What Is Borosilicate Glass? The Number That Matters
Home
Yiwu, Zhejiang, China
Mon – Fri 9:00 AM – 6:00 PM · Sat 9:00 AM – 4:00 PM
[email protected]

What Is Borosilicate Glass: The Number That Sets The Grade

Alan Xiao, Founder of Sourcing Pioneer
Alan Xiao Founder of Sourcing Pioneer
· Updated November 11, 2026 · 9 min read

Borosilicate glass is built on silica and boron trioxide.

Those two oxides form the frame of the glass. Everything else in the recipe is a small addition that changes how it melts, flows and sets.

The name in the catalogue is usually a number. Grade 3.3, high borosilicate, or just "boro". That number is the part buyers get wrong, and it is the part worth money.

This page covers what the number measures, and why three different shock figures circulate for one glass. It ends with the lines that belong in an order.

Soft-lit studio still life of a clear borosilicate beaker and glass lid on a beige surface
Clear, light and hard to shock. The label is a number.

The Short Answer

Borosilicate glass is a silicate glass in which boron trioxide does part of the frame building. The boron changes the structure enough to lower how much the glass grows when it warms up.

One published reference on borosilicate glass gives a typical laboratory recipe. About 80 percent silica, 13 percent boric oxide, 4 percent sodium or potassium oxide, and 2 to 3 percent alumina.

A glass plant’s own technical page puts the commercial grade a little wider. It lists silica at 70 to 80 percent and boric oxide at 10 to 15 percent, with small amounts of the alkali oxides. The two descriptions overlap, and both sit inside the same family.

The "3.3" in the trade name is not a purity figure and not a version number. It is the mean coefficient of linear thermal expansion, quoted in units of 10⁻⁶ per kelvin. That grade is defined in ISO 3585 and DIN 52328.

Two physical numbers are worth holding in mind. Density is about 2.23 grams per cubic centimetre. That is a little below ordinary soda-lime glass, because boron is such a light element. Highest working temperature is typically about 500 °C.

What The 3.3 Actually Measures

The coefficient of thermal expansion is a rate. It says how much a length of material grows for each degree it warms.

A value of 3.3 means one metre of glass grows by 3.3 microns per kelvin. Over a 200 degree rise that is about 0.66 millimetres. That sounds small, and it is small.

Soda-lime glass sits at about 9.0 on the same scale. That is the whole point of the comparison. The 3.3 grade grows roughly one third as much for the same heat change.

Bar chart comparing thermal expansion of fused quartz, borosilicate 3.3, aluminosilicate and soda-lime glass
Lower is better for shock. The 3.3 grade sits near the bottom.
MaterialExpansion, 10⁻⁶ per KWhat it is used for
Fused quartz0.55Optics and lab parts where cost does not decide
Borosilicate 3.33.3Bakeware, drinkware, lab glass, vials
Aluminosilicate4.5Harsh chemical and high-heat work
Soda-lime9.0Windows, bottles, most cheap drinkware

Growth matters because glass rarely heats evenly. One face of a baking dish sits on a hot shelf while the other face still holds cold food.

The warm face tries to grow and the cool face holds it back. The difference becomes strain inside the wall.

Halve the growth and you roughly halve that strain. That is what the number is worth paying for. It buys room on a bad day, not a nicer look.

Three Different Thermal Shock Numbers

Spend ten minutes on this topic and you meet three figures for how much heat the glass survives. All three are defensible. They are not the same measure.

The property figure: about 166 °C. The reference states that the glass can take a gap of roughly 330 °F before it cracks. That is 166 to 170 °C. The same entry puts soda-lime at about 100 °F, or 40 °C.

The test figure: about 220 °C. The test method is DIN ISO 718. It defines resistance as the gap at which half the samples crack in cold water. For thin-walled tube that figure is around 220 °C.

The design figure: about 120 °C. The same plant page then advises that most parts should not pass a 120 °C gap in normal use.

Three stacked figures showing 120, 166 and 220 degrees Celsius and what each one measures
Three numbers, three meanings. Only one belongs on a care label.

Read the definitions and the gap makes sense. The test figure is a failure count. Half the samples crack at that point, which means half do not.

No one designs a product so that half of them fail at the stated limit. So the design figure sits well below the test figure. The property figure sits between the two.

Wall thickness moves all three. The 220 °C value is stated for thin-walled tube. A thick pressed dish has a steeper gap through the wall, not a gentler one.

This is where a quotation hides a problem. A sheet that says "shock 220 °C" and stops there gives you a number and withholds its meaning.

Ask which line it is. A test result, a design limit, or a property claim. Only the second one is safe to print on a care label.

Borosilicate Against Soda-Lime And Tempered Glass

Soda-lime glass is the default material for glass. It is cheap because its ingredients are cheap. It works well whenever heat is not the problem.

Its expansion is 9.0 and its shock resistance is around 40 °C. Its highest short-time working temperature is about 150 °C. On the same plant page, its water rating is class 3 while grade 3.3 reaches class 1.

That rating comes from the ISO 719 test. Class 1 is the better result. It matters for anything that holds water for a long time, and it is why medicine vials use this glass.

Tempered glass is a different story, and it is where buyers get lost. Tempering does not change the recipe. It heats soda-lime glass to about 620 °C and cools it fast. That puts the outer faces into compression and the inside into tension.

The treatment makes the glass roughly four times stronger than annealed glass, on the standard description of the tempering process. It also makes it break into small rounded chunks rather than long sharp shards.

What tempering does not do is change the expansion. A tempered soda-lime dish still grows at about 9.0. It is a tougher version of the same material, not a stand-in for a low-expansion one.

There is a second difference that hits production. Tempered glass cannot be cut, drilled or ground afterwards, because any of those steps breaks the stress balance. Tube in grade 3.3 can be cut and flame-worked long after it was made.

PropertyBorosilicate 3.3Tempered soda-lime
Expansion, 10⁻⁶ per K3.39.0
Shock, thin wallAbout 220 °C in testAbout 40 °C
Working temperatureAbout 500 °CAbout 150 °C
Mechanical strengthAnnealed, ordinaryAbout four times annealed
Break patternShardsSmall rounded chunks
Can be cut after formingYesNo
Water ratingClass 1Class 3

Put simply, tempered glass wins on a drop and grade 3.3 wins on a heat jump. The two are not a swap in a specification.

Why A Pyrex Benchmark Can Mislead You

Pyrex was introduced by Corning in 1915. The brand name then became a synonym for this glass in English-speaking markets. That is where most of the confusion in this category starts.

The reference adds one sentence that matters commercially. Since the 1940s, a sizeable share of Pyrex-brand glass has been made of soda-lime rather than borosilicate.

So a buyer who says "make it like Pyrex" has named a brand, not a material. One brand name has covered two different glasses, with expansion figures of 3.3 and 9.0.

From the outside the two look alike. Both are clear, both are stiff, and both ring when tapped. A photo cannot separate them, and neither can a drop test on one sample.

What separates them is a document. An expansion figure, a grade name tied to a standard, and a water rating. If a supplier cannot put those three on paper, the word borosilicate is doing all the work.

More Boron Is Not Better

Boric oxide is the part that does the work. So it is easy to assume that more of it makes a better glass. The reference says the opposite above a set point.

These glasses split into three families, and water resistance does not rise steadily across them.

The low-expansion family holds 12 to 13 percent boric oxide with over 80 percent silica. Expansion is 3.3 and chemical durability is described as very high. This is the cookware and labware glass.

The alkaline-earth family carries about 75 percent silica and 8 to 12 percent boric oxide. It adds up to 5 percent of alkaline earth oxides and alumina, and expansion rises to the 4.0 to 5.0 band.

The high-borate family goes to 15 to 25 percent boric oxide with 65 to 70 percent silica. These soften at low heat and seal to tungsten and molybdenum. Some also pass ultraviolet light down to 180 nanometres.

The same source is blunt about the trade. Raising boric oxide above 15 percent reduces chemical resistance. In that respect the high-borate glasses differ widely from the other two.

FamilyBoric oxideExpansionWhat it is good at
Low expansion12–13 %3.3Heat and chemical resistance together
Alkaline-earth8–12 %4.0–5.0Softer, easier to work
High-borate15–25 %Very lowSealing to metal, UV transmission

So a catalogue shouting "high boron" is not promising a tougher casserole dish. It may be describing a glass chosen for a metal seal in a lamp or a sensor.

The grade that matters for cookware already has a name. It is 3.3, and it is defined by its expansion, not by how much boron went into the furnace.

Two Production Routes, Two Cost Structures

How a piece is formed decides how even its wall is. It also sets what the tooling costs and how many sizes one line can offer.

Tube drawing pulls a continuous tube out of the melt. The tube is then cut, reheated and formed into a cup, a bottle or a cylinder. The wall comes from the draw, so it stays even all the way round.

The commercial result is wide cover from shared tooling. One tube size can become many heights and diameters, because the shaping happens after the tube exists.

Moulding drops a measured gob of glass into a mould and presses or blows it to shape. The form is exact, and the surface takes the mould’s finish.

The cost sits in the mould. Every size and every profile needs its own, so a change of shape is a new tool rather than a new setting.

Diagram comparing drawn tube and moulded glass production routes and what each one costs
One route shares tooling across sizes. The other charges per shape.

Wall evenness is the quality question behind the choice. A thin patch heats faster than the rest of the wall, so the gap across that patch is larger.

The whole shock advantage comes from keeping that gap small. An uneven wall gives up part of what the material was bought for.

Flat glass takes a third route, floating on a bath of molten metal. That is how sheet for windows and sight glasses is made. It has little to do with a drinking glass.

Where Borosilicate Still Fails

Low expansion is not indestructible, and three ways to break it survive the grade.

Thick walls. The 220 °C figure is quoted for thin-walled tube. A thick part holds a steeper gap across itself, so the edge narrows as the piece gets heavier.

A gap, not a temperature. This is the usual misunderstanding in the category. A dish in grade 3.3 can sit in a 250 °C oven without trouble, because it heats slowly and evenly.

Put that hot dish on a wet stone counter and the gap between the hot base and the cool edge cracks it. The high temperature was never the problem.

Impact. Boron lowers expansion and leaves mechanical strength roughly where annealed glass sits. A knock on a hard floor does not care about the expansion figure.

A chip on the rim matters more than it looks. A chip is a crack that has already started, and the next heat change drives it inward.

So the honest claim is that the glass takes a fast change in heat better than soda-lime. It is not shatterproof, and the word unbreakable belongs in no listing.

Five Lines For The Order Sheet

"Borosilicate" on its own describes a family, not a product. Five lines turn it into something a factory can be held to.

  1. Grade and standard. Write grade 3.3 to ISO 3585, and add DIN 52328 for a European buyer. The standard is the useful half of that sentence.
  2. Expansion figure. Ask for the stated mean over 20 to 300 °C, with the tolerance. A grade claim without a number is a word.
  3. Water rating. Class 1 to ISO 719 for anything that holds liquid. This is the line that separates lab glass from decorative glass.
  4. Wall thickness. A stated figure plus a tolerance, measured at the thinnest point rather than the average. An average hides the patch that will fail.
  5. Thermal shock. Name the method, DIN ISO 718, and the gap you expect. Say whether it is a test result or a design limit.

Those five lines also make quotations comparable. Two suppliers quoting one shape can be ranked on paper, instead of on a sample that arrived in good condition.

Every line belongs on one page, and that page travels with the enquiry rather than after it. The sheet a glass factory will price from is the right home for the five.

Four Checks Before A Bulk Order

Documents settle the grade question. The goods still need a look, and four checks cover most of the risk.

1. Read the certificate against the line, not the brochure. The expansion figure, the water rating and the standard number should all appear as stated or measured values.

2. Measure the thinnest point on ten pieces. A caliper at the rim, the base centre and the shoulder. A wide spread points at the forming route.

3. Run one gap test and expect it to prove little. Boiling water into a cold piece is only about a 100 °C gap, which tempered soda-lime can survive. The standard tests a much wider gap for a reason.

4. Inspect the rim and the base edge. A chip is a crack with a head start. This is the defect that turns a fair claim into a return.

Keep one approved piece sealed with the certificate and the spec sheet. Compare the bulk delivery against that piece, not against a fresh sample sent later.

If the same container carries bakeware in other materials, check those against their own line. The notes on buying bakeware by the carton cover that family. The drinkware guide on bodies, lids and printing covers the cups that ship beside it.

What follows an inspection pass is a separate job. The steps that follow an inspection pass cover the paperwork, the freight and the customs work.

GRADE 3.3

If a quotation says borosilicate and stops there, the word is carrying the whole claim. Send me the sheet and the certificate, and I will check the expansion figure, the water rating and the wall tolerance against each other. Where the standard number is missing, I will name the three lines to add before you pay a deposit.

Send The Sheet

Questions Buyers Ask

Is borosilicate glass better than tempered glass?

It depends on the risk. Grade 3.3 handles a fast heat change far better. Tempered soda-lime handles a knock better, because tempering makes it about four times stronger. They solve different problems.

Can borosilicate glass go straight from the freezer to the oven?

Usually yes, because that gap sits inside its range. The risk is uneven heat, so a thick piece going onto a hot shelf with cold food inside is the case to avoid.

What does 3.3 mean on a borosilicate label?

It is the mean coefficient of linear thermal expansion, 3.3 in units of 10⁻⁶ per kelvin. It is also a grade defined in ISO 3585, which is why the number reads as a name.

Is more boron better in borosilicate glass?

No. Above roughly 15 percent boric oxide, water resistance falls. Those high-borate glasses exist for sealing to metal and for ultraviolet work, not for cookware.

Is Pyrex the same as borosilicate glass?

Not reliably. Pyrex started as this glass in 1915, but since the 1940s a sizeable share of glass sold under the brand has been soda-lime. Treat the brand as a brand, and ask for the expansion figure.

Is borosilicate glass safe for food contact?

It is widely used for food and for medicine containers, largely because its water rating reaches class 1. What a buyer still needs is a declaration for the market they sell in, not a family name.

Questions & Comments

We read every one and reply within 24 hours

Every cracked baking dish starts the same argument: the glass was faulty, or the dish went from a hot oven to a cold counter. Tell me the two temperatures and the wall thickness, and the argument usually settles.

Subscribe
Notify of
guest
0 Comments
Most Voted
Newest Oldest
0
Would love your thoughts, please comment.x
()
x