Open Access
Add to BibSonomy
Abstract
In this work, we proposed and experimentally demonstrated a compact and low polarization-dependent silicon waveguide crossing based on subwavelength grating multimode interference couplers. The subwavelength grating structure decreases the effective refractive index difference and shrinks the device footprint. Our designed device is fabricated on the 220-nm SOI platform and performs well. The measured crossing is characterized with low insertion loss (< 1 dB), low polarization-dependence loss (< 0.6 dB), and low crosstalk (< -35 dB) for both TE and TM polarizations with a compact footprint of 12.5 μm × 12.5 μm.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
To meet the increasing demand for ultra-high link capacity of optical interconnects, various silicon-based on-chip multiplexing technologies have been proposed, such as wavelength-division-multiplexing (WDM), mode-division-multiplexing (MDM), and polarization-division-multiplexing (PDM) [1–4]. PDM technology is attractive as it can double the optical link capacity with dual polarizations and compatible with other multiplexing technologies. However, most silicon photonics devices are polarization sensitive and need additional polarization splitters and rotators [5–7]. Hence, the polarization-insensitive photonic devices are attracting much more attention. The spatial waveguide crossings are inevitable for the large-scale planar optical interconnect system [8,9]. Efficient optical vias for multiple layers are difficult to be implemented on the SOI platform due to their limitations of optical mode coupling and fabrication cost [10]. A polarization-insensitive silicon waveguide crossing is then an essential building block for PMD optical interconnect systems.
Tremendous efforts have been devoted to dealing with the polarization sensitivity of silicon waveguide crossing. The shaped taper polarization-insensitive waveguide crossing has the lowest insertion loss to around 0.2 dB for both polarizations on 250 nm SOI platform. However, it is compromised with sophisticated taper profile and high fabrication cost [11]. Conventional multimode interference (MMI) couplers are utilized to obtain low polarization sensitivity. The simulated insertion loss is around 0.65 (0.73) dB for TE (TM) polarizations with 23×23 μm2 footprint on 220 nm SOI platform. It has easy fabrication process but the device footprint is too large. In this paper, a compact and low polarization dependent silicon waveguide crossing based on subwavelength grating (SWG) MMI couplers is proposed. The simulated crossing has low insertion loss (< 0.7 dB), low polarization-dependence loss (< 0.2 dB), and low crosstalk (< -35 dB) for both TE and TM polarizations. The MMI couplers are utilized to focus the optical signal at the intersection region according to the self-imaging effects. And the SWG structure can tune the effective refractive index of different modes in the MMI waveguide to achieve the polarization-insensitivity of the device. The SWG-based MMI crossing needs an extra etching step but can obtain a much smaller device footprint. Our proposed crossing design has great potential to be used in the advanced PDM silicon photonic circuits.
2. Design and operation principle
The polarization-insensitive waveguide crossing based on SWG MMI couplers on a 220-nm SOI platform is illustrated in Fig. 1(a). The device comprises four single-mode input/output channels and two orthogonal SWG-assisted MMI couplers. The cross-section of input and output strips have a width of 500 nm as shown in Fig. 1(b). The MMI coupler has a width of 2 μm to support the high-order optical modes and a 2 μm × 2 μm silicon square is placed at the intersection region. The SWG structures with pitch Λ and length of remaining blocks after the etching process $\alpha $ are imbedded into the two orthogonal MMI couplers. The SWG taper waveguides establish the connection between the input/output strips and the MMI couplers with the same pitch and duty cycle as given in Fig. 1(c). To prevent the significant back-reflection at the interface, the transition taper reduces linearly with its waveguide width to fulfill the effective index matching conditions [12]. The smallest tip width of the transition taper in SWG taper is set to be 100 nm under fabrication considerations.
Fig. 1. (a) Schematic diagram of the proposed silicon waveguide crossing based on SWG MMI couplers. (b) The cross-section of the silicon waveguide on SOI platform. (c) The enlarged view of the SWG taper.
The dimension of a silicon wire is typically 500 nm × 220 nm and this rectangular waveguide can only support the fundamental optical modes [13]. The field distribution of TE mode is parallel to the substrate while that of TM mode is along a perpendicular direction. Compared with the TM mode, TE mode has stronger optical confinement and higher effective refractive index. Hence, the structural birefringence of the silicon waveguide is severe and the high polarization dependence sets brings lots of difficulties for its practical PMD applications. The MMI couplers have become popular due to its attractive properties of low loss, low wavelength sensitivity, and easy fabrication [14,15]. The principle of the MMI couplers is the self-imaging effect which enables the launched optical pattern to be replicated at periodic intervals in the multimode waveguide. For the silicon waveguide crossing based on MMI couplers, the optical signal gets focused at the central crossing region with little scattering loss. The length of the MMI section is often chosen as twice of the beat length. The beat length can be given by the following formula [16,17]
The subwavelength grating structures are periodic arrangements of different materials with a period that is shorter than the wavelength of light. In deep-subwavelength regime, the SWG silicon waveguide can be approximately modeled as an equivalent homogenous material [18]. The effective refractive index of SWG waveguide is determined by the ratio of silicon and silica. For a given duty cycle, the SWG waveguide on SOI platform is also intrinsically birefringent and the effective refractive indexes in two perpendicular directions are given by Rytov [19] as
where $n_{\parallel}$ and ${n_ \bot }$ are the refractive indexes in the horizontal and perpendicular directions respectively, f is the duty cycle, ${n_1}$ and ${n_2}$ are the refractive indexes of silicon and silica. The refractive indexes of the equivalent materials in the two directions can be tuned by changing the duty cycle of the SWG waveguide.Figure 2(a) indicates the refractive indexes of the different modes in the 2 μm × 0.22 μm waveguide with SWG duty cycle varied. TE polarization is parallel to the substrate and TM polarization is perpendicular in the silicon wire waveguide. The SWG MMI waveguide has different refractive index in the two perpendicular directions for different duty cycles. According to Eq. (1), the MMI length is determined by the refractive index difference of fundamental and second-order optical modes and it is possible to overcome the polarization dependence of the SWG MMI waveguide on condition that the refractive index difference is zero. Figure 2(b) shows the effective refractive index difference of (TE0-TE2) and (TM0-TM2) in the MMI waveguide with a 2-μm width for different duty cycles. We can observe that the zero-effective refractive index difference is obtained where the duty cycle is set at 82%. In other word, the beat lengths for TE and TM polarizations keep the same for the 82% duty cycle SWG MMI waveguide. In addition, the trade-off between the transmission performance and device footprint can be released with the SWG structure. The effective refractive index of the SWG MMI waveguide is much smaller than that of the conventional silicon MMI waveguide without lithographic patterning. Hence, the calculated beat length is reduced in a further step. The 3D FDTD method is utilized to simulate the proposed device and determines the other geometry parameters including pitch, duty cycle, and beat length. The optimization goal of the device design is to achieve an excellent individual transmission performance for both TE and TM polarizations and maintain a low polarization dependence loss. To achieve this, the figure-of-merit (FOM) for this optimization is defined as
where ${T_{TE}}$ is the transmittance of TE mode and ${T_{TM}}$ is the transmittance of TM mode at 1550 nm. The term $({T_{TE}} + {T_{TM}})$ is aimed to get the best transmission performances for both TE and TM polarizations. At the same time, the term $|{{T_{TE}} - {T_{TM}}} |$ ensures that the difference between the transmittance of TE and TM polarization is limited. During the optimization process, different silicon waveguide crossings are generated and simulated in each iteration by changing the geometry variables. For the SWG structure, the pitch is optimized as 222 nm and the duty cycle is set to be 80%, which is consistent with the theoretical predicted 82% duty cycle by the effective index method as given in Fig. 2(b). The length of the SWG taper is 3 μm and the beat length of the SWG MMI coupler is 3.1 μm.Fig. 2. (a) The refractive indexes of different modes in 2 μm × 220 nm SWG waveguide with duty cycle varied on SOI platform. (b) The effective refractive index difference of (TE0-TE2) and (TM0-TM2) in the MMI waveguide with 2 μm width with different duty cycles.
The optical field distributions for TE and TM polarizations are shown in Fig. 3. The light is well confined along the propagation path and the power splitting into the perpendicular channels is negligible. The left picture of Fig. 3 shows the transmission spectra from 1530 to 1565 nm for C-band and it has a flat transmission curve over the entire wavelength range for both TE and TM polarizations. The crosstalks of TE andTM polarization are less than –45 dB and -35 dB respectively. Basically, it can be ignored since the power gets into the perpendicular channels is less than 0.1% (-30 dB) for both polarizations. The insertion losses are 0.46 dB and 0.64 dB for TE and TM polarization at 1550 nm respectively. And the polarization dependence loss is less than 0.2 dB at 1550 nm. As the intersection region of our crossing is a silicon square which has effective refractive index difference with the grating section, there exists some back reflections as the dash and solid orange lines indicated, but the back reflection is less than -13 dB for both TE and TM polarizations. The footprint of the entire structure is 12.5 μm × 12.5 μm. The polarization-insensitive silicon waveguide crossing based on traditional MMI couplers has the simulated insertion loss of 0.7 dB for TE and TM polarization at the footprint of 23 μm × 23 μm [20]. Our proposed crossing maintains a similar transmission performance but just with a quarter of the device footprint. The device can be fabricated with mature etch technique and there’re no complicated device patterns needed.
Fig. 3. The simulated insertion loss, crosstalk and back reflection transmission spectrums for TE and TM polarization of the proposed waveguide crossing (Left) and the normalized optical field distributions for TE and TM polarization (Right).
3. Experimental results and discussions
The proposed low polarization dependent silicon waveguide crossing based on SWG MMI couplers were fabricated afterwards. E-beam lithography (Raith EBPG 5000+) was used to define the patterns on the positive e-beam resists layer (AR-P 6200). The device patterns were then transferred from the photoresist to the silicon layer by inductively coupled plasma (ICP) etching [21]. It was a two-stage etching process, with full etching of a 220-nm depth for the waveguide structures in the first stage, and then shallow etching of a 70-nm depth for the input/output grating structures. Finally, the whole chip was deposited with 2 μm thick silica by plasma-enhanced chemical vapor deposition.
We have carried out the insertion loss and crosstalk measurement experimentally. A tunable semiconductor laser light source (TSl-710, Santec) has been used for testing. The laser can be tuned with wavelength from 1480 nm to 1620 nm. The polarization of the launched light is then tuned by a fiber polarization controller (PC) manually. An optical power meter (MPM-210, Santec) is implemented to collect light coupled out from chips under testing and its measurement range is from 1250 nm to 1650 nm. Our fabricated chip is placed on a movable stage and the single-mode fibers are confined in the fiber holders above the testing chip. And we can adjust the positions of fibers to get the best coupling efficiency with the assist of a CCD camera for monitoring. In addition, a motorized stage (KGW06050-L, KXC06020-G, Suruga Seiki) is utilized to realize the accurate displacement with the high resolutions of 0.05 μm/pulse and 0.0045°/pulse. The focus tube together with CCD camera is used to capture the fiber-chip image and the signal is projected to the monitor for better visualization. Figure 4 shows the scheme of our manual fiber-to-fiber measurement setup.
Figure 5 gives the microscope images of our fabricated devices. It includes the crossing, bending and plain waveguide. The plain waveguide is used for calibrate the insertion loss and the bending waveguide is for the crosstalk calibration. The bending radius in our layout are all 20 μm for low loss, which can also be replaced by inverse designed bending with small radius. Two kinds of focused grating couplers are used since both TE and TM polarizations need to considered [22]. For TE grating coupler, the central operation wavelength is 1550 nm and the grating period is 640 nm. The duty cycle is 50% and the etched depth is 70 nm. For TM grating coupler, the grating period is 1020 nm and the other parameters keep the same as the TE grating coupler. The plain waveguides are fabricated as the reference optical circuits without waveguide crossings. The insertion loss and crosstalk can be then achieved by the subtraction between the reference circuits and the crossing circuits. We also cascade three waveguide crossings in a single optical circuit to get the cumulative and averaged insertion loss. The enlarged optical view of the fabricated waveguide crossing is illustrated in Fig. 6(a). Figure 6(b) shows the fiber-to-fiber testing process for the fabricated devices. The left fiber is used to couple light from the light source to the crossing through the grating coupler and the right fiber is for coupling light out.
Fig. 6. (a) The enlarged view of our fabricated crossing waveguides based on SWG MMI coupler. (b) The fiber-to-fiber testing process for the fabricated crossing. The single-mode fibers are placed above the grating couplers.
Figure 7 shows the measured transmission spectra of the fabricated waveguide crossings. The fabricated chip has relative smooth transmission performances in the entire C-band. For the TE polarization, the transmission curve is flat and the averaged insertion loss is around 1.1 dB. For the TM polarization, a small transmission dip happens at 1542 nm for the fabrication variations and insertion loss is smaller than 3 dB in the entire wavelength range. For the simulation results, insertion loss is 0.46 dB for TE polarization and 0.64 dB for TM polarization. The measured insertion loss is higher than the simulation results due to the fabrication and measurement errors. However, it still performs well with low insertion loss and low polarization-dependence loss at 1550 nm wavelength for both TE and TM polarizations in practical. Our proposed SWG-based MMI crossing utilizes the SWG structure for eliminating the polarization-dependence loss and we believe that it can be beneficial for the future advanced PDM systems. The crosstalk is less than -35 dB for TE polarization and less than -40 dB for TM polarization. The proposed waveguide crossings are performed well in the real fabrication test.
4. Conclusions
In summary, we demonstrate a compact and polarization-insensitive silicon waveguide crossing based on SWG MMI couplers. The crossing fabrication is with simple etch processes and experimental verification on the insertion loss and crosstalk performance is carried out. For simulation, the insertion losses are 0.46 dB and 0.64 dB for TE and TM polarizations at 1550 nm respectively. And the measured insertion losses are 1 and 1.6 dB for TE and TM polarizations respectively. Our proposed SWG MMI coupler based on waveguide crossing is with low insertion loss, low crosstalk, and low polarization sensitivity with a very compact footprint of 12.5 μm × 12.5 μm. It is promising for polarization-division multiplexing applications in advanced silicon photonic integrated circuits.
Funding
Science, Technology and Innovation Commission of Shenzhen Municipality (JCYJ20170818094001391, JCYJ20180507183815699); Tsinghua-Berkeley Shenzhen Institute (TBSI) Faculty Start-up Fund; Leading Talents Program of Guangdong Province (2017BT01X121).
Disclosures
The authors declare that there are no conflicts of interest related to this article.
References
1. L. Chenlei, H. Wu, Y. Tan, S. Wang, and D. Dai, “Silicon-based on-chip hybrid (de) multiplexers,” Sci. China Inf. Sci. 61(8), 080407 (2018). [CrossRef]
2. J. M. Baumann, E.P. da Silva, Y. Ding, K. Dalgaard, L.H. Frandsen, L.K. Oxenl, and T. Morioka, “Silicon Chip-to-Chip Mode-Division Multiplexing,” Optical Fiber Communications Conference and Exposition, paper W1E.4 (2018).
3. 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(1), 3069 (2014). [CrossRef]
4. Y. Yu, G. Chen, C. Sima, and X. Zhang, “Intra-chip optical interconnection based on polarization division multiplexing photonic integrated circuit,” Opt. Express 25(23), 28330–28336 (2017). [CrossRef]
5. X. Tu, M. Li, J. Xing, H. Fu, and D. Geng, “Compact PSR based on an asymmetric bi-level lateral taper in an adiabatic directional coupler,” J. Lightwave Technol. 34(3), 985–991 (2016). [CrossRef]
6. H. Fukuda, K. Yamada, T. Tsuchizawa, T. Watanabe, H. Shinojima, and S. I. Itabashi, “Silicon photonic circuit with polarization diversity,” Opt. Express 16(7), 4872–4880 (2008). [CrossRef]
7. L. Liu, Y. Ding, K. Yvind, and J. M. Hvam, “Silicon-on-insulator polarization splitting and rotating device for polarization diversity circuits,” Opt. Express 19(13), 12646–12651 (2011). [CrossRef]
8. X. Tu, C. Song, T. Huang, Z. Chen, and H. Fu, “State of the art and perspectives on silicon photonic switches,” Micromachines 10(1), 51 (2019). [CrossRef]
9. G. Fan, R. Orobtchouk, B. Han, Y. Li, and H. Li, “8 ( 8 wavelength router of optical network on chip,” Opt. Express 25(20), 23677–23683 (2017). [CrossRef]
10. B. Wim, M. Fiers, and P. Dumon, “Design challenges in silicon photonics,” IEEE J. Sel. Top. Quantum Electron. 20(4), 1–8 (2013). [CrossRef]
11. Y. Zejie, A. Feng, X. Xi, and X. Sun, “Inverse-designed low-loss and wideband polarization-insensitive silicon waveguide crossing,” Opt. Lett. 44(1), 77–80 (2019). [CrossRef]
12. A. Densmore, D. Xu, P. Waldron, S. Janz, P. Cheben, J. Lapointe, A. Delge, B. Lamontagne, J. Schmid, and E. Post, “A silicon-on-insulator photonic wire based evanescent field sensor,” IEEE Photonics Technol. Lett. 18(23), 2520–2522 (2006). [CrossRef]
13. L. Chrostowski and H. Michael, Silicon photonics design: from devices to systems (Cambridge University Press, 2015).
14. M. Johnson, M. G. Thompson, and D. Sahin, “Low-loss, low-crosstalk waveguide crossing for scalable integrated silicon photonics applications,” Opt. Express 28(9), 12498–12507 (2020). [CrossRef]
15. H. Chen and A. W. Poon, “Low-loss multimode-interference-based crossings for silicon wire waveguides,” IEEE Photonics Technol. Lett. 18(21), 2260–2262 (2006). [CrossRef]
16. L. Soldano and E. Pennings, “Optical multi-mode interference devices based on self-imaging: principles and applications,” J. Lightwave Technol. 13(4), 615 (1995). [CrossRef]
17. M. Bachmann, P. Besse, and H. Melchior, “General self-imaging properties in N× N multimode interference couplers including phase relations,” Appl. Opt. 33(18), 3905–3911 (1994). [CrossRef]
18. R. Halir, P. Bock, P. Cheben, A. Ortega-Moñux, C. Alonso-Ramos, J. Schmid, J. Lapointe, D. Xu, J. Wangüemert-Pérez, Í Molina-Fernández, and S. Janz, “Waveguide sub-wavelength structures: a review of principles and applications,” Laser Photonics Rev. 9(1), 25–49 (2015). [CrossRef]
19. S. Rytov, “Electromagnetic properties of a finely stratified medium,” Soviet Physics JETP 2(3), 466–475 (1956).
20. J. Chen and Y. Shi, “Polarization-insensitive silicon waveguide crossing based on multimode interference couplers,” Opt. Lett. 43(24), 5961–5964 (2018). [CrossRef]
21. K. P. Yap, B. Lamontagne, A. Delâge, S. Janz, A. Bogdanov, M. Picard, E. Post, P. Chow-Chong, M. Malloy, D. Roth, and P. Marshall, “Fabrication of lithographically defined optical coupling facets for silicon-on-insulator waveguides by inductively coupled plasma etching,” J. Vac. Sci. Technol., A 24(3), 812–816 (2006). [CrossRef]
22. Y. Wang, Grating coupler design based on silicon-on-insulator (Doctoral dissertation, University of British Columbia, 2013).
