An evaluation of the binder matrix in prebaked carbon anodes used for aluminium production

Abstract

The electrolytic production of aluminium by the Hall—Heroult process employs carbon anodes which are consumed during operation. As well as the electrochemical reaction resulting in the deposition of metal, carbon is non—profitably consumed by several other mechanisms. This results in "excess" consumption which can account for between 25-45 % of the anode losses. There is consequently a considerable financial incentive to reduce the amount of carbon material lost by these means. Two important mechanisms responsible for excess carbon consumption are dusting, due to selective burning of the binder matrix component within the anode block, and thermal shock, which can occur when the anodes are set in the electrolysis cells. Thermal shocked anodes can be recovered and affect gross consumption figures whereas dusting losses represent a net carbon loss. The term "binder matrix" has been previously used to describe the intimate mixture of pitch and coke dust that acts to bond together the larger coke grains. One possibility to reduce anode excess consumption is to minimise dusting losses by improving the burning behaviour of the binder matrix. At the same time it is beneficial to improve the binder matrix mechanical properties as this can lead to an increase in the thermal shock resistance of the anode blocks. Such an evaluation of the binder matrix quality is best carried out by considering the separate influences of its two constituent materials — binder pitch and petroleum coke dust. For this study this has involved: (1) investigating the importance of the basic pitch parameters (purity, structure and porosity) on binder matrix burning behaviour. (2) investigating the importance of the dust granulometry on binder matrix burning and thermal shock behaviour. It was firstly necessary to develop a technique that allowed the production of representative, binder matrix electrodes using only the petroleum coke dust and binder pitch. This allows the effects of any variations in either of the two component fractions to be more readily measured than with an anode sample, due to the absence of coke grains and butts particles. A traditional bench—scale approach was used with a specially adapted mixer. Typically, the processing conditions involved preheating of the coke dust and pitch to 180 °C, mixing for ten minutes at 180 °C, pressing the green paste at 61.1 MPa for one minute and baking the electrodes at 1100 °C for five hours. The most critical parameter was the coke dust/pitch ratio which was found to be dependent on the pitch quinoline insolubles content. After baking appropriate sample properties were measured using standard apparatus. A case study is then presented where binder matrix electrodes were used as a means of pitch selection in an anode plant, with specific emphasis being placed on airburn. Measurements on the binder matrix composites prepared from a standard dust and each of ten, coal—tar pitches showed that two samples were inferior to the remaining population. It was recommended that the quantity of these two pitches be reduced for future operations. The reasons for the poorer performance of these two samples have been studied as part of the programme looking at pitch performance. Binder matrix electrodes were prepared from a total population of eighteen, industrial, coal—tar pitches and the influence of pitch impurities, pitch coke crystallinity, pitch coke optical texture and porosity on burning behaviour were investigated. The major findings are as follows: (1) excess sodium is the most significant impurity for catalysing airburn and both excess sodium and calcium are significant catalysts of carboxy reactivity. (2) increasing the baking temperature improves pitch coke crystallinity and binder matrix burning behaviour with significant improvements being observed above 1150 °C for impure pitches. (3) high QI pitches lead to an isotropic structure that is relatively sodium insensitive whereas low QI pitches result in more anisotropic coke structures that show a much greater deterioration in burning behaviour as excess sodium levels increase. (4) pitch nature is not able to modify the pore size distribution of the baked binder matrix when manufacturing conditions, including the pitch/dust ratio, are optimised. To evaluate the importance of dust granulometry on binder matrix performance a worldwide dust survey was undertaken where all known prebaked anode manufacturers were invited to send samples of their dust fractions. This established the range of dust targets used within the industry and gave important information on the types and frequency of dust control. Binder matrix electrodes were then prepared for a similar range of dust granulometries at the appropriate pitch/dust ratio and their performance evaluated and compared with bench—scale results. The major findings are as follows: (1) the best combination of dust parameters to give good burning behaviour are a dust fineness between 3000-4000 Blaine and a dust content (in the anode) between 30-40 %. (2) thermal shock problems may be alleviated by reducing the dust fineness, the dust content and consequently the pitch content. As these two conclusions are incompatible any recommendation on an optimum dust content or granulometry will depend on the situation within a given plant.