This very much depends on statistical power. If the effect size is big enough, 36 participants can be plenty.
I suspect the effect sizes of sleep deprivation are probably quite large.