described above, which is due to the significant
coupling which takes place in the input/output bent
waveguide sections.
3 FABRICATION AND TEST
Test structures were fabricated at CEA-LETI on
200mm wafers with a SOI/BOX thickness of
310/800nm. Waveguides were defined using 193nm-
DUV lithography and a dry etching process (B.
Szelag, 2016).
Figure 4: Measured insertion losses of best performing
fabricated design variant. Solid lines are fit lines to the data
(points), dashed lines are 3D-FDTD simulations of the
nominal structure.
Test structures with a range of nominal waveguide
widths and coupling section lengths were fabricated.
Each variant was triplicated and connected to vertical
fibre grating couplers designed to couple
1.31µmTE/1.55µmTE/1.55µmTM light into the
waveguides. In order to evaluate the insertion losses
for each channel, a reference waveguide was
measured with each type of fibre coupler. The ‘cross’
transmission (corresponding to light crossing from
one guide to another) and the ‘bar’ transmission
(corresponding to light propagation along the same
waveguide) of each variant was then measured for
each channel. The directional coupler transmission
spectra were then obtained by subtracting the
corresponding fibre coupler spectra from the
directional coupler transmission data.
The transmission spectra of the best performing
variant (rib width = 320nm, gap = 200nm, coupling
section length = 7µm) are shown in figure 4. The data
points are shown as well as polynomial fits (solid
lines). The high noise level of the data can be
attributed to the summing of the noise in the
transmission spectra and the reference spectra. The
measured insertion losses for the
1.31µmTE/1.55µmTE/1.55µmTM channels were -
0.6/-0.55/-1.2dB. The measured -1dB bandwidth is in
the region of 80nm. The use narrowband fibre grating
couplers prevented crosstalk measurements, although
FDTD simulations show values in the -10 to -15dB
range.
For comparison, the 3D-FDTD simulated
transmission spectra for this nominal structure are
also shown as dashed lines on figure 4. The observed
discrepancy as well as the fact that the best
performing variant (rib width = 320nm, L = 7µm) was
not exactly that identified via FDTD simulation may
be firstly attributed to the unsimulated optical
coupling in the waveguide routing required to access
the test structure and secondly, to departures from the
nominal waveguide geometry in the processed
devices.
4 CONCLUSIONS AND
PERSPECTIVES
We have designed, fabricated and measured an
integrated silicon low loss O-band/C-band duplexer
with polarisation insensitivity at 1550nm. The device
shows low measured losses (<1.3dB) at each design
wavelength. Further design refinements will likely
reduce insertion losses significantly below 1dB,
which compares favourably with conventional,
discrete optical componentry. Based on this positive
result, we are now fabricating a complete circuit,
including integrated laser, photodiode and broadband
edge coupler to demonstrate the potential of
integrated silicon photonics for this application.
REFERENCES
L. Vivien, J. Osmond, Jean-Marc Fédéli, Delphine Marris-
Morini, Paul Crozat, Jean-François Damlencourt, Eric
Cassan, Y. Lecunff, and Suzanne Laval, "42 GHz p.i.n
Germanium photodetector integrated in a silicon-on-
insulator waveguide," Opt. Express 17, 6252-6257
(2009).
J. Cardenas, C. B. Poitras, K. Luke, L Luo, P.A. Morton,
and M. Lipson. 2014. High Coupling Efficiency Etched
Facet Tapers in Silicon Waveguides, IEEE Photonics
Technology Letters, Vol. 26, No. 23.
K. Okamoto, 2014. Wavelength-Division Multiplexing
Devices in Thin SOI: Advances and Prospects, IEEE
Journal of Selected Topics in Quantum Electronics,
Vol. 20, NO. 4.
G. Roelkens, D. Van Thourhout, and R. Baets. 2007.
Silicon-on-insulator ultra-compact duplexer based on a