Under-Slab Insulation and Vapor Barrier

Does code require insulation under a slab?

The requirement lives in the energy code, not the building code, and the provision is headed slab-on-grade floors. The operative words are "around the perimeter": "The minimum thermal resistance (R-value) of the insulation around the perimeter of unheated or heated slab-on-grade floors shall be as specified in Table R402.1.3." The Department of Energy's own handout on the 2021 IECC titles that section Slab Edge Insulation.

So for an ordinary unheated residential slab, the energy code has never required insulation under the middle of the floor. What it requires is a perimeter band with a depth.

Three practical consequences are buried in the code's own sentence and all three save money. The insulation extends down from the top of the slab for the tabulated distance "or to the top of the footing, whichever is less", so on a shallow footing you may not have to reach the full depth at all. It is allowed to go down to the bottom of the slab and then turn horizontally, inward or outward, for the remainder of the total distance. And if it turns outward it has to be protected by pavement or at least 10 inches of soil.

One definition to check before you assume the row applies: the table covers slab-on-grade floors with a floor surface less than 12 inches below grade. A basement floor is a different element.

Note that zone 3 picked up a requirement in 2021 that it did not have before, and that zones 4 and 5 went from 2 feet to 4. If you are working to an older adopted edition, the shallower figures apply. We could not read the 2021 IECC itself, which is paywalled, so every figure above comes from a state or municipal adoption or a DOE reproduction, and one source prints the zone 3 cell differently from the other three. Check your own jurisdiction's adopted table.

Where full under-slab insulation actually comes from

It comes from a footnote, which is one reason so many people miss it:

"R-5 insulation shall be provided under the full slab area of a heated slab in addition to the required slab edge insulation R-value for slabs as indicated in the table. The slab-edge insulation for heated slabs shall not be required to extend below the slab."

So a heated slab gets R-5 under the whole floor plus the tabulated perimeter value, and in exchange the edge insulation no longer has to run below the slab, because the full-area layer already intercepts the downward path. R-5 is about one inch of extruded polystyrene, which is a lower bar than the internet's usual advice.

Two details worth knowing. First, "heated slab" is a defined term and it is broader than you would guess: a slab is heated where the heating elements, hydronic tubing or hot air distribution are "in contact with, or placed within or under, the slab". A staple-up or sand-bed system beneath the slab makes it a heated slab. Second, this was not always the rule. In the 2009 to 2015 editions the footnote raised the edge R-value for heated slabs instead; the full-area requirement arrived with the 2018 edition. And at least one state amends it upward: Washington requires R-10 under the full area on every compliance path, which roughly doubles the foam bill on a radiant floor. Our heated concrete floors guide covers the rest of that system.

One collision to watch for if you are in the Southeast. IRC R318.4 says that where the probability of termite infestation is "very heavy", extruded and expanded polystyrene, polyisocyanurate and other foam plastics "shall not be installed on the exterior face or under interior or exterior foundation walls or slab foundations located below grade." A large part of climate zone 3 sits inside that very heavy area, so a builder there can face an energy code asking for slab insulation and a building code restricting foam below grade. Louisiana resolved it by deleting the zone 3 requirement for unheated slabs outright. How your own state handles it is a question for the building official.

Does the insulation go above or below the vapor barrier?

This is the most-asked detail on the subject and the reason the answers online contradict each other is genuinely interesting: ACI reversed its own recommendation, and both camps are quoting ACI, just different editions.

The old advice existed because casting concrete on an impermeable surface means the slab can only dry upward, which makes it curl at the edges and makes finishing harder. The newer advice reflects what actually goes wrong more often in practice, which is flooring adhesive failures from moisture coming up through the slab years later. ACI changed its mind about which risk matters more.

The IRC's own wording is an implicit ordering rule that almost nobody cites: R506.2.3 puts the vapor retarder "between the concrete floor slab and the base course". Foam between the slab and the retarder is not that.

So the defensible order, bottom to top, is compacted subgrade, granular base, foam, vapor retarder, slab, with the sheet in direct contact with the underside of the concrete. That is consistent with the current ACI position, with the IRC's wording, and with the Building Science Corporation detail that puts the polyethylene "on the top of the rigid insulation, in direct contact with the concrete". No code or standard mandates that order, and we are not going to pretend one does.

There is an honest gap here worth stating. ACI's old objection to direct contact was about casting on an impermeable, non-absorptive surface, and foam is also impermeable and non-absorptive, so putting foam directly under a slab reproduces the drying problem whichever side the plastic goes. We could not find any source that addresses that, and no published measurement of curling on foam against curling on a granular base. If anyone tells you confidently which way to go, ask what they are citing.

The vapor retarder, specified properly

The standard is ASTM E1745, for plastic water vapor retarders used in contact with soil or granular fill under concrete. It has three classes, and the thing to notice is what separates them:

The permeance is the same for all three. The class is a toughness rating, not a vapor rating: how much abuse the sheet takes from a crew walking on it and from the rebar chairs sitting on it before it tears. Tensile is measured to ASTM D882 and puncture to ASTM D1709.

Which brings us to the question everybody asks. Six mil or ten mil is the wrong framing, and ACI says so directly: conventional polyethylene "(6, 8, and 10 mil) does not fully conform to the requirements of ASTM E1745 and should not be considered for use as below-slab moisture protection." The roll of clear poly at the big box is usually sold as a drop cloth, publishes no perm rating, and claims no ASTM standard. The IRC asks for a Class I vapor retarder of minimum 6 mil polyethylene or equivalent, and a product that actually carries an E1745 class is the equivalent that clears the bar.

On installation, the IRC wants seams lapped at least 6 inches and the sheet turned up at least 3 inches against the wall. If the same sheet is also doing radon duty, the lap is 12 inches, not 6, under IRC Appendix F, and the larger figure governs. Note that Appendix F is an appendix and is only in force where a jurisdiction adopts it.

One terminology note, because the words are used loosely. The IRC and ASTM both say retarder, not barrier. There is no agreed permeance at which a retarder becomes a barrier: the figures in circulation run from 0.1 perms down to 0.01, and one ACI edition puts it at 0.00 while another says the threshold is "not established". If a product calls itself a barrier, look for the perm number rather than the word.

Is it worth insulating if the code does not make you?

Two pieces of real research answer this better than any product page.

Heat leaves a slab at the edges, not through the middle. A 2021 paper in Energy and Buildings found the edges and corners of a ground floor are 1.6 to 2.1 times more thermally active than the core, that perimeter extruded polystyrene cut annual ground-floor heat loss by more than 20 percent, and, most usefully for detailing, that insulation at grade level is about 4.5 times more effective than the same insulation 150 mm lower down. That is the argument for getting the top of the edge insulation right up to grade and not worrying much about the field of the slab, and it is also why the code asks for a perimeter band.

On payback, the only published figure we could find is discouraging. A Department of Energy Building America study put the simple payback on slab insulation at 17.9 to 47.1 years and stated it is "not cost-effective", with 3.1 to 5 percent site energy savings across climate zones 4 to 7. Two caveats we will not bury: it is a retrofit and perimeter study rather than a new-build full-area one, and it is the only payback anyone has published. We could not find a payback figure for a heated slab at all, which is the case where the argument is strongest, because a radiant slab with no insulation under it is heating the ground.

The honest summary: under a heated slab, insulate, and the code agrees. Under an unheated slab, insulate the perimeter because the code says so and the physics supports it, and treat full-area insulation as a comfort decision rather than an energy one. A concrete floor in a finished room feels cold underfoot whatever the heating bill says, and nobody has published a number for that.

XPS, EPS or polyiso

The decisive property below grade is not the published R-value per inch, it is what happens to that R-value after years in contact with wet ground.

Oak Ridge National Laboratory measured it, and the result is the most important thing on this page. A 2012 ORNL study found extruded polystyrene lost 10 to 44 percent of its energy-saving performance over 15 years in ground contact, with moisture uptake of 67 to 97 percent of the material, against an ASTM C578 laboratory absorption limit of 0.3 percent by volume. The lab test does not predict field behavior. If you are sizing insulation on the strength of R-5 per inch holding forever, that is the study to read first.

Beyond that, the comparison between XPS and EPS is a live dispute between two trade associations with opposite conclusions and a commercial interest in each, and we are reporting both rather than picking. The EPS industry association says XPS absorbs 5 to 60 percent by volume in ground contact and that EPS retains its R-value better. The XPS association cites Alaskan and Quebec highway studies putting XPS at about R-4.1 per inch after 31 years against EPS at about R-2.2 after 21. ORNL is the only non-commercial measurement we found and it tested XPS only.

One common claim we checked and could not support: that polyisocyanurate must not go below grade. The polyiso trade association actively promotes below-grade use and at least one manufacturer sells a product rated to ASTM C1289 Type I at 25 to 60 psi with under 0.2 percent water absorption, explicitly approved for under-slab and slab-on-grade work. We found no manufacturer or association saying polyiso cannot go there. If you have been told otherwise, ask for the source.

Can you really build a slab on foam?

Yes, and the psi argument is almost always aimed at the wrong load.

Foam is graded to ASTM C578 by compressive resistance, and the familiar Type IV at 25 psi, Type VI at 40 and Type VII at 60 are the grades used under slabs. But the C578 number is measured at 10 percent deformation and is not a working stress. A paper presented at ASHRAE's Buildings IX conference puts the allowable long-term stress at roughly 0.10 to 0.35 of the nominal figure, which is the number you should actually design against.

Now the load. A 4 inch slab weighs about 50 pounds per square foot and a residential live load is 40, so the distributed load on the foam is around 0.625 psi. Against even the most conservative reading of Type IV that is a margin of roughly forty to one. Distributed floor load is not the problem.

What is the problem is the same thing that drives slab thickness: a point load. A column base, a car lift anchor, a jack stand or a storage rack foot concentrates load into a few square inches and that is where foam grade matters. Size the foam for the heaviest point load the floor will carry, not for the floor. We could find no published limit on how much foam may go under a slab, which is not the same as there being none, so say that is what we found rather than treating it as settled.

The gravel layer, and the order of everything

The base is a building code requirement rather than an energy one. IRC R506.2.1 asks for a base course of clean graded fill at least 4 inches thick, free of vegetation and organic material. Insulation does not replace it: the base is there for uniform support and drainage, and foam laid on a soft or uneven subgrade will telegraph every low spot into the slab.

From the bottom, then: compacted subgrade, 4 inches of clean granular base, the foam if you are using it, the vapor retarder in direct contact with the underside of the slab, then the concrete with its reinforcement chaired into the upper third. Concrete floor thickness has a cross-section of those layers and the thickness the slab itself needs.

One thing not to do: do not put a sand layer on top of the vapor retarder. It was once standard practice, sold as a blotter layer to help the slab dry evenly, and it is now understood to be a moisture trap, because the sand holds water that has nowhere to go but up through the slab. If a crew proposes it, that is the conversation to have before the pour rather than after the flooring fails.

What each layer costs

Per square foot of floor, derived from published prices and coverage rather than from a cost guide:

Notice the step from the drop-cloth poly to a sheet that actually meets the standard: even at the top of the range it is 38 cents a square foot, which on a 960 square foot floor is about $310. That is a rounding error against a flooring failure, and it is the cheapest decision on this page.

Two things about those retarder prices. The retailer spread is wider than the thickness difference: the same 10 mil roll is $780.40 at one supplier and $983.30 at another, 26 percent apart, and one brand's 15 mil at 22 cents a square foot is cheaper than another's 10 mil at 26 to 33 cents. So shop the roll, not the mil. And note that this layer is required of the slab by code rather than by your choice of floor covering, which is the point our concrete floor cost page makes when comparing concrete against vinyl and tile.

On the foam, perimeter versus full area is a bigger gap than people expect but smaller than it sounds. On a 24 by 40 foot slab, a 4 foot perimeter band is 448 of the 960 square feet, 47 percent, so code-minimum perimeter insulation costs roughly half what insulating the whole floor would. That ratio collapses as the slab gets bigger, because the band is a perimeter and the floor is an area: on a slab twice as long and twice as wide, the same band is only a quarter of the floor.

Work out your own base tonnage with the gravel calculator, and the concrete itself with the concrete floor calculator.

What goes wrong

Curling. A slab that can only dry upward shrinks more at the top than the bottom and lifts at its edges and corners. That is the mechanism behind ACI's older advice to bury the retarder, and it is a real failure, just a less expensive one than the alternative.

Flooring failures. The reason ACI changed its position: adhesives, coatings and resilient flooring fail when moisture comes up through a slab, often years after the pour and long after anyone can do anything cheap about it. This is what the retarder is for, and it is why the class matters more than the thickness.

A torn sheet. Class C poly tears at a third of the puncture resistance of Class A. The sheet gets walked on, dragged over, and has chairs and rebar set on it before the pour. Specify the class and then look at it before the concrete arrives.

Radon, if you are in an area that cares. The same membrane often serves as the radon layer, in which case the 12 inch lap governs rather than the 6 inch one, and in Washington a heated slab over an active or passive sub-slab depressurization system has to be thermally isolated from the gravel layer as well as the soil, because a depressurized layer convects.

Termites. See the code collision above. In the very heavy termite area, foam below grade is restricted by the building code whatever the energy code wants.

Sources

• 2021 Seattle Energy Code, Residential Chapter 4 - a 2021 IECC adoption, carrying the "around the perimeter" wording of R402.2.9, the depth rules and the climate zone table. Note that Washington amends the slab provisions, so this is not the model code verbatim.
• 675 IAC 14-4.4-91, 2020 Indiana Residential Code - Table N1102.1.2 and the heated-slab footnote in a 2018-generation adoption.
• DOE Building Energy Codes Program, what changed in the 2021 IECC - proposal RE32, the new zone 3 requirement and the zone 4 and 5 depth increase.
• ACI 302.1R-15, Chapter 5 - the current position that the retarder belongs in direct contact with the slab, and the finding that 6, 8 and 10 mil construction polyethylene does not conform to ASTM E1745.
• ACI 302.1R-96 - the older advice to bury the retarder under granular fill, which is what the other half of the internet is quoting.
• Kehrer and Christian, ORNL/TM-2012/159, Oak Ridge National Laboratory - extruded polystyrene losing 10 to 44 percent of its energy-saving performance over 15 years in ground contact.
• Liu and others, Energy and Buildings volume 235, 110675 (2021) - edges and corners 1.6 to 2.1 times more thermally active than the core, and insulation at grade 4.5 times more effective than the same insulation 150 mm lower.
• IRC R318.4 as adopted by South Carolina - the restriction on foam plastics below grade in very heavy termite areas.