
2
quality factors of photonic crystal microcavities [9, 10]
and automated design methods to optimize the pho-
tonic crystal microcavity structure [11, 12]. We fo-
cus on the development of a design approach which
increases the robustness of the coupling between an
emitter and a photonic crystal waveguide mode. We
propose two design approaches increasing the robust-
ness of coupling to the fabrication errors and test them
using numerical simulations.
II. PHOTONIC CRYSTAL
A typical two-dimensional photonic crystal is a pe-
riodic arrangement of circular holes etched in a thin
film of a material with high refractive index. A deleted
row of holes forms a photonic crystal waveguide (see
illustration in Fig.1(a)). A characteristic feature of a
PCW is the existence of a frequency range where the
group velocity of light decreases significantly. This
fact makes a PCW structure an extremely appealing
system for mediating interaction between light and an
isolated dipole. A PCW effectively serves as a mi-
croresonator with small mode volume and high qual-
ity factor. These systems were demonstrated to suite
the purpose of integration of A3B5quantum dot sin-
gle photon sources in a planar photonic structure [5].
Quantum dot can be considered as a dipole, which
is orientated perpendicular to the waveguide axis in
the PC plane. A PCW microresonator forms an open
cavity which can be smoothly interfaced with other in-
tegrated photonic waveguides. In this paper we study
methods to increase robustness of PCW features to
fabrication defects.
We focus our attention on a PCW created by delet-
ing a row of air holes from a 2D triangular array. The
host material is chosen to be gallium arsenide (GaAs)
because the target application is a planar semiconduc-
tor quantum dot single photon source. We start with
the description of PCW characteristics and develop-
ment of its heuristic model. Manga Rao and Hughes
[13] derived an expression for a Purcell factor Fpin
terms of PCW parameters:
Fp=3πc3a
Veff ω2
d3/2vg
,(1)
where ais the distance between air holes (if the lat-
tice is triangular a=ax, since we are focused on that
type of lattice from now on we will write ainstead
ax, but for another lattice angle axis the only cor-
rect option), Vef f is the effective mode volume and
vgis the group velocity at the resonant frequency of
the dipole ωd. The formula indicates that the largest
Fpis achieved when the wavepacket group velocity
reaches zero. Thus the design of a PCW efficiently
coupled with a single emitter resonant at ωdis equiva-
lent to engineering a PCW dispersion law to meet the
requirement dω/dk(ωd) = 0. Numerical methods for
calculation of the dispersion structure of a PCW are
well-known and straightforward [14] and can be easily
applied to a PCW with a defined geometry. However,
there exists no recipe of how to estimate geometrical
parameters of a PCW exhibiting high Purcell factor at
a wavelength of interest. We devise heuristic expres-
sions linking the target wavelength and the parameters
of a hexagonal PCW which stem upon simple optical
effects taking place inside a photonic crystal. Based
on these results we introduce methods to increase the
robustness of a PCW structure to fabrication defects.
III. A PCW PURCELL FACTOR HEURISTICS
Figure 1(b) illustrates a typical dispersion structure
of a PCW. The geometrical parameters for this exam-
ple are as follows: PC hole pattern angle θgr = 60◦,
period a= 0.238 µm, hole radius r= 0.08 µm and
membrane thickness h= 0.16 µm. These values were
chosen to put the Purcell factor FP CW peak at 925 nm.
A natural question arises whether this configuration is
unique. It turns out that the answer is negative. We
performed an extensive numerical analysis of the Pur-
cell enhancement happening in different PCW config-
urations, results are presented in Fig. 2.The 3D FDTD
simulation was carried out in Lumerical FDTD pack-
age, the details of the simulation are specified in Ap-
pendix D. We observed a continuous set of configura-
tions of a triangular PCW with the same lattice an-
gle corresponding to a peak value of FP CW at a tar-
get wavelength. The red line in Fig. 2 illustrates the
numerically computed set of (a, r) configurations cor-
responding to the most efficient coupling of a PCW
mode to dipole radiation at 925 nm. The curve in
(a, r) space closely follows the function a=c1+c2r,
where the coefficients c1and c2are weakly dependent
on aand r. The a(r) dependence is finely approx-
imated by a linear function in the region where the
FP CW reaches its highest levels. The yellow curve rep-
resents the values of aand rcorresponding to FP CW
peak which are provided by the proposed theoretical
description.
In the following subsections we provide a heuristic
theoretical description for the origins of such depen-
dence and the values of c1and c2coefficients.
A. The slope coefficient c2
The Purcell factor FP CW defines the probability
β=FP CW /(1 + FP CW ) (2)
of emitting a photon into a PCW mode. The existence
of a PCW mode is a purely interferometric effect hence
the probability pshould be related to the geometry of
a photonic crystal. We expect to derive the connec-
tion between the geometric parameters which corre-
spond to a configuration of a PCW structure reaching
maximal Purcell factor for the required wavelength.
We roughly split the emitter radiation into three cate-
gories: light exiting the PCW plane, light propagating
inside the PCW structure, and light coupled to the