Study On Yield For Random Particles In IC Design

Semiconductor manufacturing is a complex process. When it comes into the nanometer technology era, the process steps become more and more, and every step of the process may introduce random particles resulting in circuit failure and then causing yield loss. Yield is an important factor which can determine the profits of semiconductor industry. But the yield loss caused by random particles becomes more and more serious especially in nanometer node. Even in the mature manufacturing environment, the yield can not be 100%. So it is important for us to predict yield during design phase which can give direction to yield enhancement and help to shorten the production cycle. It is also critical to reduce production costs and enhance profits.This paper describes random defects which impact yield and the possible type of failure. And it also gives an overview of the size and spatial distribution characteristics of random particles. Then the current widely used prediction yield models and distribution functions are discussed, and then their limitations and shortcomings are analyzed. Combined with the distribution characteristics of random particles, an improved yield model is proposed. The simulation result given by Matlab shows that our model definitely has higher flexibility and accuracy.Meanwhile, a layout is acquired through floorplan, place, cts and route. Then critical area analysis and yield enhancement is performed based on the layout. Also this paper optimizes the conventional design flow considering that the defect density is not the same in different regions on the chip. Through that, the iterative optimization can be avoided, and production cycle can be greatly reduced. This paper uses different enhancement methods for different fault types to reduce the sensitivity of the design and improve yield, such as double via, wire spreading, and wire widening.Research results indicate that the error using the yield proposed in this paper to predict yield is relative stable, and the simulation result is similar to actual yield. Through the implementation of the enhancement, the yield increases effectively. This study has great significance and promotion for semiconductor industry and it can help manufacturing factories get more profits.