Step 1: Suppose \(\vec{A}\) is the gradient of a scalar function \(\phi\), so \(\vec{A} = \vec{\nabla}\phi\). We want the condition this places on \(\vec{A}\).
Step 2: Take the curl of both sides: \(\vec{\nabla} \times \vec{A} = \vec{\nabla} \times (\vec{\nabla}\phi)\).
Step 3: A standard vector identity states that the curl of any gradient is identically zero:
\[\vec{\nabla} \times (\vec{\nabla}\phi) = \vec{0}.\]
This holds because the second-order mixed partial derivatives of \(\phi\) commute.
Step 4: Therefore, if \(\vec{A}\) is a gradient it must be curl-free (irrotational). Conversely, a curl-free field on a simply connected region can always be written as a gradient, so this is the required condition:
\[\boxed{\vec{\nabla} \times \vec{A} = 0}\]