Open Access
Add to BibSonomy
Abstract
We demonstrate 20-μm-long twin-mirror membrane distributed-reflector (DR) lasers for chip-to-chip optical interconnects. The lasers employ distributed Bragg reflectors (DBRs) at both ends of a 20-μm-long λ/4-phase shifted distributed feedback (DFB) section. We achieve single-mode lasing in a λ/4-phase shifted DFB mode at room temperature with a threshold current of 0.39 mA. The lasing wavelength remains stable while the injected current is varied, and it is determined by the λ/4 phase-shifted DFB. The modulation current efficiency is 11.4 GHz/mA1/2, which is measured by using relative intensity noise spectra. We also demonstrate the direct modulation of the DR lasers at a bit rate of 25.8 Gbit/s with an energy cost of 163 fJ/bit.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Thanks to the rapid increase in Internet traffic, the transmission capacity in datacenters is increasing [1], and this in turn is increasing the power consumed for data transmission in facilities ranging from datacenters to CPUs. Thus, technologies designed to reduce power consumption have attracted a lot of attention, and there is a strong need for optical interconnections for chip-to-chip use to reduce the power consumption of CPUs [2]. According to [2], an operating energy of less than 34 fJ/bit is required if we apply optical links to chip-to-chip interconnects. Vertical-cavity surface-emitting lasers (VCSELs) are potential candidates because they are widely used for data transmissions ranging from a few meters to ~100 m with a low operating energy. They have already exhibited a low operating energy of 77 fJ/bit, which was achieved by reducing the diameter of the current injection region to 3.5 μm [3]. However, a further reduction in active region volume and the precise wavelength control are needed if we are to employ wavelength division multiplexing (WDM) technologies.
In this context, reducing the cavity length of in-plane lasers, such as distributed feedback (DFB) lasers, has attracted much attention because the lasing wavelength can be easily controlled by changing the grating pitch. We have already developed in-plane lasers employing photonic crystal cavities [4–6]. By employing a 2.6-μm-long cavity with a current blocking trench, we achieved an energy cost of 4.4 fJ/bit [6]. The key to realizing such an ultra-low operating energy is to obtain a large optical confinement factor by using a membrane structure in addition to an ultra-short cavity. Membrane DFB and distributed reflector (DR) lasers consisting of a DFB laser and a distributed Bragg reflector (DBR) have also been developed [7–12]. Recently, we succeeded in reducing the active region length of a DR laser to 75 μm and demonstrated single-mode lasing that had a power consumption of less than 100 fJ/bit with a 25.8-Gbit/s non-return to zero (NRZ) signal [13]. We also fabricated a DR laser with 50-μm long active region [11]. However, its threshold current was 0.8 mA and modulation-current efficiency (MCEF) was 9.4 GHz/mA0.5, which were in the same range as those of the 75-μm-long laser (0.77 mA and 9.1 GHz/mA0.5). Another group has reported an 80-μm-long DFB laser [14] whose threshold current was 0.27 mA and MCEF was 9.9 GHz/mA0.5. They reduced the active region length to 40 μm using DR structures, and its threshold current was 0.44 mA [15]. These results indicate that it is difficult to reduce the threshold current and to increase MCEF by reducing the cavity length, because of the increase in grating loss due to the large coupling coefficient, which is essential for shortening the cavity length.
However, we have to reduce the active region length so that power consumption is reduced enough to introduce optical technologies into chip-to-chip interconnects. Therefore, we consider a cavity structure to reduce the grating coupling coefficient for 20-μm-long lasers. If the active region length is reduced to around 20 μm, we can expect the energy cost to fall to ~30 fJ/bit, which is small enough for the required operating energy of chip-to-chip interconnections. A straightforward way to reduce the cavity length is to increase the coupling coefficient of the grating. In this case, we have to increase the etching depth of the grating to increase the coupling coefficient, which makes precise control of the lasing wavelength difficult. In addition, an increase in grating depth increases the scattering loss in the grating [14]. Therefore, we have to find a way to reduce the cavity length without increasing the coupling coefficient.
In this paper, we report 20-μm-long phase-shifted twin-mirror DR lasers on SiO2/Si substrates, which have DBRs at both ends of 20-μm-long λ/4-phase shifted DFB sections. In Section 2, we numerically compare the required coupling coefficient for DFB, single-mirror DR and twin-mirror DR lasers. By employing twin mirrors for DR lasers, the required coupling coefficient is greatly reduced to ~1000 cm−1, whereas the DFB lasers require a value exceeding 2700 cm−1. We also discuss the fabrication tolerances and wavelength variations of the twin-mirror DR lasers. The device fabrication process is described in Section 3. The devices consist of InP-based membranes on SiO2/Si substrates. We fabricated twin-mirror DR lasers with various lengths of DFB sections. In Section 4, we report the measurement characteristics of the twin-mirror DR lasers. We achieved single-mode lasing at the DFB Bragg wavelength with a threshold current of 0.39 mA, and an MCEF of 11.4 GHz/mA0.5. We also describe the direct modulation characteristics at a bit rate of 25.8 Gbit/s.
2. Cavity design
Our previously developed DR lasers consisted of a DFB section with a length of more than 50 μm and a single rear DBR [11]. To reduce the active region length of the single-mirror DR lasers, we have to increase the grating coupling coefficient by increasing the etching depth, which makes it difficult to control the lasing wavelength precisely. This is because a deep grating increases the variation in the etching depth, which causes deviations of both the center wavelength and the grating bandwidth. The increase in grating depth also increases the scattering loss in the grating [16], which results in reduced reflectivity. Thus, we consider twin-mirror DR lasers, which consist of a DFB section and front/rear DBRs, to reduce the required coupling coefficient of the gratings. Two mirrors increase the photon lifetime of the cavity thus allowing us to reduce the coupling coefficients. We compared three cavity designs employing λ/4-phase shifted DFB as shown in Fig. 1(a): DFB lasers, single-mirror DR lasers, and twin-mirror DR lasers. As mentioned above, we used an active length of 20 μm to achieve an energy cost of ~30 fJ/bit. We selected a rear DBR length of 50 μm because it provided almost 100% reflectivity when the grating coupling coefficient exceeded 700 cm−1. We calculated the threshold gain (Γgth) of the three cavities for different grating coupling coefficients using the coupled wave theory [17], as shown in Fig. 1(b). The threshold gain is given by
where Γ is the confinement factor, αi is the internal loss of the active region, αm is the mirror loss, and αloss is the grating scattering loss in the DFB and DBR sections. The internal loss αi and the grating loss αloss were assumed to be 0 cm−1 in the calculations. To obtain a threshold gain of 50 cm−1, the required coupling coefficients were 2700 and 2200 cm−1 when we used DFB lasers and single-mirror DR lasers, respectively. In addition, the products of the coupling coefficient and DFB-section length (κL) were 5.4 and 4.8 for the DFB lasers and single-mirror DR lasers, respectively, which could cause the problem of spatial hole burning [18].
Fig. 1 Short cavity lasers with a 20-μm-long active region. (a) Schematic of three cavity designs employing λ/4-phase shifted DFB: DFB lasers, single-mirror DR lasers, and twin-mirror DR lasers. (b) Threshold gain versus coupling coefficient of the grating. The internal loss αi and the grating loss αloss were assumed to be 0 cm−1 in the calculations.
We also calculated the dependence of the threshold gain on the front mirror length for the twin-mirror DR lasers. The front mirror length ranged from 10 to 30 μm. The coupling coefficient needed to obtain a threshold gain of 50 cm−1 was less than 1000 cm−1. This was in the same range as our previously developed DR lasers with ~100-μm-long cavities. Therefore, twin-mirror DR lasers are suitable for the 20-μm-long active region without increasing the coupling coefficient. However, the two DBRs form an additional Fabry-Pérot (FP) cavity, which might cause multiple-mode lasing in both the DFB and FP modes. Therefore, it is important to analyze lasing modes and consider the fabrication tolerance when we employ the twin-mirror DR lasers.
For our device, we fabricated the DFB and the DBR sections by employing a buried heterostructure (BH) and an InP waveguide, respectively. First, we assumed that there was a difference between the Bragg wavelengths of the DFB and DBR sections caused by fabrication error. Figure 2(a) shows the threshold gain (Γgth) as a function of the Bragg wavelength difference between the DFB and DBRs (Δλ) for twin-mirror DR lasers when threshold gain of the DFB mode was smaller than that of the FP modes. In these calculations, we assumed that the coupling coefficients were 974, 596 and 447 cm−1 for front DBR with a length of 10, 20 and 30 μm, respectively, to obtain a threshold gain of 50 cm−1. The threshold gain was increased while increasing Δλ, and the increments were smaller for a shorter DBR. Figure 2(b) shows the lasing wavelength change against Δλ. The lasing wavelength change was suppressed by reducing the front DBR length, i.e. increasing the coupling coefficient of the grating. These results show that lasers with shorter DBRs have a large tolerance in terms of threshold gain and lasing wavelength against the detuning of the center wavelength of the DFB and DBR.
Fig. 2 Fabrication error tolerance of the twin-mirror DR lasers with front DBR lengths of 10, 20, and 30 μm. We assumed there was a Bragg wavelength difference between the DFB and DBR sections. (a) Threshold gain (Γgth) and (b) lasing wavelength versus Bragg wavelength difference between the DFB and DBRs (Δλ).
Next, we considered the positional error between the active regions and the InP-DBR waveguides. As mentioned above, the DFB and DBR sections had different structures. The grating pitch of the DFB was shorter than that of the DBR because of the effective refractive index difference between the two sections. If there was shrinkage or displacement of the active region as shown schematically in Fig. 3(a), a phase change occurred because the grating pitch between the DFB and DBR sections was different. In the following calculations we assume that the active region shrank a total of 0.3 μm due to the etching process. Figure 3(b) shows the relationship between the displacement of the active region and the threshold gain (Γgth). A positive displacement means the active region shifts in a rear direction. In these calculations, we assumed the grating coupling coefficient to be 816 cm−1 for the DFB and 1095 cm−1 for the DBR. As shown in Fig. 3(b), the threshold gain was almost stable when the displacement of the active region occurred. Figures 3(c) and 3(d) show the difference between the threshold gains of the DFB and FP modes (ΔΓgth) and the change in the lasing wavelength as a function of the displacement, respectively. If the displacement is within ± 150 nm, ΔΓgth remains greater than 50 cm−1 and the wavelength change is less than 0.2 nm. These results show that our twin-mirror DR lasers have sufficient tolerance for the fabrication errors.
Fig. 3 Fabrication error tolerance of twin-mirror DR lasers. We assumed that the active region shrank by 0.3 μm with variable displacement due to lithographic error. (a) Schematic explanation of the shrinkage or displacement of the active region. (b) Threshold gain (Γgth), (c) difference between threshold gains of DFB and FP modes (ΔΓgth), and (d) lasing wavelength versus displacement of active region.
3. Device fabrication
Figure 4(a) shows a bird’s eye view of the twin-mirror DR lasers. We located DBR mirrors at both ends of a 20-μm-long λ/4-phase shifted DFB section. The lengths of the front and rear DBR mirrors were 10 and 50 μm, respectively. The asymmetric DBR lengths mean that the light can be emitted selectively from the front. Figures 4(b) and 4(c) show cross-sectional schematics of the active and passive regions, respectively. The active region had a 250-nm-thick InP-based membrane on a 2-μm-thick thermally oxidized Si substrate, which included nine-period multiple-quantum wells (9QW) with a total thickness of 150 nm. The DBR mirror was made of a 1.5-μm-wide InP waveguide.
Fig. 4 Structure of twin-mirror DR lasers: (a) Bird’s eye view of the device. Cross-sectional schematic of (b) active region and (c) passive region.
To fabricate twin-mirror DR lasers on a SiO2/Si substrate, we employed BH regrowth after the direct bonding of III-V active layers [8, 10]. First, we grew InP-based layers including the 9QW and InGaAs etch stop layers on InP substrates using metal-organic vapor phase epitaxy (MOVPE). Next, as shown in Fig. 5(b), we bonded the InP and thermally oxidized Si substrates using O2 plasma-assisted direct bonding. Then, the InP substrates and the InGaAs etch stop layers were removed by lapping and wet etching [Fig. 5(c)]. By using these processes, we transferred a thin InP-based membrane to the Si substrates. A SiO2 mask was deposited, followed by photolithography and etching to form mesa stripes [Fig. 5(d)]. Then undoped InP layers were re-grown by MOVPE to form a BH [Fig. 5(e)]. We employed lateral p-i-n junctions to inject current into the BH. n-type doping and p-type doping were carried out by Si ion implantation and Zn thermal diffusion, respectively [Fig. 5(f)]. Rather than employing embedded gratings, which are commonly used in conventional DFB lasers, we used surface gratings. The surface gratings were formed by electron-beam lithography and the dry etching of the InP surface as shown in Fig. 5(g). The etching depth was approximately 26 nm to obtain the desired coupling coefficient of 1142 cm−1 for the DBR and 852 cm−1 for the DFB. Next, InP waveguides were formed by dry etching [Fig. 5(h)], and Au-based electrodes were formed by a lift-off process [Fig. 5(i)]. Finally, the Si substrates were lapped, and the samples were cleaved into bars.
Fig. 5 Fabrication procedure of twin-mirror DR lasers on SiO2/Si substrates: (a) Epitaxial growth of QWs on InP substrates. (b) O2 plasma-assisted direct bonding of InP and SiO2/Si substrates. (c) InP substrate and InGaAs etch stop layer removal. (d) Forming mesa stripes. (e) MOVPE regrowth of InP to form BH. (f) n- and p-type doping. (g) Etching surface gratings. (h) Forming InP waveguides. (h) Electrode deposition, followed by Si substrate lapping.
4. Device characteristics
Here we discuss the lasing characteristics of twin-mirror DR lasers on a SiO2/Si substrate. The fabricated devices were mounted on a temperature-controlled stage, and all the measurements were carried out at a temperature of 25°C. Figure 6(a) shows the output power and applied voltage versus injected current (L-I-V characteristics). We achieved continuous-wave (CW) operation of the 20-μm-long twin-mirror DR lasers at room temperature with a threshold current of 0.39 mA. The maximum output power was 77 μW, which was measured with lensed fibers, and the differential resistance at an injected current of 1.8 mA was 903 Ω. The calculated fiber coupling efficiency was −7.5 dB. The lasing spectra of the device areshown in Fig. 6(b). We confirmed the single-mode operation of the device by changing the bias current and monitoring the spectra. The lasing wavelength and side-mode suppression ratio (SMSR) were 1541.3 nm and 35.7 dB, respectively, at a bias current of 2 mA. From the stopband width, the grating coupling coefficient was estimated to be 1031 cm−1. This indicates that the actual grating depth was 32.5 nm, which was deeper than designed depth of 26 nm. We also observed small ripples in the spectra, which corresponded to the light reflection from the waveguide facet. The pitch of the small ripples of ~1.23 nm corresponded to the 234-μm-long cavity, which was close to the typical distance between the waveguide facet and the front DBR of 230 μm.
Fig. 6 Static characteristics of the twin-mirror DR lasers: (a) Fiber output power and applied voltage versus injected current. (b) Lasing spectra with injected currents of 2.0 mA.
We also fabricated twin-mirror DR lasers with various DFB lengths. Figure 7(a) showsthe L-I-V characteristics of the twin-mirror DR lasers, which had a 10-μm-long front DBR with DFB lengths of 10, 15, and 20 μm. The threshold current was smaller for the shorter DFB thanks to the small active volume, but the applied voltage was larger due to the increased electrical resistance. The output power of the shorter DFB was saturated with a smaller current injection due to the thermal problem, and the 20 μm DFB provided the best result in terms of output power. The threshold currents are summarized in Fig. 7(b). The lines show calculation results. Propagation losses such as grating scattering loss αloss could reduce the reflection of the cavity, therefore we calculated the threshold current taking αloss in addition to αi. The threshold current is given by
where q is the elementary charge, V is the volume of the active region, B is a bimolecular recombination coefficient, ηi is an injection efficiency and τn is a carrier lifetime. The threshold carrier density Nth is given bywhere Ntr is a transparency carrier density, Ns is a third linearity parameter and g0 is an empirical gain coefficient.Fig. 7 (a) L-I-V characteristics of twin-mirror DR lasers with 10-μm-long front DBRs. The DFB lengths were 10, 15, and 20 μm. (b) Active region length versus threshold current. Dots show experimental values and lines show calculation results.
The employed parameters are listed in Table 1. The αloss was assumed to be same in the DFB and DBR sections. When ηi and αloss were respectively assumed to be 0.58 and 45 cm−1, the calculated results fit the experimental data well. The threshold current can be further reduced for small active region devices by reducing the grating loss and using the longer front DBR, as shown in Fig. 7(b). One possible solution to suppress the grating loss is to reduce the grating depth and improve the grating fabrication process. Note that the leakage current is also a factor limiting the threshold current; however, it is not dominant in this current range because we have already demonstrated 42-μA threshold current using photonic crystal lasers on SiO2/Si substrate [19].
Table 1. Parameters for threshold current calculation
Finally we measured the dynamic characteristics of the twin-mirror DR lasers. We used an erbium doped fiber amplifier (EDFA) and an optical bandpass filter because the fiber-coupled output power from the device was not large enough to measure high-speed characteristics. Figure 8(a) shows the relative intensity noise (RIN) spectrum of the twin-mirror DR lasers, which had a DFB and a front DBR length of 20 and 10 μm, respectively. The relaxation oscillation frequency (fr) was determined by fitting RIN spectra with a numerical model [20], and the result is shown in Fig. 8(b). The modulation-current efficiency (MCEF) was 11.4 GHz/mA1/2. This large MCEF resulted from the small active volume and the strong optical confinement in the active region.
Fig. 8 Dynamic characteristics of the twin-mirror DR laser. (a) Relative intensity noise (RIN) spectra with a bias current of 0.5 to 2.5 mA. (b) Relaxation oscillation frequency (fr) determined by RIN spectra as a function of the square root of the injected current minus the threshold current.
Figure 9 shows an eye pattern. The laser was modulated with a 25.8 Gbit/s non-return-to-zero (NRZ) pseudo-random bit sequence (PRBS) signal with a length of 231 – 1. To obtain a clear eye opening, we set the bias current (Ib) at 1.8 mA and the bias voltage (Vb) at 2.33 V. The energy cost was determined by dividing the applied power by the bit rate, and it was 163 fJ/bit. This calculation is commonly used to qualify the power consumption of VCSELs [3].
Fig. 9 Eye pattern at a bit rate of 25.8 Gbit/s. The bias current was 1.8 mA and the bias voltage was 2.33 V.
5. Conclusion
We have developed 20-μm-long λ/4-phase shifted DR lasers integrated with both front and rear DBRs. The devices consisted of thin InP-based membranes on SiO2/Si substrates. The front and rear DBRs were 10 and 50 μm long, respectively and were optimized in terms of fabrication tolerance with a DFB length of 20 μm. We achieved CW operation at room temperature with a threshold current of 0.39 mA. The lasing wavelength was determined with a λ/4-phase shifted DFB, which is suitable for future WDM applications. We also demonstrated direct modulation at a bit rate of 25.8 Gbit/s with an energy cost of only 163 fJ/bit. Our twin-mirror DR laser is the shortest DR laser ever reported. Although the energy cost is higher than the required value for chip-to-chip optical interconnects, we can reduce it by suppressing the extra scattering loss of the DBRs. We believe these single-mode short cavity DR lasers to be suitable for short distance optical interconnects.
Acknowledgments
We thank Mr. K. Ishibashi, Mr. Y. Shouji, and Mr. Y. Yokoyama for assistance in device fabrication.
References and links
1. Cisco, “Cisco Global Cloud Index: Forecast and Methodology, 2015-2020,” http://www.cisco.com/c/en/us/solutions/collateral/service-provider/global-cloud-index-gci/Cloud_Index_White_Paper.html.
2. D. A. B. Miller, “Device requirements for optical interconnects to silicon chips,” Proc. IEEE 97(7), 1166–1185 (2009). [CrossRef]
3. P. Moser, J. A. Lott, P. Wolf, G. Larisch, H. Li, N. N. Ledentsov, and D. Bimberg, “56 fJ dissipated energy per bit of oxide-confined 850 nm VCSELs operating at 25 Gbit/s,” Electron. Lett. 48(20), 1292–1294 (2012). [CrossRef]
4. S. Matsuo, A. Shinya, T. Kakitsuka, K. Nozaki, T. Segawa, T. Sato, Y. Kawaguchi, and M. Notomi, “High-speed ultracompact buried heterostructure photonic-crystal laser with 13 fJ of energy consumed per bit transmitted,” Nat. Photonics 4(9), 648–654 (2010). [CrossRef]
5. S. Matsuo, T. Sato, K. Takeda, A. Shinya, K. Nozaki, H. Taniyama, M. Notomi, K. Hasebe, and T. Kakitsuka, “Ultra-low operating energy electrically driven photonic crystal lasers,” IEEE J. Sel. Top. Quantum Electron. 19(4), 4900311 (2013). [CrossRef]
6. K. Takeda, T. Sato, A. Shinya, K. Nozaki, W. Kobayashi, H. Taniyama, M. Notomi, K. Hasebe, T. Kakitsuka, and S. Matsuo, “Few-fJ/bit data transmissions using directly modulated lambda-scale embedded active region photonic-crystal lasers,” Nat. Photonics 7(7), 569–575 (2013). [CrossRef]
7. T. Okamoto, N. Nunoya, Y. Onodera, T. Yamazaki, S. Tamura, and S. Arai, “Optically pumped membrane BH-DFB lasers for low-threshold and single-mode operation,” IEEE J. Sel. Top. Quantum Electron. 9(5), 1361–1366 (2003). [CrossRef]
8. S. Matsuo, T. Fujii, K. Hasebe, K. Takeda, T. Sato, and T. Kakitsuka, “Directly modulated buried heterostructure DFB laser on SiO2/Si substrate fabricated by regrowth of InP using bonded active layer,” Opt. Express 22(10), 12139–12147 (2014). [CrossRef] [PubMed]
9. S. Matsuo, T. Fujii, K. Hasebe, K. Takeda, T. Sato, and T. Kakitsuka, “Directly modulated DFB laser on SiO2/Si substrate for datacenter networks,” J. Lightwave Technol. 33(6), 1217–1222 (2015). [CrossRef]
10. T. Fujii, T. Sato, K. Takeda, K. Hasebe, T. Kakitsuka, and S. Matsuo, “Epitaxial growth of InP to bury directly bonded thin active layer on SiO2/Si substrate for fabricating distributed feedback lasers on silicon,” IET Optoelectron. 9(4), 151–157 (2015). [CrossRef]
11. H. Nishi, T. Fujii, K. Takeda, K. Hasebe, T. Kakitsuka, T. Tsuchizawa, T. Yamamoto, K. Yamada, and S. Matsuo, “Membrane distributed-reflector laser integrated with SiOx-based spot-size converter on Si substrate,” Opt. Express 24(16), 18346–18352 (2016). [CrossRef] [PubMed]
12. D. Inoue, T. Hiratani, K. Fukuda, T. Tomiyasu, T. Amemiya, N. Nishiyama, and S. Arai, “Low-bias current 10 Gbit/s direct modulation of GaInAsP/InP membrane DFB laser on silicon,” Opt. Express 24(16), 18571–18579 (2016). [CrossRef] [PubMed]
13. T. Fujii, K. Takeda, E. Kanno, K. Hasebe, H. Nishi, R. Nakao, T. Yamamoto, T. Kakitsuka, and S. Matsuo, “Low operating-energy directly modulated membrane distributed-reflector lasers on Si,” in Proceedings of 42nd European Conference and Exhibition on Optical Communications (2016), pp. 734–736.
14. D. Inoue, T. Hiratani, K. Fukuda, T. Tomiyasu, T. Amemiya, N. Nishiyama, and S. Arai, “High-modulation efficiency operation of GaInAsP/InP membrane distributed feedback laser on Si substrate,” Opt. Express 23(22), 29024–29031 (2015). [CrossRef] [PubMed]
15. T. Hiratani, D. Inoue, T. Tomiyasu, K. Fukuda, N. Nakamura, T. Amemiya, N. Nishiyama, and S. Arai, “High efficiency operation of GaInAsP/InP membrane distributed-reflector laser on Si,” IEEE Photonics Technol. Lett. 29(21), 1832–1835 (2017). [CrossRef]
16. T. Segawa, S. Matsuo, T. Ishii, Y. Ohiso, Y. Shibata, and H. Suzuki, “High-speed wavelength-tunable optical filter using cascaded Mach–Zehnder interferometers with apodized sampled gratings,” IEEE J. Quantum Electron. 44(10), 922–930 (2008). [CrossRef]
17. T. Numai, Fundamentals of Semiconductor Lasers (Springer, 2004).
18. K. Nakahara, Y. Wakayama, T. Kitatani, T. Fukamachi, Y. Sakuma, and S. Tanaka, “1.3 μm InGaAlAs asymmetric corrugation-pitch-modulated DFB lasers with high mask margin at 28 Gbit/s,” Electron. Lett. 50(13), 947–948 (2014). [CrossRef]
19. K. Takeda, T. Fujii, A. Shinya, T. Tsuchizawa, H. Nishi, E. Kuramochi, M. Notomi, K. Hasebe, T. Kakitsuka, and S. Matsuo, “Si nanowire waveguide coupled current-driven photonic-crystal lasers,” in Proceedings of European Conference on Lasers and Electro-Optics (2017), paper CK-9.4. [CrossRef]
20. L. A. Coldren and S. W. Corzine, Diode Lasers and Photonic Integrated Circuits (Wiley-Interscience Publication, 1995).
