Far-field emission profiles from L3 photonic crystal cavity modes

Cristian Bonato,1 Jenna Hagemeier,2 Dario Gerace,3 Susanna M. Thon,2, ∗ Hyochul Kim,2, †
Lucio C. Andreani,3 Pierre Petroff,2 Martin P. van Exter,1 and Dirk Bouwmeester1, 2
1
Huygens Laboratory, Leiden University, P.O. Box 9504, 2300 RA Leiden, the Netherlands
2
University of California Santa Barbara, Santa Barbara, California 93106, USA
3
Dipartimento di Fisica ”Alessandro Volta”, Universitá di Pavia, via Bassi 6, 27100 Pavia, Italy
We experimentally characterize the spatial far-field emission profiles for the two lowest confined
arXiv:1208.0430v1 [cond-mat.mes-hall] 2 Aug 2012

modes of a photonic crystal cavity of the L3 type, finding a good agreement with FDTD simula-
tions. We then link the far-field profiles to relevant features of the cavity mode near-fields, using
a simple Fabry-Perot resonator model. The effect of disorder on far-field cavity profiles is clarified
through comparison between experiments and simulations. These results can be useful for emission
engineering from active centers embedded in the cavity.

PACS numbers:

I. INTRODUCTION thoroughly investigated by Chalcraft and coworkers [8]

who compared the calculated resonant energies, quality
factors and emission polarizations for the lowest-order
Photonic crystals (PhC) offer unprecedented control
modes with experimental data. Most experiments cou-
over electromagnetic field confinement in all three spa-
pling single quantum dot emitters to a nanocavity exploit
tial directions [1]. In particular, two-dimensional PhC
the fundamental (i.e. lowest-energy) cavity mode [9, 10].
nanocavities in a planar waveguide have already found
However, higher order modes can still be important for,
applications in different fields such as nanolasers, non-
e.g., efficient pumping in nanocavity lasers [11], selective
linear optics and quantum information processing [2, 3].
excitation of quantum dots embedded within the cavity
Similar to any electromagnetic resonator, PhC nanocav-
[12, 13], or mutually coupling quantum dots in different
ity modes are essentially characterized by two figures of
spatial positions [14]. Several groups have studied the
merit: the cavity quality factor, Q, and the effective con-
near-field emission profiles of photonic crystal nanocav-
finement volume of each mode, Vmode [4]. The quality
ities [15, 16], even with polarization-resolving imaging
factor is proportional to the photon lifetime in the cavity
[17].
which depends on the cavity losses to the external world.
In this paper we report an experimental and theoretical
The mode volume is a quantitative measure of the spa-
investigation of the spatial far-field profile of the out-of-
tial confinement of the electromagnetic mode. In most
plane emission for the two lowest-order modes of L3-type
applications, it is crucial to maximize the Q/Vmode ratio.
PhC nanocavities. We believe the characterization of the
For example, the Purcell factor, which measures the en-
out-of-plane far-field emission for PhCs is important for
hancement of the spontaneous emission rates for atoms
two main reasons. First, for single-photon source appli-
resonant with a cavity is directly proportional to this fig-
cations the emitted radiation needs to be efficiently col-
ure of merit. In PhC nanocavities the mode is strongly
lected into a fiber, and simultaneous optimization of far-
confined to a very small volume, on the order of (λ/n)3 ,
field emission for multiple nanocavity resonances could
where λ is the mode wavelength. In a planar membrane
be useful. In addition, in the case of cavity-QED exper-
nanocavity, in-plane confinement is provided by spatial
iments in the “one-dimensional atom” approximation, a
localization of a structural defect in a perfectly periodic
perfect mode-matching is needed to get a large enough in-
PhC with a photonic band-gap, while out-of-plane con-
terference between the input light field and the field radi-
finement is given by total internal reflection between the
ated by the atom [18–20]. Recently, quite some work has
slab and the air cladding (assuming a suspended mem-
been done to get a beam-like vertical emission from PhC
brane as a planar waveguide). Very high quality factors,
nanocavities [21–24] for the fundamental mode. Here we
in the range 104 -106 [5–7] have been demonstrated in the
extend previous work by experimentally analyzing the
literature. In particular, the L3-type cavity, consisting of
far-field emission properties of both the fundamental and
three missing holes in a triangular lattice, was the first
the second-order mode, finding good agreement with nu-
PhC cavity to show quality factors larger than 104 [5].
merical simulations. We introduce a simple model, based
The spectral mode structure for L3 cavities has been on a one-dimensional Fabry-Perot resonator, to estimate
the essential far-field characteristics of a given near-field
mode profile and link them to the relevant device param-
∗ Current Address: Department of Electrical and Computer Engi-
eters. We believe that such a model can be useful for
fast parameter optimization, while full-scale numerical
neering, University of Toronto, 10 King’s College Road, Toronto,
Ontario M5S 3G4, Canada simulations can provide an accurate but time-consuming
† Current Address: Department of ECE, IREAP, University of description of the electromagnetic field in the structure.
Maryland, College Park, Maryland 20742, USA Finally, we will discuss the effect of fabrication imper-
 2

fections on the far-field cavity emission profiles. As we

will show, measurements of far-field profiles are relatively
easy to perform and they can provide insightful informa-
tion about the parameters and the quality of the cavities
under examination.
The paper is organized as follows: in Section II, we
present experimental measurements of the far-field pro-
files and a comparison with theoretical far-fields ex-
tracted from finite-difference-time-domain (FDTD) sim-
ulations; in Section III, we introduce a simple model,
based on a Fabry-Perot resonator, which is sufficient to
give indications of what the actual far-field profile looks
like for a given near-field and to link far-field properties
to actual device parameters.

II. THEORETICAL MODELING AND

EXPERIMENTAL DATA

Our sample consists of a 180 nm GaAs membrane

grown by molecular beam epitaxy on top of a 0.92 µm
Al0.7 Ga0.3 As sacrificial layer on a GaAs substrate. An
In0.4 Ga0.6 As quantum dot layer is grown at the center
of the GaAs membrane by depositing 10 periods of 0.55
Å-thick InAs and 1.2 Å-thick In0.13 Ga0.87 As. The L3 FIG. 1: Top: sketch of the L3 photonic crystal cavity used in
PhC cavities were fabricated on the sample using stan- the experiments. Given a design radius R for the holes, the
dard electron beam lithography and reactive ion etching two holes closest to and in-line with the cavity have reduced
techniques [25, 26]. The lattice constant of the triangu- design radii R′ = 0.75R [27]. The small holes are shifted
lar hole lattice is a = 240 nm. The L3 cavity design was from their lattice positions out from the center of the cavity
properly modified for Q optimization (see modified holes by a fixed quantity (55 nm). Finally, the third holes (i.e. two
in Fig. 1) [27, 28]. holes away from the small holes) are shifted from their lat-
The sample was placed in a He-flow cryostat at about tice positions out from the center of the cavity by 45 nm [28],
to further increase the theoretical Q. Bottom: experimen-
5 K and illuminated above the GaAs bandgap with a few
tally measured photoluminescence spectra for the two cavity
mW laser beam (wavelength 780 nm) on a few µm2 spot. modes.
The photoluminescence from the quantum dot layer em-
bedded in the membrane was collected in the direction
normal to the membrane using a microscope objective both modes results in a strongly linearly polarized sig-
with numerical aperture N A = 0.8 and spectrally ana- nal. However, as it is shown in the 3D FDTD simula-
lyzed with a spectrometer (resolution 5.5 GHz/pixel). An tions of Fig. 2, each near-field mode profile has x- and
example of the spectral emission is shown in Fig. 1 for y-components of the electric field of comparable inten-
a device with R = 54 nm. According to theoretical pre- sity. The reason for the detection of linearly polarized
dictions based on a guided-mode expansion method [29], light can be found by calculating the far-field projections
the cavity supports two confined modes, with resonances of such polarization-resolved near-field profiles.
(th)
respectively at λ1 = 987 nm (theoretical Q-factor The far-field profile can be obtained from the near-field
(th)
Qth ∼ 180000) and λ2 = 957 nm (Qth ∼ 15000). Ex- using the procedure introduced by Vuckovic and cowork-
perimentally, we measured the first-order mode with a ers [32]. According to the surface equivalence theorem,
Lorentzian profile centered around λ1 = 982.5 nm with all the information about the far-field profile can be ob-
a full-width at half-maximum (FWHM) of 0.195 ± 0.002 tained from equivalent electric and magnetic currents,
nm, from which we extract an experimental quality factor Js = n × H and Ms = −n × E, which depend on the
of Q ∼ 5000. On the other hand, the second-order mode in-plane near field components:
has a less perfect Lorentzian lineshape centered around Nx = −F T2 (Hy ) Ny = F T2 (Hx )
λ2 = 956.4 nm with FWHM 0.63 ± 0.03 nm, and an (1)
Lx = F T2 (Ey ) Ly = −F T2 (Ex )
experimental Q-factor Q ∼ 1500. Experimental quality
factors are lower then the predicted ones due to scatter- where F T2 denoted the two-dimensional Fourier trans-
ing from fabrication imperfections [30] and possible ab- form. These equivalent currents are used to calculate
sorption from sub-bandgap trap levels and surface states the retarded vector potential of the electromagnetic field,
[31]. which in the far-field can be related to Fourier transforms
Experimentally, the emitted radiation collected from of the near-fields. The radiation intensity per unit solid
 3

angle can be calculated as:

 2  2 !
η  L ϕ   L θ 
K(θ, ϕ) = 2 Nθ +  + Nϕ −  (2)
8λ η   η 
p
where η = µ0 /ε0 is the impedance of free-space and λ
is the mode wavelength. The radiation vectors in spher-
ical coordinates can be expressed from their cartesian
components as:
Nθ = (Nx cos ϕ + Ny sin ϕ) cos θ
(3)
Nϕ = −Nx sin ϕ + Ny cos ϕ ,
and similarly for L. The far-field profiles calculated from
the near-fields in the left panels of Fig. 2 are shown in
the same figure, on the right. In these plots, the color
scale is normalized to the totally emitted power in the
upper half-space of the PhC cavity (the same normaliza-
tion factor is used for Ex and Ey ). Most of the emission
from the x-polarization of both modes is predicted at
very large angles,and therefore is inefficiently collected
by commonly employed microscope objectives. This re-
sults in the strong linear polarization observed in the
photoluminescence spectra.
To perform a direct measurement of the far-field emit-
ted intensity at each resonant mode frequency, the fil-
tered photoluminescence at the back focal plane of the
microscope objective is imaged. Given a characteristic
size of the near-field emission, p ∼ 500 nm, at a wave-
length λ ∼ 1 µm the Fresnel number is F = p2 /(Lλ) ∼
0.01, well in the far-field regime (L ∼ 2 mm). The far- FIG. 2: FDTD simulations of polarization resolved near-field
field was imaged on an intensified CCD camera by a lens patterns (left panels) and far-field emission profiles (right pan-
with focal length 40 cm in a 2f − 2f configuration. To els) for the first and second modes of an L3 cavity. Each mode
make sure that we were looking at the microscope ob- has two linearly-polarized components. The white concentric
jective back focal plane, we adjusted the lens to see the rings correspond to a grid with numerical-aperture step of 0.2.
sharp image of the objective edge on the CCD. This sharp For far-field plots, the color scale is normalized to the totally
edge was used to calibrate the numerical aperture scale emitted power in the upper half-space of the PhC cavity (the
of the far-field images, assuming that the sharp edges same normalization factor is used for Ex and Ey ). For the x-
polarization, most of the far-field emission is at large angles,
correspond to the N A = 0.8 of the objective employed.
making collection less efficient than for y-polarization and re-
An interference filter, with a bandwidth 1 nm, was used sulting in a strongly polarized photoluminescence signal.
to spectrally select the mode of interest. Images were
collected after integrating for 30 s and the background
noise was removed by subtracting an image taken with a
slightly tilted interference filter. Finer details also appear inside the two lobes, in the
The experimental far-field spatial emission profiles form of spots separated by ∆k/k0 ∼ 0.1 − 0.2. Such
for the first-order and second-order modes are shown structures correspond, in the near-field, to light sources
in the two larger plots on the right side of Fig. 3, which are separated from the cavity about 5 − 10 times
together with the far-field projections obtained from the characteristic size of the cavity mode. The simu-
FDTD simulations (smaller insets on the left) for a lated near-field profiles shown in Fig. 2 suggest that the
direct comparison. The first-order mode exhibits a optical field decays very fast out of the cavity region,
centrally illuminated area extending to about N A ∼ 0.5, implying that such features might be due to light that
with a ring-like structure inside (N A ∼ 0.2), matching escapes from the cavity due to fabrication imperfections
the low-NA portion of the simulated far-field. The [30]. The fine details are reproducible for different mea-
simulated far-field suggests that most of the light from surements performed on the same device. Finally, we
the first-order mode is emitted in two high-NA lobes, show in Fig. 4 the far-field profiles of the second-order
which are not collected at all by our set-up. The far-field modes for different devices on the same wafer. Each plot
profile for the second-order mode consists of two lobes, shows the characteristic two-lobes profile, as expected for
whose center is at a minimal N A ∼ 0.3. this mode, with reproducible finer details that appear to
be device-dependent.
 4

FIG. 4: Experimental far-field profiles of the second-order

mode emission from four different devices on the same wafer.
Finer details are reproducible in different measurements per-
formed on the same device, but differ from device to de-
vice. The white concentric rings correspond to a grid with
numerical-aperture step of 0.2.

FIG. 3: Experimental far-field emission profiles for the two

L3 cavity modes (larger panels) and the corresponding pre-
dictions by FDTD simulations (smaller panels on the top left
of the experimental plots), resulting from the sum of x- and y-
components in Fig. 2. The white concentric rings correspond
to a grid with numerical-aperture step of 0.2. The color scale
bars show the detected photon counts per second. The counts
for the two profiles cannot be compared since they depend
not only on the device parameters, but also on the density of
quantum dots emitting in the wavelength range of the specific
cavity mode.
FIG. 5: Far-field intensity profiles for varying degrees of dis-
order, normalized by the total power radiated into the upper
half space. The images are cut-off at a numerical aperture
To investigate the cause of fine structure within the value of 0.8, indicated by the white line, to allow for better
comparison with experimental data. Data are shown for dis-
two-lobe far-field pattern of the second order mode,
order parameters of σ = 1 nm (left column), 2 nm (middle
FDTD simulations were done in which disorder was in- column), and 5 nm (right column).
troduced. It is assumed that, due to fabrication imper-
fections, the dominant type of disorder is in the hole radii
of the PhC lattice. Therefore, the hole radii R were
varied randomly according to the distribution function
P (R) ∝ exp[−(R − R̄)2 /2σ 2 )]. Some simulation data are ter σ. These results indicate that far-field measurements
given in Figure 5, showing disorder introduced to the far- could be used as an indicator of disorder in the lattice
field profiles as a result of increasing the disorder parame- structure.
 5

III. A SIMPLE FABRY-PEROT MODEL

In general, 3D FDTD simulations can provide accurate

modeling of near-field and far-field properties of PhC cav-
ity modes. However, they give little physical insight on
how the detected features of such far-field profiles can
be related to specific device parameters. In this Section,
we will show that the experimental data can be repro-
duced by a simple model, elaborated from the proposal
of Sauvan and coworkers [33].
For a line of N ≥ 3 missing holes, the PhC nanocav-
ity can be described quite accurately by a Fabry-Perot
resonator, in which the fundamental Bloch mode of a
single-line-defect PhC waveguide is trapped between two
PhC mirrors of modal reflectivity r(λ). The properties
of such a cavity are shown to depend only on three pa-
rameters, namely the group index ng of the Bloch mode,
the reflection coefficient r(λ) of the mirrors and the effec-
tive cavity length L. The Bloch mode can be calculated
as the eigenstate of the PhC waveguide in the Fourier
basis, and its modal reflectivity can be obtained with
the method described in Ref. [34]. Fabry-Perot mod-
els have been shown to be a useful tool to probe cav-
ity resonances and the group index of photonic crystal
waveguides [35, 36], and to describe acousto-mechanical
cavity tuning effects [37]. We consider the Ey near-field
profiles shown in Fig. 2. Taking the intensity distribu-
tion along the y = 0 axis, the modes show a sinusoidal
intensity distribution with nodes and antinodes in the
cavity region, decaying exponentially outside the cavity
region. From simulations, the intensity decay length for
the first-order mode is δ1 ∼ 220 ± 10 nm, while it is
δ2 ∼ 155 ± 8 nm for the second-order mode. This is
very similar to the intensity distribution in a cavity be-
tween two Distributed Bragg Reflectors (DBR). For the
Ey near-field profiles most of the radiation is emitted
along the y = 0 axis, with little structure and weaker
intensity outside. Given the relevance of such polariza-
tion in determining the far-field emission properties, to FIG. 6: Far-field results for the one-dimensional Fabry-Perot
model (first and second-order modes). For each mode, on the
a first-order approximation it makes sense to consider
left side the expected near-field profile considering a Fabry-
a one-dimensional model, which takes into account only Perot cavity with DBR mirrors with varying penetration
the structure along the y = 0 axis. Such a model has the depth δ. On the right side, the corresponding far-field profiles
advantage of being extremely simple, although able to calculated from Eq. 4. For the first-order mode, in the region
give significant hints on the main far-field profile prop- |kx |/k0 < 0.5, there is a single peak centered at kx /k0 = 0 for
erties. A two-dimensional model based on an effective small penetration depth. This peak flattens and broadens on
index approximation should be used (with no significant increasing penetration depth. For large enough penetration
difficulties) to study also the Ex profiles. depth, the central peak splits into two smaller peaks. The
Let us consider a one-dimensional Fabry-Perot res- second-order mode, in the region |kx |/k0 < 0.5, consists of
two peaks that split further and further out for increasing
onator, with the field intensity profile varying sinu-
penetration depth δ.
soidally in the cavity region and exponentially decay-
ing in the mirror regions, with characteristic penetra-
tion depth δ. The resonant frequency Ωm can be calcu-
lated by imposing the total phase acquired by the Bloch
mode traveling back and forth to be a multiple of π.
The far-field profile can be calculated using the proce- Eq. 2. Let’s consider a separable electric field distribu-
dure outlined in the previous Section (Eq. 2). Since tion Ey (x, y) = α(x)β(y). The component Lx is sep-
the modes are TE-like modes, only Ex , Ey and Hz are arable as well: Lx = α̃(kx )β̃(ky ), where α̃ and β̃ are
non-negligible and only the Lx component is relevant in the one-dimensional Fourier transforms of α(x) and β(y),
 6

the first and second-order modes for different values of

the penetration depth, δ. The first-order mode exhibits
a central peak (centered at k = 0) and two outer lobes
as predicted by the FDTD simulations. We see that, for
increasing penetration depth, the two main outer lobes
become narrower in k-space and more outward, while the
central peak broadens and flattens. A similar behavior
can be observed in the second-order mode. Here, the
far-field profile is given by two central peaks and two
outer lobes, and again, for increasing penetration depth
δ the two outer lobes move outwards and become more
localized.
The Fabry-Perot resonator model was shown to give
results for the quality factor Q in agreement with more
sophisticated FDTD simulations [33]:

k0
Q= ng (L + 2Lp ) (6)
1−R

where the group delay τ = ∂φ ∂ω experienced by the light

r

upon mirror reflection enters as a characteristic length

Lp = −λ20 /(4πng )(∂φr /∂λ)λ0 . In general, for PhC struc-
tures (for example, heterostructure mirrors) it has been
shown that Lp can be unrelated to the characteristic
damping length of the energy distribution inside the mir-
rors δ [38]. However, in the configuration under investi-
gation, the classical relation for a DBR, Lp ≤ δ, is valid,
with the equality being strictly fulfilled only in the limit-
ing case of quarter-wave mirrors with low refractive index
contrast, ∆n/n (Lp = δ = nΛ/(2∆n)).
In Fig. 7, the emission in the region |kx /k0 | < 0.5 is
FIG. 7: One-dimensional far-field profiles as a function of shown as a function of the penetration depth. In the case
the intensity penetration depths calculated using the simple of the fundamental mode, there is a single peak centered
Fabry-Perot model for the first (left-side plot) and the second around k = 0 for small penetration depth, with a quite
(right-side plot) order modes. The white dashed lines corre- flat profile and broadening for |kx /k0 | < 0.25. When the
spond to the experimental position of the center of the lobes
penetration depth increases beyond δ ∼ 150 nm, the far-
(Fig. 3) and the penetration depth from FDTD simulations.
field emission splits into two lobes, which separate more
and more on increasing δ. At δ ∼ 250 nm, the peaks are
making: centered around k ∼ ±0.25k0 , corresponding to an open-
ing angle of about 15 degrees. These findings explain the
K(θ, ϕ) ∝ |α̃(k0 sin θ cos ϕ)|2 |β̃(k0 sin θ sin 2 ring-like structure we observe in the experimental far-
2 2 2

ϕ)| × (4) field: the vertical dashed white line in the figure shows
× cos ϕ cos θ + sin ϕ
the experimental central NA for the peaks (from Fig. 3),
In a simplified one-dimensional model, β(y) is narrow which corresponds to a penetration depth of about 220
in real-space, so its Fourier transform is wide and can nm, fully compatible with the penetration depth from
be neglected (|β̃(ky )|2 ∼ 1). If we look at the far-field FDTD simulations, δ1 . For the second-order mode, in
distribution along the x-axis we select ϕ = 0, so that the the region |kx /k0 | < 0.5 there are always two emission
resulting one-dimensional far-field profile is: lobes, even at small penetration depth. On increasing
δ, the two lobes move further and further apart. From
K(θ) ∼ |α̃(k0 sin θ)|2 cos2 θ (5) Fig. 3, the lobes are centered around N A ∼ 0.35, which
in our simulations is consistent with a penetration depth
or, in terms of transverse wavevectors: K(kx ) ∼ of about 150 nm, as predicted by FDTD simulations.
|α̃(kx )|2 (k02 − kx2 ). Therefore, far-field profiles can provide useful informa-
An analytical solution can be found and is reported tion about the penetration depth δ of PhC cavity modes,
in Appendix A. However, the resulting formula is too which in this case offer a bound on the effective cavity
complicated to give intuitive insights; therefore we will length, Lp , and therefore on the Q-factor. Extensions of
just discuss the numerical results (from Eq. 12) in this the present analysis to treat coupled cavity modes [39, 40]
context. Fig. 6 shows the expected far-field profiles for can also be considered.
 7

IV. CONCLUSIONS where Π(x) is the rectangular function, Π(x) = 1 for

|x| < 1/2 and zero elsewhere. The phase φm is set by
In conclusion, we have presented an extensive charac- the boundary conditions: φm = 0 for m even (cosine-like
terization of far-field emission profiles from L3-type pho- solutions) and φm = π/2 for m odd (sine-like solutions).
tonic crystal nanocavities, introducing a simple imaging From Eq. 4:
technique as an efficient tool to give a two-dimensional
mapping of the emitted intensity. The measurements
K1 (θ) ∼ eiφm Sinc mπ
  
have been directly compared to theoretically modeled 2 (sin θ − 1) +
 cos2 θ (9)
2
+e−iφm Sinc mπ

far-field projections from the 3D FDTD near-field cavity 2 (sin θ + 1)
modes profiles, and we believe these results to be useful The two modulo-squared Sinc functions in Eq. 9 give
to PhC cavity designs for specific purposes. The effect two main peaks centered at θ = ±π/2, which correspond
of disorder on the far-field profiles was investigated via to the higher-NA peaks in the far-field in Fig. 6. The first
numerical simulations. relative maximum of Sinc2 (x) is at x ∼ 1.4303π, which
Finally,we have introduced a simple Fabry-Perot model for Eq. 9 corresponds to sin θ ∼ ±(1 − 2.861/m). The
that is able to capture the essential features of far-field FWHM of such a peak for Sinc2 (x) is 0.522π, which cor-
properties for suitably designed near-field profiles. As responds to δkx ∼ 1.044/m (δkx ∼ 0.35 for the first-order
a particular application of this model, we can envision, mode, corresponding to m = 3, and δkx ∼ 0.26 for the
for example, the simultaneous optimization of in- and second-order mode, corresponding to m = 4). Therefore
out-coupling for two different modes supported by the for the first-order mode, the first relative maxima of the
same PhC cavity, which is still an open problem that Sinc functions superpose, giving just one central peak.
might benefit from simplified models like the one pre- For the secod-order mode, on the other hand, the first
sented here. relative maxima are well separated.
Including the penetration depth δ into the model, a
simple
near-field profile can be taken as a superposition of
Acknowledgments Π Lx and two exponentially-decaying wings, as follows:

α(x) = Π Lx + H x − L


−(x−L/2)/δ
The authors acknowledge helpful discussions with 2 e  +
Morten Bakker. This work was supported by NSF NIRT +H −x − L2 e−(−x−L/2)/δ ei(km x+φm ) + c. c.


Grant No. 0304678, Marie Curie EXT-CT-2006-042580 (10)
and FOM\NWO grant No. 09PR2721-2. A portion of where H(x) is the Heaviside function (H(x) = 1 for x
>
this work was done in the UCSB nanofabrication facility, 0, H(x) = 0 for x < 0). The part around x = 0 (Π Lx )
part of the NSF funded NNIN network. gives the same Fourier-transform as in Eq. 9 (which we
label α̃1 (sin θ))), while the left and right-side exponential
decay regions give the following:
Appendix A: Analytical calculations
im(π/2)(sin θ+1) im(π/2)(sin θ−1)
e −iφm e
α̃2 (sin θ) ∼ eiφm δ/L−imπ(sin θ+1) + e δ/L−imπ(sin θ−1)
The resonance frequencies for the modes can be found
by setting the condition that the total phase acquired by e−im(π/2)(sin θ±1)
−iφm e −im(π/2)(sin θ−1)

the Bloch mode traveling back and forth is a multiple of α̃3 (sin θ) ∼ eiφm δ/L+imπ(sin θ±1) + e δ/L+imπ(sin θ−1)
π (2km L = mπ), which results in: (11)
All the quantities depend on the ratio δ/L between the
mπ penetration depth and the cavity length. The far-field
km = (7)
2L profile can be calculated to be:
Let us start with a perfectly confined mode, with no pen-  2
 3
etration into the mirrors (δ = 0). In this case the field is

X
α̃j (sin θ) cos2 θ

given by: K1 (θ) =  (12)
 j=1 
x
αδ=0 (x) = Π ei(km x+φm ) + c. c. (8)
L

[1] J. D. Joannopoulos, S. G. Johnson, J. N. Winn, R. D. tum technologies, Nat Photon 3 (2009) 687–695.
Meade, Photonic Crystals: Molding the Flow of Light, [4] M. Notomi, Manipulating light with strongly modulated
Princeton University Press, 2 edition, 2008. photonic crystals, Reports on Progress in Physics 73
[2] P. Yao, V. Manga Rao, S. Hughes, On-chip single photon (2010) 096501.
sources using planar photonic crystals and single quan- [5] Y. Akahane, T. Asano, B. Song, S. Noda, High-Q pho-
tum dots, Laser & Photonics Reviews 4 (2010) 499–516. tonic nanocavity in a two-dimensional photonic crystal,
[3] J. L. O’Brien, A. Furusawa, J. Vuckovic, Photonic quan- Nature 425 (2003) 944–947.
 8

[6] B. Song, S. Noda, T. Asano, Y. Akahane, Ultra-high-Q and bell-state analysis in the weak-coupling cavity qed
photonic double-heterostructure nanocavity, Nat Mater regime, Phys. Rev. Lett. 104 (2010) 160503.
4 (2005) 207–210. [21] S.-H. Kim, S.-K. Kim, Y.-H. Lee, Vertical beaming of
[7] E. Kuramochi, M. Notomi, S. Mitsugi, A. Shinya, T. Tan- wavelength-scale photonic crystal resonators, Phys. Rev.
abe, T. Watanabe, Ultrahigh-q photonic crystal nanocav- B 73 (2006) 235117.
ities realized by the local width modulation of a line de- [22] N.-V.-Q. Tran, S. Combrié, A. De Rossi, Directive emis-
fect, Appl. Phys. Lett. 88 (2006) 041112. sion from high- q photonic crystal cavities through band
[8] A. R. A. Chalcraft, S. Lam, D. O’Brien, T. F. Krauss, folding, Phys. Rev. B 79 (2009) 041101.
M. Sahin, D. Szymanski, D. Sanvitto, R. Oulton, M. S. [23] M. Toishi, D. Englund, A. Faraon, J. Vučković, High-
Skolnick, A. M. Fox, D. M. Whittaker, H.-Y. Liu, brightness single photon source from a quantum dot in a
M. Hopkinson, Mode structure of the l3 photonic crystal directional-emission nanocavity, Opt. Express 17 (2009)
cavity, Applied Physics Letters 90 (2007) 241117. 14618–14626.
[9] S. Strauf, K. Hennessy, M. T. Rakher, Y.-S. Choi, [24] S. L. Portalupi, M. Galli, C. Reardon, T. Krauss,
A. Badolato, L. C. Andreani, E. L. Hu, P. M. Petroff, L. O’Faolain, L. C. Andreani, D. Gerace, Planar pho-
D. Bouwmeester, Self-tuned quantum dot gain in pho- tonic crystal cavities with far-field optimization for high
tonic crystal lasers, Physical Review Letters 96 (2006) coupling efficiency and quality factor, Opt. Express 18
127404. (2010) 16064–16073.
[10] K. Hennessy, A. Badolato, M. Winger, D. Gerace, [25] A. Badolato, K. Hennessy, M. Atature, J. Dreiser, E. Hu,
M. Atatüre, S. Gulde, S. Fält, E. L. Hu, A. Imamoglu, P. M. Petroff, A. Imamoglu, Deterministic coupling of
Quantum nature of a strongly coupled single quantum single quantum dots to single nanocavity modes, Science
dotcavity system, Nature (London) 445 (2007) 896–899. 308 (2005) 1158–1161.
[11] M. Nomura, S. Iwamoto, M. Nishioka, S. Ishida, [26] S. M. Thon, M. T. Rakher, H. Kim, J. Gudat, W. T. M.
Y. Arakawa, Highly efficient optical pumping of photonic Irvine, P. M. Petroff, D. Bouwmeester, Strong coupling
crystal nanocavity lasers using cavity resonant excitation, through optical positioning of a quantum dot in a pho-
Applied Physics Letters 89 (2006) 161111. tonic crystal cavity, Applied Physics Letters 94 (2009)
[12] M. Nomura, S. Iwamoto, T. Yang, S. Ishida, Y. Arakawa, 111115.
Enhancement of light emission from single quantum dot [27] L. C. Andreani, D. Gerace, M. Agio, Gap maps, diffrac-
in photonic crystal nanocavity by using cavity resonant tion losses, and exciton-polaritons in photonic crystal
excitation, Applied Physics Letters 89 (2006) 241124. slabs, Photonics and Nanostructures - Fundamentals and
[13] R. Oulton, B. D. Jones, S. Lam, A. R. A. Chalcraft, Applications 2 (2004) 103 – 110.
D. Szymanski, D. O’Brien, T. F. Krauss, D. Sanvitto, [28] Y. Akahane, T. Asano, B.-S. Song, S. Noda, Fine-
A. M. Fox, D. M. Whittaker, M. Hopkinson, M. S. tuned high-q photonic-crystal nanocavity, Opt. Express
Skolnick, Polarized quantum dot emission from pho- 13 (2005) 1202–1214.
tonic crystal nanocavities studied under moderesonant [29] L. C. Andreani, D. Gerace, Photonic-crystal slabs with a
enhanced excitation, Opt. Express 15 (2007) 17221– triangular lattice of triangular holes investigated using a
17230. guided-mode expansion method, Phys. Rev. B 73 (2006)
[14] A. Imamoglu, S. Fält, J. Dreiser, G. Fernandez, 235114.
M. Atatüre, K. Hennessy, A. Badolato, D. Gerace, Cou- [30] S. L. Portalupi, M. Galli, M. Belotti, L. C. Andreani,
pling quantum dot spins to a photonic crystal nanocavity, T. F. Krauss, L. O’Faolain, Deliberate versus intrinsic
Journal of Applied Physics 101 (2007) 081602. disorder in photonic crystal nanocavities investigated by
[15] S. Mujumdar, A. F. Koenderink, T. Sünner, B. C. Buch- resonant light scattering, Phys. Rev. B 84 (2011) 045423.
ler, M. Kamp, A. Forchel, V. Sandoghdar, Near-field [31] C. P. Michael, K. Srinivasan, T. J. Johnson, O. Painter,
imaging and frequency tuning of a high-q photonic crys- K. H. Lee, K. Hennessy, H. Kim, E. Hu, Wavelength-
tal membrane microcavity, Opt. Express 15 (2007) and material-dependent absorption in gaas and algaas
17214–17220. microcavities, Applied Physics Letters 90 (2007) 051108.
[16] F. Intonti, S. Vignolini, F. Riboli, A. Vinattieri, D. S. [32] J. Vuckovic, M. Loncar, H. Mabuchi, A. Scherer, Opti-
Wiersma, M. Colocci, L. Balet, C. Monat, C. Zinoni, mization of the q factor in photonic crystal microcavities,
L. H. Li, R. Houdré, M. Francardi, A. Gerardino, IEEE J. Quant. Electr. 38 (2002) 850.
A. Fiore, M. Gurioli, Spectral tuning and near-field imag- [33] C. Sauvan, P. Lalanne, J. P. Hugonin, Slow-wave effect
ing of photonic crystal microcavities, Phys. Rev. B 78 and mode-profile matching in photonic crystal microcav-
(2008) 041401. ities, Phys. Rev. B 71 (2005) 165118.
[17] S. Vignolini, F. Intonti, F. Riboli, D. S. Wiersma, [34] E. Silberstein, P. Lalanne, J.-P. Hugonin, Q. Cao, Use of
L. Balet, L. H. Li, M. Francardi, A. Gerardino, A. Fiore, grating theories in integrated optics, J. Opt. Soc. Am. A
M. Gurioli, Polarization-sensitive near-field investigation 18 (2001) 2865–2875.
of photonic crystal microcavities 94 (2009) 163102. [35] S. Combrié, E. Weidner, A. DeRossi, S. Bansropun,
[18] E. Waks, J. Vuckovic, Dipole induced transparency in S. Cassette, A. Talneau, H. Benisty, Detailed analysis by
drop-filter cavity-waveguide systems, Phys. Rev. Lett. fabry-perot method of slab photonic crystal line-defect
96 (2006) 153601. waveguides and cavities in aluminium-free material sys-
[19] A. Aufféves-Garnier, C. Simon, J.-M. Gerard, J.-P. tem, Opt. Express 14 (2006) 7353.
Poizat, Giant optical nonlinearity induced by a single [36] P. Lalanne, C. Sauvan, J. Hugonin, Photon confinement
two-level system interacting with a cavity in the Purcell in photonic crystal nanocavities, Laser & Photonics Re-
regime, Phys. Rev. A 75 (2007) 053823. views 2 (2008) 514–526.
[20] C. Bonato, F. Haupt, S. S. R. Oemrawsingh, J. Gu- [37] D. A. Fuhrmann, S. M. Thon, H. Kim, D. Bouwmeester,
dat, D. Ding, M. P. van Exter, D. Bouwmeester, Cnot P. M. Petroff, A. Wixforth, H. J. Krenner, Dynamic mod-
 9

ulation of photonic crystal nanocavities using gigahertz metry in photonic structures, Physical Review Letters
acoustic phonons, Nat Photon 5 (2011) 605–609. 106 (2011) 143901.
[38] C. Sauvan, J. P. Hugonin, P. Lalanne, Difference between [40] M. Brunstein, T. J. Karle, I. Sagnes, F. Raineri, J. Bloch,
penetration and damping lengths in photonic crystal mir- Y. Halioua, G. Beaudoin, L. L. Gratiet, J. A. Levenson,
rors, Applied Physics Letters 95 (2009) 211101. A. M. Yacomotti, Radiation patterns from coupled pho-
[39] F. Intonti, F. Riboli, N. Caselli, M. Abbarchi, S. Vigno- tonic crystal nanocavities, Applied Physics Letters 99
lini, D. S. Wiersma, A. Vinattieri, D. Gerace, L. Balet, (2011) 111101.
L. H. Li, M. Francardi, A. Gerardino, A. Fiore, M. Guri-
oli, Youngs type interference for probing the mode sym-