Heat soak testing

Why tempered glass breaks spontaneously

Picture a tempered facade panel that has been hanging quietly for three years, then shatters on a calm afternoon with nobody near it. No impact, no storm, no obvious cause. That is the nightmare scenario, and almost every time, the culprit is a speck you could never have seen: a nickel sulfide (NiS) inclusion, a tiny contaminant baked into the glass back when it was still molten. What makes NiS special, and dangerous, is that it keeps changing shape long after the glass leaves the factory.

NiS comes in two crystal forms: alpha (compact, likes it hot) and beta (expanded, likes it cool). The fast quench of tempering slams the glass cold so quickly that any NiS inside gets frozen in the alpha form before it can settle into beta. Then the glass goes into service, sits at ordinary temperatures, and over months or years that trapped alpha slowly relaxes into beta, swelling roughly 2 to 4% in the process. A tiny bomb on a very long fuse.

The problem is that the inclusion is buried in rigid, pre-stressed glass with nowhere to grow. That swelling pries the surrounding glass apart, and if it happens to sit in the tension-rich core, the local stress can blow past the glass's strength and let go all at once. You get the usual tempered dice pattern, safer than annealed daggers, sure, but that is cold comfort when a three-metre lite lets go from twelve storeys up.

NiS inclusions form during float glass manufacturing when nickel-bearing contamination, most often a trace of stainless steel, comes into contact with sulphur in the molten batch. Modern float lines run cleaner than in past decades (largely because natural gas has replaced fuel oil as the furnace fuel, removing one historical sulphur source), which has measurably reduced NiS-related failures industry-wide, but the contamination pathway has never been fully eliminated. The resulting particles are tiny, typically only 100 to 200 microns across, which is why they are essentially undetectable by visual inspection before tempering.

What heat soak testing is

If the danger is a slow-fuse bomb, heat soak testing (HST) is a way to light every fuse at once, on purpose, while the glass is still safely in the building. The finished tempered lites go into a big oven, get heated to 290°C ± 10°C, held there for at least two hours, then cooled back down under control. That is the whole idea: make the bad panes fail here, not out on the tower.

At 290°C, the NiS alpha-to-beta phase transformation is greatly accelerated. An inclusion that might take five years to convert at room temperature will convert within hours inside the heat soak oven. Units containing a critical NiS inclusion will break during the heat soak cycle, inside the oven, before they are shipped. Units that survive the heat soak cycle are substantially lower-risk, though heat soak testing cannot guarantee zero future breakage: published research attributed to A. Kasper (cited via Asahi Glass technical literature) has put the residual failure rate at under roughly 1 break per 10,000 heat-soaked panes, a large improvement over untested tempered glass, but not an absolute guarantee.

Risk reduction, not elimination

Here is the honest part that gets glossed over in sales sheets: HST stacks the odds heavily in your favour, but it does not zero the risk out. The European standard reckons a heat-soaked lite is roughly 90% less likely to fail from NiS than an untested one. What is left over? The borderline inclusions, too small to force a break at 290°C in a couple of hours, but potentially just big enough to keep creeping over decades at room temperature. Rare, but not impossible.

For this reason, some high-consequence applications use laminated heat-soaked tempered glass, combining HST glass with a PVB or SentryGlas interlayer. If the tempered lite breaks, the interlayer holds the fragments in place, preventing them from falling. This is the gold standard for overhead glazing, glass floors, walkways, and canopies.

When to specify heat soak testing

You do not heat soak everything, that would be paying for a parachute on a ground-floor window. It costs money and time, and for plenty of ordinary glazing the background NiS risk is one you can live with. Where it earns its keep is anywhere a spontaneous break would be expensive, dangerous, or both:

• High-rise facade glazing: where a spontaneous breakage could drop fragments from height onto occupied areas below.
• Point-fixed structural glazing: where the glass is a structural element and its failure would compromise the assembly.
• Glass fins, spider fittings, bolted glazing: where the glass carries load and breakage cannot be contained by framing.
• Overhead and canopy glazing: always require laminated glass, often laminated HST.
• High-end residential and commercial projects: where client risk tolerance is low and post-construction breakage is especially disruptive.

Heat-strengthened glass does not require heat soak testing. Because heat strengthening does not fully lock NiS into the alpha phase, the slower cooling of heat strengthening allows more time for natural conversion, heat-strengthened glass has a significantly lower incidence of NiS-related breakage than tempered glass.

HST vs other risk mitigation strategies

Production time and cost

None of this is free, and the schedule is usually the bigger surprise. Heat soaking tacks roughly 8 to 12 hours onto the batch (ramp up, the hold, then a controlled cool), on top of the gas bill for keeping an oven at 290°C. Fabricators pass that through as a line item: a modest bump for standard commercial sizes, a lot more for oversized or oddball panels that need the oven all to themselves.

Because panels may break during the process, fabricators generally quote HST jobs with a small breakage allowance, typically 0.5–2% of the order quantity, built into the pricing. Specifiers should confirm the allowance with the fabricator and account for it in project scheduling.