On-screen fingerprint sensor with optically and electrically tailored transparent electrode patterns for use on high-resolution mobile displays

Full-screen smartphones eliminate front bezels and buttons, forcing relocation of fingerprint sensors away from the intuitive front face. While optical and ultrasonic under-display sensors are commercially used, they have drawbacks (e.g., dry-finger recognition issues, recognition time, yield, and cost). Mutual capacitive on-screen fingerprint sensing is attractive because it is proven in conventional home-button sensors, can share layers with touch sensing, works with both OLED and LCD, and suits flexible/foldable devices. However, capacitive on-screen sensors must employ transparent electrodes placed directly atop the display, risking display image degradation via Moiré patterns caused by interference between the periodic sensor electrodes and the display pixel matrix. This work asks: which transparent electrode geometries (pitch and rotation) minimize visible Moiré while maintaining sufficient capacitive sensing performance, and can such a sensor be implemented on a commercial smartphone? The study analyzes Moiré formation, searches pattern parameter spaces numerically and experimentally, and addresses sensitivity shortfalls through electrical block design, ultimately demonstrating on-display fingerprint imaging.

The paper reviews on-display fingerprint sensing modalities: optical systems (susceptible to dry-finger misclassification due to surface reflections), ultrasonic systems (challenges in recognition time, dry-finger performance, and manufacturing yield), and mutual capacitive sensors (mature technology in non-display implementations, integrable with touch, compatible with OLED and LCD, and beneficial for foldables). Prior Moiré studies for touch panels (often metal-mesh replacements for ITO) show Moiré can be mitigated by pattern geometry because touch electrode pitches (~4 mm) are much larger than display pixel pitches (~40–120 μm). In contrast, fingerprint sensor pitches (~50–100 μm) are similar to pixel pitches (e.g., 46 μm at 550 ppi), making Moiré far more challenging. The paper also references human visual contrast sensitivity function (CSF) to assess perceived visibility thresholds (~60 cycles/degree cutoff, peak sensitivity near ~8 cycles/degree).

- Sensor stack and constraints: Mutual capacitive sensor must be above the display and beneath a thin cover glass due to limited sensing distance (≤300 μm). Transparent electrodes are required; ITO chosen (~10 Ω/sq sheet resistance, ~94% transmittance in air; ~86% on substrate). Typical stack includes lower ITO (Tx/Rx), organic insulator, upper ITO, and organic passivation. A thicker lower glass (0.3–1 mm) and laminations are used to ensure mechanical robustness and suppress display noise. - Moiré analysis framework: Construct grayscale N×N images of overlapped display (diamond-pentile OLED, 570 ppi) and sensor patterns (diamond-type ITO) across pitch 20–100 μm (1 μm steps) and rotation 0–45° (1° steps; fourfold symmetry). Apply discrete Fourier transform to obtain spectral peaks corresponding to Moiré spatial frequencies. Define a visibility circle using the human CSF with a 40 cm viewing distance; threshold at 60 cycles/degree. Compute an approximate Moiré visibility (MV) metric by summing amplitude F(u,v) weighted by CSF C_s as MV = ΣΣ F(u,v)·C_s(√(u^2+v^2)). Use MV to narrow candidates, acknowledging limitations (e.g., a single strong peak vs many weaker peaks can yield similar sums). Complement with image-domain binary pattern extraction (binarize overlapped and inverted images) to visualize expected Moiré periods and directions. - Candidate selection: Based on MV maps, identify regions with minimal MV; notably around 30 μm pitch at rotation angles ~7–12°, 15–20°, and 35–40°. Select 20 candidates with low MV for empirical evaluation. - Fabrication of test coupons: Fabricate 55 sensor pattern samples on glass (10 mm × 10 mm each) using the sensor process (lower ITO/insulator/upper ITO/passivation), omitting electrical pads and interconnects. Include 35 additional patterns, mainly with 60–70 μm pitches (common in commercial sensors), to test if rotation could suppress Moiré in that range. - Optical evaluation setup: Dice the wafer to smartphone display size (Galaxy S8). Place samples on a full-white-mode OLED panel with glycerine coupling. Acquire grayscale images in a dark environment using a high-resolution industrial inspection camera, controlling for camera-induced artifacts. Quantify Moiré contrast by comparing standard deviations of pixel gray-level values across images and use mean opinion scores to align with perceived visibility. - Electrical design consideration: For selected small-pitch patterns where single-unit capacitance change is too low, electrically connect three unit cells into a block to boost signal. Implement block-wise driving in the readout circuit. - Prototype demonstration: Assemble the chosen transparent electrode pattern and block-driving architecture into a prototype integrated with a commercial smartphone display to capture fingerprint images.

- Transparent conductor selection: ITO electrodes with ~10 Ω/sq sheet resistance exhibited ~94% transmittance in free space (~86% with substrate), yet still risk visible Moiré when overlaid on displays. - Moiré predictability: DFT-based spectral analysis and binary image extraction accurately predicted Moiré directions and periods observed experimentally. Example cases: 20 μm pitch at 0° produced visible Moiré with ~190 μm period; 25 μm at 38° showed dominant period ~105 μm; 45 μm at 0° yielded very large, highly noticeable millimeter-scale Moiré periods (~2.7–3.0 mm). - Human perception framing: Using a 40 cm viewing distance and CSF threshold of 60 cycles/degree effectively delineated visible vs non-visible Moiré components. - Optimal parameter regions: MV maps indicated minimal Moiré for ~30 μm pitch at rotations around 7–12°, 15–20°, and 35–40°. Most candidates were chosen from ≤30 μm pitches. Patterns with 60–70 μm pitch (typical in conventional sensors) did not achieve acceptable Moiré suppression via rotation. - Experimental validation: Among 55 fabricated patterns, camera observations and statistical contrast analysis matched numerical predictions, enabling down-selection of low-Moiré designs suitable for on-screen integration. - Signal sensitivity solution: Because the chosen small-pitch electrodes produce insufficient single-cell capacitance change, electrically grouping three unit patterns into one block increased signal strength, enabling practical sensing. - System demonstration: With the selected pattern and block-wise driving, fingerprint sensing on a live display was demonstrated on a commercial smartphone prototype.

The study addresses the core challenge of on-screen capacitive fingerprint sensing: minimizing Moiré-induced display artifacts without sacrificing sensing performance. By uniting frequency-domain analysis (DFT with CSF weighting) and image-domain pattern extraction, the authors efficiently explored the pitch–rotation design space and identified regions where Moiré components largely fall outside human visual sensitivity. Experimental imaging confirmed the predictive power of the numerical pipeline. Recognizing that the Moiré-minimizing designs use small pitches that reduce per-cell capacitive signal, the electrical architecture was adapted by block-connecting three unit cells and employing block-wise driving, restoring adequate SNR for fingerprint imaging. The combined optical–electrical co-design thus achieves acceptable display quality and functional fingerprint acquisition on a commercial smartphone display, indicating mutual capacitive on-screen sensors can be viable alternatives to optical/ultrasonic solutions with benefits in integration, compatibility, and potential suitability for foldables.

This work develops and validates an optical–electrical co-design methodology for mutual capacitive on-screen fingerprint sensors using transparent ITO electrodes. Key contributions include: (1) a DFT- and CSF-based screening framework to predict and minimize visible Moiré across wide pitch/rotation sweeps; (2) empirical validation via fabricated pattern libraries and high-resolution imaging on OLED displays; (3) identification of low-Moiré design regions (notably around 30 μm pitch with specific rotations); and (4) an electrical block-connection strategy to recover sensing sensitivity for small-pitch electrodes. A prototype on a commercial smartphone display demonstrates practical fingerprint imaging. Future work could extend the approach to diverse pixel architectures and resolutions, optimize readout circuits for further SNR gains at small pitches, and explore alternative transparent conductors/pattern topologies that balance resistance, transparency, and manufacturability.

- Moiré inevitability: With two overlapping periodic structures, Moiré cannot be fully eliminated—only reduced below perceptual thresholds. - MV metric caveat: The summed CSF-weighted spectrum (MV) may conflate a single strong spectral component with many weaker ones; it is best for narrowing candidates rather than absolute optimization. - Display dependence: Findings are tied to the tested diamond-pentile OLED (570 ppi); optimal pitch/angle regions may shift with different pixel layouts or ppi. - Sensitivity trade-off: Small pitches that minimize Moiré reduce per-unit capacitance change, necessitating block grouping and tailored drive/readout schemes. - Transparency vs resistance: Achieving low sheet resistance (~10 Ω/sq) with ITO reduces transmittance (~86% with substrate), which can heighten contrast of residual Moiré and imposes material/process constraints.