Low-emissivity coatings and argon fills

Emissivity, and the radiation half of heat loss

Heat crosses a sealed unit three ways: conduction through the solids, convection in the gas, and radiation across the cavity. In an uncoated air filled unit, radiation is roughly two thirds of the total. The warm inner pane radiates infrared at the cold outer one, and clear glass is close to a perfect emitter.

Emissivity is the fraction of that radiation a surface actually gives off, on a scale from zero to one. Ordinary float glass sits around 0.84. A hard coat low emissivity surface is around 0.15. A modern soft coat is between 0.02 and 0.05. At 0.03 the surface is emitting about one twenty eighth of the infrared it otherwise would, and the radiation route through the cavity is very nearly closed. Nothing else in the unit changes. The glass is the same glass.

Counting the surfaces, one to four

Sealed units are described by surface number, counted from outside in. Surface 1 is the weather face. Surface 2 is the inside of the outer pane. Surface 3 is the cavity side of the inner pane. Surface 4 is the room face, the one you clean.

Convention matters here because a coating behaves differently depending on where it sits. A thermal low emissivity coating in a British climate goes on surface 3, where it faces the cavity and reflects room heat back inwards. A solar control coating goes on surface 2, where it intercepts sunlight before that energy has entered the cavity at all. A unit with both, which is common on a large rooflight, carries a coating on surface 2 and another on surface 3. Triple glazed units number one to six and usually coat surfaces 3 and 5, a point covered in more detail on triple glazing in a rooflight.

What one coating takes off the U-value

The numbers are worth holding in your head, because they show where the money goes.

A 4-16-4 unit with clear glass and air sits at a centre pane Ug of about 2.7. Change nothing but add a soft coat on surface 3 and it drops to about 1.6. That is a forty per cent improvement from a layer you cannot see, and it is the single largest step available in glazing. Fill the same coated unit with argon and it goes to about 1.1. The coating did more than twice as much work as the gas.

This ordering has a practical consequence. If a supplier is offering a choice and the budget is finite, the coating comes first every time. An argon filled uncoated unit is a poor buy.

Single, double and triple silver stacks

Soft coats are built in layers, and the count of silver layers sets what the coating can do.

A single silver stack is a thermal coating. It reflects long wave infrared, gives a low emissivity and passes most of the solar spectrum, so g-values stay high at around 0.60 and the room still gets winter gain.

Double silver adds solar control. Short wave near infrared is now reflected too, so the g-value falls to roughly 0.35 to 0.42 while light transmittance stays around 0.65. This is the workhorse for a flat rooflight over a kitchen.

Triple silver pushes further, g-values around 0.25 to 0.30, and holds light transmittance near 0.60. It is the selective coating used where a large area of glass faces the sky and the room cannot afford the heat. What that heat load actually amounts to is worked through on solar gain, overheating and shading.

Light transmittance, and the tint a stack leaves behind

Every silver layer costs some daylight and shifts the cast of the light slightly. Modern coatings are far better at this than the bronze and grey tinted glass of the nineteen eighties, but the effect is not zero.

Expect a faint blue or green cast on a double silver stack, visible mainly when you look along the glass at a shallow angle or compare it against an uncoated pane in the same room. Inside, under a rooflight, it reads as neutral. The place it shows is where two units of different specification sit in one ceiling, which is a real risk when a lantern goes into an extension and a smaller flush unit is added later. Order both from the same coating family if they will be seen together.

Why argon works, and by how much

Argon is monatomic, heavier than air and less thermally conductive. Air conducts at about 0.024 watts per metre kelvin. Argon manages 0.016, a third less. It is also more viscous in the cavity, which slows the convection loop that carries heat from the warm pane to the cold one.

In a coated 16mm cavity argon is worth roughly 0.2 to 0.3 on the Ug. It is cheap, it is a by-product of industrial oxygen production, and there is no reason to leave it out. Nitrogen fills exist but are close to air in performance and appear mainly where argon is unavailable.

The heavier gases, and the cavities they suit

Krypton conducts at about 0.0094 and xenon at 0.0055, so both beat argon on paper. Their real advantage is different: because they are denser, convection starts at a narrower gap, so their optimum cavity is much thinner.

Argon wants 15 to 16mm. Krypton peaks at about 10 to 12mm and xenon at around 8mm. That makes krypton the answer for slim units where the frame rebate will not take a wide cavity, such as a conservation profile in a Thaxted or Coggeshall roof slope where the whole unit has to sit within a slim cast frame. It is expensive, several times argon, and the cost only makes sense when the cavity is being constrained by something else.

Cavity convection when the warm pane sits underneath

This is the point where rooflights part company with wall windows, and it is rarely explained.

In a vertical unit the warm and cold panes are side by side and the gas circulates in a slow vertical loop. In a rooflight in winter the warm pane is underneath the cold one. Warm gas at the bottom sits below cold gas at the top, which is an unstable arrangement, and above a threshold it breaks into rolling convection cells that move heat far more efficiently than the vertical case does.

The consequence is that a unit installed near horizontal loses more heat than the same unit standing upright, typically 0.1 to 0.3 on the Ug, and the wider the cavity the worse it gets. A 20mm argon cavity that would be fine in a wall is a poor choice lying flat. Sixteen millimetres is close to optimal for a rooflight, and where the frame allows only a narrow gap, krypton recovers what the geometry took. This is also why declared figures should state the test inclination, a point set out on U-values and glazing specification.

Fill rate at manufacture, and the ninety per cent figure

A cavity is never one hundred per cent argon. Units are filled by displacement, with gas fed in low and air pushed out high, and the process stops when a sensor reads the target concentration.

The declared initial fill is normally 90 per cent, and thermal calculations to BS EN 673 are run at that figure rather than at pure argon. Good manufacturers achieve 92 to 95 per cent. The difference between 90 and 95 is small, around 0.02 on the Ug. The difference between 90 and a unit that was never properly filled is the whole 0.25.

Leakage, and the one per cent a year assumption

Argon molecules are small and the primary seal is not perfect. The standards work on a permitted loss of no more than one per cent of the fill per year, which means a unit at 90 per cent on day one should still be above 80 per cent after twenty years, and the thermal penalty over that period is minor.

Loss accelerates when the seal is stressed. A near horizontal unit carries a standing load on its edge seal that a vertical one does not, thermal cycling on an exposed flat roof over the Blackwater estuary is severe, and units bedded down hard onto an uneven kerb are working against a permanent twist. This is one of several reasons the perimeter build matters as much as the glass, and it is why gas loss and eventual misting tend to arrive together in units that have reached the end of their life.

Coating and gas together, and why the total is not the sum

Add the two improvements separately and you overcount. The coating removes most of the radiation path, and the gas slows conduction and convection. Once the coating has closed the radiative route, the remaining heat flow is dominated by the gas, so the gas has more leverage in a coated unit than in an uncoated one, and less absolute room to improve because the total is already small.

Worked through, an uncoated air unit at 2.7 gains about 0.15 from argon alone. The same argon in a coated unit is worth 0.4 or more. The pair are specified together for that reason, and no serious manufacturer offers one without the other.

Edge deletion, and why a soft coat is ground off the perimeter

A sputtered coating is metallic, and the sealant that bonds the unit together will not adhere reliably to it. Left in place under the seal it also corrodes, creeping inwards from the edge as a dull halo.

So the coating is ground away in a band, usually eight to twelve millimetres wide, around the perimeter of the pane before the spacer goes on. The primary and secondary seals then bond to clean glass. Edge deletion is standard practice and is one of the details that separates a proper sealed unit factory from a cheap one. It also means the coated area stops short of the sightline, which is a small part of why the edge of a unit performs worse than the middle, alongside the spacer itself covered on warm-edge spacers.

Proving the fill without opening the unit

Argon concentration can be read non-destructively. A spark emission tester fires a discharge across the cavity and reads the emission spectrum through the glass, returning a percentage in a few seconds. It is used in factory sampling and by testing houses, and it will read an installed unit in a roof provided access is possible.

For a homeowner the practical route is simpler. Ask for the sealed unit manufacturer, the glass specification string and the declared Ug with the test standard against it. A unit whose paperwork names the coating product, states 90 per cent argon and quotes BS EN 673 has been made by someone who measures. A quote that says only “argon filled low-E” has told you nothing that can be checked.

Cleaning a coated pane without harming it

Surfaces 2 and 3 are sealed inside the unit and need nothing. The coating you can reach is the self-cleaning layer on surface 1, if the unit was ordered with one, and it is pyrolytic and tough.

What damages it is abrasive: scouring pads, blade scrapers used dry, builders’ cleaning acids and cement splashes left to dry on the glass over a working day. Warm water, a soft cloth and a squeegee handle everything else. On a rooflight over a Chelmsford Victorian terrace where the pane is reached off a ladder once a year, the honest advice is water and patience rather than chemistry.

Applied films against a factory coating

Retrofit films are polyester sheets with a metallised layer, applied to surface 4, the room face. They can cut solar gain usefully and they are the only option on a unit that is staying put.

They are not equivalent to factory glazing. They sit on the warm face rather than the cavity face, so they do far less for the U-value. They absorb energy in the pane, which raises thermal stress and can matter on a toughened unit and matters a great deal on an annealed one. They scratch, and they discolour over ten to fifteen years. Where a roof is open anyway and the unit is coming out, coated glass in a new unit is the better spend, which is the usual conclusion reached on the energy efficient rooflights hub. External shading is the other lever, and the trade-off between shading outside the glass and blinds inside it has a page of its own at external shading against internal blinds.