We report a novel hybrid integrated optic device consisting of AlGaInAs/InP electroabsorption modulators and a four-arm silica-on-silicon planar lightwave circuit optical interferometer. The device is designed for generation of high spectral efficiency optical modulation formats. We demonstrate generation of 21.4 Gb/s quadrature phase shift keyed optical signals with electrical data drives of 2Vpp amplitudes, achieving a bit error rate of 10-9 with the required optical signal to noise ratio of ~18 dB in a 0.1 nm resolution bandwidth.
©2008 Optical Society of America
High spectral efficiency modulation formats have recently attracted a great deal of attention as complementary means to the traditional optical signal bandwidth multiplication in achieving data rate increase in optical communication. These modulation formats, including Quadrature Phase Shift Keying (QPSK) and various Quadrature-Amplitude Modulation (QAM) formats, are capable of encoding multiple bits of information per symbol. Hence, these modulation formats are advantageous as data rate increase can be accomplished with minimal changes to the installed base of optical filters and chromatic and polarization mode dispersion mitigation devices. Differentially-encoded QPSK (DQPSK) transmission with direct detection is especially promising due to the relative ease of its implementation and several groups have demonstrated the potentials of DQPSK transmission [1–5], including its compatibility with tight optical filtering . In addition, the recent resurgence of coherent communication techniques assisted by digital signal processing is likely to enable adoption of even more spectrally efficient formats .
In most cases, QPSK and QAM signals are generated using an array of phase modulators arranged in an optical interferometer, such as a nested Mach-Zehnder interferometer (MZI) [7–9]. For example, lithium niobate modulators were used in [1-5, 7]and GaAs phase modulators were used in . Relatively large drive-voltage requirements, which tend to get more severe with increasing baud rates, and large physical dimensions, are some of the drawbacks with the devices based on phase modulators. Traveling wave geometries are typically used for achieving high modulation bandwidth, and the matching of the optical and RF propagation speeds becomes challenging for high modulation rates. As an alternative to the phase modulators, especially at high modulation rates, we first proposed generation of high spectral efficiency modulation formats using lumped-element electroabsorption modulators (EAMs) in combination with interferometric effects, and demonstrated generation of 40Gb/s PSK signals [10, 11]. Semiconductor EAMs offer the advantages of low drive voltages, small form factors, and the integrability with semiconductor lasers and amplifiers. More recently, the concept was realized for generation of QPSK signals using monolithically integrated InP EAMs and a three-arm interferometer but with performance limitations [12, 13].
In this paper, we report fabrication of a hybrid integrated device consisting of III-V EAMs and a four-arm interferometer made of silica on silicon planar lightwave circuits (PLC) for generation of spectrally-efficient modulation formats. More specifically, we demonstrate generation of 21.4 Gb/s (10.7 Gbaud/s) QPSK signals with excellent performance comparable to the best achieved to date with QPSK modulators based on phase-only modulation.
2. Device description
Hybrid integration offers an advantage of the flexibility to choose and optimize multiple device platforms for a given functionality. We use an array of four AlGaInAs/InP EAMs as low drive voltage data modulation elements, shown in Fig. 1, which also integrates an array of semiconductor optical amplifiers (SOAs) to compensate for optical coupling losses. For implementing an optical interferometer requisite for generating the advanced modulation formats, we choose a low-loss silica PLC containing a 1×4 MMI (Fig. 2). The two components are butt-coupled as illustrated in Fig. 3, where the incoming CW light is first modulated by the EAMs and reflected at the high-reflector (HR) side of the EAM device and modulated again by the same set of data drives before it exits the device via a circulator. The temporal delay between the two instances of the modulations is less than 1 ps and its impact is negligible for ~ 30 GHz bandwidth of the EAMs. Due to this double modulation action, the reflective geometry should allow operation with smaller drive voltages than the transmissive configuration can, or alternatively operation with better extinction for the same drive voltages. The delivery of electrical drive signals is also simpler since there are no optical ingress/egress ports to contend with in delivering the high-speed electrical data to the EAMs. We note that the use of a circulator is not necessary and can be avoided by using a 2×4 MMI structure, for example.
We first provide the details of the active elements consisting of EAMs and SOAs. The device structure is composed 10 tensile strain GaInAs quantum wells and 11 compressive-strain AlGaInAs barriers. AlGaInAs quantum well material is used as it provides an advantageous trade-off between the extinction ratio and bandwidth due to its high electronic confinement. The EAM-SOA devices incorporate a single active layer manufactured by Selective Area Growth (SAG) technology. The SAG technology is used to engineer the bandgaps of the photonic components. Namely, the SOA gap is shifted +35 nm from the EAM absorption edge in order to provide efficient amplification in the modulator working spectral range. In order to obtain a low capacitance EAM, iron-doped InP is used to bury the 2-µm height active stripe. However, the AlGaInAs material is difficult to bury due to a possible contamination of aluminum-containing layers. To this aim, we apply the previously reported tertiarybutylchloride (TBCl) assisted MOVPE regrowth technique . Fe-doped InP burying layers capped by an undoped blocking layer are grown by LP-MOVPE. After removing the mask, p-type upper confinement and contact layers are regrown by MOVPE. A cross section of the structure is shown in Fig. 4.
The final component is a four element array with 70-µm length EAMs and 400-µm SOAs. A high reflection mirror using three-layer SiO2/TiO2 stacks is put on the EAM facet placed on the back side. Special care is paid to the antireflection coating at the output facet, consisting of two layers of SiO2/TiO2 stacks, in order to improve coupling efficiency with the silica waveguide as well as to reduce intracavity Fabry-Perot reflections which may lead to a spurious optical injection locking.
We use 4% high-index-contrast germanium-doped silica-on-silicon waveguides for the passive component, a detailed explanation of which is found in . Here, the index contrast is defined as (n c 2-n cl 2)/2n c 2, where nc and ncl are the core and cladding index, respectively. The high index contrast allows a tight bend radius enabling a smaller device, and the smaller mode size better matches the mode size of the EAM/SOA device. Single-mode waveguides with dimensions of 2.2 µm×2.7 µm are used in the device. Lower and upper claddings with thicknesses of 15 µm and 6 µm, respectively, are deposited around the waveguides for optical isolation. For a compact design, a minimum bend radius of 350 µm is used. Overall device size is 1.4 cm×2.3 mm, and the device length could be reduced further if desired. We use a 1×4 multi-mode interference (MMI) coupler with the nominal splitting ratio of 6 dB with the maximum imbalance less than 0.3 dB. Each arm of the splitter contains a thermo-optic phase shifter and a variable optical attenuator (VOA) consisting of a symmetric MZI with thermo-optic phase shifters in each arm. The VOAs, capable of maximum extinction of 26 dB per single pass, can be used to adjust the optical splitting ratio in a wide range. The variation of the splitting ratio of the 1×4 MMI requires only minor adjustments of the VOAs for power equalization among the four paths, which can be accomplished by measuring and equalizing the reflected output powers in one arm at a time. We use a segmented taper for the facet buttcoupling to a standard single mode fiber (SSMF) array, with the segments blazed at 7 degrees to minimize the reflection. The coupling loss at this facet is ~1 dB. The facet coupling to the active components uses an adiabatic mode squeezer with the waveguide width tapering down to 1.0 µm, with the resulting beam waist of 4 µm. The length of the adiabatic taper is 450 µm.
The passive and active components are actively aligned with respect to each other by maximizing the ASE power of the SOAs coupled to the passive component. An index matching epoxy is placed in the gap between the active and passive components. The minimum coupling loss while the alignment is being actively controlled is estimated to be 3–4 dB. At the point of the maximum coupling, the epoxy is UV cured and wire-bonding to the electrical contact pads proceeds. High speed RF data signals are delivered using an array of high-speed probes. We find that the optical alignment degrades slowly over time due to the movement of the components in curing of the epoxy. We prevent further degradation of the alignment by reinforcing the junction by bonding a thin glass straddling the passive and active components on top. The entire device is maintained at 21°C to stabilize the optical phase shifts controlled by the thermo-optical phase shifters. The EAM-SOA can operate at temperatures up to 60°C without cooling  but the phase shifters need to be readjusted for different temperatures. The device operates stably under the lab environment, whilst the device performances we report in the following sections are obtained after the coupling loss is degraded by ~10 dB from the maximum coupling. The reflection at the active-passive facet is estimated to be -30 dB. Even at such a low level of reflection, we operate SOAs with bias less than 25 mA to minimize the phase and amplitude noises associated with the spurious injection locking. Several modifications are under way to improve the alignment losses and stability. Optical spot-size converters are designed to increase the spot size of the EAM-SOA device to reduce the sensitivity to the misalignment. We have recently fabricated devices with an angled facet, reducing the reflectivity below -40 dB and achieved net fiber-to-fiber gain by running the SOAs with higher bias currents . Deployment of the devices to field applications may require a more robust packaging process, such as flip-chip bonding.
3. QPSK generation
The experimental set up to generate 21.4 Gb/s QPSK signals is schematically illustrated in Fig. 5. The data signals are directly derived from a programmable four-channel pattern generator (Anritsu MP1758C), generating two copies (D1 and D2) of 215-1, 10.7 Gb/s PRBS sequences and their complements. D1 and D2 are delayed with respect to each other to ensure the relative decorrelation between the two. QPSK signals are generated under the following drive condition: D1 and its complement modulate a pair of EAMs, while the thermo-optic phase shifters are tuned such that the silica waveguides connected to the EAM pair have π/2 phase difference per single pass, hence opposite phase after the reflection. A similar arrangement is made for the arms of the interferometer modulated by the data D2 and the complement. Finally, overall phase difference between the first pair and the second pair is set to be π/2.
The amplitudes of the drive signals are less than 2.0Vpp and the bias to the EAMs are set around -0.8 V. We use a CW laser delivering 2 dBm output powers to the hybrid device at 1561.2 nm. The bias currents to the SOAs are set around 23 mA. The output power of the QPSK signal is -19 dBm after accounting for the circulator loss. The raw insertion loss of the hybrid device is therefore 21 dB, which includes 1dB/facet coupling loss between the SSMF array and the silica PLC, and 12–13 dB/facet coupling loss between the silica PLC and the EAM device. The conversion loss of an ideal four-arm EAM QPSK device with no coupling loss would be 9 dB  and thus the total excess loss from all sources is 12 dB. With the aforementioned implementation of the spot-size converters and tilted output facet that are intended for mitigation of the coupling losses, we anticipate that QPSK signal generation with optical gains may be possible.
We plot in Fig. 6 an optical spectrum of the generated QPSK signal with virtually no sign of the CW component, indicating a good control of the interferometer phases. The maximum achieved OSNR is 28 dB as referred to a 0.1 nm resolution bandwidth. We show in Fig. 7 an optical eye diagram (left) and the associated constellation diagram (right) measured using linear optical sampling technique with a time resolution better than 1 ps . Four constellations signifying QPSK generation are clearly visible.
We further validate the QPSK signal quality by testing bit error rates using a DQPSK receiver. The optical QPSK signal is amplified by an EDFA with a noise figure of 5 dB and filtered using a 1-nm optical band pass filter. The filtered signal is further amplified using a second EDFA. At this point, the second EDFA is running in saturation and provides near constant optical power delivery to the photodetectors regardless of the received optical power as measured right after the VOA. The amplified QPSK signal is demodulated using a DQPSK demodulator implemented using a delay line interferometer filter with a matching 10.7 GHz free spectral range. We measure the demodulated signals using a balanced photo-detector with a bandwidth of 40 GHz after optically filtering the signal with a 0.25 nm optical band pass filter. Finally, the photo-detector signal is amplified using a trans-impedance amplifier with a bandwidth of 10 GHz. We program a bit error detector with the patterns calculated using the PRBS sequences used to generate the QPSK signals, according to the DQPSK encoding relations . We show in Fig. 8 an eye diagram of error-free signals from the balanced photodetector (left) and also plot BERs as a function of the required OSNR as referred in a resolution bandwidth of 0.1 nm and the raw received powers. We note that the measured performance are comparable to the best back-to-back performances achieved using phase modulators to date, after accounting for the symbol rate differences and the difference between RZ and NRZ; the OSNR requirement scales with the symbol rate and NRZ requires typically 1-3 dB more OSNR for the same BER than RZ formats [3, 4, 9, 19, 20]
We have demonstrated a novel compact hybrid integrated device for generation of advanced modulation formats. The overall size of the device is 1.5 cm×2.3 mm and the device requires less than 2Vpp drive voltages for operation. With the implementation of tilted facet designs, we expect the hybrid EAM/SOA device should be able to provide power gains for the generated signals. We validate the device by generating 21.4 Gb/s QPSK signals with excellent DQPSK BER performances. The device is very versatile with the capability of continuous tuning of the phase and amplitude of each of the four interferometer arms. Generation of more complex modulation formats including various QAM formats is under progress.
References and links
1. C. Wree, N. Hecker-Denschlag, E. Gottwald, P. Krummrich, J. Leibrich, E. D. Schmidt, B. Lankl, and W. Rosenkranz, “High spectral efficiency 1.6-b/s/Hz transmission (8×40 Gb/s with a 25-GHz grid) over 200-km SSMF using RZ-DQPSK and polarization multiplexing,” IEEE Photon. Technol. Lett. 15, 1303–1305 (2003). [CrossRef]
2. S. Chandrasekhar, et al., “Hybrid 107-Gb/s polarization-multiplexed DQPSK and 42.7-Gb/s DQPSK transmission at 1.4-bits/s/Hz spectral efficiency over 1280 km of SSMF and 4 bandwidth-managed ROADMs,” ECOC 2007, PD1.92 (2007).
3. S. Chandrasekhar and X. Liu, “Experimental investigation of system impairments in polarization multiplexed 107-Gb/s RZ-DQPSK,” OFC 2008, OThU7 (2008).
4. A. H. Gnauck, P. J. Winzer, C. Dorrer, and S. Chandrasekhar, “Linear and nonlinear performance of 42.7-Gb/s single-polarization RZ-DQPSK format,” IEEE Photon. Technol. Lett. 18, 883–885 (2006). [CrossRef]
5. M. Daikoku, I. Morita, H. Taga, H. Tanaka, T. Kawanishi, T. Sakamoto, T. Miyazaki, and T. Fujita, “100-Gb/s DQPSK transmission experiment without OTDM for 100G Ethernet transport,” J. Lightwave Technol. 25, 139–145 (2007). [CrossRef]
6. A. Leven, N. Kaneda, U.-V. Koc, and Y.-K. Chen, “Coherent receivers for practical optical communication systems,” OFC 2007, OThK4 (2007).
7. T. Kawanishi, T. Sakamoto, T. Miyazaki, M. Izutsu, T. Fujita, S. Mori, K. Higuma, and J. Ichikawa, “High-speed optical DQPSK and FSK modulation using integrated Mach-Zehnder interferometers,” Opt. Express 14, 4469–4478 (2006). [CrossRef] [PubMed]
8. H. Keang-Po and C. Han-Wei, “Generation of arbitrary quadrature signals using one dual-drive Modulator,” J. Lightwave Technol. 23, 764–770 (2005). [CrossRef]
9. R. A. Griffin, “Integrated DQPSK transmitters,” OFC 2005, OWE3 (2005).
10. I. Kang, “Interferometric operation of an electroabsorption modulator for PSK modulation and OOK modulation with performance enhancements,” ECOC 2006, We3.P.59 (2006).
12. C. R. Doerr, L. Zhang, P. J. Winzer, J. H. Sinsky, A. Adamiecki, N. Sauer, and G. Raybon, “Compact high-speed InP DQPSK modulator,” IEEE Photon. Technol. Lett. 19, 1184–1186 (2007). [CrossRef]
13. C. R. Doerr, L. Zhang, A. L. Adamiecki, N. J. Sauer, J. H. Sinsky, and P. J. Winzer, “Compact EAM-Based InP DQPSK Modulator and Demonstration at 80 Gb/s,” OFC 2007, PDP33 (2007).
14. S. Gouraud, M.-C. Cuisin, C. Kazmierski, J. Decobert, F. Alexandre, F. Elache, A. Paraskevopoulos, and D. Franke, “High speed AlGaInAs multiple-quantum- well electroabsorption modulator buried and planarized with semi-insulating Fe-doped InP grown by chloride assisted LP-MOVPE,” IPRM 2005 Glasgow TuA-1-2 (2005).
15. M. S. Rasras, I. Kang, M. Dinu, J. Jaques, N. Dutta, A. Piccirilli, M. A. Cappuzzo, E. Y. Chen, L. T. Gomez, A. Wong-Foy, S. Cabot, G. S. Johnson, L. Buhl, and S. S. Patel, “A Programmable 8-bit optical correlator filter for optical pattern recognition,” to be published in IEEE Photon. Technol. Lett. (2007).
16. N. Dupuis, et al., “10Gbit/s semi-insulating buried heterostructure loss-less reflective amplified modulators for wavelength agnostic networks,” OFC 2008, OThC2 (2008).
17. C. Dorrer, C. R. Doerr, I. Kang, R. Ryf, J. Leuthold, and P. J. Winzer, “Measurement of eye diagrams and constellation diagrams of optical sources using linear optics and waveguide technology,” J. Lightwave Technol. 23, 178–186 (2005). [CrossRef]
18. M. Serbay, C. Wree, and W. Rosenkranz, “Implementation of differential precoder for high-speed optical DQPSK transmission,” Electron. Lett. 40, 1288–1289 (2004). [CrossRef]
19. C. Wree, J. Leibrich, J. Eick, W. Rosenkranz, and D. Mohr, “Experimental investigation of receiver sensitivity of RZ-DQPSK modulation using balanced detection,” OFC2003, ThE5 (2003).
20. P. J. Winzer and R.-J. Essiambre, “Advanced modulation formats for high-capacity optical transport networks,” J. Lightwave Technol. 24, 4711–4728 (2006). [CrossRef]