We can show the claim either by direct computation or by Lie algebra theory.
In the first case let $E_{ij}$, $i\neq j$ and $H_i=E_{ii}-E_{i+1,i+1}$ for $i=1,\ldots n-1$
be a basis of $L={\frak{sl}} (n,k)$ and $A\in Z(L)$. Then $0=[A,E_{ij}]=AE_{ij}-E_{ij}A$
for all $i\neq j$ and $0=[A,H_i]=AH_i-H_iA$ for all $i$ imply that $A=0$, if $p\nmid n$.
On the other hand, we can use that $L$ is simple in characteristic zero.
This means, that every ideal, in particular the center $Z(L)$, is zero or ${\frak{sl}}(n,F)$ itself. However $L=Z(L)$ would imply that $L$ is commutative, which means $n=1$ and
$Z(L)=\frak{sl}(1)=0$. In the other case, $Z(L)=0$.
In characteristic $p$ this argument is more complicated, because for $p\mid n$
the center is in fact $1$-dimensional and $L$ then is not simple.
However, $L/Z={\frak{psl}}(n,F)$ then is simple for $n>2$.