4678 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 27, NO. 21, NOVEMBER 1, 2009
Stress-Induced Birefringence Characteristics
of Polymer Optical Rib Waveguides
M. Faruque Hossain, Hau Ping Chan, Member, IEEE, Mohammad Afsar Uddin, and R. K. Y. Li
Abstract—We report the detailed numerical investigation of
stress-induced material birefringence in polymer rib waveguide
for the design of nonbirefringent waveguide devices. To accurately
simulate the stress-induced effects we propose a more realistic
model in the finite element analysis which considers the stresses
induced over the entire sequential fabrication process. It is ob-
served that the birefringence is nonuniform, and it is different
for different etch depth and core width. The maximum birefrin-
gence in the core layer is observed near the lower cladding which
decreases to zero toward the top surface. The influence of this
material anisotropy on the modal birefringence is analyzed also
for different rib structures. We found the stress effects on the
modal birefringence to be largely affected by etch depth, while
core width has small effect. It is also found that the deeply etched
core has better birefringence stability. Finally, an accurate design
of the zero-birefringence waveguide is illustrated by taking the
stress effects into account, and the results are compared with
experimental data. Excellent agreement between calculated and
experimental results confirms the potential application of this
work to aid in the design of polarization-insensitive waveguide
devices.
Index Terms—Birefringence, optical waveguide, polymer wave-
guide, stress.
I. INTRODUCTION
P
LANAR optical waveguides are basic components in
many functional devices of optical communication
system such as, arrayed waveguide gratings (AWG), Bragg
grating filters, etc. Polymer materials are very attractive in
the fabrication of such waveguide devices due to their advan-
tages, such as simple fabrication processes using embossing
or UV techniques, easy formation of multilayer structures by
spin-coating on any surface of interest, and potential low cost at
mass production [1]–[3]. One bottleneck to the widespread ap-
plication of these devices is polarization-dependent properties
[4], [5]. An important originating source of polarization depen-
dence is stress, which is usually generated in the fabrication
process due to polymerization shrinkage and thermal-expansion
mismatch among dissimilar materials. This stress causes an
anisotropic change in the material refractive index, and thus
Manuscript received October 29, 2008; revised April 08, 2009. First pub-
lished June 16, 2009; current version published September 10, 2009.
M. F. Hossain, H. P. Chan, and M. A. Uddin are with the Department of
Electronic Engineering, City University of Hong Kong, Kowloon Tong, Hong
Kong (e-mail: eehpchan@cityu.edu.hk).
R. K. Y. Li is with the Department of Physics and Materials Science, City
University of Hong Kong, Kowloon Tong, Hong Kong.
This work was supported by the Research Grants Council of Hong Kong Spe-
cial Administrative Region, China, under project CityU 1109/05E.
Digital Object Identifier 10.1109/JLT.2009.2025518
material birefringence. The combination of stress-induced ma-
terial birefringence and geometrical birefringence results modal
birefringence, which leads to the two orthogonal polarization
modes to travel at slightly different velocities and mode field
distribution. Thus adversely affect the transmission of light and
performance of devices [6], [7].
To reduce this polarization dependence, a few methods, such
as using stress-releasing grooves or introducing some additional
layers to reduce the stress [8] or using compensating devices
[5], have been used. Generally, those methods increase the com-
plexity and cost of the devices. A simple and practical solu-
tion is to use a polarization-insensitive (i.e., nonbirefringent)
planar waveguide device. An important building block of po-
larization-insensitive devices is a zero-birefringence waveguide.
Theoretical works have been carried out in the design of such
waveguide devices by tailoring the waveguide geometry [9],
[10]. However, it is often found discrepancies between theoret-
ical design and experimental results due to the effects of induced
stresses [11], [12]. In order to minimize or control the bire-
fringence, therefore, it is paramount that the design engineers
have a good quantitative understanding of stress distributions in
planar waveguides, and their impact on the optical properties of
components.
Several empirical techniques have been developed to study
the stress-birefringence in optical waveguides [13], [14]. These
techniques are limited to uniform stress distribution only, and
not sufficient for the growing complicated optical circuit design.
On the other hand, it is impractical to measure the stresses as
well as the optical properties inside the complex structure of tiny
optical waveguide devices. A number of works have analyzed
the stress and the induced birefringence numerically using the
finite element (FE) method [15]–[17]. These works employed
simplified model in which the entire structure is assumed to
be stress free at certain temperature (e.g., SiO or glass depo-
sition temperature) and then cool down to room temperature.
This conventional approach may not be suitable to handle the
multilayer polymer devices where the material layers are de-
posited and cured at different temperature. As a result, all the
layers cannot be stress free at a certain temperature. A signif-
icant amount of stress may generate due to curing shrinkage
of polymer which cannot be explain also by simplified model.
However, although the stress-induced effects have been studied
extensively in glass waveguide [15]–[17], not many people have
analyzed stress-birefringence in polymer waveguides. This is
partly because most of the commercially available numerical
software intended for optical analysis has the limitations in the
modeling of previously mentioned practical steps. In particular,
the quantitative predictions of stress in polymer waveguides in
detail and sequential manner as well as its effects on the device
performance have not been reported yet.
0733-8724/$26.00 © 2009 IEEE