All-Subwavelength Grating Compact Three-Mode (De) Multiplexer Using Multimode Interference Coupler: Design and Simulation
All-Subwavelength Grating Compact Three-Mode (De) Multiplexer Using Multimode Interference Coupler: Design and Simulation
Abstract. We demonstrate the performance of broadband, low loss, and compact three-mode
converter and (de)multiplexer utilizing an asymmetric trijunction splitter, multimode interferom-
eter, and a phase shifter; all are designed using on subwavelength grating technology (SWG) and
built-in a silicon-on-insulator platform for the first time to the best of our knowledge. SWG
enables the flexibility of refractive index engineering for more operating bandwidth and com-
pactness compared with conventional counterparts. The designed device is capable of converting
and (de)multiplexing three different input modes TE0 , TE1 , and TE2 into a fundamental TE0
output mode. The three-dimensional finite-difference time-domain simulation results confirm
that the designed device has a low loss of 1 dB and the crosstalk reaches −65 dB over a broad
wavelength of 160 nm (1470 to 1630 nm) in the C-band range with an overall size of the
designed device is 95 × 4 μm2 . © 2021 Society of Photo-Optical Instrumentation Engineers (SPIE)
[DOI: 10.1117/1.OE.60.8.086107]
Keywords: all-subwavelength grating; three-mode (de)multiplexer; mode conversion; MMI
coupler; Rsoft photonics CAD suit.
Paper 20210618 received Jun. 11, 2021; accepted for publication Aug. 6, 2021; published online
Aug. 24, 2021.
1 Introduction
The tremendous growth in data traffic flowing through optical networks leads to the capacity of
current systems reaching the Shannon limit; and thus to meet communication demands, expan-
sion of current systems is needed.1 Optical network capacity is expanded through several multi-
plexing schemes to introduce extra links, such as wavelength-division multiplexing (WDM),
polarization-division multiplexing, mode-division multiplexing (MDM), and/or a combination
of these schemes.2–13 The rapid development in the silicon photonics and CMOS technology
offers devices with interesting features, such as high light confinement, broadband, compactness,
robustness, reduced losses, and reduced cost, compared with their counterparts used in tradi-
tional optical communication systems. This makes many researches in the past decade to design
and fabricate related all-optical devices.14 In MDM systems, the single-mode devices are modi-
fied to handle high-order modes besides fundamental mode. One of the key elements to process
waveguide modes efficiently through split, convert, and combine them in MDM is the (de)multi-
plexer.15–19 For that purpose, many devices have been designed.20 For example, Y-junctions,21,22
multimode interferometers (MMIs),23–25 adiabatic directional couplers (ADCs),26 directional
couplers (DCs),27 microring resonators (MRRs),28 and polarization beam splitters.29 Most of
these devices are based on the adiabatic transition to ensure no high-order modes are excited
that lead to an increase in their footprint, whereas the others are easy to build in small footprint
but limited in operation bandwidth.30 Among all these devices, the MMI has gained the most
attention due to its attractive performance through wider bandwidth, lower losses, and flexibility
in design tolerance compared with the rest of the devices.31–38 Still, the MMI is limited in its
performance by the fixed refractive index of silicon.39 To avoid the impact of the aforementioned
problem on size and bandwidth of these devices, designers introduced a new degree of freedom
using subwavelength grating technology (SWG), which is based on refractive index manipula-
tion and dispersion engineering of the modes of the optical waveguide leading to compact, low
loss, and broadband devices.40 SWG is a structure consisting of two or more different dielectric
media arranged periodically as grating elements with a period that is much smaller than the
operating wavelength (Λ ≪ λ) of the light propagates through the system. This results in an
overall medium that is treated as a homogeneous metamaterial combining the optical properties
of all included dielectric elements in the composite such as averaged effective index. Thus, SWG
structures offer flatter dispersion and stronger anisotropy compared with conventional ones,
which enables the ability to manipulate the optical response of the medium and control the flow-
ing light in the device by geometrical design.41–45 Light propagation through SWG waveguides is
studied by Bloch–Floquet theory.44 Recently, for the reason of SWG attractive features men-
tioned previously, a huge amount of optical structures has been designed and experimentally
fabricated based on SWG to serve different functions, such as polarization controlling,46,47 beam
splitters,48,49 phase shifters,50 DCs,51–53 MRRs,54 multimode interference couplers, waveguide
crossings, and waveguide bends.55
In this paper, a three-mode all-subwavelength grating (de)multiplexer is designed and simu-
lated for the first time, to improve the performance through achieving wider bandwidth, lower
losses, and more compactness. The proposed device is capable of mode-order conversion and
(de)multiplexing of three different input modes TE0 , TE1 , and TE2 into uniform fundamental
TE0 as output modes. The overall size of the designed device is 95 × 4 μm2 . The simulation
results confirm that the designed device has a low loss of 1 dB and crosstalk reaches
−65 dB over a broad wavelength of 160 nm (1470 to 1630 nm) in the C-band range. The sim-
ulation is done using Rsoft photonics CAD suit with mesh size setting of 1 nm in each of sim-
ulation axes for relatively high computation accuracy and using a perfectly matched layer
boundary conditions.
zero power delivered at other output port under that input mode excitation. Note that the demul-
tiplexer acts as an 1 × N-port device.
The nonlinear optical effects are negligible in the optical components used to build the
demultiplexer. One can use the transfer matrix concept, which relates the complex amplitudes
of the output field to that of the input fields to analyze the performance of the demultiplexer. The
transfer matrices of the input coupler under the n’th mode excitation, phase-shift stage, and MMI
are denoted by T cn, T Φ , and T MMI , respectively. Note that the matrix T cn depends on the order n
of the input mode, where n stands for TEn mode. The device is designed by cascaded connection
of 1 × 3 trijunction coupler where the input port is derived by a superposition of TE0, TE1, and
TE2 modes, and 3 × 3 output MMI coupler with phase shifter stage placed between them as
shown in Fig. 2. The transfer matrices of the coupler corresponding to the three modes as given,
respectively, as56
1 1 1
T c0 ¼ pffiffiffi ½ j
EQ-TARGET;temp:intralink-;e001;116;419 0 j T ; T c1 ¼ pffiffiffi ½ j 0 −j T ; T c2 ¼ pffiffiffi ½ 0 1 0 T : (1)
2 2 2
The subscriptions 0, 1, and 2 are used with matrix T c to distinguish the three input modes
π π
TE0 , TE1 , and TE2 , respectively. Note that −j ¼ e−j2 and j ¼ ej2 according to Euler’s identity.
The effect of phase accumulation of the optical field of the guided modes when propagating
through the input coupler is neglected in Eq. (1), and this effect will be introduced later.
Investigating these equations yield the following findings:
i. The trijunction coupler guides the input TE2 mode directly through the central input port to
the central output port.
ii. The trijunction coupler guides TE0 (TE1 ) mode only along with the outer ports with equal
amplitudes and is in phase (out of phase).
From the symmetric interference mechanism, the self-imaging to the central line of the MMI
coupler will occur at axial distances that are equal to positive integer times characteristics length
Lc ¼ 3Lπ ∕4.32 Thus, the signal at the central input port will be guided continuously to the central
output port of the MMI coupler when its length is LMMI ¼ 2Lc ¼ 3Lπ ∕2. It is clear from the
previous discussion that when Φ2 ¼ 0 and LMMI ¼ 3Lπ ∕2, the input TE2 mode directed to the
central output port of the MMI coupler without introducing any crosstalk with the two other
output ports of the demultiplexer. This leaves the MMI coupler to act as a 2 × 2 coupler cor-
responding to the two outer input/output arms of the original MMI and has the following transfer
matrix:56
1 j j j 1 1
T MMIð0;1Þ ¼ pffiffiffi ¼ pffiffiffi : (2)
2 −j j 2 −1 1
EQ-TARGET;temp:intralink-;e002;116;139
When Φ0 and Φ1 phases are introduced at the two outer inputs of the trijunction coupler,
Eq. (2) is modified to
Fig. 3 A simplified model for the three-mode demultiplexer when excited by TE0 and TE1 modes.
jΦ
ejΦ0 0 j e 0 ejΦ1
T0 ¼ T MMIð0;1Þ :T Φ ¼ T MMIð0;1Þ ¼ pffiffiffi : (3)
MMIð0;1Þ
0 ejΦ1 −e jΦ0
ejΦ1
EQ-TARGET;temp:intralink-;e003;116;636
Similarly, the trijunction coupler can be treated as a 1 × 2 coupler corresponding to the input/
two outer output transmission. Therefore, the transmission matrix defined for TE0 and TE1 in
Eq. (1) can be modified to
0 1 j 0 1 j
T c0 ¼ pffiffiffi ; T c1 ¼ pffiffiffi : (4)
j −j
EQ-TARGET;temp:intralink-;e004;116;563
2 2
According to that, the demultiplexer configuration shown in Fig. 2 can be modeled as shown
in Fig. 3, when the input is excited by TE0 and TE1 modes.
The outputs of the two outer ports of the demultiplexer can be calculated as
..
Eo;0 0 . 0 Ei;0
¼ T MMIð0;1Þ T c0 T c1 : (5)
Eo;1 .. Ei;1
EQ-TARGET;temp:intralink-;e005;116;481
The analysis is taken further to assess the two output fields of the 2 × 2 MMI under the
following special cases:
• Case(a): demultiplexer input mode ¼ TE0
j h jΦ0 i
e þ ejð2þΦ1 Þ ;
π
Eo;0 ¼
EQ-TARGET;temp:intralink-;e006;116;382 (6)
2
jh i
¼ −ejΦ0 þ ejð2þΦ1 Þ :
π
Eo;1
EQ-TARGET;temp:intralink-;e007;116;329 (7)
2
The phase shifter is inserted at the output of the first outer arm of the trijunction to yield a
phase shift of π∕2. Inserting Φ0 ¼ π∕2 and Φ1 ¼ 0 in Eqs. (6) and (7) yields
j h jπ i j
EQ-TARGET;temp:intralink-;sec2.1;116;265
π
Eo;0 ¼ e 2 þ ej2 ¼ ½2j ¼ −1;
2 2
j h jπ j π
i
Eo;1 ¼ −e 2 þ e 2 ¼ 0:
2
Thus, under the condition of Φ0 ¼ π∕2 and Φ1 ¼ 0, the input TE0 mode appears only at one
of the output ports of the 3 × 3 MMI coupler.
• Case(b): demultiplexer input mode ¼ TE1
j h jΦ0 i
e þ ejð−2þΦ1 Þ ;
π
Eo;0 ¼
EQ-TARGET;temp:intralink-;e008;116;153 (8)
2
jh i
¼ −ejΦ0 þ ejð−2þΦ1 Þ ;
π
Eo;1
EQ-TARGET;temp:intralink-;e009;116;100 (9)
2
j h jπ i
EQ-TARGET;temp:intralink-;sec2.1;116;735
π
Eo;0 ¼ e 2 þ e−j2 ¼ 0;
2
jh π π
i
Eo;1 ¼ −ej2 þ e−j2 ¼ 1:
2
Therefore, under this condition, the input mode TE1 is directed toward the other outer port of
the demultiplexer. Important remarks related to the previous analysis are given below
i. When Φ0 ¼ −π∕2 and Φ1 ¼ 0, Eqs. (6) and (7) become Eo;0 ¼ 0 and Eo;1 ¼ 1, respec-
tively, when TE0 is excited at the demultiplexer input. Similarly for TE1 mode, Eqs. (8)
and (9) yield the two outputs Eo;0 ¼ 1 and Eo;1 ¼ 0, respectively.
Thus, Φ0 ¼ −π∕2 and Φ1 ¼ 0 condition is also useful to separate the input TE0 and TE1
modes into two individual output ports of the 3 × 3 MMI coupler.
ii. Considering the case of Φ0 ¼ 0 and Φ1 ¼ π∕2 (i.e., the π∕2 phase shifter is placed at the
other outer port of the trijunction). Under these phase shifters, Eqs. (6) and (7) become
Eo;0 ¼ 0 and Eo;1 ¼ −1, respectively. Also, Eqs. (8) and (9) become Eo;0 ¼ j and
Eo;1 ¼ 0, respectively.
iii. When the accumulation phase θa of the optical field of the guided modes propagating
through the input 1 × 3 coupler is taken into account, then the transfer matrix of this coupler
is modified to
0
T cn
EQ-TARGET;temp:intralink-;sec2.1;116;475 ¼ T cn ejθa for n ¼ 0;1; and 2:
This makes each of the demultiplexer output fields is multiplied by the factor ejθa. Thus, the
phase θa does not alter the function of the demultiplexer but introduces an extraphase shift for all
the output fields.
a 2 Λ−a 2
n2xx ¼ :n þ n2 ; (10)
Λ 1 Λ
EQ-TARGET;temp:intralink-;e010;116;303
a −2 Λ − a −2
n−2
zz ¼ n þ n2 ; (11)
Λ 1 Λ
EQ-TARGET;temp:intralink-;e011;116;249
where nxx and nzz are the parallel and orthogonal components of the effective index of the new
resultant medium, respectively. Λ is the pitch length of the structure and is chose Λ ≪ λ, a is the
length of a single segment of the structure, n1 and n2 are the actual refractive indices of the
materials forming the SWG structure.57 In multimode waveguides design, the subwavelength
grating incorporation effect is clear in calculating the beat length of the MMI coupler is given
in the following equation:
4W 2eff n2zz
Lπ ¼ ; (12)
3λ nxx
EQ-TARGET;temp:intralink-;e012;116;663
where λ is the operating wavelength, and W eff is the effective width of the MMI coupler
λ 1
W eff ¼ W MMI þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; (13)
π n2eff þ n2cladd
EQ-TARGET;temp:intralink-;e013;116;607
W MMI is the physical width of the multimode region, neff and ncladd are the effective and cladding
refractive indices, respectively.58 The effective index in a wide SWG waveguide is
nxx λ2
neff ¼ nxx − : (14)
8W 2eff n2zz
EQ-TARGET;temp:intralink-;e014;116;537
For an SWG-based phase shifter, the effect of the structure can be expressed in terms of the
phase difference as a function of refractive index components of the periodic medium nxx and nzz
as39
πλ nxx 1 1
Φ¼ − L ; (15)
4 n2zz W 2ps W 2s ps
EQ-TARGET;temp:intralink-;e015;116;454
where W ps , W s , and Lps are the width of phase shifter waveguide, straight reference waveguide,
and length of the phase shifter, respectively.
(a) The pitch ΛT duty cycle DCT who are related as follows:
DCT ¼ aT ∕ΛT ;
EQ-TARGET;temp:intralink-;e016;116;255 (16)
ΛT values but DCT must be over the range from 70% to 100%. According to the simulation
results, it is clear that at ΛT ¼ 160 nm and DCT ¼ 70% gives the required function with better
performance.
The next step is choosing the waveguide widths. Recalling that these values must be chosen
to ensure that TE0 (TE1 ) mode is guided through the outer branches, whereas TE2 mode is
guided through the central branch only. Starting with the input stem of width W in , which needs
to be wide enough so it can handle the three input modes, thus based on the fact presented in
Fig. 2 it is suitable to choose W i > 110 nm. Also to mention that the since the trijunction will be
connected directly to the MMI coupler that specifically operates under the fundamental mode, it
is possible to set the outer branches to W o ¼ 500 nm, whereas for the central to W c ¼ 250 nm to
ensure output is observed from the central port of the trijunction only. This is confirmed through
simulating the effect of width and the results are shown in Fig. 8, which depicts the variation of
the trijunction output in response to the change of its arms width. Thus, it can be concluded that a
suitable input stem width is 1.25 μm.
The length of each arm Lb is set to values at which maximum transmission and compactness
are satisfied. Figure 9 shows the results of the simulation of trijunction output as a function of
arm length for each case of input modes. In this figure, the length is varied from 5 to 30 μm,
whereas the output of the trijunction is monitored. It is obvious that the optical field level for both
the fundamental mode TE0 (TE1 ) and TE2 modes changes with opposite behavior. Thus, each
optical mode is guided correctly to the assigned arm and no output on the other ports is detected.
Moreover, for TE2, the central port’s output field level is twice that in the case of TE0 ∕TE1 . In
this design, it is set to Lb ¼ 20 μm, since at this value the requirements mentioned earlier of
compactness and good performance can be satisfied.
Fig. 10 Phase shift difference variation with the width W ps and length Lps .
Fig. 11 MMI coupler output variation in response to the change in coupler length LMMI .
it is suitable to set W ps , and Lps to 0.3 and 7 μm, respectively. Another set of parameters may be
chosen to give the required phase shift. For example, W ps and Lps to 0.21 and 5 μm, respectively,
or W ps and Lps to 0.31 and 3 μm, respectively, and so on.
where N is the number of input (output) ports and i indicates each port I ¼ 1;2; : : : ; N (N ¼ 3
for the proposed design). These access ports are tapered to ensure maximum field injected into
and extracted from the coupler. The length MMI section LMMI is simulated and optimized for the
desired function where the MMI coupler acts as a 3-dB splitter when the field is injected from
either of the outer ports while transferring the field directly to the central output port when
injected from the central input port. Revisiting the self-imaging principle, the length of the
MMI section needs to be set at a value where double images of the input field are obtained
at the output end for the outer ports and a single image is obtained at the central output port.
Based on simulation results presented in Figs. 11 and 12 which shows the MMI coupler output
variation with the change in coupler length and the intensity distribution along the coupler for
each case of input ports, respectively. These figures confirm that the MMI coupler functions
described earlier can be satisfied at length LMMI ¼ 45 μm (excluding access waveguides length).
Also, this value matches the theoretical value calculated from Eq. (23).
4 Performance Verification
Through Rsoft photonics CAD suit, we verify the theory and evaluate the optical conversion
performance of the proposed device. A summary of the main parameters used in the simulation
is listed in Table 1. Figure 13 shows the operation corresponding to the three cases of the con-
sidered modes TE0 , TE1 , and TE2 at the input port of the proposed device where they propagate
and demultiplexed along with the device and guided to the corresponding output port success-
fully. Also, it is worthy to noteFl that the device can operate as a multiplexer recovering the same
modes from injecting the fundamental mode TE0 as input mode as shown in Fig. 14.
Fig. 12 Intensity distribution along the SWG-MMI coupler: (a) upper edge, (b) lower edge, and
(c) center; values on both axes are in μm.
Fig. 13 Intensity distribution of the guided mode along the demultiplexer when the input is (a) TE0
mode, (b) TE1 mode, and (c) TE2 mode; scales of both axes are in μm.
Fig. 14 Intensity distribution of the guided mode along the device when it works as a multiplexer to
obtain: (a) TE0 mode, (b) TE1 mode, and (c) TE2 output mode; scales of both axes are in μm.
This is confirmed by tracking mode transverse cross-section profile at the input and output
ends for both operation (de)MUX/MUX as shown in Fig. 15.
For deep evaluation of the performance of the proposed device, it is recommended to con-
sider computational approach to assess the insertion loss (I.L) and crosstalk (C.T) parameters for
each case of input modes, the equation for each parameter has the following form:
ðPout Þdesired
I:L ðdBÞ ¼ −10 log10 ; (18)
Pin
EQ-TARGET;temp:intralink-;e018;116;153
ðPout Þdesired
C:T ðdBÞ ¼ −10 log10 ; (19)
ðPout Þundesired
EQ-TARGET;temp:intralink-;e019;116;97
Fig. 15 Transverse field profile of (a) demultiplexer and (b) multiplexer. The first and second col-
umns represent the input and output ports, respectively.
ðPout ÞON
E:R ðdBÞ ¼ 10 log10 ; (20)
ðPout ÞOFF
EQ-TARGET;temp:intralink-;e020;116;735
where Pin , ðPout Þdesed , and ðPout Þundesired , ðPout ÞON , ðPout ÞOFF are the input power to the device,
power emitted from the desired output port, power emitted from the other undesired ports, power
emitted from the output port in the ON state, and the power emitted from the desired output port
Fig. 17 Crosstalk per port of the proposed demultiplexer when excited by: (a) TE0 , (b) TE1 , and
(c) TE2 .
Parameter TE0 mode (Port3 ) TE1 mode (Port2 ) TE2 mode (Portcentral )
in the OFF state, respectively. Figures 16 and 17 show the calculated insertion loss and crosstalk
as a function of the operating wavelength of the input mode of the range 1470 to 1630 nm, the
insertion loss is −3 dB, and the bandwidth is 160 nm. This is to be compared with 40 nm
reported in Ref. 56 for a conventional counterpart implemented without SWG technology.
From Figs. 16 and 17, it can be seen that the insertion loss varies from 1 to 4 dB, and the crosstalk
varies from −65 to −12 dB. At the central wavelength 1550 nm, the modes TE0 and TE1 have
the lowest losses <1.4 dB, whereas TE2 has a broadband low losses range reaching <1 dB. The
extinction ratio of the MMI coupler is estimated from the simulation results using the following
Fig. 18 Performance variation as a function of MMI width fabrication tolerance: (a) insertion loss
and (b) crosstalk.
procedure. One of the input ports is fed with 1550 nm signal and then the power at the required
output ports is recorded yielding to ðPout ÞON . The wavelength of the input signal is scanned to
yield a minimum output power at the required output port, i.e., ðPout ÞOFF . The extinction ratio is
calculated as Eq. (20). Simulation results yield extinction ratio of 32 and 31 dB, and port3 and
central port are used, respectively. Although we aware that for mode converter and demux devi-
ces, the insertion loss of <1 dB is desirable, yet, the loss can be compensated by slightly increas-
ing the gain of the booster or in-line optical amplifiers amplifier used in the optical
communication system. Table 2 gives the insertion loss, and the crosstalk obtained from each
input mode at the operating wavelength of 1550 nm.
Some defects during the process of device manufacturing can occur, which comes from the
resolution limit of the purity of supplied SOI wafer and the fabrication technology. Thus, the
fabrication tolerance study for the main parameters of the design is necessary. The impact of
fabrication tolerance on the device performance is studied through monitoring the performance
variation with the change of the main parameters as shown in Figs. 18–20.
Fig. 19 Performance variation as a function of MMI pitch length fabrication tolerance: (a) insertion
loss and (b) crosstalk.
Fig. 20 Design performance as a function of phase shifter width fabrication tolerance: (a) insertion
loss and (b) crosstalk.
5 Conclusions
A three-mode demultiplexer comprises a trijunction splitter, a π∕2 phase shifter, and a 3 × 3
MMI coupler, and all parts have been designed utilizing SWG technology for the first time.
Consequently, a compact, low-loss, and three times wider operating bandwidth compared with
the equivalent conventional DE/MUX has been obtained. Full 3D simulations using Rsoft pho-
tonics CAD suit have been conducted to evaluate the performance of the proposed device, which
shows the insertion loss of 1 dB, and crosstalk reaches −65 dB over a broad wavelength of
160 nm (1470 to 1630 nm) in the C-band range with an overall size of the designed device
Experimental
Simulation readings readings
Ref# Structure Bandwidth (nm) C.T (dB) I.L (dB) C.T (dB) I.L (dB) dimensions (μm) Approach
086107-18
61 Subwavelength grating adiabatic coupler 1260 to 1360, 1500 to 1630 <−18.5 <0.32 <−18.8 <2.6 55 × 1.15 Both
62 Angled multimode interference (de)multiplexer 1530, 15,475, and 1565 −12.67 2.44 — — 3697 × 20.5 Theoretical
63 Mode converter and demultiplexer 1530 to 1570 −15 <1 — — 43.7 × 2.7 Theoretical
64 Three-mode multiplexer and demultiplexer 1530 to 1560 −17.40 0.37 −17.40 8.51 5500 × 46.8 Both
This work All-subwavelength grating (de)multiplexer 1470 to 1630 <−65 <1.38 — — 4 × 95 Theoretical
August 2021
•
Hussain and Fyath: All-subwavelength grating compact three-mode (de)multiplexer using multimode. . .
Vol. 60(8)
Hussain and Fyath: All-subwavelength grating compact three-mode (de)multiplexer using multimode. . .
is 95 × 4 μm2 . Moreover, the designed device offers overall good performance even with fab-
rication imperfections. Finally, the presented results pave the way for compact MDM devices
that can be used in silicon photonic integrated circuits and WDM to expand the bandwidth
capacity.
Acknowledgments
No potential conflict of interest was reported by the authors.
References
1. Y. Liu et al., “Research on a novel mode division multiplexer with low crosstalk, low loss
and few mode ring-core transmission channel,” Opt. Commun. 469, 125778 (2020).
2. V. H. Nguyen, I. K. Kim, and T. J. Seok, “Silicon photonic mode-division reconfigurable
optical add/drop multiplexers with mode-selective integrated MEMS switches,” Photonics
7(4), 80 (2020).
3. G. H. Chen et al., “Compact mode division MUX/DEMUX using enhanced evanescent-
wave coupling on silicon-on-insulator (SOI) platform for 11-Tbit/s broadband transmis-
sion,” IEEE Access 8, 219881–219890 (2020).
4. H. T. Duy et al., “1 × 2 switchable dual-mode optical 90° hybrid device based on thermo-
optic phase shifters and 2 × 2 MMI couplers on SOI platform,” IEEE Photonics J. 13(1),
661016 (2021).
5. S. A. Srinivasan et al., “56 Gb/s NRZ O-band hybrid BiCMOS-silicon photonics receiver
using Ge/Si avalanche photodiode,” J. Lightwave Technol. 39 (5), 1409–1415 (2021).
6. Q. Wilmart et al., “A complete Si photonics platform embedding ultra-low loss waveguides
for O- and C-band,” J. Lightwave Technol. 39 (2), 532–538 (2021).
7. D. Zou et al., “Beyond 1.6 Tb/s net rate PAM signal transmission for rack-rack optical
interconnects with mode and wavelength division multiplexing,” J. Lightwave Technol.
39(2), 340–346 (2021).
8. J. Gao et al., “Elliptical-core highly nonlinear few-mode fiber based OXC for WDM-MDM
networks,” IEEE J. Sel. Top. Quantum Electron. 27(2), 7600511 (2021).
9. D. Kakati, M. Minz, and R. K. Sonkar, “Performance analysis of grating-assisted passive
mode-division multiplexing device using silicon photonics for high-speed RoF/RoFSO
communication,” Opt. Eng. 60 (1), 016102 (2021).
10. M. Singh and J. Malhotra, “Performance comparison of high-speed long-reach mode divi-
sion multiplexing-based radio over free space optics transmission system using different
modulation formats under the effect of atmospheric turbulence,” Opt. Eng. 58(4),
046112 (2019).
11. Y. Gao et al., “Polymer mode selecting switch based on cascaded MMI couplers, IEEE
Photonics Technol. Lett. 33(3), 147–150 (2021).
12. S. Y. Siew et al., “Review of silicon photonics technology and platform development,”
J. Lightwave Technol. 39, 4374–4389 (2021).
13. Z. Zheng et al., “Ultra-compact reconfigurable device for mode conversion and dual-mode
DPSK demodulation via inverse design,” Opt. Express 29(12), 17718 (2021).
14. M. Moradi et al., “Design and simulation of a fast all-optical modulator based on photonic
crystal using ring resonators,” Silicon 13(1), 1–7 (2021).
15. K. Mehrabi and A. Zarifkar, “Ultracompact and ultrabroadband mode division multiplexer
based on an Au nanocube array assisted directional coupler,” Appl. Opt. 59(5), 1286 (2020).
16. G. Rademacher et al., “Space-division multiplexed transmission in the S-band over 55 km
few-mode fibers,” Opt. Express 28(18), 27037–27043 (2020).
17. A. Kaushalram et al., “Ultra-broadband fabrication-tolerant mode division (de) multiplexer
on thin film lithium niobate,” Opt. Commun. 475, 126251 (2020).
18. Z. Chen et al., “Ultra-compact broadband in-line mode converter based on a width-
modulated silicon converter based on a width-modulated,” IEEE Photonics J. 13(2),
2700107 (2021).
19. H. Xu, D. Dai, and Y. Shi, “Silicon integrated nanophotonic devices for on-chip multi-mode
interconnects,” Appl. Sci. 10, 6365 (2020).
20. X. Tu et al., “State of the art and perspectives on silicon photonic switches,” Micromachines
10, 51 (2019).
21. B. Paredes et al., “Dual-band (O & C-bands) two-mode multiplexer on the SOI platform,”
IEEE Photonics J. 13, 6600309 (2021).
22. D. Q. Duy et al., “Polarization-insensitive two-mode (de)multiplexer using silicon-on-
insulator-based Y-junction and multimode interference couplers,” Opt. Eng. 58(6),
067105 (2019).
23. D. D. Tu et al., “Optical mode conversion based on silicon-on-insulator material Ψ -junction
coupler and multimode interferometer,” Opt. Laser Technol. 142, 107177 (2021).
24. X. Zhang et al., “Impact of indirect mode coupling on planar lightwave circuit based mode
converters/multiplexers,” Opt. Commun. 465, 125608 (2020).
25. D. Dai et al., “Multimode silicon photonic devices,” Proc. SPIE 11689, 116890N (2021).
26. W. U. Zezhi, “The research progress in on-chip mode (de) multiplexer,” IOP Conf. Ser.
Earth Environ. Sci. 332, 42008 (2019).
27. Y. Su et al., “Silicon photonic platform for passive waveguide devices: materials, fabrica-
tion, and applications,” Adv. Mater. Technol. 5, 1901153 (2020).
28. W. Chen et al., “Silicon mode (de) multiplexer based on cascaded particle-swarm-optimized
counter-tapered couplers,” IEEE Photonics J. 13, 6600210 (2020).
29. H. Xu et al., “Proposal for an ultra-broadband polarization beam splitter using an
anisotropy-engineered Mach–Zehnder interferometer on the x-cut lithium-niobate-on-insu-
lator,” Opt. Express 28(8), 10899–10908 (2020).
30. S. Farhadi, M. Miri, and A. Alighanbari, “Design and simulation of a compact and ultra-
wideband polarization beam splitter based on sub-wavelength grating multimode interfer-
ence coupler,” Appl. Phys. B 126, 118 (2020).
31. H. D. T. Linh et al., “Arbitrary TE0/TE1/TE2/TE3 mode converter using 1 × 4 Y-junction
and 4 × 4 MMI couplers,” IEEE J. Sel. Top. Quantum Electron. 26(2), 8300708
(2020).
32. J. Yao et al., “High performance optical 90° hybrid at 2 μm wavelength range based on 2 × 4
MMI using Si3 N4 technology,” Proc. SPIE 11617, 1161723 (2021).
33. E. Samoi, Y. Benezra, and D. Malka, “An ultracompact 3 × 1 MMI power-combiner
based on Si slot-waveguide structures,” Photonics Nanostruct.-Fundam. Appl. 39, 100780
(2020).
34. Z. Wang et al., “Compact silicon three-mode multiplexer by refractive-index manipulation
on a multi-mode interferometer,” Opt. Express 29(9), 13899–13907 (2021).
35. C. Sun et al., “Key multimode silicon photonic devices inspired by geometrical optics,” ACS
Photonics 7, 2037–2045 (2020).
36. H. D. T. Linh et al., “Mode switch for on-chip optical interconnects using multimode inter-
ference couplers,” in 6th NAFOSTED Conf. Inf. Comput. Sci., pp. 98–102 (2019).
37. C. D. Truong et al., “On-chip silicon photonic controllable 2 × 2 four-mode waveguide
switch,” Sci. Rep. 11, 897 (2021).
38. K. T. Ahmmed, H. P. Chan, and L. Binghui, “Three-mode multiplexer and demultiplexer
based on the Mach–Zehnder interferometer,” OSA Continuum 4(5), 1519–1532 (2021).
39. L. Sun et al., “Subwavelength structured silicon waveguides and photonic devices,”
Nanophotonics 9(6), 1321–1340 (2020).
40. D. González-Andrade et al., “Experimental demonstration of a broadband mode converter
and multiplexer based on subwavelength grating waveguides,” Opt. Laser Technol. 129,
106297 (2020).
41. I. Molina-Fernández et al., “Subwavelength grating metamaterial structures for integrated
photonics,” Proc. SPIE 11284, 112840K (2020).
42. W. Zhou et al., “Subwavelength engineering in silicon photonic devices,” IEEE J. Sel. Top.
Quantum Electron. 25(3), 2900113 (2019).
43. D. Zhu et al., “High-contrast and compact integrated wavelength diplexer based on subwa-
velength grating anisotropic metamaterial for 1550/2000 nm,” IEEE Photonics J. 13(2),
6600910 (2021).
Zaid Lateef Hussain received his BSc and MSc degrees in laser and optoelectronics engineering
from Al-Nahrain University, Baghdad, Iraq, in 2002 and 2005, respectively. He is currently
working toward his PhD in laser/electronics and communication engineering at the Institute
of Laser for Postgraduate Studies-University of Baghdad. His current research focuses on optical
communications.
Raad S. Fyath received his BSc degree in electrical engineering from the University of
Basrah, Iraq, in 1976, his MSc degree in electronics and communications engineering from the
University of Baghdad, Iraq, in 1987, and his PhD in electronics engineering from the University
of Wales-Bangor, United Kingdom, in 1990. Currently, he is a professor of electronics and com-
munications engineering at the College of Engineering, Al-Nahrain University, Baghdad, Iraq.
His research interests include optical and wireless communications, optoelectronics, and nano-
photonics. He published more than 100 papers in different scientific journals and conference
proceedings.