Content uploaded by Stephane Trebaol
Author content
All content in this area was uploaded by Stephane Trebaol on Feb 09, 2021
Content may be subject to copyright.
Letter Optics Letters 1
Narrow Linewidth near-UV InGaN Laser Diode based on
External Cavity Fiber Bragg Grating
ANTOINE CONGAR1, MATHILDE GAY1, GEORGES PERIN1, DOMINIQUE MAMMEZ1, JEAN-CLAUDE
SIMON1, PASCAL BESNARD1, JULIEN ROUVILLAIN2, THIERRY GEORGES2, LAURENT LABLONDE3,
THIERRY ROBIN3,AND STÉPHANE TREBAOL1
1Univ Rennes, CNRS, Institut FOTON - UMR 6082, F-22305 Lannion, France
2Oxxius, 4 rue Louis de Broglie, 22300 Lannion
3iXblue, rue Paul Sabatier, 22300 Lannion
*Corresponding author: stephane.trebaol@enssat.fr
Compiled February 9, 2021
We realize a fiber Bragg grating InGaN based laser
diode emitting at 400 nm and demonstrate its high
coherency. Thanks to the fabrication of a narrow
band fiber Bragg grating in the near-UV, we can reach
single-mode and single-frequency regimes for the self-
injection locked diode. The device exhibits 44 dB side-
mode-suppression-ratio and mW output power. De-
tailed frequency noise analysis reveals sub-MHz inte-
grated linewidth and 16 kHz intrinsic linewidth. Such
a narrow linewidth laser diode in the near-UV domain
with a compact and low-cost design could find appli-
cations whenever coherency and interferometric resolu-
tions are needed. © 2021 Optical Society of America
http://dx.doi.org/10.1364/ao.XX.XXXXXX
The InGaN-based laser diodes (LD) market is mainly driven
by the Blu-ray industry, which requires powerful and low-cost
sources. The technology is now mature, providing reliable laser
products in the blue/violet range and extends to UV. However,
the need for narrow linewidth LDs is growing in a variety of
domains ranging from industrial to metrological applications,
where linewidth requirements extends widely from tens of GHz
to sub-MHz. Compact and low-cost LDs are mandatory to
address applications such as visible light communication [
1
],
underwater LiDAR sensing [
2
], 2D and 3D holographic storage
[
3
] and industrial spectroscopy [
4
]. Furthermore, fundamental
spectroscopic applications, optical clocks, atom cooling and
atom interferometry applications require highly coherent
sources with accurate wavelength for the pumping of particular
transitions or probing hyperfine atomic structures [
5
,
6
]. Those
applications would benefit from low-cost, compact and highly
coherent LD. The research on coherence properties of GaN
edge-emitting LDs is still in its infancy. Two main designs are
considered in the literature to reach single-longitudinal-mode
(SLM) regime. The first design called "monolithic approach"
relies on the ridge waveguide effective index modulation.
Recently, first electrically pumped SLM laser diode using a
high order grating has been demonstrated [
7
]. In this work,
the authors have reported 35 dB side-mode suppression ratio
(SMSR). Despite encouraging results, this approach is not yet
mature for mass production and requires demanding technical
resources such as high-resolution e-beam lithography.
In a second approach, SLM operation is achieved thanks to
the filtering function of an external cavity. External-cavity
diode lasers (ECDL) are composed of a Fabry-Perot LD and an
external mirror providing an optical feedback. Moreover, by
transferring its spectral purity to the diode, the use of higher
quality factor cavities can induce drastic spectral narrowing
enabling single-frequency (SF) operation. These cavities are
typically high finesse Fabry-Perot resonators [
8
] or whispering
gallery mode resonators [
9
]. In most commercially available
ECDLs, optical feedback is provided by a diffraction grating
[
10
] whose position and angle with respect to the LD should
be accurately controlled. These ECDLs are thus expensive
and quite large devices because they require implementation
of high-quality electromechanical components and expensive
anti-reflection coating LDs to reach their performances.
An alternative, called fiber Bragg grating LD (FGL) scheme,
has been proposed and extensively studied with applications
in the telecom band [
11
]. Here, the LD is coupled to a narrow
band fiber Bragg grating (FBG). The stability lies on the in-fiber
optical function and no mechanical component is needed like in
conventional ECDL [
12
]. The main advantage of FGL devices
is their versatility. By design, the Bragg mirror characteristics
(center wavelength, peak reflectivity and bandwidth) can
be easily and accurately modified so that external cavity
parameters can be tailored to optimize SLM operation, where
high SMSR and high optical output power are mandatory, or
SF operation where narrow linewidth is requested in particular
for metrological applications. In the following, we consider
that the SLM operation is obtained when one mode of the laser
diode is selected by the Bragg mirror and reflected back in the
laser. On top of that, to reach SF operation, a careful tuning of
the external cavity length should be done to put in phase one
external cavity mode with the previously selected LD mode.
Then strong narrowing of the laser emission can be observed.
This two regimes address the large range of linewidth satisfying
the application needs expressed above.
arXiv:2102.04129v1 [physics.optics] 8 Feb 2021
Letter Optics Letters 2
In this paper, we demonstrate near-UV (NUV) FGL exhibiting
stable SLM emission at 400 nm with 1.3 mW optical power,
44 dB SMSR and intensity noise (IN) below -130 dB/Hz above
10 kHz. SF operation can even be obtained by accurately setting
up the FBG cavity. In this last configuration, we report intrinsic
linewidth down to 16
±
5 kHz. In our work a special attention
to use low cost components for the laser design is considered.
So far, AR-coated NUV laser diodes are dozen times more
expensive than simple cleaved facet FP laser diodes. Contrary
to what is usually reported in the literature, SLM or SF designs
have been obtained by using a simple GaN LD without any AR
coating. Our work demonstrates the possibility of achieving
state-of-the-art performances by judiciously combining low-cost
components.
The NUV FGL is schematically represented by a red box
on figure 1. The laser diode is a commercially available 400 nm
emission wavelength InGaN/GaN LD without anti-reflection
coating. Cleaved facets determine the laser cavity, which is
typically in the order of 1 mm long. The free spectral range
(FSR) is measured to be 27 pm (50.5 GHz). By design, the diode
emits on the fundamental transverse mode (TEM00) but the
beam is elliptical and highly divergent. A set of lenses is used to
circularize and collimate the beam, which is then focused into
a single-mode fiber through a fiber coupler (C). The external
Fig. 1.
Experimental setup. The device under test (DUT) is
composed of the LD, optical beam shaping lenses and a fiber
Bragg grating (FBG). The external cavity length (about 7 cm
long) is adjusted using a piezo actuator (PI1). Path 1 is used
for intensity noise (IN), low resolution spectrum using an
optical spectrum analyzer (OSA) and power measurements
(POW). For OSA and POW measurements, a fraction of the
signal is extracted. On path 2, the movement of one mirror
of the Fabry-Perot etalon (FPE) is used for laser line scanning
and frequency noise measurement, thanks to a piezo actuator
(PI2). Electrical signals generated by photodiodes (PD) are
sent to an oscilloscope (OSC) and an electrical spectrum an-
alyzer (ESA) through transimpedance amplifiers (TIA) and
DC-blocks (DCB). DC photocurrents are measured using a
multimeter (A). Coupling of light in the fiber and collimation
of the output beam are made using APC fiber couplers (C).
cavity is based on a Bragg mirror, photo-inscribed in the core of
a single-mode fiber. To eliminate parasitic reflections, fiber ends
are polished at an angle of 8
°
(APC connectors). The overall
length of the cavity is 7 cm, including the free space section
used for beam-shaping and the few centimeters fiber section
extending to the Bragg mirror. The Bragg mirror inscription
relies on the photosensitivity of germanosilicate single-mode
fiber (core diameter 2.4
µ
m at 400 nm). The fiber is transversally
exposed to a UV fringe field. No phase-mask is commercially
available to reach Bragg wavelengths lower than 405 nm by
direct inscription technique. We thus used a Talbot interfer-
ometer allowing Bragg mirrors with reflectivity wavelengths
within the range 375-405 nm [
13
]. To our knowledge, it is the
first realization of FBG at such short Bragg wavelengths.
To ensure the selection of only one longitudinal mode of the
LD cavity, the 3 dB-reflection-bandwidth of the FBG has to be
smaller than the FSR. We then designed the FBG bandwidth
to be around 20 pm. Moreover, to reach strong feedback
regime [
14
], the following expression should be satisfied :
I2RBragg >RLD
, where
I=
0.7 is the fiber coupling coefficient,
RLD =
0.18 the output LD cavity reflectivity mirror and
RBragg
the Bragg mirror reflectivity. Coefficient
RBragg
should thus be
greater than 0.4. Hence, the choice of a low-cost non-AR coated
LD implies the use of a quite high reflectivity Bragg mirror.
Thus, the peak reflectivity of the Bragg mirror is chosen to be
RBragg =70%.
L-I curves of the laser diode without (black curve) and with (red
curve) optical injection from the FBG are shown in the inset of
figure 2. As expected, the current threshold is reduced from 73 to
50 mA. Under optical feedback, the L-I curve displays a periodic
modulation of the power that corresponds to mode hopping
from adjacent LD modes. The highly multi-longitudinal mode
solitary LD spectrum is shown in black line on figure 2. The
Fig. 2.
Spectrum of the solitary laser diode (black curve) and
FGL (red curve) for a pump current close to 95 mA. Inset) L-I
curves of the solitary laser diode (black curve) and FGL (red
curve).
central wavelength is close to 400.5 nm and the spectrum
spreads over
'
4 nm (at 40 dB from the maximum). Spectral
structuration can be seen, resulting from a phenomenon, well
known in GaN-based LDs, called mode clustering [15].
When adding the FBG with a bandwidth smaller than the laser
diode FSR, one single mode is selected as it can be seen on figure
2. To reach such a SLM regime, the Bragg resonance is adjusted
to one particular LD mode by tuning the pump current and
the Bragg wavelength to improve overlap between a specific
LD mode and the Bragg central wavelength. Furthermore, this
wavelength should be as close as possible to the LD gain curve
Letter Optics Letters 3
maximum, in order to maximize the SMSR. In our experimental
conditions, we obtained an SMSR of 44 dB and an optical power
of 1.3 mW as depicted in figure 2. It is to notice that the fine
structure of the spectrum is not resolved by the 10pm resolution
OSA used and characterisations are completed by further
measurements as described in the following. Comparable SMSR
performance, close to 40 dB, has been obtained with a bulk
diffraction grating configuration in reference [
16
]. A value of
25 dB has been measured for an equivalent structure without
AR coating in the telecom range [
17
]. It is to notice that best
results for SLM lasers obtained by the monolithic approach
at NUV wavelengths are 25 dB SMSR [
7
]. Futhermore, FBG
stretching can be used to tune the SLM central frequency over
0.5 nm with mode hops between adjacent LD longitudinal
modes separated from an FSR. To this aim, one FBG extremity is
fixed, and the other is attached on a translation plate connected
to a piezo actuator. Applying a longitudinal strain onto the fiber
shifts the FBG wavelength with a sensitivity of 0.3 pm/
µ
strain.
This configuration provides stable SLM operation over a period
of hours. The device fulfills requirements for SLM applications.
To get single-frequency (SF) operation, the external cav-
ity length must be precisely controlled in order to reach good
spectral overlap between a mode of the LD cavity and a mode
of the external cavity. Figure 1displays the linewidth and
Fig. 3.
FGL linewidth estimation obtained by characteriz-
ing the Fabry-Perot etalon (FPE) transmission for 63 mA of
pump current. The red dashed line is a Lorentzian fit of the
measurement. One FPE FSR is shown in the inset to ensure
single-frequency operation and for frequency graduation of
the abscissa axis. During FN measurement, the FPE, used as a
frequency discriminator, is maintained at the quadrature point
(blue point) where response is linear (blue linear fit).
frequency noise measurement benches we used. The laser
fibered output is separated into two paths. Path 1 is used to
measure intensity noise, optical power and low-resolution
spectrum while path 2 gives access to frequency noise and
real-time high-resolution spectrum measurement. For noise
measurements (IN and FN), the signal is focused on a photo-
diode and measured as a power spectral density (PSD) using
an electrical spectrum analyzer (ESA) through a variable gain
transimpedance amplifier.
High resolution spectrum measurement is performed using
a Fabry-Perot etalon (FPE) (FSR 1GHz, finesse>500) on path
2. A FPE resonance is scanned over the laser line, applying a
voltage ramp to a piezoelectric actuator and the output signal is
observed as a function of time on an oscilloscope.
Measurement of one FSR of the 1 GHz FP etalon (inset in
figure 3) is used to ensure single-frequency operation and scale
the abscissa axis as a function of frequency. From a close look on
the laser line shown on figure 3, the linewidth is estimated to be
∆ν=2.4 MHz, which corresponds to the FPE resolution.
We then performed FN characterization to get more insights on
the SF laser frequency dynamics and obtain an estimation of the
laser linewidth for selected integration times. The FP voltage
ramp is removed and the moving mirror is now connected to
the output of a PID controller, which parameters are set in such
a way that the laser is maintained at the quadrature point (blue
dot in figure 3). In such a configuration, intensity fluctuations of
the output signal are proportional to frequency fluctuations of
the input signal and the proportionality coefficient is given by
the resonance flank slope (
P=
35
±
5 mV/MHz). By measuring
the PSD of the output signal, we can extract the frequency noise
as a function of the Fourier frequency [
18
]. Intensity noise of
the FGL are included in this measurement. IN measurement
on path 1 is thus used to guarantee that the intensity noise PSD
is negligible in comparison with the FN PSD. Indeed, IN PSD
curve is at least 20 dB below the FN PSD in the frequency range
of interest. We measure the relative intensity noise of the laser
to be -130 dB/Hz above 10 kHz.
Fig. 4.
Frequency noise measurement. Peaks in the curve are
due to external perturbations. The green dashed line shows
the low frequency 1
/f2
trend, the red dashed line the high
frequency plateau. The servolocking bandwidth is limited to
20 Hz to insure no parasitic contributions to the FN measure-
ment.
In a laser, frequency noise can usually be described by two
main contributions (see Fig. 4) expressed by
Sν=hα
fα+h0
. A
high frequency white noise, which is related to the Lorentzian
shape of the intrinsic laser-linewidth (dashed red line). This
white noise level is identified by the coefficient
h0
. This noise is
due to the random phase fluctuation of spontaneous emission
[
19
]. At lower frequencies the device is under influences of
various contributions like acoustic, thermic and electromagnetic
perturbations. Their spectral signature usually follows a
f−α
evolution on the frequency noise with
α
bounded between 1
Letter Optics Letters 4
and 2. This second contribution gives a Gaussian shape to the
integrated optical linewidth [
20
] that exhibits a Voigt profile in
the general case [21].
FN measurement plotted in figure 4is taken at pump current
I=63 mA. We find the two behaviors mentioned above. At low
frequency (<10 kHz), the FGL FN displays a 1
/f2
tendency
with
h2=
1.45
×
10
12
Hz
2
/Hz (green dashed line in the figure).
Strong perturbations probably coming from the current source
lead to an increase of FN beyond the 1
/f2
line (100 Hz-1 kHz).
Above 10 kHz, the FN reaches a plateau at
h0=
5
×
10
3
Hz
2
/Hz,
corresponding to the white intrinsic noise of the laser.
We use the beta-line approach [
20
] to estimate the integrated
optical linewidth of the device. Short-time (t=1ms) and longer-
time (t=10ms) linewidths are 250 kHz and 950 kHz respectively.
In this calculation, peaks from external perturbations have been
neglected. The former value is in the same order of magnitude
than the linewidth of 420 nm grating-ECDL laser presented in
[
6
] and 405 nm self-injection locked LD in [
9
]. Using the relation
∆ν=πh0
[
20
], we can estimate the intrinsic linewidth of the
laser to be 16
±
5 kHz. Savchenkov et al. [
22
] have reported
a similar value (
≈
30 kHz) for self-injection locked LD at 370
nm. ECDL frequency stabilizations based on external cavity
optical feedback [
8
,
23
] allow to narrow the linewidth down to
few kHz but the large external cavity (meter long) and moving
parts limit the observation time from minutes to seconds.
To reach stable SF operation, a careful consideration should be
given to the setup. External perturbations upon the external
cavity are indeed detrimental to the device linewidth perfor-
mances. Acoustic perturbations are drastically reduced using
an anti-vibration table and a Peltier module situated below the
external cavity that provides a precise temperature stabilization.
However, for the sake of compactness, improvements can
be implemented. To ensure stability of the device under
single-frequency operation, the amount of reinjected light into
the LD after a round trip in the cavity must be as constant
as possible. Because LD emission is linearly polarized, the
use of polarization-maintaining fiber during the light round
trip may ensure higher performances without demanding
implementation of other external parts. The cavity length
should be reduced as much as possible to decrease external
perturbations on the fiber. However, a trade-off has to be found
between the length and the quality factor of this cavity [24].
GaN based LDs still suffer from epitaxial growth imperfec-
tions that revealed nonlinear gain behaviors [
15
]. Moreover,
working in the strong feedback regime contributes to favor the
appearance of nonlinear effects since a relatively large amount
of light is reinjected in the laser diode. To observe SF regime,
we thus operate the laser device at low power to prevent the
apparition of unstable regimes. In a future work, optimization
of the Bragg reflectivity should allow to reach higher power
while maintaining SLM or SF regimes.
We demonstrate a reliable, all-fibered optical output, compact,
FBG NUV laser source. We have shown that single longitudinal
mode operation can be achieved using non-antireflection
coated GaN LD coupled to a fiber Bragg grating designed
at 400 nm. We measured side-mode suppression ratio up to
44 dB with 1.3 mW output power and an intensity noise level
below -130 dB/Hz above 10 kHz. Furthermore, transferring
the spectral purity of the external cavity to the diode leads to
single-frequency operation with integrated sub-MHz linewidth
and 16
±
5 kHz intrinsic linewidth. Hence, NUV FGLs could
constitute a commercial alternative to expensive grating ECDLs
with similar performances. For these reasons, they have been
used for decades in various applications at telecom wavelengths
where low–cost and compactness have to be associated to
high coherency performances. Through this work, we have
highlighted that they could also be assets for applications in the
NUV.
FUNDING
The present work is supported under projects DeepBlue
and UV4Life by the Region Bretagne (contracts N
°
16008022,
19005486) and the European Regional Development Fund (con-
tracts N° EU000181, EU000998).
DISCLOSURES
The authors declare no conflicts of interest.
REFERENCES
1.
T. C. Wu, Y. C. Chi, H. Y. Wang, C. T. Tsai, and G. R. Lin, Sci. Reports
7, 40480 (2017).
2.
L. K. Rumbaugh, K. J. Dunn, E. M. Bollt, B. Cochenour, and W. D.
Jemison, “An underwater chaotic lidar sensor based on synchronized
blue laser diodes,” in Proc.SPIE, , vol. 9827 (2016), p. 98270I.
3.
T. Tanaka, K. Takahashi, K. Sako, R. Kasegawa, M. Toishi, K. Watan-
abe, D. Samuels, and M. Takeya, Appl. Opt. 46, 3583 (2007).
4.
P. Larkin, Infrared and Raman Spectroscopy (Elsevier Science, London,
2011).
5.
C. W. Oates, F. Bondu, R. W. Fox, and L. Hollberg, The Eur. Phys. J. D
- At. Mol. Opt. Plasma Phys. 7, 449 (1999).
6. X. Zeng and D. L. Boïko, Opt. letters 39, 1685 (2014).
7.
T. J. Slight, S. Stanczyk, S. Watson, A. Yadav, S. Grzanka, E. Rafailov,
P. Perlin, S. P. Najda, M. Leszczy, and A. E. Kelly, Appl. Phys. Express
11, 112701 (2018).
8.
M. Horstjann, V. Nenakhov, and J. P. Burrows, Appl. Phys. B
106
, 261
(2012).
9.
P. S. Donvalkar, A. Savchenkov, and A. Matsko, J. Opt.
20
, 045801
(2018).
10.
F. Favre, D. Le Guen, J. C. Simon, and B. Landousies, Electron. Lett.
22, 795 (1986).
11. R. Kashyap, Fiber Bragg Gratings (Academic press, San Diego, USA,
1999).
12.
M. Ziari, J.-M. Verdiell, J.-l. Archambault, A. Mathur, H. Jeon, R.-C.
Yu, and T. L. Koch, “High-speed fiber-grating-coupled semiconductor
wavelength-division multiplexed laser,” in Conference on Lasers and
Electro-Optics, (Optical Society of America, 1997), p. CMG1.
13. K. Stump, T. K. Plant, and Y. Sun, Electron. Lett. 36, 567 (2000).
14.
K. Petermann, Laser Diode Modulation and Noise (Springer Science &
Business Media, Berlin, 1988).
15.
A. Congar, K. Hussain, C. Pareige, R. Butte, N. Grandjean, P. Besnard,
and S. Trebaol, IEEE J. Quantum Electron. 54 (2017).
16.
N. Ruhnke, A. Müller, B. Eppich, M. Maiwald, B. Sumpf, G. Erbert, and
G. Tränkle, Opt. Lett. 39, 3794 (2014).
17.
C. Park, C. Rowe, J. Buus, D. Reid, A. Carter, and I. Bennion, electronic
letters 22, 1132 (1986).
18. P. Fritschel, T. J. Kane, and A. Jeffries, Opt. Lett. 14, 993 (1989).
19.
G. P. Agrawal and N. K. Dutta, Semiconductor lasers (Kluwer Academic
Publishers, Boston, 1993), 2nd ed.
20.
G. Di Domenico, S. Schilt, and P. Thomann, Appl. Opt.
49
, 4801 (2010).
21.
G. M. Stéphan, T. T. Tam, S. Blin, P. Besnard, and M. Têtu, Phys. Rev.
A - At. Mol. Opt. Phys. 71, 1 (2005).
22.
A. A. Savchenkov, S.-W. Chiow, M. Ghasemkhani, S. Williams, N. Yu,
R. C. Stirbl, and A. B. Matsko, Opt. letters 44, 4175 (2019).
23.
P. Samutpraphoot, S. Weber, Q. Lin, D. Gangloff, A. Bylinskii, B. Braver-
man, A. Kawasaki, C. Raab, W. Kaenders, and V. Vuleti´
c, Opt. express
22, 11592 (2014).
24.
A. Naumenko, P. Besnard, N. Loiko, G. Ughetto, and J. C. Bertreux,
IEEE J. Quantum Electron. 39, 1216 (2003).
A preview of this full-text is provided by Optica Publishing Group.
Content available from Optics Letters
This content is subject to copyright. Terms and conditions apply.
