Anal. Chem. 1996, 68, 1003-1011
Characterization of SiO2-Overcoated Silver-Island Films as Substrates for Surface-Enhanced Raman Scattering William B. Lacy, John M. Williams, Lisa A. Wenzler, Thomas P. Beebe, Jr., and Joel M. Harris*
Department of Chemistry, University of Utah, Salt Lake City, Utah 84112
Vapor deposition of a 45-Å layer of silver and a subsequent 50-65-Å layer of SiO2 onto a glass slide is shown to be a viable substrate for surface-enhanced Raman spectroscopy (SERS) and a model of the silica surface suitable for adsorption kinetic studies. The chemistry of the silica is characterized by UV absorption and contact angle measurements. The physical structure is probed by atomic force microscopy to compare the surface roughness of the silver-island film and its SiO2 overcoat. Auger electron spectroscopy and X-ray photoelectron spectroscopy are used to probe the silica layer for exposure of silver through gaps in the overcoat. Comparison of SERS of dibenzyl disulfide bound to a silver-island film versus that of a silver film protected by the SiO2 overlayer tests the continuity of the SiO2 film and its ability to block access of adsorbates to the underlying silver surface. The model silica surface is used to detect the Raman spectrum of adsorbed pyridine and its adsorption isotherm; the adsorbate vibrational frequencies are consistent with previous reports of pyridine on silica. With increased interest in surface chemistry for both fundamental questions and practical applications, understanding molecular interactions at the liquid/solid interface is becoming an increasingly important goal. The ability to monitor adsorption of molecules at such interfaces is critical to understanding separation processes, catalysis, environmental transport, and many other important phenomena. The small number of molecules in a monolayer or submonolayer film of an adsorbate and in the presence of a host of molecules in the bulk overlaying solvent combine to make the goal of monitoring adsorption at the liquid/ solid interface quite difficult, especially for conventional vibrational spectroscopic techniques which have limited sensitivity. One spectroscopic tool that has helped overcome these challenges is based on the phenomenon of surface-enhanced Raman scattering (SERS) that can be observed on several roughened metallic surfaces. This effect, first reported by Fleischmann and coworkers1 in the early 1970s and subsequently characterized as a surface enhancement by Van Duyne,2 and Creighton3 and their co-workers, allows low surface concentrations of adsorbates to be detected, with relative ease, in the presence of overlaying solvent molecules. (1) Fleischmann, M.; Hendra, P. J.; McQuillan, A. J. Chem. Phys. Lett. 1974, 26, 163-166. (2) Jeanmaire, D. L.; Van Duyne, R. P. J. Electroanal. Chem. 1977, 84, 1-20. (3) Albrecht, M. G.; Creighton, J. A. J. Am. Chem. Soc. 1977, 99, 5215-5218. 0003-2700/96/0368-1003$12.00/0
© 1996 American Chemical Society
Limitations to utilizing the enhancement provided by SERS include the relatively small number of metal surfaces which exhibit the requisite plasmon resonance and the difficulty in preparing many such surfaces to be SERS-active. Of the approximately 10 metallic substrates that are effective for generating a surface enhancement, the number that are of widespread chemical interest is limited. Coupling the optical enhancement of roughened metal substrates to the surface chemistry of insulators is a difficult challenge but one that would allow adsorption studies of many other chemically important surfaces, such as silica, to be undertaken. Compared to the amount of work that has been done investigating applications of metallic SERS substrates, relatively little has been completed examining the usefulness of nonmetallic surfaces. Murray and Allara4,5 have explored the use of a silver SERS substrate separated from an adsorbate molecule by layers of Al/Al2O3 over CaF2 (used for surface roughening) and a polymer film; it has been suggested,6,7 however, that some of the surface enhancement originates from the aluminum, contributing an unknown magnitude to the total enhancement, and that the silver morphology can change with varying spacer thickness.5 In another study, Parry and Dendramis8 developed a high-vacuum SERS cell and obtained SERS spectra of thin films such as polystyrene and cetyltrimethylammonium bromide on silicon by overcoating the adsorbate with a vapor-deposited silver-island film. While this method can be used to obtain a SERS spectrum of a nonvolatile adsorbate, the adsorbate is brought into contact with a silver surface (which could perturb the adsorbate structure); furthermore, the method cannot be used for in situ monitoring of adsorption from solution. In pioneering work by Walls and Bohn,9 a metal-island SERS substrate overcoated with a thin, continuous layer of silica serves as a model surface for adsorption studies. Silver-island films on quartz substrates were sputter-coated with a thin SiO2 layer, and Raman scattering was generated by rear illumination, in a total internal reflection geometry. The dependence of Raman intensity on SiO2 thickness was studied and was found to decrease by a factor of about one-third within the first 40 Å; this result agreed with those obtained by other groups,10,11 where the distance dependence of SERS was measured by using Langmuir-Blodgett (4) Murray, C. A.; Allara, D. L.; Rhinewine, M. Phys. Rev. Lett. 1981, 46, 5760. (5) Murray, C. A.; Allara, D. L. J. Chem. Phys. 1982, 76, 1290-1303. (6) Aroca, R.; Kovacs, G. J. In Vibrational Spectra and Structure; Durig, J. R., Ed.; Elsevier: Amsterdam, 1991; pp 55-112. (7) Lopez-Rios, T.; Pettenkofer, C.; Pockrand, I.; Otto, A. Surf. Sci. 1982, 121, L541-L544. (8) Parry, D. B.; Dendramis, A. L. Appl. Spectrosc. 1986, 40 (5), 656-661. (9) Walls, D. J.; Bohn, P. W. J. Phys. Chem. 1989, 93, 2976-2982.
Analytical Chemistry, Vol. 68, No. 6, March 15, 1996 1003
(LB) films as spacer layers between a silver-island film and a Raman scatterer. A recent adaptation of the Walls/Bohn concept of employing a silica overlayer on a SERS-active substrate was described,12,13 whereby an 800-Å-thick layer of a silica sol was captured on a solid silver substrate with use of a mercaptosilane reagent acting as an adhesive; the material was used as a model for porous silica and was employed to characterize alkylsilane layers covalently bound to the silica sol. Raman scattering from alkyl chains was observed; however, integration times were long due to the limited enhancement at such large distances from the metal substrate. Thick silica layers have also been deposited over roughened silver surfaces;14 glass substrates were roughened with a spin-coated layer of alumina powder and subsequently coated with 750 Å by thermal evaporation and overcoated with SiO2 by e-beam evaporation of quartz to thicknesses ranging from 200 to 5000 Å. Monitoring SERS from pyridine adsorbing onto these substrates showed that 20-40 min was required to reach equilibrium, presumably due to slow diffusion through small pores in the structure to reach the silver surface.14 The utility of 100-Åthick SiO2-coated SERS-active silver films as supports for thin films of other metals has also been reported;15 these hybrid structures yielded a high-sensitivity SERS response for detecting adsorption of molecules from the gas phase onto metals other than silver. Our interest in silica-overcoated SERS-active silver islands is for measuring kinetics of adsorption from solution onto silica surfaces. The development of high-sensitivity charge-coupled device (CCD) detectors for spectroscopy allows Raman spectra of dilute solutes in solution to be acquired in seconds. Combining this instrumental sensitivity with SERS enhancement of adsorbate scattering makes the monitoring of adsorption kinetics possible. In this work, we characterize the surface chemistry of SiO2overcoated silver islands on glass slides and test their potential for acquiring adsorbate spectra on silica surfaces on the time scale of seconds. Raman spectra of pyridine and dibenzyl disulfide on SiO2-overcoated and bare silver-island films provided information about the SERS sensitivity and the continuity of the silica overcoat. UV-visible spectrophotometry, atomic force microscopy (AFM), X-ray photoelectron spectroscopy (XPS), and Auger electron spectroscopy (AES) were also employed to characterize the SiO2 surface chemistry and structure. Using the SiO2/Ag SERS substrates, we were able to spectroscopically probe the solid/ liquid interface on a molecular level with 1-s time resolution, suggesting future kinetic applications of this SERS substrate (model silica surface). EXPERIMENTAL SECTION Reagents. Carbon tetrachloride (Omnisolve, EM Science), pyridine (EM Science), methanol (Optima grade, Fisher), dibenzyl disulfide (Aldrich), and ammonium hydroxide (Mallinckrodt AR) were used as received. Silver (99.9999% purity) was obtained from Aldrich. Silicon monoxide (99.99% purity) was obtained from Cerac Coating Materials. Water was deionized, quartz-distilled, and filtered (Nanopure II, Barnstead) to a resistance of 18 MΩ. (10) Kovacs, G. J.; Loutfy, R. O.; Vincett, P. S.; Jennings, C.; Aroca, R. Langmuir 1986, 2, 689-694. (11) Yu, L.; Bingkun, Y.; Yinting, W.; Weiqing, Z.; Peihui, L. Chin. Phys. 1992, 12 (3), 694-700. (12) Thompson, W. R.; Pemberton, J. E. Anal. Chem. 1994, 66, 3362-3370. (13) Thompson, W. R.; Pemberton, J. E. Chem. Mater. 1995, 7, 130-136. (14) Hill, W.; Rogalla, D.; Klockow, D. Anal. Methods Instrum. 1993, 1, 89-96. (15) Wachter, E. A.; Storey, J. M. E.; Sharp, S. L.; Carron, K. T.; Jiang, Y. Appl. Spectrosc. 1995, 49 (2), 193-199.
1004 Analytical Chemistry, Vol. 68, No. 6, March 15, 1996
Figure 1. SERS substrate with model silica surface. A 45-Å layer of silver islands is deposited on glass, followed by a 50-65-Å overlayer of SiO2.
SERS Substrate Preparation. Glass microscope slides were soaked in concentrated NH4OH for 1 h and washed five times by ultrasonic cleaning in 18-MΩ water; the slides were fully dried for 1 h at 140 °C. The SERS substrates were prepared as follows: in a conventional vacuum deposition chamber (Edwards, Model E306A), the clean glass slides were placed 20 cm above a molybdenum boat (ME6A-.005Mb, R. D. Mathis Co.) containing silver and a tantalum boat (ME6A-.005Ta, R. D. Mathis Co.) containing silicon monoxide. The chamber was evacuated to a pressure of 9 × 10-6 Torr for silver deposition; the Ag and SiO were vaporized, independently, by passing currents of approximately 25 and 60 A, respectively, through the refractory boats. The mass of deposited films was monitored using a quartz-crystal microbalance thickness gauge (Edwards FTM5), by which the rate of deposition was controlled; typical rates were nominally 0.2 Å/s for silver and 1.0 Å/s for SiO (deposited as SiO2; see below). The mass thicknesses reported by the quartz microbalance were corrected for the slight difference between source/sample and source/gauge distances; the correction factor was determined by evaporating thick (1000 Å) silver films onto large area (g10 cm2) silicon substrates, producing mass changes that could be reliably weighed on an analytical microbalance. To keep the plasmon resonance frequency (as measured with UV-visible spectrophotometry; see below) of the silver-island film near the excitation frequency of the 514.5-nm laser line, deposition rates were slow, and the thicknesses of the roughened silver-island surfaces were kept between 45 and 47 Å. The deposition rate for SiO2 was generally higher than that for silver since there were no “island” size, shape, or roughness concerns; it is, though, important to keep the SiO2 deposition rate slow enough to avoid excessive heating of the substrate and destruction of the silverisland structure. As suggested by Semin and Rowlen16 to improve silver-island/glass adhesion, the silver-coated slides were allowed to remain under vacuum for 15 min prior to SiO2 deposition. The vacuum chamber was then elevated to a pressure of 5 × 10-4 Torr (or 1 × 10-4 Torr oxygen partial pressure) during the SiO deposition step to assure complete oxidation and formation of SiO2.17-19 SiO2 films were deposited at a typical mass thickness of 50-65 Å (by the quartz microbalance), thin enough to still allow enhancement of the Raman signal, yet thick enough for complete coverage of the silver islands. The silver- and silica/silver-coated slides were removed from the vacuum chamber and placed directly into methanol to keep the surface as free from water vapor and other unwanted adsorbates as possible. A cartoon of the SiO2/Ag SERS substrates can be seen in Figure 1. (16) Semin, D. J.; Rowlen, K. L. Anal. Chem. 1994, 66, 4324-4331. (17) Holland, L. Vacuum Deposition of Thin Films; John Wiley & Sons, Inc.: New York, 1956; pp 449-450. (18) George, J. Preparation of Thin Films; Marcel Dekker: New York, 1992. (19) Vacuum Deposition Chemicals & Evaporation Materials, 3rd ed.; CERAC, Inc.: Milwaukee, WI, 1988; pp 11-12.
UV-Visible Spectrophotometry. All UV-visible measurements were made on a Cary spectrophotometer (Varian, Model 17D) retrofitted to stepper-motor control (On-Line Instrument Systems). Fused silica slides were absorbance-matched prior to SiO and SiO2 deposition. All measurements were referenced to a clean silica slide and zeroed against an air background. Contact Angle Measurements. A standard contact angle goniometer (Model A-100, Rame´-Hart, Inc.) was used to make contact angle measurements on the model silica surface. 18-MΩ water was used as the sessile drop for all samples. Atomic Force Microscopy Measurements. Sample preparation of glass slides (control), slides coated with a silver-island film, and slides with the SiO2/Ag SERS substrate were prepared as described above. Surfaces were profiled using a TMX 2000 Explorer (TopoMetrix, Santa Clara, CA) atomic force microscope operating in contact mode. The tips were commercially available silicon nitride cantilevers (Digital Instruments, Santa Barbara, CA) with spring constants of 0.12 N/m. AFM images were obtained in constant force mode (5-10 nN net repulsive force on the cantilever) and are presented here as raw data except for a leastsquares background tilt subtraction. Information density of captured images was 500 × 500 pixels. All images were obtained in air. Topographic (roughness) profiles were extracted from digital image data for postacquisition analysis. AES and XPS Surface Characterization. The SiO2/Ag films were characterized by X-ray photoelectron spectroscopy and Auger electron spectroscopy in an ultra-high-vacuum (UHV) chamber that has been described in detail elsewhere.20 Data were collected in the regions of the principal oxygen and silver lines; the XPS scans, over the 62-562-eV binding energy region, employed an Al KR source operating at 13 kV, 30 mA, with a hemispherical analyzer pass energy of 150 eV. The AES scans, over the 100-600-eV region, employed an electron gun operating at 3 keV, 0.04 mA, with the analyzer operating at a constant retarding ratio of b ) 3. Surface-Enhanced Raman Scattering. The Raman instrument used for the SERS measurements is shown in Figure 2 and was configured as follows. The 514.5-nm line from an argon ion laser (Lexel, Model 95) was used for excitation. Incoherent plasma lines were eliminated by a combination of 2 Pellin-Brocha prisms (Optics for Research, ABDU-20) and a variable aperture. The beam was focused with a cylindrical lens (f ) 75 mm) to a 60-µm-wide stripe and passed through the bottom of the sample cell, which was a glass cuvette (20 mm × 10 mm × 45 mm, Type 3, NSG Precision Cells, Inc.). The cuvette contained the sample solutions and held the SiO2/Ag-overcoated glass microscope slides (25 mm × 75 mm × 1 mm, Fisher Scientific) at a 20° angle from vertical to maximize Raman scatter directed at the collection lens. Scattered light from the sample was collected and collimated at 90° from incidence by an f/1.4 camera lens (f ) 25 mm, JML Optics) and focused by an achromatic doublet (f ) 100 mm, Newport Corp.) onto the entrance slit of an f/4 single-stage spectrograph (0.5 m, Spex 1870). A holographic notch filter (Kaiser Optical Systems, Inc.) for the 514.5-nm line was placed in the collimated region, between the collection and focusing lenses, to eliminate interference from specular and Rayleigh scattering. A 1200-grooves/mm grating dispersed the scattered radiation across a CCD chip (Thomson CSF TH7895A, front (20) Leavitt, A. J.; Han, T.; Williams, J. M.; Bryner, R. S.; Patrick, D. L.; Rabke, C. E.; Beebe, T. P. Rev. Sci. Instrum. 1994, 65, 59-75.
Figure 2. Raman instrument for front-surface SERS experiments. Components are labeled as follows: Pellin Brocha prisms (PB1, PB2), mirrors (M1-M5), variable aperture (A), cylindrical focusing lens (L1), collection lens (L2), monochromator focusing lens (L3), holographic notch filter (NF), and sample/holder (S1).
illuminated) TE-cooled to -70 °C (EG&G PARC, Model 1530-P/ PUV). For all spectra, 512 rows were “binned” (on-chip signal averaging) along the vertical axis of the chip, in combination with 256 columns each comprised of two adjacent binned columns along the horizontal (wavelength) axis, to improve the signal-tonoise ratio. Data were processed off-line on a 486-PC using GRAMS/386 (Galactic) and Quattro Pro for Windows (Borland). All spectra were obtained at one monochromator setting and calibrated in the EG&G PARC software by fitting the frequencies of seven known Raman bands of indene in the 900-1400-cm-1 region to a third-order polynomial. RESULTS AND DISCUSSION Chemical Characterization of SiO2 Overlayer. In previous work, generation of a silica overlayer on silver islands had been accomplished by sputtering. A less costly means was chosen for this work based on thermal evaporation of silicon monoxide in the presence of oxygen. Thermal vapor deposition of SiO at pressures greater than 1 × 10-5 Torr is reported to result in deposition of the stoichiometric oxide, SiO2.16-19 The extent of oxidation of the evaporated SiO can be experimentally tested by optical spectroscopy. SiO is a dark solid that exhibits strong electronic absorption in the near-UV region, whereas SiO2 in this region is optically transparent. The optical absorption of a pure SiO film was compared with that deposited at high O2 partial pressure to determine the fraction of substoichiometric oxide in the model SiO2 overcoat. A 2030-Å layer of SiO was deposited onto a fused silica slide at a pressure of 5 × 10-7 Torr, where the mean free path between collisions with oxygen is several orders of magnitude longer than the source-to-sample distance, which was reduced to