S.
Ristig
a,
O.
Prymak
a,
K.
Loza
a,
M.
Gocyla
b,
W.
Meyer-Zaika
a,
M.
Heggen
b,
D.
Raabe
c and
M.
Epple
*a
aInorganic Chemistry and Center for Nanointegration Duisburg-Essen (CeNIDE), Universitaetsstr. 5-7, 45117 Essen, Germany. E-mail: matthias.epple@uni-due.de
bErnst Ruska-Center and Peter Grünberg Institute, Forschungszentrum Jülich GmbH, 52425 Jülich, Germany
cMicrostructure Physics and Alloy Design, Max-Planck-Institut für Eisenforschung, Max-Planck-Str. 1, 40237 Düsseldorf, Germany
First published on 28th April 2015
Bimetallic silver–gold nanoparticles were prepared by co-reduction using citrate and tannic acid in aqueous solution and colloidally stabilized with poly(N-vinylpyrrolidone) (PVP). The full composition range of silver:gold from 0:100 to 100:0 (n:n) was prepared with steps of 10 mol%. The nanoparticles were spherical, monodispersed, and had a diameter of ∼6 nm, except for Ag:Au 90:10 nanoparticles and pure Ag nanoparticles which were slightly larger. The size of the nanoalloys was determined by differential centrifugal sedimentation (DCS) and transmission electron microscopy (TEM). By means of X-ray powder diffraction (XRD) together with Rietveld refinement, precise lattice parameters, crystallite size and microstrain were determined. Scanning transmission electron microscopy (STEM) combined with energy-dispersive X-ray spectroscopy (EDX) and electron energy loss spectroscopy (EELS) showed that the particles consisted of a gold-rich core and a silver-rich shell. XRD and DCS indicated that the nanoparticles were not twinned, except for pure Ag and Ag:Au 90:10, although different domains were visible in the TEM. A remarkable negative deviation from Vegard's linear rule of alloy mixtures was observed (isotropic contraction of the cubic unit cell with a minimum at a 50:50 composition). This effect was also found for Ag:Au bulk alloys, but it was much more pronounced for the nanoalloys. Notably, it was much less pronounced for pure silver and gold nanoparticles. The microstrain was increased along with the contraction of the unit cell with a broad maximum at a 50:50 composition. The synthesis is based on aqueous solvents and can be easily scaled up to a yield of several mg of a well dispersed nanoalloy with application potential due to its tuneable antibacterial action (silver) and its optical properties for bioimaging.
In the case of bimetallic Ag:Au nanoparticles, an important question concerns the actual distribution of the metals inside the nanoparticles, and the resulting effects on their chemical, physical and crystallographic properties. It has been recognized that a nanoparticle is not just a small homogeneous single crystal, and that standard crystallography needs to be extended to map such structural and/or compositional inhomogeneities (“the nanostructure problem”).19 This in turn will determine their chemical and biological effects, e.g. the release of silver ions for antibacterial activity20 and the presence of gold on the nanoparticle surface for a stable subsequent covalent functionalization.
X-ray diffraction is well suited to determine the crystallite size of nanoparticles,21 but additional microscopic, spectroscopic and modelling techniques need to be employed to understand the structure of nanoparticles.4,21–24 The area of nanoalloys has been extensively reviewed by Ferrando et al.25 and also with a special focus on thermodynamics by Calvo.26
The internal structure of nanoalloys strongly depends on the synthesis route and also on their surface functionalization. As a comprehensive experimental study to supplement a variety of theoretical predictions, we have prepared a complete series of monodispersed small Ag:Au nanoalloys, reduced with citrate/tannic acid and stabilized with poly(N-vinyl pyrrolidone), over the entire composition range and analysed their structure using a wide range of methods.
The synthesis is based on aqueous solvents and can be easily scaled up to a yield of several mg of nanoalloys as required for biomedical applications, e.g. as antibacterial agents or for bioimaging.
300 mL of degassed water was heated to 100 °C. A cumulative volume of 3 mL of 10 mM solutions of HAuCl4 and AgNO3 (total metal content 30 μM), depending on the intended composition of the nanoparticles, was added. A mixture of 4.5 mL of a solution of 2 wt% trisodium citrate (77.5 mM) and 4.5 mL of a solution of 5.6 mg mL−1 tannic acid (3.3 mM) was added using a pipette under vigorous stirring. The reaction mixture was then stirred for another 10 min at 100 °C.
Further functionalization was carried out by adding 2 mL of a solution of poly(N-vinylpyrrolidone) (PVP; 30 mg mL−1) to the unpurified dispersion.27 After stirring for at least two more hours, the nanoparticles were separated from the unreacted material by ultracentrifugation (30000 rpm, 60000g, 30 min) and redispersed in ultrapure water by vortexing. The purification step was repeated twice to remove all synthesis by-products. Prior to powder diffraction analyses and transmission electron microscopy, the samples were freeze-dried.
Scanning transmission electron microscopy (STEM) was performed using two FEI Titan microscopes, both equipped with Cs-probe correctors (CEOS Company) and high-angle annular dark field (HAADF) detectors. They were operated at 200 and 300 kV, respectively. “Z-Contrast” conditions were achieved using a probe semi-angle of 25 mrad and an inner collection angle of the detector of 70 mrad. Electron energy loss (EEL) spectra were recorded using a Gatan image filter Tridiem 866ERS system analysing the Au M4,5 and Ag M4,5 edges. The elemental mapping using energy-dispersive X-ray spectroscopy (EDX) was conducted on a probe-corrected FEI Titan 80–200 “ChemiSTEM” electron microscope equipped with four symmetrical SDD detectors.
Differential centrifugal sedimentation (DCS) was performed using a CPS Instruments Disc Centrifuge DC 24000 at 24000 rpm. Two sucrose solutions (8 wt% and 24 wt%) were used for a density gradient which was supplemented with 0.5 mL of N-dodecane as a stabilizing agent. The calibration standard was a poly(vinylchloride) (PVC) latex in water with a particle size of 476 nm provided by CPS instruments. Calibration was carried out prior to each run. A sample volume of 100 μL was used for each experiment.
Ultraviolet-visible spectroscopy (UV-Vis) was performed on a Varian Cary 300 instrument. Suprasil® cuvettes with a sample volume of 3.5 mL were used.
Atomic absorption spectroscopy for elemental analysis (Ag, Au) was carried out on a Thermo Electron M-Series spectrometer with a graphite tube furnace according to DIN EN ISO/IEC 17025:2005 after dissolving the particles in aqua regia.
X-ray powder diffraction was carried out on a Bruker D8 Advance instrument in Bragg–Brentano mode with Cu Kα radiation (1.54 Å; 40 kV and 40 mA) using a silicon single crystal with a crystallographic (911)-plane as a sample holder to minimize scattering. For better homogenization, the dried powder samples were redispersed with ethanol on the silicon surface and then investigated in the range from 5 to 90° 2θ with a step size of 0.01° 2θ and a counting time of 0.6 s at each step. Rietveld refinement using the program package TOPAS 4.2 from Bruker was performed to determine the average crystallite size, microstrain, and the lattice parameters. The size of the crystallites was calculated using the Scherrer equation:28
For each Rietveld refinement, the instrumental correction as determined using a standard powder sample LaB6 obtained by NIST (National Institute of Standards and Technology) as a standard reference material (SRM 660b; a(LaB6) = 4.15689 Å) was taken into account. Each nanoparticle sample was thoroughly mixed with about the same amount of LaB6 for precise determination of the lattice parameters of the nanoparticles, including an exact determination of the sample displacement. The patterns of elemental silver (#4-0783) and gold (#4-0784) as listed in the ICDD database were used as references. As reference materials, we also studied microcrystalline powders of pure silver and gold, respectively (Fluka). The silver powder was annealed for 3 h at 700 °C to ensure microcrystallinity and the absence of microstrain; the gold powder was measured as obtained. All powder diffraction data were recorded at ambient temperature (25 ± 2 °C) to exclude a variable thermal expansion.
Targeted nominal composition Ag:Au, according to synthesis/mol%:mol% | Ag/mol% (AAS) | Au/mol% (AAS) |
---|---|---|
0:100 | 0 | 100 |
10:90 | 8 | 92 |
20:80 | 19 | 80 |
30:70 | 26 | 74 |
40:60 | 37 | 63 |
50:50 | 48 | 52 |
60:40 | 57 | 43 |
70:30 | 66 | 34 |
80:20 | 80 | 20 |
90:10 | 89 | 11 |
100:0 | 100 | 0 |
Differential centrifugal sedimentation gave the average hydrodynamic diameter of the water-dispersed particles which represents the size of the solid nanoparticle core and the hydrated layer of the PVP. The results are in good agreement with the diameter as determined by TEM, confirming that the particles are well dispersed in water and not agglomerated (Table 2).
Ag:Au (n:n, nominal composition) | D (TEM)/nm | D (DCS)/nm | Crystallite size (XRD)/nm |
---|---|---|---|
0:100 | 6.0 ± 0.7 | 4.8 | 4.7 ± 0.1 |
10:90 | 6.5 ± 1.1 | 5.5 | 8.3 ± 0.3 |
20:80 | 6.9 ± 1.0 | 5.4 | 8.9 ± 0.9 |
30:70 | 7.1 ± 1.0 | 5.5 | 9.2 ± 0.6 |
40:60 | 7.0 ± 0.5 | 5.6 | 8.3 ± 0.6 |
50:50 | 6.9 ± 0.7 | 5.7 | 6.9 ± 0.6 |
60:40 | 6.9 ± 0.7 | 5.5 | 8.5 ± 0.7 |
70:30 | 7.1 ± 0.9 | 5.3 | 6.6 ± 0.3 |
80:20 | 8.5 ± 1.0 | 6.8 | 8.5 ± 0.4 |
90:10 | 11.5 ± 0.7 | 8.1 | 8.2 ± 0.2 |
100:0 | 33.6 ± 5.7 | 29.4 | 14.0 ± 0.2 |
By X-ray powder diffraction, extended by Rietveld refinement, the average crystallite size of the particles was obtained (Table 2 and Fig. 2). Note that XRD gives the crystallite size, i.e. the average diameter of the coherently scattering domains, which is not necessarily identical with the domain size that is visible in the TEM.21 Starting from pure gold, the crystallite size is almost constant and in good agreement with the particle size obtained from TEM and DCS, respectively. This observation indicates that each particle consists of only one crystallite, i.e. is an individual single crystal. The obvious increase in size of Ag:Au 90:10 and pure silver nanoparticles then results from a change in the nucleation and growth mechanism at high silver concentrations. Twinning appears to start at 90 mol% silver, with one particle containing more than one crystallite. For pure silver, it is noteworthy that the particle size increases faster than the crystallite size, underscoring the results of TEM. However, high resolution STEM images also indicate some twinning at a lower silver content (see below).
Fig. 2 Average particle size obtained from disc centrifugal sedimentation (DCS; hydrodynamic diameter), transmission electron microscopy (TEM; metallic core), and X-ray powder diffraction (XRD; crystallite size) against the silver content from elemental analysis (AAS) (Table 1). |
The fact that the hydrodynamic diameter is almost the same as the diameter of the metallic core confirms the dispersed state of the nanoparticles when immersed in water. Note that the hydrodynamic diameter as determined by DCS is systematically underestimated because of the hydration shell.21,30 The fact that the crystallite size determined by XRD is the same as the one observed by TEM or DCS shows that the particles are not twinned up to about 80 mol% Ag, at least based on the results of X-ray powder diffraction. At very high silver content, the crystallite size is smaller than the particle diameter, indicating twinning.
Fig. 3 shows high-resolution HAADF-STEM images of Au–Ag nanoalloys at various compositions and the corresponding EDX maps. Some of the HAADF-STEM images show the atomic lattice of nanoparticles oriented along low indexed lattice orientations with respect to the electron beam and indicate the presence of twin boundaries (e.g. arrowhead in Fig. 3c). Individual particles showing twin boundaries were observed at all compositions. The EDX spectrum maps at Ag:Au 20:80 and 50:50 show pronounced Au and Ag enrichment in the centre and in the shell of the nanoparticle, respectively (Fig. 3b and d). The EEL spectrum profiles were taken across the nanoparticles (Fig. 4) and show the presence of one strong Au signal around the centre and two pronounced Ag signals at the outermost parts of the particles. The EEL spectrum profiles and the EDX maps clearly demonstrate the presence of a core–shell structure with an Au-rich core and an Ag-rich shell (Fig. 4). Cores and shells are differently pronounced at different compositions of the nanoparticles. According to EDX mapping, the chemical composition is in good agreement with the AAS data (Table 2): Ag:Au nominal 20:80, found 19:81 (Fig. 3b); Ag:Au nominal 50:50, found 58:42 (Fig. 3d); and Ag:Au nominal 80:20, found 87:13 (Fig. 3f).
High-precision X-ray diffraction gave the crystallite size and microstrain after Rietveld refinement of peak broadening (Fig. 5 and 6).
Fig. 6 Representative Rietveld refinement of Ag:Au 50:50 nanoalloys. Top: experimental curve; bottom: difference plot. |
Additional X-ray measurements were carried out after the addition of LaB6 as an internal standard. This allowed the determination of the lattice parameters of the alloyed Ag:Au with very high precision (Fig. 7; see ref. 21 for a diffractogram of a sample of silver nanoparticles, mixed with LaB6). In the face-centred cubic lattice, there is only the cubic lattice parameter that needs to be determined.
Fig. 7 Lattice parameter as a function of composition for Ag:Au nanoalloys, including own data and literature data obtained from bulk materials (Lubarda et al., bulk AgAu alloys in 11 compositions and the pure metals;34 Venudhar et al., bulk Ag:Au 50:50 alloys;35 Naidu et al., bulk Ag:Au 50:50 alloys;36 microcrystalline silver;21 microcrystalline gold, this work). |
Surprisingly, we found distinct negative deviations from Vegard's linear rule for the lattice parameter a (face-centred cubic crystal system). For Ag:Au 50:50 nanoalloys, the difference was the largest (−0.024 Å). In general, the negative deviations were much larger than comparable deviations that were found for the corresponding AuAg bulk alloys (maximum −0.005 Å for Ag:Au 50:50). This negative deviation cannot be ascribed to the nanoscopic size itself because nanocrystalline silver and gold did not show this effect, i.e. their lattice constants were the same as those of the pure microcrystalline metals (Fig. 7). This effect is clearly present in all Ag:Au alloys, but much more pronounced in the nanostate.
Deviations from Vegard's rule were also reported for PtCo nanoalloys,31 PtNi nanoalloys,32 AuCu nanoalloys,33 and AgAu nanoalloys.33 Notably, in the latter case, the as-prepared nanoparticles followed Vegard's rule, but after annealing them at 240/400 °C, a negative deviation was found.33 This observation points to a metastable crystalline state directly after synthesis and the possible presence of internal stress and strain which are relaxed during annealing, leading to a more stable state. This relaxed phase state seems to be characterized by a contracted lattice parameter in comparison to a linear atomic mixture of the individual atoms.
This lattice contraction is isotropic, as a closer analysis of the individual diffraction peaks clearly shows (Fig. 8).
This lattice contraction is reflected by a corresponding microstrain inside the nanoparticles (Fig. 9).
Fig. 9 Microstrain inside the Ag:Au nanoalloys as obtained by X-ray diffraction as a function of composition. |
UV-Vis spectroscopy shows that the particles indeed consist of alloyed Ag:Au nanoparticles and are not of a core–shell nature with a sharp border between the gold-rich core and the silver-rich shell (Fig. 8). There is a continuous change in the UV-Vis absorption maximum as a function of the Ag:Au ratio (Fig. 10 and 11), underscoring the alloy character with a gradual radial change in the composition. A core–shell nanoparticle with a sharp change in composition would show two distinct plasmon resonance bands.18,37–40
Fig. 11 Maximum of the UV-Vis absorption curve (surface plasmon resonance) as a function of particle composition. |
Core–shell structures in bimetallic nanoparticles were also found by high resolution transmission electron microscopy, together with element-sensitive techniques like EDX or EELS, e.g., for AgCo nanoalloys (Ag core, Co shell),41 FePt nanoalloys (Fe-rich core, Pt-rich shell),40,42 CoPt nanoalloys (Co core, Pt shell),32 PtNi nanoalloys (Pt core, Ni shell),32 and PdPt nanoalloys (Pd core, Pt shell).21 For Ag:Au nanoalloys, different results were reported. Particles from co-reduction of Ag+ and AuCl4− showed a gold-rich core, as found here.16 In a similar process after annealing at 240/400 °C, a random AgAu alloy was formed.23 Pulsed laser ablation from bulk alloys also led to nanoalloys with a random and uniform distribution of silver and gold.17,43 In the case of a gold seed-induced silver reduction, core (Au)–shell (Ag) nanoalloys were prepared. However, after heating to 930 °C, single-crystalline and homogeneous AgAu nanoalloys were obtained.12 Note that this may not only point to some lattice relaxation but also just represent the equilibrium structure at 930 °C which was frozen during cooling. Similar results were reported by Shore et al. for gold core–silver shell nanoparticles18 and Wang et al. for silver core–gold shell nanoparticles.40 In both cases, core–shell nanoalloys were prepared by reduction in non-aqueous solvents, followed by heating for several hours at 250 and 100 °C, respectively. After heating, homogeneous nanoalloys were formed as derived from UV spectroscopy.18,40 By sequential plasma vapour deposition and subsequent thermal treatment, polycrystalline core (Au)–shell (Ag) nanoparticles were obtained.44
The occurrence of a gold-rich core and a silver-rich shell in the nanoparticles from co-reduction of silver and gold salts indicates a faster reduction of gold due to its more noble character. Silver is reduced later because it is less noble on the surface of a gold core. However, the fact that we did not observe a sharp separation of a gold core and a silver shell but rather a smooth gradient indicates that the reduction of silver and gold is occurring simultaneously, but simply at different rates. The fact that gold and silver form a continuous row of mixed crystals certainly contributes to the formation of this smooth gradient structure. However, this leads to internal stress and a hitherto unknown lattice contraction in the case of alloys.
Theoretical studies have been widely employed to understand and predict the internal structure of nanoalloys. The reason for the deviation in bulk alloys is the fact that the effect of the surface energy cannot be neglected on the nanoscale, and that therefore the structure may deviate from the bulk phase.19,25,26,45 Rossi, Curley et al. found by genetic algorithm simulations that gold tends to enrich the surface of a 34-atom AgAu cluster.46,47 Cerbelaud et al. applied an atomistic DFT model to AgAu clusters with 34 to 201 atoms and derived a number of ordered structures.48 Miotto et al. have shown by DFT calculations in the gas phase for clusters up to 961 atoms that in the case of a bimetallic gold–silver nanoparticle, a silver core is more favourable than a gold core.49
Typically these computations were performed on naked nanoalloys or clusters, i.e. the effect of the ligand shell was not taken into account. This will of course influence the surface energy. Furthermore, modelling techniques typically yield equilibrium structures at a given set of parameters (temperature, pressure) and do not take into account kinetic effects during the synthesis (like nucleation). Nanoparticles also show a distinct reactivity and capacity for internal restructuring due to their small size.50
These differences underscore that the internal order of a bimetallic nanoparticle depends on the synthesis route (e.g. wet-chemical synthesis vs. laser ablation or seeded growth routes) and can also be changed by post-synthetic thermal treatment. This points to the mobility of the atoms in the nanocrystalline lattice in the case of a gold–silver alloy.
As the optical (absorption by plasmon resonance) and biological properties (e.g. the antibacterial action by the release of silver) strongly depend on the nanostructure, these results open a way to better understand the nature and action of custom-made silver–gold nanoalloys.
This journal is © The Royal Society of Chemistry 2015 |