We have studied at 20 Torr the very high temperature (i.e. 850°C to 1100°C) growth kinetics of SiGe in Reduced PressureChemical Vapour Deposition. The growth chemistry adopted was heavily chlorinated, i.e. SiH 2 Cl 2 + GeH 4 + HCl, in order to avoid any significant chamber quartz wall coating (and thus flaking) during the deposition of thick layers. We are mostly in a high temperature, supply-limited regime, with a small increase of the SiGe growth rate as the GeH 4 mass-flow and/or the growth temperature increases. By contrast, the Ge concentration decreases for given precursor mass-flows as the growth temperature increases. Its dependence on the F(GeH 4 ) / F(SiH 2 Cl 2 ) + F(GeH 4 ) mass-flow ratio, almost linear at 850°C and 900°C, gets parabolic for higher growth temperatures. We have used those data points to grow linearly graded (~ 9%Ge / µm) SiGe Virtual Substrates (VS) with final Ge concentrations in the 15% to 45% range. The growth temperature was gradually decreased as the Ge concentration increased, in order to obtain long misfit dislocations' segments (high glide velocity) and thus potentially lower Threading Dislocations Densities (TDDs), while keeping the surface as flat as possible. The macroscopic degree of strain relaxation of the constant composition SiGe layers on top of our VS increases as the Ge concentration increases : from 96.2% ([Ge] = 16%) up to 99.5% ([Ge] = 46%). The surface of those SiGe VS is as expected cross-hatched (undulations running along the <110> directions, with a spatial wavelength in the 2 to 3 µm range). Because of the higher surface roughness inherent to those high growth temperatures, we did not obtain the strong TDD reduction that might have been expected. We indeed ended up with TDD in the 1.5 to 2x10 5 cm -2 , i.e. values barely inferior to the ones achieved at 900°C for such a grade. Almost no dislocation pile-ups were observed, however. .