Silicon Mach-Zehnder modulators (Si MZMs) with good linearity are designed and fabricated. 6.25 Gbaud Nyquist 16, 32 and 64-Quadrature Amplitude Modulation (QAM) optical signals were successfully generated by intensity modulation from the Si MZM, and the effective data rates are 22.61 Gb/s, 28.26 Gb/s and 33.91 Gb/s respectively. The subcarrier multiplexed technique and direct detection scheme were employed in this experiment. After 53.1 km transmission, the BERs of 16-QAM and 32-QAM are both below the 7% hard-decision forward error correction limit, while the back-to-back BER of 64-QAM is well below the 20% soft-decision forward error correction limit. These results demonstrated that the Si MZM can be used in the high-capacity low-cost short-haul intensity modulation and direct detection system.
© 2014 Optical Society of America
As the rapid expansion of multimedia service continues, the demand of network capacity keeps growing. High-capacity low-cost communication systems employing advanced modulation formats such as quadrature phase shift keying (QPSK) and quadrature amplitude modulation (QAM) which have higher spectral efficiency, have been paid much attention in recent years. Coherent optical communications is a popular scheme to increase the channel capacity in long-haul transmissions. The intensity modulation and direct detection (IM-DD) optical systems provide a cost advantage in some metro, backhaul, access and interconnection applications. To beat the high cost of the discrete devices for the communication system, silicon photonics emerged as a potential candidate because of its high integration density and compatibility with complementary metal oxide semiconductor process.
As one of the most important components, silicon modulators attracted much effort from the research and development community. Silicon modulator can be realized via plasma dispersion effect which makes use of the nonlinear electro-optic effect to change the refractive index. Carrier accumulation, carrier injection and carrier depletion are three different mechanisms commonly used to electrically manipulate the free carrier concentrations in plasma-dispersion-based silicon modulators . The depletion-type silicon modulators which have higher modulation speed draw much interest, despite a bit of lower modulation efficiency . Compared to resonant modulators which are sensitive to fabrication error, temperature variation and wavelength drift, non-resonant modulators, typically Mach-Zehnder modulators (MZMs), exhibit much better tolerance and stability . Based on the depletion-type silicon Mach-Zehnder modulators (Si MZMs), the basic intensity modulation, On Off keying (OOK), was investigated previously [2,3], and the advanced modulation formats in coherent systems have been widely reported [4–8]. On the other hand, conventional IM-DD systems with advanced modulation formats were mostly using LiNbO3 MZMs which have high linearity and already show good performance. But, considering the compactness and cost effectiveness, researchers are keen on developing better silicon modulators for some high-capacity short-haul applications. As of now, the intensity modulation of 36 Mb/s 16-QAM and 54 Mb/s 64-QAM signals were generated by using depletion-type Si MZM with horizontal PN junction utilizing OFDM technique . The single carrier QPSK and 16-QAM signals using ring assisted Mach-Zehnder modulator was demonstrated at symbol rates of less than 50 Msymbol/s . Meanwhile, direct detection optical Orthogonal Frequency Division Multiplexing (DDO-OFDM) data with advanced modulation formats of QPSK, 8PSK, and 16-QAM (effective data rates are 4.38 Gb/s, 2.60 Gb/s and 1.71 Gb/s), generated by using depletion-type Si MZM, have been transmitted over 50km .
In this work, we report the intensity modulation of 6.25 Gbaud Nyquist pulse shaping 16-QAM, 32-QAM, and 64-QAM signals using a depletion-type Si MZM based on plasma dispersion effect, achieving the effective data rates of 22.61 Gb/s, 28.26 Gb/s and 33.91 Gb/s, respectively. Considering the high peak-to-average-power-ratio of OFDM, which makes the whole system need a wider linear range, an alternative approach of using single carrier subcarrier multiplexed (SCM) Nyquist QAM signals  was chosen in this work.
2. Device design and fabrication
Figure 1(a) shows cross-section of phase shifters in depletion-type Si MZM. The phase shifter waveguide of 600 nm in width is fabricated on a silicon-on-insulator (SOI) wafer with 220 nm top silicon layer. The slab of the rib waveguide is 60nm in height. The doping concentration is controlled as 5 × 1017 cm−3 for P type and 1 × 1018 cm−3 for N type. And the P type doping region is 100 nm right to the middle of the PN junction to enlarge the overlap between P doping region and light field. There is an 80 nm intrinsic region between the P and N doping in the PN junction. The distance between the edge of the rib and the heavy doping area is 0.5 μm. The doping concentration of the P + + and N + + areas are both 1 × 1020 cm−3 for good ohmic contact. The Si MZM is designed with 2 mm phase shifter. The length difference is 100 μm between the two arms and two 1 × 2 multimode interferometers (MMIs) are used to split and recombine the optical beam in the Mach-Zehnder interferometer. The coplanar waveguide (CPW) traveling-wave electrode is employed to drive the phase shifters as shown in Fig. 1(b). Four integrated termination resistances are fabricated between every G and S electrode. The propagation loss of the Si MZM is approximately 4.5 dB including 2.9 dB in phase shifter and 0.8 dB per MMI.
3. Linearity of the Si MZM
The nonlinear transfer function between applied voltage and refractive index change of Si MZM is due to the modulation mechanism of plasma dispersion effect. Taking overall consideration of the nonlinearity caused by the Mach-Zehnder structure and the plasma dispersion effect, the nonlinearity of the Si MZM shows dependence on voltage and wavelength . The linearity of Si MZM can be quantified by spurious free dynamic range (SFDR). SFDR is the difference between the carrier power and the noise level, when the distortion equals to the noise level. In this work, we studied our Si MZM and measured the SFDRs of second harmonic distortion (SHD) and third-order intermodulation distortion (IMD), which are located in the close proximity of carriers, and are generally detrimental to the transmission system (the noise level is chosen to be −165 dBm).
The measured results of the Si MZM are shown in Fig. 2(a). The highest SFDRs of SHD and third-order IMD are 79.72 dB/Hz1/2 and 91.85 dB/Hz2/3 respectively, under the bias voltage of −4 V. These results are comparable to the published date under single arm driving mode . The S-parameter of the Si MZM was measured with a vector network analyzer (Agilent N4373C). Figure 2(b) shows the measured electro-optical (E-O) modulation response under different bias voltage. The 3-dB bandwidth is larger than 18 GHz under −2 V bias voltage.
4. Intensity modulation of SCM Nyquist QAM signals on Si MZM
The experimental setup for generation, transmission, and detection of 6.25 Gbaud Nyquist 16, 32 and 64-QAM signal is shown in Fig. 3. At the transmitter, digital signal processing (DSP) is performed to form 6.25 Gbaud Nyquist 16, 32, 64-QAM signals. After four times up-sampling, the signal is filtered with a root-raised cosine (RRC) filter with a roll-off factor of 0.1. The signal comes to analog signal after filtering. Then the signal is up-converted to 6.25 GHz with the RF frequency equal to the symbol rate, which is the so-called single cycle SCM scheme . Afterwards the generated signal is sent to the arbitrary waveform generator (Tektronix AWG70002A) operating at 25GS/s. The continuous-wave (CW) laser of 1548.128 nm is employed to generate the optical signal. The Erbium doped fiber amplifier (EDFA) (Amonics AEDFA-30-B-FA) after laser compensates the insertion loss and it is helpful to achieve the required high optical signal to noise ratio (OSNR) in the experiment. In this experiment, we test Si MZM with two working modes, the single arm driving and the push pull driving. With intensity modulation, a double-sideband (DSB) Nyquist optical signal is generated. Because of the better modulation depth and linearity, the modulator under push pull driving mode achieves better performance. Therefore, under this driving mode, the transmission performance after 53.1km standard single mode fibers (SSMF) is investigated. The EDFA (Amonics AEDFA-DWDM-23-R-FC) after the modulator is used to adjust the launch power in the transmission link. The variable optical attenuator (VOA) is employed to change the optical power for calculating the BERs under different OSNRs. The last EDFA (Amonics AEDFA-PA-35-B-FA) after the VOA is employed to achieve the sensitivity of the receiver. In the transmission experiments there is no inline chromatic dispersion compensation. In DSB transmission, fiber dispersion results in different phases and self-cancellation for the two sidebands. Therefore, at the receiver, a waveshaper (Finisar-4000s) filters out one side of the received DSB optical signal and a single-sideband (SSB) signal is obtained. Meanwhile, the waveshaper can decrease the noise in the system. After the optical filter, the SSB signal is directly detected by a photo-diode and then sampled by a real time digital storage oscilloscope (Tektronix DPO72004B) operating at 50 GS/s. In the offline processing, the sampled Nyquist 16, 32 and 64-QAM signals down-converted to base band and single-carrier frequency domain equalization (SCFDE) is utilized .
The frame structure is described as following. The mapped 16,32 and 64-QAM signals are grouped into blocks with 128 symbols each, and 2 pilots are time-multiplexed into each SCFDE block to compensate the laser phase noise. In each block, 4 symbols as cyclic prefix (CP) and 4 symbols as cyclic suffix (CS) are added. In the preamble, two 63-symbol Chu-sequences are adopted for synchronization, four 128-symbol Chu-sequences are introduced as training sequences for channel estimation. The length of preamble is 670 symbols (63 × 2 + (128 + 8) × 4). In each frame, 27200 data symbols (every 128 data symbols with 4 CP and 4 CS) are included after the preamble. Therefore the effective data rate is 22.61 Gb/s (6.25 Gbaud × 4 bit/Symbol/(1 + 670/27200)/(1 + 8/128) × 126/128) for 16-QAM signal. By the same calculation method the effective data rate is 28.26 Gb/s for 32-QAM, and it’s 33.91 Gb/s for 64-QAM signal.
Figure 4 shows the BER performances of Nyquist 16-QAM versus OSNR at 0.1 nm resolution and the measured optical spectrums after 53.1 km SSMF transmission. The Si MZM is working in the best linearity condition with DC voltage of −4 V. And the radio frequency driving voltage is 8 Vpp. The high driving voltage is necessary for the required high OSNR in the experiment. Figure 4(a) shows the BER curves versus OSNR under different driving modes. The required OSNR (BER = 3.8 × 10−3) for single arm driving mode is 35.3 dB, while the required OSNR for push pull driving mode is 30.4 dB which is 4.9 dB lower than the single arm driving mode. The push pull driving mode offers better linearity and subsequently the improved performance . Meanwhile, push pull driving mode is easier to achieve deeper modulation. In this scheme, the linear region of the device is utilized more efficiently. The inset of Fig. 4(a) shows the 16-QAM constellation under push pull driving mode with the OSNR of 38.25 dB. Since the high required OSNR of this system, the transmission distance is limited. We just test the transmission performance of 53.1 km. Figure 4(b) shows the measured back-to-back and 53.1 km transmission BER curves versus OSNR under push pull driving mode. The OSNR (0.1nm) penalty of 53.1 km SSMF transmission at 7% hard-decision FEC threshold is about 4.4 dB compared with the back-to-back situation. Figure 4(c) shows the measured optical spectrums after 53.1 km SSMF transmission. The black curve shows the DSB signal which is hardly damaged after 53.1 km SSMF transmission. The red one is the SSB signal after filtering by waveshaper. The demonstrated 16-QAM’s effective data rate is tenfold higher than the previous reports [9–11]. These results extend the Si MZMs’ application to the high-capacity intensity modulation system.
The measured BER curves of 6.25 Gbaud Nyquist 32-QAM versus OSNR at 0.1 nm resolution are shown in Fig. 5. The required OSNR (BER = 3.8 × 10−3) under single arm driving mode is 38.1 dB, while the required OSNR for push pull driving mode is 36.5 dB which is 1.6 dB lower than the former. The OSNR difference is smaller than 16-QAM because of the limited driving voltage. The 32-QAM signal with more amplitude levels requires higher driving voltage to distinguish every amplitude level. Due to the limited driving voltage, the modulator may not perform at its best level. Therefore, the difference of the linearity between two driving modes does not play a major role in the system. The constellation diagram of 32-QAM under push pull driving mode with the OSNR at 40.0 dB is shown in the inset of Fig. 5(a). The measured BER curves of back-to-back and 53.1 km SSMF transmission versus OSNR under push pull driving mode are shown in Fig. 5(b). The OSNR (0.1 nm) penalty of 53.1 km SSMF transmission at 7% hard-decision FEC threshold is about 4.7 dB. To the best of our knowledge, this is the first realization of 32-QAM based on Si MZM.
Figure 6(a) shows the back-to-back BER curve versus OSNR at 0.1 nm resolution of 6.25 Gbaud Nyquist 64-QAM. Because of the limited driving voltage, the modulator only worked under push pull driving mode for better performance. The BER is well below the 20% soft-decision FEC threshold of 2 × 10−2 . The constellation diagram of 64-QAM with the OSNR at 43.4 dB is shown in the inset of Fig. 6(a). The nonlinear distortion and noise have more serious impact on the more advanced modulation format. 64-QAM signal has not transmitted. This is the first demonstration of 64-QAM in such high data rate on Si MZM. The much higher speed and more advanced modulation format are benefited from the enough 3 dB electro-optic bandwidth and the good linearity. Meanwhile push pull driving mode is a scheme to improve the performance in the IM-DD system. The performance of 64-QAM can be optimized by increasing the driving voltage. In addition, adding a transimpedance amplifier (TIA) after photo diode can further improve the system performance.
For comparison, the BER curves of Nyquist 16, 32 and 64-QAM signals, under push pull driving mode, are collectively shown in Fig. 6(b). It shows a trend that the more advanced modulation formats will require higher OSNR. The required OSNR differences are evaluated under the BER of 1 × 10−2 which is below the 20% soft-decision FEC threshold. The 32-QAM signal requires 5.6 dB higher OSNR than the 16-QAM signal. The required OSNR difference is 7.6 dB between 32 and 64-QAM signals. Although the IM-DD system shows lower complexity, one should also note that the receivers’ sensitivity of the direct detection is smaller than the coherent scheme, which means higher required OSNR. The higher driving voltage and input power should be employed in this system. And the higher required OSNR means shorter transmission. So the high-capacity low-cost IM-DD systems are usually employed in short-haul situation.
We have generated 6.25 Gbaud Nyquist 16-QAM, 32-QAM, and 64-QAM signals using depletion-type Si MZM with SCM technique and achieved the effective data rates of 22.61 Gb/s, 28.26 Gb/s and 33.91 Gb/s experimentally. Comparing the single arm driving mode, the push pull driving mode showed better performance of Si MZM in the IM-DD system. The 53.1 km transmission of 6.25 Gbaud Nyquist 16 and 32-QAM signals are well below the 7% hard-decision FEC threshold. These results prove that although the Si MZM is considered a nonlinear device, a structurally optimized device with proper bias point can provide enough linearity to show the feasibility and enormous potential for the high-capacity low-cost short-haul IM-DD systems.
This work is partially supported by the National High Technology Research and Development Program of China (863 Program) (Grants No. 2011AA010302 and 2012AA011302) and Program for New Century Excellent Talents in University.
References and links
1. G. T. Reed, G. Mashanovich, F. Y. Gardes, and D. J. Thomson, “Silicon optical modulators,” Nat. Photonics 4(8), 518–526 (2010). [CrossRef]
3. H. Yi, Q. Long, W. Tan, L. Li, X. Wang, and Z. Zhou, “Demonstration of low power penalty of silicon Mach-Zehnder modulator in long-haul transmission,” Opt. Express 20(25), 27562–27568 (2012). [CrossRef] [PubMed]
4. T. Li, J. Zhang, H. Yi, W. Tan, Q. Long, Z. Zhou, X. Wang, and H. Wu, “Low-voltage, high speed, compact silicon modulator for BPSK modulation,” Opt. Express 21(20), 23410–23415 (2013). [CrossRef] [PubMed]
5. T. Li, J. Zhang, H. Yi, W. Tan, Q. Long, Z. Zhou, X. Wang, and H. Wu, “10-Gb/s 53.1-km BPSK transmission of silicon Mach-Zehnder modulator,” in Asia Communications and Photonics Conference (Optical Society of America, 2013), AW4A.3. [CrossRef]
6. K. Goi, H. Kusaka, A. Oka, Y. Terada, K. Ogawa, T.-Y. Liow, X. Tu, G. Lo, and D.-L. Kwong, “DQPSK/QPSK Modulation at 40-60 Gb/s using Low-Loss Nested Silicon Mach-Zehnder Modulator,” in Optical Fiber Communication Conference (Optical Society of America, 2013), paper OW4J.4. [CrossRef]
7. P. Dong, X. Liu, S. Chandrasekhar, L. L. Buhl, R. Aroca, Y. Baeyens, and Y. K. Chen, “Monolithic silicon photonic circuits enable 112-Gb/s PDM- QPSK transmission over 2560-km SSMF,” in European Conference on Optical Communications (2013), paper We.2.B.1.
8. P. Dong, X. Liu, C. Sethumadhavan, L. L. Buhl, R. Aroca, Y. Baeyens, and Y.-K. Chen, “224-Gb/s PDM-16-QAM modulator and receiver based on silicon photonic integrated circuits,” in Nat. Fiber Opt. Eng. Conf., OSA Tech. Dig., Anaheim, CA (2013), Paper PDP5C.6. [CrossRef]
9. F. Vacondio, M. Mirshafiei, J. Basak, A. Liu, L. Liao, M. Paniccia, and L. A. Rusch, “A silicon modulator enabling RF over fiber for 802.11 OFDM signals,” IEEE J. Sel. Top. Quantum Electron. 16(1), 141–148 (2010). [CrossRef]
10. A. M. Gutierrez, J. V. Galan, J. Herrera, A. Brimont, D. Marris-Morini, J. M. Fedeli, L. Vivien, and P. Sanchis, “High linear ring-assisted MZI electro-optic silicon modulators suitable for radio-over-fiber applications,” in 2012 IEEE 9th International Conference on Group IV Photonics (IEEE, 2012), pp. 57–59. [CrossRef]
11. X. Ke, L. Yang, J. Sung, Y. Chen, Z. Cheng, C. Chow, C. Yeh, and H. Tsang, “Compatibility of silicon Mach-Zehnder modulators for advanced modulation formats,” J. Lightwave Technol. 31(15), 2550–2554 (2013). [CrossRef]
12. M. Erkilinc, R. Maher, M. Paskov, S. Kilmurray, S. Pachnicke, H. Griesser, B. Thomsen, P. Bayvel, and R. Killey, “Spectrally-efficient single-sideband subcarrier-multiplexed quasi-Nyquist QPSK with direct detection,” in European Conference and Exhibition on Optical Communication (2013), paper Tu3C4. [CrossRef]
13. M. Streshinsky, A. Ayazi, Z. Xuan, A. E.-J. Lim, G.-Q. Lo, T. Baehr-Jones, and M. Hochberg, “Highly linear silicon traveling wave Mach-Zehnder carrier depletion modulator based on differential drive,” Opt. Express 21(3), 3818–3825 (2013). [CrossRef] [PubMed]
14. A. O. J. Wiberg, B.-E. Olsson, and P. A. Andrekson, “Single cycle subcarrier modulation,” in Optical Fiber Communication Conference, OSA Technical Digest (2009), paper OTuE1. [CrossRef]
15. J. Li, S. Zhang, F. Zhang, and Z. Chen, “Comparison of transmission performances for CO-SCFDE and CO-OFDM systems,” IEEE Photon. Technol. Lett. 22(14), 1054–1056 (2010). [CrossRef]
16. D. Chang, F. Yu, Z. Xiao, Y. Li, N. Stojanovic, C. Xie, X. Shi, X. Xu, and Q. Xiong, “FPGA Verification of a Single QC-LDPC Code for 100 Gb/s Optical Systems without Error Floor down to BER of 10−15,” in Proceedings of OFC/NFOEC2011 (6–10 March 2011), OTuN2.