The spray foam guy has given me two quotes for the attic. One is at R21 (3 inches of foam) the other R40(5 inches, I know the math does not seem to work). He wants to see poly (vapour barrier) installed for the R21 installation, but not the R40. At what depth does it become OK not to use a different type of vapour barrier?

What type of foam is he using (just curious, it shouldn't really matter imo)

Walltite. Its a BASF product.

Most 2lb foams becomes a class-II vapor retarder at ~2" thickness. (Minimal class-II is a permeability of 1 perm.) A material's permeance (perms per unit thickness) is fairly linear- double the thickness results in half the permeability. The permeability of 6 mil poly is ~0.05 perms, making it a class-I vapor retarder( anything less than 0.1 perms). To match that with a typical 2lb foam would take quite a bit, but if the code required that it be only 0.5 perms it would be achievable @ ~4". The permeance of the BASF product is probably buried in the spec somewhere.

Whether a true class I vapor retarder is necessary to avoid moisture damage in the roofing deck varies quite a bit by climate. In the colder parts of the Canadian midwest it's probably necessary, but in southern Ontario or western B.C., probably not. (I'm not sure if local codes account for that with a specific permeability number or not.) In the lower 48 of the US only the very coldest parts would need 0.1 perms- most places anything under 1.0 is "good 'nuff", and in mixed heating/cooling climate zones vapor retarders can create as many problems as they solve- better to have both inward & outward drying if you're running air-conditioning 5+ months out of the year.

Half pound foam is highly permeable by comparison to the 2lb stuff, and needs a vapor-retardent paint to achieve class-II vapor retardency at any (reasonable) thickness.

Insulation R-value ratings can be misleading. Small voids
in irregular framing and gaps within the wall cavity between
fiberglass batt and framing of only 1 to 2% can lower the
effective R-value by 25 – 40%.
Conventional R-value tests measure conduction, which is
heat transfer between surfaces in direct contact. The cellular
structure of spray foam is a poor conductor of heat, which
gives rise to high R-values.
_____________________
spray foam roofing texas<!--Session data-->

Thats why 15" wide batts get put into 14.5" cavities, and 23" batts get put in 22.5" cavities. I have never seen a batt that didn't fit into a cavity. 

Are you spraying the roof decking or the attic floor? If you are willing to spray the attic floor, you can spray enough foam to get the appropriate vapor barrier, then blow over it with the blown insulation of your choice. This is a budget friendly way to get the air sealing of the foam and the less expensive R's from the blown insulation.

For example, get 2" of foam blown on the floor for a R-14 and an air seal, and then have 12" of blown insulation installed over it, to get another R-38 for a total R-52 that will cost much less than the R-40 sprayed foam. Remember the attic floor has less square feet than the roof deck and the gable end walls, so the foam should be cheaper and you will have less area to lose heat from.