Open Access
Add to BibSonomy
Abstract
We demonstrate a heterogeneously integrated III-V-on-SOI distributed feedback laser with a low grating strength (κ < 40 cm−1) and a narrow linewidth of Δν = 118 kHz. The laser operates single mode with a side-mode suppression ratio over 45 dB, provides a single-sided waveguide-coupled output power of 22 mW (13.4 dBm) and has a wall-plug efficiency of 17%. The dynamic characteristics were also evaluated, obtaining an intrinsic 3 dB modulation bandwidth of 14 GHz and a photon lifetime of 8 ps. Large-signal intensity modulation using a 231-1 PRBS pattern length revealed open eye diagrams up to 25 Gb/s and a penalty on the dynamic extinction ratio lower than 1 dB after transmission over a 2 km standard single mode optical fiber.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
The heterogeneous integration of III-V materials on silicon-on-insulator (SOI) wafers has experienced and enormous progression in the last few years. This heterogeneous platform offers the possibility to integrate light sources and amplifiers and combine them with the available palette of silicon photonics building blocks to demonstrate photonic integrated circuits (PICs) with added functionality [1]. Among the silicon photonics community, there is already a solid background on the integration of new materials on SOI such as germanium (Ge), silicon nitride (SiNx) or, more recently, lithium niobate (LiNBO3). Such heterogeneous platforms have served as a playground to demonstrate high-speed Ge photodiodes [2], polarization-insensitive de-multiplexers [3], low-loss waveguides [4], nonlinear photonic circuits [5], high-speed hybrid modulators [6] or integrated Pockels lasers [7]. Such hybridization of the silicon photonics platform has been possible after intensive exploration of several bonding techniques, regrowth epitaxial processes and transfer-printing technologies, which have opened a new dimension for multifunctional heterogeneous PICs [8].
Despite such remarkable advances, most of these technologies still have a long journey ahead before they can be actively considered by silicon photonics foundries, as important challenges such as the scalability, the wafer yield or the low-cost packaging have not been addressed yet. In the short-term, the oxide-based wafer-bonding of III-V wafers on SOI PICs seems to be the most viable solution to provide a sufficiently high transfer readiness level (TRL) for the emerging silicon photonics market. This technology offers good on-wafer device reproducibility and reliability while having low device degradation over time [9]. Thus, being able to control and homogenize the bonding oxide thickness over the wafer is crucial to reduce the device performance variance since it plays a major role on the evanescent coupling between the Si waveguide and the III-V materials. In that regard, we have recently demonstrated that a remarkable wafer yield improvement can be obtained in widely tunable III-V-on-SOI lasers when using a highly uniform bonding oxide thickness [10]. Similarly, other related work has also pointed out the need for a homogeneous bonding oxide thickness and has proposed a new heterogeneous integration that bonds the III-V materials at the back of the Si photonics wafer to reduce the bonding oxide thickness variation and allow other materials to be integrated at the front-side of the platform [11].
At a device level, research of such III-V-on-SOI heterogeneous platforms has been mostly focused on the demonstration of light sources and amplifiers. The integration of reliable and cost-effective lasers on silicon would open the route towards several key applications including high-speed optical interconnects, biosensors, high-performance computing or compact light detection and ranging (LIDAR) systems [8,12]. For most of these applications, simple laser architectures are highly demanded to keep the system low-cost and with a high density of integration. Integrated distributed feedback lasers are good candidates to accomplish that goal since they are compact, they can be easily driven using a single bias current and they offer a good design flexibility to adapt to the system requirements [13]. A representative example can be found in the III-V-on-SOI DFB lasers used for high-speed optical transceivers. For the intensity-modulated direct detection (IMDD) architectures employed in short-reach optical interconnects, directly modulated DFB lasers with a high modulation 3 dB bandwidth are typically used. To this end, devices are designed with a relatively short cavity length (< 400 µm), high mirror losses and a high optical confinement in the III-V materials to enhance the 3 dB modulation bandwidth [14]. As a consequence, the laser linewidth, the saturation output power and the laser’s frequency chirp are often compromised. On the other hand, continuous-wave narrow linewidth DFB lasers coupled to phase-shift modulators are usually employed in coherent optical systems that use advanced modulation formats for long-haul telecommunications [15]. In this case, since devices must have a very small frequency chirp to allow for an efficient recovery of the different constellation maps, the optimum design conditions are quite opposed to the previous example.
In this work, we propose a new low-κ III-V-on-SOI DFB laser that utilizes a backside sampled Bragg grating (SBG) to provide a high single-side waveguide-coupled output power of 22 mW, a narrow linewidth of 118 kHz and a 3 dB-bandwidth of 14 GHz simultaneously, hence enabling large-signal intensity modulation over 20 GHz. Moreover, the narrow linewidth and the high output power exhibited by these SBG-DFB lasers makes them also suitable for complex detection employing phase recovery schemes.
2. Device design and experimental setup
The low-κ III-V-on-Si DFB lasers with backside SBGs were conceived in a heterogeneous platform composed by an inversely grown III-V PIN multilayer (total thickness ≈ 3 µm) containing 6 AlGaInAs quantum wells (active medium). The III-V multilayer is then wafer-bonded onto a processed SOI wafer (with a BOX thickness of 2 µm) that contains the passive photonic circuitry (waveguides, adiabatic tapers and grating couplers). Such circuits were fabricated using a backside processing that can be summarized in six steps (see Fig. 1): i) First, the SBGs and the passive Si waveguides (300 nm-thick) were etched. ii) Then, after SiO2 cladding deposition and planarization, a bulk Si wafer was bonded on top (Si carrier). iii) Subsequently, the bonded wafer was flipped to start the backside processing. iv) The Si substrate and the silicon dioxide were removed to enable the patterning of the Si waveguides. Two different Si rib waveguides were defined: A rib waveguide with a core thickness of 500 nm for the laser section and a passive rib waveguide with a 300 nm-thick core for optical mode transmission and grating coupler definition. v) Then, a silicon dioxide layer was deposited and carefully planarized down to a thickness of ≈ 20-25 nm to enable the bonding of the III-V epitaxy. vi) A benzocyclobutene (BCB)-passivated shallow-rib III-V waveguide with a width of 2.4 µm was defined to better confine the optical mode and guarantee a low current leakage under high current injection. Finally, a front-side back-end-of-line was carried out to contact devices [16].
Fig. 1. Backside wafer processing (i-iv), heterogeneous molecular bonding (v) and front-side back-end-of-line (vi).
Thick Si rib waveguides (500 nm) were used under the gain region and at the III-V-to-Si transition to ensure a tight control of the optical confinement factor in the QWs and an efficient optical mode transfer between the III-V multi-stack and the Si rib waveguide. As specified in the first step of the backside processing, the SBG was etched at the backside of the Si rib waveguide, in the slab. A quarter wavelength shift was introduced at the center of the grating to create a resonant mode in the middle of the photonic stopband. The nominal etching depth is 30 nm, the grating width is 2.4 µm and the grating pitch (ΛDFB) is 240 nm with a duty cycle (DCDFB) of 50%. The SBG has a pitch of ΛSBG-DFB = 2.4 µm and a duty cycle of DCSG-DFB = 50%. A schematic cross-section view of the device and a top view of the DFB distribution in the Si waveguide is shown at the right hand-side of Fig. 2(a). The first design (top view) corresponds to a “typical” λ/4-shifted III-V-on-Si DFB design with a Bragg grating etched on top of the Si rib waveguide [17]. The design in the middle shows the very same λ/4-shifted DFB design but with the Bragg grating etched at the bottom of the Si rib waveguide, in the slab. Finally, a SBG-DFB laser with a backside SBG can be observed at the bottom. It is worth to mention that although Back-side-on-BOX III-V-on-Si DFB lasers in the O-band have been previously reported by Thiessen et al. [11], authors used a thinner Si rib waveguide (300 nm) that always provided a strongly confined optical mode in the III-V waveguide. An additional local amorphous Si deposition of 200 nm in the transition region was required to guarantee a good transmission coefficient from the III-V waveguide to the Si rib waveguide. In our case, the use of a thicker Si rib waveguide (500 nm) enables us to easily control the optical mode confinement in the QWs (ΓQW), in the Si waveguide ((ΓSi_wg)) or in the grating (grating strength, κ) just by adjusting the Si rib waveguide width (Wsi-rib).
Fig. 2. (a) Simulated κ as a function of WSi-rib for the classic design (black curve), the backside grating design (magenta curve) and the backside sampled Bragg grating design (grey curve). The grey square corresponds to the extracted κ value from the measured stop band in the optical spectrum (see Fig. 4(c)). A simulation of the 2D optical mode distribution of two laser designs with a Wsi-rib = 0.7 µm (top) and Wsi-rib = 0.825 is shown in the middle. Evolution of the optical mode confinement in the (b) Si waveguide and in the (c) quantum wells a function of WSi-rib.
Figure 2(a) shows the evolution of κ as a function of Wsi-rib for the three DFB laser designs mentioned before. As can be observed, the design containing the laser grating on top of the Si rib waveguide (black line, top cross-section view) provides the highest κ value regardless of Wsi-rib. As a consequence, laser designs with narrow Si rib waveguide values and a relatively short laser length need to be adopted to maintain the κL product sufficiently low and ensure a high threshold gain margin between the fundamental lasing mode and the side modes, avoiding the apparition of side-modes due to longitudinal spatial hole burning (LSHB) [18]. Moreover, the slope of the κ-WSi-rib curve is the highest around such narrow Wsi values, which makes the design very sensitive even to small bonding oxide thickness variations. To palliate that, a backside grating design can be used instead, as shown in the middle cross-section view (magenta line). In this case, the κ-Wsi-rib curve shows much lower values with a saturation around of 110 cm-1. However, this value is still too high to guarantee an optimum laser performance, and hence the same strategy than the previous design should be adopted. Finally, the SBG-DFB design meets all the requirements described before, as it provides a low κ factor even for wide Wsi-rib values (κ ≤ 50 cm-1) while enabling an optimum balance between the optical gain (g) and internal losses (αint) by adjusting ΓQW and ΓSi_wg at will. It is well-known that P-doped InP layers induce high propagation loss due to intervalence band absorption (IVBA). On the contrary, Si waveguides show low-loss propagation. Thus, the best strategy to reduce αint is to increase the optical confinement in the Si waveguide as much as possible while having a sufficient mode overlap with the QWs to guarantee the optical amplification.
To better illustrate this, two laser designs with a different optical mode cross-section energy distribution are compared. Their simulated optical mode can be observed in Fig. 2 for a laser design with a Wsi-rib = 0.7 µm (top inset, magenta box) and a Wsi-rib = 0.825 µm (bottom inset, grey box). Additionally, Figs. (b) and (c) show the evolution of ΓSi_wg and ΓQW as a function of the Si waveguide width, showing an opposite trend. As seen, the laser design with the widest Si rib waveguide provides an almost twofold increase of the ΓSi_wg parameter (60% over 33%) while offering a ΓQW of 6% (10% for the design with WSi-rib = 0.7 µm). Hence, such parameters provide a ΓQW / ΓSi_wg ratio that is three times lower for the laser design with a Wsi-rib of 0.825 µm with respect to the design with a Wsi-rib of 0.7 µm. On top of that, if we now consider using a backside SBG design, we finally obtain a κ ≈ 40 cm-1. Therefore, we have chosen a Wsi-rib of 0.825 µm for our SBG-DFB laser.
The length of the SBG is 880 µm, which provides a κL ≈ 3.5. Long laser cavities with a high mode overlap with the distributed feedback grating are desirable to narrow the laser linewidth as it decreases the laser’s mirror losses (αm), although they are also prone to show multi-longitudinal mode operation induced by LSHB under high bias current [19]. In our case, these effects start appearing for a bias current beyond 110 mA. Despite this fact, we still have a reasonably large bias current span where the laser shows an excellent performance (see next section for further details).
Figure 3 describes the experimental setup used to characterize the SBG-DFB laser (device under test, DUT). The laser is mounted on a stage equipped with a thermo-electric cooler (TEC) for temperature stabilization. The device is driven by a low-noise electrical source (BE2102, iTEST) combined with either an RF signal for small signal analysis or with a pseudorandom binary sequence (PRBS) generator (Anritsu MU183020A) for eye diagram measurements. The optical signal is coupled from the vertical grating coupler using a lensed fiber. A two-stage optical isolator is used to avoid optical feedback from the setup.
Three different optical lines are selected with an optical switch (1 × 3). The first output is coupled to a lightwave component analyzer (LCA Agilent N5242A/N4375B) for electro-optical bandwidth measurements. The second output is used for the static characterization. To do this, the optical signal is split by a 90/10 coupler to measure either the laser linewidth using a delayed self-heterodyne measurement or the L-I-V characteristic and the optical spectrum. The delayed self-heterodyne measurement is composed of an erbium-doped fiber amplifier (EDFA) that provides 10 dB of optical gain coupled to a JDS Fitel tunable band-pass filter (TBPF) with a 5 nm bandwidth and a fibered Mach-Zehnder interferometer (MZI). An acousto-optic modulator (AOM) driven at 40 MHz is placed in one arm of the MZI to shift the frequency. The other arm contains a fiber coil (50.6 km) to induce a phase delay. Both signals are coupled to a photodiode (XPDV-1120R), and their mode beating is measured by an electrical spectrum analyzer (Agilent E4448A). The second optical line of the second output is coupled to either a power meter (Thorlabs) or an optical spectrum analyzer (Advantest Q8384). Finally, the third output of the optical switch is used for transmission experiments. An optical selector provides either a direct transmission of the optical signal for back-to-back measurements (B2B) or a 2 km transmission through a fiber coil. The output is then coupled to an EDFA (10 dB gain), a TBPF, a variable optical attenuator, VOA (Agilent 81577A) and a high-speed photodiode (XPDV-2320R). A 70 GHz sampling oscilloscope (Agilent 86118A) is used to visualize the eye diagrams.
3. Static characterization
The L-I-V characteristic under continuous-wave operation has been measured for different stage temperatures from T = 20°C up to T = 45°C in steps of 5°C. The L-I curve at 25°C is shown in Fig. 4(a). The optical power follows a sub-linear trend with increasing current bias (due to gain compression) without mode jumps. A lasing threshold of Ith = 35 mA (1.65 kA/cm2) is observed, with a maximum waveguide-coupled optical power over 20 mW from each side of the SBG-DFB laser, confirming the expected symmetry of the field distribution inside the cavity. The inset presents a zoom-in of the L-I curve around Ith for various temperatures. The total wall-plug efficiency, WPE, (left + right optical power) is observed in Fig. 4(b) along with the V-I curve. A maximum value of WPE = 17% and a differential resistance of R = 4.5 Ω can be extracted. These values are well in-line with the ones previously reported in III-V/Si DFB lasers [20]. The evolution of the optical spectrum as a function of the stage temperature can be observed in Fig. 4(c), with a gain peak that drifts towards the longer-wavelength side for increasing temperatures and a side-mode suppression ratio (SMSR) over 45 dB.
Fig. 4. (a) L-I characteristic measured from either the left (grey curve) or the right hand side (magenta curve) of the SBG-DFB laser. The inset shows the evolution of Ith with increasing temperature. (b) Total wall-plug efficiency (left + right optical power) and measured voltage as a function of the injected current. (c) Optical spectra at different temperatures. (d) Evolution of the natural logarithm of Ith as a function of the temperature at the TEC stage. (e) Dissipated power as a function of the TEC stage temperature for a lasing wavelength of λ = 1543 nm.
It is worth to note that the lasing wavelength of the SBG-DFB laser (λSBG-DFB) is detuned by 7 nm (for T = 25°C) towards shorter wavelengths with respect to the gain peak wavelength (λg). This fact enhances the narrowing of the laser linewidth and increases the differential gain [21,22]. In addition, we were able to retrieve the κ value from the measured stop band in the optical spectrum (see the grey square in Fig. 2(a)), obtaining a value of κ ≈ 37 cm-1, in close agreement with the simulations. The drift of the lasing wavelength as a function of the dissipated electrical power was inspected, obtaining a value of 4 nm/W. The characteristic temperature, T0, can also be obtained from the L-I curves at different temperatures (inset of Fig. 4(a)), by using the following equation [23]:
I0 being a fitting parameter. Thus, by fitting the natural logarithm of Ith as a function of the temperature, we obtain a value of T0 = 52°C (see Fig. 4(d)), in agreement with previously reported work [24]. The thermal impedance, Zth, has also been measured using the method based on the athermalization condition reported in Ref. 25. For that, we have measured the dissipated power required to obtain a stable lasing mode at λ = 1543 nm for T = 20°C. Then, we have compensated the drift towards the longer-wavelength side induced in the lasing wavelength when the temperature is increased by a decrease in the injected current, hence maintaining the same lasing wavelength all over the temperature sweep. Results are presented in Fig. 4(e), showing an excellent fitting agreement that provides a Zth = 25°C/W.Figure 5(a) shows the measured laser linewidth using the delayed self-heterodyne setup described in Fig. 3. Figure 5(b), (c) and (d) present the measured electrical spectra for three representative current bias: I = 40 mA, I = 80 mA and I = 110 mA. A Voight function was used to separate the intrinsic Lorentzian linewidth from the Gaussian technical noise. As seen in Fig. 5(a), the laser linewidth decreases monotonically with increasing output power, in agreement with the modified Schawlow-Townes formula [26]:
Fig. 5. (a) Evolution of the measured laser linewidth as a function of the output power. The magenta line corresponds to the linear fitting of experimental data, excluding the last value at I = 120mA. Measured electrical spectra under a bias current of (b) 110 mA, (c) 80 mA and (d) 40 mA.
4. Dynamic characterization
The dynamic characteristic of SBG-DFB lasers was also evaluated. The small-signal modulation response (S21) was measured by driving the SBG-DFB laser with a combination of a DC current bias and a small RF signal generated by a vector network analyzer (VNA). The output power is sent to the LCA for the opto-electronic conversion and analyzed by the VNA to provide the S parameters (see Fig. 3). The modulation frequency is swept from 0 to 25 GHz, and the bias current is increased from 40 mA up to 110 mA in steps of 10 mA. Figure 6(a) presents the small-signal measurement, showing the characteristic peak from the relaxation oscillation frequency, ${f_r}$, followed by a frequency cut-off and a steady decay of the AM signal. The S21 parameter is composed by two contributions: the intrinsic response of the device and the parasitic response of its equivalent circuit, which is driven by the time constant (τc = RC). The product of both transfer functions describes the measured frequency response:
Fig. 6. (a) Small-signal measurements for different bias current. (b) Evolution of the relaxation oscillation frequency (${f_r}$) the measured ($f_r^{meas.}$.) and intrinsic ($f_r^{intrin.}$.) 3 dB bandwidth as a function of the square root of the bias current above the threshold current. (c) Damping factor ($\gamma $) as a function of the square of the relaxation oscillation frequency.
Figure 6(b) compares the measured ($f_r^{meas.}$) and intrinsic ($f_r^{intrin.}$) 3 dB bandwidth with the ${f_r}$ parameter, obtained from the fitting of the intrinsic small-signal response, as a function of $\sqrt {I - {I_{th}}} $. The D factor of the SBG-DFB laser is 1 GHz/mA1/2. A maximum intrinsic bandwidth of 14 GHz was extracted, with a ${f_r} = 9\; GHz$. Remarkably, the last ${f_r}$ value, obtained from a bias current of 110 mA, slightly deviates from the linear trend and undergoes saturation. This fact indicates the apparition of the thermal roll-off combined with the optical nonlinear phenomena in the laser, both of them inducing gain suppression in the laser cavity [29]. Noticeably, such bias current coincides with the beginning of the saturation of the L-I curve and the laser linewidth in Fig. 4(a) and Fig. 5(a), respectively. The ${\tau _c}$ and ${\tau _p}$ values were also extracted by fitting the measured $\gamma $ factor as a function of the square of ${f_r}$, as seen in Fig. 6(c), obtaining a ${\tau _c}\sim 120\; ps$ and a ${\tau _p}\sim 18\; ps$. These values are in agreement with the ones previously reported in similar integrated III-V/Si DFB lasers [30].
We also evaluated the large-signal intensity modulation of the SBG-DFB laser by means of data transmission experiments using a non-return-to-zero (NRZ) signal with a 231-1 PRBS pattern length. The DC bias current operating point was established at 100 mA. A voltage swing of 2 V (peak-to-peak) was used, although a much lower value (≈ 0.3 V) was finally applied to the laser due to the impedance mismatch between the laser’s access resistance (4.5 Ω) and the load (50 Ω). Figure 7 shows open eye diagrams for a B2B configuration and after transmission over a 2 km standard single mode fiber (SSMF), with a gradual degradation of the dynamic extinction ratio (ER) from 6 dB at 10 Gb/s down to 3 dB for 25 Gb/s modulation speed. A moderate penalty of < 1 dB is observed when comparing the B2B measurements with the transmission over the SSMF. It is worth to remark that we did not use any electrical amplification at the receiver, nor equalization.
Fig. 7. Measured optical eye diagrams obtained from an NRZ modulation signal with a PRBS pattern length of 231-1 for various modulation speeds (10 Gb/s, 15 Gb/s, 20 Gb/s and 25 Gb/s) and for a back-to-back configuration or after transmission over a 2 km optical fiber.
These results put forward the use of SBG-DFB lasers as directly modulated laser (DML) sources in high-speed intensity-modulated direct detection transceivers. Moreover, the narrow linewidth characteristic will certainly be of interest in hybrid approaches that combine DMLs for transmission with intensity and phase recovery schemes at the receiver side [31].
5. Conclusions
We demonstrate a low-κ, high-power and narrow linewidth distributed feedback laser designed with a backside-etched sampled Bragg grating (SBG) to allow for an effective reduction of the grating strength and the intracavity loss simultaneously. The SBG-DFB laser has a lasing threshold of Ith = 35 mA, a maximum single-side waveguide-coupled optical power of 22 mW (13.4 dBm), a wall-plug efficiency of WPE = 17% and a characteristic temperature of T0 = 52°C. A lasing mode linewidth as low as Δν = 118 kHz was measured. Small-signal measurements were also carried out, showing a 3 dB bandwidth of 14 GHz that lead to the measurement of open eye diagrams at 25 Gb/s after transmission over a 2 km single mode optical fiber with a penalty below 1 dB in the dynamic extinction ratio. These narrow-linewidth SBG-DFB lasers with good dynamical behavior could be of interest in applications requiring from hybrid schemes that combine intensity-modulated transmitters with complex receivers retrieving the intensity and the phase simultaneously.
Funding
H2020 Excellent Science (H2020-ICT-2017-1).
Acknowledgments
The authors are grateful to Théo Verolet for his early contribution in designing the adiabatic tapers and the mask layout and to Jean-Guy Provost, Fabrice Blache and Filipe Jorge for their technical advice and fruitful discussions.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. P. Kaur, A. Boes, G. Ren, T. G. Nguyen, G. Roelkens, and A. Mitchell, “Hybrid and heterogeneous photonic integration,” APL Photonics 6(6), 061102 (2021). [CrossRef]
2. C. T. DeRose, D. C. Trotter, W. A. Zortman, A. L. Starbuck, M. Fisher, and M. R. Watts, “Ultra compact 45 GHz CMOS compatible Germanium waveguide photodiode with low dark current,” Opt. Express 19(25), 24897–24904 (2011). [CrossRef]
3. J. C. Mikkelsen, A. Bois, T. Lordello, D. Mahgerefteh, S. Menezo, and J. K. S. Poon, “Polarization-insensitive silicon nitride Mach-Zehnder lattice wavelength demultiplexers for CWDM in the O-band,” Opt. Express 26(23), 30076–30084 (2018). [CrossRef]
4. M. J. Heck, J. F. Bauters, M. L. Davenport, D. T. Spencer, and J. E. Bowers, “Ultra-low loss waveguide platform and its integration with silicon photonics,” Laser Photonics Rev. 8(5), 667–686 (2014). [CrossRef]
5. S. Fathpour, “Heterogeneous nonlinear integrated photonics,” IEEE J. Quantum Electron. 54(6), 1–16 (2018). [CrossRef]
6. P. O. Weigel, J. Zhao, K. Fang, H. Al-Rubaye, D. Trotter, D. Hood, and S. Mookherjea, “Bonded thin film lithium niobate modulator on a silicon photonics platform exceeding 100 GHz 3-dB electrical modulation bandwidth,” Opt. Express 26(18), 23728–23739 (2018). [CrossRef]
7. M. Li, L. Chang, L. Wu, J. Staffa, J. Ling, U. A. Javid, and Q. Lin, “Integrated Pockels Laser,” arXiv preprint arXiv:2204.12078 (2022).
8. D. Liang and J. E. Bowers, “Recent progress in heterogeneous III-V-on-silicon photonic integration,” Light: Adv. Manu. 2(1), 59–83 (2021). [CrossRef]
9. S. Bao, Y. Wang, K. Lina, L. Zhang, B. Wang, W. A. Sasangka, and K. H. Lee, “A review of silicon-based wafer bonding processes, an approach to realize the monolithic integration of Si-CMOS and III–V-on-Si wafers,” J. Semicond. 42(2), 023106 (2021). [CrossRef]
10. K. Hassan, J. M. Ramirez, V. Ramez, S. Malhouitre, C. Jany, S. Guerber, and D. Bitauld, “On the yield assessment and improvement of III-V on silicon lasers,” In Optical Interconnects XXII (p. PC120070D). SPIE. (2022, March).
11. T. Thiessen, J. C. Mak, J. Da Fonseca, K. Ribaud, C. Jany, J. K. Poon, and S. Menezo, “Back-side-on-BOX heterogeneously integrated III-V-on-silicon O-band distributed feedback lasers,” J. Lightwave Technol. 38(11), 3000–3006 (2020). [CrossRef]
12. G. Roelkens, A. Abassi, P. Cardile, U. Dave, A. De Groote, and Y. De KoninckD. Van Thourhout, “III-V-on-silicon photonic devices for optical communication and sensing,” Photonics 2(3), 969–1004 (2015). [CrossRef]
13. G. Morthier and P. Vankwikelberge, Handbook of distributed feedback laser diodes, Artech House (2013).
14. A. Abbasi, J. Verbist, J. Van Kerrebrouck, F. Lelarge, G. H. Duan, and X. YinG. Morthier, “28 Gb/s direct modulation heterogeneously integrated C-band InP/SOI DFB laser,” Opt. Express 23(20), 26479–26485 (2015). [CrossRef]
15. K. Zou, Z. Zhang, P. Liao, H. Wang, Y. Cao, and A. AlmaimanA. E. Willner, “Higher-order QAM data transmission using a high-coherence hybrid Si/III–V semiconductor laser,” Opt. Lett. 45(6), 1499–1502 (2020). [CrossRef]
16. G. H. Duan, C. Jany, A. Le Liepvre, A. Accard, M. Lamponi, and D. MakeG. T. Reed, “Hybrid III-V on Silicon Lasers for Photonic Integrated Circuits on Silicon,” IEEE J. Sel. Top. Quantum Electron. 20(4), 158–170 (2014). [CrossRef]
17. S. Stankovic, R. Jones, M. N. Sysak, J. M. Heck, G. Roelkens, and D. Van Thourhout, “Hybrid III–V/Si distributed-feedback laser based on adhesive bonding,” IEEE Photonics Technol. Lett. 24(23), 2155–2158 (2012). [CrossRef]
18. G. Morthier, K. David, P. Vankwikelberge, and R. Baets, “A new DFB-laser diode with reduced spatial hole burning,” IEEE Photonics Technol. Lett. 2(6), 388–390 (1990). [CrossRef]
19. S. Ogita, Y. Kotaki, K. Kihara, M. Matsuda, H. Ishikawa, and H. Imai, “Dependence of spectral linewidth on cavity length and coupling coefficient in DFB laser,” Electron. Lett. 24(10), 613–614 (1988). [CrossRef]
20. S. Menezo, T. Thiessen, G. El-Zammar, J. Mak, J. Da Fonseca, and P. BrianceauJ. K. Poon, “Back-side-on-BOX heterogeneous laser integration for fully integrated photonic circuits on silicon,” 45th European Conference on Optical Communication (2019).
21. M. Oberg, O. Kjebon, S. Lourdudoss, S. Nilsson, L. Backbom, K. Streubel, and J. Wallin, “Increased modulation bandwidth up to 20 GHz of a detuned-loaded DBR laser,” IEEE Photonics Technol. Lett. 6(2), 161–163 (1994). [CrossRef]
22. T. Yamamoto, “High-speed directly modulated lasers,” In Optical Fiber Communication Conference (pp. OTh3F-5). Optica Publishing Group (2012, March).
23. M. N. Sysak, D. Liang, R. Jones, G. Kurczveil, M. Piels, and M. FiorentinoJ. E. Bowers, “Hybrid silicon laser technology: A thermal perspective,” IEEE J. Sel. Top. Quantum Electron. 17(6), 1490–1498 (2011). [CrossRef]
24. M. N. Sysak, H. Park, A. W. Fang, J. E. Bowers, R. Jones, and R. O. CohenM. Paniccia, “Experimental and theoretical thermal analysis of a hybrid silicon evanescent laser,” Opt. Express 15(23), 15041–15046 (2007). [CrossRef]
25. D. Mickus, C. Murphy, R. Mckenna, and J. F. Donegan, “Method to measure thermal impedance for all-active lasers using the athermalisation condition,” Opt. Contin. 1(3), 556–564 (2022). [CrossRef]
26. C. Henry, “Phase noise in semiconductor lasers,” J. Lightwave Technol. 4(3), 298–311 (1986). [CrossRef]
27. J. E. Carroll, J. Whiteaway, D. Plumb, and R. G. S. Plumb, “Distributed feedback semiconductor lasers,” (Vol. 10). IET (1998).
28. P. A. Morton, T. Tanbun-Ek, R. A. Logan, A. M. Sergent, P. F. Sciortino, and D. L. Coblentz, “Frequency response subtraction for simple measurement of intrinsic laser dynamic properties,” IEEE Photonics Technol. Lett. 4(2), 133–136 (1992). [CrossRef]
29. S. Yamaoka, N. P. Diamantopoulos, H. Nishi, R. Nakao, T. Fujii, K. Takeda, and S. Matsuo, “Directly modulated membrane lasers with 108 GHz bandwidth on a high-thermal-conductivity silicon carbide substrate,” Nat. Photonics 15(1), 28–35 (2021). [CrossRef]
30. A. Gallet, “Hybrid III-V on silicon lasers for optical communications,” (Doctoral dissertation, Université Paris-Saclay, 2019).
31. Q. Hu, D. Che, Y. Wang, F. Yuan, Q. Yang, and W. Shieh, “Complex modulation and detection with directly modulated lasers,” Optics Express 23(25), 32809–32819 (2015). [CrossRef]
