Let A be a square matrix. In order to find the inverse of matrix A using elementary row operations, we write A = IA.
Now, perform a sequence of elementary row operations successively on A on the LHS and the pre-factor I on RHS, till we get I = BA. Here, B is the inverse of of matrix A.
However, in applying one or more row operations on A = IA while finding A–1 by elementary row operations, if we obtain all zeroes in one or more row of the matrix A on the LHS, then the inverse of matrix A would not exist as we will not get I = BA in this case.
In applying one or more row operations while finding A–1 by elementary row operations, we obtain all zeroes in one or more row, then A–1 __does not exist__.