Open Access
Add to BibSonomy
Abstract
We propose and demonstrate a type of high-performance transverse magnetic (TM) multimode interferometer (MMI) in Z-cut thin film lithium niobate (TFLN). Both 1 × 2 and 4 × 4 MMI designs are demonstrated. Simulation results show that the insertion losses (ILs) are nominally about 0.157 and 0.297 dB for the 1 × 2 and 4 × 4 MMI, respectively, with wide fabrication tolerances. Based on the designed structure, the MMIs are fabricated using an argon based induced coupled plasma (ICP) etching method in Z-cut TFLN. The measured ILs are 0.268 and 0.63 dB for these two kinds of devices. The presented TM mode MMI featuring compact size and low loss can be used for both multifunctional devices and on-chip integrated circuits on a Z-cut TFLN platform.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Single crystal lithium niobate (LN) plays a great role in photonics due to its large second order susceptibilities and wide transparency [1,2]. The typical bulk LN crystal has been widely used in modulators for fiber-optical communications [2,3]. In these devices, ion diffusion is used to introduce a refractive index difference, which results in poor light confinement and a larger device size [2,3]. The recently developed high quality thin film LN (TFLN) has greatly encouraged the development of on-chip integrated devices, and many different kinds of high performance TFLN-based devices have been demonstrated [4–11]. However, most of the reported results are based on X-cut LN, especially for modulators. It’s expected that a Z-cut LN based device with metal electrodes placed parallel with the crystal orientation will maximize the second order susceptibility as the electrical field has more overlapped area with the optical field [2] when compared to the X-cut case. Accordingly, the TM mode is desirable in a Z-cut LN based device as r33 will only maximize its effect when light is polarized vertically [4,12]. In this scenario, a high-performance transverse magnetic (TM) multimode interferometer (MMI) serves as a power splitter or hybrid MMI, which is indispensable. The MMI is also very important for making future larger scale on-chip integrated circuits, just as it is demonstrated in its silicon photonics (SiPs) and III-V counterparts [13–15]. Although a 1 ${\times} $ 2 MMI is demonstrated both in bulk LN and TFLN, its large size [16] and significant loss [5] limit its potential in large scale integration. In addition, most published results are based on the transverse electric (TE) mode, suitable for the application of X-cut TFLN-based devices. Additionally, a 4 ${\times} $ 4 MMI with application as a multiple-branch splitter and hybrid MMI [17–19] also has good prospects and there has been no demonstration in either bulk LN and TFLN up to now.
Therefore, we demonstrate compact and low loss TM mode MMIs based on Z-cut TFLN in this manuscript. Both the design of 1 ${\times} $ 2 and 4 ${\times} $ 4 MMIs are described in detail. The simulated insertion losses (ILs) for the 1 ${\times} $ 2 and 4 ${\times} $ 4 MMI are about 0.157 and 0.297 dB, respectively. Both types of MMIs have a large fabrication tolerance according to the simulation results. As a proof-of-principle demonstration, the devices are fabricated in 700 nm-thick Z-cut TFLN using an argon-based induced coupled plasma (ICP) etching method. The measured ILs for 1 ${\times} $ 2 and 4 ${\times} $ 4 MMI are found to be about 0.268 and 0.63 dB. Compared to devices discussed in current literature, this device features a compact size and low IL, which validates its superior performance and suitability for application of integration in the Z-cut TFLN platform. The proposed 1 ${\times} $ 2 MMI can be used as an equal power splitter for many important application scenarios, while the 4 ${\times} $ 4 MMI can serve as a 90-degree hybrid mixer for coherent signal demultiplexing.
2. Principle and design
The schematic structure of the 1 ${\times} $ 2 and 4 ${\times} $ 4 MMI is shown in Fig. 1. An MMI can be divided into three parts: input single mode region, multimode region and output single mode region. The input single mode light will activate several higher-order modes inside the multimode region. These higher-order modes with different propagation constants will periodically interfere and be imaged at specific positions, a process which is usually called self-imaging and is the general working principle of an MMI [17]. Several outputs can be obtained if the single mode waveguide is put at a particular image plane. The self-imaging can be divided into symmetric and general imaging cases. For symmetric imaging, the input single mode waveguide is positioned at the center of the multimode region, as shown in Fig. 1 (a). The Mth N-fold self-imaging position can be described by formula (1) according to guided-mode propagation analysis [17],
Fig. 1. Schematic structures of (a) 1 ${\times} $ 2 and (b) 4 ${\times} $ 4 MMIs. Ii and Oi denote the ith input and output ports, respectively. The inset coordinate axis shows the crystal orientation.
Before starting to design the device, we first calculated the basic waveguide structure using the 3D finite difference time domain (FDTD) method to realize better mode confinement. As mentioned before, all the simulation and measurements in this manuscript use TM-polarized light. During the simulation, the waveguide width and length are set as 1.3 and 10 μm, respectively. Figure 2(a) shows the simulated transmission results with respect to different waveguide depths when the etch ratios (defined as etching depth divided by total depth) are 40% and 50%, respectively. If the waveguide depth were to be too small, most of the light would radiate out and result in high (radiative) loss. It can be noted that there is not significant improvement when the waveguide depth is more than about 700 nm under either etch ratios. Figure 2(b) shows the transmission results with different etch ratios when the waveguide depth is 700, 900 and 1100 nm. It can be observed in Fig. 2(b) that a higher etch ratio will result in lower radiative loss, as the optical mode is confined better. The improvement of transmission reaches saturation when the etch rate is above 40%. However, too high etch ratio will create a big challenge for processing due to the hard-to-etch property of TFLN, and resulting sidewall roughness can subsequently create detrimental propagation loss. In any case, a ridge waveguide is desired as it is compatible with the modulator process in terms of electrode location [7,10]. Therefore, we choose a waveguide depth and etch depth as 700 and 300 nm (corresponding to a 42.86% etch ratio), respectively, after considering both mode confinement and fabrication process.
Fig. 2. Simulated transmission results with respect to (a) waveguide depth and (b) etch ratio. (Inset in Fig. 2(b) shows the TM mode profile under 700 nm height and 300 nm etch depth.)
For a 1 ${\times} $ 2 MMI, the length (LM) and width (WM), which are the key parameters, are simulated for different cases, as shown in Fig. 3(a). Based on the 3D FDTD method, the transmission spectra of the 1 ${\times} $ 2 MMI when LM changes from 15.0 to 16.5 μm are simulated, with results shown in Fig. 3(a). For a short device, the IL will be large and the center wavelength (WL) will shift to the longer band region. While the center WL will shift to the short wavelength range with LM above 16.0μm, it still will only result in a small improvement in the loss. LM is thus set at 15.5 μm to realize a 1550 nm center WL and reasonable IL. Figure 3(b) shows the simulated transmission spectra with respect to different WM. The center WL will deviate from 1550 nm when the WM diverges from 5.0 μm. Therefore, WM is set at 5.0 μm. A comparison of the results of the 1 ${\times} $ 2 MMI with and without tapers is also analyzed, as shown in Fig. 3(c). It can be clearly observed from Fig. 3(c) that introducing a taper will result in better performance in terms of loss. Here, the taper length is 10 $\mathrm{\mu}\textrm{m}$, and width is changed from 1.3 to 2.0 $\mathrm{\mu}\textrm{m}$. Figure 3(d) shows the simulated mode profile for a 1 ${\times} $ 2 MMI when LM and WM are 15.5 and 5.0 μm, respectively. A clear mode distribution validates its high performance. Based on the optimized structure, the total IL of the 1 ${\times} $ 2 MMI at the 1550 nm wavelength is calculated at about -0.157 dB.
Fig. 3. Simulated transmission spectra of 1 ${\times} $ 2 MMI for different multimode waveguides of (a) length LM and (b) width WM. (c) Simulated transmission spectra comparison for 1 ${\times} $ 2 MMI with and without tapers. (d) Simulated mode distribution of the 1 ${\times} $ 2 MMI when LM and WM are 15.5 and 5.0 $\mathrm{\mu}\textrm{m}$, respectively. (Inset in Fig. 3(a) shows a schematic top view of the 1 ${\times} $ 2 MMI.)
The same design methodology is used for the 4 ${\times} $ 4 MMI design. The difference is that the 2.5D variational FDTD simulation method is used instead due to its larger size. Figure 4(a) shows the simulated transmission spectra with respect to different LM for a fixed WM when TM-polarized light is injected from Port 2 (as shown in Fig. 1(b)). All four ports have equal output power and the center WL is near 1550 nm when LM is 547 μm. The center WL will shift and the output of different ports will lose consistency if the LM diverges from such value. Therefore, the LM is set at 547 μm. Figure 4(b) shows the simulated transmission spectra with relation to WM when LM is fixed. The optimum width is around 20 μm according to Fig. 4(b). For the 4 ${\times} $ 4 MMI, the introduce of a taper will also result in better device performance. The simulated different taper width regarding to transmission spectra is shown in Fig. 4(c). We found that the optimum taper width is around 3.1 μm, which results in uniformed output spectra. Therefore, the LM, WM and taper width of the 4 ${\times} $ 4 MMI are 547, 20 and 3.1 μm, respectively. Under such conditions, the total IL of the 4 ${\times} $ 4 MMI is calculated as -0.297 dB at 1550 nm wavelength. The simulated mode distribution of the 4 ${\times} $ 4 MMI is shown in Fig. 4(d).
Fig. 4. Simulated transmission spectra of the 4 ${\times} $ 4 MMI for different multimode waveguides (a) length LM, (b) width WM and (c) taper width. (d) Simulated mode distribution of the 4 ${\times} $ 4 MMI when LM and WM are 547 and 20 $\mathrm{\mu}\textrm{m}$, respectively.
There is always unavoidable error during the fabrication process of photonics devices. Therefore, the tolerance is analyzed for both the 1 ${\times} $ 2 and 4 ${\times} $ 4 MMIs. Figure 5(a) shows the transmission coefficient under different LM for the 1 ${\times} $ 2 MMI. It is observed that a short LM will result in large loss while the center wavelength of the spectrum shifts away from 1550 nm for longer LM. About ${\pm} \,$ 0.5 μm tolerance is expected with consideration of loss and center wavelength when the LM is set at 15.5 μm in our design. Similarly, about ${\pm} \,$ 0.1 μm tolerance can be observed for the WM of 1 ${\times} $ 2 MMI from Fig. 7(b). For the 4 ${\times} $ 4 MMI, the LM tolerance is ∼1 μm according to the simulated results in Fig. 7(c), under which low loss and acceptable equal output power can be obtained. Figure 7(d) shows that all four ports have uniformed output power when the WM ranges from 19.5 to 20.1 μm. For both the cases of 1 ${\times} $ 2 and 4 ${\times} $ 4 MMIs, the tolerance of WM is lower than that of the LM, which means the width of the multimode waveguide region is more sensitive to processing errors. But at least 150 nm process tolerance exists, which can guarantee success of device fabrication.
Fig. 5. Simulated transmission spectra with respect to different multimode regions (a) length LM and (b) width WM for 1 ${\times} $ 2 MMI. Simulated transmission spectra with respect to different multimode regions (c) length LM and (d) width WM for 4 ${\times} $ 4 MMI.
3. Fabrication and characterization
The devices are fabricated on a 700 nm thick Z-cut TFLN wafer with a 2 μm silicon oxide isolating the top layer and substrate. The process flow of the fabrication is shown in Fig. 6 (a). After cleaning, a thin layer of chrome (Cr) is deposited by electron beam (e-beam) evaporation for the facilitation of electron beam lithography (EBL). Then, ma-N 2403 resist is spin-coated for EBL, which has 75 keV energy. Both the resist and Cr serve as a shadow mask for subsequent dry etching after development and Cr wet etch. Then the pattern is transferred to LN through an inductively coupled plasma (ICP) etching method with only argon gas. An Oxford PlasmaPro 100c Cobra is used for the ICP etching with around 320 W bias power. The LN etch rate is about 25 nm/min, and the selectivity is around ∼1:1. The shadow mask after dry etching is stripped in hot N-methyl-2-pyrrolidone and Cr etchant. The scanning electron microscope (SEM) image of the etched waveguide and device cross section is shown in Figs. 6(b) and (c). At each end, the input/output waveguides are adiabatically tapered to 5 μm to serve as edge couplers after dicing. Here, a two-step dicing method is used with a shallow trench first and then is cut past the substrate. Figure 6(d) shows an SEM image of the edge coupler facet after dicing.
Fig. 6. (a) Process flow of the fabrication. SEM images of the (b) etched waveguide, (c) device cross section (platinum (Pt) is deposited as the protection layer for focused ion beam cutting) and (d) edge coupler facet after dicing.
The 1 ${\times} $ 2 MMI is first characterized with measured results shown in Fig. 7. To evaluate the IL, the transmission power of different numbers of cascaded MMIs are measured (as shown in the inset of Fig. 7(a)). During the measurement, the input light is fixed at the TM polarized state using a polarization controller (PC), and injected into the device through a lensed fiber. A 1550 nm wavelength light from a tunable laser with a stable output power is used to measure different numbers of cascaded MMIs. It can be concluded from Fig. 7(a) that the slope of the fitted line is about 0.268 dB/MMI, which corresponds to 0.268 dB IL for the 1 ${\times} $ 2 MMI. Figure 7(b) shows the transmission spectra of the two outputs for a single MMI. During the measurement, a supercontinuum laser (NKT Photonics SuperK Versa) is used as a light source and the output is collected through a lensed fiber to an optical spectrum analyzer (OSA). The spectra of the edge coupler and other background are deducted from the measured results using a reference straight waveguide. Nearly the same transmission performance can be observed from Fig. 7(b) for both the two outputs of the 1 ${\times} $ 2 MMI.
Fig. 7. (a) Measured loss for cascaded 1 ${\times} $ 2 MMIs (Inset image is the microscope image of the cascaded 1 ${\times} $ 2 MMIs). (b) Measured transmission spectra from the two outputs of a single 1 ${\times} $ 2 MMI (Inset image is the microscope image of the single MMI under measurement with labeled input direction).
For the 4 ${\times} $ 4 MMI, the IL is measured using a supercontinuum laser and calculated by deducting the output of a reference straight waveguide with the same length. The coupling loss in each facet is about 5 dB, and the propagation loss of the waveguide is estimated at about 3.7 dB/cm. The cascaded structure is not used here because its size is a little too large and we cannot cascade too many given the limitation of our sample’s surface area. The light is also polarized at the TM state using a PC during the measurement. Figure 8(a) shows the measured transmission spectra of the four output ports. Figure 8(b) shows an optical microscope image when the second port (as shown in Fig. 1(b)) is input with light. All four ports show uniform spectra near the C band (1530 to 1565 nm). The inset of Fig. 8(a) shows the detailed results at 1550 nm. ILs from Port 1 to Port 4 are 6.6, 6.85, 6.54, 6.62 dB, respectively. The total IL of the 4 ${\times} $ 4 MMI is about 0.63 dB. The slightly larger IL compared to the simulation results can be attributed to fabrication process errors.
Fig. 8. (a) Measured experimental transmission spectra of the 4 ${\times} $ 4 MMI (inset shows details at 1550 nm). (b) Optical microscope image of the 4 ${\times} $ 4 MMI when Port 2 is input with light (the small red arrow shows the input light’s direction).
4. Discussion
An MMI has many applications in photonics, especially for on-chip integrated devices. One example is the Mach-Zehnder interferometer (MZI) structure modulator, in which a 1 ${\times} $ 2 MMI is used as a power splitter that divides input light into two equal paths [10]. Although a Y-splitter can also play the same role, its critical fabrication requirements can greatly affect subsequent device performance [20]. The proposed TM mode 1 ${\times} $ 2 MMI can thus be used as a high-performance power splitter to realize high performance modulators and other related devices in Z-cut TFLN, where the r33 is expected to be maximized compared to the X-cut case. In addition, the proposed 4 ${\times} $ 4 MMI can be used both as a multi-port power splitter and 90-degree hybrid mixer for coherent signal demultiplexing [17–19]. Although TFLN on its own cannot realize high performance photodetection in the telecommunications band, multifunctional integrated on-chip circuits are possible with the help of the heterogeneous integration method [21].
Table 1 shows a comparison of the results of the demonstrated MMIs in this manuscript and literature. Both the device size and IL show advantages when compared to the reported results. It’s worth noting that the propagation loss of TM mode is larger than that of the TE mode under the same conditions. It is probably due to the weak mode confinement compared with TE case for the same waveguide geometry [22]. Therefore, both the simulated and measured ILs for TM mode are larger than the reported TE cases. Based on the same design methods, under 0.05 dB loss is obtained for the 1${\times} $2 TE mode MMI, which accords well with literature [10]. A little deviation of the measured results compared with the simulated ones can be minimized by further optimizing fabrication processes, such as using gas clustered ion beam smoothening [6]. High yield and low processing sensitivity are expected for the proposed device as its fabrication tolerance is large. It is noted that similar results can be obtained for the TE mode case based on the method shown in this manuscript.
Table 1. Comparison of MMIs in literature and this work.
5. Conclusion
We have proposed and demonstrated high performance TM mode 1${\times} $2 and 4${\times} $4 MMIs based on TFLN. The design methods for these two kinds of MMIs were described in detail with simulated IL of 0.157 and 0.297 dB for 1${\times} $2 and 4${\times} $4 MMIs, respectively. The (spatial) process tolerance for the proposed MMIs is over 150 nm according to the simulation results. By using an optimized argon-based ICP etching method, the designed devices were fabricated and characterized. The measured ILs for 1${\times} $2 and 4${\times} $4 MMIs were 0.268 and 0.63 dB, respectively. The slightly-larger IL compared to the simulation results can be further reduced by optimizing process technology. Compared to the existing literature, our proposed devices feature compact size and low IL. We believe that the strong performance of the MMIs demonstrated in this manuscript can be used for both functional devices and on-chip integrated circuits on the Z-cut TFLN platform.
Funding
National Research Foundation Singapore (NRF-CRP15-2015-01, QEP-P2, QEP-P3); Ministry of Education - Singapore (MOE2018-T2-1-137).
Acknowledgements
The authors acknowledge the assistances of the Spin and Energy Lab (SEL) and E6NanoFab facilities at National University of Singapore (NUS). The authors thank Mengyuan Ye at China University of Geoscience (Wuhan) for assistance and useful discussions.
Disclosures
The authors declare that they have no competing interests.
References
1. R. S. Weis and T. K. Gaylord, “Lithium Niobate: Summary of Physical Properties and Crystal Structure,” Appl. Phys. A: Solids Surf. 37(4), 191–203 (1985). [CrossRef]
2. E. L. Wooten, K. M. Kissa, A. Yi-Yan, E. J. Murphy, D. A. Lafaw, P. F. Hallemeier, D. Maack, D. V. Attanasio, D. J. Fritz, G. J. McBrien, and D. E. Bossi, “A review of lithium niobate modulators for fiber-optic communications systems,” IEEE J. Sel. Top. Quantum Electron. 6(1), 69–82 (2000). [CrossRef]
3. H. Lu, W. Qiu, C. Guyot, G. Ulliac, J.-M. Merolla, F. Baida, and M.-P. Bernal, “Optical and RF Characterization of a Lithium Niobate Photonic Crystal Modulator,” IEEE Photonics Technol. Lett. 26(13), 1332–1335 (2014). [CrossRef]
4. X. P. Li, K. X. Chen, and L. F. Wang, “Compact and electro-optic tunable interleaver and lithium niobate thin film,” Opt. Lett. 43(15), 3610–3613 (2018). [CrossRef]
5. M. Prost, G. Liu, and S. J. B. Yoo, “A Compact Thin-Film Lithium Niobate Platform with Arrayed Waveguide Gratings and MMIs,” in 2018 Optical Fiber Communications Conference and Exposition (OFC), Tu3A.3 (2018).
6. 6. S. Y. Siew, E. J. H. Cheung, H. Liang, A. Bettiol, N. Toyoda, B. Alshehri, E. Dogheche, and A. J. Danner, “Ultra-low loss ridge waveguides on lithium niobate via argon ion milling and gas clustered ion beam smoothening,” Opt. Express 26(4), 4421–4430 (2018). [CrossRef]
7. C. Wang, M. Zhang, X. Chen, M. Bertrand, A. Shams-Ansari, S. Chandrasekhar, P. Winzer, and M. Lončar, “Integrated lithium niobate electro-optic modulators operating at CMOS-compatible voltages,” Nature 562(7725), 101–104 (2018). [CrossRef]
8. I. Krasnokutska, R. J. Chapman, J.-L. J. Tambasco, and A. Peruzzo, “High coupling efficiency grating couplers on lithium niobate on insulator,” Opt. Express 27(13), 17681–17685 (2019). [CrossRef]
9. A. N. R. Ahmed, S. Shi, A. Mercante, S. Nelan, P. Yao, and D. W. Prather, “High-efficiency lithium niobate modulator for K band operation,” APL Photonics 5(9), 091302 (2020). [CrossRef]
10. M. Xu, M. He, H. Zhang, J. Jian, Y. Pan, X. Liu, L. Chen, X. Meng, H. Chen, Z. Li, X. Xiao, S. Yu, S. Yu, and X. Cai, “High-performance coherent optical modulators based on thin-film lithium niobate platform,” Nat. Commun. 11(1), 3911 (2020). [CrossRef]
11. J. Zhao, C. Ma, M. Rüsing, and S. Mookherjea, “High quality entangled photon pair generation in periodically poled thin-film lithium niobate waveguides,” Phys. Rev. Lett. 124(16), 163603 (2020). [CrossRef]
12. R. W. Boyd, Nonlinear optics (Academic Press, 2019).
13. M. K. Smit, X. Leijtens, H. Ambrosius, E. Bente, J. van der Tol, B. Smalbrugge, T. de Vries, E.-J. Geluk, J. Bolk, R. van Veldhoven, L. Augustin, P. Thijs, D. D’Agostino, H. Rabbani, K. Lawniczuk, S. Stopinski, S. Tahvili, A. Gilardi, W. Yao, K. Williams, P. Stabile, P. Kuindersma, J. Pello, S. Bhat, Y. Jiao, D. Heiss, G. Roelkens, M. Wale, P. Firth, F. Soares, N. Grote, M. Schell, H. Debregeas, M. Achouche, J.-L. Gentner, A. Bakker, T. Korthorst, D. Gallagher, A. Dabbs, A. Melloni, F. Morichetti, D. Melati, A. Wonfor, R. Penty, R. Broeke, B. Musk, and D. Robbins, “An introduction to InP-based generic integration technology,” Semicond. Sci. Technol. 29(8), 083001 (2014). [CrossRef]
14. C. Sun, M. T. Wade, Y. Lee, J. S. Orcutt, L. Alloatti, M. S. Georgas, A. S. Waterman, J. M. Shainline, R. R. Avizienis, S. Lin, B. R. Moss, R. Kumar, F. Pavanello, A. H. Atabaki, H. M. Cook, A. J. Ou, J. C. Leu, Y.-H. Chen, K. Asanović, R. J. Ram, M. A. Popović, and V. M. Stojanović, “Single-chip microprocessor that communicates directly using light,” Nature 528(7583), 534–538 (2015). [CrossRef]
15. G. Chen, Y. Yu, and X. Zhang, “Monolithically mode division multiplexing photonic integrated circuit for large capacity optical interconnection,” Opt. Lett. 41(15), 3543–3546 (2016). [CrossRef]
16. Y. Yao, W. Wang, and B. Zhang, “Designing MMI structured beam-splitter in LiNbO3 crystal based on a combination of ion implantation and femtosecond laser ablation,” Opt. Express 26(15), 19648–19656 (2018). [CrossRef]
17. L. Soldano and E. Pennings, “Optical multi-mode interference devices based on self-imaging: principles and applications,” J. Lightwave Technol. 13(4), 615–627 (1995). [CrossRef]
18. S. Farwell, P. Aivaliotis, Y. Qian, P. Bromley, R. Griggs, J. N. Y. Hoe, C. Smith, and S. Jones, “InP coherent receiver chip with high performance and manufacturability for CFP2 modules,” in 2014 Optical Fiber Communications Conference and Exposition (OFC), W1I.6 (2014).
19. M. Seimetz and C.-M. Weinert, “Options, feasibility, and availability of 2 × 4 90◦ hybrids for coherent optical systems,” J. Lightwave Technol. 24(3), 1317–1322 (2006). [CrossRef]
20. H. Sasaki and I. Anderson, “Theoretical and experimental studies on active Y-junctions in optical waveguides,” IEEE J. Quantum Electron. 14(11), 883–892 (1978). [CrossRef]
21. G. Chen, J. Goyvaerts, S. Kumari, J. Van Kerrebrouck, M. Muneeb, S. Uvin, Y. Yu, and G. Roelkens, “Integration of high-speed GaAs metal-semiconductor-metal photodetectors by means of transfer printing for 850 nm wavelength photonic interposers,” Opt. Express 26(5), 6351–6359 (2018). [CrossRef]
22. R. Pafchek, R. Tummidi, J. Li, M. A. Webster, E. Chen, and T. L. Koch, “Low-loss silicon-on-insulator shallow ridge TE and TM waveguides formed using thermal oxidation,” Appl. Opt. 48(5), 958–963 (2009). [CrossRef]
