Process · Materials · Construction Materials
How float glass is made
Nearly every flat pane you look through was made by letting molten glass spread out on molten tin. The surface of a liquid is flat by itself, so nothing has to be ground or polished.
From the book Glass: Sand, Fire, LightChapters 25 to 28. Read the full book free as a PDF.
The problem it solved
Before float, flat glass was a compromise. Sheet glass was drawn upward from the furnace in a ribbon, which was cheap but left faint waves, enough to distort a reflection or a view through a shop window. Plate glass was cast thick and then ground and polished flat, which gave a true surface at a huge cost in labour, abrasives and wasted glass. The twentieth century wanted plate quality at sheet prices, and car windscreens made the demand urgent.
The idea
Pour molten glass onto a bath of molten metal. The metal is liquid and dense, so the glass floats on it and spreads out. The metal surface is perfectly level, so the underside of the glass is flat. The top surface is flat too, from gravity and from the heat above, and because the glass never touches anything solid while it is soft, both faces come out fire-polished with no grinding at all.
The metal has to stay liquid across the whole working range of the glass, be dense enough to float it, and be affordable by the hundred tonnes. Tin fits. It melts at a low temperature and does not boil away in the heat, which is why it has been the float metal ever since.
Where it came from
Sir Alastair Pilkington and his team at Pilkington in St Helens, England, developed the process. Pilkington dates the start of experiments to December 1952. After about seven years of work and heavy cost, the company announced it to the world on 20 January 1959, and then licensed it widely. A popular story says Alastair Pilkington thought of the idea while washing dishes. The company says that story is a myth and that he rejected it himself. What is documented is the long, expensive grind of making it work continuously.
Today the float process is the world standard for high-quality flat glass, and Pilkington counts around 500 float lines worldwide.
Walking down the line
- Melting furnace. Sand, soda ash, limestone, dolomite and cullet are weighed and melted in a regenerative tank furnace.
- Tin bath. Glass enters at about 1,050 degrees Celsius and leaves at about 600, when it is stiff enough to lift onto rollers without a mark. The bath is a sealed steel case under a protective gas, usually nitrogen with a little hydrogen, so the tin does not oxidise into the glass. The ribbon is held wide and its thickness set by rollers gripping its edges.
- Annealing lehr. The ribbon rides on rollers through a long oven that cools it on a controlled profile. Uneven cooling stores stress that can burst a sheet later.
- Inspection and cutting. Lasers and cameras scan the ribbon, computer-controlled wheels cut it into sheets, robots stack it. Failed glass is crushed on the spot and goes back to the start as cullet.
Why the line never stops
A float line runs day and night, because the glass in the furnace cannot be allowed to cool. That is why float plants are large, costly and sited near their customers and fuel. Flat glass is heavy and low in value per tonne, so it does not travel well.
What happens after the line
The float line produces a commodity ribbon. Most value is added downstream. Tempering heats the glass and then chills the surface so that it freezes in compression. Laminating bonds two sheets around a plastic interlayer so that a cracked windscreen stays in one piece. Coatings add low-emissivity or solar-control behaviour. A modern double-glazed unit insulates much better than the single pane of a century ago.
Float glass starts from the same raw materials as every other glass. For that part, see how glass is made from sand.
Quick answers
Why is tin used in float glass?
Tin is liquid across the glass working range, dense enough to float the glass and cheap enough to hold in large quantities.
Who invented float glass?
Sir Alastair Pilkington and a team at Pilkington in St Helens, England. The process was announced on 20 January 1959 after about seven years of development.
