talk to your building official

and look at some of joe lstiburek's details for other options

Is the wall already poured? Sometimes I bond out the top of the wall to receive the foam. Can you raise the slab 1-2"? Probably not, but that would allow you to float the slab over the foam. As you said, in residential, a common approach is to bevel the top of the foam; it's worth asking your AHJ if they will allow it for commercial work. 

Another example of the building code micromanaging the project with no though given to how this will work in the real world.  Running the finished floor (probably carpet or vinyl) over the 2" thick insulation doesn't work for attachment.  Also, the slab can't be tied to the foundation wall and will move up and down at will.  All that to save the tiny energy loss at the edge of the 4" slab.  Ridiculous.

I tend to agree with proto, but could you be trying to go beyond the code. Is there any way to add dowels or use an epoxy wedge to keep the slab from moving(which could be expensive, a change order). I am not sure about the floor maybe run a furring strip along the perimeter.

Geezertect, I would counter that it's ridiculous to NOT address the significant thermal bridging where the slab meets the wall. If the sub-slab fill is prepped properly, there will not be any settling. We put point loads on floating footings all the time, and don't worry about settling, and there is almost no load at the edge of a slab. If your slab is "moving up and down at will" then you have made a serious mistake at some point in the process.

It could not be easier to design a thermal bridge free slab; to wait until this stage of the game to address it is unfortunate, but to ignore it entirely would be shameful. 

My point is that the amount of thermal loss along the 4" edge of the slab doesn't justify the flooring attachment issue and possible movement of the slab.

 If the sub-slab fill is prepped properly, there will not be any settling  That's a big "if".  If you are on expansive soil or if you have a less than competent contractor and/or superintendent on the job, you can have a problem.  The guys in the field usually have enough sense to place footings on undisturbed soil, but they tend to be sloppy when it comes to the slab.  Yes, if everybody does everything right, then everything should work out.  But, that's belaboring the obvious.  We need to be realistic about the quality of people building this stuff, and try to design so that our buildings are as idiot-proof as possible.  If it doesn't get done right, the architect is going to be pulled into the fight whether or not the design was correct.  My background was in residential and lowest common denominator commercial, so I may be a tad touchy on this aspect.  Just my opinion.

Geezertect, I understand what you're saying. I've always worked in high end residential and have no experience with commercial work, so my viewpoint may be skewed. However, I have found ways to insulate slabs many times, and the same has been done by others countless times, so to take a walk on what really is an important detail--the thermal bridging--is a cop-out, in my opinion.

For example, one work-around I've done is to make the bottom part of the concrete wall thick enough to provide a shelf for both the foam and the slab above. High density foam can easily support the slab load, and prevents the differential settlement you're concerned about. Another approach is to use exterior foam on the walls. Another is to use a fully insulated, thickened-edge slab or a raft slab. 

Past generations not worrying about heat loss and energy use has gotten us into a difficult spot, so I feel obligated to call anyone out who says that taking energy efficiency into account is too difficult. It's not too difficult, it just requires different techniques than you might be used to.

i'm partial to this last one, although, if you've got a good concrete gal, have her do the beveled edge, slick.

Our typical detail is the bevel cut shown above. It can be tricky if you don't have a good concrete person though (can telegraph through flooring if not covered by trim/wall).

Looking at the four small details above, the obvious flaw is the thickness of the exterior wall versus the thickness of the foundation wall.  They depict a stud wall, but typically (in my area at least) the foundation wall is 8", meaning the inside faces of the two walls don't align.  You will have a 4" strip of the foundation wall exposed, then the edge of the insulation exposed, and then finally the floor slap itself.  The ledge detail solves part of the problem, but you still have the damned insulation edge to deal with when attaching the finished floor.

The large one above shows a 9" thick foundation wall with (I guess) a 2x6 stud wall above it, so the base will--barely-- cover the ragged edge of the insulation.  It works, although trying to convince some cheapskate clients to buy the extra wall thicknesses might present a challenge.  :)

I still think it's a lot of dicking around for very little energy loss.  Natural gas just isn't that expensive.  But..... I know I'm on the losing side of that argument philosophically.

You're obviously just not throwing enough money at the problem. 

Use Dow Corning's aerogel blanket insulation at the slab edge. Material is 10 mm thick and boasts R-9.8/inch. Three layers of the material will give you roughly R-11.5 in 1-3/16 inches of gap. That's small enough you could probably manage it with gyp and wall base. You might also be able to use some type of epoxy joint filler over the top if it and be done. 

I may have missed this, but what is the flooring material? Seems like 2" of rigid insulation wouldn't matter for certain types of flooring. I assume you would install the factory-cut edge flush with the slab as well, so it wouldn't be that ragged.

^  If it's carpet, you have to be able to put down a tack strip.  If it's vinyl, I don't know if the mastic will stick to the insulation.  Same question if it's some kind of ceramic tile.  I guess if we can put a man on the moon, we can do it but it sure looks like a place for a crack or failure.

Have you considered just using an underlayment (plywood, cement board, etc. depending on the flooring)?

This is an energy code requirement? Never done anything like it before

EverydayIntern:  You still have the problem of the attachment of the plywood or cement board edges into the insulation.

shuellmi:  I never did anything like it either.  Maybe it's something that came in after I left.  In any case, I hope the local building departments are exercising discretion in enforcing it, given the obvious practical problems and (I repeat) the minimal energy savings achieved.

thats a truly shit dtl for all the reasons Geezer mentions. not to mention as Geezer points out the diagrams are just wrong..... subsidence will happen, some of us may be dead when it does but for the sake of a thermal break you have limited the structural performance of the building which also means the water and air tightness of the building will be eventually compromised.....you should be addresding the flooring materials thermal break and provide proper air and water barriers at the potential location for condensation, thats ultimately the price you pay here. the heat loss can be kept to a minimum without putting 2" insulation between slab and foundation via interior side detailing. see the Aerogel as an example.....all yhis over essentially condensation, the heat loss again, here is minimal..

beveling of concrete is also dumb for practical reasons as noted by Josh above. AS an Architect it is your job to make the best decision and not the "right" decision purely based on codes etc...in other words as a professonal you decide of all the factors which matters the least and which the most.