Warning: This page is very long. Windows are a crucial part of passive-solar design, and there’s a lot to say about them. If you persist to the end, you’ll know a lot about how to choose the right windows.
Windows need to let in the sun’s warmth in winter. At the same time, windows are often the main source of heat loss in winter in an otherwise well insulated house. There’s a ton of information about windows on the internet. Most of it is garbage. Even at its best, available online information is confusing, misleading, and of little help to someone designing a passive-solar house.
We need some physics to understand windows. Energy from the sun reaches us in the form of electro- magnetic waves that include not only visible light but also ultraviolet and infrared radiation. The figure on the left below shows a somewhat simplified picture of the solar spectrum at ground level. (The atmosphere has filtered out some of the sun’s energy, especially ultraviolet.) Ground-level solar energy is distributed roughly
Ultraviolet radiation 5%
Visible light 45%
Infrared radiation 50%
It’s important to note that half of the sun’s energy is infrared radiation – wavelengths longer than our eyes can see. This is called near-infrared radiation because the wavelengths are only slightly beyond the visible spectrum. A passive-solar house needs to capture this energy but, as we’ll see, standard windows filter out most of the near-infrared radiation. Altogether, essentially all the sun’s energy falls between wavelengths of 0.3 μm (300 nm) and 2.5 μm (2500 nm).
All objects both absorb and emit electromagnetic waves as what’s called thermal radiation, but the wavelengths of that radiation depend on the object’s temperature. The sun radiates visible light, along with UV and infrared, because it’s an extremely hot ball of gas. You and all the objects around you also emit thermal radiation, but at much longer wavelengths because you’re much cooler than the sun. The figure on the right above shows the thermal radiation from a room-temperature object. The shape of the spectrum is the same as solar radiation but the wavelengths are roughly a factor of 10 longer, what we call far-infrared radiation.
The most intense thermal radiation is at wavelengths near 10 μm, and essentially all the emitted energy is found between wavelengths of 5 μm and 25 μm. The much shorter wavelengths of visible light are shown on the left for reference. The main thing to note is that there is absolutely no overlap of the wavelengths of solar radiation and room-temperature radiation. This is going to be important for understanding windows.
What about a hot day when the temperature is 100°F rather than room temperature? Or a cold 0°F day? These changes nudge the thermal radiation spectrum slightly to the left or right, but hardly enough to notice. The main conclusion – that all the emitted energy is found between wavelengths of 5 μm and 25 μm – remains true.
Energy in the form of heat passes through windows in three ways:
Enter low-e glass. Essentially all windows today are made from what’s called low-e glass, which stands for low emissivity. To a physicist, emissivity is a number that characterizes an object’s ability to absorb and emit thermal radiation. It seems surprising, but good absorbers of thermal radiation – dark materials – are also good emitters. A dark material has a high emissivity. (A perfect absorber, which physicists call a black body, has an emissivity of 1.)
A piece of glass is not an absorber at the wavelengths of visible light – you know this because visible light passes through – but glass is a quite good absorber at the much longer far-infrared wavelengths of thermal radiation. That is, the radiation emitted by everything in your house sees a piece of glass not as transparent but as opaque. The long-wavelength emissivity of glass is about 0.9 – not a perfect absorber of thermal radiation but close. So a window made with untreated glass absorbs thermal radiation from the room and re-radiates it to the cold outdoors. A heat loss.
Since good absorbers are good emitters, we could prevent – or at least minimize – radiation losses by causing a window to be a bad absorber of far-infrared radiation. The easiest way to not absorb electromagnetic waves is to reflect them – to be a mirror. And indeed, shiny metal surfaces – good reflectors – have very low emissivity. Thermos bottles are made of polished metal so that heat is reflected rather than absorbed.
A window made from a piece of shiny metal would reflect long-wavelength thermal radiation back into the room rather than absorbing it, so there would be almost no heat loss to the outside by radiation. But there is the obvious problem that you can’t see through a metal window.
Except that, under the right circumstances, you can. An interesting property of metals is that extremely thin sheets of metal are transparent to visible and near-infrared light while remaining reflective to far-infrared thermal radiation. (This is a consequence of how electromagnetic waves of different wavelengths interact with the electrons in the metal.) By “extremely thin,” I mean less than 0.5 μm thickness, which is less than 1% of the diameter of a human hair.
If an extremely thin metal coating – less than 0.5 μm thickness – is applied to a piece of glass, the glass remains transparent to visible light (there’s a slight loss of transmission) but it becomes a reflecting mirror to far-infrared thermal radiation. Light gets through, but radiative heat losses are drastically reduced. High reflectivity means low emissivity, and this is low-e glass: glass with an extremely thin metal coating – usually silver – that you don’t see. It would make more sense to call it high-reflectivity glass.
Either way, the low emissivity and high reflectivity refer only to long-wavelength, far-infrared radiation. Where many web sites go wrong is confusing the near-infrared radiation of the sun with the far-infrared thermal radiation of the interior of your house. The graphs above showed that there’s no overlap between the near-infrared wavelengths of the solar spectrum and the far-infrared wavelengths of thermal radiation. Low-e glass is about blocking thermal radiation; it has nothing to do with solar energy.
Well, almost nothing. Any thin metal coating on the window will function as a low-e coating. But a clever design of the low-e coating can change the amount of near-infrared solar radiation that passes through the window. If you wear glasses, you probably have anti-reflection coatings on the lenses. The lenses of cameras and binoculars have anti-reflection coatings. An anti-reflection coating is a very thin coating on the lens – less than 0.5 μm thickness – that utilizes the constructive and destructive interference of light waves to control reflections. They are an optical version of noise-cancelling headphones, which work on the same principle.
The low-e coating on a window can be designed to do something similar. In fact, the coating on a window is usually not a single layer but a multi-layer sandwich of several very thin coatings – the silver that provides long-wavelength reflectivity along with very thin coatings of other transparent materials. A proper choice of materials and thicknesses can exploit the constructive and destructive interference of light waves to control the range of visible and near-infrared wavelengths that are allowed to pass through.
There are two basic classes of low-e coatings on windows for residential buildings, called hard coat and soft coat. A hard coat is applied to the glass during manufacturing and becomes integrated with the glass, so it more durable – i.e., harder. A soft coat is applied after manufacturing and, as you would expect, is softer and less durable. It’s placed on an inside surface of a double-pane window where it won’t be damaged.
A hard coat (also called a passive low-e coating) window has somewhat higher transmission of visible light – about 85%. But much more important is that a hard coat window passes a substantial portion of the sun’s infrared radiation – and recall that 50% of the sun’s energy is infrared.
You have to have this infrared energy in a passive-solar home! In other words, passive-solar design – except maybe in near-tropical climates where only minimal heat is needed – requires hard-coat low-e windows on the south side in order to gain enough solar energy. A soft-coat window would block half of the heat that you need.
Windows are characterized by what is known as the solar heat gain coefficient, abbreviated SHGC. It is the fraction of the sun’s energy that gets through. A passive-solar house needs windows with a high SHGC so that you maximize winter heating with the minimal amount of window area. A high-SHGC window would admit way too much solar energy in the summer if the sun is incident on it, but that’s not relevant because passive-solar design ensures that the south-facing windows are shaded in the summer. A typical soft-coat window, which blocks the sun’s infrared, has a SHGC of 0.37. That is, 37% of the sun’s energy gets through. In contrast, a typical hard-coat window has a SHGC of 0.69. That is, a hard-coat window admits almost twice as much solar energy as a soft-coat window of the same size! That’s why hard-coat windows are essential for successful passive-solar design.
Now the sad truth: Sales people in window stores – even top-quality windows – know absolutely none of this. Having read this far, you now know far more about the technical aspects of windows than they do. And that’s only the first sad truth. The second is that the default glass used in windows has a soft coat, exactly what you don’t want. That’s not surprising since most houses have given no thought to solar issues and have windows that are blasted by direct sun in the summer.
If you choose windows based on the architectural design, the quality of the workmanship, etc. and don’t say otherwise, you’re going to get soft-coat windows that will fail to provide the solar energy that you need for a passive-solar house. (And then, of course, some people who have unwittingly made this error will proclaim that passive-solar design doesn’t work.)
You, the consumer, have to specify that the factory install hard-coat glass on the south-facing windows. (Soft coat is OK for windows on the other sides and is perhaps marginally better because a soft coat has marginally lower emissivity and thus marginally better ability to minimize radiation losses.) The sales person will probably be clueless, but all window manufacturers can do this. The best approach is to find out who manufactures the glass used in their windows and get a catalog with technical specifications.
For example, we have Semco Windows (since acquired by Sierra Pacific Windows), but that’s simply the company that makes the window units. The glass in Semco Windows was made by Cardinal Glass Industries, one of just a handful of companies that manufacture window glass. Their catalog shows a wide variety of coatings that can be supplied – as well as windows with clear glass that have no coating. The coating names – such as LOE272 – are not helpful. By reading the specs, you can see that some are tinted or dark to decreases visible transmission. By looking at visible light transmission (slightly higher with hard coat), the solar heat gain coefficient, and the U-factor (more on this shortly), you can easily identify the coating numbers of the main hard-coat and soft-coat windows. By the time you’ve finished this page, you’ll be able to do this. Then insist that the order be written for windows with the appropriate coating number.
A low-e coating – and keep in mind that both hard-coat and soft-coat windows are low-e glass – significantly reduces heat loss through windows by radiation. The next task for energy-efficient windows is to reduce heat loss by conduction. Glass is a pretty decent heat conductor, so a single-pane glass window is basically a highway for heat to be conducted from the inside of your house to the outside.
The insulation value of a wall or ceiling is specified by its R-value. (And no, you really don’t want to know what the truly arcane technical definition of R is.) A higher R-value means more insulation, less energy loss by heat conduction. The Construction page noted that a well-insulated wall with 2 x 4 studs is typically R-10; a wall with 2 x 6 studs is more like R-15. The insulation itself has a higher R-value, but the net R-value of the entire wall unit is lowered because the studs conduct heat. The fact that a “heat leak” like the studs lowers the net R-value will be relevant to windows.
For reasons lost in the mists of time, the insulating value of window glass is specific not by its R-value but by its U-value. For practical purposes, the U-value is simply the reciprocal or inverse of the R-value. That R-10 wall has a U-value of 0.1. (To be fair, U-values consider heat loss by both conduction and radiation whereas R-values are concerned with conduction only, so one is not exactly the reciprocal of the other if radiation losses are significant. But radiation losses are not significant for low-e windows, so the reciprocal relationship works just fine.)
A simple single-pane window has an R-value of about 1, so it’s U-value is also about 1. A calculation based on the thermal conductivity of glass would given an even lower R-value, but a simple window benefits from the fact that a thin layer of “dead air” is trapped next to the glass. Air is a good insulator, and the R-value of a single-pane window is due more to the trapped air than to the glass itself.
The fact that air is a good insulator is the basis of the most common window: a double-pane window, also called double glazing. Two sheets of glass, one with a low-e coating, are held apart by spacers, trapping a layer of air between them. The R-value of the window unit has little to do with the glass and is due almost entirely to the insulation of the trapped air. But there’s a limit. The insulation of air increase with the thickness of the air layer up to about half an inch. Thicker layers of air begin to develop convective motions – less dense warmer air rises, more dense cooler air falls – and heat transfer due to convection offsets any gain from increased thickness. So the spacing between the two panes is typically right at half an inch.
But one can do better than air. Air conducts heat because of the diffusion of air molecules. A gas with less diffusion would be a better insulator. Enter argon-filled windows. Argon is one of the “inert gases,” like helium, on the right-hand edge of the periodic table. Argon molecules are heavier than air molecules, so diffusion is slower and argon is a better insulator. All window manufacturers now offer argon-filled double-pane windows.
And then there are triple-pane windows. They should be a serious consideration for a passive-solar house in colder climates.
The solar heat gain coefficient (SHGC) is where things start to get interesting. Whether two panes or three, a window with a hard coat has a much larger SHGC than a window with a soft coat. A double-pane window with a soft coat has a SHGC of 0.37, meaning that it blocks out 63% of the sun’s energy – decidedly what you don’t want if you’re trying to heat your house with the sun. A double-pane window with bare glass, no coatings, does even better than a hard coat – but it’s downfall is in the next column.
The manufacturer gives the insulating value of the window in terms of its U-factor, but I’ve converted those to R-factors that will seem more familiar and can be compared to the insulating value of the wall. There’s an entry for air-filled windows and one for argon filled windows. (Except that argon is not an option for uncoated windows from Cardinal.) A single uncoated pane has the R-value of 1 noted above. The insulation value goes up with a double pane window due to the trapped gas, and goes even higher with a triple pane window.
A hard coat and a soft coat are substantially better insulators than an uncoated window. That’s because both coatings are low-e coatings that reflect far-infrared thermal radiation and reduce the radiation losses. Here you can see that a low-e coating boosts the R-value by about 50% over that of bare glass. It’s true that an uncoated double-pane window would let in even more solar energy than a hard-coat window, but that small increase is solar energy is more than negated by much larger heat losses through the uncoated window. A soft-coat window has a very slightly larger R-value than a hard-coat window, due to its low-e coating having a very slightly smaller emissivity, but – for energy-gathering windows – the difference doesn’t make up for the huge drop in SHGC of a soft coat. Soft-coat windows are appropriate on the north side where their slightly larger R-value might help a bit.
Finally, you can see that an argon fill gives a pretty substantial increase in the R-value compared to an air fill.
Note: These are the R-values of the glass. The R-value of the entire window unit also has to take account of heat conduction through the framing and other materials, so the R-value of the entire unit is usually somewhat smaller than that of the glass alone. But that’s really getting into the weeds, so I’m going to stay focused on the glass. But be aware that inexpensive windows often use lots of heat-conducting metal parts that can substantially degrade the window’s thermal performance. It’s especially important to avoid metal spacers separating the panes. In general, higher quality (i.e., more expensive) windows use higher quality materials and have better thermal performance
A primary observation is that the R-value of even double-pane, argon-filled windows is substantially less than the R-value of the wall that they’re placed in. Hence the earlier comment that windows are typically the main source of heat loss from a well insulated house. They’re a heat leak in a well insulated well.
As an example, one of our upstairs rooms has a single large window unit that occupies 20% of the south-facing wall. The wall, with 2 x 6 studs and blown-in insulation, is about R-15. The double-pane, argon-filled, hard-coat window is, according to the table, about R-3.8. You can calculate the net R-value of the entire wall as follows:
Rnet = [fwall / Rwall + fwindow / Rwindow]–1
Where fwall is the fraction consisting of solid wall (0.80 in our case) and fwindow is the fraction consisting of window (0.20). If you do this calculation, don’t forget to invert (the –1 power) after adding the fractions!
A calculation finds that the window reduces the net R-value of this wall from R-15 to R-9.4. That’s a huge loss. Calculations like these clearly show that you don’t want large windows on the other three sides of the house. Large windows on the south side are acceptable only because of the solar energy they can harvest during the day.
If we had a simple, single-pane window, like windows used to be most houses, our R-15 wall would be reduced to a net R-value of 3.9. Might as well just leave the window open at that point! On the other hand, going to a triple-pane, argon-filled, hard coat window would increase the net R-value to R-11.5, a 20% increase in the insulation value over double pane. I’ve not priced triple-pane windows compared to double-pane, but it’s an interesting proposition that in colder climates the extra cost of triple-pane windows might be more than made up for by needing less supplemental heat at nights and on cloudy days.
Cardinal Industries is now offering a new coating, of indium tin oxide, that is applied to one of the other surfaces (in addition to the hard coat) to reduce the emissivity and the radiation losses even further. This must not have been an option when we ordered our windows, because we certainly would have selected it. With this additional coating, the R-value of a double-pane, argon-filled, hard-coat window increases from R-3.8 to R-4.8. That would increase our wall’s net R-value from R-9.4 to R-10.5. A triple-pane, argon-filled, hard-coat window goes from R-5.9 to R-6.7. Including this second coating seems a no-brainer unless it’s outrageously expensive.
But, you’re no doubt thinking, you can deal with the heat loss through windows at night by closing the drapes. Right? Sadly, no. Room air circulates – convection. The slightly cooler air between the drapes and the window is denser, so it sinks and flows out the bottom while pulling in warmer room air around the top and sides of the drapes. Despite what window-covering ads may claim, numerous studies have found that window coverings have essentially no insulating value because of this convection. You might as well leave the drapes open at night.
With one minor exception: cellular shades. These are the honeycombed paper shades that are pulled down. The claim is that the air inside the cells provides the same type of insulation as the air trapped between the panes of a double-pane window. Sellers of cellular shades make some pretty astounding claims about a shade’s ability to provide insulation. Many web sites claim that a double-cell, light-filtering shade, probably the most common type, has an insulation value of R-2.8. This would add to the window’s insulation, raising my R-3.8 windows to R-6.6. A few web sites give R-4.5 as the insulation value of a double-cell light-filtering shade. These are store-front websites, selling window coverings, so not surprisingly none of them provide any information as to where these numbers come from.
Are these numbers accurate? Sadly, no. If you’re really into engineering analysis, check out this web site where a residential-energy engineer measured the R-values. He looked at four options.
Standard double-cell, light-filtering shades.
Double-cell, light-filtering shades that fit into side tracks to seal the edges.
Double-cell, blackout shades.
Double-cell, blackout shades with side tracks.
And here’s what he found:
Double-cell, light-filtering shades R-0.6
Double-cell, light-filtering shades with side tracks R-1.2
Double-cell, blackout shades R-0.9
Double-cell, light-filtering shades with side tracks R-2.0
This web site references another technical study that found similar results. Just as with ordinary drapes, convection around the edges and even convection within the cells – they’re not sealed – seriously limits the purported “air insulation” of the shades. The actual R-values are nowhere near the advertised values. Side tracks seal the edges and reduce convection, but they’re apparently somewhat of a pain and still don’t provide much insulation. Blackout shades do somewhat better, although still not great, but they are reportedly cumbersome to raise and lower due to their weight, not to mention being an aesthetic issue for many people. A standard double-cell, light-filtering shade adds essentially no additional insulation to my R-3.8 windows.
And I can prove it. We installed double-cell, light-filtering shades upstairs. After noticing that the early morning winter temperatures upstairs seemed about the same whether I lowered the shades overnight or not, I started making measurements. Over a period of a couple of winter months I measured the overnight temperature drop in two upstairs rooms, in one of which I pulled the shades down tight and the other I left the shades up. And I switched back and forth which room was which. There was a lot of day-to-day variation, but on average lowering the shades made only about half a degree difference in the overnight temperature drop – very little difference. I then made a mathematical model of heat loss from the rooms – through the windows, walls, and ceiling. I can’t claim great accuracy for the model because I had to make a number of simplifying assumptions, but I concluded that the insulating value of the shades was no higher than R-1. That’s consistent with the results of the people doing serious measurements. And more relevant to homeowners, I demonstrated that you can barely discern a difference – about half a degree – between closing the shades overnight versus leaving them open.
Bottom line: Unless you go to extremes, like blackout shades with side seals, window coverings have no meaningful insulation value. They do not prevent your windows from being your primary heat loss in winter. That argues all the more strongly for installing the best windows you can afford, knowing that money spent up front on windows with good R-values will largely be recovered over the years in terms of less need for supplemental heat.
© 2021 Randy Knight