What is the polarization extinction ratio of a 0.23 inch waveguide module?

The polarization extinction ratio (PER) of a typical 0.23 inch optical waveguide module, such as the 0.23 inch optical waveguide module, is specified at a minimum of 20 dB under standard operating conditions, with typical values ranging from 22 dB to 25 dB at a wavelength of 550 nm (green light) when measured at 25°C. This PER value is critical because it directly determines the contrast ratio and color purity in augmented reality (AR) displays; a PER below 15 dB leads to noticeable ghosting and reduced image sharpness. For this module, the waveguide design uses a polarization-maintaining (PM) fiber pigtail and a specific grating coupler geometry that achieves a polarization extinction ratio of 23 dB ± 2 dB across the full field of view (FOV) of 30 degrees diagonal. This number is not just a spec sheet figure—it’s the difference between a usable AR experience and a washed-out mess. Let’s break down the real-world implications, the measurement methods, the materials science behind it, and how it compares to competing modules.

What exactly is PER and why does it matter in a 0.23 inch waveguide? PER is the ratio of the optical power in the desired polarization state (usually TE, transverse electric) to the orthogonal polarization state (TM, transverse magnetic), expressed in decibels. In a waveguide-based AR display, the micro-OLED emits unpolarized light, but the waveguide’s diffractive gratings are designed to couple only TE-polarized light efficiently. If the PER is low, some TM-polarized light leaks into the waveguide, creating stray light, reducing contrast, and causing color shifts. For the 0.23 inch module, the micro-OLED itself has a pixel pitch of 4.5 µm and a resolution of 640×480 pixels (nHD), and the waveguide’s input grating is optimized for a PER of 22 dB at the center wavelength of 532 nm. At 20 dB, the leaked TM power is 1% of the TE power, which is acceptable for most indoor AR applications. At 15 dB, the leaked power jumps to 3.16%, and you start seeing a 10% drop in contrast ratio—from 1000:1 down to 900:1, which is noticeable in side-by-side comparisons.

How is the PER measured for this specific module? The standard measurement setup uses a polarized laser source (e.g., a 532 nm DPSS laser with a PER > 30 dB), a half-wave plate to rotate the polarization, and a power meter with a polarizer at the output. The waveguide module is mounted on a precision rotation stage. The input fiber is a PM fiber with a 0.14 NA (numerical aperture) and a 5.5 µm mode field diameter, aligned to the slow axis. The measurement is done at 25°C and 50% relative humidity. The output light is collected from the exit pupil (a 10 mm × 8 mm eyebox) using a 2 mm diameter aperture. The PER is calculated as 10 log10(P_TE / P_TM). For the 0.23 inch module, the typical test results show a PER of 24.1 dB at the center of the FOV, dropping to 21.3 dB at the edge (15° off-axis). This drop is due to the angular dependence of the grating efficiency—the diffraction efficiency for TE mode decreases by about 3% at the edge, while the TM mode leakage increases by 0.5%. The module’s datasheet guarantees a minimum of 20 dB across the entire FOV, but batch testing of 100 units shows a mean PER of 23.2 dB with a standard deviation of 1.1 dB, indicating excellent manufacturing consistency.

Material and design choices that affect PER The waveguide substrate is made of Schott D263T borosilicate glass with a refractive index of 1.53 at 550 nm. The grating layer is a binary surface-relief grating etched into a TiO₂ thin film (n = 2.4 at 550 nm) with a depth of 150 nm and a period of 380 nm. This grating is designed for a 1D pupil expansion in the horizontal direction. The polarization selectivity comes from the grating’s form birefringence—the TiO₂ ridges have a higher refractive index for TE polarization (electric field parallel to the grating lines) than for TM polarization. The effective index difference is about 0.15, which gives a polarization extinction ratio of roughly 25 dB at the design wavelength. However, the actual PER is lower because of scattering from surface roughness (RMS roughness of 3 nm) and absorption in the TiO₂ layer (extinction coefficient k = 0.001). The input coupler uses a polarization beam splitter (PBS) film with a 100 nm thick silver coating, which has a measured PER of 30 dB at 550 nm, but the overall module PER is limited by the grating’s angular bandwidth. The PM fiber pigtail is a Fujikura PANDA fiber with a beat length of 2.5 mm at 550 nm, ensuring that the input polarization is maintained within 0.5 dB over 1 meter of fiber. The fiber-to-waveguide coupling loss is 0.8 dB, and the polarization crosstalk at the coupling point is less than -25 dB.

Temperature and wavelength dependency PER is not a static number. At 60°C, the PER of this module drops by about 2 dB because the refractive index of the TiO₂ grating changes by 0.003 per °C (dn/dT = 0.0001/°C for TiO₂, but the polymer overcoat has a larger dn/dT of -0.0001/°C, creating stress birefringence). At 0°C, the PER increases by 1 dB due to reduced thermal expansion mismatch. Wavelength-wise, the PER peaks at 550 nm (23 dB) and drops to 18 dB at 450 nm (blue) and 17 dB at 650 nm (red). This is because the grating’s resonance condition is wavelength-dependent—the TE diffraction efficiency is 85% at 550 nm, 70% at 450 nm, and 65% at 650 nm, while the TM leakage is 0.5% at 550 nm, 1.5% at 450 nm, and 2% at 650 nm. For a full-color AR display using a white micro-OLED with color filters, the overall PER is a weighted average across the RGB spectrum. With the 0.23 inch module, the measured PER for white light (D65 illuminant) is 20.5 dB, which is acceptable for most applications but not ideal for high-contrast outdoor use. A competing module from another manufacturer using a slanted grating design achieves a white-light PER of 22 dB, but at the cost of a 10% lower optical efficiency (15% vs. 17% for this module).

Comparison with other waveguide modules Let’s put this in perspective with a table of PER values for different waveguide sizes and technologies:

The 0.23 inch module has a PER that is competitive for its size class. The 0.3 inch module uses a double-layer grating that improves PER by 2 dB but reduces efficiency by 2% due to absorption in the second layer. The 0.5 inch module has a lower PER because it uses a simpler polymer grating (n = 1.7) with lower form birefringence. The 1.0 inch module achieves the highest PER but uses a bulky PBS cube and a multi-layer coating, making it unsuitable for compact AR glasses. For the 0.23 inch module, the trade-off between PER and efficiency is deliberate—the 17% efficiency is among the best for this size, and the 23 dB PER is sufficient for indoor AR with ambient light levels below 500 lux.

Real-world performance and measurement pitfalls In actual AR glasses, the PER you experience depends on the alignment of the module to the user’s eye. The eyebox is 10 mm × 8 mm, and the PER varies by up to 3 dB across the eyebox due to the waveguide’s exit pupil uniformity. At the center of the eyebox, the PER is 24 dB, but at the edge (5 mm from center), it drops to 21 dB. This is because the grating’s angular response is not perfectly uniform—the TE diffraction efficiency varies by 5% across the eyebox, while the TM leakage varies by 2%. Also, the polarization state of the micro-OLED itself matters. The OLED used in this module has a circular polarizer with a PER of 1000:1 (30 dB) at normal incidence, but at off-axis angles (30°), the PER drops to 100:1 (20 dB) due to the polarizer’s angular dependence. This means the module’s overall PER is limited by the OLED’s polarizer at large angles. In practice, the system-level PER for a complete AR glasses prototype using this module is measured at 19 dB at the center of the FOV and 16 dB at the edge, which is still acceptable for most applications but requires careful optical design to avoid ghosting.

How to verify the PER spec yourself If you’re an engineer evaluating this module, you can measure the PER with a simple setup: a 532 nm laser pointer (with a polarizer to ensure > 30 dB PER), a half-wave plate, and a photodiode with a polarizer. Mount the module on a rotation stage, align the input fiber to the slow axis, and measure the output power with the polarizer oriented for TE and TM. Make sure to use a 2 mm aperture at the exit pupil to simulate the eye’s pupil. The measured PER should be within 2 dB of the datasheet spec. If it’s lower than 18 dB, check for fiber misalignment—the PM fiber’s slow axis must be aligned to within 1° of the waveguide’s input grating axis. Also, check for dust on the grating surface, which can scatter light and reduce PER by up to 5 dB. In one batch test, a module showed a PER of 14 dB because of a 10 µm dust particle on the input grating, which was cleaned with isopropyl alcohol and restored to 22 dB.

Long-term stability and reliability The PER of this module degrades by less than 0.5 dB after 1000 hours of continuous operation at 25°C and 50% humidity, based on accelerated aging tests. At 85°C and 85% humidity (damp heat test), the PER drops by 2 dB after 500 hours due to moisture absorption in the polymer overcoat (a 5 µm thick UV-curable epoxy with a water absorption rate of 0.5% by weight). The TiO₂ grating itself is stable up to 300°C, but the silver PBS coating can tarnish if exposed to sulfur-containing air, reducing PER by 1 dB per year in industrial environments. The module is sealed with a glass lid and a desiccant, but for outdoor use, an additional protective coating is recommended. The PER after 10,000 thermal cycles (-40°C to 85°C) is within 1 dB of the initial value, showing good mechanical robustness.

Impact on image quality metrics PER directly affects the modulation transfer function (MTF) of the display. At 20 dB PER, the MTF at 30 cycles per degree (the resolution limit of the human eye) is 0.45, compared to 0.55 at 25 dB PER. This is because leaked TM light creates a veiling glare that reduces contrast. In a dark room, a PER of 20 dB gives a contrast ratio of 800:1 for a checkerboard pattern, while 15 dB gives 400:1. For text readability, 20 dB PER allows reading 8-point font at 40 cm distance, while 15 dB PER requires 10-point font. The color gamut also shrinks—at 20 dB PER, the sRGB coverage is 95%, but at 15 dB, it drops to 85% because the leaked TM light shifts the white point by 0.01 in CIE 1931 coordinates. For AR applications like navigation or notifications, a PER of 20 dB is fine, but for medical imaging or design visualization, a PER of 25 dB is preferred.

Cost and manufacturing considerations Achieving a PER of 23 dB on a 0.23 inch waveguide requires a wafer-level process with a yield of 85% for the grating etching step. The TiO₂ deposition is done by atomic layer deposition (ALD) at 250°C, with a thickness tolerance of ±2 nm. The grating pattern is defined by nanoimprint lithography with a 100 nm alignment accuracy. The PM fiber pigtail is fusion-spliced to the waveguide with a polarization alignment accuracy of ±0.5°, which is achieved using a polarization analyzer in the splicing setup. The cost of the module is about $45 per unit in volumes of 10,000, with the PER spec being the main cost driver—relaxing the PER to 18 dB would reduce the cost to $35 because it allows use of a simpler polymer grating (n = 1.6) and a standard single-mode fiber instead of PM fiber. For high-volume production (100,000 units), the cost drops to $30, with a PER yield of 90% for the 20 dB spec.