GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter Instagram YouTube Icon Navigation Search Icon Main Search Icon Video Play Icon Plus Icon Minus Icon Picture icon Hamburger Icon Close Icon Sorted

Community and Q&A

Roof deck: open-cell or closed-cell foam?

EightNineTen | Posted in General Questions on

I’ve been going back and forth with my spray foam company.  My foamer originally pushed open cell but is now telling me closed cell, which was my decision, is the best application for my situation.  He originally said that open cell was overall a better value.

I have 2 by 9’s at the roof.  So I’m limited on space for spraying.

My final plan was to put r21 of closed cell spray in with r21 of regular batts underneathe it to achieve r40+.

I live in NY.

After talking to an energy guy, a pro, he said open cell 100%.  It lets the water come through so you know if there’s a leak.  But my sprayer said so what?  He said the open cell foam would absorb 20lbs of water before that happened and if you had to fix the area, you would have to remove the wet open cell foam.

So. Anyone have some guidance on this?

GBA Prime

Join the leading community of building science experts

Become a GBA Prime member and get instant access to the latest developments in green building, research, and reports from the field.


  1. Expert Member
    Dana Dorsett | | #1

    Is that NYC or some other part of NY? (It matters!)

    By 2x9" do you really you really mean 1.5" x 9.25" ( the nominal dimensions of a milled 2x10)?

    In NYC/US climate zone 4 just 2" of blown closed cell foam (HFO blown, preferred) would be sufficient for dew point control on R30 rock wool (7.25" nominal thickness), which would be a better overall value. That would be a nominal R42-44.

    If you want to take it up to the full code minimum performance, installing 3" of closed cell foam (R18-R21) and cap-nailing 1.5" wide strips of 1" polyiso on the rafter edges plus R30 rock wool gets it to R48-R51 at center cavity, and with the R6 poliso cuts thermally breaking the rafters it will beat a code-min R49 on performance even if only R48 at center cavity.

    The whole leak detection canard about open cell foam is really a 1-sigma type event. With a full cavity fill of open cell foam it would be lower R-performance but it would also allow more moisture into the roof deck from interior moisture drives, risking rot over large sections of the roof (particulaly shaded or north facing roof pitches.)

    I zones 5 & 6 it will take more closed cell than 2" foam to achieve dew point control at the foam/fiber boundary. At 3" it would be fine for R30 rock wool in zone 5, which would work with the 1" polyiso edge strips. For zone 6 it takes 4" to have dew point control on 5.5" /R23 rock wool, with half inch polyiso edge strips.

    Cutting foil faced polyiso is dead easy using a steel wallboard taping knife that has been sharpened on the edges. See:

    1. EightNineTen | | #3

      Dana - that reply below was for you!

  2. EightNineTen | | #2

    Dana -

    I live on Long Island in Nassau County.

    Can you tell me what you suggest specifically? Are you saying to do 2 inch closed cell foam under roof deck with r30 roxul under it? Was my original plan of 3 inches of closed cell with a r19 kraft eco batt under it going to perform the same as the roxul? Tell me what you think is the optimal setup.

    1. Expert Member
      Dana Dorsett | | #6

      Long island is zone 4A. In zone 4A as long as ~30% of the total R is closed cell foam at the roof deck (either above or below the roof deck) the risk of moisture accumulation is low.

      HFO blown foam runs about R7/inch, and is considerably more environmentally friendly than closed cell foam blown with HFCs. At 2" you would have R14 foam, and with R30 rock wool you'd be at R44 total, with R14/R44= 32% of the R as low-permeance foam on the cold-in-winter side of the assembly. With R6/inch closed cell foam it would be R12/R42= 28.5%- a bit under spec, but still not a big moisture hazard. (Installing more than 2" of HFC blown foam at a time without a cooling period between lifts is a fire hazard as it cures, but most HFO foam can be installed of lifts of 4" or more.) The low vapor permeance of the closed cell foam keeps the roof deck from loading up with moisture over the winter, but is still sufficiently vapor open to provide seasonal drying when the roof deck is warm/hot over the spring & summer.

      Installing 9.25" of half-pound open cell foam has to happen in two lifts with a cooling period between, and would result in only R34. Most 0.7lb open cell foam would be ~R38 @ 9.25" . Both are too vapor open to be risk-free, and would need half perm paint (aka "vapor barrier latex") on gypsum board or 2-mil nylon (Certainteed MemBrain) "smart" vapor retarder on the interior side to mitigate that risk.

      In my area 9.25" thick half-pound foam runs $2.75-$3.25 per square foot, as does 2" of HFO blown closed cell foam. Most HFC blown closed cell about $2-$2.25 per square foot @ 2". The R30 rock wool runs $1.25-$1.50 per square foot (even at box store pricing) So even with 2" of HFO blown foam and R30 rock wool you'd be no more than $4.75 per square foot for R44. ($0.93/R-foot, worst-case)

      Closed cell 3" HFC foam would run about $3 per square foot + $0.60 for the batts, for $3.60 for R37. ($0.97/R-foot, typical)

      Closed cell 3" of HFO blown runs about $4 per square foot, + $0.60 batts for about$4.60 for R40. (($0.87/R-foot, typical.)

      It's all pretty similar in terms of $/R-foot, but the 2" HFO blown foam + R30 rock wool brings it quite a bit closer to current code minimum. ( R44 with the 2" HFO blown foam instead of R34 with the 3" HFC blown foam.) The biggest savings is the 2" foam instead of 3".

      1. EightNineTen | | #7

        Then you're saying:

        2 inch closed cell
        r30 rockwool underneathe


      2. EightNineTen | | #8

        Wait, what is HFO? I'm getting lost.

        1. Expert Member
          Dana Dorsett | | #9

          Yes, 2" of foam (preferably blown with HFO blowing agents), with R30 rock wool underneath. If there are sections where it's not going to have gypsum board nailed to the underside of the rafters, perforated fabric type radiant barrier stapled to the rafter edge to keep them up during an earthquake installs quickly & cheaply, and will not form a moisture trap at about 15 cents per square foot. (Perforated only- other RBs are true vapor barriers, and will trap moisture.) Kraft facers don't have the fire ratings to be safe when left exposed, but RB does. The RB adds another R1 or so of performance when it's facing air, nothing at all if it's under gypsum board.

          HFO1234ze (tradename Honeywell Solstice) is a refrigerant now used for blowing polyurethane foams that has an extremely low global warming potential, comparable to that of CO2. Several vendors are now using it, but it's a cost adder, more expensive per inch than HFC blown goods. The industry standard HFC245fa has a global warming potential on the order of 1000 x CO2.

          A non-exhaustive list of HFO blown foams are:

          Foam-Lok 2000-4G

          Demilec HeatLok HFO High Lift

          Icynene ProSeal HFO

          GacoOnePass Low GWP

          I'm sure there are others. See:

          1. EightNineTen | | #10

            Keep in mind, I am a newbie with this stuff.

            Please confirm:

            A) 2 inch closed cell spray foam (I don't believe he offers the products you're referring to, just the standard closed cell stuff)

            B) R30 Rockwool under 2 inches of closed cell

            That's the way to install it. Please confirm.

            As for sheetrocking underneathe, we will not be rocking at the roof rafters. It's going to be completely open. Let me know if this is a problem.

  3. BrianPontolilo | | #4


    For a while, lots of builders and foam installers were making the suggestion that we should be using open cell foam when insulating roof decks for the exact reason your installer suggested it: so that we'll know if the roof was leaking. We heard a lot of that reasoning over the years at FHB. But it turns out that it is an unnecessary caution and that closed cell can be used and will perform better in that application. Here is an article that may be helpful with your situation:

    1. EightNineTen | | #5

      Thank you Brian! Can you look at how i'm trying to install it and let me know what you believe is optimal? Did you see my reply to Dana?

  4. Expert Member
    Dana Dorsett | | #11

    >"Please confirm:

    A) 2 inch closed cell spray foam (I don't believe he offers the products you're referring to, just the standard closed cell stuff)

    B) R30 Rockwool under 2 inches of closed cell

    That's the way to install it. Please confirm."

    That's it!

    If the product is rated less than R13 @ 2" it's a bit marginal from a dew point perspective (the outermost part of the rock wool could get damp by the end of winter) but not enough to worry about. The IRC prescriptive for zone 4 calls out R15 minimum exterior side foam for R49 total, which is 30.6%. R13 out of R43 is 30.2% (good 'nuff for the kind o' girlz I go with! :-) It's "in the noise".) R12 out of R42 would be 28.6% still not enough be a real problem unless using high solar reflective index roofing (eg: bright white Galvalume metal roofing.)

    If it's only R12 @ 2" and you're the kind of person who would lie awake night thinking about it, install 2 mil nylon (Certainteed MemBrain) on the underside of the rafters and make it air tight. Nylon membrane has variable permeance to water vapor and will behave as a vapor barrier when the roof is at it's coldest, but will become vapor open if there's enough moisture in the rock wool to support mold (not that mold really grows on rock wool or foam anyway.) At 15 cents per square foot it's a lot cheaper than going with an other inch of closed cell, which would have to be installed after a cooling period if its HFC blown foam, making the foam even more expensive.

    Closed cell foam installers are really accustomed to installing 2" lifts, but many would want to push it to 3" in the first pass (to save time) if that was the target thickness. That's both a fire hazard and a quality issue with HFC blown foam- don't even THINK of going there, even if the installer says it would be fine. (Most manufacturers would not back that up.)

  5. Expert Member
    Dana Dorsett | | #12

    For a sanity check on the relative safety of the 2" foam + R30 rock wool stackup, see Tables 3 & 4 of this set of WUFI simulations by one of the better building science groups out there:

    See the "2" ccSPF + spray fiberglass" column in Table 3. At R49 (total, which means R36 fiberglass) in climates as cold as Minneapolis or International Falls MN it's only a risk if the roofing is a light colors highly solar reflective material. In Table 4, simulations at high indoor humidity, it's at most a slight risk for anything but a solar reflective roof or climate much colder than yours. R30 rock wool is about as air retardent as the 1.8 lbs density fiberglass simulated- a decent comparable.

  6. EightNineTen | | #13

    "R30 rock wool is about as air retardent as the 1.8 lbs density fiberglass simulated- a decent comparable."

    You're way beyond me with this stuff. Can you put this in layman's terms?

  7. Expert Member
    Dana Dorsett | | #14

    Fiber insulation isn't air-tight, but the degree to which it resists air flow varies by density. Low density fiberglass R19s fit the same wall cavities as a high density R21, but allow orders of magnitude more air flow than the R21s. The term "air retardency" refers to how much it resists air flow.

    In the case of the roof insulation stackup it can matter. When the roof is cold the entrained air in the insulation at the top of the fiber layer is denser than the warm air in the bottom, and air will convect within the batt as the warm air rises, cold air falls. Those convection currents can potentially move more moisture to condense on the cold foam if the batt has no air barrier on the bottom side. At 1.8lbs per cubic foot density blown fiberglass is sufficiently resistant to those flows that it doesn't really matter, and the same is true for rock wool at the density found in R30 batts.

    At those air densities the water vapor diffusion is more important than the convecting air, and both fiberglass & rock wool are very vapor-open to water vapor diffusion.

    Thus even though the insulation I'm recommending is not what was simulated in the referenced paper, they are quite comparable in all respects that we care about for the moisture discussion. (From a fire resistance point of view they differ by quite a bit- fiberglass melts at temperatures still relevant to house fires, rock wool does not.) But from strictly a moisture point of view it's a decent comparison, not different enough to change the result of the modeling.

  8. EightNineTen | | #15

    Dana, all over my head. All I need to know is yes or no, is the rockwool going to perform better in terms of air tightness and moisture defense compared to an eco batt or regular pink insulation?

    1. Expert Member
      Dana Dorsett | | #17

      Fiberglass vs. rock wool: Moisture performance- NO difference, thermal performance- a measurable difference, fire resistance- HUGE difference.

      With fiberglass it would need another layer in the assembly to protect the foam from catching fire to meet code. With rock wool as little as 1.5" is enough- with R30s you'd have 7.25", about 5x the minimum necessary.

      1. EightNineTen | | #18

        Bottom line, rockwool = better

        1. Expert Member
          Dana Dorsett | | #20

          Usually more expensive, but in general, better- especially if any of it will be left exposed to the attic (even when behind perforated radiant barrier holding it up.)

          1. Deleted | | #21


          2. EightNineTen | | #22


            Spray foamer is pushing this formula below. Do you believe it will perform the same as the setup you and I discussed above (the r30 rock under 2 inches of closed cell spray)

            R21 ext walls cc
            R21 3.5 in cc roof
            R19 batts for roof under spray to achieve r40 (owens corning brand)

  9. GBA Editor
    Martin Holladay | | #16

    Q. "Is the rockwool going to perform better in terms of air tightness and moisture defense compared to an Eco batt or regular pink [fiberglass] insulation?"

    A. No, but the mineral wool will be easier to install, and will have a slightly higher R-value per inch, than the other two products you list.

  10. jberks | | #19


    Just from my experience, you'll know if there is a leak right way with closed cell foam just as well, if not quicker.

    Last time I didn't double check my guys roof penetration flashing, well we got a leak dripping down through the closed cell spray foam. So Despite what the "pros" say (ie the guy who sprays the shit and leaves, never to experience the aftermath) closed cell foam does not stop water infiltration. If there's a leak in the roof membrane, water will at some point find it's way to a low point and drip. (From my limited experience, as always I could be wrong)

    It's actually better in this case I used closed cell, because open cell would have absorbed a whole bunch of water before it got fully saturated and started to drip.

  11. Expert Member
    Dana Dorsett | | #23

    >"R21 ext walls cc
    >R21 3.5 in cc roof
    >R19 batts for roof under spray to achieve r40 (owens corning brand)"

    Only a FOAM guy would be pushing that!

    Line by line:

    >R21 ext walls cc

    With the typical 25% framing fraction closed cell foam between framing is a waste, robbed of it's performance by the thermal bridging. You can't actually get 3.5" of cc foam in a 2x4 wall anyway- it's too tough to trim. The best they're going to do is 3.25" (but they'll charge you for 3.5". :-) )

    Install R15 rock wool instead (R23 if it's 2x6 framing.) The 3.25" of closed cell foam adds less than R1 to the whole-wall performance after factoring in the thermal briding, and in the walls you don't need any cc foam for dew point control in your climate zone. For the math on the whole-wall R performance see:

    For less money and higher performance you could install 1.75" thick Bonfiglioli edge strips on the framing and compress R19s or R20s or R23 rock wool into the 5.25" space:

    >R21 3.5 in cc roof
    >R19 batts for roof under spray to achieve r40

    You only need R8 foam to have dew point control on R19. With just R12 (2") you would have sufficient dew point control for up to R30, ending up at R42 in 9.25" total thickness (a 2x10 rafter). Shaving off the cost of the "extra" 1.5" of foam and R19 batts more than pays for the R30 rock wool. And the rock wool is a sufficient thermal barrier for fire safety on the foam whereas the R19 fiberglass is not.

    Seriously go with the previously discuseed...

    ...B) R30 Rockwool under 2 inches of closed cell...

    ...not because I came up with it, but because it's the best value, greener, and equal or higher higher performance than 3.5" cc + R19 fiberglass.

    1. Deleted | | #24


    2. Deleted | | #25


    3. EightNineTen | | #26

      99% doing what we discussed but have TWO quick questions:

      1) It's easier to work with r15 roxul, so is it the same to do 2x r15 roxul to reach r30?

      2) what do you think of 2 inches of closed cell with 8 inches of open cell under it to get past rafters and deal with the thermal bridging? still not as good of value right?

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |