Open Access
Add to BibSonomy
Abstract
Symmetric polarization rotator-splitter (PRS) is proposed and experimentally demonstrated on InP for the first time. Instead of integrating a mode-selective splitter, we employ a symmetric multimode-interference (MMI) splitter at the output of an adiabatic taper section to extract the linear superpositions of the transverse-electric (TE) and the transverse-magnetic (TM) components of the input signal. As a result, the entire device functions as a PRS with its basis on the S2-S3 plane of the Poincaré sphere, whereas we can fully eliminate complicated asymmetric structures that are challenging to fabricate on InP. Moreover, the adiabatic taper, which operates as a mode-evolution-based polarization converter, is designed judiciously to minimize the overall length. The designed InP PRS with a total length of 750 µm is fabricated by a simple single-etching process. A polarization extinction ratio of more than 16.3 dB and a polarization-dependent loss of 0.67 dB are demonstrated experimentally at a 1550-nm wavelength.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Compact polarization rotator-splitter (PRS) is an essential component for realizing on-chip polarization (de)multiplexing and manipulation inside a photonic integrated circuit (PIC) [1–9]. It has been a conventional procedure to use the fundamental transverse-electric (TE) and transverse-magnetic (TM) modes as the orthogonal basis in realizing a PRS. However, in order to split the TE and TM modes and rotate the TM mode to a TE mode, complicated structures are inevitably required, such as laterally asymmetric waveguides and/or mode-selective directional couplers. While such structures can be implemented compactly on silicon photonic platforms [1–5], realization on InP has been challenging and has required complicated fabrication procedures [6–8]. On the other hand, monolithic InP PIC provides a unique advantage that lasers and amplifiers can be integrated with a minimal insertion loss [10–14]. Furthermore, InP-based modulators and detectors generally exhibit superior performance compared with the silicon photonic counterparts for high-speed coherent transceiver applications [15–18]. While polarization-diversity detectors have been demonstrated by integrating polarization-selective multiple-quantum-well photodiodes (PDs) [19,20], more general PSR would be an essential building block for versatile applications, including both the transmitter and receiver PICs. It is therefore desirable to realize a compact and simple PRS that can easily be integrated on InP PIC.
In many practical applications, such as polarization (de)multiplexing in digital coherent systems [12,13,15–19], it is not mandatory to split or combine TE0 and TM0 modes. Instead, any orthogonal polarization modes can be used to define the basis since the state of polarization (SOP) is scrambled anyway during the fiber transmission and unscrambled automatically at the receiver by the digital signal processing (DSP). For other applications, such as constructing a polarization diversity circuit to eliminate the polarization-dependent loss (PDL) inside the chip [1,2], it is also not necessary to use the TE0/TM0 basis to split and combine two polarization components. Instead of TE0/TM0-based PRSs, ±45° linear SOPs or any other mutually orthogonal SOPs on the S2-S3 plane of the Poincaré sphere can be used as the orthogonal basis to enable significantly simpler symmetrical PRS with a wider design space. While such device has been demonstrated on silicon [21], it has not been realized on InP to our knowledge.
In this work, we demonstrate a monolithic InP PRS, which splits the input light into two orthogonal SOPs on the S2-S3 plane, rotates both of them to TE modes, and output them to two independent ports. The device consists of an adiabatic taper section followed by a symmetric multimode-interference (MMI) splitter, which are all fabricated in a simple single etching process without the need for complicated laterally asymmetric waveguides. The taper section, which contains only a vertical asymmetry, is designed judiciously to minimize the length and enable mode-evolution-based polarization conversion with a large fabrication error tolerance. Using the fabricated PRS with a total length of 750 µm, a polarization extinction ratio over 16.3 dB with a PDL of 0.67 dB is demonstrated at 1550-nm wavelength.
2. Operational principle and design
2.1 Entire device configuration
Figure 1 illustrates the schematic of the symmetrical PRS demonstrated in this work. It consists of an adiabatic taper section and a multi-mode interferometer (MMI) splitter. As shown in the inset, the entire device has an identical layer profile with 350-nm-thick lattice-matched InGaAsP core layer (Q1.37) and 230-nm-thick upper InP cladding layer, which is etched by 210 nm to form a ridge waveguide.
Fig. 1. Schematic of the symmetric PRS, which consists of the adiabatic taper section and the 1 × 2 MMI splitter. Evolution of the mode profile for TE0 and TM0 input are also shown.
When a polarization-multiplexed light is incident to the device, its x- and y-polarization components, denoted by Ex and Ey, respectively, excite the TE0 and TM0 modes of the input ridge waveguide, respectively. Inside the adiabatic taper section, the waveguide width is increased gradually. With a proper design of the waveguide as described in Section 2.2, we can let the TM0 mode to adiabatically evolve to TE1 mode, while keeping the TE0 mode unchanged [3]. Then, both modes are input to the MMI splitter. Here, the MMI splitter is designed judiciously as explained in Section 2.3, so that it operates as a mode-independent splitter; the TE0 and TE1 modes at the input are split into two TE0 modes at the output ports with minimal loss. Due to the even and odd symmetries of the TE0 and TE1 modes at the input, the total optical fields at the two output ports would be proportional to ${E_x} + {E_y}{e^{i\theta }}$ and ${E_x} - {E_y}{e^{i\theta }}$, respectively. Here, the factor ${e^{i\theta}}$accounts for the total optical phase retardance between the two modes and consists of the effective refractive index between the TE0 and TM0 at the input ridge waveguide, that between the TE0 and the adiabatically converting mode (from TM0 to TE1) inside the tapered section, and the different phase shifts experienced by the respective modes at the MMI splitter. When θ = πm, where m is an integer, the output fields from the two ports are represented as ${E_x} + {E_y}$ and ${E_x} - {E_y}$, so that this device functions as a 45° PRS that splits ±45° linear polarization modes into two ports. More generally, when θ ≠ 0, the incident light is split with respect to an orthonormal basis, defined by two orthogonal SOPs on the S2-S3 plane of the Poincaré sphere.
In the previously demonstrated PRSs based on the mode-evolution adiabatic tapers, asymmetric mode-selective couplers need to be attached at the output to separate TE0 and TE1 modes [3–9]. This has imposed a great technical challenge for the InP platform, requiring complicated fabrication procedures and/or large footprint. Here, instead, we employ a laterally symmetric MMI splitter to extract the linear superposition of the TE0 and TE1 modes. If the mode-dependent loss of the splitter is negligible (i.e., unitary), the entire device operates as an ideal PRS with just a different basis, which would not be a problem for many applications in practice. On the other hand, we can significantly simplify the fabrication procedure and minimize the device footprint.
2.2 Adiabatic taper section
In order to minimize the overall device length without degrading the loss and crosstalk properties, each section needs to be optimized judiciously. This is especially challenging for InP devices compared with silicon-photonic devices due to its weaker optical confinement.
First, the thickness of the upper InP cladding is carefully designed to avoid the use of a pre-converter at the input, which would be necessary when a thin or no InP cladding is used to ensure efficient coupling of the TM0 mode [6–9]. Figure 2(a) shows the fraction of TE-like component excited when a TM0 mode is coupled to the waveguide, calculated as a function of the upper InP cladding thickness. From this result, we selected the thickness to be 230 nm as shown in Fig. 1 inset, so that a pure TM0 mode is excited directly without the need for a pre-converter. This rather thick upper InP cladding of our device, on the other hand, would generally increase the required length of the taper section due to the reduced vertical asymmetry. To cope with this issue, we apply a method similar to the fast quasiadiabatic dynamics [22] to design a non-uniform taper instead of a simple linear taper, as described below.
Fig. 2. (a) Fraction of TE-like component in the TM0 mode as a function of the upper InP cladding thickness. (b) Effective indices of the three highest index modes, n1, n2, and n3 and (c) 1/Δn2, where $\Delta n \equiv {n_2} - {n_3}$, as a function of the waveguide width W. The wavelength is 1550 nm.
Figure 2(b) shows the effective refractive indices of the first three guiding modes, ${n_1}$, ${n_2}$, and ${n_3}$, of the ridge InP/InGaAsP waveguide shown in the inset of Fig. 1 with various width W. Here, the color scale describes the fraction of the Ex component contained in each mode; the blue (red) region denotes that the mode has a TE-like (TM-like) profile. In Fig. 2(c), we also plot 1/Δn2 as a function of W, where we define $\Delta n \equiv {n_2} - {n_3}$. This will be used later when we design a compact taper.
From Fig. 2(b), we can see that the first mode (n1) exhibits the TE0 profile, independent on the width. In contrast, the second and the third modes (n2 and n3) exhibit TM0 and TE1 profiles when the width is narrow, but transform to TE1 and TM0 modes, respectively, as the width increases. Therefore, by gradually tapering the width from 2.0 µm to 3.2 µm, the input TM0 mode can be converted to TE1 mode.
To ensure adiabatic evolution of the second mode from TM0 to TE1 without coupling to the third mode, the waveguide width needs to be tapered at a sufficiently slow rate compared with $\Delta n$. To minimize the overall length, therefore, it is essential to adjust the tapering rate depending on $\Delta{n}$ [22–25]. For convenience, we define $z(W )$ to represent the position along the z axis where the width of the taper becomes $W$. As the boundary conditions, $z({{W_i}} )= 0$ and $z({{W_f}} )= L$, where ${W_i}$ and ${W_f}$ are the initial and final widths of the taper, and L is the total length. Then, the entire taper structure is designed as
Here, $f({\Delta n} )$ (${\propto} \partial z/\partial W$) describes the inverse of the normalized tapering rate and should be defined to decrease monotonically with $\Delta n$ to achieve efficient conversion. Since $\Delta n$ changes with W as shown in Fig. 2(c), $f({\Delta n} )$ is also a function of W.
The wave propagation inside the adiabatic taper section is simulated by the eigenmode expansion (EME) method for various taper designs. Figure 3(a) shows the conversion efficiency from TM0 to TE1 for various values of L when Wi and Wf are set to 2.0 µm and 3.2 µm, respectively. Here, we compare three taper designs: (i) when $f({\Delta n} )$ is a constant, corresponding to a linear taper, (ii) when $f({\Delta n} )= 1/\mathrm{\Delta }n$, and (iii) when $f({\Delta n} )= 1/\mathrm{\Delta }{n^2}$. Figure 3(b) depicts the taper geometry derived from Eq. (1) for the three cases. From Fig. 3(a), we can confirm that a non-uniform taper with $f({\Delta n} )= 1/\mathrm{\Delta }{n^2}$ achieves the best performance, enabling a conversion efficiency of 95% with L of only 700 µm. This is in clear contrast to a linear taper, which requires L to be more than 2300 µm to achieve 95% efficiency. Note that when $L$ = 0, the 2.0-µm-wide input waveguide is connected directly to the 3.2-µm-wide waveguide without a taper section. Since the modes inside these two waveguides are not perfectly orthogonal with each other, the conversion efficiency does not drop down to zero but is around 10% in Fig. 3(a).
Fig. 3. (a) Simulated conversion efficiency from TM0 to TE1 inside the taper section as a function of taper length, L. Wi and Wf are 2.0 µm and 3.2 µm, respectively. Three cases of $f({\Delta n} )$ are plotted: (i) $f({\Delta n} )$ = constant (linear taper), (ii) $f({\Delta n} )= 1/\mathrm{\Delta }n$, and (iii) $f({\Delta n} )= 1/\mathrm{\Delta }{n^2}$. (b) Taper geometries for the three cases.
To investigate the fabrication tolerance, Fig. 4 shows the simulated conversion efficiency under deviations in the waveguide width and the etching depth. We can see that performance is almost insensitive for a width variation of 200 nm and an etching-depth variation of 40 nm. This large fabrication tolerance is due to the adiabatic mode-evolution mechanism used for the polarization conversion [3,26], which only depends weakly on Δn and is inherently insensitive to the absolute values of n2 and n3, unlike the mode-coupling-based polarization rotators [27–29].
Fig. 4. Fabrication tolerance of the taper section against the deviation in (a) the waveguide width (etching depth = 210 nm) and (b) the etching depth (Wi = 2.0 µm, Wf = 3.2 µm).
2.3 MMI splitter
We then design the MMI splitter, so that it operates as an ideal 1 × 2 splitter with a minimal insertion loss for both TE0 and TE1 at the input. As shown in Fig. 1, we assume that the adiabatic taper section is connected to the center (x = 0) of the MMI input. At the MMI output, two single-mode waveguides are connected at symmetric locations from the center.
First, the width of the MMI coupler is optimized so that only TE0, TE1, and TE2 modes exist inside the MMI. Then, the length of the MMI is set to the half beat length between the TE0 and TE2 modes. In such a case, when a TE0 mode from the adiabatic taper is input, TE0 and TE2 modes are excited inside the MMI coupler due to the symmetry of the modes. After propagating inside the MMI, they interfere destructively at the center and generate symmetric in-phase two intensity peaks as shown in Fig. 1. As a result, they can easily be separated without scattering and coupled to the TE0 mode of the two output ports. From the even symmetry in x direction, they should have an equal optical phase. In contrast, the TE1 mode from the adiabatic taper would only excite the TE1 mode in the MMI due to the odd symmetry. At the output, the light is coupled to the two output ports with π phase difference. As a result, we obtain the total optical fields, which are proportional to Ex + Eyeiθ and Ex – Eyeiθ, respectively, at the two output ports as explained in Section 2.1.
From the eigenmode analysis of the MMI splitter assuming the same cross-sectional profile as shown in Fig. 1, the width and the length of the MMI are determined to be 4.3 µm and 28 µm, respectively, to satisfy the above-mentioned conditions. For this structure, the total transmission through the 1 × 2 MMI splitter is simulated to be 99.4% and 99.8% for the TE0 and TE1 input, respectively, at a wavelength of 1550 nm.
Finally, Fig. 5 shows the simulated light propagation of the entire device, consisting of the adiabatic taper section with L = 700 µm and the MMI splitter as designed above. Two cases are shown, where TE0 and TM0 modes at 1550-nm wavelength are launched at the input. We can confirm that the TE0 mode is split equally into two output ports without changing its polarization. In contrast, when the TM0 mode is input, the electric-field orientation is rotated to x components, and then split into two output ports. Due to the symmetry in x direction, the two outputs exhibit in-phase and out-of-phase amplitudes for the TE0 and TM0 inputs, respectively. The entire device, therefore, functions as a PRS with the orthogonal bases on the S2-S3 plane as explained in Fig. 1.
Fig. 5. Simulated light propagation through the entire device with L = 700 µm when TE0 (a) or TM0 (b) mode is launched. Intensities of x- and y-components of the electric field are plotted.
3. Fabrication and characterization
The device designed in the previous section was fabricated by a simple single-etching process. The entire waveguide was formed by CH4/H2-based inductively-coupled-plasma reactive-ion etching (ICP-RIE) using an SiO2 hard mask, which was patterned by electron-beam lithography. In this work, we have carefully controlled the etching time to adjust the etching depth to 210 nm as designed in Fig. 1; however, a thin etching-stop layer [26,27] may be inserted in practice to enable precise control of the etching depth. The microscopic image of the fabricated PRS is shown in Fig. 6(a). The total length of the PRS itself is 750 µm, which consists of the 700-µm-long taper section and the 28-µm-long MMI splitter as designed in the previous section, as well as a short transitional section between the two. The total length of the entire device, including the S-shaped output fan-outs, which was not minimized in this work, is around 1100 µm. This is nearly half the length of the previously demonstrated InP-based PRS with a total length of 2000 µm [6].
Figure 6(b) shows the experimental setup to characterize the fabricated PRS. To avoid the influence of coherent interference due to the residual reflections at the device facets, we first employed an incoherent light source at 1550-nm wavelength with the 3-dB bandwidth of 2 nm, which was generated by spectrally slicing the amplified spontaneous emission (ASE) from an erbium-doped fiber amplifier (EDFA) using arrayed-waveguide gratings (AWG) [28]. A sequence of a polarizer (POL), a half-wave plate (HWP), and a quarter-wave plate (QWP) was used to control the SOP of the incident light. The power and SOP of the output light from the device were measured by an optical power meter and a polarization analyzer (Thorlabs, TXP5004). Unitary transformation of SOP at the input fiber pigtail was calibrated by a fiber-based polarization controller (PC2) before the measurement, so that we could deterministically control the input SOP to the device under test (DUT) by rotating the HWP and QWP.
We first input TE0 and TM0 light into the device and measured the insertion loss. Due to the symmetric structure of the entire device, the input power is split equally to two output ports with the power difference of less than 0.8 dB. It is confirmed that the output light is polarized to the TE mode with more than 98% for both ports. By removing the fiber-to-chip coupling loss, which was measured from test straight waveguides, the excess on-chip loss is derived to be less than 2.5 dB and 3.8 dB for the TE0 and TM0 input, respectively.
Then, we fix the QWP to 45° and change the angle of the HWP, ϕ. In this case, as shown in Fig. 7(b), the input SOP rotates around a circle on the S2-S3 plane from the right-handed circular (RHC), +45° linear, left-handed circular (LHC), and then to −45° linear states. Figure 7(a) shows the measured output power at 1550 nm from two ports. We can see that the maximum and the minimum transmissions are obtained when ϕ = 17.5° and 62.5°, corresponding to a pair of orthogonal SOPs on the S2-S3 plane. The polarization extinction ratio (PER), defined as the ratio between these two values, is 18.2 dB and 16.3 dB for port 1 and 2, respectively. The PDL at the maximum transmission is 0.67 dB.
Fig. 7. PER measurement at 1550-nm wavelength. (a) Output power from each port as a function of HWP angle. (b) Input SOP to the device when the HWP is rotated to an angle ϕ. The QWP is fixed to 45° in this measurement.
Finally, the wavelength dependence of the device was characterized by using a tunable laser source (TLS) as shown in Fig. 6(b). The effect of Fabry-Perot resonance at the device facets was eliminated by averaging the measured data over a 1-nm wavelength window. Due to the birefringence of the waveguides, θ defined in Section 2.1 is generally nonzero and changes with the wavelength. As a result, the orthogonal basis of the PRS rotates on the S2-S3 plane as we sweep the wavelength. Therefore, we rotated the HWP and measured the transmission spectra at the two output ports for each case. The QWP was fixed to 45° as in the previous measurement.
Figure 8(a) shows the wavelength dependence of the transmitted power ratio between the two output ports for various HWP angles. We can confirm that with the proper orthogonal basis, which changes with the wavelength, the input light is split into two output ports with more than 14 dB in a wavelength range from 1540 to 1560 nm. For each wavelength, we can define the HWP angle ϕmax that maximizes the transmission to one port and minimize at the other. Similar to Fig. 7(a), we can confirm that at ϕmin ≡ ϕmax + 45°, the transmission is switched between the two output ports. By dividing these values, we can obtain the PER as a function of the wavelength, which is plotted in Fig. 8(b). A PER of more than 18 dB and 14 dB are obtained for port 1 and port 2, respectively, in a wavelength range of 1550 ± 10 nm.
We should note that the change of the orthonormal basis of the PRS with the wavelength would not be a problem in practice when this device is used for polarization (de)multiplexing in dual-polarization coherent transceivers. This is because the SOP would rotate randomly inside the transmission fiber anyway and such rotation is automatically removed by the DSP at the receiver. It would also not cause any problem for a polarization diversity application; since the input light to this device is separated by an orthonormal basis with a negligible PDL and rotated to the TE0 modes at the two output ports, the actual basis is not important in practice. This device can, therefore, be used in wide ranges of dual-polarization applications.
Fig. 8. Measured wavelength dependence of the device. (a) Output power ratio between port 1 and port 2 for various input SOPs around the S2-S3 plane by rotating the HWP. (b) PER (ratio between the maximum transmission and minimum transmission) for each port as a function of the wavelength.
4. Conclusion
We have proposed, fabricated, and experimentally demonstrated a symmetric PRS on monolithic InP platform for the first time. By carefully designing the adiabatic taper structure and the symmetric MMI splitter, highly efficient PRS with a length of 750 µm was realized. Using the fabricated device, a PER of more than 16.3 dB with a PDL of 0.67 dB was obtained at 1550-nm wavelength. Wideband operation was also confirmed with the PER of more than 14 dB in the wavelength range of 1550 ± 10 nm. Unlike the previously demonstrated PRSs on InP, our device consists of a simple single-etch laterally symmetrical ridge waveguide structure without a need for an asymmetric directional coupler or a pre-converting section, significantly simplifying the fabrication procedure as well as reducing the overall device footprint. The presented device should, therefore, be useful to realize a wide range of dual-polarization PICs on InP.
Acknowledgment
Portions of this work were presented at the European Conference on Optical Communication (ECOC) in 2021, “Mode-evolution-based symmetrical polarization splitter-rotator on monolithic InP platform” (paper We4D.4). The authors acknowledge Eisaku Kato, Warakorn Yanwachirakul, and Ryota Tanomura for their support in device fabrication.
Disclosures
The authors declare no conflict 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. T. Barwicz, M. R. Watts, M. A. Popovic, P. T. Rakich, L. Socci, F. X. Kartner, E. P. Ippen, and H. I. Smith, “Polarization-transparent microphotonic devices in the strong confinement limit,” Nat. Photonics 1(1), 57–60 (2007). [CrossRef]
2. 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]
3. 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]
4. 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]
5. W. D. Sacher, Y. Huang, L. Ding, T. Barwicz, J. C. Mikkelsen, B. J. Taylor, G.-Q. Lo, and J. K. Poon, “Polarization rotator-splitters and controllers in a Si3N4-on-SOI integrated photonics platform,” Opt. Express 22(9), 11167–11174 (2014). [CrossRef]
6. S. Keyvaninia, H. Boerma, M. Wössner, F. Ganzer, P. Runge, and M. Schell, “Highly efficient passive InP polarization rotator-splitter,” Opt. Express 27(18), 25872–25881 (2019). [CrossRef]
7. H. Boerma, S. Keyvaninia, F. Ganzer, P. Runge, and M. Schell, “Monolithically integrated InP-based polarization rotator-splitter with simplified fabrication process,” in OSA Advanced Photonics Congress (AP), JTh2A.5 (2020).
8. M. Perestjuk, H. Boerma, A. Schindler, S. Keyvaninia, P. Runge, and M. Schell, “Inverse-designed InP-based polarization rotator-splitter,” in Opt. Fiber Commun. Conference (OFC), W6A.49 (2021).
9. W. Yuan, K. Kojima, B. Wang, T. Koike-Akino, K. Parsons, S. Nishikawa, and E. Yagyu, “Mode-evolution-based polarization rotator-splitter design via simple fabrication process,” Opt. Express 20(9), 10163–10169 (2012). [CrossRef]
10. S. Arafin and L. A. Coldren, “Advanced InP photonic integrated circuits for communication and sensing,” IEEE J. Select. Topics Quantum Electron. 24(1), 1–12 (2018). [CrossRef]
11. M. Smit, K. Williams, and J. van der Tol, “Past, present, and future of InP-based photonic integration,” APL Photonics 4(5), 050901 (2019). [CrossRef]
12. V. Lal, J. Summers, N. Kim, S. W. Corzine, P. Evans, M. Lauermann, A. Nguyen, A. Hosseini, M. Lu, J. T. Rahn, M. R. Chitgarha, J. Zhang, J. Osenbach, T. Vallaitis, P. Samra, C. Park, M. Kuntz, J. Tang, C. Tsai, H. Sun, R. Schmogrow, D. Pavinski, B. Behnia, P. Mertz, T. Butrie, K.-T. Wu, M. Mitchell, M. Ziari, M. Reffle, D. Welch, and F. Kish, “Extended C-band tunable multi-channel InP-based coherent transmitter PICs,” J. Lightwave Technol. 35(7), 1320–1327 (2017). [CrossRef]
13. H. Yagi, T. Kaneko, N. Kono, Y. Yoneda, K. Uesaka, M. Ekawa, M. Takechi, and H. Shoji, “InP-based monolithically integrated photonic devices for digital coherent transmission,” IEEE J. Sel. Top. Quantum Electron. 24(1), 1–11 (2018). [CrossRef]
14. I. M. Soganci, T. Tanemura, and Y. Nakano, “Integrated phased-array switches for large-scale photonic routing on chip,” Laser Photonics Rev. 6(4), 549–563 (2012). [CrossRef]
15. M. C. Larson, J. Sonkoly, A. Mizrahi, M. Ayliffe, and J. Kemp, “InP vs Si photonic integrated circuit platforms for coherent data center interconnects,” in Proc. European Conf. on Opt. Commun. (ECOC), (2018).
16. Y. Ogiso, J. Ozaki, Y. Ueda, H. Wakita, M. Nagatani, H. Yamazaki, M. Nakamura, T. Kobayashi, S. Kanazawa, Y. Hashizume, H. Tanobe, N. Nunoya, M. Ida, Y. Miyamoto, and M. Ishikawa, “80-GHz bandwidth and 1.5-V V( InP-based IQ modulator,” J. Lightwave Technol. 38(2), 249–255 (2020). [CrossRef]
17. K. Nagashima, Y. Ishikawa, A. Izawa, M. Nishita, N. Matsubara, H. Ishii, T. Saeyang, Y. Ogiso, Y. Ueda, and M. Kohtoku, “Transmitter-receiver optical sub assembly using ultra-compact tunable DBR/ring laser,” in Proc. Opt. Fiber Commun. Conference (OFC), F2A.2 (2021).
18. M. Kurokawa, K. Nakayama, M. Takechi, Y. Mizuno, T. Misawa, E. Banno, H. Uemura, Y. Sugimoto, S. Kumagai, T. Okimoto, N. Kono, T. Ishikawa, H. Hara, T. Kato, K. Tanaka, M. Ekawa, and K. Uesaka, “High optical output power and high-responsivity IC-TROSA for 800 Gbps applications,” in Proc. European Conf. on Opt. Commun. (ECOC), We4G.1 (2021).
19. S. Keyvaninia, T. Beckerwerth, G. Zhou, M. Gruner, F. Ganzer, W. Ebert, S. Mutschall, A. Seeger, P. Runge, and M. Schell, “Novel photodetector chip for polarization diverse detection,” J. Lightwave Technol. 37(16), 3972–3978 (2019). [CrossRef]
20. T. Suganuma, S. Ghosh, M. Kazi, R. Kobayashi, Y. Nakano, and T. Tanemura, “Monolithic InP Stokes vector receiver with multiple-quantum-well photodetectors,” J. Lightwave Technol. 36(5), 1268–1274 (2018). [CrossRef]
21. Y. Ma, Y. Liu, H. Guan, A. Gazman, Q. Li, R. Ding, Y. Li, K. Bergman, T. Baehr-Jones, and M. Hochberg, “Symmetrical polarization splitter/rotator design and application in a polarization insensitive WDM receiver,” Opt. Express 23(12), 16052–16062 (2015). [CrossRef]
22. 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]
23. D. Guo and T. Chu, “Broadband and low-crosstalk polarization splitter-rotator with optimized tapers,” OSA Continuum 1(3), 841–850 (2018). [CrossRef]
24. A. Horth, P. Cheben, J. H. Schmid, R. Kashyap, and N. J. Quitoriano, “Ideal, constant-loss nanophotonic mode converter using a Lagrangian approach,” Opt. Express 24(6), 6680–6688 (2016). [CrossRef]
25. A. Melikyan and P. Dong, “Adiabatic mode converters for silicon photonics: Power and polarization broadband manipulators,” APL Photonics 4(3), 030803 (2019). [CrossRef]
26. A. E. Elfiqi, R. Tanomura, D. Yu, W. Yanwachirakul, H. Shao, Y. Suzuki, T. Tanemura, and Y. Nakano, “Robust InP/InGaAsP polarization rotator based on mode evolution,” IEEE Photonics Technol. Lett. , 34(2), 109–112 (2022). [CrossRef]
27. A. E. Elfiqi, R. Kobayashi, R. Tanomura, T. Tanemura, and Y. Nakano, “Fabrication-tolerant half-ridge InP/InGaAsP polarization rotator with etching-stop layer,” IEEE Photonics Technol. Lett. 32(11), 663–666 (2020). [CrossRef]
28. M. Zaitsu, T. Tanemura, A. Higo, and Y. Nakano, “Experimental demonstration of self-aligned InP/InGaAsP polarization converter for polarization multiplexed photonic integrated circuits,” Opt. Express 21(6), 6910–6918 (2013). [CrossRef]
29. M. Zaitsu, T. Tanemura, and Y. Nakano, “Numerical study on fabrication tolerance of half-ridge InP polarization converters,” IEICE Trans. Electron. E97.C(7), 731–735 (2014). [CrossRef]
