Page 18 - Chip Scale Review_March-April_2024-digital
P. 18
deposited onto various bond pad sizes
to determine and maintain a consistent
height for efficient optical coupling to
grating couplers. The translational and
angular tolerances for both the free-form
couplers and tapered waveguides were
determined via simulation.
Simulation
3D finite difference time domain
(FDTD) simulations were performed
to determine the coupling efficiency
(CE) for both presented approaches. The
alignment tolerances of the structure
as a function of translation between
the two chips were calculated. For
the evanescent coupling approach,
si mulat ion result s usi ng A nsys-
Lumerical’s 3D FDTD solver are shown
in Figure 2. The gap length in the Z
direction was input to be 0µm for the
X/Y tolerance results and 500nm for
both tilt and twist tolerance simulations.
The results from the data in Figure
2 show this design has a lateral (Y)
tolerance of ±2.7μm, vertical (Z)
tolerance of 2.8μm, longitudinal (X)
tolerance of over ±150μm, twist rotation
tolerance of ±2.3°, and tilt rotation
tolerance of ±0.4°. For this simulation,
the underfill epoxy refractive index
is assumed to be 1.457 to match the
refractive index of silica being used
[10]. Further simulations indicated the
Z-tolerance can be extended to near
3.75µm if an epoxy with refractive
index closer to 1.5 is used. However, as
the epoxy refractive index increases,
the light becomes less confined in the
waveguides and will require longer
tapers to compensate for the scattering
loss. This simulation data demonstrates
this design has comparable lateral
a nd a ng u la r tole r a nces to t hose
achievable using high-speed pick-and-
place die bonders (approximately ±3-
10µm) and has a vertical tolerance
comparable to back-end-of-line (BEOL)
thicknesses typical of active PICs. This
indicates there is potential use for this
evanescent coupler design in flip-chip
bonded packages.
In the case of chip-to-chip freeform
reflectors, possible misalignments or
tilts that may be introduced during
assembly are necessary to determine as
they are important metrics to evaluate
Figure 2: 3D FDTD simulation results showing: a) coupler parameter definitions where the red is Si and the the practicality of the device. FDTD
blue is Si 3 N 4 , b) the translational alignment tolerance in X/Y/Z, and c) the rotational alignment tolerance where simulations indicate that the reflectors
tilt indicates rotation about the Y-axis and twist indicates rotation about the Z-axis. Reprinted with permission are very tolerant to both translation and
from [1] © The Optical Society
16
16 Chip Scale Review March • April • 2024 [ChipScaleReview.com]