Linewidth narrowing in self-injection-locked on-chip

lasers
Yating Wan (  yating.wan@kaust.edu.sa )
King Abdullah University of Science and Technology https://orcid.org/0000-0003-2157-2406
Emad Alkhazraji (  emad.alkhazraji@kaust.edu.sa )
King Abdullah University of Science and Technology https://orcid.org/0000-0001-9684-5996
Weng Chow (  wwchow@sandia.gov )
Sandia National Laboratories
Frederic Grillot (  frederic.grillot@telecom-paris.fr )
Télécom ParisTech
John Bowers (  bowers@ece.ucsb.edu )
University of California, Santa Barbara https://orcid.org/0000-0003-4270-8296

Article

Keywords:

DOI: https://doi.org/

License:   This work is licensed under a Creative Commons Attribution 4.0 International License.
Read Full License

Additional Declarations: (Not answered)

Linewidth narrowing in self-injection-locked on-chip lasers

Emad Alkhazraji1, Weng W. Chow2*, Frédéric Grillot3, John E. Bowers4, and Yating Wan1*
1
Integrated Photonics Lab, King Abdullah University of Science and Technology, Thuwal, Saudi Arabia.
2
Sandia National Laboratories, Albuquerque, NM 87185-1086, U.S.A.
3
LTCI, Télécom Paris, Institut Polytechnique de Paris, 91120 Palaiseau, France
4
Department of Electronic and Computer Engineering, University of California – Santa Barbara, Santa Barbara,
California 93106, U.S.A.

* wwchow@sandia.gov, yating.wan@kaust.edu.sa

Keywords: semiconductor lasers, laser linewidth, heterogeneous integration, silicon

photonics, quantum dot lasers, quantum well lasers, laser optimization

Abstract: Stable laser emission with narrow linewidth is of critical importance in many
applications, including coherent communications, LIDAR, and remote sensing. In this work,
the physics underlying spectral narrowing of self-injection-locked on-chip lasers to Hz-level
lasing linewidth is investigated using a composite-cavity structure. Heterogeneously-integrated
III-V/SiN lasers operating with quantum-dot and quantum-well active regions were analyzed
with a focus on the effects of carrier quantum confinement. The intrinsic differences were
associated with gain saturation and carrier-induced refractive index, which were directly
connected with 0- and 2-dimensional carrier densities of states. Results from parametric studies
were presented for tradeoffs involved with tailoring linewidth, output power, and injection
current with different device configurations. A multi-objective optimization analysis was
provided to optimize the operation and design parameters, serving as an analytical tool for
parametric studies to produce timely results in engineering design.

1. Introduction
Quantum-confinement-based quantum well (QW) and quantum dot (QD)semiconductor
laser diodes are the primary options for solid state light sources, owing to their excellent
characteristics of power efficiency, high-temperature operation, small form-factors, etc. While
QWs have been adopted for years in commercialized products, QDs are of interest due to their
zero-dimensional (0-D) density of states and atom-like degeneracy.1-4 Since their first
demonstration in 1994,5 QD lasers have long surpassed their QW counterparts in terms of low
transparency current density, high-temperature stability, and reduced sensitivity to external
feedback and material defects.1,2,6,7 The gain bandwidth of QDs can be engineered to emit a
1
wide range of wavelengths throughout the near-infrared window by leveraging the inherent
line-broadening effects.8,9 Moreover QDs are associated with a theoretical-zero linewidth
enhancement factor (𝛼𝛼𝐻𝐻 ). The much reduced 𝛼𝛼𝐻𝐻 in QDs compared to QWs leads to larger
damping rates (thus higher coherence collapse thresholds), less frequency chirping, ultrafast
gain dynamics, suppressed filamentation, less vulnerability to noise and optical feedback, and
easier frequency/phase stabilization in QD devices.1,10,11
That said, although homogeneous and inhomogeneous broadening phenomena in QDs
are advantageous from the aforementioned viewpoints, they can adversely impact the purity of
emission with comparatively high phase (or frequency) noise and, in turn, wide lasing intrinsic
linewidths. For decades, several techniques have been investigated and employed to stabilize
resonators. Compared to electronic means of laser stabilization, optical injection-locking12,13 is
a more compact and cost-effective alternative in which a stable narrow-linewidth emission is
injected into the laser’s cavity to stabilize (‘lock’) its subsequent emissions14,15. Injection
locking has been achieved and demonstrated in different forms, including the “master-slave”
configuration via an external active pure seeding source,16 the Pound-Drever-Hall technique via
a stable cavity,17 optical feedback injection (a.k.a. self-injection locking) via external optical
circuits,18 Bragg19 or holographic20 gratings in Littrow or Littman configurations, high-finesse
Fabry-Perot (FP) cavities, etc.21
In an integrated photonic setting, self-injection-locking can be realized by locking the
emission of a laser diode to one or more eigenmodes of a high-Q integrated microring resonator.
Since the first demonstration in 199822 and based on the theoretical foundation,23-26 high-Q
crystalline and integrated microring resonators have been extensively utilized to achieve narrow
linewidths,22,27-30 low-noise photonic microwave oscillators,31 and soliton comb generators.14,32
In such a paradigm, the laser diode, along with the microring resonator, may be perceived as a
composite resonator whose eigenmodes extend into free space. This composite resonator/free-
space combination possesses its own resonance modes with additional benefits of selective
mode suppression, power enhancement, and frequency stabilization against chirping effects and
cavity-length fluctuation.33-35 When a component of the laser’s emission is in resonance with
one of the so-called whispering gallery modes of the microresonator, i.e., a composite cavity
mode, optical feedback injection occurs in the form of intracavity Rayleigh backscattering off
the surface and volumetric nonidealities of the microring resonator.
In microring resonators, the Silicon Nitride (SiN) platform has gained a great deal of popularity
due to its comparatively low guiding losses, low thermo-optic coefficients, compatibility with

2
the CMOS platform, being free from two-photon absorption in the telecom window, and high
degree of Kerr nonlinearity for modulation and microcomb generation but simultaneously
negligible Raman and Brillouin nonlinearities that limit the maximum allowed optical power.36-
38
With the aid of self-injection locking, hybrid integrated III-V lasers to chip-based SiN
microresonators have been widely utilized to realize chip lasers and soliton microcomb
sources14,30,39-43 with linewidths as narrow as 40 mHz30, competing with state-of-the-art fiber
lasers. That said, heterogeneous integration of III-V lasers with SiN microresonators offers
more stability, higher-volume production, compactness, and extra features with demonstrated
performance far exceeding the solitary III-V lasers grown on native platforms.36,44,45 More than
30 dB laser frequency noise reduction was achieved due to self-injection locking of the III-V
laser to the SiN microresonator44. The achieved linewidths at the tens of Hz level greatly surpass
its intrinsic limitation of spectral impurity. This maturating of heterogeneous laser technology
intensified the exploration of complex integrated III-V and Si/SiN optical configurations.
In this work, we perform a first-of-its-kind parametric analysis on self-injection locking
in III-V/SiN composite cavities made up of QW- and QD-lasers and ultrahigh-Q SiN resonators.
As such complex quantum structures, we investigate the effects of active medium parameters,
namely, the number of QW and QD layers and the average QD density per layer, over the
optical and spectral performance characteristics. Unlike traditional theoretical works treating
linewidth narrowing via external cavities in traditional lasers23-26 by focusing on the cavity-
coupling dynamics, this work uniquely investigates linewidth narrowing enabled by their
quantum-confined active mediums. This work analyzes and compares QW and QD active
medium archetypes in terms of the major differences in injection locking and linewidth
narrowing arising from carrier quantum confinement and associated density of state functions,
gain saturation, and carrier-induced refractive index fluctuations (typically cast in terms of
𝛼𝛼𝐻𝐻 ).46,47
Lastly, we perform a multi-objective optimization study considering different laser
performance criteria (objective functions). Previous reports exist in literature on optimizing the
stability and locking range of self-injection-locking via high-Q resonators focusing on
parameters concerning the injection-locking mechanism itself,48,49 such as the coupling strength,
roundtrip time, frequency detuning, and pump coupling efficiency. Instead, in this work, we
focus on the design parameters of the QW/QD active region in addition to the injection current
density as optimization variables to optimize the laser performance in the locked regime in
terms of linewidth-narrowing and power efficiency. The aim of this work is to uncover the

3
effects of growth design parameters of such complex systems and facilitate design and
engineering decision-making in the future while shedding light on underlying physics involved
in the observed effects.

2. Results
To investigate the differences between the QD and QW gain regions in terms of
linewidth narrowing resulting from self-injection-locking in the coupled III-V/SiN compound
cavity, a generic 1-d compound cavity structure was utilized. As depicted in Fig. 1, two coupled
resonators with cavity-quality factors 𝑄𝑄1 and 𝑄𝑄2 represent the III-V and SiN sections,
respectively. Both resonators are coupled via an effective coupling transmission of 𝑇𝑇1 = 0.01
and an outcoupling transmission of 𝑇𝑇2 = 0.02. Meanwhile, the cavity length and refractive
index of the SiN resonator were set as 𝐿𝐿1 = 4 𝑚𝑚𝑚𝑚 and 𝑛𝑛1 = 2.1, respectively, and as 𝐿𝐿2 =
600 µ𝑚𝑚 and 𝑛𝑛2 = 3.6, for the III-V section, respectively.

Fig. 1 Basic III-V/SiN coupled-cavity configuration used in the calculations. Input to the laser
theory and connection to experimental devices is through the passive cavity resonances and Q
factors, as well as an effective coupling between cavities 𝑇𝑇1 . The gain and carrier-induced
refractive index (𝐺𝐺 and 𝛿𝛿𝑛𝑛, respectively) are calculated from the laser theory. The top left inset
shows two resonances of the coupled system in the III-V and SiN cavities. Each composite
mode contributes to a point in the resonances. Together, they reproduce the finite-width (Fox-
Li) quasi-mode resonances caused by outcoupling. The top right inset shows illustrations of the
two active region structures of QWs and QDs.

The top left inset of Fig. 1 shows the resonance profile for each of the III-V and SiN
cavities in resonance with each other, showing two closely spaced composite-cavity resonances,
i.e., resonant narrow (rn) and resonant wide (rw), with splitting determined by 𝑇𝑇1 . Each
4
resonance is composed of multiple composite-cavity modes. The linewidths and Q factors of
the uncoupled cavities are 550 MHz and 3.9×105 for the III-V cavity and 60 MHz and 3.6×106
for the SiN cavity. It was assumed that an intracavity filter, such as a Distributed Bragg reflector
(DBR), enabled a single quasi-Fox-Li mode operation at the rn resonance. First, the composite-
cavity mode frequencies and eigenfunctions were calculated. For the rn resonance, the
resolution of the Fox-Li mode required a few hundred composite-cavity modes. The composite-
cavity eigenfunctions were used to compute the mode confinement and overlap factors. Then,
mode intensities and lasing frequencies were computed.

2.1. Integrated III-V/SiN QW laser

Fig. 2a shows the steady-state solution for the full-width-at-half-maximum (FWHM) of
the laser linewidth as a function of the injection current densities (𝐽𝐽) and different numbers of
QW layers. The corresponding output optical power at each operation point on the curve is
color-coded. At any fixed 𝐽𝐽, the output power increases with the number of QW layers due to
the increased active medium volume, leading to larger differential gain and smaller quasi-
Fermi-level separation50. To better visualize the trends with increasing the QW layers, Fig. 2b
depicts the device’s optical power and FWHM at 𝐽𝐽 = 3𝐽𝐽𝑡𝑡ℎ along with the wall-plug efficiency
(WPE), which is defined as the total output optical power divided by the total input electrical
power. The optical power at 𝐽𝐽 = 3𝐽𝐽𝑡𝑡ℎ shows a nearly-linear increase with the QW layers at a
rate of 0.85. On the other hand, WPE surprisingly decreased from ~2.5% to ~1% as the number
of QW layers increased from 2 to 7. The efficiency degradation is indicative of increased QW
losses and can adversely impact the device’s performance.50,51 The actual experimental device
may also suffer from effects stemming from the poor transport of carriers across multi-QWs.
Carrier distribution nonuniformity can lead to dissimilar populations of carriers in different
wells; each is independently in equilibrium with its own quasi-Fermi level.52
From a spectral perspective, the linewidth narrowing in an integrated III-V/SiN laser
can be observed in Fig. 2a to take place in three stages: Immediately pass the lasing threshold
from gain clamping, according to the Schawlow-Townes description; an intermediate region,
where locking to the high-Q SiN passive resonator begins to take place; and at sufficiently high
injection current, total locking of III-V laser and SiN resonator is achieved, leading to
significant linewidth reduction. However, it can also be observed from Fig. 2a that the locking
mechanism becomes more demanding and takes place at higher currents with increased QW

5
layers. In other words, as the active medium’s volume increases, locking conditions become
more stringent due to the elevated injection ratio requirement.

Fig. 2 a Linewidth and lasing characteristics of the integrated III-V/SiN QW laser as a function
of the injected current density and number of QW layers. b The optical power and FWHM at a
current density injection of 3Jth for different numbers of QW layers alongside the WPE.

With that said, in terms of an absolute FWHM, more QW layers resulted in narrower
linewidths for any given 𝐽𝐽. Specifically, increasing the number of QW layers from 2 to 4
reduced the linewidth FWHM from 13 to 4.5 Hz. We ascribe this to the appreciable reduction
in the 𝛼𝛼𝐻𝐻 due to the increased differential gain (𝜕𝜕𝐺𝐺/𝜕𝜕𝑛𝑛𝑐𝑐 ), owing to the increased carrier density
of states at the band edge. 𝛼𝛼𝐻𝐻 quantifies the amplitude-phase fluctuation-coupling of
semiconductor lasers under current injections that alter the carrier distribution. It is a key
parameter in semiconductor lasers that characterizes not only the linewidth broadening effects
but also many other aspects of the laser dynamics, such as frequency chirp, optical feedback
effects, the formation of optical frequency combs, and the self-injection-locking detuning. In
the late 1960s, the concept of phase-amplitude coupling in semiconductor lasers was
observed53,54, which was later quantified by Henry in his renowned paper55 as 𝛼𝛼𝐻𝐻 . It was defined
as the ratio between the derivatives of the real and imaginary parts of the susceptibility 𝜒𝜒, which
in essence are the coupled derivatives (with respect to carrier density 𝑛𝑛𝑐𝑐 ) of the refractive index
and the optical gain of the medium, respectively, as necessitated by causality and formulated
by the Kramers-Kronig well-known relation:

𝜕𝜕𝜒𝜒′/𝜕𝜕𝑛𝑛𝑐𝑐 4𝜋𝜋 𝜕𝜕𝑛𝑛/𝜕𝜕𝑛𝑛𝑐𝑐

𝛼𝛼𝐻𝐻 = =− ∙ (1)
𝜕𝜕𝜒𝜒′′/𝜕𝜕𝑛𝑛𝑐𝑐 𝜆𝜆 𝜕𝜕𝐺𝐺/𝜕𝜕𝑛𝑛𝑐𝑐

6
In other words, 𝛼𝛼𝐻𝐻 is further reduced with more QW layers due to the increased differential
gain, i.e., the dominator of Equation 1, leading to more narrowing in the linewidth of the locked
mode. However, when the number of QW layers was increased beyond four, diminishing
returns in the FHWM reduction were observed with the lowest FWHM of ~3 Hz with 7 QW
layers, compared to the 4.5 Hz with 4 QW (see Fig. 2b). Such behavior suggests that, a tradeoff
is in play in the enhancement of the linewidth between the carrier-induced change in refractive
index and the differential gain. This hinders the attainable benefits of increasing the number
QW layers beyond a certain point. In a later part of this manuscript, we perform a multi-
objective optimization investigation considering different laser performance criteria (objective
functions).

2.2. Integrated III-V/SiN QD laser

Next, a similar examination was followed on the performance of the integrated III-
V/SiN QD laser with two design parameters, viz., the number of QD layers in the active medium
and the average dot density per layer. With a fixed dot density of 4×1014 m-2, Fig. 3a shows the
change of the linewidth FWHM as a function of the number of QD layers. Similarly, at a fixed
number of QD layers of five, Fig. 3b shows the change of the linewidth FWHM as a function
of the QD density. Like the QW device, increasing the QD layers or dot density reduced
linewidth FWHM. For instance, increasing the number of QD layers from 5 to 6 appreciably
reduced the FWHM (at 𝐽𝐽 = 3𝐽𝐽𝑡𝑡ℎ ) from ~200 Hz to ~75 Hz (−62.5%), while increasing the QD
layers beyond that point resulted in diminishing returns, i.e., narrowing from ~75 to ~45 Hz at
the cost of the extra 4 QD layers. Similarly, the linewidth FWHM linearly decreased from ~97
to ~30 Hz (−70% decline at a rate of −17×10-14 Hz/m-2), when doubling the dot density from
4×1014 to 8×1014 m-2. In both QW and QD devices, after full locking, the linewidth decreases
with intracavity intensity, which increases with the number of layers for both QW and QD
devices.
On the other hand, contrary to the observed trend in the QW device, attainment of
locking becomes easier, i.e., taking place at a lower current density, with increased QD
layers/density (notice the point of steep linewidth reduction in Figs. 3a and 3b). We attribute
the difference in the linewidth performance between the QW and QD lasers to the difference in
their 𝛼𝛼𝐻𝐻 , being nearly zero in the latter. This is a result of the frequency pulling in QWs being
dependent on carrier density (due to their asymmetric carrier density distribution) while being

7
independent of carrier density in QDs (due to their symmetric inhomogeneously broadened
carrier density distribution).

Fig. 3 a,b Linewidth FWHM of the III-V/SiN QD as a function of the injection current density
for different QD layers (a) and QD densities (b). c,d Colormaps of the output power (left) and
wall-plug efficiency (right) as functions of the QD layers (c) and QD density (d).

QDs are typically grown via the Stranski-Krastanov mode. The resulting inherent size
dispersive nature gives rise to homogeneous and inhomogeneous broadening phenomena.
Therefore, in terms of absolute linewidth, QDs are typically associated with broader linewidths
compared to the less quantum-confined nanostructures, e.g., QWs. However, self-injection-
locking can drastically reduce their intrinsic linewidth FWHM, owing mainly to their near-zero
𝛼𝛼𝐻𝐻 - a distinctive feature of QDs. Via self-injection-locking, the threshold current of the locked
mode to which the bias current contributes the most is decreased. Meanwhile, the modal gain
of the other modes is reduced due to the decreased carrier occupation probabilities.56 𝛼𝛼𝐻𝐻
strongly depends on the occupation probability of carriers in the individual QDs, such that the
lower the occupation probability, the lower 𝛼𝛼𝐻𝐻 is.11 In other words, increasing the QD layers or
density translates into a lower carrier occupation probability per QD group as the total number
of dots increases. Consequently, this leads to a smaller 𝛼𝛼𝐻𝐻 , and ultimately a stronger linewidth
reduction and easier attainment of injection locking.

8
 On another front, in terms of power and efficiency, Fig. 3c shows 2D colormaps of the
output power (left) and wall-plug efficiency (right) of the QD device as functions of the
injection current density and the number of layers. Fig. 3d shows the same, albeit for different
QD densities, instead of number of layers. Much like the case of the QW laser, either increasing
the number of QD layers or dot density per layer resulted in higher optical powers for any fixed
𝐽𝐽 (any vertical slice of Fig. 3c (left) or Fig. 3d (left), which can be attributed to the increased
effective volume of the active medium.57,58 Furthermore, any horizontal slice of Fig. 3c (left)
or Fig. 3d (left) gives the L−I relation for that particular number of QD layers/density. That
said, Fig. 3c (right) and Fig. 3d (right) show that the efficiency increases with increasing the
QD layers/density for any fixed current density (any vertical slice of Figs. 3c and 3d). This is
in stark contrast to the observed reduction in efficiency with more layers in the case of the QW
counterpart. In addition, any horizontal slice of Fig. 3c (right) or Fig. 3d (right) also indicates
that the efficiency is maximum right after the locking point and starts to decay as the current
increases (L−I rollover). Like bulk and QW structures, QDs suffer from defect losses, yet to a
much lesser degree due to their strong confinement associated with their 0-D density of states.

2.3 Parameter optimization and engineering considerations

The complexity of the optimal operation point and design parameters have been analyzed
in the integrated III-V/SiN QW and QD lasers, with several tradeoffs in play between competing
parameters with a few conflicting performance criteria. In the following, we perform a multi-
objective optimization of the performance parameters of the III-V/SiN QW and QD lasers,
taking into account four objective functions (criteria) with respect to the number of QW/QD
layers, QD surface density per layer, and injection current density as optimization variables. In
this optimization process, the adopted objective functions are: minimizing the input power,
maximizing the output optical power, maximizing the wall-plug efficiency, and minimizing the
intrinsic linewidth FWHM (maximizing the linewidth narrowing due to self-injection locking).
Fig. 4 shows the 4-dimensional design space of the III-V/SiN QW and QD lasers showing the
FWHM and input and output powers of each point. Furthermore, the shown colormaps indicate
the number of layers (or QD density) of each point. The projections of each point on the XY
and XZ planes are also depicted in grey for clarity. It is worth mentioning that only points
beyond the locking point are considered here. Firstly, a genetic algorithm was employed to
obtain the Pareto frontier for each device (traced by the green curves in Fig. 4) consisting of a
set of Pareto optimality points, defined as the set of points where no performance criterion can

9
objectively be made better off. In other words, for each point outside of the Pareto frontier,
there exists at least one point on the Pareto frontier that is objectively better in at least one
performance criterion. Thereafter, we followed with the Technique for Order of Preference by
Similarity to Ideal Solution (TOPSIS), a multi-criteria decision-making algorithm developed
by Hwang and Yoon.59 In the realm of integrated photonics, power consumption remains an
essential attribute in order to satiate the demand for reduced switching and operation power
consumption in the tele- and data-com industries, which gave rise to Si photonics in the first
place.60 Meanwhile, although the intrinsic linewidth is of great significance, all the considered
points here are already within the locking range of the III-V/SiN coupled cavity with
substantially low FWHM for all intents and purposes. Therefore, minimizing power
consumption was adopted as the most pressing concern.

Fig. 4 Optimization of the integrated III-V/SiN QD and QW lasers showing the 4D design
space, the Pareto frontier, and the TOPSIS optimal points for different QW layers, in the case
of the integrated III-V/SiN QW laser, and for different QD layers densities for the integrated
III-V/SiN QD laser.

Following this, the obtained optimal operation point for each case is shown in Fig. 4. For
the integrated III-V/SiN QW laser, the optimal point was found to be that of an input power of
~54 mW, an output optical power of ~1.4 mW, a linewidth FWHM of ~18 Hz, and a wall-plug
efficiency of ~2.6%. This point is obtained with 2 QW layers and an injection current density
of ~2 kA/cm2. The corresponding input power of ~54 mW would translate into a switching
energy consumption of ~5.4 pJ/bit given a 10 Gbit/s direct modulation in an ON-OFF Keying
(OOK) modulation format. This was estimated based on the fact that 54 mW means 54 mJ of
10
input energy is distributed among 10 billion bits every second. Considering the associated
power consumption, minimizing the number of QW layers and maintaining moderate current
density injection are recommended.
Conversely, the optimal point for the integrated III-V/SiN QD laser was found to be
manifested with 8 QD layers and an injection current density of ~0.5 kA/cm2, resulting in an
output optical power of ~0.43 mW and a linewidth FWHM of ~67 Hz. Similarly, in terms of
the dot density, the TOPSIS optimal point was found to be of an output optical power of ~0.96
mW and a linewidth FWHM of ~30 Hz, obtained with 8×10-14 m-2 QD density and an injection
current density of ~0.25 kA/cm2. In other words, contrary to the III-V/SiN QW case, the
analysis shows that maximizing the number of QD layers or QD density yields optimal
performance parameters when injected with low current density.

3. Discussion
In this section, we compare the locking phenomenon in both integrated III-V/SiN QW
and QD devices. To that end, Fig. 5a illustrates the effect of increasing the number of layers in
both devices on the locking threshold condition, i.e., the onset of self-injection-locking, in terms
of the required current density, alongside the corresponding output optical power. Similarly,
Fig. 5b shows the effect of increasing the QD density while fixing the number of QD layers to
five. At first glance, the selling point of each device becomes rather apparent. On the one hand,
the coupled integrated III-V/SiN QW laser offers more optical power, which only grows larger
with more added QW layers. However, this is achieved at the expense of drastically increased
current density and input power requirements, adversely diminishing the quantum and wall-
plug efficiency. This is where the integrated III-V/SiN QD laser shines. While it indeed lags
behind the QW counterpart in terms of absolute output optical power, injection locking and the
steep reduction in the linewidth of the integrated III-V/SiN QD laser take place at a much lower
current density (and input power) with increased QD layers or QD density. Improving the
efficiency and reducing locking requirements can therefore be achieved with increased QD
density or layers, unlike the QW counterpart. Nonetheless, although increasing either the QD
layers or QD density yielded similar trends in the examined performance parameters, these
trends were much stronger with increasing the dot density per layer. We attribute this to the
substantially higher QD packing density (dots per unit volume) in the former approach. This is
because increasing the QD layers entails more QW layers, in which each QD layer is embedded,
in addition to capping/barrier layers. This, in turn, would drastically dilute the active medium's

11
dot packing density compared to increasing the QD surface density per layer, allowing for better
realization and utilization of the QD attractive features.

Fig. 5 a,b Comparison of the lock conditions where the steep reduction in the intrinsic linewidth
takes place in terms of the required injection current density and corresponding output power
as functions of the number of layers (a) for the integrated III-V/SiN QW and QD lasers and as
functions of the QD density for the QD laser (b).

The results also indicate an important point. Typically, standalone freerunning QW

lasers suffer from high 𝛼𝛼𝐻𝐻 that are further elevated with injection current. At the same time,
QD lasers suffer from wide intrinsic linewidths. However, introducing the SiN microring
resonator here enabled both integrated III-V/SiN devices − when locked − to circumvent these
linewidth-related drawbacks, as both devices are able to achieve similar linewidth performance,
even at higher current injections. Yet, the QD-based device does so more efficiently, while the
QW emits at higher optical powers.

4. Methods
To carry out the modal intensities and lasing frequencies within the compound cavity,
61
we followed the formulated theoretical analysis given in Ref. . The theoretical approach
includes a) multimode laser interaction to treat mode competition and wave mixing, b)
quantum-optical contributions from spontaneous emission and c) composite laser/free-space
eigenmodes to describe outcoupling and coupling among components within an extended cavity.
Table 1 depicts the common composite cavity structure and gain medium parameters of the
integrated III-V/SiN QD and QW lasers. The gain medium parameters were extracted from the
62
QW and QD DFB lasers presented in Ref. and Ref. 63, respectively. The fact that the gain
medium parameters of both devices are identical is predicted by quantum kinetic calculations,
owing to the balance among different Coulomb correlation contributions.64,65 It also indicates
12
that the obtained results in this analysis are exclusively based on the difference between 0-D
and 2-D carrier densities of states.

Table 1 Parameters for integrated III-V/SiN QD and QW lasers.

Region Parameter Symbol QD QW

QW height hqw (nm) 8 nm

Waveguide height hwg (µm) 0.2 µm

Stripe width w (µm) 4.9 µm

Composite
SiN cavity Q factor QSiN 3.6×106
cavity
III-V cavity Q factor QIII-V 3.9×105

Spontaneous-emission factor β 0.003

QW height hqw (nm) 8 nm

Dephasing rate γ (s-1) 1012 s-1

Population relaxation rate γab (s-1) 1012 s-1

Defect loss rate γnr (s-1) 109 s-1

Gain
medium
Bimolecular recombination rate B3d (m3s-1) 10-16 m3s-1

Lateral confinement factor Γxy 0.20 0.28

Inhomogeneous broadening ∆inh (meV) 10 −

5. Conclusion

This paper analyzes heterogeneously-integrated III-V/SiN lasers operating with QD and

QW active regions. We focus on the effects of carrier quantum confinement on linewidth
narrowing. The intrinsic differences are associated with gain saturation and carrier-induced
refractive index (customary cast in 𝛼𝛼𝐻𝐻 ), which are directly connected with 0- and 2-D carrier
densities of states. When comparing both integrated III-V/SiN QW and QD structures, the
former shows higher output optical powers that increase with more added QW layers, though
at the cost of degraded efficiency. On the other hand, the integrated III-V/SiN QD counterpart
displayed a more energy-efficient self-injection-locking that only improved with more QD
layers/density. The results also show that introducing the SiN microring resonator to be locked
with either the QD or QW III-V laser aided in circumventing their linewidth-related drawbacks,
as both devices are able to achieve similar linewidth performance, even at higher current
injections. However, the QD-based device was far more energy-efficient to be locked with SiN

13
resonator, thus narrowing linewidth, whereas integrated III-V/SiN QW laser resulted in higher
optical powers.

Furthermore, results from parametric studies are presented for tradeoffs involved with
tailoring linewidth, output power, and injection current with different device configurations.
Lastly, a multi-objective optimization analysis was provided to optimize the operation and
design parameters, with energy efficiency taken as the highest priority concern.

Acknowledgments
This research was supported by Advanced Research Projects Agency-Energy (ARPA-E) No.
DE-AR000067, the U.S. Department of Energy under Contract No. DE-AC04-94AL85000 and
the American Institute for Manufacturing (AIM) Integrated Photonics. This work was
performed, in part, at the Center for Integrated Nanotechnologies, an Office of Science User
Facility operated for the U.S. Department of Energy (DOE) Office of Science.

Conflict of Interests
The authors declare no conflict of interest

Contributions
EA and YW performed the parametric study, optimization, and data analysis. WWC
conceptualized the idea and extended the multimode laser theory to QW and QD lasers. FG and
JEB contributed to connecting simulation results to QD experiments and applications, as well
as connection to III-V/Silicon integration.

References
1. Bimberg, D. & Pohl, U. W. Quantum dots: promises and accomplishments. Materials
Today 14, 388-397, doi:https://doi.org/10.1016/S1369-7021(11)70183-3 (2011).
2. Mirin, R. P., Ibbetson, J. P., Nishi, K., Gossard, A. C. & Bowers, J. E. 1.3 μm
photoluminescence from InGaAs quantum dots on GaAs. Applied Physics Letters 67,
3795-3797, doi:10.1063/1.115386 (1995).
3. Coldren, L. A., Corzine, S. W. & Mashanovitch, M. L. Diode lasers and photonic
integrated circuits. (John Wiley & Sons, 2012).
4. Norman, J. C., Jung, D., Wan, Y. & Bowers, J. E. Perspective: The future of quantum
dot photonic integrated circuits. APL Photonics 3, 030901, doi:10.1063/1.5021345
(2018).
5. Kirstaedter, N. et al. Low threshold, large To injection laser emission from (InGa)As
quantum dots. Electronics Letters 30, 1416-1417 (1994). <https://digital-
library.theiet.org/content/journals/10.1049/el_19940939>.

14
6. Shang, C. et al. Perspectives on Advances in Quantum Dot Lasers and Integration with
Si Photonic Integrated Circuits. ACS Photonics 8, 2555-2566,
doi:10.1021/acsphotonics.1c00707 (2021).
7. Wan, Y., Norman, J., Liu, S., Liu, A. & Bowers, J. E. Quantum Dot Lasers and
Amplifiers on Silicon: Recent Advances and Future Developments. IEEE
Nanotechnology Magazine 15, 8-22, doi:10.1109/MNANO.2020.3048094 (2021).
8. Besombes, L., Kheng, K., Marsal, L. & Mariette, H. Acoustic phonon broadening
mechanism in single quantum dot emission. Physical Review B 63, 155307,
doi:10.1103/PhysRevB.63.155307 (2001).
9. Asryan, L. V. & Suris, R. A. Inhomogeneous line broadening and the threshold current
density of a semiconductor quantum dot laser. Semiconductor Science and Technology
11, 554-567, doi:10.1088/0268-1242/11/4/017 (1996).
10. Smowton, P. M., Pearce, E. J., Schneider, H. C., Chow, W. W. & Hopkinson, M.
Filamentation and linewidth enhancement factor in InGaAs quantum dot lasers. Applied
Physics Letters 81, 3251-3253, doi:10.1063/1.1516236 (2002).
11. Melnik, S., Huyet, G. & Uskov, A. V. The linewidth enhancement factor α of quantum
dot semiconductor lasers. Opt. Express 14, 2950-2955, doi:10.1364/OE.14.002950
(2006).
12. Lang, R. Injection locking properties of a semiconductor laser. IEEE Journal of
Quantum Electronics 18, 976-983, doi:10.1109/JQE.1982.1071632 (1982).
13. Hadley, G. Injection locking of diode lasers. IEEE Journal of Quantum Electronics 22,
419-426, doi:10.1109/JQE.1986.1072979 (1986).
14. Voloshin, A. S. et al. Dynamics of soliton self-injection locking in optical
microresonators. Nature Communications 12, 235, doi:10.1038/s41467-020-20196-y
(2021).
15. Ratkoceri, J. & Batagelj, B. Injection-locked range and linewidth measurements at
different seed-laser linewidths using a Fabry–Pérot laser-diode. Optical and Quantum
Electronics 50, 402, doi:10.1007/s11082-018-1676-9 (2018).
16. Khan, M. T. A. et al. 100 Gb/s Single Channel Transmission Using Injection-Locked
1621 nm Quantum-Dash Laser. IEEE Photonics Technology Letters 29, 543-546,
doi:10.1109/LPT.2017.2664719 (2017).
17. Drever, R. W. P. et al. Laser phase and frequency stabilization using an optical resonator.
Applied Physics B 31, 97-105, doi:10.1007/BF00702605 (1983).
18. Shemis, M. A. et al. Broadly Tunable Self-injection Locked InAs/InP Quantum-dash
Laser Based Fiber/FSO/Hybrid Fiber-FSO Communication at 1610 nm. IEEE
Photonics Journal 10, 1-10, doi:10.1109/JPHOT.2018.2809566 (2018).
19. Volodin, B. L. et al. Wavelength stabilization and spectrum narrowing of high-power
multimode laser diodes and arrays by use of volume Bragg gratings. Opt. Lett. 29, 1891-
1893, doi:10.1364/OL.29.001891 (2004).
20. Steckman, G. J. et al. Volume Holographic Grating Wavelength Stabilized Laser Diodes.
IEEE Journal of Selected Topics in Quantum Electronics 13, 672-678,
doi:10.1109/JSTQE.2007.896060 (2007).
21. Laurent, P., Clairon, A. & Breant, C. Frequency noise analysis of optically self-locked
diode lasers. IEEE Journal of Quantum Electronics 25, 1131-1142,
doi:10.1109/3.29238 (1989).
22. Vassiliev, V. V. et al. Narrow-line-width diode laser with a high-Q microsphere
resonator. Optics Communications 158, 305-312, doi:https://doi.org/10.1016/S0030-
4018(98)00578-1 (1998).

15
23. Kazarinov, R. & Henry, C. The relation of line narrowing and chirp reduction resulting
from the coupling of a semiconductor laser to passive resonator. IEEE Journal of
Quantum Electronics 23, 1401-1409, doi:10.1109/JQE.1987.1073531 (1987).
24. Henry, C. Phase noise in semiconductor lasers. Journal of Lightwave Technology 4,
298-311, doi:10.1109/JLT.1986.1074721 (1986).
25. Fleming, M. & Mooradian, A. Spectral characteristics of external-cavity controlled
semiconductor lasers. IEEE Journal of Quantum Electronics 17, 44-59,
doi:10.1109/JQE.1981.1070634 (1981).
26. Henry, C. H. Theory of linewidth of semiconductor lasers. IEEE J Quantum Electron
18, 259-264 (18).
27. Pavlov, N. G. et al. Narrow linewidth diode laser self-injection locked to a high-Q
microresonator. AIP Conference Proceedings 1936, 020005, doi:10.1063/1.5025443
(2018).
28. Liang, W. et al. Ultralow noise miniature external cavity semiconductor laser. Nature
Communications 6, 7371, doi:10.1038/ncomms8371 (2015).
29. Kondratiev, N. M. et al. Self-injection locking of a laser diode to a high-Q WGM
microresonator. Opt. Express 25, 28167-28178, doi:10.1364/OE.25.028167 (2017).
30. Li, B. et al. Reaching fiber-laser coherence in integrated photonics. Opt. Lett. 46, 5201-
5204, doi:10.1364/OL.439720 (2021).
31. Liang, W. et al. High spectral purity Kerr frequency comb radio frequency photonic
oscillator. Nature Communications 6, 7957, doi:10.1038/ncomms8957 (2015).
32. Pavlov, N. G. et al. Narrow-linewidth lasing and soliton Kerr microcombs with ordinary
laser diodes. Nature Photonics 12, 694-698, doi:10.1038/s41566-018-0277-2 (2018).
33. Santis, C. T., Vilenchik, Y., Satyan, N., Rakuliic, G. & Yariv, A. Quantum control of
phase fluctuations in semiconductor lasers. Proc National Acadamy of Sciences 115,
E7896-E7904 (2018).
34. Shakir, S. A. & Chow, W. W. Semiclassical theory of coupled lasers. Opt. Lett. 9, 202-
204 (1984).
35. Wang, X.-G., Zhao, B.-B., Grillot, F. & Wang, C. Frequency noise suppression of
optical injection-locked quantum cascade lasers. Opt. Express 26, 15167-15176,
doi:10.1364/OE.26.015167 (2018).
36. Xiang, C. et al. High-performance lasers for fully integrated silicon nitride photonics.
Nature Communications 12, 6650, doi:10.1038/s41467-021-26804-9 (2021).
37. Liu, J. et al. High-yield, wafer-scale fabrication of ultralow-loss, dispersion-engineered
silicon nitride photonic circuits. Nature Communications 12, 2236,
doi:10.1038/s41467-021-21973-z (2021).
38. Gyger, F. et al. Observation of Stimulated Brillouin Scattering in Silicon Nitride
Integrated Waveguides. Physical Review Letters 124, 013902,
doi:10.1103/PhysRevLett.124.013902 (2020).
39. Stern, B., Ji, X., Okawachi, Y., Gaeta, A. L. & Lipson, M. Battery-operated integrated
frequency comb generator. Nature 562, 401-405, doi:10.1038/s41586-018-0598-9
(2018).
40. Raja, A. S. et al. Electrically pumped photonic integrated soliton microcomb. Nature
Communications 10, 680, doi:10.1038/s41467-019-08498-2 (2019).
41. Jin, W. et al. Hertz-linewidth semiconductor lasers using CMOS-ready ultra-high-Q
microresonators. Nature Photonics 15, 346-353, doi:10.1038/s41566-021-00761-7
(2021).
42. Fan, Y. et al. Hybrid integrated InP-Si3N4 diode laser with a 40-Hz intrinsic linewidth.
Opt. Express 28, 21713-21728, doi:10.1364/OE.398906 (2020).

16
43. Stern, B., Ji, X., Dutt, A. & Lipson, M. Compact narrow-linewidth integrated laser based
on a low-loss silicon nitride ring resonator. Opt. Lett. 42, 4541-4544,
doi:10.1364/OL.42.004541 (2017).
44. Xiang, C. et al. Laser soliton microcombs heterogeneously integrated on silicon.
Science 373, 99-103, doi:doi:10.1126/science.abh2076 (2021).
45. Guo, J. et al. Chip-based laser with 1-hertz integrated linewidth. Science Advances 8,
eabp9006, doi:doi:10.1126/sciadv.abp9006 (2022).
46. Arakawa, Y. & Yariv, A. Quantum well lasers - gain, spectra, dynamics. IEEE J.
Quantum Electron. 22, 1887-1899 (1986).
47. Asada, M., Miyamoto, Y. & Suematsu, Y. Gain and the threshold of three-dimensional
quantum-box lasers. IEEE J. Quantum Electron. 22, 1915-1921 (1986).
48. Galiev, R. R., Kondratiev, N. M., Lobanov, V. E., Matsko, A. B. & Bilenko, I. A.
Optimization of Laser Stabilization via Self-Injection Locking to a Whispering-Gallery-
Mode Microresonator. Physical Review Applied 14, 014036,
doi:10.1103/PhysRevApplied.14.014036 (2020).
49. Galiev, R. R., Kondratiev, N. M., Lobanov, V. E., Matsko, A. B. & Bilenko, I. A.
Mirror-Assisted Self-Injection Locking of a Laser to a Whispering-Gallery-Mode
Microresonator. Physical Review Applied 16, 064043,
doi:10.1103/PhysRevApplied.16.064043 (2021).
50. Smowton, P. M. et al. Non-uniform carrier distribution in multi-quantum-well lasers.
Applied Physics Letters 83, 419-421, doi:10.1063/1.1593818 (2003).
51. Nagarajan, R., Fukushima, T., Corzine, S. W. & Bowers, J. E. Effects of carrier transport
on high ‐ speed quantum well lasers. Applied Physics Letters 59, 1835-1837,
doi:10.1063/1.106213 (1991).
52. Tessler, N. & Eistenstein, G. On carrier injection and gain dynamics in quantum well
lasers. IEEE Journal of Quantum Electronics 29, 1586-1595, doi:10.1109/3.234409
(1993).
53. Haug, H. & Haken, H. Theory of noise in semiconductor laser emission. Zeitschrift für
Physik A Hadrons and nuclei 204, 262-275, doi:10.1007/BF01326200 (1967).
54. Lax, M. Quantum Noise. X. Density-Matrix Treatment of Field and Population-
Difference Fluctuations. Physical Review 157, 213-231, doi:10.1103/PhysRev.157.213
(1967).
55. Henry, C. Theory of the linewidth of semiconductor lasers. IEEE Journal of Quantum
Electronics 18, 259-264, doi:10.1109/JQE.1982.1071522 (1982).
56. Mirdadashi, H., Sheikhey, M. M. & Baghban, H. Dynamics and spectral characteristics
of quantum dot semiconductor lasers under optical injection-locking. Physica Scripta
95, 025802, doi:10.1088/1402-4896/ab42a6 (2020).
57. Solomon, G. S., Trezza, J. A., Marshall, A. F. & Harris, J. J. S. Vertically Aligned and
Electronically Coupled Growth Induced InAs Islands in GaAs. Physical Review Letters
76, 952-955, doi:10.1103/PhysRevLett.76.952 (1996).
58. Mesaritakis, C. et al. Effect of the number of quantum dot layers and dual state emission
on the performance of InAs/InGaAs passively mode-locked lasers. Applied Physics
Letters 101, 251115, doi:10.1063/1.4772592 (2012).
59. Tzeng, G.-H. & Huang, J.-J. Multiple attribute decision making: methods and
applications. (CRC press, 2011).
60. Miller, D. A. B. Device Requirements for Optical Interconnects to Silicon Chips.
Proceedings of the IEEE 97, 1166-1185, doi:10.1109/JPROC.2009.2014298 (2009).
61. Chow, W. W., Wan, Y., Bowers, J. E. & Grillot, F. Analysis of spontaneous emission
limited linewidth of an integrated III-V/SiN laser. Laser Photonics Rev, 21006620
(2022).
17
62. Fang, A. & al., e. A distributed feedback silicon evanescent laser. Opt Exp 16, 4413
(2008).
63. Wan, Y. & al., e. High speed evanescent quantum-dot lasers on Si. Laser Photonics Rev.,
2100057 (2021).
64. Chow, W. W., Knorr, A., Hughes, S., Girndt, A. & Koch, S. W. Carrier correlation
effects in a semiconductor laser medium. IEEE J. Selected Topics in Quantum Electron.
3, 136 (1997).
65. Lorke, M., Jahnke, F. & Chow, W. W. Excitation dependences of gain and carrier-
induced refractive index change in quantum-dot lasers. Appl. Phys. Lett. 90, 051112
(2007).

18
Figures

Figure 1

Figure 4
Figure 2

Figure 3

Figure 3

Figure 5
Figure 4

Figure 2

Figure 5

Figure 1