticles are formed. In the presence of the seeds, the silver nanostructures adopt either a disk-like or a triangular shape depending on the chemical environment. 1427

dx.doi.org/10.1021/la4039705 | Langmuir 2014, 30, 1424−1434

Langmuir

Article

showed that they are mostly monocrystalline and contain a limited number of twin planes (Figures S1 and S2). To further elucidate how the growth proceeds (via enlargement of the seeds by mutual combination and/or via deposition of silver atoms), silver precursor was substituted by a gold precursor (KAuCl4) in the step growth. Figure 4 (bottom) shows TEM micrographs of the gold nanostructures produced utilizing tiny Agn seeds (2.5 nm) in a growth solution containing a gold ionic precursor. Two populations of particles are present: large flat nanoparticles and spherical particles of smaller size produced by new nucleation. Close inspection of the flat nanostructures shows that they possess a hole of 2−3 nm diameter in their core, i.e., close to the dimension of the preformed metallic seeds. The phenomenon is due to the wellknown galvanic oxidation process of Agn seeds by gold salt.43 In contrast, the spherical particles, produced from homogeneous nucleation, do not exhibit such a hole. These observations suggest that Au nanoplates emerge from the growth of a single seed and not from agglomerated seeds. A similar scenario might be operative in the case of the formation of Ag nanoplates wherein introduction of a tiny seed leads to the selective reduction of Ag+ on the seed surface. Hence, these overall results suggest each tiny seed gives rise to a single nanoplate. Role of the Capping Agents. As illustrated in Figure 3 (right), the chemical environment of the growth medium has a strong effect on the shape of the plate-like silver nanostructures. In a medium rich in citrate ions, thin nanostructures with sharp corners and high aspect ratio are obtained. The use of citrate ions alone results in triangular nanoparticles, but in order to control the size distribution and avoid the coalescence, it is important to induce the growth in the presence of a weak amount of a protective polymer. In the absence of a protective agent, monodispersed triangular particles are prepared only in an extremely diluted solutions. These observations suggest that the citrate ions strongly affect the growth rate of the {100} faces vs the one of {111} faces of the growing seeds. In a medium rich in PVP, one also obtains plate-like nanostructures, but with highly rounded aspects. This indicates that the PVP also alters the kinetics of the relative growth rates of the {100} and {111} faces, but the difference from each other is lower than that induced by the citrate species. This is assigned to the weak interaction of the macromolecules with the metal precursor and the seeds in comparison to the ones induced by the citrate species as evidenced from pAg measurements (Figure S3). Upon addition of PVP to an aqueous solution of silver nitrate, a slight decrease of the relative concentration of

nm) affect the rate of reaction by at least 1 order of magnitude. For example, it takes a few minutes for complete reduction of a 2.0 × 10−4 M Ag+ solution in the presence of tiny seeds (2.5 nm), while about 60 h is required when large seeds (10 nm) are present in the reactive medium. Moreover, we observed from HRTEM analysis that the large seeds possess a large number of imperfections (stacking faults or twins) (Figure S1). Considering the large amount of energy necessary to change these microstructures, they still remain defective and consequently stuck in their initial morphology. When used during the step growth, the large seeds slowly produce spherical particles which absorb around 400 nm (Figure 4 (top)). In contrast, small

Figure 4. (top) UV−vis spectrum of Ag nanoparticles grown from citrate-stabilized seeds (10 nm), PVP-stabilized seeds (10 and 5 nm), PSS/citrate-stabilized seeds (2.5 nm). (bottom) TEM images of gold nanoparticles grown from PSS-citrate stabilized seeds. Note that in this latter case the experimental conditions have not been optimized to avoid secondary nucleation.

preformed seeds (