best case= o(nlogn)
If we choose central element as pivot then array is divided into two equal part. And recurrence relation become T(n)=2T(n/2)+n which will take o(nlogn) time.
Worst case =o(n^2).
If we choose central element as pivot then array is divided into two part one part contain 0 element and other part is n-1 element. And recurrence relation become T(n)=T(n-1)+1 which will take o(n^2).