Abstract

Detailed petrographical analysis of mineralised samples from the Early Iron Age mine site at Campolungo show them to be vein quartz with abundant chalcopyrite and minor amounts of pyrite and trace amounts of fine-grained electrum. The restricted mineralogy and simple textures suggest that the mineralisation could be successfully beneficiated to give a copper-rich feed and that any copper metal smelted from that feed would have iron as its only impurity. The trace amounts of fine-grained gold cannot be successfully mechanically separated and so the samples are not gold ores. However, their presence may indicate that elsewhere at Campolungo there were richer or more coarse-grained, gold-bearing ores that could have been exploited.

Introduction

The Early Iron Age Silter di Campolungo copper mine lies in the Val Grigna south-east of Bienno in Lombardy, northern Italy. Research into the mine began in 1994 and is ongoing. Many of the initial findings, including a detailed description of the mine site, its geology and archaeology are given in Ancel et al (1998) but more especially in Tizzoni and Tizzoni (1999). Descriptions of the local and regional geology of the area and its mineralisation are given by Rodeghiero et al in Tizzoni and Tizzoni (1999).

At Campolungo metasediments of the Lower Permian Collio Sedimentary Formation dip 15-20^oN and strike 250^o. The sequence comprises thinly bedded, fine-grained, green-grey micaceous sandstone/siltstones interbedded with medium to coarse-grained sandstones. The sediments have been metamorphosed to lower greenschist facies and the finer grained lithologies have a pronounced slatey fabric. The rocks have been tectonised and so are cut by a northeast-southwest, northwest-southeast conjugate fault set. The northwest or more accurately the west northwest-trending (304^o) faults of this set are mineralised. It is these, and in particular a series of subparallel, 15-25cm wide, epigenetic, chalcopyrite-bearing quartz veins that were exploited during the early Iron Age when approximately 400 tonnes of 25% copper ore were extracted (Ancel et al 1999).

It is a well known irony that one of the most difficult materials to find and recognise at an abandoned mine site (of any age) is the ore or ores that were being exploited (Ixer, 1999; 2001). In the absence of in situ ore that clearly was about to be extracted all other mineralised matter needs to be interpreted carefully in order to avoid confusing sub-economic/non-ores with what was mined (Ixer, 2001). Spoil heaps, for example, by their very nature, comprise discarded wallrocks, gangue and subeconomic and non-economic mineralised samples but very little ore. However, in the absence of genuine ore, mineralised samples taken from a mine site have much to say about what that ore was like, notably in terms of its probable mineralogy and textures (but neither grade nor tonnage). Once a thorough understanding of the ore petrography (minerals and their textures) has been established then meaningful discussions on beneficiation and subsequent metal production are possible (Ixer, 2001).

The following detailed petrography of the copper-gold mineralisation and its associated country rocks is given as a contribution towards a greater understanding of the origin and exploitation of the mineralisation at Campolungo.

Sampling and methods of study

Five polished blocks, four polished thin sections and two standard thin sections were prepared after visual inspection of a 10kg collection of typical spoil material collected by Prof. Marco Tizzoni. They were chosen to represent vein-style mineralisation and its immediately adjacent wallrocks. Blocks were made from coarse-grained chalcopyrite samples and polished thin sections from the mineralisation-country rock junctions and two thin sections of barren quartz-carbonate-bearing veins. The material was investigated in transmitted and reflected light, the latter using x8 air and x16 oil- and x40 oil-immersion lenses. All opaque phases greater than 2µm were identified optically, as were all non-opaque phases other than fine-grained clay minerals. However, only the more common secondary copper and iron minerals, many of which are post mining in age, were characterised.

Results

Country Rocks.

Macroscopic descriptions.
Pale to dark green argillites/ phyllites are fissile and the finer lithologies show a good cleavage. They have been heavily tectonised and are cut by a number of generations of quartz veins some of which have been folded. There is much barren, milky-white quartz.

Microscopic descriptions.
The wallrocks are fine-grained, metamorphosed siliciclastics, some of which have been silicified. The angular nature of the mineral grains and the composition of the rock clasts suggest that some of these rocks were originally volcanosedimentary perhaps crystal-lithic tuffs. The main wallrock lithologies have a strong phyllitic fabric, shown by alternating quartz-feldspar- and muscovite-sericite-TiO₂ -rich layers and by a pervasive, subparallel alignment of the silicate, phyllosilicate and rock clasts. A later fabric delineated by muscovite, chlorite and thin quartz veins cuts the main fabric at a high angle, suggesting there has been more than one generation of deformation.

Petrographically the rocks comprise angular to subangular rock clasts with lesser amounts of quartz and feldspar accompanied by lath-shaped phyllosilicates set within a very fine ?chloritic, or less commonly sericitic, matrix. Angular quartz showing straight and undulose extinction; lath-shaped, zoned, polysynthetically twinned, calcium-poor plagioclase; large, untwinned or simply twinned feldspars including perthite and microcline lie with their long axes along the main fabric of the rock. The feldspars vary in size and are unaltered to slightly turbid, others show replacement by white mica and ?chlorite along their cleavage planes or crystal edges and locally the replacement is intense.

Most of the rock clasts are very fine-grained ?chlorite, difficult to characterise and are cut by quartz-sericite-albite-chalcedony veinlets. Volcanic and volcaniclastic clasts are common and include micro- or fine-grained porphyritic rocks with quartz, plagioclase and untwinned feldspar phenocrysts; others display faint volcanic textures including spherules, silicified glass sherds and fiamme. Less common igneous rocks include coarse-grained quartz-feldspar or potassium feldspar-plagioclase plutonics including cuneiform granite. Metamorphic rock clasts, although less abundant than the igneous ones, include quartz-muscovite and quartz-chlorite schists alongside lesser amounts of stretched quartz and metaquartzites. Siltstone/fine-grained sandstone clasts are very rare.

Abundant coarse-grained muscovite, minor amounts of coarse-grained chlorite and trace amounts of mixed muscovite-chlorite form lath-shaped crystals and largely define the fabrics within the wallrocks, especially their cleavage/fissility. Locally phyllosilicate laths are deformed around rock clasts or are kinked, but essentially they lie in two main directions, one defining the main fabric of the rock, the other at a high angle to the main fabric.

Detrital heavy minerals are widely distributed but only occur in trace amounts. Dark green to green-brown, 20 - 120µm diameter crystals of tourmaline, euhedral to subhedral, 10-80µm long, zoned zircon crystals, sphene and apatite are present, as are single crystals of 20µm diameter chromite and 10µm long ilmenite. Graphite wisps and laths up to 120 x 10µm in size lie parallel to phyllosilicate cleavage planes and show sygmoidal or crumpled textures. Haematite is rare, forming very fine, up to 1µm long, acicular crystals within 20-80µm diameter clasts or single crystals up to 60µm in diameter. Both graphite and irregular haematite are associated with 10µm long TiO₂. Detrital TiO₂ grains include 20 - 150µm diameter leucoxene, some of which is broken and rare, 30 - 80µm diameter, spongy TiO₂ and rare rock clasts, up to 200µm in diameter, comprising very fine-grained TiO₂-haematite intergrowths. Orange, equant or broken, lath-shaped crystals, 5 - 100µm in size and 120µm long, dark brown TiO₂ are also probably detrital.

The rocks have suffered lower greenschist metamorphism with the growth of albite, white mica, chlorite, TiO₂ minerals and quartz. Neomorphic, simply twinned albite rims and replaces detrital plagioclase or forms thin, quartz-albite veinlets or knots of radiating crystals infilling void spaces including tension gashes. White mica, both as muscovite and sericite, replaces earlier silicates or forms veinlets often associated with quartz. Locally fine-grained sericite is the main matrix to feldspathic clasts in the wallrocks but very fine-grained chlorite is the main matrix. TiO₂ minerals are widespread and dominate the opaque assemblage and most are authigenic. Small, 5 - 10mm diameter, colourless to pale yellow TiO₂ is common and includes octahedrite crystals; larger crystals, 20 - 150µm but up to 200µm long, are tabular and are broken. Other neomorphic TiO₂ includes clusters of 5 - 20mm diameter crystals and 20 - 150µm long laths; these include pale blue TiO₂ that overgrows earlier yellow to orange TiO₂ grains.

The wallrocks have been heavily silicified and hence carry numerous generations of quartz veinlets, many associated with white mica, that are parallel, or at high angles, to the main fabric of the rock. Much quartz infills void spaces including tension gashes.

Biotite, amphibole, epidote and base metal sulphides, notably pyrite, chalcopyrite, sphalerite and galena were not recognised from the wallrocks at Campolungo. Indeed the wallrocks were barren of any significant amounts of opaque minerals other than TiO₂.

Mineralised samples.

Macroscopic descriptions.
Most of the mineralised specimens are very similar and comprise angular to sub-rounded grey-green metasiltstones/metasandstones and tabular, green-grey, slatey phyllites cemented by numerous generations of vuggy quartz. Although minor amounts of chalcopyrite are present infilling small voids in the country rocks, most is associated with cross-cutting milky-white quartz as random knots and bunches of chalcopyrite or as thin chalcopyrite stringers lying parallel to the vein walls. Limonite-stained quartz and coarse-grained, pale pink dolomite, both of which are intergrown with milky quartz, are devoid of significant sulphide mineralisation. Much of the dolomite is weathered and limonite-stained.

Post-mining oxidation of the sulphides is widespread and much of the spoil material is stained brown, green or blue. Chalcopyrite alters to limonite accompanied by radiating clusters of malachite. Elsewhere a variety of green and blue-green secondary copper carbonates and sulphates infill void spaces or form thin coatings and veinlets associated with pale-coloured gypsum.

Microscopic descriptions.
The primary ore is simple, comprising quartz (with minor dolomite) veins carrying chalcopyrite accompanied by small amounts of pyrite and trace amounts of tetrahedrite group minerals and sphalerite. Pale-coloured native gold/electrum is present. A wide range of secondary copper sulphides have replaced chalcopyrite including digenite, pale blue 'chalcocite', covelline and spionkopite. Limonite, green and blue copper minerals, including malachite, azurite and members of the langite/brochantite group are abundant, late and typical of the supragossan zone (Ixer, 1999)

Early vein/segregation quartz cementing shattered wallrocks is coarse-grained, shows strained extinction, growth zones and has abundant fluid inclusions/fine-grained 'dust'. It is associated with dolomite and minor amounts of white mica but is barren of mineralisation. This generation of quartz is cut by grain size reduction zones or thin, clear, inclusion-free quartz veins. The same clear quartz generation lies immediately adjacent to chalcopyrite.

Coarse-grained, twinned dolomite infills void spaces between large quartz crystals. It is inclusion-free and devoid of sulphides but shows limonite- and wad-staining along cleavage planes. Neither barite nor fluorite were recognised.

Pyrite (FeS₂), as 5-100µm diameter, euhedral, unzoned crystals, shows little alteration and commonly is enclosed within chalcopyrite as are very rare 20µm diameter, spongy marcasite (FeS₂) crystals. Locally pyrite is almost totally pseudomorphed by pale blue limonite that is paler and has a higher reflectance than limonite replacing chalcopyrite.

Chalcopyrite (CuFeS₂) is the most abundant sulphide forming veinlets or centimetre-wide, irregular clusters/bunches in quartz. Many larger chalcopyrite masses have 1-10µm diameter satellite chalcopyrite grains associated with trace amounts of a 10µm diameter tetrahedrite group mineral ((CuFeZn)₁₂(SbAs)₄S₁₃) in the clear quartz surrounding them. Chalcopyrite encloses euhedral pyrite and trace amounts of 60µm diameter, pale brown sphalerite (ZnS). Most chalcopyrite shows some degree of alteration/weathering. The alteration sequence is the normal one of chalcopyrite altering sequentially to copper sulphides followed by limonite and finally malachite. Relict chalcopyrite up to 80µm in diameter is present in limonite.

Pale yellow to very pale yellow native gold/electrum (AuAg) is present as 1-10µm long grains infilling fractures in quartz or as 1-2µm diameter grains within limonite or covelline replacing chalcopyrite. This confirms the earlier visual and SEM identification of electrum by Rodeghiero et al (1999).

The secondary mineral assemblage is varied but much is probably post-mining in age. Chalcopyrite alters along cleavage planes, fractures or grain boundaries to 5-20µm wide, pale blue chalcocite (Cu₂S) or 5-60µm wide, deep blue digenite (Cu₉S₅). Elsewhere, small, 1-40µm diameter, pale blue 'chalcocite' or very white copper sulphide, some of which is anisotropic, are present within malachite crystals or form 5-10µm diameter crystals intergrown with chalcopyrite in quartz. Both covelline (CuS) and spionkopite (Cu₃₉S₂₈) replace chalcopyrite directly as 10-20µm wide rims but elsewhere spionkopite forms 5-10µm long, lath-like crystals intergrown with, and replacing, digenite.

Very minor amounts of a pale lavender-grey, 10-40µm diameter, highly anisotropic phase with blood red internal reflections, is visually identified as cuprite (Cu₂O). It is enclosed as 60-80µm diameter crystals within limonite itself within chalcopyrite or forms the centre of limonite veinlets associated with malachite.

Green malachite (Cu₂(CO₃)(OH)₂ replaces chalcopyrite. Elsewhere, thin, 5-10µm wide malachite veinlets cross-cut quartz; malachite mosaics greater than 200µm in diameter comprise 20-40µm diameter crystals and 100µm long, radiating, tabular crystals infill vughs. Deep blue azurite (Cu₃(CO₃)₂(OH)₂ is less common than malachite and forms thin 10-20µm wide veinlets locally enclosed within malachite.

Limonite replaces chalcopyrite initially along 111 cleavage planes so giving a distinctive, lattice-like relict texture of 2µm long 'grains'. A later generation of botryoidal limonite infills void spaces up to 200µm across or forms thin, up to 20µm wide, cross-cutting veins associated with malachite as does 5 - 40µm diameter, botryoidal haematite.

Very pale green, slightly pleochroic, spherulitic or radiating brochantite (Cu₄(SO₄)(OH)₆) forms thin films enclosing quartz crystals or lies along cleavage planes in feldspar crystals. Most brochantite, however, infills voids, including small tension gashes. Locally it is associated with jarosite but mainly with limonite. Many voids have a thin limonite rim followed by brochantite with high and then low birefringence and finally gypsum. Gypsum (CaSO₄.2H₂O) with crenulated extinction is associated with malachite, limonite and brochantite as part of the alteration of chalcopyrite. It forms cross-cutting veinlets in quartz and is often the last supergene mineral to infill void spaces.

Genesis of the mineralisation.

Rodeghiero et al (1999) suggest that the mineralisation may be associated with Permian igneous activity or alternatively be part of a Triassic hydrothermal event. The presence of trace amounts of gold suggests a third possibility, that the mineralisation was formed by regional metamorphic fluids scavenging metals from the country rocks and later precipitating them into veins. This suggestion is partially supported by the lack of any detrital iron-titanium oxide minerals or haematite in the wallrocks showing that fluids have removed iron and perhaps other metals from them. However, the simple mineralogy and lack of any fluid inclusion data from the quartz gangue makes it very difficult to say more about the age of the mineralisation or the origin of the mineralising fluids other than the veins are epigenetic, hydrothermal and younger than the enclosing Lower Permian Collio Formation metaclastics.

Campolungo mineralisation as ore.

As copper ore
This microscopical study confirms the published mineralogical descriptions (Rodeghiero et al., 1999). The primary mineralisation is very simple, comprising the copper-iron sulphide chalcopyrite plus minor amounts of the iron sulphides pyrite and marcasite, and trace amounts of base metals found in sphalerite and tetrahedrite/tennantite. The secondary mineral assemblage, although more mineralogically complex, remains chemically simple comprising copper and iron sulphides, oxides, hydroxides, carbonates and sulphates.

The almost monomineralic nature of the mineralisation and lack of complex mineral textures/intergrowths between the ore minerals and the gangue suggest that the copper mineralisation at Campolungo was a copper ore. Also it suggests that beneficiation to produce a copper-rich "smelter-charge" would be simple and effective, requiring simple crushing and sorting/separation. It is worth noting that any copper metal produced solely from this mine would be free of significant trace metals other than iron.

As gold ore.
Although no gold is visible to the naked eye, trace amounts of very fine-grained (10µm) electrum are associated with the alteration of chalcopyrite or are present within fractures in vein quartz. Liberation of gold of this size would have required fine crushing followed by very fine grinding prior to any subsequent attempts at mechanical separation and these would have had little chance of success (R. Chapman pers. comm.). Indeed, today gold of this size is not mechanically, but chemically, recovered.

Despite the samples not being gold ore, they show that discrete gold is present at Campolungo, hinting that there may be, or may have been, exploitable gold in some of the primary or secondary copper ores.

Characteristically very low-grade gold ores suffer from the so-called 'nugget effect' namely that the gold is not evenly distributed in the primary ores but occurs in very small areas of high concentration (the 'nuggets') amid large volumes of barren ore. The existence of gold-rich primary ores can only be demonstrated by a major programme of chemical analyses of bulk samples followed by very extensive mineralogical examination of any gold-rich samples.

However, gold is not only present in the primary ores in gold-bearing deposits but is also associated with secondary ores. In particular gold will be enriched within any oxidised, iron-rich gossan zone that overlies gold-bearing primary ore. Unfortunately at Campolungo the development of a gossan zone was weak. Although there is much copper-staining it is superficial and probably post-mining in age. That and the paucity of pyrite in the primary mineralisation suggest that any gossan was small, for it is the initial oxidation of pyrite that initiates gossan formation and without much pyrite little happens.

Finally, rivers draining gold-bearing mineralisation often concentrate enough eroded gold that a gold placer is created. These placers form when gold and other heavy minerals become trapped in the river sediments in sufficient quantities that they become exploitable. Even very low-grade gold mineralisation can produce exploitable placers and since the mineralisation and mine at Campolungo are drained by a river the former existence of a worked, downstream placer or placers is worth exploring.

Acknowledgements.

Dr. Marco Tizzoni is thanked for introducing me to the ores of Campolungo and for supplying the rock samples and background material (including his unpublished notes) to this mine. Dr Rob Chapman is thanked for useful discussions and advice on the beneficiation of these ores. Rosie Ixer is thanked for preparing this manuscript with her usual care.

References.

B. Ancel, M. Cottet, E Kammenthaler, D. Morin, P. Oppizzi and M. Tizzoni. 1998. La mine de cuivre de Campolungo (BS): une exploitation révelutionnaire de l'Age du Fer. Premières hypothèses et observations. Notizie Archeologiche Bergomensi, 6, 163 - 183.

R.A. Ixer. 1999. The Role of Ore Geology and Ores in the Archaeological Provenancing of Metals. In: 'Metals in Antiquity'. S.M.Young, M.Pollard, P.Budd and R.A.Ixer (eds). British Archaeological Reports, 792, 43-52

R. A. Ixer. 2001. An assessment of copper mineralisation from the Great Orme Mine, Llandudno, North Wales, as ore in the Bronze Age. Proceedings of the Yorkshire Geological Society, 53, 213-219.

F. Rodeghiero, S. de Donatis and M. Moroni. 1999. Lineamenti geologici e giacimentologici del territorio di Bienno in C.C.Tizzoni and M.Tizzoni (eds). La Miniera Perduta, Cinque anni di ricerche archeometallurgiche nel territorio di Bienno.

C.C.Tizzoni and M.Tizzoni. 1999. La Miniera Perduta, Cinque anni di ricerche archeometallurgiche nel territorio di Bienno.