Reviewers' comments:
Reviewer #1 (Remarks to the Author):
This paper demonstrated a visible frequency comb based on the silica micro
-
disk resonator with
high Q factor, which decreased the threshold of the pump power. The physical process is four
-
wave
-
mixing. The dispersion of the TM mode in visible range is contr
olled to anomalous by
changing the shape the edge of the micro
-
cavity. It is helpful for optical clock cause the rubidium
atom lines are visible. In addition, the stable comb can be realized if locking the comb line to a gas
cell. So, this is interesting w
ork, and there are some questions:
1.For visible wavelength, the micro
-
cavity is thinner (1.5 micron at 778nm). Is it very difficult to
control the dispersion by changing the shape of the edge? Is the dispersion sensitive to wedge
angle? In this experimen
t, the tapered fiber is used to couple pump into the micro
-
cavity. The
thinness of the micro
-
cavity is 1.5 micron, and does it affect the stability of the comb cause the
coupling condition is difficultly controllable. If so, I think this kind of micro
-
cavi
ty is not best choice
for visible comb.
2.What is the stable range of pump detuning for stable comb output? How long you can keep the
stable mode
-
locking comb by using servo?
3.The low amplitude noise means the mode
-
locking state, but the autocorrelation
can directly give
pulse information. Could you measure the autocorrelation of the visible comb?
4.As shown in Fig.1, the comb spectrum of 1064nm is different with 1550nm and 778nm. What is
the reason? Strong ASE from YDFA?
5.We can see that the dispersi
on D2/2π=3.3KHz is very small in Fig. 2b. What is the resolution of
the Mach
-
Zehnder interferometer? The results are average?
6.For the optical clock, the stability is key point. I think that it will be better if the authors can lock
the visible comb to g
as cell for more stable comb. And I suggest that the authors can realize the
visible comb using more integrated structure like Ref. 45.
In my opinion, the manuscript can be published after the authors have considered these
comments.
Reviewer #2 (Rema
rks to the Author):
This paper written by S. H. Lee et al reports on soliton microcomb generation at IR and NIR
wavelength, achieved by controlling the dispersion of the WGM microcavity. I think this paper
presents an important step toward the generation
of the soliton comb in a microcavity at shorter
wavelength. Although the idea on controlling the geometric dispersion is not new, the tuning of the
dispersion by the use of TM
-
TE coupling is nice. Also, the obtained result (achieving the soliton
comb at 1
064 and 780 nm) is having impact, so I support this paper to appear in Nature
Communications. However, the effect of the TM
-
TE coupling is not clearly shown as claimed by the
authors. So this point must be clarified.
1. According to the plot III in Fig. 1
b, the system exhibits already small analogous dispersion even
without the presence of TE
-
TM coupling. Therefore the effect of TE
-
TM coupling to the comb
generation is not clear. I would love to know whether the authors obtained comb for the cavity
shown i
n Fig. 3b, where not TE
-
TM coupling occurs.
2. I would like to see the beta_2 curve for the structure with theta=90 degree in Fig. 3d. Then we
can directly compare the dispersions for theta=90 and 40. I think this is essential to claim that the
TE
-
TM effe
ct is sufficiently affecting the dispersion.
3. Related to the previous point, I would like to see the curves for different wavelength. If I
understand correctly, the curves in Fig. 3d are for the wavelength close at the pump. How about at
the edge of the
spectrum (for instance the curves for relative mode number 200)? I assume the
peak of the curve shifts towards a smaller t for a shorter wavelength mode. I think this information
is important, in order to know the information on the possible bandwidth of
the comb generation.
In addition, the following points must be clarified.
4. The authors should add more plots in Fig. 3d. Three plots are not sufficient. The data appears
me that the dispersion is monotonously increasing (or decreasing) rather than exh
ibiting a peak. I
would like to see data points in between 1.47 and 1.49 um. If this is difficult, why not 1.5 um?
5. What is the reason that you do not observe Raman shift (wavelength shift of the comb envelope
towards lower frequency) in III in Fig. 1e
6. I think 778 nm is rather NIR than visible. I suggest the authors to change the title.
7. I see strong mode coupling close at the pump in Fig. 2b. Isn't it difficult to obtain smooth comb
with such strong peak close at the pump? What is the reason that
you do not suffer from this
inter
-
mode coupling in Fig. 2b.
Reviewer #3 (Remarks to the Author):
The paper by Lee and colleagues reports on the generation of bright temporal dissipative cavity
solitons in microresonators at wavelength of 1550 nm, 106
4 nm and 778 nm. The latter represents
the shortest temporal dissipative soliton generated so far in a microresonator and the first one in
the visible wavelength regime (VIS ~ 400
-
800 nm).
The challenge in accomplishing soliton formation at 1064 nm and 78
0 nm was overcoming the
‘dispersion barrier’ imposed by increasingly normal material dispersion towards shorter wavelength
(as bright solitons require anomalous dispersion).
In order to achieve anomalous dispersion the authors employ both geometrical disp
ersion
engineering (for demonstration at 1064 nm) as well as additional mode
-
hybridization effects
between TE and TM modes (for demonstration at 780 nm). Importantly, all three cases (1550 nm,
1064 nm and 778 nm) are based on the same technology platform,
which illustrates the successful
dispersion engineering (to the extent required for soliton generation) over a factor of two in optical
wavelength. Another important aspect is that it was possible to maintain the same and
electronically accessible soliton
repetition rate of 20 GHz in all three cases.
Extending the operating wavelength range of microresonator solitons beyond initial
demonstrations in the near
-
infrared towards visible and mid
-
infrared is potentially enabling a
variety of novel applications
e.g. in the bio
-
medical domain. As such the present demonstration of
microresonator soliton generation at the edge of the visible regime is a novel and relevant to the
field of microresonators and beyond. The presented data are of high quality and conclusi
vely
support the manuscript that is written in a clear manner.
Before I can recommend the manuscript without reservation for publication I would like to bring up
the following points (in random order) for consideration by the authors. These points do not
question the general novelty and importance of the work but are intended to help improving the
manuscript:
-
While not immediately related to bright solitons, the authors could include a comment on dark
soliton or dark pulse generation, which provides a
n alternative method of creating microresonator
frequency comb in the normal dispersion regime (Xue et al., Nature Photonics 9, 594, 2015).
-
While threshold power levels are mentioned it is not clear whether they refer to the parametric
threshold power
or the threshold power for soliton generation (both notions seem to appear in the
manuscript). Also it is not clear whether these power values specify the power in the tapered fiber
or the coupled power when the threshold is reached.
-
Besides threshold
power levels, it would be interesting to know how much power in the tapered
fiber was required to achieve the shown soliton spectra and what the coupling efficiency was in the
respective modes (especially in the 778 nm case). It is clear that this first de
monstration has not
necessarily been optimized for efficiency or ideal coupling to the resonator; nevertheless the power
levels (in the tapered fiber) as well as the coupling efficiency are important characteristics of the
setup and should be reported alon
g with the generated optical spectra.
-
Dispersion engineering via waveguide diameter, very similar to the ‘thickness’ parameter here,
has already been employed in early work on microresonators (Del’Haye et al., PRL 107, 063901,
2011). While this work wa
s not aiming at soliton formation the authors might want to consider
citing this work to better put their work into context.
-
The discussion in the manuscript focuses on whether the dispersion is anomalous or normal (sign
of beta2 or D2). It would be go
od if the authors could also discuss the (absolute) value of D2 and
the contribution of D3 (as visible in Fig. 3g), which both supposedly increase width reduced
operating wavelength. Estimating the values of D2 and D3 as well as the expected soliton
bandwi
dth would be of interest, in particular as the authors suggest that the results could even
extend across the visible into the ultraviolet bands. Helpful references in this context could be a
study on the tolerance of solitons against non
-
zero D3 (Herr et a
l., PRL 113, 123901, 2014) and
universal scaling laws for e.g. the 3 dB bandwidth (Coen et al., Optics Letters 38, 11, 2013).
-
Again related to further extending the results into the visible and potentially the UV bands: Can
the authors comment on how t
hin the disks could be fabricated and operated before running into
mechanical problems (vibrations or even collapse of the structure)?
-
Using the mode hybridization is a novel an interesting approach. Can the authors comment on
the achievable spectral b
andwidth of this approach (which must be limited as the approximate
mode degeneracy is probably restricted to a certain wavelength interval)?
-
The dispersion measurements in Fig. 2a, 3d show error bars. I assume this is a result from the
parabolic fit o
f the resonance frequencies. Can the authors comment on how the error bars were
obtained in the presence of the outliers due to mode crossings?
-
The RF beatnotes are narrow and show an impressive signal
-
to
-
noise ratio. It would however be
interesting to
choose a smaller scale e.g. 1 MHz span instead of 8 MHz, so that the small
sidebands are better visible. Can the authors speculate on the origin of these noise (?) sidebands?
-
As mentioned the manuscript is written in a clear manner but several minor l
anguage issues
(plural/singular, misplaced words etc.) should be corrected.
Reviewer #1 (Remarks to the Author):
This paper demonstrated a visible frequency comb based on t
he silica micro-disk resonator with high Q factor, which
decreased the threshold of the pump power. The physical
process is four-wave-mixing. The dispersion of the TM
mode in visible range is controlled to anomalous by changing
the shape the edge of the micro-cavity. It is helpful for
optical clock cause the rubidium atom lines are visible. In
addition, the stable comb can be realized if locking the
comb line to a gas cell. So, this is interesting work, and there are some questions:
1.For visible wavelength, the micro-cavity is thinner (1.5 micr
on at 778nm). Is it very difficu
lt to control the dispersion
by changing the shape of the edge? Is
the dispersion sensitive to wedge angle?
In this experiment, the tapered fiber
is used to couple pump into the micro-cavity. The thinness
of the micro-cavity is 1.5 mi
cron, and does it affect the
stability of the comb cause the
coupling condition is difficultly controllable.
If so, I think this kind of micro-cavity is not
best choice for visible comb.
Reply:
Thanks for the comment. In fact, we have an ability
to precisely control and predict the dispersion of
fabricated devices. The dispersion of the wedge resonators
are largely determined by the two factors : the thickness
and the wedge angle. Precise microfabricat
ion provides excellent control of t
hese parameters (See Ref 41 and 46).
To illustrate this control we have provided measurements and si
mulation data of dispersion
versus thickness in figure
2a along with banded regions that should be effect of
wedge angle variation from 30 - 40 degrees. Our process
control is typically of order several degrees. Thickness
control is extremely good becaus
e the oxidation occurs over
43 hours. To clarify we have added a note in the text indica
ting that the oxidation is cali
brated. The coupling condition
is also very controlled. For the 778 nm experiment, we
use a commercial single mode fiber 780HP to fabricate the
taper with a minimum width around 1um. The length of t
he the thinnest part of the taper is around 2 mm and the
transmission is usually above 90%. Since the width of the
taper continuously changes around the taper center, we
tune the position on the taper from which the light is coup
led into the microcavity to approach the phase-matching
condition. Using this method, we can ac
hieve critical coupling condition for the
resonators of various thicknesses with
only one tapered fiber. Besides a stable coupling conditi
on, the feedback loop which locks the pump laser frequency
to a certain soliton power further allows long-term operation
of the soliton (see Ref 5). In this experiment, both 1 um
soliton and 778 nm soliton remained locked several hours until
we turn off the pump laser. We are also working on a
fully integrated visible soliton system on-chip which can
eventually eliminate the need for tapered fiber coupling.
2.What is the stable range of pump
detuning for stable comb output? How long you can keep the stable mode-locking
comb by using servo?
Reply:
The pump detuning for stable soliton operation depends on th
e pump power, but it is in the range of 10 cavity
linewidths (tens of MHz in our case). The soliton comb
can be stably operated for ho
urs using the capture lock
method (ref 53) until the pump laser was turned off. We have
previously reported a record
of soliton parameters over
the duration of the mode-lo
cking (see X. Yi et al.
Optica
2, 1078 (2015) for a 24-hour measurement of a 1550 nm
soliton - this is ref. 5 in this paper)
.
3.The low amplitude noise means the mode-l
ocking state, but the autocorrelation
can directly give pulse information.
Could you measure the autocorrelation of the visible comb?
Reply:
We are not set up to measure autocorrelation at th
ese shorter wavelengths. However, we have measured
autocorrelation for 1550 nm soliton combs generated in silica
wedge resonators similar to those used in this paper
(X. Yi et al.
Optica
2, 1078 (2015) - ref. 5 in this paper
). Significantly those measur
ements have confirmed the close
agreement between the measured autocorrelation pulse
width and the pulsewidth computed using the hyperbolic
secant shape of the soliton spectrum. We are thus very
confident in the prediction of pulsewidth provided in the
paper by using the hyperbolic secant envelope provided in t
he data. We have included text in the manuscript that
makes clear how we are calculating the pulsewidth and also
reference the appropriate paper in case readers would
like to check.
4.As shown in Fig.1, the comb spectr
um of 1064nm is different with 1550nm
and 778nm. What is the reason? Strong
ASE from YDFA?
Reply:
Thanks for the comment. The difference is caused by
the resolution of the OSA
in frequency units, which
decreases with decreasing wavelength.
Therefore, the spectrum of 1064 nm has
a lower contrast compared with the
spectrum of 1550 nm. On the other hand, the 778 nm comb
was measured as second-order diffraction at 1550 nm in
the OSA and accordingly has a better resolution. We have adde
d a comment in the revision to clarify this point.
5.We can see that the dispersion D2/2
π
=3.3KHz is very small in Fig. 2b. What
is the resolution of the Mach-Zehnder
interferometer? The results are average?
Reply:
The dispersion is obtained by parabolically fitting the mode family spectrum over a large wavelength span to
reduce the measurement error. The FSR of the MZI
is around 40 MHz, which is calibrated to <10
-5
accuracy. We
have added a sentence to indicate how the measurement was
performed and to give the characteristics of the Mach-
Zehnder interferometer.
6.For the optical clock, the stability is key point. I think t
hat it will be better if the author
s can lock the visible comb to
gas cell for more stable comb. And I suggest that the author
s can realize the visible comb using more integrated
structure like Ref. 45.
Reply:
Thanks for the suggestion. We have included a comment along these lines (i.e. ref 45 - now ref. 47) in the
concluding paragraph of our manuscript. Furthermore, in the
future we are working with other groups to ultimately
lock this device to a Rb gas cell as suggested by the referee.
In my opinion, the manuscript can be published
after the authors have considered these comments.
Reply:
We thank the reviewer for their comments which have improved our manuscript.
Reviewer #2 (Remarks to the Author):
This paper written by S. H. Lee et al
reports on soliton microcomb generation
at IR and NIR wavelength, achieved by
controlling the dispersion of the WGM
microcavity. I think this paper presents
an important step toward the generation
of the soliton comb in a microcavity at shorter wavelengt
h. Although the idea on control
ling the geometric dispersion
is not new, the tuning of the dispersion
by the use of TM-TE coupling is nice. Also, the obtained result (achieving the
soliton comb at 1064 and 780 nm) is having impact, so I
support this paper to appear in Nature Communications.
However, the effect of the TM-TE coupl
ing is not clearly shown as claimed by
the authors. So this point must be
clarified.
1. According to the plot III in Fig.
1b, the system exhibits al
ready small analogous dis
persion even without the
presence of TE-TM coupling. Therefore t
he effect of TE-TM coupling to the comb generation is not clear. I would love
to know whether the authors obtained
comb for the cavity shown in Fig. 3b, where not TE-TM coupling occurs.
Reply
: Thanks for the comment. Indeed some modes can exhi
bit small amount of anomalous dispersion in the
absence of the TE-TM coupling. However, as shown in the re
vised Fig. 3d, the dispersion is very close to zero and
therefore making the soli
ton generation difficult
, because the system is more sensitive to distortions in mode family
dispersion.
Also,
we do not have means to fabricate the devices with th
eta = 90 degree for dispersion measurement because of
the nature of the wet-chemical
etching process used to create the high-Q
silica wedge devices. As a result, the TE -
TM mode coupling is always present to some degree in our
samples. We have added a sentence to indicate that the
90 degree sidewall case is not possible with the current etch process.
2. I would like to see the beta_2 curve for the structure
with theta=90 degree in Fig.
3d. Then we can directly
compare the dispersions for theta=90 and 40. I think this is
essential to claim that the
TE-TM effect is sufficiently
affecting the dispersion.
Reply
: Thanks for the comment. We added the calculated dispersion of theta=90 degree resonators in Fig. 3d
(horizontal line) which makes clear that
the TE-TM coupling greatly increases t
he amount of dispersion at a certain
thickness. . As noted above (and now in a comment in the
manuscript) we do not have means to fabricate high-Q
devices with theta = 90 degree for dispersion measuremen
t because of the nature of t
he wet-chemical etching
process used to create the high-Q silica wedge devices. We agr
ee that this would be an interesting measurement. On
the other hand, the agreement of
measured dispersion versus thickness with
modeling provides strong evidence that
the mode hybridization is providing the
intended dispersion. Also, we note that additional resonator thicknesses have
been added (per your comment below) to the data which further confirm the effect.
3. Related to the previous point, I would like to see the
curves for different wavelength. If I understand correctly, the
curves in Fig. 3d are for the wavelength close at the pu
mp. How about at the edge of t
he spectrum (for instance the
curves for relative mode number 200)? I assume the peak
of the curve shifts towards a smaller t for a shorter
wavelength mode. I think
this information is
important, in order to know the information on the possible bandwidth of
the comb generation.
Reply:
Thanks for the comment. We have added a new simulation in
figure 3g that directly addresses the referee’s
comment. It plots the second order dispersion versus wavel
ength at a series of thicknes
ses. The bandwidth of the
hybridization effect can be di
rectly seen in these plots.
In addition, the following points must be clarified.
4. The authors should add more plots
in Fig. 3d. Three plots are not su
fficient. The data appears me that the
dispersion is monotonously increasing (or decreasing) rather
than exhibiting a peak. I would like to see data points in
between 1.47 and 1.49 um. If this is difficult, why not 1.5 um?
Reply:
Thanks for the suggestion. We added more data points.
They agree well with the simulation and importantly
verify that the effect is not monotonic.
5. What is the reason that you do not
observe Raman shift (wavelength shift of the comb envelope towards lower
frequency) in III in Fig. 1e
Reply:
The Raman shift for the 778 nm soliton microcomb in Fig.
1e and Fig. 4c is minimal because of relatively large
pulse width (145 fs derived from sech
2
-fit) and small spectral bandwidth. The calculated Raman SSFS for this case
using the formula from Ref. 50 is
0.3 nm. We added a sentence to clarify this point.
6. I think 778 nm is rather NIR than visible.
I suggest the authors to change the title.
Reply:
Although the pump wavelength 778 nm is not within t
he visible range, a portion of the soliton frequency comb
is actually visible. Indeed, we can obs
erve red light emitting from the microcav
ity using bare eyes (also see fig. 1d).
Furthermore, we have added a new result (Fig. 4e) showi
ng an even broader soliton spectrum whose wavelength can
reach as low as 755 nm.
7. I see strong mode coupling close at the
pump in Fig. 2b. Isn't it difficult to obtain smooth comb with such strong
peak close at the pump? What is the
reason that you do not suffer from this inter-mode coupling in Fig. 2b.
Reply:
Although the mode crossing is close to the pump, it
only affects one mode with minor distortions (see the
optical spectra in Fig. 2d). Moreover, the Raman effect sh
ifts the soliton envelope center towards longer wavelength
and thereby further minimizes the influence of this mode cros
sing to the soliton. Therefore,
the soliton is stable and
easy to generate.
We thank the reviewer for their comm
ents which have improved our manuscript.
Reviewer #3 (Remarks to the Author):
The paper by Lee and colleagues reports on the generati
on of bright temporal dissipative cavity solitons in
microresonators at wavelength of 1550 nm, 1064 nm and
778 nm. The latter represents the shortest temporal
dissipative soliton generated so far in a microresonator and the first one in the visible wavelength regime (VIS ~ 400-
800 nm). The challenge in accomplishing soliton formation at 1064 nm and 780 nm was overcoming the ‘dispersion
barrier’ imposed by increasingly normal material dispersion towards shorter wavelength (as bright solitons require
anomalous dispersion). In order to achieve anomalous disp
ersion the authors employ both geometrical dispersion
engineering (for demonstration at 1064 nm) as well as
additional mode-hybridization effects between TE and TM
modes (for demonstration at
780 nm). Importantly, all three cases (1550 nm, 1064 nm and 778 nm) are based on the
same technology platform, which illustrates the successful
dispersion engineering (to t
he extent required for soliton
generation) over a factor of two in optic
al wavelength. Another important aspect is
that it was possible to maintain the
same and electronically accessible soliton re
petition rate of 20 GHz in all three cases.
Extending the operating wavelength range of microresonat
or solitons beyond initial demonstrations in the near-
infrared towards visible and mid-infrared is potentially enabli
ng a variety of novel applicati
ons e.g. in the bio-medical
domain. As such the present demonstrati
on of microresonator soliton generation at
the edge of the visible regime is a
novel and relevant to the field of microresonators
and beyond. The presented data
are of high quality and
conclusively support the manuscript that
is written in a clear manner.
Before I can recommend the manuscript without reservation
for publication I would like to bring up the following points
(in random order) for consideration by the authors. Thes
e points do not question the general novelty and importance
of the work but are intended to help improving the manuscript:
- While not immediately related to brig
ht solitons, the authors could include
a comment on dark soliton or dark pulse
generation, which provides an altern
ative method of creating mi
croresonator frequency comb in the normal dispersion
regime (Xue et al., Nature Photonics 9, 594, 2015).
Reply:
Thanks for the comment. We have included this citation and added a comment in the revision.
- While threshold power levels are mentioned it is not clear
whether they refer to the parametric threshold power or
the threshold power for soliton generation (both notions seem
to appear in the manuscript). Also it is not clear
whether these power values specify t
he power in the tapered fiber or the coupled power when the threshold is
reached.
Reply:
The threshold power mentioned in Fig. 1c refers to par
ametric oscillation threshold. These values are the
power launched in the tapered fiber while
the resonator is critically coupled. We have clarified it in the revision.
- Besides threshold power levels, it would be interesting to
know how much power in the tapered fiber was required to
achieve the shown soliton spectra and what the coupling effi
ciency was in the respective modes (especially in the
778 nm case). It is clear that this first demonstration has
not necessarily been optimized for efficiency or ideal
coupling to the resonator; nevertheless the power levels (i
n the tapered fiber) as well as the coupling efficiency are
important characteristics of the se
tup and should be reported along with
the generated optical spectra.
Reply:
For 1um soliton, the minimum pump power is 100mW, while for 778 nm soliton, it is 135 mW. We have added
the numbers in the revision.
- Dispersion engineering via waveguide diameter, very simila
r to the ‘thickness’ parameter here, has already been
employed in early work on microresonators (Del’Haye et al
., PRL 107, 063901, 2011). While this work was not aiming
at soliton formation the authors might w
ant to consider citing this work to
better put their work into context.
Reply:
Thanks for the comment. We have added the citation.
- The discussion in the manuscript focuses on whether the di
spersion is anomalous or normal (sign of beta2 or D2). It
would be good if the authors could also di
scuss the (absolute) value of D2 and t
he contribution of D3 (as visible in
Fig. 3g), which both supposedly increase
width reduced operating wavelength. Esti
mating the values of D2 and D3 as
well as the expected soliton bandwidth would be of interest,
in particular as the authors s
uggest that the results could
even extend across the visible into the ultraviolet bands. He
lpful references in this context could be a study on the
tolerance of solitons against non-zero D3 (Herr et al., PR
L 113, 123901, 2014) and univer
sal scaling laws for e.g. the
3 dB bandwidth (Coen et al., Optics Letters 38, 11, 2013).
Reply:
We agree with the referee and have added a new fi
g. 3g which provides a simulation of the GVD versus
wavelength at a series of oxide thicknesses. These
plots provide information on the useful bandwidth of the
hybridization effect and also illustrate t
he required thickness in order to extend the effect across the visible band. We
have added the fitted D2 and D3 values in Fig. 4b and Fig.
4d. We have also added a comment in the text relating to
higher order dispersion and have also added the citations suggested by the referee.
- Again related to further extending the results into
the visible and potentially the UV bands: Can the authors
comment on how thin the disks could be fabricated and operat
ed before running into mechanical problems (vibrations
or even collapse of the structure)?
Reply:
As noted above, figure 3g has been added to show that
a thickness of about 1 micron is required for soliton
operation in the blue end of the visible spectrum. At the same
time we have fabricated a series of disks with oxide
thickness down to this value. They are mechanically stable
with respect to silicon undercut to levels that can provide
high Q operation. We have accordingly added a comment to
this effect and also cited our earlier paper on stress
buckling (
Chen, Tong, Hansuek Lee, and Kerry J. Vahala. "The
rmal stress in silica-on-silicon disk resonators."
Applied Physics Letters
102.3 (2013): 031113.).
- Using the mode hybridization is a novel an interesti
ng approach. Can the authors
comment on the achievable
spectral bandwidth of this approach (which must be
limited as the approximate mode degeneracy is probably
restricted to a certain wavelength interval)?
Reply:
As noted above, we have added fig. 3g which answers this question.
- The dispersion measurements in Fig. 2a, 3d show error bars
. I assume this is a result from the parabolic fit of the
resonance frequencies. Can the authors comment on how the e
rror bars were obtained in t
he presence of the outliers
due to mode crossings?
Reply:
Error bars were obtained by measuring dispersion of
various samples for each thickness. Fitting errors are
much smaller than sample variations
and are therefore ignored. We have not
ed how error bars were determined in
Fig. 2a. In Fig. 3d, the error bars were so small (absolut
e dispersion is much larger compared with Fig. 2a) that we
have decided to omit them in favor of larger data points.
- The RF beatnotes are narrow and show an impressive sig
nal-to-noise ratio. It would however be interesting to
choose a smaller scale e.g. 1 MHz span instead of 8 MHz, so
that the small sidebands are better visible. Can the
authors speculate on the origin of
these noise (?) sidebands?
Reply:
Thanks for the comment. We have rescaled our plot
as suggested by the reviewer. The sidebands near 10
kHz originate from the feedback loop,
i.e., the pump laser piezo tuning bandwidth. We have added a comment to this
effect in the figure caption.
- As mentioned the manuscript is written in a clear ma
nner but several minor language issues (plural/singular,
misplaced words etc.) should be corrected.
Reply:
Thanks for the comment. We have made improvements in
the revised manuscript to correct typographical
errors.
We also thank the reviewer for their comments which have improved our manuscript.
REVIEWERS' COMMENTS:
Reviewer #1 only submitted Remarks to the Editor
Reviewer #2 (Remarks to the Author):
I am satisfied with the revision made by the authors. I have nothing more to add.
Congratulations!
Reviewer #3 (Remarks to the Author):
The authors have fully addressed all my previous comments and questions; from my perspective
the manuscript can be published without any further delay.