Critical Value
The cutoff from the t-distribution that leaves in each tail. A test statistic beyond it leads to rejecting the null, and it sets the width of a confidence interval.
When to use
Look it up in the t-table using the right degrees of freedom and (use the one-tailed row for one-sided tests), then compare it with the t-statistic or multiply it by the standard error to build an interval.
See Lec 5.