What Mössbauer Spectroscopy Can Tell us About Precious Metal Working in Celtic Times*
R. Gebhard1, G. Große2, G.Lehrberger2, J. Riederer3, F.E. Wagner2, U. Wagner2
1 Prähistorische Staatssammlung, D-80535 München, 2 Technische Universität München, D-85747 Garching, 3 Rathgen-Forschungslabor, D-14059 Berlin, Germany
Abstract: In Celtic coin moulds iron-bearing components differ from the top to the bottom, showing that the top has been heated in a strongly reducing atmosphere to about 1200 °C, while the bottom was mainly oxidized below 700 °C. From this information the procedure for making coin blanks in Celtic times can be reconstructed.
During its heyday (200 - 40 B.C.) Manching, a Celtic Oppidum south of the Danube in Bavaria was the site of highly advanced metal working techniques. The production of iron tools, arms, fibulae, and bronze jewelry was well developed. Gold was mainly used for coinage, while gold jewelry is scarce. We report on a study of technical ceramics used in Manching for the production of coin blanks. As methods of investigation, Mössbauer spectroscopy, thin section microscopy, and neutron activation analysis were applied. Moulds used for gold and silver coin production, tuyères from a furnace for gold and iron work, bricks from furnaces with the remainders of tube canals, and samples of mudplaster were studied with the aim to reconstruct metallurgical processes used by the Celts.
Coin moulds are flat ceramic plates measuring about 7 x 12 x 2 cm, with small circular or square pits on the upper side (fig.1). They were widely used in the Oppida of Central and Western Europe. The connection of these moulds with precious metal working is confirmed by gold and silver inclusions, which can be seen, often with the bare eye, on the surface of the coin moulds and on fragments of furnace linings [1,2].
Fig.1: Some coin moulds from the Oppidum of Manching.
A prominent feature of the coin moulds is their layer structure. The surface usually shows a noticeable degree of vitrification, the core is of gray colour, while the bottom is often reddish. The individual layers of ten coin moulds were studied by Mössbauer spectroscopy. Some results are shown in fig.2.
Fig. 2: Room temperature Mössbauer spectra of material from the surface, the core and the bottom part of coin moulds 19/607 (top) and 19/613 (middle). Spectra of material from the bottom of coin mould 19/613 after reduction at 800, 900 and 950 °C for 3 h are shown in the bottom row.
The Mössbauer spectra of the surface, the core and the bottom of the coin moulds can be interpreted consistently, though there are interesting differences between individual coin moulds. Coin mould 19/607 is somewhat uncharacteristic, as its bottom part is gray. The fractional area of Fe2+ in the RT Mössbauer spectra of its surface layer (fig.2, top row) is 72 % and does not vary much between the core and the bottom, indicating that this mould was strongly reduced from top to bottom. An interesting feature is the presence of iron metal in the surface layer and in the core. The metallic iron can easily be identified by its hyperfine field of 33 T. Its fractional area is 12 % for the core. In order to survive 2000 years of burial in the ground without oxidation, the iron particles must have been protected by a tight cover.
For coin mould 19/613, only the Mössbauer spectrum of the surface layer shows 5.6 % of iron metal (fig.2, middle row). In 19/613 the surface is totally reduced with a fractional area of 72 % Fe2+-silicates and an additional Fe2+-component which, according to its Mössbauer parameters, may be wustite. The core and the bottom are less strongly reduced, the Fe2+-species being only 22 and 12 %, respectively. An iron-bearing component with a large quadrupole splitting of 3.6 mm/s is observed in material from the core and from the bottom. This component was identified as almandine by its Mössbauer parameters  and by thin section microscopy. Almandine is a garnet stable during firing up to about 950 °C in oxidizing as well as reducing atmosphere. The presence or absence of almandine in different layers of a coin mould can therefore be used as an indicator for the temperature experienced by the mould in antiquity.
Table 1: Changes in fractional areas of iron-bearing species (magnetic, nonmagnetic Fe3+, nonmagnetic Fe2+) in the material from the bottom part of coin mould 19/613 after heating in air for 48 h to the given temperatures.
Material from the bottom of coin mould 19/613 was used for refiring experiments at different temperatures and in varying atmospheres. Up to 500 °C, refiring in air (Table 1) reflects the oxidation of the Fe2+-component, and increases the fractional area of the Fe3+-doublet. The important observation, however, is that the Mössbauer patterns still change on heating above 700 °C, indicating that the temperature experienced by the bottom part of the coin mould in antiquity had not exceeded 700 °C. Almandine disappears between 900 and 950 °C.
Samples from the bottom of coin mould 19/613 were also fired in a reducing environment by heating them with charcoal in closed vessels for 3 h. Spectra taken after reduction at 800, 900 and 950 °C are shown in the bottom row of fig.2. With 89 % of the iron in the form of Fe2+-components, reduction is almost complete already at 800 °C. Iron metal begins to form at 900 °C and is up to about 12 % after reduction at 950 °C. As during refiring in air, almandine disappears between 900 and 950 °C.
From these data the following conditions for the melting of coin blanks can be inferred. The coin moulds were used in a strongly reducing atmosphere. The difference in the temperature reached at the top and at the bottom of the coin moulds suggests that they were embedded in glowing charcoal and then air was blown in from above with bellows to briefly reach the temperature required for melting the metal pieces or dust into bulk ingots.
Fig.3: Temperatures recorded during the field firing of a replica furnace in the top, the core and the bottom of a replica coin mould. There is good agreement between temperatures measured in the core (8 mm from the top), and the temperatures simulated by solving the heat transfer equation.
Fig. 4: Comparison of Mössbauer spectra from the three layers of a replica coin mould used in the field firing experiment with spectra of samples of the same clay treated in the laboratory at appropriate temperatures in reducing (R) and oxidizing (O) atmospheres for 3 h.
To confirm this notion, field experiments were performed in a replica furnace using replica coin moulds with thermocouples embedded in the top, the core, and the bottom. The coin moulds were covered with red glowing charcoal and air was blown into the furnace. Pieces of gold and silver could be melted within three to five minutes. The temperatures recorded during the firing process (fig.3) confirm the conclusions drawn from the Mössbauer spectra. The observed temperature distribution in the coin moulds can be simulated by solving the one dimensional heat transfer equation using the measured temperatures at the top and at the bottom of the mould as boundary conditions and a value of l = 0.0034 W/(cm K) for the thermal conductivity (fig.3).
Finally, Mössbauer spectra were taken of the three layers of one of the replica coin moulds and of the same clay laboratory fired under appropriate conditions (fig. 4). No iron metal was formed during the field experiment, even though iron metal shows up in the laboratory emulation, but else the comparison confirms that the conditions are reducing on the top and in the core of the mould, but oxidizing at the bottom. As predicted from the results of the Mössbauer studies of the Celtic coin moulds, the procedure of melting the coin blanks lasted only a few minutes. In our field experiment 50 % of the starting metals were melted. Much higher yields might have been obtained by the experienced and skilled Celtic craftsmen. The final product, the rainbow cups, have been left for us to marvel at.
* This study was supported by the Bundesminister für Forschung und Technologie under Contract number 03WA9-TUM.
 Gebhard, R. - In: Prehistoric Gold in Europe. G. Morteani and J.P. Northover (edts.), Kluver Academic Publishers, Dordrecht, 1995, p261.
 Gebhard, R., G. Lehrberger, G. Morteani, Ch. Raub, F.E. Wagner, U. Wagner. - cf. , p.273.
 Wagner, F.E. and E. Murad. - Phys. Chem. Minerals, 14 (1987) 264.
Last update: 12.03.2004