Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Minerals Engineering 160 (2021) 106664

Contents lists available at ScienceDirect

Minerals Engineering
journal homepage: www.elsevier.com/locate/mineng

Lime properties and dose effects on causticisation of synthetic Bayer liquor


C.A. du Plessis a, *, H. Lambert a, E. Hoummady a, R.G. McDonald b, D. Bedell b
a
Lhoist Recherche et Développement SA, Rue de l’industrie 31, B-1400 Nivelles, Belgium
b
CSIRO Mineral Resources, 7 Conlon St, Karawara, WA 6152, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: The production of alumina (Al2O3) from bauxite via the Bayer process is underpinned by the use of lime in a
Lime variety of process reactions. One of the most important is the causticisation of process liquors, in which lime is
Slaking used to convert soluble sodium carbonate into caustic, i.e. sodium hydroxide. Causticisation is complicated by
Causticisation
unproductive side reactions with sodium aluminate that limit both the extent of causticisation and the efficiency
Alumina
Bayer
of lime use. Substantial industrial effort and literature has been devoted to optimising causticisation process
Tri-calcium aluminate (TCA) parameters (e.g. temperature, reaction time, liquor concentrations, and reactor mixing regimes) but little
Monocarbonate attention has been given to determining the impacts of lime properties and dose. In this study a test procedure
Hemicarbonate was developed and used to evaluate the impact of lime dose, in both slaked and dry quicklime formats, on
CO2 emissions causticisation performance. This was used to determine the effects of various lime properties, including (1)
Greenhouse gas (GHG) chemical composition (purity), (2) particle size and (3) reactivity on the causticisation of synthetic Bayer liquors.
The results show that, of these parameters, the differences in lime chemical composition had the most significant
impact. The methodology can be used to facilitate optimal lime reagent choice and dosage in industrial alumina
Bayer refineries. The potential impact of lime utilisation efficiency is also illustrated in the context of overall CO2
emission intensity.

1. Introduction 1.2. Reactions

1.1. Lime in the Bayer process Caustic soda, i.e. sodium hydroxide (NaOH), is the key reagent used
in alkaline digestion of bauxite ore (Whittington et al. 1997) which re­
Lime is an essential reagent required in various process steps of the lies on the dissolution of solid aluminium hydroxide and oxy-hydroxide
Bayer process, in which bauxite ore is converted into alumina (Whit­ minerals to form aluminate ions in solution, i.e. Al(OH)-4 (Sipos, 2009).
tington et al. 1997), as illustrated in Fig. 1. It is used for improving In addition to the mineral digestion reaction (Eq. (1)), NaOH also reacts
conversion of goethite to hematite and dissolution of boehmite and with organic matter contained in bauxite, and to a minor extent with
diaspore during digestion; control of liquor impurities such as silica, atmospheric CO2, resulting in the formation of soluble sodium carbonate
oxalate, titanium and phosphorous; minimising soda losses to red mud; (Power et al. 2012) illustrated in simplified form in Eq. (2):
providing filter aid for liquor clarification and for causticisation (Arıkan
Al(OH)3 + NaOH → NaAl(OH)4 (1)
et al. 2019; Wellington, 2015; Whittington, 1996; Whittington and
Cardile, 1996). Alumina Bayer refineries, accordingly, consume large Organic matter + NaOH → Organic acid anions + Na2CO3 (2)
quantities of lime, ranging from approximately 50 kt per annum per
million tonnes of alumina production, for a plant processing high quality This presence of sodium carbonate reduces the effectiveness of the
lateritic bauxite, to at least twice that quantity for Bayer refineries using liquor to digest alumina minerals from the bauxite. Bayer process liquors
diasporic bauxites (Arıkan et al. 2019; Chaplin, 1971). are maintained at a controlled maximum level of soluble anions, so
removal of soluble carbonate is essential to the maintenance of an
acceptable concentration of NaOH. The only known way to economi­
cally remove the soluble sodium carbonate is by causticisation with lime

* Corresponding author.
E-mail address: chris.duplessis@lhoist.com (C.A. du Plessis).

https://doi.org/10.1016/j.mineng.2020.106664
Received 10 July 2020; Accepted 5 October 2020
Available online 26 October 2020
0892-6875/© 2020 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
C.A. du Plessis et al. Minerals Engineering 160 (2021) 106664

Fig. 1. Simplified Bayer process diagram, indicating causticisation location and other lime addition points.

(sometimes also referred to as re-causticisation), which precipitates space of the LDH (Tóth et al. 2014). The extent of carbonate incorpo­
solid calcium carbonate and proportionately returns (i.e. regenerates) ration, i.e. normally 0.5 or 1 carbonate for each [Ca2Al(OH)6]+ unit,
hydroxide ions (in the form of NaOH) to the solution. The basic reaction results either in a hemicarbonate (Eq. (6)), or monocarbonate (Eq. (7))
(Eq. (3)) is preceded by a hydration reaction (Eq. (4)) in scenarios where (Runčevski et al. 2012; Smith 2017). The hemicarbonate has been re­
dry quicklime is used directly in the causticisation reaction (Xu et al. ported as the most likely product from the reaction of lime with indus­
1998). trial Bayer liquors, while the use of synthetic liquor typically results in
monocarbonate as the main LDH in laboratory studies (Rosenberg et al.
Ca(OH)2 + Na2CO3 → 2NaOH + CaCO3 (3) 1999).
CaO + H2O → Ca(OH)2 (4) 4Ca(OH)2 + 2NaAl(OH)4 + 0.5Na2CO3 + 5.5H2O →
These equations are sufficient for the prediction of reaction rates and [Ca2Al(OH)6]2.(CO3)0.5.OH.(5.5H2O) + 3NaOH (6)
efficiencies of causticisation of alkaline sodium carbonate solutions in 4Ca(OH)2 + 2NaAl(OH)4 + Na2CO3 + 5H2O →
the paper pulp industry (Wang et al. 1994). However, in processing
[Ca2Al(OH)6]2.CO3⋅5H2O + 4NaOH (7)
bauxite to alumina by the Bayer process, the situation is complicated by
the presence of sodium aluminate, NaAl(OH)4, and several impurity Depending on the complex set of prevailing equilibrium conditions,
anions including phosphate, titanate, oxalate and fluoride (Chaplin, the LDH intermediates may decompose into either TCA or CaCO3
1971) that can form a range of compounds with lime. The most (Buttery et al. 2002; Sizyakov and Tikhonova 2012; Smith 2017). The
important (and unproductive) side-reaction with lime (Fig. 2) is the reactions producing TCA or LDHs (e.g. Eq. (5), 6 and 7) are detrimental
reaction with aluminate to form TCA, tri-calcium aluminate, Ca3[Al for two reasons: (1) the precipitated alumina species are lost to residue
(OH)6]2 (Eq.5). and (2) lime is inefficiently used for low productivity reactions, i.e. less
Na2CO3 is converted per mol of Ca(OH)2 consumed. Based on these two
3Ca(OH)2 + 2NaAl(OH)4 → Ca3[Al(OH)6]2 + 2NaOH (5) criteria, TCA is the most undesirable precipitate product, followed by
Previous studies have suggested that TCA exists in equilibrium with hemicarbonate and then monocarbonate. The equilibrium between the
CaCO3, and that the degree of causticisation under given conditions can most desired reaction product of lime use (CaCO3) and aluminium-
be predicted thermodynamically (Whittington, 1996; Roach, 2000). The containing precipitates (TCA and LDHs) is influenced by a number of
equilibrium conditions are, however, complex due to the described factors including the solution aluminate concentration, combined
formation of intermediate layered double hydroxide precipitates (LDHs, caustic and sodium carbonate concentration, and temperature (de Witt
also referred to as hydrocalumite as a generic term for compounds with et al. 2000; Roach, 2000; Rosenberg et al. 2001). Other strategies that
the formula Ca2Al(OH)6.X⋅nH2O that are based on repeating [Ca2Al have successfully been employed to shift the equilibrium towards the
(OH)6]+ units), and the reported metastability of these LDHs (Rosenberg desired product include the use of additives such as gluconate to inhibit
et al. 2001). Anions such as carbonate are intercalated into the interlayer the formation of TCA and facilitate the formation of calcite (Eq. (3))

Fig. 2. Simplified diagrammatic illustration of lime reactions into productive


and unproductive products, TCA (tri-calcium aluminate) and LDHs (layered Fig. 3. Diagrammatic illustration of the links between the three operational
double hydroxides). causticisation process objectives.

2
C.A. du Plessis et al. Minerals Engineering 160 (2021) 106664

(Rosenberg et al. 1999; Rosenberg 2008). with low purity lime) was the best reagent for causticisation (Young,
1982; Libby, 1983) and that slaking, prior to or during causticisation, is
1.3. Operational context detrimental to both the extent of causticisation and the efficiency of lime
use, suggesting that a layer of Ca(OH)2 coating on CaO may inhibit
In an alumina refinery the objectives of causticisation are to: (1) causticisation. Conversely, other studies found that a high purity lime
convert as much of the sodium carbonate as possible to caustic while (2) sample performed better in causticisation than a less pure industrial lime
minimising the loss of soluble alumina in the form of precipitates that (Wellington and Valcin, 2007). Fundamental studies have demonstrated
normally deport to residue disposal and (3) using lime reagent as effi­ that it is highly unlikely that Ca(OH)2 coating formation on CaO parti­
ciently as possible (Fig. 3). The primary objective is to convert sodium cles could inhibit causticisation. Explanations for inhibition are more
carbonate, although this cannot be independently optimised without likely to be found in the formation of surface coatings of calcium car­
impacting the other two parameters, due to their intrinsic linkage. For bonate and/or calcium aluminate/alumino-carbonate solid species. Xu
example, the extent of causticisation can be increased by increasing the et al. (1998) demonstrated that, for CaO and Ca(OH)2, the rate of
lime dose, but that would lead to inefficient lime use and alumina losses causticisation of sodium carbonate solutions is controlled by the diffu­
due to unproductive side-reactions, as outlined earlier. Similarly, lime sion of CO2–3 ions to the solid surface. In both cases it is the formation of
efficiency and alumina losses can be optimised (by using small lime dose CaCO3 that progressively blocks the surface and inhibits the reaction,
or very dilute liquors, for example), but at the expense of reduced thereby occluding and rendering unreacted a portion of the Ca(OH)2.
Na2CO3 conversion. Operationally, it is important to balance these ob­ With aluminate solutions the situation is further complicated by the
jectives for an economically optimal overall plant outcome. The opera­ formation of surface deposits of calcium aluminates (Xu et al. 1997). It is
tional setpoints to achieve such optimisation are influenced by the possible to form TCA in this way (Whittington et al. 1997), but in Bayer
specific lime reagent properties, the lime dose response and the pre­ causticisation the intermediate is generally a calcium alumino-
vailing causticisation process conditions. carbonate, variously identified as a hemicarbonate or monocarbonate
To optimise performance, causticisation in most modern Bayer re­ depending on conditions (Buttery et al. 2002; Roach, 2000; Rosenberg
fineries occurs in so-called “side-steam” or “outside” causticisation, et al. 2001). Nevertheless, it has been suggested that direct formation of
which has largely replaced the original “inside causticisation” process CaCO3 from Ca(OH)2 is possible in Bayer causticisation (de Witt et al.
due to inherent inefficiencies of the latter, i.e. high lime consumption 2000; Libby, 1983). Surface deposits and coatings influence overall
and limited carbonate removal capability (Chaplin, 1971). Much causticisation rates by blocking the reacting surface from the liquor.
research has been undertaken to optimise the side-stream causticisation Differences in reaction rates between types of lime are therefore likely to
process (Roach, 2000; Rosenberg et al. 2001; Whittington, 1996; be related to the propensity of their solid surfaces to attach and grow
Whittington and Cardile, 1996; Xu et al. 1997; Xu et al. 1998; Young, layers of calcium carbonate and calcium aluminate/alumino-
1982) described in early patents (Cagnolatti et al. 1965; Cundiff, 1950). carbonates. There appears to be a scarcity of published studies that
Generally, best causticisation with least TCA formation is achieved with have systematically evaluated lime properties in the context of deter­
a more dilute aluminate liquor at close-to-boiling temperature (Chaplin, mining the impacts of lime dose on the extent of causticisation and lime
1971). The liquor concentration used as causticisation feed is a efficiency.
compromise between the causticisation process efficiency and the
amount of carbonate available for removal (Roach, 2000). At the initial 1.5. Slaked lime versus dry lime trade-off
washer stages, the solution temperature is high and sodium carbonate
very concentrated, whereas further down the washer train (Fig. 1) the Simplicity of reagent sourcing and operation generally mean that
temperature is generally lower and the solution more dilute due to the Bayer refineries use a single source of lime, even though the optimal
counter-current wash water. In side-stream causticisation practice, this lime reagent performance parameters may be different for different
generally means that the causticiser feed is derived from the overflow of process units within the Bayer process circuit (Suss et al. 2008). Lime is
the first or second mud washer (Fig. 1). The causticiser solids residue, commonly used in the form of slaked lime or milk of lime (MoL). The
which is a mixture of calcium compounds of carbonate, aluminate, advantages of this practice are that slaked lime can be readily distrib­
alumino-carbonates and lime impurities (i.e. grits comprising mainly uted around the plant, typically in a ring-main, with off-take points at
quartz, MgO, and minerals containing iron and other trace impurities) the various process locations and the dose accurately controlled in
reports to the bauxite residue. Hence improvements to lime efficiency response to process requirement. The disadvantage of using slaked lime
(which reduce the amount of causticiser residue produced and is the associated water (typically 80 wt%) injected into the plant liquor
aluminium lost) also contribute to the process and environmental out­ circuit. To overcome the dilution effect of this water, evaporation is
comes related to bauxite residue production and storage. Efforts to required elsewhere in the circuit (Fig. 1), with significant associated
improve efficiency have included separating the causticisation process energy requirements. The option of using a higher Ca(OH)2 solids con­
into two stages, the use of additives such as sodium gluconate (Rosen­ tent in the slaked lime (i.e. reduced water content) is constrained
berg et al. 1999; Rosenberg 2008) and high temperature conditions up to because slaked lime pumpability is significantly reduced as solids con­
170◦ C (de Witt et al., 2000). tent increases above 20 wt%, due to viscosity and shear-thinning effects
(Ruiz-Agudo and Rodriguez-Navarro, 2010). To overcome the high en­
1.4. Influence of lime properties on causticisation – Current state of ergy cost associated with evaporation, some Bayer refineries opt to use
knowledge dry quicklime dosing in some locations within the plant circuit. Using
dry quicklime, however, may be sub-optimal, depending on the lime
Most of the technical and patent literature on Bayer causticisation source and process conditions (de Witt et al. 2000) and conventional
focuses on the effects of process conditions (such as temperature, con­ agitator reactors may not provide optimal dispersion and mixing of dry
centration, and mixing), but little attention has been given to the in­ quicklime into process liquors.
fluence of the properties of lime. The main quicklime properties of
relevance are: reactivity (measured by exothermic reaction rate with 1.6. Objectives and experimental rationale
water), CaO content (measured by XRF) and availability (measured by
titration), and impurities (Australian Standard AS 1672.1, 1997; Euro­ Alumina Bayer refineries often use the available CaO % content of
pean Standard EN 12485, 2017). The limited literature on the influence quicklime as the basis for calculating lime performance. It is important
of lime properties also reflects conflicting views. For example, some to note that this metric is determined via an acid titration method that
studies concluded that low reactivity quicklime (typically associated does not account for slaking, particle size, or reactivity effects and may

3
C.A. du Plessis et al. Minerals Engineering 160 (2021) 106664

Fig. 4. Sample preparation summary with red dotted lines indicating the comparison pairs for evaluating various lime properties.

therefore be a poor predictor of performance in the target reaction(s), as this metric is that it is readily determined by solution assay (titration
has been demonstrated for other lime applications (Ineich et al. 2017). methods) and standard lime analyses methods. LE only determines the
The purpose of this study is therefore to use a standardised causticisation lime efficiency inherent in its properties as a reagent. It does not give an
dose–response test, with a view to meeting the need for information indication of the productive versus unproductive use of lime (in terms of
beyond simple CaO content to enable more informed process and eco­ the formation of precipitated products: CaCO3, TCA or LDHs), which is
nomic decision-making on lime sourcing for the Bayer process. In predominantly a consequence of operational process conditions (such as
developing the test, the influences of lime properties and dose on extent lime dosage, liquor composition and temperature). The relative forma­
of causticisation and lime efficiency are evaluated. The results highlight tion of productive versus low productivity precipitation products is
the effects of individual lime properties, including lime purity, and ad­ determined by solid analyses, such as quantitative X-ray diffraction.
ditions such as dry quicklime, dry hydrated lime and MoL. The condi­ Maximum extent of causticisation (MEC, %) is defined as the
tions of the test simulate conditions typical of current side-stream plateau in EC % as a function of CaO equivalent dose. Maximum
causticisation practice. The examples shown are for causticisation of effective dose (MED, %) is defined as the CaO equivalent dose (in g/L)
synthetic Bayer liquor using a limited range of lime samples, but the test above which no further increase in EC % occurs. The CaO equivalent
is equally applicable to actual plant liquors and all types of lime. dose is the mass of lime, adjusted for the total CaO% content as
measured by XRF.
1.7. Definitions used
2. Materials and methods
Lime is the generic term used to describe various forms of calcium
oxide, including: quicklime - a dry reagent of mainly CaO, produced 2.1. Lime samples
from the calcination of limestone of which CaCO3 is the main compo­
nent; hydrate or hydrated lime - a dry reagent of mainly Ca(OH)2, The quicklime and hydrated lime samples used in this study were
produced from the stoichiometric hydration of quicklime with water; from two different carbonate sources. The designated label for each
and slaked lime or milk of lime (MoL) – a slurry of Ca(OH)2 particles sample and preparation pathway are summarised in Fig. 4. In each test,
typically in the range of 10-22 wt% solids content, produced from comparison pairs were used, indicated by the red dotted lines. The
slaking quicklime in a stoichiometric excess of water. interpretation of results was mainly confined within a comparison pair,
Soluble sodium carbonate and sodium hydroxide concentrations are rather than between pairs. Most of the lime property comparisons were
important processing parameters. In this regard A, C and S in this work conducted using subsamples of carbonate source 1, thereby enabling a
have the meanings accepted as standard in the alumina industry: A = comparison of effects of physical characteristics and slaking on samples
aluminium concentration in units of g/L Al2O3, C = total caustic soda of identical chemistry. Quicklime sample B2 was compared to quicklime
concentration in units of g/L Na2CO3 equivalent, and S = (C + Na2CO3) sample D, originating from a different source (carbonate source 2), to
in units of g/L Na2CO3 equivalent. facilitate a comparison of different quicklime chemical compositions.
Causticisation process performance in this study was assessed by The key chemical and physical characteristics of all the samples are
evaluating the extent of causticisation (EC, %), lime efficiency (LE, summarised in Table 1.
%), and alumina losses via unproductive reactions. Extent of caustici­ Quicklime sample B2 was prepared from industrial lime manufac­
sation (EC, %), in this study, is defined by the proportion of soluble tured from limestone Carbonate Source 1 (Fig. 4) using conventional
carbonate removed from solution by causticisation, expressed as a per­ methods (Boynton 1980). The calcination product, with a particle size
centage, as determined by solution titration. The nature of solids, range of + 10/-60 mm, was crushed to 100% <5 mm. The < 2 mm
including the precipitates formed, is determined separately by various fraction was discarded to yield a + 2/-5 mm fraction, which was then
analyses. This definition of EC is preferred to the more commonly used further crushed to 100% <2 mm to produce B2.
equivalent term “causticisation efficiency” to avoid confusion with LE. Quicklime samples C1 and C3 were also prepared from Carbonate
Importantly, EC does not consider the amount of lime used to achieve Source 1 under laboratory conditions, to produce quicklime samples
the conversion reaction. LE in this study is defined as the molar ratio of with the same chemistry but with different reactivities. These were
soluble carbonate removed from solution relative to Ca added (total CaO achieved by calcining under different temperature conditions (Zanin
% as determined by XRF), expressed as a percentage. The advantage of et al. 2019) resulting in a high reactivity (C1) and low reactivity (C3)

4
C.A. du Plessis et al. Minerals Engineering 160 (2021) 106664

Table 1
Compositions of lime samples (% w/w dry). Av CaO = available lime.
Sample label CaO % MgO % SiO2 % Al2O3 % Fe2O3 % SO3 % CO2 % Av CaO % t60 min Size

B2 95.1 1.5 0.3 0.1 0.1 0.1 0.7 93.1 2.2 − 2 mm


D 81.5 4.9 8.3 0.5 0.3 0.8 1.1 80.6 11.3 P80 310 µm
H1 74.6 0.2 0.2 <0.1 <0.1 <0.1 2.4 N/A N/A P50 3 µm
H2 73.0 0.26 1.4 <0.1 <0.1 <0.1 2.9 N/A N/A P50 54 µm
C1 96.9 0.8 0.5 0.1 0.1 <0.1 0.2 95.1 0.2 − 2 mm
C3 96.9 0.8 0.5 0.1 0.1 <0.1 0.1 96.4 8.7 − 2 mm

quicklime sample, respectively. Quicklime reactivity was determined by 2.2.3. Slaked lime
the European standard t60 reactivity test, in which t60 is the time in Slaked lime (milk of lime) was prepared as a 20wt% solids slurry
minutes to reach 60◦ C from a starting temperature of 20◦ C in a slaking (after slaking) by adding deionised water to the dry lime solids in a
test (Ineich et al. 2017; Zanin et al. 2019). stirred vessel and heating to approximately 90◦ C under nitrogen. The
Quicklime D was prepared from Carbonate Source 2, to yield mixture was stirred at temperature until fully slaked. The dilution effect
quicklime with a different chemical composition and higher levels of of water added via the slaking process was taken into consideration in
impurities compared to the other quicklime samples. Carbonate Source the interpretation of the results.
2 was crushed to < 1 mm and calcined. The final product was used as is,
with a measured granulometry of 80% passing 310 µm. 2.2.4. Bottle roll batch test
Hydrated lime samples H1 and H2 were prepared using conven­ A standard bottle roll batch test (BRBT) was used for dose–response
tional hydration methods (Boynton, 1980), also originating from Car­ studies. These tests were conducted in wide-necked Nalgene high-
bonate Source 1. Sub-samples were separated using a laboratory air density polyethylene bottles mounted in a custom-made rotating
classifier to produce a fine Ca(OH)2 particle size sample (H1, 50% <3 carousel positioned within a water bath heated to 95◦ C (Fig. 5). A batch
µm), and a coarse particle size sample (H2, 50% <54 µm). of bottles was freshly prepared for each test run, with each batch con­
taining a series of increment doses for a selected lime. Each experiment
2.2. Procedures was therefore a comparison of two batches. This setup enabled indi­
vidual batches with a range of incrementally increased doses and
2.2.1. Lime dose different lime sample comparison pairs (red dotted lines in Fig. 4) to be
Comparisons were conducted as batches of individual test bottles run simultaneously under identical conditions (i.e. with the same
containing incrementally increasing lime doses. Lime dose increments starting SBL and at the same temperature) and the results interpreted as
were based on a mass equivalent basis, but expressed as CaO equivalent a function of lime dose. Lime (dry or pre-slaked) was added to each pre-
content (based on the CaO % content as determined by XRF, Table 1), to heated bottle, which was then topped up with synthetic liquor at tem­
normalise for the effect of different chemical compositions of samples perature. The bottles were sealed, shaken and secured in the carousel.
used (Table 1) and to compared slaked and unslaked formats as well as These were then rotated in the water bath at 95◦ C. A standard time of
hydrates on an equivalent basis. two hours was chosen to be consistent with common industrial practice
(Roach, 2000). After removal from the carousel, samples of liquor and
2.2.2. Synthetic Bayer liquor solids were taken, and analysed as described below.
The synthetic Bayer liquor (SBL) was prepared as a pure sodium
aluminate/sodium carbonate solution by dissolving pure aluminium
2.3. Analysis of samples
hydroxide (Alcan Superwhite) and AR sodium hydroxide in hot deion­
ised water at 90◦ C, followed by the addition of pure sodium carbonate
A, C & S titrations - A, C and S concentrations were determined by
solution. A 5 L 18/10 stainless steel reactor was used with a magnetic
liquor titration (Connop, 1996), using Metrohm autotitration equipment
stirrer hotplate and Teflon stirrer bar. The solution was made up to a
and software. Special care was taken to minimise carbon dioxide uptake
fixed weight (estimated to give the required concentration) prior to use.
by the liquor prior to titration. C and S titrations data were used to
The hot mixture was passed through a 0.45 µm membrane filter and
calculate EC and LE respectively. Titration A was monitored but not
diluted to the required concentration. The target concentrations in the
reported because it is a relatively insensitive metric. Rather, the loss of
liquor were: A = 62 g/L, C = 96 g/L, S = 120 g/L, to give target ratios for
aluminium from the SBL was monitored by XRF analysis of the solids,
A/C = 0.65 and C/S = 0.80.
while quantitative x-ray diffraction (QXRD) methods were used to
indicate its speciation in the solids.
Elemental and TIC analyses - Diluted liquor samples were analysed
for Al, Ca and Na by ICP-OES. Solids were analysed for Al, Ca, S, Fe, Mg,
Na and Si by borate-fusion wavelength dispersive x-ray fluorescence.
Total inorganic carbon (TIC) analyses were determined for solids using a
LabFit CS-2000 Carbon and Sulphur Analyzer, and for liquors with a

Table 2
Standard deviation values (expressed in %) of the various parameters measured
in experimental reproducibility tests.
Solids addition

Parameter 15 g/L 21 g/L

A 0.18 0.23
C 0.15 0.30
S 0.06 0.10
EC 1.1 1.4
Fig. 5. Diagrammatic representation of bottle roll batch test in a temperature-
LE 1.1 1.5
controlled water bath with rotating carousel.

5
C.A. du Plessis et al. Minerals Engineering 160 (2021) 106664

Fig. 6. Causticisation dose–response curves comparing quicklime samples B2 (a highly reactive, high purity sample) and D (a lower reactivity, lower purity sample)
in slaked (milk of lime) and dry quicklime formats in terms of (a) extent of causticisation (EC) and (b) lime efficiency (LE).

Shimadzu TOC-VCSH Total Organic Carbon Analyzer. plateau, are relevant features of the dose–response curve. To facilitate
Quantitative x-ray diffraction (QXRD) analyses - Dried solids visualisation and interpretation, experimental data (both for EC and LE)
were analysed by x-ray diffraction (XRD) using a Panalytical high- were fitted with an empirical five-parameter logistic equation (Cum­
resolution multi-purpose powder diffractometer (Empyrean) using co­ berland et al. 2015; Gadagkar and Call, 2015; Ineich et al. 2017),
balt-Kα radiation at 40 kV and 40 mA, a Bragg-Brentano high-definition (Eq.10), where: y = extent of causticisation (EC %) or lime efficiency (LE
incident monochromator and PIXcel3D photon counting x-ray detector. %); x = lime dose; d = initial value of y; a = plateau (maximum value of
Data were collected for two hours over an angular range of 5–130◦ 2θ in y); c = inflection point; b = curve sharpness (negative for LE plots); m =
continuous scan mode. The Rietveld method with Bruker TOPAS 5 curve skewness.
software was used for QXRD. The predominant phases detected in the ⎛ ⎞
causticisation products corresponded to the published diffraction pat­ ⎜ ⎟
a− d
terns for calcite, CaCO3 (Markgraf and Reeder, 1985), TCA (katoite), y=⎜
⎝[ ( )− b ]m ⎟
⎠+d (10)
Ca3Al2(OH)12 (Lager et al. 2002), monocarbonate Ca4Al2(OH)12.CO3. 1+ cx

(5.5H2O), (François et al. 1998), and portlandite, Ca(OH)2 (Chaix-


Pluchery et al. 1987). The XRD diffractograms were not consistent with
2.4.1. Experimental reproducibility
the formation of hemicarbonate (Runčevski et al. 2012), which was
Quicklime sample B2 was selected to investigate the test reproduc­
expected based upon previous work (Rosenberg et al. 1999). The QXRD
ibility. Five tests each with dry solids additions of 15 g/L and 21 g/L
data were normalised to give optimum correlation between the calcu­
equivalent CaO to SBL were run as described previously. The standard
lated and analysed element contents.
deviations (%) for the liquor titration values, extent of causticisation,
and lime efficiency were calculated from the results and are shown in
2.4. Dose-response curves for EC and LE Table 2. It is apparent that as the dose increases, so does the standard
deviation. The standard deviation values, however, are excellent and
The EC (Eq. (8)) and LE (Eq. (9)) were calculated from S and C confirm that the test procedure can be applied with confidence and
titration data and CaO equivalent dose (in g/L). provide context for the interpretation and comparison of the results in
subsequent tests.
100 × [1 − (S − C)]
EC(%) = (8)
(S0 − C0 ) 3. Results and discussion
( )
S∙ C
S
− C0
S0
3.1. Quicklime origin and chemical composition – Comparing B2 and D in
56.1
LE(%) = 100 × × (9) dry and slaked formats
CaO equivalent dose (g/L) 106
The dose–response curves for EC % generally follow a pattern of an The influence of lime type (reflected by differences in chemical
increased response relative to dose, up to a plateau, corresponding to composition and reactivity) was investigated using the BRBT described in
pseudo-equilibrium conditions (Roach, 2000; Whittington, 1996). The the experimental section. The extent of causticisation (EC, %) achieved by
dose at which such a plateau is reached and the maximum EC % of the adding dry quicklime to synthetic Bayer liquor (SBL) is shown in Fig. 6a as
a function of lime dose for quicklimes B2 and D, compared in dry
Table 3 quicklime and in slaked dosage formats. Quicklime B2 is high purity, high
Summary of causticisation dose performance parameters for B2 and D. reactivity (i.e. rapid-slaking) quicklime, whereas quicklime D is a lower
reactivity (i.e. slower-slaking), lower purity quicklime containing signif­
Parameter Lime B2 Lime D
icant magnesium and silicon (Table 3). Maximum effective dose (MED) is
Slaked Dry Slaked Dry
defined as the lime dose required to reach 99% of the maximum (plateau)
Maximum extent of causticisation (MEC) 71% 65% 66% 64% value. Lime efficiency (LE, %) is similarly shown as a function of dose in
Maximum effective dose (MED) 15 g/L 17 g/L >24 g/L 22 g/L Fig. 6b. The key performance parameters for these two limes are shown in
Lime efficiency (LE) at MED 62% 49% 31%
Table 3. The relative mass proportions of precipitated mineral phases in
<27%

6
C.A. du Plessis et al. Minerals Engineering 160 (2021) 106664

Fig. 7. Residue composition obtained by QXRD analysis for tests dosed with (a) slaked B2, (b) dry quicklime B2, (c) slaked D, and (d) dry quicklime D, with the
vertical dotted line indicating the Maximum Effective Dose (MED) as determined based on extent of causticisation (EC).

the residues, as determined by QXRD are shown in Fig. 7. solution. Under these conditions monocarbonate appears to be a stable
Slaked format comparison - Slaked lime sample B2 reaches a MED of reaction product, rather than an intermediate metastable product, at
71% at a dose of 15 g/L CaO, with an LE of 62%. In contrast, the least within the timeframe of these experiments (i.e. two hours) (De
causticisation curve for slaked lime D does not reach a plateau within Silva et al. 2012). The most relevant evaluation of the proportional
the studied conditions, and at a dose of 24.5 g/L CaO only achieves solids composition is, however, at the respective MED (vertical dotted
65.8% causticisation. LE is also substantially lower for slaked lime D line) dose concentration for each sample, where the proportion of pro­
throughout the dosing range. Based on these dose–response curves, a ductive (calcite) versus low productivity (monocarbonate and TCA)
substantially lower dose of slaked lime B2 would be required, compared products, are more favourable (at the MED) when using slaked B2
to slaked lime sample D, to achieve the same causticisation outcome. compared to slaked D. Consistent with other performance parameters,
The superior performance of slaked lime B2 correlates with the higher slaked lime D shows detectable amounts of unreacted lime (portlandite)
purity and/or greater reactivity, as measured by the temperature rise throughout the dose range, and a significant higher proportion of TCA
during slaking, t60 (Table 1). The finer particle size of lime sample D is formation at equivalent doses above 9 g/L CaO. These results suggest
unlikely to be relevant as finer particle size generally favours reactivity, that coarse Ca(OH)2 particles become coated with TCA, which prevents
a factor that did not sufficiently favour lime sample D in this scenario. further reaction with carbonate in solution (Xu et al. 1997). It may also
The corresponding mineral phases (in Fig. 7a and 7b) are significantly be due in part to the fact that lime D contains a higher MgO content than
different between the two lime reagents, with the most striking differ­ lime B2, although the specific effect of MgO on TCA formation is un­
ence in the predominance of monocarbonate precipitation at higher known. The effect of dose on alumina loss from solution is shown in
dose levels for slaked lime B2. Calcite is the predominant product at Fig. 8a, indicating a substantially greater extent of alumina loss at the
doses below the MED, while overdosing (i.e. beyond the MED) results in MED dose level when using lime D, compared to B2. The effect of lime
predominance of monocarbonate, due to depletion of carbonate in dose optimisation on alumina losses can be contextualised by

7
C.A. du Plessis et al. Minerals Engineering 160 (2021) 106664

Fig. 8. Extent of alumina lost from solution, as a function of CaO equivalent dose for samples B2 and D under (a) slaked and (b) dry quicklime addition regimes.

considering an indicative industrial causticiser feed flow rate of 1700 range, even at low dosages, but is not present at similar dosages when
m3/h and alumina concentration of 62 g/L. Under these conditions, each slaked lime is used. This suggests that the hydration reaction of CaO into
1% delta in alumina loss represents 9,200 tonnes of alumina per year. Ca(OH)2 and monocarbonate formation can occur in parallel, and that
Dry quicklime comparison – The EC and LE performances of the two calcite most likely forms via Ca(OH)2 in preference to directly from CaO.
limes are less pronounced when dosed as dry quicklime; dry quicklime Comparing the same figures also shows TCA formation even at very low
B2 is only marginally better than dry quicklime D (Fig. 6), although dry dosages in the case of dry quicklime. This suggests that dry-added CaO
quicklime B2 performed substantially better than dry quicklime D in may be a better substrate for initial TCA nucleation than Ca(OH)2,
terms of alumina loss from solution (Fig. 8b). Comparing each lime whereas Ca(OH)2 preferentially nucleates calcite or monocarbonate.
sample in dry versus slaked format, the following observations are TCA formation is known to form by secondary nucleation, requiring a
made: Lime D performs better as dry quicklime compared to its slaked suitable substrate to trigger the nucleation (Salimi and Vaughan, 2016).
form, in terms of EC and LE but worse when considering alumina loss
(comparing Fig. 8a and 8b). Lime B2 performs substantially better when 3.2. Ca(OH)2 particle size – Hydrated lime H1 vs. H2
slaked in comparison metrics (i.e. EC, LE and alumina loss). Greater
alumina loss from solution occurred for lime samples when used in dry Hydrated limes H1 and H2 were produced from the same source to
compared to slaked format. This demonstrates that reported observa­ have a similar chemical composition. H1 is finer (50% − 3 µm) than H2
tions of superior performance of dry lime dosage over slaked (or the (50% –54 µm). The EC and LE curves are shown in Fig. 9 and Fig. 10, and
converse) might not be generally applicable, and highlights the impor­ the overall performance results are summarised in Table 4.
tance of explicitly determining the lime dose–response for each appli­ The effects of Ca(OH)2 particle size were investigated using hydrated
cation scenario and lime type. A comparison of Fig. 7a with 7c illustrates lime so that samples with identical chemistry could be prepared with
that, for dry lime dosage, monocarbonate is present throughout the dose different, quantifiable particle size distributions. The fact that hydrates

Fig. 9. Causticisation dose–response curves comparing Ca(OH)2 hydrate sample H1 (an ultra-fine hydrate) and H2 (a coarser hydrate) in terms of (a) extent of
causticisation (EC) and (b) lime efficiency (LE), where both hydrates were used in dry format.

8
C.A. du Plessis et al. Minerals Engineering 160 (2021) 106664

Fig. 10. Residue composition obtained from QXRD analysis for tests dosed with (a) hydrate sample H1 (an ultra-fine hydrate) and (b) H2 (a coarser hydrate).

were used meant that the influence of Ca(OH)2 particle size could be difference is, however, observable between the dry quicklime dose for­
evaluated independent of any slaking effects. The results (Figs. 9 & 10 mats, with somewhat better performance shown by the lower reactivity
and Table 4) show that finer Ca(OH)2 particles (H1, 50% <3 µm) quicklime C3. The relative insensitivity of causticisation performance to
perform somewhat better than the coarser H2 (50% <54 µm) Ca(OH)2 lime reactivity is consistent with the observation that causticisation is
particles at intermediate doses, but there is no significant difference at diffusion controlled (Xu et al. 1998) prior to the formation of TCA on Ca
higher dose levels. The similarity of the EC performance curves (Fig. 9a), (OH)2 particle surfaces (Salimi and Vaughan, 2016). These results sug­
make the mathematically determined MED less relevant for interpreta­ gest that, for high purity lime in the normal range of reactivity (t60
tion purposes, and is therefore not plotted as vertical lines in Fig. 10. An ranging between 0.5 and 9 min) (Zanin et al. 2019), there is no signif­
interesting observation of the solids assay (Fig. 10) is the higher pro­ icant effect of lime reactivity on causticisation performance. This also
portion of the precipitated monocarbonate phase generated using finer indicates that the differences observed between sample B2 and D, as
particles (H1) at the same dose levels, or even when comparing at the described earlier, cannot be attributed only to reactivity differences,
calculated MEDs. Monocarbonate overtakes TCA precipitation at a dose rather highlighting the effect of composition differences, while
of 15 g/L, and overtakes calcite precipitation at 21 g/L in the case of the acknowledging that chemical composition and reactivity may be
fine H1 particles, while for the coarser H2, monocarbonate is signifi­ related. For both samples C1 and C3, slaking resulted in a slightly
cantly less pronounced while more unreacted portlandite remains. This improved performance compared to unslaked dry quicklime addition,
indicates that at dose levels where causticisation reaction demand for mirroring the effect observed for sample B2.
lime is exceeded, finer Ca(OH)2 particles stimulate monocarbonate
formation, compared with coarser Ca(OH)2 particles. There are a range 4. Industrial implications
of factors that may influence Ca(OH)2 particle size in industrial appli­
cations (Ineich et al. 2017). However with all other factors being equal, The results presented throughout were expressed as CaO equivalent.
slaking conditions influence Ca(OH)2 particle size and hence also, the The actual quicklime dosage required would be higher, based on the
subsequent causticisation effects. CaO content of the lime, and might amplify the differences between
quicklime sources when comparing a quicklime dose requirement basis
3.3. Reactivity of quicklime – C1 and C3 rather than CaO equivalent content.
The influence of lime dose and properties on the nature of products
Quicklime samples C1 and C3 are both high purity quicklimes with in the residues suggests potential future avenues of investigation,
an identical chemical composition, but have different reactivities particularly in relation to LDH generation and use. The very high pro­
because of differing calcination conditions, as reflected by different t60 portion of seemingly stable LDH formation demonstrated using lime B2
values (Table 1). C1 is an ultra-reactive (t60 = 0.2 min) quicklime, while (Fig. 7a) with increasing lime dosage, may be of particular interest. If
C3 is a moderately reactive quicklime (t60 = 8.7 min). These were these findings are confirmed under actual plant conditions, the forma­
evaluated in a comparison pair BRBT, both in slaked and dry quicklime tion of LDH may be explicitly targeted and utilised in the Bayer pro­
addition formats to evaluate their dose-responses. cesses. The LDH properties that facilitate impurity removal from
The data in Fig. 11 and Table 5 show a minor difference in the alkaline leaching solutions are well known (Perotta and Williams, 1996;
causticisation performances of slaked limes prepared from these two Rosenberg and Armstrong, 2005; Nguyen and Lee, 2019). Deliberate
quicklime samples with different reactivities. A larger, but still small LDH generation could be promoted for impurity removal. It may also
enable flow sheet modification and optimisation, as suggested by
Rosenberg and Armstrong (2005), in which a portion of the spent liquor
Table 4
(after alumina precipitation but containing residual sodium aluminate
Summary of causticisation dose performance parameters for H1 and H2.
and sodium carbonate, Vind et al. 2018) is reacted with lime to generate
Parameter H1 H2
LDH (Eq. (6) and (7)) resulting in hemicarbonate and monocarbonate,
Maximum extent of causticisation (MEC) 67% 68% respectively. Hence, the LDH could then be recovered and used instead
Maximum effective dose (MED) 24 g/L 27 g/L of lime in the causticisation reactor (Eq. 11 or 12). These reactions also
Lime efficiency (LE) at MED 35% 31%
enable the recovery of aluminium.

9
C.A. du Plessis et al. Minerals Engineering 160 (2021) 106664

Fig. 11. Causticisation dose–response curves comparing an ultra-reactive lime (C1) with a moderately reactive lime (C3) in terms of extent of causticisation (EC) (a,
b) and lime efficiency (LE) (c, d) in slaked (a, c) and dry quicklime (b, d) formats.

The method described in this study allows for evaluation and po­
[Ca2Al(OH)6]2.(CO3)0.5.(OH).(5.5H2O) + 3.5Na2CO3 → tential optimisation of dry quicklime reagent dosing to be compared
4CaCO3 + 2NaAl(OH)4 + 5NaOH + 5.5H2O (11) with slaked lime reagent addition. The incentive for optimisation of dry
quicklime dosing is significant. Practically, approximately 160 kg water
[Ca2Al(OH)6]2.CO3⋅5H2O + 3Na2CO3 → is injected into a typical outside causticiser process per tonne of alumina
4CaCO3 + 2NaAl(OH)4 + 4NaOH + 5H2O (12) produced. An alumina plant with production capacity of 4 million
The advantages of this flow sheet modification would be: (1) that the tonnes Al2O3, would thereby inject 640,000 tonnes of water into the
NaOH content of the spent liquor stream to digestion is increased, (2) the causticiser per annum. To overcome the dilution effect of this water,
alumina concentration is reduced, and (3) water is removed from the evaporation is needed elsewhere in the circuit (Fig. 1), requiring
spent liquor by the hydration reaction (Eq. (6) and (7)) (Rosenberg and approximately 480,000 GJ per annum in evaporation energy. Assuming
Armstrong, 2005). a modest energy price of USD 5 per GJ, this represents a cost of USD 2.4
million per annum.

Table 5 4.1. CO2 emission considerations


Summary of causticisation dose performance parameters for C1 and C3 in slaked
and dry format. The efficient use of lime also warrants consideration in the context of
greenhouse gas (GHG) emissions. Reduced lime consumption will
Parameter Lime C1 Lime C3
reduce an alumina refinery’s overall GHG emissions due to the CO2
Slaked Dry Slaked Dry
footprint inherently associated with lime reagents. In addition to the
Maximum extent of causticisation (MEC) 69% 64% 71% 69% different properties of lime that influence lime utilisation efficiency, the
Maximum effective dose (MED) 20 g/L 21 g/L 16 g/L 17 g/L differences in CO2 footprint of various lime sources are also relevant and
Lime efficiency (LE) at MED 41% 37% 53% 49%
may compound the impact on CO2 emission reduction. The CO2

10
C.A. du Plessis et al. Minerals Engineering 160 (2021) 106664

Table 6 significant. Given the operational effort involved in the slaking process
Indicative comparison of CO2 intensity impact, for a scenario (B) where a 20% and the impact of slaked lime dosing on the water balance of Bayer re­
reduction in lime consumption is combined with a lower CO2 intensity lime fineries (and subsequent evaporation energy requirements), the
reagent. methods described in this study may be used to assess and compare the
Scenarios relative techno-economic effectiveness of various modes of lime dosing.
A B
Declaration of Competing Interest
Lime consumption 0.047 0.038 t lime / t alumina
Lime reagent CO2 intensity 1.3 1.1 t CO2 / t lime
Lime CO2 contribution 0.061 0.041 t CO2 / t alumina The authors declare that they have no known competing financial
Non-lime CO2 contribution 0.503 0.503 t CO2 / t alumina interests or personal relationships that could have appeared to influence
Overall plant CO2 intensity 0.564 0.544 t CO2 / t alumina the work reported in this paper.
Lime contribution to CO2 intensity 10.8 7.6 %
Overall CO2 footprint reduction 3.5 %
Acknowledgements

emission intensity of lime reagents range from 0.95 to 1.4 t CO2 / t lime The research was funded by Lhoist. The authors wish to acknowledge
and are mainly dependent upon the energy efficiency of calcination the valuable contributions of Fiona Benn, Tuyen Pham, Steve Peacock
kilns, fuel properties used as heat source and the extent of carbonate and Sophia Surin at the CSIRO, and Catherine Bouchard, Stéphane
conversion (Miner and Upton, 2002). In general, the lowest CO2 foot­ Meganck and Laurent Duvivier at the Lhoist Business Innovation Centre.
prints are achieved by lime calcined in parallel flow regenerative shaft
kilns, while rotary kilns generally result in larger CO2 intensity (Pir­ References
inger, 2017). The optimisation of lime utilisation efficiency combined
with CO2 intensity differences can be illustrated by way of an indicative Alcoa, 2003. Environmental Protection Statement. Pinjarra Refinery Efficiency Upgrade.
1 December 2003. Alcoa World Alumina Australia. http://www.epa.wa.gov.au/
scenario comparison (Table 6) based on industrially relevant parameters sites/default/files/Referral_Documentation/A1498_R1122_EPS_Efficiency%
(Alcoa, 2003). 20Upgrade-Final%201Dec03.pdf.
Arıkan, H., Demir, G.K. and Vural, S., 2019. Investigation of lime usage impacts on
bauxite processability at ETI Aluminyum Plant. Int J Ind Chem. 10, 57-66 (2019).
5. Conclusions
https://doi.org/10.1007/s40090-019-0171-x.
European Standard EN 12485, 2017. Chemicals used for treatment of water intended for
This study examined the influence of lime properties on causticisa­ human consumption – Calcium carbonate, high-calcium lime, half-burnt dolomite,
magnesium oxide, calcium magnesium carbonate and dolomitic lime – Test methods.
tion performance using a synthetic Bayer liquor (SBL) in a two-hour
Australian Standard. AS 1672.1, 1997. Lime and Limestones.
batch causticisation test. The indicators of performance used were the Boynton, R.S., 1980. Chemistry and Technology of Lime and Limestone, second ed. Wiley
extent of causticisation, lime efficiency, and proportions of phases in the & Sons, New York.
residues (where applicable). The lime properties investigated were: (1) Buttery, J.H.N., Patrick, V.A., Rosenberg, S.P., Heath, C.A., Wilson, D.J., 2002.
Thermodynamics of hydrocalumite formation in causticisation. Light Metals. 2002,
quicklime origin (reflecting different chemistries), (2) slaked versus dry 185–190.
quicklime dosage regimes, (3) Ca(OH)2 particle size, and (4) quicklime Cagnolatti, L.A., Kellogg, K.B. and Abboud, H.I., 1965. Process for treating aluminum
reactivity. containing ores. US Patent 3,210,155.
Chaix-Pluchery, O., Pannetier, J., Bouillot, J., Niepce, J.C., 1987. Structural
prereactional transformations in Ca(OH)2. Journal of Solid State Chemistry. 67 (2),
• Chemistry – A high purity lime performed significantly better than a 225–234. https://doi.org/10.1016/0022-4596(87)90358-6.
low purity lime, particularly when both were used in slaked format, i. Chaplin, N.T., 1971. Reaction of lime in sodium aluminate liquors. Light Metals. TMS
47–61. https://doi.org/10.1007/978-3-319-48176-0_27.
e. 71% extent of causticisation was achieved at 62% lime efficiency Connop, W.L., 1996. A new procedure for the determination of alumina, caustic and
with dose of 15 g/L CaO compared to 66% causticisation at 27% lime carbonate in bayer liquors. Fourth International Alumina Quality Workshop,
efficiency with a dose of 25 g/L CaO. Darwin, Australia, pp. 322–330.
Cumberland, W.N., Fong, Y., Yu, X., Defawe, O., Frahm, N., De Rosa, S., 2015. Nonlinear
• Slaking versus dry addition – Slaking generally resulted in calibration model choice between the four and five-parameter logistic models.
improved performance when using high purity quicklime, although Journal of Biopharmaceutical Statistics. 25 (5), 972–983. https://doi.org/10.1080/
the contrary effect of slaking was found for the low purity quicklime. 10543406.2014.920345.
Cundiff, W.H., 1950. Process for extraction of alumina from aluminous ores. US Patent
• Particle size – Ca(OH)2 particle size distribution of reagents dosed in
2,522,605.
hydrate format was found to have a modestly beneficial effect on De Silva L, Smith P and Xu B., 2012. Recovery of soda from Bayer residue via mud
lime efficiency: dry hydrated lime with 50% <3 µm yielded 67% causticisation (complex causticisation). 9th International Alumina Quality Workshop
causticisation and 35% lime efficiency at the maximum effective (AQW 2012); 18-22 March 2012; Perth, WA, Australia. AQW Inc; 2012. Poster.
https://publications.csiro.au/rpr/download?pid=csiro:EP116275&dsid=DS2.
dose, compared to 67% causticisation and 31% lime efficiency for De Witt, G.C., Roach, G.I.D. and Reid, G., 2000. Method for causticisation of alkaline
the same lime with 50% –54 µm. solutions. International Patent Application No. PCT/AU00/00346.
• Reactivity – Quicklime reactivity was found to have no significant François, M., Renaudin, G. and Evrard, O., 1998. A cementitious compound with
composition 3CaO.Al2O3.CaCO3.11H2O. Acta Crystallographica Section C: Crystal
effect when used in slaked format, and a slightly beneficial effect Structure Communications. 54(9), pp. 1214-1217. Crystallographic Information File
when dosed in dry quicklime format. (CIF). http://scripts.iucr.org/cgi-bin/paper?S0108270198004223.
Gadagkar, S.R., Call, G.B., 2015. Computational tools for fitting the Hill equation to
dose–response curves. Journal of Pharmacological and Toxicological Methods. 71,
The data demonstrate that lime properties can significantly influence 68–76. https://doi.org/10.1016/j.vascn.2014.08.006.
causticisation performance and optimal reagent dose in synthetic Bayer Ineich, T., Degreve, C., Karamoutsos, S., du Plessis, C., 2017. Utilization efficiency of
plant liquor. It is likely that performance variations would also be found lime consumption during magnesium sulfate precipitation. Hydrometallurgy. 173,
241–249. https://doi.org/10.1016/j.hydromet.2017.09.001.
when using actual plant liquors and under plant operating conditions. Lager, G.A., Downs, R.T., Origlieri, M., Garoutte, R., 2002. High-pressure single-crystal
The method demonstrated in this study may be used to assess a wider X-ray diffraction study of katoite hydrogarnet: Evidence for a phase transition from
range of processing and liquor conditions, to evaluate and compare the Ia3d→ I4‾3 symmetry at 5 GPa. American Mineralogist. 87 (5–6), 642–647. https://
doi.org/10.2138/am-2002-5-606.
performance of different lime sources, chemistries and formulations
Libby, S.C., 1983. The effects of lime/limestone characteristics in causticization, Light
during causticisation of alumina plant liquors. Metals 1983. TMS, Warrendale, PA, pp. 275–293.
Additionally, the method may be used to evaluate the potential Markgraf, S.A., Reeder, R.J., 1985. High-temperature structure refinements of calcite and
performance benefit of slaking compared with dry quicklime addition. magnesite. American Mineralogist. 70 (5–6), 590–600.
Miner, R., Upton, B., 2002. Methods for estimating greenhouse gas emissions from lime
The data from this study indicated that the differences between quick­ kilns at kraft pulp mills. Energy. 27 (8), 729–738. https://doi.org/10.1016/S0360-
limes when dosed in slaked versus dry reagent addition formats may be 5442(02)00017-8.

11
C.A. du Plessis et al. Minerals Engineering 160 (2021) 106664

Nguyen, T., Lee, M.-S., 2019. The removal of silicate(iv) by adsorption onto Smith, P., 2017. Reactions of lime under high temperature Bayer digestion conditions.
hydrocalumite from the sodium hydroxide leaching solution of black dross. Hydrometallurgy. 170, 16–23. https://doi.org/10.1016/j.hydromet.2016.02.011.
Processes. 7, 612. https://doi.org/10.3390/pr7090612. Suss, A., Panov, A., Kuznetzova, N., Kuvyrkina, A., Damaskin, A., Damaskina, A.,
Perotta A.J. and Williams, F., 1996. Layered double hydroxide formation in Bayer liquor Shmigidin, Y., 2008. In: The impact of magnesium on tricalcium aluminate (TCA)
and its promotional effect on oxalate precipitation. Light Metals 1996. TMS, filter aid properties. AQW Inc., Darwin, Australia, pp. 37–43.
Warrendale, PA. pp. 17-28. Tóth, V., Ádok, M., Pallagi, A., Kukovecz, A., Kónya, Z., Sipos, P., Palinko, I., 2014.
Piringer, H., 2017. Lime shaft kilns. Energy Procedia 120, 75–95. https://doi.org/ Synthesis and properties of CaAl-layered double hydroxides of hydrocalumite-type.
10.1016/j.egypro.2017.07.156. Chemical Papers. 68 (5), 633–637. https://doi.org/10.2478/s11696-013-0500-z.
Power, G., Loh, J.S.C., Vernon, C., 2012. Organic compounds in the processing of lateritic Vind, J., Alexandri, A., Vassiliadou, V., Panias, D., 2018. Distribution of selected trace
bauxites to alumina Part 2: Effects of organics in the Bayer process. elements in the Bayer process. Metals. 8 (5), 327. https://doi.org/10.3390/
Hydrometallurgy. 127, 125–149. https://doi.org/10.1016/j.hydromet.2012.07.010. met8050327.
Roach, G.I.D., 2000. The equilibrium approach to causticisation for optimising liquor Wang, L., Tessier, P.J.C., Englezos, P., 1994. Thermodynamics and kinetics of the Kraft
causticity, Light Metals 2000. TMS, Warrendale, PA. pp. 97-103. https://doi.org/ causticizing reaction. Canadian Journal of Chemical Engineering. 72 (2), 314–320.
10.1007/978-3-319-48176-0_30. https://doi.org/10.1002/cjce.5450720219.
Rosenberg, S.P., Armstrong, L., 2005. Layered double hydroxides in the Bayer process: Wellington, M., 2015. Effect of lime causticization on silica, calcium, oxalate and organic
Past, present and future. Light Metals. TMS 1, 157–161. https://doi.org/10.1007/ carbon levels in Bayer process washer overflow liquor. Chem Xpress. 8 (3), 208–213.
978-3-319-48176-0_31. Wellington, M., Valcin, F., 2007. Impact of Bayer process liquor impurities on
Rosenberg, S.P., Wilson, D.J., Heath, C.A., 1999. Improved Bayer Causticisation. causticization. Industrial & Engineering Chemistry Research. 46 (15), 5094–5099.
International Patent Application No. PCT/AU99/00757. https://doi.org/10.1021/ie070012u.
Rosenberg, S.P., Wilson, D.J. and Heath, C.A., 2001. Some aspects of calcium chemistry Whittington, B.I., 1996. The chemistry of CaO and Ca(OH)2 relating to the Bayer process.
in the Bayer process. Light Metals 2001. TMS, Warrendale, PA. pp. 19-25. https:// Hydrometallurgy. 43 (1–3), 13–35. https://doi.org/10.1016/0304-386X(96)00009-
doi.org/10.1007/978-3-319-48176-0_28. 6.
Rosenberg, S.P., 2008. Alumina recovery. US Patent Application No. US 2004/ Whittington, B.I., Cardile, C.M., 1996. The chemistry of tricalcium aluminate
0170546A1. hexahydrate relating to the Bayer industry. International Journal of Mineral
Runčevski, T., Dinnebier, R.E., Magdysyuk and O.V., Pöllmann, H., 2012. Crystal Processing 48 (1–2), 21–38. https://doi.org/10.1016/S0301-7516(96)00011-7.
structures of calcium hemicarboaluminate and carbonated calcium Whittington, B.I., Fallows, T.M., Willing, M.J., 1997. Tricalcium aluminate hexahydrate
hemicarboaluminate from synchrotron powder diffraction. Acta Crystallographica. (TCA) filter aid in the Bayer industry: Factors affecting TCA preparation and
B68, 493-500. https://doi.org/10.1107/S010876811203042X. morphology. International Journal of Mineral Processing. 49 (1–2), 1–29. https://
Ruiz-Agudo, E., Rodriguez-Navarro, C., 2010. Microstructure and rheology of lime putty. doi.org/10.1016/S0301-7516(96)00035-X.
Langmuir. 26 (6), 3868–3877. https://doi.org/10.1021/la903430z. Xu, B.A., Giles, D.E., Ritchie, I.M., 1997. Reactions of lime with aluminate-containing
Salimi, R., Vaughan, J., 2016. Crystallisation of tricalcium aluminate from sodium solutions. Hydrometallurgy. 44 (1–2), 231–244. https://doi.org/10.1016/S0304-
aluminate solution using slaked lime. Powder Technology. 294, 472–483. https:// 386X(96)00051-5.
doi.org/10.1016/j.powtec.2016.03.017. Xu, B.A., Giles, D.E., Ritchie, I.M., 1998. Reactions of lime with carbonate-containing
Sipos, P., 2009. The structure of Al(III) in strongly alkaline aluminate solutions – A solutions. Hydrometallurgy. 48 (2), 205–224. https://doi.org/10.1016/S0304-386X
review. Journal of Molecular Liquids. 146, 1–14. https://doi.org/10.1016/j. (97)00079-0.
molliq.2009.01.015. Young, R.C., 1982. Chemistry of Bayer Liquor Causticisation. Light Metals. TMS,
Sizyakov, V.M., Tikhonova, E.V., 2012. Physical and chemical transformations of Warrendale, PA, pp. 97–117.
calcium in aluminate solutions from alumina production. Russian Journal of Applied Zanin, M., Lambert, H., du Plessis, C.A., 2019. Lime use and functionality in sulphide
Chemistry. Zhurnal Prikladnoi Khimii. 85 (11), 1746–1750. https://doi.org/ mineral flotation: A review. Minerals Engineering. 143, 105922 https://doi.org/
10.1134/S1070427212110043. 10.1016/j.mineng.2019.105922.

12

You might also like