Open Access
Add to BibSonomy
Abstract
Polarization management, and in particular polarization rotation, is becoming increasingly important for photonic integrated circuits (PICs). While fiber-optic networks are generally polarization insensitive, the large aspect ratio of high-index-contrast PIC waveguides leads to a large polarization-dependent response of integrated components such as waveguides, optical cavities, couplers, etc. Although foundry-processed polarization rotators operating at telecom and datacom wavelengths (C- and O-band) have been demonstrated, to date, there have been few reports of devices operating at shorter wavelengths. This work demonstrates silicon nitride (SiN) polarization rotators operating from λ=700-1000 nm (the I/Z-band) that take advantage of optical coupling between two waveguiding layers in a standard foundry process. We demonstrate a broadband white-light polarization measurement setup that enables precise characterization of the polarization-dependent transmission of photonic waveguide devices. Measurements on foundry-processed devices confirm full TE-to-TM rotation exhibiting a maximum polarization extinction ratio (PER) approaching 20 dB (limited by our measurement setup), and an exceptionally large bandwidth of up to 160 nm with an insertion loss less than 0.2 dB. Beam propagation method (3D-BPM) simulations show good agreement with experimental data and enable the device parameters to be adjusted to accommodate different operating wavelengths and geometries with no changes to the existing foundry process. This work opens up opportunities for applications in quantum information and bio-sensing where operation at λ<1000nm is needed.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Polarization rotation is a key function for many photonic integrated circuits (PICs) [1] that are connected to fiber-optic networks in which the polarization may not be known or precisely controlled. Although polarization-maintaining (PM) fiber exists, the majority of deployed optical networks are non-PM. In contrast, PICs generally consist of thin film dielectrics that have been patterned into large aspect ratio waveguides. This asymmetric cross-section results in a large birefringence and a large polarization dependence in waveguides and other PIC components (e.g., couplers, microring cavities, etc.). Silicon PICs, in particular are highly-polarization-dependent due to the high aspect ratio waveguides and the large index contrast between the silicon core and the silicon oxide (SiO$_2$) cladding. Recently, silicon nitride (SiN) [2] has emerged as a core material for PICs [3] that enables ultra-low propagation losses [4] in a foundry platform at wavelengths ranging from the visible to the mid-infrared [5]. Despite the lower index contrast between the SiN core and SiO$_2$ cladding, modal birefringence is significant so that polarization-dependence can still be substantial.
A number of approaches have been proposed for polarization management in PICs. In general, these rely either on minimizing the birefringence in waveguides [6] or splitting the two orthogonal polarizations (TE and TM) into separate signal paths on a chip [7], thereby enabling optimized components for one specific polarization to be used (polarization diversity). Controlling the birefringence in waveguides may not always be possible due to material and geometric constraints imposed by the foundry process. In contrast, polarization splitting can generally be achieved by using a directional coupler [7] optimized to exploit polarization-dependent coupling, although this approach generally exhibits a limited bandwidth. A drawback with polarization splitting, however, is the need for optimized components for each polarization (TE and TM), thereby increasing the number of components on-chip.
Beyond splitting, polarization management also requires the ability to rotate the polarization state of on-chip optical signals. This can be achieved using slanted rib waveguides [8,9] or via mode evolution using a patterned feature overlaid on a waveguide [10,11]. Alternatively, a symmetry-breaking feature can be etched into the waveguide to achieve polarization rotation [12]. These demonstrations have shown efficient polarization rotation in the C-band and the O-band [13] with operating bandwidths of 70 nm and insertion loss <1 dB [11]. However, polarization rotation at shorter wavelengths, such as 780 nm [14] and further to the visible wavelength range [15,16] is required for quantum information or biosensing applications. A previous experimental demonstration has been reported at 780 nm using electron-beam lithography to pattern SiN rib waveguides, with reported losses in the 1 dB-range and a device length larger than 750 $\mu$m [14].
In this paper, we experimentally demonstrate a broadband, foundry-compatible SiN polarization rotator at wavelengths from 700 nm to 1000 nm (the I/Z-band), a wavelength region not addressed in previously-reported integrated polarization rotators that have focused on the C-band [8–12]. We utilize a broadband photonic integrated circuit (PIC) polarization testing setup with a white light source and a spectrometer to characterize the polarization-dependent response of fabricated devices. White-light spectroscopy can characterize PIC devices over a wavelength range much larger than what is possible with tunable lasers. Measurements on foundry-fabricated devices demonstrate full TE-to-TM rotation near $\lambda$=780 nm, and a maximum polarization extinction ratio (PER) approaching 20 dB (limited by our measurement setup) with losses as low as 0.2 dB, a bandwidth of 160 nm, and a maximum device length of 300 $\mu$m. We investigated the effects of various device parameters, including rotator length, rotator width, and device layer thicknesses on polarization rotation angle and bandwidth, and show excellent agreement between measurement and 3D-BPM simulation. The devices demonstrated in this work enable applications in quantum information and bio-sensing where polarization-controlled operation at $\lambda$<1000 nm is required.
2. Design approach
2.1 Silicon nitride platform
Our polarization rotator is designed for the AIM Photonics SiN-Only Passive PIC platform. This platform enables many passive components with demonstrated operation from 700-1625 nm [17] and losses as low as 3.2 dB/meter [5]. The layer structure of the SiN-only passive PIC platform includes two low-loss and low-fluorescence SiN layers of thickness $t_{SiN}$=150 nm or 220 nm, as shown in Fig. 1(a). These SiN waveguide layers are separated by a thin SiO$_2$ layer with thickness $t_{SiO2}$=50-100 nm. In the remainder of this paper, we refer to the first (waveguide) SiN layer as "FN" and the second SiN layer as "SN." The FN SiN layer used here is deposited via low-pressure chemical vapor deposition (LPCVD), while the SN SiN is deposited via either LPCVD or plasma-enhanced chemical vapor deposition (PECVD). The properties of the wafers used for measurements are shown in Table 1.
Fig. 1. Foundry-processed, broadband polarization rotator: a) Cross-sectional layer structure of the Silicon Nitride-Only Passive PIC platform (AIM Photonics), b) top-view schematic of a polarization rotator designed for the SiN-Only Passive PIC platform. The first SiN (waveguide) layer is called "FN" while the second SiN layer is "SN."
Table 1. Wafers used for the polarization rotation measurements.
2.2 Device design and simulation
The TE$_{00}$ and TM$_{00}$ modes in a rectangular waveguide are predominantly polarized with the electric field in the in-plane and out-of-plane transverse directions, respectively. Thus, a waveguide device that converts one of these eigenmodes to the other can be thought of as a linear polarization rotator. This PIC component is analogous to a bulk half-wave plate.
The basic architecture for the mode evolution [10] based polarization rotator is shown in Fig. 1(b). The structure consists of a single-layer input SiN waveguide ($w_{wg}$=800 nm) and width taper segment, a bi-layer rotator segment, a bi-layer combined width taper and mode transition segment, and a single-layer output waveguide. The waveguide width narrows down to w$_{rot, max.}$<w$_{wg}$/2 in the rotator region.
The rotator segment and the width/mode taper segment utilize the second SiN layer (SN), while the other sub-components utilize only the first SiN layer (FN). The SN overlay tapers in width as shown in Fig. 1(b) resulting in a SN-overlay w$_{rot}$=0 to w$_{rot, max.}$. Polarization rotation occurs in the rotator segment region due to a symmetry-breaking interaction between the FN waveguide mode and the slanted SN layer. Following the rotator segment is a waveguide width- (FN) and mode-taper (SN) that adiabatically transitions the bi-layer waveguide mode to the output waveguide (FN layer) mode. Unlike the symmetry-breaking rotator SN segment, the mode taper is symmetric and is thus not expected to affect the polarization rotation angle beyond enabling a reduced insertion loss. Our design does not require partial (rib) etches and defines the polarization rotator in the standard "FN" and "SN" layers of the foundry process.
We simulated the polarization rotator using a three-dimensional beam propagation method (BPM) solver (Synopsys RSoft BeamPROP). The lower index of SiN compared to Si requires longer device lengths (i.e., $\leq$300 $\mu$m for the SiN polarization rotators in the present work vs $\approx$10 $\mu$m for Si in previous work [18]). These device lengths are computationally prohibitive to simulate using three-dimensional finite-difference-time-domain (FDTD) solvers [18], and 3D-BPM is therefore used in this work.
First, the TE$_{00}$ and TM$_{00}$ polarized modes of the input and output waveguide segment are computed as shown in Fig. 2(a). Next, the TE$_{00}$-mode is launched into the polarization rotator and the overlaps between the local power and the previously-computed TE$_{00}$ and TM$_{00}$ modes are computed along the length of the structure [Fig. 2(b)]. The simulations clearly show conversion from an input TE$_{00}$ mode to an output TM$_{00}$ mode along the device length, corresponding to polarization rotation from horizontal (launched TE$_{00}$-mode) to vertical (TM$_{00}$ output mode) as shown in Fig. 2(c). We can convert the simulated overlaps to a polarization angle using $\alpha =\mathrm {atan}(P_{TM00}/P_{TE00})$, where P is the power in the TM$_{00}$ and the TE$_{00}$ mode, respectively. The plotted polarization angle in Fig. 2(d) confirms the design enabling complete polarization rotation from TE to TM. Although the simulations in Fig. 2 are for a device operating at $\lambda$=840 nm, the 3D model can be readily adapted for other wavelengths, including visible operation.
Fig. 2. Simulated polarization rotation: a) simulated TE$_{00}$- and TM$_{00}$-modes at $\lambda$=840 nm, b) simulated polarization rotation for an input TE$_{00}$-mode at $\lambda$=840 nm showing 90-degree rotation to TM at the output, c) extracted mode overlap with the computed TE$_{00}$- and TM$_{00}$-modes along the device length, d) polarization rotation angle extracted from the TE$_{00}$- and TM$_{00}$-mode components in (c).
3. Experimental
3.1 Broadband measurement setup
The devices were measured using the setup shown in Fig. 3(a), similar to our previous work [19] but modified to enable the measurement of both orthogonal polarization states (TE and TM) at the device output. We use a $\lambda$=780 nm laser to align tapered polarization-maintaining (PM) fibers to the rotator’s input and output waveguides while measuring the device transmittance using a photodetector. Subsequently, a broadband white light source is coupled to the device via a 2x1 PM fiber coupler and the polarization rotator’s spectral response is measured using a spectrometer. The white light source’s polarization is set using an interchangeable polarizer aligned for TE or TM before coupling to the fiber. At the output of the device, light is collected and sent to a U-bench with a second polarizer that functions as an analyzer before measuring the transmission spectrum in the spectrometer. Both the laser and white light sources are polarized parallel to the fiber’s slow or fast axes. Our setup enables broadband measurement and spectral analysis of the polarization rotation covering $\lambda$=700 nm to 1000 nm using a Si detector and PM780-HP fibers and 950 nm to 1550 nm using an InGaAs detector and PM980-XP fibers.
Fig. 3. Broadband polarization rotation measurement setup. Measurements of each sub-component in the setup enables a calibrated polarization extinction to be obtained. Measurements on a test waveguide inform the ultimate calibrated polarization extinction ratio (PER) that can be measured in our setup. Inset: Optical microscope device images.
We calibrated the individual components in the setup to obtain their polarization extinction ratio (PER) using a photodetector with an aligned linear polarizer. The white light source enables broadband polarized light with PER>30 dB, while the U-bench polarization analyzer has a measured PER>25 dB. The two lensed fibers can be rotated to ensure horizontal or vertical alignment in order to excite the TE$_{00}$ and TM$_{00}$ modes in the device, respectively. Combined, the two lensed fibers give a measured PER=18 dB. Finally, a straight PIC test waveguide (no polarization rotator) resulted in a measured maximum PER=15.5-16.7 dB. This PER sets the maximum polarization extinction that can be reliably measured in our setup after calibration and also determines the maximum polarization rotation angle, $\alpha =\mathrm {atan}(P_{TM00}/P_{TE00})$$\approx$88$^o$, for a TE$_{00}$ input. The rotator’s polarization extinction ratio for an input TE$_{00}$ mode is obtained from PER = −10log($P_{TE00}$/$P_{TM00}$) enabling us to relate PER to rotation angle, $\alpha$.
3.2 Device characterization
The experimental characterization of the SiN polarization rotators consists of three separate measurements, as shown in Fig. 4, for a device from wafer A. We first measure the transmittance of a device with a 75 $\mu$m long rotator segment with no polarizer in the output U-bench, followed by measurements with a linear polarizer set to measure horizontally- (TE) or vertically-polarized (TM) light, respectively [Fig. 4(a)]. As with 3D-BPM simulations in Fig. 2(c),(d), we can extract the polarization rotation angle from the measured TE and TM signal at the device output. The measurements in Fig. 4(b) show a peak rotation angle of 87.6$^o$ at $\lambda$=882 nm, i.e. close to the measurement resolution in our setup predicted from calibration [Fig. 3]. We measure a maximum PER$\approx$20 dB and an average PER$\approx$15 dB (see system calibration discussion). The measured polarization rotation angle is >80$^o$ over an 80 nm wavelength span ($\lambda$=853-933 nm), confirming broadband operation. These results are in good agreement with 3D-BPM simulations [Fig. 4(c)].
Fig. 4. Polarization rotation measurement (wafer A): a) measured broadband transmission without polarizer and with U-bench polarizer (analyzer) set to TE and TM, b) extracted polarization rotation angle, c) 3D-BPM simulated polarization rotation angle.
We also measured devices from wafer D, as shown in Fig. 5. The measurements for a device with a 225 $\mu$m rotator length show a consistent peak rotation angle approaching 90$^o$ near $\lambda$=780 nm. More importantly, we observe an exceptionally broadband spectrum with bandwidth exceeding 100 nm [Fig. 5(a)]. As with the rotator in Fig. 4, we find good agreement between our measurement and 3D-BPM simulations. A second polarization rotator with a longer rotation segment (300 $\mu$m) exhibited a bandwidth of 160 nm with PER>10 dB ($\lambda$=790-950 nm), as shown in Fig. 5(b). The peak rotation angle is 87.2$^o$ at $\lambda$=840 nm, i.e. close to the measurement resolution in our setup predicted from calibration (Fig. 3). These results (Fig. 4 and Fig. 5) demonstrate the versatility of our polarization rotator, while modeling enables accurate design for different layer structures, bandwidth, and wavelengths.
Fig. 5. a) Polarization rotation measurement and simulation (225 $\mu$m rotator length, wafer D), b) measurement of second device with 300 $\mu$m long rotator segment. Inset: extracted polarization extinction ratio (PER).
3.3 Device loss
We investigated the insertion loss of devices with different lengths corresponding to the layer structures of the devices in Fig. 4 and Fig. 5. The loss was extracted by taking the direct measurement [no polarizer in the output U-bench as in Fig. 3(a)] and dividing by the measured transmittance of a reference waveguide on-chip. The measurements correspond to an input TE$_{00}$-mode with a device transmittance that only considers the total output power and not the polarization thereby enabling loss measurements without having to account for partial rotation.
The results from wafer A in Fig. 6(a) show that the thinner SiN layer (t$_{FN,SN}$=150 nm) devices show an exceptionally low loss of around 0.2 dB or less over the peak rotation angle region ($\lambda$=800-925 nm). Beyond this wavelength range the excess loss increases slightly, although it remains below 0.5 dB over the range 725-1000 nm. The devices from wafer D measured in Fig. 6(b) (with t$_{FN,SN}$=150/220 nm and t$_{SiO_2}$=100 nm) show a slightly larger excess loss of around 0.6 dB at shorter wavelengths ($\lambda$=750-800 nm). This may be in part due to the PECVD-SiN material used for the t$_{SN}$=220 nm layer in wafer D compared to the LPCVD-SiN used for the t$_{SN}$=150 nm layer in wafer A. Our previous work has shown that PECVD-SiN exhibits a larger loss compared to LPCVD-SiN for $\lambda$=725-1000 nm [5]. At longer wavelengths, the loss also appears to depend on the rotator length. Although further investigation is needed, the thicker second SiN layer (t$_{SN}$=220 nm) compared to the first SiN layer (t$_{FN}$=150 nm) implies that the effective index of the rotator can be larger than the effective index of the waveguide resulting in substantial optical mode overlap with the second SiN layer, potentially resulting in lost power. We emphasize that the overall excess loss of the measured polarization rotators is low (<0.2 dB to $\approx$0.6 dB). Comparing the results from Figs. 4, 5 and Fig. 6, we clearly observe a tradeoff between polarization rotation angle, device bandwidth, and insertion loss, especially for the devices from wafer D with t$_{SN}$=220 nm.
Fig. 6. Measured polarization rotator insertion loss vs. wavelength for various rotator lengths: a) wafer A (the black line corresponds to the device from Fig. 4), b) wafer D (the blue line corresponds to the device from Fig. 5(a), the red line from Fig. 5(b)).
3.4 Dependence on rotator length
We performed additional characterization of devices with different rotator lengths (wafer D). Measurements show a clear increase in polarization rotation angle with increasing rotator length accompanied by a redshift in the center wavelength [Fig. 7(a)]. This length-dependence is explained by mode evolution in which e.g. the TE$_{00}$-mode is continuously-rotated along the device length and hence experiences a larger rotation for longer device lengths assuming all other parameters are identical [10]. More importantly, the bandwidth also increases with an extracted polarization rotation angle of 80$^o$ or greater over a bandwidth of around 200 nm [$\lambda$=765-965 nm, black curve in Fig. 7(b)]. Looking at Fig. 7(b), the increased bandwidth is well-reproduced by our 3D-BPM model as is the increase in peak rotation angle with increasing rotator length. Although additional investigation is needed, simulations have shown that even longer rotators (i.e. >300 $\mu$m) may result in a further increase in the device bandwidth along with a slight redshift in the center wavelength.
Fig. 7. Polarization rotation spectrum vs. rotator length: a) measured devices from wafer D; the black line corresponds to the device from Fig. 5(a), the red line corresponds to the device from Fig. 5(b), b) 3D-BPM simulations.
3.5 Dependence on SiO$_2$ spacer thickness
We also investigated the dependence of the rotation angle on the $SiO_2$ spacer thickness. We measured a number of identical rotators but with different values of $t_{SiO2}$. Figure 8 shows measurements from identical rotators from three different wafers whose oxide spacer was $t_{SiO2}$=50 nm, 75 nm, and 100 nm (wafers A, B, and C). The measurements show a clear polarization rotation dependence on $t_{SiO2}$ with full TE-to-TM conversion (90$^o$ rotation) for $t_{SiO2}$=50 nm and progressively smaller rotation angle as $t_{SiO2}$ increases. The thicker $SiO_2$ spacer between the FN- and SN-layers results in a reduced evanescent field interaction of the propagating mode (FN-waveguide) with the SN-rotator and hence a smaller rotation. These results are consistent with our proposal for micro-electro-mechanically (MEMS) tunable polarization rotators [18] in which the second SiN layer can be displaced vertically to adjust the rotation angle.
Fig. 8. Measured and simulated polarization rotation angle vs. $SiO_2$ thickness (wafers A, B, C); measurement $\lambda =$ 890 nm (peak rotation angle); simulation $\lambda =$ 780 nm (design wavelength). Inset: rotation angle vs. wavelength.
4. Discussion
We observe good agreement between our measurements and 3D-BPM simulations, as shown in Fig. 4 and Fig. 5. Our simulations did not include material absorption and generally predict low losses of <0.1 dB. The loss measurements in Fig. 6(a) (t$_{SN}$=150 nm, wafer A) are in good agreement with the predicted loss obtained via 3D-BPM modeling and are consistent with the low absorption obtained in our LPCVD-SiN waveguides. As previously mentioned, the devices with t$_{SN}$=220 nm (wafer D) utilized PECVD-SiN, which has been found to exhibit larger absorption compared to LPCVD-SiN. And indeed, the measured loss for these devices is substantially larger [Fig. 6(b)] compared to the t$_{SN}$=150 nm that utilized LPCVD-SiN [Fig. 6(a)].
The measured polarization rotation angle’s dependence on the SiN-layer spacing (i.e. $t_{SiO2}$), as in Fig. 8, is also well-predicted by our 3D-BPM model. Indeed, our previous computational study on micro-electro-mechanically (MEMS) tunable polarization rotators [18] showed that dynamic control of the separation between the rotator structure (second SiN) and the waveguide (first SiN layer) enables a continuously-tunable polarization rotation angle to be obtained. Although our previous work focused on silicon devices, the current work on SiN devices shows that a MEMS-tunable rotator can be realized in other material systems besides silicon. Actuation of all-dielectric structures such as SiN can be realized using gradient electric forces [20]. Additional modeling and measurements have shown that the device in Fig. 4 from wafer A is more sensitive to small variations in separation between the first and second SiN layers due to the small $t_{SiO2}$=50 nm. In contrast, the devices in Fig. 5 from wafer D are more robust to small variations in SiN layer separation due to the larger $t_{SiO2}$=100 nm. For this reason, the polarization rotator shown in Fig. 5 with t$_{FN}$=150 nm / $t_{SiO2}$=100 nm / t$_{SN}$=220 nm, may be preferred, as it offers both robustness to fabrication variations as well as an increased bandwidth of operation. Although the loss is higher in the t$_{SN}$=220 nm devices due to the PECVD-SiN material used, we expect future improvements in materials deposition to minimize optical losses, e.g. by using LPCVD-SiN or via annealing of the PECVD-SiN.
Beyond adjusting the rotator length (as in Fig. 8) we can also take advantage of the rotator region’s FN waveguide width to tune the interaction strength with the symmetry-breaking SN perturber. A narrower waveguide results in a less-confined mode (TE$_{00}$ and TM$_{00}$) so that the peak rotation angle is shifted towards shorter wavelengths. Simulations have confirmed the shift towards shorter wavelengths when the rotator width is narrowed, but the operating bandwidth is also reduced. Although additional investigation is needed, increasing the rotator length appears to compensate for this by enabling rotation over larger wavelength ranges. Figure 9 shows the TE-to-TM polarization rotation for a device similar to those in Fig. 5, but with a slightly narrower FN waveguide width in the rotator segment along with a slight increase in the rotator length. The results indicate a shift of the peak rotation to $\lambda$=740-780 nm compared to $\lambda$=840 nm (Fig. 5) and a peak rotation of $\approx$85$^o$ – i.e. close to our setup’s measurement resolution.
Fig. 9. Measured polarization rotation angle for a device with a narrower rotation region (blue curve) compared to device in Fig. 5(a) (red curve) resulting in a shift towards shorter wavelength (layer structure: wafer D).
5. Conclusion
This work has demonstrated a foundry-processed, broadband SiN polarization rotator in the I/Z-bands (700 nm<$\lambda$<1000 nm). The design is fully compatible with the foundry’s existing layer structure and can be readily incorporated into PIC designs. We have accurately modeled the polarization rotation spectrum and behavior using 3D-BPM simulations. A broadband measurement setup enables accurate polarization measurement over $\lambda$=700-1000 nm. Measurements on a variety of devices with different SiN layer thicknesses, SiO$_2$ spacers, and rotator lengths have illustrated some of the design parameters and how they influence the device behavior. The bi-layer SiN structure can be adapted for operation over a broad wavelength range, and simulations show that foundry-compatible polarization rotators operating at visible wavelengths are possible with only small changes to the existing design.
Funding
U.S. Naval Research Laboratory (WU 1AB2); Air Force Research Laboratory (FA8650-21-2-1000).
Acknowledgment
The authors thank SF Preble and VS Sundaram (RIT) for helpful discussions. The U.S. Government is authorized to reproduce and distribute reprints for governmental purposes notwithstanding any copyright notation thereon. The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the United States Air Force, the Air Force Research Laboratory or the U.S. Government.
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time, but may be obtained from the authors upon reasonable request. Data release is subject to approval by the US Naval Research Laboratory.
References
1. P. Dong, “Silicon photonic integrated circuits for wavelength-division multiplexing applications,” IEEE J. Sel. Top. Quantum Electron. 22(6), 370–378 (2016). [CrossRef]
2. C. Xiang, W. Jin, and J. E. Bowers, “Silicon nitride passive and active photonic integrated circuits: trends and prospects,” Photonics Res. 10(6), A82–A96 (2022). [CrossRef]
3. T. Sharma, J. Wang, B. K. Kaushik, et al., “Review of recent progress on silicon nitride-based photonic integrated circuits,” IEEE Access 8, 195436–195446 (2020). [CrossRef]
4. J. Liu, G. Huang, and R. N. Wang, “High-yield, wafer-scale fabrication of ultralow-loss, dispersion-engineered silicon nitride photonic circuits,” Nat. Commun. 12(1), 2236 (2021). [CrossRef]
5. N. F. Tyndall, T. H. Stievater, D. A. Kozak, et al., “A low-loss SiN photonic integrated circuit foundry platform for waveguide-enhanced Raman spectroscopy,” in Smart Photonic and Optoelectronic Integrated Circuits XXIII, vol. 11690S. HeL. Vivien, eds., International Society for Optics and Photonics (SPIE, 2021), pp. 40–46.
6. D. Dai, L. Liu, and S. Gao, “Polarization management for silicon photonic integrated circuits,” Laser Photonics Rev. 7(3), 303–328 (2013). [CrossRef]
7. P. Dong, “A compact silicon-on-insulator polarization splitter,” IEEE Photonics Technol. Lett. 17(1), 100–102 (2005). [CrossRef]
8. Z. Huan, R. Scarmozzino, and G. Nagy, “Realization of a compact and single-mode optical passive polarization converter,” IEEE Photonics Technol. Lett. 12(3), 317–319 (2000). [CrossRef]
9. P. Dong, “Slanted-rib waveguide InGaAsP-InP polarization converters,” J. Lightwave Technol. 22(5), 1352–1357 (2004). [CrossRef]
10. M. Watts and H. Haus, “Integrated mode-evolution-based polarization rotators,” Opt. Lett. 30(2), 138–140 (2005). [CrossRef]
11. H. Zhang, S. Das, and J. Zhang, “Efficient and broadband polarization rotator using horizontal slot waveguide for silicon photonics,” Appl. Phys. Lett. 101(2), 021105 (2012). [CrossRef]
12. M. Aamer, A. M. Gutierrez, and A. Brimont, “CMOS compatible silicon-on-insulator polarization rotator based on symmetry breaking of the waveguide cross section,” IEEE Photonics Technol. Lett. 24(22), 2031–2034 (2012). [CrossRef]
13. M. J. Hossain, M. R. Uddin, L. G. Carpenter, et al., “Bilayer silicon nitride based on-chip polarization rotator for o-band photonic applications,” in IEEE Photonics Conference, (IEEE, Orlando, FL, 2023).
14. K. Gallacher, P. F. Griffin, and E. Riis, “Silicon nitride waveguide polarization rotator and polarization beam splitter for chip-scale atomic systems,” APL Photonics 7(4), 046101 (2022). [CrossRef]
15. Z.-S. Hou, X. Xiong, and J.-J. Cao, “On-chip polarization rotators,” Adv. Opt. Mater. 7(10), 1900129 (2019). [CrossRef]
16. T. Sneh, A. Hattori, M. Notaros, et al., “Design of integrated visible-light polarization rotators and splitters,” in Frontiers in Optics + Laser Science (FiO+LS), (Optica Publishing Group, Rochester, NY, 2022).
17. “Aim photonics silicon nitride passive pic,” (Accessed 21 November 2023).
18. M. W. Pruessner, N. F. Tyndall, and T. H. Stievater, “MEMS-tunable polarization management in photonic integrated circuits,” Opt. Express 31(19), 31316–31328 (2023). [CrossRef]
19. N. F. Tyndall, T. H. Stievater, and D. A. Kozak, “Mode-crossing spectroscopy for photonic waveguide characterization,” APL Photonics 4(10), 106107 (2019). [CrossRef]
20. M. W. Pruessner, D. Park, and T. H. Stievater, “Broadband opto-electro-mechanical effective refractive index tuning on a chip,” Opt. Express 24(13), 13917–13930 (2016). [CrossRef]
