Previous post: Introduction to Slag Analysis: What is it and why bother?
Whilst writing this series I realised that without understanding how iron is made in a furnace, all this talk about slag is a bit confusing. The following is an attempt to state the process clearly and concisely. As with many aspects of the archaeological record, iron production is not a simple process, so aspects of this may be confusing. Bear with me – and do feel free to ask questions.
The basic idea
Iron in the environment is almost always found combined with other elements (usually oxygen or sulphur) in rocks, excluding meteorites and native iron which have limited archaeological significance. When a rock contains enough iron for it to be profitable for us to attempt to use it to create iron, the rock is referred to as an ore. In order to make them easier to smelt, and to drive off unwanted elements like sulphur, ores can be roasted at around 800C in pits.
The process of turning ores into metal is called smelting, and involves heating the ores to high temperatures (usually around 1200C) in an atmosphere that is at least 75% carbon monoxide (Killick and Gordon 1989, 120). The carbon monoxide strips away the oxygen in the iron ore fragments, reducing the iron oxide to iron metal. Any unwanted minerals are melted down into a liquid rock-like mass called slag.
In order for these high temperatures to be reached, the reaction must take place in a container: this is the furnace. Fuel, usually charcoal, is burnt within the furnace creating both the heat and the carbon monoxide necessary for the reaction. To create the high temperatures enough air must reach the charcoal, and this is usually provided by manually forcing air into the furnace using bellows. The bellows are connected to a hole or tube into the furnace referred to by archaeologists as a tuyure.
The furnace is usually made of a ceramic, often tempered with silica in the form of sand grains or crushed quartz, but it can also be made of stones, or may include stone pieces in its construction. Silica helps to protect the ceramic of the furnace from the high temperatures, but in many cases the ceramic melts during the process of smelting.
But what exactly happens inside the furnace?
Exactly what is going on inside a furnace during smelting is not exactly known. The widely accepted theoretical model is:
[Hydrated iron oxides, FeOOH] – - – > haematite (Fe2O3) – - – > magnentite (Fe3O4) – - – > wüstite (FeO) – - – >iron (Fe)
If we assume the ore has been roasted, and the water driven off any hydrated iron oxides, the chemical reaction can be expressed:
- 3 Fe2O3 + CO – - – > 2 Fe3O4 +CO2
- Fe3O4 + CO – - – > 3 FeO + CO2
- FeO + CO – - – > Fe + CO2
At the same time, some iron is lost to the production of slag. As well as iron oxides, many ores contain unwanted ‘gangue’ oxides like silica, and as well as the melting furnace wall, the fuel itself may also contain numerous other oxides. In the high temperature conditions of the furnace any silica present is likely to combine with some of the iron (II) oxide, creating a olivine mineral known as fayalite:
2FeO + SiO2 – - – > Fe2SiO4 or 2FeO.SiO2
Fayalite, as well as the lower levels of other oxides, combine to form a liquid with a melting temperature below 1200C. Collectively this is known as slag, and drips through the furnace, sometimes being collected in a pit at the base of the furnace, sometimes being allowed to flow out of the furnace in a process known as ‘tapping’.
Depending on the exact conditions inside the furnace, the iron oxide reduction may take place in stages within the slag, or independently within the individual ore particles.
The difference between iron and copper
The reaction which reduced copper oxides down to copper metal happens in a very similar manner, using similar equipment and raw materials. However copper has a melting point of 1084C, which means that at the temperatures above the copper metal will melt and form a liquid at the bottom of the furnace. This makes it relatively easy to get out of the furnace.
In contrast, the melting temperature of pure iron is 1535C. This means that if the iron remains pure during the smelting process, it can’t melt in the furnace. If the above pathway of reduction is correct, the ore is reduced to lots of microscopic fragments of iron metal. How do these particles end up as useful lumps of iron?
Iron particles, slag and making the bloom
This is where things start to get a bit tricky. Killick and Gordon wrote a short but very interesting paper in 1989 that dealt with how ore particles actually get reduced to iron, and perhaps most importantly how that iron actually conglomerates at the base of the furnace.
I’ve found the paper quite hard going, and I’ve tried in the past to translate it into something visual which would be more accessible, but with limited success. At its most basic we theorise that there are two different theoretical pathways for ore fragments to be reduced. One uses a lot of fuel to create a really reducing atmosphere which turns the ore fragments to a mixture of metal and slag. The other uses less fuel to turn the ore fragments to slag first, which is then reduced to metal.
There are major differences between the raw material use and the outcomes of these methods. In the first method where lots of fuel is used to reduce ore directly to metal, minimal iron is lost to slag production and consequently rather lean (say less than 40% iron) ores can be used. However fuel usage is high, which is an important consideration when the fuel in use is charcoal which is costly and time consuming to produce and is difficult to transport. Additionally the powerful reduction atmosphere is thought to be likely to encourage reduction of other unwanted elements into the metal (Killick and Gordon 1989, 121). Additionally the availability of carbon and the early reduction of iron means that there is the chance that carbon-iron alloys may be created. This might be desirable if a harder iron is desired, but at high levels of carbon (2-4%) cast iron is created. This brittle, hard and not workable by standard smithing techniques.
By contrast, the lower fuel methodology dissolves all the ore to the slag before reducing available FeO to metal out of the slag. The major downside of this methodology is that it relies on having a very rich ore as much slag is created and there is the risk of loosing considerable iron to the slag. Due to the decarburising nature of the slag (Killick and Gordon 1989, 121) these blooms are usually low in carbon, but by controlling the frequency of tapping (Percy, 1864) it is suggested that producers could control how much carbon was taken in and consequently how steely the bloom was.
One of the aims of the analysis of slag is to work out whether these theories can be verified, and whether a particular sample shows evidence of particular production techniques.
And in the end?
So once the bloom is formed at the base of the furnace, what happens? As we’ve discussed, it’s not easy to get to that point. It’s important to remember that you can’t know what’s happening inside the furnace during the smelt – you can guess, and experience informs, but you can’t actually look. All this talk of how producers intentionally controlled the smelt is largely done on faith – only a handful of people in the world have sufficient experience in this method of iron production and all of them were self-taught.
However we do know that as the slag levels build, they will eventually reach the point where they cover the air blast intake holes. At that point anecdotal evidence suggests that the furnace will start booming and bubbling. At this time you have a choice – if you have the sort of furnace with a hole in it for letting slag out, you can proceed to do that. If not, then you have to stop the process there, before the liquid slag flows into your bellows and burns them and you!
If you decide to tap the slag out, you can keep the process going as long as you want, adding more ore and charcoal at regular intervals. If you’re stopping, then in the majority of cases it looks like the producers would break down the front of the furnace and pull out the bloom of iron whilst it was still hot.
At that point the race is on to consolidate the bloom – that is, to drive out all the liquid slag that is likely to be entrapped within the bloom. The bloom is very hot and it is likely that producers wouldn’t have wanted to waste this heat, which takes a lot of fuel to get, so they would have pulled the bloom out and started working it immediately. This involves smashing it with a large sledgehammer and forcing the slag out in a shower of glowing sparks. It’s at this point that the smith can get a feel for the quality of the iron. If the iron has taken in any unwanted elements – phosphorus, sulphur, high levels of carbon – things will start to go wrong and the bloom might break, crack or just be too hard to work.
After that, what happens to the bloom depends on who wants the iron and what it’s being used for. It’s likely to be turned into bars at some point, but whether this always happens at the same place as the smelt is unclear. What we do know is that it’s extremely rare to ever find iron at a smelting site, and if you do find iron bloom pieces it’s likely to have been discarded as unusable rubbish!
Killick, D. & Gordon, R.B., 1989. The mechanism of iron production in the bloomery furnace. In R. M. Farquhar, R. G. V. Hancock, & L. A. Pavlish, eds. Proceedings of the 26th International Archaeometry Symposium, held at University of Toronto, Toronto, Canada, May 16th to May 20th 1988. Toronto: University of Toronto, pp. 120-123.