Open Access
Add to BibSonomyAbstract
A polarization-independent 2 × 2 switch based on silicon-wire waveguides has been realized with a compact size of 600 × 500 μm2. Polarization-independent operation was achieved with a polarization-diversity technique which implements polarization splitters, TE-TM intersections, and Mach–Zehnder switches. The extinction ratios of the 2 × 2 switch for TE, TM, and a mixed polarization at a wavelength of 1550 nm were measured to be larger than 30 dB, 25 dB, and 30 dB, respectively. The measured switching powers for the TE and TM polarizations were 25 and 55 mW, respectively. The measured polarization-dependent loss was lower than 1 dB. The differential group delay (DGD) between the TE and TM modes was also evaluated using the Mueller matrix method, which was in good agreement with the values estimated from the path lengths for each mode. A path-length–compensated switch was fabricated, whose DGDs for all paths were indeed as small as ~2 ps, mainly from the access waveguides. The switch could provide an important route to develop ultra-compact polarization-independent integrated circuits based on silicon-wire waveguides.
© 2014 Optical Society of America
1. Introduction
A large–port-number optical matrix switch is essential for the next-generation optical networks, including flexible multi-degree node systems [1], densely integrated computing systems and datacenters [2], and dynamic optical path networks [3]. Silicon-based optical waveguides are expected to play a crucial role for realizing more compact optical matrix switches with ultra-dense integration of optical elements compared to the conventional silica-based optical waveguides. Silicon waveguides make optical devices very compact exploiting the strong confinement factor owing to the high refractive index contrast between Si (~3.48) and SiO2 (~1.444). On the other hand, they introduce a large polarization dependence to silicon photonics devices, including polarization mode dispersion, polarization-dependent loss (PDL), and polarization dependence of the operating wavelength, which is one of the big challenges remaining in silicon photonics.
One method to overcome the problem of polarization dependence is to make waveguides optically isotropic. Using 1.5-μm-height rib waveguides with a large modal field, a polarization-independent thermo-optic (TO) switch with an extinction ratio of 25 dB has been demonstrated [4]. However, channel-type waveguides using much thinner silicon-on-insulator (SOI) wafers (~200 nm) are more important for realizing monolithic integration with electronic circuits [2, 5] as well as a much smaller footprint [6]. In such silicon “wire” waveguides, achieving isotropy in the field distribution, effective index, and group index is much more difficult [7, 8]. Another method is the polarization-diversity technique [9–14], where an input with arbitrary polarization is split into the TE and TM polarizations by a polarization splitter and one polarization is converted to the other by a polarization rotator. With this approach, the development of integrated optical filters, such as arrayed waveguide gratings (AWGs) [10], ring resonators [11–13], and more complicated coherent receivers [14], has been reported.
In this paper, we demonstrate a polarization-diversity TO 2 × 2 switch based on silicon-wire waveguides as a building block for ultra-compact polarization-independent large-scale optical matrix switches, which was preliminary presented in a conference paper [15]. The multimode interference (MMI) type intersections for a TE propagating waveguide and a TM propagating waveguide is the key component that allows us to configure the switch without a polarization rotator. The intersection is carefully designed since sophisticated waveguide intersections are crucial to scale toward a large port number with the overall crosstalk kept sufficiently low. The polarization-independent switching operation, the PDL, and the differential group delay (DGD) between the TE and TM modes are evaluated.
2. Device configuration and component design
Figure 1 shows the configuration of the polarization-diversity 2 × 2 switch, which consists of spot-size converters (SSCs), polarization splitters (PSs), TE-TM intersections (ISs), and Mach-Zehnder (MZ) switches for the TE and TM polarizations. All the components are based on silicon-wire waveguides with a size of 430 nm x 220 nm. The input signals, coupled through the SSCs, are divided by the PSs into TE and TM polarization signals; these are incident to the corresponding MZ switches designed for the TE and TM modes, after propagating through the ISs. They are switched by the two polarization-dependent MZ switches and sent to the two output ports after passing through the same ISs and PSs. Our polarization-diversity scheme consists of the TE and TM switches with different designs. However, it is not so complicated to have just two designs for each component. Also, the fabrication process is completely common for the TE and TM switches. The advantage of our polarization-diversity scheme, compared to the previous scheme [9–14], is to avoid use of polarization rotators, whose polarization extinction ratio is not as good as that for the polarization splitters. Therefore, our polarization-diversity scheme will be more advantageous in terms of device performance.
Fig. 1 Configuration of the polarization-diversity switch. SSC: spot-size converter, PS: polarization splitter, IS: TE-TM intersections, SW: Mach–Zehnder switch. All the components are based on silicon-wire waveguides with a size of 430 nm x 220 nm.
For the PSs, we used a directional coupler whose design is similar to the one reported by NTT [16]. The ISs are the key component that allows our diversity switch to be configured without polarization rotators. Their design is based on the tilted MMI structure that we proposed in [17], which demonstrated crosstalk as low as –40 dB for both modes and an insertion loss as low as 0.3 dB. Each of the two MZ switches consists of two 3-dB directional couplers and TO phase shifters, which induce a change of refractive index by applying current to the heater on the waveguide. The TO phase shifter is suitable for circuit switching [1, 6] owing to its low excessive loss. The response time of our TO switches was previously measured about 30 μs, which could be the case also in this work. In the SSCs for low-loss coupling at each end of the waveguides, the mode expansion is achieved only by a narrowed core [18].
3. Fabrication and measurement
For the device fabrication, a commercial SOI with a 220-nm–thick silicon layer on a 3-μm–thick buried oxide layer was used. A 50-nm–thick SiO2 layer was deposited as a hard mask on the SOI wafer by plasma chemical vapor deposition (PCVD). The mask pattern was drawn by an electron-beam lithography system. We used simple reactive ion etching (RIE) with a sulfur hexafluoride (SF6)-trifluoromethane (CHF3) gas mixture to transfer the resist mask pattern to the 50-nm–thick SiO2 layer and the silicon layer, followed by PCVD for the 2-μm–thick SiO2 cladding layer. Subsequently, 70-μm–long, 5-μm–wide, and 50-nm–thick Pt heaters were deposited on the SiO2 cladding at the MZ interferometer’s (MZI) arms for the TO effect, as shown in Fig. 2. Finally, Au electrodes were deposited for access to the heaters. The resistance of the heaters was measured to be 64 Ω. An optical microscope image of the fabricated device is shown in Fig. 3. The total size of the switch was 600 × 500 μm2.
Fig. 2 Cross-sectional diagram of the waveguide with the Pt heater (70-μm long, 5-μm wide, and 50-nm thick, 64Ω). Au electrodes were deposited for access to the heaters.
Fig. 3 Microscopic image of the polarization-diversity switch. The total size of this switch is 600 × 500 μm2.
The light generated from a wavelength-tunable laser was introduced to a single-mode optical fiber (SMF) and then coupled to the waveguide on the switch chip with a focusing lens module. The input lens module had a polarizer which makes the input polarization TE, TM, or an arbitrary orientation by adjusting the angle of the polarizer. Coupling losses of 4.5 dB and 3.4 dB were measured for the TE and TM polarizations, respectively. The output light from the chip was coupled to an SMF with a lensed fiber and detected by an optical power meter and an optical spectrum analyzer to evaluate the switching operation and its wavelength and polarization dependences. From cutback measurements on reference waveguides, the propagation loss of our silicon-wire waveguide was estimated to be 4.3 (3.5) dB/cm for TE (TM) polarization.
The DGD between the TE and TM polarizations for each path was measured by an optical component analyzer (Agilent N7788B), based on the Mueller matrix method [19]. The DGD or the first-order polarization mode dispersion at each frequency is described by dϕ/dω, where ϕ is the rotation angle made by the frequency dispersion of the output polarization state in the Stokes space. The value of dϕ/dω is numerically determined using the elements of dM/dω · M−1 at each frequency, where M is the Mueller matrix of the device under test. Therefore, the focus of the measurement in the optical component analyzer is to obtain the frequency dispersion of M, using three fixed input polarizations and the three corresponding output polarizations at each frequency, scanned with a step of dω. For this purpose, the measurement system was equipped with a wavelength-tunable light source, a polarization controller, and a polarimeter. Note that the Mueller matrix method can be applied even to a non-unitary system, i.e., a system with a PDL, thus the DGD as well as the PDL are simultaneously determined.
4. Results and discussion
The transmittances at the bar and cross ports when the TE or TM light was introduced to the input-2 port are plotted in Fig. 4 as a function of the power applied to the heater. The wavelength was fixed at 1550 nm. The switch state could be controlled by changing the heater power. The switch takes the “cross (bar)” state at a power of 40 (15) mW for TE polarization, whereas it takes the “cross (bar)” state at 30 (95) mW for TM polarization. It should also be noted that the power required to switch between the “cross” and the “bar” states is larger in the TM case than in the TE case. This is a consequence of the lower effective index of the TM mode. Since the TO coefficient is much larger in Si than in SiO2, the phase shift at a specific injected power is smaller for the TM mode with the small confinement to the silicon wire. The diversity circuit gives a polarization-independent “cross” or “bar” connection when both the TE- and TM-MZ switches take their “cross” or “bar” states. Figure 4 illustrates the operation conditions.
Fig. 4 Transmittance at 1550 nm as a function of applied power for (a) TE and (b) TM polarization. The switch takes the “cross (bar)” state at a power of 40 (15) mW for TE polarization, whereas it takes the “cross (bar)” state at 30 (95) mW for TM polarization. The insertion loss of the diversity switch was evaluated to be 13 (12.1) dB for TE (TM) polarization, resulting in a PDL of 0.9 dB.
From the above results, the insertion loss of the diversity switch was evaluated to be 13 (12.1) dB for TE (TM) polarization. Thus, the measured PDL was 0.9 dB. The on-chip loss was estimated to be 4 (5.3) dB for TE (TM) polarization by subtracting the coupling loss, whose breakdown was estimated as 0.6 (0.6) dB for the ISs, 1.3 (1.1) dB for waveguide propagation, and 2.7 (4.2) dB for some excess loss that may be generated in the PSs and the 3-dB couplers in the MZ switches. The large insertion loss can be significantly reduced by introducing a SSC with a lower loss [20].
The transmittance spectra of the polarization-diversity TO switch for TE, TM, and a mixed polarization are shown in Figs. 5(a), 5(b), and 5(c), respectively. As seen in Figs. 5(a) and 5(b), the cross-state extinction ratios, i.e., T22/T21, at 1550 nm were 34 dB and 26 dB for the TE and TM polarizations, respectively. The extinction ratio for TM polarization was smaller than that for TE polarization. In addition, the wavelength showing the highest extinction (30 dB at 1555 nm) for TM polarization shifted from the design wavelength of 1550 nm, in contrast to that for TE polarization. From these results, the MZ switch for TM polarization is expected to have larger fabrication errors in the directional couplers. Figure 5(c) shows the transmission spectra for an input linear polarization that forms an angle of approximately 45 degrees to the TE and TM polarization directions. An extinction ratio as large as 30 dB was obtained for the bar state at 1550 nm. A better extinction in the bar state than in the cross state is a commonly observed trend in MZ switches, since the interference visibility at the cross port is in principle always unity or in practice insensitive to the error of the 3-dB splitting ratios.
Fig. 5 Transmittance as a function of wavelength for (a) TE, (b) TM, and (c) mixed polarization. The cross-state extinction ratios, i.e., T22/T21, for the TE and TM polarizations at 1550 nm were 34 dB and 26 dB, respectively.
The splitting ratios of the PSs for the TE and TM polarizations were measured to be about −15 dB and −25 dB, respectively. Since the leaked components at the first PSs reach the discarded ports at the second PSs, the residual leaked components at the output ports of the second PSs experience twice the amount of each extinction. In other words, the limitations for the extinction ratio imposed by the two PSs will be 30 dB and 50 dB for TE and TM polarizations, respectively. The measured extinction ratio of 34 dB for TE polarization is comparable to the limitation of 30 dB, indicating that the leakage in the PSs limits the extinction ratio for TE polarization. On the other hand, the measured extinction ratio of 30 dB at 1555 nm for TM polarization is much smaller than 50 dB, indicating that the leakage in the TM-MZIs, caused by some fabrication error, limits the extinction ratio for TM polarization. Note that the crosstalk at the ISs is as low as −40 dB and it is not the limiting factor in this case. For better extinction, a PS based on cascaded directional couplers, which has demonstrated a splitting ratio larger than −30 dB, is a good candidate [21]. With this method, the extinction ratio in our diversity switch will not be limited by the total leakage (60 dB) of the PSs. Regarding broadband operation, the V-shaped spectral response originates from the wavelength dependence of the 3-dB couplers in the MZ switches. Operation over a broader wavelength range would be possible by introducing a wavelength-insensitive design for the 3-dB couplers [22, 23].
Silicon-wire waveguides have strong birefringence between the TE and TM modes, which introduces a significant temporal walk-off, characterized by the DGD. It is important to evaluate the DGD in our polarization-diversity switch in order to extend it to a large-scale matrix switch. The group delay is characterized by ngL/c, where ng and L represent the group index and the physical waveguide length, respectively. For our silicon-wire waveguide with cross-sectional dimensions of 430 × 220 nm2, the ng for the TE and TM modes was calculated to be 4.18 and 3.44 at 1550 nm, respectively. From these values, the DGDs of four possible paths (1 to 1’, 1 to 2’, 2 to 1’, and 2 to 2’), including the input and output access waveguides (see Fig. 6), were estimated from the designed waveguide lengths, as shown in Table 1. For comparison, we fabricated another sample with equalized path lengths for all paths, also listed in Table 1. The microscope image for the new sample is presented in Fig. 7. The estimated DGDs ranged from 0.4 ps to 0.7 ps for the path-length–compensated switch and from 4.6 ps to 10.2 ps for the uncompensated switch. The values for the compensated switch are sufficiently small even for 40 Gbps transmission, and can be further smaller by precise adjustment of the lengths. Note that the physical length differences between the compensated and uncompensated switches in Table 1 originated from the length differences of the access waveguides due to the different chip sizes. The longer access waveguides for the uncompensated switch resulted in an offset DGD of ~3ps. It must also be noted that the access waveguides in the compensated switch had dimensions of 220 × 220 nm2 in order to eliminate the offset DGD from the access waveguides. This access waveguides have a larger propagation loss due to the small dimensions. This is why the physical lengths were made shorter for the compensated switch.
Fig. 6 Configuration of the four possible paths of 1 to 1’, 1 to 2’, 2 to 1’, and 2 to 2’ in the diversity switch. In this simple figure, the TM path has a longer physical length for 1-1’ while the TE path has a longer physical length for 2-2’.
Table 1. Calculated physical length and DGD for different paths, including the input and output access waveguides.
Fig. 7 Microscopic image of the path-length–compensated polarization-diversity switch. The path for TM was lengthened to compensate the small group index for TM, c.f. Fig. 3.
Figure 8 shows the measured DGDs without and with compensation as a function of wavelength, which is varied from 1520 nm to 1600 nm. For the switch without compensation, the measured values at 1550 nm were 4.1 ps, 6.9 ps, 7.4 ps, and 10.2 ps for the paths 1-1’, 1-2’, 2-1’, and 2-2’, respectively, which are in good agreement with the estimated values presented in Table 1. For the switch with compensation, the measured values at 1550 nm were 2.3 ps, 2.7 ps, 2.6 ps, and 3.2 ps for the paths 1-1’, 1-2’, 2-1’, and 2-2’, respectively; these are indeed reduced by the path length equalization, however they are larger than the estimated values in Table 1. These residual DGDs are attributed to fabrication errors in the access waveguides with dimensions of 220 × 220 nm2. An error of +/− 10 nm in the thickness and/or width of the waveguides results in a DGD of 1 ps or larger, which is possible in our fabrication process. Such a residual DGD from the access waveguides would not increase further even when extending our diversity switch to a large scale matrix switch, since the access waveguides are placed only at each input and output port.
Fig. 8 Measured DGD of the 1 to 1’, 1 to 2’, 2 to 1’, and 2 to 2’ paths in the diversity switch (a) without compensation and (b) with compensation. The measured values at 1550 nm for the paths 1-1’, 1-2’, 2-1’, and 2-2’ were 4.1 ps, 6.9 ps, 7.4 ps, and 10.2 ps without compensation, 2.3 ps, 2.7 ps, 2.6 ps, and 3.2 ps with compensation, respectively.
As we have shown, the DGD estimation from the waveguide lengths agrees well with the experiment, and the compensation of the DGD has been accomplished simply by the adjustment of the path length because each polarization mode independently propagates in the diversity circuit. Future experimental work will mainly include data transmission at 40 Gbps or higher speeds for an extended-port-number matrix switch.
5. Conclusions
A polarization-independent 2 × 2 switch based on silicon-wire waveguides has been realized with a compact size of 600 × 500 μm2. Polarization-independent operation was achieved with a polarization-diversity technique implementing PSs, TE-TM ISs, and MZ switches. The measured switching powers for the TE and TM polarizations were 25 and 55 mW. The extinction ratios of the 2 × 2 switch for the TE and TM modes at a wavelength of 1550 nm were measured to be larger than 30 dB and 25 dB, respectively. The extinction ratio for a mixed polarization was measured to be larger than 30 dB at the same wavelength. The measured polarization-dependent loss was lower than 1 dB. The DGD between the TE and TM modes, evaluated using the Mueller matrix method, was in good agreement with the calculated values and thus can be easily compensated for by adjusting the waveguide lengths for the two modes. The switch could provide an important route to polarization-independent integrated circuits based on silicon-wire waveguides.
Acknowledgments
This study was supported in part by the Project for Developing Innovation Systems of MEXT, Japan.
References and links
1. S. Nakamura, S. Takahashi, M. Sakauchi, T. Hino, M. Yu, and G. Lo, “Wavelength selective switching with one-chip silicon photonic circuit including 8 × 8 matrix switch,” Optical Fiber Communication Conference, OSA Technical Digest, paper OTuM.2 (2011). [CrossRef]
2. B. G. Lee, A. Rylyakov, W. M. J. Green, S. Assefa, C. W. Baks, R. Rimolo-Donadio, D. Kuchta, M. Khater, T. Barwicz, C. Reinholm, E. Kiewra, S. Shank, C. Schow, and Y. A. Vlasov, “Four- and Eight-Port Photonic Switches Monolithically Integrated with Digital CMOS Logic and Driver Circuits,” Optical Fiber Communication Conference, OSA Technical Digest, paper PDP5C.3 (2013). [CrossRef]
3. S. Namiki, T. Kurosu, K. Tanizawa, J. Kurumida, T. Hasama, H. Ishikawa, T. Nakatogawa, M. Nakamura, and K. Oyamada, “Ultrahigh-definition video transmission and extremely green optical networks,” IEEE J. Sel. Top. Quantum Electron. 17(2), 446–457 (2011). [CrossRef]
4. S. Nakamura, S. Takahashi, I. Ogura, J. Ushida, K. Kurata, T. Hino, H. Takeshita, A. Tajima, M.-B. Yu, and G.-Q. Lo, “High extinction ratio optical switching independently of temperature with silicon photonic 1×8 switch, ” Optical Fiber Communication Conference, OSA Technical Digest, paper OTu2I.3 (2012).
5. G. W. Cong, T. Matsukawa, T. Chiba, H. Tadokoro, M. Yanagihara, M. Ohno, H. Kawashima, H. Kuwatsuka, Y. Igarashi, M. Masahara, and H. Ishikawa, “Large current MOSFET on photonic silicon-on-insulator wafers and its monolithic integration with a thermo-optic 2 × 2 Mach-Zehnder switch,” Opt. Express 21(6), 6889–6894 (2013). [CrossRef] [PubMed]
6. K. Suzuki, K. Tanizawa, T. Matsukawa, G. W. Cong, S.-H. Kim, S. Suda, M. Ohno, T. Chiba, H. Tadokoro, M. Yanagihara, Y. Igarashi, M. Masahara, S. Namiki, and H. Kawashima, “Ultra-compact 8 × 8 strictly-non-blocking Si-wire PILOSS switch,” Opt. Express 22(4), 3887–3894 (2014). [CrossRef] [PubMed]
7. P. Dumon, G. Priem, L. R. Numes, W. Bogaerts, D. V. Thourhout, P. Bienstman, T. K. Liang, M. Tsuchiya, P. Jaenen, S. Beckx, J. Wouters, and R. Baets, “Linear and nonlinear nanophotonic devices based on silicon-on-insulator wire waveguides,” Jpn. J. Appl. Phys. 45(8B), 6589–6602 (2006).
8. E. Dulkeith, F. Xia, L. Schares, W. M. J. Green, and Y. A. Vlasov, “Group index and group velocity dispersion in silicon-on-insulator photonic wires,” Opt. Express 14(9), 3853–3863 (2006). [CrossRef] [PubMed]
9. Q. Wang and J. Yao, “A high speed 2x2 electro-optic switch using a polarization modulator,” Opt. Express 15(25), 16500–16505 (2007). [CrossRef] [PubMed]
10. W. Bogaerts, D. Taillaert, P. Dumon, D. Van Thourhout, R. Baets, and E. Pluk, “A polarization-diversity wavelength duplexer circuit in silicon-on-insulator photonic wires,” Opt. Express 15(4), 1567–1578 (2007). [CrossRef] [PubMed]
11. T. Barwicz, M. R. Watts, M. A. Popović, P. T. Rakich, L. Socci, F. X. Kärtner, E. P. Ippen, and H. I. Smith, “Polarization-transparent microphotonic devices in the strong confinement limit,” Nat. Photonics 1(1), 57–60 (2007). [CrossRef]
12. H. Fukuda, K. Yamada, T. Tsuchizawa, T. Watanabe, H. Shinojima, and S. Itabashi, “Silicon photonic circuit with polarization diversity,” Opt. Express 16(7), 4872–4880 (2008). [CrossRef] [PubMed]
13. J. Zhang, H. Zhang, S. Chen, M. Yu, G. Q. Lo, and D. L. Kwong, “A tunable polarization diversity silicon photonics filter,” Opt. Express 15(25), 16500–16505 (2007). [PubMed]
14. C. R. Doerr, P. J. Winzer, Y. K. Chen, S. Chandrasekhar, M. S. Rasras, L. Chen, T. Y. Liow, K. W. Ang, and G.-Q. Lo, “Monolithic polarization and phase diversity coherent receiver in silicon,” J. Lightwave Technol. 28(4), 520–525 (2010). [CrossRef]
15. S.-H. Kim, Y. Shoji, G. Cong, H. Kawashima, T. Hasama, and H. Ishikawa, “Polarization-diversity 2 × 2 switch with silicon-wire waveguide,” European Conference and Exhibition on Optical Communication, Technical Digest, paper We4B.5 (2013).
16. H. Fukuda, K. Yamada, T. Tsuchizawa, T. Watanabe, H. Shinojima, and S. Itabashi, “Ultrasmall polarization splitter based on silicon wire waveguides,” Opt. Express 14(25), 12401–12408 (2006). [CrossRef] [PubMed]
17. S.-H. Kim, G. Cong, H. Kawashima, T. Hasama, and H. Ishikawa, “Tilted MMI crossings based on silicon wire waveguide,” Opt. Express 22(3), 2545–2552 (2014). [CrossRef] [PubMed]
18. V. R. Almeida, R. R. Panepucci, and M. Lipson, “Nanotaper for compact mode conversion,” Opt. Lett. 28(15), 1302–1304 (2003). [CrossRef] [PubMed]
19. H. Dong, P. Shum, M. Yan, J. Q. Zhou, G. X. Ning, Y. D. Gong, and C. Q. Wu, “Generalized Mueller matrix method for polarization mode dispersion measurement in a system with polarization-dependent loss or gain,” Opt. Express 14(12), 5067–5072 (2006). [CrossRef] [PubMed]
20. T. Shoji, T. Tsuchizawa, T. Watanabe, K. Yamada, and H. Morita, “Low loss mode size converter from 0.3um square Si wire waveguides to singlemode fibres,” Electron. Lett. 38(25), 1669–1670 (2002). [CrossRef]
21. L. Chen, C. R. Doerr, and Y.-K. Chen, “Compact polarization rotator on silicon for polarization-diversified circuits,” Opt. Lett. 36(4), 469–471 (2011). [CrossRef] [PubMed]
22. K. Jinguji, N. Takato, A. Sugita, and M. Kawachi, “Mach-Zehnder interferometer type optical wavelength-flattened coupling ratio,” Electron. Lett. 26(17), 1326–1327 (1990). [CrossRef]
23. J. Van Campenhout, W. M. J. Green, and Y. A. Vlasov, “Design of a digital, ultra-broadband electro-optic switch for reconfigurable optical networks-on-chip,” Opt. Express 17(26), 23793–23808 (2009). [CrossRef] [PubMed]
