Formally it should not depend of potential. See for example Schelling
.... Phys Rev B65,144306. This authors
demonstrated that for three-body potentials it's not important how you
include three-body part in the expression for
heat flux. Right now I'm testing it for the same set of potentials you
mentioned for Si and SiC. However I have confusing
results, still testing. I'll be really appreciate any comments.
The coefficient 1.60e6 is coming from the text of compute_stress_atom.cpp.
There stress is multiplied by -force->nktv2pr so that stress/volume is
in Bars, see manual.
For the units "metal" this coefficient equals -1.6e6, and it depends of units.