The apparatus used during the experiment was a graphite tube furnace of an inhouse design, shown in Fig. 3. The furnace was kept under an argon atmosphere at 1.151.3 bar to ensure that neither the sample or the graphite was in contact with an oxidizing atmosphere in the hot zone. Two B-type thermocouples were used. The one inserted from the top regulated the temperature, while the second was used as an insurance. This furnace was used due to its special crucible insertion and withdrawal system. It allows the operator to insert and extract samples while the furnace is still running and kept inert. This keeps disturbances to the furnace atmosphere to a minimum, allowing for all the samples in one series to be subjected to as similar conditions as possible.

12 g of silicon pellets were put into 15 IG-110 graphite crucibles with dimensions 030 mm/22 x 45 mm and an inner height of 40 mm. The silicon pellets were premelted to minimize inconsistencies in contact area between metal and slag. This gave a smoother surface morphology with a contact area approximately equal to the crucible opening. The furnace was evacuated to < 0.3 mbar and refilled with 5.0 argon. Heating to 1600 °C was started after the internal furnace pressure

reached 1.2 bar, at a rate of 40 min. Each crucible was held in the hotzone for 7 min before quenching. The holding time at the chamber separation mechanism was 5 min for the first series, but was later increased to 7 min due to the withdrawal rods heat. 11 g of slag was then added to each crucible, with compositions listed in Table 1. Three different slags were used with five samples in each slag series, totaling 15 samples. Five time steps were chosen and they were 5, 10, 20, 30 and 180 min. The first four time steps measure the mass transfer kinetics, while the 180 min sample was expected to reach equilibrium concentration as proposed by Jakobsson [9]. The same procedure as used above was performed for each sample

Graphite tube furnace

Fig. 3 Graphite tube furnace

Table 1 Intended slag concentrations

Slag ID



AI2O3 [wt%]













set. To keep the experimental conditions as constant as possible each sample set was run on the same day and kept in a desiccator after cooling to avoid contamination.

The silicon used in this experiment had a purity of at least 8 N. Initial master slags were made from commercial oxides with a purity of > 99.5%. Homogeneity was achieved by mixing the oxides in different proportions, melting, quenching and crushing, with the last three steps repeated at least twice. The master slags contained approximately 0.1 wt% Fe2O3, with the total initial concentration of all other measured oxides > 0.1 wt%. By mixing the master slags at different proportions the desired slag compositions were obtained. These can be seen in Table 1. To ensure homogeneity the new slag mixtures were heated to 1873 K, kept there for 1 h, before quenched and crushed. The mass loss after heating was only measurable for the 180 min samples where the mass loss was close to 0.6% due to reactions with the crucible.

Before analysis the graphite crucible was removed and the slag and metal separated. Great care was taken to ensure that no metal was present in the slag samples and no slag in the metal samples. As to not contaminate the final product pieces were discarded rather than kept. Over 95 wt% of the metal and at least 60 wt% of the slag was kept after separation. After separation the metal samples were crushed to a fine powder using a tungsten carbide disk mill. The slag samples they were crushed with a polytetrafluoroethylene ball mill.

Both metal and slag were analyzed by ICP-MS to find the concentration of Ca and Al in the metal, and to check for contaminants in the slag. This was done by the HR-ICP-MS lab at the Department of Chemistry at NTNU, with the instrument type Element 2 from Thermo Electronics. Two sub-samples, between 25-45 mg, were taken from each sample and dissolved in a 1.5 mL ultra pure 68% HNO3 + 0.5 mL 40% HF mixture. After dissolution the samples are diluted, with deionized water, to a final volume between 216-220 mL, giving 0.1 M HNO3 and 0.23% v/v HF. Three samples of NIST metallurgical silicon standard reference, three standard slag references and three blank samples were analyzed together with the sub samples.

< Prev   CONTENTS   Source   Next >