In this paper, four-channel cascaded Mach-Zehnder interferometer-based wavelength (de)multiplexers in the O-band are demonstrated experimentally by utilizing silicon nitride (SiN) optical waveguides. By reference to the commonly used 100 Gigabit Ethernet standards, two types of (de)multiplexer devices with different channel spacings are designed and fabricated. Both the devices exhibit low insertion loss and flat passbands. The lower thermo-optical coefficient provided by SiN brings benefits of reduction in thermal sensitivity. The fabricated (de)multiplexers show a temperature-dependent wavelength shift of about 18.5 pm/°C, which is reduced by 75% compared to the standard silicon-based devices.
© 2017 Optical Society of America
Wavelength-division multiplexing (WDM) scheme, as an effective solution to utilize the unprecedented bandwidth scalability of photonic technology, has been widely adopted in optical communication systems . The wavelength parallelism in a single mode optical fiber which allows for the extension of network capacity sustained the explosive traffic demand in the long-haul networks over the past decades. Recently, rapid traffic growth within the local area networks has propelled the deployment of WDM optical links in short reach . With the ever-increasing market size in data center applications, 100 Gigabit Ethernet (GbE) has been standardized . In 100 GbE, 4 × 25 Gbit/s WDM electro-optic transceiver solutions are commonly employed and several independent norms and multi-source agreements (MSAs) have been released [4, 5]. Due to the ability to realize monolithic integration of photonics and electronics, silicon photonics platform is now an intriguing candidate for enabling ultra-compact, low-cost and power-efficient electro-optic transceivers [6–9]. Optical (de)multiplexers, as the key functional components in WDM transceivers, can be realized by using different device structures in silicon photonics, such as arrayed-waveguide gratings (AWGs) , echelle diffraction gratings (EDGs) , or cascaded Mach-Zehnder interferometers (MZIs) [12, 13]. Considering only four wavelength channels are (de)multiplexed in common 100 GbE applications, the cascaded MZI-based device exhibits benefits in respect of size, insertion loss and spectral flatness compared with the other two device types. During the past few years, considerable progress has been made toward developing the cascaded MZI-type WDM (de)multiplexers on silicon-on-insulator (SOI) platform [14, 15]. However, as a result of the high refractive index contrast in this platform and large thermo-optical (TO) coefficient of silicon core, a major challenge encountered by this device is the wavelength shift response when fabrication variations and environment temperature fluctuations take place. Several strategies have been proposed to alleviate this deterioration of filtering patterns. For example, the waveguides in the MZI arms can be widened to reduce the sensitivity to process variations and passive temperature compensation based on the asymmetric design of optical confinement is able to enhance the tolerance to temperature changes [12, 16, 17]. However, in these proposals additional adiabatic tapers need to be used for avoiding excitation of the higher order modes in the widened waveguide section and precise structure design and dimension control are required for the accurate temperature compensation, which leads to the increment of the total device size and complexity.
Silicon nitride (SiN), as another common material in CMOS fabs, provides moderate index contrast and relatively low TO coefficient, which is capable of enabling the reduction of process sensitivity and temperature dependence. Additionally, SiN waveguides can be conveniently integrated onto SOI platform using chemical vapor deposition (CVD) and lithography . Motivated by these potentials, SiN-based optical components encountered intensive research efforts in various applications. For (de)multiplexing functions, SiN AWG-type devices have been successfully demonstrated and monolithically integrated with silicon modulators and germanium photodetectors for constructing the transmitters and receivers in dense WDM (DWDM) systems [19, 20]. However, to our best knowledge, the SiN (de)multiplexers targeted for 100 GbE applications have not been reported previously. In this paper, we experimentally demonstrated the SiN (de)multiplexers designed for the 100GBASE-LR4 norm and CWDM4 MSA, based on the cascaded MZIs. The thermal sensitivities of the fabricated devices are measured around 18.5 pm/°C, which is reduced by ~75% compared to the standard silicon-based devices.
2. Device structure
The cascaded MZI-type (de)multiplexers are based on a series of four-port filter stages connected in a binary tree configuration. To implement 4-channel (de)multiplexing, two filter stages with different free spectral ranges (FSRs) are required for stepwise filtering. Figure 1(a) shows the schematic layout of the presented 4-channel demultiplexer. A single MZI, as a simple four-port filter with only one delay segment, exhibits a sine-like transmission spectrum. To achieve an improved spectral response with flat transmission passband and sharp roll-off at the band edge, lattice filter consisting of more delay segments with optimized coupling coefficients and delay line lengths can be employed. Here, considering the filter response of the demultiplexer is primarily determined by that of the input filter stage, a four-port lattice filter with two delay segments is leveraged in the first stage. At the output side (i.e. the second stage), two single MZIs perform the filtering function with the footprint saving benefit.
Figure 1(b) shows the cross section of the SiN waveguides used in the (de)multiplexing devices. The width W and height H are selected as 1.1 μm and 340 nm, respectively, for ensuring single-mode propagation. The waveguides sit on top of a 3-μm-thick buried oxide (BOX) layer. Using the finite-difference method, the mode profile of the SiN waveguide is simulated and shown in Fig. 1(c). The calculated effective index and group index values at the wavelength of 1302 nm are about 1.60 and 2.05, respectively. In each filter stage, the delay line length differences are determined in terms of the required FSR value and relative frequency offset. A more detailed description about the design of the delay line length differences can be found in Ref . Considering the relatively lower index contrast of the SiN platform as compared to SOI contributes to the reduction in the sensitivity of the waveguide effective index to the width variations induced by fabrication errors, uniform waveguide width of 1.1 μm is utilized throughout the SiN (de)multiplexers. The selection of the geometrical parameters of the directional couplers is based on the three-dimensional finite-difference time-domain simulation. To achieve the power cross-coupling coefficients around 0.5, 0.29 and 0.08 , the coupling lengths of the directional couplers are selected as 31 μm, 18 μm and 3 μm, respectively, with the coupling gap fixed at 0.26 μm.
3. Fabrication and characterization
Fabrication is carried out by first conducting the thermal oxidation of a standard silicon wafer to form the BOX layer, which is followed by the deposition of SiN device layer using low-pressure CVD (LPCVD). Electron-beam lithography (EBL) and induced coupled plasma (ICP) etching are then performed to define waveguide patterns on the SiN layer. The propagation loss of the fabricated SiN waveguide is about 2.8 dB/cm. To couple light in and out of the waveguides, a fully etched apodized grating coupler is designed which allows for the patterning of all device components in a single etch step and supports operation with TE-polarized light only. Figures 2(a) and 2(b) show the scanning electron microscope (SEM) image of a fabricated apodized grating coupler and the optical microscope image of a completed demultiplexer, respectively. About the device size, the footprint of the fabricated SiN demultiplexers is approximately 1 × 0.6 mm2. In the demultiplexers, all the bent waveguides have a radius of 60 μm.
The fabricated demultiplexers are characterized using the test setup depicted in Fig. 2(c). Light from a broadband amplified spontaneous emission (ASE) source is first sent into an optical polarizer and a polarization controller (PC) for achieving TE-polarized input. Then the light of TE polarization is coupled in and out of the fabricated devices assisted by grating couplers. The devices under test (DUT) are placed on a sample stage equipped with a thermoelectric cooler which is used to control the temperature of the devices. The temperature changes are monitored using a thermistor mounted on this stage. The transmission spectrum at each output port of the demultiplexers is measured by an optical spectrum analyzer (OSA).
4. Experimental results
4.1 Room temperature
Figure 3(a) plots the measured room-temperature transmission spectra at all four outputs of the demultiplexer device designed for 100GBASE-LR4 norm, labeled as Type-Ι device. The loss introduced by the apodized grating coupler has been assessed to be about 7 dB and removed in the above transmission spectra. The test results indicate that low insertion loss (IL) of less than 1.8 dB is achieved for each wavelength channel. Concerning the wavelength registration, the central wavelengths (λcenter) of four channels precisely match the standard with spectral shifts ranging from 0.02 nm to 0.1 nm, well below the ± 1 nm norm range. At four λcenter positions, the crosstalk (XTcenter) from the suppressed wavelength channels is less than −20 dB. The measured 1 dB bandwidth of all four channels exceeds 2.87 nm, which is about 63% of the channel spacing. When looking at 1 dB bandwidth, the worst crosstalk (XT1 dB) from the suppressed channels is about −7.5 dB. This crosstalk value could be improved by doubling the filter stages, but at the expense of a slight reduction in bandwidth . Table 1 lists the detailed performance characteristics for Type-Ι device.
The demultiplexer designed in accordance with 100G CWDM4 MSA (Type-ΙΙ device) is also fabricated and characterized. Structural distinctions between these two device types exist solely in the lengths of delay line waveguides, which originates from the different channel spacing defined by two standards. Figure 3(b) shows the measured normalized room-temperature transmission spectra at four outputs of the Type-ΙΙ device. The spectral ripples in the longer-wavelength side are caused by the reflection from the grating couplers . The IL value of all four transmission channels is below 1.8 dB. With respect to the ITU-T grids for CWDM, the channel center wavelengths exhibit spectral shifts from 0.37 nm to 1.62 nm, within the norm range of ± 6.5 nm. The absolute wavelength shifts are relatively larger than those of the Type-Ι device, which is likely due to the extended influence of the phase error induced by increased filter period and the enlarged group velocity dispersion over a wider spectral range in the Type-ΙΙ device. As for the crosstalk, XTcenter below −22 dB is achieved for all channels except the one with the longest wavelength (XTcenter <-15 dB), which is caused by the wavelength dependence of the directional couplers. The 1 dB bandwidth of four wavelength channels is more than 11.86 nm, or over 59% of the channel spacing. The corresponding XT1 dB value ranges from −8.02 dB to −9.54 dB. Table 2 summarizes the optical performance of Type-ΙΙ device.
4.2 Thermal sensitivities
To characterize the thermal sensitivities of these two types of demultiplexers, transmission spectra of the fabricated devices are recorded at different temperatures ranging from 20°C to 50°C with an interval of 5°C. Figures 4(a) and 4(b) show the measured spectra at 20°C and 50°C for the Type-Ι and Type-ΙΙ devices, respectively. In both figures, a spectral red shift of around 0.55 nm is observed with negligible variations of the shape of filtering curves for the 30-degree increase in temperature. In spite of the almost identical spectral shift, the performances of the two demultiplexers encounter different degrees of degradation due to the separate channel spacings, as seen in Figs. 4(a) and 4(b). For the CWDM device, the larger channel spacing and 1 dB bandwidth permit handling a greater change in temperature. By linear fitting the channel center wavelength shift with temperature, the thermal sensitivities of the two demultiplexers are extracted. Figure 5 shows the results of the channel at the minimum wavelength. For the Type-Ι device, the temperature-dependent wavelength shift is 18.4 pm/°C for the Ch. 1, as shown in Fig. 5(a). For the Type-ΙΙ device, the thermal sensitivity is 18.7 pm/°C for the Ch. 1, as shown in Fig. 5(b). Similar temperature dependences are observed for the other channels in both devices. Compared to the traditional silicon-based devices (70-80 pm/°C), the presented SiN demultiplexers achieve a reduction in thermal sensitivities of about 75% [16, 22].
5. Discussion and conclusion
The decreased temperature dependence contributes to the proper function of two (de)multiplexers in a realistic thermally variable environment. As mentioned, the CWDM device with a larger channel spacing allows for a greater operating temperature range. However, considering the temperature controllers are commonly used in the 100GBASE-LR4 optical transceiver modules, both devices are expected to work well in the commercial temperature range from 0°C to 70°C. In order to meet the industrial requirements, further improvement of the device performance need be performed for both demultiplexers, especially regarding the crosstalk. As mentioned above, doubling the filter stages is validated as a functional way to enhance the suppression of unwanted channels . To realize monolithic integration of the (de)multiplexers and active devices, (de)multiplexers with silica cladding are usually required. For the oxide-clad SiN (de)multiplexers, since the TO coefficient of silica is about four times smaller than that of SiN , the thermal sensitivity is still expected to be low. Additionally, considering a relatively large bending radius of 60 μm has been utilized in the presented air-clad (de)multiplexers, the device footprint could be largely maintained for the (de)multiplexers with oxide cladding. Besides the (de)multiplexers, combination of the reduced thermal sensitivity and improved process robustness enables the SiN platform to be a potential alternative for the fabrication of other silicon photonics passive devices.
In summary, we experimentally demonstrate two types of SiN four-channel (de)multiplexers based on the cascaded MZIs and designed for the 100GBASE-LR4 norm and CWDM4 MSA, respectively. Both devices exhibit low insertion loss and flat passbands. Due to the lower TO coefficient provided by SiN than that of silicon, the fabricated (de)multiplexers achieve a 75% reduction in thermal sensitivities compared to the standard silicon-based devices. The robustness to process variations is also expected to be improved by the reduced index contrast of SiN platform. All these features make the presented (de)multiplexers promising for 100 GbE applications.
National High Technology Research and Development Program of China (2015AA016904); National Natural Science Foundation of China (NSFC) (61335002, 11574102); Major State Basic Research Development Program of China (2013CB632104, 2013CB933303).
References and links
1. C. A. Brackett, “Dense wavelength division multiplexing networks: Principles and applications,” IEEE J. Sel. Areas Comm. 8(6), 948–964 (1990). [CrossRef]
2. J. P. Turkiewicz, P. Mazurek, and H. de Waardt, “Towards 1 Tbit/s SOA-based 1310 nm transmission for local area network/data centre applications,” IET Optoelectron. 9(1), 1–9 (2015). [CrossRef]
3. C. Cole, D. Allouche, F. Flens, B. Huebner, and T. Nguyen, “100GbE-Optical LAN Technologies [Applications & Practice],” IEEE Commun. Mag. 45(12), 12–19 (2007). [CrossRef]
6. L. Chen, C. Doerr, R. Aroca, S. Park, J. Geyer, T. Nielsen, C. Rasmussen, and B. Mikkelsen, “Silicon photonics for 100G-and-beyond coherent transmissions,” in 2016 Optical Fiber Communication Conference (2016), paper Th1B.1.
7. A. Narasimha, B. Analui, Y. Liang, T. J. Sleboda, and C. Gunn, “A Fully Integrated 4 × 10 Gb/s DWDM Optoelectronic Transceiver in a standard 0.13 μm CMOS SOI,” in Proceedings of IEEE International Solid State Circuits Conference (IEEE, 2007), pp. 42–43.
8. S. Assefa, S. Shank, W. Green, M. Khater, E. Kiewra, C. Reinholm, S. Kamlapurkar, A. Rylyakov, C. Schow, F. Horst, H. Pan, T. Topuria, P. Rice, D. M. Gill, J. Rosenberg, T. Barwicz, M. Yang, J. Proesel, J. Hofrichter, B. Offrein, X. Gu, W. Haensch, J. Ellis-Monaghan, and Y. Vlasov, “A 90nm CMOS integrated nano-photonics technology for 25Gbps WDM optical communications applications,” Electron Devices Meeting (IEDM), 2012 IEEE International, postdeadline session 33.8 (2012). [CrossRef]
9. H. F. Liu, “Demonstration of a 4λ × 12.5 Gb/s fully integrated silicon photonic link,” In Microoptics Conference (MOC), 2011 17th, 1–3. IEEE, 2011.
10. D. Dai, Z. Wang, J. F. Bauters, M.-C. Tien, M. J. Heck, D. J. Blumenthal, and J. E. Bowers, “Low-loss Si3N4 arrayed-waveguide grating (de)multiplexer using nano-core optical waveguides,” Opt. Express 19(15), 14130–14136 (2011). [CrossRef] [PubMed]
11. D. Feng, W. Qian, H. Liang, C.-C. Kung, J. Fong, B. J. Luff, and M. Asghari, “Fabrication insensitive echelle grating in silicon-on-insulator platform,” IEEE Photonics Technol. Lett. 23(5), 284–286 (2011).
12. F. Horst, W. M. Green, S. Assefa, S. M. Shank, Y. A. Vlasov, and B. J. Offrein, “Cascaded Mach-Zehnder wavelength filters in silicon photonics for low loss and flat pass-band WDM (de-)multiplexing,” Opt. Express 21(10), 11652–11658 (2013). [CrossRef] [PubMed]
13. M. S. Hai, A. Leinse, T. Veenstra, and O. Liboiron-Ladouceur, “A thermally tunable 1 × 4 channel wavelength demultiplexer designed on a low-loss Si3N4 waveguide platform,” Photonics 2(4), 1065–1080 (2015). [CrossRef]
14. S.-H. Jeong, D. Shimura, T. Simoyama, T. Horikawa, Y. Tanaka, and K. Morito, “Si-nanowire-based multistage delayed Mach-Zehnder interferometer optical MUX/DeMUX fabricated by an ArF-immersion lithography process on a 300 mm SOI wafer,” Opt. Lett. 39(13), 3702–3705 (2014). [CrossRef] [PubMed]
15. S. Dwivedi, P. De Heyn, P. Absil, J. Van Campenhout, and W. Bogaerts, “Coarse wavelength division multiplexer on silicon-on-insulator for 100 GbE,” in IEEE International Conference on Group IV Photonics (GFP), WC2 (2015). [CrossRef]
16. K. Hassan, C. Sciancalepore, J. Harduin, T. Ferrotti, S. Menezo, and B. B. Bakir, “Toward athermal silicon-on-insulator (de)multiplexers in the O-band,” Opt. Lett. 40(11), 2641–2644 (2015). [CrossRef] [PubMed]
18. R. Baets, A. Z. Subramanian, S. Clemmen, B. Kuyken, P. Bienstman, N. Le Thomas, G. Roelkens, D. Van Thourhout, P. Helin, and S. Severi, “Silicon Photonics: silicon nitride versus silicon-on-insulator,” in 2016 Optical Fiber Communication Conference (2016), paper Th3J.1.
19. L. Chen, C. R. Doerr, L. Buhl, Y. Baeyens, and R. A. Aroca, “Monolithically integrated 40-wavelength demultiplexer and photodetector array on silicon,” IEEE Photonics Technol. Lett. 23(13), 869–871 (2011). [CrossRef]
20. P. Dong, “Silicon photonic integrated circuits for Wavelength-Division-Multiplexing applications [Invited],” IEEE J. Sel. Top. Quantum Electron. 22(6), 6100609 (2016). [CrossRef]
21. Y. Wang, X. Wang, J. Flueckiger, H. Yun, W. Shi, R. Bojko, N. A. Jaeger, and L. Chrostowski, “Focusing sub-wavelength grating couplers with low back reflections for rapid prototyping of silicon photonic circuits,” Opt. Express 22(17), 20652–20662 (2014). [CrossRef] [PubMed]
23. X. Tu, J. Song, T.-Y. Liow, M. K. Park, J. Q. Yiying, J. S. Kee, M. Yu, and G.-Q. Lo, “Thermal independent silicon-nitride slot waveguide biosensor with high sensitivity,” Opt. Express 20(3), 2640–2648 (2012). [CrossRef] [PubMed]