J. Phys. Chem. B 2000, 104, 9683-9688
9683
Electrochemical Synthesis of Silver Nanoparticles L. Rodrı´guez-Sa´ nchez,* M. C. Blanco, and M. A. Lo´ pez-Quintela Department of Physical Chemistry, UniVersity of Santiago de Compostela, E-15706 Santiago de Compostela, Spain ReceiVed: May 11, 2000; In Final Form: July 21, 2000
An electrochemical procedure, based on the dissolution of a metallic anode in an aprotic solvent, has been used to obtain silver nanoparticles ranging from 2 to 7 nm. By changing the current density, it is possible to obtain different silver particle sizes. The influence of the different electrochemical parameters on the final size was studied by using different kinds of counter electrodes. The effect of oxygen presence in the reaction medium as well as the type of particle stabilizer employed have also been investigated. In some conditions an oscillatory behavior is observed. Characterization of particles was carried out by TEM and UV-vis spectroscopy. The maximum and the bandwidth of the plasmon band are both strongly dependent on the size and interactions with the surrounding medium. The presence of different silver clusters was detected by UVvis spectroscopy. By using this technique, the existence of an autocatalytic step in the synthesis mechanism is proposed.
1. Introduction
2. Experimental Section
Metal small particles have received increasing attention in the last years.1,2 These particles are usually prepared by chemical reduction of metal salts3. Reetz et al.4 have described an electrochemical procedure to obtain particles in which a metal sheet is anodically dissolved and the intermediate metal salt formed is reduced at the cathode, giving rise to metallic particles stabilized by tetraalkylammonium salts. Some of the advantages of this method are the high purity of the particles and the possibility of a precise particle-size control achieved by adjusting the current density. The primary aim of this work is to adjust the general scheme to the synthesis of silver nanoparticles. For the optimization, it is necessary to take into account the following parameters: the choice of the right solvent, supporting electrolyte, type of electrode, and the current density. It is wellknown that colloidal dispersions of metals exhibit absorption bands in the UV-vis region, due to collective excitations of the free electrons (surface plasmon band).5 Optical properties of clusters have been investigated for many years.6 For a single cluster, the width and position of the plasmon modes are influenced by the shape, volume, and surface/interface effects. In many cluster samples, interactions between clusters and the surrounding media, due to the presence of stabilizers or to the existence of cluster size or shape distributions, made the understanding of the optical behavior difficult.7 Charge transfer from/toward the ligands may change the electron density, and lead to blue or red shifts of the plasmon resonance band. Henglein and co-workers8 have demonstrated the influence of redox reactions upon the position when reduction or oxidation take place, showing blue and red shift, respectively. In the present work, the optical properties of the obtained silver colloids are correlated with the silver particle size. The evolution of the optical properties has been used for the understanding of some kinetic aspects of the synthesis mechanism.
All chemicals were of analytical grade and used without further purification: H2SO4 from Merck, tetrabutylammonium bromide (TBABr) and acetate (TBAAcO) from Aldrich, aluminum oxide, alpha; 99.99%, 1.0 µm from Alfa, were employed. An Autolab PGSTAT 20 potentiostat was used both in the synthesis and the electrochemical study. Temperature was kept at 25 ( 0.1 °C using a Grant thermostatic bath. All potentials were measured against Ag/AgCl reference electrode. The experiments were carried out in a standard Metrohm electrolysis beaker containing a sacrificial silver sheet as anode (counter electrode), and the same size platinum sheet was used as cathode (working electrode). These two electrodes were vertically placed face-to-face inside the cell. Platinum electrode was handpolished with 1 µm alumina powder to a mirror-like finish. Then, the electrode was activated by triangular potential cycling between 1.35 and -0.15 V, at a scan rate of 500 mV s-1 for 5 min in 1.0 M H2SO4. Before each experiment, the silver electrode was hand polished by fine grade emery paper and washed with bidistillated water and a small amount of acetone. The electrolyte solution consisting of tetrabutylammonium bromide 0.1 M dissolved in acetonitrile was deaerated by bubbling nitrogen for about 15 min, keeping an inert atmosphere during the whole process. A freshly prepared dissolution was used in each experiment. Strong stirring was kept during the galvanostatic electrolysis. TEM measurements were performed on a transmission electron microscope JEOL 2000, working at 200 kV and equipped with an EDXA detector. Samples were prepared by adding acetonitrile to a fraction of the obtained sol, and a droplet of it was placed on a carbon-coated copper grid covered with an acetatecellulose polymer. Silver particle formation was confirmed by dispersive X-ray energy analysis. Size data were averaged over 100-200 particles from different TEM micrographs To obtain the absorption spectra, ultracentrifugation of the sol at 10000 rpm, for 10 min in a Sigma 2-15 ultra-centrifuge, was carried out to make faster the sedimentation process of the
* Author to whom correspondence should be addressed. Fax: 00 34 81 595 012. E-mail:
[email protected].
10.1021/jp001761r CCC: $19.00 © 2000 American Chemical Society Published on Web 09/23/2000
9684 J. Phys. Chem. B, Vol. 104, No. 41, 2000
Rodrı´guez-Sa´nchez et al.
particle aggregates, which are always present during the particle synthesis. Spectra from the supernatant were recorded on a diode-array Hewlett-Packard HP8452 spectrophotometer in 1 cm light path length cuvettes. 3. Results and Discussion Optimization of Synthesis Variables. (1) SolVent and Supporting Electrolyte. Although the solvent proposed in the method described by Reetz et al.4 was a mixture of acetonitrile and tetrahydrofurane, this mixture was found unsuitable for the silver synthesis, because the presence of tetrahydrofurane induces the aggregation of the metal particles obtained, which can be detected by the blue color of the solution. The same color was found when silver colloidal particles are agglomerated in the presence of hydrogen peroxide.9,10 For this reason, pure acetonitrile was used as solvent. Its aprotic character is necessary because when protons are in the medium, passivation of the silver anode occurs, and the synthesis cannot take place. It is necessary to have a supporting electrolyte as well as a stabilizer of the particles in the medium. When synthesis is carried out in 0.1 M NaClO4 (a supporting electrolyte recommended in electrochemical literature when acetonitrile is the solvent), no particles are obtained; the electroreduction produces silver deposition on the cathode. By using tetrabutylammonium bromide or acetate, which act as supporting electrolytes and as stabilizers, silver nanoparticles are obtained. Cyclic voltammetries of these two salts were recorded. As it can be seen in Figure 1a, both salts are electrochemically inert in the range 0.8-1.8 V. The processes that occur at higher and lower potentials are the following: Anodic potentials: -
Cathodic potentials: ‚
(1) 2 CH3CO2 f 2 CH3CO2 + 2 e -
‚
(2) Br f Br + 1 e
-
NBu4+ + 1e- f NBu3 + Bu‚
-
2 Br‚ f Br2
The solvent must be oxygen free, to avoid the oxidation of the small metal particles produced; but this is not the only reason: in silver reduction, oxygen interferes with the electrochemical process. Figure 1b shows the potential evolution when a supporting electrolyte 0.1 M tetrabutylammonium bromide is used, the solution initially deoxygenated without keeping the inert atmosphere during the electrolysis. In the beginning, the synthesis occurs at -2.75 V but as the time goes by (about 1000 s), oscillations in the potential can be observed, and finally (about 2200 s) the potential is not high enough to let the silver reduction take place and no particles are obtained. Figure 1c displays linear voltammetries of the system, using two platinum electrodes, in the presence of different amounts of O2. The presence of a reduction peak about -1.5 V, shows clearly the oxygen reduction at this potential, therefore the oscillations in the potential can be attributed to the diffusion of small amounts of O2 from the atmosphere into the solution. The oscillatory behavior disappears when tetrabutylammonium acetate is employed, which clearly shows a participation of bromide in this oscillatory phenomenon. It is also observed that the solution becomes slightly yellow as the synthesis takes place at any of the current densities employed. This also happens when silver is not in the medium. Figure 2 shows the spectrum of the bromide salt solution after electrolysis using platinum electrodes, under the same experimental synthesis conditions. A wide absorption band, attributed to the anodic formation of Br2, at 364 nm is evident. Taking
Figure 1. A) Linear voltammetries of supporting electrolytes using Pt electrodes. B) Potential variation with time in TBA bromide 0.1 M, without inert atmosphere. Working electrode: Pt, Counter electrode: Ag. Current density: -1.25 mA cm-2. C) Linear voltammetries using Pt electrodes in TBA bromide 0.1 M. (A) O2 free. Bubbling O2: (B) 10 s. (C) 15 s.
Figure 2. Spectrum of tetrabutylammonium bromide after Br2 electrolysis at current density of -3 mA cm-2.
into account that the silver plasmon band appears at about 400 nm, this secondary reaction is undesirable for any spectroscopic study of silver particles. For this reason, acetate salt was employed in this work. A typical synthesis of silver nanoparticles is carried out in acetonitrile using 0.1 M tetrabutylammonium acetate. Under these conditions, upon applying the current, the electrolyte
Electrochemical Synthesis of Silver Nanoparticles
J. Phys. Chem. B, Vol. 104, No. 41, 2000 9685
Figure 3. SEM photographs of the platinum surface after silver reduction at (A) low current density, (-1.4 mA cm-2); (B) high current density, (-7 mA cm-2); and (C) magnification of B).
TABLE 1: Summary of the Reduction Products Obtained at the Working Electrode Depending on the Experimental Conditions cathode nature platinum platinum aluminum
conditions low current density (-1.4 mA cm-2) (Figure 3.A). high current density (-7 mA cm-2) (Figure 3.B) -2 mA cm-2
products silver nanoparticles and thin white film silver nanoparticles and dendritic deposition of black film no particles, only thin white film
becomes dark yellow and as the reduction process goes on, a black precipitate is formed. This precipitate can be redispersed by dilution with acetonitrile and the dark yellow color reappears. Therefore, this color change can be explained because, when the concentration of silver colloidal particles is too high, interactions between the stabilizer chains increase and flocculation takes place. These flocculated particles can be redispersed again when enough solvent is added. (2) Current Density and Type of Electrode. Table 1 shows the reduction products obtained in different conditions, employing platinum or aluminum as cathode. The morphology of the cathodic deposits obtained using platinum as cathode at different current densities is shown in Figure 3. These results are in accordance with other results already reported in the literature: 11,12,13 when the growth of a new phase takes place close to the thermodynamic equilibrium (low current densities), a rather compact phase is formed. Conversely, when the growth conditions are far from equilibrium (high current densities), irregular aggregates may be formed. The study of these parameters shows the existence of a competition between two different cathode surface processes that are summarized in Figure 4: the particle formation, by reduction and stabilization of silver ions by the tetrabutylammonium salt, and the film deposition at the cathode surface (see Table 1). This second process limits the yield of the particle synthesis, and must be minimized, because when the electrode surface is totally covered by silver deposition, the only process that occurs is the silver deposition. Table 2 shows the crystallographic characteristics and atomic radius of silver and the different materials employed. The similarity between silver and aluminum may explain the high tendency to the silver deposition on aluminum and the fact that, in this case, no particles are obtained. On the contrary, the difference in radius and lattice parameters for platinum and silver leads mainly to particle formation. Characterization. Figure 5 shows TEM photographs as well as the corresponding size distributions of the biggest and the smallest particles obtained. To confirm the composition, EDXA was carried out, showing that only the black images come from silver (particles), whereas the gray areas come from the
Figure 4. Schematic picture showing the competition of two processes: (1) silver particle formation, (2) silver deposition.
TABLE 2: Some Relevant Characteristics of Silver and the Employed Cathodic Materials species
crystallographic structure/ lattice parameter (Å)
atomic radius (Å)
Ag Al Pt
fcc; a ) 4.08626 fcc; a ) 4.04959 fcc; a ) 3.9240
1.444 1.431 1.380
stabilizer. Table 3 shows the average size of the particles obtained for different current densities. The observed size decreases as the current density is increased. This is an expected result, previously found and explained by Reetz4 for other metals. Spectroscopic Study. The optical behavior of silver colloid was studied by UV-vis spectroscopy. This shows a clearly resolved surface plasmon resonance, well separated from the interband transition.14 In Figure 6 a typical absorption spectrum from the silver sol is shown. A broad plasmon absorption band centered at 444 nm can be observed. This peak appears in the range 420-444 nm, depending on the average size of the colloidal particle. Table 4 shows the maximum of the plasmon band position as a function of the current density employed. A red shift can be observed as the current density is increased and therefore as the particle size decreases. This red shift, usually found in small metal particles, can be explained by the spilling out of the
9686 J. Phys. Chem. B, Vol. 104, No. 41, 2000
Rodrı´guez-Sa´nchez et al.
Figure 5. TEM images and size distribution of silver nanoparticles synthesized in TBA acetate 0.1 M in acetonitrile, at various current densities.
TABLE 3: Mean Particle Size Obtained from Different Current Densities j (mA cm-2)
L (nm)
-1.35 -2.85 -4.14 -6.90
6 ( 0.7 4.5 ( 0.8 3.2 ( 0.6 1.7 ( 0.4
conduction electrons, which is more important as the particle size decreases.7 To obtain the bandwidths, spectra were fitted to Lorentzians according to the simple free-electron theory.15,16 Figure 7 shows the bandwidth linear increase with the particle size. This linear dependence is in agreement with a modified version of Drude’s theory, introduced by Doyle17 and Kreibig et al.18 For small particles, these authors assume that the mean free path of the electrons is limited by the particle size. It should be noted that other theories (quantum confinement,15,16 quantum box model,19 etc.) also predict the 1/R law of the plasmon bandwidth.
Figure 6. Absorption spectrum, taken from the electroreduced sample at -5.8 mA cm-2.
We have also studied the evolution with time of the spectra during the synthesis carried out at a current density of -5.83 mA cm-2. More than thirty spectra were recorded, some of them are shown in Figure 8. In the beginning, a band at about 315 nm, a shoulder at 375 nm, and a weak absorption band at 437
Electrochemical Synthesis of Silver Nanoparticles
J. Phys. Chem. B, Vol. 104, No. 41, 2000 9687
Figure 8. Spectra obtained during the nanoparticle synthesis at -5.83 mA cm-2. Inset: Plot of ln(a/(1 - a)) vs time (a ) Abst/Abs∝). Figure 7. Linear variation of the half-width (HW) with the inverse of the particle radius.
TABLE 4: Variation of λmax with Current Density j (mA cm-2)
λmax
-1.35 -1.43 -2.85 -3.00 -3.29 -5.83 -6.90
422 426 430 430 432 438 444
nm are observed. The intensity of this last band increases as time goes by. The band at 315 nm could be attributed20 to the existence of Ag2+, and the band at 375 nm to long-lived clusters.21 The band about 420-440 nm is commonly attributed to metallic silver particles. All spectra were fitted to three Lorentzians. Table 5 summarizes some of the results. In the beginning no particles are observed (band at 400 nm), and the main contribution to the absorption is due to the small silver clusters. At long times the relative contribution due to the absorption of this kind of clusters is small, but even at longer times three Lorentzians are needed for the obtaining of good fits. This fact shows the existence of the initial clusters at any time, what is in agreement with the suggestion that in the synthesis method employed, clusters are continuously generated. The evolution of the optical density at 430 nm means a sigmoidal shape. This suggests that an autocatalytic growth could be involved.22 A possible explanation can be given in the following terms. The first silver ions are reduced at the platinum surface, forming the initial clusters. Some of these clusters diffuse to the bulk and can be detected by UV-vis spectroscopy (see Figure 8). As reaction proceeds, the number of clusters increases and the probability that some clusters be located near the electrode surface is higher. These clusters can be charged at the same potential (by tunneling). The reduction on those charged clusters will be easier than on the platinum
surface, giving rise to an autocatalytic effect. This explanation is in agreement with the observed shift of the cathodic potential toward less negative values as the reaction proceeds. The inset of Figure 8 shows a typical autocatalytic behavior. From the fit, a rate constant of kobs ) (2.6 ( 0.2) × 10-3 s-1 is obtained. It is interesting to note that similar kinetic results were obtained9 for silver chemical reduction in aqueous media: (k ) (1-5) × 10-3 s-1), as well as for silver reduction in microemulsions.23,24 It has been pointed out that silver electrodeposition on the cathode takes place simultaneously, diminishing the effective surface for particle production. This is responsible for the leveling off observed in the absorbance at long time. When the surface is totally covered by the silver electrodeposit, the production of particles finally stops. 4. Conclusions Silver nanoparticles obtained by electroreduction of anodically solved silver ions in acetonitrile containing tetrabutylammonium salts (TBA bromide or TBA acetate) have been studied. An interesting oscillatory behavior is observed in some conditions. The optimization of the general synthesis method has been carried out. The current density plays an important role, not only on particle size but also on the efficiency of the process. The cathode nature seems to be also a decisive parameter, because only particles have been obtained when platinum instead of aluminum was employed. The spectra of silver sols show the presence of two different silver clusters. The smallest clusters are present from the beginning to the end of the process, and this suggests that clusters are continuously generated. From the plot of the optical density of the plasmon band versus time an autocatalytic effect has been found, which is explained assuming that the reduction can proceed easier on charged clusters located near the surface. A linear variation of the bandwidth with the inverse of the particle radius (predicted by Drude’s model) as well as a red shift of the plasmon band (due to the electron spill out) with decreasing particle size are observed.
TABLE 5: Fits of the UV-Vis Spectra Obtained during the Synthesis Performed at - 5.83 mA cm-2 t (s) 250 500 860 1100 1500 2300 2500 2900 3200 3600 a
λmax1 (nm)a
441.7 ( 0.8 442.5 ( 0.5 438.4 ( 0.9 437.2 ( 0.7 435.4 ( 0.8 437 ( 1 434.7 ( 0.9 433.1 ( 0.4
HW1 (nm)d
58 ( 5 47 ( 3 79 ( 6 83 ( 3 91 ( 3 89 ( 7 100 ( 3 104 ( 13
λmax2 (nm)b
372 ( 1 372.4 ( 0.7 372 ( 1 375 ( 1 373 ( 1 383 ( 9 373 ( 9 381 ( 25
λmax1 ) 437.5 nm. b λmax2 ) 375.2 nm. c λmax3 ) 315.5 nm. d HW) half-width.
HW2 (nm)d
λmax3 (nm)c
HW3 (nm)d
62 ( 6 40 ( 3 74 ( 7 88 ( 9 93 ( 9 118 ( 50 106 ( 27 231 ( 95
316 ( 3 314 ( 1 306 ( 12 314 ( 2 307 ( 9 321 ( 30 324 ( 21 316 ( 38 320 ( 40 317 ( 50
46 ( 8 48 ( 4 66 ( 17 41 ( 6 59 ( 16 30 ( 18 30 ( 15 42 ( 27 43 ( 29 40 ( 30
9688 J. Phys. Chem. B, Vol. 104, No. 41, 2000 Acknowledgment. Partial financial support by the Xunta de Galicia is gratefully acknowledged (PGIDT99PXI20905B). References and Notes (1) Ozin, G. A. AdV. Mater. 1992, 4, 612. (2) Henglein, A. Chem. ReV. 1989, 89, 1861. (3) Lo´pez-Quintela, M. A.; Rivas, J. Curr. Opin. Colloid Interface Sci. 1996, 1, 806. (4) Reetz, M. T.; Helbig, W. J. Am. Chem. Soc. 1994, 116, 7401. (5) Wilcoxon, J. P.; Williamson, R. L.; Baughman, R. J. Chem. Phys. 1993, 12, 9933. (6) a) Hughes, A. E.; Jain, S. C. AdV. Phys. 1979, 20, 717, and references therein. (b) Kreibig, U.; Fragstein, C. Z. Phys. 1962, 224, 307. (7) Kreibig, U.; Vollmer, M. Optical Properties of Metal Clusters; Springer-Verlag: Berlin, Heidelberg, 1995. (8) Henglein, A.; Mulvaney, P.; Linnert, T. Faraday Disc. 1991, 92, 31. (9) Huang, Z.-Y.; Mills, G.; Hajek, B. J. Phys. Chem. 1993, 97, 11542. (10) Kreibig, U. Z. Phys. B: Condens. Matter Quanta 1978, 31, 39. (11) Stranski, J.; Krastanov, L. Akad. Wiss. Math. Nat. K111b 1938, 797.
Rodrı´guez-Sa´nchez et al. (12) Franck, F. Van der Merwe, J. Proc. R. Soc. London, Ser. A 198 1949, 205. (13) Volmer, M.; Weber, A. Z. Phys. Chem. 1926, 119, 277. (14) Henglein, A. J. Phys. Chem. 1993, 97, 5457. (15) Kawata, A.; Kubo, R. J. Phys. Soc. Jpn. 1966, 21, 1765. (16) Kubo, R. J. Phys. Soc. Jpn. 1962, 17, 1765. (17) Doyle, W. T. Phys. ReV. 1958, 111, 1067. (18) Kreibig, U.; Fragstein, C. V. Z. Phys. 1969, 224, 307. (19) Genzel, L.; Martin, T. P.; Kreibig, U. Z. Phys. B 1975, 21, 339. (20) Ershov, B. G.; Janata, E.; Henglein, A.; Fojtik, A. J. Phys. Chem. 1993, 97, 4589. (21) Henglein, A.; Linnert, T.; Mulvaney, P. Ber. Bunsen-Ges. Phys. Chem. 1990, 94, 1449. (22) Boudart, M. Kinetics of Chemical Processes; Prentice Hall: Englewood Cliffs, NJ, 1968; Chapter 6. (23) Rivadulla, J. F.; Vergara, M. C.; Blanco, M. C.; Lo´pez-Quintela, M. A.; Rivas, J. J. Phys. Chem. 1997, 101, 8997. (24) Tojo, C.; Blanco, M. C.; Rivadulla, J. F.; Lo´pez-Quintela, M. A. Langmuir 1997, 13, 1970.