Tempered Glass Self-Breakage at 0.3%: What Heat Soak Testing Really Buys

A 0.3% self-breakage rate is the figure most often quoted for fully tempered glass, and it routinely fails to match what actually happens on site. On a 500 m2 project, 0.1% suggests almost nothing should break, yet three panes did. Two things usually explain the gap: the rate is counted per pane, not per square metre, and the higher the tempering stress, the smaller the inclusion needed to trigger a break. At 1.5 m2 per pane, 500 m2 is about 333 panes, so three failures is an observed rate of 0.90% - three times the 0.3% rule of thumb. This article works through the fracture mechanics behind that number and what heat soak testing (HST) actually buys.

NiS Phase Transformation and Volume Expansion

Nickel sulfide (NiS) inclusions are a primary cause of spontaneous breakage in tempered glass. During the tempering process, NiS in its high-temperature phase is quenched and frozen. Over time, these inclusions slowly transform into their low-temperature phase, causing a volume expansion of 2-4%. This expansion generates significant internal stress, which can lead to spontaneous breakage.

Be careful with a number that circulates widely: multiplying a 3% expansion by 900 MPa to get 27 MPa has no physical basis - 900 MPa is not a recognised modulus for soda-lime glass (E is about 70,000 MPa), and the product is meaningless as a stress. The expansion ratio is also the wrong variable to begin with. In linear elasticity the stress field around a transforming spherical inclusion depends on the expansion strain, the elastic mismatch and the constraint - but not on the absolute inclusion size. What size does control is whether the flaw is critical, and that is a fracture mechanics question.

• K_IC = 0.75 MPa*m^0.5 for soda-lime glass; Y = 2/pi = 0.637 for a penny-shaped flaw
• Inverted for the critical radius: a_c = (1/pi) * [ K_IC / (Y * sigma_t) ]^2
• At sigma_t = 45 MPa: a_c = (1/pi) * [ 0.75 / (0.637 * 45) ]^2 = 2.18e-4 m = 0.218 mm
• Critical inclusion diameter d_c = 2 * a_c = 0.436 mm
• Check back: K_I = 0.637 * 45 * sqrt(pi * 2.18e-4) = 0.75 MPa*m^0.5 = K_IC, closed

Impact of Different Inclusion Sizes

One caveat on that 0.436 mm figure: it accounts for the residual tensile stress only, and ignores the local stress concentrated at the inclusion interface by the transformation itself. Including that term pulls the critical size down sharply, so 0.436 mm is a conservative upper bound for the residual-stress-only case, not the value that governs real projects:

The two bases differ by close to an order of magnitude because the dominant term differs: residual tension is only an amplifier, while the stress generated at the NiS-glass interface is the driver, reported as high as 125,000 psi (about 860 MPa). Design against the 0.05-0.2 mm range, not against 0.436 mm.

So inclusion size matters, but not by scaling the stress. Doubling the diameter from 0.5 mm to 1.0 mm does not double a local stress - the elastic stress field around a transforming inclusion is largely size-independent. What changes is how close the flaw sits to the critical value computed above. A 0.5 mm inclusion is already past d_c = 0.436 mm and is expected to break; a 0.3 mm one sits below it and should survive. That step change, not a gradual stress climb, is what makes NiS breakage look so random in the field: it is governed by a threshold, not by a slope.

Field Consequences of Inadequate Control

Inadequate control over NiS inclusions can have severe field consequences. For example, a building with 1,000 tempered glass panels, each 10 mm thick, might experience 3-10 spontaneous breakages per year without HST. These breakages not only pose safety risks but also result in significant maintenance and replacement costs. By implementing HST, the number of breakages can be reduced to just 1 per year, significantly improving safety and reducing costs.

Residual Stresses and Impurity Amplification

Tempered glass has a surface compressive stress of at least 90 MPa, which provides strength and resistance to thermal and mechanical shocks. However, this also means that any impurities, such as NiS inclusions, are subjected to even greater internal tensile stresses. These internal stresses can amplify the effects of the NiS phase transformation, making the glass more susceptible to spontaneous breakage.

One more correction: surface compression and internal tension must never be added together. The tempering stress profile through the thickness is self-equilibrating - surface in compression, core in tension, two ends of the same stress field, separated in space. For an ideal parabolic profile the core tension is about half the surface compression:

The local tension from a NiS inclusion superimposes on that 45 MPa at the inclusion location - not on 90 MPa, and certainly not on a fabricated 140 MPa sum.

Comparative Analysis of Stress Levels

The variable that actually moves the threshold is the tempering stress. Higher surface compression means higher core tension, which means a smaller critical inclusion - so more of the inclusion population becomes dangerous:

Pushing surface stress from 90 to 120 MPa shrinks the critical inclusion by 44%. That is the real cost of over-tempering.

Field Consequences of High Stress Concentrations

The trend in the table above - higher stress, smaller critical inclusion - is independently confirmed: Liu Xiaogen and co-workers at the China Building Materials Academy state plainly that the critical NiS diameter decreases as surface stress increases, and international work records that heat-strengthened glass, with far lower residual stress, has very little chance of NiS breakage. Read the table as a relative trend rather than absolute thresholds: going from 90 to 120 MPa shrinks the critical size by about 44%, and that ratio holds; the absolute values sit high because the transformation stress is not included.

One counterweight: raising stress also improves the fragmentation count and impact resistance required by GB 15763.2. Maximum surface stress is a trade-off, not a one-way quality target.

The last row is the useful one. Heat-strengthened (semi-tempered) glass has a critical diameter of 1.414 mm, some 3.24 times the fully tempered minimum. Inclusions that large are vanishingly rare in production, which is why heat-strengthened glass is effectively immune to NiS breakage. That is a conclusion derived from the critical-size calculation, not a rule of thumb.

Industry Self-Breakage Rate Statistics

The industry standard for self-breakage rates in tempered glass is typically between 0.1% and 0.3%, based on the number of panels. This range is derived from extensive field data and is a widely accepted figure in the industry. However, this rate can be significantly reduced through the use of heat soak testing (HST).

HST holds the glass at 290°C ± 10°C for 2 hours, which greatly accelerates the metastable alpha to beta transformation. Inclusions already past the critical size transform and burst inside the furnace; those below it transform harmlessly and never expand again. Note that HST is a detonation process, not an inspection process - it does not locate NiS, it simply makes the dangerous panes fail on the factory floor instead of over a street.

Cost-Benefit Analysis of HST

The count itself is straightforward: 10,000 panes at 0.1-0.3% gives 10-30 failures; at 0.01% it gives 1. But pane counts only compare like for like at a single thickness. NiS enters at the melting stage and is spread roughly uniformly through the glass, so the more robust basis is by weight: lambda, the number of critical inclusions per tonne, sits at about 1-3 for ordinary tempered glass. The probability that a given pane contains one follows a Poisson distribution:

At a density of 2500 kg/m3, a pane weighs thickness(mm) x 2.5 kg per square metre, so one tonne covers 80 m2 at 5 mm but only 40 m2 at 10 mm. The practical consequence: at equal area, thicker glass breaks more often, with 10 mm running about twice the rate of 5 mm. Comparisons across thicknesses must go back to the weight basis, otherwise thick panels get under-specified.

Whether HST pays for itself then depends on the replacement cost per pane, Whether HST pays for itself depends entirely on the replacement cost per pane, and that number swings by more than an order of magnitude between a ground-floor shower screen and a 40th-floor curtain wall panel requiring cradle access and road closure. Work it out with your own unit cost rather than any published figure:

• Break-even when: cost of HST per pane < (rate before - rate after) x replacement cost per pane
• Ground-floor pane, low replacement cost: HST often does not pay back
• High-rise or overhead glazing: replacement cost dominates, HST pays back many times over

This is why the decision should be made per elevation, not per project.

Heat Soak Testing (HST) Parameters

The standard parameters for HST, as defined by EN 14179-1, are 290°C ± 10°C for 2 hours. This temperature and duration are carefully chosen to ensure that any NiS inclusions will undergo the phase transformation and either break the glass or become stable. The process is highly effective in reducing the risk of spontaneous breakage in the field.

Consider a batch of 10,000 tempered glass panels. Without HST, we might expect 10-30 panels to spontaneously break. With HST, this number is reduced to just 1 panel. This not only improves safety but also reduces maintenance and replacement costs.

Optimal Temperature and Duration

Both bounds have a real physical reason, and neither is about glass being damaged at 310°C - glass is tempered at 620-650°C, so 310°C cannot deform it. The genuine upper-bound concern is stress relaxation: above roughly 300°C the tempering stress begins to relax measurably, so an over-hot soak quietly eats the surface compression you are paying for. The lower bound is kinetic: at 270°C the alpha to beta transformation is too slow for a 2-hour hold to complete, so dangerous inclusions survive the furnace and break later. 290°C ± 10°C sits between those two failure modes.

Field Consequences of Inadequate HST

HST only addresses the NiS mechanism. Three installation-stage causes are untouched by it, and on site they are frequently as common as NiS: thermal stress breakage from over-tight edge restraint, unpolished edge damage from handling, and hard-point contact at setting blocks. A specification that relies on HST alone will still see breakage, and the residual rate will be blamed on the glass when the real cause is the frame.

GB 15763.2 Requirements and Surface Stress

GB 15763.2 sets three hard requirements: fragment count, surface stress and impact resistance. The fragment count is not a flat 40 - it is graded by thickness, which is the clause most often misquoted:

• 3 mm: at least 30 fragments in a 50 x 50 mm area
• 4-12 mm: at least 40 fragments in a 50 x 50 mm area
• 15 mm and above: at least 30 fragments in a 50 x 50 mm area
• Surface compressive stress: at least 90 MPa at any thickness

For a 10 mm thick tempered glass panel, the surface compressive stress of 90 MPa provides excellent resistance to thermal and mechanical stresses. The particle density requirement ensures that the glass is free from large inclusions that could act as stress concentrators.

GB 15763.2 mandates a surface compressive stress of 90 MPa and a particle density of 40 particles per 50x50 mm area. These values are based on industry practice and are code-mandated to ensure the quality and safety of tempered glass.

Impact of Particle Density on Glass Quality

Note also that the fragment count and the NiS risk pull in opposite directions. Raising surface stress improves the fragment count and impact resistance, but it also raises core tension, which shrinks d_c from 0.552 mm to 0.245 mm as tabulated above - so the NiS breakage probability goes up. Maximum surface stress is a trade-off, not a one-way quality target.

Field Consequences of Non-Compliance

Non-compliance with GB 15763.2 can have severe field consequences. For instance, a 10 mm thick tempered glass panel with a surface compressive stress of only 80 MPa instead of the required 90 MPa would be more susceptible to thermal and mechanical stresses. This could lead to premature breakage, especially in high-stress environments such as high-rise buildings or areas with extreme temperature variations. Ensuring that the glass meets the required standards is crucial for maintaining its structural integrity and safety.

Shower Enclosure Design Considerations

Shower enclosures commonly use 8 mm or 10 mm thick tempered glass. For frameless doors, the glass must be fully tempered and certified to 3C standards. This ensures that the glass meets the required safety and performance criteria.

Alternative designs include using semi-tempered laminated glass, which does not shatter upon breakage, or fully tempered glass with an anti-explosion film (PET >= 0.1 mm). These options provide additional safety and durability, especially in high-traffic areas.

• 8 mm or 10 mm thick tempered glass for shower enclosures
• Frameless doors must be fully tempered and 3C certified
• Semi-tempered laminated glass as an alternative
• Fully tempered glass with an anti-explosion film (PET >= 0.1 mm)

Comparative Analysis of Shower Enclosure Designs

When designing shower enclosures, it is important to consider the trade-offs between different glass types. For example, 8 mm thick tempered glass is lighter and easier to handle than 10 mm thick glass, but it may be more susceptible to breakage in high-stress environments. On the other hand, 10 mm thick tempered glass provides greater strength and durability, making it a better choice for high-traffic areas. Semi-tempered laminated glass offers the advantage of not shattering upon breakage, but it may be more expensive and less transparent than fully tempered glass. Fully tempered glass with an anti-explosion film (PET >= 0.1 mm) provides an excellent balance of safety and transparency, making it a popular choice for many applications.

Field Consequences of Poor Design Choices

Poor design choices in shower enclosures can have serious field consequences. For example, using non-tempered glass for frameless doors can lead to a higher risk of breakage and potential injuries. Similarly, using 8 mm thick tempered glass in a high-traffic area may result in premature breakage, requiring frequent replacements and increasing maintenance costs. By choosing the appropriate glass type and design, the risk of breakage can be minimized, ensuring the safety and longevity of the shower enclosure.

Common Mistakes and Their Costs

1. Mistake: Not performing HST. Cost: Increased risk of spontaneous breakage, leading to potential injuries and higher maintenance costs. Correct Approach: Always perform HST to reduce the self-breakage rate to 0.01%.
2. Mistake: Using non-tempered glass for frameless doors. Cost: Higher risk of breakage and potential injuries. Correct Approach: Use fully tempered glass with 3C certification for frameless doors.
3. Mistake: Ignoring particle density requirements. Cost: Increased likelihood of impurities and stress concentrations. Correct Approach: Ensure the glass meets the GB 15763.2 particle density requirement of 40 particles per 50x50 mm area.
4. Mistake: Using glass with insufficient surface compressive stress. Cost: Reduced strength and increased risk of breakage. Correct Approach: Ensure the surface compressive stress is at least 90 MPa, as mandated by GB 15763.2.
5. Mistake: Not considering alternative designs. Cost: Missed opportunities for enhanced safety and durability. Correct Approach: Evaluate semi-tempered laminated glass or fully tempered glass with an anti-explosion film for added safety.

Heat soak testing (HST) is a cost-effective method to reduce the self-breakage rate of tempered glass from 0.3% to 0.01%, significantly improving safety and reducing maintenance costs.

FAQ

What is the typical surface compressive stress for tempered glass and how does it affect nickel sulfide inclusions?

Tempered glass has a surface compressive stress of at least 90 MPa. This internal tension magnifies the stress concentration around nickel sulfide inclusions, which can expand 2-4% as they transform, potentially leading to spontaneous breakage.

How effective is heat soak testing (HST) in reducing the field breakage rate of tempered glass?

Heat soak testing, conducted at 290 ± 10°C for 2 hours, reduces the field breakage rate from 0.1-0.3% to approximately 0.01%, effectively lowering the rate by 1-2 orders of magnitude.

What are the fragmentation requirements for tempered glass under GB 15763.2?

According to GB 15763.2, in a 50x50 mm area, the fragmentation must result in at least 40 particles, ensuring that the glass breaks into smaller, safer pieces upon impact or failure.