| dc.description.abstract | In this paper, we randomly sprinkle dots to generate synthetic wafers for testing, and
construct a boomerang model. The number of defects in the synthetic wafer is further
analyzed through the characteristic parameters obtained from the previous research. We are
hoping that by the characteristic parameters of the data compared to the real average number
of defects, see if we can accurately calculate the average number of defects. Then, by
comparing the average number of defects, the equivalent defect efficiency value for judging
the phenomenon of wafer clustering is obtained, and it is applied to synthetic and real wafers.
The Poisson model is used as the basis for this research. First get the true average defect
number λ0, and then use various characteristic equations in previous research [11][16] to
estimate the number of defects. By adjusting different parameter conditions: wafer size, fixed
defect number, defect number in a specific interval, etc., analyze the estimated defect value
and its distribution, and finally make statistics.
Next, according to the equivalent defect efficiency value(EDE) obtained from the
previous research [16], we will make some slight corrections. Using the corrected results to
analyze the clustering phenomenon of synthetic and real wafers, and observe the results of the
wafers in different EDE. Then compared with the previous study [4] B-score to get an
equation that can be used to derive the relationship between the EDE and the standard
deviation.
Finally, we apply EDE to the homemade special wafer in [13], and observe the
relationship between the two and judge EDE we obtained from the result. Then compare these
parameters with B-score to verify the correctness of the equation we derive and the
correlation between the two parameters. | en_US |