Abstract
A Y-junction is commonly used in on-chip systems because of its excellent broadband characteristic. However, due to the lack of regulation methods, in most cases Y-junctions are used as passive components. In this work we propose a reconfigurable broadband Y-junction based on phase change material. When Sb2Se3 layers on two branches are at different states, the Y-junction is asymmetric and works as a reconfigurable dual-mode (de)multiplexer. When both Sb2Se3 layers are amorphous, the Y-junction is symmetric and works as a dual-mode 3-dB power splitter. To achieve quasi-adiabatic evolution for both states in a short device length, we propose a segmented fast quasi-adiabatic method. By dividing the gap region into multiple segments and optimizing the geometry and length of each segment, the proposed device achieves bandwidth > 100 nm (crosstalk < -20 dB) in a compact footprint of 19.3 × 3 µm2. The simulation result shows that at center wavelength of 1550 nm, the crosstalk and insertion loss of our device are < -41 dB and <0.12 dB, respectively, under asymmetric mode (de)multiplex state, and the excess loss is within 0.06 dB under symmetric power splitting state. The proposed device may contribute to the realization of a high-bandwidth, flexible mode-division-multiplexing network.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
To cope with the ever-growing demand for data transmission capacity, technologies such as wavelength-division multiplexing (WDM) [1], polarization-division multiplexing (PDM) [2] and mode-division multiplexing (MDM) [3] have been extensively investigated. Among them, the MDM does not need multiple laser sources and have sufficient available mode channels. Besides, MDM can be combined with WDM and PDM, making MDM a promising approach to further increase communication capacity.
In MDM system, multimode power splitter [4–5] and mode (de)multiplexer [6–7] are two fundamental components. Currently, symmetric and asymmetric Y-junction have been used to construct ultra-broadband multimode power splitters [8–9] and mode (de)multiplexer [7,10] respectively due to their excellent broadband characteristic. If symmetric power splitting state and asymmetric mode (de)multiplex state can be realized in one reconfigurable Y-junction, the function of multimode power splitter and reconfigurable mode (de)multiplexer can be integrated into a single device, which will significantly improve the density of integration and flexibility of on-chip MDM system. However, due to the lack of regulation method, most Y-junctions are passive. Although some reconfigurable Y-junctions based on electro-optic [11] or thermos-optic [12] effect are demonstrated, limited by the small electro-optical or thermos-optic coefficient, those devices exhibit either small tunning range or extreme large footprint. Thus, the key issue of realizing reconfigurable Y-junction is how to achieve sufficient refractive index change in the branch waveguides.
In recent years, phase-change materials (PCMs) have been used in construction of reconfigurable photonic devices due to their drastic refractive index change during phase transition (Δn > 1) [13]. Besides, the phase change of PCM can be nonvolatile, which can reduce the power consumption of reconfigurable device. To achieve strong phase modulation of light, PCM is usually deposited on the waveguide to form hybrid waveguide. In this case absorption of PCM will cause excess loss in device. 20 nm thick GST deposited on 500 nm wide silicon waveguide will cause excess loss of 4.038 dB/um when GST is crystalline [14]. Thus, strong absorption of conventional PCM such as GST restrict the application of PCM in photonic devices. Although novel PCM such as Ge2Sb2Se4Te1 significantly reduce the absorption at amorphous state, the absorption coefficient k of Ge2Sb2Se4Te1 in crystalline state is still as high as 0.035 [15]. More recently, Sb2Se3 and Sb2S3 were reported as an ultra-low loss PCM which exhibits negligible intrinsic absorption losses (k < 10−3) in both states. The crystallization temperature of Sb2Se3 is about 200°C and the amorphization temperature is about 600°C [16]. Phase change can be conducted either optically [17] or electrically. Electrically driven phase change mainly based on Joule-heating. In Ref. [18], a 6.2 V pulse of 1 ms duration is used to crystallize Sb2Se3, while a 21 V pulse of 400 ns duration is used to heat Sb2Se3 above melting temperature and quench to avoid recrystallization. Currently, Sb2Se3 and Sb2S3 have been applied in optical switch [18], programmable MMI [19], reconfigurable metasurface [20], MZI [21,22] and reconfigurable ADC [23].
Here, we demonstrate a Sb2Se3-assisted reconfigurable broadband Y-junction. By controlling the states of Sb2Se3 layers deposited on both branches, the Y-junction can be switched between symmetric and asymmetric states. In the symmetric state, the Y-junction works as 3-dB power splitter for TE0 and TE1. While in the asymmetric state, the Y-junction works as a reconfigurable mode demultiplexer for TE0 and TE1. Compared with conventional Y-junction, introducing high refractive index Sb2Se3 layers into Y-junction makes the change of effective refractive index (neff) in gap region steeper, which means a larger footprint may be needed to achieve quasi-adiabatic evolution. Besides, large refractive index difference between amorphous and crystalline Sb2Se3 results in different effective refractive index distribution when Y-junction works under symmetric and asymmetric states. To achieve quasi-adiabatic evolution for both symmetric and asymmetric states in one device structure while reducing device length, we propose a segmented fast quasi-adiabatic (SFQA) method . First, we divide the gap region into N segments and design the geometry of N-1 interfaces to minimum average effective refractive index (neff) change for both TE0 and TE1. Then we optimize the length of segments in gap region using fast quasi-adiabatic dynamics [24]. The principle of fast quasi-adiabatic dynamics protocol is to keep the coupling between modes in a low level by homogeneously distributing adiabaticity over the device to achieve low crosstalk (XT) in a short device length. Our device has a compact footprint of 19.3 × 3 µm2 and exhibits a broad bandwidth > 100 nm for crosstalk < -20 dB. At the center wavelength of 1550 nm, the XTs and insertion losses (ILs) for both modes under asymmetric state are lower than -41 dB and 0.12 dB respectively, and the excess loss (EL) under symmetric state is lower than 0.06 dB. To the best of our knowledge, it is the first time that PCM is used to construct reconfigurable Y-junction, and this structure can be expanded to other functions such as reconfigurable power splitter.
2. Principle and design
The schematic of the proposed dual-mode (de)multiplexer is depicted in Fig. 1(a), the main structure of which is a symmetric silicon Y-junction. Two identical Sb2Se3 layers are deposited on the center of both branches and two indium tin oxide (ITO) layers are deposited on each Sb2Se3 layer (with the same geometry as Sb2Se3 layers) acting as heaters. When the Sb2Se3 on both branches are amorphous, the Y-junction is under symmetric state and the device works as a 3-dB power splitter for both TE0 and TE1 as shown in Fig. 1(d). When the Sb2Se3 on two branches are amorphous and crystalline respectively, the Y-junction is under asymmetric state and TE0 and TE1 are demultiplexed as TE0 in crystalline branch and amorphous branch respectively as shown in Fig. 1(c). The width of silicon stem waveguide Ws is 900 nm and the width of silicon waveguide in S-band region Wb is 400 nm. In S-band region, the width Wpcm and height Hpcm of the Sb2Se3 layers are 300 nm and 60 nm respectively, while the thickness of ITO Hito is 30 nm. The length SL and width SW of S-band are 13 um and 1 um respectively. In this work we adopt SOI platform with 220 nm thick silicon layer and 2 µm silica box layer. A 2 um thick SiO2 upper cladding is deposited on the device after the lift-off of Sb2Se3 and ITO. At the center wavelength of 1550 nm, the refractive indices of Sb2Se3 are na = 3.285 + 0i and nc = 4.505 + 0.0002i at amorphous and crystalline states respectively [19], and the refractive indices of Si, SiO2 and ITO are 3.476, 1.444 and 1.9615 + 0.0056i [25] respectively.
Since an additional Sb2Se3 layer is deposited on silicon Y-junction, to avoid abrupt neff change, in design region Sb2Se3 layer need to be designed as a taper structure. Compared with conventional Y-junction, apart from tapered gap between two branches, the structure of tapered Sb2Se3 layer also need to be designed. Therefore, the optimization target is designing the geometry and relative position of Sb2Se3 taper and tapered gap. The initial device structure for optimization is an ideal Y-junction whose branch gap starts from 0 as shown in Fig. 2(a). We define the waveguide structure between the input stem waveguide and output S-bands as design region, Figs. 2(c) and (d) show two typical cross sections of waveguide in that region. In the design region we have two variables for design, the width of Sb2Se3 layer w increases from 0 to 300 nm, and the width of branch gap g increases from 0 to 100 nm.
To achieve quasi-adiabatic evolution of TE0 and TE1 in design region, we have to find a change route from (w,g) = (0,0) to (w,g) = (300,100) in which neff of TE0 and TE1 evolve smoothly under both symmetric and asymmetric states. To simplify the optimization procedure, we set the minimum increments of w and g to be 10 nm, and w and g can only increase continuously or remain the same. Thus, the first step of SFQA method is to divide the gap region into N segments (N varies from 30 to 40 according to different change routes) and design the two parameters of w and g at the interface of each two segments. According to definition above, if a segment starts from (w,g) = (w0,g0), at the end of that segment (w,g) can only be (w0 + 10,g0), (w0,g0 + 10) or (w0 + 10,g0 + 10). A figure of merit (FOM) defined as below is used to evaluate the evolution smoothness:
Here N denotes the total number of segments in design region from (w,g) = (0,0) to (w,g) = (300,100), $\varDelta n$i means the difference of neff between the end and start of ith segment, while superscript TE0a, TE0s, TE1a, TE1s denotes TE0 and TE1 under asymmetric and symmetric states respectively. Using 3D FDTD simulation [28], we simulate the neff profiles of TE0 and TE1 as w and g vary under asymmetric and symmetric states, the results are plotted in Figs. 3(a)-(d). We traverse all change routes of (w,g) that satisfy the constraint (as shown in Fig. 3(e)), and the FOM of all 41 routes are shown in Fig. 3(f). The change route with minimum FOM is demonstrated by black dashed line in Figs. 3(a)-(d), and the corresponding waveguide structure is shown in Fig. 3(g). Notably, considering that ideal triangular taper is unachievable in fabrication, in practice w start from 60 nm (equal to Hpcm) instead of 0 nm, so the number of segments N in our structure is 25.
After the change route of w and g are decided, the second step of SFQA method is to determine the length of each segment using fast quasi-adiabatic dynamics. According to Ref. [26], an adiabaticity parameter c(z) is defined as below:
Since in our work the gap region consists of discrete 25 segments, in the proposed $\textrm{SFQA}$ method we modify the adiabaticity parameter c as shown in Eq. (3)
The ${c_i}$ and ${L_i}$ after applying $\textrm{SFQA}$ is shown by orange line in Fig. 4(a) and Fig. 4(b) respectively.
We simulate and plot the performances of linear case and SFQA case as a function of the total length of 25 segments L in Fig. 5. As expected, compared with linear case, Y-junction using SFQA method achieve lower XTs for both modes under asymmetric state at a shorter length (XTs < -40 dB when L = 6.3 um). Under symmetric state two cases show no significant difference, that is because symmetric Y-junction is much easier to achieve adiabatic evolution compared with asymmetric Y-junction. Thus, we choose L = 6.3 um, the total length of first 15 segments is 3 um and the total length of last 10 segments is 3.3 um, the proportion of each segment is marked in Fig. 4(b).
In our current structure the gap width starts from zero and increase gradually. However, the non-zero gap width caused by the limited feature size achievable by lithography will deteriorate the IL and XT property. To avoid this problem, we use circular-hole-based chirped subwavelength slot to replace part of the ideal gap as mentioned in Ref. [27]. To determine the radius R and period Λ of each hole in the subwavelength slot, we simulate the effective refractive index of TE0 in subwavelength slot waveguide as the function of R and Λ. The result is demonstrated in Fig. 6(a), together with the effective refractive index of TE0 in ideal gap waveguide as the function of gap width (black solid line). According to the result shown in Fig. 6(b), in region 1 R is set to be 55 nm and Λ increases from 140 nm to 180 nm to mimic ideal gap when the gap width g decreases from 40 nm to 20 nm. While in region 2 Λ is fixed to be 180 nm and R decrease from 55 nm to 25 nm to mimic the ideal gap when the g decreases from 20 nm to ideal zero. In total 8 subwavelength holes are used to replace the ideal gap.
After replacing ideal gap with subwavelength holes, the performance of the proposed reconfigurable Y-junction is investigated using 3D FDTD simulation [28]. The XTs and ILs of our device under asymmetric mode demultiplex state and ELs under symmetric 3 dB power splitting state are demonstrated in Fig. 7(a) and Fig. 7(b) respectively. Compared with ideal gap Y-junction, under asymmetric state, the deterioration of XTs and ILs at the center wavelength of 1550 nm are below 1.3 dB and 0.09 dB respectively. While under symmetric state, the deterioration of ELs at the center wavelength of 1550 nm are less than 0.06 dB. The small variation indicates that our replacement of ideal gap using subwavelength holes is successful.
3. Result and discussion
According to simulation results in Fig. 7, under asymmetric state, the XTs and ILs of TE0 and TE1 at the center wavelength of 1550 nm are -41.9 dB, 0.06 dB and -41 dB, 0.12 dB respectively. While under symmetric state, the ELs of TE0 and TE1 at the center wavelength of 1550 nm are 0.03 dB and 0.05 dB respectively. It is noteworthy that the proposed device exhibits large band width. At asymmetric state, the XTs of both modes are lower than -20 dB in a large wavelength range of >100 nm, and the ILs of both modes are lower than 0.2 dB.
Next, the fabrication tolerance of our device at center wavelength of 1550 nm is investigated. We test the fabrication tolerance for the variation of Sb2Se3 layer width Δw, silicon waveguide width Δd and hole radius ΔR, the results are shown in Fig. 8 and Table 1. As shown in Fig. 8(a), for ± 40 nm Sb2Se3 layer width variation, ILs of TE0 and TE1 under asymmetric state experience maximum degradation of 0.11 dB and 0.09 dB respectively. Meanwhile, even in the worst case (Δw = -40 nm) XTs of both modes are still less than -17 dB. According to Fig. 8(d), under symmetric state device shows better tolerance for Δw, the variations of ELs are less than 0.04 dB for both modes when Δw change from -40 nm to 40 nm. Then, we test the tolerance for Δd by changing the waveguide and gap widths to ${W_s} \pm \varDelta d,\; {W_b} \pm \varDelta d$ and $g \mp \varDelta d$. From Fig. 8(b) we can see under asymmetric state as $\varDelta w$ change from – 40 nm to 40 nm, XTs and ILs of both modes experience a maximum degradation of 27 dB and 0.44 dB respectively. Under symmetric state, according to Fig. 8(e) EL of both modes remain less than 0.4 dB as $\varDelta w$ varies. Lastly, we analyze the performance of device against the variation of hole radius ΔR. From Fig. 8(c) we find XTs of both modes under asymmetric state improve around 2 dB when ΔR = 2 nm, which may be caused by the error exists in the process of equating ideal gap waveguide to chirped subwavelength holes. Under symmetric state, Y-junction shows strong tolerance to ΔR, ELs of both modes are below 0.07 dB as ΔR varies from -10 nm to 10 nm. Generally, the proposed Y-junction exhibits good tolerance to fabrication error under both states.
4. Conclusion
In conclusion, we proposed a Sb2Se3-assisted reconfigurable Y-junction. By controlling the state of Sb2Se3 layers deposited on top of the branch waveguides, we modify the effective refractive index of the PCM-Si hybrid waveguide and thus switch the proposed device between asymmetric state and symmetric state. To achieve quasi-adiabatic evolution under both states in a short device length, we propose a segmented fast quasi-adiabatic (SFQA) method. By dividing gap region into multiple segments and optimizing the geometry and length of each segments, we propose a reconfigurable Y-junction with a bandwidth of more than 100 nm for crosstalk < -20 dB in an ultra-compact footprint of around 19.3 × 3 µm2. Under asymmetric state the Y-junction acts as a reconfigurable mode (de)multiplexer and the XTs and ILs of TE0 and TE1 at center wavelength of 1550 nm are -41.9 dB, 0.06 dB and -41 dB, 0.12 dB respectively. Under symmetric state the Y-junction works as a 3-dB power splitter, the ELs of TE0 and TE1 are 0.03 dB and 0.05 dB respectively. The proposed device may contribute to the construction of future high bandwidth reconfigurable on chip MDM network. Moreover, PCM assisted Y-junction structure demonstrated in this work provides a new solution for realizing broadband reconfigurable devices, and our SFQA method may be useful in designing other device based on Y-junction with multilayer structures.
Funding
National Natural Science Foundation of China (51911530159, 62175076).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. B. G. Lee, X. Chen, A. Biberman, X. Liu, I. W. Hsieh, C.-Y. Chou, J. I. Dadap, F. Xia, W. M. Green, L. Sekaric, Y .A. Vlasov, R. M. Osgood Jr, and K. Bergman, “Ultrahigh-bandwidth silicon photonics nanowire waveguides for on-chip networks,” IEEE Photonics Technol. Lett. 20(6), 398–400 (2008). [CrossRef]
2. X. Zhang, M. Karlsson, P. A. Andrekson, and E. Kolltveit, “Polarization-division multiplexed solitons in optical fibers with polarization-mode dispersion,” IEEE Photonics Technol. Lett. 10(12), 1742–1744 (1998). [CrossRef]
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. L. Han, B. P. P. Kuo, N. Alic, and S. Radic, “Ultra-broadband multimode 3 dB optical power splitter using an adiabatic coupler and a Y-branch,” Opt. Express 26(11), 14800–14809 (2018). [CrossRef]
5. H. Xu, D. Dai, and Y. Shi, “Ultra-broadband on-chip multimode power splitter with an arbitrary splitting ratio,” OSA Continuum 3(5), 1212–1221 (2020). [CrossRef]
6. J. Xing, Z. Li, X. Xiao, J. Yu, and Y. Yu, “Two-mode multiplexer and demultiplexer based on adiabatic couplers,” Opt. Lett. 38(17), 3468–3470 (2013). [CrossRef]
7. W. Chen, P. Wang, T. Yang, G. Wang, T. Dai, Y. Zhang, and J. Yang, “Silicon three-mode (de) multiplexer based on cascaded asymmetric Y junctions,” Opt. Lett. 41(12), 2851–2854 (2016). [CrossRef]
8. Z. Wang, Y. Liu, Z. Wang, Y. Liu, J. Du, Q. Song, and K. Xu, “Ultra-broadband 3 dB power splitter from 1.55 to 2 µm wave band,” Opt. Lett. 46(17), 4232–4235 (2021). [CrossRef]
9. H. Xu and Y. Shi, “Ultra-broadband dual-mode 3 dB power splitter based on a Y-junction assisted with mode converters,” Opt. Lett. 41(21), 5047–5050 (2016). [CrossRef]
10. N. Riesen and J. D. Love, “Design of mode-sorting asymmetric Y-junctions,” Appl. Opt. 51(15), 2778–2783 (2012). [CrossRef]
11. Y. Yamaguchi, S. Nakajima, A. Kanno, T. Kawanishi, M. Izutsu, and H. Nakajima, “High extinction ratio characteristics of over 60 dB Mach-Zehnder modulator with asymmetric power-splitting Y-branches on X-cut Ti:LiNbO3,” Jpn. J. Appl. Phys. 53(8S2), 08MB03 (2014). [CrossRef]
12. A. Rajandekar and R. Singhal, “Voltage-controlled all-polymer reconfigurable optical power splitter,” IEEE 4th International Conference on Photonics (ICP), 244–246 (2013).
13. M. Wuttig, H. Bhaskaran, and T. Taubner, “Phase-change materials for non-volatile photonic applications,” Nat. Photonics 11(8), 465–476 (2017). [CrossRef]
14. H. Zhang, L. Zhou, B. M. A. Rahman, X. Wu, L. Lu, Y. Xu, and J. Chen, “Ultracompact Si-GST hybrid waveguides for nonvolatile light wave manipulation,” IEEE Photonics J. 10(1), 1–10 (2018). [CrossRef]
15. Y. Zhang, J. B. Chou, J. Li, H. Li, Q. Du, A. Yadav, S. Zhou, M. Y. Shalaginov, Z. Fang, H. Zhong, C. Roberts, P. Robinson, B. Bohlin, C. Ríos, H. Lin, M. Kang, T. Gu, J. Warner, V. Liberman, K. Richardson, and J. Hu, “Broadband transparent optical phase change materials for high-performance nonvolatile photonics,” Nat. Commun. 10(1), 1–9 (2019). [CrossRef]
16. J. R. Erickson, V. Shah, Q. Wan, N. Youngblood, and F. Xiong, “Designing fast and efficient electrically driven phase change photonics using foundry compatible waveguide-integrated microheaters,” Opt. Express 30(8), 13673–13689 (2022). [CrossRef]
17. M. Delaney, I. Zeimpekis, D. Lawson, D. W. Hewak, and O. L. Muskens, “A New Family of Ultralow Loss Reversible Phase-Change Materials for Photonic Integrated Circuits: Sb2S3 and Sb2Se3,” Adv. Funct. Mater. 30(36), 2002447 (2020). [CrossRef]
18. C. Ríos, Q. Du, Y. Zhang, C.-C. Popescu, M. Y. Shalaginov, P. Miller, C. Roberts, M. Kang, K. A. Richardson, T. Gu, S. A. Vitale, and J. Hu, “Ultra-compact nonvolatile photonics based on electrically reprogrammable transparent phase change materials,” arXiv:2105.06010 (2021).
19. M. Delaney, I. Zeimpekis, H. Du, X. Yan, M. Banakar, D. J. Thomson, D. W. Hewak, and O. L. Muskens, “Nonvolatile programmable silicon photonics using an ultralow-loss Sb2Se3 phase change material,” Sci. Adv. 7(25), eabg3500 (2021). [CrossRef]
20. L. Chen, Y. Hao, L. Zhao, R. Wu, Y. Liu, Z. Wei, N. Xu, Z. Li, and H. Liu, “Multifunctional metalens generation using bilayer all-dielectric metasurfaces: erratum,” Opt. Express 29(6), 9332–9345 (2021). [CrossRef]
21. J. Faneca, I. Zeimpekis, S. T. Ilie, T. D. Bucio, K. Grabska, D. W. Hewak, and F. Y. Gardes, “Towards low loss non-volatile phase change materials in mid index waveguides,” Neuromorph. Comput. Eng. 1(1), 014004 (2021). [CrossRef]
22. L. Lu, S. F. Reniers, Y. Wang, Y. Jiao, and R. E. Simpson, “Reconfigurable InP waveguide components using the Sb2S3 phase change material,” J. Opt. 24(9), 094001 (2022). [CrossRef]
23. T. Teo, M. Krbal, J. Mistrik, J. Prikryl, L. Lu, and R. Simpson, “Comparison and analysis of phase change materials-based reconfigurable silicon photonic directional couplers,” Opt. Mater. Express 12(2), 606–621 (2022). [CrossRef]
24. S. Martínez-Garaot, A. Ruschhaupt, J. Gillet, T. Busch, and J. G. Muga, “Fast quasiadiabatic dynamics,” Phys. Rev. A 92(4), 043406 (2015). [CrossRef]
25. Z. Ma, Z. Li, K. Liu, C. Y. e, and V. J. Sorger, “Indium-tin-oxide for high-performance electro-optic modulation,” Nanophotonics 4(1), 198–213 (2015). [CrossRef]
26. H. C. Chung, K. S. Lee, and S. Y. Tseng, “Short and broadband silicon asymmetric Y-junction two-mode (de) multiplexer using fast quasiadiabatic dynamics,” Opt. Express 25(12), 13626–13634 (2017). [CrossRef]
27. L. Lu, D. Liu, M. Yan, and M. Zhang, “Subwavelength adiabatic multimode Y-junctions,” Opt. Lett. 44(19), 4729–4732 (2019). [CrossRef]
28. Lumerical FDTD solutions, https://www.lumerical.com.