2132 Chem. Mater.

1999, 11, 2132-2140

Single-Crystal Colloidal Multilayers of Controlled

Thickness
P. Jiang, J. F. Bertone, K. S. Hwang, and V. L. Colvin*
Department of Chemistry and Center for Nanoscale Science and Technology, Rice University,
MS-60, 6100 Main Street, Houston, Texas 77005
Received February 5, 1999. Revised Manuscript Received April 19, 1999

Materials whose dielectric constant varies spatially with submicrometer periodicity exhibit
diffractive optical properties which are potentially valuable in a number of existing and
emerging applications. Here, such systems are fabricated by exploiting the spontaneous
crystallization of monodisperse silica spheres into close-packed arrays. By reliance on a
vertical deposition technique to pack the spherical colloids into close-packed silica-air arrays,
high quality samples can be prepared with thicknesses up to 50 µm. These samples are
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

planar and thus suitable for optical characterization. Scanning electron microscopy (SEM)
Downloaded via INDIAN INST OF TECH KANPUR on September 11, 2019 at 14:49:47 (UTC).

of these materials illustrates the close-packed ordering of the spherical colloids in planes
parallel to the substrate; cross-sectional SEM micrographs of the arrays as well as optical
methods are used to measure sample thickness and uniformity. Normal-incidence transmis-
sion spectra in the visible and near-infrared regions show distinct peaks due to diffraction
from the colloidal layers. While these basic optical characteristics are similar to thicker and
polycrystalline gravity-sedimented colloidal crystals, the systematic control over the number
of colloidal layers allows the effect of sample thickness on the optical spectrum to be studied
for the first time.

Introduction prepared arrays of alumina rods, for example, were

recently used as a planar waveguide device operating
Many novel properties can be engineered in periodic in the gigahertz region;4 alternating layers of AlGaAs
dielectric materials by controlling the symmetry and and AlAs deposited through chemical vapor provide a
length scale of a sample’s microstructure.1-6 This idea stacked dielectric cavity used in distributed feedback
has been used to engineer microwave optics,4 and lasers.17 While these methods for creating dielectric
recently these dielectric structures have been termed structures have found use in commercial products, the
“photonic band gap” materials. The exciting potential potential applications of photonic band gap materials
of these structured materials for optical applications has at optical wavelengths have been limited by the dif-
motivated efforts to shrink the length scale of the ficulty in fabricating three-dimensionally periodic ma-
periodicity to submicrometer sizes.7-16 Lithographically terials with nanometer-scale structures.
Recently, colloidal crystals have received renewed
* To whom correspondence should be addressed. E-mail: colvin@ attention as an avenue to achieving controlled dielectric
rice.edu. periodicity.18-21 Colloidal crystals are regular crystalline
(1) Petrov, E. P.; Bogomolov, V. N.; Kalosha, I. I.; Gaponenko, S.
V. Phys. Rev. Lett. 1998, 81, 77.
arrays of highly monodisperse spheres of dielectric
(2) Lin, S. Y.; Fleming, J. G.; Hetherington, D. L.; Smith, B. K.; materials such as silica or polymers.22 These materials
Biswas, R.; Ho, K. M.; Sigalas, M. M.; Zubrzycki, W.; Kurtz, S. R.; Bur, have been the subject of study for several decades due
J. Nature 1998, 394, 251.
(3) Gupta, S.; Tuttle, G.; Sigalas, M.; Ho, K. M. Appl. Phys. Lett. to their unusual optical23-28 and thermodynamic
1997, 71, 2412.
(4) Lin, S. Y.; Chow, E.; Hietala, V.; Villeneuve, P. R.; Joannopoulos, (16) Joannopoulos, J. D.; Villeneuve, P. R.; Fan, S. Nature 1997,
J. D. Science 1998, 282, 274. 386, 143.
(5) Holtz, J. H.; Asher, S. A. Nature 1997, 389, 829. (17) Tanaka, K.; Nakamura, T.; Takamatsu, W.; Yamanishi, M.;
(6) Lin, V. S. Y.; Motesharei, K.; Dancil, K. P. S.; Sailor, M. J.; Lee, Y.; Ishihara, T. Phys. Rev. Lett. 1995, 74, 3380.
Ghadiri, M. R. Science 1997, 278, 840. (18) Asher, S. A.; Holtz, J.; Weissman, J.; Pan, G. MRS Bull. 1998,
(7) Velev, O. D.; Jede, T. A.; Lobo, R. F.; Lenhoff, A. M. Nature 1997, October, 44.
389, 447. (19) Miguez, H.; Meseguer, F.; Lopez, C.; Blanco, A.; Moya, J. S.;
(8) Zakhidov, A. A.; Baughman, R. H.; Iqbal, Z.; Cui, C.; Khayrullin, Requena, J.; Mifsud, A.; Fornes, V. Adv. Mater. 1998, 10, 480.
I.; Dantas, S. O.; Marti, J.; Ralchenko, V. G. Science 1998, 282, 897. (20) Bogomolov, V. N.; Prokof′ev, A. V.; Shelykh, A. I. Phys. Solid
(9) Holland, B. T.; Blanford, C. F.; Stein, A. Science 1998, 281, 538. State 1998, 40, 594.
(10) Davis, S. A.; Burkett, S. L.; Mendelson, N. H.; Mann, S. Nature (21) Lidorikis, E.; Li., Q.; Soukoulis, C. M. Phys. Rev. E 1997, 55,
1997, 385, 420. 3613.
(11) Imhof, A.; Pine, D. J. Nature 1997, 389, 948. (22) Pieranski, P. Contemp. Phys. 1983, 24, 25.
(12) Yablonovitch, E. J. Opt. Soc. Am. B 1993, 10, 283. (23) Yoshiyama, T.; Sogami, I. Phys. Rev. Lett. 1984, 53, 2153.
(13) Wijnhoven, J. E. G.; Vos, W. L. Science 1998, 281, 802. (24) Sanders, J. V. Acta Crystallogr. 1968, A24, 427.
(14) Pan, G.; Kesavamoorthy, R.; Asher, S. A. J. Am. Chem. Soc. (25) Goodwin, J. W.; Ottewil, R. H.; Parentich, A. J. Phys. Chem.
1998, 120, 6525. 1980, 84, 1580.
(15) Joannopoulos, J. D.; Meade, R. D.; Winn, J. N. Photonic (26) Krieger, I. M.; O’Neill, F. M. J. Am. Chem. Soc. 1968, 90, 613.
Crystals: Molding the Flow of Light; Princeton University Press: (27) Hiltner, P. A.; Krieger, I. M. J. Phys. Chem. 1969, 73, 2386.
Princeton, NJ, 1995. (28) Monovoukas, Y.; Gast, A. P. Langmuir 1991, 7, 460.

10.1021/cm990080+ CCC: $18.00 © 1999 American Chemical Society

Published on Web 07/15/1999
Single-Crystal Colloidal Multilayers Chem. Mater., Vol. 11, No. 8, 1999 2133

properties.29-35 Colloidal crystals offer a starting point spectra can be collected over some regions.45 Nonethe-
for the fabrication of periodic dielectric materials created less, gravity-sedimented samples generally contain
through self-assembly rather than lithographic means. polycrystalline domains of unknown sizes.46,47 In addi-
Here, one has the potential to create samples with full tion, during the gravity sedimentation method, crystal
three-dimensional ordering, a feature important for formation can only occur at specific volume fractions of
engineering larger and more complete photonic band colloids.22 As a result, array growth is difficult to adjust,
gaps. Also, self-assembly methods can create thicker and thus film thickness is not easily controlled.
materials than those made using lithography. While the Other methods for ordering colloids into arrays have
role of sample thickness in the optical properties of been explored by investigators seeking to make dense
colloidal crystals has not been investigated, it is clear monolayer films of close-packed colloids for nonoptical
that control over this important parameter would be applications. Under the right conditions, colloid solu-
quite valuable in materials engineering. In addition, tions that are spun-coated onto surfaces provide high
self-assembly methods are not limited in their length quality monolayers.48-52 Convective self-assembly of
scale, and it is feasible to create materials that are wetting solutions is also a route for making ordered
active in the ultraviolet and even the soft X-ray range monolayers and structures of nanocrystals on a variety
using colloids with smaller diameters. Finally, as has of substrates.53 Another method applied to polymer
been demonstrated recently, colloidal crystals provide colloids uses the flow of solvent through micromachined
an ideal scaffold for the creation of both polymeric channels to create dense colloidal arrays of multiple
and inorganic samples with complex porous layers.37,38,54,55 Finally, recent work by Nagayama et
structures.7-9,13,36-40 Such samples may exhibit more al.56-65 has exploited capillary forces to drive the as-
complete photonic band gaps than silica-air crystals as sembly of larger (d > 1 µm) polystyrene colloids on flat
higher index contrasts are possible; in addition, they surfaces. This method, which is conceptually similar to
also may find applications in catalysis and separation the Langmuir-Blodgett methods for film deposition,
technologies. The success of these nonlithographic routes creates ordered monolayer films on nearly any vertical
to the formation of periodic dielectric structures requires surface. Under the appropriate solvent and colloid
the development of robust methods for creating high conditions Nagayama et al. showed that patches of bi-
quality, uniform, and mechanically stable colloidal and trilayer arrays could also be formed. All of these
crystals. methods have the advantage of producing well-ordered
A popular method for creating colloidal crystals is the films on planar surfaces quite rapidly; however, in most
gravity sedimentation of colloids from dispersions;19,22,32,41 instances the deposition techniques have been optimized
recent progress in the area has focused on strategies for the formation of monolayer films.
for strengthening the crystals and generating sample In this paper a self-assembly technique which relies
formats well suited for optical testing. In one instance, on capillary forces to organize colloids is used to
silica colloids were crystallized between parallel plates, fabricate colloidal crystal multilayers. These arrays are
and the solvent was then polymerized so as to trap their ordered over long (∼1 cm) length scales and can be
crystalline order.42-44 This stabilization comes at a cost, fabricated with thicknesses ranging from two layers to
however, as the matrix material has a refractive index several hundred layers. Moreover, the thickness of the
closely matched to that of the colloids, thus reducing resulting sample can be precisely dictated through
their diffractive properties. In other work, silica colloids
were crystallized in capillaries, and the solvent was (45) Vos, W. L.; Sprik, R.; van Blaaderen, A.; Imhof, A.; Lagendijk,
allowed to evaporate. The resulting silica-air structures A.; Wegdam, G. H. Phys. Rev. B 1996, 53, 16231.
(46) Clark, N. A.; Hurd, A. J.; Ackerson, B. J. Nature 1979, 281,
have strong diffractive properties, and transmission 57.
(47) Okubo, T. Langmuir 1994, 10, 1695.
(29) Hartl, W.; Klemp, R.; Versmold, H. Phase Transitions 1990, (48) Deckman, H. W.; Dunsmuir, J. H. Appl. Phys. Lett. 1982, 41,
21, 229. 377.
(30) Smits, C.; van Duijneveldt, J. S.; Dhont, J. K.; Lekkerkerker, (49) Winzer, M.; Kleiber, M.; Dix, N.; Wiesendanger, R. Appl. Phys.
H. N. W. Phase Transitions 1990, 21, 157. A 1996, 63, 617.
(31) Aastuen, D. J. W.; Clark, N. A.; Swindal, J. C.; Muzny, C. D. (50) Burmeister, F.; Schafle, C.; Keilhofer, B.; Bechinger, C.;
Phase Transitions 1990, 21, 139. Boneberg, J.; Leiderer, P. Adv. Mater. 1998, 10, 495.
(32) Davis, K. E.; Russel, W. B.; Glantschnig, W. J. J. Chem. Soc., (51) Deckman, H. W.; Dunsmuir, J. H. J. Vac. Sci. Technol. B 1983,
Faraday Trans. 1991, 87, 411. 1, 1109.
(33) Imhof, A.; van Blaaderen, A.; Maret, G.; Mellema, J.; Dhont, (52) Hulteen, J. C.; Van Duyne, R. P. J. Vac. Sci. Technol. 1995,
J. K. G. J. Chem. Phys. 1994, 100, 2170. 13, 1553.
(34) Kesavamoorthy, R.; Super, M. S.; Asher, S. A. J. Appl. Phys. (53) O’hara, P. C.; Leff, D. V.; Heath, J. R.; Gelbart, W. M. Phys.
1992, 71, 1116. Rev. Lett. 1995, 75, 3466.
(35) Rundquist, P. A.; Jagannathan, S.; Kesavamoorthy, R.; Brnardic, (54) Park, S. H.; Qin, D.; Xia, Y. Adv. Mater. 1998, 10, 1028.
C.; Xu, S.; Asher, S. A. J. Chem. Phys. 1991, 94, 711. (55) Park, S. H.; Xia, Y. Langmuir 1999, 15, 266.
(36) Velev, O. D.; Jede, T. A.; Lobo, R. F.; Lenhoff, A. M. Chem. (56) Denkov, N. D.; Velev, O. D.; Kralchevsky, P. A.; Ivanov, I. B.;
Mater. 1998, 10, 3597. Yoshimura, H.; Nagayama, K. Nature 1993, 361, 26.
(37) Park, S. H.; Xia, Y. Adv. Mater. 1998, 10, 1045. (57) Dimitrov, A. S.; Nagayama, K. Langmuir 1996, 12, 1303.
(38) Park, S. H.; Xia, Y. Chem. Mater. 1998, 10, 1745. (58) Nagayama, K. Phase Transitions 1993, 45, 185.
(39) Jiang, P.; Hwang, K. S.; Mittleman, D. M.; Bertone, J. F.; (59) Dimitrov, A. S.; Dushkin, C. D.; Yoshimura, H.; Nagayama,
Colvin, V. L. J. Am. Chem. Soc., submitted for publication. K. Langmuir 1994, 10, 432.
(40) Johnson, S. A.; Ollivier, P. J.; Mallouk, T. E. Science 1999, 283, (60) Dushkin, C. D.; Nagayama, K.; Miwa, T.; Kralchevsky, P. A.
963. Langmuir 1993, 9, 3695.
(41) Mayoral, R.; Requena, J.; Moya, J. S.; Lopez, C.; Cintas, A.; (61) Adachi, E.; Dimitrov, A. S.; Nagayama, K. Langmuir 1995, 11,
Miguez, H.; Meseguer, F.; Vazquez, L.; Holgado, M.; Blanco, A. Adv. 1057.
Mater. 1997, 9, 257. (62) Nagayama, K. Colloids Surf. A 1996, 109, 363.
(42) Jethmalani, J. M.; Ford, W. T. Chem. Mater. 1996, 8, 2138. (63) Dimitrov, A. S.; Nagayama, K. Chem. Phys. Lett. 1995, 243,
(43) Jethmalani, J. M.; Sunkara, H. B.; Ford, W. T.; Willoughby, 462.
S. L.; Ackerson, B. J. Langmuir 1997, 13, 2633. (64) Denkov, N. D.; Velev, O. D.; Kralchevsky, P. A.; Ivanov, I. B.;
(44) Jethmalani, J. M.; Ford, W. T.; Beaucage, G. Langmuir 1997, Yoshimura, H.; Nagayama, K. Langmuir 1992, 8, 3183.
13, 3338. (65) Yamaki, M.; Higo, J.; Nagayama, K. Langmuir 1995, 11, 2975.
2134 Chem. Mater., Vol. 11, No. 8, 1999 Jiang et al.

control of sphere size and concentration in solution. This For evaluation of the role of sphere size on film formation,
deposition behavior can be adequately simulated using six samples (215.2 (3.9%), 298.6 (3.8%), 324.1 (7.2%), 396.6
existing models of film formation developed for mono- (5.4%), 437.4 (4.2%), and 508.6 nm (4.3%)) were used. Numbers
in parentheses following the sphere diameters refer to the
layer samples. The resulting planar colloidal crystals relative standard deviation of the measured diameters. All the
are highly ordered, and evaporation of the solvent leaves samples were placed under the same crystallizing dish to
a silica-air array with good optical contrast and me- ensure identical ethanol evaporation rates. Samples with
chanical stability. The optical properties of these samples volume fractions of 0.1%, 0.5%, 1.0%, 1.5%, 2.0%, and 3.0%
are measured at normal incidence and their qualitative and the same particle size (298.6 nm (3.8%)) were used to
features and trends compared to gravity sedimented determine the effect of volume fraction on the film thickness.
Samples (215.2 (3.9%) and 279.2 nm (4.6%)) with volume
colloidal crystals.
fractions of 1% were used to determine the effect of solvent
evaporation rate on the film thickness. Duplicates of each
Experimental Methods sample were tested under three different evaporation condi-
Materials and Substrates. All solvents and chemicals are tions: open to ambient atmosphere, covered in a crystallizing
of reagent quality and filtered by 0.22 µm syringe filters dish, and covered in a crystallizing dish with a small amount
(Gelman) except for tetraethoxysilane (99%, Alfa) and 3-(tri- of ethanol (20 mL in a 50 mL beaker). One of the duplicates
methoxysilyl)propyl methacrylate (TPM, 98%, Aldrich) which was used to determine the evaporation rate by weighing the
are freshly vacuum distilled before use. The 200 proof ethanol vial on a digital analytical balance. The typical solvent
is obtained from Pharmaco Products, and 29.6% ammonium evaporation rate, given in units of volume per time, was found
hydroxide is purchased from Fisher. Ultrapure water (18.2 to be 9 × 10-4 cm3/min.
MΩcm-1) is used directly from a Milli-Q water system. Thicker films are fabricated by successive dip coatings. After
Microslides (75 × 25 × 1 mm, Fisher) are cut into two equal each single coating is deposited, the film is air-dried for at least
halves along the long sides and are used as substrates. Glass 24 h and dipped again into another purified silica alcosol with
scintillation vials (20 mL, Fisher) are used as experimental the same particle size. The coating and drying cycle can be
cells. repeated to systematically increase film thickness.
Instrumentation. Scanning electron microscopy is carried Characterization of Colloidal Crystal Films. Scanning
out on a software-controlled LEO 440 SEM designed specifi- electron microscopy (SEM) is used to determine the film
cally for electron-beam lithography. Transmission spectra are thickness. A thin layer of gold is sputtered onto the samples
obtained using an Ocean Optics ST2000 fiber optic UV-near- prior to imaging. To reveal an edge appropriate for cross-
IR spectrometer. A Coulter N4 Plus laser dynamic light sectional analysis, samples are scraped using a sharp razor
scattering (DLS) instrument is used to size the particles. A blade. By tilting the specimen 30-40°, it is possible to image
CrC-100 sputtering system is used to sputter a thin layer (3-4 this edge and quantify film thickness; 10 random locations
nm) of gold on samples before SEM analysis. from the center square centimeter of the sample are analyzed
Colloid Synthesis and Purification. Monodisperse SiO2 for thickness. Optical properties of the silica colloidal single-
nanospheres are synthesized following the Stober-Fink-Bohn crystal films are evaluated by measuring their transmission
method.66 Nanospheres with diameters ranging from 200 to spectra at normal incidence, using an Ocean Optics ST2000
700 nm and relative standard deviations smaller than 7% are fiber optic UV-near-IR spectrometer.
obtained through strict control of the reaction conditions.67 The
sizes and size distributions of these samples are obtained from Results
SEM and DLS measurements. Over 200 spheres are sized
using SEM in order to arrive at the reported diameters. Before Crystalline quality is among the most important
deposition, the silica alcosols are washed with 200 proof parameters in determining the performance of colloidal
ethanol by repeated centrifugation and ultrasonic dispersion crystals in optical applications. The formation of point
cycles in order to remove impurities, such as ammonia, water,
and unreacted tetraethoxysilane. Six cycles are usually per- defects or domains can have an enormous impact on the
formed. 3-(trimethoxysilyl)propyl methacrylate (TPM) coated diffraction properties. Figure 1A shows a typical SEM
silica is synthesized using a published method68 and purified image of a single crystal of colloidal silica with a 298.6
as above. All microscopy data presented in this work, however, nm diameter, grown using the methods described above.
are from particles without TPM coating. The solvent of This sample exhibits an ordered close-packed arrange-
deposition, rather than the coating surface, is the key param- ment of silica colloids over a sample area of 10 µm which
eter in controlling film deposition.
Experimental Apparatus and Procedures. Prior to use
extends over an even larger area; this can be illustrated
all microslides and scintillation vials are soaked in a chromic- by the two-dimensional Fourier transform of a low-
sulfuric acid cleaning solution overnight, rinsed well with magnification SEM image. Such data is shown in the
ultrapure water from a Milli-Q water system, and dried in a inset of Figure 1A. The sharp peaks confirm the pres-
stream of nitrogen. A clean microslide is then placed into 15 ence of long-range crystalline order, extending over the
mL of purified silica alcosol in a clean scintillation vial. The largest length scales (40 × 40 µm2) accessible in a single
vial is covered by a 1200 mL crystallizing dish to keep out
low-magnification image. Perfectly ordered films can be
external airflow and contamination. The entire apparatus is
placed on a vibration-free bench in a temperature-controlled made for a variety of sphere diameters. Parts A and B
laboratory (22 ( 1 °C). For the preparation of these multilayer of Figure 4 show SEM images of two crystals with
samples, no substrate withdrawl is necessary. The volume different particle sizes (206.4 and 437.4 nm, respec-
fractions of different samples are determined by drying 6 mL tively). As in Figure 1A, long-range order is confirmed
of purified silica alcosol in 80 °C vacuum oven overnight and by Fourier transform of low magnification images
then weighing the residual solid. 2.04 g/mL is used as the (insets to Figure 4).
density of silica,67 though its actual density may be somewhat
lower. Then, the samples are diluted or concentrated using a It is a challenge to use microscopy for proof of sample
centrifuge-redisperse cycle to the required volume fractions. order over centimeter distances. Lower magnification
images do not have the spatial resolution needed to
(66) Stober, W.; Fink, A.; Bohn, E. J. Colloid Interface Sci. 1968, identify distinct colloids; one approach is to compare
26, 62. micrographs of 40 µm2 areas collected as a sequence
(67) Bogush, G. H.; Tracy, M. A.; Zukoski IV, C. F. J. Non-Cryst.
Solids 1988, 104, 95. during the translation of the sample. If domain bound-
(68) Philipse, A. P.; Vrij, A. J. Colloid Interface Sci. 1989, 128, 121. aries are observed during this movement, then the
Single-Crystal Colloidal Multilayers Chem. Mater., Vol. 11, No. 8, 1999 2135

Figure 2. Fourier transforms (FFT) of four 11.75 × 11.75 µm2

regions of a 1 cm2 colloidal silica single crystal. These images
were part of a series taken every 50 µm over a 4 × 4 mm
square region; they represent the four corners and image c
was closest to the edge of the sample. The four parallel dashed
lines indicate one reference direction, and demonstrate that
the four images are roughly of the same orientation, as were
all intermediate images. Within a 4 × 4 mm2 region, the
largest deviation from this orientation is ∼1.6 degrees, rep-
Figure 1. (A) Typical scanning electron micrograph (SEM) resented in image c. This demonstrates that the crystal-
of a sample (top view) showing spheres of 298.6 nm diameter. lographic orientation of the sample is preserved over this entire
The inset shows a Fourier transform of a 40 × 40 µm2 region. region.
(B) Typical SEM side view of the same sample at the same
a narrow size distribution exhibits order over a long
magnification (×12 000), showing a perspective view of the
cleaved face and the underlying substrate. length scale and sharp peaks in the Fourier transform
of a low-magnification SEM image; the polydisperse
hexagonal registry of the images evident in their FFTs sample has only short-range order and exhibits rings
will not be aligned. Figure 2 shows such data for four in the Fourier transform. Films that lack crystalline
representative images taken at the edges of a 4 mm quality, such as the one shown in Figure 5B, even when
square. The alignment of the spots in all four images relatively thin, are white and opaque. This is in sharp
illustrates the registry of the array lattice over these contrast to the brilliant color and transparency apparent
long length scales. Morevoer, analysis of many inter- in more ordered films (Figure 3). Data collected from
mediate images also showed perfect alignment and no over 20 samples suggest that when this method is used,
evidence of any grain boundaries. colloids must have a size distribution with relative
While order within the plane parallel to the glass standard deviation less than 8% (on the diameter) in
substrate is important for the optical properties of the order to form high quality close-packed crystals and
films, especially when they are evaluated away from thus good optical samples.
normal incidence, the stacking of close-packed layers For this deposition process, two parameters allow for
perpendicular to the substrate is responsible for the control of film thickness (or number of colloidal layers):
diffraction observed in normal-incidence transmission particle diameter (d) and particle volume fraction (V.F.).
measurements. To evaluate this issue, colloid films can Figure 6 shows how film thickness varies with these
be abraded to reveal their stacking perpendicular to the two parameters. For these measurements layer thick-
substrate as shown in Figure 1B. This cross-section ness is determined using cross-sectional SEM images
shows that the close-packed structure extends uniformly of the types shown in Figures 1B, 7, and 8. Particle size
over the 15 layers shown; Figure 8B shows a thicker affects the film thickness in a systematic way: for the
50 layer sample which also exhibits a regular arrange- same volume fraction, a solution of larger particles
ment from the top surface to the bottom surface. The yields fewer layers. Two examples are shown in Figure
visual appearance of the samples (see Figure 3) is a 7, and the relationship is quantified in Figure 6A.
testament to this high crystalline quality and the Particle volume fraction also affects film thickness
uniform thickness of these films. (Figure 6B); more concentrated colloidal solutions leave
To form high quality planar crystals, particle size thicker multilayer deposits (Figure 8). Because film
distribution (S.D.) is a crucial factor. This is illustrated uniformity degrades significantly when colloidal solu-
in Figure 5, which shows SEM images of two samples tions of more than 4% volume fraction are used, the
with different size distributions (3.8% and 14.2% rela- formation of thicker films requires multiple coating. In
tive standard deviations, respectively). The sample with this approach, samples are allowed to dry and then are
2136 Chem. Mater., Vol. 11, No. 8, 1999 Jiang et al.

Figure 3. (A) Photograph of three samples with different

sphere diameters (307.2, 353.0, and 383.7 nm, from left to
right). The samples are clear and highly uniform. (B) Photo-
graph of the same samples illuminated with white light.
Figure 4. SEM images of two samples with different particle
diameters (×12 000 magnification). The insets show Fourier
reimmersed in the colloidal solutions; new colloidal transforms of 40 × 40 µm2 regions. Key: (A) d ) 206.4 nm;
multilayers form easily on top of the original ones and (B) d ) 437.4 nm.
the process can be repeated many times with the
thickness increasing linearly after each deposition. touching six others in one layer (Figures 1 and 4). This
Figure 9 shows a scanning electron micrograph of a close-packing arrangement is well-known in crystals
portion of a multicoated sample. comprised of colloids, such as these silica spheres, whose
The optical properties of these films can be evaluated ionic interactions are minimal.69,70 In this close-packed
using visible-near-IR transmission in normal incidence. geometry, whether the structure is face-centered cubic
Figure 10 shows scaled absorption spectra vs particle (ABCABC...), hexagonal close-packed (ABABAB...), or
size for a wide range of colloidal sphere diameters. Two randomly stacked, these images illustrate that the
notable features of the spectra are sharp peaks due to samples are oriented with their (111) axes parallel to
the Bragg diffraction of visible light from the ordered the substrate.
spheres and a background rising at shorter wave- Exact characterization of the crystal structure of these
lengths. The systematic change in color seen in trans- samples from SEM micrographs alone is problematic.
mission and reflection of uniform films of different Though images such as these confirm a regular close-
particle sizes (see Figure 3) is due to Bragg diffraction. packed arrangement (Figures 1B, 7, and 8), they cannot
Figure 11 shows the changes in the optical transmission be used to distinguish between the fcc and hcp struc-
spectra of films of different thickness made using equal tures. This is because it is difficult to determine the
sized particles. The background has been subtracted in exact angle of the cross-sectional view, which in turn
these data so that they can be compared easily. The makes it difficult to assess which crystal face is visual-
intensity of the diffracted peak increases approximately ized in these images. Confocal microscopy of gravity
linearly with the number of layers, although this is not sedimented colloidal crystals71 has indicated that the
evident in Figure 11 because the data have been scaled stacking between (111) planes is neither fcc nor hcp but
to uniform height. Also, the peak widths narrow as is instead a random arrangement. Theoretical calcula-
sample thickness is increased. Finally, Fabry-Perot tions, though, have indicated that the fcc structure is
fringes (indicated by the arrows in Figure 11) become stabilized slightly even with completely noninteracting
smaller and more closely spaced in thicker samples. spheres.72,73 In future studies, where the transmission

Discussion (69) Tsunekawa, S.; Barnakov, Y. A.; Poborchii, V. V.; Smoilovich,

S. M.; Kasuya, A.; Nishina, Y. Microporous Mater. 1996, 8, 275.
Crystal Structure of Planar Arrays. SEM images (70) Miguez, H.; Meseguer, F.; Lopez, C.; Mifsud, A.; Moya, J. S.;
Vazquez, L. Langmuir 1997, 13, 6009.
of the top view of these samples show that the colloids (71) van Blaaderen, A.; Ruel, R.; Wiltzius, P. Nature 1997, 385, 321.
are arranged in a close-packed fashion, with each sphere (72) Woodcock, L. V. Nature 1997, 385, 141.
Single-Crystal Colloidal Multilayers Chem. Mater., Vol. 11, No. 8, 1999 2137

Figure 7. SEM cross-sectional images of two samples with

Figure 5. SEM images of two samples with different particle different particle diameters, grown from solutions with the
size distributions (S.D.) (×12 000 magnification). The insets same volume fraction (0.5%). The 206.4 nm particle size
show Fourier transforms of 40 × 40 µm2 regions. (A) S.D. ) produced a film with 14 layers (A), and a 396.6 nm particle
3.8%; (B) S.D. ) 14.2%. size produced a film with 7 layers (B).
Domain Size of Planar Arrays. An important
feature of these samples is their long-range order within
the plane parallel to the substrate. In gravity-sedi-
mented colloidal crystals, there can be some preferential
alignment of the crystal axes with the capillary side-
walls; however, such samples are multicolored under
white light illumination74 and are thought to contain
colloidal crystal domains whose (111) axes are not
aligned. In contrast, the multilayer samples described
here are deposited onto a flat substrate which forces
their (111) axes to be aligned over distances of mil-
limeters (Figure 2). For this reason, when viewed at a
variety of angles, the color of the entire film is uniform
(Figure 3). Thus, unlike the gravity-sedimented colloidal
crystals which are polycrystalline in most instances,46,47
these samples consist of a single crystalline close-packed
array.75 We note that in this work when samples with
slightly broader size distributions (>8%) are deposited,
grain boundaries in the plane parallel to the substrate
are observed; similarly, Nagayama et al. observed grain
Figure 6. (A, top) Number of layers vs the inverse of the boundaries in monolayers of colloids with 10-15%
particle diameter, for samples grown from a 1% V.F. solution. distributions in size.57 This suggests that the monodis-
(B, bottom) Number of layers vs the particle volume fraction,
d ) 298.6 nm.
persity of the colloids, rather than the deposition process
itself, is responsible for their long-range ordering.
is investigated as a function of angle away from the Perfection of Colloidal Crystals. Though these
normal, the exact nature of the crystal structure will samples are single crystals, they are not perfect crystals
be a more important issue.26,27
(74) Gast, A. P.; Russel, W. B. Phys. Today 1998, 51 (12), 24.
(73) Bruce, A. D.; Wilding, N. B.; Ackland, G. J. Phys. Rev. Lett. (75) Parker, S. P. Dictionary of Scientific and Technical Terms, 4th
1997, 79, 3002. ed.; McGraw-Hill Book Company: New York, 1989.
2138 Chem. Mater., Vol. 11, No. 8, 1999 Jiang et al.

Figure 10. Normal incidence transmission spectra (optical

density) of samples with different sphere diameters, deter-
mined using SEM analysis. These curves have been scaled to
a constant height, and vertically offset for clarity. The arrows
indicate the expected positions of the peaks for each curve,
calculated using Bragg’s law at normal incidence and the
diameters shown. Typically at the peak maximum only 10-
20% of the incident light is transmitted.

Figure 8. SEM cross-sectional images of two samples grown

from solutions with different particle volume fractions. In both
parts A and B, the spheres are 298.6 nm in diameter. (A)
V.F. ) 0.8%, 18 layers; (B) V.F. ) 3.0%, ∼50 layers.

Figure 11. Normal incidence transmission spectra (optical

density) of three samples (d ) 279.2 nm) with increasing film
thickness. These curves have been scaled to a constant height,
and vertically offset for clarity. Thicker samples exhibit more
closely spaced (and smaller) Fabry-Perot fringes. The arrows
Figure 9. Typical SEM image of a sample (d ) 279.2 nm) indicate local maxima, λp, chosen for thickness analysis, with
deposited with four successive coatings. Each deposition the longest wavelength as p ) 0 (see supplemental Figures 1
provides 12 colloidal monolayers. and 2). In addition, a rising background apparent in the data
of Figure 10 has been subtracted from these data; for samples
as is clear from the SEM photographs. Two types of of more than 20 layers, this background contributes at most
defects are observed routinely in the SEM images of 20% of the rejection of light at the peak maximum.
these samples. The first is sphere vacancies that appear
every ∼10 µm on average. The second is vertical cracks
in the crystal, perpendicular to the glass substrate. high vacuum of the SEM damages their structure and
Examples may be seen in Figures 7A and 8A. Cracks induces cracking; to evaluate this issue, atomic force
are typically observed every 10 µm in most samples and microscopy (AFM) of the same materials was performed
unlike a classic grain boundary, the registry of the under ambient conditions on the as-prepared surfaces.
hexagonal packing is preserved across these features. Though cracks were still observed, they were fewer in
One explanation for the cracking is that the deposition number and typically found every 100 µm. The drying
of metal and the subsequent exposure of films to the tension in thin colloidal films is well-known to cause
Single-Crystal Colloidal Multilayers Chem. Mater., Vol. 11, No. 8, 1999 2139

cracking of much thinner deposits;76 it is likely that such solvent evaporation rate. This is because, while faster
drying-induced shrinkage is responsible for this problem solvent evaporation does lead to faster array formation,
as well. Strategies for minimizing this effect include the it also leads to increased solution influx into the array
addition of surfactants, and better control over the growth region. This increased influx of silica colloids
drying conditions. balances the faster growth rate, leading to a film
Controlling Film Thickness. One of the great thickness independent of evaporation rate. This predic-
advantages of this method of colloidal crystal formation tion was verified experimentally by creating multilayer
is its ability to create samples with precisely controlled samples under different evaporation conditions. While
thickness. By controlling such parameters as colloidal film uniformity was somewhat dependent on evapora-
concentration and sphere size, it is possible to vary the tion rate, film thickness was not. In addition, film
thickness of the films from a few layers to hundreds of thickness was insensitive to the surface derivatization
layers. The representative images in Figures 7 and 8 of the colloids.
suggest that film thickness, T, is related to the deposi- An important issue in this method is the role that the
tion parameters in a systematic way. Figure 6 il- solvent plays in controlling the deposition of colloidal
lustrates the quantitative relationships between the multilayers. The solvent enters into the model through
number of colloidal layers, k, and the deposition pa- the parameter L that is defined as the height of the
rameters φ (volume fraction of colloids in solution) and meniscus over which evaporation occurs. In the case of
d (colloid diameter). the ethanolic solutions used in this work, this parameter
These trends can be understood using a model devel- can be extracted from linear fits to the data shown in
oped by Nagayama et al.56-65 for the self-assembly of Figure 6. Assuming a β of 1, the data in Figure 6A
monolayer colloidal films. In his experiments, which are provides an L value of 316 µm while the data in Figure
reminiscent of Langmuir-Blodgett methods for self- 6B gives an L value of 301 µm. Similar evaporation
assembly, a vertical or slanted substrate is withdrawn lengths, on the order of 200 µm, were reported for the
from a solution of polystyrene colloids at a constant rate. growth of polystyrene colloidal monolayers from water.63
A meniscus region is formed on the substrate due to While this parameter cannot be exactly computed, an
wetting by the solution; evaporation of the solvent out upper limit for its value can be found by considering
of this thin meniscus leads to a constant solution influx, the meniscus height of ethanol on a flat glass surface.
which draws colloids into the area of film formation. Using the surface tension of ethanol and its contact
During the solvent evaporation these colloids experience angle for wetting glass, it is possible to calculate a
interparticle capillary forces which organize them into meniscus height of 3500 µm.78 As expected, the observed
close-packed arrays. The resulting thickness is then meniscus height of ∼310 µm is smaller than the value
entirely dependent on the flux of particles into the for a flat substrate as the film must wet a rough surface
meniscus, and mass balance can be used to derive a of colloids. This analysis suggests that solvents such as
simple formula for array thickness. water, which have large surface tensions and even lower
The film formation conditions used by Nagayama et contact angles, should deposit thicker films than the
al. were optimized for the production of uniform colloidal corresponding ethanolic solutions. Attempts to deposit
monolayers; however, it is possible to adapt this formal- silica colloids from aqueous solutions did lead to thicker
ism to the formation of multilayer films. To create thick films, but these samples had poor uniformity.
films, the array growth rate, which is a balance of the Measuring Multilayer Thickness Using Optical
solvent evaporation and substrate withdrawal rates, Techniques. Multilayer thickness is an important
must be as small as possible. Thus, the slowest possible parameter for quantitative assessment of sample optical
array growth results when the substrate is not me- properties. While scanning electron microscopy (SEM)
chanically withdrawn. Under this circumstance the provides a measure of this parameter, it is a destructive
array growth rate and the solvent evaporation rate are method as it requires the deposition of a thin gold
equal. This leads to a prediction that the final multilayer coating on the sample. A less time-consuming and
thickness depends on the deposition parameters accord- nondestructive method is an optical technique which
ing to77 uses the Fabry-Perot (FP) fringe79 positions to analyze
for film thickness. These fringes result from interference
βLφ between reflections from the top and bottom surfaces
k) (1)
0.605d(1 - φ) of the sample and are indicated by the arrows in Figure
11. An analysis, using Bragg’s law, of the spectral
where k is the number of layers, L is the meniscus positions of these local maxima permits an accurate
height, β is the ratio between the velocity of a particle determination of sample thickness (Supporting Figure
in solution and the fluid velocity and is taken to be 1, d 1). The thickness determined from this optical technique
is the particle diameter, and φ is the particle volume is in excellent agreement with thickness found from
fraction in solution. The data shown in Figure 6 clearly cross-sectional SEM (Supporting Figure 2).
demonstrate that the basic trends predicted by this Thickness Uniformity in Multilayer Samples. An
equation are valid. A surprising result from the model important feature of this deposition method is that it
is that the final array thickness is not dependent on the allows for the systematic control over multilayer thick-
ness (Figure 6). Indeed, the preparation of such well-
(76) Bergna, H. E. The Colloid Chemistry of Silica; Bergna, H. E.,
Ed.; American Chemical Society: Washington, DC, 1994; Vol. 234, p
376. (78) Hiemenz, P. C. Principles of Colloid and Surface Chemistry;
(77) This equation was derived from eq 7 from Nagayama et al.57 Marcel Dekker: New York, 1984; p 278.
by allowing the evaporation rate, je, to be equal to the array growth (79) Hecht, E. Optics, 2nd ed.; Addison-Wesley Publishing Com-
rate, vc. pany: Reading, MA, 1987.
2140 Chem. Mater., Vol. 11, No. 8, 1999 Jiang et al.

controlled materials now makes it possible to evaluate plete treatments of scattering from periodic dielectrics.
many thickness-dependent properties of these systems. Theories of the optical properties of photonic band gap
For this work to proceed, however, it is important to materials provide an avenue for this understanding; this
assess the uniformity of the film thickness over the issue has been explored more fully elsewhere.84
entire area of the sample. One way to characterize the An important feature of these spectra is the rising
thickness distribution is to rely on the statistical background that grows at shorter wavelengths; this
analysis of cross-sectional SEM images. Films are causes the samples to look slightly hazy and could place
typically abraded along a diagonal, and it is possible to limits on their use in many optical applications. This
collect cross-sectional data at many points over the background has been observed, but rarely discussed, in
entire central region (∼5 mm) of the sample. The gravity-sedimented crystals as well, and its origins
standard deviation found for thickness in these mea- remain obscure. While it does not clearly fit the λ-4
surements is on the order of 10%. A complementary dependence expected for Rayleigh scattering, defects or
method for analyzing film thickness uniformity is to rely sphere surface roughness could give rise to this phe-
on the optical techniques to assess film thickness nomenon. Alternatively, it may represent the collective
(Supporting Figures 1 and 2). Simulations of the inter- effect of diffraction off of all of the lattice planes other
ference effect indicate that if the thickness of the sample than the (111) planes.85 If so, it is an intrinsic property
in the optical path varies by more than (10%, then well- of any periodic array of spheres, not merely of self-
defined maxima and minima will not be observed. Thus, assembled colloidal crystals.
the observation of well-defined fringes (Figure 11) An interesting feature of the optical spectra is the
demonstrates that film thickness deviates less than 10% increase in peak width as sample thickness is decreased
on average. (Figure 11). This effect is similar to the Debye-Scherrer
The Optical Properties of Colloidal Crystal Mul- broadening of X-ray diffraction peaks in small crystal-
tilayers. The basic features of the normal incidence lites;86 however, the widths reach a limiting value of
optical spectrum of these colloidal crystal multilayers only ∼40 nm in thick samples. A full treatment of the
are in good agreement with those observed for gravity- thickness-dependent optical effects of these samples is
sedimented samples. Figure 10 shows how the peak in the subject of a forthcoming publication.87
the optical spectum depends on sphere diameter. When
Conclusion
the wavelength of incident light satisfies the Bragg
condition, it is diffracted away from the propagation Using a simple method, it is possible to deposit planar
axis, leading to a decrease in the transmission and thus single crystals of nanosphere arrays on a wide variety
a peak in the absorption spectrum. The position of this of substrates. Scanning electron microscopy (SEM) of
peak can be related to the sphere diameter and the the samples shows a close-packed structure and single-
effective refractive index of the medium using λmax ) crystal quality throughout the film. The number of
2neffd111, where d111 is the interlayer spacing.80 The nanospheres deposited can be widely varied with this
arrows that are shown in Figure 10 are peak positions method; film thicknesses ranging from single monolay-
predicted from this equation using sphere sizes as ers to 100 layers are easily created through control over
determined from SEM; agreement with experimental the particle volume fraction and sphere size. The planar
data is quite good. format and uniformity of these materials make them
While this simple diffraction theory can predict the ideally suited for optical characterization. The qualita-
peak positions, it is unable to provide a quantitative tive features of their spectra are in good agreement with
explanation for any of the other features in the spectra. simple models of diffraction.
Recently, several groups have attempted to adapt the Acknowledgment. The authors would like to ac-
dynamical theory of X-ray diffraction to this knowledge funding from the NSF (CHE-9702520) and
purpose.42,43,45,81-83 However, this theory is based on the the Welch Foundation (C-1342). V.L.C. would like to
approximation that the refractive index contrast be- thank D. M. Mittleman for many useful discussions
tween the scattering centers and the interstitial medium concerning the optical properties of these materials.
is quite small, on the order of 10-4, an approximation
not valid for these silica-air crystals. For this reason, Supporting Information Available: Figures showing an
a detailed analysis of the shapes and intensities of the analysis, using Bragg’s law, of sample thickness (Supporting
optical transmission function must rely on more com- Figure 1) and correlation of SEM with Bragg’s Law thickness
(Supporting Figure 2). This material is available free of charge
via the Internet at http://pubs.acs.org.
(80) For close-packed spheres, the interlayer spacing is related to
the sphere diameter, D, by d111 ) (2/3)1/2D. CM990080+
(81) Monovoukas, Y.; Fuller, G. G.; Gast, A. P. J. Chem. Phys. 1990,
93, 8294. (84) Mittleman, D. M.; Bertone, J. F.; Jiang, P.; Hwang, K. S.;
(82) Rundquist, P. A.; Photinos, P.; Jagannathan, S.; Asher, S. A. Colvin, V. L. J. Chem. Phys. 1999, 111, 345.
J. Chem. Phys. 1989, 91, 4932. (85) Liu, L.; Li, P.; Asher, S. A. J. Am. Chem. Soc. 1997, 119, 2729.
(83) Sunkara, H. B.; Jethmalani, J. M.; Ford, W. T. Hybrid (86) Cullity, B. D. Elements of X-ray Diffraction, 2nd ed.; Addison-
Organic-Inorganic Composites; Mark, J. E., Lee, C. Y.-C., Bianconi, Wesley Publishing Company, Inc.: Reading, MA, 1978.
P. A., Ed.; American Chemical Society: Washington, DC, 1995; Vol. (87) Bertone, J. F.; Jiang, P.; Hwang, K. S.; Mittleman, D. M.;
585, p 181. Colvin, V. L. Phys. Rev. Lett., in press.