Microchemical Journal 63, 7191 (1999) Article ID mchj.1999.1769, available online at http://www.idealibrary.

com on

Adsorption Kinetics of Au and Ag Nanoparticles on Functionalized Glass Surfaces

Sung-Ho Park, Jung-Hyuk Im, Jun-Wan Im, Byung-Hoon Chun, and Jae-Ho Kim 1
Department of Applied Chemistry, Ajou University, Suwon 442749, Korea Accepted June 7, 1999 Well-dened two- or three-dimensional arrangements of nanosize Au and Ag particles were fabricated for surface-enhanced Raman scattering substrate applications and their unique optical properties. Two-dimensional arrays of colloidal lms of Au and Ag were prepared on glass slides modied with silane compounds having bifunctional moieties that have specic afnity for gold or silver. During the formation of colloidal lms of Au and Ag on glass slides, UVvis absorption spectroscopy was used to monitor the progress of the surface immobilization reaction of colloidal particles on solid substrates. Adsorption characteristics could be accurately modeled by the wellknown MichaelisMenten kinetics. It was found that the k 3 / k 1 values were proportional to particle size. This indicates that the rate of formation of the rst monolayer with relatively small particles is faster than that of the multilayers, whereas for larger colloids the multilayer formation rate is much faster. SERS spectra of p -aminothiophenol (PATP) on a silver colloidal lm were obtained with 514.5-nm excitation. SERS intensities of PATP increased signicantly with morphological change of the colloidal lm. This morphological alteration, parallel to water evaporation from the colloidal lm, was presumably induced by the difference in dielectric constants of air and water. 1999 Academic Press Key Words: adsorption kinetics; Au colloid; Ag colloid; colloidal metal lm; self-assembled monolayer; surface-enhanced Raman scattering spectroscopy.

INTRODUCTION Since 1928, when Sir. Chandrasekhra Venkata Raman discovered the Raman scattering phenomenon that bears his name, Raman spectroscopy has steadily become a more useful tool for the characterization of chemical materials and contributed to the elucidation of their structures. Improvements in the various components of Raman spectroscopic instrumentation including lasers, detectors, electronics, and computers have given Raman spectroscopy wide applications and enhanced its analytical sensitivity. The most notable examples are Fourier transform Raman, resonance Raman, and surface-enhanced Raman spectroscopy. The discovery of surface-enhanced Raman scattering (SERS) has prompted numerous theoretical and experimental studies (1 4). SERS is a spectroscopic technique that allows one to obtain vibrational spectra of molecules adsorbed on metal surfaces. Determination of the nature of the adsorbed molecules at the solid surface is important for understanding the physiochemical properties of intermolecular or adsorbatesubstrate interactions at the interface. Numerous studies have reported that many molecules adsorbed on appropriately prepared metal surfaces display Raman cross sections several orders of magnitude greater than the corresponding quantity for an isolated molecule or from a solution (5, 6). Because of the large enhancement, small sample volumes can be used; detection limits of SERS technique lie in the picomole-to-femtomole range (7).
1

To whom correspondence should be addressed. 71 0026-265X/99 $30.00

Copyright 1999 by Academic Press All rights of reproduction in any form reserved.

72

PARK ET AL.

The understanding of the mechanism responsible for SERS was concentrated in the early study. There are two mechanisms that are now considered as the most appropriate explanations for surface-enhanced Raman scattering. One is related to relatively longrange electronic resonance effects of the optical elds with the surface plasmon of the SERS-active substrate, namely, the electromagnetic theory. In the electromagnetic model, a molecule near an active conducting SERS surface experiences a stronger electromagnetic eld compared with a free molecule (6). The other explanation is the chemical enhancement theory related to the formation of charge-transfer complexes between adsorbed molecules and the substrate, causing distortion in the polarizability of the molecules (7). In the chemical enhancement model, the Raman scattering cross section of the molecule is amplied by molecule/metal interaction. Ample experimental evidence has been reported for each of these mechanisms to suggest that more than one mechanism is responsible for the overall enhancement (6, 8). Although several different mechanisms for the SERS effect have been proposed, a general agreement has been reached that the surfaces of metals need to be roughened properly to obtain large surface enhancement. Probably the most common substrates used for SERS are colloidal suspensions and silver or gold particles and electrochemically roughened silver or gold electrodes because their preparation requires no specialized instrumentation (9 12). Among the many SERS substrates providing appropriate characteristics, colloid hydrosols are the most commonly employed for several reasons: (a) their easy preparation, handling, and storage; (b) the low electric potential existing at the surface (11); and (c) the minimum inuence of these systems on the structures of the adsorbate on the surface, especially when biological molecules are used (13). The preparation of colloidal metals appears simple, but great care must be exercised to produce a reproducible and stable colloid (4). The purity of the water and reagents, as well as the cleanliness of the glassware, is critical (4). Small metal particles exhibit novel physical and chemical properties that arise from size and surface effects (14 16). In the case of novel metal colloids, the surface plasmon band of particles changes with variation in particle diameter (16 18). Metal particles are surrounded by anions which produce an interparticle repulsive force. When a molecule is exposed to metal hydrosol, it is adsorbed on the particle by replacing the adsorbed anion, which reduces the charge of the colloid particle and increases the possibility of the occulation of the colloidal systems (19). The characteristics of this process are strongly inuenced by the size and shape of the particle, the ionic strength, the pH, and the concentration of the hydrosol solution. These parameters have contributed to rather poor reproducibility of SERS effects of species adsorbed on metal hydrosols. Therefore, to fulll the potential of the SERS technique as an established analytical method, a well-characterized SERS substrate and a reliable and reproducible preparation procedure are required. Although many approaches have been reported on this account, preparation of a well-dened, stable SERS substrate having uniform roughness has proven difcult (20). Recently the fabrication of nanostructures or nanoparticles has increased interest throughout the elds of pure and applied science and technology due to their unique physical, optical, and chemical properties (21). A new method of generating macroscopic metal surfaces, based on self-assembly of nanometer-scale colloidal gold and silver particles from solution onto various solid substrates, has been reported by many research groups in the past several years (22, 23). It has been found that a well-dened and uniform

Au AND Ag NANOPARTICLES ON GLASS

73

nanometer-scale architecture of silver or gold colloid, called colloidal metal lms (CMFs), can be generated and many possible applications, including SERS substrates, have been demonstrated. CMFs as SERS substrates show unique advantages over other substrates since they can be characterized at both the macroscopic and microscopic levels, are highly reproducible, are electrochemically addressable, and are simple to prepare in large numbers (2224). Furthermore, the surface roughness of CMFs can be dened and controlled by the colloid diameter, which provides not only the condition to optimize SERS effect but also an opportunity to understand surface enhancement theory. However, little information is available on the critical parameters that determine surface roughness. The rate of formation of CMFs, depending on particle diameter, is particularly important in control of the nal topological properties, interparticle distance, and surface coverage of colloidal metal particles for SERS application. In this study, we report on the kinetics of colloidal particle chemisorption on organosilane-modied glass substrates. The adsorption rate constants were calculated by using four different sizes of colloidal Au and Ag particles. Using the MichaelisMenten equation and UVvis absorption spectroscopy, we demonstrated that it is possible to control the surface density of colloidal particles by changing immersion time and particle size. This result is useful for fabricating reproducible colloidal metal lms with specied physiochemical properties, which is crucial in using nanoparticle-based devices including SERS substrates. MATERIALS AND METHODS Materials. The following chemicals were obtained from Aldrich: HAuCl 4 3H 2O, 3-aminopropyltrimethoxysilane, 3-mercaptopropyltrimethoxysilane, trisodium citrate dihydrate, acetone, methanol, 2-propanol, p -aminothiophenol, and H 2O 2. Silver nitrate was purchased from Kojima Chemicals. The following materials were obtained from J. T. Baker: HCl, H 2SO 4, and HNO 3. All chemicals were used as received. H 2O (18 M) was obtained from Milli-Q. Glass slides were obtained from Gebruder Rettberg. Colloid preparation. Various sizes of Au colloids were prepared according to the previously described procedure (25). A brief description of the procedure is as follows. To an initial solution of 50 ml 0.01% (w/v) HAuCl 4 that is at a boil, 1.75 ml of 1% (w/v) trisodium citrate is added. The solution initially develops a gray color, which changes to a lavender and nally red in 13 min of continuous boiling. All processes were done under reux. By controlling the amount of trisodium citrate initially added, various sizes of colloidal particles were formed. The maximum absorption peak in the UVvis spectra shifts from 520 nm to a longer wavelength as the citric acid/Au mole ratio decreases. Citrate-reduced Ag colloids were prepared according to a modied Lee and Meisel method (26). All glassware used in the colloid preparation was rigorously cleaned in a bath of freshly prepared 3:1 HCl:HNO 3 (aqua regia) (caution! aqua regia is extraordinarily hazardous; it must be handled with the utmost care) and thoroughly rinsed with ultrapure water. An aqueous solution (500 ml) of silver nitrate (90 mg) was heated to 45C, and then the temperature was increased rapidly to the boiling point while rigorously stirring. As soon as the boiling commenced, a 1.0% (w/w) solution of sodium citrate (10 ml) was rapidly added, and heat was reduced, but the solution was kept boiling gently for 90 min and continuously stirred.

74

PARK ET AL.

Surface derivation with organosilane coupling agent on a glass slide. The surface of a glass slide was modied with 3-aminopropyltrimethoxysilane (APTMS) or 3-mercaptopropyltrimethoxysilane (MPTMS). All glass slides used in this report were washed with detergent and rinsed with distilled water followed by sonication in acetone and methanol for 2 min each subsequently and dried in an oven before cleaning with a piranha solution (H 2O 2:H 2SO 4 1:3, v/v) (Caution! Piranha solutions are extremely dangerous, reacting explosively with organic materials.) After exposure to a piranha solution for 30 min, the plates were rinsed thoroughly with distilled water and stored in methanol (HPLC grade) until use. An APTMS (or MPTMS) solution was prepared by adding 2 ml APTMS (or MPTMS) to 25 ml of 2-propanol. The clean plates were placed in the prepared solutions for 1 day. The modied plates were rinsed with 2-propanol and annealed at 110C for 10 min and stored in an inert atmosphere (27). Colloid immobilization. Each modied glass slide was immersed in colloid suspension. After they were constructed on the plates, the CMFs were rinsed with distilled water and then stored in distilled water. Particle size was determined by analysis of SEM photographs and atomic force microscopy (AFM) images. The morphology of the CMFs was measured by AFM. Instrumentation. A JASCO V-550 UV/VIS spectrophotometer, working in the range 190 850 nm, with a resolution of 0.5 nm per element, was used to carry out UVvis spectroscopy. AFM images were obtained by normal noncontact mode measurement with V-shaped silicon nitride cantilevers [spring constant: 0.067 N/m, SPM-LS, Park Scientic Instruments (PSI), Sunnyvale, CA]. Surface-enhanced Raman spectra were collected using a triple monochromator coupled with a blue intensied CCD array detector (Spex Industries, Edison, NJ). A Coherent Innova 70 argon ion laser was used as an excitation source (514.5 nm). The laser light was passed through a premonochromator to remove plasma emission lines. The scattered light was collected in a 45 angle geometry. Spectra obtained with 514.5-nm excitation were recorded using a 1200 groove/mm grating. RESULTS AND DISCUSSION We report on the kinetic behavior of the adsorption of colloid particles to a surface modied with organosilanes containing either thiol or amino-terminal groups for selfassembling Au and Ag colloids. The schematic diagram of the adsorption process of colloid particles is shown in the Fig. 1. To our knowledge, there is one report regarding the kinetics of immobilization of colloids onto solid surfaces (28). Natan et al. showed that the surface coverage is roughly proportional to the square root of immersion time within 45 min, and colloidal particle coverage is limited by repulsive interparticle interaction inhibiting additional particle immobilization on the Au particle. They observed that the absorbance of CMFs constructed by longer exposure of the substrate to colloidal Au was not signicantly changed even for periods of 6 weeks. In our study, an apparent plateau of the increase in absorbance at longer reaction time was observed, and this feature was consistently observed from immobilization reactions of Au colloids of various particle sizes. To elucidate the rate constant of the surface immobilization reaction of the colloidal particle, the well-known MichaelisMenten model was used to describe the relationship of maximal absorbance at the absorption peak to immersion time. The MichaelisMenten model, which is often adopted to describe the kinetics of an

Au AND Ag NANOPARTICLES ON GLASS

75

FIG. 1. Schematic diagram of colloidal metal lm preparation. k 1 and k 2 are adsorption and desorption rate constants for the rst layer. k 3 and k 4 are adsorption and desorption rate constants for the multilayers, respectively. The bars represent the glass substrate, and the circles, the Au or Ag colloid.

enzyme-catalyzed reaction, allowed us to accurately describe the relationship of peak absorbance with immersion time and to interpret the adsorption mechanism of Au colloids onto modied solid substrates. Contrary to the previous report, the multilayer formation of Au colloidal lm was observed by AFM. To understand the nature of multilayer formation, it is necessary to elucidate detailed reaction pathways of chemisorption of Au particles on modied surfaces. The adsorption reaction can proceed either by ion dissociation of the thiol or by formation of H as follows (29): glass SH Au 3 glass SAu H , glass SH Au 3 glass SAu H . (1) (2)

Reaction (2) is favored considerably over (1) according to the previously reported electrochemical evidence (30 32) as well as conductometric analysis (29). This was afrmed by spectroscopic measurement in this study. If the adsorption reaction proceeds as indicated in Eq. (1), terminal SAu will induce signicant colloid aggregation as occurred after addition of anions such as Cl and OH . However, no colloid aggregation was observed during the immobilization reaction. As shown in the Fig. 2, the absorption peak in UVvis spectra indicated that the number of adsorbed colloids is limited. If adsorption proceeds by reaction pathway (2), the surface charge will not develop. Instead, a slight partial change may appear on the surface of an immobilized colloidal particle due to the difference in electronegativity of Au and S atoms. Since the electronegativity of S is slightly higher than that of Au, a partial positive charge appears in Au particles which act as nuclei for additional particle adsorption. This was veried by AFM studies. Monodispersed Au colloids were covalently adsorbed on surface-conned MPTMS-coated glass as shown in Fig. 3. If the initial negative charge of

76

PARK ET AL.

FIG. 2. Absorption spectra of Au colloidal lms immobilized on MPTMS-coated glass slides: The immersion times in a suspension of 16-nm-diameter Au colloid particles were 5 min, 10 min, 30 min, 2 h, 18 h, and 24 h (from bottom to top, respectively).

Au colloidal particles was signicantly weakened, or at least converted to positive, a three-dimensional multilayer would form rather than a monolayer of Au colloids. Au particles in suspension approach two different regions of the substrate as depicted in reactions (3) and (4). An Au particle diffuses toward surface-conned thiol functional groups as in case (3) or to already adsorbed Au particles on the glass surface as in (4): k1 0 glassSAu H , glassSH Au | k2 k3 glassSAu Au | 0 glassSAun . k4

(3)

(4)

During the initial adsorption period, reaction (3) is favored over (4) because more thiol functional groups are available than adsorbed Au particles on the substrate. Most of the hydrogen radicals generated by immobilization of Au particles were converted to H 2 molecules and consequently stored as H 2O 2 since a trace amount of O 2 is available in the reaction (28). It is reasonable to consider that the reverse rate constant k 2 in reaction (3) is negligible due to a strong covalent bond between the sulfhydryl and the Au particle. Furthermore, it was found that CMFs stored in water for several weeks maintain a constant optical density, indicating that k 4 is small enough to be neglected too. Since the result of

Au AND Ag NANOPARTICLES ON GLASS

77

FIG. 3. AFM image of Au colloidal lm prepared by immersing a MPTMS-coated glass slide in Au colloid ( d 16 nm) suspension for 30 min.

reaction (4) is the formation of a multilayer of Au colloidal lm, the rate equation of the multilayer formation reaction can be expressed as follows with the rate constant k 3: V k 3 glass SAu Au . (5)

Although it appears that the reaction rate depends on both [glass SAu] and [Au] in Eq. (5), it was found that [glass SAu] governs the reaction rate of multilayer formation. The effective concentration of the Au colloid is estimated to be in the range of a few nanomoles, whereas Au surface coverage at a fully adsorbed state was around 20 fmol. Based on the experimental ndings and the concentration ratio, it is reasonable to neglect [Au] and simplify Eq. (5) to the rate of formation of a multilayer, V k 3 glass SAu , the rate of formation of the rst monolayer, glass SAu k 1 glass SH Au , (7) (6)

78 and the rate of breakdown of

PARK ET AL.

glass SAu k 3 glass SAu . The following relationships can be established under steady-state conditions: k 1 Au glass SH k 3 glass SAu , glass SAu If K M is dened as K M k 3/ k 1 Eq. (10) can be expressed as glass SAu Au glass SH . KM Au glass SH . k 3/ k 1

(8)

(9) (10)

(11)

(12)

It is reasonable to say that the Au colloidal concentration in the suspension is considered constant throughout the reaction since the decrease in concentration of the colloid by adsorption on the solid substrate is near zero. Therefore, the following relationship can be established: glass SH glass SH total glass SAu . Substitution of the expression in Eq. (12) then gives glass SAu Solving for [glass SAu], glass SAu or glass SAu glass SH total Au . Au K M glass SH total Au / K M 1 Au / K M glass SH total glass SAu Au . KM (13)

(14)

(15)

(16)

Au AND Ag NANOPARTICLES ON GLASS

79

Therefore, the rate of formation of the Au colloidal multilayer can be expressed as V k 3 glass SH total Au . Au K M

(17)

If [Au] is much larger than K M, the maximal rate of the adsorption reaction can be expressed as V max k 3 glass SH total if Au K M. (18)

Substitution of the expression for V max into Eq. (17) then gives V V max or 1 1 KM 1 . V V max V max Au From Beers law, A abc , If the value of ab is constant, dA dc ab , dt dt dA abV , dt dA abVdt , A abVt . Therefore, VA 1 abt V max A max 1 . abt dc V, dt dA d abc , dA d abc . dt dt Au Au K M

(19)

(20)

and

(21)

80

PARK ET AL.

FIG. 4. Plot of the reciprocal of the absorbance of Au colloidal lms at max versus the reciprocal of the immersion time ( t ). The diameter of Au particles is 16 nm and the absorbance at the saturation point is 0.089. K M was calculated to be 0.319. The immersion time for each lm preparation was the same as in Fig. 2.

If the above expression is substituted into Eq. (20), 1 1 KM 1 . A/ abt A max/abt A max/ abt Au (22)

Since [Au] is directly related to the time of immersion of the substrate in the colloidal suspension, Eq. (22) can be expressed as 1 1 KM 1 . A A max A max t (23)

Figure 4 is a plot of the reciprocal of the absorbance maxima versus the reciprocal of immersion time. All of the data points dene a nearly straight line in accordance with Eq. (23). From the data in Fig. 4, we obtained the best-t value of A max 0.089, which complied well with the experimental results. The relationship k 3 / k 1 0.319 demonstrates that the rate of formation of the rst monolayer is kinetically three times more facile than that of the multilayer. Many studies have proven that the adsorption kinetics of Au on amine and thiol functionalized surfaces are almost identical (23). The adsorption characteristics of Au colloid of various particle sizes were investigated. We studied colloid suspensions having maximum absorbance at max 519.5, 526.5, 535.0, and 540.5 nm, corresponding to particle diameters of 12, 41, 71, and 147 nm, respectively. We used APTMS as a coupling agent and induced Eq. (23) to analyze adsorption behavior. The difference in wavelength of absorption peaks indicates the energy level of surface plasmon of the particle according to size. For each particle size, the absorbance saturation point was observed at longer

Au AND Ag NANOPARTICLES ON GLASS

81

immersion times as shown in Fig. 5. It was found that k 3 / k 1 values were proportional to particle size. For instance, K M values varied from 0.13 to 3.15 to 5.41 as the diameter of the colloid particle increased to 12, 41, and 71 nm, respectively. The K M value estimated from the adsorption of 16-nm Au colloid on an MPTMS-modied surface is well-tted with those obtained from adsorption of different-sized Au colloid on an APTMS-modied surface. Therefore, it is reasonable to say that the adsorption kinetic behavior of Au colloid of different sizes on an MPTMS-modied surface will closely resemble that on an APTMS surface. This result indicates that the adsorption kinetics of Au colloid varies drastically with particle size. For example, for smaller Au particles, the rate of formation of the rst monolayer is relatively faster than that of the multilayer. On the other hand, for larger colloids rst layer formation occurs at a slow rate compared with multilayer formation. Facilitated multilayer formation for larger colloids reects the fact that surface-conned larger particles act as nucleation sites more effectively for adhesion of Au particles on the top of the rst colloidal layer. Large metal particles tend to develop high surface charge as well as high polarizability compared with smaller particles. This observation is consistent with the well-known characteristics of the Au colloid, in that smaller colloids tend to be more stable and resistant against aggregation compared with larger particles. This is presumably responsible for the observation of larger k 3 / k 1 values from largeparticle reactions compared smaller-particle reactions. That is to say, the stability difference varied with the sizes of the colloid and is directly related to the characteristics of the surface adsorption reaction. In the initial period of immobilization, a larger particle is less effectively immobilized on the surface by the functional groups of thiol or amine, which reects smaller k 1 with relatively large k 3 . Therefore, for larger Au colloids the immobilization reaction shows a large k 3 / k 1 value, indicating more facilitated multilayer formation. The extent of aggregation of gold colloid on the substrate can be measured by UVvis spectroscopy, especially in the region 600 to 700 nm. Prolonged immersion time leads to an increase in the low-energy band around 640 nm as shown in Fig. 5. The new band at 640 nm indicates a typical electronic transition of aggregated Au particles (33). However, this feature was not seen when MPTMS was used as a coupling agent between glass surface and Au colloid. It may be due to the electronegativity difference between nitrogen and sulfur atoms. Nitrogen is more electronegative than sulfur. Therefore, the amine group serves as a nucleus more effectively than the thiol group, leading to effective particle occulation. By observation of the low-energy band near 640 nm, it was found that particle size was limited. Occurrence of the low-energy band is also related to colloidal size. When small Au colloid particles with a particle diameter of 12 nm were immobilized on the APTMS-modied surface, the low-energy band did not appear, even at long immersion times. As evidenced in Fig. 6, the longer immersion time leads only to higher surface coverage on the substrate with nearly identical electronic transition. It is interesting to compare AFM images of CMFs composed of different sizes of Au nanoparticles. Figure 7 clearly shows that the adsorption rate of small particles in the rst layer is larger than that of large particles, which is consistent with the UVvis studies. Even though CMFs were constructed with the same immersion times, the surface coverage with colloid particles 41 nm in diameter was higher than that with 71-nm particles.

FIG. 5. Absorption spectra of Au colloidal lms and their plots according to Eq. (23). The diameters of Au particles are 12 nm (a), 41 nm (b), and 71 nm (c). The immersion times for Au CMF construction were controlled at 15 min, 20 min, 30 min, 1 h, 2 h, 3 h, 18 h, and 24 h (from bottom to top, respectively). 82

Au AND Ag NANOPARTICLES ON GLASS

83

FIG. 6. AFM images of Au colloidal lms prepared by immersing APTMS-coated glass in a suspension of 41-nm-diameter Au colloid particles for (a) 15 min, (b) 20 min, (c) 1 h, (d) 3 h, (e) 6 h, and (f) 24 h, respectively.

84
PARK ET AL.

FIG. 7. AFM images of Au colloidal lms prepared by immersing APTMS-coated glass slides for 6 h in various sizes of gold particles. d (a) 41 nm, (b) 71 nm, (c) 147 nm.

Au AND Ag NANOPARTICLES ON GLASS

85

FIG. 8. Absorption spectra of 50- to 60-nm-diameter Ag colloidal lms on APTMS-coated glass. The immersion time in Ag colloidal suspension was varied: 15 min, 30 min, 1 h, 2 h, 3 h, 6 h, 24 h, and 48 h (from bottom to top, respectively).

It is well known that the silver particle is nearly identical in adsorption behavior to gold toward thiol or amide moieties. However, the adsorption kinetics of Ag nanoparticles on such terminal groups on the solid substrate has not been investigated. When a surfacemodied glass slide with an amide group was exposed to Ag colloid suspension, the peak absorbance of Ag colloidal lm in UVvis spectra was proportional to the immersion time as indicated in Fig. 8. The kinetic behavior of Ag colloid immobilization on an APTMSmodied glass surface is illustrated in Fig. 9, which is a plot of 1/ A versus 1/ t . The tendency of the plot is almost identical as in the case of the Au colloid. It indicates that the adsorption characteristics of Ag are quite similar to those of Au. Figure 10 shows the evolution of absorption bands of Ag colloid lm on a glass slide in air at different immersion times. It was interesting that a blue shift in the absorption maximum and a new energy band were observed in UVvis spectra of Ag colloidal lm when the intermediate material was changed from water to air by evaporation of water from the colloid lms. Two broad bands near 500 and 600 nm appeared at longer immersion times, reecting the formation of an aggregate multilayer of Ag nanoparticles on the surface. This blue shift may stem from the change in the aggregate state and in interparticle interaction as observed in a previous report (35). Interparticle conglomeration is also responsible for peak broadening and appearance of a new energy band. Although the Ag colloidal lms of Fig. 10 were prepared under conditions identical to those under which the lms in Fig. 8 were prepared, the absorption spectra in Fig. 10 were acquired in a nitrogen atmosphere, not in water. Therefore, differences in the absorption spectra in Figs. 8 and 10 stem from the degree of aggregation, which was induced by the substitution of intermediate materials between the colloid particles. The dielectric constant of a solvent

86

PARK ET AL.

FIG. 9. Plot of the reciprocal of the absorbance of Ag colloidal lms at max versus the reciprocal of the immersion time ( t ): 1/ A vs 1/ t . The absorbance at the saturation point is 2.782 and K M is 1.97.

is often used as a measure of its capacity to cause separation of charged particles. Since water has a larger dielectric constant compared with air, water will separate unstable charged silver particles more effectively than air. Consequently, a change in intermediate material from water to air would cause colloid particle aggregation on the substrate. To conrm this interpretation silver colloidal lm was observed by AFM. As shown in Fig. 11, silver colloid particles generally have a relatively broader size distribution than gold colloid particles. As immersion time increased from 30 min to 3 h and nally to 48 h, the particle size in the lms also appeared to increase. In the case of Ag colloidal lms constructed by 48 h immersion, it was difcult to obtain AFM images due to the irregular surface morphology. This observation of an increase in particle size with longer immersion times is analogous to the aggregation of Ag particles induced by replacing water with nitrogen in the drying process of the lms. Surface-enhanced Raman scattering spectroscopy was used to investigate the effects of morphology and surface plasmon variation by replacing intermediate materials. SERS spectra of p -aminothiophenol (PATP) were obtained only on the multilayered silver colloidal metal lm, not on the monolayer of Ag CMFs (Fig. 12). The prepared silver colloidal lm was immersed for 15 min in a 10 3 M ethanol solution of PATP, and then washed subsequently with ethanol and water. To obtain SERS spectra, it was necessary to increase surface particle density to the multilayer extent. There was a signicant difference in SERS spectra of PATP measured on a dried colloidal lm and those on a wet lm. When a wet colloidal lm was used as a SERS substrate, Raman bands at 1573, 1440, 1391, and 1142 cm 1 were tremendously increased as water evaporated from the surface. All those modes have been assigned as b 2, in-plane modes (34). Since Raman bands associated with the in-plane mode selectively show strong intensities, PATP is adsorbed with a higher tilted angle to the substrate compared with the other two cases. On the other hand, the SERS spectrum in Fig. 12a indicates that

Au AND Ag NANOPARTICLES ON GLASS

87

FIG. 10. Absorption spectra of dried Ag lms in a nitrogen atmosphere. Films were prepared by immersing APTMS-coated glass slides in Ag colloid suspension for 15 min, 30 min, 1 h, 2 h, 3 h, 6 h, and 24 h (from bottom to top, respectively).

PATP is oriented at when it is adsorbed on a dried Ag colloid lm. The selective enhancement of the b 2 modes can also be explained by the charge-transfer mechanism (7). The charge transfer between metal particle and adsorbate induces a so-called resonance Raman-like scattering process on the Ag CMF surface. In the charge-transfer mechanism, the increase in SERS intensity of the adsorbate in the vicinity of the substrate is explained by the change in the energy gap of the constituents. When the energy gap between the energy state of the Fermi resonance ( E f) and the electronic transition energy level of the adsorbate becomes equal to the excitation energy, SERS is operative. As a consequence of the appearance of new energy bands and peak broadening in UVvis spectra of the dried lms, the energy of the excitation source of 514.5 nm is matched by the energy gap between the E f of silver colloidal lm and the electronic level of p -aminothiophenol on the Ag colloidal lm. However, if this is the sole reason for the selective increase in b 2 modes, the SERS spectrum of PATP on a silver colloidal lm, which was dried previously to expose PATP, must show a spectrum similar to that obtained from a silver CMF lm on which PATP was adsorbed in the wetted state and then dried. The three SERS spectra in three different states of Ag CMFs in Fig. 12 are apparently different. This difference in the spectra can be explained by the electromagnetic enhancement effect, which involves the enhancement of the optical elds near the surface. Electromagnetic (EM) enhancement requires small surface roughness to permit surface plasmon resonance, resulting in strong local enhancement of the electric eld of both incoming and scattered radiation. SERS intensity decays exponentially as the distance between the metal and the sample increases. It has been known that when the

88
PARK ET AL.

FIG. 11. AFM images of Ag colloidal lms prepared by immersing APTMS-coated glass slides in Ag colloid suspension for (a) 30 min, (b) 3 h, and (c) 48 h.

Au AND Ag NANOPARTICLES ON GLASS

89

FIG. 12. SERS of p -aminothiophenol on Ag colloidal lms: (a) on a dried Ag colloidal lm, (b) on a wet Ag colloidal lm, (c) on a wet Ag colloidal lm that was dried in a nitrogen atmosphere.

sample is sandwiched between metal particles, maximal SERS intensity can be obtained. If the sample adsorbs to an irregular metal surface, it will not be easy to be sandwiched between metal particles. However, if the sample adsorbs on a relatively smooth colloid aggregate surface, and then the surface morphology changes to an irregular morphology by the gradual evaporation of water on the surface, it will be sandwiched between metal particles. Therefore, distinctive adsorption patterns of PATP induced by different morphologies of the SERS substrates are reected in the SERS spectra of Fig. 12. CONCLUSIONS Au and Ag colloidal lms with well-dened two- or three-dimensional structures were successfully constructed on glass slides modied with self-assembled monolayers (SAMs) of bifunctional silanes containing either amine or thiol groups on the opposite end of the molecule. The specic afnity of Au and Ag colloids to the amine functional groups of the SAMs on the glass surface allowed the fabrication of well-ordered, stable, and reproducible colloidal lms with careful control of experimental conditions. The relationship between surface coverage and adsorption time was precisely described by using the well-known MichaelisMenten model. The adsorption characteristics were investigated with various sizes of colloidal particles. It was found that the k 3 / k 1 values were proportional to particle size. This implies that the rate of formation of the rst monolayer of relatively small particles is faster than the rate of multilayer formation. Contrary to smaller particles, for large colloid particles the rate of formation of the rst layer is slower than the rate of multilayer formation, suggesting large colloid particles aggregate more effectively.

90

PARK ET AL.

Prolonged immersion time led to an increase in the low-energy band in the absorption spectra of Au colloid lms between 600 and 700 nm. This was found only with APTMS, not with MPTMS, as a coupling agent. The amine group has a stronger effect on particle occulation than the thiol group. Surface-enhanced Raman scattering spectra of p -aminothiophenol adsorbed on a Ag colloidal lm were obtained with 514.5-nm excitation. The SERS intensity of p -aminothiophenol varied signicantly with the topology of the substrates. The surface morphology was monitored by AFM. The morphological variation affects the Fermi level of the colloidal lm. The increase in SERS intensities was explained by both electromagnetic and charge-transfer effects. As a consequence of H 2O evaporation, the space between Ag particles is replaced by air and the morphology of the Ag colloidal lm is changed. This morphological alteration parallel to H 2O evaporation is presumably induced by the difference in dielectric constants of air and of H 2O. ACKNOWLEDGMENTS
The authors are grateful for the nancial support of the Korean Research Foundation made in the program year of 1997. We also thank Mr. Kyung-Ho Lee and Ms. Jin- Ho Park for their excellent experimental contributions to this work.

REFERENCES
1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. Fleischmann, M.; Hendra, P. J.; McGuillan, A. J. Chem. Phys. Lett., 1974, 26, 163. Jeannair, D. J.; Van Duyne, R. P. J. Electroanal., 1997, 84, 1. Albrecht, M. G.; Creighton, J. A. J. Am. Chem. Soc., 1977, 99, 5215. Cotton, T. M.; Kim, J.-H.; Uphaus, R. A. J. Microchem. J., 1990, 42, 44. Prog, Gi. X. Polym. Sci., 1994, 19, 317. Ritchie, G.; Chen, C. Y. Pure Appl. Chem., 1980, 52, 361. Cotton, T. M.; Kim, J.-H.; Chumanov, G. D. J. Raman Spectrosc., 1991, 22, 729. Otto, A. Top. Appl. Phys., 54.; Cardona, M.; Guntherodt, G. (Eds.) Light Scattering in Solids, Vol. 4, p. 289. Springer, Berlin, 1984. Moskovits, M. Rev. Mod. Phys., 1985, 57, 783. Creighton, J. A. Spectroscopy of Surfaces. Wiley, Chichester, 1988. Garrell, R. L. Anal. Chem., 1989, 61, 401A. Koglin, E.; Sequaris, J.-M. Topics in Current Chemistry, p. 8. Springer-Verlag, Berlin/Heidelberg, 1986. De Groot, J.; Hester, R. E. J. Phys. Chem., 1987, 91, 1693. Halperin, W. P. Rev. Mod. Phys., 1986, 58, 533. Henglein, A. J. Phys. Chem., 1993, 97, 5457. Kreibig, U. In Contribution of Cluster Physics to Materials Science and Technology (J. Davenas and P. M. Rabette, Eds.), NATO ASI Series E, No. 104, p. 373. Martinus Nijhoff, Dordrecht, 1986. Turkevich, J.; Garton, G.; Stevenson, P. C. J. Colloidal Sci. Suppl., 1954, 1, 26. (a) Kreibig, U. J. Phys. Colloq., 1977, 38, C229. (b) Doremus, R. H. J. Chem. Phys., 1964, 40, 2389. (c) Hampe, W. Z. Phys., 1958, 152, 476. Creighton, J. A. Surface Studies with Lasers (F. R. Ausseneg, A. Leitner, and M. E. Lippish, Eds.), p. 55. Springer-Verlag, New York, 1983. (a) Pettinger, B.; Bao, X.; Wilcock, I. C.; Muhler, M.; Ertl, G. Phys. Rev. Lett., 1994, 72, 1561. (b) Van Duyne, R. P.; Hulteen, J. C.; Treichel, D. A. J. Chem. Phys., 1993, 99, 2101. (c) Roark, S. E.; Rowlen, K. L. Chem. Phys. Lett., 1993, 212, 50. (d) Schueler, P. A.; et al. Anal. Chem., 1993, 65, 3177. (e) Mullen, K. I.; Carron, K. T. Anal. Chem., 1991, 63, 2196. (f) Byahut, S.; Furtak, T. E. Langmuir, 1991, 7, 508. (a) Schnur, J. M. Science, 1993, 262, 1669. (b) Huber, C. A.; et al., Science, 1994, 263, 800. (c) Whitesides, G. M.; Mathias, J. P.; Seto, C. T. Science, 1991, 254, 1312. Grabar, K. K.; Freeman, R. G.; Hommer, M. B.; Natan, M. J. Anal. Chem., 1955, 67, 735743.

21. 22.

Au AND Ag NANOPARTICLES ON GLASS

91

23. Freeman, R. G.; Grabar, K. C.; Allison, K. J.; Bright, R. M.; Davis, J. A.; Guthrie, A. P.; Hommer, M. B.; Jackson, M. A.; Smith, P. C.; Walter, D. G.; Natan, M. J. Science, 1995, 267, 1629 1632. 24. Bright, R. M.; Walter, D. G.; Musick, M. D.; Jackon, M. A.; Allison, K. J.; Natan, M. J. Langmuir, 1996, 12, 810 817. 25. Hayat, M. A. Colloidal Gold. Academic Press, San Diego, 1989. 26. Munro, C. H.; Smith, W. E.; Garner, M.; Clarkson, J.; White, P. C. Langmuir, 1995, 11, 3713. 27. Goss, C. A.; Charych, D. H.; Majda, M. Anal. Chem., 1991, 63, 85. 28. Grabar, K. C.; Smith, P. C.; Musick, M. D.; Davis, J. A.; Walter, D. G.; Jackson, M. A.; Guthrie, A. P.; Natan, M. J. J. Am. Chem. Soc., 1996, 11, 1148. 29. Schessler, H. M.; Karpovich, D. S.; Blanchard, G. J. J. Am. Chem. Soc., 1996, 118, 9645. 30. Karpovich, D. S.; Blanchard, G. J. Langmuir, 1994, 10, 3315. 31. Widrig, C. A.; Chung, C.-J.; Porter, M. D. J. Electroanal. Chem., 1991, 310, 335. 32. Weishaar, D. E.; Lamp, B. D.; Porter, M. D. J. Am. Chem. Soc., 1992, 114, 5860. 33. Chumanov, G.; Sokolov, K.; Gregory, B. W.; Cotton, T. M. J. Phys. Chem., 1995, 99, 9466. 34. Osawam, M.; Matsuda, N.; Yoshii, K.; Uchida, I. J. Phys. Chem., 1994, 98, 12702. 35. Chumanov, G.; Sokolov, K.; Gregory, B. W.; Cotton, T. M. J. Am. Chem. Soc., 1995, 9466.