Langmuir 2001, 17, 6375-6379 6375

Sol-Gel Processing of Silica-Coated Gold Nanoparticles

Yoshio Kobayashi, Miguel A. Correa-Duarte, and Luis M. Liz-Marzán*
Departamento de Quı́mica Fı́sica, Universidade de Vigo, 36200, Vigo, Spain

Received May 17, 2001. In Final Form: July 13, 2001

Two procedures are presented which allow the homogeneous incorporation of silica-coated gold
nanoparticles within transparent silica gels. Both UV-visible absorption spectra and transmission electron
microscopy show that there is no aggregation of the metal particles during sol-gel transition. The optical
properties of the gels are compared with those of the starting sols and interpreted on the basis of the porous
structure of the gels and standard optical theories. Evidence is also shown of the potential of these methods
for the preparation of gels with different particle sizes and shapes as well as with very high nanoparticle
concentrations, which can be useful for several applications.

Introduction We present here a procedure that overcomes these

The immobilization of nanoparticles of various nature difficulties by means of surface modification of the guest
within gels is very important for a number of technological nanoparticles via deposition of a thin silica shell,10 so as
applications, such as decorative coatings,1 catalysis,2 to preserve their colloidal stability during the whole sol-
optical filters,3 nonlinear optical materials,4 and so forth. gel transition. This permits full control of the nanoparticle
To obtain a homogeneous distribution of the particles morphology (and properties) prior to gelation and at the
within the gel, many approaches to immobilized nano- same time ensures a homogeneous distribution within
particles involve the synthesis in situ,5-7 which makes it the gel. It also allows the colloid to be cleaned by standard
rather difficult to control both monodispersity and particle purification techniques, such as dialysis or centrifugation,
size, especially when high nanoparticle concentrations are so that other substances do not affect the properties of the
to be achieved. In addition, in some procedures, other gel.
chemical species derived from the chemical reactions We have specifically chosen gold nanoparticles to
performed on the precursor molecules are hard to remove demonstrate the performance of this method, since their
from the gel, since they are also firmly incorporated within optical properties are very sensitive to aggregation and
the gel structure. An example is the reduction of AuCl4- to the environment in which they are embedded.11,12
ions in situ, from which Cl compounds normally remain However, the same procedures can be used for the
in the gel. incorporation of other sorts of nanoparticles, both with
An alternative process is the synthesis of nanoparticles simple compositions such as CdS13 and with complex
with the composition, morphology, and properties suitable morphologies such as CdSe@CdS.14 Results are shown here
for the desired application, followed by their incorporation on the formation of gels from two different routes: either
in a rigid gel.8,9 This process is highly desirable in the case by controlled pH changes in aqueous sodium silicate
of nanoparticles with complex structures or specific shapes, solutions or by hydrolysis of tetramethoxysilane in
such as alloys, core-shell particles, or nanorods. An issue methanol/water mixtures.
that needs to be addressed in wet chemical methods is We show that in both cases we can control the particle
that the change of conditions associated with the sol-gel morphology and concentration over a wide range basically
transition is very likely in many cases to affect irreversibly without affecting the stability and the optical properties.
the dispersed state of the nanoparticles. In such cases,
the properties of the final material will be different from Experimental Section
those of the nanoparticles and in general cannot be Materials and Methods. Tetrachloroauric acid (HAuCl4‚
predicted in advance. 3H2O), 3-aminopropyl trimethoxysilane (APS), tetramethoxy-
silane (TMOS), ACS grade methanol, and sodium silicate solution
* To whom correspondence should be addressed. E-mail: (Na2O(SiO2)3-5, 27 wt % SiO2) were purchased from Aldrich and
lmarzan@uvigo.es. Fax: +34 986 812382. used as received. Milli-Q water with a resistivity higher than
(1) Weyl, W. A. Coloured Glasses; Society of Glass Technology: 18.2 MΩ cm was used in all the preparations.
Sheffield, U.K., 1951. UV-visible spectra were measured with a HP 8453 diode array
(2) Braun, S.; Rappoport, S.; Zusmen, R.; Avnir, D.; Ottolenghi, M. spectrophotometer in 1 cm path length cuvettes. Transmission
Mater. Lett. 1990, 10, 1.
(3) Böhmer, M. R.; Balkenende, A. R.; Bernards, T. N. M.; Peeters, electron microscopy (TEM) was performed on a Philips CM20
M. P. J.; van Bommel, M. J.; Boonekamp, E. P.; Verheijen, M. A.; Krings, microscope operating at 200 kV. Samples for TEM were prepared
L. H. M.; Vroon, Z. A. E. P. In Handbook of Advanced Electronic and by grinding a piece of gel (either wet or dry), suspending the
Photonic Devices; Nalwa, H. S., Ed.; Academic Press: San Diego, 2001. sample in ethanol by sonication, and depositing a drop of the
(4) Nogami, M.; Selvan, S. T.; Song, H. In Handbook of Advanced suspension on a carbon-coated TEM copper grid. Scanning
Electronic and Photonic Devices; Nalwa, H. S., Ed.; Academic Press: electron microscopy (SEM) was performed with a Philips XL-30
San Diego, 2001.
(5) Kozuka, H.; Sakka, S. Chem. Mater. 1993, 5, 222. microscope operating at 30 kV, equipped with a setup for element
(6) Gwak, J.-H.; Kim, S.-J.; Lee, M. J. Phys. Chem. B 1998, 102,
7699. (10) Liz-Marzán, L. M.; Giersig, M.; Mulvaney, P. Langmuir 1996,
(7) Armelao, L.; Bertoncello, R.; De Dominicis, M. Adv. Mater. 1997, 12, 4329.
9, 737. (11) Mulvaney, P. Langmuir 1996, 12, 788.
(8) Morris, C. A.; Anderson, M. L.; Stroud, R. M.; Merzbacher, C. I.; (12) Henglein, A. J. Phys. Chem. 1993, 97, 5457.
Rolison, D. R. Science 1999, 284, 622. (13) Correa-Duarte, M. A.; Kobayashi, Y.; Caruso, R. A.; Liz-Marzán,
(9) Anderson, M. L.; Morris, C. A.; Stroud, R. M.; Merzbacher, C. I.; L. M. J. Nanosci. Nanotechnol. 2001, 1, 95.
Rolison, D. R. Langmuir 1999, 15, 674. (14) Correa-Duarte, M. A.; Liz-Marzán, L. M. Unpublished results.

10.1021/la010736p CCC: $20.00 © 2001 American Chemical Society

Published on Web 09/08/2001
6376 Langmuir, Vol. 17, No. 20, 2001 Kobayashi et al.

analysis by X-ray energy dispersion (EDS) Microspec WDX 3PC.

Prior to the measurements, the samples were coated with a 2 nm
thick gold film by means of a vacuum sputterer, to improve
electrical conductivity. X-ray fluorescence was measured with a
Siemens SRS 3000 using a rhodium anode and an 8 mm mask.
Intensity was converted into concentration using suitable
standards.
Synthesis of Au@SiO2. Silica-coated Au nanoparticles
(Au@SiO2) were prepared by the method previously reported.10
Typically, a gold sol is prepared by boiling 5 × 10-4 M HAuCl4
in the presence of 1.6 × 10-3 M sodium citrate.15 This results in
a stable dispersion of gold particles with an average diameter of
around 15 nm and 10% polydispersity. The Au particle size was
varied through citrate concentration as described by Turkevich.15
A freshly prepared aqueous solution of APS (2.5 mL, 1 mM) is
added to 500 mL of the gold sol under vigorous magnetic stirring.
The mixture of APS and gold dispersion is allowed to stand for
15 min, and then 20 mL of a 0.54 wt % sodium silicate solution
at pH 10-11 (adjusted with a cation-exchange resin) is added,
again under vigorous magnetic stirring. The resulting dispersion
(pH ≈ 8.5) is then allowed to stand for 1 week, so that the active
silica polymerizes onto the primed gold particle surface. The Figure 1. Time evolution of the UV-visible spectra of a
silica shell thickness is then about 5-7 nm thick. When citrate-Au colloid upon addition of sodium silicate solution.
appropriate, the Au concentration was increased by centrifugation
at 3000 rpm for several hours and removal of the supernatant.
This also resulted in removal of excess sodium silicate from no visual effect at all and a minimum change in the
solution. spectrum, basically due to dilution. Aggregation of gold
Sol-Gel Transition. Two different methods were used to particles is not so apparent upon mixture of citrate-Au
prepare silica gels loaded with nanoparticles, based on the sol- with the TMOS-methanol mixture used for the TMOS
gel literature.16,17 The description of the methods follows, and method, but there is a clear red-shift of the plasmon band.
several aspects will be subsequently compared. Similar stabilizing effects were previously found with
Sodium Silicate Method. To 5 mL of the starting nanoparticle respect to electrolyte addition18 and temperature increase19
dispersion with the desired concentration, 0.15 mL of sodium and are due to the anomalously high stability of amorphous
silicate solution (27% w/w) was added under magnetic stirring. silica in water,16,20 thus further confirming that the
Once silicate addition was completed, the pH was adjusted to 7
by titration with HCl (2 M). The sample was then poured into
colloidal behavior of Au@SiO2 is governed by the silica
a cuvette for further characterization and quickly shielded. The shell. With respect to gel formation, this enhanced stability
sol-gel transition occurs over several hours, but the rate of gel is the key to obtaining homogeneous gels, even at elevated
formation (as well as turbidity) can be adjusted by careful choice nanoparticle loadings, so that the gel retains the optical
of silicate concentration and final pH. The parameters given properties of the starting colloid.
here were found to be optimal for minimum turbidity. B. Comparison of Optical Properties. In Figure 2,
TMOS Method. TMOS was chosen as a silica source because we compare the spectral evolution during sol-gel transi-
a faster sol-gel transition is normally achieved, in comparison tion for both methods. We see that in both cases there is
to other Si alkoxides with longer alkoxy chains. In a typical an increase of intensity, mainly at higher energies, which
preparation procedure, first 0.32 mL of TMOS was added to 1.33
is an indication of an increase in light scattering10 arising
mL of methanol. Next, 2 mL of the nanoparticle dispersion was
added to the mixture of TMOS and methanol, and then the final from the formation of mesopores within the gel structure.
mixture was poured into a cuvette which was shielded with a This is confirmed by TEM (Figure 3) and nitrogen
rubber sheet to avoid unwanted evaporation. After several hours, adsorption as was reported elsewhere.13 All these results
a rigid gel was obtained. The final molar ratio of TMOS, water, agree that the pore size is larger for gels formed by the
and methanol is 1:50:15, assuming that 2 mL of colloid contains sodium silicate method than for those from the TMOS
2 mL of water. method. This can be easily reasoned through the respective
formation mechanism: while in the sodium silicate method
Results and Discussion the silica polymerization arises from bonding of previously
A. Relevance of Silica Coating. The first important existing (branched) oligomers, in the TMOS method it
observation is the major stabilizing role played by the originates from silicic acid due to hydrolysis of TMOS.17
silica shell during sol-gel transition. Figure 1 shows the Additionally, no basic or acid catalyst is added in the
time evolution of UV-visible spectra of a citrate-stabilized procedure outlined above, so that the conditions for gel
gold colloid upon addition of 0.1 mL of sodium silicate formation are very mild, thus favoring uniform structures.
solution. The quick formation of a second band, which The actual catalyst for sol-gel transition must be residual
progressively red-shifts, while the initial band drops in citrate ions remaining in solution, which was tested by
intensity, clearly indicates the irreversible aggregation mixing the TMOS/methanol mixture with a dilute sodium
of the starting Au particles,11 long before gel formation. citrate solution. A clear, transparent gel was obtained
This is due to a sudden increase in both pH and ionic within a similar time scale.
strength. Visually, this is observed as a color change from The insets of Figure 2 show the time traces for the
red to blue. maximum position of the plasmon band for both methods.
A similar experiment performed on an Au@SiO2 colloid The trend is opposite: while in the sodium silicate method
formed from the same citrate-Au nanoparticles produces the change is very small (of the same order of the
spectrophotometer resolution) and toward lower wave-
(15) Turkevich, J.; Garton, G.; Stevenson, P. C. J. Colloid Sci., Suppl.
1 1954, 26. (18) Ung, T.; Liz-Marzán, L. M.; Mulvaney, P. Langmuir 1998, 14,
(16) Iler, R. K. The Chemistry of Silica; John Wiley & Sons: New 3740.
York, 1979. (19) Liz-Marzán, L. M.; Mulvaney, P. New J. Chem. 1998, 1285.
(17) Brinker, C. J.; Scherer, G. W. Sol-Gel Science; Academic Press: (20) Mulvaney, P.; Liz-Marzán, L. M.; Giersig, M.; Ung, T. J. Mater.
San Diego, 1990. Chem. 2000, 10, 1259.
Sol-Gel Processing of Gold Nanoparticles Langmuir, Vol. 17, No. 20, 2001 6377

Figure 2. Time evolution of the UV-visible spectra of Au@SiO2

colloids during the sol-gel transition for the sodium silicate
method (a) and the TMOS method (b). The insets show the time
traces for the maximum position of the plasmon band.

lengths, in the TMOS method there is a clear red-shift Figure 3. TEM micrographs of silica gels loaded with Au@SiO2
during the initial stages of the process. nanoparticles, prepared according to the sodium silicate method
(a) and the TMOS method (b). A larger porosity is clearly
Interpretation of these optical changes can be made on observed for the sodium silicate method.
the basis of Mie theory.11,21 From Mie theory, we know
that the position of the plasmon band is sensitive to
changes in the refractive index of the medium in which With dielectric constants of 2.13 for silica and 1.82 for
the particles are embedded, as methanol and the respective concentrations (assuming
that volumes are additive), we have for the sodium silicate
method φ ) 0.004 and eff ) 1.7815. For the TMOS method,
λmax2 ) λP2(∞ + 2m) (1) φ ) 0.044 and eff ) 1.8042. Thus, the “solvent” refractive
index basically does not change for the sodium silicate
where ∞ is the high-frequency value of the dielectric method, while there is a small increase for the TMOS
function (13.2 for gold), m is the dielectric constant of the method. Meanwhile, scattering is larger for the sodium
medium (assumed to be nonabsorbing), and λP is the silicate method than for the TMOS method, because of a
metal’s bulk plasma wavelength, given by larger pore size. This results in a very small blue-shift of
the plasmon band position for the sodium silicate method,
λP2 ) 4π2c2m0/Ne2 (2) while the refractive index effect dominates for the TMOS
method, yielding a global red-shift of the maximum.
Thus, either method can be chosen depending on what
where N is the concentration of free electrons on the is more convenient for each specific application (retaining
surface, m is the effective mass of conduction electrons, the plasmon band position or having minimum scattering).
and 0 is vacuum permittivity.
C. Influence of Au Particle Size. For bare Au
For sols, the solvent dielectric constant is that of water
nanoparticles, the precise plasmon band position is
(1.78), while for the gels there will be an average dielectric
dependent (among other parameters) on particle size and
constant given for the sodium silicate method by
shape, mainly due to a modification of the electron mean
free path on the surface.11 Therefore, using starting colloids
eff ) (H2O(1 - φ) + SiO2φ) (3) with different particle sizes should allow a fine-tuning of
the optical properties of the gels. Figure 4 shows the
where φ is the volume fraction of silica within the gel. For spectra of sols and (wet) gels (prepared according to the
the TMOS method, we have TMOS method) based on Au@SiO2 particles with core

eff ) (H2OφH2O + MeOHφMeOH + SiO2φSiO2) (4) (21) Bohren, C. F.; Huffman, D. F. Absorption and Scattering of Light
by Small Particles; Wiley: New York, 1983.
6378 Langmuir, Vol. 17, No. 20, 2001 Kobayashi et al.

Figure 5. UV-visible spectra of thin xerogels obtained after

drying silica gels loaded with Au@SiO2 nanoparticles of
increasing concentrations: 0.010 (a), 0.019 (b), 0.25 (c), and
0.60 (d) Au w/w%. Absorbance values were normalized to an
optical path length of 10 mm.

thin xerogels with increasing nanoparticle concentration

in Petri dishes because of the simplicity of this method,
though other techniques such as spin-coating, dip-coating,
or spraying can be used as well if reproducible thicknesses
must be achieved. The thickness of the resulting xerogels
was determined by SEM at several spots of the samples
and was always in the range of 0.2-0.3 microns.
The Au concentration was modulated by controlled
centrifugation of the starting Au@SiO2 colloid, and the
final concentration in the film was determined by X-ray
fluorescence. Figure 5 shows the UV-visible spectra of
four films with Au concentrations of 0.06, 0.14, 0.29, and
0.61 w/w%. The spectra were normalized to an optical
Figure 4. UV-visible spectra of sols and gels prepared path length of 10 mm. In every case, we find the same
according to the TMOS method with Au@SiO2 particles with plasmon band position, and no secondary maximum or
average core diameters of 15 (a), 22 (b), and 30 nm (c). shoulder shows up, even for the most concentrated sample,
in which the spectrophotometer sensitivity was insufficient
diameters of 15, 23, and 30 nm. TEM measurements for measuring a clean spectrum at wavelengths below 550
further confirmed that for the different particle sizes the nm. These results clearly show that nanoparticle con-
particle distribution within the gel is homogeneous and centration can be varied without loss of uniformity and
there is no sign of aggregation. In the spectra of the sols, with retention of the optical properties of single nano-
we found a red-shift of the plasmon band (larger than particles.
expected for this size range) when increasing particle size, Figure 6 shows representative TEM micrographs of the
as well as band broadening, which is most likely due to samples described in Figure 5. These micrographs further
deviations from the spherical shape. In the corresponding confirm a homogeneous nanoparticle distribution for the
gels, we measured a further red-shift from the maximum various concentrations, showing that a decrease of inter-
position in the sol, due to the refractive index increase as particle distance is the only effect that arises from
discussed above. This confirms that the properties of the increasing the concentration.
gels are dictated by those of the starting sol, with no
alterations due to aggregation or particle-particle inter- Conclusions
action. In this paper, we have demonstrated the viability of
D. Nanoparticle Concentration. As stated above, two sol-gel processes for the homogeneous incorporation
some of the applications of metal nanoparticle loaded silica of silica-coated gold nanoparticles within transparent silica
gels, such as optical filters or nonlinear optical materials, gels. Both UV-visible absorption spectra and transmis-
require a high nanoparticle concentration within the gel, sion electron microscopy show that there is no aggregation
while retaining the optical identity of the individual of the metal particles during the sol-gel transition, so
particles. This is one of the major deficiencies of most of that the optical properties of the starting colloid are fully
the current preparation methods, in which high concen- retained in the gel. Evidence is also shown of the potential
trations lead to aggregated particles and poor control of of these methods for the preparation of gels with various
particle size and monodispersity. nanoparticle morphologies as well as with very high
The difficulty for the optical characterization of highly concentrations, which can be useful for several applica-
concentrated gels is the very high absorbance, which tions. In particular, measurement of nonlinear optical
exceeds the detection limit of standard spectrophotometers properties is in progress and will be reported elsewhere.
(absorbance values above 3). To overcome this problem Additionally, since the silica coating technique can be
and for demonstration purposes, we decided to prepare applied to a large variety of nanoparticles, such as Ag, Co,
Sol-Gel Processing of Gold Nanoparticles Langmuir, Vol. 17, No. 20, 2001 6379

Figure 6. TEM micrographs for Au@SiO2-loaded xerogels with increasing Au concentration (see Figure 5 for details). The scale
bar is valid for all four micrographs.

Pt, Fe3O4, CdS, and so forth, gel materials doped with performed the X-ray fluorescence measurements. This
these nanocrystalline materials can be prepared. Work work has been supported by the Spanish Ministerio de
toward this generalization is also in progress. Educación y Cultura (Project No. PB98-1088) and Xunta
de Galicia (Project No. PGIDT00PXI30108PN). Y.K.
Acknowledgment. The authors are indebted to J. B. acknowledges the Spanish Ministerio de Educación y
Rodrı́guez of the CACTI from Vigo University for as- Cultura for a personal grant.
sistance with TEM and SEM measurements. J. Millos LA010736P