Optical Probes inside Photonic Crystals
model inspired by studies of atoms in
Fabry–Pérot cavities,3 in which the relative
change in radiative lifetime is assumed to
be equal to the total solid angle subtended
by the Bragg reflections divided by a 4ꢅ
solid angle. An advanced theoretical de-
scription has recently been made, wherein
the DOS was calculated numerically.22 For
silica or polystyrene colloidal crystals and
opals, a change in emission rate of less
than 5% was found, in good agreement
with the data of Figure 3. Apparently, a
large change in the fluorescence spectrum
can coincide with a minor change in fluo-
rescence lifetime.
To observe large modifications of
spontaneous-emission rates and even full
suppression of spontaneous emission, light
sources must be placed in a photonic crys-
tal that strongly interacts with light. One
way to produce such crystals is through
self-organization, resulting in inverse opals
or air-sphere crystals in materials with high
refractive indices (see the article by Colvin
in this issue). We have measured angle-
resolved emission spectra from a laser dye
(Nile blue) infiltrated in a titania inverse
opal.23 Figure 4a shows spectra for various
detection angles ꢆ from the normal to the
(111) surface plane.24 At ꢆ ꢀ 0ꢇ, the emis-
sion is greatly suppressed over a wide fre-
quency range by the first-order stop gap.
The bandwidth of the stop band is com-
parable to the broad spectral range of the
dye; hence, the spectrum is significantly
altered. The large relative bandwidth of
15% is a signature of a well-ordered pho-
tonic crystal that strongly interacts with
the light. It is apparent from Figure 4 that
with increasing ꢆ, the frequency range of
suppressed emission moves toward high
frequencies, as expected for simple Bragg
diffraction. For emission in excess of
ꢆ ꢀ 60ꢇ, the stop band no longer overlaps
the emission spectrum.
To obtain directional emission proper-
ties, the emission spectra are divided by
spectra that are not modified by Bragg dif-
fraction (see Figure 4b). At ꢆ ꢀ 0ꢇ, the stop
gap is readily apparent, and the emission
is attenuated by 75% near the center of the
gap. The attenuation is limited by a mecha-
nism akin to the one described earlier, in
agreement with the estimate (1–lB/l),
where lB is the Bragg attenuation length
and l is the mean free path for diffusion.
At low ꢆ, single, broad stop bands are re-
vealed in Figure 4b. Interestingly, at higher
angles there appears to be a transition to
a double stop band. The lower-frequency
stop bands are denoted by S1, and the
higher-frequency ones by S2. With increas-
ing ꢆ, the S2 stop band becomes more ap-
parent, while S1 decreases in amplitude.
Further analysis reveals that the frequen-
tion shows a broad emission spectrum in
the near-infrared when optically pumped
at ꢃ ꢀ 488 nm (see solid curve in Fig-
ure 5).28 The luminescence is attributed to
optical transitions at defects or grain
boundaries in the Si. In this way, the pho-
tonic crystal itself acts as a light source,
which makes it ideal for studies of the
modification of spontaneous emission in
these structures. Figure 5 (dashed curve)
also shows the luminescence spectrum
measured on a photonic crystal composed
of five layers of Si “logs” (i.e., 1.25 unit
cells). Clearly, the emission is suppressed
over a broad spectral range, consistent
with the calculated bandgap region (indi-
cated on top of the figure). To study the
effect on spontaneous emission of Er3ꢀ
probe ions, Er3ꢀ was implanted in both the
reference layer and the photonic crystal.
The Er3ꢀ peak at 1.54 ꢁm is clearly ob-
served in Figure 5 and is reduced in the
photonic crystal. More measurements are
required to further identify the influence
of edge effects and the local DOS on the
changes in spontaneous emission.
While two-dimensional photonic crystals
do not possess confinement in the third di-
mension, they have two great advantages:
they can be integrated with standard two-
dimensional integrated-optics technology,
and external probes (incident from the third
dimension) can be used to probe their
local DOS. Calculations show that strong
modifications in the local optical DOS can
be attained in these structures.13,29 We have
fabricated two-dimensional photonic crys-
tals based on a cubic array of Si pillars using
electron cyclotron resonance-etching tech-
niques (see Figure 4 in the introductory
article in this issue).30 Next, we have de-
veloped two different wet-chemical proc-
esses to coat the Si photonic crystals with
optical probe ions. In a first approach,
Si pillars were coated with a thin SiO2
film doped with an eosin dye. A modified
base-catalyzed sol-gel process based on
the decomposition of tetra-ethoxysilane
was used, leading to the growth of a
45-nm-thick dye-doped oxide film on the
pillars. The inset in Figure 6 shows a cross-
sectional image of the photonic crystal,
made using a fluorescence confocal opti-
cal microscope.31 It can be seen that the full
surface of the structure is coated with the
fluorescent layer. Next, we developed a
method to coat a photonic crystal with
luminescent Er3ꢀ ions.32 The crystal is
dipped in an ErCl3 solution and subse-
quently oxidized and annealed. Figure 6
shows photoluminescence spectra of Si
pillar structures with a relatively large
pitch (4 ꢁm, 8 ꢁm, and 16 ꢁm). The lumi-
nescence intensity increases with decreas-
ing pitch due to the larger surface area
Figure 4. (a) Normalized emission
spectra as a function of frequency for
Nile blue dye in titania inverse opal.
(b) Relative intensities, obtained from
the spectra in (a). Solid curves are for
ꢆ ꢀ 0ꢇ, dashed curves for ꢆ ꢀ 25ꢇ,
dotted curves for ꢆ ꢀ 45ꢇ, and
dash-dotted curves for ꢆ ꢀ 60ꢇ. The
S1 and S2 stop bands in (b) at ꢆ ꢀ 25ꢇ
are indicated by thin dashed lines,
and the centers of these stop bands
are indicated with arrows (dashed for
ꢆ ꢀ 25ꢇ, dotted for ꢆ ꢀ 45ꢇ). (From
Reference 24.)
cies of the S1 and S2 stop-band edges show
a so-called avoided crossing as a function
of ꢆ, which is the result of multiple Bragg
diffractions: the (200) Bloch mode mixes
with the (000) and (111) modes. The emis-
sion data agree well with reflectivity data
and with stop gaps calculated by the plane-
wave expansion method.26
Optical Probes in Photonic
Crystals Made Using Lithography
The titania-based inverted opal structure
described in the last section shows strong
interaction with light, but probably does
not possess a full bandgap. We have stud-
ied the photoluminescence from Si-based
three-dimensional photonic crystals based
on the “woodpile” structure made by Lin
et al.,27 which do possess a full bandgap.
A scanning electron microscopy (SEM)
image of the structure is shown in the
inset of Figure 5. It was found that the
polycrystalline silicon used in the fabrica-
644
MRS BULLETIN/AUGUST 2001