Layer-by-layer Self-Assembly of Colloidal Gold-Silica Multilayers

Introduction

Significant effort is currently being invested to assemble nanostructured materials onto substrates, to study their novel properties, and to investigate their potential device applications[1–3]. One long term goal is the creation of “artificial” solids whose optical and electrical properties can be “tuned” by varying the identity of component “atoms” and their interparticle interactions [4]. Among the many nanostructured materials available [5–9], Au nanoparticles (NPs) are attractive candidates because of their well-controlled syntheses, narrow size distributions, and unique optical/electrical properties.

Several strategies have been developed to assemble Au nanoparticles onto substrates. Organic functional groups have been used to covalently link Au NPs to surfaces [10–14]. Alternatively, polyelectrolytes have been employed to immobilize Au particles via electrostatic interactions [15]. These methods possess advantages over other techniques such as vapor deposition or Langmuir–Blodgett assembly in terms of their low cost and high throughput while conserving fine control over the composite product.

For potential applications, the formation of dense NP monolayers is also desirable. Various methods have therefore been explored to achieve high particle densities; however, coverage of immobilized NPs is often limited by repulsive forces stemming from charges on stabilizers passivating their surfaces. These same charges prevent particles from agglomerating in solution. In the current experiment, the student minimizes this problem by removing most of the stabilizer from the Au sol using a mixed-bed ion-exchange resin. This leads to robust NP monolayers that are among the densest to date.

Subsequent layer-by-layer (LBL) self-assembly is then utilized to construct three dimensional (3D) multilayer films. Both the interparticle distance within Au monolayers and the separation between these layers control the optical and electrical properties of such “artificial” solids. The current experiment tunes monolayer spacings through variable thickness silica layers. Other ways exist for controlling this parameter. One approach involves coating each component Au NP with a silica shell of variable thickness [13]. Another involves varying the number of polymer bilayers between Au NP monolayers [14, 15b]; however, irrespective of the approach, the ability to obtain NP multilayers is the first step towards true designer materials.

Experimental

Au Nanoparticle Preparation. Au NPs are prepared using the Turkevich [16] method. This generates ~15 nm diameter Au particles. Specifically, 400 mL of deionized (DI) H₂O containing 0.1 g HAuCl₄ 3H₂O is heated to boiling (Note that quantities can be scaled down as needed). When ready, 50 mL of a 1% (by weight) sodium citrate solution is added to the solution under vigorous stirring. The mixture is then boiled for an additional 20 min whereupon it is left to cool. The final volume is adjusted to 500 mL using DI water. The obtained sol is wine red and the Au concentration is approximately 0.5 mM (in mol atoms per L). Representative low- and high-resolution transmission electron microscope (TEM) images of the particles are shown in Figure 1. Systematic particle sizing can be conducted by analyzing such micrographs or by dynamic light scattering.

The gold NP absorbance exhibits a strong band around 520 nm. The origin of this resonance stems from the collective excitation of free electrons in the material and is referred to as a surface plasmon. Its peak wavelength depends on particle size and shape as well as on the dielectric constant of the surrounding medium. More about the properties of plasmons can found below as well as in Reference 17.

To enhance the adsorption of Au NPs onto glass, most of the citrate anions on the particle surface must be removed. This is achieved by adding 8 g of ion exchange resin (Amberlite MB-150, Aldrich) to the 500 mL Au sol while stirring. The resulting suspension is not stable for long periods of time and its color changes to a blue hue after several days. This indicates aggregation of the particles; therefore, any

Table 1. Effect of solvent and accompanying dielectric constant on the Au NP monolayer peak plasmon band position.

Solvent	Optical dielectric constant	Peak wavelength (nm) +/- 5%
Air	1.000	516.2
Water	1.786	525.7
Acetone	1.847	521.0
Ethanol	1.850	521.0
Hexane	1.891	525.5
Ethylene glycol	2.048	525.7
1,3 Propanediol	2.074	526.7
Toluene	2.238	531.4
Nitrobenzene	2.406	532.3
Bromobenzene	2.430	534.3

Figure 1. (a) Low and (b) high resolution TEM images of Au nanoparticles.

NP deposition or UV/visible absorption measurements should be conducted quickly once the particles have had their anion layer stripped. Subsequent adsorption of the particles onto pre-conditioned glass surfaces is fast and yields dense coverages.

Conditioning the Glass and Forming a Gold NP Monolayer. Rectangular slides (quartz, from Gold Seal Products, or Premium Cover Glass, from Fisher 0.9 ´ 5 cm) are cleaned in 95% ethanol and 1% by weight KOH. The substrates are then rinsed with DI water. Next, to activate their surfaces, the coverslips are immersed into a 0.54% (by weight) silicate solution overnight. Once complete, the slides are immersed in a 1% (by volume) solution of 3-aminopropyltrimethoxysilane (NH₂(CH₂)₃Si(OCH₃)₃, APS) for 3–4 hours. They are subsequently rinsed with deionized (DI) water. When fresh, the APS solution is turbid; however, it becomes transparent when significant (unwanted) condensation has occurred. To immobilize Au NPs and to create a gold monolayer, the APS functionalized glass is immersed into the resin-treated sol. Monolayer deposition is complete in approximately 20 minutes but the student can follow its accumulation with time as demonstrated in Figure 2c.

Effects of the Medium on the Spectrum. At this point an experiment can be conducted to illustrate the tunable optical/electrical properties of the eventual 3D assembly. In this sense, the plasmon absorption of the above Au monolayer is exquisitely sensitive to its local environment. Specifically, the peak position of the Au plasmon resonance varies with solvent dielectric constant. To observe this property, the Au NP monolayer is immersed into a chosen solvent for 30 minutes. It is then transferred into a cuvette containing the same solvent and the absorption spectrum is measured. Both the peak absorption wavelength and solvent dielectric constant are recorded. Between experiments, the slide is rinsed and immersed in a solvent miscible with both the previous solvent and the following one. Using this strategy, the same NP monolayer can be used for multiple measurements. Whereas various methods have previously been used to demonstrate the effects of the medium on the NP plasmon spectrum [17], the present experiment ensures that any observed spectral changes are solely due to variations in the medium’s dielectric constant as opposed to NP structural or chemical variations resulting from agglomeration, for example.

Table 1 lists various solvents that have been used in this experiment along with their respective optical dielectric constants. The observed peak wavelength of the plasmon resonance is also listed, showing systematic changes with solvent dielectric constant. Such sensitivities illustrate that above NP monolayers may be used for sensing purposes [5, 17]. In this regard, the student should realize that this sensitivity to the dielectric constant of the medium is the basis of popular analytical methods for identifying adsorbed biomolecules including proteins and DNA; however, the experiment ultimately highlights the control one has over the optical and electrical properties of the NP assembly.

Layer-by-Layer Self Assembly of a Au-Silica NP Multilayer. Now, to form a multilayer, the glass slide with the initial Au NP monolayer is immersed in 1% APS for 3 hours. This is followed by immersion in tetraethyl orthosilicate (TEOS, Si(OCH₂CH₃)₄) for 15 hours. The latter solution is prepared by adding 0.4–0.5 mL TEOS and 2 mL ammonia into 500 mL 4:1 ethanol:DI water solution. This process leads to coverage of the first Au monolayer by a silica layer. Next, another layer of Au particles is deposited onto the insulating silica layer by immersing the slide in 1% APS for 3 hours followed by immersion in the resin-treated Au sol for an additional 3–4 hours. Note that the glass slide must be rinsed thoroughly with DI water after each immersion. The above steps are repeated to obtain sequential alternating Au NP-silica monolayers. The absorption spectrum of the slide can be taken after each Au monolayer has been added to follow the process spectroscopically (see Figure 4). This can then be used to verify the quality of the layer deposition based on UV-Visible measurements described below.

Discussion

Gold Nanoparticle Self Assembly. Au NPs have little affinity for glass. This is because their surfaces are highly hydrophobic while those of the glass are hydrophilic. Furthermore, silica and most other oxides in water are covered by hydroxide or hydronium ions while those of the NP contain ions originating from precursors in their synthesis (citrate in our case). Therefore, a bridging primer molecule is used to link Au to SiO₂. Primers are generally bifunctional. In this case, one side of the molecule contains a silane group while the other possesses a mercapto or amine group. The current experiment employs aminosilane bridges such as APS, NH₂(CH₂)₃Si(OCH₃)₃, or alternatively, 3-aminopropyltriethoxysilane [APTES, NH₂(CH₂)₃Si(OC₂H₅)].

The first step in attaching the linker to glass is the hydrolysis of the methoxy terminus of the primer to yield a silanol head group. This is then followed by condensation polymerization of the silanol group with surface-bound OH groups. Both steps are described by Reactions 1 and 2 below where the subscript s denotes “surface”.

(MeO)₃Si-R + 3H₂O 3MeOH + R-Si(OH)₃ (1)

R-Si(OH)₃ + 3(SiO₂)_s-OH 3H₂O + R-Si(-O)₃-(SiO₂)_3s (2)

Figure 2. (a) UV/visible absorption spectrum of a monolayer of 15-nm diameter Au NPs. (b) AFM image of the same monolayer. (c) Time evolution of the absorbance at 520 nm showing saturation in approximately 20 minutes.

The relative rates of hydrolysis (Reaction 1) versus condensation (Reaction 2) are determined by the polarity of the solvent as well as its pH. Such rates determine the amount of surface binding versus self condensation among the silanol head groups of APS, and thus the quality of the desired surface binding. Subsequent exposure of the treated substrate to a NP sol initiates monolayer formation via the free amine-terminus binding to gold.

Figure 2a shows the absorption spectrum of the resulting monolayer. The plasmon peak at 520 nm is apparent and is nearly identical to that of the original Au NPs in solution. No near infrared absorption is seen. This indicates that within the monolayer, the particles behave more or less independently. Figure 2b shows an atomic force microscope (AFM) image of the product monolayer. From the size of the Au NPs, determined by TEM measurements (Figure 1a; Note that particle sizes can also be determined by dynamic light scattering; however, obtained values are somewhat distorted because the latter technique determines the average NP volume in solution.), and from the number of particles that can be counted in the AFM image (Figure 2b), one can calculate the fraction of the total surface area occupied by the particles. If an AFM is available, the student can also conduct such measurements firsthand. Otherwise, the particle density measured here can be used. In spite of the seemingly dense coverage in Figure 2b, only 25% of the available surface is actually occupied.

Figure 2c shows that the plasmon absorbance of the Au film levels off approximately 20 minutes after the beginning of the deposition. The student may check this during the actual experiment by measuring the UV/visible absorbance periodically. The eventual saturation of the absorption shows the self-limiting behavior of the monolayer.

Dielectric Sensitivity of the Plasmon Resonance. To explain the sensitivity of the Au NP plasmon resonance to the dielectric constant of its environment, we refer to Mie theory. Mie (1908) first described the correlation between the peak wavelength of the plasmon band with the dielectric constant of the surrounding medium [18]. For metal particles, when the radius, R, is much smaller than the wavelength of incident light, Mie’s theory simplifies to the following expression:

(3)

which relates the NP cross section to various optical parameters. In eq 3, s_ext is the Au NP extinction cross section (in cm²), w is the angular frequency of the incident light, c is the speed of light, e_m is the frequency-dependent dielectric constant of the surrounding medium, R is the NP radius and e₁ (e₂) is the real (imaginary) part of the material’s frequency-dependent dielectric constant. This expression shows the dependence of the material’s absorption on the dielectric constant of the surrounding medium

()

It can further be shown that the wavelength of the peak plasmon absorption depends on e_m through:

(4)

where l_p is the metal’s bulk plasma wavelength, e^¥ is the material’s high-frequency dielectric constant and e_m is again the surrounding medium’s dielectric constant. From the equation it is apparent that the position of the Au plasmon band increases with increasing dielectric constant of the medium. This provides the guiding rationale for sensing based on surface plasmon resonance (SPR) spectroscopy [17].

Layer-by-Layer Self-Assembly of a Au NP Multilayer. The underlying basis for multilayer formation in the current experiment is polycondensation of silicate. This is a sol-gel process applied to the preparation of glass ceramics and to the synthesis of silica particles. The procedure was first developed by Stober in 1968 and is often called the Stober process [19]. Here, it is first utilized in the attachment of primer molecules to the silica glass surface and then to generate layers of silica that separate Au NP monolayers from one another. The overall strategy is illustrated in Figure 3.

In the experiment, the employed silicate and alcohol are tetraethyl-orthosilicate (TEOS, Si(OCH₂CH₃)₄), and ethanol. The ammonia used in the reaction provides basic pH conditions which catalyzes the reaction. The complete hydrolysis of TEOS and the complete condensation of the

Figure 3. Schematic strategy by which to obtain layer-by-layer deposition of Au/SiO₂ on glass. (a) Conditioning the glass with APS primer. (b) Attaching Au particles to the amine head groups of APS. This yields monolayer coverage. (c) Treating the Au monolayer with additional APS where the amine functional group attaches to the particles. TEOS is then added to generate a layer of silica. (d) APS treatment followed by soaking in Au sol generates the 2nd Au NP layer. Steps (c) and (d) can be repeated to obtain a multilayer.

Figure 4. (a) Absorption spectra of 9 consecutively deposited Au-SiO₂ layers on glass. (b) The absorbance at the wavelength of maximum absorption after each deposited Au layer (solid circles; the linear least-squares fit slope is m = 0.1 absorbance units per layer). The linearity further supports the presence of self-limiting monolayer deposition at each step. Triangles are calculated from literature values of the absorption cross section as outlined in the text. A fit to this data yields m = 0.07 absorbance units per layer. The difference between slopes can be attributed to scattering by the silica and its interfaces.

silicic acid derivatives can be represented by equations analogous to eqs 1 and 2:

Si(OCH₂CH₃)₄ + 4H₂O ® Si(OH)₄ + 4CH₃CH₂OH (1a)

nSi(OH)₄ ® nSiO₂ + 2nH₂O (2a)

In all cases, the condensation reaction rate can be controlled by changing the alcohol chain length, the pH, and the identity of the silane.

When the experiment is carried out properly, the sequential adsorption of multiple layers can be achieved. Figure 4a shows sequential absorption spectra of up to 9 monolayers of Au on glass. The first 7 were deposited as described in the Experimental section. By contrast, the last two involved modifications of the procedure to illustrate the effects of varying the interlayer thickness. Figure 4b shows the corresponding peak absorbance at ~520 nm plotted versus number of monolayers. It reveals a linear trend with a slope of m = 0.1 absorbance units per monolayer.

The student can further establish this linear correlation by comparing predictions from a simple monolayer absorbance calculation where only the NP diameter is assumed. To illustrate, a value for the NP molar extinction coefficient is obtained from the literature and is at ~520 nm [20]. Next, to carry out the comparison, e is first converted to its associated extinction cross section (s) through the well known relationship [21]:

(5)

A value of s(cm²) = 1.606 ´ 10^–12 cm² is obtained. We now define a “linear” cross sectional density b = s/d where d is the NP diameter (d = 1.5 ´ 10^–6, b = 1.07 ´ 10^–6cm). The attenuation of light through the monolayer or multilayer can then be expressed in terms of an exponential attenuation relation of the form

(6)

where b is our linear cross sectional density (cm), G is the NP surface density (number/cm²) and l is the optical pathlength (cm). The student will recognize b and G as essentially equivalent to the molar extinction coefficient and concentration in the Beer–Lambert law. A value of G, estimated by examining Figure 2, is G = 1.0 ´ 10¹¹ particles/cm². Values of l can be assumed to be integer numbers of NP diameters (l = kd where k = 1,2,3…). By rearranging eq 6 and relating it to the absorbance () the student obtains the following relationship for either the monolayer or multilayer absorbance

. (7)

Inserting values for the expression and using l = d one determines that the single monolayer absorbance should be A = 0.07. This is in remarkable agreement with the spectra in Figure 4a and, in principle, with the student acquired data. A subsequent comparison over the entire number of monolayers in the multilayer also shows good agreement. The experimentally obtained slope of m = 0.1 absorbance unit per monolayer (circles in Figure 4b) and the difference between experimental and predicted values (m = 0.07 absorbance units per monolayer) can be explained by light scattering due to the silica particles, as well as the silica/glass and silica/gold interfaces. Furthermore, differences due to interlayer interactions arise and will be described below.

Near identical experimental/theoretical slopes and monolayer absorbances demonstrate the homogeneity of each layer. They also support the assertion that the 3D assembly consists of stacked NP monolayers. On a deeper level, though, the quantitative agreement between the predicted and observed absorbances must be carefully considered because the extinction cross section of Au NPs is size- and shape-dependent. As a consequence, it generally cannot be carried over from one NP preparation to the next.

As further extension of this work, 3D assemblies can be constructed using other molecular “bridges.” For example, dithiol or diamine molecules can be used to link synthesized Au monolayers. In either case, the interlayer thickness is determined by the length of the molecule’s alkyl chain.

Interlayer Interactions. Interactions between particles within a single monolayer or across layers in a multilayer can have pronounced effects on the plasmon spectrum.[22] In this regard, despite the relatively high NP density measured here, the peak position of the plasmon band does not change significantly from that observed at lower densities. This can be verified by comparing sequential extinction spectra recorded during monolayer formation in Figure 2c. The student may therefore conclude that at ~25% total surface area coverage distances between particles in a monolayer are too large to allow for interparticle interactions, which affect the overall spectroscopic behavior of the assembly. This conclusion, however, does not exclude the potential role of interlayer interactions.

In this respect, a careful examination of Figure 4a shows that the peak position of the plasmon band does indeed shift from 520 nm to 560 nm within the first seven layers (constructed using the procedure described above). The observed “redshift” is attributed to interlayer particle-particle interactions when the silica layer is thin and reveals tuning of the structure’s overall plasmon spectrum. To further demonstrate control over the 3D assembly’s optical properties, the final 2 monolayers are constructed by increasing the intervening silica layer thickness using more TEOS in the experiment (from 0.4–0.5 mL to 2 mL, Experimental section) and a longer immersion time in the NP sol (from 15 to 40 hours). The end result is a reversal of the plasmon redshift in the 8th and 9th layers, which moves backwards towards 548 nm. The student may likewise demonstrate this effect, showing that increasing the silica layer thickness weakens any interlayer interactions, resulting in a blueshift of the multilayer spectrum towards the single monolayer limit.

Summary

The layer-by-layer self-assembly of Au NP multilayers has been demonstrated. Using ion exchange resins, students create dense Au NP monolayers on glass substrates. Subsequent application of well-developed silicate condensation procedures, enables them to insert intervening SiO₂ layers between self-assembled NP monolayers. This leads to the controlled synthesis of a 3D assembly whose optical and electrical properties can be controlled by varying the dielectric constant of the local environment as well as the thickness of intervening SiO₂ layers. Students show proof of principle results of this effect by monitoring the absorbance of a single monolayer as a function of the local dielectric environment. Later they monitor the controlled assembly of the 3D structure spectroscopically and also have the opportunity to demonstrate control over the optical properties of the solid by varying the spacing between monolayers. This experiment represents a readily accessible example of the creation of potential designer materials made of well-defined nanostructures and was conducted by a group of approximately 40 2^nd-year undergraduate chemistry majors at the University of Notre Dame during the spring semesters of 2007 and 2008.

Supporting Materials. Supporting information describing the sequence of steps followed in the class and laboratory is available (http://dx.doi/10.1333/s00897082135a).

Acknowledgments. M. Kuno thanks the NSF CAREER (CHE-0547784) and NIRT (ECS-0609249) programs for funding. Masaru Kuno is a Cottrell Scholar of Research Corporation. We thank the Notre Dame Radiation Laboratory and the Department of Energy Office of Basic Energy Sciences for funding and use of their facilities. This is contribution number 4736 from the Notre Dame Radiation Laboratory.

References and Notes

1. Mulvaney, P. Langmuir, 1996, 12, 788–800.

2. Liz-Marzan, L.M.; Mulvaney, P. J. Phys. Chem. B, 2003, 107, 7312–7326.

3. Henglein, A. Chem. Rev., 1989, 89, 1861–1873.

4. Shevchenko, E. V.; Talapin, D. V.; Kotov, N. A.; O’Brien, S.; Murray, C. B. Nature, 2006, 439, 55–59.

5. McFarland, A. D.; Haynes, C. L.; Mirkin, C. A.; Van Duyne, R. P. J. Chem. Educ. 2004, 81, 544A–544B.

6. Dungey, K. E.; Muller, D. P.; Gunter, T. J. Chem. Educ. 2005, 82, 769–770.

7. Winkler, L. D.; Arceo, J. F.; Hughes, W. C.; DeGraff, B. A.; Augustine, B. H. J. Chem. Educ. 2005, 82, 1700–1702.

8. Kippeny, T.; Swafford, L. A.; Rosenthal, S. J. J. Chem. Ed. 2002, 79, 1094–1100.

9. Boatman, E. M.; Lisensky, G. C.; Nordell, K. J. J. Chem. Ed. 2005, 82, 1697–1699.

10. 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.

11. Doron, A.; Katz, E.; Willner, I. Langmuir 1995, 11, 1313–1317.

12. Alejandro-Arellano, M.; Ung, T.; Blanco, A.; Mulvaney, P.; Liz-Marzan, L. M. Pure Appl. Chem.2000, 72, 257–267.

13. Liz-Marzan, L. M.; Giersig, M.; Mulvaney, P. Langmuir1996, 12, 4329–35; Mulvaney, P. Langmuir1996, 12, 788–800.

14. Jiang, C.; Markutsya, S.; Tsukruk, V. V., Langmuir 2004, 20, 882–890.

15. (a) Decher, G., Science 1997, 277, 1232-1237;( b) Schmitt, J.; Decher, G.; Dressick, W. J.; Brandow, S. L.; Geer, R. E.; Shashidhar, R.; Calvert, J. M., Adv. Mater. 1997, 9, 61–65.

16. Turkevich, J.; Stevenson, P. L.; Hillier, J Discuss. Faraday Soc. 1951, 11, 55–75.

17. Keating, C. D.; Musick, M. D.; Keefe, M. H.; Natan, M. J. J. Chem. Educ. 1999, 76, 949–955.

18. Bohren, C. F.; Huffman, D. R., Absorption and Scattering of Light by Small Particles; Wiley: New York, 1983; p 530.

19. Stober, W.; Fink, A.; Bohn, E. J. Colloid Interf. Sci. 1968, 26, (1), 62–69.

20. Demers, L. M.; Mirkin, C. A.; Mucic, R. C.; Reynolds, R. A.; Letsinger, R. L.; Elghanian, L.; Viswanadham, G. Anal. Chem. 2000, 72, 5535–5541.

21. Protasenko, V.; Hull, K. L.; Kuno, M. Chem. Educ. 2005, 10, 269–282.

22. Gunnarsson, L.; Rindzevicius, T.; Prikulis, J.; Kasemo, B.; Käll, M.; Zou, S.; Schatz, G.C. J. Phys. Chem. B, 2005, 109, 1079–1087.