Slow-light-based Variable Symbol-rate Silicon Photonics Dqpsk

Transcript

Slow-light-based variable symbol-rate silicon photonics DQPSK receiver Keijiro Suzuki,1 Hong C. Nguyen,1,2 Takemasa Tamanuki1, Fumihiro Shinobu,1 Yuji Saito,1,2 Yuya Sakai,1,2 and Toshihiko Baba1,2,* 1 Department of Electrical and Computer Engineering, Yokohama National University, 79-5 Tokiwadai, Hodogayaku, Yokohama, 240-8501, Japan 2 Core Research for Evolutional Science and Technology, Japan Science and Technology Agency, 5 Sanbancho, Chiyoda-ku, Tokyo, 102-0075, Japan * [email protected] Abstract: We report a silicon DQPSK receiver whose symbol rate can be varied by a tunable one-bit delay line including an all-pass micro-ring slowlight device. It also consists of Si-wire waveguides with spot-size converters, optimized splitters/couplers, heater-controlled Mach-Zehnder attenuators and phase shifters, 90° hybrid with a low-loss crossing and balanced Ge photodiodes, all of which are fabricated by using CMOScompatible process. Demodulation was confirmed at symbol-rates of 7.4 − 9.0 Gbaud, corresponding to bit-rates of 14.8 − 18.0 Gb/s. ©2012 Optical Society of America OCIS codes: (230.3120) Integrated optics devices; (250.5300) Photonic integrated circuits; (200.4650) Optical interconnects; (130.0250) Optoelectronics. References and links 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. T. Baba, “Slow light in photonic crystals,” Nat. Photonics 2(8), 465–473 (2008). J. Adachi, N. Ishikura, H. Sasaki, and T. Baba, “Wide range tuning of slow light pulse in SOI photonic crystal coupled waveguide via folded chirping,” IEEE J. Sel. Top. Quantum Electron. 16(1), 192–199 (2010). N. Ishikura, T. Baba, E. Kuramochi, and M. Notomi, “Large tunable fractional delay of slow light pulse and its application to fast optical correlator,” Opt. Express 19(24), 24102–24108 (2011). J. Yang, N. K. Fontaine, Z. Pan, A. O. Karalar, S. S. Djordjevic, C. Yang, W. Chen, S. Chu, B. E. Little, and S. J. B. Yoo, “Continuously tunable, wavelength-selective buffering in optical packet switching networks,” IEEE Photon. Technol. Lett. 20(12), 1030–1032 (2008). W. M. J. Green, H. F. Hamann, L. Sekaric, M. J. Rooks, and Y. A. Vlasov, “Ultra-compact reconfigurable silicon optical devices using micron-scale localized thermal heating,” in Tech. Dig. Opt. Fiber Commun. Conf., (IEEE Photonics Society, 2007), paper OtuM3. J. Cardenas, M. A. Foster, N. Sherwood-Droz, C. B. Poitras, H. L. R. Lira, B. Zhang, A. L. Gaeta, J. B. Khurgin, P. Morton, and M. Lipson, “Wide-bandwidth continuously tunable optical delay line using silicon microring resonators,” Opt. Express 18(25), 26525–26534 (2010). A. Melloni, F. Morichetti, C. Ferrari, and M. Martinelli, “Continuously tunable 1 byte delay in coupled-resonator optical waveguides,” Opt. Lett. 33(20), 2389–2391 (2008). A. Melloni, A. Canciamilla, C. Ferrari, F. Morichetti, L. O'Faolain, T. F. Krauss, R. De La Rue, A. Samarelli, and M. Sorel, “Tunable delay lines in silicon photonics: coupled resonators and photonic crystals, a comparison,” IEEE Photonics J. 2(2), 181–194 (2010). F. Shinobu, N. Ishikura, Y. Arita, T. Tamanuki, and T. Baba, “Continuously tunable slow-light device consisting of heater-controlled silicon microring array,” Opt. Express 19(14), 13557–13564 (2011). K. K. Lee, D. R. Lim, H.-C. Luan, A. Agarwal, J. Foresi, and L. C. Kimerling, “Effect of size and roughness on light transmission in a Si/SiO2 waveguide: Experiments and model,” Appl. Phys. Lett. 77(11), 1617 (2000). F. Ohno, T. Fukazawa, and T. Baba, “Mach-Zehnder interferometers composed of µ-bends and µ-branches in a Si photonic wire waveguide,” Jpn. J. Appl. Phys. 44(7A), 5322–5323 (2005). T. Fukazawa, F. Ohno, and T. Baba, “Very compact arrayed-waveguide-grating demultiplexer using Si photonic wire waveguides,” Jpn. J. Appl. Phys. 43(No. 5B), L673–L675 (2004). G. T. Reed, G. Mashanovich, F. Y. Gardes, and D. J. Thomson, “Silicon optical modulators,” Nat. Photonics 4(8), 518–526 (2010). J. Michel, J. Liu, and L. C. Kimerling, “High-performance Ge-on-Si photodetectors,” Nat. Photonics 4(8), 527– 534 (2010). S. Assefa, F. N. Xia, W. M. J. Green, C. L. Schow, A. V. Rylyakov, and Y. A. Vlasov, “CMOS-integrated optical receivers for on-chip interconnects,” IEEE J. Sel. Top. Quantum Electron. 16(5), 1376–1385 (2010). A. Alduino, L. Liao, R. Jones, M. Morse, B. Kim, W.-Z. Lo, J. Basak, B. Koch, H.-F. Liu, H. Rong, M. Sysak, C. Krause, R. Saba, D. Lazar, L. Horwitz, R. Bar, S. Litski, A. Liu, K. Sullivan, O. Dosunmu, N. Na, T. Yin, F. #161674 - $15.00 USD (C) 2012 OSA Received 18 Jan 2012; revised 6 Feb 2012; accepted 7 Feb 2012; published 10 Feb 2012 13 February 2012 / Vol. 20, No. 4 / OPTICS EXPRESS 4796 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. Haubensack, I. w. Hsieh, J. Heck, R. Beatty, H. Park, J. Bovington, S. Lee, H. Nguyen, H. Au, K. Nguyen, P. Merani, M. Hakami, and M. Paniccia, “Demonstration of a high speed 4-channel integrated silicon photonics WDM link with hybrid silicon lasers,” in Integrated Photonics Research, Silicon and Nanophotonics, OSA Technical Digest (CD) (Optical Society of America, 2010), paper PDIWI5. X. Zheng, J. Lexau, Y. Luo, H. Thacker, T. Pinguet, A. Mekis, G. Li, J. Shi, P. Amberg, N. Pinckney, K. Raj, R. Ho, J. E. Cunningham, and A. V. Krishnamoorthy, “Ultra-low-energy all-CMOS modulator integrated with driver,” Opt. Express 18(3), 3059–3070 (2010). R. Nagarajan, J. Rahn, M. Kato, J. Pleumeekers, D. Lambert, V. Lal, H.-S. Tsai, A. Nilsson, A. Dentai, M. Kuntz, R. Malendevich, J. Tang, J. Zhang, T. Butrie, M. Raburn, B. Little, W. Chen, G. Goldfarb, V. Dominic, B. Taylor, M. Reffle, F. Kish, and D. Welch, “10 Channel, 45.6 Gb/s per channel, polarization-multiplexed DQPSK, InP receiver photonic integrated circuit,” J. Lightwave Technol. 29(4), 386–395 (2011). C. R. Doerr, L. Zhang, and P. J. Winzer, “Monolithic InP multiwavelength coherent receiver using a chirped arrayed waveguide grating,” J. Lightwave Technol. 29(4), 536–541 (2011). C. R. Doerr and L. Chen, “Monolithic PDM-DQPSK receiver in silicon,” in European Conference and Exhibition on Optical Communication, (IEEE Photonics Society, 2010), paper PD3.6. K. Suzuki, H. C. Nguyen, T. Tamanuki, F. Shinobu, Y. Saito, Y. Sakai, and T. Baba, “Symbol-rate-variable DQPSK receiver fabricated using CMOS-compatible process,” in Microoptics Conference, (Optical Society of Japan / Japan Society of Applied Physics, 2011), paper PD-1. V. Mikhailov, R. I. Killey, and P. Bayvel, “Experimental investigation of partial demodulation of 85.3 Gb/s DQPSK signals,” in European Conference and Exhibition on Optical Communication, (IEEE Photonics Society, 2008), paper We.1.E.5. M. Shinkawa, N. Ishikura, Y. Hama, K. Suzuki, and T. Baba, “Nonlinear enhancement in photonic crystal slow light waveguides fabricated using CMOS-compatible process,” Opt. Express 19(22), 22208–22218 (2011). F. Shinobu, Y. Arita, and T. Baba, “Low-loss simple waveguide intersection in silicon photonics,” Electron. Lett. 46(16), 1149–1150 (2010). J. Wang, W. Y. Loh, K. T. Chua, H. Zang, Y. Z. Xiong, T. H. Loh, M. B. Yu, S. J. Lee, G. Q. Lo, and D. L. Kwong, “Evanescent-coupled Ge p-i-n photodetectors on Si-waveguide with SEG-Ge and comparative study of lateral and vertical p-i-n configurations,” IEEE Electron Device Lett. 29(5), 445–448 (2008). S. Park, T. Tsuchizawa, T. Watanabe, H. Shinojima, H. Nishi, K. Yamada, Y. Ishikawa, K. Wada, and S. Itabashi, “Monolithic integration and synchronous operation of germanium photodetectors and silicon variable optical attenuators,” Opt. Express 18(8), 8412–8421 (2010). D. Ahn, C. Y. Hong, J. Liu, W. Giziewicz, M. Beals, L. C. Kimerling, J. Michel, J. Chen, and F. X. Kärtner, “High performance, waveguide integrated Ge photodetectors,” Opt. Express 15(7), 3916–3921 (2007). 1. Introduction Since slow-light has much lower group-velocity than the speed of light in vacuum, it gives rise to large delay in a compact device. Slow-light devices that make the delay tunable are expected toward on-chip time-domain optical signal processing, such as optical buffering, retiming, de/multi-plexing, fast correlation, and so on [1]. For these purposes, various slowlight devices, for example photonic crystal waveguides (PCWs) [2,3], micro-ring all-passfilters (APF) [4–6] and micro-ring coupled-resonator optical waveguides (CROWs) [7,8] have been studied. Among these devices, we focus on the micro-ring APF because high quality micro-ring APF can be fabricated easily with Si CMOS-compatible process and it is more robust against structural disordering than PCWs and CROWs. Previously, we demonstrated a heater-controlled 300-ps tunable delay with 1 nm bandwidth [9]. In this paper, we report an application of this slow-light device as a tunable delay line in an optical coherent receiver. The optical coherent receiver is important for the rapidly increasing data network traffic at data centers and high-performance computers. In such networks, the signal delay and loss at electric wires are the major issues that limit their performance. To overcome this, low-cost and high-performance optical transceivers have been studied extensively and even on-board and on-chip optical interconnects are being studied in Si photonics with their optical components such as passive circuits [10–12], optical modulators [13], photodetectors (PDs) [14], and their integrations [15–17]. In future advanced optical interconnects with higher transmission capacities, downsizing of devices and high-density integration will be required, as well as reduced fiber cables and connectors. In this context, we focus on multilevel phaseshift-keying devices in Si photonics. It is suitable for high bit-rate transmission with low baud-rate, which is easier to handle in Si electronics. In this paper, we present the fabrication and characterization of differential quadrature phase-shift-keying (DQPSK) receiver, which is monolithically integrated on SOI substrate. Since the DQPSK format does not require any local oscillators at the receiver, it is relatively easy to fabricate. So far, InP-based and Si- #161674 - $15.00 USD (C) 2012 OSA Received 18 Jan 2012; revised 6 Feb 2012; accepted 7 Feb 2012; published 10 Feb 2012 13 February 2012 / Vol. 20, No. 4 / OPTICS EXPRESS 4797 based monolithic DQPSK receivers have been reported [18–21]. The Si-based receiver is more suitable for optical interconnects because of its large potential for opto-electronic integrations and low-cost mass production. The Si-based device reported in Ref [20]. was designed for long-haul transmissions. Here, we focus more on the applications to optical interconnects and add a new functionality −variability of symbol rates− by integrating the slow-light-based tunable one-bit delay line. This variability gives the flexibility to DQPSK receivers; the flexibility is still advantageous even under partial-demodulation, which improves the tolerance to the chromatic dispersion [22]. We demonstrate the demodulation of DQPSK optical signal at bit rates of 14.8 − 18.0 Gb/s. In the following sections, we present the layout of the receiver, then evaluate its components and operation as a DQPSK receiver. Tunable 1-bit Delay (a) MZI Attenuator Signal Input Monitor Monitor Heater Ge PD 90o Hybrid µ-Ring … Monitor Crossing Monitor Monitor Monitor (b) Ge PD Signal Input (from SSC) Tunable 1-bit Delay MZI Attenuator 90o Hybrid 200 µm Fig. 1. Si DQPSK receiver. (a) Configuration. (b) Optical microscope image of fabricated device. Taper w: 0.6 → 0.4 µm L: 2 µm Port 1 (b) 5.3 µm 2.6 µm 2.2 µm 0.9 µm Port 2 R = 5 µm Port 2 4 2 0 0 −2 −4 −2 −6 0.45 dB −8 −4 1.50 1.52 1.54 1.56 1.58 1.60 Wavelength λ [µm] Imbalance [dB] Transmission [dB] R = 5 µm Port 1 Port 2 Port 1 0.9 µm 1.8 µm Transmission [dB] Input Input 4 2 0 0 −2 −4 −2 −6 0.84 dB −8 −4 1.50 1.52 1.54 1.56 1.58 1.60 Wavelength λ [µm] Imbalance [dB] (a) Fig. 2. Fabricated splitter/coupler and their branching characteristics. (a) 1×2 MMI splitter. (b) 2×2 MMI coupler. w and L are the width and length of taper, respectively. 2. Fabricated receiver The receiver was fabricated by using CMOS-compatible process (8-inch SOI wafer, KrF stepper exposure). A Si wire waveguide which is a basic component, is 0.40 µm wide and 0.22 µm high, and buried by SiO2 cladding. Light is coupled from lens-attached fiber to the waveguide through a spot-size converter (SSC) located at the edge of the device chip. The #161674 - $15.00 USD (C) 2012 OSA Received 18 Jan 2012; revised 6 Feb 2012; accepted 7 Feb 2012; published 10 Feb 2012 13 February 2012 / Vol. 20, No. 4 / OPTICS EXPRESS 4798 coupling loss is typically 3 dB/facet [23]. Figure 1(a) illustrates the layout of the fabricated receiver including variable optical attenuators (VOAs), tunable one-bit delay line, 90° hybrid and balanced PDs. The VOA is a heater-controlled asymmetric Mach-Zehnder interferometer (MZI). The tunable one-bit delay line consists of a long Si-wire waveguide for a fixed delay and a heater-controlled all-pass micro-ring array slow–light device [9] for the tunable delay. The 90° hybrid includes 1 × 2 and 2 × 2 multi-mode interference (MMI) splitters/couplers, a mode-expansion-type crossing [24] and heater-controlled phase-shifters. The PD is composed of n-type Ge epitaxially grown on a p-type Si slab [25]. In addition, 3 dB tap monitors were included at the outputs of VOAs and the 90° hybrid to evaluate the transmitted power and fine-tune each device. All devices described above are designed to operate under transverseelectric (TE) polarization. For fiber communications, the polarization-insensitive operation or some particular polarization control is generally required for optical devices. For on-chip optical interconnects, on the other hand, such polarization control is not necessary as the polarization from emitters to receivers is fixed to either one. All devices are monolithically integrated, as shown in the optical microscope image in Fig. 1(b). The device footprint is as small as ~1.0 × 2.5 mm2. The footprint is mainly dominated by the wide separation between phase-shifters, which is set in this preliminary experiment so as to avoid thermal crosstalk. The footprint will be reduced further if these separations and electrode pads are optimized. A. Splitter and coupler The MMI splitters and couplers were designed using three-dimensional finite-difference timedomain method, taking into account the optical characteristics, size and robustness against fabrication error. Figure 2 shows structural parameters of the fabricated devices and the measured transmission and branching characteristics. The transmission spectrum of the 1 × 2 splitter in Fig. 2(a) is almost flat, and its excess loss is 0.45 dB at λ = 1.55 µm. The spectrum of the branching ratio exhibits small oscillation with a wavelength interval ∆λ ~0.5 nm and an amplitude of less than ± 0.25 dB. Provided that the group index of Si-wire is 4.2, the interval corresponds to a 570 µm cavity length, which is in good agreement with the length between the waveguide facet at the SSC and the input end of the splitter, which is 550 µm. This indicates that the oscillation is not caused by the MMI itself but by the Fabry-Perot resonance between the SSC and splitter. Neglecting the oscillation by averaging the spectrum, the branching ratio falls into ± 0.20 dB. On the other hand, the transmission spectrum of the 2 × 2 coupler in Fig. 2(b) is approximately flat, and its excess loss is slightly increased to 0.84 dB at λ = 1.55 µm. The branching ratio is within ± 0.25 dB at λ = 1.530 − 1.565 µm (C band). The performance in Fig. 2 is sufficient for the receiver operation. B. Crossing In the 90° hybrid, a waveguide crossing is necessary. We used the mode-expansion-type crossing that we further optimized from that in Ref [24]. As shown in Fig. 3(a), the expanded waveguide is composed of four ellipses. The excess loss and crosstalk caused by the diffraction around the center are suppressed by slightly narrowing the expanded waveguide near the center. We fabricated N-cascaded crossings on another chip for evaluation and measured the transmission spectra, as shown in Fig. 3(b). The spectrum is almost flat even at larger N. Figure 3(c) shows the N-dependence of transmission intensity at λ = 1.55 µm. The slope indicates a loss of 0.12 dB/crossing. #161674 - $15.00 USD (C) 2012 OSA Received 18 Jan 2012; revised 6 Feb 2012; accepted 7 Feb 2012; published 10 Feb 2012 13 February 2012 / Vol. 20, No. 4 / OPTICS EXPRESS 4799 (a) 2.76 µm 1.0 µm Transmission [dB] 3.6 µm (b) 0 (c) N =2 −5 5 10 20 0.12 dB/Crossing −10 −15 1.50 1.52 1.54 1.56 1.58 Wavelength λ [µm] 5 10 15 20 25 Number of Crossing N 0 Fig. 3. (a) Fabricated crossing and its loss characteristics. (b) Transmission spectra. (c) Loss at λ = 1.55 µm. (a) 200 µm Trench Heater Input Heater Output Transmission [dB] 5 µm (b) 0 Pheat [mW] 0, 17 (c) −10 27 dB −20 −30 1.53 1.54 1.55 1.56 Wavelength λ [µm] 0 20 40 60 80 100 120 Electric Power Pheat [mW] Fig. 4. (a) Fabricated heater controlled MZI attenuator. (b) Transmission spectra with and without heating. (c) Transmission intensity with heating power at λ = 1.55 µm. C. VOA The VOA is necessary for balancing input powers to the 90° hybrid. To realize the on-chip VOA, high-density carrier injection into the pn-junction has been used [26]. In this study, we employed a heater-controlled asymmetric MZI because the pn doping process can be neglected and the electrical power consumption can be reduced. Figure 4(a) shows the fabricated device. We used the splitter in Fig. 2(a) as the branch and confluence of the device. Each arm of the MZI has a heater placed above the Si-wire, and a deep trench is formed around the MZI to increase heating efficiency and suppress the thermal crosstalk. (b) shows the transmission spectra of the device. The spectra exhibits an oscillation with ∆λ = 7 nm, corresponding to the asymmetric arm length of 81.4 µm. The excess loss of the device estimated from the total loss of the branch and confluence is 0.90 dB. (In Fig. 4(b), the maximum intensity of the oscillation is −2 dB because the fiber-to-fiber transmission intensity through a shorter Si wire waveguide is used as a reference in this figure.) When only the single arm of the MZI is heated, the heating power Pheat = 17 mW is required for the π phase shift. (c) shows the transmission intensity with Pheat at λ = 1.55 µm. The attenuation can be varied from 0 to 27 dB. #161674 - $15.00 USD (C) 2012 OSA Received 18 Jan 2012; revised 6 Feb 2012; accepted 7 Feb 2012; published 10 Feb 2012 13 February 2012 / Vol. 20, No. 4 / OPTICS EXPRESS 4800 D. Slow-light-based tunable one-bit delay line An important feature of this receiver is the variable symbol rate, which is achieved by integrating a tunable one-bit delay line. It is composed of a long Si-wire waveguide for a fixed delay and the heater-controlled all-pass micro-ring slow-light device [9] for a tunable delay, as shown in Fig. 5. The long Si wire waveguide is 7.13-mm long, giving a 100 ps delay for a group index of 4.2. The slow-light device consists of 10 race track rings, each of which is directionally coupled with the bus waveguide and ring has a heater directly above. As the heaters can be controlled individually, their resonance wavelengths are tuned. As a result, the delay at a specific wavelength can be changed from 100 ps (off resonance) to 150 ps (on resonance). The loss at the fixed delay line is 2.1 dB, which corresponds to the total loss of the one-bit delay line under off resonance. On resonance, on the other hand, a 6 dB loss is added due to the tunable delay-line. We evaluated the one-bit delay tunable range of the fabricated device. A OOK optical signal at λ = 1.546 µm (off-resonance) was modulated by non return to zero (NRZ) 27−1 pseudo random bit sequence (PRBS) generated from a pulse pattern generator (PPG) (Anritsu, MU-181020A) in a LiNbO3 modulator. Then the optical signal was launched to the receiver. VOAs were adjusted to pass the optical signal only through the delay line arm. In this case, no interference occurs at the 90° hybrid, which allows us to observe the delayed signal at the subsequent monitor port. Pattern-locked waveforms of the output signal were measured by a sampling oscilloscope (Agilent, 86100C / 54754A) through an O/E converter. One-bit delay was achieved at 7.4, 8.0 and 9.0 Gbaud when 3.0, 0.96 and 0 mW heating power were injected to each ring, as shown in Fig. 6(a)-(c), respectively. The delay could be extended further for lower symbol rates. However, since a longer delay increased the loss and degraded the signal quality, we set 7.4 Gbaud as the lowest symbol rates in the demodulation experiment shown in Section 3. (a) Output Input Input 10 µm Heater-Controlled Micro-Ring Delay Line Output Heater (b) 50 µm Micro Ring 5 m (Below Heater) µ Normalized Intensity Fig. 5. Fabricated tunable one-bit delay line. (a) Long Si wire waveguide for fixed delay. (b) Heater-controlled all-pass micro-ring slow-light device for tunable delay. 1.0 (a) Delay 0.8 0.6 0.4 0.2 0 0 0.5 1.0 1.5 Time [ns] (b) 2.0 (c) 0.5 1.0 1.5 Time [ns] 2.0 0.5 1.0 1.5 Time [ns] 2.0 Fig. 6. Pattern-locked waveform of delayed 27−1 NRZ PRBS signal. Gray and black lines depict those with and without delay. (a) 7.4 Gbaud. (b) 8.0 Gbaud. (c) 9.0 Gbaud. #161674 - $15.00 USD (C) 2012 OSA Received 18 Jan 2012; revised 6 Feb 2012; accepted 7 Feb 2012; published 10 Feb 2012 13 February 2012 / Vol. 20, No. 4 / OPTICS EXPRESS 4801 Al wire (a) Anode Input Cathode Anode Ge 5 µm 0.7 3.5 0.6 3.0 (c) 0.5 2.5 0.4 2.0 0.3 1.5 0.2 1.0 0.1 0.5 0 0 0 1 2 3 4 5 6 7 Coupled Optical Power [mW] Responsivity [A/W] 10−1 (b) Pin [mW] 10−2 4.3 1.4 10−3 0.33 10−4 0.10 −5 10 0.2 µA Dark 10−6 10−7 −4 −3 −2 −1 0 1 2 Voltage [V] Photocurrent [mA] Current [A] 20 µm Si Fig. 7. (a) Optical microscope image of fabricated Ge PD. (b) Current-voltage characteristic. (c) Photocurrent and responsivity characteristics for a bias voltage of −4 V. (a) (b) (c) Fig. 8. Eye-pattern of NRZ PRBS signal output from PD. Input optical power is 6 dBm. Bias voltage is −4 V. (a) 2 Gb/s. (b) 5 Gb/s. (c) 10 Gb/s. E. Ge PD The Ge PD is monolithically integrated on Si slab of SOI substrate using direct epitaxial growth [25]. The fabricated device is shown in Fig. 7(a). The tapered Si wire is connected with p-doped wide Si area. Ge is grown on this area, and its upper surface is n-doped to form the PIN structure in the vertical direction. (b) shows the current-voltage characteristics. The dark current at zero-bias is 0.2 µA, which is lower than a critical value of ~1 µA required for multi-Gb/s detection [27]. The dark current increases with increasing voltage, which should be caused by the carrier recombination through deep levels [25]. (c) shows the photocurrent characteristic at a bias voltage of −4 V. The photocurrent increases almost linearly with the input optical power. The responsivity is ~0.45 A/W. To evaluate the high speed operation, OOK modulated optical signal (231−1 NRZ PRBS) was coupled to the device. The coupled optical power into the Ge PD was 6 dBm, and the bias voltage was set at −4 V. Eye-pattern of the output electrical signal was observed by using the sampling oscilloscope, as shown in Fig. 8. Although the noise level at each “0” and “1” state becomes larger at higher bit-rates, the eye opening is confirmed at 10 Gb/s. 3. DQPSK demodulation We evaluated the demodulation of DQPSK optical signal in the fabricated receiver. Figure 9 illustrates the measurement setup. Two synchronized 231−1 NRZ PRBS signals were generated from the PPG. They modulated cw light from a tunable laser into DQPSK optical signals through LiNbO3 DQPSK modulator (Sumitomo Osaka Cement Co., Ltd., T.SBZH1.520PD-ADC-P-FN). The modulated optical signal was amplified by Er-doped-fiber amplifier, and its optical power was adjusted by one VOA. The signal was adjusted to TE-polarization, and coupled to the SSC of the receiver. The sum of the coupling loss at the SSC and the propagation loss in the entire Si wire waveguide was ~6 dB. Subsequent VOAs were adjusted so that intensity levels at the two input ports of the 90° hybrid are balanced. The tunable one- #161674 - $15.00 USD (C) 2012 OSA Received 18 Jan 2012; revised 6 Feb 2012; accepted 7 Feb 2012; published 10 Feb 2012 13 February 2012 / Vol. 20, No. 4 / OPTICS EXPRESS 4802 bit delay was optimized for the symbol-rate of the input signal. Phase-shifters of the 90° hybrid were adjusted to obtain the clearest eye-pattern at an output electrical signal. The heaters used for the above adjustments were controlled individually through external heater controller. The total excess loss at the VOA and 90° hybrid including four splitters, one coupler, one crossing and the fixed delay line is 4.9 dB. In addition, two 3-dB tap monitors give a loss of 6.8 dB. Adding other optical wiring losses, the total excess loss of the receiver without the tunable delay line is approximately 20 dB. The Ge PD was accessed with RF probe, and −4 V bias was applied through a bias-tee. Electrical output from the balanced PD was amplified by RF amplifiers, and the eye-pattern of differential output signal was observed on the sampling oscilloscope by subtracting one single output from the other output. Constellation patterns were also observed by using a coherent analyzer (Agilent, 93204A / N4391A), where each of the orthogonal signals was not the differential output but the single PD output due to the restriction of the RF probe, resulting in twice more noisy patterns than the eye patterns. The optical power incident to the Ge PD was −16 dBm when all components were optimized. Figure 10 shows eye-patterns and constellations of demodulated DQPSK signals. It is observed that eye-patterns of the phase modulated signal are converted to ones of the amplitude modulated signal, indicating that the receiver operates as a DQPSK receiver. The eye opening is confirmed at a symbol-rate of 7.4 Gbaud. Although it is difficult to confirm the eye opening at 9.0 Gbaud, the shape of the eye pattern can be recognized. These results show that the receiver can demodulate DQPSK signals ranging from 14.8 ( = 7.4 × 2) to 18.0 ( = 9.0 × 2) Gb/s in bit-rates. This is also confirmed from the constellations at 7.4 and 9.0 Gbaud. Although their convergences are not so good, the demodulated signals were split into four parts, which is a clear evidence of the demodulation. This receiver has a 6.8 dB excess loss from monitor ports, which limits the optical input power into the Ge PD. This leads to small amplitude electrical output and hence the eyepatterns and constellations affected by noise. If the tap output is designed to be small, clearer eye-patterns, constellations and higher speed operation will be obtained. Signal Generator Electric Amp Ch 1 Ch 1 NRZ PRBS Pulse Pattern Generator (Anritsu, MU181020A) Ch 2 Ch 2 DC Supply EDFA VOA Tunable Laser LN DQPSK Modulator ( Sumitomo Osaka Cement, T.SBZH1.5 ) Polarization Controller Input Signal (10 Gb/s) Fabricated Receiver … Electric Bias Tee Amp I1 I2 Q1 Q2 Lens-Attached Fiber Heater Controller ( Sampling Oscilloscope Agilent, 86100C/54754A ) DC Supply Fig. 9. DQPSK demodulation setup. #161674 - $15.00 USD (C) 2012 OSA Received 18 Jan 2012; revised 6 Feb 2012; accepted 7 Feb 2012; published 10 Feb 2012 13 February 2012 / Vol. 20, No. 4 / OPTICS EXPRESS 4803 (a) I Channel Q Channel (b) I Channel Q Channel Fig. 10. Eye-patterns and constellations of demodulated DQPSK signal. (a) 7.4 Gbaud. (b) 9.0 Gbaud. 4. Conclusion We fabricated a symbol-rate-variable DQPSK receiver by integrating various optical components including tunable slow-light device into a footprint of ~1 × 2.5 mm2 on SOI substrate using CMOS-compatible process. The tunable slow-light device successfully achieved variable symbol-rate detection. We confirmed that splitters/couplers, crossing, MZI attenuators, 90° hybrid and balanced Ge PDs have sufficient performance for receiver operation. The DQPSK demodulation experiment showed that the fabricated receiver operates at bit-rates of 14.8 − 18.0 Gb/s. These results demonstrate the potential of high-density photonic integration by Si photonics and useful application of slow-light. Acknowledgments This work was partly supported by the FIRST Program of JSPS. The authors would like to thank Sumitomo Osaka Cement Co., Ltd. for providing DQPSK LN modulator, and Agilent Technologies Inc. for providing coherent analyzer. #161674 - $15.00 USD (C) 2012 OSA Received 18 Jan 2012; revised 6 Feb 2012; accepted 7 Feb 2012; published 10 Feb 2012 13 February 2012 / Vol. 20, No. 4 / OPTICS EXPRESS 4804