Waveguide crossings employing tilted MMI structures on silicon wire waveguide are proposed and demonstrated. Intersecting angle of the two MMI waveguides is optimized for low crosstalk. The optimization is carried out with input polarizations specified. On the fabricated MMI crossings, crosstalk lower than −38 dB in the C-band was experimentally confirmed. A novel polarization-insensitive crossing based on a diversity circuit was fabricated. Crosstalk lower than −30 dB in the C-band is demonstrated.
© 2014 Optical Society of America
There has been a continuing demand for low-crosstalk and low-loss waveguide crossings in planar light circuits (PLCs). Table 1 shows a summary of the relevant studies reported to date. In a simple waveguide crossings based on a typical silicon-wire waveguide, light is strongly diffracted to yield significant crosstalk. One common approach to reduce the diffraction is to decrease the effective index or expand the optical mode in space. This can be done by tapered widening of the waveguides toward the center of the crossings. The taper shapes tried so far include ellipses [1,2], a parabola with a rib formed by double etching [3,4], and a more complicated one determined by numerical optimization method . As nontrivial designs, we note the offset waveguides , subwavelength diffraction gratings , and directional coupler (DC) -based crossings . An orthodox design [9–11] is the one that utilizes the idea of spot-size converter , which includes a high-index dielectric as a second core. Another common design is that including two 1 × 1 multimode interference (MMI) structures [13–16]. The MMI-crossings’ characteristics depend on the angle between the two MMIs that intersect each other , [17,18]. This aspect was discussed on silica PLC . The silica PLC is a weakly guiding system (2% of Δ) and the properties predicted in  are unlikely to hold for silicon-wire waveguides. Motivated in part by this, we are interested in applying the MMI-crossing to silicon-wire waveguides. The primary advantage that we note is its fabrication simplicity in that MMI-crossings can be formed in a single patterning step. Another advantage is that it is straightforward to design polarization selective crossings.
In this paper, we first present results of theoretical analysis of the tilt-angle dependence of the crosstalk of the MMI-crossing. The experimental results are then presented, including crosstalk as good as the lowest crosstalk [4,5] reported to date. Also reported is a crossing that can cross TE and TM polarized light with low crosstalk. We further demonstrate a novel polarization-insensitive crossing that is formed as a polarization-diversity circuit.
Figure 1 schematically illustrates the crossing that we have in mind. The crossing includes two 1 × 1 MMI structures that have a common center and intersect each other with an angle of θ. In each of the MMIs, the optical field at the input port is first formed at the center and is again formed at the exit . For designing, use was made of a two-dimensional finite difference time domain method. In this calculation, we assigned to the SOI layer an effective index of 2.88 for TE polarization and 2.07 for TM polarization, taking account of its structure with the 220-nm thick SOI on a 3.0-μm thick SiO2 clad.
The first step of design is to make a single 1 × 1 MMI have low excess loss. We presume that the MMI dimension is 1.3 μm (width, WMMI) × 6.2 μm (length, LMMI) for TE and 1.5 μm × 6.2 μm for TM polarization. These dimensions satisfy a requirement LMMI = 2n(WMMI)2/λ  for the MMI to serve as 1 × 1 MMI.
We also let the input and output waveguides be tapered. We set the width of the waveguide core outside of the taper to 0.45 μm and the taper length to 3 μm. We calculated the transmission loss while varying the taper width at its end. Figure 2 shows the calculation result. It is to be noticed that the transmission loss is sensitive to the choice of the taper width. The optimal taper width is 1.0 μm for the TE polarization and 1.5 μm for the TM polarization.
In 90 degree-crossing, two side ports see an equal power of optical leakages because it is symmetric. This symmetry is broken if the intersecting waveguide is tilted. One side port then sees less leakage while the other more. We set the former port to an output port at the bar position. This is an essential idea behind our MMI crossing.
If the two MMIs are for a common polarization, the combined structure serves as a crossing for that polarization. If one MMI is for TE-polarization and another for TM-polarization, the resultant structure allows a TE-polarized input to cross a TM-polarized input. Polarization selectivity by combining the two MMIs is a primary advantage of this design. Assuming that the two inputs are both TE-polarized, we calculated transmission (from port 1 to 1’) and crosstalk (from port 1 to 2’) of the crossing, by varying the angle θ. Figures 3(a) and 3(b) show the result of the calculation with wavelength fixed to 1550 nm. It is to be noticed in Fig. 3(a) that crosstalk becomes lowest, −44 dB, at θ = 110þ. The optimal MMI length could shift from that for a single MMI. With the intersection angle θ fixed to 110þ, we then calculated transmission and crosstalk while varying the MMI length from 5.8 to 6.8 μm. The calculation result is shown in Fig. 3(b), proving that the optimal MMI length stays at 6.2 μm. Figure 3(c) shows wavelength-dependence of the performances for the crossing with θ = 110þ and MMI length of 6.2 μm. It is to be noticed that the crosstalk level is lower than −40 dB over a 100-nm wavelength-range, reaching its minimum, −50 dB, at 1560 nm and that the transmission loss at 1560 nm is 0.24 dB. In the same manner, we calculated on an MMI crossing for TM polarization. The optimal angle θ was 105þ and the optimal MMI length was 6.3 μm. The wavelength-dependent response of this crossing is shown in Fig. 3(d).
Next, we consider a case that one input is TE-polarized and the other TM-polarized. We again calculated the performance while varying θ. The optimal angle was found to be 105þ. Checking whether or not the optimal MMI length stays at 6.2 μm, we have found that the optimal length shifts to 6.0 μm for both TE and TM polarizations. Wavelength-dependent performance of this optimal crossing is plotted in Fig. 4..
To fabricate a silicon-wire waveguide, a commercial SOI wafer which includes a 220 nm-thick-silicon layer on the 3000 nm-thick-SiO2 layer was used. 50 nm-thick-SiO2 layer was deposited on the SOI wafer by a plasma chemical vapor deposition (PCVD) for hard mask. The mask pattern was drawn by electron-beam lithography. To transfer resist mask pattern to silicon layer, we used a simple reactive ion etching (RIE) with sulfur hexafluoride (SF6)-trifluoromethane (CHF3) gas mixture. After forming a mesa waveguide structure with small roughness, 2500 nm-thick-SiO2 layer was deposited by using a PCVD technique. We laid on a waveguide chip patterns including not only a single MMI crossing but also multiple MMI crossings connected in series to determine the transmission loss per crossing. A spot-size-converter using a narrow waveguide core is formed at each of the input and output ends of the waveguide chip. Figures 5(a), 5(b), and 5(c) show SEM photographs of the fabricated TE-TE, TM-TM, and TE-TM crossings, respectively. For optical characterization, a wavelength-tunable laser diode was used. Light was focused by a lens on the input end of the waveguide chip. The output light from the chip was collected by a lensed fiber to measure the transmitted power with a photodiode. The coupling loss per I/O port was estimated to be 4.5 dB for TE-polarization and 3.4 dB for TM-polarization.
4. Results and discussions
We measured transmission from port 1 to 1’ for waveguides including multiple TE-TE crossings connected in series. The result is shown in Fig. 6. From this cutback plot, we estimate the insertion loss per crossing to be 0.29 dB. The primary focus of the present study is on low crosstalk rather than on low loss. The longer taper will improve insertion loss, which is a theme of future study. Measured wavelength-dependent transmission (from port 1 to 1’) and crosstalk (from port 1 to 2’) of a single TE-TE and TM-TM crossing are shown in Fig. 7. We fitted a polynomial curve to the crosstalk spectrum that revealed fringes due to some Fabry-Perot effects. It is to be noted that crosstalk for TE-TE (TM-TM) crossing is lower than −38(−48) dB, which occurs at 1530 nm, in the whole C-band. This is in good agreement with the theoretical prediction that we presented.
In the TE-TM crossing, port 1 serves as an input port for TE-polarized light and port 2 for TM-polarized light. Figures 8(a) and 8(b) show the measurement results for TE-polarized input and TM-polarized input, respectively. It is to be noted that for both the TE and TM polarizations, the crosstalk level is −42 dB, which occurs for TM input at 1560 nm, or lower in the whole C-band. This is consistent with the calculation. The insertion loss per crossing was estimated to be 0.31 dB for TE polarization and 0.26 dB for TM polarization.
By forming a diversity circuit, as shown in Fig. 9(a), it is further possible to make the crossing polarization-insensitive. It is to be noticed that the diversity circuit consists of all the three types of crossings that we discussed so far and two polarization beam splitters. We fabricated such a diversity circuit, as shown in Fig. 9(b). Measurement results of the spectral response are shown in Fig. 10(a) for TE polarization as the input polarization and 10(b) for TM input, indicating that crosstalk of −30 dB, which occurs for TE input at 1560 nm, or lower is retained for both the polarizations over the whole C-band. Crosstalk is not as good as the three element crossings. The deterioration is caused by optical leak at the polarization beam splitters. This is the first demonstration of the diversity approach to polarization insensitive crossing, to the best knowledge of ours. The diversity circuit in Fig. 9(a) would serve as a polarization-insensitive 2 × 2 switch , if the TE-TE and TM-TM crossings are replaced by 2 × 2 polarization-sensitive switches. It is to be noticed that the TE-TM crossing is essential to the diversity circuit.
Waveguide crossings based on a tilted MMI structure on a silicon wire waveguide were proposed and experimentally demonstrated. The single-layer structure is advantageous for fabrication. With input polarizations specified, TE-TE, TM-TM, and TE-TM crossings were designed. The intersection angle between the two 1 × 1 MMIs was optimized for low crosstalk, whereas the taper shape of the input/output port was optimized for low loss. Crosstalk of the fabricated crossings is −38 dB or lower in the entire C-band. A polarization-diversity crossing using these crossings was demonstrated, crosstalk of which is −30 dB or lower.
Authors are grateful to Mr. K. Tashiro for his technical assistance in device fabrication. This study was supported in part by Project for Developing Innovation Systems of MEXT, Japan.
References and links
1. T. Fukazawa, T. Hirano, F. Ohno, and T. Baba, “Low loss intersection of Si photonic wire waveguides,” Jpn. J. Appl. Phys. 43(2), 646–647 (2004). [CrossRef]
2. F. Shinobu, Y. Arita, and T. Baba, “Low-loss simple waveguide intersection in silicon photonics,” Electron. Lett. 46(16), 1149–1150 (2010). [CrossRef]
3. W. Bogaerts, P. Dumon, D. V. Thourhout, and R. Baets, “Low-loss, low-cross-talk waveguide crossings for silicon-on-insulator on nanophotonic waveguides,” Opt. Lett. 32(19), 2801–2803 (2007).
4. W. Bogaerts, S. K. Selvaraja, P. Dumon, J. Brouckaert, K. De Vos, D. Van Thourhout, and R. Baets, “Silicon-on insuluator spectral filters fabricated with CMOS technology,” IEEE J. Sel. Top. Quantum Electron. 16(1), 33–44 (2010). [CrossRef]
5. P. Sanchis, P. Villalba, F. Cuesta, A. Håkansson, A. Griol, J. V. Galán, A. Brimont, and J. Martí, “Highly efficient crossing structure for silicon-on-insulator waveguides,” Opt. Lett. 34(18), 2760–2762 (2009). [CrossRef] [PubMed]
6. D. Tanaka, Y. Ikuma, and H. Tsuda, “Low-loss, small crosstalk offset crossing structure for large-scale planar lightwave circuits,” IEICE Electron. Express 6(7), 407–411 (2009). [CrossRef]
7. P. J. Bock, P. Cheben, J. H. Schmid, J. Lapointe, A. Delâge, D.-X. Xu, S. Janz, A. Densmore, and T. J. Hall, “Subwavelength grating crossings for silicon wire waveguides,” Opt. Express 18(15), 16146–16155 (2010). [CrossRef] [PubMed]
8. Y. Shoji, K. Kintaka, S. Suda, H. Kawashima, T. Hasama, and H. Ishikawa, “Low-crosstalk 2 x 2 thermo-optic switch with silicon wire waveguides,” Opt. Express 18(9), 9071–9075 (2010). [CrossRef] [PubMed]
9. Y. Wakayama, T. Kita, and H. Yamada, “Optical crossing and integration using Si-Wire/silica waveguides,” Jpn. J. Appl. Phys. 50(4), 201–204 (2011). [CrossRef]
11. A. M. Jones, C. T. DeRose, A. L. Lentine, D. C. Trotter, A. L. Starbuck, and R. A. Norwood, “Ultra-low crosstalk, CMOS compatible waveguide crossings for densely integrated photonic interconnection networks,” Opt. Express 21(10), 12002–12013 (2013). [CrossRef] [PubMed]
12. T. Shoji, T. Tsuchizawa, T. Watanabe, K. Yamada, and H. Morita, “Low loss mode size converter from 0.3 μm square Si wire waveguides to single mode fibers,” Electron. Lett. 38(25), 1669–1670 (2002). [CrossRef]
13. 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]
14. F. Xu and A. W. Poon, “Silicon cross-connect filters using microring resonator coupled multimode-interference-based waveguide crossings,” Opt. Express 16(12), 8649–8657 (2008). [CrossRef] [PubMed]
15. C.-H. Chen and C.-H. Chiu, “Taper-integrated multimode-interference based waveguide crossing design,” IEEE J. Quantum Electron. 46(11), 1656–1661 (2010). [CrossRef]
16. Y. Zhang, S. Yang, A. E.-J. Lim, G. Q. Lo, C. Galland, T. Baehr-Jones, and M. Hochberg, “A CMOS-compatible, and low-crosstalk silicon waveguide crossing,” IEEE Photonics Technol. Lett. 25(5), 422–425 (2013). [CrossRef]
17. Y. Xie, J. Xu, and J. Zhang, “Elimination of cross-talk in silicon-on-insulator waveguide crossings with optimized angle,” Opt. Eng. 50(6), 0646011–0646014 (2011).
18. P. Sanchis, J. V. Galan, A. Griol, J. Marti, M. A. Piqueras, and J. M. Perdigues, “Low-crosstalk in silicon-on insulator waveguide crossings with optimized-angle,” IEEE Photonics Technol. Lett. 19(20), 1583–1585 (2007). [CrossRef]
19. H. Liu, H. Tam, P. K. A. Wai, and E. Pun, “Low-loss waveguide crossing using a multimode interference structure,” Opt. Commun. 241(1–3), 99–104 (2004). [CrossRef]
20. L. B. Soldano and E. C. M. Pennings, “Optical multi-mode interference deivices based on self-imaging: Principles and applications,” J. Lightwave Technol. 13(4), 615–627 (1995). [CrossRef]
21. S.-H. Kim, Y. Shoji, G. W. Cong, H. Kawashima, T. Hasama, and H. Ishikawa, “Polarization diversity 2×2 switch with silicon-wire waveguide,” in European Conference on Optical Communication 2013 (2013), paper We.4.B.5.