This is an issue with Sekilc formulation not working well for non-intermediate fluids being the hottest one.
I want to look at the Ural method to see if it works better and it seems like a generalized version of the Selick formulation in terms of boundary conditions.