Multijunction solar cells based on III-V compounds have by far the highest conversion efficiency. However, the fabrication cost is very high. An attractive solution to lower the cost while maintaining high efficiencies is to design multijunctions on cheaper substrates such as silicon. Different routes have been taken to cope with lattice mismatch issues between Si and III-Vs: epitaxial growth of GaAs on Si with or without buffer layers or non-epitaxial techniques such as mechanical stacking. A novel approach has been recently presented consisting in direct epitaxial growth of thin Si or SiGe layers on GaAs by low temperature PECVD, where the high crystal quality reported opens a new way for making III-V/Si tandem solar cells. Yet, due to the low deposition rate, the epi-Si layers are still rather thin (~10μm), and though they are assumed to be of high quality, some defects might be still present at the interface or in the bulk. These critical aspects still have to be carefully investigated to evaluate the potential benefits of this novel concept. Here, numerical simulation has definitely a role to play. Therefore, we propose to simulate these novel structures using TCAD tools and to estimate the defect density inside the thin epi-layer based on experimental inputs. Realistic efficiencies achievable by this novel concept will thus be presented and discussed.