Using first principles methods (quantum espresso code) I would like to find if a monolayer of a material is flat or buckled. Using the van der Waals interaction in the calculation I found that the buckled configuration has the lowest energy whereas based on the literature it should be the flat one?
You can safely use a dispersion corrected DFT method for 2D systems if your vacuum space is large enough, but there is another issue that must be explained.
Most of the current dispersion correction methods such as DFT-D(Grimmes D2/D3/D4), TS, MBD, etch methods are overestimating interlayer interaction, exfoliation energies. They are not safe to use if you are trying to decide if a structure is more stable with just few meVs.
If you are going to use dispersion correction anyway, I'd suggest you use SCAN+rVV10 or ACFDT-RPA. ACFDT-RPA is considered the fifth rung method on Jacob's Ladder and is computationally very expensive. SCAN-rVV10, on the other hand, got so much attention lately due to its ACFDT-RPA like results for layered materials just with METAGGA level!