A spatially coherent source, such as laser or an antenna surface, of characteristic length $D$, say diameter, operating at wavelength $\lambda$ has a beamwidth of $k_0\frac{\lambda}{D}$, where $k_0$ is factor of order $\tfrac{1}{2} \text{to} \tfrac{3}{2}$, i.e., $\mathcal O(k_0)=1$, and it depends on the shape of the emitting surface, its illumination by the source and how the beam width itself is defined. In this context, by coherence of the emitting surface, for laaer it is output lens, I mean that every point is of the same phase, in other words the emitting surface is an equiphase surface coinciding with a phase front. The geometric factor $k_0$ depends on the amplitude of each emitting point source, each point on the aperture plane, this amplitude can be varying from point to point but no its phase. If the area is large then the each point that act as a Huygens (hemi)-spherical interfere destructively in most directions except the one orthogonal to the surface. In other directions away from the central one the Huygens wavelets interfere partially leading to "sidelobes" that are much smaller than the central "main" lobe in whose direction all points are in phase. Antenna engineers, by convention, usually specify the beamwidth as the half power direction, physicists may prefer first nulls around the peak in whose direction totally destructive interference occur. In either the case, the beamspread in steradians may also be written as $k_1\frac{4\pi \lambda^2}{A_{eff}}$ where now $\mathcal O(k_1)=1$ and $A_{eff}$ is the effective aperture area.
