90 Nm Cmos Image Sensor Production Process Optimization

hi recent years, Complementary Metal Oxide Semiconductor Image Sensor(CIS) is developed well, it is more advanced than CCD image sensor with lower power, smaller chip size, higher integration, which make it used in more and more aspect, such as mobile phone, computer, survalline camera and so on. There are many CIS design, development, fabrication companies emerged for the market demand is getting much bigger in all over the world. The CIS process which is derived from logic process is not only sensitive to the device process, but also the photo diode-its unique device structure, is very sensitive to light and dark current. The photo diode requires more precise, optimized process to meet the market needs of higher image quality. This them is to discuss to optimize the 90nm CIS process which developed from 90nm logic process, so that we can get the improved image.First, in logic process, the STI top corner is sharp after STI process which will induce the higher leakage. To improve the STI corner rounding, will suppress effectively the edge leakage. There are 2 steps in the process will impact the STI corner rounding:one is STI dry etching process, the second is STI linear pre-clean process. In this them, the 2 steps of split demonstrated, so that we get the proper process condition for the rounding corner.Photo Diode is the most important part of CIS device, its basic function is to absorb the light energy to product electrical signal. The more light energy absorbed, the more efficient PD is. So how to improve the light absorption is the hot topic for CIS process developing. In this them, we thin down the physical film thickness above the photo diode to improve the light absorption.The optimization to CIS process is complex and endless. This them is focused on United Microelectronic Corporation (Singapore Branch) 90nm CIS process. Currently, UMC CIS process fab out over 3000 pieces monthly, and keeps stale yield. But with the continues emergence of new technique, the optimization of CIS process will continue.