LIVE image
A state-of-the-art, multiple-low-e-coating Serious Window with xenon gas fill being installed at the Rocky Mountain Institute. Photo: Rocky Mountain Institute. Click on image to enlarge.

An in-depth look at the fascinating world of low-conductivity gas-fill for high-performance windows.

Over the last two weeks I've covered the major strategies for improving the energy performance of windows: adding extra layers of glass, increasing the thickness of the airspace between the layers of glass, and adding low-emissivity coatings. Another important strategy is to use a low-conductivity gas instead of air in the space between the layers of glass. Most commonly argon is used, though krypton is available for the highest-performance windows, and xenon is occasionally used.

Low-conductivity gas-fills don't make as much difference as adding an additional layer of glazing, increasing the spacing between the layers of glass from a quarter- to a half-inch, for adding a low-e coating, but they are nonetheless significant--and definitely worth it when choosing new windows that have low-e coatings. Adding argon is the most cost-effective improvement you can make to a window. But what are these gas-fills and how do they work?

Why low-conductivity gases make sense

To understand how argon works, we have to go back to how heat moves through windows. There are three modes of heat transfer: conduction, convection, and radiation. With clear (non-low-e) double-glazed windows, radiation accounts for about half of the heat transfer, with conduction and convection each accounting for about 25%. When a low-e coating is added to a window (see last week's blog), the radiant component of that heat loss is significantly reduced, and as a result the conductive and convective portions become much more significant. As the name implies, low-conductivity gas fills reduce conductive heat flow. Most of us think of conduction, which is the transfer of kinetic energy from molecule to molecule, as occurring only through solids--think of a hot cast-iron skillet handle--but conduction also occurs across gases. Sometimes we refer to this as gas-phase conductivity.

The Noble gases

Air has a thermal conductivity of 0.014 Btus per square foot per hour for every degree Fahrenheit difference in temperature (don't worry about those units) at room temperatures. If we can replace that air with a lower-conductivity gas, we can slow the heat loss through windows. Argon is a great option. It has a conductivity of 0.0092--34% lower than that of air--and it is, by far, the most common low-conductivity gas for windows.

Some of the highest-performing windows use the more exotic gas, krypton, in the space between the glass. Krypton has a conductivity of 0.0051, which is 63% lower than that of air. Xenon, an even rarer gas, has a conductivity that's 79% lower. These gases--found in the far-right-hand column of the Periodic Table--are all highly stable and unreactive, an attribute that earned them the moniker of "noble gases" (so named because, like nobility, they don't interact with commoners).
A chart from FDR Design, Inc. showing the impact of gas fills and the spacing between the layers of glass in an IGU. Graphic: FDR Design, Inc. Click on image to enlarge.
All of these gases are components of the air we breathe. Argon makes up a little less than one percent of our atmosphere (third after nitrogen and oxygen) and is produced quite inexpensively as a byproduct of extracting oxygen out of the air. Krypton is present in air at a concentration of about one part per million (one ten-thousandths of a percent), and xenon is present at an even smaller concentration. As a result, these exotic gases are far more expensive to extract.

Buying a window filled with krypton instead of argon adds about $100 to the price, according to a Marvin Windows and Doors rep I spoke with recently, while there is little if any additional cost for argon. At a manufacturing cost of only about 10¢ per window, it's one of the best deals around, according to Randi Ernst, president of FRD Design, Inc., which sells gas-filling equipment to the window industry.

The benefit of low-conductivity gases

Adding argon to a double-glazed window reduces the U-factor by about 0.05 (reducing the U-factor means reducing heat flow). With non-low-e glass, adding argon drops the U-factor from 0.50 to 0.45, a 10% reduction in heat loss (assuming optimal spacing for the glass). When there's a low-e coating, that same argon improves the U-factor from 0.30 to 0.25--a much more respectable 17% improvement in performance. Using krypton with an optimal spacing drops the U-factor by another 0.025, so the total improvement over air is 25%.

The optimal thickness for gas-fill

With an insulating glass unit (IGU), there is an optimal thickness that varies according to the gas fill. With a thicker air space there's less conductive heat loss, but if the spacing gets too deep convective loops form that begin increasing heat loss (see my blog two weeks ago). With air, the optimal thickness for the air space is about a half-inch--assuming the standardized temperature conditions used for modeling window performance in this country. Argon is about the same--just a few millimeters thinner.

Significantly, if we assumed a lower difference in temperature (delta-T) between the indoors and outdoors, as they assume in Europe, there would be less convection between the glass and the optimal thickness would be greater--as we find on European windows. Because most of the U.S. actually experiences a significantly lower delta-T than the 70°F assumed in U.S. standards, a thicker glazing spacing actually makes sense.

With krypton, though, the optimal thickness is significantly less: about 5/16th of an inch (with U.S. delta-T assumptions). This is because krypton is more slippery than air or argon. It forms convective loops more easily, which increases that convective component of heat flow.  

Do we really want radioactive windows?

It is a relatively little-known fact that krypton is somewhat radioactive. There are a lot of isotopes of krypton; krypton-85 with a half-life of 10.8 years, is the one that raises concern. Krypton-85 is produced by the fission of uranium and plutonium, and it gets released in the atmosphere through nuclear bomb testing, releases by nuclear power plants, and by the reprocessing of spent nuclear fuel.

The latter source is the most significant, and a majority of that comes from the French reprocessing plant, Cogema La Hague, which has been operating since 1976. The concentration of krypton-85 in the atmosphere has increased several-hundred-fold since the early 1940s, and some of that krypton-85 ends up in the krypton we extract from the atmosphere. As a result, canisters of krypton gas have measurable levels of radioactivity.

Is this significant for us, though? Probably not very. In most areas, the radioactivity from krypton in our windows will be lower than background radiation. If we're willing to live with other sources of radiation in buildings, such as concrete foundations and granite countertops, we probably shouldn't worry too much about krypton. However, ionizing radiation is cumulative, and when we can avoid exposure we should try to do so.

Does the gas stick around?

The question of whether the low-conductivity gas lasts in an IGU is huge. If it leaks out in a few years, it wouldn't be worth spending more for it. The rule-of-thumb, based on laboratory testing, is that 1% of the gas will be lost per year. Oddly, there has been very little research done on gas retention rates in the field.

Fortunately, the research that has been done offers generally good news. Randi Ernst has done about the only field testing of gas retention rates that I know of. From repeatedly testing several dozen windows over a period of years, he has found that about 0.6% of the gas leaks out per year.

That's a pretty low leakage rate: a window starting with 95% argon would be down to 79% argon after 30 years and 70% argon after 50 years. Even assuming 1% annual loss, after 30 years, there will still be 70% of the original argon, and after 50 years 58%. Most windows don't last 50 years for other reasons, so I'm comfortable with the gas retention.

Bottom line

It's always worth adding low-conductivity gas fill to an IGU. While I'm not terribly worried about the radioactivity of krypton, it does give me pause, and we get far more bang for the buck with argon. If I want better energy performance than can be achieved with low-e and argon in an IGU, rather than replace the argon with krypton, I'll specify a third layer of glass with another low-e coating or a second low-e coating on the inner (#4 surface) of a double-glazed IGU (see last week's blog).

Alex is founder of BuildingGreen, Inc. and executive editor of Environmental Building News. Archives of the Energy Solutions blog can be found here. To keep up with his latest articles and musings, you can sign up for his Twitter feed

If you enjoyed this article, sign up for BuildingGreen email updates:



1 When you state that "Adding a posted by Robert Riversong on 04/03/2012 at 09:32 am

When you state that "Adding argon to a double-glazed window reduces the U-factor by about 0.05", it is important to qualify that as center-of-glass U-factor. The reduction in whole unit U-factor will not be as significant, since most conductive losses are through glass edge spacers and window frame.

Your statement that "No matter what the gas is in an insulating glass unit (IGU), there is an optimal thickness" suggests that there is an optimum thickness regardless of which gas is included. While the rest of the paragraph makes clear that this is not the case, the opening sentence confuses the matter and would be better stated as "There is an optimum thickness for each different gas fill."

It may be that the ionizing alpha radiation from krypton fills in windows don't alarm you because it is often" less than background levels", but radiation exposure is cumulative and any additional source adds to the background level. Also, windows are in every room of the house, unlike granite countertops or concrete foundations. If krypton adds $100 to the cost of a window, increases radiation exposure and it's primary advantage is allowing the use of thinner IGUs, it doesn't seem like a wise investment to me.

2 Robert, Thanks. I fixed the c posted by Alex Wilson on 04/03/2012 at 09:57 am

Robert, Thanks. I fixed the confusing point about optimal glass separation and clarified my thoughts about the radiation exposure.

3 Alex, Further to my previous posted by Tony Marshallsay on 04/04/2012 at 07:16 am


Further to my previous comments about seal leakage in IGUs, I would like to point out the significance of the phrases "laboratory testing", "field testing", "From repeatedly testing several dozen windows over a period of years" and "in the field".

When it comes to testing gas leakage rates from sealed IGUs, it seems to me that this could only be reasonably accurately done by first measuring the production fill mass of the gas in question (a value probably arrived at by dividing the amount of gas used to fill a large number of IGUs by that number), then measuring the amount of gas left when the seal is broken on a sample IGU in a previously evacuated chamber. If that were the case, the leakage rate results could be strongly influenced by variability in the initial fill and in quality of the seal, with it being practically impossible to determine which had the greater influence.

In any case, since it is impossible for a single IGU to tested multiple times over a period of years, as this test method is by definition destructive, and I cannot imagine anyone taking the trouble to remove several dozens of IGUs from a building over multiple years, my conclusion is that the results obtained over time by Randi Ernst were probably from tests on multiple random samples from a large stock of similar IGUs held in some convenient warehouse. If so, in the light of my previous comments on seal life under weathering and thermal cycling, I would have to consider them unreliable. If, however, the samples were set out in racks in an open field somewhere (within weatherproof frames), they could be taken as valid.

Please let me know where I can find this information.

4 Tony, Actually, Randi Ernst u posted by Alex Wilson on 04/04/2012 at 09:11 am

Tony, Actually, Randi Ernst uses specialized equipment that allows windows to be tested in place. I didn't get into the detail of this testing equipment. So, indeed, the same windows can be tested over and over, and there is no loss of gas in the testing process. These windows are installed in his office building.

5 Alex, Thanks for your reply. posted by Tony Marshallsay on 04/05/2012 at 12:55 am


Thanks for your reply. If that is the case, then I really would like to know how Randi manages to assess gas loss while leaving the IGUs in place and intact.

6 Nice article, Alex. On the ra posted by Allison A. Bailes III, PhD on 04/05/2012 at 07:32 am

Nice article, Alex. On the radioactivity of Kr-85, it's a ?? emitter, and the ?? particles (electrons) are of relatively low energy (251 keV average, 687 keV max). As long as the gas is enclosed in the IGU, the radiation is most likely all absorbed by the glass and window frame and never gets a chance to ionize the atoms in anyone's body. You could hold a Geiger counter up next to the window, and it would probably not register a single count due to the Kr-85 inside the IGU.

With the low leakage rate you cited above, only a little bit of Krypton will get into the air in the house, and that little bit contains only a small amount of Kr-85. Once in the house, it has a greater-than-zero chance of damaging cells -- but it's a chance that's barely above zero because of the tiny, tiny amount mixed in with all the house air. That's another reason for a good ventilation system, but even without it, the chances of Kr-85 causing harm are slim to none.

So, you're right not to worry about the radioactivity of Kr-85. I doubt I'd use windows filled with Krypton, though, just because of the extra expense.

7 Can you say a little more abo posted by Derek Roff on 04/05/2012 at 07:42 am

Can you say a little more about the radiation risk from Krypton? If it emits alpha particles, then they probably can't penetrate the glass panes. I checked several websites, which said that alpha particles won't penetrate more than a few centimeters of air, nor the outer layer of your skin. and Wikipedia, for example. This makes it sound like there would be no exposure to radioactivity, except through the krypton that leaks out of the window, and is breathed into someone's lungs. I'm guessing that this is a very low exposure.

If we really wanted to fill our windows with radioactive gas, what about using radon? We could start a little cottage industry, pumping radon out from under houses, and putting it into windows. It's a wonder that such windows aren't available in the market.

8 Oops. I guess either my brows posted by Allison A. Bailes III, PhD on 04/05/2012 at 07:57 am

Oops. I guess either my browser or this site didn't recognize the special character I put in there for the type of radiation emitted by Kr-85. Those two places where you see ?? should have the symbol for beta-minus.

9 windows with argon are common posted by Traonvouez on 04/05/2012 at 10:39 pm

windows with argon are common feature here in France, and has been for a long time, so it's hard to understand how it can be new in the US. When ordering for a window we are asked questions like: "with air or special glass inside?"," low emissivity coating?", and possibly double or triple glazing?" What we don't have is something like Serious Windows multi-mylar film, an installers would love it because it's lighter than triple glazing

10 Response to Tony on non-invas posted by Alex Wilson on 04/09/2012 at 05:53 am

Response to Tony on non-invasive testing of gas fill:

Randi Ernst uses a highly specialized meter, called the GasGlass Handheld, made by the Finnish company Sparklike Ltd.

This is how the technology is described on the Sparklike website:

"The technology is based on plasma emission spectroscopy. A high voltage spark is launched in the IG units cavity causing a light emission which is observed and analyzed further.

The user can simply place the device against the unit, press the button and receive an instant result with high accuracy."

You can learn more at

11 Alex, I downloaded and read t posted by Tony Marshallsay on 04/14/2012 at 08:27 am


I downloaded and read through the information on the device Randi uses. As far as I can see, it may be a very good tool when in the hands of a highly-skilled university lab technician working in a clean, climate-controlled environment; but evidently the manufacturers decided that they needed to spend more effort on a smart "design" that somehow justified what is probably a very high price than on making it a robust device more tolerant of installation conditions in the "real world".

Maybe that's being too cynical but there are so many warnings and caveats in the manual that I, for one, would have second thoughts about even taking it out of the box. And I really don't like the idea of creating a spark within the IGU by means of a device placed up against it.

Since so many factors can affect the readings, it is hard to see how repeatability can be achieved, and hence there must be doubt as to the accuracy of readings taken in anything less than perfect conditions. I also wonder how well it would perform when used on an IGU with insulating rather than conducting spacers.

I refer to comments I made on a previous article regarding the long-term adhesion performance of IGU sealants under the high temperatures and wide thermal cycling we experience here in the Middle East, which were based on conversations with people in the trade and lead me to believe that more representative leakage results would be obtained by destructive testing as I described - albeit at considerably higher cost than using Randi's tester.

— Share This Posting!

Recent Discussions

posted by behrlich
on Nov 23, 2015

PLAs do not have to be made from corn, of course, but in this case it is, which means it probably does have some GMO, unfortunately. The product...

posted by priya
on Nov 20, 2015

Hard to rate this product without knowing if it's GMO.

Recent Comments

Autoclaved Aerated Concrete (AAC): Will the U.S. Ever Lighten Up?

Simon Hatton says, "Merhis Projects are an Australian Developer and Construction company and have been using their patented Merhis Building System for the last 4 years...." More...

Getting to Know Spider Insulation

Robert Haverlock says, "Cynthia, All wood has formaldehyde. And as far as I know, Canada does not allow added formaldehyde in their building products, Roxul included. Ask..." More...

Alex Wilson says, "Cynthia, yes that's correct; there is formaldehyde in mineral wool (urea-extended phenol formaldehyde). I'm not as worried about that as many others..." More...

Cynthia Crawford says, "

Alex-Sorry- I meant to say sprayed- fiberglass. Thank you for your answer in any case, for both foam and fiberglass. It's the first time I've...

" More...

Alex Wilson says, "Cynthia, Spider is a spray-fiberglass product, not a spray-foam, but neither material could be considered rodent-proof. In our home, rodent entry at..." More...