Crystal Growth Modifying Reagents; Nucleation Control Additives or Agglomeration Aids?

James A Counter1 1 Nalco Australia, 2 Anderson St., Botany, NSW, 2019, Australia

Keywords: precipitation, crystallisation, nucleation, agglomeration, precipitation additive

Abstract Crystal growth modifiers (CGM) are commonly used in the Bayer process to increase the average gibbsite particle size and improve the particle size distribution during crystallisation from supersaturated sodium aluminate solutions. With respect to the mechanism of action there has been much debate as to whether this result could be due to the additives aiding agglomeration or reducing the number of secondary nuclei. Presented here is work completed on both synthetic and real Bayer plant liquors, over various experimental conditions in order to study the mechanism of action of CGM additives. Crystallisation kinetics, secondary electron microscopy (SEM) and atomic force microscopy (AFM) data are collated to examine the impact that these surface active additives have on the precipitation process.

Introduction Crystal growth modifiers were introduced to the Bayer industry over 20 years ago [1] and have gained general acceptance as additional control tools within the crystallisation section of the Bayer process. These products have exhibited widespread utility under high-agglomeration and all-growth precipitation strategies. The effect of addition of CGM has lead many plants to a reduction in ultra fines, tougher particles, liquor productivity increases via either reduced fill temperatures or increased seed charges and alumina trihydrate coarsening along with many other benefits using doses ranging from 5-50 ppm. Three major crystallisation mechanisms, nucleation, growth and agglomeration of crystals are in action for producing gibbsite as the preferred phase. The yield of gibbsite is largely determined by nucleation and growth of particles, while agglomeration is responsible for producing a coarse crystal of commercial interest. The effect of optimising the Bayer plant precipitation parameters in each of these crystallisation scenarios is the most important part of the entire industrial process. The relationship between the operating parameters and gibbsite yield has been detailed in a number of publications over many years [2-6]. Optimum conditions for maximum yields do not always coincide with optimum product size specifications, demanded by the smelters. CGM additives have allowed Bayer plants around the world to operate under optimum conditions without product size compromise. Although there have been a number of recent publications postulating the CGM mechanism of action, [7-9] there still exists a degree of uncertainty as to the impact of these programs and their influence under various agglomeration and secondary

nucleation strategies practiced within the industry. The main thrust of the present studies is to examine the influence of a CGM on each of the precipitation scenarios in order to better understand the mechanism of action, leading to further development of new and improved products. This paper presents combined work of the impact of CGM on both synthetic and real Bayer plant liquors under specific crystallisation scenarios over a temperature range experienced in industry. Results from a number of experimental techniques are collated to answer the questions posed by industry and propose a general mechanism of action for these additives.

Experimental Synthetic Bayer Liquor Preparation Pure synthetic, supersaturated sodium aluminate solutions were used in this investigation. They were prepared from analytical grade and high purity grade reagents: aluminium trihydroxide, Al(OH)3, (C31 grade, 0.01% SiO2, 0.004% Fe2O3, 0.15% Ba2O; ALCOA, Arkansas USA); sodium hydroxide, (99.0% pure, 0.01% Si, Merck, Australia); Milli-Q water (surface tension 72.8 μNm-1 at 20ºC, specific conductivity < 0.5 μscm-1 and pH = 5.6). The synthetic liquors used in the crystallisation measurements were prepared in a 2.5 dm3 well sealed stainless steel digester immersed in an oil bath at 150ºC. A known mass of NaOH was dissolved in 1.3 dm3 of Milli-Q water in the digester with an agitation rate of 250 rpm. This was followed by slow addition of a known mass of the aluminium trihydroxide. After complete dissolution of 3 – 4 hours, the solution was transferred to a 2.0 dm3 volumetric flask and made up to the mark with Milli-Q water. Bayer Plant Liquor Preparation The Bayer plant liquor experiments were conducted using reconstituted plant spent liquor. Spent liquor is the term used in the Bayer process to describe the liquor after the final classification stage which returns back to digestion. A desired weight of spent liquor was measured into a stainless steel beaker and the volume was reduced by evaporation to approximately 30%. To this a set weight of Al(OH)3 solid was added and the mixture stirred until it was dissolved. This solution was removed from the hot-plate and placed on a weighing balance and de-ionised water added until a desired weight was attained. The pregnant liquor was filtered to remove any insoluble material. Final liquor composition was such that A/C = 0.66 ± 0.05 where: A (aluminium hydroxide) = 150 ± 10 g/L of Al2O3 C (total caustic) = 230 ± 10 g/L of Na2CO3 S (soda, total alkali) = 260 ± 10 g/L of Na2CO3

131

Light Metals 2006 Edited by Travis J. Galloway TMS (The Minerals, Metals & Materials Society), 2006

Seed Preparation and Characterization for Agglomeration Seed used for the agglomeration experiments was the well characterized Alcoa C31, having BET surface area of 0.381 m2/g and d(0,1) = 6.3 μm, d(0,5) = 37.1 μm and d(0,9) = 92.3 μm, determined by a Malvern Mastersizer. Seed Preparation and Characterization for 2º Nucleation Broken seed fragments, formed via secondary nucleation, provide the main source of new crystals in the Bayer process [10,11]. In this type of secondary nucleation process, fluid shearing action, crystal-crystal and crystal-vessel wall/impeller blade collisions are dominantly responsible for removing nuclei from the growing seed crystal surfaces into solution. It has been stated that particles > 20 μm in diameter are less likely to agglomerate as they do not have sufficient surface energy to do so [2]. Therefore in order to focus on the effect of the CGM additive on secondary nucleation alone, especially in the early stages of crystallisation it was necessary to use a large screened seed, high supersaturation, low seed charge and enough agitation to allow the seed to remain in the bulk solution, without settling. Uniform coarse, gibbsite seed crystals were prepared from high purity Al(OH)3, (C31) by wet screening using both 55 and 90 μm sieves to give crystals in the size range 30 – 105 μm. These crystals were then treated with ultra-sonication, followed by decantation to remove the fine particles (<10 μm). The resulting product (washed and dried at room temperature in a desiccator) consisted of agglomerated, pseudo hexagonal-shaped crystals, in the size range 30 - 100 μm with reasonably smooth surfaces (Figure 1) and a corresponding BET surface area of 0.1445 m2g-1.

Figure 1. SEM photomicrograph of gibbsite seed crystals for 2º nucleation experiments. Batch Crystalliser A baffled, well sealed, 2.5 dm3, 316 stainless steel vessel was used for the crystallisation experiments. A central, 4 blade, 45° – pitch turbine impeller driven by a 70W, multi-speed motor provided a constant agitation speed at 400 rpm as well as fully developed axial flow, moderate shear and a high degree of

suspension uniformity within the crystallizer. A thermocouple sensor and a conductivity probe were fitted through the lid of the crystallizer. The entire vessel was submerged in a 15 dm3, thermostatically-controlled oil bath, maintaining a constant temperature to within ± 0.05°C. The crystal growth modifier used in this study was the commercially available Nalco product, 7837, unless otherwise stated. Models for the Kinetic Study Quantification of the crystallisation kinetics data produced from the crystallisation experiments was performed using the semi-empirical power law model for secondary nucleation and growth of gibbsite as shown below:

( )nkdtd

Sσσ

=−1 (1)

where σ is the Al(III) relative supersaturation which is defined as (C – Ce)/Ce, where C and Ce are the instantaneous Al(III) concentration and Al(III) equilibrium solubility, respectively; S is the total particle surface area; k is the nucleation rate constant and n is the reaction order. To quantify the influence of temperature and estimate the activation energy from the rate constant k, the Arrhenius expression was used.

⎟⎠⎞

⎜⎝⎛−=

RTEkk a

o exp (2)

where ko is the pre-exponential factor, Ea is the activation energy, R is the gas constant and T is the absolute temperature. Atomic Force Microscope (AFM) Measurements Interaction force measurements between two gibbsite particles (sphere and prism) were measured using a Nanoscope III Atomic Force Microscope (Veeco). A gibbsite spherical particle (hydrated alumina sphere [12]) was glued onto a triangular silicon nitride cantilever and an aluminium disk with Epikote heat sensitive resin. Particle interaction forces were measured using a fluid cell filled with sodium aluminate liquor at 25oC. The cantilever spring constant, 0.1 Nm-1 was determined using the Cleveland method [13].

Results and Discussion Agglomeration Experiments The majority of plant test work over the last decade has been completed by simulating agglomeration conditions of the particular Bayer site. Fine seed charge into pregnant liquor at high temperature is the standard method for CGM evaluation. From the work completed here over a temperature range (58-78ºC) typical to Bayer plants, coarsening of the fine C31 seed was evident after 18 hours. The results given in Figure 2 below and taken from the 68ºC test, indicate the increase in the d(0,1) values with respect to the CGM dose. It is evident that the increase in particle size and the subsequent decrease in available seed surface area results in an overall decrease in yield in the isothermal batch experiment. Therefore in continuous Bayer plant conditions,

132

process changes are made to increase yield up to a higher level without size compromise.

444546474849505152

Control 1 3 5

CGM Dosage mg/m2

d(0,

1)

59

59.5

60

60.5

61

Yiel

d g/

L

d(0,1)Yield

Figure 2. d(0,1) and yield results for 68ºC agglomeration test. SEM images given below show the difference between the control and CGM dosed test at 68ºC. The control, given in Figure 3 below exhibits large amounts of fine particles, yet to be fully incorporated into the crystal structure of the parent seed. It is clearly evident that over the 18 hour experiment the CGM additive has reduced the number of fine particles and achieved a more robust, block-like hexagonal morphology, under these conditions.

Figure 3. SEM images of control agglomeration test for 68ºC.

Figure 4. SEM images of CGM treated agglomeration test for 68ºC. Data from the kinetic analysis of this system is given in Table I. The reaction order at each temperature was found to be n = 2, corresponding to values reported extensively in literature for systems under comparable conditions. Activation energies for the control and CGM additive tests were estimated from Arrhenius plots. Values fall comfortably in the 53 -83 range reported in the literature for pure sodium aluminate solutions [3, 6, 14].

Table I. Kinetic parameters for seeded gibbsite agglomeration experiments at 58, 68 and 78oC.

Terms T(°C) Control 7831 Treated

7837 Treated

ko (m-2h-1) 58,68,78 1.10 ×105 32.6 ×105 7.78 × 105

58 2.0 ± 0.2 2.0 ± 0.2 2.0 ± 0.2 68 2.0 ± 0.2 2.0 ± 0.2 2.0 ± 0.2 Reaction

Order (n) 78 2.0 ± 0.2 2.0 ± 0.2 2.0 ± 0.2

Ea (kJmol-1) 61.2 70.3 66.6

Higher activation energy means greater sensitivity to temperature in the presence of the additive. The increase in the pre-exponential factor of ko indicates the additive aids in increasing the successful collision rate of fine seed particles during the agglomeration stage. In this case the product 7831 has a higher ko possibly indicating a greater successful collision rate than 7837. Consider the information thus far in terms of the mechanism of agglomeration put forward by Low [15] and given in Figure 5 below. The action of the CGM additive would be aiding the cementation process during agglomeration and in terms of attrition the additive appears to successfully reduce the breakdown of particles, forming a fully cemented agglomerate, which is evident in the morphology of the SEM images given in Figure 4.

Individual particles

Floc (no cementing)

Agglomerate (fully cemented)

Breakdown

Aggregate (cementing by hydrate deposition)

Figure 5. Mechanism of agglomeration, Low [15]. These tests have only considered one of the possible crystallisation mechanisms present under Bayer conditions. It was necessary to perform experiments under nucleating conditions in order to establish a more conclusive mechanism of the action of the additive. Secondary Nucleation Experiments To investigate the influence of the additive on secondary nucleation, isothermal batch crystallisation experiments were carried out over 24 hours. The tests were conducted in pure synthetic liquors, under constant conditions of NaOH = 4 M, initial σAl = 1.5, seed surface area = 7.23 m2dm-3 at temperatures of 60, 65 and 70oC. Results are discussed in terms of new crystal formation, kinetics and morphology. New Crystal Formation The observation of new fine particles (secondary nuclei) in the size range of 0.1 – 30 μm during all of the crystallisation tests

133

over the first hour are summarised in Table II. There is a marked difference between the control and treated samples for the 60 and 65°C runs, however no discernable difference could be detected over the first hour for the higher 70°C temperature run. Table II. Number of new particles generated for all crystallisation experiments under constant conditions over the first hour.

New Particles < 30 μm (1010h-1 per 25 mL) T (°C)

Control With CGM 60 3.1 1.1 65 10.1 3.7 70 17.5 18.4

The induction times for the appearance of secondary nuclei were measured by conductivity. At a constant initial σ = 1.5 and seed charge of 7.23 m2dm-3 the induction time was found to decrease with increasing temperature for both the control and treated experiments. However, the induction time was significantly shorter in the control experiment than in the treated solutions at 60 and 65oC. At 70oC there was no detectable difference in the induction time measurement for the two scenarios. This observation would imply that the CGM additive could be affecting solution speciation and/or crystal surface processes during the secondary nucleation. It is plausible that the CGM additive initially deactivated the seed surface to a high extent resulting in a lengthened time for generating sufficient active sites for fast growth. The negative effect of additive on the processes was negated at the higher temperature. The net, total crystal surface area with time, is shown in Figure 6 and appears to contradict the number of new particles generated given in Table II. The control tests displayed a lower total surface area than that of CGM treated tests at the same temperature, which would tend to indicate the possible impact of agglomeration during the crystallisation. However, it is proposed that initially a larger number of secondary nuclei formed in the treated solutions and these nuclei immediately were either adsorbed back on the seed surface or rapidly agglomerated with each other to form aggregates with a high porosity. This resulted in a low number and a high net surface area of the crystals.

0

5

10

15

20

25

30

35

40

45

50

0 1 2 3 4 5

Time (hours)

Surf

ace

Are

a (m

^2/L

)

Control at 70CCGM treated at 70CControl at 65CCGM treated at 65CControl at 60CCGM treated at 60C

Figure 6. The variation of total available surface area for the crystallising slurry as a function of temperature for seeded, control and CGM treated tests over time.

SEM analysis strongly supported the above observations. The images of the product crystals revealed the existence of significant differences in the development and growth of the new surface layer and microstructures at the seed crystal surfaces, between the control and CGM treated tests. After 4 hours the SEM images for the control experiments exhibited expected temperature dependences. At 60°C only 2° nucleation was observed, at 65°C a mixture of 2° nucleation and separate aggregates and at 70°C the majority of particles formed were separate aggregates. The primary crystals were pseudo-hexagonal plates (Figure 7b) which agglomerated and grew to secondary crystals with a size range of 10 – 30 μm. Very few, if any secondary nuclei were found to attach and grow on the parent seed surface (Figure 7a). In general for the control solutions it was found that as the isothermal batch temperature was increased the aggregation of the small nuclei, broken off the parent seed surface, increased. These aggregates are evident in the Figure 7a, with sizes of up to ~30 μm

(a)

(b)

Figure 7. SEM image for the control test after 4 hours at 65ºC. Low (a) and high (b) magnification. In contrast, the SEM images for the CGM dosed experiments showed that at 60°C only 2° nucleation on the surface was observed, at 65°C a mixture of 2° nucleation and aggregates, but not separate from the parent seed, they had re-adsorbed and at 70°C the majority of particles were secondary nuclei or aggregates on the seed surface. The primary crystals predominately exhibited pseudo-hexagonal plates with some elongated hexagonal shaped crystals produced. The parent seed surfaces were partly or fully covered by fine particles (Figure 8b). These particles exhibited stick or needle-like morphology along with the pseudo-hexagonal plates growing on the surface of seeds. The size of these primary crystals were in the range of 0.5 – 10 μm, implying quick attachment as soon as the nuclei formed. In the presence of CGM a majority of the fine particles adhered and grew on the seed surface, the degree to which the nuclei

134

remained on the parent seed surface increased with increasing isothermal batch temperature. The remaining fraction of fine particles formed aggregates in the bulk solution, with some re-attaching to the parent seed surface. This is clear evidence for the observation of the higher net surface area and correlates well with the lower number of new crystals in the treated systems compared with the control.

(a)

(b) Figure 8. SEM image for the CGM dosed test after 4 hours at 65ºC. Low (a) and high (b) magnification. The conclusion from this evidence is that the involvement of the CGM additive in the seeded gibbsite crystallisation for all temperatures not only enhanced the parent seed growth mechanism, through accelerated secondary nucleation, with less attrition and hence particle surface roughening but also enhanced the secondary nuclei agglomeration behaviour. Kinetics of Secondary Nucleation As the growth rate of gibbsite is relatively slow when compared with other inorganic crystals [3,10,16-18] it is feasible to assume the initial kinetics to be dominated by secondary nucleation at high σ (> 1.15). Therefore, the birth rate of 3-dimensional nuclei, reflected in a dramatic crystal population increase, is considered to far exceed the rate of growth (2-dimensional nucleation process) of both parent seed crystals and the nuclei [16,17]. The crystallisation kinetics data produced from the seeded crystallisation at a high initial σ = 1.5 for which secondary nucleation occurred were analysed. Plots of ln[(-1/S)(dσ/dt)] versus lnσ for both the control and CGM treated solutions over 4 hours at each of the specific temperatures yielded linear relationships. The secondary nucleation dominated-reaction order (n) was derived from the slopes of these plots and reproducibly found to be equal to 4. The results are summarised in Table III. Similar values have been reported for secondary nucleation of gibbsite from sodium aluminate solutions under comparable conditions [14,19].

Table III. Kinetic parameters for seeded gibbsite secondary nucleation experiments at 60, 65 and 70oC.

Terms T (°C) Control CGM Treated ko (m-2h-1) 60, 65, 70 3.81 × 1022 4.50 × 1034

60 4.2 ± 0.2 4.1 ± 0.2 65 3.9 ± 0.2 4.0 ± 0.2 React Order (n) 70 4.0 ± 0.2 4.1 ± 0.2

Ea (kJmol-1) 162 ± 30 239 ± 40 Activation energies for gibbsite secondary nucleation for the control and CGM additive tests were estimated from Arrhenius plots. Ea values of 162 ± 30 and 239 ± 40 kJmol-1 for secondary nucleation were calculated for the control and additive tests, respectively. These values are higher than the 132 ± 15 kJmol–1 reported for secondary nucleation from synthetic sodium aluminate solutions under similar conditions to the control [19], but are much closer to the value of 160 kJmol-1, reported in earlier work [14]. The increase in Ea in the presence of the additive suggests that more temperature-induced, chemical perturbations occur in these treated solutions. It is proposed that the additive is not only involved in the clustering of aluminate ions in supersaturated sodium aluminate solution but also the interactions between the clusters and surfaces of seed crystal, which are responsible for the development of microscopic growth units. As was the case for the agglomeration work, the pre-exponential factor of ko was found to be much higher for the CGM treated gibbsite secondary nucleation than for the control tests. It is therefore obvious that the successful collision frequency between the nuclei forming, Al(III) containing species was enhanced by the presence of additive. Interaction Forces The direct interaction forces between two gibbsite particles under Bayer conditions were measured in-situ and the effect of a crystal growth modifier determined, using the atomic force microscope (AFM). Synthetic liquor was prepared containing 4 M NaOH and σ = 1.5 at 25ºC. The fluid cell was allowed to equilibrate for 20 minutes prior to collection of data. Initially the interaction force between the gibbsite particles was first measured in pure sodium aluminate liquor. As illustrated by the force curves (repeat approaches) in Figure 9, there is only a small repulsion on approach between the particles, which would be expected as the electrostatic repulsive forces are screened as a result of the high ionic strength of the sodium aluminate liquor. On separation of the surfaces, a very weak adhesive force is observed. In addition, there appears to be some hysteresis in the approach and retract curves. This may be a result of the viscous nature of the solution introducing some hydrodynamic forces into the system. The CGM was mixed with the synthetic liquor for 10 minutes prior to addition into the fluid cell. It is clearly evident the difference the CGM additive has on the interaction forces by examining Figure 10. On approach, as for the case in the control experiment, there was very little interaction. However, upon separation of the gibbsite surfaces an adhesion is observed that varies with position on the surface. This is a common observation in systems where a chemical additive is used that does not form an even monolayer of coverage.

135

-0.01

-0.006

-0.002

0.002

0.006

0.01

0 20 40 60 80

Apparent separtion, nm

F/R

, mN

/m

Figure 9. Interaction forces on approach (filled symbols) and separation (crosses) between gibbsite particles in-situ at 25oC.

-0.02

-0.015

-0.01

-0.005

0

0.005

0.01

0 20 40 60 80

Apparent separtion, nm

F/R

, mN

/m

Figure 10. Interaction forces measured on approach (filled symbols) and separation (crosses) of gibbsite particles in CGM treated liquor at 25oC.

-0.01

-0.006

-0.002

0.002

0.006

0.01

0 20 40 60 80

Apparent separtion, nm

F/R

, mN

/m

liquorliquoradditiveadditive

Figure 11. Comparison of average interaction force curves between gibbsite particles in the control (liquor) and dosed (additive) tests at 25oC.

A comparison of the interaction between gibbsite particles in the control and CGM dosed experiments is illustrated in Figure 11. On approach there is little difference between the interaction force curves. The repulsive force is very small and would allow the particles to contact each other under normal processing conditions. On separation however, there is a significant increase in the adhesion observed between the gibbsite particles in the presence of the CGM additive. Under process conditions, the gibbsite particles would therefore be likely to agglomerate in the presence of a CGM.

Conclusion In a process where the chemical inclusion of the particle onto the surface of the seed is the rate limiting step and where particles are colliding and continually breaking off parent seed crystals, the addition of CGM may just allow the particle sufficient time to chemically bond with the seed surface and be incorporated into the structure. Evidence has been presented here which shows the increase in the successful collision frequency of not only Al(III) containing species but also small fractured gibbsite particles and fine seed crystals in the presence of the additive. Two gibbsite particles will agglomerate if they have sufficient surface energy to do so and also have sufficient time to allow cementation to take place [2]. It is postulated that the CGM modified seed surfaces possessed a high surface energy interacting with the nuclei strongly with little or no repulsive energy barrier. Therefore to conclude, addition of a CGM additive induces strong adhesive interaction forces between the gibbsite particles and nuclei, and under specific crystallization process conditions, this would markedly enhance both agglomeration and coarsening of the particles and also allow for secondary nucleation control.

References 1. W.J. Roe, D.O. Owen and J.A. Jankowski, “Crystal Growth Modification: Practical and Theoretical Considerations for the Bayer Process”, First International Alumina Quality Workshop, Gladstone (Australia), (1988). 2. C Misra, “The Precipitation of Bayer Aluminium Trihydroxide”, (Ph.D. Dissertation, University of Queensland, 1970). 3. C. Misra and E.T, White, “Kinetics of Crystallisation of Aluminium Trihydroxide from Seeded Caustic Aluminate Solutions.” American Institute of Engineers Symposium Series 438, 67(110), (1971), 53-65. 4. M Kanehara, “Formulation of Alumina Hydrate Precipitation Rate in Bayer Process for Plant Design and Operation”, Light Metals, (1971), 87-105. 5. J Anjier and H Roberson, “Precipitation Technology”, Light Metals, (1985), 367-375. 6. E.T. White, and S.H. Bateman, “Effect of Caustic Concentration on the Growth Rate of Al(OH)3 Particles.” Light Metals, (1988), 157-162. 7. Y Xie et al., “Study on the Application and Mechanism of Cationic Surfactant on the Precipitation of Sodium Aluminate Liquor”, Light Metals, (2001), 135-137. 8. Y Xie et al., “Research on the Application and Mechanism of Crystal Growth Modifier on the Precipitation Process in Sodium Aluminate Liquors, Light Metals, (2002), 157-160.

136

9. Y Xie et al., “Research on the Mechanism and Optimum Adding Method of Additives in Seed Precipitation”, Light Metals, (2003), 87-91. 10. P.I.W. Loh, H.M. Ang and E.A. Kirke, "Secondary Nucleation of Alumina Trihydrate in a Batch Crystallizer", Australia's Bicentennial International Conference for the Process Industries, Sydney, (1988), 304-309. 11. A.R. Hind, S.K. Bhargava and S.C. Grocott, “The Surface Chemistry of Bayer Process Solids: A Review”, Colloids and Surfaces, 146, (1999), 359-374. 12. J. Addai-Mensah et al., “Particle Interactions and Surface Forces During Gibbsite Precipitation From Bayer Liquors”, 6th International Alumina Quality Workshop, Bunbury (Australia), (1999). 13. J.P. Cleveland et al., Rev. Sci. Instru., 64, (1993), 403-405. 14. A.V. Nesterov and V.G. Teslya, “Mathematical Model for Crystallisation of Aluminium Hydroxide from Aluminate Solutions”, Zhurnal Prikladnoi Khimii, 62, (1989), 1999-2004. 15. G.C. Low, “Agglomeration Effects in Aluminium Trihydroxide Precipitation”, (Ph.D. Dissertation, University of Queensland, 1975). 16. A Halfon and S Kaliaguine, “Alumina Trihydrate Crystallisation, Part 1 Secondary Nucleation and Growth Rate Kinetics”, Canadian J. Chem. Eng., 54, (1976), 160-167. 17. A Halfon and S Kaliaguine, “Alumina Trihydrate Crystallisation, Part 2 A Model of Agglomeration”, Canadian J. Chem. Eng., 54, (1976), 168-172. 18. W.R. King, “Some Studies in Alumina Trihydrate Precipitation Kinetics”, Light Metals, 2, (1973), 551-563. 19. J Li et al., “Secondary Nucleation of Gibbsite Crystals from Synthetic Bayer Liquors: Effect of Alkali Metal Ions”, Journal of Crystal Growth, 219, (2000), 451-464.

137