Using the second moment to estimate parameter $p$ for a Geometric distribution, it is unbiased or not? If it is biased, how to compute it?

Question

Using the second moment (method of moment) to estimate parameter $p$ for a Geometric distribution with $\mu=1/p$ and $var = 1/p^2$, it is unbiased or not? If it is biased, how to compute the bias? Here is what I did.enter image description here