A 10 × 10 Gb/s DFB laser diode array fabricated using a SAG technique Oh Kee Kwon,* Yong Ahn Leem, Young Tak Han, Chul Wook Lee, Ki Soo Kim, and Su Hwan Oh Photonics-Wireless Convergence Components Research Department, Electronics and Telecommunications Research Institute, 138 Gajeongno, Yuseong-gu Daejeon, South Korea * [email protected]
Abstract: We present a ten-channel distributed feedback laser diode array (DFB-LDA) developed for the transmission of 100-Gb/s (10 × 10 Gb/s) signals separated by an 8 nm wavelength grid at a center wavelength of 1.55 μm. For the fabrication of this type of laser array, a selective area growth (SAG) technique, electron-beam lithography, and a reverse-mesa ridge waveguide LD processing technique were adopted to offer a tailored gain spectrum to each channel, providing both accurate lasing-wavelength control and excellent single-mode yield over all channels, and reducing the fabrication cost and electrical and thermal resistances. To evaluate the operational performance of the fabricated chip systematically, we also developed a sub-assembly module containing a ten-channel λ/4-shifted DFB-LDA, ten matching resistors, flexible printed circuit board (FPCB) wiring, and a thermistor on a metal optical bench. The static and dynamic properties of all channels of the fabricated array are examined in this paper. The developed sub-assembly module shows a side-mode suppression ratio (SMSR) of > 50 dB, a modulation bandwidth of > 10 GHz, and a clear eyeopening before and after a 2-km transmission with dynamic extinction ratio of > 5 dB. ©2014 Optical Society of America OCIS codes: (140.2010) Diode laser arrays; (140.3490) Lasers, distributed-feedback.
References and links 1. 2. 3.
IEEE 802.3ba 40 Gb/s and 100 Gb/s Ethernet Task Force Public Area, http://www.ieee802.org/3/ba/index.html. http://10x10msa.org/documents.htm. G. P. Li, T. Makino, A. Sarangan, and W. Huang, “16-wavelength gain-coupled DFB laser array with fine tenability,” IEEE Photon. Technol. Lett. 8(1), 22–24 (1996). 4. R. Nagarajan, C. H. Joyner, R. P. Schneider, J. S. Bostak, T. Butrie, A. G. Dentai, V. G. Dominic, P. W. Evans, M. Kato, M. Kauffman, D. J. H. Lambert, S. K. Mathis, A. Mathur, R. H. Miles, M. L. Mitchell, M. J. Missey, S. Murthy, A. C. Nilsson, F. H. Peters, S. C. Pennypacker, J. L. Pleumeekers, R. A. Salvatore, R. K. Schlenker, R. B. Taylor, Huan-Shang Tsai, M. F. Van Leeuwen, J. Webjorn, M. Ziari, D. Perkins, J. Singh, S. G. Grubb, M. S. Reffle, D. G. Mehuys, F. A. Kish, and D. F. Welch, “Large-scale photonic integrated circuits,” IEEE J. Sel. Top. Quantum Electron. 11(1), 50–65 (2005). 5. C. Zhang, S. Liang, H. Zhu, L. Han, and W. Wang, “Multichannel DFB laser arrays fabricated by upper SCH layer SAG technique,” IEEE J. Quantum Electron. 50(2), 92–97 (2014). 6. T. Schrans, G. Yoffe, Y. Luo, R. Narayan, S. Rangarajan, D. Hui, F. Kusnadi, A. Hanjani, and B. Pezeshki, “100Gb/s 10km link performance of 10×10Gb/s hybrid approach with integrated WDM array of DFB lasers,” in Proc. OFC/NEOEC 2009 (2009), NThA4. 7. O. K. Kwon, K. H. Kim, E. D. Sim, J. H. Kim, H. S. Kim, and K. R. Oh, “Broadly wavelength-tunable external cavity lasers with extremely low power variation over tuning range,” IEEE Photon. Technol. Lett. 17(3), 537– 539 (2005). 8. T. Fujii, M. Ekawa, and S. Yamazaki, “Growth pressure dependence of selective area metalorganic vapor phase epitaxy on planar patterned substrates,” J. Cryst. Growth 156(1–2), 59–66 (1995). 9. J. H. Song, K. Kim, Y. A. Leem, H. J. Kim, and G. Kim, “Strain-controlled selective-area growth of InGaAsP films on InP,” Jpn. J. Appl. Phys. 46(33), L783–L785 (2007). 10. O. K. Kwon, Y. A. Leem, D. H. Lee, C. W. Lee, Y. S. Baek, and Y. C. Chung, “Effects of asymmetric grating structures of output efficiency and single longitudinal mode operation in λ/4-shifted DFB laser,” IEEE J. Quantum Electron. 47(9), 1185–1194 (2011).
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9073
11. Y. T. Han, O. K. Kwon, D. H. Lee, C. W. Lee, Y. A. Leem, J. U. Shin, S. H. Park, and Y. Baek, “A costeffective 25-Gb/s EML TOSA using all-in-one FPCB wiring and metal optical bench,” Opt. Express 21(22), 26962–26971 (2013). 12. O. K. Kwon, Y. A. Leem, C. W. Lee, K. S. Kim, “Simple technique for analyzing bandgap changes of SAGgrown MQW structures,” will be submitted in IEEE Photon. Technol. Lett. 13. Aoyagi, S. Shirai, K. Takagi, T. Takiguchi, Y. Mihashi, C. Watatani, and T. Nishimura, “Uncooled directly modulated 1.3μm AlGaInAs-MQW DFB laser diodes,” Proc. SPIE 5595, 228–233 (2004). 14. K. Uomi, T. Tsuchiya, H. Nakano, M. Aoki, M. Suzuki, and N. Chinone, “High-speed and ultra-chirp 1.55 μm multiquantum well λ/4-shifted DFB lasers,” IEEE J. Quantum Electron. 27(6), 1705–1713 (1991). 15. P. A. Morton, T. Tanbun-Ek, R. A. Logan, A. M. Sergent, P. F. Sciortino, Jr., and D. L. Coblentz, “Frequency response subtraction for simple measurement of intrinsic laser dynamic properties,” IEEE Photon. Technol. Lett. 4(2), 133–136 (1992). 16. O. K. Kwon, C. W. Lee, D. H. Lee, E. D. Sim, J. H. Kim, and Y. S. Baek, “InP-based polarization-insensitive planar waveguide concave grating demultiplexer with flattened spectral response,” ETRI J. 31(2), 228–230 (2009).
1. Introduction There has been substantial interest in the development of 100-Gb/s Ethernet systems for relieving the ever-increasing data traffic congestion in local area networks (LANs), such as data centers, server networks, and lab/enterprising networks. The standardization of 100-Gb/s Ethernet (100-GbE) was completed in 2010 [1], and the adoption of 100-GbE Ethernet transceivers has started to grow. In addition, the specifications of 100-GbE multi-source agreement (MSA) were recently proposed and revised for a transmission of 100-Gb/s Ethernet signals over single mode fiber (SMF) with a length of 2 m to at least 2 km, 10 km, or 40 km (Rev 2.5) [2]. The transmitters operating in this specification require the use of ten 1.55 μm light sources separated by an 8 nm wavelength grid operating at a modulation rate of 10 Gb/s. These light sources should also be compact and cost-effective, and have good performance uniformity over all channels. A ten-channel 10-Gb/s distributed feedback laser diode array (DFB-LDA) is considered to be one of the most promising candidates for this type of light source. Although many studies have been conducted to implement multi-wavelength DFB-LDAs [3–6], with the exception of Ref [6], most of them have focused on the development of dense wavelength division multiplexing (DWDM) applications. When we fabricate this type of laser employing a conventional multiple quantum well (MQW) structure, the channel power can be severely limited owing to the relatively narrow gain spectrum (i.e., ~30 nm). Instead of the use of an asymmetric MQW structure [7], the gain spectrum of each channel can also be modified by utilizing the selective area growth (SAG) technique (i.e., a growth process on a maskpatterned semiconductor substrate) [8, 9]. This technique appears to be attractive since it can offer the tailored gain spectrum to each channel by simply adjusting the opening width between adjacent mask patterns under the specific growth temperature and pressure. It was previously reported that the photo-luminescence (PL) spectrum can be red-shifted by 88 nm for a 1.5-μm QW structure without degradation of the spectral quality [9]. From this viewpoint, it is feasible to realize a ten-channel DFB-LDA that can have almost the same optical gain properties for all channels (i.e., a wavelength range of 72 nm). In this paper, we fabricate and demonstrate a ten-channel DFB-LDA with an 8 nm wavelength grid for application of a 100-Gb/s Ethernet system. The SAG technique and Ebeam lithography are used to optimize the channel gain, and control the channel lasingwavelength and grating phase-shift, respectively. A reverse-mesa ridge waveguide (RMRWG) LD processing technique was adopted for fabrication of a λ/4-shifted DFB-LDA [10]. To evaluate the operational performances of the fabricated chip systematically, we developed a sub-assembly module for the array structure and examined the static and dynamic properties of each channel. The rest of this paper is organized as follows. The device and module structures, along with the SAG process results, are described in Section 2. Section 3 provides the experimental results obtained from the ten-channel DFB-LDA chip and its module. Finally, Section 4 summarizes the results of this research.
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9074
2. Device and module structures and SAG process results Figure 1 shows a photograph of an optical sub-assembly module for a 10 × 10 Gb/s DFBLDA. The module contains a ten-channel λ/4-shifted DFB-LDA, ten surface-mountable device (SMD) resistors, a ten-channel flexible printed circuit board (FPCB), and a thermistor on a copper-tungsten (CuW) metal optical bench (MOB). In this module, the FPCB has an arc-bent-shape in the form of a grounded coplanar waveguide (GCPW) [11], where the distances between both signal-lines at the input (right) and output (left) ports were designed to be 800 and 500 μm, respectively. For the fabricated 6-mm long FPCB, the RF return loss, insertion loss, and channel cross-talk level were measured to be < −24 dB, < 0.32 dB, and < −54 dB at 15 GHz, respectively. Considering the series resistance of each DFB-LD (i.e., ~5 Ω), the resistance of the SMD resistor was selected to be 45 Ω to match the line impedance of the FPCB.
FPCB SMD resistor
10 channel DFB-LDA
Thermistor
CuW MOB
Fig. 1. A photograph of a sub-assembly module of a 10 × 10 Gb/s DFB-LDA.
The epitaxial layers were grown using a lateral-flow-type metal-organic chemical vapor deposition (MOCVD). Trimethylindium (TMIn), trimethylgallium (TMGa), PH3, and AsH3 were used as source material. The layer stack before the SAG process consists of an n-InP buffer, a grating layer with a band-gap wavelength of 1.3 μm, an n-InP space layer, and a lattice-matched outer separate confinement hetero-structure (SCH) layer of 1.08 μm. The grating patterns were designed to have a channel-spacing of 8 nm for all channels by changing the grating period according to Bragg condition (i.e, λB = 2Λneq, where λB, Λ, and neq are the Bragg wavelength, the grating period, and the equivalent refractive index (with the material dispersion), respectively). Nearly rectangular-shaped 50% duty gratings were formed through E-beam writing and dry-etching. The mask patterns used for the SAG process have bilaterally symmetric shapes with a unit cell period P of 500 μm, as shown in Fig. 2. The spacing between adjacent mask stripes with a width of Wm within the unit cell is the same as the opening width Wo. For this pattern, we introduce an additional geometric parameter M to create the relation between Wm and Wo (i.e., Wm + Wo = M/2). Ten SAG mask patterns with different values of Wo ( = M/2−Wm, M = 400 μm) were designed to make the gain spectrums positioned near the respective channel wavelengths. After the mask-patterning, the InGaAsP SAG layers were grown at the temperature of 630 °C and the pressure of 100 mbar as follows: a 10-nm thick outer SCH layer, a 20-nm thick inner SCH layer with a band-gap wavelength of 1.24 μm, 7-pair QWs (i.e., 0.6% compressively strained 6-nm thick wells of 1.62 μm, and 0.45% tensile-strained 7.5-nm thick barriers of 1.3 μm), a 20-nm thick inner SCH layer, and a 10-nm thick outer SCH layer. After the SAG process, room-temperature PL measurements were performed at the centers of the opened areas. Figures 3(a) and 3(b) show the PL spectra and their peak-wavelength shifts for the mask patterns with different values of Wo. The PL spectra have almost the same shapes as a full width at half maximum (FWHM) of
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9075
~32 meV, as shown in the inlet of Fig. 3(a). As Wo decreases, its peak-intensity tends to be reduced to ~0.7, unlike the test result of the SAG layers grown on the mask patterns with the same Wo (i.e., no peak-intensity reduction in PL), and the peak-wavelength is shifted to a longer wavelength side with a wavelength interval of about 8 nm. It is well known that this red-shift is mainly originated from the lowering of the quantum level (caused by the increased growth rate) and the reduction of the band-gap energy (caused by the increased indium composition). For this result, the structures of SAG layers were examined. It was found that, at the Wo of 100 μm, the thicknesses of well and barrier become about 10 nm and 12.4 nm with the indium/gallium growth rate enhancement of 1.74/1.46, respectively. Detailed theoretical and experimental analyses related to this are explained in [12]. Bilateral symmetry
Bilateral symmetry
P
Mask
Mask
M
Wm
Wo+Wm=M/2
Wo
(P-M)/2
Fig. 2. Cross-sectional view of the mask pattern used for the SAG. 1.4
PL peak-wavelength shift [nm]
48 1.0
PL intensity [A.U.]
1.2
32 meV 0.5
1.0 0.8
0.0 150
100
50
0
-50
-100
Energy difference [meV]
0.6 0.4 0.2 0.0 1350
1400
1450
1500
1550
Wavelength [nm]
1600
1650
40 32 24 16 8 0 -8 -16 -24 -32 -40 90
100 110 120 130 140 150 160 170 180 190
Wo [μm]
Fig. 3. (a) Measured PL spectra and (b) PL peak-wavelength shifts for mask patterns with different values of Wo. The inlet of 3(a) is the synchronized superposition of the normalized PL spectra with respect to the (band-gap) energy difference from the peak-intensity energy.
After the removal of the SAG mask, a 30-nm thick outer SCH layer, a 100-nm thick p-InP residual cladding layer, a 20-nm thick p-InGaAsP etch-stop layer, a 2-μm thick p-InP upper cladding layer, and a 0.2-μm thick p + InGaAs layer were grown in sequence. RM-RWGs with a ridge-neck width of about 2 μm were fabricated. After the typical LD fabrication processes (e.g., benzo-cyclobutene (BCB) process, contact-layer opening, p-metallization, lapping, n-metallization, and scribing), both facets of 300-μm long DFB-LDA chip-bars were anti-reflection (AR) coated with an ion beam deposition of TiO2 and SiO2. A reflectivity of about 0.53% was obtained at 1.55 μm. To examine the effects of SAG on the gain properties, we first tested as-cleaved FabryPerot (FP) LDAs (without a grating layer). Figure 4 shows the net modal gain spectra for a 300-μm long FP-LDA at the thresholds. The gain spectra were obtained from the amplified spontaneous emission (ASE) spectra using the well-known Hakki-Paoli method. They are all nearly identical, and have a spectral width of about 35 nm at a net modal gain of −10 cm−1. In the L-I measurements, there were no considerable changes in the threshold current (i.e., < 2 mA) with slight reductions in the slope efficiency (i.e., ~0.025 W/A).
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9076
10
5
Net modal gain [1/cm]
0
0
35 nm
-5 -10 -15
-10
-20 -60
-40
-20
0
20
40
60
Wavelength difference [nm]
-20
-30
-40 1350
1400
1450
1500
1550
1600
1650
Wavelength [nm] Fig. 4. Net modal gain spectra for the fabricated ten-channel FP-LDA chip. The inlet shows the synchronized superposition of the gain spectra with respect to the (band-gap) wavelength difference from the intensity-peak wavelength.
3. Results of 10 × 10 Gb/s DMLA Figures 5(a) and 5(b) show typical fiber-coupled output spectra and L-I characteristics for the fabricated ten-channel DFB-LDA chips, respectively. All spectra were measured at a current of 50 mA. They show side-mode suppression ratios (SMSRs) of > 50 dB with an average channel-(wavelength) spacing of 8.2 nm. The DFB-LDA shows different channel properties on the L-I curve, unlike an FP-LDA. When the channel number is increased, the threshold current is increased gradually and the slope efficiency is decreased slightly with some variations (see Fig. (6)). This result may be due to the degradations of various material and structural parameters related to the threshold current in a DFB-LD employing SAG layers (e.g., differential gain, scattering and radiation losses caused by the grating, wavelength detuning from the gain peak, and so on). Nevertheless, we note here that as the number of channels increases (i.e., Wo decreases), the thicknesses and refractive-indices of the SAG layers increase owing to the increases in growth rate and Indium composition, respectively, and therefore the optical mode confined in this waveguide narrows. As a result, this narrowed optical field can reduce the coupling coefficient. To confirm this, we analyzed the optical modes for waveguide structures containing SAG layers, and then extracted their coupling coefficients using a perturbation method [10]. The calculations show a reduction of 0.52 in the normalized coupling coefficient (κL) of ch. 10. From this result, we can conclude that one of the main reasons for the increase in threshold current in a DFB-LDA is a reduction of the coupling coefficient through a decrease in the grating reflectivity. 1 2 3 4 5 6 7 8 9 10
0
16
Output power [mW]
Output spectrum [dB]
14 -20
-40
-60
12 10 8 6 4 2
-80 1500
1520
1540
1560
1580
Wavelength [nm]
1600
1620
0
0
20
40
60
80
100
120
Injection current [mA]
Fig. 5. (a) Typical fiber-coupled output spectra and (b) L-I characteristics for the fabricated ten-channel DFB-LDA chips. All measurements were performed at a temperature of 25°C. In (b), threshold currents and slope efficiencies are obtained to be 13(#1)~28(#8) mA and 0.165(#3)~0.143(#10) W/A (at 50 mA) for all channels, respectively.
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9077
dP/dI
30
0.18
25
0.16
20
0.14
15
0.12
10
0.10
1
2
3
4
5
6
7
8
9
10
Channel-spacing differnece [nm]
0.20
Ith
Slope efficiency dP/dI [W/A]
Threshold current Ith [mA]
35
2.0 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 -2.0
1
2
3
Channel #
4
5
6
7
8
9
10
Channel #
Fig. 6. (a) Threshold currents and slope efficiencies and (b) channel-spacing differences for five ten-channel DFB-LDA chips. In (a), all slope efficiencies were measured at a current of 50 mA. In (b), wavelengths of ch. 5 were used as a reference, and channel-spacing differences were obtained from the differences between the measured intensity-peak wavelengths and the channel wavelengths obtained from the reference λc (i.e., λc = λr + Δλ(i-r), where λr, Δλ, i, and r are the reference wavelength, channel-spacing (8 nm), channel number (1~10), and reference channel number (5), respectively).
On the other hand, it is shown in Fig. 6(b) that, as the channel number is increased, the channel-spacing is getting wider with the difference of ± 1.5 nm. This result can be explained by the increase in the refractive-indices of SAG layers which makes the channel Bragg wavelength red-shifted. As a result, to implement the SAG-grown DFB-LDA with the accurate channel spacing, the index changes of SAG layers should be considered in the design of the laser array. Figure 7 shows the electro-optic (EO) responses of a ten-channel DFB-LDA module. The −3 dB modulation bandwidths appear to be > 10 GHz for all channels. However, as the number of channels increases (i.e., Wo decreases), the bandwidth decreases from 12 to 10 GHz, and the over-shoot increases from 2 to 6 dB. For the reason of these results, it may be thought of as the changes of various parameters related to the modulation properties in a DFB-LD employing SAG layers (e.g., differential gain, carrier transport effect, non-uniform carrier distribution across the active region, and so on.) The important thing to note here is that a high κL structure (i.e., ch. 1) contains large amounts of photon densities within the cavity, and in particular, these photon densities are concentrated at the phase-shift. These attributes in the high κL structure increase the resonance frequency through a small-signal analysis [13] and suppress the optical gain through a nonlinear gain mechanism [14], which in turn enlarge the modulation bandwidth and increase the damping under the modulation, and vice versa. To confirm this experimentally, intrinsic responses were extracted from the measured EO responses at different bias currents using a subtraction method [15]. 9
EO response [dB]
6 3 0 -3 -6 -9 -12 -15
2
4
6
8
10
12
14
16
18
Modulation frequency [GHz] Fig. 7. Measured EO responses of a ten-channel DFB-LDA module. The measurement was performed individually for all channels. The bias currents were 80 mA for ch. 1 through ch. 8, and 100 mA for ch. 9 and ch.10.
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9078
Figures 8(a) and 8(b) show the resonance frequency fr and damping factor γ for the different bias currents for ch. 1 and ch. 10, respectively. It is clear that, compared to the results (of ch. 1) shown in Fig. 7(a), both fr and γ are reduced considerably. According to the relation between fr2 and γ, a K-factor of 0.282 ns for ch. 1, and 0.168 to 1.95 ns for ch. 10, was obtained. As a result, to improve the channel performance uniformity during static and dynamic conditions, it is necessary to optimize the coupling coefficient and channel-spacing of each channel by adjusting the grating duty and period, respectively. Figure 9 shows the measured eye patterns before and after a 2-km transmission. Owing to the large modulation bandwidth of the fabricated device and its module, the eye patterns are clearly opened with minimum dynamic extinction ratios (DERs) of 5.3 dB and 5 dB for before and after the transmission, respectively.
Fig. 8. Resonance frequency (black) and damping parameter (red) with different bias currents for (a) ch. 1 and (b) ch. 10. Ch.
BtoB
2 km transmission
Ch.
1
2
3
4
5
6
7
8
9
10
BtoB
2 km transmission
Fig. 9. Measured eye patterns at the back-to-back (BtoB) and after 2-km transmission for all channels. The measurement was performed individually for all channels. The bias currents were 80 mA for ch.1 through ch.8, and 100mA for ch. 9 and ch. 10. The modulation currents were ± 30 mA for all channels.
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9079
4. Summary We fabricated and demonstrated a ten-channel 10 Gb/s DFB-LDA based on the SAG technique and E-beam lithography. The measured PL spectra of the SAG layers grown on the designed mask pattern showed almost the same shapes with a wavelength interval of about 8 nm. For the FP-LDA, nearly identical spectra with a spectral width of about 35 nm were obtained for all channels without considerable changes in the threshold current or slope efficiency. For the DFB-LDA, while high SMSRs of > 50 dB were obtained for all channels, a gradual increase in the threshold current (i.e., 13 to 28 mA), a widening in the channelspacing (with the difference of ± 1.5 nm), a reduction in the modulation bandwidth (i.e., 12 to 10 GHz), and an enhancement in the over-shoot (i.e., 2 to 6 GHz) were found. We confirmed that these changes are closely related to a reduction of the coupling coefficient and an increase of the refractive index in the fabricated waveguide structure containing an SAG layer. Owing to the large modulation bandwidth of > 10 GHz for all channels, the developed module shows a clear eye opening before and after the 2-km transmission (with a DER of > 5 dB) for all channels. According to these results, we concluded that our DFB-LDA is capable of operating at a data rate of 10 Gb/s, and can be used as a low-cost light source for 100-Gb/s Ethernet transmitters. Additional improvement in the channel performance uniformity and simultaneous operation for all channels will be the subjects of future investigations. In addition, a DFB-LDA with a single output port can be realized through integration with an optical MUX [16]. Work realizing this type of laser is ongoing. Acknowledgment This work was supported by the “Energy Efficient Power Semiconductor Technology for Next Generation Data Center” IT R&D Project (No. 10038766) of the Korea Ministry of Knowledge Economy (MKE/KEIT).
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9080
Abstract: We present a ten-channel distributed feedback laser diode array (DFB-LDA) developed for the transmission of 100-Gb/s (10 × 10 Gb/s) signals separated by an 8 nm wavelength grid at a center wavelength of 1.55 μm. For the fabrication of this type of laser array, a selective area growth (SAG) technique, electron-beam lithography, and a reverse-mesa ridge waveguide LD processing technique were adopted to offer a tailored gain spectrum to each channel, providing both accurate lasing-wavelength control and excellent single-mode yield over all channels, and reducing the fabrication cost and electrical and thermal resistances. To evaluate the operational performance of the fabricated chip systematically, we also developed a sub-assembly module containing a ten-channel λ/4-shifted DFB-LDA, ten matching resistors, flexible printed circuit board (FPCB) wiring, and a thermistor on a metal optical bench. The static and dynamic properties of all channels of the fabricated array are examined in this paper. The developed sub-assembly module shows a side-mode suppression ratio (SMSR) of > 50 dB, a modulation bandwidth of > 10 GHz, and a clear eyeopening before and after a 2-km transmission with dynamic extinction ratio of > 5 dB. ©2014 Optical Society of America OCIS codes: (140.2010) Diode laser arrays; (140.3490) Lasers, distributed-feedback.
References and links 1. 2. 3.
IEEE 802.3ba 40 Gb/s and 100 Gb/s Ethernet Task Force Public Area, http://www.ieee802.org/3/ba/index.html. http://10x10msa.org/documents.htm. G. P. Li, T. Makino, A. Sarangan, and W. Huang, “16-wavelength gain-coupled DFB laser array with fine tenability,” IEEE Photon. Technol. Lett. 8(1), 22–24 (1996). 4. R. Nagarajan, C. H. Joyner, R. P. Schneider, J. S. Bostak, T. Butrie, A. G. Dentai, V. G. Dominic, P. W. Evans, M. Kato, M. Kauffman, D. J. H. Lambert, S. K. Mathis, A. Mathur, R. H. Miles, M. L. Mitchell, M. J. Missey, S. Murthy, A. C. Nilsson, F. H. Peters, S. C. Pennypacker, J. L. Pleumeekers, R. A. Salvatore, R. K. Schlenker, R. B. Taylor, Huan-Shang Tsai, M. F. Van Leeuwen, J. Webjorn, M. Ziari, D. Perkins, J. Singh, S. G. Grubb, M. S. Reffle, D. G. Mehuys, F. A. Kish, and D. F. Welch, “Large-scale photonic integrated circuits,” IEEE J. Sel. Top. Quantum Electron. 11(1), 50–65 (2005). 5. C. Zhang, S. Liang, H. Zhu, L. Han, and W. Wang, “Multichannel DFB laser arrays fabricated by upper SCH layer SAG technique,” IEEE J. Quantum Electron. 50(2), 92–97 (2014). 6. T. Schrans, G. Yoffe, Y. Luo, R. Narayan, S. Rangarajan, D. Hui, F. Kusnadi, A. Hanjani, and B. Pezeshki, “100Gb/s 10km link performance of 10×10Gb/s hybrid approach with integrated WDM array of DFB lasers,” in Proc. OFC/NEOEC 2009 (2009), NThA4. 7. O. K. Kwon, K. H. Kim, E. D. Sim, J. H. Kim, H. S. Kim, and K. R. Oh, “Broadly wavelength-tunable external cavity lasers with extremely low power variation over tuning range,” IEEE Photon. Technol. Lett. 17(3), 537– 539 (2005). 8. T. Fujii, M. Ekawa, and S. Yamazaki, “Growth pressure dependence of selective area metalorganic vapor phase epitaxy on planar patterned substrates,” J. Cryst. Growth 156(1–2), 59–66 (1995). 9. J. H. Song, K. Kim, Y. A. Leem, H. J. Kim, and G. Kim, “Strain-controlled selective-area growth of InGaAsP films on InP,” Jpn. J. Appl. Phys. 46(33), L783–L785 (2007). 10. O. K. Kwon, Y. A. Leem, D. H. Lee, C. W. Lee, Y. S. Baek, and Y. C. Chung, “Effects of asymmetric grating structures of output efficiency and single longitudinal mode operation in λ/4-shifted DFB laser,” IEEE J. Quantum Electron. 47(9), 1185–1194 (2011).
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9073
11. Y. T. Han, O. K. Kwon, D. H. Lee, C. W. Lee, Y. A. Leem, J. U. Shin, S. H. Park, and Y. Baek, “A costeffective 25-Gb/s EML TOSA using all-in-one FPCB wiring and metal optical bench,” Opt. Express 21(22), 26962–26971 (2013). 12. O. K. Kwon, Y. A. Leem, C. W. Lee, K. S. Kim, “Simple technique for analyzing bandgap changes of SAGgrown MQW structures,” will be submitted in IEEE Photon. Technol. Lett. 13. Aoyagi, S. Shirai, K. Takagi, T. Takiguchi, Y. Mihashi, C. Watatani, and T. Nishimura, “Uncooled directly modulated 1.3μm AlGaInAs-MQW DFB laser diodes,” Proc. SPIE 5595, 228–233 (2004). 14. K. Uomi, T. Tsuchiya, H. Nakano, M. Aoki, M. Suzuki, and N. Chinone, “High-speed and ultra-chirp 1.55 μm multiquantum well λ/4-shifted DFB lasers,” IEEE J. Quantum Electron. 27(6), 1705–1713 (1991). 15. P. A. Morton, T. Tanbun-Ek, R. A. Logan, A. M. Sergent, P. F. Sciortino, Jr., and D. L. Coblentz, “Frequency response subtraction for simple measurement of intrinsic laser dynamic properties,” IEEE Photon. Technol. Lett. 4(2), 133–136 (1992). 16. O. K. Kwon, C. W. Lee, D. H. Lee, E. D. Sim, J. H. Kim, and Y. S. Baek, “InP-based polarization-insensitive planar waveguide concave grating demultiplexer with flattened spectral response,” ETRI J. 31(2), 228–230 (2009).
1. Introduction There has been substantial interest in the development of 100-Gb/s Ethernet systems for relieving the ever-increasing data traffic congestion in local area networks (LANs), such as data centers, server networks, and lab/enterprising networks. The standardization of 100-Gb/s Ethernet (100-GbE) was completed in 2010 [1], and the adoption of 100-GbE Ethernet transceivers has started to grow. In addition, the specifications of 100-GbE multi-source agreement (MSA) were recently proposed and revised for a transmission of 100-Gb/s Ethernet signals over single mode fiber (SMF) with a length of 2 m to at least 2 km, 10 km, or 40 km (Rev 2.5) [2]. The transmitters operating in this specification require the use of ten 1.55 μm light sources separated by an 8 nm wavelength grid operating at a modulation rate of 10 Gb/s. These light sources should also be compact and cost-effective, and have good performance uniformity over all channels. A ten-channel 10-Gb/s distributed feedback laser diode array (DFB-LDA) is considered to be one of the most promising candidates for this type of light source. Although many studies have been conducted to implement multi-wavelength DFB-LDAs [3–6], with the exception of Ref [6], most of them have focused on the development of dense wavelength division multiplexing (DWDM) applications. When we fabricate this type of laser employing a conventional multiple quantum well (MQW) structure, the channel power can be severely limited owing to the relatively narrow gain spectrum (i.e., ~30 nm). Instead of the use of an asymmetric MQW structure [7], the gain spectrum of each channel can also be modified by utilizing the selective area growth (SAG) technique (i.e., a growth process on a maskpatterned semiconductor substrate) [8, 9]. This technique appears to be attractive since it can offer the tailored gain spectrum to each channel by simply adjusting the opening width between adjacent mask patterns under the specific growth temperature and pressure. It was previously reported that the photo-luminescence (PL) spectrum can be red-shifted by 88 nm for a 1.5-μm QW structure without degradation of the spectral quality [9]. From this viewpoint, it is feasible to realize a ten-channel DFB-LDA that can have almost the same optical gain properties for all channels (i.e., a wavelength range of 72 nm). In this paper, we fabricate and demonstrate a ten-channel DFB-LDA with an 8 nm wavelength grid for application of a 100-Gb/s Ethernet system. The SAG technique and Ebeam lithography are used to optimize the channel gain, and control the channel lasingwavelength and grating phase-shift, respectively. A reverse-mesa ridge waveguide (RMRWG) LD processing technique was adopted for fabrication of a λ/4-shifted DFB-LDA [10]. To evaluate the operational performances of the fabricated chip systematically, we developed a sub-assembly module for the array structure and examined the static and dynamic properties of each channel. The rest of this paper is organized as follows. The device and module structures, along with the SAG process results, are described in Section 2. Section 3 provides the experimental results obtained from the ten-channel DFB-LDA chip and its module. Finally, Section 4 summarizes the results of this research.
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9074
2. Device and module structures and SAG process results Figure 1 shows a photograph of an optical sub-assembly module for a 10 × 10 Gb/s DFBLDA. The module contains a ten-channel λ/4-shifted DFB-LDA, ten surface-mountable device (SMD) resistors, a ten-channel flexible printed circuit board (FPCB), and a thermistor on a copper-tungsten (CuW) metal optical bench (MOB). In this module, the FPCB has an arc-bent-shape in the form of a grounded coplanar waveguide (GCPW) [11], where the distances between both signal-lines at the input (right) and output (left) ports were designed to be 800 and 500 μm, respectively. For the fabricated 6-mm long FPCB, the RF return loss, insertion loss, and channel cross-talk level were measured to be < −24 dB, < 0.32 dB, and < −54 dB at 15 GHz, respectively. Considering the series resistance of each DFB-LD (i.e., ~5 Ω), the resistance of the SMD resistor was selected to be 45 Ω to match the line impedance of the FPCB.
FPCB SMD resistor
10 channel DFB-LDA
Thermistor
CuW MOB
Fig. 1. A photograph of a sub-assembly module of a 10 × 10 Gb/s DFB-LDA.
The epitaxial layers were grown using a lateral-flow-type metal-organic chemical vapor deposition (MOCVD). Trimethylindium (TMIn), trimethylgallium (TMGa), PH3, and AsH3 were used as source material. The layer stack before the SAG process consists of an n-InP buffer, a grating layer with a band-gap wavelength of 1.3 μm, an n-InP space layer, and a lattice-matched outer separate confinement hetero-structure (SCH) layer of 1.08 μm. The grating patterns were designed to have a channel-spacing of 8 nm for all channels by changing the grating period according to Bragg condition (i.e, λB = 2Λneq, where λB, Λ, and neq are the Bragg wavelength, the grating period, and the equivalent refractive index (with the material dispersion), respectively). Nearly rectangular-shaped 50% duty gratings were formed through E-beam writing and dry-etching. The mask patterns used for the SAG process have bilaterally symmetric shapes with a unit cell period P of 500 μm, as shown in Fig. 2. The spacing between adjacent mask stripes with a width of Wm within the unit cell is the same as the opening width Wo. For this pattern, we introduce an additional geometric parameter M to create the relation between Wm and Wo (i.e., Wm + Wo = M/2). Ten SAG mask patterns with different values of Wo ( = M/2−Wm, M = 400 μm) were designed to make the gain spectrums positioned near the respective channel wavelengths. After the mask-patterning, the InGaAsP SAG layers were grown at the temperature of 630 °C and the pressure of 100 mbar as follows: a 10-nm thick outer SCH layer, a 20-nm thick inner SCH layer with a band-gap wavelength of 1.24 μm, 7-pair QWs (i.e., 0.6% compressively strained 6-nm thick wells of 1.62 μm, and 0.45% tensile-strained 7.5-nm thick barriers of 1.3 μm), a 20-nm thick inner SCH layer, and a 10-nm thick outer SCH layer. After the SAG process, room-temperature PL measurements were performed at the centers of the opened areas. Figures 3(a) and 3(b) show the PL spectra and their peak-wavelength shifts for the mask patterns with different values of Wo. The PL spectra have almost the same shapes as a full width at half maximum (FWHM) of
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9075
~32 meV, as shown in the inlet of Fig. 3(a). As Wo decreases, its peak-intensity tends to be reduced to ~0.7, unlike the test result of the SAG layers grown on the mask patterns with the same Wo (i.e., no peak-intensity reduction in PL), and the peak-wavelength is shifted to a longer wavelength side with a wavelength interval of about 8 nm. It is well known that this red-shift is mainly originated from the lowering of the quantum level (caused by the increased growth rate) and the reduction of the band-gap energy (caused by the increased indium composition). For this result, the structures of SAG layers were examined. It was found that, at the Wo of 100 μm, the thicknesses of well and barrier become about 10 nm and 12.4 nm with the indium/gallium growth rate enhancement of 1.74/1.46, respectively. Detailed theoretical and experimental analyses related to this are explained in [12]. Bilateral symmetry
Bilateral symmetry
P
Mask
Mask
M
Wm
Wo+Wm=M/2
Wo
(P-M)/2
Fig. 2. Cross-sectional view of the mask pattern used for the SAG. 1.4
PL peak-wavelength shift [nm]
48 1.0
PL intensity [A.U.]
1.2
32 meV 0.5
1.0 0.8
0.0 150
100
50
0
-50
-100
Energy difference [meV]
0.6 0.4 0.2 0.0 1350
1400
1450
1500
1550
Wavelength [nm]
1600
1650
40 32 24 16 8 0 -8 -16 -24 -32 -40 90
100 110 120 130 140 150 160 170 180 190
Wo [μm]
Fig. 3. (a) Measured PL spectra and (b) PL peak-wavelength shifts for mask patterns with different values of Wo. The inlet of 3(a) is the synchronized superposition of the normalized PL spectra with respect to the (band-gap) energy difference from the peak-intensity energy.
After the removal of the SAG mask, a 30-nm thick outer SCH layer, a 100-nm thick p-InP residual cladding layer, a 20-nm thick p-InGaAsP etch-stop layer, a 2-μm thick p-InP upper cladding layer, and a 0.2-μm thick p + InGaAs layer were grown in sequence. RM-RWGs with a ridge-neck width of about 2 μm were fabricated. After the typical LD fabrication processes (e.g., benzo-cyclobutene (BCB) process, contact-layer opening, p-metallization, lapping, n-metallization, and scribing), both facets of 300-μm long DFB-LDA chip-bars were anti-reflection (AR) coated with an ion beam deposition of TiO2 and SiO2. A reflectivity of about 0.53% was obtained at 1.55 μm. To examine the effects of SAG on the gain properties, we first tested as-cleaved FabryPerot (FP) LDAs (without a grating layer). Figure 4 shows the net modal gain spectra for a 300-μm long FP-LDA at the thresholds. The gain spectra were obtained from the amplified spontaneous emission (ASE) spectra using the well-known Hakki-Paoli method. They are all nearly identical, and have a spectral width of about 35 nm at a net modal gain of −10 cm−1. In the L-I measurements, there were no considerable changes in the threshold current (i.e., < 2 mA) with slight reductions in the slope efficiency (i.e., ~0.025 W/A).
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9076
10
5
Net modal gain [1/cm]
0
0
35 nm
-5 -10 -15
-10
-20 -60
-40
-20
0
20
40
60
Wavelength difference [nm]
-20
-30
-40 1350
1400
1450
1500
1550
1600
1650
Wavelength [nm] Fig. 4. Net modal gain spectra for the fabricated ten-channel FP-LDA chip. The inlet shows the synchronized superposition of the gain spectra with respect to the (band-gap) wavelength difference from the intensity-peak wavelength.
3. Results of 10 × 10 Gb/s DMLA Figures 5(a) and 5(b) show typical fiber-coupled output spectra and L-I characteristics for the fabricated ten-channel DFB-LDA chips, respectively. All spectra were measured at a current of 50 mA. They show side-mode suppression ratios (SMSRs) of > 50 dB with an average channel-(wavelength) spacing of 8.2 nm. The DFB-LDA shows different channel properties on the L-I curve, unlike an FP-LDA. When the channel number is increased, the threshold current is increased gradually and the slope efficiency is decreased slightly with some variations (see Fig. (6)). This result may be due to the degradations of various material and structural parameters related to the threshold current in a DFB-LD employing SAG layers (e.g., differential gain, scattering and radiation losses caused by the grating, wavelength detuning from the gain peak, and so on). Nevertheless, we note here that as the number of channels increases (i.e., Wo decreases), the thicknesses and refractive-indices of the SAG layers increase owing to the increases in growth rate and Indium composition, respectively, and therefore the optical mode confined in this waveguide narrows. As a result, this narrowed optical field can reduce the coupling coefficient. To confirm this, we analyzed the optical modes for waveguide structures containing SAG layers, and then extracted their coupling coefficients using a perturbation method [10]. The calculations show a reduction of 0.52 in the normalized coupling coefficient (κL) of ch. 10. From this result, we can conclude that one of the main reasons for the increase in threshold current in a DFB-LDA is a reduction of the coupling coefficient through a decrease in the grating reflectivity. 1 2 3 4 5 6 7 8 9 10
0
16
Output power [mW]
Output spectrum [dB]
14 -20
-40
-60
12 10 8 6 4 2
-80 1500
1520
1540
1560
1580
Wavelength [nm]
1600
1620
0
0
20
40
60
80
100
120
Injection current [mA]
Fig. 5. (a) Typical fiber-coupled output spectra and (b) L-I characteristics for the fabricated ten-channel DFB-LDA chips. All measurements were performed at a temperature of 25°C. In (b), threshold currents and slope efficiencies are obtained to be 13(#1)~28(#8) mA and 0.165(#3)~0.143(#10) W/A (at 50 mA) for all channels, respectively.
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9077
dP/dI
30
0.18
25
0.16
20
0.14
15
0.12
10
0.10
1
2
3
4
5
6
7
8
9
10
Channel-spacing differnece [nm]
0.20
Ith
Slope efficiency dP/dI [W/A]
Threshold current Ith [mA]
35
2.0 1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 -2.0
1
2
3
Channel #
4
5
6
7
8
9
10
Channel #
Fig. 6. (a) Threshold currents and slope efficiencies and (b) channel-spacing differences for five ten-channel DFB-LDA chips. In (a), all slope efficiencies were measured at a current of 50 mA. In (b), wavelengths of ch. 5 were used as a reference, and channel-spacing differences were obtained from the differences between the measured intensity-peak wavelengths and the channel wavelengths obtained from the reference λc (i.e., λc = λr + Δλ(i-r), where λr, Δλ, i, and r are the reference wavelength, channel-spacing (8 nm), channel number (1~10), and reference channel number (5), respectively).
On the other hand, it is shown in Fig. 6(b) that, as the channel number is increased, the channel-spacing is getting wider with the difference of ± 1.5 nm. This result can be explained by the increase in the refractive-indices of SAG layers which makes the channel Bragg wavelength red-shifted. As a result, to implement the SAG-grown DFB-LDA with the accurate channel spacing, the index changes of SAG layers should be considered in the design of the laser array. Figure 7 shows the electro-optic (EO) responses of a ten-channel DFB-LDA module. The −3 dB modulation bandwidths appear to be > 10 GHz for all channels. However, as the number of channels increases (i.e., Wo decreases), the bandwidth decreases from 12 to 10 GHz, and the over-shoot increases from 2 to 6 dB. For the reason of these results, it may be thought of as the changes of various parameters related to the modulation properties in a DFB-LD employing SAG layers (e.g., differential gain, carrier transport effect, non-uniform carrier distribution across the active region, and so on.) The important thing to note here is that a high κL structure (i.e., ch. 1) contains large amounts of photon densities within the cavity, and in particular, these photon densities are concentrated at the phase-shift. These attributes in the high κL structure increase the resonance frequency through a small-signal analysis [13] and suppress the optical gain through a nonlinear gain mechanism [14], which in turn enlarge the modulation bandwidth and increase the damping under the modulation, and vice versa. To confirm this experimentally, intrinsic responses were extracted from the measured EO responses at different bias currents using a subtraction method [15]. 9
EO response [dB]
6 3 0 -3 -6 -9 -12 -15
2
4
6
8
10
12
14
16
18
Modulation frequency [GHz] Fig. 7. Measured EO responses of a ten-channel DFB-LDA module. The measurement was performed individually for all channels. The bias currents were 80 mA for ch. 1 through ch. 8, and 100 mA for ch. 9 and ch.10.
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9078
Figures 8(a) and 8(b) show the resonance frequency fr and damping factor γ for the different bias currents for ch. 1 and ch. 10, respectively. It is clear that, compared to the results (of ch. 1) shown in Fig. 7(a), both fr and γ are reduced considerably. According to the relation between fr2 and γ, a K-factor of 0.282 ns for ch. 1, and 0.168 to 1.95 ns for ch. 10, was obtained. As a result, to improve the channel performance uniformity during static and dynamic conditions, it is necessary to optimize the coupling coefficient and channel-spacing of each channel by adjusting the grating duty and period, respectively. Figure 9 shows the measured eye patterns before and after a 2-km transmission. Owing to the large modulation bandwidth of the fabricated device and its module, the eye patterns are clearly opened with minimum dynamic extinction ratios (DERs) of 5.3 dB and 5 dB for before and after the transmission, respectively.
Fig. 8. Resonance frequency (black) and damping parameter (red) with different bias currents for (a) ch. 1 and (b) ch. 10. Ch.
BtoB
2 km transmission
Ch.
1
2
3
4
5
6
7
8
9
10
BtoB
2 km transmission
Fig. 9. Measured eye patterns at the back-to-back (BtoB) and after 2-km transmission for all channels. The measurement was performed individually for all channels. The bias currents were 80 mA for ch.1 through ch.8, and 100mA for ch. 9 and ch. 10. The modulation currents were ± 30 mA for all channels.
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9079
4. Summary We fabricated and demonstrated a ten-channel 10 Gb/s DFB-LDA based on the SAG technique and E-beam lithography. The measured PL spectra of the SAG layers grown on the designed mask pattern showed almost the same shapes with a wavelength interval of about 8 nm. For the FP-LDA, nearly identical spectra with a spectral width of about 35 nm were obtained for all channels without considerable changes in the threshold current or slope efficiency. For the DFB-LDA, while high SMSRs of > 50 dB were obtained for all channels, a gradual increase in the threshold current (i.e., 13 to 28 mA), a widening in the channelspacing (with the difference of ± 1.5 nm), a reduction in the modulation bandwidth (i.e., 12 to 10 GHz), and an enhancement in the over-shoot (i.e., 2 to 6 GHz) were found. We confirmed that these changes are closely related to a reduction of the coupling coefficient and an increase of the refractive index in the fabricated waveguide structure containing an SAG layer. Owing to the large modulation bandwidth of > 10 GHz for all channels, the developed module shows a clear eye opening before and after the 2-km transmission (with a DER of > 5 dB) for all channels. According to these results, we concluded that our DFB-LDA is capable of operating at a data rate of 10 Gb/s, and can be used as a low-cost light source for 100-Gb/s Ethernet transmitters. Additional improvement in the channel performance uniformity and simultaneous operation for all channels will be the subjects of future investigations. In addition, a DFB-LDA with a single output port can be realized through integration with an optical MUX [16]. Work realizing this type of laser is ongoing. Acknowledgment This work was supported by the “Energy Efficient Power Semiconductor Technology for Next Generation Data Center” IT R&D Project (No. 10038766) of the Korea Ministry of Knowledge Economy (MKE/KEIT).
#206949 - $15.00 USD (C) 2014 OSA
Received 24 Feb 2014; revised 14 Mar 2014; accepted 23 Mar 2014; published 7 Apr 2014 21 April 2014 | Vol. 22, No. 8 | DOI:10.1364/OE.22.009073 | OPTICS EXPRESS 9080
