Theoretically, due to the large d (the diameter of the specimen over which the illumination is coherent, see page1708), parallel-beam TEM with α ≈ 0 mrad can be used to record lattice fringes with diffraction contrast (so-called HRTEM). However, the lattice fringes in BF-STEM mode are not as good as that in TEM mode. In principle, DF STEM almost doesn't present diffraction contrast due to the negligible d.
In practice, the purity of Z(atomic number)-contrast in HAADF-STEM images can be affected by coherent diffraction contrast if the specimens are periodically arranged atoms (crystals) instead of pure amorphous phase. The crystalline specimen scatters electrons coherently at Bragg angles. The Bragg scattering is very sensitive to the orientation of crystals or crystallites as well as their strains. Crewe et al.  suggested a ratio method to reduce the coherent diffraction contrast effect. 
The Z-contrast in HAADF-STEM images is almost pure in the following two cases:
i) The materials are in amorphous phase. In this case, there is no Bragg scattering.
iii) The crystal structures, consisting of different elements, are the same. In this case, the diffraction contributions from different elements are the same.
 Crewe AV, Langmore JP, Isaacson MS. Siegel B, Beaman D, editors. Physical aspects of
electron microscopy and microbeam analysis. New York: John Wiley; 1975; p. 47–62.