January 1, 2015 / Vol. 40, No. 1 / OPTICS LETTERS
69
Ground-state lasing in high-power InAs/GaAs quantum dots-in-a-well laser using active multimode interference structure Yuanbing Cheng,1,* Jian Wu,2 Lingjuan Zhao,3 Xianshu Luo,4 and Qi Jie Wang1,5 1
Photonics Center of Excellence (OPTIMUS), School of Electrical & Electronic Engineering, Nanyang Technological University, 50 Nanyang Ave, 639798, Singapore 2 State Key Laboratory of Information Photonics and Optical Communications, Beijing University of Posts and Telecommunications, 10 Xitucheng Road, Beijing 100876, China 3 Key Laboratory of Semiconductor Materials, Institute of Semiconductors, Chinese Academy of Sciences, 35A Qinghua East Road, Beijing 100083, China 4 Institute of Microelectronics, A*STAR (Agency of Science and Technology Research), 11 Science Park Road, 117685, Singapore 5 e-mail: [email protected] *Corresponding author: [email protected] Received October 29, 2014; revised November 13, 2014; accepted November 21, 2014; posted November 24, 2014 (Doc. ID 225973); published December 22, 2014 We have designed and demonstrated InAs/GaAs quantum dots-in-a-well laser diodes for short cavities with transverse fundamental mode operation by using an active multimode interferometer (MMI) structure for the first time to the best of our knowledge. Room-temperature continuous-wave ground-state lasing at 1280 nm has been achieved with an output power of 116 mW per facet, which is 2.4 times higher than that of the conventional ridge laser diodes. By using the MMI structures, the excited-state (ES) lasing is effectively suppressed with no ES lasing, even at a high injection current of 400 mA. This device has great potential for high-power single-mode laser emission with low electric power consumption and simple fabrication processes. © 2014 Optical Society of America OCIS codes: (140.5960) Semiconductor lasers; (250.5590) Quantum-well, -wire and -dot devices; (230.7370) Waveguides. http://dx.doi.org/10.1364/OL.40.000069
In recent years, self-organized InAs/GaAs quantum dots (QD) have been of considerable interest because of their atomic-like properties and delta-function-like discrete density of states [1,2]. QD lasers are expected to exhibit superior performances over other kinds of semiconductor lasers, including bulk, quantum well (QW), and quantum dash lasers. InAs/GaAs, QD lasers have demonstrated the lowest threshold currents and highest operating temperatures by using the dots-in-a-well (DWELL) structure and p-doping, respectively [3–7]. The QDs offer the possibility for fabricating 1310 nm low cost GaAsbased laser sources suitable for cost-effective optical fiber interconnects. However, it is very difficult to realize a high-power QD laser with ground-state (GS) lasing. This is because that the QD laser has a low modal gain because of the low QD density in the active region [8,9]. The ground-state modal gain of QD laser is typically 3 cm−1 per QD layer, which is about one order lower than that of QW lasers. Therefore, at high injection current, the excited state (ES) lasing at shorter wavelength can easily occur in QD lasers because of the limited density of states combined with slow intraband relaxation [10]. In addition, ES lasing will eventually surpass GS lasing and become the dominant lasing peak if the injection current is to be further increased. To achieve GS lasing in a QD laser with high output power, one of the methods is to increase the QD density and uniformity during material growth, but the highest modal gain is around 7 cm−1 [11], which is still far from satisfaction. Another method is to reduce the laser cavity loss by high-reflectivity facet coatings and/or long cavities [12,13]. This method has a shortcoming in that the ES lasing is only delayed by 10°C by 0146-9592/15/010069-04$15.00/0
optimal facet coating, and the ES lasing still appears at high temperature [12]. A distributed feedback (DFB) reflector can be used to efficiently suppress the ES emission in QD lasers; however, the output power can’t be improved by this structure [14]. In this Letter, we propose a novel approach to realize high-power QD lasers for short cavities with GS emission. As is well known, MMI structures have the property of wavelength-dependent transmissivity and are widely used in passive waveguides. If a section of an MMI structure is inserted in the QD lasers, a different cavity loss for ES and GS wavelengths can be brought about, depending on the MMI’s dimensions. By careful design for the MMI’s dimensions, much higher cavity loss for ES can be achieved compared to that for GS, so that a high-output power QD laser with GS emission can be realized. Moreover, the active MMI offers simultaneous high power and fundamental transverse mode output, which can be easily coupled into standard single-mode fiber without using an additional output mode trimming device. Active MMI structures have been used in QW-based semiconductor laser diodes, and light emitting diodes to increase the output power and keeping the fundamental transverse mode emission [15–18]. In this work, for the first time to the best of our knowledge, we demonstrate GS emission in a high-power InAs/GaAs DWELL laser using an active MMI structure. The schematic structure of the active-MMI QD laser diode (MMI-LD) is shown in Fig. 1(a). It consists of a 1 × 1 MMI section which is connected to the conventional single-mode waveguides. It was designed to operate at a wavelength of 1270 nm (GS wavelength) with a © 2015 Optical Society of America
70
OPTICS LETTERS / Vol. 40, No. 1 / January 1, 2015
Fig. 1. (a) Schematical structure of the MMI-LD and (b) calculated transmission spectra to illustrate the principle of suppression of ES lasing in the device.
waveguide height H of 1.8 μm, and an input/output waveguide width W of 3 μm. The width of the MMI region W mmi is 10 μm. The length of the MMI region Lmmi requires careful design to suitably image the light from the input port onto the output ports. According to the dispersion equation, the propagation constant of the vth guide mode can be written as βv ≈ 2πneff ∕λ − v 12 πλ0 ∕4neff w2eff ;
(1)
by 2-D approximation [19], where neff and weff are the effective refractive index and effective width of the MMI section, respectively. λ0 represents the free space wavelength of the light. Then, the beat length Lπ of the two lowest-order mode is Lπ π∕β0 − β1 4neff w2eff ∕3λ0 :
(2)
For the proposed structure as shown in Fig. 1(a), the input single-mode waveguide is located at the center of the start wall of the MMI section, and only evensymmetric modes are excited in the MMI section. Lπ increases quadratically with weff and a symmetric single-fold self-image length Lmmi is given by Lmmi
3Lπ : 4
layers with different compositions are calculated from the Drude model. It can be seen that the transmissivity for GS and ES wavelength is 92.0% and 61.8%, respectively, as shown by the dashed line in Fig. 1(b) when Lmmi 300 μm. This means that the MMI structure leads to additional cavity loss at the ES wavelength, while the cavity loss at GS wavelength is nearly unchanged if we choose Lmmi 300 μm. The transmissivity for the ES wavelength can be reduced further to 38.2% for the cascaded MMI-LD, which includes two cascaded 1 × 1 MMI sections connected through a single-mode waveguide. Therefore, it is possible to suppress ES lasing using the active MMI-LD structure. As reported in [20,21], the refractive index of the QD active region nc is changed with injection carriers and temperature. However, this carrier induced refractive index is three times smaller in QD than that in QW. Hence, it is reasonable that the variation of the refractive index is set to less than 0.05 in the simulation. Figure 2(a) shows the calculated transmission curves at GS wavelength under different nc . It can be seen that the transitivity is nearly unchanged at Lmmi 300 μm as nc varies between 3.480 and 3.490. The carrier induced refractive index change is too small to lead to a significant power reduction. This result suggests that the active MMI can be treated as a passive one in the simulation. Relaxed fabrication tolerances, as well as operating conditions, are important for MMI. Figure 2(b) shows the effect of W mmi on the transitivity of the center of the GS wavelength (1270 nm). The offset of W mmi is set to be 0.2 μm in the simulation considering the errors of pattern dimensions during the fabrication. The transitivity of the center of the GS wavelength is changed from 92.0% to 91.8% when W mmi varies between 9. 80 and 10.20 μm, which indicates the large fabrication tolerance of this approach. The QD laser wafers were grown on an Si-doped GaAs (100) substrate by molecular beam epitaxy. InAs dots are embedded in an InGaAs quantum well which forms the DWELL structure. The active layer consists of a sevenlayer InAs DWELL. Each QD layer is composed of 3 ML of InAs grown on 2 nm In0.15 Ga0.85 As and covered with 6 nm In0.15 Ga0.85 As. The QDs are separated by 50 nm GaAs spacer layers. It has been contended that an InGaAs QW around the dots is very important in helping the quantum dots to capture carriers [3]. The atomic force microscope image shows a QD density of 2.65 × 1010 cm−2 and a lateral diameter of about 50 nm. A 75 nm undoped and a 1500 nm Be-doped In0.35 Ga0.65 As
(3)
The operation principle is illustrated by the schematic transmission spectra at both ES (∼1180 nm) and GS zones (∼1270 nm) in Fig. 1(b). The transitivity versus Lmmi in lasing center wavelengths of GS and ES is simulated by a 3D beam propagation method (BPM). The material structure of the LD used in the simulation is shown in Table 1. The quantum dot active laser is simply treated as one single layer. Refractive indexes of the
Fig. 2. Calculated transmission curves at GS wavelength under (a) different nc and (b) different W mmi with the change of the length of MMI section.
January 1, 2015 / Vol. 40, No. 1 / OPTICS LETTERS
71
Table 1. Specifications for the InAs QD Wafers Layer P-GaAs P-Al0.35 Ga0.65 As i- Al0.35 Ga0.65 As InAs QD active layers i-Al0.35 Ga0.65 As n-Al0.35 Ga0.65 As n-GaAs n-GaAs substrate
Doping Concentration 18
Be:2 × 10 Be:1 × 1018 Uid Uid Uid Si:5 × 1017 Si:1 × 1018 S:1 × 1018
Thickness
Refractive Index@1270 nm
Refractive Index@1180 nm
300 nm 1.5 μm 75 nm 457.8 nm 75 nm 1.5 μm 0.3 μm 350 μm
3.413 3.220 3.220 3.485(3.480–3.490) 3.220 3.220 3.413 3.413
3.434 3.232 3.232 3.485(3.480–3.490) 3.232 3.232 3.434 3.434
layer forms the upper waveguide, while a 75 nm undoped and a 1500 nm Si-doped In0.35 Ga0.65 As layer forms the lower waveguide. A 200 nm p-type GaAs is used as the contact layer. The InAs/GaAs QD structures were fabricated into MMI-LD with varied cavity length. The fabrication starts with defining a shallow shallow ridge using the conventional contact lithography. Wet etching was terminated when the upper n-doped GaAs and n-doped AlGaAs layers above the QD active region were removed to a depth of 1.8 μm. A 400 nm thick layer of SiO2 was deposited on the sample surface by plasma enhanced chemical vapor deposition (PECVD), and contact windows were opened on the ridge top. A Ti/Au metal stack was deposited as contact metals for both p- and n-type electrodes. Finally, the wafer was cleaved to form a Fabry–Pérot laser cavity after being thinned to 100 μm. No coating was done on the laser facets. For comparison, conventional single-stripe LDs with a waveguide width of 3 μm were also fabricated simultaneously on the same wafer. The light output power/injection current/voltage (L-I-V) characteristics of the fabricated devices under continuous-wave (CW) operation at room temperature are summarized in Fig. 3. The actual width W mmi and length of the fabricated MMI region Lmmi are 10 μm and 345 μm, respectively. Among the three kinds of devices (conventional ridge LD, MMI-LD, and cascaded MMI-LD) with the same device length of 1 mm, the conventional ridge LD has the lowest threshold current of 22 mA. The cascaded MMI-LD has the lowest threshold current density J th of 486.1 A∕cm2 , which corresponds to 69.4 A∕cm2 for each of the seven QD layers. By using MMI configurations, the threshold current density J th is reduced by 33.7%. Moreover, a significant maximum output power increase of more than 40 mW and 67 mW for the MMI and cascaded MMI configurations have been
Fig. 3. Light output power versus current and voltage versus current for different kinds of InAs/GaAs QD laser diodes.
achieved, respectively, compared with conventional ridge LDs. This indicates that the output power for MMI-LD and cascaded MMI-LD at the injection current of 420 mA is 1.8 and 2.4 times higher than that of conventional ridge LDs, respectively. The maximum output power of 116 mW per facet for the cascaded MMI-LD has been achieved at room temperature. This result is comparable to the reported output power of 136.3 mW per facet in a GS lasing QD laser with a long cavity length (2 mm), considering more layers of QDs (10 versus 7) and lower working temperature (10°C versus RT) used in [22]. The slope efficiency of the L-I curve for the MMILD is a bit larger than that of the nonMMI reference. This is believed to result from the larger extra sidewall loss in the single-mode waveguide than that in the MMI section. The difference in the output power among these three kinds of devices could be associated with the different active areas. The large active area contributes to the reduction of series resistance and thermal resistance for MMI-LDs [16]. Low thermal resistance results in a high power for LD working at room temperature. The voltage/current (V–I) characteristics of the QD lasers are also shown in Fig. 3. It can be seen that the operating voltages at threshold are 1.40 V with a series resistance of 2.85 Ω, 1.55 V with a series of 5.22 Ω for the cascaded MMI-LD and MMI-LD, respectively, compared with 1.68 V and 8.94 Ω for the reference conventional ridge LD. The relationship of the driving voltage and series resistance R of the LD can be written as follows: V V th IR R ≈ R1 R2 R1
ρd ρd 1 ≈ ∝ ; A A A
(4) (5)
where V th is the turn-on voltage of the laser diode; I is the injection current; R1 and R2 denote the contact and bulk resistance, respectively; A is the active area; ρ is the bulk resistivity; and d is the bulk thickness. Therefore, the larger active area contributes to smaller series resistance R which leads to a lower driving voltage. The series resistances of the devices can be further reduced if Ni/Ge/Au alloy are adopted instead of Ti/Au to form n-contact of the devices. The wall-plug power consumption (P eletric UI) in active MMI-LDs is calculated. As a sequence of reduction in driving voltage, a significant 47.1% reduction in electric power consumption to 0.593 W at output power of 55 mW (at 300 mA) was achieved for MMI-LD, while 68.3% reduction in electric power consumption to 0.355 W was achieved for cascaded MMI-LD.
72
OPTICS LETTERS / Vol. 40, No. 1 / January 1, 2015
The authors would like to express their sincere appreciation to Dr. Liu Zhihong for useful discussions and comments. This work is supported by A*STAR SERC Future Data Center Technologies, Thematic Strategic Research Programme under Grant No. 112 280 4038, and A*STAR -MINDEF Science and Technology Joint Funding Programme under Grant No. 122 331 0076.
Fig. 4. Emission spectra of (a) the conventional ridge InAs/GaAs QD LD, and (b) MMI-LD and cascaded MMI-LD for different drive currents above the threshold.
Figure 4(a) shows a series of room-temperature lasing spectra corresponding to different working conditions of the conventional LDs measured by Fourier transform infrared spectroscopy (FTIR). GS lasing emission with a peak wavelength of 1280 nm can be observed at an injection current of 60 mA. When the injection current increases to 150 mA, the first ES lasing spectrum appears and the two-state lasing coexists simultaneously. With further increase of the injection current to 180 mA, only ES lasing peak remained, while the GS lasing disappeared. However, for the MMI-LD, no transition between GS and ES lasing was observed with an increase of injection current. GS lasing is still observed, even at a high injection current of 400 mA, and no ES lasing appears as shown in Fig. 4(b), giving significant evidence of the suppression of ES lasing by the MMI structure. The cascaded MMI-LD with two MMI sections exhibits higher output power than the MMI-LD with a single MMI section because of the large active area. The result suggests that output power could be further increased with more MMI sections. The QD lasers with active MMI configurations can demonstrate not only GS lasing at high injection current, but also high output power. It is worth mentioning that ES lasing appears more easily with a short cavity length (L < 1.2 mm), less QD stacks, and high injection current [13,23]. The proposed active MMI structure is quite promising for various QD devices, including laser diodes, super-luminescent light emitting diodes, bistable laser diodes, and mode-locked laser diodes. It is one of the most useful and simplest methods to suppress or delay ES lasing without sacrificing the output power of GS. In conclusion, we have demonstrated to the best of our knowledge the first high-power InAs/GaAs QD lasers with transverse fundamental mode operation and low electric power consumption using active MMI structures. Room-temperature CW GS lasing at 1,280 nm has been achieved with an output power of 116 mW per facet and a low threshold current density J th of 486.1.2 A∕cm−1 for a seven-layer QD lasers. By using the MMI structures, the ES lasing is effectively suppressed. GS-ES transition is obviously observed at an injection current of 150 mA for the conventional ridge LD, while no ES lasing appears, even at a high injection current of 400 mA for both MMI-LDs and cascaded MMI-LDs. These results show the great potential of QD lasers with MMI structures for highpower, single-mode emission with low electric power consumption and simple fabrication processes.
References 1. M. Sugawara and M. Usami, Nat. Photonics 3, 30 (2009). 2. P. Michler, A. Kiraz, C. Becher, W. V. Schoenfeld, P. M. Petroff, L. D. Zhang, E. Hu, and A. Imamoglu, Science 290, 2282 (2000). 3. A. Stintz, G. T. Liu, H. Li, L. F. Lester, and K. J. Malloy, IEEE Photon. Technol. Lett. 12, 591 (2000). 4. H. Y. Liu, M. Hopkinson, C. N. Harrison, M. J. Steer, R. Frith, I. R. Sellers, D. J. Mowbray, and M. S. Skolnick, J. Appl. Phys. 93, 2931 (2003). 5. A. Y. Liu, C. Zhang, A. Snyder, D. Lubyshev, J. M. Fastenau, A. W. K. Liu, A. C. Gossard, and J. E. Bowers, J. Vac. Sci. Technol. B 32, 02C108 (2014). 6. C. Sandall, P. M. Smowton, C. L. Walker, T. Badcock, D. J. Mowbray, H. Y. Liu, and M. Hopkinson, Appl. Phys. Lett. 88, 111113 (2006). 7. S. Fathpour, Z. Mi, P. Bhattacharya, A. R. Kovsh, S. S. Mikhrin, I. L. Krestnikov, A. V. Kozhukhov, and N. N. Ledentsov, Appl. Phys. Lett. 85, 5164 (2004). 8. K. Mukai, Y. Nakata, K. Otsubo, M. Sugawara, N. Yokoyama, and H. Ishikawa, IEEE J. Quantum. Electron. 36, 472 (2000). 9. P. W. Fry, L. Harris, S. R. Parnell, J. J. Finley, A. D. Ashmore, D. J. Mowbray, M. S. Skolnick, M. Hopkinson, G. Hill, and J. C. Clark, J. Appl. Phys. 87, 615 (2000). 10. Y. Kaptan, H. Schmeckebier, B. Herzog, D. Arsenijevic, M. Kolarczik, V. Mikhelashvili, N. Owschimikow, G. Eisenstein, D. Bimberg, and U. Woggon, Appl. Phys. Lett. 104, 261108 (2014). 11. K. Nishi, T. Kageyama, M. Yamaguchi, Y. Maeda, K. Takemasa, T. Yamamoto, M. Sugawara, and Y. Arakawa, J. Cryst. Growth 378, 459 (2013). 12. Y. L. Cao, T. Yang, P. F. Xu, H. M. Ji, Y. X. Gu, X. D. Wang, Q. Wang, W. Q. Ma, and L. H. Chen, Appl. Phys. Lett. 96, 171101 (2010). 13. V. V. Korenev, A. V. Savelyev, A. E. Zhukov, A. V. Omelchenko, M. V. Maximov, and Y. M. Shernyakov, Proc. SPIE V, 8432 (2012). 14. M. Stubenrauch, G. Stracke, D. Arsenijevic, A. Strittmatter, and D. Bimberg, Appl. Phys. Lett. 105, 011103 (2014). 15. W. J. Lai, Y. B. Cheng, Y. Chen, D. B. Zhou, J. Bian, L. J. Zhao, and J. Wu, J. Semicond. 32, 054007 (2011). 16. Z. G. Zang, K. Mukai, P. Navaretti, M. Duelk, C. Velez, and K. Hamamoto, Appl. Phys. Lett. 100, 031108 (2012). 17. H. S. Jiang, H. A. Bastawrous, T. Hagio, S. Matsuo, and K. Hamamoto, in 22nd IEEE International Semiconductor Laser Conference (ISLC) (IEEE, 2010), p. 123. 18. T. Fukuda, K. Okamoto, Y. Hinokuma, and K. Hamamoto, IEEE Photon. Technol. Lett. 21, 176 (2009). 19. L. B. Soldano and E. C. M. Pennings, IEEE/OSA J. Lightwave. Technol. 13, 615 (1995). 20. M. T. Crowley, J. Houlihan, T. Piwonski, I. O, 'Driscoll, D. P. Williams, E. P. O, 'Reilly, A. V. Uskov, and G. Huyet, Appl. Phys. Lett. 103, 021114 (2013). 21. N. Y. Gordeev, O. I. Rumyantsev, I. G. Savenko, A. S. Payusov, F. I. Zubov, M. V. Maximov, and A. E. Zhukov, J. Nanophoton. 7, 073087 (2013). 22. Q. Cao, S. F. Yoon, C. Y. Liu, and C. Y. Ngo, Nanoscale Res. Lett. 2, 303 (2007). 23. Z. Y. Zhang, Q. Jiang, and R. A. Hogg, Electron. Lett. 46, 1155 (2010).
69
Ground-state lasing in high-power InAs/GaAs quantum dots-in-a-well laser using active multimode interference structure Yuanbing Cheng,1,* Jian Wu,2 Lingjuan Zhao,3 Xianshu Luo,4 and Qi Jie Wang1,5 1
Photonics Center of Excellence (OPTIMUS), School of Electrical & Electronic Engineering, Nanyang Technological University, 50 Nanyang Ave, 639798, Singapore 2 State Key Laboratory of Information Photonics and Optical Communications, Beijing University of Posts and Telecommunications, 10 Xitucheng Road, Beijing 100876, China 3 Key Laboratory of Semiconductor Materials, Institute of Semiconductors, Chinese Academy of Sciences, 35A Qinghua East Road, Beijing 100083, China 4 Institute of Microelectronics, A*STAR (Agency of Science and Technology Research), 11 Science Park Road, 117685, Singapore 5 e-mail: [email protected] *Corresponding author: [email protected] Received October 29, 2014; revised November 13, 2014; accepted November 21, 2014; posted November 24, 2014 (Doc. ID 225973); published December 22, 2014 We have designed and demonstrated InAs/GaAs quantum dots-in-a-well laser diodes for short cavities with transverse fundamental mode operation by using an active multimode interferometer (MMI) structure for the first time to the best of our knowledge. Room-temperature continuous-wave ground-state lasing at 1280 nm has been achieved with an output power of 116 mW per facet, which is 2.4 times higher than that of the conventional ridge laser diodes. By using the MMI structures, the excited-state (ES) lasing is effectively suppressed with no ES lasing, even at a high injection current of 400 mA. This device has great potential for high-power single-mode laser emission with low electric power consumption and simple fabrication processes. © 2014 Optical Society of America OCIS codes: (140.5960) Semiconductor lasers; (250.5590) Quantum-well, -wire and -dot devices; (230.7370) Waveguides. http://dx.doi.org/10.1364/OL.40.000069
In recent years, self-organized InAs/GaAs quantum dots (QD) have been of considerable interest because of their atomic-like properties and delta-function-like discrete density of states [1,2]. QD lasers are expected to exhibit superior performances over other kinds of semiconductor lasers, including bulk, quantum well (QW), and quantum dash lasers. InAs/GaAs, QD lasers have demonstrated the lowest threshold currents and highest operating temperatures by using the dots-in-a-well (DWELL) structure and p-doping, respectively [3–7]. The QDs offer the possibility for fabricating 1310 nm low cost GaAsbased laser sources suitable for cost-effective optical fiber interconnects. However, it is very difficult to realize a high-power QD laser with ground-state (GS) lasing. This is because that the QD laser has a low modal gain because of the low QD density in the active region [8,9]. The ground-state modal gain of QD laser is typically 3 cm−1 per QD layer, which is about one order lower than that of QW lasers. Therefore, at high injection current, the excited state (ES) lasing at shorter wavelength can easily occur in QD lasers because of the limited density of states combined with slow intraband relaxation [10]. In addition, ES lasing will eventually surpass GS lasing and become the dominant lasing peak if the injection current is to be further increased. To achieve GS lasing in a QD laser with high output power, one of the methods is to increase the QD density and uniformity during material growth, but the highest modal gain is around 7 cm−1 [11], which is still far from satisfaction. Another method is to reduce the laser cavity loss by high-reflectivity facet coatings and/or long cavities [12,13]. This method has a shortcoming in that the ES lasing is only delayed by 10°C by 0146-9592/15/010069-04$15.00/0
optimal facet coating, and the ES lasing still appears at high temperature [12]. A distributed feedback (DFB) reflector can be used to efficiently suppress the ES emission in QD lasers; however, the output power can’t be improved by this structure [14]. In this Letter, we propose a novel approach to realize high-power QD lasers for short cavities with GS emission. As is well known, MMI structures have the property of wavelength-dependent transmissivity and are widely used in passive waveguides. If a section of an MMI structure is inserted in the QD lasers, a different cavity loss for ES and GS wavelengths can be brought about, depending on the MMI’s dimensions. By careful design for the MMI’s dimensions, much higher cavity loss for ES can be achieved compared to that for GS, so that a high-output power QD laser with GS emission can be realized. Moreover, the active MMI offers simultaneous high power and fundamental transverse mode output, which can be easily coupled into standard single-mode fiber without using an additional output mode trimming device. Active MMI structures have been used in QW-based semiconductor laser diodes, and light emitting diodes to increase the output power and keeping the fundamental transverse mode emission [15–18]. In this work, for the first time to the best of our knowledge, we demonstrate GS emission in a high-power InAs/GaAs DWELL laser using an active MMI structure. The schematic structure of the active-MMI QD laser diode (MMI-LD) is shown in Fig. 1(a). It consists of a 1 × 1 MMI section which is connected to the conventional single-mode waveguides. It was designed to operate at a wavelength of 1270 nm (GS wavelength) with a © 2015 Optical Society of America
70
OPTICS LETTERS / Vol. 40, No. 1 / January 1, 2015
Fig. 1. (a) Schematical structure of the MMI-LD and (b) calculated transmission spectra to illustrate the principle of suppression of ES lasing in the device.
waveguide height H of 1.8 μm, and an input/output waveguide width W of 3 μm. The width of the MMI region W mmi is 10 μm. The length of the MMI region Lmmi requires careful design to suitably image the light from the input port onto the output ports. According to the dispersion equation, the propagation constant of the vth guide mode can be written as βv ≈ 2πneff ∕λ − v 12 πλ0 ∕4neff w2eff ;
(1)
by 2-D approximation [19], where neff and weff are the effective refractive index and effective width of the MMI section, respectively. λ0 represents the free space wavelength of the light. Then, the beat length Lπ of the two lowest-order mode is Lπ π∕β0 − β1 4neff w2eff ∕3λ0 :
(2)
For the proposed structure as shown in Fig. 1(a), the input single-mode waveguide is located at the center of the start wall of the MMI section, and only evensymmetric modes are excited in the MMI section. Lπ increases quadratically with weff and a symmetric single-fold self-image length Lmmi is given by Lmmi
3Lπ : 4
layers with different compositions are calculated from the Drude model. It can be seen that the transmissivity for GS and ES wavelength is 92.0% and 61.8%, respectively, as shown by the dashed line in Fig. 1(b) when Lmmi 300 μm. This means that the MMI structure leads to additional cavity loss at the ES wavelength, while the cavity loss at GS wavelength is nearly unchanged if we choose Lmmi 300 μm. The transmissivity for the ES wavelength can be reduced further to 38.2% for the cascaded MMI-LD, which includes two cascaded 1 × 1 MMI sections connected through a single-mode waveguide. Therefore, it is possible to suppress ES lasing using the active MMI-LD structure. As reported in [20,21], the refractive index of the QD active region nc is changed with injection carriers and temperature. However, this carrier induced refractive index is three times smaller in QD than that in QW. Hence, it is reasonable that the variation of the refractive index is set to less than 0.05 in the simulation. Figure 2(a) shows the calculated transmission curves at GS wavelength under different nc . It can be seen that the transitivity is nearly unchanged at Lmmi 300 μm as nc varies between 3.480 and 3.490. The carrier induced refractive index change is too small to lead to a significant power reduction. This result suggests that the active MMI can be treated as a passive one in the simulation. Relaxed fabrication tolerances, as well as operating conditions, are important for MMI. Figure 2(b) shows the effect of W mmi on the transitivity of the center of the GS wavelength (1270 nm). The offset of W mmi is set to be 0.2 μm in the simulation considering the errors of pattern dimensions during the fabrication. The transitivity of the center of the GS wavelength is changed from 92.0% to 91.8% when W mmi varies between 9. 80 and 10.20 μm, which indicates the large fabrication tolerance of this approach. The QD laser wafers were grown on an Si-doped GaAs (100) substrate by molecular beam epitaxy. InAs dots are embedded in an InGaAs quantum well which forms the DWELL structure. The active layer consists of a sevenlayer InAs DWELL. Each QD layer is composed of 3 ML of InAs grown on 2 nm In0.15 Ga0.85 As and covered with 6 nm In0.15 Ga0.85 As. The QDs are separated by 50 nm GaAs spacer layers. It has been contended that an InGaAs QW around the dots is very important in helping the quantum dots to capture carriers [3]. The atomic force microscope image shows a QD density of 2.65 × 1010 cm−2 and a lateral diameter of about 50 nm. A 75 nm undoped and a 1500 nm Be-doped In0.35 Ga0.65 As
(3)
The operation principle is illustrated by the schematic transmission spectra at both ES (∼1180 nm) and GS zones (∼1270 nm) in Fig. 1(b). The transitivity versus Lmmi in lasing center wavelengths of GS and ES is simulated by a 3D beam propagation method (BPM). The material structure of the LD used in the simulation is shown in Table 1. The quantum dot active laser is simply treated as one single layer. Refractive indexes of the
Fig. 2. Calculated transmission curves at GS wavelength under (a) different nc and (b) different W mmi with the change of the length of MMI section.
January 1, 2015 / Vol. 40, No. 1 / OPTICS LETTERS
71
Table 1. Specifications for the InAs QD Wafers Layer P-GaAs P-Al0.35 Ga0.65 As i- Al0.35 Ga0.65 As InAs QD active layers i-Al0.35 Ga0.65 As n-Al0.35 Ga0.65 As n-GaAs n-GaAs substrate
Doping Concentration 18
Be:2 × 10 Be:1 × 1018 Uid Uid Uid Si:5 × 1017 Si:1 × 1018 S:1 × 1018
Thickness
Refractive Index@1270 nm
Refractive Index@1180 nm
300 nm 1.5 μm 75 nm 457.8 nm 75 nm 1.5 μm 0.3 μm 350 μm
3.413 3.220 3.220 3.485(3.480–3.490) 3.220 3.220 3.413 3.413
3.434 3.232 3.232 3.485(3.480–3.490) 3.232 3.232 3.434 3.434
layer forms the upper waveguide, while a 75 nm undoped and a 1500 nm Si-doped In0.35 Ga0.65 As layer forms the lower waveguide. A 200 nm p-type GaAs is used as the contact layer. The InAs/GaAs QD structures were fabricated into MMI-LD with varied cavity length. The fabrication starts with defining a shallow shallow ridge using the conventional contact lithography. Wet etching was terminated when the upper n-doped GaAs and n-doped AlGaAs layers above the QD active region were removed to a depth of 1.8 μm. A 400 nm thick layer of SiO2 was deposited on the sample surface by plasma enhanced chemical vapor deposition (PECVD), and contact windows were opened on the ridge top. A Ti/Au metal stack was deposited as contact metals for both p- and n-type electrodes. Finally, the wafer was cleaved to form a Fabry–Pérot laser cavity after being thinned to 100 μm. No coating was done on the laser facets. For comparison, conventional single-stripe LDs with a waveguide width of 3 μm were also fabricated simultaneously on the same wafer. The light output power/injection current/voltage (L-I-V) characteristics of the fabricated devices under continuous-wave (CW) operation at room temperature are summarized in Fig. 3. The actual width W mmi and length of the fabricated MMI region Lmmi are 10 μm and 345 μm, respectively. Among the three kinds of devices (conventional ridge LD, MMI-LD, and cascaded MMI-LD) with the same device length of 1 mm, the conventional ridge LD has the lowest threshold current of 22 mA. The cascaded MMI-LD has the lowest threshold current density J th of 486.1 A∕cm2 , which corresponds to 69.4 A∕cm2 for each of the seven QD layers. By using MMI configurations, the threshold current density J th is reduced by 33.7%. Moreover, a significant maximum output power increase of more than 40 mW and 67 mW for the MMI and cascaded MMI configurations have been
Fig. 3. Light output power versus current and voltage versus current for different kinds of InAs/GaAs QD laser diodes.
achieved, respectively, compared with conventional ridge LDs. This indicates that the output power for MMI-LD and cascaded MMI-LD at the injection current of 420 mA is 1.8 and 2.4 times higher than that of conventional ridge LDs, respectively. The maximum output power of 116 mW per facet for the cascaded MMI-LD has been achieved at room temperature. This result is comparable to the reported output power of 136.3 mW per facet in a GS lasing QD laser with a long cavity length (2 mm), considering more layers of QDs (10 versus 7) and lower working temperature (10°C versus RT) used in [22]. The slope efficiency of the L-I curve for the MMILD is a bit larger than that of the nonMMI reference. This is believed to result from the larger extra sidewall loss in the single-mode waveguide than that in the MMI section. The difference in the output power among these three kinds of devices could be associated with the different active areas. The large active area contributes to the reduction of series resistance and thermal resistance for MMI-LDs [16]. Low thermal resistance results in a high power for LD working at room temperature. The voltage/current (V–I) characteristics of the QD lasers are also shown in Fig. 3. It can be seen that the operating voltages at threshold are 1.40 V with a series resistance of 2.85 Ω, 1.55 V with a series of 5.22 Ω for the cascaded MMI-LD and MMI-LD, respectively, compared with 1.68 V and 8.94 Ω for the reference conventional ridge LD. The relationship of the driving voltage and series resistance R of the LD can be written as follows: V V th IR R ≈ R1 R2 R1
ρd ρd 1 ≈ ∝ ; A A A
(4) (5)
where V th is the turn-on voltage of the laser diode; I is the injection current; R1 and R2 denote the contact and bulk resistance, respectively; A is the active area; ρ is the bulk resistivity; and d is the bulk thickness. Therefore, the larger active area contributes to smaller series resistance R which leads to a lower driving voltage. The series resistances of the devices can be further reduced if Ni/Ge/Au alloy are adopted instead of Ti/Au to form n-contact of the devices. The wall-plug power consumption (P eletric UI) in active MMI-LDs is calculated. As a sequence of reduction in driving voltage, a significant 47.1% reduction in electric power consumption to 0.593 W at output power of 55 mW (at 300 mA) was achieved for MMI-LD, while 68.3% reduction in electric power consumption to 0.355 W was achieved for cascaded MMI-LD.
72
OPTICS LETTERS / Vol. 40, No. 1 / January 1, 2015
The authors would like to express their sincere appreciation to Dr. Liu Zhihong for useful discussions and comments. This work is supported by A*STAR SERC Future Data Center Technologies, Thematic Strategic Research Programme under Grant No. 112 280 4038, and A*STAR -MINDEF Science and Technology Joint Funding Programme under Grant No. 122 331 0076.
Fig. 4. Emission spectra of (a) the conventional ridge InAs/GaAs QD LD, and (b) MMI-LD and cascaded MMI-LD for different drive currents above the threshold.
Figure 4(a) shows a series of room-temperature lasing spectra corresponding to different working conditions of the conventional LDs measured by Fourier transform infrared spectroscopy (FTIR). GS lasing emission with a peak wavelength of 1280 nm can be observed at an injection current of 60 mA. When the injection current increases to 150 mA, the first ES lasing spectrum appears and the two-state lasing coexists simultaneously. With further increase of the injection current to 180 mA, only ES lasing peak remained, while the GS lasing disappeared. However, for the MMI-LD, no transition between GS and ES lasing was observed with an increase of injection current. GS lasing is still observed, even at a high injection current of 400 mA, and no ES lasing appears as shown in Fig. 4(b), giving significant evidence of the suppression of ES lasing by the MMI structure. The cascaded MMI-LD with two MMI sections exhibits higher output power than the MMI-LD with a single MMI section because of the large active area. The result suggests that output power could be further increased with more MMI sections. The QD lasers with active MMI configurations can demonstrate not only GS lasing at high injection current, but also high output power. It is worth mentioning that ES lasing appears more easily with a short cavity length (L < 1.2 mm), less QD stacks, and high injection current [13,23]. The proposed active MMI structure is quite promising for various QD devices, including laser diodes, super-luminescent light emitting diodes, bistable laser diodes, and mode-locked laser diodes. It is one of the most useful and simplest methods to suppress or delay ES lasing without sacrificing the output power of GS. In conclusion, we have demonstrated to the best of our knowledge the first high-power InAs/GaAs QD lasers with transverse fundamental mode operation and low electric power consumption using active MMI structures. Room-temperature CW GS lasing at 1,280 nm has been achieved with an output power of 116 mW per facet and a low threshold current density J th of 486.1.2 A∕cm−1 for a seven-layer QD lasers. By using the MMI structures, the ES lasing is effectively suppressed. GS-ES transition is obviously observed at an injection current of 150 mA for the conventional ridge LD, while no ES lasing appears, even at a high injection current of 400 mA for both MMI-LDs and cascaded MMI-LDs. These results show the great potential of QD lasers with MMI structures for highpower, single-mode emission with low electric power consumption and simple fabrication processes.
References 1. M. Sugawara and M. Usami, Nat. Photonics 3, 30 (2009). 2. P. Michler, A. Kiraz, C. Becher, W. V. Schoenfeld, P. M. Petroff, L. D. Zhang, E. Hu, and A. Imamoglu, Science 290, 2282 (2000). 3. A. Stintz, G. T. Liu, H. Li, L. F. Lester, and K. J. Malloy, IEEE Photon. Technol. Lett. 12, 591 (2000). 4. H. Y. Liu, M. Hopkinson, C. N. Harrison, M. J. Steer, R. Frith, I. R. Sellers, D. J. Mowbray, and M. S. Skolnick, J. Appl. Phys. 93, 2931 (2003). 5. A. Y. Liu, C. Zhang, A. Snyder, D. Lubyshev, J. M. Fastenau, A. W. K. Liu, A. C. Gossard, and J. E. Bowers, J. Vac. Sci. Technol. B 32, 02C108 (2014). 6. C. Sandall, P. M. Smowton, C. L. Walker, T. Badcock, D. J. Mowbray, H. Y. Liu, and M. Hopkinson, Appl. Phys. Lett. 88, 111113 (2006). 7. S. Fathpour, Z. Mi, P. Bhattacharya, A. R. Kovsh, S. S. Mikhrin, I. L. Krestnikov, A. V. Kozhukhov, and N. N. Ledentsov, Appl. Phys. Lett. 85, 5164 (2004). 8. K. Mukai, Y. Nakata, K. Otsubo, M. Sugawara, N. Yokoyama, and H. Ishikawa, IEEE J. Quantum. Electron. 36, 472 (2000). 9. P. W. Fry, L. Harris, S. R. Parnell, J. J. Finley, A. D. Ashmore, D. J. Mowbray, M. S. Skolnick, M. Hopkinson, G. Hill, and J. C. Clark, J. Appl. Phys. 87, 615 (2000). 10. Y. Kaptan, H. Schmeckebier, B. Herzog, D. Arsenijevic, M. Kolarczik, V. Mikhelashvili, N. Owschimikow, G. Eisenstein, D. Bimberg, and U. Woggon, Appl. Phys. Lett. 104, 261108 (2014). 11. K. Nishi, T. Kageyama, M. Yamaguchi, Y. Maeda, K. Takemasa, T. Yamamoto, M. Sugawara, and Y. Arakawa, J. Cryst. Growth 378, 459 (2013). 12. Y. L. Cao, T. Yang, P. F. Xu, H. M. Ji, Y. X. Gu, X. D. Wang, Q. Wang, W. Q. Ma, and L. H. Chen, Appl. Phys. Lett. 96, 171101 (2010). 13. V. V. Korenev, A. V. Savelyev, A. E. Zhukov, A. V. Omelchenko, M. V. Maximov, and Y. M. Shernyakov, Proc. SPIE V, 8432 (2012). 14. M. Stubenrauch, G. Stracke, D. Arsenijevic, A. Strittmatter, and D. Bimberg, Appl. Phys. Lett. 105, 011103 (2014). 15. W. J. Lai, Y. B. Cheng, Y. Chen, D. B. Zhou, J. Bian, L. J. Zhao, and J. Wu, J. Semicond. 32, 054007 (2011). 16. Z. G. Zang, K. Mukai, P. Navaretti, M. Duelk, C. Velez, and K. Hamamoto, Appl. Phys. Lett. 100, 031108 (2012). 17. H. S. Jiang, H. A. Bastawrous, T. Hagio, S. Matsuo, and K. Hamamoto, in 22nd IEEE International Semiconductor Laser Conference (ISLC) (IEEE, 2010), p. 123. 18. T. Fukuda, K. Okamoto, Y. Hinokuma, and K. Hamamoto, IEEE Photon. Technol. Lett. 21, 176 (2009). 19. L. B. Soldano and E. C. M. Pennings, IEEE/OSA J. Lightwave. Technol. 13, 615 (1995). 20. M. T. Crowley, J. Houlihan, T. Piwonski, I. O, 'Driscoll, D. P. Williams, E. P. O, 'Reilly, A. V. Uskov, and G. Huyet, Appl. Phys. Lett. 103, 021114 (2013). 21. N. Y. Gordeev, O. I. Rumyantsev, I. G. Savenko, A. S. Payusov, F. I. Zubov, M. V. Maximov, and A. E. Zhukov, J. Nanophoton. 7, 073087 (2013). 22. Q. Cao, S. F. Yoon, C. Y. Liu, and C. Y. Ngo, Nanoscale Res. Lett. 2, 303 (2007). 23. Z. Y. Zhang, Q. Jiang, and R. A. Hogg, Electron. Lett. 46, 1155 (2010).
