In semiconductor manufacturing, the gate critical dimension is a critical parameter that directly impacts device performance and yield. Maintaining tight control over the dimension is essential for:
This analysis employs statistical hypothesis testing to evaluate process performance and determine appropriate sample sizes for quality control studies.
In the gate lithography step of semiconductor manufacturing, the target gate critical dimension (CD) is 32.0 nm. Engineers want to test whether the process mean significantly deviates from this target. A schematic of the gate is provided below:
presample<-c(31.6, 32.1, 31.9, 31.7, 32.2, 31.8, 32.0, 31.5, 31.9, 31.7)
sd(presample)
## [1] 0.2221111
We now have the standard deviation. We will use this to determine the number of samples we need using the t-power test.
power.t.test(n = NULL, delta = 0.1, sd = .2221111, sig.level = 0.05,power = 0.85,type = c("one.sample"), alternative = c("two.sided"))
##
## One-sample t test power calculation
##
## n = 46.25363
## delta = 0.1
## sd = 0.2221111
## sig.level = 0.05
## power = 0.85
## alternative = two.sided
We get the value of n = 46.253. We round it up to 47
A fab is evaluating two different lithography tools to determine whether they produce wafers with the same mean gate CD.
Population 1 (Tool 1): Data are the measurements from Problem
1.
Population 2 (Tool 2): A second dataset of measurements is to be collected from another tool.
We will perform a two-sample pooled-variance t-test to compare the
tools. The test should be performed at a significance level of The test should be powerful enough to detect
a mean difference of 0.20 nm between tools with 90% probability (power =
0.90). How many samples should be collected from the second population
(Tool 2) and used in the two-sample t-test with Tool 1?
samples<-c(31.6, 32.1, 31.9, 31.7, 32.2, 31.8, 32.0, 31.5, 31.9, 31.7,32.1, 32.0, 32.1, 32.3, 31.9, 31.9, 32.3, 32.2, 31.9, 32.1,31.9, 31.9, 32.1, 31.6, 31.6, 31.9, 31.8, 32.1, 31.8, 31.7,32.3, 32.0, 32.0, 31.7, 31.9, 32.0, 31.7, 32.1, 31.9, 31.9,31.9, 32.4, 32.0, 31.8, 32.2, 31.7, 31.9)
One - sided
t.test((samples),mu=32)
##
## One Sample t-test
##
## data: (samples)
## t = -2.1047, df = 46, p-value = 0.04081
## alternative hypothesis: true mean is not equal to 32
## 95 percent confidence interval:
## 31.87512 31.99722
## sample estimates:
## mean of x
## 31.93617
dat<-rnorm(40,10,2)
dat
## [1] 10.910249 12.595461 11.299137 8.363784 5.674494 9.938236 11.350505
## [8] 9.656028 12.813092 8.751224 8.602887 7.896202 8.084090 9.680444
## [15] 10.775182 6.732601 10.674856 12.714164 10.832880 10.895693 5.934505
## [22] 9.990453 9.634380 6.372859 11.803924 10.351361 8.080458 9.110627
## [29] 11.168839 9.012711 12.327372 8.840692 11.314044 11.105124 9.515748
## [36] 12.943549 9.268618 11.262450 14.039960 8.246597
Two sided
power.t.test(delta = 0.2,sd=0.222111,sig.level = 0.05,
power=0.9,type = "two.sample",
alternative = "two.sided")
##
## Two-sample t test power calculation
##
## n = 26.91595
## delta = 0.2
## sd = 0.222111
## sig.level = 0.05
## power = 0.9
## alternative = two.sided
##
## NOTE: n is number in *each* group
We will use 27 samples
hist(dat,main="histogram of collected data",col="gold")
boxplot(dat,main="Boxplot of data",col="purple")
The complete R code
presample<-c(31.6, 32.1, 31.9, 31.7, 32.2, 31.8, 32.0, 31.5, 31.9, 31.7)
sd(presample)
power.t.test(n = NULL, delta = 0.1, sd = .2221111, sig.level = 0.05,power = 0.85,type = c("one.sample"), alternative = c("two.sided"))
samples<-c(31.6, 32.1, 31.9, 31.7, 32.2, 31.8, 32.0, 31.5, 31.9, 31.7,32.1, 32.0, 32.1, 32.3, 31.9, 31.9, 32.3, 32.2, 31.9, 32.1,31.9, 31.9, 32.1, 31.6, 31.6, 31.9, 31.8, 32.1, 31.8, 31.7,32.3, 32.0, 32.0, 31.7, 31.9, 32.0, 31.7, 32.1, 31.9, 31.9,31.9, 32.4, 32.0, 31.8, 32.2, 31.7, 31.9)
t.test((samples),mu=32)
dat<-rnorm(40,10,2)
dat
power.t.test(delta = 0.2,sd=0.222111,sig.level = 0.05,
power=0.9,type = "two.sample",
alternative = "two.sided")
hist(dat,main="histogram of collected data",col="gold")
boxplot(dat,main="Boxplot of data",col="purple")