Mode-division multiplexing technology using the high-order modes of multimode waveguides enables high-bandwidth data transmission. High-speed mode channel switching is a pivotal function for these optical networks. Here, we propose a modal switching scheme on a silicon-on-insulator platform and demonstrate a high-speed two-mode switch that exploits a Y-junction and multimode interference structure. The design allows for simultaneous switching of two optical modes. A PN-doped junction-based phase shifter in one branch of a Y-junction enables dynamic switching in 2.5 ns. The measured switching extinction ratio is 12.5 dB or better with an open eye diagram for a 10 Gb/s on–off key optical payload signal. The optical power penalty is within 0.5 dB for the two-mode switching at a bit error rate of . This two-mode switch could enable on-chip mode-based switching network topology for greater aggregated throughput capacity.
© 2017 Optical Society of America
Silicon photonics is a promising platform for high-bandwidth optical interconnects due to their compatibility with existing complementary metal-oxide semiconductor technologies and dense integration . To offer high data transmission capacity, wavelength-division multiplexing and polarization-division multiplexing using fundamental optical transverse electric (TE) and transverse magnetic (TM) modes have been investigated extensively [2–5]. Recently, mode-division multiplexing (MDM) has attracted attention to further increase data capacity by introducing high-order modes for each single wavelength carrier. To achieve MDM for on-chip optical communication, various building blocks have been demonstrated, such as mode (de)multiplexers [6–9], high-order mode filters , multimode bends , two-mode power splitters , and modal switches. Among them, optical modal switches are capable of exchanging payload data between multimodal channels, which is indispensable in achieving full functionality of on-chip optical networks using MDM technology. A few optical modal switches were reported for MDM-based systems. An optical modal switch using microring resonators has been proposed and experimentally demonstrated . Another class of optical modal switches using Mach–Zehnder interferometers (MZIs), Y-junctions, and multimode interference (MMI) structures with phase shifters has been proposed [14–17]. Nevertheless, dynamic and high-speed optical modal switching remains to be experimentally demonstrated.
In this work, we propose and experimentally demonstrate high-speed and dynamic switching of modes using a novel optical two-mode switch structure, which simultaneously handles both modes by combining a MMI structure with a Y-junction. Similar structures have been used for two-mode (de)multiplexers [6,7]. Here, an additional phase shifter based on a PN junction is used to realize a 50 MHz switching between the fundamental mode (TE0) and the first-order mode (TE1). The MMI structure simultaneously enables high-speed switching and demultiplexing of two modes a step further compared to reported work (e.g., [14–17]). Furthermore, by using a multimode optical splitter and an MMI structure (, 3, and 4), the proposed switching scheme can be scaled for multimode switching.
2. TWO-MODE SWITCH DESIGN
The proposed optical two-mode switch is schematically shown in Fig. 1(a) with a top view shown in Fig. 1(b), while a block diagram of the proposed device is illustrated in Fig. 1(d). It consists of a Y–junction, a phase shifter, and a MMI structure. Waveguide widths, and , enable operation on two modes, simultaneously. The input stem of the Y-junction () supports both fundamental and first-order transverse electric modes (TE0 and TE1), while output branches of the Y-junction () support only the TE0 mode. As such, the TE1 mode is converted to TE0 by splitting its mode profile in two TE0 beams (). Limited by a fabrication technology limitation, a 200 nm gap is required between the two branches of the Y-junction, which introduces additional insertion loss of the order of 0.2 dB based on simulation. The S bend radius of the Y-junction branch is to reduce bending loss. The TE0 mode input into the stem of the Y-junction is split into two TE0 beams () with the same phase propagating through the two branches of the Y-junction, while the TE1 input is split into two TE0 beams () with a phase difference.
The phase shifter, based on a PN junction, employs the free-carrier injection method to realize dynamic and high-speed phase shifting. Although the free-carrier depletion method using a reverse-biased PN junction achieves higher switching speed (up to GHz), a forward-biased PN junction shortens the required length of the phase shifter and lowers the applied bias voltage . To simplify the design and device test, only one branch of the Y-junction is made active to change the phase difference between the two Y-junction branches. Ideally, the two branches should be doped for a balanced structure. The length of the phase shifter () with a lateral PN junction is 267 μm to achieve a (or ) phase shift by applying a bias voltage . A cross section of the junction is shown in Fig. 1(c), where a PN junction is formed at the edge of waveguide, reducing the loss of propagating modes due to carrier injection. The height of the waveguide is 220 nm with p- and n-doped rib heights being 90 nm. Heavily doped regions ( and ) acting as electrodes to connect electrical pads for dynamic switching are 750 nm away from the waveguide edges to further reduce any introduced insertion loss due to carrier injection.
A MMI structure with paired interference effectively and simultaneously couples the TE0 and TE1 modes to two distinct TE0 outputs (Arms A and B of the MMI). To accomplish this, the length of the MMI waveguide () is 23.86 μm, which is half of the beating length between the two modes . Outputs (Arms A and B) are located at from the center, where is the width of the MMI region. To reduce coupling loss, tapered waveguides at the input and output expand modes with widths from 500 nm () to 1.2 μm () over 10 μm (). All parameter values of the MMI structure are determined by an eigenmode expansion (EME) method  by using a commercial tool (Lumerical MODE solutions).
Figures 2(a) and 2(b) show block diagrams and simulated mode propagations along the optical two-mode switch in bar and cross states. The operation principle is as follows. When the phase shifter introduces an extra phase difference of between the two branches of the Y-junction, the TE0 input will go through the MMI structure and output from Arm A as a TE0 mode (), while the TE1 input will output from Arm B of the MMI structure as a TE0 mode (). The routing path () is established as a bar state. When the phase shifter introduces an extra phase difference of between the two branches of the Y-junction, the routing path of the TE0 (TE1) input will be switched to () as a cross state. Figures 2(c) and 2(d) demonstrate simulated transmission spectra in bar and cross states. In this mode behavioral assessment, waveguide propagation loss is not considered. Definitions of insertion loss (IL) and crosstalk (XT) are and , respectively, where , , and are input power and output power in desired and undesired routing paths, respectively. Simulated insertion loss in the bar state is less than 0.3 dB for both TE0 and TE1 inputs over the wavelength range from 1530 nm to 1570 nm with a crosstalk of less than for the TE0 input and for the TE1 input, while insertion loss and crosstalk in the cross state are less than 0.2 dB and , respectively, for both TE0 and TE1 inputs for the same wavelength range. Simulated bar-state extinction ratios (ERs) are greater than 27 dB for both and routing paths within the wavelength range from 1530 nm to 1570 nm, while cross-state ERs are greater than 32 dB for both and routing paths in the same wavelength range. Ripples of curves in the bar state and curves in the cross state are mainly from the MZI effect of the designed structure. For the TE0 input, the two beams in the two branches of the Y-junction are identical TE0 modes from the single TE0 input source; output fringes of the MMI coupler are produced due to the relative phase difference between the two branches of the Y-junction. On the other hand, two beams in the branches, though derived by splitting the same TE1 input, are antiphase from different parts of the TE1 input. Therefore, ripples are not apparent for the curve in the bar state and the curve in the cross state, as shown in Figs. 2(c) and 2(d).
In the intended application, the optical two-mode switch would directly handle TE1 modes on-chip. In the experimental demonstration, a two-mode multiplexer is used to enable the TE1 mode in demonstrating the fabricated optical modal switch. The multiplexer is based on a tapered directional coupler . The width of the tapered waveguide decreases linearly from to along a coupling length () of 92.5 μm (Fig. 3) with a width of a wide bus waveguide designed to 930 nm (), and a coupling gap between waveguides maintains 200 nm (). The bending radius of the S-bend () is set to 50 μm.
3. FABRICATION AND CHARACTERIZATION
The designed optical two-mode switch with a two-mode multiplexer was fabricated in a 220 nm silicon-on-insulator platform. An optical microscopic image of the fabricated optical two-mode switch is shown in Fig. 4 along with other test structures. Light is coupled in and out of the switch through grating coupler arrays with a pitch of 127 μm to match commercial fiber arrays. Electrodes of the two-mode switch are connected out to electrical pads. The electrical pad size is with a separation of 125 μm to match a ground-signal-configured high-frequency 50 ohm terminated probe. The total footprint of the device, including electrical pads and a two-mode multiplexer, is .
To characterize performance of the two-mode switch, input light is adjusted to TE polarization by a polarization controller before coupling to the chip via grating couplers. An electrical probe is positioned on electrical pads to provide forward bias voltages. The 3 dB optical bandwidth of the grating couplers is approximately 30 nm with a fiber-to-grating-coupler coupling loss of approximately 6 dB at the central wavelength of 1550 nm. By subtracting the fiber-to-grating-coupler coupling loss, Figs. 5(a) and 5(b) illustrate measured normalized transmissions of the two-mode switch at a wavelength of 1550 nm as a function of the applied bias voltage. Here, insertion loss of the two-mode multiplexer is included. A test structure, which consists of two identical tapered directional couplers to simultaneously realize a two-mode multiplexer and a demultiplexer, is used to obtain the insertion loss of the two-mode multiplexer. Half of the total insertion loss of the test structure is insertion loss of the two-mode multiplexer, which is approximately 1.5 dB for both TE0-TE0 and TE0-TE1 channels (Fig. 3) in the wavelength range from 1530 nm to 1570 nm. A phase shift voltage () of 0.24 V switches between bar and cross states with a bias () of 0.98 V. Insertion losses of routing path at the bar state and routing path at the cross state are 3.5 dB and 4.2 dB, respectively, while insertion losses of the routing path at the cross state and the routing path at the bar state are 2.6 dB and 4.8 dB, respectively. At the idle state with no bias voltage, insertion losses for the , , , and routing paths are 4.5 dB, 6.5 dB, 5.1 dB, and 6.5 dB, respectively, because the device is not under an appropriate operation situation. Ideally, all insertion losses for the four routing paths in the idle state should be 4.5 dB by including the insertion loss of the two-mode multiplexer. However, due to the two different branches of the Y-junction, the input power cannot be perfectly split into two outputs of the MMI coupler. The switching ER for the TE0 input in the bar state is 14.7 dB, while it is 21.7 dB in the cross state. As for the TE1 input, ERs in bar and cross states are 12.5 dB and 23.1 dB, respectively. These ER values are sufficiently high for data transmission switching . To investigate the bar state ER reduction, Fig. 6(a) illustrates simulated bar state ERs when the phase shifter in one branch of the Y-junction for the optical two-mode switch has different insertion losses (0 dB, 3 dB, and 6 dB). Bar state ERs deteriorate from 34 dB to 10 dB for both and routing paths at a wavelength of 1550 nm when the insertion loss of the phase shifter increases from 0 dB to 6 dB. The reduced bar state ER is attributed to the unbalanced insertion loss of the Y-junction with a phase shifter in only one of the two branches since injected carriers in the phase shifter increase the propagation loss of one branch while the propagation loss of the other branch remains the same. Drawn currents for bar and cross state applied voltages of 0.98 V and 1.22 V are 6.5 mA (6.4 mW) and 22 mA (26.8 mW), respectively.
Figures 5(c) and 5(d) show measured normalized transmission spectra in bar and cross states by including the insertion loss of the two-mode multiplexer. Insertion loss in the bar state is less than 3.5 dB for both TE0 and TE1 inputs over the wavelength range from 1530 nm to 1570 nm with a crosstalk of less than for the TE0 input and for the TE1 input, while the insertion loss and crosstalk in the cross state are less than 5.5 dB and , respectively, for both TE0 and TE1 inputs in the same wavelength range. Measured insertion losses are higher than simulated ones in Figs. 2(c) and 2(d) since the measured ones include additional insertion losses from waveguide, two-mode multiplexer, phase shifter, Y-junction and MMI coupler. The ripples of the curve in Fig. 5(c) and the curve in Fig. 5(d), which are consistent with simulated results in Figs. 2(c) and 2(d), respectively, are mainly from the MZI effect of the designed structure. The measured values of crosstalk for both bar and cross states are worse than the simulated ones in Figs. 2(c) and 2(d) because an unbalanced Y-junction with only one branch doped introduces different insertion losses for the two branches. To confirm, the simulated crosstalk of and routing paths in the cross state with different phase shifter insertion losses is demonstrated in Fig. 6(b). The crosstalk is deteriorated by 21 dB at a wavelength of 1550 nm when the insertion loss of the phase shifter increases from 0 dB to 6 dB. Figure 6 demonstrates that a balanced Y-junction, which can be achieved by doping both branches of the Y-junction identically, would lead to higher ER and lower crosstalk.
4. SWITCHING PERFORMANCE
Switching performance of the two-mode switch is measured using the experimental setup shown in Fig. 7. A continuous wave laser light at a wavelength of 1550 nm propagates through an amplitude modulator driven by a pulse-pattern generator (PPG). A generated 10 Gb/s pseudo-random bit sequence (PRBS) optical signal is first amplified by an erbium-doped fiber amplifier (EDFA) and set to TE polarization by a polarization controller before coupling into an input port of the optical two-mode switch. Light output from the switch chip is amplified by another EDFA, filtered by an optical bandpass filter with an optical bandwidth of 0.7 nm, and followed by a variable optical attenuator before being injected into a photodetector (PD). An optical power meter is used to monitor optical power received by the PD through a optical coupler with a 10/90 split ratio. A clock synthesizer is used to provide an external clock to a PPG and to simultaneously synchronize an error detector and trigger a digital communication analyzer for recording bit error rates (BERs) and eye diagrams. A 50 MHz electrical square-wave gating signal is applied to the optical two-mode switch for validating dynamic switching performance.
The measured dynamic switching response to the 50 MHz gating signal is shown in Fig. 8 for the , , , and routing paths, respectively. Open eye diagrams of 10 Gb/s payload are observed from both bar and cross routing paths. The switching time (10%–90%) is less than 2.5 ns, enabling a potential switching speed of hundreds of megahertz.
Measured BERs as functions of the average received total optical power at the photodetector are shown in Fig. 9. The 12.5 Gb/s error detector (manufactured by Anritsu) has a required setup time of approximately 2 μs, limiting the BER measurement on packets to a gating speed of 200 kHz. Reference ungated and back-to-back (B2B) measurements were taken. For the B2B measurement, the two-mode optical switch is replaced by an attenuator with a loss corresponding to the insertion loss associated with the and channels. Different BER curves for 200 kHz switching, no switching, and back-to-back scenarios are within 0.5 dB at BER of , indicating negligible power penalties from the optical two-mode switch when it switches.
In summary, we propose and experimentally demonstrate a novel optical two-mode switch with a relatively high switching speed for mode-division multiplexing system applications. The optical two-mode switch comprises a Y-junction, a phase shifter, and a MMI structure. The dynamic switching is achieved by controlling a PN-junction-based phase shifter in one branch of a Y-junction. The achieved switching time is less than 2.5 ns, while switching extinction ratios for bar and cross states range from 12.5 dB to 23.1 dB at a wavelength of 1550 nm. Open eye diagrams of the switched 10 Gb/s payload are observed with less than a 0.5 dB power penalty in BER measurements, indicating good switching performance. The high-speed optical two-mode switch has potential for future high-speed multimode-based optical networks with nanosecond-scale switching times.
Canada Research Chairs; Discovery Grants Program from Natural Sciences and Engineering Research Council of Canada (NSERC).
The authors would like to acknowledge CMC Microsystems for subsidizing and managing silicon photonics multiproject wafer (MPW) fabrication at the Institute of Microelectronics (IME) in Singapore.
1. W. N. Ye and Y. Xiong, “Review of silicon photonics: history and recent advances,” J. Mod. Opt. 60, 1299–1320 (2013). [CrossRef]
2. A. Liu, L. Liao, Y. Chetrit, J. Basak, H. Nguyen, D. Rubin, and M. Paniccia, “Wavelength division multiplexing based photonic integrated circuits on silicon-on-insulator platform,” IEEE J. Sel. Top. Quantum Electron. 16, 23–32 (2010). [CrossRef]
3. C. Xie, “WDM coherent PDM-QPSK systems with and without inline optical dispersion compensation,” Opt. Express 17, 4815–4823 (2009). [CrossRef]
4. Y. Xiong, D. X. Xu, J. H. Schmid, P. Cheben, S. Janz, and W. N. Ye, “Fabrication tolerant and broadband polarization splitter and rotator based on a taper-etched directional coupler,” Opt. Express 22, 17458–17465 (2014). [CrossRef]
5. B. Milivojevic, C. Raabe, A. Shastri, M. Webster, P. Metz, S. Sunder, B. Chattin, S. Wiese, B. Dama, and K. Shastri, “112 Gb/s DP-QPSK transmission over 2427 km SSMF using small size silicon photonics IQ modulator and low power CMOS driver,” in Optical Fiber Communication Conference, OSA Technical Digest (online) (2013), paper OTh1D.1.
6. Y. Li, C. Li, C. Li, B. Cheng, and C. Xue, “Compact two-mode (de)multiplexer based on symmetric Y-junction and multimode interference waveguides,” Opt. Express 22, 5781–5786 (2014). [CrossRef]
7. F. Guo, D. Lu, R. Zhang, H. Wang, and C. Ji, “A two-mode (de)multiplexer based on multimode interferometer coupler and Y-junction on InP substrate,” IEEE Photon. J. 8, 2700608 (2016). [CrossRef]
8. 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, 3743–3746 (2016). [CrossRef]
9. N. Riesen and J. D. Love, “Design of mode-sorting asymmetric Y-junctions,” Appl. Opt. 51, 2778–2783 (2012). [CrossRef]
10. X. Guan, Y. Ding, and L. H. Frandsen, “Ultra-compact broadband higher order-mode pass filter fabricated in a silicon waveguide for multimode photonics,” Opt. Lett. 40, 3893–3896 (2015). [CrossRef]
11. D. Dai, “Multimode optical waveguide enabling microbends with low inter-mode crosstalk for mode-multiplexed optical interconnects,” Opt. Express 22, 27524–27534 (2014). [CrossRef]
12. Y. Luo, Y. Yu, M. Ye, C. Sun, and X. Zhang, “Integrated dual-mode 3 dB power coupler based on tapered directional coupler,” Sci. Rep. 6, 23516 (2016). [CrossRef]
13. B. Stern, X. Zhu, C. Chen, L. Tzuang, J. Cardenas, K. Bergman, and M. Lipson, “On-chip mode-division multiplexing switch,” Optica 2, 530–535 (2015). [CrossRef]
14. C. Sun, Y. Yu, G. Chen, and X. Zhang, “Integrated switchable mode exchange for reconfigurable mode-multiplexing optical networks,” Opt. Lett. 41, 3257–3260 (2016). [CrossRef]
15. R. Imansyah, T. Tanaka, L. Himbele, H. Jiang, and K. Hamamoto, “Electrically controlled optical-mode switch for fundamental mode and first order mode,” Jpn. J. Appl. Phys. 55, 08RB06 (2016). [CrossRef]
16. H. Xiao, L. Deng, G. Zhao, Z. Liu, Y. Meng, X. Guo, G. Liu, S. Liu, J. Ding, and Y. Tian, “Optical mode switch based on multimode interference couplers,” J. Opt. 19, 025802 (2016). [CrossRef]
17. D. Melati, A. Alippi, and A. Melloni, “Reconfigurable photonic integrated mode (de)multiplexer for SDM fiber transmission,” Opt. Express 24, 12625–12634 (2016). [CrossRef]
18. G. T. Reed, G. Mashanovich, F. Y. Gardes, and D. J. Thomson, “Silicon optical modulators,” Nat. Photonics 4, 518–526 (2010). [CrossRef]
19. D. F. G. Gallagher and T. P. Felici, “Eigenmode expansion methods for simulation of optical propagation in photonics: pros and cons,” Proc. SPIE 4987, 69–82 (2003). [CrossRef]
20. D. Nikolova, S. Rumley, D. Calhoun, Q. Li, R. Hendry, P. Samadi, and K. Bergman, “Scaling silicon photonic switch fabrics for data center interconnection networks,” Opt. Express 23, 1159–1175 (2015). [CrossRef]
21. Y. Ding, J. Xu, F. Da Ros, B. Huang, H. Ou, and C. Peucheret, “On-chip two-mode division multiplexing using tapered directional coupler-based mode multiplexer and demultiplexer,” Opt. Express 21, 10376–10382 (2013). [CrossRef]
22. B. G. Lee, A. Biberman, J. Chan, and K. Bergman, “High-performance modulators and switches for silicon photonic networks-on-chip,” IEEE J. Sel. Top. Quantum Electron. 16, 6–22 (2010). [CrossRef]