Design and Simulation of a Vertical Waveguide Based Polarization

Color Routing Device

F. Ali

1,2

, B. Figeys

, O. Shramkova

, R. MacCiarnain

, X. Rottenberg

1,3

R. Gehlhaar

and J. Genoe

1,2

Imec, Kapeldreef 75, 3001 Leuven, Belgium

KU Leuven, ESAT Department, Kasteelpark Arenberg 10 - bus 2440, 3001 Leuven, Belgium

ULB, Ecole Polytechnique de Bruxelles, Av. Franklin Roosevelt 50, 1050 Brussels, Belgium

Keywords: Color Routing, Polarization Routing, Vertical Waveguides, Image Sensors, Polarization-Dependent Imaging.

Abstract: We present a high index vertical-waveguide based optical device which offers polarization and color routing

with sub-micron spatial resolution over the visible spectrum (400-700 nm). The device exploits a polarization,

color splitting principle to effectively separate orthogonal linear polarization states, colors of the incident light

beam respectively. Finite difference time domain simulations are performed to optimize the design parameters

and analyse the device performance under different illumination conditions. The unique ability to

simultaneously manipulate polarization and color with remarkable efficiency can allow a new era for high-

resolution optical imaging devices.

1 INTRODUCTION

Image sensors are the backbone of modern imaging

systems, which convert light into electrical signals,

enabling the formation of a digital image equivalent.

The pursuit of ultra-high resolution and high-quality

imaging has driven significant advancements in

image sensor technology, catering to a surging

demand in an array of consumer and industrial

applications such as smartphones, automotive

systems, robotics, and surveillance (Park et al., 2022;

Theuwissen A., 2021; Takahashi et al., 2017).

Conventional image sensors are comprised of a

Bayer pattern color filter array to render color

information for color images. These sensors generally

detect only the intensity of the light but lack the

ability to capture polarization data which can convey

information about important characteristics and

properties of the imaged scene. Nonetheless,

incorporation of a polarization filter array into the

image sensor allows for the realization of

polarization-sensitive imaging, enabling a more

comprehensive understanding of the captured scene

(Gruev et al., 2010; Chen et al., 2016; Wu et al.,

2023).

The functionality of these image sensors relies

heavily on the filtering process, which unfortunately

discards a substantial amount of light and reduces the

signal-to-noise ratio, posing a major limitation to the

pixel downsizing trend and thus hindering the

continued miniaturization of imaging technology

(Arbabi et al., 2018; Lee et al., 2023; Scott-Thomas

J., 2023). To overcome this limitation, researchers

have proposed several color and polarization routing

devices, which split and direct the incident light into

designated pixels based on spectral content and

polarization state, without significant losses

(Nishiwaki et al., 2013; Yun et al., 2021; Kang et al.,

2023; Khorasaninejad et al., 2015; Rubin et al. 2019;

Nesic et al., 2023). Although these approaches have

successfully addressed the inherent efficiency

limitations of filter-based image sensors, none of

them have been able to integrate polarization and

color splitting functionalities into a single device for

highly-efficient polarization-dependent imaging.

In this paper, we demonstrate a design for

simultaneous polarization and color routing to

achieve a highly efficient polarization-sensitive

image sensor. The proposed device comprises of

vertical-waveguides and employs a splitting principle

to achieve its intended functionality. Numerical

simulations reveal that the device yields a total

transmission efficiency of over 95%. Our design

overcomes the long-standing efficiency limitations of

traditional sensor designs, and boasts a compact

footprint, with sub-micron pixel sizes making it a

Ali, F., Figeys, B., Shramkova, O., MacCiarnain, R., Rottenberg, X., Gehlhaar, R. and Genoe, J.

Design and Simulation of a Vertical Waveguide Based Polarization Color Routing Device.

DOI: 10.5220/0013121100003902

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 13th International Conference on Photonics, Optics and Laser Technology (PHOTOPTICS 2025), pages 15-19

ISBN: 978-989-758-736-8; ISSN: 2184-4364

suitable candidate for advanced optical imaging

applications.

2 DESIGN AND SIMULATION

Figure 1 shows a schematic representation of the

proposed polarization color router (PCR). A sketch of

the unit cell design with a pitch P is depicted in Figure

1a. The PCR is composed of two optical components

with distinct functionalities: a polarization splitter

(PS) unit on top and color splitter (CS-X, CS-Y) unit

below. Both units are composed of high index

vertical-waveguides with TiO

as a core and a SiO

cladding. The corresponding geometric parameters

and cross-sectional views are presented at multiple

planes (A–D) along the propagation direction.

The primary function of the top PS is to route the

input light beam to distinct output channels according

to its polarization state. It comprises of two segments:

a top square-shaped input waveguide (plane A) with

thickness H2 and width W1, which tapers down

asymmetrically to two partially overlapping

orthogonal waveguides of dimension W1×W2

forming an L-shaped cross section at plane B. The

bottom segment consists of an L-shaped waveguide

of thickness H3, which gradually tapers down to two

orthogonal square-shaped waveguides of width W2

(at plane C). The input waveguide adiabatically

transmits the light beam to the L-shaped waveguide

at plane B. Then, the beam is routed towards either of

the square-shaped waveguides X or Y at exit plane of

the PS (plane C), based on the incident polarization

vector's orientation (horizontal or vertical

polarization: indicated by green arrows).

Figure 1: Schematic of the Polarization Color Router: (a) A

unit cell design comprising vertical-waveguides with cross-

sectional views at different planes. (b) A configuration for

the pixel design at the detector plane for a unit cell with a

pitch P = 1 µm.

Bottom color splitting elements (CS-X, CS-Y)

separate the incident light signal into its constituent

wavelengths or colors at sub-micron spatial

resolution. To achieve this functionality multimode

vertical-waveguides are employed, a technology

previously validated in our earlier research (Kang et

al., 2023). The color splitting segment consists of two

identical rectangular waveguides with dimension W3

× W4 × H4, connected to the two orthogonal output

channels of the top PS unit at plane C. The PS exit

waveguide (X or Y) injects light asymmetrically into

the color splitting element (CS-X or CS-Y), exciting

both fundamental and second-order modes within the

waveguide. As these modes propagate, their speeds

vary with frequency causing spatial separation of

colors at the waveguide exit (plane D). It is worth to

mention, a subtle tuning of the waveguide’s

geometric parameters enables a precise control over

the optimal color spectrum, demonstrating the

versatility of our device. Finally, the separated colors

(polarization dependent) are captured by

corresponding pixels in the detector plane.

The device also features an H1 thick anti-

reflection coating (ARC) layer on waveguide

entrances to effectively reduce the back reflections

and enhance device efficiency. When illuminated

with a focused light beam from the top, device splits

the beam into constituent color bands according to its

polarization state, subsequently recorded in the

relevant pixel. Here, the visible spectrum is divided

into two bands (Blue/Red), with the potential to

expand to three bands (RGB) by adjusting the

waveguide’s dimension. The proposed pixel design is

composed of four pixels in a compact layout at the

detector plane, labelled X1, X2, Y1, and Y2, with

dimensions W5ൈ W6, as shown in Figure 1b. The

pixel pairs (X1, X2) and (Y1, Y2) are designed to

capture blue and red spectral bands for X-polarized

and Y-polarized light, respectively.

Finite difference time domain simulations in a

commercial solver, were employed to perform a

comprehensive optimization of design parameters

and analyse the optical performance of PCR (Ansys

Lumerical FDTD). A broadband Gaussian beam of

numerical aperture (NA) 0.8 was used to excite the

proposed device across the visible spectrum 400-700

nm. Unit cell simulations were performed with Bloch

boundary conditions in X and Y directions, while

perfectly matched layer (PML) boundary conditions

were set for the Z axis (propagation direction). The

optical constants of SiO

and TiO

used for

simulations

were taken from in-house measurements

and presented in the Figure 2.

PHOTOPTICS 2025 - 13th International Conference on Photonics, Optics and Laser Technology

The optimal device parameters were determined

through a two-stage nested sweeps simulation

process, targeting maximum efficiency and low pixel

cross-talk as main figure of merit. Initially, the top PS

unit was optimized for maximum polarization routing

efficiency, yielding optimal design parameters: P =1

μm, W1 = 550 nm, W2 = 200 nm, H2 = 2.8 μm, and

H3 = 3.7 μm. Subsequently, optimization of color

splitting elements enabled the separation of the

polarization-routed beam into two spectral bands,

leading to corresponding parameters: W3 = 200 nm,

W4 = 400 nm, and H4 = 3.1 μm. ARC design

principle and numerical optimization led to the

selection of a 90 nm thick (H1) SiO

layer as the

optimal anti-reflective coating. Meanwhile,

subwavelength pixel dimensions of W5 = 225 nm and

W6 = 500 nm were employed, yielding high spatial

resolution.

Figure 2: Material Optical Constants: Refractive Index (n,

k) of SiO

and TiO

3 RESULTS AND DISCUSSION

Figure 3 shows the simulated optical performance of

the proposed device under broadband visible

illumination (400-700 nm). The simulated electric-

field intensity distribution at the detector plane is

presented in Figure 3a, showing the device’s response

to blue (450 nm) and red (650 nm) wavelengths for

both X and Y polarization states. For X-polarization,

the incident light is primarily directed to the X output

channel at the PS exit (plane C: Figure 1a), then

routed to the CS-X splitting element where blue and

red wavelengths propagate at different speeds and are

separated into X1 and X2 pixels as evident from the

top row of Figure 3a. In contrast, Y1 and Y2 capture

the aforementioned colors for the orthogonal (Y)

polarization state (bottom row). Meanwhile, it is

worth to mention that for all four possible

configurations, the field is highly localized in the

designated pixels ensuring high efficiency with

minimal polarization and color crosstalk.

Figure 3b shows the spectral response of the PCR

for orthogonal polarization states of the incident light

beam. It can be observed that for X-polarization, the

majority (≈50%) of light in the blue band (400-550

nm) is collected by the X1 pixel, meanwhile X2

captures the red band (550-700 nm) with a similar

efficiency. The geometric symmetry induces a similar

response for the orthogonal polarization state (on the

right side of Figure 3b but now with Y1 and Y2 active

pixels. Average cross-talk between the active pixels

across the two spectral bands is 35-40%, which can

be further supressed by introducing deep trench

isolation (DTI) between the pixels. The device

achieves remarkable efficiency, featuring high

(>95%) total transmission and low (<5%) reflection.

Moreover, the device exhibits a low power leakage,

with under 5% of power directed to pixel-free region.

Figure 3: Optical performance of the PCR: (a) Normalized

electric-field intensity distribution profiles at the detector

plane for orthogonal polarization states (X and Y) at 450 nm

and 650 nm. The white (solid) and yellow (dashed)

rectangles define the four pixel and bottom color splitters

perimeters. (b) Transmission and reflection efficiency for X

and Y-polarized incident light over the visible spectrum.

Next, we demonstrate the device's versatility by

tuning the short-axis width (W3) of color splitting

units (CS-X, CS-Y) to support multiple color

compositions. For this purpose, simulations were

performed by sweeping W3 from 150 nm to 250 nm,

using an X-polarized Gaussian beam (NA = 0.8).

Figure 4 illustrates the dependance of pixel (X1, X2)

transmission efficiency over incident beam’s

wavelength and waveguide’s short-axis width. These

transmission maps indicate a gradual shift of the

crossing point (identified by black arrows) of two

separated spectral bands, defining the color

Design and Simulation of a Vertical Waveguide Based Polarization Color Routing Device

information in the relevant pixel. With a short-axis

width of 150 nm, wavelengths between 400-500 nm

were directed to pixel X1 and X2 received 500-700

nm, whereas the 250nm width split the spectrum into

400-570 nm (X1) and 570-700 nm (X2), without any

significant loss in the routing efficiency. Thus, by

precise control over short-axis width, the desired

color composition can be achieved. Furthermore,

tailoring thickness (H4) and long-axis width (W4) can

offer additional degree of freedom to achieve

customized spectral splitting.

Figure 4: Variation of pixel’s spectral composition: Power

transmission to the X1 and X2 pixel as a function of

wavelength and short axis width of bottom color splitting

unit.

To analyze the device’s capability to operate

under different illumination conditions, it was excited

with an X-polarized Gaussian beam by varying the

numerical aperture between 0.1-0.9. Figure 5 depicts

the power routed to the corresponding pixels (X1, X2)

as a function of wavelength and numerical aperture.

It is evident that the device performs as intended and

separates the two wavelength bands (blue and red) for

all illumination numerical apertures, albeit with

changes in efficiency. Although, pixel’s transmission

efficiency drops as NA decrease, yet remains ≈25%

for the lowest designed NA of 0.1. Nonetheless, we

anticipate that by following the optimization process,

the device can be tailored to operate at desired NA

with high routing efficiencies.

Further investigation will focus on expanding the

device's capabilities to separate polarization-

dependent visible light into red, green, and blue

(RGB) spectral bands by tailoring the color splitting

waveguide's geometric dimensions. While the bottom

color splitters have already been successfully

fabricated using standard backend processing

techniques on 300 mm wafers (Kang et al., 2023), we

will explore advanced lithography and fabrication

techniques to develop top polarization splitting

waveguides to realize complete device stack.

Ultimately, optical characterization will be performed

to determine the device's true efficiency and validate

its performance.

Figure 5: Device performance under a wide range of

illumination numerical aperture (NA) settings: Power

transmitted to the X1 and X2 pixel as a function of designed

numerical aperture and wavelength.

4 CONCLUSION

In conclusion, a novel high index vertical-waveguide

based polarization color routing device at visible

wavelengths is proposed and investigated through

numerical simulations. The key innovation lies in

simultaneously splitting the polarization and color at

sub-micron spatial resolution with near unity

transmission and a high pixel routing efficiency of

50%. The design’s versatility enables a tuneable color

composition via geometric parameter modulation and

adoptability to perform well over a wide range of

numerical aperture settings. We firmly believe this

device shows great promise to be implemented for

high NA polarization-sensitive image sensors with

possible applications in high-end professional

cameras, smartphones, and machine vision devices.

REFERENCES

Park, S., Lee, C., Park, S., Park, H., Lee, T., Park, D., Heo,

M., Park, I., . . . & Kim, T. (2022). A 64Mpixel CMOS

Image Sensor with 0.56μm Unit Pixel Separated by

Front Deep-Trench Isolation. IEEE International Solid-

State Circuits Conference (ISSCC), San Francisco, CA,

USA, pp. 1-3. https://doi.org/10.1109/ISSCC42614.

2022.9731750.

Theuwissen A. (2021). There’s More to the Picture Than

Meets the Eye, and in the future it will only become

more so. IEEE International Solid-State Circuits

Conference (ISSCC), San Francisco, CA, USA, pp. 30-

35. https://doi.org/10.1109/ISSCC42613.2021.936605.

Takahashi, S., Huang, Y-M., Sze, J-J., Wu, T-T., Guo, F-

S., Hsu, W-C., . . . & Yaung, D-N. (2017). A 45 nm

Stacked CMOS Image Sensor Process Technology for

Submicron Pixel. Sensors, 17(12), 2816. https://doi.org

/10.3390/s17122816.

Gruev, V., Perkins, R., & York, T. (2010). CCD

polarization imaging sensor with aluminum nanowire

PHOTOPTICS 2025 - 13th International Conference on Photonics, Optics and Laser Technology

optical filters. Opt. Express 18, 19087-19094. https://

doi .org/10.1364/OE.18.019087.

Chen, W. T., Török, P., Foreman, M. R., Liao, C. Y., Tsai,

W-Y., Wu, P. R., & Tsai, D. P. (2016). Integrated

plasmonic metasurfaces for spectropolarimetry.

Nanotechnology 27 224002. https://doi.org/10.1088/

0957-4484 /27/22/224002.

Wu, Y., Yang, Y., Yin, Y., Dai, L., Li, X., Huang, H., &

Wen, S. (2023). Flexible high-resolution thin

micropolarizers for imaging polarimetry. Chin. Opt.

Lett. 21, 031301. https://doi.org/10.3788/COL202321

.031301.

Arbabi, E., Kamali, S. M., Arbabi, A., & Faraon, A. (2018).

Full-Stokes Imaging Polarimetry Using Dielectric

Metasurfaces. ACS Photonics, 5 (8), 3132–3140.

https://doi. org /10.1021/acsphotonics.8b00362.

Lee, G-D. R., Kim, D-H., Kwon, D., Park, J-E., Cho, D.,

Kang, J., . . . & Song, J. (2023). A 0.5µm Pixel 3-layer

Stacked CMOS Image Sensor with Deep Contact and

In-pixel Cu-Cu Bonding Technology. International

Electron Devices Meeting (IEDM), San Francisco, CA,

USA, pp. 1-4. https://doi.org/10.1109/IEDM45741.

2023.10413687.

Scott-Thomas, J. (2023). Trends and Developments in

State-of-the-Art CMOS Image Sensors. https://www.

imagesensors.org/Past%20Workshops/2023%20Work

shop/2023%20Papers/R1.pdf

Nishiwaki, S., Nakamura, T., Hiramoto, M., Fujii, T., &

Suzuki, M. (2013). Efficient colour splitters for high-

pixel-density image sensors. Nature Photon, vol. 7, no.

3, pp. 240–246. https://doi.org/10.1038/nphoton.

2012.345.

Yun, S., Roh, S., Lee, S., Park, H., Lim, M., Ahn, S., &

Choo, H. (2021). Highly Efficient Color Separation and

Focusing in the Sub-micron CMOS Image Sensor.

IEEE International Electron Devices Meeting (IEDM),

San Francisco, CA, USA, pp. 30.1.1-30.1.4. https://doi.

org/10.1109/IEDM19574.2021.9720592.

Kang, S., Benelajla, M., Ciamain, R. M., Ali, F.,

Papadopoulou, A., . . . & Genoe, J. (2023). Wafer-level-

integrated vertical-waveguide sub-diffraction-limited

color splitters. International Electron Devices Meeting

(IEDM), San Francisco, CA, USA, pp. 1-4. https://doi.

org/10.1109/IEDM45741.2023.10413746.

Khorasaninejad, M., Zhu, W., & Crozier, K. B. (2015).

Efficient polarization beam splitter pixels based on a

dielectric metasurface. Optica 2, 376-382. https://doi.

org /10.1364/OPTICA.2.000376.

Rubin, N. A., D’Aversa, G., Chevalier, P., Shi, Z., Chen,

W. T., & Capasso, F. (2019). Matrix Fourier optics

enables a compact full-Stokes polarization camera.

Science365, eaax1839. https://doi.org/10.1126/science

.aax1839.

Nesic, A., Blaicher, M., Marin-Palomo, P., Füllner, C.,

Randel, S., Freude. W., & Koos, C. (2023). Ultra-

broadband polarisation beam splitters and rotators

based on 3D-printed waveguides. Light: Advanced

Manufacturing 4, 23. https://doi.org/10.37188/lam.

2023.022.

Ansys Lumerical FDTD. https://www.ansys.com/products

/optics/fdtd.

Design and Simulation of a Vertical Waveguide Based Polarization Color Routing Device