High-gain 1.3 μm GaInNAs semiconductor optical amplifier with enhanced temperature stability for all-optical signal processing at 10 Gb/s D. Fitsios,1,* G. Giannoulis,2 V.-M. Korpijärvi,3 J. Viheriälä,3 A. Laakso,3 N. Iliadis,2 S. Dris,2 M. Spyropoulou,2 H. Avramopoulos,2 G. T. Kanellos,4 N. Pleros,1,4 and M. Guina3 1 2
Department of Informatics, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece
School of Electrical and Computer Engineering, National Technical University of Athens, Athens, Greece 3 4
Optoelectronics Research Centre (ORC), Tampere University of Technology, Tampere, Finland
Information Technologies Institute, Center for Research and Technology Hellas, Thessaloniki, Greece *Corresponding author: [email protected] Received 2 September 2014; revised 13 November 2014; accepted 14 November 2014; posted 17 November 2014 (Doc. ID 222131); published 23 December 2014
We report on the complete experimental evaluation of a GaInNAs/GaAs (dilute nitride) semiconductor optical amplifier that operates at 1.3 μm and exhibits 28 dB gain and a gain recovery time of 100 ps. Successful wavelength conversion operation is demonstrated using pseudorandom bit sequence 27 − 1 non-return-to-zero bit streams at 5 and 10 Gb∕s, yielding error-free performance and showing feasibility for implementation in various signal processing functionalities. The operational credentials of the device are analyzed in various operational regimes, while its nonlinear performance is examined in terms of four-wave mixing. Moreover, characterization results reveal enhanced temperature stability with almost no gain variation around the 1320 nm region for a temperature range from 20°C to 50°C. The operational characteristics of the device, along with the cost and energy benefits of dilute nitride technology, make it very attractive for application in optical access networks and dense photonic integrated circuits. © 2014 Optical Society of America OCIS codes: (250.5980) Semiconductor optical amplifiers; (130.4815) Optical switching devices; (160.6000) Semiconductor materials. http://dx.doi.org/10.1364/AO.54.000046
1. Introduction
Semiconductor optical amplifiers (SOAs) have already been established as key elements in telecom and datacom environments, due to their small footprint, fast gain recovery, broadband gain, and ability for integration [1–6]. SOAs offer a wide range of optical functionalities, with some indicative examples including optical amplification with SOAs as standalone modules [1] or integrated with electro-absorption 1559-128X/15/010046-07$15.00/0 © 2015 Optical Society of America 46
APPLIED OPTICS / Vol. 54, No. 1 / 1 January 2015
modulators [2], as well as colorless reflective modulation [3] using reflective semiconductor optical amplifiers (RSOAs). Moreover, SOAs have been shown to enable advanced functionalities such as optical random access memory (RAM) [4]. As the number of elements on-chip increases [5], operational temperatures are expected to significantly increase, affecting the circuit’s operation and requiring power-hungry external cooling. The majority of SOA devices reported so far have relied on the rather expensive InP for the active material platform [6]. On the other hand, GaInNAs/ GaAs (dilute nitride) technology has emerged as an
attractive alternative for conventional InP-based materials, due to its significantly decreased cost combined with competitive performance characteristics [7]. Moreover, an additional advantage of dilute nitrides is the large conduction band offset between the GaInNAs quantum well and GaAs barrier, which is manifested as less temperature-sensitive operation [8]. Excellent performance and temperature characteristics have already been demonstrated by a number of GaInNAs-based laser devices [9,10]; however, only a few dilute nitride edge-emitting SOAs have been presented so far. Pozo et al. [11] demonstrated a 1300 nm dilute nitride SOA capable of multiwavelength amplification. Studies by Piwonski et al. [12] on ultrafast gain dynamics proved that SOAs operating near 1550 nm can have a gain recovery time of 370 ps. Additionally, the 1.3 μm travelling-wave SOA presented by Hashimoto et al. in [13] had a peak chip gain of 14 dB, while exhibiting much smaller temperature dependence than conventional InP SOAs. Nevertheless, only recently, we demonstrated a dilute nitride SOA operating at 1.3 μm, providing simultaneously both high gain and low gain-recovery time values, while offering enhanced temperature stability [14]. The device exhibited a gain value of 28 dB and gain recovery time of around 100 ps, while it showed almost no gain variation around the 1320 nm region for a temperature range from 20°C to 50°C, raising the potential for uncooled operation. In this paper, we provide a thorough analysis on our SOA device reported in [14], offering further experimental verification and detailed insight into its performance credentials. Successful wavelength conversion operation at 5 and 10 Gb∕s is reported, while in each case the respective Q-factor values are measured, indicating error-free operation. The gain recovery time of the device is measured in various operational regimes, supported by statistical distribution values. Finally, the nonlinear performance of the device is assessed by exploring fourwave-mixing (FWM) phenomena. Taking into account the enhanced temperature stability and the cost advantages of GaAs-based materials, we believe that the proposed dilute nitride SOA technology could potentially form a good candidate for nextgeneration optical access network and datacom environments. 2. Fabrication and Experimental Setup
The GaInNAs/GaAs semiconductor heterostructure can be seen in Fig. 1(a). It was grown on n-GaAs (100) substrate by plasma-assisted molecular beam epitaxy (PA-MBE). A radio frequency plasma source was used to produce active nitrogen for the growth. A valved cracker was used for arsenic and normal effusion cells for group-III elements and silicon and beryllium dopants. The structure consists of two 7-nm-thick GaIn0.31 N0.02 As quantum wells (QWs), which are surrounded by tensile-strained GaNAs for strain compensation. The thickness of GaNAs
Fig. 1. (a) SOA heterostructure and respective bandgap energy. (b), (c) SEM images of (b) facet of the SOA and (c) surface of the SOA.
is 10 nm between the QWs as well as on the outer sides of both QWs. This QW group is embedded in a waveguide region of 103 nm of GaAs on both sides of the group. The intrinsic waveguide region is further clad by Al0.60 GaAs layers, which are n-doped below the QWs and p-doped above. On top of the p-AlGaAs cladding, a 200 nm p-GaAs contact layer finalizes the structure. Following, the material was processed into a ridge waveguide by patterning a 200 nm thick SiN etch mask with UV lithography and etching the SOA structure with inductively coupled plasma reactive ion etching (ICP-RIE), due to its ability to provide etch uniformity across the semiconductor surface. A scanning electron microscope (SEM) image of the waveguide cross section can be seen in Fig. 1(b). The 2 μm width and 1.7 μm height of the ridge were designed to optimize the optical confinement structure of the device. Ridge height was controlled using in situ reflectance measurement. Consequently, the etch mask was removed and a new round of photolithography and dry etching was performed to define the current insulation layer. Ti/Pt/Au ohmic metal contact layers were deposited on the front surface of the sample. After mechanical lapper thinning and polishing, Ni/Au/Ge/Au ohmic metal contact layers were deposited on the back surface and annealed. The device was then cleaved to form a 1 mm long cavity. As can be seen in the top view SEM image of the device in Fig. 1(c), facets on the ends of the waveguide were cleaved on a 7° angle to reduce reflectivity and allow for single-pass amplification. Both facets were then antireflection (AR) coated using a SiO2 ∕TiO2 film pair to minimize reflections, further reducing reflectivity below 0.1%. The SOA was subsequently bonded p side up on C mounts for experimental evaluation. A list of the main SOA parameters can be found in Table 1. A fiber alignment probe station was employed for the SOA chip experimental evaluation. Light was coupled in and out of the 2 μm ridge width SOA 1 January 2015 / Vol. 54, No. 1 / APPLIED OPTICS
47
Table 1.
Group refractive index Total confinement factor for the two QWs in vertical direction Effective refractive index Current density at transparency Carrier density at transparency Internal losses Internal quantum efficiency
SOA Parameters
3.62 4.3%
3.244 739 A∕cm2 1.3718 cm−3 at T 300 K (26.85°C) 1.5318 cm−3 at T 325 K (51.85°C) 1.6718 cm−3 at T 350 K (76.85°C) 31 cm−1 0.6
waveguide by using lensed fibers with a spot size diameter of 2.5 μm in order to ensure minimum coupling losses of around 5 dB per facet. The complete experimental setup can be seen in Fig. 2. For the static gain measurements a CW signal (probe signal) at 1309.95 nm (λ2) was fed into the SOA after passing through a variable optical attenuator (VOA) utilized for controlling the input signal power level. A polarization controller (PC) was used for adjusting the signal polarization state to compensate for the SOA polarization-dependent gain (PDG). Moreover, isolators were employed at the input and output of the chip to eliminate backreflections of light at the fiber/air and waveguide/air interfaces. The output signal was then inserted in a 4 nm Bragg grating (BG) band-pass filter to reject ASE noise, and its power was measured with a powermeter. Wavelength conversion operation was achieved by means of pump–probe cross-gain modulation (XGM), where a CW signal (pump signal) at 1314.01 nm (λ1) was launched into a Ti:LiNbO3 modulator (MOD) driven by a programmable pattern generator (PPG) so as to produce a 27 − 1 pseudo-random bit sequence (PRBS)
Fig. 2. (a) Experimental setup. (b) Picture of SOA chip on fiber alignment probe station. 48
APPLIED OPTICS / Vol. 54, No. 1 / 1 January 2015
of non-return-to-zero (NRZ) data pulses at 5 or 10 Gb∕s. For the dynamic gain performance measurements of the SOA, the pump signal was reprogrammed to include at least 18 consecutive “0”s between successive logical “1” pulse streams so as to allow sufficient time for full gain recovery. The wavelength converted signal at λ2 was then inserted into the 4 nm band-pass filter and analyzed at the oscilloscope. The respective statistical values were calculated using the MATLAB statistics toolbox [15]. Finally, the temperature dependence of the SOA performance was evaluated by varying the operating temperature of the chip using a Peltier element that was placed beneath the chip mount. 3. Experimental Results
Figure 3 illustrates the ASE spectra of the SOA for various injected currents, with the temperature set at 21.5°C. As can be seen, the emission peak experiences a blue shift as the injected current increases, reaching a maximum shift of 5.4 nm for increasing the injected current from 200 to 400 mA. For an injection current value of 400 mA, the −3 dB spectral bandwidth reaches a value of almost 30 nm. Figure 4 shows the chip gain as a function of input power with the temperature set at 20°C. As can be seen, a peak gain value of 28 dB is achieved, which is to our knowledge the highest ever reported for dilute nitride SOAs. Moreover, the saturated output power as a function of the signal input power can be also seen in Fig. 4. The maximum output power was measured at around 11 dBm for an input signal of −10 dBm. It should be noted that fiber-to-facet and facet-to-fiber coupling losses of 5 dB each have been taken into account for calculating the above values. The PDG of the device was 9.7 dB, measured at the small-signal gain region. However, this value was not optimized for the device in this initial fabrication run and will be a target for investigation in future fabrication runs, where we will examine the applicability of techniques widely used in SOAs so far [16]. The noise figure (NF) of the device was calculated using the well-known formula NF OSNRin∕OSNRout, while the OSNR was estimated using the polarization nulling method,
Fig. 3. SOA ASE spectra for various injected currents.
Fig. 4. SOA gain curve as a function of the input power and SOA output power as a function of the input power. Temperature set at 20°C.
reported in [17]. For these measurements, a signal at 1309.454 nm was modulated to produce a 10 Gb∕s 27 − 1 PRBS sequence and inserted into a commercially available linear SOA, fabricated by Kamelian, for amplification. Consequently, the signal was injected into the dilute nitride SOA under test. The NF was calculated for an input power of −15 dBm, corresponding to a gain value of 25 dB, close to the small signal gain region. The NF value was NF 7.5 dB. The respective signal spectra for the obtained NF measurements can be seen in Fig. 5, where Fig. 5(a) depicts the spectrum of the input signal after amplification through the linear SOA, while Fig. 5(b) illustrates the spectrum of the dilute nitride SOA output signal in the case of a small signal gain of 25 dB. The NF performance of our device is similar to that reported for a 1310 nm InGaAsP–InP quantum-well SOA with 25 dB gain, reported in [18], as well as with commercial quantum-well SOAs developed by Aeon and Kamelian [19,20]. Figure 6 depicts the wavelength converted signal at λ2 at the SOA output providing an evaluation of the gain recovery times for three different gain saturation regimes. Transition into different saturation regimes is achieved by adjusting the power of the probe signal. As can be observed, the measured 10%–90% gain recovery time reduces from 203 to 156 ps as we move into saturation; however, the SOA’s gain recovery is shortened in the deep saturation regime reaching a value of approximately 100 ps, a value that can allow for operation
Fig. 5. Optical spectrum of (a) input signal after amplification through the linear SOA and (b) GaInNAs SOA output signal.
Fig. 6. (a)–(c) SOA gain recovery dynamics for various gain saturation regimes and (d)–(f) respective statistical distributions.
in various signal processing functionalities at 10 Gb∕s. It should be noted that the low extinction ratio in Fig. 6(c) owes to the low differential gain in the deep saturation regime. The validity of the results was checked with a statistically adequate number of 104 measured samples for each case, the statistical distribution of which is shown in Figs. 6(d)–6(f), corresponding to the three gain saturation regimes shown in Figs. 6(a)–6(c), respectively. The statistical distribution in Fig. 6(d) has a mean value of 203 ps and a standard deviation of 25.8 ps, while the statistical distribution in Fig. 6(e) has a mean value of 156 ps and a standard deviation of 36.6 ps, and in Fig. 6(f) a mean value of 103 ps and a standard deviation of 33.3 ps. The effect of the injected current on the SOA gain recovery times was examined in different cases, specifically for 310, 345, and 390 mA, as can be seen in Figs. 7(a)–7(c), respectively. The mean values for the gain recovery time were 219.9, 215.2, and 203.2 ps, respectively, calculated through the respective statistical distributions in Figs. 7(d)–7(f). For a current of 310 mA, the standard deviation of the measured samples was 48.7 ps, while for currents of 345 and 390 mA the standard deviation values were 39.2 and 25.8 ps, respectively. Figures 8(a) and 8(b) illustrate the obtained eye diagrams of the wavelength converted signal carrying the complementary PRBS streams at 5 and 10 Gb∕s PRBS, respectively. The 100 ps gain recovery time of the device can be also confirmed by the 10 Gb∕s eye diagram of the wavelength converted signal displayed in Fig. 8(b). At both operational speeds, clear eye openings are obtained; however, a triangular pulse shape is observed at 1 January 2015 / Vol. 54, No. 1 / APPLIED OPTICS
49
Fig. 7. (a)–(c) SOA gain recovery dynamics for various injected currents and (d)–(f) respective statistical distributions.
10 Gb∕s as the device is near its operational speed limits due to the gain recovery time of 100 ps. Due to having the receiver in an AC-coupled configuration, it was not possible to extract any extinction ratio information. On the other hand, some information about the amplitude modulation of the eye diagram
can be given by the eye evaluation metrics, specifically by the standard deviation “σ 1 ” of level “1” of the eye diagrams. In particular, σ 1 had values of 0.118 at 5 Gb∕s and 0.168 at 10 Gb∕s, respectively, indicating a rather small divergence in the “1” level. A reason for the noise appearing in level “1” of the eye diagrams is the spectral bandwidth of the 4 nm BG band-pass filter that is employed at the SOA output to reject the ASE noise, due to the lack of any narrower filters at our disposal. Consequently, a significant portion of outband noise passes through the filter and is contributing to this noise level. Moreover, the wavelength converted signal at the output of the SOA appears with inversed polarity, and, hence, the noise at the level of ones originates from the unsaturated gain of the SOA in the presence of an input bit with logic ‘0’ value. Finally, for the case of 10 Gb∕s wavelength conversion, the gain recovery time of the SOA induces some amplitude modulation, as it is near its practical speed limit. Bit error rate (BER) measurements were performed for 5 and 10 Gb∕s using offline processing in MATLAB. The electrical signals from the photoreceiver were digitized using a 80 GSa∕s, 33 GHz realtime oscilloscope. Digital signal processing (DSP) was used for clock recovery and finding the optimum sampling point, after which errors could be directly counted and the BER calculated. No further processing was applied (i.e., no equalization or any other DSP that could improve signal quality). The measured BERs, as well as the 95% confidence interval (CI), are shown in Table 2. Note that the number of bits that could be processed was limited by the memory depth of the oscilloscope (5 · 105 for 5 Gb∕s and 1 · 106 for 10 Gb∕s). The BER was measured at 9.1 · 10−5 for the 10 Gb∕s case, while no errors were observed for 5 Gb∕s, indicating that the BER is better than 7.4 · 10−6 (the upper limit of the CI). For both cases, the measured BER values were found to be below 10−4, well below the standard forward error correction (FEC, 7% overhead) of 2 · 10−3 [21]. These results were taken with the probe and pump signals reaching the SOA with powers of −3 and 3.6 dBm, respectively. Since the SOA device was designed for DC operation, its electrical bandwidth is only 0.5 GHz. However, targeting on high-speed direct modulation schemes using the 1.3 μm GaInNAs SOA, alternative integration concepts [9] and reported techniques [22] could be followed to essentially improve the EO bandwidth of the SOA. In order to examine the device’s nonlinear performance in terms of FWM, we injected two copropagated CW signals at 1310.4 and at 1310.8 nm, respectively. As can be seen in Fig. 9, the measured downconversion FWM efficiency was Table 2.
Wavelength Conversion Speed Fig. 8. Obtained wavelength converted signal eye diagrams for PRBS stream at (a) 5 Gb∕s and (b) 10 Gb∕s. 50
APPLIED OPTICS / Vol. 54, No. 1 / 1 January 2015
5 Gb∕s 10 Gb∕s
Bit Error Rate Measurements
Q-Factor Measured (dB) BER 11.73 9.93
0 9.1 · 10−5
95% CI (Lower Limit)
95% CI (Upper Limit)
0 7.3 · 10−5
7.4 · 10−6 1.1 · 10−4
Fig. 9. Spectrum for four-wave-mixing operation.
around −36 dB, showing that the device holds potential for applications requiring nonlinear operation. As dilute nitride materials exhibit low temperature sensitivity credentials, an examination of the device’s operation in different temperatures was performed. Figure 10(a) depicts the ASE spectra of the SOA for different operating temperatures up to 50°C, with the injection current set at 350 mA. The 3 dB bandwidth of the SOA’s ASE profile is 53.8 nm and varies only within 0.6 nm for the whole temperature range. Moreover, a spectral redshift of around 18 nm is observed by increasing the operating temperature from
Fig. 10. (a) ASE spectra of the SOA for different operating temperatures. (b) Measured ASE power at the SOA facet versus injected current for various temperatures.
20°C to 50°C. Nevertheless, the ASE peak amplitude at around 1290 nm varies within only 3.9 dB for a temperature shift from 20°C to 50°C, while the ASE amplitude level around the 1320 nm region remains almost constant within the entire temperature range, revealing excellent thermal stability. It is worth noting that the temperature dependence of the ASE output characteristics of the device is significantly smaller compared to that of conventional InP-based SOAs with similar cavity length reported in the literature [13]. Figure 10(b) shows the measured ASE power at the SOA facet versus the injected current for various temperatures. As can be seen, a value of around 0 dBm was obtained for a current of around 390 mA, with the temperature set at 20°C. This value saturates for currents above 390 mA. As the temperature of the SOA increases, the SOA ASE power starts to drop. However, this decrease in ASE power is limited to only 3.8 dB for a temperature increase to 50°C, once again confirming the good thermal stability of the device. 4. Conclusions
We have performed a thorough experimental evaluation of the performance of a SOA based on GaInNAs/GaAs and operating at 1.3 μm. The device exhibited 28 dB gain, an output power of 11 dBm, and a gain recovery time of approximately 100 ps, record values for SOAs based on this material system. These values are more than adequate for operation as an amplification medium in metro and access networks [23], while making this a candidate material system for several on-chip functionalities in photonic integrated circuits (PICs). Moreover, wavelength conversion operation was performed successfully at both 5 and 10 Gb∕s, confirming the device’s credentials as both an amplifier and a medium for all-optical signal processing functionalities. The list of applications that could be enabled also includes the optical cache scheme reported in [24], where SOA-based switches are employed to constitute optical memory schemes, as well access enabling elements. The bit-level buffering solutions that can be offered by SOAs [4] can enable low-latency and line-rate performing optical routing solutions, while a whole list of optical processing applications [25,26] can be utilized in network environments [27]. As GaInNAs/GaAs technology has the advantages of significantly decreased cost and enhanced temperature stability, the proposed SOA device could be potentially utilized in optical access network applications, where operation without the need for cooling is highly beneficial energy- and cost-wise. Moreover, the device’s enhanced insensitivity to increased temperature can facilitate the exploitation of stateof-the-art hybrid integration of multiple active elements on a silicon-on-insulator (SOI) platform [28], allowing for higher integration densities and for more advanced on-chip functionalities. The work of D. Fitsios was supported by the IKY Foundation through the SIEMENS Fellowship of 1 January 2015 / Vol. 54, No. 1 / APPLIED OPTICS
51
Excellence Program for Post-Graduate Studies in Greece. This work has also been supported in part by the European Commission through the FP7-ICTFET Open project RAMPLAS (Contract No. 270773). References 1. H. Wang, J. Wu, and J. Lin, “Spectral characteristics of optical pulse amplification in SOA under assist light injection,” J. Lightwave Technol. 23, 2761–2771 (2005). 2. K. Ławniczuk, O. Patard, R. Guillamet, N. Chimot, A. Garreau, C. Kazmierski, G. Aubin, and K. Merghem, “40Gb/s colorless reflective amplified modulator,” IEEE Photon. Technol. Lett. 25, 341–343 (2013). 3. Z. Al-Qazwini, M. Thollabandi, and H. Kim, “Colorless optical transmitter for upstream WDM PON based on wavelength conversion,” J. Lightwave Technol. 31, 896–902 (2013). 4. D. Fitsios, C. Vagionas, G. T. Kanellos, A. Miliou, and N. Pleros, “Dual-wavelength bit input optical RAM with three SOA-XGM switches,” IEEE Photon. Technol. Lett. 24, 1142–1144 (2012). 5. S. C. Nicholes, M. L. Masanovic, B. Jevremovic, E. Lively, L. A. Coldren, and D. J. Blumenthal, “An 8 × 8 InP monolithic tunable optical router (MOTOR) packet forwarding chip,” J. Lightwave Technol. 28, 641–650 (2010). 6. M. Cabezon, A. Villafranca, D. Izquierdo, J. J. Martinez, and I. Garces, “Integrated all-optical 4-input NOR logic gate based on InP technology,” in Conference on Lasers and Electro-Optics (CLEO), San Jose, California, 6–11 May 2012, paper JW4A.7. 7. X. Yang, M. J. Jurkovic, J. B. Heroux, and W. I. Wang, “Molecular beam epitaxial growth of InGaAsN:Sb/GaAs quantum wells for long-wavelength semiconductor lasers,” Appl. Phys. Lett. 75, 178–180 (1999). 8. M. Kondow, S. Nakatsuka, T. Kitatani, Y. Yazawa, and M. Okai, “Room-temperature pulsed operation of GaInNAs laser diodes with excellent high-temperature performance,” Jpn. J. Appl. Phys. 35, 5711–5713 (1996). 9. M. Dumitrescu, M. Wolf, K. Schulz, Y. Q. Wei, G. Adolfsson, J. Gustavsson, J. Bengtsson, M. Sadeghi, S. Wang, A. Larsson, J. Lim, E. Larkins, P. Melanen, P. Uusimaa, and M. Pessa, “10 Gb/s uncooled dilute-nitride optical transmitters operating at 1.3 μm,” in Proceedings of the Optical Fiber Communication Conference (IEEE, 2009), paper OWJ7. 10. M. Kondow, T. Kitatani, S. Nakatsuka, M. C. Larson, K. Nakahara, Y. Yazawa, M. Okai, and K. Uomi, “GaInNAs: a novel material for long-wavelength semiconductor lasers,” IEEE J. Sel. Top. Quantum Electron. 3, 719–730 (2007). 11. J. Pozo, N. Vogiatzis, O. Ansell, P. J. Heard, J. M. Rorison, P. Tuomisto, J. Konttinen, M. Saarinen, C. Peng, J. Viheriälä, T. Leinonen, and M. Pessa, “Fabrication and characterization of GaInNAs/GaAs semiconductor optical amplifiers,” Proc. SPIE 6997, 69970C (2008). 12. T. Piwonski, J. Pulka, G. Madden, G. Huyet, J. Houlihan, J. Pozo, N. Vogiatzis, P. Ivanov, J. M. Rorison, P. J. Barrios, and A. J. Gupta, “Ultrafast gain and refractive index dynamics in GaInNAsSb semiconductor optical amplifiers,” J. Appl. Phys. 106, 083104 (2009). 13. J. Hashimoto, K. Koyama, T. Katsuyama, Y. Iguchi, T. Yamada, S. Takagishi, M. Ito, and A. Ishida, “1.3 μm travelling-wave GaInNAs semiconductor optical amplifier,” in Proceedings of Optical Amplifiers and Their Applications (OSA, 2003), paper WB3.
52
APPLIED OPTICS / Vol. 54, No. 1 / 1 January 2015
14. D. Fitsios, G. Giannoulis, N. Iliadis, V.-M. Korpijärvi, J. Viheriälä, A. Laakso, S. Dris, M. Spyropoulou, H. Avramopoulos, G. T. Kanellos, N. Pleros, and M. Guina, “High gain 1.3-μm GaInNAs SOA with fast-gain dynamics and enhanced temperature stability,” Proc. SPIE 8982, 898208 (2014). 15. MATLAB, and Statistics Toolbox Release 2012b, The MathWorks, Inc., Natick, Massachusetts, United States. 16. S. Tanaka, N. Hatori, A. Uetake, S. Okumura, M. Ekawa, G. Nakagawa, and K. Morito, “Compact, very-low-electric-powerconsumption (0.84 W) 1.3 μm optical amplifier module using AlGaInAs MQW-SOA,” in Proceedings of the European Conference on Optical Communications, Torino, Italy (2010), paper Th10D3. 17. M. Rasztovits-Wiech, M. Danner, and W. R. Leeb, “Optical signal-to-noise ratio measurement in WDM networks using polarization extinction,” in Proceedings of the European Conference of Optical Communications (ECOC), Madrid, Spain, 20–24 September 1998, pp. 549–550. 18. J. Jin, D. Tian, J. Shi, and T. Li, “Fabrication and complete characterization of polarization insensitive 1310 nm InGaAsP–InP quantum-well semiconductor optical amplifiers,” Semicond. Sci. Technol. 19, 120–126 (2004). 19. http://www.kamelian.com/data/oa_13_ds.pdf. 20. http://03799fb.netsolhost.com/img/pdf/O‑Amp_8.5x11_p.pdf. 21. H. Hu, H. Ji, M. Galili, M. Pu, H. C. H. Mulvad, K. Yvind, J. M. Hvam, P. Jeppesen, and L. K. Oxenløwe, “All-optical wavelength conversion of a high-speed RZ-OOK signal in a silicon nanowire,” in Proceedings of the IEEE Photonics Conference (PHO), Arlington, Virginia, 9–13 October 2011. 22. K. E. Zoiros and P. Morel, “Enhanced performance of semiconductor optical amplifier at high direct modulation speed with birefringent fiber loop,” AIP Adv. 4, 077107 (2014). 23. L. Spiekman, “Semiconductor optical amplifiers in access networks,” in Proceedings of the OptoElectronics and Communications Conference, Hong Kong, China, 13–17 July 2009, paper TuH1. 24. P. Maniotis, D. Fitsios, G. T. Kanellos, and N. Pleros, “Optical buffering for chip multiprocessors: a 16 GHz optical cache memory architecture,” J. Lightwave Technol. 31, 4175–4191 (2013). 25. E. Kehayas, D. Tsiokos, P. Bakopoulos, D. Apostolopoulos, L. Stampoulidis, A. Poustie, R. McDougall, G. Maxwell, Y. Liu, S. Zhang, H. J. S. Dorren, J. Seoane, P. Van Holm-Nielsen, P. Jeppesen, and H. Avramopoulos, “40-Gb/s all-optical processing systems using hybrid photonic integration technology,” J. Lightwave Technol. 24, 4903–4911 (2006). 26. E. Kehayas, J. Seoane, Y. Liu, M. Martinez, J. Herrera, P. V. Holm-Nielsen, S. Zhang, R. McDougall, G. Maxwell, F. Ramos, J. Marti, H. J. S. Dorren, P. Jeppesen, and H. Avramopoulos, “All-optical network subsystems using integrated SOA-based optical gates and flip-flops for label-swapped networks,” IEEE Photon. Technol. Lett. 18, 1750–1752 (2006). 27. T. Nguyen-Cac, J. Bauwelinck, X. Yin, E. Tangiongga, and T. Koonen, “Demonstration of long-reach PON using 10 Gb/s 3R burst-mode wavelength converter,” IEEE Photon. Technol. Lett. 25, 1492–1495 (2013). 28. D. Fitsios, T. Alexoudi, G. T. Kanellos, K. Vyrsokinos, N. Pleros, T. Tekin, M. Cherchi, S. Ylinen, M. Harjanne, M. Kapulainen, and T. Aalto, “Dual SOA-MZI wavelength converters based on III-V hybrid integration on a μm-scale Si platform,” IEEE Photon. Technol. Lett. 26, 560–563 (2014).
Department of Informatics, Aristotle University of Thessaloniki, 54124 Thessaloniki, Greece
School of Electrical and Computer Engineering, National Technical University of Athens, Athens, Greece 3 4
Optoelectronics Research Centre (ORC), Tampere University of Technology, Tampere, Finland
Information Technologies Institute, Center for Research and Technology Hellas, Thessaloniki, Greece *Corresponding author: [email protected] Received 2 September 2014; revised 13 November 2014; accepted 14 November 2014; posted 17 November 2014 (Doc. ID 222131); published 23 December 2014
We report on the complete experimental evaluation of a GaInNAs/GaAs (dilute nitride) semiconductor optical amplifier that operates at 1.3 μm and exhibits 28 dB gain and a gain recovery time of 100 ps. Successful wavelength conversion operation is demonstrated using pseudorandom bit sequence 27 − 1 non-return-to-zero bit streams at 5 and 10 Gb∕s, yielding error-free performance and showing feasibility for implementation in various signal processing functionalities. The operational credentials of the device are analyzed in various operational regimes, while its nonlinear performance is examined in terms of four-wave mixing. Moreover, characterization results reveal enhanced temperature stability with almost no gain variation around the 1320 nm region for a temperature range from 20°C to 50°C. The operational characteristics of the device, along with the cost and energy benefits of dilute nitride technology, make it very attractive for application in optical access networks and dense photonic integrated circuits. © 2014 Optical Society of America OCIS codes: (250.5980) Semiconductor optical amplifiers; (130.4815) Optical switching devices; (160.6000) Semiconductor materials. http://dx.doi.org/10.1364/AO.54.000046
1. Introduction
Semiconductor optical amplifiers (SOAs) have already been established as key elements in telecom and datacom environments, due to their small footprint, fast gain recovery, broadband gain, and ability for integration [1–6]. SOAs offer a wide range of optical functionalities, with some indicative examples including optical amplification with SOAs as standalone modules [1] or integrated with electro-absorption 1559-128X/15/010046-07$15.00/0 © 2015 Optical Society of America 46
APPLIED OPTICS / Vol. 54, No. 1 / 1 January 2015
modulators [2], as well as colorless reflective modulation [3] using reflective semiconductor optical amplifiers (RSOAs). Moreover, SOAs have been shown to enable advanced functionalities such as optical random access memory (RAM) [4]. As the number of elements on-chip increases [5], operational temperatures are expected to significantly increase, affecting the circuit’s operation and requiring power-hungry external cooling. The majority of SOA devices reported so far have relied on the rather expensive InP for the active material platform [6]. On the other hand, GaInNAs/ GaAs (dilute nitride) technology has emerged as an
attractive alternative for conventional InP-based materials, due to its significantly decreased cost combined with competitive performance characteristics [7]. Moreover, an additional advantage of dilute nitrides is the large conduction band offset between the GaInNAs quantum well and GaAs barrier, which is manifested as less temperature-sensitive operation [8]. Excellent performance and temperature characteristics have already been demonstrated by a number of GaInNAs-based laser devices [9,10]; however, only a few dilute nitride edge-emitting SOAs have been presented so far. Pozo et al. [11] demonstrated a 1300 nm dilute nitride SOA capable of multiwavelength amplification. Studies by Piwonski et al. [12] on ultrafast gain dynamics proved that SOAs operating near 1550 nm can have a gain recovery time of 370 ps. Additionally, the 1.3 μm travelling-wave SOA presented by Hashimoto et al. in [13] had a peak chip gain of 14 dB, while exhibiting much smaller temperature dependence than conventional InP SOAs. Nevertheless, only recently, we demonstrated a dilute nitride SOA operating at 1.3 μm, providing simultaneously both high gain and low gain-recovery time values, while offering enhanced temperature stability [14]. The device exhibited a gain value of 28 dB and gain recovery time of around 100 ps, while it showed almost no gain variation around the 1320 nm region for a temperature range from 20°C to 50°C, raising the potential for uncooled operation. In this paper, we provide a thorough analysis on our SOA device reported in [14], offering further experimental verification and detailed insight into its performance credentials. Successful wavelength conversion operation at 5 and 10 Gb∕s is reported, while in each case the respective Q-factor values are measured, indicating error-free operation. The gain recovery time of the device is measured in various operational regimes, supported by statistical distribution values. Finally, the nonlinear performance of the device is assessed by exploring fourwave-mixing (FWM) phenomena. Taking into account the enhanced temperature stability and the cost advantages of GaAs-based materials, we believe that the proposed dilute nitride SOA technology could potentially form a good candidate for nextgeneration optical access network and datacom environments. 2. Fabrication and Experimental Setup
The GaInNAs/GaAs semiconductor heterostructure can be seen in Fig. 1(a). It was grown on n-GaAs (100) substrate by plasma-assisted molecular beam epitaxy (PA-MBE). A radio frequency plasma source was used to produce active nitrogen for the growth. A valved cracker was used for arsenic and normal effusion cells for group-III elements and silicon and beryllium dopants. The structure consists of two 7-nm-thick GaIn0.31 N0.02 As quantum wells (QWs), which are surrounded by tensile-strained GaNAs for strain compensation. The thickness of GaNAs
Fig. 1. (a) SOA heterostructure and respective bandgap energy. (b), (c) SEM images of (b) facet of the SOA and (c) surface of the SOA.
is 10 nm between the QWs as well as on the outer sides of both QWs. This QW group is embedded in a waveguide region of 103 nm of GaAs on both sides of the group. The intrinsic waveguide region is further clad by Al0.60 GaAs layers, which are n-doped below the QWs and p-doped above. On top of the p-AlGaAs cladding, a 200 nm p-GaAs contact layer finalizes the structure. Following, the material was processed into a ridge waveguide by patterning a 200 nm thick SiN etch mask with UV lithography and etching the SOA structure with inductively coupled plasma reactive ion etching (ICP-RIE), due to its ability to provide etch uniformity across the semiconductor surface. A scanning electron microscope (SEM) image of the waveguide cross section can be seen in Fig. 1(b). The 2 μm width and 1.7 μm height of the ridge were designed to optimize the optical confinement structure of the device. Ridge height was controlled using in situ reflectance measurement. Consequently, the etch mask was removed and a new round of photolithography and dry etching was performed to define the current insulation layer. Ti/Pt/Au ohmic metal contact layers were deposited on the front surface of the sample. After mechanical lapper thinning and polishing, Ni/Au/Ge/Au ohmic metal contact layers were deposited on the back surface and annealed. The device was then cleaved to form a 1 mm long cavity. As can be seen in the top view SEM image of the device in Fig. 1(c), facets on the ends of the waveguide were cleaved on a 7° angle to reduce reflectivity and allow for single-pass amplification. Both facets were then antireflection (AR) coated using a SiO2 ∕TiO2 film pair to minimize reflections, further reducing reflectivity below 0.1%. The SOA was subsequently bonded p side up on C mounts for experimental evaluation. A list of the main SOA parameters can be found in Table 1. A fiber alignment probe station was employed for the SOA chip experimental evaluation. Light was coupled in and out of the 2 μm ridge width SOA 1 January 2015 / Vol. 54, No. 1 / APPLIED OPTICS
47
Table 1.
Group refractive index Total confinement factor for the two QWs in vertical direction Effective refractive index Current density at transparency Carrier density at transparency Internal losses Internal quantum efficiency
SOA Parameters
3.62 4.3%
3.244 739 A∕cm2 1.3718 cm−3 at T 300 K (26.85°C) 1.5318 cm−3 at T 325 K (51.85°C) 1.6718 cm−3 at T 350 K (76.85°C) 31 cm−1 0.6
waveguide by using lensed fibers with a spot size diameter of 2.5 μm in order to ensure minimum coupling losses of around 5 dB per facet. The complete experimental setup can be seen in Fig. 2. For the static gain measurements a CW signal (probe signal) at 1309.95 nm (λ2) was fed into the SOA after passing through a variable optical attenuator (VOA) utilized for controlling the input signal power level. A polarization controller (PC) was used for adjusting the signal polarization state to compensate for the SOA polarization-dependent gain (PDG). Moreover, isolators were employed at the input and output of the chip to eliminate backreflections of light at the fiber/air and waveguide/air interfaces. The output signal was then inserted in a 4 nm Bragg grating (BG) band-pass filter to reject ASE noise, and its power was measured with a powermeter. Wavelength conversion operation was achieved by means of pump–probe cross-gain modulation (XGM), where a CW signal (pump signal) at 1314.01 nm (λ1) was launched into a Ti:LiNbO3 modulator (MOD) driven by a programmable pattern generator (PPG) so as to produce a 27 − 1 pseudo-random bit sequence (PRBS)
Fig. 2. (a) Experimental setup. (b) Picture of SOA chip on fiber alignment probe station. 48
APPLIED OPTICS / Vol. 54, No. 1 / 1 January 2015
of non-return-to-zero (NRZ) data pulses at 5 or 10 Gb∕s. For the dynamic gain performance measurements of the SOA, the pump signal was reprogrammed to include at least 18 consecutive “0”s between successive logical “1” pulse streams so as to allow sufficient time for full gain recovery. The wavelength converted signal at λ2 was then inserted into the 4 nm band-pass filter and analyzed at the oscilloscope. The respective statistical values were calculated using the MATLAB statistics toolbox [15]. Finally, the temperature dependence of the SOA performance was evaluated by varying the operating temperature of the chip using a Peltier element that was placed beneath the chip mount. 3. Experimental Results
Figure 3 illustrates the ASE spectra of the SOA for various injected currents, with the temperature set at 21.5°C. As can be seen, the emission peak experiences a blue shift as the injected current increases, reaching a maximum shift of 5.4 nm for increasing the injected current from 200 to 400 mA. For an injection current value of 400 mA, the −3 dB spectral bandwidth reaches a value of almost 30 nm. Figure 4 shows the chip gain as a function of input power with the temperature set at 20°C. As can be seen, a peak gain value of 28 dB is achieved, which is to our knowledge the highest ever reported for dilute nitride SOAs. Moreover, the saturated output power as a function of the signal input power can be also seen in Fig. 4. The maximum output power was measured at around 11 dBm for an input signal of −10 dBm. It should be noted that fiber-to-facet and facet-to-fiber coupling losses of 5 dB each have been taken into account for calculating the above values. The PDG of the device was 9.7 dB, measured at the small-signal gain region. However, this value was not optimized for the device in this initial fabrication run and will be a target for investigation in future fabrication runs, where we will examine the applicability of techniques widely used in SOAs so far [16]. The noise figure (NF) of the device was calculated using the well-known formula NF OSNRin∕OSNRout, while the OSNR was estimated using the polarization nulling method,
Fig. 3. SOA ASE spectra for various injected currents.
Fig. 4. SOA gain curve as a function of the input power and SOA output power as a function of the input power. Temperature set at 20°C.
reported in [17]. For these measurements, a signal at 1309.454 nm was modulated to produce a 10 Gb∕s 27 − 1 PRBS sequence and inserted into a commercially available linear SOA, fabricated by Kamelian, for amplification. Consequently, the signal was injected into the dilute nitride SOA under test. The NF was calculated for an input power of −15 dBm, corresponding to a gain value of 25 dB, close to the small signal gain region. The NF value was NF 7.5 dB. The respective signal spectra for the obtained NF measurements can be seen in Fig. 5, where Fig. 5(a) depicts the spectrum of the input signal after amplification through the linear SOA, while Fig. 5(b) illustrates the spectrum of the dilute nitride SOA output signal in the case of a small signal gain of 25 dB. The NF performance of our device is similar to that reported for a 1310 nm InGaAsP–InP quantum-well SOA with 25 dB gain, reported in [18], as well as with commercial quantum-well SOAs developed by Aeon and Kamelian [19,20]. Figure 6 depicts the wavelength converted signal at λ2 at the SOA output providing an evaluation of the gain recovery times for three different gain saturation regimes. Transition into different saturation regimes is achieved by adjusting the power of the probe signal. As can be observed, the measured 10%–90% gain recovery time reduces from 203 to 156 ps as we move into saturation; however, the SOA’s gain recovery is shortened in the deep saturation regime reaching a value of approximately 100 ps, a value that can allow for operation
Fig. 5. Optical spectrum of (a) input signal after amplification through the linear SOA and (b) GaInNAs SOA output signal.
Fig. 6. (a)–(c) SOA gain recovery dynamics for various gain saturation regimes and (d)–(f) respective statistical distributions.
in various signal processing functionalities at 10 Gb∕s. It should be noted that the low extinction ratio in Fig. 6(c) owes to the low differential gain in the deep saturation regime. The validity of the results was checked with a statistically adequate number of 104 measured samples for each case, the statistical distribution of which is shown in Figs. 6(d)–6(f), corresponding to the three gain saturation regimes shown in Figs. 6(a)–6(c), respectively. The statistical distribution in Fig. 6(d) has a mean value of 203 ps and a standard deviation of 25.8 ps, while the statistical distribution in Fig. 6(e) has a mean value of 156 ps and a standard deviation of 36.6 ps, and in Fig. 6(f) a mean value of 103 ps and a standard deviation of 33.3 ps. The effect of the injected current on the SOA gain recovery times was examined in different cases, specifically for 310, 345, and 390 mA, as can be seen in Figs. 7(a)–7(c), respectively. The mean values for the gain recovery time were 219.9, 215.2, and 203.2 ps, respectively, calculated through the respective statistical distributions in Figs. 7(d)–7(f). For a current of 310 mA, the standard deviation of the measured samples was 48.7 ps, while for currents of 345 and 390 mA the standard deviation values were 39.2 and 25.8 ps, respectively. Figures 8(a) and 8(b) illustrate the obtained eye diagrams of the wavelength converted signal carrying the complementary PRBS streams at 5 and 10 Gb∕s PRBS, respectively. The 100 ps gain recovery time of the device can be also confirmed by the 10 Gb∕s eye diagram of the wavelength converted signal displayed in Fig. 8(b). At both operational speeds, clear eye openings are obtained; however, a triangular pulse shape is observed at 1 January 2015 / Vol. 54, No. 1 / APPLIED OPTICS
49
Fig. 7. (a)–(c) SOA gain recovery dynamics for various injected currents and (d)–(f) respective statistical distributions.
10 Gb∕s as the device is near its operational speed limits due to the gain recovery time of 100 ps. Due to having the receiver in an AC-coupled configuration, it was not possible to extract any extinction ratio information. On the other hand, some information about the amplitude modulation of the eye diagram
can be given by the eye evaluation metrics, specifically by the standard deviation “σ 1 ” of level “1” of the eye diagrams. In particular, σ 1 had values of 0.118 at 5 Gb∕s and 0.168 at 10 Gb∕s, respectively, indicating a rather small divergence in the “1” level. A reason for the noise appearing in level “1” of the eye diagrams is the spectral bandwidth of the 4 nm BG band-pass filter that is employed at the SOA output to reject the ASE noise, due to the lack of any narrower filters at our disposal. Consequently, a significant portion of outband noise passes through the filter and is contributing to this noise level. Moreover, the wavelength converted signal at the output of the SOA appears with inversed polarity, and, hence, the noise at the level of ones originates from the unsaturated gain of the SOA in the presence of an input bit with logic ‘0’ value. Finally, for the case of 10 Gb∕s wavelength conversion, the gain recovery time of the SOA induces some amplitude modulation, as it is near its practical speed limit. Bit error rate (BER) measurements were performed for 5 and 10 Gb∕s using offline processing in MATLAB. The electrical signals from the photoreceiver were digitized using a 80 GSa∕s, 33 GHz realtime oscilloscope. Digital signal processing (DSP) was used for clock recovery and finding the optimum sampling point, after which errors could be directly counted and the BER calculated. No further processing was applied (i.e., no equalization or any other DSP that could improve signal quality). The measured BERs, as well as the 95% confidence interval (CI), are shown in Table 2. Note that the number of bits that could be processed was limited by the memory depth of the oscilloscope (5 · 105 for 5 Gb∕s and 1 · 106 for 10 Gb∕s). The BER was measured at 9.1 · 10−5 for the 10 Gb∕s case, while no errors were observed for 5 Gb∕s, indicating that the BER is better than 7.4 · 10−6 (the upper limit of the CI). For both cases, the measured BER values were found to be below 10−4, well below the standard forward error correction (FEC, 7% overhead) of 2 · 10−3 [21]. These results were taken with the probe and pump signals reaching the SOA with powers of −3 and 3.6 dBm, respectively. Since the SOA device was designed for DC operation, its electrical bandwidth is only 0.5 GHz. However, targeting on high-speed direct modulation schemes using the 1.3 μm GaInNAs SOA, alternative integration concepts [9] and reported techniques [22] could be followed to essentially improve the EO bandwidth of the SOA. In order to examine the device’s nonlinear performance in terms of FWM, we injected two copropagated CW signals at 1310.4 and at 1310.8 nm, respectively. As can be seen in Fig. 9, the measured downconversion FWM efficiency was Table 2.
Wavelength Conversion Speed Fig. 8. Obtained wavelength converted signal eye diagrams for PRBS stream at (a) 5 Gb∕s and (b) 10 Gb∕s. 50
APPLIED OPTICS / Vol. 54, No. 1 / 1 January 2015
5 Gb∕s 10 Gb∕s
Bit Error Rate Measurements
Q-Factor Measured (dB) BER 11.73 9.93
0 9.1 · 10−5
95% CI (Lower Limit)
95% CI (Upper Limit)
0 7.3 · 10−5
7.4 · 10−6 1.1 · 10−4
Fig. 9. Spectrum for four-wave-mixing operation.
around −36 dB, showing that the device holds potential for applications requiring nonlinear operation. As dilute nitride materials exhibit low temperature sensitivity credentials, an examination of the device’s operation in different temperatures was performed. Figure 10(a) depicts the ASE spectra of the SOA for different operating temperatures up to 50°C, with the injection current set at 350 mA. The 3 dB bandwidth of the SOA’s ASE profile is 53.8 nm and varies only within 0.6 nm for the whole temperature range. Moreover, a spectral redshift of around 18 nm is observed by increasing the operating temperature from
Fig. 10. (a) ASE spectra of the SOA for different operating temperatures. (b) Measured ASE power at the SOA facet versus injected current for various temperatures.
20°C to 50°C. Nevertheless, the ASE peak amplitude at around 1290 nm varies within only 3.9 dB for a temperature shift from 20°C to 50°C, while the ASE amplitude level around the 1320 nm region remains almost constant within the entire temperature range, revealing excellent thermal stability. It is worth noting that the temperature dependence of the ASE output characteristics of the device is significantly smaller compared to that of conventional InP-based SOAs with similar cavity length reported in the literature [13]. Figure 10(b) shows the measured ASE power at the SOA facet versus the injected current for various temperatures. As can be seen, a value of around 0 dBm was obtained for a current of around 390 mA, with the temperature set at 20°C. This value saturates for currents above 390 mA. As the temperature of the SOA increases, the SOA ASE power starts to drop. However, this decrease in ASE power is limited to only 3.8 dB for a temperature increase to 50°C, once again confirming the good thermal stability of the device. 4. Conclusions
We have performed a thorough experimental evaluation of the performance of a SOA based on GaInNAs/GaAs and operating at 1.3 μm. The device exhibited 28 dB gain, an output power of 11 dBm, and a gain recovery time of approximately 100 ps, record values for SOAs based on this material system. These values are more than adequate for operation as an amplification medium in metro and access networks [23], while making this a candidate material system for several on-chip functionalities in photonic integrated circuits (PICs). Moreover, wavelength conversion operation was performed successfully at both 5 and 10 Gb∕s, confirming the device’s credentials as both an amplifier and a medium for all-optical signal processing functionalities. The list of applications that could be enabled also includes the optical cache scheme reported in [24], where SOA-based switches are employed to constitute optical memory schemes, as well access enabling elements. The bit-level buffering solutions that can be offered by SOAs [4] can enable low-latency and line-rate performing optical routing solutions, while a whole list of optical processing applications [25,26] can be utilized in network environments [27]. As GaInNAs/GaAs technology has the advantages of significantly decreased cost and enhanced temperature stability, the proposed SOA device could be potentially utilized in optical access network applications, where operation without the need for cooling is highly beneficial energy- and cost-wise. Moreover, the device’s enhanced insensitivity to increased temperature can facilitate the exploitation of stateof-the-art hybrid integration of multiple active elements on a silicon-on-insulator (SOI) platform [28], allowing for higher integration densities and for more advanced on-chip functionalities. The work of D. Fitsios was supported by the IKY Foundation through the SIEMENS Fellowship of 1 January 2015 / Vol. 54, No. 1 / APPLIED OPTICS
51
Excellence Program for Post-Graduate Studies in Greece. This work has also been supported in part by the European Commission through the FP7-ICTFET Open project RAMPLAS (Contract No. 270773). References 1. H. Wang, J. Wu, and J. Lin, “Spectral characteristics of optical pulse amplification in SOA under assist light injection,” J. Lightwave Technol. 23, 2761–2771 (2005). 2. K. Ławniczuk, O. Patard, R. Guillamet, N. Chimot, A. Garreau, C. Kazmierski, G. Aubin, and K. Merghem, “40Gb/s colorless reflective amplified modulator,” IEEE Photon. Technol. Lett. 25, 341–343 (2013). 3. Z. Al-Qazwini, M. Thollabandi, and H. Kim, “Colorless optical transmitter for upstream WDM PON based on wavelength conversion,” J. Lightwave Technol. 31, 896–902 (2013). 4. D. Fitsios, C. Vagionas, G. T. Kanellos, A. Miliou, and N. Pleros, “Dual-wavelength bit input optical RAM with three SOA-XGM switches,” IEEE Photon. Technol. Lett. 24, 1142–1144 (2012). 5. S. C. Nicholes, M. L. Masanovic, B. Jevremovic, E. Lively, L. A. Coldren, and D. J. Blumenthal, “An 8 × 8 InP monolithic tunable optical router (MOTOR) packet forwarding chip,” J. Lightwave Technol. 28, 641–650 (2010). 6. M. Cabezon, A. Villafranca, D. Izquierdo, J. J. Martinez, and I. Garces, “Integrated all-optical 4-input NOR logic gate based on InP technology,” in Conference on Lasers and Electro-Optics (CLEO), San Jose, California, 6–11 May 2012, paper JW4A.7. 7. X. Yang, M. J. Jurkovic, J. B. Heroux, and W. I. Wang, “Molecular beam epitaxial growth of InGaAsN:Sb/GaAs quantum wells for long-wavelength semiconductor lasers,” Appl. Phys. Lett. 75, 178–180 (1999). 8. M. Kondow, S. Nakatsuka, T. Kitatani, Y. Yazawa, and M. Okai, “Room-temperature pulsed operation of GaInNAs laser diodes with excellent high-temperature performance,” Jpn. J. Appl. Phys. 35, 5711–5713 (1996). 9. M. Dumitrescu, M. Wolf, K. Schulz, Y. Q. Wei, G. Adolfsson, J. Gustavsson, J. Bengtsson, M. Sadeghi, S. Wang, A. Larsson, J. Lim, E. Larkins, P. Melanen, P. Uusimaa, and M. Pessa, “10 Gb/s uncooled dilute-nitride optical transmitters operating at 1.3 μm,” in Proceedings of the Optical Fiber Communication Conference (IEEE, 2009), paper OWJ7. 10. M. Kondow, T. Kitatani, S. Nakatsuka, M. C. Larson, K. Nakahara, Y. Yazawa, M. Okai, and K. Uomi, “GaInNAs: a novel material for long-wavelength semiconductor lasers,” IEEE J. Sel. Top. Quantum Electron. 3, 719–730 (2007). 11. J. Pozo, N. Vogiatzis, O. Ansell, P. J. Heard, J. M. Rorison, P. Tuomisto, J. Konttinen, M. Saarinen, C. Peng, J. Viheriälä, T. Leinonen, and M. Pessa, “Fabrication and characterization of GaInNAs/GaAs semiconductor optical amplifiers,” Proc. SPIE 6997, 69970C (2008). 12. T. Piwonski, J. Pulka, G. Madden, G. Huyet, J. Houlihan, J. Pozo, N. Vogiatzis, P. Ivanov, J. M. Rorison, P. J. Barrios, and A. J. Gupta, “Ultrafast gain and refractive index dynamics in GaInNAsSb semiconductor optical amplifiers,” J. Appl. Phys. 106, 083104 (2009). 13. J. Hashimoto, K. Koyama, T. Katsuyama, Y. Iguchi, T. Yamada, S. Takagishi, M. Ito, and A. Ishida, “1.3 μm travelling-wave GaInNAs semiconductor optical amplifier,” in Proceedings of Optical Amplifiers and Their Applications (OSA, 2003), paper WB3.
52
APPLIED OPTICS / Vol. 54, No. 1 / 1 January 2015
14. D. Fitsios, G. Giannoulis, N. Iliadis, V.-M. Korpijärvi, J. Viheriälä, A. Laakso, S. Dris, M. Spyropoulou, H. Avramopoulos, G. T. Kanellos, N. Pleros, and M. Guina, “High gain 1.3-μm GaInNAs SOA with fast-gain dynamics and enhanced temperature stability,” Proc. SPIE 8982, 898208 (2014). 15. MATLAB, and Statistics Toolbox Release 2012b, The MathWorks, Inc., Natick, Massachusetts, United States. 16. S. Tanaka, N. Hatori, A. Uetake, S. Okumura, M. Ekawa, G. Nakagawa, and K. Morito, “Compact, very-low-electric-powerconsumption (0.84 W) 1.3 μm optical amplifier module using AlGaInAs MQW-SOA,” in Proceedings of the European Conference on Optical Communications, Torino, Italy (2010), paper Th10D3. 17. M. Rasztovits-Wiech, M. Danner, and W. R. Leeb, “Optical signal-to-noise ratio measurement in WDM networks using polarization extinction,” in Proceedings of the European Conference of Optical Communications (ECOC), Madrid, Spain, 20–24 September 1998, pp. 549–550. 18. J. Jin, D. Tian, J. Shi, and T. Li, “Fabrication and complete characterization of polarization insensitive 1310 nm InGaAsP–InP quantum-well semiconductor optical amplifiers,” Semicond. Sci. Technol. 19, 120–126 (2004). 19. http://www.kamelian.com/data/oa_13_ds.pdf. 20. http://03799fb.netsolhost.com/img/pdf/O‑Amp_8.5x11_p.pdf. 21. H. Hu, H. Ji, M. Galili, M. Pu, H. C. H. Mulvad, K. Yvind, J. M. Hvam, P. Jeppesen, and L. K. Oxenløwe, “All-optical wavelength conversion of a high-speed RZ-OOK signal in a silicon nanowire,” in Proceedings of the IEEE Photonics Conference (PHO), Arlington, Virginia, 9–13 October 2011. 22. K. E. Zoiros and P. Morel, “Enhanced performance of semiconductor optical amplifier at high direct modulation speed with birefringent fiber loop,” AIP Adv. 4, 077107 (2014). 23. L. Spiekman, “Semiconductor optical amplifiers in access networks,” in Proceedings of the OptoElectronics and Communications Conference, Hong Kong, China, 13–17 July 2009, paper TuH1. 24. P. Maniotis, D. Fitsios, G. T. Kanellos, and N. Pleros, “Optical buffering for chip multiprocessors: a 16 GHz optical cache memory architecture,” J. Lightwave Technol. 31, 4175–4191 (2013). 25. E. Kehayas, D. Tsiokos, P. Bakopoulos, D. Apostolopoulos, L. Stampoulidis, A. Poustie, R. McDougall, G. Maxwell, Y. Liu, S. Zhang, H. J. S. Dorren, J. Seoane, P. Van Holm-Nielsen, P. Jeppesen, and H. Avramopoulos, “40-Gb/s all-optical processing systems using hybrid photonic integration technology,” J. Lightwave Technol. 24, 4903–4911 (2006). 26. E. Kehayas, J. Seoane, Y. Liu, M. Martinez, J. Herrera, P. V. Holm-Nielsen, S. Zhang, R. McDougall, G. Maxwell, F. Ramos, J. Marti, H. J. S. Dorren, P. Jeppesen, and H. Avramopoulos, “All-optical network subsystems using integrated SOA-based optical gates and flip-flops for label-swapped networks,” IEEE Photon. Technol. Lett. 18, 1750–1752 (2006). 27. T. Nguyen-Cac, J. Bauwelinck, X. Yin, E. Tangiongga, and T. Koonen, “Demonstration of long-reach PON using 10 Gb/s 3R burst-mode wavelength converter,” IEEE Photon. Technol. Lett. 25, 1492–1495 (2013). 28. D. Fitsios, T. Alexoudi, G. T. Kanellos, K. Vyrsokinos, N. Pleros, T. Tekin, M. Cherchi, S. Ylinen, M. Harjanne, M. Kapulainen, and T. Aalto, “Dual SOA-MZI wavelength converters based on III-V hybrid integration on a μm-scale Si platform,” IEEE Photon. Technol. Lett. 26, 560–563 (2014).
