Articles
https://doi.org/10.1038/s41566-018-0132-5
Bridging ultrahigh-
Q
devices and photonic circuits
Ki Youl Yang
1,2
, Dong Yoon Oh
1,2
, Seung Hoon Lee
1,2
, Qi-Fan Yang
1
, Xu Yi
1
, Boqiang Shen
1
,
Heming Wang
1
and Kerry Vahala
1
*
1T. J. Watson Laboratory of Applied Physics, California Institute of Technology, Pasadena, CA, USA. 2These authors contributed equally: Ki Youl Yang,
Dong Yoon Oh and Seung Hoon Lee. *e-mail: vahala@caltech.edu
© 2018 Macmillan Publishers Limited, part of Springer Nature. All rights reserved.
SUPPLEMENTARY INFORMATION
In the format provided by the authors and unedited.
Nature PHO
tONicS
|
www.nature.com/naturephotonics
Supplemental Section: Bridging ultra-high-Q devices and photonic circuits
Ki Youl Yang*, Dong Yoon Oh*, Seung Hoon Lee*, Qi-Fan Yang, Xu Yi, Boqiang Shen, Heming Wang,
and Kerry Vahala
†
T. J. Watson Laboratory of Applied Physics, California Institute of Technology, Pasadena, California 91125, USA.
*These authors contributed equally to this work.
†
Corresponding author: vahala@caltech.edu
I. INTRINSIC CAVITY Q FACTOR MEASURED FROM 1520-1560 NM
Fig. S1a shows a plot of the measured intrinsic cavity Q as a function of wavelength (extended data of Fig. 2 in main
text). The Q factors were measured by cavity ring-down traces of the fundamental TM/TE modes, and Fig.S1b-c
shows the superposition of multiple cavity ring-down scans for the fundamental TM mode with the corresponding
decay time and loaded and intrinsic Q factors. The measured Q factors are in the range 214 - 230 million and 114 -
130 million for the fundamental TM and TE, respectively.
FIG. S1:
Intrinsic cavity Q as a function of wavelength.
(
a
) Intrinsic Q factors of the integrated ridge res-
onator measured versus wavelength. The Q factors of the fundamental TM and TE modes are indicated (black cir-
cle: TM0, red circle: TE0). The Q factors were measured from cavity ring-down traces. (
b
) Superposition of mul-
tiple cavity ring-down scans for the fundamental TM mode at 1520 nm. The red line is an exponential fit, and the
corresponding decay time and Q factors are indicated in the panel. (
c
) As in
b
but for the fundamental TM mode
at 1540 nm. The ring-down scans at 1560 nm were provided in the manuscript.
II. INVESTIGATION OF CAVITY LOSS MECHANISM
PECVD SiN is deposited over the entire disk and then removed in the fabrication process (Fig.1d in main text).
This material often has an absorption peak in the 1500 nm band due to the residual N-H bonds. Therefore, any
SiN residue on the cavity may cause Q degradation. Accordingly, we processed a phosphoric acid etching for slightly
longer than the required duration for the film removal in order to ensure complete eliminatation of the PECVD SiN
layer. Phosphoric acid was chosen due to its excellent etching selectivity, and negligible impact on silica Q factor.
2
FIG. S2:
Microfabrication process for silica ridge resonators.
Two variations of the fabrication process were
studied to create silica ridge resonators without the SiNx waveguide coupler. The first process (upper) proceeds
without the SiNx process (PECVD deposition and SiNx etching). In this case, the thermal oxidation process on the
etched silica layer was followed by the additional lithography, wet etch of the silica, and then XeF
2
etching to re-
move the silicon through the ring aperture. The second process (below) includes the PECVD silicon nitride process,
however the entire PECVD SiNx layer was removed by RIE and phosphoric acid etchings so that the waveguide
was not present in the final structure.
The effects of the SiNx deposition and removal process on Q factor were studied by fabricating two additional types
of structures shown in Figure S2: (1) silica ridge resonators without the silicon nitride step (i.e., only silica ridge is
fabricated); (2) with the SiNx step, but with the SiNx completely removed. The intrinsic Q factors of these structures
were measured by using a tapered fiber coupler. Both sets of Q factors showed consistent results with those measured
from the waveguide integrated device, thereby confirming that any parasitic losses from the nitride residue and the
waveguide coupler are not a significant contribution to loss. The low waveguide-induced parasitic losses (i.e., not
coupling related) might be related to the already very weak interaction between waveguide and resonator required to
obtain critical coupling.
III. WAVEGUIDE-RESONATOR COUPLING
FIG. S3:
Waveguide-resonator coupling as a function of pulley waveguide length.
(
a
) Schematic of pul-
ley version waveguide and ridge resonator.
R
wg
=
R
ridge
+
g
+
w/
2 is the bending radius of pulley waveguide,
and the waveguide is wrapped around the ring over
L
pully
. (
b
) Waveguide transmission coupling to the fundamen-
tal TE resonator mode. The dashed line is the numerically calculated transmission coupling
S1–S3
, and the dots are
experimental results. Lower coupling at longer pulley length is caused by over-coupling. (
c
) As in
b
but for the fun-
damental TM resonator mode. Smaller mode field overlap between the TM (hybrid) resonator mode and TE waveg-
uide mode requires longer pulley length to achieve coupling comparable to the TE fundamental resonator mode.
As noted in the main text and Fig. S3, we have implemented the pulley version of the waveguide to control the
coupling of the resonator and waveguide. In comparison to the resonator circumference (which is over a centimeter)
the pulley is relatively short. The waveguide-pulley coupling was studied both numerically and by measurement of
3
FIG. S4:
Wavelength dependence of coupling.
Measured waveguide to resonator coupling using the fundamen-
tal TE mode in the resonator and a 200 micron long pulley length. Unity coupling indicates critical coupling.
coupling in structures having various pulley lengths. Fig. S3 shows the waveguide transmission coupling to modes as
a function of pulley waveguide length measured for two polarizations. The numerical predictions (dashed lines in the
panels) are in a close agreement with the measurement results (dots). To calculate the coupling we have used a mode
field overlap calculation presented elsewhere
S1–S3
. It is important to note that the lower transmission coupling to the
TE fundamental mode at longer pulley length (
>
200
μ
m) is a result of over-coupling.
Fig. S4 investigates the wavelength dependence of the power coupled to the resonator. Unity in this plot represents
critical coupling. The measurement is performed using the fundamental TE mode and a 200 micron long pulley
length. This result shows that the coupling constant as well as the intrinsic Q values are quite uniform over this band
of wavelengths. The uniformity of the Q values was previously confirmed in Fig. S1.
IV. HIGH TEMPERATURE ANNEALING EFFECT ON CAVITY Q
FIG. S5:
Distribution of Q deviations as a result of high temperature annealing.
Q
1
is cavity intrinsic Q
factor prior to the annealing step, and
Q
2
is the intrinsic Q factor measured after the annealing process.
We proposed a hermetic capping structure on the integrated resonator (Fig.5 in the main text) and the fabrication
of the capping structure might require device compatibility with high temperature steps. As a preliminary test of
the effect of annealing on cavity Q factor, integrated ridge resonators (17 EA) were fabricated, annealed at 1000
◦
C in a nitrogen ambient, and the cavity Q factors were measured before and after the annealing step via end-fire
coupling. The measurement results in Fig.S5 show that annealing generally improves the optical Q factor (
Q
1
and
Q
2
are intrinsic Q factor before/after annealing, respectively).
V. MODE FILTERING
It is interesting to note that there is an enormous amount of spatial mode filtering that occurs in the integrated
ridge resonator as compared to the earlier non-integrated wedge resonator design. Fig. S6 shows a measured mode
dispersion for the wedge (non integrated and taper coupled) versus the waveguide coupled ridge. Many modes are
4
knocked-out by the ring aperture and leakage into the exterior silica layer. These degrees of freedom can potentially
be used to control the mode spectrum in the future.
FIG. S6:
Transverse mode filtering in the silica ridge resonator.
(
a
) Upper panel depicts a non-integrated
silica wedge resonator
S4
and a false-colored SEM cross-sectional image with calculated fundamental mode profile in
color (bar: 10
μ
m). Lower panel gives the measured mode spectrum of the wedge resonator (FSR = 15.228 GHz).
(
b
) As in panel
a
except for the silica ridge resonator (FSR = 15.228 GHz). Compared to the wedge resonator, the
ridge resonator filters out a large number of transverse mode families.
VI. ADDITIONAL COMMENTS
In the measurement, the power coupling efficiency to the silicon nitride waveguide was approximately 25 % using
the lensed optical fiber. No attempt to improve coupling was made, but in the future a tapered waveguide end can
be introduced to improve the coupling efficiency
S5
.
The Mach Zehnder interferometer used for spectral calibrations (see figure 2b caption in main text) was, itself,
calibrated using an adaptation of the radio-frequency spectroscopy method
S6
.
[S1] Chin, M. & Ho, S. Design and modeling of waveguide-coupled single-mode microring resonators.
J. Lightwave Technol.
16
, 1433 (1998).
[S2] Hosseini, E. S., Yegnanarayanan, S., Atabaki, A. H., Soltani, M. & Adibi, A. Systematic design and fabrication of high-Q
single-mode pulley-coupled planar silicon nitride microdisk resonators at visible wavelengths.
Opt. Express
18
, 2127–2136
(2010).
[S3] Spencer, D. T., Bauters, J. F., Heck, M. J. & Bowers, J. E. Integrated waveguide coupled
Si
3
N
4
resonators in the
ultrahigh-Q regime.
Optica
1
, 153–157 (2014).
[S4] Lee, H.
et al.
Chemically etched ultrahigh-Q wedge-resonator on a silicon chip.
Nat. Photon.
6
, 369–373 (2012).
[S5] Almeida, V. R., Panepucci, R. R. & Lipson, M. Nanotaper for compact mode conversion.
Opt. Lett.
28
, 1302–1304 (2003).
[S6] Li, J., Lee, H., Yang, K. Y. & Vahala, K. J. Sideband spectroscopy and dispersion measurement in microcavities.
Opt.
Express
20
, 26337–26344 (2012).