Backside Illumination (BSI) Image Sensors

Keywords: backside illumination sensor,bsi image sensor,cmos image sensor,bsi process,image sensor fabrication

Backside Illumination (BSI) Image Sensors are the CMOS image sensor architecture where light enters from the back of the silicon wafer (opposite the metal wiring) โ€” eliminating the optical obstruction caused by metal interconnect layers above the photodiodes, increasing quantum efficiency by 30-90% compared to front-side illumination (FSI), and enabling smaller pixel sizes (down to 0.56 ยตm pitch) that are essential for the high-resolution cameras in modern smartphones, automotive, and surveillance systems.

FSI vs. BSI Architecture

``
Front-Side Illumination (FSI): Backside Illumination (BSI):
Light โ†“ Light โ†“
[Micro-lens] [Micro-lens]
[Color filter] [Color filter]
โ”Œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ” โ”Œโ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”
โ”‚ Metal 3 โ”‚ โ”‚ Photodiode (silicon) โ”‚ โ† Light hits
โ”‚ Metal 2 โ”‚ โ† Light โ”‚ Thin silicon (~3 ยตm) โ”‚ directly
โ”‚ Metal 1 โ”‚ must pass โ””โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”˜
โ”‚ Photodiode (silicon)โ”‚ through โ”‚ Metal 1 โ”‚
โ””โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”˜ wiring โ”‚ Metal 2 โ”‚
โ”‚ Metal 3 โ”‚
โ”‚ Carrier wafer โ”‚
โ””โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”€โ”˜

FSI: Light blocked/scattered by metal โ†’ low QE at small pixels
BSI: Light hits photodiode directly โ†’ high QE regardless of pixel size
`

BSI Performance Advantage

| Metric | FSI | BSI | Improvement |
|--------|-----|-----|------------|
| Quantum efficiency (green) | 40-55% | 70-85% | +50-90% |
| Quantum efficiency (blue) | 25-40% | 60-80% | +100-140% |
| Angular response | Poor at edges | Uniform | Significant |
| Minimum pixel pitch | ~1.4 ยตm | 0.56 ยตm | Much smaller |
| Crosstalk | Medium | Low (with DTI) | Better color |

BSI Fabrication Process

`
Step 1: Standard CMOS process on bulk wafer (front-side)
- Photodiodes, transfer gates, readout transistors
- Full BEOL metal stack (M1-M5+)

Step 2: Wafer bonding
- Bond CMOS wafer (face-down) to carrier wafer or logic wafer
- Oxide-oxide or hybrid bonding

Step 3: Wafer thinning
- Grind and CMP the original substrate
- Thin silicon to ~3-5 ยตm (need photodiode but not more)

Step 4: Backside processing
- Anti-reflection coating (ARC)
- Color filter array (Bayer pattern RGB)
- Micro-lens array (one lens per pixel)
- Deep trench isolation (DTI) between pixels

Step 5: Backside pad opening and interconnect
- TSV or bond pad connections to front-side circuits
``

Key Technologies in Modern BSI Sensors

| Technology | What It Does | Impact |
|-----------|-------------|--------|
| Deep Trench Isolation (DTI) | Oxide-filled trench between pixels | Prevents optical/electrical crosstalk |
| Stacked BSI | Pixel array wafer bonded to logic wafer | Pixel + CPU in one package |
| 2-layer stacked | Pixel + ISP logic | Faster readout, HDR |
| 3-layer stacked | Pixel + DRAM + logic | Global shutter, extreme speed |
| Phase detection AF | Split photodiodes for autofocus | DSLR-like AF in phones |

Pixel Size Evolution

| Year | Pixel Pitch | Resolution (phone) | Sensor |
|------|-----------|--------------------|---------|
| 2010 | 1.75 ยตm | 5 MP | FSI |
| 2015 | 1.12 ยตm | 13 MP | BSI |
| 2020 | 0.8 ยตm | 48-108 MP | BSI stacked |
| 2023 | 0.56 ยตm | 200 MP | BSI stacked + DTI |

Major Manufacturers

| Company | Market Share (2024) | Key Products |
|---------|--------------------|--------------|
| Sony | ~45% | IMX series (iPhone, Sony cameras) |
| Samsung | ~25% | ISOCELL (Galaxy, HP2) |
| OmniVision | ~10% | OV series (automotive, security) |
| ON Semiconductor | ~8% | Automotive image sensors |

BSI image sensors are the enabling technology behind the smartphone camera revolution โ€” by solving the fundamental optical limitation of front-side illumination where metal wiring blocked light from reaching photodiodes, BSI architecture made sub-micron pixels practical, enabling 200-megapixel sensors in devices thin enough to fit in a pocket while capturing images that rival dedicated cameras.

Want to learn more?

Search 13,225+ semiconductor and AI topics or chat with our AI assistant.

Search Topics Chat with CFSGPT