Archaeometallurgical slag
Ancient iron production is different from modern industrial iron production in many aspects. The archaeological investigation of early iron production usually depends on the debris in the site to figure out the technology and working process. Slag, waste of iron-working processes such as smelting or smithing, is one of the powerful research objects. After the iron working process, the iron products were always moved and processed, while slag was left at the working site. The size, shape, chemical composition and microstructure of slag are related to the iron-working process. In addition, given its resistance to weathering, slag is usually well preserved in the archaeological site. Therefore, it could provide clues to the iron working technology and process.
Overview
Ores used in ancient smelting process are rarely pure metal compounds and the worthless material has to be removed through slagging. Slag is the necessary material to collect impurities from ores, gangue, the furnace lining and charcoal ash, which can yield numerous aspects of information about smelting process (Bachmann 1982).[1] Slag is also direct evidence of ancient smelting activities. Through slag analysis, archaeologists can reconstruct ancient human activities concerned with metal and try to discuss the organization and specialization in metal production (Maldonado et al. 2009).[2]
In modern view, slag should have three functions. The first is to protect the melt from contamination. The second is to accept unwanted liquid and solid components. Finally, it can help to control the supply of refining media to the melt. In order to approach these objectives, a good slag should have lower melting temperature, lower gravity and better fluidity to ensure a liquid slag which can be separated well with metal melting. Moreover, slag should maintain the correct composition so that it can collect more impurities and be immiscible in the melt (Moore1990, 152).[3] Yet this is not always the case for ancient slag. In the investigating of ancient metallurgy the intention and constraints of ancient people should be carefully considered.
There are several kinds of non-ferrous metal that have been used by ancient people. Slag analysis is usually applied in the research about smelting and refining processes of copper, tin, lead and zinc. Slag can be created in many steps of metal production. Based on the mechanism of creation, slags can be catalogued as furnace slag, tapping slag, crucible slag, etc. In a smelting furnace, as many as four different phases could co-exist. From top to bottom, they are slag, matte, speiss, and liquid metal (Thornton 2009).[4] The separation between these phases would not be perfect and therefore, three other phases can usually be observed in the slag if they ever existed. Through chemical and mineralogical analysis of the slag, things can be learnt like the identity of metal smelted, kinds of ore used and technical parameters such as working temperature, gas atmosphere, slag viscosity, etc.
Forming of slag
Natural iron minerals are mixtures of iron and unwanted impurities, or gangue. These impurities are usually removed by “slagging” during the smelting process.[5] Slag was removed by liquation, that is, the solid gangue was converted into liquid slag. Therefore, the smelting process must be operated at or above the temperature at which the slag is fluid enough to be removed from the ores.
Smelting could be conducted in various types of furnaces, and the condition within the furnace may differ, thus affect the morphology, chemical composition and the microstructure of slag. Take bloomery and blast furnace, which were two common methods for smelting iron, for example. In the bloomery process, a solid state of iron was produced. This is because the bloomery process was conducted at a temperature higher than that which the pure iron oxide could be reduced to iron metal, but lower than the melting point of iron metal; therefore, solid iron metal was obtained. Blast furnaces were used to produce liquid iron. Generally speaking, the difference between bloomery and blast furnace production is that the blast furnace was operated at higher temperature and a more reducing condition than the bloomery. Because of the higher reducing environment, which was achieved by increasing the fuel to ore ratio, more carbon reacted with iron ore, and thus resulted in the production of cast iron rather than the plain iron. Besides, the more reducing condition within the blast furnace also generated less iron-rich slag.
Many other factors also influence the composition and morphology of slag during the smelting process, in which the charcoal was exclusively added to the furnace, reacted with oxygen, and generated carbon monoxide, which was responsible for reducing the iron ore into iron metal. The liquefied slag was separated from the ore, and was removed through the tapping arch of the furnace wall.[6] This is called tapped slag. However, the methods of removing slag are somewhat different. Some slag may be left inside the furnace rather than being tapped, therefore resulting in various morphologies of slag, which serves as a useful indicator to investigate the smelting process and furnace type. Smelting, the flux, the charcoal ash and the furnace lining may also contribute to the chemical composition of slag, which could be useful to infer the yield of production.
In addition to the smelting process, slag may also form while smithing and refining. The product of the bloomery process is heterogeneous blooms of entrapped slag. Therefore, smithing is necessary to cut up and remove the trapped slag by reheating, softening the slag and then squeezing it out. On the other hand, the refining process is conducted to refine the cast iron produced in the blast furnace process. By re-melting the cast iron in an open hearth, the carbon is oxidized and removed from the iron. Liquid slag is formed and removed in this process.
To sum up, different furnace types, working conditions and processes result in various types of slags. Through the investigation of slag, either in the macro-scale or the micro-scale analysis, many archaeological questions about metallurgy may be answered. In the following section, the analysis of slag will be addressed in macro-analysis, micro-analysis and yield evaluation respectively.
Slag analysis
Analysis of slag is mainly based on its shape, texture, isotopic signature, chemical and mineralogical characters. Analytical tools like Optical Microscope, scanning electron microscope (SEM), X-ray Fluorescence (XRF), X-ray diffraction (XRD) and inductively coupled plasma-mass spectrometry (ICP-MS) are widely employed in the study of slag.
Macro analysis
The first step in the investigation of archaeometallurgical slag is usually the identification and macro analysis of slag in the field. Physical properties of slag such as shape, colour, porosity and even smell can be used to identify slags in archaeological site and catalogue them primarily. For example, tap slag usually has wrinkling upper face and flat lower face which contacted with soil (Tumiati 2005).[7] This primary catalogue can be quite useful in designing sampling strategy which should make sure that samples collected for micro analysis can represent the character and variety of slag heaps. In addition, total weight of slag heaps could also be estimated through macro analysis and then the production scale might be determined.
Bulk chemical analysis
Chemical composition of slag can reveal a lot of crucial information about smelting process and XRF is the most common tool used in this study. Hauptmann (2007, 20)[8] stated that through the composition of slag, four smelting parameters could be detected. They are the composition of charge, the firing temperature, the gas atmosphere and the reaction kinetics. In order to obtain low melting slag, most ancient slag composition is consistent with the eutectics parts of quaternary system CaO-SiO2-FeO-Al2O3. In most cases this system can be simplified to CaO-SiO2-FeO2 (Hauptmann 2007, 21).[8] In some areas, proportion of silicates to metal oxides in gangue with ore and furnace lining is consistent with this eutectics parts and form slag by itself. Otherwise, flux is needed to get the right propertion (Craddock 1989).[9] Plotting result of chemical analysis in the ternary phase diagram, the melting temperature of slag can be learnt. This is a quite typical procedure in slag analysis and has nearly been used in every case of this investigation. For example, Chiarantini et al.'s (2009)[10] analyzed the copper smelting slags from Populonia, southern Tuscany and estimated the melting point of these slags were about 1125-1250°C. Another quite important parater, viscosity of slag, can also be calculated through its chemical composition with equation:
- Kv=CaO+MgO+FeO+MnO+Alk2O/Si2O3+Al2O3 where Kv is the index of viscosity.[11]
Additionally, composition of slag can determine the metal smelted and provide information about ores. In early stage of smelting, separation between melting metal and slag is usually not complete. Thus, the smelted non-ferrous metal would usually be rich in the slag created in this process. For example, copper slag often contain more than 0.1% copper in it. For slag rich in lead and sometimes tin with over 100ppm of silver, it might be considered as the waste of silver production (Craddock 1989).[9] The main, minor and trace elements of slag can be indicators about type of the ore used in smelting as well(Hauptmann 2007, 24).[8] For example, presence of sulfur usually suggests that sulphidic ores had been used (Bachmann 1982,16)[1]
Mineralogical analysis
Optical microscope, SEM, and XRD can tell us the identity and distribution of minerals in slag. Minerals present in slag are often good indicators of gas atmosphere in furnace, cooling rate of slag and homogeneity of slag. If there is un-decomposed charge trapped in slag, more information about ore and flux can be learnt. Slag minerals are generally catalogued as silicates, oxides and sulphides. Moreover, metal pills are always concerned in this study as well. Bachmann (1982)[1] classified main silicates in slag according to the ration between metal oxides and silica.
- Ratio MeO : SiO2 silicate examples
- 2 : 1 fayalite
- 2 : 1 monticellite
- 1.5 : 1 melilites
- 1 : 1 pyroxene
- Ratio MeO : SiO2 silicate examples
Fayalite (Fe2SiO4) is the most common mineral in ancient slag. Through the shape of fayalite, cooling rates of the slag can be roughly estimated (Donaldson 1976, cited in Etller et al. 2009).[12][13] In addition, fayalite can react with oxygen to form magnetite:
- 3Fe2SiO4 + O2= 2FeO·Fe2O3 + 3SiO2
Therefore, the gas atmosphere in furnace can be calculated through the ratio of magnetite/fayalite in slag (Hauptmann 2007, 22).[8] Metal sulphides usually suggest that sulphidic ores has been used. These metal sulphides survived from oxidizing stage before smelting and therefore may also indicate a multi-stage smelting process. Pure metals are always hoped to be removed from slag by smelter yet this process is never complete. Left metal prills can tell archaeologists what kind of metal was extracted from ore.
Mineral composition can also be used to estimate the chemical composition primarily. For example, monticellite and pyroxene would form when fayalite can not take more CaO into it. They could be the indicator of relatively high calcium content (Bachmann 1982).[1]
Moreover, in the early stage of metallurgy, the original charges of furnace may not decompose completely and therefore some component of ores or flux would remain in slag. These material can be discerned by petrographic analysis and help to identify ores and fluxing agent used in the process. Hauptmann (2007, 171)[8] has ever investigated copper slag from Faynan, Jordan and interpreted quartz inclusions in the slag as a charge coming from a quartz-rich host rock.
Lead isotope analysis
Lead isotope analysis is a useful technique to determine the ore source of ancient smelting activities. Lead isotope composition is the signature of certain metal ores deposit and varies very little throughout the whole deposit. It has also proved that lead isotope composition will pass unchanged through smelting process (Stos-Gale 1989).[14] Therefore, the metal rich non-ferrous slag is usually proper material for this analysis. On one hand, it can provide information about origin of ores used in the smelting site and facilitate the research about production organization. On the other hand, it can play a significant role in tracing raw material source of certain metal artefact.
The content of four stable isotopes of lead are usually used in this analysis. They are 204Pb, 206Pb, 207Pb and 208Pb. Ratios: 208Pb/207Pb, 207Pb/206Pb and 206Pb/204Pb are measured through mass spectrometry and widely employed to express the character of certain ore deposit. Except 204Pb, other lead isotopes are products of radioactive decay of uranium and thorium. During deposit forming process, uranium and thorium would be separated from ores. Thus, deposits formed in different geological period would have various lead isotope signature.
- 238U →206Pb
- 235U →207Pb
- 232Th→208Pb
Hauptmann (2007, 79)[8] analysed lead isotopic composition of slags from Faynan, Jordan and the result revealed the same signature as the ores from the Dolomite-Limestone-Shale Units in Wadi Khalid and Wadi Dana.
Physical dating
Date of slag is usually thought to be a complex problem for there is few chronological indicators like pottery or radiocarbon dating material like charcoal stratigraphic associated with slags. Direct physical dating of slag through thermoluminescence dating could be a good method to solve this problem. Thermoluminescence dating (TL dating), which is often used in investigation about pottery age, is actually measuring the time since last heating event of the sample. Agencies which can emit luminescence signals are crystals like quartz and feldspar.
Slag has good potential for TL dating because it forms through high temperature process and usually contains crystals that can emit luminescence signal. However, the complex composition of slag always affect the quality of TL measurement. Haustein (2003)[15] suggested that if quartz separated from out from slag was used as dating material, the problem could be solved.
Early copper smelting slag
At dawn of copper smelting, the shape and composition of copper smelting slag vary considerably from modern impression of slag. The traditional way to separate slag of non-ferrous metallurgy as smelting slag and crucible slag is not applicable anymore. Charges are only partially decomposed and the whole slag has never been fully molten. It can not be tapped out of furnace and would solidified in it as furnace slag. People have to crush it into pieces to recover metal prills embedded in it. Moreover, gas atmosphere in early furnaces is much less reducing and fayalite may not be able to form in this condition. The whole texture of slag could be quite inhomogeneous and less vitrified (Craddock 1995, 126-127).[16]
These slags are usually nut-size and easily overlooked by excavator yet are the most crucial indicator and evidence of early copper smelting activities. Hauptmann (2007, 157)[8] listed several cases of this finding. For example, slag granules from Catal Huyuk (Anatolia) are usually thought to be the first evidence of copper ore smelting.
Conclusion
Slag analysis can always be a useful tool for archaeologist in reconstructing ancient metal production process and detecting ancient human activities in this process. Primarily, through macro analysis, slag should be catalogued to facilitate sample selection and production scale may also be estimated. Then, chemical and mineralogical composition of slag can be investigated through micro analysis. Various parameters of metal production process can be estimated. Measuring of lead isotope would link slag with both ores and metal objects. Physical dating is quite useful in determining the precise age of slag. Finally, early copper smelting slag is always worth more attention though it may look quite different from normal slags.
See also
References
- ^ a b c d Bachmann, H.G., 1982, The Identification of Slags from Archaeological Sites, London: Institute of Archaeology
- ^ Maldonado, B., Rehren, Th., 2009, Early Copper Smelting at Itziparátzico, Mexico, Journal of Archaeological Science 36, 1998-2006
- ^ Moore, J.J., 1990, Chemical Metallurgy Second Edition, Oxford: Butterworth-Heinemann Ltd
- ^ Thornton, C.P., Rehren, Th., Pigott, V.C., 2009, The production of speiss (iron arsenide) during the Early Bronze in Iran, Journal of Archaeological Science 36, 308-316
- ^ Craddock, P. T. 1995. Early Metal Mining and Production. Edinburgh: Edinburgh University Press.
- ^ Archaeometallurgy. 2001. Centre for Archaeology Guidelines [Brochure]. Wiltshire: English Heritage.
- ^ Tumiati, S., Casartelli, P., Mambretti, A., Martin, S., Frizzo, P., and Rottoli, M., 2005, The ancient mine of Servette (Saint-Marcel, Cal d’Aosta, Western Italian Alps): a mineralogical, metallurgical and charcoal analysis of furnace slags, Archaeometry, 47, 317–40.
- ^ a b c d e f g Hauptmann, A., 2007, The archaeo-metallurgy of Copper:Evidence from Faynan, Jordan, Berlin; Heidelberg; New York: Springer
- ^ a b Craddock, P., 1989, The Scientific Investigation of Early Mining and Smelting, In Henderson, J.(ed), Scientific Analysis in Archaeology, Oxford : Oxford University Committee for Archaeology, Institute of Archaeology ; Los Angeles, Calif. : UCLA Institute of Archaeology ; Oxford : distributed by Oxbow Books, 178-212
- ^ Chiarantini, L, Benvenuti M., Costagliola, P., Fedi, M.E., Guideri, S., Romualdi, A., 2009, Copper production at Baratti (Populonia, southern Tuscany) in the early Etruscan period (9th–8th centuries BC), Journal of Archaeological Science 36, 1626-1636
- ^ The lower the Kv is, the higher the viscosity is.
- ^ Donaldson, C. H., 1976, An experimental investigation of olivine morphology, Contributions to Mineralogy and Petrology, 57, 187–95.
- ^ Ettler, v., Ervinkal, R.C., Johan, Z.., 2009, Mineralogy of Medieval Slags from Lead and Silver Smelting (Bohutín, Príbram district, Czech Republic): Towards Estimation of Historical Smelting Conditions, Archaeometry 51/6, 987-1007
- ^ Stos-Gale, Z,A., 1989, Lead Isotope Studies of Metals and the Metal Trade in the Bronze Age Mediterranean, In Henderson, J.(ed), Scientific Analysis in Archaeology, Oxford : Oxford University Committee for Archaeology, Institute of Archaeology ; Los Angeles, Calif. : UCLA Institute of Archaeology ; Oxford : distributed by Oxbow Books, 274-301
- ^ Haustein, M., Roewer, G., Krbetschek, M.R., 2003, Dating Archaeometallurgical slags using thermoluminescence, Archaeometry 45/3, 519-530
- ^ Craddock, P., 1995, Early Metal Mining and Production, Edinburgh: Edinburgh University Press Ltd