Infrared Myths

An Eccentric Anomaly: Ed Davies's Blog

I see a lot of confusion about infrared when it comes to interactions with buildings, astronomy and so on. Here's a bit of a brain dump to refer to in such cases which I hope will throw some light, of various wavelengths, on the matter.

Of course, it's quite possible I've got the wrong end of the ray when it comes to some of this stuff - if you can see how this page could be improved please get in touch.

At the root there are two common misunderstandings:

The word “light” is used by physicists to refer to all electromagnetic radiation with shorter wavelengths than radio: infrared, visible light, ultraviolet, gamma waves, X-rays and so on. That's how I'll use it here.

Light can be characterized by its frequency or its wavelength. The higher the frequency the shorter the wavelength, and vice-versa. Despite the frequency being the more conserved property (eg, the wavelength changes when light enters a medium, like glass, where its speed is lower) it's usual to refer to the wavelength of visible and infrared light.

For infrared, wavelengths are normally given in micrometres (microns, µm, millionths of a metre) whereas for visible light nanometres (nm, billionths of a metre) are often used. For consistency I'm going to use µm throughout here but, for example, the wavelength of green light used as a reference for the sensitivity of the eye is often given as 555 nm which is 0.555 µm.

There are four bands of electromagnetic radiation which are typically of interest. In decreasing order of wavelength (and therefore increasing frequency and increasing energy per photon):

Thermal IR 12…10 µm “Heat” radiation given off by things at around “room” temperature: above freezing but below boiling.
Near IR 2.5…0.8 µm Short wavelength infrared, too long for the human eye to see, (usually, perhaps?).
Visible 0.8…0.4 µm Light the human eye happens to be sensitive to.
Ultraviolet 0.4…0.1 µm Light too short for the human eye to see.

The bounds of these bands are given in very round numbers and are subject to much discussion and variation of definition. E.g., somebody could come up with a convincing argument that near IR should be considered as 3 µm down to 0.7 µm or whatever as there are a lot of different classification schemes depending on purpose.

The wavelength range of IR is very wide - over 1000 times. The shortest, bounding with visible, is about 750 nm (shown as 0.8 µm above) so less than 1 µm whereas the longest, bounding with the short end of the microwave radio spectrum, is 1 mm (1000 µm) in wavelength. That end of the IR spectrum, though, is only of interest when dealing with very cold things like the cosmic background radiation (at around 2 to 3 K, ie, 2 or 3 degrees Celsius above absolute zero).

When a photon of light of any of these wavelengths (IR, visible or UV) hits a surface it can be reflected, transmitted through the material or absorbed. If it's absorbed then it might have special effects like pumping an electron to flow round a circuit connected to a photovoltaic panel but typically it will just cause the material to be warmed. The shorter the wavelength, so the higher the energy, the more warming an individual photon will cause.

When a material is warmer than absolute zero it will emit electromagnetic radiation spread across a band of the spectrum. The warmer an object is the higher the energy of photons it tends to emit and therefore the shorter the wavelength they are on average. The average wavelength is inversely proportional to the temperature.

The peak wavelength of light emitted by a body is roughly 3 mm (millimetres), or 3000 µm, divided by its absolute temperature in kelvins. The freezing point of water is 273.15 K so buildings tend to be around 300 K so emit radiation at around 10 µm - hence the term thermal infrared for the band of the spectrum around this wavelength.

Thermal imaging cameras and IR thermometers detect light at these sort of wavelengths. Regrettably, I don't have a thermal camera but I do have a few IR thermometers. My favourite for robustness and apparent accuracy, if not for compactness, is this one from Lidl for less than £20:

This effect is modified by the emissivity of the material which can vary with wavelength. An ideal emitter would also be an ideal absorber so would appear completely black and is referred to as a black body. Real materials typically have emissivities around 0.95 (95%) for thermal infrared though they can be much lower and widely different from their visible light emissivities.

The surface of the Sun is at around 5500 K so its radiation peaks at about 0.5 µm - by happy coincidence in the middle of the visible spectrum. Sunlight as it reaches the Earth's surface consists of about:

Proportionally, there's very little thermal IR in solar radiation when it leaves the Sun and what there is tends to be absorbed by the atmosphere on the way in. Similarly, a lot of the UV in the sunlight which reaches the top of the atmosphere gets absorbed by the ozone in the stratosphere so doesn't reach the surface.

Here's a good diagram from Wikimedia:

Note that this spectrum only includes UV, visible and near IR. To scale, thermal IR would be about 5 or 6 chart widths to the right by which point the energy would have tailed off almost completely.

One of the most common forms of confusion is conflating the thermal-IR and near-IR bands when they actually behave quite differently. All they really have in common is that their wavelengths are too long for them to be detected by the human eye.

Near IR acts pretty much the same way as visible light, apart from not being visible to humans of course. It is visible to the CMOS and CCD sensors in digital cameras which typically have an IR-cut filter to stop it causing weird colour effects. Still, some gets through so that bright near-IR sources, such as TV remote controls, are visible to such cameras. Go on, try it - this is the pointy end of a TV remote taken with my phone camera:

This is how “black light” security cameras work - they're typically monochrome CMOS or CCD sensors with no IR-cut filter and a surrounding ring of bright IR LEDs to provide illumination. This is quite different from the “FLIR”-type thermal-imaging cameras used, for example, on police helicopters at night.

Thermal IR, on the other hand, is not detected by normal camera sensors even with the IR-cut filter removed. Special sensors are needed which makes thermal imaging cameras significantly more expensive. About once a year some bright spark on some green building or alternative energy forum or other suggests that thermal imaging cameras can be made by removing the IR-cut filters from web cams or whatever. Not surprisingly, nobody has ever reported success with this sort of scheme.

Another important area of difference between near and thermal IR is transmission by glass. Near IR gets through pretty much as well as visible light (though I think low-iron glass transmits near IR relatively better than ordinary glass does - not sure though) whereas most glass is effectively opaque to thermal IR.

(This is another reason that thermal imaging cameras are expensive; they need lenses made of materials transparent to thermal IR: typically germanium or quartz are used. IR thermometers I've looked inside seem to use plastic Fresnel lenses which I imagine are fine for focusing on a single sensor but not so great for giving a good image.)

An easy way to see that thermal IR is not transmitted well by normal glass is to point an IR thermometer at the inside of a window in a heated room on a cold day. If it's single glazed then the glass will have a temperature about half way between indoors and outdoors and this will be shown by the thermometer, rather than the lower temperature of the outside background.

Conversely, some materials can be surprisingly transparent to thermal IR. For example, I've found that an IR thermometer can “see” a radiator quite clearly through a white plastic carrier bag which is pretty much opaque to visible light. It wasn't just the effect of the radiator warming the bag, the thermometer could see the difference between the radiator and the cooler wall beside it.

Consider photons of solar radiation which have made it all the way from the Sun and through the atmosphere to impinge on a window. 10% or so will simply be reflected by the window. If they hit normal (ie, perpendicular) to the surface of the window then fewer will be; if it's at a very shallow angle then quite a lot more. (See my calculator page.)

A few percent of the photons will be absorbed by the glass as they pass through. As noted above, most near IR and visible light will get through though UV is, I think, absorbed a lot more. What does get absorbed, though, will directly warm the window.

Windows can be coated with materials to reduce the transmission of solar radiation. This is not generally a pressing concern in the north of Scotland so I haven't looked into the matter much but, as I understand it, the idea is that the near IR is largely cut out. This reduces the heating of the room significantly. Note, though, that reducing the transmission of the same amount of visible light would have similar effects on the heating, but would adversely affect the windowiness of the window.

Most of the solar radiation photons get through the glass and are absorbed in the room, perhaps after bouncing around a few times first, where they release their energy as heat.

Heat gets back out of the room through the window via the normal convection and conduction mechanisms. Also, objects in the room emit thermal IR which, when it impinges on the window glass, is absorbed. Some is, of course, transferred back into the room by convection, conduction or thermal IR radiation. The rest is conducted through the glass and lost to the outside via, you guessed it, convection, conduction and thermal radiation.

Depending on the window geometry it will also be exposed to a view of some surrounding objects: ground, trees, other buildings, etc. These will be giving off some thermal IR too so will help to replace some of the heat lost from the window. A clear sky, particularly at night, won't so will cool the window quite a bit more effectively.

If glass doesn't transmit thermal IR well what's the point of low-emissivity coatings in double and triple glazed windows? Many descriptions are rather confusing but here's what I think is going on. Heat gets transferred from the inner to outer layers of glass both by conduction and convection in the gas filling (air, argon or whatever) and by thermal-infrared radiation.

If the low-e coating is on an outward-facing glass surface then its effect is obvious, it reduces the amount of thermal IR transfer from the inner glass pane to the next outer one raising the temperature of the inner pane so decreasing the conductive flow out of the room and increasing the thermal IR emitted from the window back to where it's useful.

If, as is usually the case I think, the low-e coating is on the inside face of an outer pane then its effect will be to cause thermal IR photons from the inner glass to be reflected back. As glass is pretty much opaque to thermal IR it can't be transmitted so the only possibilities are absorption or reflection. The low-e coating reduces the absorption so increases the reflection.

In general, all this stuff is pretty straightforward but it would help a lot if people were a bit clearer about what sort of infrared they were talking about.