Image sensors are required in cameras, from phones to telescopes, capturing light and turning it into images. CMOS (Front-Side Illuminated) and BSI (Backside-Illuminated) sensors work on similar principles but differ in structure, affecting light capture, noise, and color quality. This article explains their designs, performance, uses, and future developments in detail.
CC7. From BSI to Stacked CMOS Architectures
CMOS vs BSI Sensor Overview
Every camera, from the smartphone in your pocket to the telescopes exploring distant galaxies, depends on how efficiently its image sensor captures light. Both CMOS and BSI sensors follow similar semiconductor principles, but their structural differences lead to major variations in light sensitivity, noise performance, and image quality. In traditional CMOS (Front-Side Illuminated, FSI) sensors, metal wiring and transistors sit above the photodiodes, partially blocking incoming light and reducing overall sensitivity. This design makes CMOS sensors cost-effective and easier to manufacture, but limits low-light performance. In contrast, BSI (Back-Side Illuminated) sensors flip the structure, positioning the photodiode on top so that light reaches it directly without obstruction. This improves quantum efficiency, reduces noise, and enhances performance in compact or high-end imaging systems, from DSLR cameras to scientific instruments.
CMOS Sensor Architecture
A Front-Side Illuminated (FSI) CMOS sensor represents the earlier and more conventional image sensor structure used in digital cameras and smartphones. In this architecture, incoming light must pass through multiple layers of materials before reaching the photodiode, the light-sensitive region responsible for converting photons into electrical signals.
Working Process
Each pixel in the display operates through a coordinated process involving microlenses, color filters, metal interconnects, transistors, and a photodiode layer. The microlenses first focus incoming light through the red, green, and blue color filters, ensuring that only specific wavelengths reach each subpixel. Above the photodiode, metal interconnects and transistors manage the pixel’s electrical control and signal readout, though their position can partially block some of the incoming light. Beneath these layers lies the photodiode, which captures the remaining light and converts it into an electrical charge, forming the pixel’s basic image signal.
Limitations of FSI Design
• Reduced Light Sensitivity: A portion of the light is reflected or absorbed by the wiring and transistor layers before it can reach the photodiode.
• Lower Fill Factor: As pixel sizes shrink, the ratio of the light-sensitive area to the total pixel area decreases, leading to more noise.
• Weaker Low-Light Performance: FSI sensors struggle in dim environments compared to modern alternatives like BSI sensors.
Inside the BSI CMOS Sensor
The Backside-Illuminated (BSI) CMOS sensor revolutionized digital imaging by addressing the major drawback of traditional Front-Side Illuminated (FSI) designs, light blockage from metal wiring and transistors. By reversing the sensor’s structure, BSI allows incoming light to reach the photodiode directly, dramatically improving light efficiency and image quality.
BSI Technology Function
• The silicon wafer is thinned down to just a few micrometers to expose the photosensitive layer
• The photodiode layer is positioned on the top side, directly facing incoming light
• The metal wiring and transistor circuits are relocated to the back side, preventing them from obstructing light paths
• Advanced microlenses are aligned precisely over each pixel to ensure optimal light focusing
Advantages of BSI Sensors
• Higher Light Absorption Efficiency: Up to 30–50% improvement compared to FSI sensors, resulting in brighter and cleaner images.
• Superior Low-Light Performance: Reduced photon loss enhances sensitivity and minimizes noise in dark environments.
• Improved Color Accuracy: With unobstructed light paths, color filters produce more accurate and vivid tones.
• Compact Pixel Design: BSI supports smaller pixel sizes while maintaining image quality, ideal for high-resolution sensors.
• Enhanced Dynamic Range: Better signal capture in both bright and dim regions of a scene.
Light Efficiency and Sensitivity Comparison
| Feature | FSI CMOS Sensor | BSI Sensor |
|---|---|---|
| Light Path | Light passes through wiring → partial loss | Direct to photodiode → minimal loss |
| Quantum Efficiency (QE) | 60–70% | 90–100% |
| Low-Light Performance | Moderate | Excellent |
| Reflection & Crosstalk | High | Low |
| Image Clarity | Average | Sharp and bright in dim light |
Pixel Shrink and Fill Factor
In FSI CMOS Sensors
As the pixel size drops below 1.4 µm, the metal interconnects and transistors occupy a larger surface area. The fill factor decreases, resulting in less light being captured per pixel and increased image noise. The result is darker images, reduced contrast, and weaker performance in low-light conditions.
In BSI CMOS Sensors
The photodiode is positioned above the wiring, allowing light to strike it directly. This configuration achieves a nearly 100% fill factor, meaning almost the entire pixel area becomes light-sensitive. BSI sensors maintain uniform brightness and a higher signal-to-noise ratio (SNR) across the image frame. They also deliver superior low-light performance, even in compact modules like smartphone or drone cameras.
Crosstalk, Noise, and Backside Diffusion
| Aspect | Potential Issues in CMOS (FSI) Sensors | Potential Issues in BSI Sensors | Engineering Solutions | Impact on Image Quality |
|---|---|---|---|---|
| Optical Crosstalk | Light is scattered or blocked by metal wiring before reaching the photodiode, causing uneven illumination. | Light leaks into neighboring pixels due to backside exposure. | Deep Trench Isolation (DTI): Creates physical barriers between pixels to prevent optical interference. | Sharper images, better color separation, and reduced blurring. |
| Charge Recombination | Charge carriers are lost within thick silicon or metal layers, lowering sensitivity. | Backside Recombination: Carriers recombine near the exposed surface before collection. | Passivation Layers & Surface Treatment: Reduce defects and improve charge collection. | Enhanced sensitivity and reduced signal loss. |
| Blooming Effect | Overexposure in one pixel causes adjacent pixels to saturate due to front-side diffusion. | Overexposure spreads charge beneath the thinned silicon layer. | Surface Doping & Charge Barriers: Contain charge and prevent overflow. | Reduced white streaks and smoother highlights. |
| Electrical & Thermal Noise | Heat from on-pixel transistors generates noise in the signal path. | Higher shot noise due to thin silicon and dense circuitry. | Low-Noise Amplifiers & On-Chip Noise Reduction Algorithms. | Cleaner images, improved low-light performance. |
| Fill Factor Limitation | Metal layers and transistors cover a large pixel area, reducing light sensitivity. | Almost eliminated - photodiode fully exposed to light. | BSI Structure & Microlens Optimization. | Maximum light capture and uniform brightness. |
From BSI to Stacked CMOS Architectures
Structure of a Stacked CMOS Sensor
| Layer | Function | Description |
|---|---|---|
| Top Layer | Pixel Array (BSI Design) | Contains the light-sensitive photodiodes that capture incoming light, using a BSI structure to maximize sensitivity. |
| Middle Layer | Analog/Digital Circuitry | Handles signal conversion, amplification, and image processing tasks separately from the pixel array for cleaner outputs. |
| Bottom Layer | Memory or Processor Integration | May include embedded DRAM or AI processing cores for rapid data buffering and real-time image enhancement. |
Advantages of Stacked CMOS Sensors
• Ultra-Fast Readout: Enables high-speed continuous shooting and actual video capture up to 4K or 8K resolutions with minimal rolling shutter distortion.
• Enhanced On-Chip Processing: Integrates logic circuits that perform HDR merging, motion correction, and noise reduction directly on the sensor.
• Energy Efficiency: Shorter data paths and independent power domains improve • throughput while reducing power consumption.
• Smaller Form Factor: Vertical stacking allows compact module design ideal for smartphones, automotive cameras, and drones.
• AI and Computational Imaging Support: Some stacked sensors include dedicated neural processors for intelligent autofocus, scene recognition, and real-time enhancement.
Dynamic Range and Color Performance in CMOS vs BSI Sensors
BSI (Backside-Illuminated) Sensors
By eliminating metal wiring above the photodiode, BSI sensors allow photons to reach the light-sensitive area directly. This structure increases full-well capacity, improving light absorption and minimizing highlight clipping. As a result, BSI sensors offer superior HDR performance, better color depth, and finer shadow gradation, making them best for HDR photography, medical imaging, and low-light surveillance.
FSI (Front-Side Illuminated) Sensors
In contrast, FSI sensors require light to pass through several layers of circuitry before reaching the photodiode. This causes partial reflection and scattering, which limits dynamic range and tone mapping capability. They are more prone to overexposure in bright conditions and often produce less accurate color in deep shadows.
Applications of CMOS vs BSI Sensors
CMOS (FSI) Sensors
• Machine vision
• Industrial inspection
• Medical endoscopy
• Surveillance cameras
BSI Sensors
• Smartphones
• Digital cameras
• Automotive ADAS
•Astronomy and scientific imaging
• 8K video recording
Future Developments in CMOS vs BSI Sensors
• 3D-stacked designs combine pixel, logic, and memory layers for ultra-fast readout and AI-driven imaging.
• Global shutter BSI sensors eliminate motion distortion for robotics, drones, and automotive systems.
• Organic CMOS and quantum-dot sensors deliver higher sensitivity, broader spectral response, and richer colors.
• On-sensor AI processing enables real-time noise reduction, object detection, and adaptive exposure control.
• Hybrid imaging platforms merge CMOS and BSI advantages, improving dynamic range and reducing power use.
Conclusion
CMOS and BSI sensors have reshaped modern imaging, with BSI offering higher light sensitivity, less noise, and better color accuracy. The rise of stacked CMOS and AI-integrated sensors further enhances speed, image clarity, and dynamic range. Together, these technologies continue to advance photography, surveillance, and scientific imaging with greater precision and efficiency.
Frequently Asked Questions
What materials are used in CMOS and BSI sensors?
Both use silicon wafers. BSI sensors also include thinned silicon layers, microlenses, and metal interconnects for better light absorption.
Which sensor type uses more power?
BSI sensors consume more power due to their complex design and faster data processing, although modern designs are improving efficiency.
Why are BSI sensors more expensive than CMOS?
BSI sensors require additional manufacturing steps, such as wafer thinning and precise layer alignment, which makes them more expensive to produce.
How do these sensors handle heat?
High temperatures increase noise in both sensors. BSI designs often include better thermal control to keep image quality stable.
Can CMOS and BSI sensors detect infrared light?
Yes. When equipped with IR-sensitive coatings or filters removed, both can detect infrared, with BSI showing better IR sensitivity.
What is the purpose of microlenses on image sensors?
Microlenses guide light directly into each pixel’s photodiode, improving brightness and efficiency in smaller BSI sensors.