15605 S5003130

6

f rom the fire to pour th* mol eon copper into a mould. The product, a cast har ingot, was anaTysod and found to contain    Iron, O.02% nickel,

0.0081 Łlnc and <10 ppm tin, lead, antimony, bismut h and arsenie; the rest

was copper,

A major difficulty in the experiments may be attrlbutod to the lack of an iron ore flux as the malachlte was not self-fluxlng. This error seems to have orlginated from the few misleading Chemical analyses of archaeological slag speclmens. Most of the 'slag* analyses were actually slagged fumace linlng and slagged tuyeres. This led Friede and Steel (1975, 223) to consider only calcium, sodium and potassium as possible flux additions and the importance of the iron flux was not recognized.

The experimenta1 slags were consequently rather viscous.

The excavations of Beno Rothenberg (1962) on the Timna sites in the years 1960-75 inspired the work started by Boydell (Tylecote 6 Boydell 1978) and continued by Ghaznavi (Tylecote et al.1977) and by Merkel (see beIow).

Two furnaces have been used; the large one, A (Fig. 2), which was 30 cm in diameter and the smaller one, B, which was only 20 cm diameter (Fig. 3). Most of the earlier work was carried out on the smaller furnace and pure copper oxide, ox±de and sulphide ores were smelted with roasting where reauired.

Experimentation started with the larger shaft furnace, called type A, based on archaeological evidence from Timna Slte 2 (Tylecote & Boydell 1978). A column of segmental firebricks, measuring 32 cm in diameter and 65 cm high, was built upon a rectangular platform and surrounded with a sand seal and insulation brick. The furnace was lined with a mixture of clay, sand and charcoal dust. Seven experiments were conducted with a single, lncllned tuyere. The tuyere was about 25 cm above the furnace bottom. A mullite tubę with a 25 mm diameter served as a tuyere; lt did not protrude into the furnace from the wali. After five experiments the tuyere diameter was reduced to 19 mm, which improved operation. Airflow rat es were varied between 200 and 370 litres/min.

However,these experiments were not successful and attention was now concentrated on the operation of a smaller bowl furnace, called Type B (Tylecote & Boydell 1978). Archaeological evidence from Timna Site 39b was used to select the furnace dimensions. Smaller firebricks formed the fumace column, complemented with the same type of sand seal and insulating brlcks. The bowl furnace measured 22 cm in diameter and 30 cm high; it used a single, lncllned tuyere 19 mm ln diameter. Usually the tuyere entered through the furnace wali at a height which was varied between 16 cm and 18 cm above the bottom. On the second experiment, the tuyere was placed vertlcally into the furnace 12 cm above the hearth. A total of thlrteen experiments were completed with this bowl furnace.

The first five experiments used copper oxide or copper metal plus hematlte flux, at various ratios. Airflows from 100 litres/min to 150 litres/min were tested. Preheat times were about 1 hour 30 minutes. Post-heat charcoal additions after the last ore charge, varied up to 2 hours 30 minutes. A temperaturę profile was obtained for experiment B3, and the lsotherms showed that the bottom temperatures were too Iow to melt the slag. At best the smeltlng products were copper prllls mixed with slag. Recovery of input copper ranged from 42% to 94.5%.

A second series of expcriments with tbe bowl furnace by Tylecote and Boydell (1978) used crushed (<u mm) nodular ore from Tirana. The optimum rat i o of ore to hcmatite flux war. found te be 1:2. In Ejcperlment 69 no hematite was added so as to demonstrate that the nodular ore was not self-fluxlng and the copper could not be scparatcd from the slag. Varlous rat los of charcoal fuel to mlxed ore and flint were tested; emplrically the ratlo of 1:2 was found superior. A fuel/ore ratlo of 1:3 was tried ln Experlment B7, but the lncreese ln ore caused the tuyere to błock with slag. The subsequent decrease ln temperaturę resulted ln poor separation of molten slag and copper.

Plano-convex copper lngots were not produced la sny of the experiments The copper prills contalned Iron and sulphur as the main impurities. Recovery of arsenie was about 50X of the lnput values. Tracę element analysis was madę on the copper prills from Experlments B12 and B13 by Tylecote and Boydell (1978) and llerkel (1977).

Tylecote and Boydell (1978) also reported on elght reflnlng experi-ments for the Impure smelted copper. Iron was the main lmpurlty which needed to be removed. Due to the difference in denslty and the immisci-billty between copper and iron. Iron floats to the top la a moltea mixture. Under laboratory conditlons, the Iron metal could be recovered from the top of the copper. Tylecote and Boydell (1978, 45) propose this process may be the source for early finds of iron objeets. However, under oxidi-zing conditlons the iron will form ferrous oxlde which forms a slag with the charcoal ash and crucible. It was demonstrated that remeltlng refines the Impure copper.

Ghaznavi (1976) completed further experimentation with the bowl furnace. A total of seventeen experiments were undertaken to smelt seven different copper ores and two artlflclal copper ores. The purpose of these experlments was to study tracę element partitioning durlng copper smelting under simulatlon conditlons.

The first four experlments were conducted uslng a single lncllned tuyere positioned about 11 cm above the bottom. The firebrlck furnace measured 23 cm in diameter and was 30 ca high. The airflow was ISO litres/ min, which was wlthln a proposed rangę for bag—bellows. Copper oxide ore from Timna was smelted only ln the second and third experiaents. Fuel/ ore plus flux rat los were tested at 1:1 and 1:2. Only copper prills were produced from the experlments at report ed efficlencles of 69X and 80.21.

The experiments with Timna ore were conducted with only one tuyere, not two as stated in Tylecote, et al.(1977). The analytical results for the smelted copper from Timna ore by Tylecote and Boydell (1978) and Ghaznavl (1976) are not that different. Thus conclusions based on compositional dlfferences relating to the nunber of tuyeres reąulre further experimenta-tlon.

An additlonal, second tuyere was placed at the furnace top for twelve subseąuent experiments uslng sulphlde ores by Ghaznavl (1976). This change was introduced, accordlng to the author, to inerease the burnlng of carbon monoxide in the exhaust gases.

The beneflt of the second tuyere, inereasing both airflow and temper-atures, was an improved copper and slag liquation. It inereased the temperaturę distributlon considerably and allowed the format ion of soae-thlng resembllng an ingot ratber than isolated prills of copper enveloped in slag. (Tylecote, Ghasnavi & Boydell 1977).