Open Access
Add to BibSonomy
Abstract
A III/V-on-Bulk-Si DFB laser with a long phase shift section optimized for single-mode stability is presented. The optimized phase shift allows stable single-mode operations up to 20 times a threshold current. This mode stability is achieved by a gain difference between fundamental and higher modes maximized by sub-wavelength-scale tuning of the phase shift section. In SMSR-based yield analyses, the long-phase-shifted DFB laser showed superior performance compared to the conventional λ/4-phase-shifted ones.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Distributed-feedback (DFB) laser diodes (LDs) are single-wavelength light sources that are most widely used in a variety of applications including optical communication and optical sensing [1–6]. The main performance of the DFB LDs includes single-mode purity measured by side-mode suppression ratios(SMSRs) and single-mode operation range, in addition to general light source performance including high power, high efficiency, low linewidth, and high modulation speed [1]. In particular, it is well-known that manufacturing yield, which determines the DFB LD cost, is most affected by the single-mode characteristics. Therefore, in the DFB LD development, it is important to improve the general performance without sacrificing the single-mode performance and vice versa, as a trade-off may exist between the two performances. For the general DFB performance such as output power and efficiency, it is necessary to increase its external efficiency and suppress the spatial hole burning (SHB) [2,3]. To this end, there have been various proposals incorporating asymmetric phase shift (PS) or grating [4], corrugation-pitch-modulation [5], multi-PS [6], and long PS [7–10]. Although the simplest long PS structure has been widely adopted, the solution against the single-mode instability accompanying the structure has been unknown for the DFB LDs.
From an industry perspective, a major trend in the development of photonic integrated circuits (PICs) including the DFB LD is the emergence of silicon(Si) photonics, which leverages the complementary-metal-oxide-semiconductor (CMOS) infrastructure of the Si industry [10–23]. In particular, the recently emerging III/V-on-Si platform enables single-chip integration including the light source, so expectations are rising for the commoditization of PICs in various applications including datacenter and light detection and ranging (LiDAR) [17–24]. Therefore, recent light source development tends to pursue single-chip integration along with various photonic devices on top of a Si substrate instead of a conventional III/V one. In the incumbent CMOS industry, the integration tendency has been further strengthened so that the generic bulk-Si(BS) substrate is preferred instead of the specialty Si-on-insulator(SOI) substrate for light source development to facilitate future integration with volume CMOS products such as DRAM [25–29].
This paper presents design and verifying experimental results that avoid single-mode instability from the long PS in the III/V-on-BS DFB LD. We present the change in single-mode stability over super-λ-scale and sub-λ-scale tuning of the PS length with a good agreement between design and measurement, and improved yield with the optimized design. This paper discusses on design and measurement only, because the fabrication processes on the BS substrate have been previously published [25–29]. This paper also focuses on the single-mode performance only, since the general performances including output power, wall plug efficiency(WPE), modulation speed, and thermal resistance have been also previously reported [9,10]. This result is expected to contribute to mass production of low-cost single-wavelength light sources in the CMOS-based Si photonics industry.
2. Concept
The vertical and lateral structure of the III/V-on-BS DFB LD is illustrated in Fig. 1. The vertical structure of Fig. 1(a) and 1(b) includes the Si waveguide patterned in the local SOI structure formed on the BS substrate, the III/V epitaxial stack containing the multiple quantum wells(MQWs), and the current channel defined by the proton implantation. Since the III/V-on-BS waveguide confines the light in the horizontal and vertical directions, the optical mode is formed around the bonding interface of Si and III/V as shown in the inset of Fig. 1(a). Depending on the design of the Si waveguide and the III/V stack, the vertical distribution of the optical mode can be adjusted. As the III/V portion of the optical mode distribution increases, the optical gain increases but the internal optical loss also increases. The local SOI structure on the BS substrate was implemented through a proprietary solid-phase epitaxy (SPE) process, and its details have been published in the literature [9,10,25–29]. In the horizontal structure in Fig. 1(c), the DFB LD including the long PS is connected to the grating coupler through the mode converter and Si waveguide. As the III/V portion of the optical mode distribution increases, the optical loss of the mode conversion also increases. The grating coupler for performance evaluation inevitably affected the device yield measurement result as shown in the measurement section because the optical reflection was not completely suppressed despite its low reflection design.
Fig. 1. Conceptual illustration of III/V-on-bulk-Si DFB LD. (a) Vertical structure. The inset is a simulated optical mode. (b) SEM image of cross-section of the fabricated device. (c) Planar structure.
Figure 2 illustrates the DFB cavities of no PS, λ/4 PS, and long PS for the III/V-on-Si DFB LD. The characteristics of each cavity are well-known and can be qualitatively understood by the corresponding S21 spectrum [1]. The S21 is the (2, 1) element of the 2×2 scattering matrix of the DFB cavity and relates to the cavity transmittance. The cavity without PS in Fig. 2(a) has the low transmittance in the Bragg reflection band, and lasing occurs at the two edges of the reflection band when optical gain is provided, making it difficult to secure single-mode stability. In this cavity, single mode is possible for discrete devices that can exploit reflections at chip facets, but single mode is impossible for integrated devices without the facets. The λ/4-PS cavity in Fig. 2(b) has the high transmission mode in the center of the Bragg reflection band where lasing occurs when optical gain is provided, making it the most stable single-mode cavity. However, as the longitudinal optical mode distribution is concentrated in the short PS section, it becomes nontrivial to improve the general performance such as optical power, WPE, and modulation speed of the DFB LD. While various cavity structures to mitigate the longitudinal mode concentration have been proposed, and the long PS has been widely used thanks to its simplicity. In this structure, the longitudinal mode concentration is alleviated by the long PS, so it is relatively easier to improve the general performance, but it is critical to retain the single-mode stability that can be degraded without special measures. As shown in the S21 spectrum, the lasing mode in the center of the Bragg reflection band still exists in the long PS structure, but the lasing possibility of the band-edge modes also increases as the super-λ-scale PS length(Lps) increases. Moreover, the position of the lasing mode in the reflection band shifts over the change of the sub-λ-scale PS length(δps). Therefore, it is critical to precisely optimize the PS lengths(Lps + δps) for the single-mode stability. The next design section presents how the mode selectivity of the long-PS cavity changes over the Lps and δps, and the optimal single-mode design.
Fig. 2. Conceptual illustrations of DFB cavities. (a) Standard DFB cavity without phase shift. (b) DFB cavity with λ/4 phase shift. (c) DFB cavity with long phase shift. Lps and δps are super-λ-scale and sub-λ-scale PS lengths. The corresponding S21 spectrum is shown in the right side for each cavity.
3. Design
Figure 3 describes the DFB cavity model used in this paper and its comparison to the model from the literature [1]. The model described in Fig. 3(a) is based on the transfer matrix method (TMM) extended for the long-PS DFB cavity. The DFB transfer matrix, TDFB, is the matrix multiplication of the transfer matrixes of the gratings and PS. The threshold modal gain (TMG) of the DFB cavity is obtained numerically under the condition that the (1, 1) element of TDFB is zero. The model was validated through the comparison in Fig. 3(b) showing the TMGs in good agreement with the literature for the λ/4-PS cavity. The small discrepancy is presumably due to structural differences between III/V (literature) and III/V-on-Si (this paper) or the nature of numerical calculations with finite precision. The most uncertain part in the TMM modeling is the refractive index calculations according to the optical mode distribution of the III/V-on-Si waveguide. The finite-difference time-domain (FDTD)-based refractive index calculations of the gratings are also inherently uncertain because the III/V stack consists of multiple layers of different material combinations with large uncertainties in the refractive index profiles. Therefore, the DFB LD has been developed with calibrations through design-measurement iterations.
Figure 4 summarizes the single-mode stability change from the TMM model over the super-λ-scale length of Lps and sub-λ-scale length of δps. The single-mode stability can be analyzed from the TMG difference between the fundamental mode and the first higher-order mode. That is, the single mode is stable if the TMG of the fundamental mode is significantly lower than that of the higher-order mode. The effect of the PS length is a combination of the slow envelop over Lps in Fig. 4(a) and the fast oscillation over δps in Fig. 4(b). For example, from the 0-PS cavity to the λ/4-PS cavity, the single-mode stability dramatically changes with the insertion of the PS of λ/4, which repeats throughout the PS length change. Figure 4(a) shows the slow super-λ-scale envelop with the TMG difference continuously decreasing over Lps. Here, to decouple the sub-λ-scale oscillations from the super-λ-scale envelop, the fundamental mode is located at the center of the Bragg reflection band in all the calculated points. The dotted line in the TMG plot is from the previously reported TMG criterion for single-mode stability [4]. Since Lps is preferred to be less than ∼60µm, Lps was designed to be 25 µm and 50 µm in this paper. Figure 4(b) shows the slow sub-λ-scale oscillation with the TMG difference oscillates significantly with the small change of δps. For Lps of 25 µm and 50 µm, the optimal δps are found to be 0 nm and half of the grating period (Pg), respectively. The wavelength difference and the SMSR show the decreasing envelopes over Lps and the fast oscillations over δps similar to the TMG difference, as shown in the middle and bottom of Fig. 4(a) and 4(b). Note that the wavelength difference decreases measurably with Lps, while the SMSR does not. In the SMSR modeling, the operating current is five times its threshold current. Figure 4(c) plots the TMGs of a few lowest modes for the main combinations of Lps and δps. The third plot in the first row (Lps = 0, δps = Pg/2) is the traditional λ/4-PS cavity, where the TMG difference is maximized with the highest single-mode stability. The first plot of the second row (Lps = 25 µm, δps = 0) and the third plot of the third row (Lps = 50 µm, δps = Pg/2) are the long-PS cavities chosen in this paper for maximized single-mode stabilities.
Fig. 4. Single-mode stability simulation over phase shift length. Normalized TMG difference, lasing wavelength difference, and SMSR over phase shift lengths, Lps(a), and δps(b). Normalized TMGs of fundamental and a few higher-order modes over Lps and δps. The green box indicates the λ/4 cavity. The blue boxes indicate the cavities chosen for single-mode stability.
4. Measurement
Figure 5 summarizes the measurement results of the single-mode stability of the III/V-on-BS DFB LDs designed with the long-PS cavities discussed above. In this paper, the single-mode stability analyses are based on spectral characterizations, and the linewidth characterization needs to be added in the future for next-generation Datacenter and LiDAR applications. The average threshold current of the fabricated DFB LDs was ∼14 mA at 25°C, and the single-mode performance was evaluated at 2-mA intervals from 20 to 180 mA of operating current, which corresponds to 1.4 to 20 times the threshold current. The output light of the DFB LD was input to an optical spectrum analyzer through the on-chip grating coupler and an optical fiber probe, and the wavelengths of the first and second highest peaks and SMSR were acquired. The full spectrum was also acquired at each operating current interval of 10 mA. Figure 5(a) ∼ (d) and Fig. 5(e) ∼ (h) show the single-mode stability change over the δps change at Lps of 25 µm and 50 µm, respectively. As discussed in the design section, the single-mode retention range decreases with increasing Lps, and the single-mode stability strongly depends on δps. The optimal δps are found to be 0 and Pg/2 at Lps of 25 µm and 50 µm respectively, which are in good agreement with the model prediction. From Lps of 25 µm to 50 µm, the wavelength difference decreases but the SMSR did not, which are also well-agreed with the model.
Fig. 5. Measured single-mode stability of fabricated DFB LDs over sub-λ-scale PS tuning around Lps of 25 µm [(a) ∼ (d)] and 50 µm[(e) ∼ (h)] with target kLg of 5.9. The blue boxes indicate the cavities chosen for single-mode stability in Fig. 4.
Figure 6 shows the early-stage results of the single-mode yield from the III/V-on-BS DFB LDs with 50-µm PS. The SMSR of >30 dB is considered single-mode in this paper. From the SMSRs of 287 DFB LDs having the same cavity of 50-µm PS, the single-mode yields were 56% and 22% with the average and the minimum SMSRs over the operating range, respectively. Figure 6(b) compares the single-mode yields of the 50-µm-PS cavity and λ/4-PS cavity. With 70 DFB LDs of the two cavities co-fabricated on the same wafer, the single-mode yields based on the minimum SMSR were 30% and 12% for the 50-µm-PS cavity and λ/4-PS cavity, respectively. The 50-µm-PS cavity exhibited the unexpectedly higher single-mode yield compared to the λ/4-PS cavity despite its lower TMG difference. It is presumably because the optical reflection at the on-chip grating coupler degraded overall certainty of the yield analysis and the device count was also not large enough due to low process yield at the moment. Further yield analyses are planned after increasing the process yield and suppressing the optical reflection in the future.
Fig. 6. Cumulative probability plots with SMSR performance. (a) Analysis with 287 DFB LDs with the optimized PS of 50 µm. (b) Analysis with 70 DFB LDs co-fabricated with the PSs of λ/4 and 50 µm.
5. Summary
A Single-mode light source should have both general performance and single-mode stability. In particular, it is important to secure the single-mode stability that determines the yield and cost of the single-mode light source. For the III/V-on-BS DFB LD, this paper presents the design and experimental verification of maximizing the single-mode stability while adopting the long-PS cavity that is advantageous for general performance improvement. By optimizing the super-λ-scale and sub-λ-scale PS lengths, the single-mode stability was achieved over the wide range of operating currents, resulting in the higher single-mode yield measurements compared to the typical λ/4-PS cavity. The DFB LD in this paper has been developed on the generic BS substrate for integration into volume CMOS products and is therefore expected to be used in various applications including DRAM interfaces, Datacenter interconnects, co-packaged optics, and chip-scale LiDAR.
Disclosures
The authors declare no conflict 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. L. A. Coldren, S. W. Corzine, and M. L. Mashanovitch, Diode lasers and photonic integrated circuits, 2nd ed. (Wiley-Interscience, 2012).
2. M. Aoki, K. Uomi, T. Tsuchiya, S. Sasaki, M. Okai, and N. Chinone, “Quantum size effect on longitudinal spatial hole burning in MQW λ/4-shifted DFB lasers,” IEEE J. Sel. Top. Quantum Electron. 27(6), 1782–1789 (1991). [CrossRef]
3. P. Doussiere, C. Shieh, S. Demars, and K. Dzurko, “Very high power 2007 nm InP single mode distributed feedback laser diode with reduced linewidth,” Proc. SPIE 6485, 64850G (2007). [CrossRef]
4. O. Kwon, Y. Leem, D. Lee, C. Lee, Y. Baek, and Y. Chung, “Effects of asymmetric grating structures on output efficiency and single longitudinal mode operation in λ/4-shifted DFB laser,” J. Lightwave Technol. 47(9), 1185–1194 (2011). [CrossRef]
5. M. Okai, “Spectral characteristics of distributed feedback semiconductor lasers and their improvements by corrugation-pitch-modulated structure,” J. Appl. Phys. 75(1), 1–29 (1994). [CrossRef]
6. S. Ogita, Y. Kotaki, M. Matsuda, Y. Kuwahara, and H. Ishikawa, “Long-cavity multiple-phase-shift distributed feedback laser diode for linewidth narrowing,” J. Lightwave Technol. 8(10), 1596–1604 (1990). [CrossRef]
7. S. Srinivasan, A. W. Fang, D. Liang, J. Peters, B. Kaye, and J. E. Bowers, “Design of phase-shifted hybrid silicon distributed feedback lasers,” Opt. Express 19(10), 9255–9261 (2011). [CrossRef]
8. C. Zhang, S. Srinivasan, Y. Tang, M. J. R. Heck, M. L. Davenport, and J. E. Bowers, “Low threshold and high speed short cavity distributed feedback hybrid silicon lasers,” Opt. Express 22(9), 10202–10209 (2014). [CrossRef]
9. D. Shin, J. Cha, S. Kim, Y. Shin, K. Cho, K. Ha, G. Jeong, H. Hong, K. Lee, and H. Kang, “O-band DFB laser heterogeneously integrated on bulk-silicon platform,” Opt. Express 26(11), 14768–14774 (2018). [CrossRef]
10. D. Shin, J. Cha, Y. Shin, K. Ha, C. Lee, C. Shin, D. Shim, B. Lee, H. Hong, K. Lee, and H. Kang, “Heterogeneously integrated light sources on bulk-silicon platform,” in Proc. IEEE Int. Electron Devices Meeting, 2018, pp. 23.6.1–23.6.4.
11. D. Liang and J. E. Bowers, “Recent progress in lasers on silicon,” Nat. Photonics 4(8), 511–517 (2010). [CrossRef]
12. J. E. Bowers, “Heterogeneous photonic integration on silicon,” Proc. of IPC2016, WC2.1 (2016).
13. M. J. R. Heck, J. F. Bauters, M. L. Davenport, J. K. Doylend, S. Jain, G. Kurczveil, S. Srinivasan, Y. Tang, and J. E. Bowers, “Hybrid silicon photonic integrated circuit technology,” IEEE J. Sel. Top. Quantum Electron. 19(4), 6100117 (2013). [CrossRef]
14. H. Park, M. N. Sysak, H. Chen, A. W. Fang, D. Liang, L. Liao, B. R. Koch, J. Bovington, Y. Tang, K. Wong, M. Jacob-Mitos, R. Jones, and J. E. Bowers, “Device and integration technology for silicon photonic transmitters,” IEEE J. Sel. Top. Quantum Electron. 17(3), 671–688 (2011). [CrossRef]
15. S. Stankovic, R. Jones, M. N. Sysak, J. M. Heck, G. Roelkens, and D. V. Thourhout, “Hybrid III-V/Si distributed-feedback laser based on adhesive bonding,” IEEE Photonics Technol. Lett. 24(23), 2155–2158 (2012). [CrossRef]
16. S. Keyvaninia, S. Verstuyft, L. V. Landschoot, F. Lelarge, G. H. Duan, S. Messaoudene, J. M. Fedeli, T. D. Vries, B. Smalbrugge, E. J. Geluk, J. Bolk, M. Smit, G. Morthier, D. V. Thourhout, and G. Roelkens, “Heterogeneously integrated III-V/silicon distributed feedback lasers,” Opt. Lett. 38(24), 5434–5437 (2013). [CrossRef]
17. J. C. Hulme, J. K. Doylend, M. J. R. Heck, J. D. Peters, M. L. Davenport, J. T. Bovington, L. A. Coldren, and J. E. Bowers, “Fully integrated hybrid silicon two dimensional beamscanner,” Opt. Express 23(5), 5861–5874 (2015). [CrossRef]
18. J. K. Doylend and S. Gupta, “An overview of silicon photonics for LiDAR,” Proc. SPIE 11285, 112850J (2020). [CrossRef]
19. C. Shin, D. Shin, H. Byun, J. Kim, J. Lee, C. Lee, K. Son, I. Hwang, D. Shim, B. Jang, E. Lee, H. Choo, and K. Ha, “Spatially-chirped optical phased array design and its 10-m LiDAR operation,” in Proc. Conf. Lasers Electro-Opt., OSA Tech. Dig., 2021, paper STu4D.2.
20. H. Byun, J. Lee, B. Jang, C. Lee, E. Lee, I. Hwang, C. Shin, D. Shin, D. Shim, T. Otsuka, S. Hwang, H. Choo, and K. Ha, “A gain-enhanced silicon-photonic optical phased array with integrated O-band amplifiers for 40-m ranging and 3D scan,” in Proc. Conf. Lasers Electro-Opt., 2020, pp. 1–2.
21. J. Lee, D. Shin, B. Jang, H. Byun, C. Lee, C. Shin, I. Hwang, D. Shim, E. Lee, J. Kim, K. Son, T. Otsuka, K. Ha, and H. Choo, “Single-chip beam scanner with integrated light source for real-time light detection and ranging,” in Proc. IEEE Int. Electron Devices Meeting, 2020, pp. 7.2.1–7.2.4.
22. H. Byun, Y. Cho, I. Hwang, B. Jang, J. Kim, C. Lee, E. Lee, J. Lee, T. Otsuka, D. Shim, C. Shin, D. Shin, K. Son, H. Choo, and K. Ha, “Single-chip beam scanner LiDAR module for 20-m imaging,” in Proc. IEEE Int. Electron Devices Meeting, 2021, pp. 14732–14745.
23. D. Shin, I. Hwang, E. Lee, C. Shin, J. Lee, H. Byun, D. Shim, B. Jang, C. Lee, K. Son, T. Otsuka, H. Choo, and K. Ha, “Commoditizing the uncommoditized: Chip-scale LiDAR,” in Proc. SPIE Photon. West, Opt. Interconnects XXII, 2022, pp. 12007–12012.
24. K. Son, D. Shin, J. Lee, B. Jang, D. Shim, H. Byun, C. Lee, Y. Cho, T. Otsuka, C. Shin, I. Hwang, E. Lee, K. Ha, and H. Choo, “Palm-sized LiDAR module with III/V-on-Si optical phased array,” in Proc. Sym. VLSI, 2022.
25. D. Shin, K. Cho, H. Ji, B. Lee, S. Kim, J. Bok, S. Choi, Y. Shin, J. Kim, S. Lee, K. Cho, B. Kuh, J. Shin, J. Lim, J. Kim, H. Choi, K. Ha, Y. Park, and C. Chung, “Integration of silicon photonics into DRAM process,” Proc. of OFC2013, OTu2C (2013).
26. K. Ha, D. Shin, H. Byun, et al., “Si-based optical I/O for optical memory interface,” Proc. of SPIE OPTO2012 (2012).
27. H. Byun, J. Bok, K. Cho, et al., “Bulk-Si photonics technology for DRAM interface,” Photonics Res. 2(3), A25–A33 (2014). [CrossRef]
28. D. Shin, H. Byun, D. Shim, J. Cha, Y. Shin, C. Shin, C. Lee, E. Lee, B. Jang, J. Lee, I. Hwang, J. Kim, K. Son, K. Ha, and H. Choo, “III/V-on-bulk-Si technology for commercially viable photonics-integrated VLSI,” in Proc. Sym. VLSI, JFS1.4, 2020.
29. D. Shin, H. Byun, D. Shim, J. Cha, Y. Shin, C. Shin, C. Lee, E. Lee, B. Jang, J. Lee, I. Hwang, K. Son, Y. Cho, T. Otsuka, H. Choo, and K. Ha, “Bulk-Si platform: Born for DRAM, upgraded with on-chip lasers, and transplanted to LiDAR,” J. Lightwave Technol. 40(10), 3137–3148 (2022). [CrossRef]
