November 1, 2014 / Vol. 39, No. 21 / OPTICS LETTERS
6197
Narrow linewidth 1560 nm InGaAsP split-contact corrugated ridge waveguide DFB lasers Kais Dridi,1,* Abdessamad Benhsaien,1 Jessica Zhang,2 and Trevor J. Hall1 1
Photonics Technology Laboratory, Centre for Research in Photonics at the University of Ottawa, 800 King Edward Avenue, Ottawa, Ontario K1N 6N5, Canada 2
CMC Microsystems, 1200 Montreal Road, M50-IPF, Ottawa, Ontario K1A 0R6, Canada *Corresponding author: [email protected]
Received August 11, 2014; revised September 28, 2014; accepted September 28, 2014; posted September 29, 2014 (Doc. ID 220751); published October 22, 2014 We demonstrate a split-contact corrugated ridge waveguide InGaAsP distributed feedback laser at 1560 nm. The laser cavity has been defined with uniform third-order gratings etched along the sidewalls of the ridge waveguide. The gratings were fabricated using a standard I-line stepper lithography technique along with an inductively coupled reactive ion-etching process. Stable single-mode operation has been achieved with side-mode suppression ratios ≥50 dB, output powers ≥7 mW, a wavelength tuning range ≥2.3 nm, and narrow linewidths (≤140 kHz) for different biasing conditions, with a minimum of 70 kHz. The effect of p-contact partition on device performance is also studied. © 2014 Optical Society of America OCIS codes: (140.3490) Lasers, distributed-feedback; (250.5960) Semiconductor lasers. http://dx.doi.org/10.1364/OL.39.006197
Semiconductor split-contact (or multielectrode) distributed feedback (DFB) lasers have received considerable attention since their inception by Yoshikuni and Motosugi [1]. Indeed, it has been demonstrated that partitioning the top electrode into multiple electrically separated sections would result in pure frequency modulation and chirping suppressed amplitude modulation, in wide tuning range, and in narrow linewidths [1–5]. Additionally, it is strongly believed that the injection current control through the different electrodes has a great impact on reducing the adverse spatial hole burning (SHB) to which single-contact DFB lasers are prone [6]. Although this solution would add significant enhancements to DFB lasers, they are still subject to fabrication limitation particularly when it comes to the gratings definition. In fact, it is well known that in standard semiconductor DFB lasers the gratings are being buried through epitaxial regrowth. This has been found to be challenging especially when it comes to the interface between the gratings and upper cladding layers [7,8]. To overcome the problem of regrowth, the gratings can rather be defined on the ridge sidewalls, as exemplified in Fig. 1, once all the epitaxial layers have been grown. This describes what is known as laterally coupled or corrugated ridge-waveguide (CRW) DFB lasers. It is hence obvious that this solution does not demand an overgrowth process, ultimately leading to a simplified fabrication process [7,8]. In addition, the use of low-cost fabrication techniques, such as stepper optical lithography, opens the doors for high-yield and low-cost manufacturing of CRW-DFB lasers making them ideal for commercial mass production. Moreover, CRW-DFB lasers offer an outstanding monolithic integrate-ability, which is a target key for affordable photonic integrated circuits, for example. We recently demonstrated a symmetrical split-contact CRW-DFB laser at 1560 nm [9]. In this Letter, we zero in on the optical characterization of CRW-DFB lasers with three electrodes. Stable single-mode operation with wavelength tuning range ≥2.3 nm and narrow linewidths 0146-9592/14/216197-04$15.00/0
≤140 kHz have been recorded. The effect of the p-contact splitting on the device performance has also been studied by considering two different 1500-μm-long devices as exemplified by Fig. 2. The first device, L1 , has three equal 494-μm-long electrodes separated by 9-μmwide grooves. The second device, L2 , has a 740-μm-long center section and two equal-side electrodes 370 μm each, with 10-μm-wide separation grooves. The measured isolation resistances between adjacent electrodes are 1630 Ω and 1800 Ω for devices L1 and L2 , respectively. The devices have been fabricated with a ridge waveguide that has the same dimensions as in [9]. The fabricated third-order gratings showed a normalized coupling coefficient of 1.5. The tested devices were as-cleaved and have been mounted on an accurate temperaturecontrolled stage. Figure 3 shows the light/current characteristic of both devices, L1 and L2 , under uniform current injection at 25° C. The continuous wave (CW) threshold currents are 68 and 60 mA for L1 and L2 , respectively. The maximum slope efficiencies η are 0.03 and 0.036 for L1 and L2 , respectively. Both devices can emit output powers ≥7 mW. These values are comparable to the ones obtained with one-electrode devices, and reasons behind the low efficiency and power values have been discussed in [8]. Nevertheless, we have noticed that the threshold values are lower than the ones obtained either for one-electrode
Fig. 1. 3D cutaway structure for the split-contact corrugated ridge waveguide with three electrodes. © 2014 Optical Society of America
6198
OPTICS LETTERS / Vol. 39, No. 21 / November 1, 2014
Fig. 4. Variations of the output power: (a) in the nonuniform injection case (2), and (b) as a function of I 2 ∕I T at 25°C (I T 280 mA). Fig. 2. Side view of the three-electrode CRW-DFB lasers (L1 and L2 ) used in this work. Injection current schemes and cases are also given.
Fig. 5. SMSR variations: (a) in the uniform injection case; (b) and (c) in the nonuniform injection cases (1) and (2), respectively; (d) as a function of temperature. Fig. 3. Light-power characteristics for L1 and L2 at 25°C under uniform injection. The inset shows the modal gain against the detuning (Δλ) from the lasing wavelength (1560 nm).
[8] or two-electrode [9] CRW-DFB lasers. This is a significant sign that the SHB effect has been reduced by considering the three-electrode configuration, as has been demonstrated [10]. Figure 4(a) shows the variations of the output power at 25°C in the nonuniform injection case (2), whereas Fig. 4(b) shows the variations of the optical output power against the ratio I 2 ∕I T while I T 280 mA at 25°C. Moreover, stable single-mode operation and high sidemode suppression ratios (SMSRs) for a wide range of injection currents and at different temperatures have been obtained. Indeed, the measured SMSRs exceed 50 dB under different biasing schemes: in the uniform injection case as shown in Fig. 5(a), and the nonuniform injection cases (1) and (2) as shown in Figs. 5(b) and 5(c). In these three cases, device L2 shows better results in terms of side-mode discrimination in comparison to device L1 . In terms of temperature dependence, device L2 shows better SMSR performance than L1 in either injection schemes, as can be seen in Fig. 5(d). In Fig. 5(d), the total current I T was kept at 280 mA: In the uniform case,
device L1 has been driven by considering I 1 I 2 I 3 I T ∕3, whereas device L2 has been driven at I 1 I 3 70 mA and I 2 140 mA; in the nonuniform case, a biasing at I 1 I 3 60 mA and I 2 160 mA has been applied for both L1 and L2 . Stable single-mode operation with SMSR > 40 dB has been observed for L2 at temperature ≥80°C. Additionally, continuous wavelength tunability without mode hopping has been obtained depending on the p-contact partition and current injection scheme. In fact, under uniform injection and at 25°C, both devices, L1 and L2 , show a tuning range of ∼1.7 nm as can be seen in Fig. 6(a). When considering the nonuniform case (1) the tuning range has increased to ∼1.8 and 2.3 nm at 25°C for L1 and L2 respectively as shown in Fig. 6(b). In the nonuniform injection case (2) a tuning range of ∼1.45 nm has been measured for both L1 and L2 at 25°C as shown in Fig. 6(c). This shows that by tailoring the biasing injected current through the three electrodes wide frequency tunability may be obtained. As for the temperature dependency, it is clear as in Fig. 6(d), that L2 can still be tuned at high temperatures when compared to L1 while both devices were being
November 1, 2014 / Vol. 39, No. 21 / OPTICS LETTERS
Fig. 6. Lasing emission wavelengths variations: (a) in the uniform injection case; (b) and (c) in the nonuniform injection cases (1) and (2), respectively; (d) as a function of temperature.
biased at 60∕160∕60 mA. The increase in temperature showed a red-shift lasing wavelength at a rate dλ∕dT of 0.0927 and 0.0946 nm∕°C for L1 and L2 , respectively. The optical spectral linewidth has been carried out using the common delayed self-heterodyne interferometric (DSHI) technique [11]. The experimental setup as well as the procedure of the extraction of the intrinsic linewidth has been described in [8,9]. Figure 7(a) shows an example of a beat note for laser L2 biased at 180 mA in the nonuniform case (2), defined in Fig. 2. The fitted Voigt profile gives a full-width at half-maximum (FWHM) of 0.969 MHz, from which the extracted Lorentzian and
Fig. 7. (a) Example of a delayed self-heterodyne RF beat note with the Voigt fitting for the laser L2 biased at 180 mA [nonuniform injection case (2)] at 25°C; linewidth variations (b) in the uniform case; (c) and (d) in the nonuniform cases (1) and (2), respectively (at 25°C).
6199
Gaussian FWHMs are 0.522 and 0.645 MHz, respectively. Keeping in mind that the DSHI technique assumes that the intrinsic linewidth is the half of the Lorentzian part, the final linewidth is 0.261 MHz. Following this procedure [8,9] we have studied the dependence of the linewidth on various injection conditions. Figure 7(b) shows the intrinsic linewidth (δν) dependence on the total current I T for L1 and L2 while being uniformly biased. Both devices showed almost the same linewidths with lower values for L2 at total injection exceeding 260 mA. The minimum recorded δυ is 0.1805 and 0.162 MHz for L1 at 260 mA and L2 at 280 mA, respectively. When considering the nonuniform injection, L2 shows lower linewidths than L1 . Indeed, Fig. 7(c) shows the variations of δυ against I T in the nonuniform injection case (1): In this case the minimum δν is 0.1 MHz for L1 and 0.070 MHz for L2 both at 290 mA. Figure 7(d) shows the δν variations in the nonuniform injection case (2): In this case, L2 shows lower δν than L1 for all the I 2 injection range, with a recorded minimum of 0.250 MHz and 0.090 MHz for L1 at 60∕70∕60 mA and L2 at 60∕110∕60 mA, respectively. These results demonstrate how useful the consideration of CRW-DFB lasers with multielectrode could be in terms of narrow linewidths, as was demonstrated earlier with standard DFB lasers [3–5]. Although the output power for these devices was relatively low for the observed linewidth, the large detuning between the lasing wavelength and the gain peak wavelength (see inset of [Fig. 3]) could be an important factor behind the narrow linewidth measured in this study as has been found in [12]. Moreover, the low measured linewidth enhancement factor (α ≤ 1.8) around the lasing wavelength (1560 nm) compensates for the low output powers. This low α-factor is in reasonable agreement with the observed linewidth, given the large detuning of ∼ − 30 nm (i.e., the lasing wavelength is at the shorter side of the gain peak) [12,13]. Moreover, in comparison to the results obtained for two-electrode CRW-DFB lasers [9], we have noticed that the devices in this study show: (i) more stable linewidth variations (i.e., in other words, no linewidth jumps have been observed under different current injection conditions); (ii) in general narrower linewidth values depending on the biasing conditions and the electrode partition. These low values could also be attributed to the fact that the current injection through the three electrodes creates some sort of symmetric distribution for the optical-field profile inside the laser cavity. This distribution would effectively reduce the SHB effect, which in turn gives rise to reduced spectral linewidths [6,10]. Equally important, we have to mention that the observed low linewidths were possible thanks to the use of the ultralow noise source LDX-3620, which have shown small drive-current noise impact on the measured linewidth [14]. We have demonstrated InGaAsP/InP split-contact corrugated ridge waveguide distributed feedback lasers at 1560 nm. Third-order gratings have been defined and etched uniformly along the cavity using the I-line optical stepper lithography plus a reactive ion etching. The experimental characterization revealed that stable single-mode operation has been achieved with side-mode suppression ratios ≥50 dB, output powers ≥7 mW, and a
6200
OPTICS LETTERS / Vol. 39, No. 21 / November 1, 2014
tuning coefficient of 0.095 nm∕°C, depending on the electrode partition. Moreover, under different biasing conditions, narrow linewidths (≤140 kHz) with a minimum of 70 kHz have been recorded. In addition to these features, these sources offer easy monolithic integrability with other devices, which make them distinctive sources for the recent advanced coherent optical communication systems. The authors would like to thank Joe Seregelyi for his helpful discussions and for providing the isolator and the ultralow noise source, and Michel Poulin of Teraxion for his fruitful discussions and suggestions. The authors gratefully acknowledge the Ontario Graduate Scholarship program. They are also thankful to the Natural Sciences and Engineering Research Council (NSERC) for its support of this research and CMC Microsystems for its support of the fabrication process. Dr. Trevor Hall is grateful to the Canada Research Chair (CRC) program for their support of his CRC-I in Photonic Network Technology. References 1. Y. Yoshikuni and G. Motosugi, in Optical Fiber Communication, OSA Technical Digest Series (Optical Society of America, 1986), paper TuF1.
2. Y. Yoshikuni and G. Motosugi, J. Lightwave Technol. 5, 516 (1987). 3. Y. Kotaki, S. Ogita, M. Matsude, Y. Kuwahara, and H. Ishikawa, Electron. Lett. 25, 990 (1989). 4. M. Fukuda, K. Sato, Y. Kondo, and M. Nakao, J. Lightwave Technol. 7, 1504 (1989). 5. D. V. Eddolls, S. J. Vass, R. M. Ash, and C. A. Park, Electron. Lett. 28, 1057 (1992). 6. X. Pan, H. Olesen, and B. Tromborg, IEEE Photon. Technol. Lett. 2, 312 (1990). 7. L. M. Miller, J. T. Verdeyen, J. J. Coleman, R. P. Bryan, J. J. Alwan, K. J. Beernink, J. S. Hughes, and T. M. Cockerill, IEEE Photon. Technol. Lett. 3, 6 (1991). 8. K. Dridi, A. Benhsaien, J. Zhang, and T. J. Hall, IEEE Photon. Technol. Lett. 26, 1192 (2014). 9. K. Dridi, A. Benhsaien, J. Zhang, K. Hinzer, and T. J. Hall, Opt. Express 22, 19087 (2014). 10. M. Usami and S. Akiba, IEEE J. Quantum Electron. 25, 1245 (1989). 11. T. Okoshi, K. Kikuchi, and A. Nakayama, Electron. Lett. 16, 630 (1980). 12. S. Ogita, M. Yano, H. Ishikawa, and H. Imai, Electron. Lett. 23, 393 (1987). 13. K. Y. Liou, N. K. Dutta, and C. A. Burrus, Appl. Phys. Lett. 50, 489 (1987). 14. G. H. Duan and P. Gallion, IEEE Photon. Technol. Lett. 3, 302 (1991).
6197
Narrow linewidth 1560 nm InGaAsP split-contact corrugated ridge waveguide DFB lasers Kais Dridi,1,* Abdessamad Benhsaien,1 Jessica Zhang,2 and Trevor J. Hall1 1
Photonics Technology Laboratory, Centre for Research in Photonics at the University of Ottawa, 800 King Edward Avenue, Ottawa, Ontario K1N 6N5, Canada 2
CMC Microsystems, 1200 Montreal Road, M50-IPF, Ottawa, Ontario K1A 0R6, Canada *Corresponding author: [email protected]
Received August 11, 2014; revised September 28, 2014; accepted September 28, 2014; posted September 29, 2014 (Doc. ID 220751); published October 22, 2014 We demonstrate a split-contact corrugated ridge waveguide InGaAsP distributed feedback laser at 1560 nm. The laser cavity has been defined with uniform third-order gratings etched along the sidewalls of the ridge waveguide. The gratings were fabricated using a standard I-line stepper lithography technique along with an inductively coupled reactive ion-etching process. Stable single-mode operation has been achieved with side-mode suppression ratios ≥50 dB, output powers ≥7 mW, a wavelength tuning range ≥2.3 nm, and narrow linewidths (≤140 kHz) for different biasing conditions, with a minimum of 70 kHz. The effect of p-contact partition on device performance is also studied. © 2014 Optical Society of America OCIS codes: (140.3490) Lasers, distributed-feedback; (250.5960) Semiconductor lasers. http://dx.doi.org/10.1364/OL.39.006197
Semiconductor split-contact (or multielectrode) distributed feedback (DFB) lasers have received considerable attention since their inception by Yoshikuni and Motosugi [1]. Indeed, it has been demonstrated that partitioning the top electrode into multiple electrically separated sections would result in pure frequency modulation and chirping suppressed amplitude modulation, in wide tuning range, and in narrow linewidths [1–5]. Additionally, it is strongly believed that the injection current control through the different electrodes has a great impact on reducing the adverse spatial hole burning (SHB) to which single-contact DFB lasers are prone [6]. Although this solution would add significant enhancements to DFB lasers, they are still subject to fabrication limitation particularly when it comes to the gratings definition. In fact, it is well known that in standard semiconductor DFB lasers the gratings are being buried through epitaxial regrowth. This has been found to be challenging especially when it comes to the interface between the gratings and upper cladding layers [7,8]. To overcome the problem of regrowth, the gratings can rather be defined on the ridge sidewalls, as exemplified in Fig. 1, once all the epitaxial layers have been grown. This describes what is known as laterally coupled or corrugated ridge-waveguide (CRW) DFB lasers. It is hence obvious that this solution does not demand an overgrowth process, ultimately leading to a simplified fabrication process [7,8]. In addition, the use of low-cost fabrication techniques, such as stepper optical lithography, opens the doors for high-yield and low-cost manufacturing of CRW-DFB lasers making them ideal for commercial mass production. Moreover, CRW-DFB lasers offer an outstanding monolithic integrate-ability, which is a target key for affordable photonic integrated circuits, for example. We recently demonstrated a symmetrical split-contact CRW-DFB laser at 1560 nm [9]. In this Letter, we zero in on the optical characterization of CRW-DFB lasers with three electrodes. Stable single-mode operation with wavelength tuning range ≥2.3 nm and narrow linewidths 0146-9592/14/216197-04$15.00/0
≤140 kHz have been recorded. The effect of the p-contact splitting on the device performance has also been studied by considering two different 1500-μm-long devices as exemplified by Fig. 2. The first device, L1 , has three equal 494-μm-long electrodes separated by 9-μmwide grooves. The second device, L2 , has a 740-μm-long center section and two equal-side electrodes 370 μm each, with 10-μm-wide separation grooves. The measured isolation resistances between adjacent electrodes are 1630 Ω and 1800 Ω for devices L1 and L2 , respectively. The devices have been fabricated with a ridge waveguide that has the same dimensions as in [9]. The fabricated third-order gratings showed a normalized coupling coefficient of 1.5. The tested devices were as-cleaved and have been mounted on an accurate temperaturecontrolled stage. Figure 3 shows the light/current characteristic of both devices, L1 and L2 , under uniform current injection at 25° C. The continuous wave (CW) threshold currents are 68 and 60 mA for L1 and L2 , respectively. The maximum slope efficiencies η are 0.03 and 0.036 for L1 and L2 , respectively. Both devices can emit output powers ≥7 mW. These values are comparable to the ones obtained with one-electrode devices, and reasons behind the low efficiency and power values have been discussed in [8]. Nevertheless, we have noticed that the threshold values are lower than the ones obtained either for one-electrode
Fig. 1. 3D cutaway structure for the split-contact corrugated ridge waveguide with three electrodes. © 2014 Optical Society of America
6198
OPTICS LETTERS / Vol. 39, No. 21 / November 1, 2014
Fig. 4. Variations of the output power: (a) in the nonuniform injection case (2), and (b) as a function of I 2 ∕I T at 25°C (I T 280 mA). Fig. 2. Side view of the three-electrode CRW-DFB lasers (L1 and L2 ) used in this work. Injection current schemes and cases are also given.
Fig. 5. SMSR variations: (a) in the uniform injection case; (b) and (c) in the nonuniform injection cases (1) and (2), respectively; (d) as a function of temperature. Fig. 3. Light-power characteristics for L1 and L2 at 25°C under uniform injection. The inset shows the modal gain against the detuning (Δλ) from the lasing wavelength (1560 nm).
[8] or two-electrode [9] CRW-DFB lasers. This is a significant sign that the SHB effect has been reduced by considering the three-electrode configuration, as has been demonstrated [10]. Figure 4(a) shows the variations of the output power at 25°C in the nonuniform injection case (2), whereas Fig. 4(b) shows the variations of the optical output power against the ratio I 2 ∕I T while I T 280 mA at 25°C. Moreover, stable single-mode operation and high sidemode suppression ratios (SMSRs) for a wide range of injection currents and at different temperatures have been obtained. Indeed, the measured SMSRs exceed 50 dB under different biasing schemes: in the uniform injection case as shown in Fig. 5(a), and the nonuniform injection cases (1) and (2) as shown in Figs. 5(b) and 5(c). In these three cases, device L2 shows better results in terms of side-mode discrimination in comparison to device L1 . In terms of temperature dependence, device L2 shows better SMSR performance than L1 in either injection schemes, as can be seen in Fig. 5(d). In Fig. 5(d), the total current I T was kept at 280 mA: In the uniform case,
device L1 has been driven by considering I 1 I 2 I 3 I T ∕3, whereas device L2 has been driven at I 1 I 3 70 mA and I 2 140 mA; in the nonuniform case, a biasing at I 1 I 3 60 mA and I 2 160 mA has been applied for both L1 and L2 . Stable single-mode operation with SMSR > 40 dB has been observed for L2 at temperature ≥80°C. Additionally, continuous wavelength tunability without mode hopping has been obtained depending on the p-contact partition and current injection scheme. In fact, under uniform injection and at 25°C, both devices, L1 and L2 , show a tuning range of ∼1.7 nm as can be seen in Fig. 6(a). When considering the nonuniform case (1) the tuning range has increased to ∼1.8 and 2.3 nm at 25°C for L1 and L2 respectively as shown in Fig. 6(b). In the nonuniform injection case (2) a tuning range of ∼1.45 nm has been measured for both L1 and L2 at 25°C as shown in Fig. 6(c). This shows that by tailoring the biasing injected current through the three electrodes wide frequency tunability may be obtained. As for the temperature dependency, it is clear as in Fig. 6(d), that L2 can still be tuned at high temperatures when compared to L1 while both devices were being
November 1, 2014 / Vol. 39, No. 21 / OPTICS LETTERS
Fig. 6. Lasing emission wavelengths variations: (a) in the uniform injection case; (b) and (c) in the nonuniform injection cases (1) and (2), respectively; (d) as a function of temperature.
biased at 60∕160∕60 mA. The increase in temperature showed a red-shift lasing wavelength at a rate dλ∕dT of 0.0927 and 0.0946 nm∕°C for L1 and L2 , respectively. The optical spectral linewidth has been carried out using the common delayed self-heterodyne interferometric (DSHI) technique [11]. The experimental setup as well as the procedure of the extraction of the intrinsic linewidth has been described in [8,9]. Figure 7(a) shows an example of a beat note for laser L2 biased at 180 mA in the nonuniform case (2), defined in Fig. 2. The fitted Voigt profile gives a full-width at half-maximum (FWHM) of 0.969 MHz, from which the extracted Lorentzian and
Fig. 7. (a) Example of a delayed self-heterodyne RF beat note with the Voigt fitting for the laser L2 biased at 180 mA [nonuniform injection case (2)] at 25°C; linewidth variations (b) in the uniform case; (c) and (d) in the nonuniform cases (1) and (2), respectively (at 25°C).
6199
Gaussian FWHMs are 0.522 and 0.645 MHz, respectively. Keeping in mind that the DSHI technique assumes that the intrinsic linewidth is the half of the Lorentzian part, the final linewidth is 0.261 MHz. Following this procedure [8,9] we have studied the dependence of the linewidth on various injection conditions. Figure 7(b) shows the intrinsic linewidth (δν) dependence on the total current I T for L1 and L2 while being uniformly biased. Both devices showed almost the same linewidths with lower values for L2 at total injection exceeding 260 mA. The minimum recorded δυ is 0.1805 and 0.162 MHz for L1 at 260 mA and L2 at 280 mA, respectively. When considering the nonuniform injection, L2 shows lower linewidths than L1 . Indeed, Fig. 7(c) shows the variations of δυ against I T in the nonuniform injection case (1): In this case the minimum δν is 0.1 MHz for L1 and 0.070 MHz for L2 both at 290 mA. Figure 7(d) shows the δν variations in the nonuniform injection case (2): In this case, L2 shows lower δν than L1 for all the I 2 injection range, with a recorded minimum of 0.250 MHz and 0.090 MHz for L1 at 60∕70∕60 mA and L2 at 60∕110∕60 mA, respectively. These results demonstrate how useful the consideration of CRW-DFB lasers with multielectrode could be in terms of narrow linewidths, as was demonstrated earlier with standard DFB lasers [3–5]. Although the output power for these devices was relatively low for the observed linewidth, the large detuning between the lasing wavelength and the gain peak wavelength (see inset of [Fig. 3]) could be an important factor behind the narrow linewidth measured in this study as has been found in [12]. Moreover, the low measured linewidth enhancement factor (α ≤ 1.8) around the lasing wavelength (1560 nm) compensates for the low output powers. This low α-factor is in reasonable agreement with the observed linewidth, given the large detuning of ∼ − 30 nm (i.e., the lasing wavelength is at the shorter side of the gain peak) [12,13]. Moreover, in comparison to the results obtained for two-electrode CRW-DFB lasers [9], we have noticed that the devices in this study show: (i) more stable linewidth variations (i.e., in other words, no linewidth jumps have been observed under different current injection conditions); (ii) in general narrower linewidth values depending on the biasing conditions and the electrode partition. These low values could also be attributed to the fact that the current injection through the three electrodes creates some sort of symmetric distribution for the optical-field profile inside the laser cavity. This distribution would effectively reduce the SHB effect, which in turn gives rise to reduced spectral linewidths [6,10]. Equally important, we have to mention that the observed low linewidths were possible thanks to the use of the ultralow noise source LDX-3620, which have shown small drive-current noise impact on the measured linewidth [14]. We have demonstrated InGaAsP/InP split-contact corrugated ridge waveguide distributed feedback lasers at 1560 nm. Third-order gratings have been defined and etched uniformly along the cavity using the I-line optical stepper lithography plus a reactive ion etching. The experimental characterization revealed that stable single-mode operation has been achieved with side-mode suppression ratios ≥50 dB, output powers ≥7 mW, and a
6200
OPTICS LETTERS / Vol. 39, No. 21 / November 1, 2014
tuning coefficient of 0.095 nm∕°C, depending on the electrode partition. Moreover, under different biasing conditions, narrow linewidths (≤140 kHz) with a minimum of 70 kHz have been recorded. In addition to these features, these sources offer easy monolithic integrability with other devices, which make them distinctive sources for the recent advanced coherent optical communication systems. The authors would like to thank Joe Seregelyi for his helpful discussions and for providing the isolator and the ultralow noise source, and Michel Poulin of Teraxion for his fruitful discussions and suggestions. The authors gratefully acknowledge the Ontario Graduate Scholarship program. They are also thankful to the Natural Sciences and Engineering Research Council (NSERC) for its support of this research and CMC Microsystems for its support of the fabrication process. Dr. Trevor Hall is grateful to the Canada Research Chair (CRC) program for their support of his CRC-I in Photonic Network Technology. References 1. Y. Yoshikuni and G. Motosugi, in Optical Fiber Communication, OSA Technical Digest Series (Optical Society of America, 1986), paper TuF1.
2. Y. Yoshikuni and G. Motosugi, J. Lightwave Technol. 5, 516 (1987). 3. Y. Kotaki, S. Ogita, M. Matsude, Y. Kuwahara, and H. Ishikawa, Electron. Lett. 25, 990 (1989). 4. M. Fukuda, K. Sato, Y. Kondo, and M. Nakao, J. Lightwave Technol. 7, 1504 (1989). 5. D. V. Eddolls, S. J. Vass, R. M. Ash, and C. A. Park, Electron. Lett. 28, 1057 (1992). 6. X. Pan, H. Olesen, and B. Tromborg, IEEE Photon. Technol. Lett. 2, 312 (1990). 7. L. M. Miller, J. T. Verdeyen, J. J. Coleman, R. P. Bryan, J. J. Alwan, K. J. Beernink, J. S. Hughes, and T. M. Cockerill, IEEE Photon. Technol. Lett. 3, 6 (1991). 8. K. Dridi, A. Benhsaien, J. Zhang, and T. J. Hall, IEEE Photon. Technol. Lett. 26, 1192 (2014). 9. K. Dridi, A. Benhsaien, J. Zhang, K. Hinzer, and T. J. Hall, Opt. Express 22, 19087 (2014). 10. M. Usami and S. Akiba, IEEE J. Quantum Electron. 25, 1245 (1989). 11. T. Okoshi, K. Kikuchi, and A. Nakayama, Electron. Lett. 16, 630 (1980). 12. S. Ogita, M. Yano, H. Ishikawa, and H. Imai, Electron. Lett. 23, 393 (1987). 13. K. Y. Liou, N. K. Dutta, and C. A. Burrus, Appl. Phys. Lett. 50, 489 (1987). 14. G. H. Duan and P. Gallion, IEEE Photon. Technol. Lett. 3, 302 (1991).
