Broadband and low-crosstalk polarization splitter-rotator with optimized tapers

Defen Guo; Tao Chu

doi:10.1364/OSAC.1.000841

1. Introduction

Silicon photonic integration is considered one of the most promising techniques in realizing high-density and high-speed optoelectronic integrated circuits, because it offers compact device size and is compatible with complementary metal-oxide-semiconductor (CMOS) fabrication processes. However, due to the large index contrast of silicon waveguides, silicon photonic devices are affected by the polarization mode dispersion and polarization-dependent loss [1]. Polarization transparent circuits or polarization diversity schemes [2,3], where both orthogonal polarizations are converted to identical ones, e.g., TE polarizations, and pass through a pair of identical photonic circuits in parallel, have been proposed to solve this issue. Thus, a high-performance polarization splitter-rotator (PSR) with low insertion loss, low polarization crosstalk, broad bandwidth, small device size, and large fabrication tolerance is preferred for realizing the polarization diversity scheme.

Various types of PSRs, based on either mode coupling or mode evolution mechanisms, have been proposed, including asymmetrical directional coupler (ADC) [4–7], adiabatic coupler [8–10], and TE1-assisted structure [11–20]. The ADC-based devices, which belong to the mode coupling types, are mostly compact and sensitive to the wavelength, coupling length, and waveguide width. The adiabatic-coupler-based PSRs usually have broad bandwidths and large fabrication tolerance. However, the device length (300 μm) needs to be sufficient for maintaining the slow evolution of local modes [8]. Relatively compact adiabatic 1550-nm PSR (100 μm) based on an asymmetric bi-level lateral taper has been proposed, but no experimental results were reported [9]. A compact adiabatic 1310-nm PSR (60 μm) has also been reported, but the feature size is only 60 nm [10]. The TE1-assisted PSRs usually contain a TM0–TE1 bi-level taper and a TE0-TE1 mode demultiplexer, in which the mode demultiplexer can be ADCs [11,12], multimode interference couplers (MMIs) [13,14], adiabatic couplers [15,16] and asymmetric Y junctions [17,18]. The ADC- and MMI-based devices still have the problems of limited bandwidth and fabrication tolerance. The adiabatic-coupler-based devices have large sizes (> 450 μm [15,16]). It is difficult for the asymmetrical Y-junctions to achieve low-loss corners in fabrication. Ultracompact and broadband TE1-assisted PSR based on fast quasiadiabatic dynamics has been proposed [20]. But the crosstalk is large (> 11.4 dB) and no experimental results are reported. None of the PSRs mentioned above have been experimentally demonstrated to have a preferred performance of <1 dB polarization conversion loss (PCL) and <–20 dB polarization crosstalk (CT) within >80 nm wavelength range.

In this letter, we experimentally demonstrated broadband, low-crosstalk, low-loss, and fabrication-tolerant TE1-assisted PSRs at 1550-nm and 1310-nm wavelength bands. The TM0–TE1 bi-level tapers and the TE0–TE1 mode demultiplexers are designed to have optimized tapers using the particle swarm optimization (PSO) and shortcuts to adiabaticity (STA) method, respectively. Recently, the protocols of STA have been proposed to design different kinds of coupled waveguide devices to shorten the adiabatic process, like asymmetric Y-junctions [21], power splitters [22,23], and mode (de)multiplexers [24]. In this work, we use the STA method to design efficient TE0-TE1 demultiplexer. Ridge waveguides for the demultiplexer are adopted to increase the coupling coefficient and reduce the coupling length. The total lengths of the 1550- and 1310-nm PSR devices are approximately 130- and 120-μm, respectively, and could be further reduced in size. For the 1550-nm PSR, the simulated PCL and polarization CT are less than 0.5 and –50 dB, respectively, over a wide wavelength range of 1450–1650 nm. The experimental results showed that the PCL and polarization CT are less than 0.74 and –20 dB, respectively, over the wavelength range measured from 1500 to 1600 nm. For the 1310-nm PSR, the simulated PCL and polarization CT are less than 0.2 and –46 dB, respectively, over the whole O band from 1260 to 1360 nm. The experimental results showed that the PCL and polarization CT are less than 1 and –23 dB, respectively, over the wavelength range measured from 1260 to 1340 nm. We have reported the results in [25] and we will give an introduction of the design process and discuss about the characteristics in this manuscript.

2. Design and simulation

2.1 Structure and principle

Figure 1

Fig. 1 Schematic diagram of the proposed PSR. W₁, W₂…W_n mean the waveguide widths of the PSO-based taper. W_a(z) means the access waveguide width. W_b means the bus waveguide width. D(z) means the center-separation between access and bus waveguides.

Download Full Size | PDF

schematically shows the structure of the proposed PSR, which consists of a bi-level taper and an asymmetric coupler. The PSR is based on a silicon-on-insulator (SOI) substrate. The silicon core thickness is 220 nm and slab thickness is 90 nm for the PSR ridge waveguide. The thicknesses of the SiO₂ cladding above and below are 2 and 3 μm, respectively. In operation, the input TM0 polarization light is rotated to the TE1 mode light when passing through the bi-level taper. Then, the TE1 mode light is converted to TE0 polarization and outputs from the cross port when passing through the asymmetrical coupler. The input TE0 polarization light will experience no mode conversion and output from the bar port when passing through the whole device. Thus, the device accomplishes the functions of polarization rotation and splitting.

2.2 Design of the bi-level taper

The PSO method is an efficient way to solve the problem of multi-parameter optimization and has been used successfully to design different kinds of silicon photonic devices [26,27]. Here we use the PSO method to optimize the parameters of bi-level taper, as shown in Fig. 2(a)

Fig. 2 (a) Schematic of the PSO-optimized bi-level taper. (b) Simulated TM0-TE1 conversion efficiency for the 1550-nm device. (c) Simulated TM0-TE1 conversion efficiency for the 1310-nm device. The inset pictures show the power distribution when TM0 is input.

Download Full Size | PDF

. The taper, which is 20-μm long, is divided into 4 and 10 segments with equal lengths for the 1550-nm and 1310-nm PSR, respectively. The values of the average TM0-TE1 conversion efficiency over wavelengths from 1500 to 1600 nm and from 1260 to 1360 nm are set to the figure of merit of the PSO, respectively. The optimized results obtained with the Lumerical 3D finite difference time domain (3D FDTD) software, are shown in Fig. 2(b) and Fig. 2(c). It can be seen that the input TM0 is turned to the TE1 mode and the conversion efficiency is greater than 90% from 1450 to 1650 nm for the 1550-nm device and greater than 97% from 1260 to 1360 nm for the 1310-nm device. The device parameters for the 1550- and 1310-nm tapers are listed in Table 1

Table 1. Parameters of the 1550-nm bi-level taper (nm).

View Table | View all tables in this article

and Table 2

Table 2. Parameters of the 1310-nm bi-level taper (nm).

View Table | View all tables in this article

, respectively.

2.3 Design of the TE0–TE1 demultiplexer

The TE0-TE1 demultiplexer shown in Fig. 1 consists of a single-mode access waveguide and a multimode bus waveguide. The width W_b of the multimode bus waveguide is fixed. The width W_a(z) of the single-mode access waveguide and the separation D(z) between the access and bus waveguides keep changing continuously along the propagation direction z. The design goal of a high-performance demultiplexer is to optimize the taper shapes of W_a(z) and D(z) using the protocol of STA. We give a brief introduction of the design here.

For the 1550-nm mode demultiplexer, the values of the initial access-waveguide width, bus-waveguide width, coupling length, minimum edge-gap, maximum coupling coefficient κ_max, and parameter c are listed in Table 3

Table 3. Values of parameters for the 1550-nm and 1310-nm mode demultiplexer.

View Table | View all tables in this article

[24]. The width of the single-mode access waveguide for phase matching is 432 nm when the width of the bus waveguide is 1000 nm. Then the propagation-constant mismatch is calculated around the width of 432 nm and fitted by a linear function. The coupling coefficients are calculated for the phase-matching widths and fitted by an exponential function. The minimum width of 200 nm for the edge-gap between the access and bus waveguides is chosen considering the fabrication limits of the 180-nm CMOS processes. The corresponding value of the maximum coupling coefficient κ_max is 0.11328 μm⁻¹. The coupling length is chosen as 80 μm. By using the method proposed in [24], the parameters of the demultiplexer can be derived, and are shown in Fig. 3(a)

Fig. 3 (a) Derived parameters of the 1550-nm mode demultiplexer. (b) Derived parameters of the 1310-nm mode demultiplexer. (c) FDTD simulation results of the 1550-nm mode demultiplexer. (d) FDTD simulation results of the 1310-nm mode demultiplexer.

Download Full Size | PDF

. For the 1310-nm mode demultiplexer, the values of the initial access-waveguide width, bus-waveguide width, coupling length, minimum edge-gap, maximum coupling coefficient κ_max, and parameter c are also summarized in Table 3. Following the same process, the parameters of the 1310-nm demultiplexer can be derived, and are shown in Fig. 3(b).

Figures 3(c) and 3(d) show the 3D FDTD simulation results of the 1550-nm and 1310-nm mode demultiplexer with the optimized parameters, respectively. For the 1550-nm device, the conversion losses of the input TE1 mode light to output TE0 mode light in the cross port and the residual TE1 mode light in the bar port are less than 0.1 dB and −24 dB, respectively, over 1450–1650 nm. For the 1310-nm device, the conversion losses of the input TE1 mode light to output TE0 mode light in the cross port and the residual TE1 mode light in the bar port are less than 0.1 dB and −23 dB, respectively, over 1260–1360 nm. The insets in Fig. 3(c) and Fig. 3(d) show the power distribution of the input mode light to the desired output mode light at 1550 and 1310 nm, respectively.

2.4 Simulation results of the PSR

Here, strip waveguides were chosen to follow the demultiplexer. Two 10-μm-long bi-level tapers were used to connect the ridge waveguides to 400-nm-wide strip waveguides. The simulation results of the total 1550-nm PSR with the optimized parameters are shown in Fig. 4

Fig. 4 FDTD simulation results of the total 1550-nm PSR with optimized parameters. (a) TE0 input. (b) TM0 input. The inset pictures display the power distribution at 1550 nm.

Download Full Size | PDF

. The performance of PSR can be characterized by PCL and polarization CT [5,7]. The PCL for input TE0 and TM0 mode lights is defined as

P C L_{b a r}^{T E 0 - T E 0} = 10 \times \log (P_{b a r}^{T E 0 - T E 0} / P_{i n}^{T E 0})

and

P C L_{c r o s s}^{T M 0 - T E 0} = 10 \times \log (P_{c r o s s}^{T M 0 - T E 0} / P_{i n}^{T M 0})

, respectively. The CT evaluates the crosstalk level of the TE0 mode between the two output ports and can be defined as

C T_{b a r}^{T E 0 - T E 0} = 10 \times \log (P_{b a r}^{T E 0 - T E 0} / P_{c r o s s}^{T E 0 - T E 0})

and

C T_{c r o s s}^{T M 0 - T E 0} = 10 \times \log (P_{c r o s s}^{T M 0 - T E 0} / P_{b a r}^{T M 0 - T E 0})

for input TE0 and TM0 mode lights, respectively. Here,

p_{p o r t}^{\mod e 1 - \mod e 2}

indicates the power of mode2 in the port when the mode1 light provides the input. Figures 4(a) and 4(b) show that the PCLs and CTs for both the TE0 and TM0 mode are less than 0.5 and −50 dB, respectively, over a wide wavelength range from 1450 to 1650 nm. The simulation results of the total 1310-nm PSR are shown in Fig. 5

Fig. 5 FDTD simulation results of the total 1310-nm PSR with optimized parameters. (a) TE0 input. (b) TM0 input. The inset pictures display the power distribution at 1310 nm.

Download Full Size | PDF

, in which the PCLs and CTs for both the TE0 and TM0 mode are less than 0.2 and −46 dB, respectively, over a wide wavelength range from 1260 to 1360 nm.

3. Fabrication, measurements, and discussion

3.1 Fabrication and characterization method

To experimentally validate the characteristics, a series of PSRs with optimized parameters were fabricated and measured. The electron beam lithography, inductively coupled plasma, and plasma-enhanced chemical vapor deposition processes were used to fabricate the devices. Figures 6(a) and 6(b)

Fig. 6 Microscopic and scanning electron microscope pictures of the fabricated PSR. (a) 1550-nm PSR. (b) 1310-nm PSR.

Download Full Size | PDF

show the microscopic and SEM pictures of the fabricated 1550- and 1310-nm PSRs, respectively. The measurements of the PSR were conducted using systems with a tunable laser (Santec TLS-510c), a polarization controller, fiber alignment stages (Suruga-ES3700 with 50 nm alignment resolution), an optical spectrum analyzer (Yokogawa AQ6370c), and an optical power meter (Yokogawa AQ2200-221). TE and TM type grating couplers, which also work as TE and TM type on-chip polarizers, respectively, were used to receive or transmit light from/into the fibers.

3.2 Test results

The transmission spectra of the straight reference waveguides are measured first and the results are shown in Figs. 7(a), 7(b), 8(a), and 8(b)

Fig. 7 Measured results of the 1550-nm PSR; (a) four TE-type reference links; (b) four TM-type reference links; (c) PCL versus wavelength; (d) polarization CT versus wavelength.

Download Full Size | PDF

Fig. 8 Measured results of the 1310-nm PSR; (a) two TE-type reference links; (b) two TM-type reference links; (c) PCL versus wavelength; (d) polarization CT versus wavelength.

Download Full Size | PDF

. Taking the 1550-nm reference waveguides as an example, the central wavelength of TE-type and TM-type reference link are 1563.8 nm and 1563.4 nm, respectively. The 1-dB and 3-dB bandwidths for the TE-type reference links are about 21.2 nm and 39.2 nm, respectively. The 1-dB and 3-dB bandwidths for the TM-type reference links are about 20.8 nm and 41.1 nm, respectively. Four reference links with same configuration at different position of the chip are measured. Generally, the transmission spectra are almost coincident near the central wavelength. However, the spectra far away from the central wavelength have a maximum deviation bout 2 dB. A slight misalignment when testing the grating couplers will lead to big difference at the marginal wavelengths. When characterizing the performance of PSRs, we adjusted the test platform to ensure that the spectra of the PSRs overlap as much as possible with the corresponding reference grating couplers. The operation is reasonable because the variation of the transmission for central wavelengths is below 0.4 dB. Figures 7(c) and 7(d) show the measured results of PCL and CT for the 1550-nm PSR, respectively. The transmission spectra were normalized to straight reference waveguides. Over the entire measured wavelength range from 1500 to 1600 nm, the PCL shows no obvious degradation and the maximum PCL is approximately 0.74 dB. The ripples in the spectra of PCL mainly result from the grating couplers. The figures show that the CTs are less than −20 dB from 1500 to 1600 nm. Figures 8(c) and 8(d) show the measured results of PCL and CT for the 1310-nm PSR, respectively. The PCL and CTs are less than 1 and −23 dB from 1260 nm to 1340 nm, respectively. Considering that the maximum simulated CT is less than −46 dB for both the 1550 and 1310-nm PSRs, our experimental results show higher values than those of the simulated results. This may be attributed to the fabrication errors and the inaccurate calibration of the input polarization state. The demonstrated broadband operation is limited by the available bandwidth of grating couplers.

3.3 Fabrication tolerance analysis

The fabrication tolerance is investigated by varying the parameters of the PSR. For example, for the 1550-nm PSR, Figs. 9(a)–9(h)

Fig. 9 Measured PCLs of the 1550-nm PSR when the variation of waveguide separation D is (a) 10 nm, (c) 20 nm, (e) −10 nm, and (g) −20 nm, when the coupling length of the TE0–TE1 demultiplexer is (i) 70 μm and (k) 90 μm. Measured CTs when the variation of waveguide separation D is (b) 10 nm, (d) 20 nm, (f) −10 nm, and (h) −20 nm, when the coupling length of the TE0–TE1 demultiplexer is (j) 70 μm and (l) 90 μm.

Download Full Size | PDF

show the experimental results of the PSRs when the variation of parameter D is within ± 20 nm. The performance is clearly seen to be insensitive to the variation of parameter D. The maximum PCL and CT are still less than 1.3 and −20 dB, respectively, from 1500 to 1600 nm. It is worth noting that the experimental results are meaningful on the condition that all the other parameters, like slab thickness, width of waveguides, don’t change. Figures 9(i)–9(l) show the experimental results when the coupling length of the TE0–TE1 demultiplexer are 70 and 90 μm. The PSRs still display high performance. The maximum PCL and CT are approximately 0.6 and −20 dB, respectively. Figure 10

Fig. 10 Simulated TM0–TE0 PCL for the possible fabrication errors of (a) waveguide separation, (b) slab-layer height, (c) access-waveguide width, and (d) bus-waveguide width.

Download Full Size | PDF

depicts the simulated results of TM0–TE0 PCL from 1500 to 1600 nm for the possible fabrication errors of waveguide separation D, slab-layer height H_slab, access-waveguide width W_a, and bus-waveguide width W_b. The PCLs are less than 0.8 dB for the variations of D as large as ± 20 nm, W_a as large as ± 20 nm, and W_b as large as ± 20 nm, within a broadband bandwidth from 1500 to 1600 nm. The simulation results in Fig. 10(b) shows that the PSR still has high performance when the variations of slab-thickness are within ± 10 nm. However, the PCL deteriorates when the variation is −20 nm. To investigate the reason, we simulate the demultiplexer part and bi-level part separately. And the results are shown in Figs. 11(a) and 11(b)

Fig. 11 (a) Simulated TE1-TE0 loss of the demultiplexer when the slab height is 70 nm. Simulated TM0-TE1 loss of the bi-level taper for the possible fabrication errors of (b) slab-layer height, (c) waveguide width, and (d) slab waveguide width.

Download Full Size | PDF

, respectively. It can be clearly seen that the deterioration can be attributed to both the decrease coupling of the demultiplexer and the decrease TM0-TE1 conversion efficiency of the bi-level taper. The increase of the length for each section can compensate for the deterioration. Figures 11(c) and 11(d) show the simulated TM0-TE1 efficiency of the bi-level taper when the variation of waveguide width W_{n_wg} and slab width W_{n_slab} are within ± 20 nm. Clearly the bi-level taper has large fabrication tolerance against taper width.

3.4 Comparison

According to Tables 4 and 5

Table 4. Comparison of the experimentally demonstrated 1550-nm PSR.

View Table | View all tables in this article

Table 5. Comparison of the experimentally demonstrated 1310-nm PSR.

View Table | View all tables in this article

, our device shows high performance with respect to PCL, CT, and bandwidth compared with other experimentally demonstrated PSRs. Both the 1-and 20-dB bandwidths for PCL and CT, respectively, are larger than 80 nm. In addition, the PSR length is less than 130 μm and only longer than the adiabatic PSR in [10], which has a 60-nm feature size. The feature size of our 1550- and 1310-nm PSRs is 200 and 160 nm, respectively, which can be fabricated with the 180-nm CMOS processes. The upper cladding material of our PSR is SiO₂, which is compatible with most metal back-end-of-line processes.

4. Conclusion

We proposed and experimentally demonstrated high-performance TE1-assisted PSRs based on an SOI platform. The PSO and STA methods are used to optimize the TM0–TE1 bi-level taper and TE0–TE1 demultiplexer of the PSR, respectively. The measured PCLs are less than 1 dB and the crosstalks are less than −20 dB within 80-nm bandwidth for both the 1550- and 1310-nm PSRs. By using a similar method, the PSR operating at other wavelength ranges, such as in the mid-infrared area, can also be designed and fabricated. Our method provides a practical solution to design passive polarization-handling devices.

Funding

National Key Research and Development Program of China (No. 2016YFB0402505); National Natural Science Foundation of China (61575189 and 61635011).

References

1. B. Shen, P. Wang, R. Polson, and R. Menon, “An integrated-nanophotonics polarization beamsplitter with 2.4 ×2.4 µm² footprint,” Nat. Photonics 9(6), 378–382 (2015). [CrossRef]

2. 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]

3. 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]

4. 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] [PubMed]

5. H. Guan, A. Novack, M. Streshinsky, R. Shi, Q. Fang, A. E.-J. Lim, G.-Q. Lo, T. Baehr-Jones, and M. Hochberg, “CMOS-compatible highly efficient polarization splitter and rotator based on a double-etched directional coupler,” Opt. Express 22(3), 2489–2496 (2014). [CrossRef] [PubMed]

6. Y. Xiong, J. G. Wangüemert-Pérez, D.-X. Xu, J. H. Schmid, P. Cheben, and W. N. Ye, “Polarization splitter and rotator with subwavelength grating for enhanced fabrication tolerance,” Opt. Lett. 39(24), 6931–6934 (2014). [CrossRef] [PubMed]

7. K. Tan, Y. Huang, G.-Q. Lo, C. Yu, and C. Lee, “Experimental realization of an O-band compact polarization splitter and rotator,” Opt. Express 25(4), 3234–3241 (2017). [CrossRef] [PubMed]

8. L. Socci, V. Sorianello, and M. Romagnoli, “300 nm bandwidth adiabatic SOI polarization splitter-rotators exploiting continuous symmetry breaking,” Opt. Express 23(15), 19261–19271 (2015). [CrossRef] [PubMed]

9. 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]

10. W. D. Sacher, B. Peng, J. C. Rosenberg, M. Khater, Y. Martin, J. S. Orcutt, Y. A. Vlasov, W. M. Green, and T. Barwicz, “An O-band polarization splitter-rotator in a CMOS-integrated silicon photonics platform,” in Frontiers in Optics 2016 (Optical Society of America, Rochester, New York, 2016), p. FTu2D.2.

11. D. Dai and J. E. Bowers, “Novel concept for ultracompact polarization splitter-rotator based on silicon nanowires,” Opt. Express 19(11), 10940–10949 (2011). [CrossRef] [PubMed]

12. D. Dai and H. Wu, “Realization of a compact polarization splitter-rotator on silicon,” Opt. Lett. 41(10), 2346–2349 (2016). [CrossRef] [PubMed]

13. Y. Ding, H. Ou, and C. Peucheret, “Wideband polarization splitter and rotator with large fabrication tolerance and simple fabrication process,” Opt. Lett. 38(8), 1227–1229 (2013). [CrossRef] [PubMed]

14. J. Wang, M. Qi, Y. Xuan, H. Huang, Y. Li, M. Li, X. Chen, Q. Jia, Z. Sheng, A. Wu, W. Li, X. Wang, S. Zou, and F. Gan, “Proposal for fabrication-tolerant SOI polarization splitter-rotator based on cascaded MMI couplers and an assisted bi-level taper,” Opt. Express 22(23), 27869–27879 (2014). [CrossRef] [PubMed]

15. W. D. Sacher, T. Barwicz, B. J. Taylor, and J. K. Poon, “Polarization rotator-splitters in standard active silicon photonics platforms,” Opt. Express 22(4), 3777–3786 (2014). [CrossRef] [PubMed]

16. W. D. Sacher, Y. Huang, L. Ding, T. Barwicz, J. C. Mikkelsen, B. J. F. Taylor, G.-Q. Lo, and J. K. S. Poon, “Polarization rotator-splitters and controllers in a Si₃N₄-on-SOI integrated photonics platform,” Opt. Express 22(9), 11167–11174 (2014). [CrossRef] [PubMed]

17. J. Wang, B. Niu, Z. Sheng, A. Wu, W. Li, X. Wang, S. Zou, M. Qi, and F. Gan, “Novel ultra-broadband polarization splitter-rotator based on mode-evolution tapers and a mode-sorting asymmetric Y-junction,” Opt. Express 22(11), 13565–13571 (2014). [CrossRef] [PubMed]

18. C. Sun, Y. Yu, G. Chen, and X. Zhang, “A low crosstalk and broadband polarization rotator and splitter based on adiabatic couplers,” IEEE Photonics Technol. Lett. 28(20), 2253–2256 (2016). [CrossRef]

19. G. Hang, A. Novack, M. Streshinsky, S. Ruizhi, L. Yang, F. Qing, A. E. J. Lim, L. Guo-Qiang, T. Baehr-Jones, and M. Hochberg, “High-efficiency low-crosstalk 1310-nm polarization splitter and rotator,” IEEE Photonics Technol. Lett. 26(9), 925–928 (2014). [CrossRef]

20. H.-C. Chung and S.-Y. Tseng, “Ultrashort and broadband silicon polarization splitter-rotator using fast quasiadiabatic dynamics,” Opt. Express 26(8), 9655–9665 (2018). [CrossRef] [PubMed]

21. S. Martínez-Garaot, S.-Y. Tseng, and J. G. Muga, “Compact and high conversion efficiency mode-sorting asymmetric Y junction using shortcuts to adiabaticity,” Opt. Lett. 39(8), 2306–2309 (2014). [CrossRef] [PubMed]

22. S.-Y. Tseng, R.-D. Wen, Y.-F. Chiu, and X. Chen, “Short and robust directional couplers designed by shortcuts to adiabaticity,” Opt. Express 22(16), 18849–18859 (2014). [CrossRef] [PubMed]

23. D. Guo and T. Chu, “Compact broadband silicon 3 dB coupler based on shortcuts to adiabaticity,” Opt. Lett. 43(19), 4795–4798 (2018). [CrossRef] [PubMed]

24. D. Guo and T. Chu, “Silicon mode (de)multiplexers with parameters optimized using shortcuts to adiabaticity,” Opt. Express 25(8), 9160–9170 (2017). [CrossRef] [PubMed]

25. D. Guo and T. Chu, “Polarization splitter-rotators with optimized taper structures,” in 7th International Multidisciplinary Conference on Optofluidics (IMCO2017), Singapore, (2017).

26. H. Guan, Y. Ma, R. Shi, A. Novack, J. Tao, Q. Fang, A. E.-J. Lim, G.-Q. Lo, T. Baehr-Jones, and M. Hochberg, “Ultracompact silicon-on-insulator polarization rotator for polarization-diversified circuits,” Opt. Lett. 39(16), 4703–4706 (2014). [CrossRef] [PubMed]

27. Y. Zhang, M. Streshinsky, A. Novack, Y. Ma, S. Yang, A. E.-J. Lim, G.-Q. Lo, T. Baehr-Jones, and M. Hochberg, “A compact and low-loss silicon waveguide crossing for O-band optical interconnect,” Proc. SPIE 8990, 899002 (2014). [CrossRef]

	W₁	W₂	W₃	W₄	W₅
Ridge	400	400	537	545	1000
Slab	400	758	1119	1235	2600

	W₁	W₂	W₃	W₄	W₅	W₆	W₇	W₈	W₉	W₁₀	W₁₁
Ridge	400	439	475	509	540	570	598	625	651	676	700
Slab	400	738	963	1145	1302	1442	1570	1688	1798	1902	2000

	Initial W_a (μm)	W_b (μm)	Coupling length (μm)	Minimum edge-gap (μm)	κ_max (μm⁻¹)	c
1550-nm	0.432	1	80	0.2	0.11328	0.5758
1310-nm	0.347	0.8	70	0.16	0.12272	0.6119

Reference	Structure or principle	PCL (dB)	CT (dB)	Bandwidth (nm)	Feature size (nm)	Length (μm)	CMOS-compatibility
[4]	Mode interference	0.6	12	35	100	36.8	No
[5]	Mode interference	0.5	20	30	200	27	Yes
[12]	Bi-level taper + ADC + mode filter	1.5	20	47	120	~70	No
[13]	Bi-level taper + MMI	2.5	12	100	y-corner	190	No
[15]	Adiabaticity	1.5	13	50	200	475	Yes
[16]	Adiabaticity	1.5	19	80	200	> 450	Yes
[18]	Adiabaticity	1	14	100	y-corner	272	Yes
This work*	Adiabaticity	0.74	20	100	200	130	Yes
This work*	Adiabaticity	0.6	20	100	200	120	Yes

Reference	Structure or principle	PCL (dB)	CT (dB)	Bandwidth (nm)	Feature size (nm)	Length (μm)	CMOS-compatibility
[19]	Mode interference	2	20	40	y-corner	-	Yes
[7]	Mode interference	-	18	100	200	~20	Yes
[10]	Adiabaticity	1	27	60	60	~110	Yes
This work*	Adiabaticity	1	23	80	160	120	Yes

Broadband and low-crosstalk polarization splitter-rotator with optimized tapers

Abstract

1. Introduction

2. Design and simulation

2.1 Structure and principle

2.2 Design of the bi-level taper

2.3 Design of the TE0–TE1 demultiplexer

2.4 Simulation results of the PSR

3. Fabrication, measurements, and discussion

3.1 Fabrication and characterization method

3.2 Test results

3.3 Fabrication tolerance analysis

3.4 Comparison

4. Conclusion

Funding

References

Cited By

Figures (11)

Tables (5)

OSA Continuum