Silver Clusters onto Nanosized Colloidal Silica as Novel Surface-Enhanced Raman Scattering Active Substrates

Silver Clusters onto Nanosized Colloidal Silica as Novel Surface-Enhanced Raman Scattering Active Substrates M AURIZIO MUNIZ-M IRANDA Dipartimento di Chimica, Universita` di Firenze, Via della Lastruccia 3, Sesto Fiorentino, Italy

A new method is proposed for obtaining surface-enhanced Raman scattering (SERS)-active substrates by photochem ical reduction of silver nitrate onto colloidal silica. Transmission electron microscopy (TEM ) and UV–visible absorption spectroscopy are employed to investigate the nanoscale structure of the materials. H igh quality SERS spectra are obtained from different organic ligands to check the efŽ ciency of these substrates. A marked stability of the colloidal suspension is ensured by the scarce tendency of the Ag-doped silica particles to aggregate by either aging or adsorption of ligand. Index Headings: Surface-enhanced Raman scattering; SERS; Colloid; Silica; Silver.

INT RODUCTIO N Since 1979 1 colloidal dispersions of silver, gold, and copper have been employed as surface-enhanced Raman scattering (SERS)-active substrates because of their capacity to produce suitable electromagnetic enhancements. Thus, most of the SERS experiments perform ed in the past years were carried out with these substrates. Actually, colloidal dispersions of metal particles in aqueous or non-aqueous solutions are more resistant than solid surfaces to the damage due to the impact of the laser beam. The continuous agitation of the metal particles under the exciting radiation, because of the Brownian motion, allows the sample to be irradiated for longer times with stronger laser powers. Moreover, the presence of small particles dispersed in a liquid closely agrees with the description of the Raman enhancement according to the electromagnetic theory. The surface-selection rules 2 based on this theory closely hold for metallic colloidal substrates. The behavior of bumps in a conducting bulk under the electric Ž eld of the radiation is, instead, m ore closely interpreted on the basis of the increased local Ž eld deriving from small isolated metal spheres. Transmission electron m icroscopy (TEM) and absorption spectroscopy in the UV–visible region, where the plasmon bands of the dispersed particles occur, allow shape and size of the colloidal particles to be investigated. Finally, these substrates are easily prepared by following simple procedures, applied m ainly to silver hydrosols and based on the reduction of silver nitrate with borohydride 1 or citrate.3 The colloidal dispersions, however, undergo aggregation of the metal particles, depending on their intrinsic instability in the aqueous m edium. The addition of ligand or aging induces aggregation, which increases progressively with time. This process can alter the adsorption of ligand and consequently the SERS spectrum and can lead to the collapse of the colloidal particles. Stabilizing compounds like poly(vinylpyrrolidone) increase the colloid Received 5 November 2002; accepted 23 January 2003. E-mail: [email protected]Ž .it.

Volume 57, Number 6, 2003

stability, but can affect the adsorption of organic ligand. Ions, deriving from the reducing agent used in the colloid preparation, can produce spectral interference with the Raman signal of the adsorbate. For example, by oxidization of borohydride by silver nitrate, borate anions are formed in solution. They, when adsorbed on silver particles, produce two intense Raman bands, 4 commonly observed at about 615 and 930 cm 2 1. In addition, the presence of residual borohydride, added in excess to obtain the complete reduction of the Ag 1 ions, can provoke undesired reactions, as in the case of diazines.5 For this reason, the use of silver colloids obtained by photochemical reduction, without the presence of chemical reagent, assum es peculiar importance. In the past, Ahern and Garrell 6 proposed a m ethod for obtaining SERS-active ‘‘photocolloids’’, in the presence of organic compounds, from silver nitrate solutions irradiated by the 514.5-nm laser line. The SERS effect was also observed by photoreduction of silver ions onto TiO 2 ,7,8 due to formation of small silver clusters onto the surface of the nanoparticles constituting the substrate. EfŽ cient SERS substrates were prepared by thermally evaporating silver on fumed silica materials with a subm icrometer structure.9,10 In this latter case, the colloidal silica, which is able to disperse and stabilize metal particles, supplied the surface roughness necessary to ensure Raman enhancement of the adsorbate. More recently, dried SERS substrates have been obtained by impregnating silica with silver nanoparticles.11 In the present study, silica colloidal particles with nanoscale structure have been used for obtaining SERSactive substrates in an aqueous suspension. The photoreduction of silver ions occurs onto the silica surface, without the presence of organic ligands. Preliminary results have showed SERS performances comparable with those of the usual silver hydrosols;12 now, a m ore extended investigation should conŽ rm this evidence, showing the main advantages of these novel substrates with respect to the pure silver hydrosols. M oreover, useful information on the silver photoreduction m echanism and the size of the silver clusters are obtained by TEM investigation and UV–visible extinction spectra. EXPERIMENTAL Preparation of SERS-Active Substrates. Pure silver colloids were prepared by reduction of AgNO 3 (Aldrich, 99.9999% purity) with excess NaBH 4 (Aldrich, 99.995% purity) according to Creighton et al.’s procedure.1 To obtain, instead, photoreduced silver, a AgNO 3 solution was added dropwise to colloidal silica (Aldrich, Ludox TM40), resulting in a sample with 10 wt % as SiO 2 content and 2.4 3 10 2 4 M as Ag 1 concentration like in the pure silver colloids obtained by Creighton’s procedure. The

0003-7028 / 03 / 5706-0655$2.00 / 0 q 2003 Society for Applied Spectroscop y

APPLIED SPECTROSCOPY

655

F IG . 1. UV –visible extinction spectra of Ag-doped silica colloidal dispersions, (A) without ligand and (B) with 10 2 3 M pyrazole. (C ) refers to a pure silica colloid.

aqueous suspension of negatively charged silica particles (average diameter, 22 nm) was alkaline (pH ;9 at 25 8C) like Creighton’s silver hydrosols. NaCl and Na 2 SO 4 were present as 0.03 wt % and 0.08 wt %, respectively. No interference with the adsorption of organic ligands was expected from the presence of sulfate anions, which are not strongly adsorbed on silver. 13 Chloride anions, instead, usually improve the Raman enhancement of the adsorbate. A defocused beam of the 514.5-nm laser line with 50 mW power density irradiated the sample under vigorous stirring. During irradiation, the colorless suspension became brown-yellow, indicative of the presence of silver aggregates on the silica particles. In the UV–visible extinction spectrum, a broad band at about 410 nm occurred, due to the surface plasmon absorption (SPA) of the m etal particles. The invariance of the extinction spectrum conŽ rmed the complete silver photoreduction. Transmission Electron M icroscopy and UV–Visible Absorption M easurements. Transmission electron m icroscopy (TEM ) measurem ents were obtained by using a Philips EM 201 instrument with an electron beam emitted at 80 kV after placing a drop of colloidal sample on a carbon–Cu grid. Large m agniŽ cations (up to 46 000) were adopted in order to investigate the aggregates at the level of nanoparticles. The UV–visible extinction spectra were recorded with a Cary 5 spectrophotometer. Raman Spectroscopy. Raman spectra were obtained by using the 514.5-nm exciting line of an Ar 1 laser, a Jobin-Yvon HG-2S monochromator, a cooled RCAC31034A photomultiplier, and a data acquisition system. A defocused beam with 50 m W power density, as m onitored at the base of a 10-mm quartz cell containing the samples, was used to reduce thermal effects. The powerdensity measurements were perform ed with a power meter instrument (m odel 362, Scientech, Boulder, CO), giving ;5% accuracy in the 300 –1000 nm spectral range. UV–VISIBLE EXTINCTION SPECTRA In the UV–visible extinction spectra of metal colloids the wavelength of the maximum of the extinction bands, 656

Volume 57, Number 6, 2003

F IG . 2. UV–visible extinction spectra of Ag hydrosols, (A) without ligand and (B) with 10 2 3 M pyrazole.

assigned to plasmon resonances, as well as the proŽ le and the number of these bands, depends on the shape and the size of the m etal clusters.14 –16 During the photoreduction process of silver nitrate with green light, in the UV– visible extinction spectrum a broad band occurs at about 410 nm (Fig. 1), progressively increasing with the irradiation time. This band, which is not present in the pure colloidal silica, is attributable to the surface plasmon absorption (SPA ) band of metal nanoclusters, obtained by photoreduction onto the silica particles, as well as found for Ag clusters form ed on nanocrystalline TiO 2 particles. 8 By observing the m aximum of the SPA band, a small red shift is obser ved with respect to the plasmon band of the pure silver particles, as obtained by following Creighton’s procedure, and detected at about 390 nm (Fig. 2). Ho¨ vel et al. 17 also obser ved broadening and red shift of the SPA band of silver clusters embedded into a SiO 2 matrix. More recently, a sm all red shift (about 20 nm) has been found on the surface plasm on of silver island Ž lms when coated with a thin SiO 2 overlayer. 18 By considering that the silver particles in Creighton’s hydrosols have an average diameter of 15 nm,19 the silver aggregates in the Ag-doped silica colloids should have slightly larger diameters. However, TEM investigation can offer more detailed information on the size and shape of the particles present in the silica colloidal suspension. By aging or addition of an organic ligand like pyrazole, which strongly adsorbs on silver, the SPA band doesn’t signiŽ cantly change (Fig. 1), with little evidence of particle aggregation. In silver hydrosol, instead, by addition of pyrazole, a large secondary band is observed at about 690 nm (Fig. 2), attributed to aggregated metal particles. This accounts for a higher stability of the Ag-doped silica colloids with respect to the pure silver hydrosols, resulting in more reproducible SERS results. TRANSM ISSION ELECTRO N M ICROSCOPY M EASUREM ENTS In order to estimate size and dimension of the Agdoped silica colloidal particles, TEM m easurements have been perform ed. In Fig. 3 the silica particles appear as

F IG . 3.

TEM micrographs of a Ag-doped silica colloid.

APPLIED SPECTROSCOPY

657

F IG . 4. SERS spectra of 2,2 9-bipyridine (10 2 4 M ): (A) in Ag-doped silica colloidal dispersion; (B) in Ag hydrosol with chloride anions; and (C ) in salt-free Ag hydrosol.

F IG . 5. SERS spectra of 1,10-phenanthroline (10 2 4 M): (A) in Agdoped silica colloidal dispersion; (B) in Ag hydrosol with chloride anions; and (C ) in salt-free Ag hydrosol.

light-colored spheroidal spots, whose size is 20 –30 nm, as well as reported in the Aldrich protocol. The darkcolored spots are, instead, indicative of silver aggregates, which appear slightly larger than the particles made up of pure silica. This points to a photoreduction m echanism consisting of the formation of silver layers on the silica nanoparticles. This hypothesis is validated by the following considerations. In an alkaline medium, as in the present case, silver ions are hydrolyzed to Ag–OH, which reacts with the silica surface, giving rise to .Si–O–Ag. Thus, when the colloidal suspension is irradiated with visible light, silver bound to silica is photoreduced and metal clusters cover the surface of the silica nanoparticles. In conclusion, the presence of nanostructured silica provides the Ag surface with the roughness necessary for the electromagnetic enhancement of the m etal surface.

to silver via the two nitrogen atoms of 2,29-bipyridine in cis conformation. 21 In the Ag-doped silica colloidal dispersion (Fig. 4), only Type I, as adsorbed species, is detected, because of the alkaline m edium, with identical band wavenumbers and relative intensities. 1,10-Phenanthroline. This compound adsorbed in salt-free Ag hydrosols through the lone pairs of the two nitrogen atoms 22 as well as 2,2 9-bipyridine. In aged colloids the ligand showed a tilted orientation with respect to the metal surface, while in the presence of halide anions a more perpendicular orientation was favored. The SERS observed in the Ag-doped silica colloidal dispersion (Fig. 5) closely corresponds to that observed in Ag hydrosol with added Cl 2 , as shown by strong enhancement of the band at ;720 cm 2 1 and an intensity decrease of the band at ;420 cm 2 1. 2-Amino, 5-Nitropyridine. In Ag hydrosols this ligand was very sensitive to the presence of chloride or bromide anions because they were able to remove the hydroxide ions from the silver surface, which favored the adsorption of the compound in the anionic form. 23 This latter, instead, occurred in salt-free Ag colloid, as shown by the strong enhancement of the in-plane ring vibration at 838 cm 2 1. The SERS of the Ag-doped silica colloidal dispersion (Fig. 6) is practically identical to that observed

SURFACE-ENHANCED RAMAN SCATTERING SPECTRA O F ORG ANIC LIGANDS Different organic compounds have been employed in the Ag-doped silica colloidal dispersions. High-quality SERS spectra have been always obtained, indicating strong interaction between ligand and silver. The spectral results are reported in Figs. 4 –7 and are compared with those detected in pure Ag hydrosols. The ligands used in the present experiments are generally able to give rise to different adsorbed species, depending on the surrounding medium (pH, electrolytes) or substrate characteristics. 2,2 9-Bipyridine. Two different adsorbed species were found by SERS in Ag hydrosols, depending on the pH and the presence of halide anions such as Cl 2 or Br 2 . At the usual pH value (;9) of the Ag colloids, the ligand adsorbed as a species named Type I. 20 This form, considered similar to the Ag(I)[bpy] 2 coordination compound, showed characteristic bands at ;355, 1015, 1308, and 1596 cm 2 1. Another species, named Type III, usually observed in silver hydrosol at acidic pH with halide anions, exhibited the same bands as Type I, but upshifted to ;374, 1025, 1321, and 1606 cm 2 1 , respectively. Moreover, Type I showed strong enhancement for the ring band at 1015 cm 2 1 and weak bands in the 1500 –1600 cm 2 1 spectral region. The opposite pattern was observed for Type III. The two different species are both bonded 658

Volume 57, Number 6, 2003

F IG . 6. SERS spectra of 2-amino,5-nitropyridine (10 2 4 M): (A) in Agdoped silica colloidal dispersion; (B) in Ag hydrosol with chloride anions; and (C ) in salt-free Ag hydrosol.

(2)

(3)

(4)

F IG . 7. SERS spectra of pyrazole (10 2 3 M): (A) in Ag-doped silica colloidal dispersion; (B) as A, but with addition of 10 2 3 M NaCl; and (C ) as A, but with addition of 10 2 2 M NaC l. (D ) refers to 10 2 3 M pyrazole in salt-free Ag hydrosol.

in Ag colloid with coadsorbed chloride anions, showing adsorption of the ligand as a neutral molecule. Pyrazole. Pyrazole was adsorbed on salt-free Ag colloid as pyrazolide anion, 24 as shown by the occurrence of two strong SERS bands at 1129 and 1484 cm 2 1 . W hen Cl 2 anions were added to the colloidal dispersion or the pH was down-shifted to acidic values, these bands were progressively replaced by the bands of the neutral molecule at 1150 and 1508 cm 2 1 , respectively. In the SERS experiments performed in the Ag-doped silica colloidal dispersion (Fig. 7), the presence of adsorbed pyrazolide is detected, but together with pyrazole as minority species. By additions of chloride ions, the neutral molecule was predominantly adsorbed. All these measurem ents have been performed to check the SERS efŽ ciency of the Ag-doped silica substrates. The spectral data are completely comparable with those obtained from the usual silver colloidal dispersions, as regards the Raman enhancements and the overall spectral patterns. The presence of chloride anions in solution not only improves the SERS intensities, but also promotes the spectral changes observed in pure silver sols. These changes correspond to different orientations of the adsorbate with respect to the surface or to different adsorbed species formed at the interfaces. The most evident spectral modiŽ cations are observed for 2-amino,5-nitropyridine, which is very sensitive to the halide-induced effect. Pyrazole is adsorbed as a neutral molecule only by consistent additions of NaCl, without sol collapse, which, instead, occurred in pure silver colloids. Actually, as shown by the UV–visible extinction spectra (Fig. 1), the adsorption of a ligand such as pyrazole produces little aggregation of the colloidal particles, with more sol stability than in Ag hydrosols (Fig. 2). CONCLUSION The employment of these new SERS-active colloidal dispersions provides m any advantages with respect to the hydrosols constituted of pure silver. The more relevant properties of the present substrates are summ arized below. (1) A m arked stability of the colloidal suspension is

(5)

ensured by the scarce tendency of the Ag-doped silica particles to aggregate, by aging or by adsorption of ligand. The absence of reducing agent for obtaining SERS-active silver clusters avoids undesired reactions or interference with the Raman scattering of the adsorbate. The photoreduction process is obtained in an easy and extemporaneous way by irradiation with the same laser line (514.5 nm) employed in the SERS measurements. The colloidal silica used here is a commercial product, with tested particle size and electrolyte content, and this is a warranty to have a standard protocol and reproducible results. High-quality SERS spectra of organic ligands are obtained owing to the presence of nanosized silica particles that ensure the electromagnetic enhancement effect and owing to chloride anions that further increase the Raman enhancement by formation of surface charge-transfer complexes.13

Finally, new and interesting perspectives of research are opened by employing these nanostructured m aterials: (1) different spectral results can be achieved by varying the surrounding medium (pH, electrolytes), avoiding sol collapse; (2) the Ag-doped silica particles can be used as dried substrates by gelation in SERS experiments without aqueous medium or for water-insoluble ligands; (3) it is possible to study the third-order nonlinear susceptibility, which usually assumes large values for m etal-dispersed systems and has been experimentally proved 25 to be proportional to the SERS intensity in silver sols prepared by the photochemical m ethod. ACK NOW LEDGM ENTS The author gratefully thanks the Italian Ministero dell’Istruzione, Universita` e Ricerca (M IUR), and the Consiglio Nazionale delle Ricerche (CNR) for the Ž nancial support to the ‘‘Progetto Finalizzato Materiali Speciali II’’.

1. J. A. Creighton, C. G. Blatchford, and M . G. Albrecht, J. Chem. Soc., Faraday Trans. 2 75, 790 (1979). 2. J. A. Creighton, Surf. Sci. 124, 209 (1983). 3. P. C. Lee and D. Meisel, J. Phys. Chem. 88, 3391 (1982). 4. G. Sbrana, N. Neto, M. Muniz-Miranda, and M. Nocentini, J. Phys. Chem. 94, 3706 (1990). 5. M . M uniz-Miranda, N. Neto, and G. Sbrana, J. Mol. Struct. 143, 275 (1986); and 174, 351 (1988). 6. A. M. Ahern and R. L. Garrell, Anal. Chem. 59, 2813 (1987). 7. L. M . Sudnik, K. L. Norrod, and K. L. Rowlen, Appl. Spectrosc. 50, 422 (1996). 8. Y. Liu, C.-Y. Liu, Z.-Y. Zhang, and C.-Y. Wang, Spectrochim. Acta, Part A 57, 35 (2001). 9. A. Alak and T. Vo-Dinh, Anal. Chem. Acta 206, 333 (1988). 10. A. Alak and T. Vo-Dinh, Anal. Chem. 61, 656 (1989). 11. F. X. Liu, Y. Xiao, and Y.-S. Li, J. Ram an Spectrosc. 32, 73 (2001). 12. M . Muniz-Miranda, J. Raman Spectrosc. 33, 295 (2002). 13. M . Muniz-Miranda and G. Sbrana, J. Raman Spectrosc. 27, 105 (1996). 14. E. J. Zeman and G. S. Schatz, J. Phys. Chem. 91, 634 (1987). 15. U. Kreibig and M. Vollmer, Optical Properties of Metal Clusters (Springer, New York, 1995). 16. R. M. Bright, M. D. Musick, and M. J. Natan, Langmuir 14, 5695 (1998). 17. H. Ho¨ vel, S. Fritz, A. Hilger, U. Kreibig, and M. Vollmer, Phys. Rev. B 48, 18178 (1993).

APPLIED SPECTROSCOPY

659

18. P. J. Tarcha, J. DeSaja-Gonzalez, S. Rodriguez-Llorente, and R. Aroca, Appl. Spectrosc. 53, 43 (1999). 19. I. R. Nabiev, V. A. Savchenko, and E. S. Efrem ov, J. Ram an Spectrosc. 14, 375 (1983). 20. M . Kim and K. Itoh, J. Phys. Chem. 91, 126 (1987). 21. A. De Bonis, G. Compagnini, R. S. Cataliotti, and G. Marletta, J. Raman Spectrosc. 30, 1067 (1999).

660

Volume 57, Number 6, 2003

22. M . Muniz-Miranda, J. Phys. Chem. A 104, 7803 (2000). 23. M . M uniz-Miranda, N. Neto, and G. Sbrana, J. Mol. Struct. 348, 261 (1995). 24. M . M uniz-Miranda, N. Neto, and G. Sbrana, J. Mol. Struct. 482, 207 (1999). 25. T. Sato, T. Ichikawa, T. Ito, Y. Yonezaw a, K. Kadono, T. Sakaguchi, and M. Miya, Chem. Phys. Lett. 242, 310 (1995).