(This must have been asked somewhere before but for some reason I could not find an answer)
In physics one often meets the criterion that a vector field has zero divergence, e.g.
$$\nabla\cdot\mathbf{v}= 0 \quad (1)$$
Based on this one often claim that the vector field can be written as the curl of another field,
$$\mathbf{v}=\nabla\times\mathbf{A} \quad (2)$$
since $\nabla\cdot (\nabla \times \mathbf{f})= 0$ for any (smooth) $\mathbf{f}$. But the implication here is only obvious from (2) -> (1) and not the other way around.
I tried the Helmholtz decomposition theorem where the vector field can be decomposed into curl-free and divergence-free components
$$\mathbf{v} = \nabla \phi + \nabla\times \mathbf{A}$$
Taking the divergence of this and combining with (1) gives
$$\nabla\cdot\mathbf{v} = \nabla \cdot\nabla \phi = 0$$
But I can't see how this would imply $\nabla \phi = 0$ which I guess would thus imply (2).
Am I missing something or completely on the wrong track?