R29
Dark fixed pattern noise generation by Negative-Bias-Temperature (NBT) stress
on CMOS imager pixel transfer gate
Hirofumi Yamashita, Motohiro Maeda, Shogo Furuya, Takanori Yagami Toshiba Corporation, 2-5-1, Kasama, Sakae-ku, Yokohama, 247-8585, Japan, Phone : +81-45-890-2383, Fax : +81-45-890-2492, hirofumi.jp
Introduction
Negative bias on transfer gate is an efficient way to reduce dark current by Shockley-Read-Hall (SRH) mechanism generated via Si-SiO2 interface states at a transfer gate channel [1] [2] [3]. But it has also been reported that lower negative gate bias increases dark current noise, such as dark current generated by Gate-Induced-Leak (GIL) Trap-Assisted-Tunneling (TAT) mechanisms [2] [3] [4] [5], thus setting appropriate negative gate bias is an important issue to avoid additional noise
generation. This contribution reports newly discovered dark Fixed-Pattern-Noise (FPN) generation caused by Negative-Bias-Temperature (NBT) stress on transfer gate. The dark FPN has similar behavi
or to Vth shift after NBT stress usually observed in p-type MOS-FET [6][7][8]. The dark FPN dependency on stress condition was investigated with test pixel array AC operation, and the root cause model of the dark FPN is proposed.
Test Procedure
The test pixel array with 1.75um pitch 2-way shared pixel was used for the dark FPN measurements. The pixel schematic used in the test pixel array is shown in Figure 1. The negative gate bias is applied from external power supply. The pixel raw data are taken after the on-chip Correlate Double Sampling (CDS) circuitry and Analog to Digital Converter (ADC). Operational timing is controlled by an
on-chip timing controller. Sensor chip temperature is varied by controlling the test chip atmosphere temperature. The dark image analysis is done after averaging 10 dark images so that a temporal noise in dark image can be minimized.
Experimental Results
Figure 2 compares dark signal distribution before
NBT stress to dark signal distribution after NBT stress. The NBT stress applied is -2.5V negative gate
bias, 60 ℃ temperature, and 30 minutes stress
time. Dark signal distribution was taken by normal dark image readout operation with -1.5V negative gate bias as a transfer gate low level. As shown in Figure 2, the dark signal distribution tail in lower dark signal level is slightly varied after NBT stress is applied. Figure 3 shows dark image from test pixel array after NBT stress. Darker pixels, which appear to be black defects, generated by NBT stress are observed. It also is observed that dark distribution shift is more obvious when negative gate bias is lower or temperature is higher. In order to see the dark distribution shift behavior is caused by the similar mechanisms with ordinary pMOS-FET Vth shift after NBT stress, dynamic behavior of off-stress dark signal distribution right after the NBT stress is observed. Figure 4 shows dark signal distribution recovery after NBT stress. Fast recovery of dark distribution (about ~50% of dark level shift) is observed as is often observed in pMOS-FET Vth shift after NBT stress. This result
suggests that the root cause of the dark FPN generation has similar mechanisms to that of pMOS-FET Vth shift after NBT stress. Figure 5 shows black defect probability dependency both on negative gate bias and on temperature. BT stress time is 30 minutes. As shown in Figure 5, Black defect probability is 0 when stress bias is -1.5V but increases when stress bias is lower than -1.5V, especially when temperature is higher than 60℃. The proposed mechanism of black defect generation by NBT stress is
shown in Figure 6. Since the dark FPN dependency on NBT stress and the dark FPN dynamic recovery behavior after NBT stress are similar to those of the ordinary pMOS-FET Negative Bias Temperature instability (NBTI), root cause of the dark FPN is probably similar to that of the pMOS-FET NBTI. Key phenomenon is the interaction of accumulated holes with Si-SiO2 interface. During NBT stress, accumulated holes and H atoms terminating Si dangling bond react and generate interface states[6][7][8]. Reaction products are captured in gate dielectric and produce positive fixed charge. The positive fixed charge generates potential pockets in a transfer gate channel. After the NBT stress, when dark electrons accumulated in photo-diode are readout while transfer gate is ON, a part of dark electrons are captured in the potential pocket and then recombine with accumulated holes via generated interface states when transfer gate goes OFF. In order to see the transfer gate Si-SiO2 surface has a critical role in the dark FPN generation, dark signal distribution dependency on negative gate bias is compared between before NBT stress and after NBT stress. Figure 7 compares dark signal cumulative distribution before NBT stress and after stress. The NBT stress are -3.5V negative gate bias, 30 ℃temperature, and 30 minutes stress time. Dark current increase is observed after NBT stress when negative gate bias is -0.8V, whereas less dark current increase is observed after NBT stress when negative bias is -1.5V. The results indicate that the number of interface states at transfer gate channel is increased by NBT stress, and dark current with SRH mechanism is increased via gene
rated interface states. It can be concluded that NBT stress on transfer gate generates electron trap level at transfer gate channel and produce dark FPN, if NBT stress with large negative bias with high temperature is used.
Conclusion
Dark FPN generation caused by NBT stress was found. Cause analysis of the dark FPN was investigated with test pixel array AC operation, and the root cause model of the dark FPN is proposed. The proposed root cause for the dark FPN is the generation of interface states by interaction of accumulated holes with Si-SiO2 interface, as is often observed in pMOS-FET NBTI.
References
[1] H. Han, et al, International Image Sensor Workshop, 2007 pp238
[2] B. Mheen, et al, Electron Device Letters V ol.29, No.4, April, 2008, pp347
[3] H. Yamashita, et al., International Image Sensor Workshop 2009, Session 04-1
[4] B. Pain, et al, “International Image Sensor Workshop, 2005 pp145
[5] N. V. Loukianova, et al, Trans. Electron Devices V ol.50, No.1, 2003, pp77
[6] O. Jeppson, et al., J.appl.Phys., p2004, 1977
[7] V. Huard, et al, 41st IRPS, 2003,, p178, 2003
react to stress的中文翻译[8] T. Yang, et al. VLSI Tech., p92, 2003,
PD PD
Xfer Xfer
S i g .
V D D
Figure 1 Test pixel schematic.
Figure 2 Dark signal distributions comparison between before and after NBT stress. NBT stress are -2.5V negative gate bias and 60 ℃ temperature, and stress time is 30 minutes. Dark signal distributions are taken with Vxfer=-1.5V and with 30 degree temperature. (a) Dark histogram, (b) Enlarged dark histogram.
Figure 3 Enlarged dark image taken after NBT stress. Circles show black defect pixel.
Figure 4P +FD
Xfer gate
N -
N +
hole
①Before NBT stress
××××poly
poly
poly
Gate insulator
Si
Si
Si
②During NBT stress
③After NBT stress
×
×P o t e n t i a l
Gate insulator
Gate insulator
×
Dynamic behavior of off-stress dark distribution right after NBT stress. NBT stress are Vxfer=-2.5V , Temperature is 60degree, and stress time is 30minutes. Figure 5 Black defect probability dependencies on negative gate bias and on temperature. Stress time is 30 minutes. Dark signal distributions are taken with Vxfer=-1.5V and with 30 degree temperature.
between before and after stress. NBT stress are
Vxfer=-3.5V and 35 ℃ temperature, and stress time is 30 minutes. Dark signal distributions are taken with
Vxfer=-1.5V and with 30 degree temperature. Obvious change of dark signal distribution is observed after NBT stress when dark signal distribution is taken with Vxfer=-0.8V .
版权声明:本站内容均来自互联网,仅供演示用,请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198,我们将在24小时内删除。
发表评论