subtract average values with SEM

#1
I have values of two groups, each has its own mean and calculated standad error of the mean (SEM):

A +-SEM(A)
B +-SEM(B)


I wouls like to present the substruction of these values with its calculated SEM: (A-B) +-SEM(A-B)

Is it right that:

SEM(A-B)=sqrt( [SEM(A)]^2 + [SEM(B)]^2 ) ?

Can i use it even if n is not the same for both groups?