Summary

The crystallisation of gibbsite, $\gamma$-Al(OH)₃, is an important step in the production of alumina, Al₂O₃. The latter is the material from which aluminium is extracted. Nowadays, the production of alumina is most widely achieved via the Bayer process. In this process, gibbsite is crystallised from seeded sodium aluminate solution, NaAl(OH)₄, and is subsequently calcined to alumina. The quality of gibbsite determines the quality of alumina. For that reason, the particle size, the morphology, strength, amount of agglomeration of the gibbsite crystals and the incorporation of impurities are important parameters. Fundamental knowledge of the crystallisation process is necessary to optimise the crystallisation step and to improve the quality of the crystals. The results of a laboratory study of the crystallisation of gibbsite according to the Bayer process are described in this thesis.

Experiments on the crystallisation of gibbsite from unseeded sodium aluminate solutions reveal gibbsite crystals and agglomerates with various morphologies. Many of the crystals are twinned. The various kinds of twinning and the effects on the growth morphology and growth kinetics are discussed in chapter 2. It is shown that single non-twinned crystals are lozenge-shaped, mainly bounded by {001} top and {110} side faces. The relative occurrence of the various faces shows that the morphological importance (MI) of the faces follows the sequence MI$_{\{001\}}$$\gg$ MI$_{\{110\}}$ > MI$_{\{100\}}$ > MI$_{\{101\}}$$\approx$ MI$_{\{112\}}$. Single or multiple twinning, dislocations and the presence of impurities lead to the formation of other crystal shapes: large hexagonal plates and prisms. Most of the hexagonal plates are twinned sixfold on {110}, with {001} as basal faces and {100} as side faces. The twinning already occurs at the initial stage of nucleation. The combination of fast growing {100} faces and a nucleation mechanism at the reentrant corners results in large crystals with sizes up to 200 micrometers in diameter. The small prismatic crystals are formed when the lateral growth is inhibited, apparently by impurities. Because the lozenge-shaped gibbsite crystals are single crystalline, it is concluded that these crystals exhibit the basic morphology instead of the hexagons, normally considered as the basic morphology in the literature.

To understand the growth and interface properties of the three main types of gibbsite crystals, i.e. lozenges, hexagons and prisms, a study of the surface topography using optical and atomic force microscopy is done and described in chapter 3. Many of the lozenges show no dislocation sources. This suggests that the growth of these crystals proceeds exclusively by a 2D nucleation mechanism and subsequent step advancement. Lozenges which grow to somewhat thicker crystals show one or a few screw dislocations ending on the basal face. The second type of crystals, the sixfold twinned hexagons, has a complex surface topography as a result of many defects. The various growth features on the {001} faces are shown to be the result of defects in the crystals and inhomogeneities in their environment. The lowest steps observed are about 5 Å$\,$ high, which is equal to half of the unit cell dimension along c in accordance with the selection rules of the space group. 2D nucleation at the reentrant corners at the outcrops of twin planes is shown, which verifies the lateral growth mechanism for sixfold twinned crystals as suggested in chapter 2. Crystals of the third major crystal morphology found, prisms, also exhibit many defects. Mosaicity is observed and related to the presence of misaligned crystallites or impurities. It is shown that most gibbsite crystals contain many defects and that each type of morphology reveals a different surface structure.

In-situ optical microscopy has been used to study the growth rates of the individual faces of the different types of gibbsite single crystals growing from aqueous sodium aluminate solutions. The growth rate measurements are presented in chapter 4. The growth rates are measured for the {100} and {001} faces in case of twinned hexagons and {110} faces for single crystalline lozenges. To interpret these results, a definition of the driving force adapted for gibbsite crystallisation from caustic aluminate solution is derived in this chapter. Furthermore, besides the well-known crystal growth models for birth and spread and spiral growth, a new analytical model for enhanced 2D nucleation at the intersection lines of contacting crystals is introduced. Although the crystals were grown under the same external conditions, a significant growth rate dispersion was observed for crystallographically equivalent faces of crystals of the same type. The average growth rates measured as a function of the driving force are fitted with the growth rate equations for the various growth mechanisms. It is shown that a birth and spread-type growth rate equation as well as the equation for contact nucleation growth derived in this chapter can be used to describe the growth rate of gibbsite under various growth conditions for all crystallographic directions. This is in accordance with the surface topography studies reported in chapter 3.

In chapter 5 the effect of a wide range of growth conditions and inorganic impurity additions on the growth morphology of gibbsite crystals is characterised using scanning electron microscopy. These observations lead to the description of a general morphology evolution during the crystallisation process. Crystallisation starts with the formation of thin, rounded hexagons and faceted lozenges, which upon further growth develop into faceted plates and blocks with well-formed basal, prismatic and chamfered faces. Finally, large blocks with many defects and stress regions are formed. This morphology evolution has only a weak dependence on the growth conditions. Changing the external conditions, i.e. increasing driving force and/or caustic concentrations, just leads to a faster development of the morphology evolution. Moreover, increasing the driving force or the caustic concentration leads to larger crystals. The influence of small amounts of inorganic impurities on the growth of gibbsite crystals turns out to be negligible. In contrast, gibbsite crystals grown from potassium aluminate solutions show a morphology which is elongated along the c-axis. This indicates that the alkali ions of the solution have a major influence on the morphology of gibbsite crystals.

Experiments show that despite the fact that the structure of gibbsite is pseudohexagonal, a lozenge-shaped morphology is energetically the most favoured one. The theoretical morphology of gibbsite on the basis of a detailed connected net analysis is the subject of chapter 6. This method relies on determining crystal faces parallel to connected nets in the crystal structure, followed by calculation of the edge energy of 2D nuclei for each face. By subscribing the highest morphological importance to the faces with the highest value of edge (free) energy, i.e. the highest nucleation barrier, the growth morphology is obtained. Growth units are defined as Al(OH)_6/2, specifically, an Al atom with half of the six surrounding OH groups. It shows that 251 different connected nets corresponding with 12 crystal faces can be identified. Monte Carlo simulations based on 2D nucleation growth of these faces show that their MI follows: MI$_{\{002\}}$$\gg$ MI$_{\{110\}}$> MI$_{\{200\}}$> MI$_{\{112\},\{11\overline{2}\}}$$\approx$ MI$_{\{101\},\{10\overline{1}\}}$$\gg$ MI$_{\{011\}}$$\approx$ MI$_{\{211\},\{21\overline{1}\}}$. This corresponds well with the experimental observations. In this chapter, it is shown, with the complete connected net analysis of gibbsite, that the growth rate of faces is no longer proportional to the attachment energy, but is determined by the actual edge (free) energy together with the relevant 2D growth mechanism.

In conclusion, gibbsite crystals have many different growth morphologies. It is fascinating that crystals with different morphologies grow simultaneously in the same batch. This study reveals that gibbsite crystals free from dislocations mainly expand laterally, resulting in an ultra thin lozenge-shaped morphology. Here, growth probably proceeds by a 2D nucleation mechanism. Furthermore, this study indicates that the variation in growth morphology of gibbsite crystals is primarily the result of intrinsic properties of the material, like twinning and accumulation of growth imperfections, but also the continuous formation of nuclei during crystallisation, agglomeration and probably restructuring of the solution. This also explains the dispersion in growth rate of the crystals. External conditions, such as driving force, caustic concentration and impurities, seem to have only a small effect on the growth mechanism of gibbsite crystallisation. There is still an open question why the lateral growth rate of prisms is lower than that of plates and lozenges. This and several other aspects of gibbsite crystal growth are still shrouded in mystery and need clarification in the future.