Chapter 3 of 10 · 10 min read
Scale and fluxes
See how scale forms on hot iron and how much metal each heat costs, why sand and borax let a weld take by turning scale into a runny slag, when to flux, how much to use and how to weld without it.
Scale is the forge welder’s oldest opponent. Hot iron oxidises in air, and at welding heat it does so fast enough to grow a skin of oxide while the smith reaches for the tongs. Left between the two faces of a weld, that skin keeps them apart, and the weld fails. Smiths have dealt with scale for thousands of years by sprinkling the hot metal with something that melts and dissolves it, so that the hammer can squeeze it out as a liquid. Cravens’s treatise of 1921 recommended clean, sharp sand for iron and potter’s clay mixed with brine for steel; Swingle’s book of 1915 named white sand and borax, and noted that proprietary compounds were also sold.
This chapter explains how scale grows and what it costs in metal, how fluxes turn solid scale into a fluid slag, how the common fluxes compare, when and how much to flux, and how some smiths weld without flux at all.
How scale grows
When iron is hot, oxygen from the air reacts with its surface to form iron oxides, and the layer of oxide, the scale, thickens as iron atoms move out through it to meet the oxygen. Because the scale itself slows the movement, it grows quickly at first and then more slowly: its thickness rises roughly with the square root of the time. The rate also rises steeply with temperature, so that scale grows many times faster at welding heat than at a red heat.
- Surface heated
- 4 × 25 × 150 = 15,000 mm²
- Depth consumed
- about 0.05 mm after 1 minute; about twice that after 4 minutes
- metal lost in 1 minute ≈ 15,000 × 0.05 = 750 mm³, about 6 g of a 0.74 kg heated length, about 0.8%
- after 4 minutes at heat, about twice as much, about 1.6%
- three heats instead of one lose roughly three times the metal and dimension
A percent or so of metal per heat sounds small, but it comes off the surface where the weld is, it shrinks the stock’s section, and it adds up over repeated heats. Hart’s advice that a second heating should be avoided if possible was partly about the time lost and partly about this: every return to the fire costs metal, adds scale, and risks burning.
The scale itself
The scale on iron is not one substance but layers of different iron oxides. The layer next to the metal, the most iron-rich oxide, melts at a temperature close to welding heat, while the outer oxides melt higher. At welding heat, then, the scale is partly softened but not fluid enough to be squeezed out cleanly; it smears and stays. A flux’s job is to combine with the oxide to form something that melts well below welding heat, is runny enough to flow out under the hammer, and wets the iron so that it covers and protects the surfaces until the moment they meet.
The chart shows why fluxes work. Sand, which is mostly silica, reacts with the iron oxide at the surface to form an iron silicate that melts below welding heat. Hart described the reaction as rapid, forming easily fusible silicates that coat the fresh iron in a thin glass and stop further oxidation. Borax melts far lower still, dissolves iron oxide readily, and forms a glassy coating over the work long before welding heat, which is why it is the commonest modern flux.
Fluxes compared
| How it works | Suits | Watch for | |
|---|---|---|---|
| Clean silica sand | reacts with scale to form a fusible iron silicate | wrought iron and mild steel at full welding heat | needs a high heat to work; can leave inclusions if coarse |
| Anhydrous borax | melts low, dissolves oxide, coats the surface | mild and higher-carbon steel; general use | fluid slag runs off fast; spits and bubbles if not dehydrated |
| Borax with iron filings | borax as above; filings help the faces grip and bond | difficult steels; laminates | filings must be clean and fine |
| Clay with brine | forms a protective glaze | an older practice for steel | messy; not widely used now |
| No flux | relies on a reducing fire and very quick, clean work | skilled smiths; small clean welds; laminated billets in a sealed can | little margin for error |
Cravens judged borax “not good for the metal”, a view some smiths of the time held, possibly because of the heavy, glassy slag that ordinary borax leaves or the sodium it introduces, and preferred sand for iron. Modern smiths use borax widely, and its low melting point makes it especially suitable for steels, which must be welded at lower heats than wrought iron. Borax sold as a household cleaning product contains water of crystallisation, which boils off with violent spitting when it hits hot metal; smiths either buy anhydrous borax or dehydrate it by heating it in a pan until it stops bubbling, then grinding the glassy result.
When to flux
Flux is applied when the work is hot enough for it to melt and stick, usually at a bright red to orange heat, and before the scale has grown thick. If it is put on too early, at a dull red, it does not melt and falls off; too late, near welding heat, the scale it must dissolve has already grown. The smith takes the work from the fire, sprinkles the flux on the faces to be joined, returns it to the fire and brings it to welding heat. Large or difficult welds may be fluxed twice. Hart described smiths dipping the hot bars into a box of sand kept beside the forge, and sprinkling more by hand over large work during hammering.
How much flux
Not much. A thin, even coating that turns to a clear, runny glaze over the faces is all that is needed; heaped flux simply runs off into the fire, where it attacks the hearth lining and forms clinker. A teaspoon, about five grams, covers the faces of a small scarf weld. The smith judges the coating by its look at heat: a smooth, wet, bubbling glaze with the iron’s colour showing through is right; a dry crust or a thick pool is wrong.
What the flux looks like at heat
Experienced smiths read the flux as much as the colour. As the work nears welding heat, the flux becomes very fluid and the surface looks wet, almost greasy, and slightly bubbling. On some steels, the first few small sparks appear from the surface just as the flux runs freely, and that is the moment to take the heat. If the flux has run off and the surface looks dry, it has gone too far, or the flux was too little; the faces are scaling again and should be refluxed before the weld.
Squeezing it out
The fluid slag must leave the joint, and the joint is shaped to let it. Cravens explained that the faces of a scarf weld are rounded slightly, so that when they first meet at the middle the slag and scale squeeze out towards the edges. Flat faces trap a film of slag in the middle; concave faces trap a pool. The first hammer blows, light and quick, land at the middle and push outwards. A spray of glowing slag from the joint’s edges at the first blow is the sign that the flux has done its work.
Inclusions
Slag that does not get out stays in the weld as an inclusion: a line or patch of glassy oxide between the two pieces, which shows in a broken or polished weld as a dark seam. Small inclusions are common in wrought iron, which is itself full of fine slag fibres from its manufacture, and do little harm when they lie along the bar’s length. A slag sheet across the joint, at right angles to the load, weakens the weld greatly. Coarse sand, too much flux, and faces that do not squeeze out properly are the usual causes.
Welding without flux
Cravens noted that iron can be welded without flux if the work is done quickly. A deep reducing fire keeps scale thin, and a skilled smith who takes the heat at exactly the right moment, makes the transfer in a second or two and strikes cleanly can make a sound weld on clean stock without any. Some smiths prefer it for small work, because it leaves no glassy slag to clean off. Modern pattern-welders sometimes make laminated billets without flux by sealing the stack in a steel can or welding round its edges so that air cannot reach the interfaces, and some spray the stack with a little oil or kerosene that burns off the oxygen inside, a practice covered in Chapter 9.
Flux and the hearth
Flux that runs off into the fire attacks the hearth and the firebrick lining of a gas forge, dissolving them as it dissolves scale. Gas forges used for welding need a floor that flux can collect on without damage, such as a replaceable kiln shelf or a coating resistant to borax, and the floor is scraped or replaced as the glassy flux builds up. In a coal or coke forge, flux joins the clinker and is removed with it.
Health and flux
Borax and proprietary fluxes are chemicals with safety data sheets; borax is classed as a reproductive hazard by some regulators, and fluxes containing fluorides produce irritant fumes at welding heat. The smith keeps flux in labelled, closed containers, avoids breathing its dust or fumes, washes after handling it, and follows the safety data sheet for each product. Flux spitting from wet borax can burn eyes and skin, so eye protection is worn whenever flux is applied to hot work.
Clean stock before the fire
Flux deals with the scale that forms in the fire, not with everything the stock brings with it. Rust, mill scale, paint, oil and grease should be removed before the stock goes into the fire, by wire brushing, grinding or a pass through the fire followed by brushing at a red heat. Thick mill scale on new bar is especially troublesome, because it is already a layer of oxide before the fire adds its own. Galvanised or plated steel must have its coating removed completely, well away from the weld area, because zinc fume from a hot galvanised bar causes metal fume fever, and zinc and cadmium contaminate the weld. Heritage iron often carries many layers of old paint, some containing lead, which must be removed with the controls that lead paint requires before the iron is heated, as Chapter 10 explains.
Wrought iron’s own slag
Wrought iron welds more easily than steel partly because it already contains slag. It was made by working a pasty mass of iron and slag in a puddling furnace, then rolling and reheating it, so that fine threads of iron silicate slag run all through it along the direction of rolling. Hart described the process and its purpose, to squeeze out and work the slag and give the iron a fibrous structure. At welding heat, the slag threads at the surface melt and act as a flux of their own, so wrought iron is more forgiving of a thin flux or a slightly slow transfer than steel. The same slag threads show in a broken wrought iron weld as a fibrous, woody surface, and a weld across the fibres must be worked enough to carry them through the joint, which is why the heritage repairs of Chapter 10 are planned around the direction of the fibre.
The running case: the shop’s flux
This is an illustration. The shop’s owner tried three fluxes in a season of class welds: ordinary borax dehydrated in a pan, a commercial anhydrous borax, and a proprietary compound with iron filings. Students’ first welds took more often with the anhydrous borax than with the home-dehydrated kind, which spat and ran off unevenly; the compound with filings gave the best results on the higher-carbon steel used for tool welds. The shop now uses anhydrous borax for general welding and the filings compound for tools and laminates, each in a labelled shaker that gives a thin, even sprinkle.
A failure: the borax crust
This is an illustration. A student fluxed a pair of scarfs at a dull red heat, heaping the borax on thickly, then brought them up to heat. The borax formed a thick, lumpy crust that ran off one scarf entirely and pooled on the other. The weld took on one side and not the other, and broke when it was twisted.
Applied too early and too thickly, the flux had melted unevenly and left one face unprotected. The owner showed the student how to flux at a bright red with a thin sprinkle and watch for the even, wet glaze; the next weld took all round.
Exercise: scale and the second heat
A smith brings two pieces of 20 mm square stock, each heated over 100 mm of its length, to welding heat, but misses the first heat and must return the work to the fire for a second. Taking a depth of metal consumed of about 0.05 mm per heat over the four faces of each heated length, estimate the metal lost per heat and the change in the bar’s section after two heats. What else does the second heat cost?
Each piece’s heated surface is 4 × 20 × 100 = 8,000 mm², so a heat consumes about 8,000 × 0.05 = 400 mm³ per piece, about 3 g, and about 800 mm³ for the pair.
After two heats, each face has lost about 0.1 mm, so the section has shrunk from 20 mm to about 19.8 mm square, roughly a 2% loss of area. That is small for most work, but the second heat also adds new scale that must be fluxed again, raises the risk of burning the thinner scarf tips, and costs another few minutes of fire and labour, which is why getting the first heat right matters.
