I am trying to solve this using Hilbert's paradox.
Say I have a collection of countable sets: $A_1,...,A_n$.
I want to show that there is a bijective function from then union, $A=\cup{A_i}, 1\leq i\leq k$ to $\mathbb{N}$.
I'm thinking to map the elements from each ${A_i}, 1\leq i\leq k-1$ to a prime number $p_i$ raised to a power according to the element's order in its group ${A_i}$ (which can be ordered, since it's countable), and then use elements from group ${A_k}$ to fill in the "gaps" in $\mathbb{N}$.
But then I have a problem with elements from ${A_k}$ whose image is a power of the prime numbers $p_1,...,p_k-1$ which I've used before. I could map them to powers of the prime $p_k$ but then I have the same problem with other elements whose image is a power of $p_k$...
How may I solve this problem (or maybe my approach is incorrect)?
I think this should work.
First, let me translate the "countable" hypothesis into $A_k\simeq \mathbb{N}$ for all $k=0,...,n-1$. Now let $n\mathbb{N}+k$ be $\{{n*j+k | j\in\mathbb{N}\}}$. It is clear that for all $i=0,...,n-1$, $n\mathbb{N}+k\simeq\mathbb{N}$ (via the function $x\mapsto\frac{x-k}{n}$). Now we get: $\mathbb{N}=\bigsqcup_{k=0}^{n-1}(n\mathbb{N}+k)\simeq\bigsqcup_{k=0}^{n-1}(A_k)$ for what we've just said (the first step is properly an equality: the union returns $\mathbb{N}$; the second follows from the hypothesis as seen).
Here, "$\simeq$" means is in bijection with and "$\bigsqcup$" is the disjoint union. We should also notice that all the bijections are explicit (apart from the ones given by the hypothesis).
I suggest you to take $n=3$ to easily see how this works.