Open Access
Add to BibSonomy
Abstract
We present a silicon thermo-optic 2☓2 four-mode optical switch optimized for optical space switching plus local optical mode switching. Four asymmetric directional couplers are utilized for mode multiplexing and de-multiplexing. Sixteen 2☓2 single-mode optical switches based on balanced thermally tunable Mach-Zehnder interferometers are exploited for switching function. The measured insertion losses are 8.0~12.2 dB and the optical signal-to-noise ratios are larger than 11.2 dB in the wavelength range of 1525~1565 nm. The optical links in “all-bar” and “all-cross” states exhibit less than 2.0 dB and 1.4 dB power penalties respectively below 10−9 bit error rates for 40 Gbps data transmission.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Photonic network-on-chip (NoC) is regarded as a promising scheme to meet the demands on large-capacity, low-latency and low-power-consumption interconnect of high-performance multi-core processors [1–3]. Optical switch, which is responsible for optical link switching at the nodes, is a key component of photonic NoC. Silicon photonics is thought to become a very promising platform for the realization of on-chip optical switches because of its compactness in footprint and compatibility with the standard complementary metal-oxide semiconductor process [2]. In the past several years, various optical switches optimized for different photonic NoCs, such as Fat-Tree [4], Mesh [5, 6], Clos [7] and Crossbar [8], have been proposed and demonstrated in silicon photonics platform. Wavelength-division multiplexing (WDM) has been exploited to improve the information capacity of photonic NoC [9]. Mode-division multiplexing (MDM) has attracted widespread attention, since it offers a new dimension to increase the capacity seamlessly by transmitting multiple optical signals simultaneously in orthogonal spatial modes of a waveguide [10–17].
Different from traditional optical switches in photonic NoCs, the switching system with the introduction of MDM is more complex and demands new scheme to realize the switching function among optical signals carried on different spatial modes [15–17]. A silicon 1☓2 two-mode optical switch based on microrings has been demonstrated by B. Stern et al. [15] Recently, a silicon 2☓2 four-mode optical switch based on Mach-Zehnder switches has been demonstrated [16]. Besides, a silicon 2☓2 two-mode dual-polarization optical switch based on Mach-Zehnder switches has also been demonstrated [17]. While the optical switches demonstrated in [16] and [17] focus on switching data between mode or polarization groups and are not capable of switching data in the same mode group.
In this paper, we demonstrate a silicon thermo-optic 2☓2 four-mode optical switch based on Mach-Zehnder switches. The optical switch is capable of switching data not only between different mode groups but also in the same mode group. The device is composed of two ADC based mode multiplexers, sixteen 2☓2 Mach-Zehnder optical switches and two ADC based mode de-multiplexers within a footprint of 4000 µm☓1200 µm. The insertion losses (ILs) for all measured optical links of the device are within 12.2 dB in the wavelength range of 1525~1565 nm. The optical signal-to-noise ratios (OSNRs) are larger than 11.2 dB in the same wavelength range. To verify the data transmission capability, eye diagrams and bit error rates (BERs) for 40 Gbps data transmission are measured.
2. Device design
Figures 1(a) and 1(b) show the principle and architecture of the 2☓2 four-mode optical switch, respectively. The optical switch has two input multimode waveguides I1, I2 and two output multimode waveguides O1, O2. To simplify the description of the switching functionalities, we mark the multiple spatial modes in the input multimode waveguide I1/I2 as mode group 1/2. We denote the group switching of optical signals in mode groups 1 and 2 as optical space switching. Note that the mode sequence remains constant. We denote the switching of optical signals among different spatial modes in the same mode group as local optical mode switching.
Fig. 1 (a) Principle and (b) architecture of the 2☓2 four-mode optical switch; structures of (c) mode multiplexers based on asymmetric directional couplers and (d) thermo-optic 2☓2 SM-OS. (M-MUX, mode multiplexer; M-DEMUX, mode de-multiplexer; M-MUX A, auxiliary mode multiplexer; M-DEMUX A, auxiliary mode de-multiplexer; SM-OS, single-mode optical switch; OSU, optical switching unit; MMI, multimode interference; TiN, titanium nitride).
First, the four spatial modes in mode group 1 or 2 from an input multimode waveguide are converted to four fundamental modes by the corresponding four-channel mode de-multiplexer, which are then space switched to four output single-mode waveguides by the follow-on 4☓4 single-mode optical switch (SM-OS). The eight outputs of two 4☓4 SM-OS are then shuffled to eight input ports of four 2☓2 SM-OS. We name each 2☓2 SM-OS optical switching unit (OSU) in the following sections. The eight output ports of the four OSUs are connected to two four-channel mode multiplexers. When the four OSUs (S13~S16) are in the identical states (“cross” or ‘bar”), the entire mode group of an input multimode waveguide can be switched to a specific output multimode waveguide simultaneously. The functionality of the 4☓4 SM-OS is to realize the local optical mode switching among the optical signals in mode group 1 or 2.
The ILs of an optical switch are mainly caused by the OSUs and waveguide crossings. It is reported that the insertion loss of a waveguide crossing can be down to 0.05 dB, while that of a thermo-optic OSU is about 0.3 dB [18]. The insertion loss uniformity is also important for an optical switch as its performance is usually decided by the optical link with the largest insertion loss. So we adopt Beneš [19] network to construct the 4 × 4 non-blocking SM-OS, as it has a better performance in insertion loss and insertion loss uniformity than Spanke-Beneš [20] and crossbar [21] networks. Since each 4☓4 SM-OS has 4! routing states and the cascaded four OSUs (S13~S16) have identical states (“cross” or ‘bar”), the 2☓2 four-mode optical switch has (4!)2 × 2 = 1152 routing states. Several structures have been exploited for mode multiplexing and de-multiplexing, such as adiabatic coupler [22, 23], asymmetric Y-junction [24, 25], multimode interference coupler [26] and asymmetric directional coupler (ADC) [10–17, 27, 28]. Here ADCs are chosen to construct mode multiplexers/de-multiplexers for their better compactness and scalability. To be compatible with the electro-optic tuning scheme in the future, silicon rib waveguides with 220 nm height and 70 nm slab thickness are used to construct the ADCs. As shown in Fig. 1(c), the width of the rib waveguides carrying the TE0, TE1, TE2 and TE3 mode are 0.40 µm, 0.92 µm, 1.42 µm and 1.92 µm respectively. The optimized coupling lengths for the TE1, TE2 and TE3 modes are 13 µm, 15 µm and 19 µm, respectively. The gaps between single-mode waveguides and the multimode waveguides are 0.26 µm for TE1, TE2 and TE3 modes. Adiabatic tapers connecting the waveguides with different widths are 10 µm length. Previous work indicates that the optical bandwidths of the mode multiplexers/de-multiplexers based on ADCs are relatively large, which are consequently suitable for WDM application [10–14, 16,28] and the ADCs in [28] are used here. Considering device characterization, optical fibers should be introduced for input and output of optical signals. Nevertheless, the coupling between multimode fibers and multimode waveguides is still a challenge work. Consequently, two auxiliary mode multiplexers and two mode de-multiplexers are integrated with the device. We denote the input/output single-mode waveguide for the ith-order spatial mode of the auxiliary mode multiplexer/de-multiplexer m as / (i = 0, 1, 2, 3 and m = 1, 2), as shown in Fig. 1(b). Although both Mach-Zehnder optical switch and microring optical switch are capable of manipulating WDM signals, the former is adopted for its larger tolerance to fabrication imperfection and lower temperature sensitivity [29, 30]. Two multimode interference (MMI) couplers with 6 μm width and 43 μm length and two thermo-optic phase shifters with 200 μm length are utilized to construct the 2☓2 Mach-Zehnder optical switch, which is used as the OSU in the 2☓2 four-mode optical switch, as shown in Fig. 1(d). A silicon inverse taper covered by an air-bridge silicon dioxide intermediate transition waveguide is used to reduce the coupling loss between the silicon waveguide and a lensed fiber with a spot size of 5 μm.
3. Device fabrication
The device is fabricated on an 8-inch silicon-on-insulator wafer with a 220-nm-thick top silicon layer and a 3-µm-thick buried silicon dioxide layer by 180 nm CMOS foundry process at the Institute of Microelectronics, Singapore. 248-nm deep ultraviolet photolithography is utilized to define the patterns and inductively coupled plasma etching is employed to form the silicon waveguides. A 1500-nm-thick silicon dioxide layer is deposited on the silicon layer by plasma-enhanced chemical vapor deposition (PECVD) to prevent the absorption of the optical field by the metal. Then a 150-nm-thick titanium nitride is sputtered on the separate layer, and 1-µm-wide and 200-µm-long TiN micro-heaters are fabricated on the two arms of the Mach-Zehnder optical switch to realize the thermal tuning. Via holes are etched after depositing a 300-nm-thick silicon dioxide layer by PECVD. Finally, aluminum wires and pads are fabricated. Figure 2 shows the micrograph of the fabricated device with a footprint of 4000 µm☓1200 µm.
4. Device characterization
4.1 Transmission spectra
As mentioned in Sec. 2, there exists 1152 routing states for the device. Here we select four representative routing states for the performance characterization of the optical switch. The established optical links and states of 16 OSUs in the chosen routing states are shown in Table 1.
Table 1. Measured optical links and states of 16 OSUs
The experimental setup for characterizing the transmission spectra is shown in the bottom half of Fig. 3. An amplified spontaneous emission source and an optical spectrum analyzer are utilized to measure the transmission spectra of the device. The driving voltages and states of each OSU are calibrated by direct-current powers. If all 16 OSUs are in the “bar” state (State 1, “all-bar” state), the optical signals will be guided from I1/I2 to O1/O2 and the mode sequences in mode groups 1 and 2 will remain constant. In other words, neither local optical mode switching nor optical space switching will occur in State 1. If S13~S16 are in the “cross” state and S1~S12 are in the “bar” state (State 2), the optical space switching will occur and the optical signals will be guided from I1/I2 to O2/O1 while the switching among optical signals carried on different spatial modes in mode group 1 or 2 will not occur. Contrarily, if S1~S12 are in the “cross” state and S13~S16 are in the “bar” state (State 3), local optical mode switching will occur in mode groups 1 and 2, but optical space switching will not happen. If all 16 OSUs are in the “cross” state (State 4, “all-cross” state), both local optical mode switching and optical space switching will occur.
Fig. 3 Experimental setup for characterizing the device (TL, tunable laser; ASE, amplified spontaneous emission; MD, modulator; PC, polarization controller; PPG, pulse pattern generator; AFG, arbitrary function generator; DCP, direct-current power; DUT, device under test; DCA, digital communication analyzer; RTO, real-time oscilloscope; VOA, variable optical attenuator; OSA, optical spectrum analyzer; PM, power meter).
The transmission spectra for the signal and noise of the optical links in the shown routing states in Table 1 are normalized with the intrinsic transmission spectrum of ASE and shown in Fig. 4. Straight waveguides with the same coupling structures as those of the device are utilized to estimate the coupling loss between the device and a lensed fiber with a spot size of 5 μm. The experimental result shows that it is about 2.4 dB/butt. Therefore, the on-chip insertion loss of the 2☓2 four-mode optical switch is within 7.4 dB (12.2-2.4☓2 = 7.4 dB) in the wavelength range of 1525~1565 nm. Slight wavelength dependence in the transmission spectra is mainly caused by the dispersion of the mode multiplexers/de-multiplexers based on ADCs and the OSUs based on Mach-Zehnder interferometers. It is noted that fabrication imperfection makes the central wavelengths of the OSUs and the mode multiplexers/de-multiplexers deviate from their theoretical values, which further results in the transmission fluctuation. The ILs of the optical links come from the coupling losses of the device with the fibers, the propagation losses of connected waveguides and the ILs of waveguide crossings,mode multiplexers/de-multiplexers and OSUs. The measured insertion loss of the waveguide crossing is ~0.05 dB and the measured propagation loss of the connected waveguide is 2.5 dB/cm. The insertion loss of the mode multiplexer/de-multiplexer is 0.2~0.5 dB in the wavelength of 1525-1565 nm, which is a little bit larger than its calculated values (~0.03 dB for TE0 mode, 0.04~0.16 dB for TE1 mode, 0.05~0.16 dB for TE2 mode, and 0.05~0.16 dB for TE3 mode). The insertion loss per OSU is 0.4~1.2 dB in the wavelength range of 1525~1565 nm.
Fig. 4 Transmission spectra for the signal and noise of the measured optical links in the shown routing states in Table 1 (IL, insertion loss; OSNR, optical signal-to-noise ratio).
The total noise for one specific routing path is the summation of the noise from all input ports, since the light beams from different input ports are incoherent with each other. The total noise is shown in the bold blue lines in Fig. 4. The OSNRs are larger than 11.2 dB for all optical links in the wavelength range of 1525~1565 nm. The detailed ILs and OSNRs for all measured optical links of the device in the wavelength range of 1525-1565 nm are shown in Table 2.
Table 2. ILs and OSNRs for the measured optical links in the wavelength range of 1525-1565 nm
The random errors caused by fabrication imperfection deviate the performance of ADCs from their design targets and further increase the inter-modal crosstalk among different mode channels in the same mode group. And the fabrication imperfection deteriorates the performance of the 3 dB MMI couplers in OSUs, which increases the crosstalk between their two input/output ports and further deteriorates the OSNRs of the switch. The OSNRs of the device can be improved by improving the fabrication or adopting OSU with high OSNR to construct the optical switch [31].
Generally, the ILs and OSNRs of the optical switch are mainly decided by the performance of ADCs, OSUs and waveguide crossings. To expand the optical switch to a larger scale than two mode groups demonstrated here, more ADCs, OSUs and waveguide crossings are required, which further deteriorates the IL and OSNR of each mode channel. And the crosstalk among different mode channels becomes more complex. An ADC capable of (de)multiplexing more than four modes can also be used to expand the optical switch [14], but it leads to higher intel-modal crosstalk and requires a more complex device design. To address the deterioration of ILs and OSNRs caused by the expanding of the optical switch, a more advanced fabrication technology than 180 nm CMOS foundry process used here can be considered for device fabrication [32].
4.2 Eye diagrams and bit error rates
The top half of Fig. 3 illustrates the experimental setup used for data transmission and BER characterization. 40 Gbps pseudo-random binary sequence with a length of 231-1 generated by a pulse pattern generator is utilized to drive a LiNbO3 optical modulator. Continuous-wave light emitted from a tunable laser is first modulated by the LiNbO3 optical modulator and then sent to a polarization controller. The light derived from the polarization controller is sent to the digital communication analyzer for back to back (B2B) eye diagram observation and the real-time oscilloscope for BER measurement, respectively. Similarly, the light derived from the polarization controller is coupled into and out of the device by a lensed fiber with a spot size of 5 μm. The output optical signal is then sent to a digital communication analyzer for eye diagram observation and a real-time oscilloscope with a receiver for BER measurement. A variable optical attenuator and an optical power meter are used to control and monitor optical power, respectively.
The eye diagrams and BERs for 40 Gbps data transmission of the device in states 1 [Figs. 5(a) and 5(b)] and 4 [Figs. 5(c) and 5(d)] at 1545 nm and 1565 nm are measured. Clear and open eye diagrams are observed for all measured optical links, which verifies the data transmission function of the device with the data rate of 40 Gbps. As shown in Figs. 5(a) and 5(b), the power penalties of state 1, in which all the 16 OSUs are in the “bar” state, with respect to B2B are 1.2~2.0 dB and 1.3~1.8 dB at 1545 nm and 1565 nm, respectively. The power penalties of state 4, in which all the 16 OSUs are in the “cross” state, with respect to B2B are 1.1~1.2 dB and 1.1~1.4 dB at 1545 nm and 1565 nm respectively, as shown in Figs. 5(c) and 5(d). The BER curves for eight measured optical links are within 0.8 dB at BER of 10−9 at 1545 nm and 1565 nm, indicating that the optical switch has a good equilibrium characteristic for data transmission.
Fig. 5 Eye diagrams and BERs for 40 Gbps data transmission of the device in the “all-bar” state at 1545 nm (a), 1565 nm (b) and “all-cross” state at 1545 nm (c), 1565 nm (d).
4.3 Power consumptions and response time
The driving voltages and power consumptions of all the 16 OSUs in the “cross” and “bar” states are listed in Table 3. The power variances are due to different initial phase differences between the two arms for different OSUs caused by the fabrication imperfection. The power consumptions of the device are 408 mW in the “all-bar” state and 502 mW in the “all-cross” state, respectively. The number of OSUs increases when the optical switch is expanded to a large scale, which further increases the power consumption. The power consumption can be reduced by improving fabrication which can reduce the random phase errors in the waveguides to reduce the extra phase-shift voltages. And air trench can be implemented to improve the heat efficiency [33].
Table 3. Driving voltages and power consumptions of all the 16 OSUs
A real-time oscilloscope is used to measure the response time of each OSU by applying a 10 KHz square-wave electrical signal from an arbitrary function generator to it. Since the response time of each optical link is decided by all OSUs that it passes through, the response time is limited by the slowest OSU, whose 10% - 90% rising time and 90% - 10% falling time are about 12 μs and 9 μs, respectively. The responses of the other OSUs are around this value. An electro-optic tuning mechanism with higher switching speed can be applied to improve the relatively low switching speed in the future [34].
5. Conclusion
In conclusion, we present a silicon thermo-optic 2☓2 four-mode optical switch specially optimized for optical space switching plus local optical mode switching. ADCs are used for mode multiplexing/de-multiplexing and thermal tunable Mach-Zehnder switches are utilized to construct the switching network. The experimental results show that the ILs are within 8.0~12.2 dB and the OSNRs are larger than 11.2 dB, which can be further optimized by improving the fabrication. The optical links of the device in the “all-bar” and “all-cross” states exhibit 1.2~2.0 dB and 1.1~1.4 dB power penalties respectively below 10−9 BERs for 40 Gbps data transmission, which indicates a good equilibrium characteristic for data transmission of the optical switch. The optical switch demonstrated here is compatible with WDM and MDM applications and shows promising potential for future high capacity optical networks.
Funding
National Key R&D Program of China (2017YFA0206402); National Natural Science Foundation of China (NSFC) (61575187, 61505198, 61535002, 61704168, 61235001).
References and links
1. R. G. Beausoleil, P. J. Kuekes, G. S. Snider, S. Y. Wang, and R. S. Williams, “Nanoelectronic and nanophotonic interconnect,” Proc. IEEE 96(2), 230–247 (2008). [CrossRef]
2. D. A. B. Miller, “Device requirements for optical interconnects to silicon chips,” Proc. IEEE 97(7), 1166–1185 (2009). [CrossRef]
3. L. Thylén and L. Wosinski, “Integrated photonics in the 21st century,” Photon. Res. 2(2), 75–81 (2014). [CrossRef]
4. H. Gu, J. Xu, and W. Zhang, “A low-power fat tree-based optical network-on-chip for multiprocessor system-onchip,” in Proceedings of the conference on Design, Automation and Test in Europe (2009), pp. 3–8.
5. H. X. Gu, K. H. Mo, J. Xu, and W. Zhang, “A low-power low-cost optical router for optical networks-on-chip in multiprocessor systems-on-chip,” in 2009 IEEE Computer Society Annual Symposium on VlSI (2009), pp. 19–24 . [CrossRef]
6. G. Hendry, E. Robinson, V. Gleyzer, J. Chan, L. P. Carloni, N. Bliss, and K. Bergman, “Time-division-multiplexed arbitration in silicon nanophotonic networks-on-chip for high-performance chip multiprocessors,” J. Parallel Distrib. Comput. 71(5), 641–650 (2011). [CrossRef]
7. A. Joshi, C. Batten, Y. J. Kwon, S. Beamer, I. Shamim, K. Asanovic, and V. Stojanovic, “Silicon-photonic clos networks for global on-chip communication,” in Proceedings of the 3rd ACM/IEEE International Symposium on Networkson-Chip (2009), pp. 124–133. [CrossRef]
8. S. Han, T. J. Seok, N. Quack, B. W. Yoo, and M. C. Wu, “Large-scale silicon photonic switches with movable directional couplers,” Optica 2(4), 370–375 (2015). [CrossRef]
9. R. Aguinaldo, A. Forencich, C. DeRose, A. Lentine, D. C. Trotter, Y. Fainman, G. Porter, G. Papen, and S. Mookherjea, “Wideband silicon-photonic thermo-optic switch in a wavelength-division multiplexed ring network,” Opt. Express 22(7), 8205–8218 (2014). [CrossRef] [PubMed]
10. L. W. Luo, N. Ophir, C. P. Chen, L. H. Gabrielli, C. B. Poitras, K. Bergmen, and M. Lipson, “WDM-compatible mode-division multiplexing on a silicon chip,” Nat. Commun. 5, 3069 (2014). [CrossRef] [PubMed]
11. Y. D. Yang, Y. Li, Y. Z. Huang, and A. W. Poon, “Silicon nitride three-mode division multiplexing and wavelength-division multiplexing using asymmetrical directional couplers and microring resonators,” Opt. Express 22(18), 22172–22183 (2014). [CrossRef] [PubMed]
12. J. Wang, Y. Xuan, M. Qi, H. Huang, Y. Li, M. Li, X. Chen, Z. Sheng, A. Wu, W. Li, X. Wang, S. Zou, and F. Gan, “Broadband and fabrication-tolerant on-chip scalable mode-division multiplexing based on mode-evolution counter-tapered couplers,” Opt. Lett. 40(9), 1956–1959 (2015). [CrossRef] [PubMed]
13. D. Dai, “Silicon Nanophotonic Integrated Devices for On-Chip Multiplexing and Switching,” J. Lightwave Technol. 35(4), 572–587 (2017). [CrossRef]
14. D. Dai, C. Li, S. Wang, H. Wu, Y. Shi, Z. Wu, S. Gao, T. Dai, H. Yu, and H. K. Tsang, “10‐Channel Mode (de) multiplexer with Dual Polarizations,” Laser Photonics Rev. 12(1), 1700109 (2018). [CrossRef]
15. B. Stern, X. Zhu, C. P. Chen, L. D. Tzuang, J. Cardenas, K. Bergman, and M. Lipson, “On-chip mode-division multiplexing switch,” Optica 2(6), 530–535 (2015). [CrossRef]
16. H. Jia, T. Zhou, L. Zhang, J. Ding, X. Fu, and L. Yang, “Optical switch compatible with wavelength division multiplexing and mode division multiplexing for photonic networks-on-chip,” Opt. Express 25(17), 20698–20707 (2017). [CrossRef] [PubMed]
17. Y. Zhang, Y. He, Q. Zhu, C. Qiu, and Y. Su, “On-chip silicon photonic 2× 2 mode-and polarization-selective switch with low inter-modal crosstalk,” Photon. Res. 5(5), 521–526 (2017). [CrossRef]
18. S. Zhao, L. Lu, L. Zhou, D. Li, Z. Guo, and J. Chen, “16× 16 silicon Mach–Zehnder interferometer switch actuated with waveguide microheaters,” Photon. Res. 4(5), 202–207 (2016). [CrossRef]
19. V. E. Beneš, “Algebraic and topological properties of connecting networks,” Bell Labs Tech. J. 41(4), 1249–1274 (1962). [CrossRef]
20. R. A. Spanke and V. E. Beneš, “N-stage planar optical permutation network,” Appl. Opt. 26(7), 1226–1229 (1987). [CrossRef] [PubMed]
21. H. S. Hinton, “A nonblocking optical interconnection network using directional couplers,” in Proceedings of the IEEE Global Telecommunications Conference (1984), pp. 885–889.
22. M. Greenberg and M. Orenstein, “Multimode add-drop multiplexing by adiabatic linearly tapered coupling,” Opt. Express 13(23), 9381–9387 (2005). [CrossRef] [PubMed]
23. C. Sun, Y. Yu, G. Chen, and X. Zhang, “Silicon mode multiplexer processing dual-path mode-division multiplexing signals,” Opt. Lett. 41(23), 5511–5514 (2016). [CrossRef] [PubMed]
24. J. B. Driscoll, R. R. Grote, B. Souhan, J. I. Dadap, M. Lu, and R. M. Osgood, “Asymmetric Y junctions in silicon waveguides for on-chip mode-division multiplexing,” Opt. Lett. 38(11), 1854–1856 (2013). [CrossRef] [PubMed]
25. W. Chen, P. Wang, and J. Yang, “Mode multi/demultiplexer based on cascaded asymmetric Y-junctions,” Opt. Express 21(21), 25113–25119 (2013). [CrossRef] [PubMed]
26. T. Uematsu, Y. Ishizaka, Y. Kawaguchi, K. Saitoh, and M. Koshiba, “Design of a compact two-mode multi/demultiplexer consisting of multimode interference waveguides and a wavelength-insensitive phase shifter for mode-division multiplexing transmission,” J. Lightwave Technol. 30(15), 2421–2426 (2012). [CrossRef]
27. Y. Sun, Y. Xiong, and W. N. Ye, “Experimental demonstration of a two-mode (de)multiplexer based on a taper-etched directional coupler,” Opt. Lett. 41(16), 3743–3746 (2016). [CrossRef] [PubMed]
28. H. Jia, L. Zhang, J. Ding, L. Zheng, C. Yuan, and L. Yang, “Microring modulator matrix integrated with mode multiplexer and de-multiplexer for on-chip optical interconnect,” Opt. Express 25(1), 422–430 (2017). [CrossRef] [PubMed]
29. R. Ji, L. Yang, L. Zhang, Y. Tian, J. Ding, H. Chen, Y. Lu, P. Zhou, and W. Zhu, “Five-port optical router for photonic networks-on-chip,” Opt. Express 19(21), 20258–20268 (2011). [CrossRef] [PubMed]
30. K. Padmaraju and K. Bergman, “Resolving the thermal challenges for silicon microring resonator devices,” Nanophotonics 3(4–5), 269–281 (2014).
31. K. Suzuki, K. Tanizawa, S. Suda, H. Matsuura, T. Inoue, K. Ikeda, S. Namiki, and H. Kawashima, “Broadband silicon photonics 8 × 8 switch based on double-Mach-Zehnder element switches,” Opt. Express 25(7), 7538–7546 (2017). [CrossRef] [PubMed]
32. K. Tanizawa, K. Suzuki, M. Toyama, M. Ohtsuka, N. Yokoyama, K. Matsumaro, M. Seki, K. Koshino, T. Sugaya, S. Suda, G. Cong, T. Kimura, K. Ikeda, S. Namiki, and H. Kawashima, “Ultra-compact 32 × 32 strictly-non-blocking Si-wire optical switch with fan-out LGA interposer,” Opt. Express 23(13), 17599–17606 (2015). [CrossRef] [PubMed]
33. Y. Sun, Y. Cao, Y. Yi, L. Tian, Y. Zheng, J. Zheng, F. Wang, and D. Zhang, “A low-power consumption MZI thermal optical switch with a graphene-assisted heating layer and air trench,” RSC Advances 7(63), 39922–39927 (2017). [CrossRef]
34. M. Yang, W. M. J. Green, S. Assefa, J. Van Campenhout, B. G. Lee, C. V. Jahnes, F. E. Doany, C. L. Schow, J. A. Kash, and Y. A. Vlasov, “Non-blocking 4x4 electro-optic silicon switch for on-chip photonic networks,” Opt. Express 19(1), 47–54 (2011). [CrossRef] [PubMed]
