In the current work, the metallothermic reduction of natural quartz by magnesium has been studied at 1373 K under different reaction conditions, i.e. quartz type, quartz particle size, Mg:SiO2 mole ratio and reaction time. The microstructure of reaction products was studied to illustrate the reaction progression through scanning and transmission electron microscopy techniques. X-ray diffraction analysis with Rietveld phase quantification was used to calculate the change in the amount of phases at different reaction conditions. The results showed that the Mg:SiO2 mole ratio strongly affects reaction mechanism and product characteristics such as phase content and microstructure. At lower Mg:SiO2 mole ratios, the reaction rate is fast at the beginning and the formation of a product layer consisting of different phases such as MgO, Si, Mg2Si, Mg2SiO4 and MgSiO3 around quartz particles limits the Mg diffusion. This phenomenon is more noticeable for larger quartz particle sizes where Mg should diffuse longer distance towards the quartz core to react with it. At higher Mg:SiO2 mole ratios, a significant amount of Si–Mg liquid alloy is formed during reaction where the high mobility of Mg in this liquid phase and cracking of quartz particles result in significantly higher reaction rate. Here the formation of intermediate phases is not significant and the products would be the mixture of MgO, Mg2Si, and either Si or Mg phases.

In this work, the kinetics of natural quartz reduction by Mg to produce either Si or Mg2Si was studied through quantitative phase analysis. Reduction reaction experiments were performed at various temperatures, reaction times and Mg to SiO2 mole ratios of 2 and 4. Rietveld refinement of X-ray diffraction patterns was used to obtain phase distributions in the reacted samples. SEM and EPMA examinations were performed to evaluate the microstructural change during reduction. The results indicated that the reduction reaction rate was slower at a mole ratio of 2 than 4 at the same temperature, as illustrated by the total amount of Si formed (the percent of Si that is reduced to either Si or Mg2Si to total amount of Si) being 59% and 75%, respectively, after 240 min reaction time for mole ratios of 2 and 4. At the mole ratio of 4, the reaction rate was strongly dependent on the reaction temperature, where SiO2 was completely reduced after 20 min at 1273 K. At the lower temperatures of 1173 and 1073 K, total Si formed was 75% and 39%, respectively, after 240 min reaction time. The results of the current work show that Mg2Si can be produced through the magnesiothermic reduction of natural quartz with high yield. The obtained Mg2Si can be processed further to produce silane gas as a precursor to high purity Si. The combination of these two processes offers the potential for a more direct and low carbon method to produce Si with high purity.

Silica fume is an important byproduct from the silicon production process, with mainly concrete and refractory applications. In this work, the effect of different combustion gas atmospheres on the properties of silica fume has been investigated through small-scale experiments and a pilot-scale experiment. In the small-scale experiments, SiO gas was formed and reacted with four mixtures of nitrogen fixed at 79 vol %, oxygen, carbon dioxide, and carbon monoxide. No significant difference between the particulate matter (PM) formed was found when replacing O 2 with CO 2 using a holding temperature of 1700 °C, when measuring the specific surface area (SSA), particle size distribution, or carbon content. PM formed at a holding temperature of 1650 °C was found to have a significantly higher SSA of 32.9 m 2 /g compared to the SSA of 22.3 m 2 /g formed at 1700 °C with the same gas mixtures. Condensation products were only found when replacing O 2 with CO, where brown PM was formed through SiO reacting with SiO 2 , SiC, and Si. Using a pilot-scale setup, the combustion gas for the silicon process was altered by recirculating different ratios of flue gas back to the furnace hood. Silica fume samples were collected during several tapping cycles with varying flue gas recirculation (FGR) rates. The SSA was measured for the sampled PM and correlated to a selection of operational parameters. A significant negative correlation was found between the SSA of PM and the concentration of H 2 O and SO 2 in the off-gas, while a significant positive correlation was found between the SSA and the temperature of the off-gas, the concentration of dust in the off-gas, and the NO X generation during the process. These results indicate that the temperature and SiO concentration during combustion are more important than the CO 2 (or O 2 ) concentration in the combustion atmosphere. Using FGR to increase the CO 2 concentration in the off-gas should not change the silica fume particle size, if the H 2 O content is controlled.

Fundamental studies have been carried out experimentally and theoretically on the magnesiothermic reduction of silica with different Mg/SiO2 molar ratios (1–4) in the temperature range of 1073 to 1373 K with different reaction times (10–240 min). Due to the kinetic barriers occurring in metallothermic reductions, the equilibrium relations calculated by the well-known thermochemical software FactSage (version 8.2) and its databanks are not adequate to describe the experimental observations. The unreacted silica core encapsulated by the reduction products can be found in some parts of laboratory samples. However, other parts of samples show that the metallothermic reduction disappears almost completely. Some quartz particles are broken into fine pieces and form many tiny cracks. Magnesium reactants are able to infiltrate the core of silica particles via tiny fracture pathways, thereby enabling the reaction to occur almost completely. The traditional unreacted core model is thus inadequate to represent such complicated reaction schemes. In the present work, an attempt is made to apply a machine learning approach using hybrid datasets in order to describe complex magnesiothermic reductions. In addition to the experimental laboratory data, equilibrium relations calculated by the thermochemical database are also introduced as boundary conditions for the magnesiothermic reductions, assuming a sufficiently long reaction time. The physics-informed Gaussian process machine (GPM) is then developed and used to describe hybrid data, given its advantages when describing small datasets. A composite kernel for the GPM is specifically developed to mitigate the overfitting problems commonly encountered when using generic kernels. Training the physics-informed Gaussian process machine (GPM) with the hybrid dataset results in a regression score of 0.9665. The trained GPM is thus used to predict the effects of Mg-SiO2 mixtures, temperatures, and reaction times on the products of a magnesiothermic reduction, that have not been covered by experiments. Additional experimental validation indicates that the GPM works well for the interpolates of the observations.