Open Access
Add to BibSonomy
Abstract
By using colliding-pulse mode-locking (CPM) configuration with asymmetric cladding layer and coating, 1.5-µm AlGaInAs/InP multiple quantum well (MQW) CPM lasers with high-power and ultra-short pulse generation capability at a repetition rate of 100 GHz are reported. The laser adopts a high-power epitaxial design, with four pairs of MQWs and an asymmetrical dilute waveguide cladding layer to reduce the internal loss, maintaining good thermal conductivity while increasing the saturation energy of the gain region. The asymmetric coating is introduced, as compared to conventional CPM laser with symmetric reflectivity, to further increase the output power and shorten the pulse width. With a high reflection (HR) coating of 95% on one facet and another facet as cleaved, 100-GHz sub-picosecond optical pulses with peak power on a Watt level are demonstrated. Two mode-locking states, the pure CPM state and the partial CPM state, are investigated. Pedestal-free optical pulses are obtained for both states. For the pure CPM state, a pulse width of 564 fs, an average power of 59 mW, a peak power of 1.02 W, and an intermediate mode suppression ratio over 40 dB are demonstrated. For the partial CPM state, a pulse width of 298 fs is demonstrated.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Compact semiconductor mode-locked laser diodes (MLLDs) with high repetition frequency, ultra-short pulse width, and high output power are desirable in the fields of optical fiber communications, photonic THz communications, optical analog to digital conversions, microwave photonics, and frequency conversions [1]. For an MLLD, its repetition frequency is inversely proportional to its resonant cavity length, resulting in multi-gigahertz to tens of gigahertz pulsed operation due to their short cavity length on the order of hundreds of microns to several millimeters. For applications requiring a repetition frequency beyond 100 GHz, the cavity length of a conventional two-section MLLD needs to be less than 400 µm in the 1.5-µm band. Such a short cavity can hardly support a high-power operation as compared to a long cavity laser, due to the influence of the Auger recombination, carrier leakage, and heat dissipation issue at a high pump level.
Several solutions can be adopted to increase the output power while maintaining a high repetition frequency. One method is to increase the number of quantum wells to increase the differential gain to realize efficient lasing at a short cavity length. Recently, Hou et al. reported a two-section 100-GHz MLLD with six pairs of asymmetric MQWs. Ultra-short optical pulses with a pulse width of 440 fs, average power of 10.15 mW, and peak power of 202 mW were realized at a cavity length of 4 µm. It represents a record short pulse width for 100-GHz MLLD with high-power output using a two-section configuration with uncoated facets. Another method is to adopt a colliding-pulse mode-locking (CPM) configuration, where the cavity length can be doubled or multiplied while maintaining a high repetition rate. It is a common practice for mode-locked lasers operating beyond 100 GHz, with the consideration to address the cleaving, thermal management, and gain saturation issues associated with a short cavity. With a long cavity, the active region of the CPM laser can adopt a high-power epitaxial design with fewer quantum wells (QW) as compared to conventional design to reduce the internal loss introduced by the QWs. A slab-coupled optical waveguide (SCOW) structure [2] or a far-field reduction layer (FRL) [3] can also be introduced to further increase the saturation energy and reduce the internal loss due to the increased optical mode area and reduced confinement factor. The gain saturation effect at a high intra-cavity intensity level can thus be alleviated with the high saturation energy design, resulting in a less prominent gain-induced nonlinear self-phase modulation (SPM) during the formation of the short pulses. The dominant factor influencing the pulse evolution will be the gain-induced group velocity dispersion (GVD) which only introduces linear chirp and can be easily compensated.
From the viewpoint of shortening the pulse width, it is desirable to increase the intracavity pulse energy and the loading (absorber Q normalized to cavity Q or the normalized absorption modulation depth) of the saturable absorber [4,5], which can be enhanced using an HR-HR coating [4,6]. However, there is a tradeoff between high intracavity energy and high output power when using a symmetric HR-HR coating design. The output power of CPM lasers operating beyond 100 GHz is mostly limited to a couple of hundred milliwatts, with pulse width on a 500 fs∼1 ps level.
This paper proposes and demonstrates a CPM laser with a high-power and ultra-short pulse generation capability using AlGaInAs QWs with both the cladding layer and coating adopting an asymmetric design. A high-power laser epitaxial design with four pairs of AlGaInAs strained quantum wells and a dilute waveguide layer are used to reduce the cavity loss, increase the gain saturation energy, and suppress the SPM effect in the CPM laser. The asymmetric coating increases the intracavity pulse energy and reduces the pulse width while achieving a high output power. With an HR (95%)-CL coating scheme, 100-GHz sub-picosecond optical pulses with pulse width ranging from ∼300 fs to ∼600 fs are demonstrated. Average power of 59 mW, peak power of 1.02 W, a pulse width of 564 fs, and an intermediate mode suppression ratio over 40 dB are demonstrated for a pure CPM state. An ultra-short pulse width of 298 fs is demonstrated for a partial CPM state. Both CPM states manifest a pedestal-free 100-GHz pulse generation capability.
2. Device design and fabrication
The effect of facet reflectivity on the pulse performance of the CPM laser was first investigated using FreeTWM, free software based on the traveling wave model [6]. In the simulation, the fundamental repetition frequency was set to be 50-GHz, the saturable absorber (SA) section was positioned in the center of the cavity, and its length was set to 5% of the cavity length. The peak power and the pulse width of the CPM were evaluated by changing the reflectivity of the facet. During the simulation, the reflectivity of the output facet was fixed at 0.3, while the reflectivity of the other facet varied from 0.3 to 1. The peak power, pulse width and fundamental frequency suppression ratio were calculated, as shown in Fig. 1. The pulse peak power increases as the reflectivity of the facet increases. When the facet reflectivity increases from 0.3 to 1, the peak power of the output pulse increases by a factor of nearly 5, as shown in Fig. 1(a). The pulse width shows an opposite tendency, the pulse width decreases from 2.7 ps to 1 ps as the facet reflectivity increases from 0.3 to 1, as shown in Fig. 1(b). The suppression of the fundamental mode is evaluated from the RF spectrum. Figure 1(c) shows a typical 100-GHz CPM spectrum for an HR (95%)-CL coating case, with the 100-GHz frequency tone being 37-dB higher than the 50-GHz tone, indicating a good colliding pulsed state. The simulation confirms that simultaneous pulse narrowing, and power scaling can be obtained for the CPM laser with an HR-CL coating design.
Fig. 1. Simulation results of the output pulse performance of the CPM laser with a fixed reflectivity of 0.3 at the output facet, while that of the other facet varies from 0.3 to 1. (a) the peak power (b) the pulse width and (c) the RF spectrum of the output pulse (Cleaved/HR-0.95).
Then CPM lasers were fabricated using an epitaxial structure following a high-power design, with the consideration to simultaneously increase the average power and reduce the SPM effects. For a mode-locked laser, the saturation energy ${E_{sat}}$ is expressed as [7,8] :
where $h\nu $ is the photon energy, A is the mode area, $\varGamma $ is the optical confinement factor, and $\textrm{d}g/\textrm{d}N$ is the differential gain. A high ${E_{sat}}$ requires a low $\varGamma $, a small $\textrm{d}g/\textrm{d}N$ and a large mode area. The small $\textrm{d}g/\textrm{d}N$ is realized by reducing the number of QWs, which also contributes to a reduced $\varGamma $ and low internal loss. The schematic of ethe pitaxial wafer structure is shown in Fig. 2(a). Four compressively strained AlGaInAs QWs and five tensile strained barriers were adopted as compared to conventional laser design with six or more QWs. The increase in mode area A and further reduction of the confinement factor $\varGamma $ was realized by placing a dilute waveguide with three alternations of a 500-nm InP layer and a 90-nm 1.15-Q InGaAsP layer beneath the lower separate confinement heterostructure (SCH) layer. The dilute waveguide functions similarly to a SCOW structure or an FRL to expand the optical mode area and pull the mode away from the p-cladding layer toward the n-cladding layer. The $\varGamma $ of the quantum well was calculated to be 2.56%. With such a low $\varGamma $, it can be expected that the internal loss contributed from the free-carrier absorption [9] in the p-cladding layer can also be further reduced. Compared to SCOW structure consisting of thick InGaAsP quaternary material, the dilute waveguide has a much better thermal dissipation performance due to the contribution of the InP layer (thermal conductivity 68W/(m·K)) in between the InGaAsP layer (thermal conductivity 4.4 W/(m·K)) [10]. This dilute waveguide can be regarded as multiple FRLs capable of smoothly pulling the mode field down to the n-cladding layer without exciting higher-order transverse mode, so as to maintain a large fundamental mode area. A double-trench ridge structure with a ridge width of 3.3 µm and trench width of 6.3 µm was adopted to suppress the higher-order lateral modes while maintaining a large fundamental lateral mode area. The scanning electron microscope (SEM) picture of the device cross-section is shown in Fig. 2(b). After fabrication following a standard semiconductor laser manufacturing process, the laser was cleaved to 850 µm, corresponding to a fundamental mode spacing of 50 GHz. An SA section of 42 µm was placed in the center to facilitate the 100-GHz CPM operation. Figure 2(c) shows the schematic of the CPM laser. The gain sections and the SA were electronically isolated by removing the InGaAs contact layer, and an electrical isolation of 2 kΩ was obtained for the 40-µm isolation gap of the CPM laser.Fig. 2. (a) The schematic of the epitaxial wafer, (b) the SEM picture of the double-trench ridge structure. (c) The schematic of the CPM laser.
3. Device characterization
Lasers with asymmetric coating (HR 95% and cleaved, HR-CL) facets and uncoated facets (CL-CL) were compared following the fabrication. The single-facet power-current-voltage (P-I-V) curves of the two lasers were measured using an integrating sphere. During the measurement, the gain section was forward-biased at ${I_{Gain}}$, and the SA section was reverse-biased at ${V_{SA}}$. As shown in Fig. 3(a), for the CL-CL CPM laser the threshold current is around 39 mA when the SA was left unbiassed (0 V) and increases to 50 mA when the SA was biased at −1.4 V. The average power is 13.1 mW at 300 mA for 0-V reverse bias and reduces to 8.8 mW at a reverse bias voltage of −1.4 V. For the CPM laser with an asymmetric coating, as shown in Fig. 3(b), the threshold current of the HR-CL CPM laser is around 29 mA when the SA was left unbiassed (0 V) and increases to 35 mA when the SA was biased at −1.6 V. Its average power reaches 66.3 mW and 49.7 mW, respectively, at ${I_{Gain}} = 300\,\textrm{mA}$ with a reverse bias voltage of 0 V and −1.4 V. The output power of the CPM laser with an asymmetric coating is about 5∼6 times higher than that of the laser with uncoated facets under the same experimental conditions, which is consistent with the simulation results. For the CL-CL laser, at a pump current of 300 mA, the laser can work in a mode-locking state at a bias voltage ranging from −1 V to −1.4 V. While the HR-CL laser shows an increased bias voltage ranging from −0.8 V to −1.6 V. The threshold current of the HR-CL laser also decreases compared to the CL-CL laser.
Fig. 3. Typical P-I-V characteristics of (a) the CPM laser with uncoated facets (CL-CL) and (b) the CPM laser with asymmetrically coated (HR 95% and CL) facets.
The pulse characterization setup for the CPM laser is shown in Fig. 4. The laser chips were tested on a temperature-controlled heatsink set at 25 °C. Light emitted from the facet was coupled into an AR-coated lensed fiber followed by a fiber-pigtailed isolator. The laser output was split through a 9:1 fiber coupler to a second-harmonic generation (SHG) autocorrelator (APE pulseCheck Autocorrelators 150) and a frequency-resolved optical gating pulse analyzer (FROG Coherent Solutions HR-150) for time domain analysis, and an optical spectrum analyzer (Advantest Q8384) for spectrum analysis. Before being fed into the intensity autocorrelator, the pulses were amplified by an erbium-doped fiber amplifier (EDFA) to boost the laser power. The fiber length, including the EDFA connecting the chip and the measurement equipment, was about 30 meters.
Fig. 4. Measurement set-up of the CPM laser. TEC: Temperature controller; EDFA: Erbium-doped fiber amplifier; PC: Polarization controller; OSA: Optical spectrum analyzer; FROG: Frequency-resolved optical gating.
Multiple tests were carried out to compare the pulse width of the CL-CL and the HR-CL CPM lasers fabricated in the same batch. The pulse widths of the CL-CL CPM lasers are generally around 1 ps while that of the HR-CL lasers are generally below 600 fs in the experiments, which is also consistent with the trend in the simulation.
Further investigation revealed that there are two typical mode-locked states for the HR-CL CPM laser. Figure 5 shows the two cases when the laser was forward biased at ${I_{Gain}}$ of 300 mA, with different reverse bias voltages. Figure 5(a)-(c) correspond to a pure CPM state when the ${V_{SA}}$ was biased at −0.9 V, and the optical spectrum of the CPM laser shows a well-suppressed intermediate mode between the 100-GHz modes, with the best intermediate mode suppression ratio over 40 dB. The pulse train was clean and sharp, with no obvious pedestals, as shown in Fig. 5(b) and (c). The pulse width was calculated to be 564 fs from the autocorrelation trace, assuming a sech2 shape. The time-bandwidth product (TBP) was calculated to be 0.43. The average power and peak power reached a level of 59 mW and 1.02 W, respectively, with a pulse energy of 0.58 pJ. Figure 5(d)-(f) correspond to a partial CPM state when the ${V_{SA}}$ increased to −1.1 V. In the partial CPM state, the optical spectrum shows a severely deteriorated suppression of the intermediate modes, as shown in Fig. 5(d). However, as shown in Fig. 5(e) and (f), the autocorrelation trace is still clean and sharp, with no feature of spiking or obvious pulse pedestal. The autocorrelation trace also shows no obvious deterioration as compared to the pure CPM state. The pulse width was calculated to be 596 fs, assuming a sech2 fit, with a TBP of 0.43. The spectrum and pulse characteristics of the laser may result from two different possibilities. One is a 100-GHz CPM pulse train mixed with a continuous wave (CW) incoherent Fabry-Perot (FP) cavity modes. The other is a 100-GHz CPM pulse train mixed with a 50-GHz fundamental mode-locking pulse train. Both origins can result in similar spectral and time domain performance.
Fig. 5. The optical spectrum and autocorrelation traces of the CPM laser at two reverse bias voltages of −0.9 V and −1.1 V when ${I_{Gain}} = 300\,mA$ (a) optical spectrum at −0.9 V (b) autocorrelation traces at −0.9 V in a 50-ps span and (c) 4-ps span (d) optical spectrum at −1.1 V (e) autocorrelation traces at −1.1 V in a 50-ps span and (f) 4-ps span
To further investigate the phase relationship among the modes as shown in Fig. 5(d), a photonic down conversion technique was used to examine the coherence of the laser modes. As schematically shown in Fig. 6(a), if two adjacent modes A and B with a mode spacing of 50 GHz are in coherence, the beating between the +1 sideband of mode A and the −1 sideband of mode B will generate a sharp tone at an intermediate frequency (IF) detectable using low-frequency components and equipment. Otherwise, the incoherent sidebands will not generate a definite IF beating tone.
Fig. 6. (a) Schematic of the photonic down conversion technique for determining the coherent property of the 50-GHz FP modes. (b) RF spectrum when a 20-GHz RF signal was applied to the SA section.
Figure 6(b) shows the RF spectrum converted from the optical pulses when a 20-GHz RF signal was applied to the SA section. No signal of a 10-GHz IF beating tone was observed from the spectrum. While a 20-GHz tone could be clearly observed, corresponding to a coherent mode beating from the sidebands of the adjacent 100-GHz modes as well as the modulation sidebands themselves. We can conclude that the laser with a poor intermediate mode suppression ratio actually worked in a mixed state comprising a coherent 100-GHz CPM pulsed output and an incoherent F-P continuous wave (CW) output. The time domain performance should be a train of 100-GHz pulses along with a CW floor. The CW floor has no observable influence on the pulse shape and repetition rate of the CPM in the time domain. For situations where only the pulse width and shape matters, the CW floor may not pose a major problem. A post-stage pulse reshaping can also be introduced to remove the DC component resulting from the CW contribution.
The CL-CL CPM laser also showed similar spectral performance, indicating that the HR-CL coating was not the reason for the partial CPM state. We tend to attribute the feature to the influence of the SA position. The physical reason needs further investigation.
For the batch of HR-CL CPM lasers under test, the shortest pulse width of 298 fs was obtained for a laser with a forward current bias of 200 mA and a reverse bias of -1.4 V. Figure 7 shows the autocorrelation traces and the spectrum of the device. A sech2 fit is assumed for pulse width estimation, as shown in Fig. 7(a). It worked at a repetition rate of 102.4 GHz, calculated from the pulse separation of 9.765 ps, as shown in Fig. 7(b). The spectrum shown in Fig. 7(c) indicates that the laser works in a partial CPM state.
Fig. 7. The autocorrelation traces and spectrum of an HR-CL CPM laser when ${I_{Gain}} = 200\,mA$ and ${V_{SA}} = \, - 1.4\,V$ in a measurement span of (a) 50 ps. (b) 150 ps. (c) The spectrum of the HR-CL CPM laser.
A FROG measurement was carried out to further characterize the spectrogram and chirp of the laser. Figure 8(a) shows a well-confined pulse energy in the ±500-fs region of the spectrogram. Figure 8(b) shows the chirp is relatively flat and linear across the main part of the pulse, with the rising and falling part located only on the wings of the pulse, indicating the pulse is still dominated by the GVD instead of the nonlinear SPM effect. The pulse width was calculated to be 459 fs, assuming a sech2 shape, from the FROG measurement. The pulse was broadened compared to that shown in Fig. 7(a) due to the dispersion introduced by an extra single-mode fiber of about 2-3 meters to connect the optical signal to the FROG. The time-bandwidth product (TBP) was calculated to be 0.354 using the built-in function of the FROG.
Fig. 8. FROG measurement of the HR-CL CPM laser when ${I_{Gain}} = 200\; mA$ and ${V_{SA}} = \; - 1.4\; V$, (a) spectrogram; (b) time and phase retrieval.
4. Conclusion
We reported a CPM laser scheme with asymmetric cladding and coating for high-power short pulses generation. The effect of facet reflectivity on the pulse performance of the CPM laser was investigated, showing an overall performance improvement in both the peak power and pulse width when the CPM laser adopts an HR-CL coating design. A dilute waveguide structure and a high-power design were used to increase the gain saturation energy, reduce the intracavity loss, and suppress the SPM effect. With an asymmetric coating of HR (95%)-CL, 100-GHz sub-picosecond optical pulses were obtained with the peak power reaching a level of over 1 Watt. In a pure CPM state, short pulses with a peak power of 1.02 W, a pulse width of 564 fs, and an intermediate mode suppression ratio over 40 dB were demonstrated. In a partial CPM state, ultra-short optical pulses with pulse width down to 298 fs were obtained.
Funding
National Key Research and Development Program of China (2019YFB2203800); National Natural Science Foundation of China (62074141).
Acknowledgment
This work was supported by the National Key Research and Development Program of China under Grant No. 2019YFB2203800 and the National Natural Science Foundation of China (NSFC) under Grant No. 62074141
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.
Reference
1. J. H. Marsh and L. P. Hou, “Mode-Locked Laser Diodes and Their Monolithic Integration,” IEEE J. Sel. Top. Quantum Electron. 23(6), 1–11 (2017). [CrossRef]
2. F. R. Ahmad and F. Rana, “Fundamental and Subharmonic Hybrid Mode-Locking of a High-Power (220 mW) Monolithic Semiconductor Laser,” IEEE Photonics Technol. Lett. 20(15), 1308–1310 (2008). [CrossRef]
3. L. Hou, M. Haji, J. Akbar, B. Qiu, and A. C. Bryce, “Low divergence angle and low jitter 40 GHz AlGaInAs/InP 1.55 µm mode-locked lasers,” Opt. Lett. 36(6), 966–968 (2011). [CrossRef]
4. S. Arahira and Y. Ogawa, “480-GHz subharmonic synchronous mode locking in a short-cavity colliding-pulse mode-locked laser diode,” IEEE Photonics Technol. Lett. 14(4), 537–539 (2002). [CrossRef]
5. H. Haus, “Theory of mode locking with a slow saturable absorber,” IEEE J. Quantum Electron. 11(9), 736–746 (1975). [CrossRef]
6. J. Javaloyes and S. Balle, “Anticolliding design for monolithic passively mode-locked semiconductor lasers,” Opt. Lett. 36(22), 4407–4409 (2011). [CrossRef]
7. Y. Liu, R. Zhang, H. Wang, D. Lu, and L. Zhao, “25-GHz Semiconductor Mode-locked Laser with Subpicosecond Pulse Output in the 1.5-µm Band(Invited),” Acta Photonica Sinica 51, 0251211 (2022). [CrossRef]
8. D. J. Derickson, R. J. Helkey, A. Mar, J. R. Karin, J. G. Wasserbauer, and J. E. Bowers, “Short Pulse Generation Using Multisegment Mode-Locked Semiconductor-Lasers,” IEEE J. Quantum Electron. 28(10), 2186–2202 (1992). [CrossRef]
9. G. P. Agrawal and N. A. Olsson, “Self-Phase Modulation and Spectral Broadening of Optical Pulses in Semiconductor-Laser Amplifiers,” IEEE J. Quantum Electron. 25(11), 2297–2306 (1989). [CrossRef]
10. M. Faugeron, M. Tran, O. Parillaud, M. Chtioui, Y. Robert, E. Vinet, A. Enard, J. Jacquet, and F. V. Dijk, “High-Power Tunable Dilute Mode DFB Laser With Low RIN and Narrow Linewidth,” IEEE Photonics Technol. Lett. 25(1), 7–10 (2013). [CrossRef]
