First of all let $a \in R$
I tried using the root test and got nowhere. Also tried by comparisson but I get stuck. The last thing I tried is to rewrite the series as follows: $$ \sum (\sqrt{n+a} - \sqrt[4]{n^2 + 2n}) = \sum \frac{n+a - \sqrt{n^2 + 2n}}{\sqrt{n+a} + \sqrt[4]{n^2 + 2n}} $$ since I've seen lots of problems similar to this being treated this way. I assume that I need to compare it to a $p$-series. I need a hint.
I would use the fact that $x^4-y^4=(x-y)(x^3+x^2y+xy^2+y^3)$. So:\begin{multline}\sqrt{n+a}-\sqrt[4]{n^2+2n}=\sqrt[4]{n^2+2an+a^2}-\sqrt[4]{n^2+2n}=\\=\frac{2an-2n+a^2}{\sqrt[4]{n^2+2an+a^2}^3+\sqrt[4]{n^2+2an+a^2}^2\sqrt[4]{n^2+2n}+\sqrt[4]{n^2+2an+a^2}\sqrt[4]{n^2+2n}^2+\sqrt[4]{n^2+2n}^3}\end{multline}Since the denominator behaves as $4n^{\frac32}$, the series converges if $a=1$ and diverges otherwise.