Open Access
Add to BibSonomy
Abstract
In this paper, a compact high-resolution two-stage interleaved arrayed waveguide grating (AWG) system with a 3D structure on a silicon nitride (Si3N4) platform is proposed. The device is comprised of a 7-channel primary AWG with a 0.4-nm resolution and seven 26-channel second-stage AWGs, each with a 2.8-nm resolution. Different arrayed waveguide widths are utilized to achieve the wavelength tuning of the second-stage AWGs. The AWGs have a greater fabrication tolerance than conventional AWGs. A taper-MMI input structure is utilized to make the -3 dB pass-band reach 0.4 nm, which is 100% of the channel spacing. Also, the horizontal slot arrayed waveguides are individually introduced into the AWG to reduce the inter-layer crossing and bending losses, which is also found to greatly reduce the footprint of the device. The proposed AWGs have an average crosstalk of about -24 dB in the 2-µm band, demonstrating the feasibility of the on-chip 3D optoelectronic integration design.
© 2020 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Arrayed waveguide grating (AWG) is a type of multiplexer/de-multiplexer with the best-rated comprehensive performance, and plays an important role in many dense wavelength division multiplexing (DWDN) systems and modules [1,2]. AWG is also widely utilized in optical systems, such as optical sensors [3] and spectrometers [4]. Many researchers are committed to developing high-performance AWG based on different materials and structures [5,6,7]. Guido Maier and Mario Martinelli et al. theoretically proposed and proved the multi-stage network structure in [8], which was found to provide high resolution and large free spectral ranges [9]. Two structures exist for the design of two-stage AWGs. The first is the two-stage banded structure [10], which utilizes a low-resolution AWG as the first stage to output a set of broadband signals, and several fine-pitch filters act as the second stage to further separate these groups into individual channels. The second structure, namely the two-stage interleaved structure [11], is the opposite; a high-resolution AWG is utilized as the first stage, and several coarse AWGs are utilized in the second stage as the filters to selectively separate the pre-filtered input spectra from the first-stage AWG with relatively low adjacent-channel crosstalk, and the final partial signal is then output. Compared with banded AWGs, interleaved AWGs achieve lower crosstalk due to the two-stage filtering of their interleaved structure; however, this also increases the fabrication accuracy requirement, with banded AWGs requiring a greater spectral band and flatness of the first-stage outputs [11]. B. I. Akca and C. R. Doerr demonstrated a two-stage interleaved AWGs spectrometer with 1 nm resolution over a 75-nm bandwidth centered at 1550 nm that was actualized on a silicon nitride (Si3N4) platform [9]. Recently, A. V. Wijk et al. [12] presented a two-stage banded AWGs on a Si3N4 platform, which had a cascaded configuration with a 1 × 3 flat-pass-band AWG as the primary filter and three 1 × 70 AWGs as secondary filters, along with a multiple-input multi-mode interference (MMI) coupler utilized as the input to achieve an ultra-broad-bandwidth.
Via the utilization of the multi-stage structure and the design of an AWG with an appropriate number of channels and spacing, the on-chip layout can be optimized to decrease the footprint of the device; however, the device size remains an important problem for on-chip optoelectronic integration [12]. A good solution is combining the multi-stage structure with the multi-layer structure [13], and distributing each stage in 3D-direction. In previous research, the vertical couplers for inter-layer coupling were designed, and an attempt was made to implement the multi-layer design of two-stage AWGs [14]. In the present study, a two-stage interleaved AWGs centered at 1950nm with a tri-layer structure is demonstrated. The method of changing the width of the arrayed waveguides to achieve the wavelength tuning of the second-stage AWGs is utilized, which greatly improves the fabrication tolerance. A taper-MMI structure is utilized as the first-stage AWG input and the output channel bandwidth (-3 dB) reaches 100% of the channel spacing. Also, to reduce the crossing loss caused by the frequent crossing of the arrayed waveguides between AWGs of different layers, the slot arrayed waveguides are introduced into the device, which not only decreases the transmission loss, but also optimizes the footprint of the device. Finally, the proposed device is simulated with an insertion of -3.1 dB and an average crosstalk of about -24 dB. The present study provides an exploration of 2 µm band Si3N4 and 3D photonic platform optical devices, which is greatly significant and can be utilized for reference to develop 2-µm optical communications [15,16], infrared spectrometers [17], and astrophotonics devices [18,19].
2. Principle and structural design
2.1 Interleaved AWGs system
Figure 1 illustrates the layout of multi-layer two-stage interleaved AWGs. The entire device is divided into three layers in the vertical direction, and the different structures of waveguides in each layer are shown in Figs. 1(b) and (c). The first-stage AWG with slot arrayed waveguides and a taper-MMI input structure is designed in the middle layer. Seven second-stage AWGs are distributed in the upper and lower layers of the first-stage AWG, with three in the first layer (S2, S4, S6) and four in the third layer (S1, S3, S5, S7). The first-stage and second-stage AWGs are connected via an inverted taper vertical coupler, as indicated by the black-dotted box in Fig. 1(a).
Fig. 1. (a) Layout of the multi-layer two-stage interleaved AWGs. (b) Waveguides (rectangular and slot) structure of the second layer. (c) Waveguides structure of the first and third layers.
2.2 AWG and waveguide designs
For the two-stage interleaved AWGs, the wavelength difference between the corresponding output channels of the second-stage AWGs is also the spectral resolution of the entire device. In the AWG design, the central wavelength shifts of the second-stage AWGs are actualized by changing the length increment between the adjacent arrayed waveguides of each AWG. For high-resolution interleaved AWGs, this length increment variation is minuscule [9], which undoubtedly places a high demand on the layout design and fabrication. In essence, the wavelength shift between the corresponding output channels of the second-stage AWGs is achieved by changing the optical path difference between the arrayed waveguides. Therefore, the width of the arrayed waveguides could be changed to affect the effective index of refraction (nc), thereby actualizing wavelength drift [20]. Figure 2(a) presents the variation of the waveguide width for every 1 × 10−4 change in nc under different thicknesses for a Si3N4 rectangular waveguide. For an ultrathin waveguide, such as a 0.05 µm × 4.0 µm Si3N4 waveguide, this variation is considerable. Simultaneously, the ultrathin waveguide also introduces a large bending radius and transmission loss, as shown in Fig. 2(b); therefore, the waveguide selection requires a trade-off. In the present study, to maximize the effect, Si3N4 waveguides were chosen with the first stage of 0.15 µm × 1.5 µm and the second stage of 0.05 µm × 4.0 µm, as shown in Fig. 2(c). Simultaneously, the proposed two-stage interleaved AWGs (1 + 7) worked at 2-µm band TE0 mode and centered at 1950nm.
Fig. 2. (a) The variation of the width of a Si3N4 rectangular waveguide for every 1 × 10−4 change in nc under different thicknesses; the red line indicates the initial width of the waveguides with the corresponding thickness. (b) The confinement factor and bending radius of waveguides with different thicknesses. (c) Cross-sections of the two-stage AWGs waveguides.
The spectral resolution of the device was designed as 0.4 nm, so the width increment of the second-stage AWGs arrayed waveguides was approximately 400 nm according to Fig. 2(a). In this design, the central wavelengths (λ0) of the 7 second-stage AWGs (S1-S7) were respectively 1950.0, 1950.4, 1950.8, 1951.2, 1951.6, 1952.0 and 1952.4 nm. The corresponding arrayed waveguide widths (aw) were obtained via simulation at 4.00, 4.39, 4.79, 5.24, 5.75, 6.32 and 7.00 µm, as shown in Fig. 3.
Finally, the design parameters (i.e., the central wavelength (λ0), spectral resolution (Δλ), diffraction order (m), free spectral range (FSR), focal length (R), length increment between adjacent arrayed waveguides (ΔL), width of arrayed waveguides (aw), number of arrayed waveguides (Ng), and number of output channels (Nch)) of the first-stage and second-stage AWGs are given in Table 1.
Table 1. Design parameters of the two-stage interleaved AWGs
2.3 Taper-MMI input structure
The output spectrum could be flattened by introducing the multi-mode interferometer (MMI) structure at the end of the input waveguide [21]. However, the introduction of the square MMI means that the waveguide will have saltation in the light transmission direction [22], which will increase the insertion loss. A trigonometric combined taper-MMI structure was therefore introduced into the input waveguide of the first-stage AWG, as shown in Fig. 4(a), which reduces the loss caused by the saltation in the waveguide without deteriorating the spectral performance. z is the direction of light transmission, y is the direction of waveguide width expansion, and the taper satisfied the equation shown in Formula (1). The entire length of the structure is Lt, the taper length is Ltaper, and the input and output widths are respectively Win and Wout. According to the self-mapping principle, the input light field could form two images on the MMI output image surface, and the distance between the two images is Y, as shown in Fig. 4(a). The two self-mappings in the AWG output channel represent spectral broadening, which is associated with Y. Figure 4(b) shows that the spectral simulation of the first-stage AWG central output channel varied with Y. In particular, when Y was 5.5 µm, the -3 dB bandwidth of the output spectral reached 0.4 nm, which was 100% of the entire channel spacing.
Fig. 4. (a) The structural schematic diagram of the taper-MMI input structure. (b) The spectral simulation of the first-stage AWG central output channel varies with Y.
The values of Lt, Win and Wout were respectively chosen as 50, 1.5 and 10 µm. The variations of the transmission and Y with the change of Ltaper from 0 to 50 µm were simulated, and are shown in Fig. 5(a). The transmission loss of the device was found to decrease with the increase of Ltaper. However, when Ltaper was greater than 40 µm, Y began to decrease from 5.5 µm, so the taper length was set as 40 µm. The spectral characteristics of the device were also simulated, as shown in Fig. 5(b). The device maintained a reliable performance in the bandwidth range of 1910-1990nm. Figure 5(c) shows the optical field of the taper-MMI input at the wavelength of 1950nm. The outputs of the central channel of first-stage AWG with different input structures are shown in Fig. 5(d), and the flatness of the output was found to significantly improved.
Fig. 5. (a) Simulation of the transmission and Y varies with Ltaper. (b) Simulation of the transmission and Y varies with the wavelength from 1910 to 1990nm. (c) The optical field of the taper-MMI input. (d) The output of the central channel of the first-stage AWG with different input structures.
2.4 Vertical coupler
The selection of ultra-thin waveguides inevitably leads to the increase of the device size, which can be effectively solved by the design of the first-stage and second-stage AWGs of the two-stage interleaved AWGs in different waveguide layers. The vertical coupler [23] was utilized to solve the problem of light coupling between the different layers of multi-layer platforms. In the research by Bai et al. [14], an inverted taper coupler was proposed to actualize the interconnection between two stages of AWGs in different layers, as shown in Fig. 6, and was utilized in the multi-layer AWGs design in the present study. A tri-layer structure was chosen to reduce the crosstalk and crossing loss between the upper and lower layers. The gradient-varying waveguide structure achieved better phase matching. For the TE0 mode with a gap of 2400 nm, the coupler realized the low-loss inter-layer coupling from 1910 to 1990nm.
Fig. 6. (a) The spatial structure of the tri-layer gradient vertical coupler based on the Si3N4/SiO2 platform. (b) Design parameters of each vertical coupler.
2.5 Slot arrayed waveguides
To reduce the crossing loss caused by the frequent crossing of the arrayed waveguides between the AWGs of different layers, the slot arrayed waveguide structure was introduced to the design of the first-stage AWG. The strong constraint on the light field of the slot waveguide [24] simultaneously solves the bending loss problem caused by the ultra-thin waveguide, which further compacts the device layout. Figure 7(a) shows the cross-section of a slot waveguide with a horizontal structure. A low-refractive-index horizontal sandwich (0.15 µm × 0.7 µm Si3N4) was added between two high-refractive-index waveguides (0.22 µm × 0.7 µm Si). For the TE0 mode, the light field was symmetrically distributed in the upper and lower high-refractive-index waveguides, while the light field of the TM0 mode was restricted to the middle region with the low-refractive-index [25], as shown in Figs. 7(b) and (c). Figure 7(d) shows the crossing losses between the lower slot waveguide, as well as the original rectangular waveguide (0.15 µm × 1.5 µm), and the upper rectangular waveguide (0.05 µm × 4.0 µm), with the variation of the crossing angle from 0° to 90°. The results demonstrate that, compared with the original waveguide [14], the inter-layer crossing loss was negligible when the slot arrayed waveguides were utilized. The bending losses of the slot waveguide under different bending radii are shown in Fig. 7(e). The slot waveguide remained loss-free when the bending radius was only 10 µm. Also, at the wavelength of 1950nm, the TE0 mode had an effective refractive index of 1.4658 in the rectangular waveguide and 2.5475 in the slot waveguide; therefore, the length increment between the adjacent arrayed waveguides (ΔL) could be reduced from 851.41 µm (in Table 1) to 489.89 µm with the slot arrayed waveguides. Therefore, the introduction of the slot waveguide structure into the arrayed waveguide not only reduced the crossing and transmission losses, but also eliminated the limitation of the waveguide bending radius on the AWG layout design and reduced the AWG footprint.
Fig. 7. (a) Cross-section of the slot waveguide with a horizontal structure. (b) The TE0 mode field in the slot waveguide. (c) The TM0 mode field in the slot waveguide. (d) The crossing losses of the slot case and the original case vary with the crossing angle. (e) The bending loss of the slot waveguide varies with the bending radius.
Furthermore, an inverted tapered coupler was designed to actualize the coupling between the Si3N4 rectangular waveguide and the Si/Si3N4 slot waveguide, as shown in Fig. 8(a). The Si3N4 rectangular waveguide (0.15 µm × 1.5 µm) gradually transitioned into the slot region of the Si slot waveguide (0.59 µm × 0.7 µm) and merged with the slot waveguide via two inverted tapers. Figure 8(b) shows the variations of the transmission of the coupler with the taper width (at) and length (Lc). Considering the device size and fabrication, at was set as 0.05µm and Lc was set as 80 µm, which achieved the coupling loss of 0.31 dB. The simulation of the light field from the rectangular waveguide to the slot waveguide, and then coupled to the rectangular waveguide again, is exhibited in Fig. 8(c).
Fig. 8. (a) The structure of the slot waveguide coupler. (b) The relationship between the transmission efficiency and coupling length. (c) The optical field of the coupler.
In view of the complex multilayer structure of the device, the process fabrication, especially the vertical alignment, is a challenge that will be investigated in-depth in future research. For practical fabrication processing, the device will be fabricated by deep-UV lithography stepper technology on SiO2 wafers. First, the stoichiometric Si3N4 waveguide layer of 50-nm core thicknesses on the first layer of the platform can be obtained by low-pressure chemical vapor deposition (LPCVD) at 800 °C, and inductively-coupled plasma (ICP) etching will then be used to define the waveguides. For the second layer, in addition to the 150-nm Si3N4 layer, the Si layer of the slot arrayed waveguides can be deposited by plasma-enhanced chemical vapor deposition (PECVD). Finally, the Si3N4 waveguides of the third-layer will be the same as those of the first layer. Between each layer, SiO2 will be deposited as interlayer cladding and upper cladding, and will be planarized by chemical mechanical polishing (CMP) to achieve a smooth surface.
3. Simulation of the device
The transmission spectrum of the AWGs from 1910 to 1990nm is shown in Fig. 9(a). The proposed device has uniformity over the entire FSR. In particular, the results around 1930, 1950, and 1970nm, as indicated by the red-dotted boxes in Fig. 9(a), are specifically shown in Figs. 9(a), 9(b) and 9(c). The flatness of the output spectrum at 1950nm was greatly improved by utilizing the taper-MMI input structure, and exhibited a slight deterioration in the range around 1930 and 1970nm. This was due to the design error between the two stages of the AWGs, which could be improved by selecting the appropriate number and spacing of the channels and the diffraction order. The channel crosstalk of seven second-stage AWGs are presented in Fig. 10. The average crosstalk of each AWG was approximately -24 dB and exhibited a slight fluctuation in the edge channels, which is consistent with Fig. 9. In addition, the performance and parameters of the proposed device relative to those of the devices proposed in other studies are listed in Table 2. Compared with previous design, the proposed device has both a large number of channels (182) and a narrow channel spacing (0.4 nm), and exhibits an excellent crosstalk performance. Due to the limitation of the waveguide and working wavelength, the size of the device may be a problem and worthy of further study to expand its application advantages.
Fig. 9. (a) The output of the two-stage interleaved AWGs simulated from 1910 to 1990nm. (b-d) The locally enlarged views of A (1930nm), B (1950nm), and C (1970nm).
Table 2. Performance comparison of AWGs
4. Conclusion
In summary, a compact and high-resolution two-stage interleaved AWGs with a 3D structure was proposed, and its performance was simulated. The wavelength tuning of the second-stage AWGs was actualized by changing the arrayed waveguide width, which not only improved the fabrication tolerance, but also reduced the complexity of the layout design. A taper-MMI structure was implemented at the entrance of the first-stage AWG to make the pass-band (-3 dB) of the output be 100% of the channel spacing. In particular, slot arrayed waveguides were introduced in the first-stage AWG to reduce the crossing loss with the second-stage AWGs in the vertical direction and optimize the layout. A coupler was also proposed to actualize the low-loss coupling of light from a rectangular waveguide to a horizontal slot waveguide. Finally, the tri-layer two-stage interleaved AWGs layout with a footprint of 2.0 cm × 2.1 cm was demonstrated. The device centered at 1950nm was found to have a high spectral resolution of 0.4 nm and an FSR of 72.8 nm, and it exhibited a great performance from 1920 to 1980nm with crosstalk of approximately -24 dB.
Funding
National major scientific research instrument development project (61627802); National Key Research and Development Program of China (2019YFB2205300).
Disclosures
The authors declare no conflict of interest.
References
1. N. Juhari, M. F. Yusof, M. H. A. Wahid, and N. S. Ismail, “Design of arrayed waveguide grating (AWG) demultiplexer based PMMA for narrow channel spacing,” AIP Conf. Proc. 2203(1), 020056 (2020). [CrossRef]
2. Z. Y. Zhang, Y. Wang, and H. K. Tsang, “Ultracompact 40-Channel Arrayed Waveguide Grating on Silicon Nitride Platform at 860 nm,” IEEE J. Quantum Electron. 56(1), 1–8 (2020). [CrossRef]
3. D. H. Geuzebroek, A. van Rees, E. J. Klein, and K. Lawniczuk, “Ultra-wide band (400-1700nm) integrated spectrometer based on arrayed waveguide gratings for spectral tissue sensing,” Proc. IEEE 14th Int. Conf. Group IV Photon, 83–84 (2017).
4. R. A. Crocombe, “Miniature optical spectrometers: The art of the possible, Part IV: New near-infrared technologies and spectrometers,” Spectroscopy 23(6), 26 (2008).
5. Q. Han, J. S. Yves, Y. X. Chen, M. Menard, and W. Shi, “Polarization-insensitive silicon nitride arrayed waveguide grating,” Opt. Lett. 44(16), 3976–3979 (2019). [CrossRef]
6. X. Xia, T. T. Lang, L. B. Zhang, and Z. H. Yu, “Reduction of non-uniformity for a 16 16 arrayed waveguide grating router based on silica waveguides,” Appl. Opt. 58(4), 1139–1145 (2019). [CrossRef]
7. P. Yuan, X. G. Zhang, J. M. An, P. G. Yin, Y. Wang, and Y. D. Wu, “Improved Performance of a Wavelength-Tunable Arrayed Waveguide Grating in Silicon on Insulator,” Chinese Phys. Lett. 36(5), 054204 (2019). [CrossRef]
8. G. Maier, M. Martinelli, A. Pattavina, and E. Salvadori, “Design and cost performance of the multistage WDM-PON access networks,” J. Lightwave Technol. 18(2), 125–143 (2000). [CrossRef]
9. B. I. Akca and C. R. Doerr, “Interleaved Silicon Nitride AW G Spectro meters,” IEEE Photonics Technol. Lett. 31(1), 90–93 (2019). [CrossRef]
10. K. Takada, M. Abe, T. Shibata, and K. Okamoto, “10-GHz-Spaced 1010-Channel Tandem AWG Filter Consisting of One Primary and Ten Secondary AWGs,” IEEE Photonics Technol. Lett. 13(6), 577–578 (2001). [CrossRef]
11. T. K. Chan, J. Karp, R. Jiang, N. Alic, S. Radic, C. F. Marki, and J. E. Ford, “2007-Channel 2-D Array Demultiplexer for Ultralarge Data Bandwidth,” J. Lightwave Technol. 25(3), 719–725 (2007). [CrossRef]
12. A. V. Wijk, C. R. Doerr, Z. Ali, M. Karabiyik, and B. I. Akca, “Compact ultrabroad-bandwidth cascaded arrayed waveguide gratings,” Opt. Express 28(10), 14618–14626 (2020). [CrossRef]
13. S. D. Nicolás and M. Lipson, “Scalable 3D dense integration of photonics on bulk silicon,” Opt. Express 19(18), 17758–17765 (2011). [CrossRef]
14. N. F. Bai, X. T. Zhu, Y. J. Zhu, W. Hong, and X. H. Sun, “Tri-layer gradient and polarization-selective vertical couplers for interlayer transition,” Opt. Express 28(15), 23048–23059 (2020). [CrossRef]
15. H. Zhang, N. Kavanagh, Z. Li, J. Zhao, N. Ye, Y. Chen, N. V. Wheeler, J. P. Wooler, J. R. Hayes, S. R. Sandoghchi, F. Poletti, M. N. Petrovich, S. U. Alam, R. Phelan, J. O. Carroll, B. Kelly, L. Grüner-Nielsen, D. J. Richardson, B. Corbett, and F. C. Garcia Gunning, “100 Gbit/s WDM transmission at 2 µm: transmission studies in both low-loss hollow core photonic bandgap fiber and solid core fiber,” Opt. Express 23(4), 4946–4951 (2015). [CrossRef]
16. H. Zhang, M. Gleeson, N. Ye, N. Pavarelli, X. Quyang, J. Zhao, N. Kavanagh, C. Robert, H. Yang, P. E. Morrissey, A. Gocalinska, E. Pelucchi, O. Brien, F. Peters, B. Corbett, and F. Gunning, “Dense WDM transmission at 2 µm enabled by an arrayed waveguide grating,” Opt. Lett. 40(14), 3308–3311 (2015). [CrossRef]
17. E. Ryckeboer, A. Gassenq, M. Muneeb, N. Hattasan, S. Pathak, L. Cerutti, J. B. Rodriguez, E. Tournie, W. Bogaerts, R. Baets, and G. Roelkens, “Silicon-on-insulator spectrometers with integrated GaInAsSb photodiodes for wide-band spectroscopy from 1510 to 2300 nm,” Opt. Express 21(5), 6101–6108 (2013). [CrossRef]
18. P. Gatking, S. Veilleux, Y. W. Hu, J. B. Hawthorn, and M. Dagenais, “Arrayed waveguide grating spectrometers for astronomical applications: new results,” Opt. Express 25(15), 17918–17935 (2017). [CrossRef]
19. P. Gatkine, S. Veilleux, and M. Dagenais, “Astrophotonic Spectrographs,” Appl. Sci. 9(2), 290 (2019). [CrossRef]
20. X. T. Zhu, W. Hong, N. F. Bai, and X. H. Sun, “A multi-stage high resolution sub-wavelength arrayed waveguide grating,” Proc. SPIE 11455, 114555D (2020). [CrossRef]
21. N. R. Huang, Q. Wang, and S. L. He, “Study of AWG basedon MMI,” Acta Photonica Sinica 32(4), 413–416 (2003).
22. L. Y. Zhang, Y. D. Wu, J. M. An, Y. Wang, L. L. Wang, P. Pan, J. S. Zhang, X. G. Zhang, and X. W. Hu, “Flat Spectural Response of Arrayed Waveguide Grating,” Acta Photonica Sinica 42(4), 379–385 (2013). [CrossRef]
23. G. Liu, K. Shang, S. J. B. Yoo, S. Pathak, S. Feng, S. Li, and W. Lai, “Silicon nitride tri-layer vertical Y-junction and 3D couplers with arbitrary splitting ratio for photonic integrated circuits,” Opt. Express 25(9), 10474–10483 (2017). [CrossRef]
24. E. Samoi, Y. Benezra, and D. Malka, “An ultracompact 3×1 MMI power-combiner based on Si slot-waveguide structures,” Photonics Nanostruct.-Fundamentals Appl. 39, 100780 (2020). [CrossRef]
25. G. Q. Chen, L. X. Chen, W. Q. Ding, F. K. Sun, and R. Feng, “Ultrashort slot polarization rotator with double paralleled nonlinear geometry slot crossings,” Opt. Lett. 38(11), 1984–1986 (2013). [CrossRef]
26. Y. Liu, Z. Li, D. Li, Y. Yao, J. B. Du, Z. Y. He, and K. Xu, “Thermo-optic tunable silicon arrayed waveguide grating at 2-um wavelength band,” IEEE Photonics J. 12(4), 1–8 (2020). [CrossRef]
