Content uploaded by Xinliang Zhang
Author content
All content in this area was uploaded by Xinliang Zhang on Oct 05, 2021
Content may be subject to copyright.
JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 30, NO. 11, JUNE 1, 2012 1595
An Ultracompact DP-QPSK Demodulator Based on
Multimode Interference and Photonic Crystals
Lei Xiang, Yameng Xu, Yu Yu, and Xinliang Zhang, Member, IEEE
Abstract—We propose a novel scheme that combines a polariza-
tion-beam splitter (PBS) and a 90 hybrid into a single device to
demodulate the dual-polarization quadrature phase-shift keying
(DP-QPSK) signal by introducing a photonic crystal structure into
a4 4 multimode interference coupler. Since no separated PBS
is utilized, the demodulator has much smaller chip size compared
with a conventional DP-QPSK demodulator which contains two
PBSs and two 90 hybrids. Simulation with 2-D finite-difference
time-domain method indicates that the demodulator can operate
very well for demodulating the DP-QPSK signal. The numerical
simulation demonstrates a common-mode rejection ratio larger
than 20 dB and phase errors smaller than of the device over
the C-band wavelength range.
Index Terms—Coherent detection, multimode interference
(MMI), photonic crystal (PC).
I. INTRODUCTION
WITH the rapid growth of information exchange all over
the world, it is urgent to increase the speed of signal
transmission, in other words, increasing the spectral efficiency
(SE) as high as possible. For this reason, more and more
multilevel modulation formats have been proposed, such as
quadrature phase-shift keying (QPSK), quadrature amplitude
modulation, and so on [1]. Accordingly, the integrated devices
for these advanced modulation formats have also received
more and more attentions [2]–[4]. Dual-polarization QPSK
(DP-QPSK) using coherent detection with digital signal pro-
cessing is a promising solution to achieve 100-Gb/s channel
rate in wavelength division multiplexing systems because
of its high SE (4 bits/symbol), high receiver sensitivity, and
high resilience to the linear impairments such as chromatic
dispersion and differential group delay [5]–[7]. Usually, two
polarization-beam splitters (PBSs) and two 90 optical hybrids
are required in the receiver to demodulate the DP-QPSK signal
[8]–[11]. Although many methods to achieve the PBS function
had been proposed [12], [13], the footprints for these schemes
were still too large and as a result became a problem for
Manuscript received November 17, 2011; revised January 28, 2012; accepted
February 12, 2012. Date of publication March 13, 2012; date of current ver-
sion April 04, 2012. This work was supported in part by the National Basic
Research Program of China under Grant 2011CB301704, the National Nat-
ural Science Foundation of China under Grant 61007042, the National Science
Fund for Distinguished Young Scholars under Grant 61125501, and the Doc-
toral Program Foundation of Institutions of Higher Education of China under
Grant 20090142110052.
The authors are with the Wuhan National Laboratory for Optoelectronics,
Huazhong University of Science and Technology, Wuhan 430074, China
(e-mail: xlzhang@mail.hust.edu.cn).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/JLT.2012.2188277
Fig. 1. Conventional DP-DQPSK demodulator.
monolithic integration. At the same time, various kinds of 90
hybrids have been reported such as 4 4or2 4 multimode
interference (MMI) [14], [15], star coupler [16], and 3-dB
couplers with a phase shifter [17]. And 4 4or2 4 MMI is
a good demodulation candidate for its wide optical bandwidth,
simple structure, and large fabrication tolerance.
Previously, In [18] and [19], the authors have attached a pho-
tonic crystal (PC) structure into a 2 2 MMI coupler to realize
broadband demultiplexers and an ultracompact PBS. Here, we
propose an ultracompact DP-QPSK demodulator that combines
the PBS with the 90 optical hybrid by introducing a PC struc-
ture into a 4 4 MMI coupler. Comparing with conventional
demodulator, the chip length can be reduced.
II. DESIGN AND ANALYSIS
The conventional DP-QPSK demodulator is shown in Fig. 1.
Two PBSs are used to split the signal and the local oscillator
(LO) into TE-polarization (TE-pol) and TM-polarization (TM-
pol). Then, the TE (TM)-pol signal transmits into a 90 optical
hybrid with the corresponding LO signals, respectively. At the
output, the in-phase and quadrature (I/Q) can be obtained with
balanced detection in TE-pol and TM-pol, respectively.
From Fig. 1, it can be seen that in order to get a DP-QPSK
demodulator, the PBS and the 90 optical hybrid should be de-
signed and fabricated separately and then combined together to
form a demodulator. As for the 90 hybrid, a 4 4 MMI had
been proposed and demonstrated successfully to act as a 90
hybrid [14], [15], and the operation principle is shown in Fig. 2.
As Fig. 2 shows, when the input signals have different phase
differences, the four output parts of 4 4 MMI have different
outputs. Assisting by the balanced detectors, the I and Q signal
can be detected. In Fig. 2(a)–(d), the left parts of the figures
are the qualitative simulation results, while the right parts are
the quantitative simulation results. The four output ports are
numbered from right to left. Fig. 2(a) shows the output when
the phase difference of the two input signals is 0 . As can be
seen, the first and the third output ports receive much larger
0733-8724/$31.00 © 2012 IEEE
Authorized licensed use limited to: Huazhong University of Science and Technology. Downloaded on March 10,2020 at 02:00:00 UTC from IEEE Xplore. Restrictions apply.
1596 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 30, NO. 11, JUNE 1, 2012
Fig. 2. Characteristic of 4 4MMIasa90 hybrid. (a) Phase difference of
the two input signals: 0 . (b) Phase difference of the two input signals: 90 .
(c) Phase difference of the two input signals: 180 . (d) Phase difference of the
two input signals: 270 .
power than the second and the forth ports. Then, the first and the
forth output parts are connected to a pair of photodiodes (PDs),
while the second and the third outputs are connected to another
pair of PDs. Hence, the I and Q can be obtained for 1 and 0.
Fig. 2(b)–(d) shows the outputs when the phase differences of
the two input signals are 90 ,180,and270,inthesimilar
way, the I/Q can be obtained for 0/0, 0/1, and 1/1. In this way,
the 4 4 MMI can be used as a 90 hybrid to demodulate the
QPSK signal.
On the other hand, to form the DP-QPSK demodulator, the
PBS is a critical part. It is obvious that if the extinction ratio
of the PBS is not good enough, the demodulation result may be
influenced. T. Inoue et al. have proposed a method by cascading
two PBSs to improve the polarization extinction ratio of the
TE/TM signals, which further enlarges the chip size to be 25
mm 21 mm [10]. To solve the problem, we propose a novel
scheme to realize the DP-QPSK demodulator by combining the
PBS with the 90 hybrid into a single device, which means that
the phase-modulated signal can be demodulated while the TE
and TM modes can be split simultaneously. Fig. 3 shows the
structure diagram in which the functions of both PBS and 90
hybrid are illustrated.
The PC structure located in the middle of the MMI coupler
is specially designed as a polarization sensitive filter, which en-
ables the TE mode passing through, while the TM mode being
obstructed, as shown in Fig. 3(a). On the other hand, another
PC structure can be designed to ensure the TM mode passing
through, while the TE mode is obstructed, as shown in Fig. 3(b).
As a result, the PC-assisted 4 4 MMI can replace the PBS and
Fig. 3. Schematic configuration of PC-assisted 4 4 MMI. (a) PC-assisted
MMI that demodulates the TE-pol signal. (b) PC-assisted MMI that demodu-
lates the TM-pol signal.
Fig. 4. Ultracompact DP-DQPSK demodulator.
the 90 hybrid, which means the DP-QPSK demodulator can be
realized in an ultracompact size as shown in Fig. 4.
The operational principle of the demodulator shown in Fig. 4
can be analyzed as follows. The signal and the LO are split by
two 1 2 MMI couplers, respectively. Then, the split signal
and LO transmit into the PC-assisted 4 4MMI.IfthePCis
TE-transparent and TM-obstructed, only the TE-pol signal and
the TE-pol LO can interact in the 4 4MMIandthenbede-
tected by the balanced detector. In this case, the TE-pol QPSK
signal can be demodulated, while the TM-pol QPSK cannot,
and vice versa in the other case. Compared with the conven-
tional demodulator shown in Fig. 1, the proposed demodulator
eliminates the separate PBS part, which can obviously reduce
the chip size and it is more competitive in fabrication tolerance
because there are less devices to fabricate. On the other hand,
the two 1 2 MMI couplers and two 4 4 MMIs can be easily
monolithic integrated, which is very helpful for increasing the
stability and reducing the transmission and coupling loss.
Authorized licensed use limited to: Huazhong University of Science and Technology. Downloaded on March 10,2020 at 02:00:00 UTC from IEEE Xplore. Restrictions apply.
XIANG et al.: ULTRACOMPACT DP-QPSK DEMODULATOR 1597
As Chung and Lee and Shi et al. have assumed, the PC
structure in the middle of the MMI has the same transmission
coefficient for all the guided modes [18], [19]. In Soldano
and Pennings’ opinion [20], the field profile at a distance L
can be written as a superposition of all the guided mode field
distributions
(1)
where and are the excitation coefficient and modal
field profile ofthe th order mode, and are the propagation
constants of the fundamental and th order mode, respectively,
and the propagation constants spacing can be written as
(2)
where is the beat length of the two lowest order modes.
Then, the can be written as
(3)
For 4 4 MMI, the first fourfold image distance should be
(4)
Since the size of the PC in the middle of the MMI is very
tiny, its influence on the propagation constants of all the guided
modes can be ignored. However, it cannot be ignored for phase-
modulated formats, since the PC structure may introduce phase
shifts to the phase-modulated formats and the imaging phenom-
enon may be changed. Assuming that the phase shifts of all the
guided modes caused by the PC are the same, which is defined
as , to ensure that the PC-assisted 4 4 MMI can work as a
90 hybrid. Then, (1) should be amended
(5)
Then, the first fourfold image distance should be
(6)
From (6), it can be seen that if the number of the excited guide
modes is large enough, (6) can be approximately equal to (4),
and the influence of the PC on the imaging length of MMI can
be ignored. It should be noted that if the size of MMI is very
small and the number of the excited guide modes is quite few,
the influence of the PC cannot be ignored, and it may shorten
the imaging length of the MMI; in this case, we may slightly
shorten the length of the designed MMI.
III. SIMULATION RESULTS AND DISCUSSION
In our numerical simulation, a silicon-on-insulator sand-
wiched structure is employed. The thickness for the Si guide
and SiO layer is 0.22 and 2 m, and the refractive index is
3.48 and 1.46, respectively. The effective index method is used
Fig. 5. Band structures of the PC that the radius of the air hole is for
TE and TM polarizations.
in the simulation; thus, the 3-D structure can be treated as a
2-D structure. The multimode section has a width of
12 m to ensure that the influence of the PC can be ignored.
The effective index of the slab is 2.83 for TE-pol and 1.9
for TM-pol. The input and output waveguides have a same
width of 0.35 m to guarantee the single-mode
operation. To reduce the insertion loss (IL), we further add
linear taper structures between the input/output waveguides and
the multimode section. The width and length of the taper are set
to be 0.45 and 10 m, respectively. Using the 2-D beam-prop-
agation method, we get the first fourfold image length of 4 4
MMI for TE-pol mand mfor
TM-pol.
After designing the 4 4 MMI, it is crucial todesignthePC
structure as required. In this paper, we design the PC struc-
ture with a hexagonal pattern of air holes (with a period of
) in the multimode section. In order to realize the TE-trans-
parent PC, the radius of the air hole is . The bandgap di-
agram for such a PC structure is shown in Fig. 5. Such a PC
structure was found to have a bandgap in the spectral range
for TM polarization as shown in
Fig. 5. For the communication wavelength m, we
set which falls inside the spectral range; then, we
get the PC period m. We can get the wavelength range
that the TM-pol cannot pass through.
The length of the PC is chosen to be six periods to ensure large
enough reflection for TM-pol. And the width of the PC should
be m m periods to cover the whole width of the
multimode section.
Meanwhile, the TM-transparent PC-assisted 4 4MMI
structure can be obtained by changing the parameters of the
PC and MMI. Fig. 6 shows the bandgap diagram for a PC
structure with the air hole . Such a PC structure has
a bandgap in the spectral range
for TE polarization as shown in Fig. 6. With the same method
mentioned previously, the parameters of the TM-transparent
PC-assisted 4 4 MMI structure can be optimized. At last,
the PC period is chosen to be m. We can get the
wavelength range that the TE-pol cannot
Authorized licensed use limited to: Huazhong University of Science and Technology. Downloaded on March 10,2020 at 02:00:00 UTC from IEEE Xplore. Restrictions apply.
1598 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 30, NO. 11, JUNE 1, 2012
Fig. 6. Band structures of the PC that the radius of the air hole is for
TE and TM polarizations.
pass through. It can be seen that the PC structures may not in-
fluence the operating wavelength range of MMI.
It is obvious that the PC structures have tight fabrication tol-
erance, and the undesigned PC structures may influence the
working performance of the demodulator. As we have explained
in the prior section, the influence of the PC on the imaging length
of MMI can be ignored, while the PC structures have the sig-
nificant influence on the wavelength dependence of the device
across the C-band wavelength range. The radius of the air holes
is the most important factor, and we simulate the bandgap of
the PC structures with different radiuses as shown in Fig. 7. For
TE-transparent PC, the period m, if we expect the PC
to work across the C-band wavelength range, it should have a
bandgap for TM in the range .From
Fig. 7(a), it can be seen that the radius can vary from to
, which means from 0.1602 to 0.198 m, the fabrication
tolerance is 37.8 nm. While, for TM-transparent PC, the period
m and the PC should have a bandgap for TE in the
range . From Fig. 7(b), it can be seen
that the radius can vary from to , which means
from 0.5029 to 0.5194 m, the fabrication tolerance is 16.5 nm.
It can be seen that the TE-pol demodulator has the larger fabrica-
tion tolerance than the TM-pol demodulator, and in fabrication,
we should accurately control the dimension of the PC structure.
The 2-D finite-difference time-domain (FDTD) method [21]
is used to simulate the whole structure. Without loss of gener-
ality, only the TE-transparent PC-assisted 4 4 MMI has been
simulated. First, two signals in TM-pol whose phase difference
is 0 are injected into the proposed structure and the simulation
result is shown in Fig. 8.
Fig. 8 shows the light field intensity of the TM-pol signals,
as we can see that the TM-pol signals were obstructed by the
TE-transparent PC and they cannot transmit to the output parts.
So the TM-pol signals cannot be demodulated in the TE-trans-
parent PC-assisted 4 4 MMI. For comparison, two signals in
TE-pol with phase difference of 0 ,90 , 180 ,and270 are in-
jected into the proposed structure, respectively. Fig. 9 shows the
simulation results.
Fig. 9(a)–(d) shows the light field intensity in the PC-assisted
MMI when the two TE-pol input signals with different phase
Fig. 7. Band structures of the PC with the radius of the air hole increasing from
(a) to and (b) to for TE and TM polarizations. The red parts
denote TM mode while blue parts denote TE mode.
Fig. 8. Two-dimensional FDTD simulation of the field distributions at
m in the TE-transparent PC-assisted 4 4 MMI demodulator with two
TM-pol signals injected whose phase difference is 0 .
differences. From the simulation results, it can be seen that in the
TE-transparent PC-assisted 4 4 MMI demodulator the TE-pol
Authorized licensed use limited to: Huazhong University of Science and Technology. Downloaded on March 10,2020 at 02:00:00 UTC from IEEE Xplore. Restrictions apply.
XIANG et al.: ULTRACOMPACT DP-QPSK DEMODULATOR 1599
Fig. 9. Two-dimensional FDTD simulation of the field distributions at m in the TE-transparent PC-assisted 4 4 MMI demodulator. (a) Phase
difference of the two input signals in TE-pol: 0 . (b) Phase difference of the two input signals in TE-pol: 90 . (c) Phase difference of the two input signals in
TE-pol: 180 .(d
) Phase difference of the two input signals in TE-pol: 270 .
Fig. 10. Output power of the four output waveguides. (a) Phase difference of the two input signals in TE-pol: 0 . (b) Phase difference of the two input signals in
TE-pol: 90 . (c) Phase difference of the two input signals in TE-pol: 180 . (d) Phase difference of the two input signals in TE-pol: 270 .
Authorized licensed use limited to: Huazhong University of Science and Technology. Downloaded on March 10,2020 at 02:00:00 UTC from IEEE Xplore. Restrictions apply.
1600 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 30, NO. 11, JUNE 1, 2012
Fig. 11. IL difference between the conventional 4 4MMIandthePC-as-
sisted 4 4MMI.
signals can transmit to the output parts. From Fig. 9(a)–(d), we
can see that when the phase difference of the two input signals
is different, the four outputs can get different permutation and
combination, which means that the TE-transparent PC-assisted
44 MMI has the potential to demodulate the TE-pol QPSK
signals. It should be noted that the TM-pol QPSK signals can
also be demodulated by the TM-transparent PC-assisted 4 4
MMI demodulator which is not simulated here. The simulation
results shown in Fig. 9 are qualitative, in order to know whether
the PC-assisted 4 4 MMI can demodulate the DP-QPSK per-
fectly, Fig. 10 shows the details of the quantitative simulation
results.
Fig. 10 shows the state of the four output waveguides with
monitors, the x-coordinate represents the simulation stop size
and the y-coordinate represents the normalized output power.
In Fig. 10(a), the two input signals have a phase difference of
0,thepowerofthefirst and the third output waveguides are
larger than the other two, and the four outputs of the PC-as-
sisted demodulator have almost the same intensity contrast as
the conventional demodulator shown in Fig. 2(a). Using the
balanced detector, the I/Q can be obtained for 1/0 in TE-pol.
Similarly, Fig. 10(b)–(d) proves the analysis in Fig. 2(b)–(d),
and with the balanced detector, the I/Q can be obtained for 0/0,
0/1, and 1/1 in TE-pol, respectively. It can be observed that the
PC structure has no influence on the 4 4 MMI as a 90 hy-
brid, and our previous assumption that the phase shifts of all
the guided modes caused by the PC are the same is correct.
Thus, the TE-transparent PC-assisted 4 4 MMI can work well
for TE-pol QPSK demodulation. Similarly, the TM-transparent
PC-assisted 4 4 MMI can work well for TM-pol QPSK de-
modulation. This proves that the structure shown in Fig. 4 can
work well for DP-QPSK demodulation.
From Fig. 9, it can be seen that some TE-pol signals have dif-
fused because of the PC structure in the TE-transparent PC-as-
sisted 4 4MMI.Inordertoknowtheinfluence of the PC struc-
ture, we simulate the IL of the 4 4 MMI with and without
PC and give their difference value. In Fig. 11, it can be seen
that the IL difference value is less than 2 dB across the C-band
range, especially the IL difference value is almost less than 1
dB across the wavelength range from 1540 to 1570 nm. Thus,
the PC structure influences the transmission quality of the 4 4
MMI to some degree but its influence to the demodulation re-
sults can be ignored.
Fig. 12. Simulated CMRR of the proposed device as a function of wavelength:
(a) in TE demodulator and (c) in TM demodulator. Simulated phase error of the
proposed device as a function of wavelength: (b) in TE demodulator and (d) in
TM demodulator.
As a 90 hybrid, the common-mode rejection ratio (CMRR)
and phase error are two critical performance parameters
[22]–[25]. The CMRR indicates the symmetry of the four
outputs when power is launched from either input. Defining
Pi is the power at each output, then the CMRR for I channel
is defined as dB ,
while the CMRR for Q channel is defined as dB
. The phase error indicates the
phase deviation from the phase quadrature condition. In order
to get the optimal CMRR results, the width and length of the
output parts taper are set to be 1 and 10 m, respectively.
Fig. 12(a) and (c) shows the corresponding CMRR for the sim-
ulated device across the C-band wavelength range in TE-pol
and TM-pol, respectively. The simulation result confirms that
the device can exceed the typical system requirement of 20
dB over C-band. Fig. 12(b) and (d) shows the phase errors as
defined in [25] are within over the C-bandbothinTEand
TM demodulator. From the simulated results, it can be seen
that the PC-assisted 4 4 MMI can match the performance
parameters as a 90 hybrid. Fig. 13 shows the simulated IL in
TE mode of the proposed structure in Fig. 4 from the signal port
to the output ports. The 6 dB intrinsic loss of the 90 hybrid
is not counted in Fig. 13, this means the IL is less than 10 dB.
The simulated loss breakdown is the following: 4 dB the 1 2
MMI IL, 2–5 dB the PC-assisted MMI IL. The propagation
length is so short that propagation loss has been ignored. With
the proposed structure, the coherent receiver chip size can be
less than 0.5 1mm, considering the chip size and the IL
with the previous work [10], [17], [26], our proposed structure
is promising for further investigating.
IV. CONCLUSION
We have proposed and demonstrated an ultracompact
DP-QPSK demodulator, using the FDTD numerical simulation.
Comparing with the conventional DP-QPSK demodulator that
has two PBSs and two 90 hybrids, the novel demodulator that
combines the PBS and the 90 hybrid into a single device by
Authorized licensed use limited to: Huazhong University of Science and Technology. Downloaded on March 10,2020 at 02:00:00 UTC from IEEE Xplore. Restrictions apply.
XIANG et al.: ULTRACOMPACT DP-QPSK DEMODULATOR 1601
Fig. 13. IL of the proposed structure in TE mode.
introducing a PC structure into a 4 4 MMI is in ultracompact
size. Since the TE-pol and TM-pol can be split completely in
the PC structure, we can believe that the demodulation results
are more reliable and this device can be more advantageous in
monolithic integrated circuit and its characteristics are suffi-
cient for coherent detection over the C-band.
REFERENCES
[1] Y. Han and G. Li, “Theoretical sensitivity of direct-detection multi-
level modulation formats for high spectral efficiency optical commu-
nications,” IEEE J. Sel. Topics Quantum Electron., vol. 12, no. 4, pp.
571–580, Jul./Aug. 2006.
[2] E. Cassan et al., “On-chip optical interconnects with compact and low-
loss light distribution in silicon-on-insulator rib waveguides,” IEEE J.
Sel. Topics Quantum Electron., vol. 9, no. 2, pp. 460–464, Oct. 2003.
[3] G. Rasigade et al., “High extinction ratio 10 Gbit/s silicon optical mod-
ulator,” Opt. Exp., vol. 19, no. 7, pp. 5827–5832, Mar. 2011.
[4] L. Vivien et al., “42 GHz pin germanium photodetector integrated
in a silicon-on-insulator waveguide,” Opt. Exp., vol. 17, no. 8, pp.
6252–6257, Apr. 2009.
[5] C. Wree et al., “High spectral efficiency 1.6-b/s/Hz transmis-
sion (8 40 Gb/s with a 25-GHz grid) over 200-km SSMF using
RZ-DQPSK and polarization multiplexing,” IEEE Photon. Technol.
Lett., vol. 15, no. 9, pp. 1303–1305, Sep. 2003.
[6] D. van den Borne et al., “1.6-b/s/Hz spectrallyefficient transmis-
sion over 1700 km of SSMF using 40 85.6-Gb/s POLMUX-RZ-
DQPSK,” J. Lightw. Technol., vol. 25, no. 1, pp. 222–232, Mar. 2007.
[7] P. Boffiet al., “Measurement of PMD tolerance in 40-Gb/s polar-
ization-multiplexed RZ-DQPSK,” Opt. Exp., vol. 16, no. 17, pp.
13398–13404, Aug. 2008.
[8] Y. Sakamaki et al., “Dual polarisation optical hybrid using silica-based
planar lightwave circuit technology fo r digitalcoherent receiver,” Elec-
tron. Lett., vol. 46, no. 1, pp. 58–60, Jan. 2010.
[9] R. Nagarajan et al., “10 channel, 45.6 Gb/s per channel, polariza-
tion-multiplexed DQPSK, InP receiver photonic integrated circuit,” J.
Lightw. Technol., vol. 29, no. 4, pp. 386–395, Feb. 2011.
[10] T. Inoue et al., “Double-pass PBS-integrated coherent mixer using
silica-based PLC,” presented at the presented at the Opt. Fiber
Commun. Conf., San Diego, CA, 2010, Paper OThB2.
[11] C. R. Doerr et al., “Monolithic InP multiwavelength coherent receiver
using a chirped arrayed waveguide grating,” J. Lightw. Technol., vol.
29, no. 4, pp. 536–541, Feb. 2011.
[12] Y. Hashizutne et al., “Integrated polarisation beam splitter using wave-
guide birefringence dependence on waveguide core width,” Electron.
Lett., vol. 37, no. 25, pp. 1517–1518, 2001.
[13] M. R. Watts et al., “Integrated mode-evolution-based polarization
splitter,” Opt. Lett., vol. 30, no. 9, pp. 967–969, May 2005.
[14] M. Seimetz and C. M. Weinert, “Options, feasibility, and availability of
2490 hybrids for coherent optical systems,” J. Lightw. Technol.,
vol. 24, no. 3, pp. 1317–1322, Mar. 2006.
[15] S. H. Jeong and K. Morito, “Novel optical 90 hybrid consisting of
a paired interference based 2 4 MMI coupler, a phase shifter and a
22 MMI coupler,” J. Lightw. Technol., vol. 28, no. 9, pp. 1323–1331,
May 2010.
[16] C. R. Doerr et al., “Monolithic demodulator for 40-Gb/s DQPSK using
a star coupler,” J. Lightw. Technol., vol. 24, no. 1, pp. 171–174, Jan.
2006.
[17] Y. Sakamaki et al., “One-chip integrated dual polarization optical hy-
brid using silica-based planar lightwave circuit technology,” in Proc.
Eur. Conf. Opt. Commun., 2009, pp. 1–2.
[18] L. W. Chung and S. L. Lee, “Multimode-interference-based broadband
demultiplexers with internal photonic crystals,” Opt. Exp., vol. 14, no.
11, pp. 4923–4927, May 2006.
[19] Y. Shi et al., “Proposal for an ultracompact polarization-beam splitter
based on a photonic-crystal-assisted multimode interference coupler,”
IEEE Photon. Technol. Lett., vol. 19, no. 11, pp. 825–827, Jun. 2007.
[20] L. B. Soldano and E. C. M. Pennings, “Optical multi-mode interference
devices based on self-imaging: Principles and applications,” J. Lightw.
Technol., vol. 13, no. 4, pp. 615–627, Apr. 1995.
[21] Y. Jiao et al., “Accurate and efficient simulation for silicon-nanowire-
based multimode interference couplers with a 3-D finite-element mod e-
propagation analysis,” JOSA B, vol. 27, no. 9, pp. 1813–1818, Sep.
2010.
[22] E. Pennings et al., “Ultracompact, all-passive optical 90 hybrid on
InP using self- imaging,” IEEE Photon. Technol. Lett.,vol.5,no.6,
pp. 701–703, Jun. 1993.
[23] M. Boudreau et al., “An integrated InP coherent receiver for 40 and 100
Gb/Sec telecommunication systems,” in Proc. Opt. Fiber Commun.
Conf., 2009, pp. 1–3.
[24] S. H. Jeong and K. Morito, “Optical 90 hybrid with broad operating
bandwidth of 94 nm,” Opt. Lett., vol. 34, no. 22, pp. 3505–3507, Nov.
2009.
[25] R. Halir et al., “High-performance 90 hybrid based on a silicon-on-
insulator multimode interference coupler,” Opt. Lett.,vol.36,no.2,pp.
178–180, 2011.
[26] C. R. Doerr et al., “Monolithic polarization and phase diversity
coherent receiver in silicon,” J. Lightw. Technol., vol. 28, no. 4, pp.
520–525, Feb. 2010.
Author biographies not included at authors’ request due to space
constraints.
Authorized licensed use limited to: Huazhong University of Science and Technology. Downloaded on March 10,2020 at 02:00:00 UTC from IEEE Xplore. Restrictions apply.
