1. CMOS IMAGE SENSOR SEIED MANOOCHEHR HOSEINI University of TEHRAN-IRAN
In the Name of GOD advanced VLSI
class presentation
CMOS IMAGE SENSOR
SEIED MANOOCHEHR HOSEINI
University of TEHRAN-IRAN
Supervisor : Dr. FAKHRAEI

2. Applications
This growth in not only due to the ability to integrate sensing array with analog- and digital-processing circuits in CMOS, but also the persistently improving image quality of CIS that is comparable to that of CCD [1]. In addition, the pixel size of CIS has been reduced and become comparable to CCD with the help of pixel-sharing architecture that results in the high capacity of photodiode and thus, the high SNR even with the small-size
pixel.

3. The imaging system
[1]

4. FILL Factor
Optically Sensitive Area
~30% of Total Pixel Area
[3]

5. Increase of FILL Factor
[2]

6. Microlens and Colour Filter
[3]

7. Colour Filter Response
[3]

8. CMOS Image Sensor Trend
2.0
3.0
4.0
5.0
1.0
10.0
’02
’03
’04
’05
’06
’07
’08
Pixel pitch, um
5.6um (0.3M pixel)
3.8um (1.3M pixel)
2.8um (2M pixel)
< 1.8um
2.2um (7.2M pixel)
< 1.5um
Year
This Work
[2]

9. A typical 3T pixel

10. 4T Architecture

11. 4T Shared Pixel Architecture

12. Pixel-level charge summation for sub-sampling.
[2]

13. By pixel summation SNR increase by 3dB
Sum of signal =
Sum of noise =
[2]

14. By clumn wise averanging SNR increase by 3dB
Average signal =
Average noise =
Savg
Ramp Gen.
Savg
[2]

15. Old and new GNOX process; improvement of random noise by reducing the trap effect on SF transistor.
The gate oxide fabrication process is changed to reduce the interface trap density of SF transistor. In conventional GNOX process (Fig ), to prevent boron segregation from silicon to gate poly, gate oxidation is formed at high temperature in nitrogen atmosphere. In this case, nitrogen ions spread out from gate oxide to silicon surface due to its diffusion, so that trapping probability of signal electrons in the channel increases, causing flicker noise. To reduce the trapping effect of nitrogen ion, a nitrogen film has been formed under low temperature condition so that the nitrogen ion can be localized far from the oxide-silicon interface.
[2],[4]

16. Increase light gathering power by 20% pixel height shrink
As the pixel size shrinks and the pixel array becomes larger, the distance between micro-lens and photodiode becomes increasingly important in the viewpoints of light-gathering efficiency and crosstalk.
[2]

17. SNR Increase With New Sub-Sampling
[2]

18. Implementation of average filter by MATLAB software
Use average filtering
Original image

19. Implementation of average filter by MATLAB software
Use average filtering
Original image

20. 4-shared pixel architecture
High fill factor (57%)
with SNR max of 41dB under the full-resolution operation.
Pixel level charge summation
Increase 6dB in SNR
Noise reduction technology
SF trap treatment to reduce 1/f noise

21. REFRENSES
[1] A. El Gamal and H. Eltoukhy, “CMOS Image Sensor,” IEEE Circuit and Devices Magazine, vol. 21, no. 3, pp. 6-20, May-June, 2005.
[2] Young Chan Kim, Yi Tae Kim, Sung Ho Choi, Hae Kyung Kong, Sung In Hwang, Ju Hyun Ko, Bum Suk Kim, Tetsuo Asaba, Su Hun Lim, June Soo Hahn, Joon Hyuk Im, Tae Seok Oh, Duk Min Yi, Jong Moon Lee, Woon Phil Yang, Jung Chak Ahn, Eun Seung Jung, Yong Hee Lee.
Samsung Electronics, Kiheung, Korea
“1/2-inch 7.2MPixel CMOS Image Sensor with 2.25µm Pixels Using 4-Shared Pixel Structure for Pixel-Level Summation .” ISSCC 2005.
[3] CMOS Image Sensors : Lenses and Filters on Silicon // Lindsay Grant, ST Microelectronics Robert Henderson, University of Edinburgh
[4] J.Y. Kim, et al., “Characterization and Improvement of Random Noise in 1/3.2” UXGA CMOS Image Sensor with 2.8µm Pixel using 0.13µm Technology,” IEEE Charge-Coupled Devices and Advanced Image Sensors, pp , 2005.