When it comes to adsorption kinetics, there is a factor called the "sticking coefficient" which is temperature dependent. This coefficient accounts for the fact that not every atom/molecule which hits a surface also stays on that surface. So, in your case the temperature dependence is apparently a decrease with temperature.
Additionally, a change in temperature may affect the residual composition of the gas phase, see fig. 5 and the surrounding explanations in this work by my colleagues:
Article Influence of growth temperature on the properties of aluminu...
Oh, one more addition: if you use ellipsometry for thickness measurements, check the mean square error. For my oxide films, if the material quality is bad, the fitting MSE deteriorates and the thickness seems to increase despite not actually doing so.
So, it would actually be possible that it's to some extent a fit model issue.
Check out the paper "A structure zone diagram including plasma-based deposition and ion etching" by Andre Anders. The density and crystallinity of your film depends on the deposition rate, pressure, temperature, energy of arriving atoms, energy of ions also arriving, and other factors. If you're depositing the same volume of material in a way that makes a more dense film, the thickness will be less.