KyL's right.
Since TG does not offer shader models which can make a distinction between dielectrics (sand, plastic) and conductors we assume the visibility of a metal is due to its reflectivity and thus diffuse is 0.
Increasing IOR then assures we get an entirely reflective surface.
That's where KyL's suggestions came from.
When using a shader model capable of handling conductors then IOR is not needed to increase the reflectivity and a more physically plausible value can be used.
In the end the index of refraction is the difference in speed of light between a vacuum and a material and it's not that the speed of light is 20 or 50 times slower in metals, it's just that TG's shader model is dielectric, so we need to cheat.
Here's a nice little read which popped up in my mind:
https://www.chaosgroup.com/blog/understanding-metalness@KyL I believe the reflective shader model can be chosen? Blinn or Blinn-Beckmann if I recall correctly.
I don't know the default method for speculars in -say- a default shader...hmmm...makes me wonder.
It's an approximated BRDF I'm pretty sure of, since we can't control fresnel for instance.
You can make a fresnel-like shader though from blue nodes and feed that into a reflective shader's roughness input (with roughness set to 1 iirc)