You have correctly identified what you need to do: you need to show that there exists $\frac{n}{3^k}\in\mathbb{Z}_{(3)}$ with the property that there does not exist an $\frac{m}{3^{\ell}}\in\mathbb{Z}_{(3)}$ such that $\frac{n}{3^k}\cdot \frac{m}{3^{\ell}} = 1$.
However, I do not think you have correctly done this. The obvious way to accomplish this is, again as you have tried to do, to simply exhibit a specific $n$ and a specific $k$ and show it has the desired property. That means showing that for this specific choice of $n$ and $k$, no choice of $m$ and $\ell$ will have the property that $\frac{n}{3^k}\cdot \frac{m}{3^{\ell}} = 1$. That suggests that your answer should:
- Specify an $n$ and a $k$; and
- Prove that no choice of $m$ and $\ell$ will work for that $n$ and $k$.
However, what you do instead is specify $n$ and $m$.
Moreover, for your choice of $n$, namely $n=1$, it is not true that $\frac{n}{3^k}$ has no inverse! If $n=1$ and $k$ is an arbitrary nonnegative integer, then $\frac{n}{3^k} = \frac{1}{3^k}$ does have an inverse in $\mathbb{Z}_{(3)}$: namely,$\frac{3^{k+1}}{3}$ is an inverse for $\frac{1}{3^k}$.
You are close, though: if you take $n=2$ and $k=1$, say, to get $\frac{2}{3}$, then you want to show that there is no choice of $m$ and $\ell$ that will satisy $\frac{2}{3}\cdot \frac{m}{3^{\ell}} = 1$. Do that.