Sunspire SkylightsChelmsford
01245 527275 Get a quote
Guide · Energy Efficient Rooflights

Warm-edge spacers and where condensation starts

17 sections 11 minute read

Condensation on a rooflight almost never starts in the middle of the glass. It starts as a narrow band around the perimeter, usually along the bottom edge, and it works inwards from there. The reason is a strip of material about fifteen millimetres wide that nobody sees once the unit is fitted: the spacer bar that holds the two panes apart.

That strip is a bridge from the outer pane to the inner one. Whatever it is made of decides how cold the inside face of the glass gets near its edge, and the edge is where the dew point is crossed first. This page covers what the edge seal is built from, what the different spacer materials do to the numbers, and why on a rooflight the edge matters more than it does on a window in a wall.

The perimeter is a different thermal component from the middle

Thermal calculation treats a sealed unit as two zones. The centre pane zone is glass, cavity, glass and nothing else, and it is what the quoted Ug describes. The edge zone is a band around the perimeter where the spacer, the seals and the frame all influence the heat flow, and it performs considerably worse.

The convention in calculation is that the edge zone extends about 63.5 millimetres in from the sightline. Everything inside that band is affected by the spacer to some degree. On a large lantern pane that band is a small fraction of the area. On a 550 by 780 roof window sized to fit between trusses on a Beaulieu Park estate roof, the edge zone is most of the glass. Small units are dominated by their edges, which is why they never achieve the whole product figures that large ones do.

The sky dome Vertical window Rooflight
A rooflight faces the whole dome of sky and is not shaded by the fence or the neighbouring house, which is why it delivers far more than a vertical window of the same area.

What is actually in the edge of a sealed unit

Cut through the perimeter of a double glazed unit and there are five things in a space about twenty millimetres deep.

The spacer bar sits between the panes, set back from the glass edge. Inside the spacer is desiccant. Between the spacer and each pane is the primary seal, a thin bead of polyisobutylene which is the actual gas and moisture barrier. Outside the spacer, filling the channel to the edge of the glass, is the secondary seal, a structural adhesive of polysulphide, polyurethane or silicone that holds the sandwich together. And on the room side of all of it is the frame rebate and its gasket, which decides how much of this assembly is protected and how much is exposed.

Why aluminium was the default for so long

Aluminium spacer is easy to roll, easy to bend, dimensionally stable, cheap and holds desiccant well. For thirty years it was in effectively every sealed unit made.

Its problem is thermal conductivity of around 160 watts per metre kelvin, which is roughly a hundred and sixty times that of glass and four thousand times that of the argon in the cavity. Every millimetre of aluminium is a conduit from the cold outer pane to the warm inner one. In an era when the glass itself had a U-value near 3, the spacer was a rounding error. Now that the coating and the gas have pushed the centre pane below 1.1, the spacer is a large fraction of what remains, and the industry moved.

Psi values, and what the numbers look like

The edge effect is quantified as a linear thermal transmittance, Ψg, in watts per metre kelvin. It is multiplied by the total length of the glass perimeter and added into the whole product calculation.

Spacer type Typical Ψg (W/mK) Edge glass temperature at 20°C in, 0°C out
Aluminium box spacer 0.07 to 0.09 Around 9 to 10 °C
Thin walled stainless steel 0.045 to 0.055 Around 11 to 12 °C
Composite or glass fibre 0.035 to 0.045 Around 12 to 13 °C
Thermoplastic or silicone foam 0.030 to 0.040 Around 12.5 to 13.5 °C

Three degrees does not sound like much until you set it against a dew point. A room at 20 degrees and 60 per cent relative humidity has a dew point of 12 degrees. An aluminium edge at 9.5 degrees is wet. A composite edge at 12.5 degrees is dry. The same room, the same night, the same glass, and a different outcome, decided by fifteen millimetres of material.

The difference between a wet edge and a dry one is often two degrees, and two degrees is exactly what a warm edge spacer delivers.

The definition a spacer has to meet

“Warm edge” is a marketing phrase with a technical definition behind it, and the definition is worth knowing because the phrase gets used loosely.

Annex E of BS EN ISO 10077-1 sets the criterion. Summing, for every component of the spacer, the wall thickness multiplied by its thermal conductivity, the total must not exceed 0.007 watts per kelvin. A steel spacer with walls thin enough passes. A conventional aluminium box section does not, by a wide margin. When a supplier says a unit has warm edge, the fair follow up is which product, because the answer separates a genuine thermoplastic or composite spacer from a slightly thinner metal one.

Where the first water appears, and why

On a roof window set in a pitched slope, the first condensation appears along the lowest edge of the pane. Two effects put it there rather than anywhere else.

The lower edge of a roof window frame is the part most deeply buried in the roof construction and often the part with the least insulation dressed up to it, so it runs colder than the head. And any water that forms anywhere on the pane runs down to that edge and collects. On a flat rooflight the whole perimeter is a low point relative to the middle of the pane, which is why flush units show a continuous frame of dampness rather than a line. The frame of a well made flat unit has a drainage channel and weep holes to deal with that, and blocking those with sealant during installation guarantees a problem later.

Corners, and the joints that leak first

A spacer frame is either four straight lengths joined by corner keys, or a single length bent at three corners and joined once.

Every joint is a discontinuity in the primary seal and a potential path for argon out and water vapour in. Bent corners have one joint rather than four, which is why they are the better construction and why good manufacturers use them. Thermoplastic spacers are extruded as a continuous bead directly onto the glass by robot, with no joints at all, which is the best answer available. On a unit going into a roof, where the seal is being pressure cycled by the sun every single day, the corner construction is worth asking about.

Stainless steel and the thin walled metals

The first generation of warm edge products kept the metal and made it thinner. Stainless steel conducts at about 17 watts per metre kelvin against aluminium’s 160, and rolling it to a wall thickness of two tenths of a millimetre reduces the cross section further.

These spacers are rigid, easy for a fabricator to handle on existing equipment, hold desiccant in the conventional way and are dimensionally reliable at large sizes. They are a genuine improvement on aluminium and they are the most common warm edge product in the United Kingdom. They are not the best performing, and on a rooflight where the edge is doing more work than usual there is a case for going further.

Composite, foam and thermoplastic spacers

The higher performing products remove the metal altogether or reduce it to a foil.

Composite spacers use a rigid plastic or glass fibre body with a thin metal foil bonded to the outer face, the foil being there as a gas barrier rather than as structure. Silicone foam spacers are a compressible closed cell foam with an integral desiccant and an applied vapour barrier, which accommodates movement well. Thermoplastic spacer, usually written TPS, is a butyl compound with desiccant mixed through it, extruded hot onto the glass as one continuous bead with no corners and no separate primary seal. TPS gives the best psi figures and the best argon retention, and it is what to ask for on a large or highly specified unit.

The desiccant, and how long it lasts

The cavity of a sealed unit is filled with gas that contains a small amount of water vapour at the moment of manufacture, and the desiccant absorbs it so the unit is optically clear from cold.

The material is a 3A molecular sieve, a synthetic zeolite with a pore size chosen to admit water molecules and exclude argon and nitrogen. Its capacity is finite and it is sized to handle the initial moisture plus whatever creeps through the seals over the design life. Once it is saturated, any further ingress condenses in the cavity, which is the misting that signals a unit at the end of its life. Nothing can be done to the desiccant afterwards, and the unit wants replacing rather than reviving.

Primary and secondary seals, and which one matters when

The two seals do different jobs and people frequently attribute one’s function to the other.

The primary seal of polyisobutylene is the barrier. It is what keeps argon in and water vapour out, and its performance depends on the width and continuity of the bead. The secondary seal is structural: it bonds the panes to the spacer so the unit behaves as one component under wind load and thermal movement. Polysulphide is the traditional secondary seal and it retains gas well but degrades under ultraviolet. Silicone resists ultraviolet indefinitely and is required where the seal is exposed, but it is more permeable to gas. Which one is correct depends entirely on how much of the edge the frame covers.

Daylight from a single large rooflight over a kitchen extension
Daylight from a single large rooflight over a kitchen extension

Edge cover, and why it decides the life of the unit

Edge cover is the depth by which the frame overlaps the edge seal. It is typically 12 to 15 millimetres, and on a rooflight it is a life expectancy issue rather than a detail.

A unit lying in a roof plane receives ultraviolet from the whole sky for the whole day, at an intensity a vertical window never sees. Where the frame does not shade the secondary seal, a polysulphide seal will chalk, craze and eventually lose adhesion, and the unit fails long before its time. This is the reason structurally bonded flush glazed rooflights, where the glass is bonded to the frame and the edge is visible from outside, are made with silicone secondary seals and a black ceramic border printed onto the glass. If a unit with an exposed edge is quoted without either, that is the question to ask.

What the standards test and declare

BS EN 1279 is the sealed unit standard and it runs to several parts, each testing something different.

Part 2 is the moisture penetration test, which cycles units through humidity and temperature and measures the moisture uptake of the desiccant. Part 3 covers the gas leakage rate and the initial gas concentration, with a permitted loss rate of not more than one per cent a year. Part 4 covers test methods for the edge seal itself, and parts 5 and 6 cover conformity and factory production control. A unit declared to BS EN 1279 with an argon concentration of 90 per cent and a compliant leakage rate is a unit that will still be performing in twenty years. That declaration should be in the paperwork rather than assumed.

Reading the spacer on a unit already in the roof

It is often possible to tell what is in an existing rooflight without taking anything apart, which is useful at survey.

Mill finish aluminium is bright, and it is usually printed along its length with the maker’s name, the build-up in millimetres and a manufacturing date code. Stainless warm edge spacers are duller and often supplied in black or grey to match the sightline. Composite and thermoplastic spacers read as a matt black or dark grey band with no visible seam at the corners. Where the band is bright metal and the date code reads before about 2010, the unit is on an aluminium spacer and any condensation complaint about its perimeter has an obvious explanation.

What the spacer changes in the whole product number

The edge effect enters the whole product calculation as Ψg multiplied by the glass perimeter length, divided by the total area, so its weight depends on the shape of the unit.

On a 1340 by 1400 roof window with about 5.1 metres of glass perimeter and 1.6 square metres of area, moving from a 0.08 aluminium spacer to a 0.035 composite one improves the Uw by roughly 0.14 W/m²K. On a small unit the improvement is larger, because the perimeter to area ratio is higher. The way that figure is assembled from Ug, Uf and Ψg is set out on the page covering U-values and glazing specification. It is a bigger gain than most of the upgrades that cost considerably more.

The spacer is only half of the cold edge

A warm edge spacer raises the glass edge temperature by two or three degrees. It cannot do anything about the frame it sits in, or about the kerb the frame sits on.

Where a rooflight is built onto an uninsulated timber upstand, or where the insulation stops at the face of the kerb rather than continuing up and returning to meet the frame, the coldest surface in the assembly is not the glass at all. It is the internal lining a few centimetres away from it, and that is where the mould appears. That subject has a page to itself under thermal bridging at the rooflight perimeter, and it is the reason Sunspire treats the upstand detail as part of the glazing decision rather than as separate joinery.

Specifying the edge, in one line

The glazing line on a quote should name the spacer product and its material, not simply the phrase warm edge. It should state the declared Ψg, the argon fill percentage and the standard the unit is manufactured to. It should say whether the corners are bent or keyed. And on any flush glazed or structurally bonded unit it should confirm the secondary seal is silicone and the edge is protected from ultraviolet.

That is five short items. Every reputable sealed unit manufacturer publishes all of them, and a supplier who cannot produce them is selling a unit they have not specified. Where the edge fits into the wider case for spending money on glazing is covered across the energy efficient rooflights hub.

Keep reading

More on energy efficient rooflights

All about energy efficient rooflights →

Get a quote

Tell us about your roof

We start with the building, the covering and the slope. Then we tell you what will suit it, and what it will cost, as a fixed written number.

  • Surveyed before it is priced
  • 10-year workmanship guarantee
  • Building Control notification handled
  • New installations and replacements
Get a quote Call us