by Vaclav Smil
Perhaps the most surprising case of using meteoritic iron was discovered in 1818 when an English expedition, led by John Ross searching for the Northwest Passage, encountered the Inuit in northwest Greenland who had iron knives and iron spear points. The origin of these metal objects was puzzling, but eventually they were traced to iron flakes removed by hammer stones from meteorites of the Cape York fall (Wayman, 1988). Egyptian meteoritic beads predate the emergence of iron smelting by nearly 2000 years. Small iron objects are first documented from Mesopotamia of 2600–2500 BCE, but larger items and ceremonial weapons (the metal was still too rare to produce functional designs) became more common only after 1900 BCE; the metal was in everyday use after 1400 BCE and became widely affordable only after 1000 BCE.
Bloomery Iron
Smelting of iron followed the practices established for the production of color metals that had been going on in some parts of the Middle East for nearly 2000 years. Simple bowl-shaped hearths—shallow and usually clay- or stone-lined pits—were encircled by low circular clay walls. These walls were sometimes only knee high (Romans made most of their metal in furnaces no more than 1 m tall and less than half a meter of internal diameter), but in some parts of the Old World (including Central Africa) they eventually reached heights of more than 2 m (Van Noten & Raymaekers, 1988). Furnaces were filled with charcoal and crushed (and often roasted) iron ore, and relatively high temperatures were achieved by blowing in air through tuyères, narrow clay tubes inserted near the surface level (see Appendix B for definitions of some major technical terms associated with the production of iron and steel).
Tuyères were connected to leather bellows to force air into the hearth and to raise smelting temperature. Small bellows were operated by hand, larger ones by a man’s weight (using a treadle or a rocking bar), and the most powerful bellows were eventually powered by waterwheels. Temperature inside these charcoal-fueled furnaces usually did not reach more than 1100–1200°C (and often it was less than 900°C), high enough to reduce iron oxide and far from enough to melt the metal and produce liquid iron (pure Fe liquefies at 1535°C): the final product of this smelting was a bloom, a spongy mass made up of iron and iron-rich slag composed of nonmetallic impurities (Bayley, Dungworth, & Paynter, 2001). Hence the common name of these furnaces, bloomery, and of the product, bloomery iron.
Modern experiments demonstrated a relatively narrow range of conditions required for successful smelting (Tylecote, Austin, & Wraith, 1971). When the conditions inside the furnace are insufficiently reducing there is no metal produced, just iron-rich slag, but when they are too reducing slag becomes too viscous and cannot be easily separated from the metal. Intermediate conditions produce a good bloom; most of the slag comes from iron ore, about 30% originates from siliceous furnace lining, and less than 5% is fuel ash (Paynter, 2006). Blooms made in the smallest early furnaces weighed less than 1 kg, more typical medieval range was 5–15 kg, and the bloom mass increased to 30–50 kg (or even to more than 100 kg) only with the introduction of taller furnaces and waterwheel-powered bellows.
Bloomery iron contained typically between 0.3% C and 0.6% C, and in Europe it was the only ferrous material available in significant quantities during the antiquity and until the later medieval period. The iron produced by bloomeries was consolidated and shaped by subsequent smithing: repeated reheating and hammering of the bloom was required to produce a mass of wrought iron that contained just 0.04–0.08% C and that was ductile, malleable, and weldable. Wrought iron was used to make an increasing range of weapons and utilitarian and ornamental objects, ranging from arrowheads to bolts and axes (Ashkenazi, Golan, & Tal, 2013; Barrena, Gómez de Salazar, & Soria, 2008), and modern metallurgical examinations find small amount of slag trapped in these products.
Bloomeries supplied all of Europe’s iron during the continent’s first notable increase of demand for the metal that started in the eleventh century—with the introduction of iron mail, originally as small metal plaques, later as hand-forged and riveted knots—and expanded during the twelfth and thirteenth centuries. There was an increase in the production of hand weapons (ranging from knives to maces) and helmets, as well as agricultural and transportation tools and implements, with iron turned into plows, pitchforks, sickles, hoes, cart axles, hoops (for casks, wagons, and windmills), and horseshoes. The first documented use of powerful forge tilt hammers driven by waterwheels dates from 1135 in the famous Cistercian monastery of Clairvaux. More iron also went into construction as bolts, grills, bars, and clasps, and in the thirteenth century metal bands were used in Notre Dame de Paris. A century later the papal palace in Avignon consumed 12 t of the metal (Caron, 2013).
Bloomery smelting was practiced by virtually all Old World cultures, and thousands of these simple, temporary hearths (sometimes with parts of walls still intact) were excavated in regions ranging from both Sahelian and sub-Saharan Africa (Haaland & Shinnie, 1985) to nomadic societies on the steppes of Central Asia (Sasada & Chunag, 2014), and from coastal Sri Lanka (Juleff, 1996; 2009) to Scandinavia (Olsson, 2007; Svensson et al., 2009) and Korea, where the practice may have been transferred from what is now Russia’s Pacific coast region rather than from China where cast iron was dominant (Park & Rehren, 2011).
Most of the evidence of the earliest Euroasian iron smelting has been known for a long time, with numerous remains of simpler and lower structures (often called Corsican forges) and sturdier and taller furnaces (called Catalan forges) found from the Atlantic to the Urals. In contrast, new excavations of ancient bloomeries and new carbon datings have been changing our views on the development of iron metallurgy in Africa (Holl, 2009; Zangato & Holl, 2010). These findings indicate early smelting activities in regions ranging from the Middle Senegal Valley in the west to the Nile Valley in the east, and from Niger’s Eghazzer basin to the Great Lakes region of East Africa, with the many dates going to more than 2500 years before present and with inferred furnace temperatures of 1100–1450°C.
Persistence of this smelting technique is attested by the fact that the Spanish bloomeries at San Juan Capistrano (built during the 1790s) were the oldest ironworks in California, and operating bloomeries survived in parts of England into the eighteenth century; in parts of Spain and in southern France they were still present by the middle of the nineteenth century. Bloomery smelting was just the first step in obtaining useful metal: the ferrous sponge mixed with slag had to be processed by being repeatedly worked (wrought) by alternate heating and hammering (requiring as many as 30–50 cycles) in order to remove the interspersed impurities and to produce wrought iron that could be forged into weapons, horseshoes, colter tips, nails, and other small iron objects. For centuries all of this hot and hard labor was done everywhere manually, and only the adoption of larger waterwheels made it possible to build mechanized forges using heavier hammers. Even so, this traditional combination of bloomeries and forges had its obvious production limits.
Being small-scale batch operation—every heat was terminated in order to remove relatively small masses of the solid bloom—iron smelting in traditional low-rise bloomeries could never supply large-scale demand for the metal in an economic way, and labor-intensive (and also highly energy-intensive) forging added to the cost (further increased by substantial losses of iron during the forging process). Not surprisingly, with rising demand some European bloomeries, exemplified by medieval German and Austrian Stucköfen, became taller (Technisches Museum in Vienna has a fine model). These furnaces still produced small masses of metal (Stuck) whose removal required tearing the front wall of the structure, but because the smelting process lasted a bit longer and because waterwheel-driven bellows supplied more powerful blast and temperatures in lower parts of the furnace were higher, the resulting bloom was often a mixture of sponge iron and steel.
Blast Furnaces
Japan’s traditional tatara is the best known example of a furnace that could be both a bloomery producing a solid sponge iron and a blast furnace yielding liquid iron or steel
(Hitachi Metals, 2014; Iida, 1980). Tatara furnaces were rather low and rectangular (height of just over 1 m, width of 1 m, and length of 3 m) and (since the late seventeenth century) the blast was delivered by cross-blowing bellows. Given the scarcity of Japanese iron ores, abundant iron sands were charged with charcoal and the furnaces were operated in two modes, as zuku-oshi (pig iron-pressing) and kera-oshi (steel pressing, which will be described in the following section). Pig iron production used akome iron sand (with high titanium content, as 5% or more of TiO2) which was added after the charcoal charge; the furnace’s tuyères were placed low and at a low incline so the blast could penetrate the entirety of the furnace’s lower part; and, in order to achieve high temperatures needed to produce liquid iron, the heats lasted 4 days. The liquid metal was cast, or, after cooling, taken to a workshop for decarburization to produce sage-gane (low carbon steel) used to make a variety of everyday items.
Experiments with a replica of a 1.3-m tall tatara furnace built in Miyagi prefecture and charged with magnetite ore (rather than with iron sand) confirmed that it reaches very high smelting temperatures (in excess of 1500°C) during the last stage of its operation, and found increasing metal yield with higher metal content in the ore (Matsui, Terashima, & Takahashi, 2014). But even with the best available ore the yield was no higher than 62%, and with poorer ores more than half of iron present in charged iron sand could be lost in slag. Analysis of the produced metal showed carbon content of 0.15% and silicon content of 0.03%, with negligible amounts of Mn, P, S, Ti, and Cu, corresponding to low carbon steel.
But the first liquid iron was produced long before Japanese tatara became common. Unlike in Europe, where solid-state reduction of iron in bloomeries dominated the metal’s production until the late medieval period, there is only sparse and inconclusive evidence of bloomery smelting in ancient China, but plentiful evidence of producing liquid iron for casting. Chinese metallurgists, after centuries of experience with bronze castings, became the pioneers of liquid iron production during the Spring and Autumn Period (770–473 BCE), and smelting of relatively large quantities of liquid iron in fairly large furnaces is well documented during the Han dynasty (207 BCE–220 CE).
Chinese smelting took advantage of phosphorus-rich iron ores (with a lower melting point) and good refractory clays. The tallest refractory clay structures were just over 5 m high and they were often strengthened on the outside by vine cables or heavy timbers, were charged with nearly 1 tonne of iron ore, and the process yielded two tappings of cast iron a day (Needham, 1964; Wagner, 1993). In larger furnaces, reduction of iron oxide began already during the early stage of the descent through the combustion zone, and once the right temperature is reached iron will rapidly absorb carbon (its share rises to 2% and to as much as 4.3%); its melting temperature is lowered as the eutectic point falls to just 1150°C. Another essential component of the early Chinese success was the use of double-acting bellows whose strong air blast helped to raise the smelting temperature. A later Chinese innovation was to use coal packed around tube-like crucibles charged with iron and larger bellows powered by waterwheels.
Chinese mastered the casting of liquid iron into interchangeable stacked clay molds (introduced during the Bronze Age for bronze casting) as well as into iron molds (introduced about 2000 years ago) in order to mass-produce a variety of farming and manufacturing tools as well as thin-walled cooking pots and pans. Ancient Chinese ironmasters had also found ways to overcome the inherently brittle nature of cast iron: they discovered that prolonged heating of the alloy at high temperature causes it to lose its brittleness, a process we now explain by conversion of cementite to graphite embedded within ferritic or pearlitic matrix.
Adz and hammer heads, hoe blades, and spade and plough tips were produced by liquid iron casting by the middle of the first millennium of the common era, and excavations of worksites from the Han dynasty (at the very beginning of the common era) show production of large quantities of coins, belt buckles, horse bits, vehicle parts, and smaller decorative items (Hua, 1983). Some fairly large iron statues were also cast during the Han dynasty, and even larger ones came later. They included eight iron oxen used to anchor the ends of cables supporting the Pujin bridge in Shanxi built around 720 during the Tang dynasty—the largest ones were 3.32-m long and weighed 70 t— and a 40–50 t lion in Cangzhou in Hebei hollow cast in 953 (Derui & Hanping, 2011). But this early deployment of liquid iron production and skilled casting soon ended in a technical cul-de-sac. As with so many other innovations in ancient China, there were hardly any improvements for centuries to come and modern blast furnaces eventually arose from designs used in medieval Europe.
European production of liquid (cast) iron took off only with the diffusion of blast furnaces. After controversies regarding the dating of the oldest found blast furnace in Sweden (Lapphyttan, in the mining region of Norberg), it is now accepted that the furnace operated for two centuries starting in the second half of the twelfth century, putting the country’s first liquid iron production two centuries earlier than previously assumed (Rydén & Ågren, 1993). The other region with early blast furnaces was the lower Rhine Valley, starting during the latter part of the thirteenth century (to distinguish them from Stucköfen, their German name was Flüssöfen, liquid-producing furnaces, or Hochöfen, tall furnaces), and there are numerous reference to the casting of cannons from fourteenth-century Italy (Buchwald, 1998; Williams, 2009). In England, the first reliably documented blast furnace dates only to 1491.
In Europe and North America, cast iron from these furnaces has been widely known as pig iron (see Appendix B). In the late medieval and early modern Europe, it was often used directly as a substitute for bronze to cast large objects, above all guns, but for most other uses it had to be first converted into wrought iron or steel (see the following section). As with the early Chinese structures, gradual increase of European blast furnace stack heights allowed for a longer contact between the ore and the fuel, raised smelting temperatures, and reduced the melting point and charcoal consumption (Williams, 2009).
Georgius Agricola (1494–1555) gives a description of the prevailing practice of the sixteenth-century iron ore smelting in his famous De re metallica:
A certain quantity of iron ore is given to the master, out of which he may smelt either much or little iron. He being about to expend his skill and labour on this matter, first throws charcoal into the crucible, and sprinkles over it an iron shovel-full of crushed iron ore mixed with unslaked lime. Then he repeatedly throws on charcoal and sprinkles it with ore, and continues this until he has slowly built up a heap; it melts when the charcoal has been kindled and the fire violently stimulated by the blast of the bellows (Agricola, 1556, p. 420).
European blast furnaces of the seventeenth century were usually square stone structures, the largest ones being more than 5 m tall and almost 2 m in diameter at the center, and until coke replaced charcoal and coal-fired steam power displaced water power, they were subject to what has been termed the dual tyranny of wood and water. Because of high preindustrial costs of bulk land transportation and the need to power furnace bellows, all larger blast furnaces had to be located not only as close to all key material inputs as practicable—the most desirable location was close both to forests (or coppiced tree plantations) and to iron ore mines—but also at sites where running water could supply sufficient kinetic energy.
And, as Biringuccio (1480–1539) reminded any would-be builder, larger furnaces should be placed into “a hillside so that the ore and charcoal can be put in easily from above on level ground, which bears the load the animals bring there, for these blast furnaces are never so small that they need less than fifty or sixty sacks of charcoal and six or eight loads of ore continually” (Biringuccio, 1540, p. 152). Molten iron accumulating at the furnace bottom was periodically discharged into sand molds forming characteristic pigs. Pig iron was converted to bar (wrought) iron in forges; these manufactories could be located closer to the market, but they were often ad
jacent or not too distant from blast furnaces, because they also needed a great deal of charcoal for their operation, because the refining process entailed high metal losses (up to 30%), and because the most productive forges used waterwheel-powered hammers.
Excavations of some old forge sites—be it from the Roman-era France (Domergue, 1993) or from late medieval and early modern England (Belford, 2010)—show how long-lived and how extensive some of these operations were, with what sophistication they used water power for their operations, and what a wide range of products they were manufacturing. During the fourteenth century, European cast iron found a new large market in an increasingly larger scale production of field guns (cannons) and cannonballs (displacing the traditional stone shots), but most pig iron continued to be converted laboriously into wrought iron.
Charcoal
Charcoal was the only source of carbon used to reduce ores in preindustrial societies (Biringuccio, 1540; Porter, 1924). Thermodynamics explains why the fuel was produced by pyrolysis (heating in the absence of oxygen) already in prehistoric times, thousands of years ago: thermochemical equilibrium calculations indicate that carbon is a preferred product of pyrolysis at moderate temperatures, with water, CO2, CH4, and traces of CO as byproducts (Antal & Grønli, 2003). Traditional charcoaling producing the fuel inside simple earthen kilns (wood covered by mud or turf) was very inefficient, with only 15–25% of the mass of the dry charged wood converted to charcoal; in volume terms more than 20 m3, and only rarely less than 10 m3 of wood were consumed to produce a tonne of charcoal (Fig. 1.1). Similarly, iron smelting during the antiquity was also extraordinarily inefficient and medieval practices brought only marginal improvements. Johannsen (1953) estimated that medieval bloomery hearths required at least 3.6 times, and as much as 8.8 times, weight of fuel than the mass of ore.