1. Introduction

The wavelength division multiplexed passive optical network (WDM-PON) is one of the most promising candidates for next-generation access networks because they are capable of handling increasing demands on data bandwidth as well as providing enhanced security and scalability to support several local subscribers [1,2]. In implementing the practical WDM-PON system, the most critical issue among many is how to reduce the cost. Therefore, various WDM-PON features for accessing networks have been proposed [3–9] for optical network units (ONUs) employing such sources as the Fabry–Perot laser diode (FP-LD) [4,5], the reflective SOA (RSOA) [7,8], and the reflective electro-absorption modulator (REAM) [9]. However, there have been only a few studies for cost-effective sources of an optical line terminal (OLT) in the WDM-PON system. In previous reports [10–12], we showed cost-effective OLT sources such as multi-wavelength lasers (MWLs) [10,11] and T-ECLs [12,13]. In particular, the T-ECL was very attractive as a cost-effective OLT source because it can be easily fabricated for mass production. Therefore we reported on the T-ECL with compact C-band 16 channels based on the SLD and the polymeric waveguide with Bragg reflector for a cost-effective WDM-PON system. The operation of the T-ECL was successful in direct modulation for 1.25 and 2.5 Gbit/s over a 20 Km transmission [12,13].

The previous result suggests that the proposed T-ECL should be useful for WDM-PON systems. In addition, if the T-ECL can be used for both ONU and ONT sources in WDM-PON systems, it can reduce installation cost and complexity as well as simplify the system configuration and network management. Besides the C-band T-ECL, there have been strong demands for another source of tunable light with a different lasing wavelength. Among the several suggested light sources, an L-band T-ECL is one of the most promising candidates as source in WDM-PON systems. However, in the L-band wavelength region of around 1.6 μm for which InGaAsP/InP materials are used, the material properties are degraded, mainly due to nonradiative recombination. This result then induces degradations of laser characteristics such as the slope efficiency, saturation power, and characteristic temperature of threshold current. Therefore, the performance level of the L-band SLD, which is used for the light source of the L-band T-ECL, is generally lower than that of the C-band SLD. Therefore, it is necessary to design and fabricate an L-band SLD with a new type of structure so as to show a similar performance as the C-band SLD as given in [14].

In this paper we propose an L-band SLD integrated with an SSC to improve the performance of the L-band T-ECL. The structure of the L-band SLD is in general composed of a double-waveguide core, a planar buried heterostructure (PBH), and an SSC with a BD-RWG. The length of the SSC region is intentionally shortened only to decrease the propagation loss of a wave in the SLD, but the structure of the SSC region is optimized to make a narrow and circular beam and to obtain an FFP of less than 15 degrees. The output power of the L-band T-ECL is higher than that of the C-band T-ECL if the optimized L-band SLD is used as a gain source of the L-band T-ECL.

2. Design and fabrication

For optical access networks, performance of the C-band SLD is discussed in [12]. In the paper, the SLD is utilized as a source of a T-ECL operating at 1.25 Gbit/s over a 20 Km transmission in the C-band. The structure of the SLD, however, should be improved to obtain high output power in the L-band T-ECL, because the gain decrease in the L-band SLD leads to a reduction of the output power when compared with the output power of the C-band SLD. The output power of the L-band SLD is increased simply by reducing the length of the SSC in the SLD by 200 μm in comparison with that in [14] because shortening the length suppresses the propagation loss in the SSC region. The FFP characteristic for the ridge width of the SSC region is shown in Fig. 1 . Though the angle of the FFP decreases along with the increase of the ridge width of SSC, it does not form a circular beam because of both the reduction of the SSC length and the characteristic of the ridge waveguide. The FFP angle is not varied if the ridge width is over 10 μm. On the other hand, to suppress the lasing characteristics of SLD the waveguide is tilted by 7° with respect to the cleaved facet. Therefore, the angle of emission beam is tilted about 24° with respect to the output facet of the SLD. It is very difficult to achieve high coupling efficiency between SLDs with 24° tilted emission angles and a polymer Bragg reflector, even though aspheric microlenses in a transistor outline (TO) packing can are used. The FFP of the SLD is the most important factor in the structure of hybrid integration, such as the T-ECL for high coupling efficiency. The circular beam of the FFP in the SLD is indispensible for increasing the output power of the T-ECL by means of improving the coupling efficiency between the SLD and the polymer waveguide. Therefore, we propose that the structure of the L-band SLD should possess an SSC with a BD-RWG, as shown in Fig. 2 . Figure 2(a) represents a schematic configuration of the SSC SLD, and Fig. 2(b) shows a schematic diagram of the BD-RWG in the SSC region. The width of the passive waveguide core is set to 2 μm to reduce the FFP angle of the vertical direction, while the width of the BD-RWG is designed at 10 μm. The present SLD is fabricated as a BD-RWG type instead of a B-RWG type [16] in order to make the SLD generate a more circular beam at the front facet.

 

Fig. 1 FFP characteristic with the ridge width of SSC region.

Download Full Size | PDF

 

Fig. 2 (a) Schematic configuration of the L-band SSC SLD having novel SSC; (b) schematic configuration of the front facet in SSC region.

Download Full Size | PDF

The SLD is fabricated to possess the structure of a double-waveguide core with an active waveguide in a PBH and a passive waveguide in a BD-RWG shape. The SLD with a cavity length of 600 μm includes four sections: a straight active, a bending active, an SSC, and a passive waveguide with lengths of 400 μm, 60 μm, 110 μm, and 30μm, respectively. The SSC (λ_g = 1.3 μm) is butt-coupled to an active layer (λ_g = 1.60 μm) of multiple quantum wells (MQWs) with a strained separate confinement heterostructure (SCH). According to our numerical simulation, maximum coupling efficiency between the SLD and the polymeric waveguide with a Bragg reflector in the T-ECL is achieved if the FFP angle of the SLD is less than 15°.

In this work, the width of the passive waveguide core is set to be 2 μm, and the width of ridge waveguide is formed to be 10 μm to make the FFP angle less than 15°. In this way, the circular-shaped FFP is achieved. The thickness of the passive waveguide is 150 nm, and the corresponding gain wavelength λ_g is 1.1 μm. For the present study, only one more process is necessary to fabricate the passive waveguide core when compared with the fabrication process of the SLD described in the previous report [14].

For high-speed operation over 1.25 Gbit/s, the parasitic capacity of the SLD should decrease. Therefore, trenches are formed in the lateral sides of the active region by using selective wet etching. The depth and width of the BD-RWG region are defined as 6 and 10 μm, respectively. To reduce reflectivity and to obtain a ripple of less than 3 dB, the passive waveguide is tilted by 7° towards the [110] direction of the cleavage facet, and the front facet is AR-coated with two layers of SiO₂/TiO₂. The reflectivity of the SSC facet is estimated to be less than 10⁻⁴. The rear facet in the active region is high-reflection (HR)-coated with multilayers of SiO₂/TiO₂ with a reflectivity of 98%.

3. Result and Discussion

Figure 3(a) represents the light output power versus the injected current (L-I) of the SSC SLD. The black and red lines indicate the L-I characteristic of the C-band SLD in [14] and the L-band SLD in this work, respectively. As shown in the figure, saturation of the output power is not observed at a high current level. The average output power is 9 mW, which is similar to 9.3 mW of the C-band SLD at the injection current of 60 mA. Figure 3(b) shows the output spectra of the C- and the L-band SLDs at an injection current of 50 mA. The output spectra are measured by using a tapered single-mode fiber. The peak intensity of the C-band SLD is just a little bit higher than that of the L-band SLD. The difference between two peak intensities of the output power is less than 1 dB. The center wavelengths of the C- and the L-band SLDs are around 1.55 μm and 1.575 μm, respectively. The 3 dB spectral widths of the C- and L-band emission spectrum are about 50 nm ranging from 1520 nm to 1570 nm and 70 nm ranging from 1540 nm to 1610 nm at the operating current of 50 mA, respectively, covering the full C- and L-bands. Note that the ripple is measured to be less than 3 dB up to 100 mA, over which the present C- and L-band SLDs are not lasing. Especially, to obtain the center wavelength of about 1.58 μm, we used the gain wavelength λ_g of 1.6 μm for the MQW active layer of the L-band SLD. In the wavelength region of 1.6 μm, the optical characteristics of the laser are degraded mainly due to nonradiative recombination. However, performance of the present optimized L-band SLD is comparable to that of the C-band SLD in the previous reports [12]. These results demonstrate that the performance of the SLD is more reliable and better than that of earlier-reported SLDs.

 

Fig. 3 (a) L-I characteristics of the fabricated SSC SLDs as a function of the injection current and (b) output spectra of SSC SLD.

Download Full Size | PDF

The FFPs observed from the SSC facet at the injection current of 50 mA are shown in Fig. 4 . In the figure, the black and red symbols represent the horizontal and vertical FFPs whose FWHMs are 14° and 13°, respectively. The result is almost consistent with the designed value. When compared with the FFPs in [12], more narrow and circular FPPs are obtained from the proposed BD-RWG of the SSC SLD with a passive waveguide core of 2 μm width and a B-RWG of 10 μm width.

 

Fig. 4 FFP characteristics from the SSC facet of SLD

Download Full Size | PDF

The L-band T-ECL is fabricated by using the L-band SLD, an aspheric microlens, and a tunable polymer Bragg reflector. The structure of the T-ECL is described in [12] in which the thermo-electric cooler (TEC) in a TO can is not included. The fabrication procedure and the performance of the tunable polymer Bragg reflector are detailed in [15]. The L-I curves of the T-ECL are shown in Fig. 5 . The temperature of the polymer Bragg grating is fixed by using a TEC, but the SLD temperature is not intentionally controlled. The black and red lines stand for the L-I characteristics of the C-band T-ECL [11] and the L-band T-ECL in the present study, respectively. The L-band T-ECL shows small kinks due to mode hopping at the current level of 47 mA and 62 mA. The kinks are caused by the temperature instability of the SLD by dint of the injection current. It is possible to reduce these kinks by using a TEC in the SLD TO can, as explained in [12]. The maximum output power is 7 mW and the slope efficiency is 0.138 mW/mA at 50 mA. Consequently, the output power is 0.6 mW higher and the slope efficiency is 0.03 mW/mA higher than those of the C-band T-ECL because of the less divergent beam.

 

Fig. 5 L-I characteristics of T-ECL with no TEC.

Download Full Size | PDF

Figure 6 shows the superimposed CW spectra of 16 channels spaced by 1 nm with a resolution of 0.01 nm. The SLD current is set to be 50 mA and the temperature of the Bragg reflector in the polymer waveguide is set to be 25°C. Figure 6(a) presents the output spectra of the C-band T-ECL, while Fig. 6(b) shows the superimposed spectra of the L-band T-ECL. The lasing wavelength of both T-ECLs are blueshifted linearly over a 15 nm wavelength range due to the thermo-optic effect of the Bragg reflector in the polymer waveguide, if the electric power is applied to the polymer waveguide from 0 to about 71 mW. The tuning range of the C- and the L-band T-ECLs is from 1545.8 to 1530.8 nm and from 1587.3 nm to 1572.3 nm, respectively. During the tuning, the output power of the C-band T-ECL varied from 3.5 to 2.46 dBm, with a variation of about 1 dB. The spectral peak power of the L-band T-ECL is higher than 3.73 dBm, with a small variation of 0.6dB. This result indicates that the spectral characteristics of the L-band are better than those of the C-band, and that is because the coupling efficiency is improved by using the L-band SLD with a more narrow and circular beam. The 20 dB bandwidth in both of the T-ECLs is less than 0.1 nm, and the side-mode suppression ratio (SMSR) is larger than 40 dB.

 

Fig. 6 Wavelength tuning characteristics of the polymer Bragg grating tunable laser for 16 channels; (a) C-band T-ECL, (b) L-band T-ECL.

Download Full Size | PDF

4. Conclusion

We have developed an L-band SLD with a SSC that possesses the characteristics of high output power and a narrow and circular beam for a light source of a T-ECL. The characteristics of the SLD are improved through optimizing the SSC structure of the SLD with a passive waveguide core of 2 μm in width and a BD-RWG of 10 μm in width. The average output power is 9 mW at the injection current of 60 mA, and the ripple is less than 3 dB. The horizontal and vertical FWHMs are 14° and 13°, respectively. Remarkably, the fabricated L-band T-ECL employing the novel SLD operates well enough for high performance. The device shows a maximum output power of 7 mW, and the slope efficiency is 0.138 W/A. These results imply that the proposed design of the device is useful as a light source for a T-ECL.