organic compounds in the processing of lateritic bauxites

Hydrometallurgy 127-128 (2012) 125–149

Contents lists available at SciVerse ScienceDirect

Hydrometallurgy

j ourna l homepage: www.e lsev ie r .com/ locate /hydromet

Organic compounds in the processing of lateritic bauxites to alumina Part 2: Effects oforganics in the Bayer process

Greg Power a,⁎, Joanne S.C. Loh b, Chris Vernon b

a Arriba Consulting Pty Ltd, PO Box 975, Canning Bridge, Applecross WA 6153, Australiab CSIRO Process Science and Engineering/Parker Centre CRC for Integrated Hydrometallurgy Solutions, PO Box 7229, Karawara WA 6152, Australia

⁎ Corresponding author. Tel.: +61 8 9334 8940.E-mail addresses: [email protected]

(G. Power).

0304-386X/$ – see front matter © 2012 Elsevier B.V. Alhttp://dx.doi.org/10.1016/j.hydromet.2012.07.010

a b s t r a c t
a r t i c l e i n f o
Article history:Received 3 May 2012Received in revised form 23 July 2012Accepted 23 July 2012Available online 1 August 2012

Keywords:Bayer processAlumina productionOrganic compoundsBauxite

This review is the second of a series examining the history and current state of knowledge of the science related toorganic compounds in the Bayer process for the extraction of alumina from lateritic bauxites. The series of reviewsprovides a compilation of the information available from the public literature, and critical analysis of the researchthat has been carried out in this area in the past 50 years. It points the way to future opportunities for research toassist inmitigating the high costs that organics impose on the industry globally ($500 Mpa in Australia alone). Part2 provides unique insights into the effects on the Bayer process of organic compounds in the liquor.

© 2012 Elsevier B.V. All rights reserved.

1. Introduction

It was noted in Part 1 of this review series that organic mattercontained in bauxite is the main source of the organic compounds inBayer process liquors which cost the Australian alumina industry alonein excess of $500 m (US$520 m) per year (Power and Loh, 2010). This es-timate is consistent with figures provided by Teas and Kotte (1980) for arange of other bauxites. It was further noted that:

• Organic compounds build up in the recirculating process liquor toconcentrations determined by a complex interaction of inputs, out-puts and reactions;

• The presence of organic compounds leads to increased rawmaterialsusages, reduced production efficiency and lower product quality;

• Mitigation of the effects of organic compounds requires the installa-tion of additional processing equipment which increases capital andoperating costs; and

• Volatile organic compounds emitted fromdigestion and other parts ofthe refinery process may raise environmental and health concerns.

It is the purpose of Part 2 of the review series to provide a system-atic compilation of the current state of knowledge and research intothe effects that organic compounds have on the Bayer refinery pro-cess. Part 2 examines the effects that organics have on the physicaland chemical nature of the liquor, specific effects on the productivityof the liquor through adsorption to the surfaces of precipitation seedand the formation of complexes with the aluminate ion, effects on the

.au, [email protected]

l rights reserved.

purity, particle size distribution and other physical properties of theproduct alumina, scale formation on the surfaces of vessels andpipes, and the production of volatile gases including hydrogen.

2. Liquor properties

2.1. Solution composition and nomenclature

The first step of the Bayer process is the digestion of aluminium hy-droxideminerals by reaction with caustic soda. Digestion of aluminiumtri-hydroxide (gibbsite) for example, consumes hydroxide as follows:

AlðOHÞ3 þ OH−→½AlðOHÞ4�− ð1Þ

This causes a reduction in the “free caustic” (OH−) concentration inthe liquor by binding one OH− ion (“bound caustic”) for each Al unitdissolved. Bound caustic is released later in the process, when this equa-tion is reversed to recover pure Al(OH)3 in the precipitation stage of theBayer process circuit. The sum of the free caustic and bound caustic isused by Bayer process operators as a measure of the overall causticityof the solution, and is given the title “total caustic” with the symbol“TC”, or simply “C” (shaded box, Fig. 1). The convention within theindustry is to express TC in units of its sodium carbonate equivalent(rather than moles), hence:

TC ¼ ½OH−� þ ½AlðOH4Þ−�; expressed as equivalent g=LNa2CO3 ð2Þ

The chemistry of the formation of organic compounds in Bayer pro-cess digestion due to the degradation of complex organic materials in

http://dx.doi.org/10.1016/j.hydromet.2012.07.010

mailto:[email protected]

mailto:[email protected]

http://dx.doi.org/10.1016/j.hydromet.2012.07.010

http://www.sciencedirect.com/science/journal/0304386X

Total

Soda

(TS) Total Alkali(TA)

Non-Alkaline Soda

SodiumCarbonate

Total Caustic

(TC)

Inorganicnon-alkaline soda

Sodium oxalate

Otherorganics

“Humates”

Aromatics

Formate & Acetate

Na2SO4

NaCl

NaF

Na2P2O5

Na2V2O5

All sodium salts

NaOH+

NaAl(OH)4

+Na2CO3

NaOH+

NaAl(OH)4

NOOC

TOC

Fig. 1. Schematic representation of distribution of species in a Bayer process liquor, and summary of nomenclature (adapted from Smith (2011)). Note: TS, TA, TC and TOS areexpressed as equivalent g/L Na2CO3.

126 G. Power et al. / Hydrometallurgy 127-128 (2012) 125–149

bauxite is described in detail in Part 1 (Power and Loh, 2010). The diges-tion of organic compounds also consumes hydroxide ions, however inthis case the products are organic acid anions and carbonate whichare not recovered in the precipitation stage of the process. For the pur-pose of determining the impacts this has on the properties of the liquor,it is useful to consider the overall process of the digestion of organicmatter in the following general terms:

Complex OrganicMatter þ OH−→Organic Acid Anions þ CO

2−3 ð3Þ

In addition to organic compounds, there are inorganic compoundswhich are formed in digestion either by dissolution of salts or mineralsor from degradation of organic compounds containing sulphur or nitro-gen. Hence, the sodium ion content of any liquor, which is convention-ally expressed as the total soda content (TS), may be considered to be acombination of alkaline and non-alkaline soda, as summarised in theleft-hand box in Fig. 1.

The non-alkaline soda is made up of inorganic and organic sodiumsalts, as summarised in the middle box of Fig. 1. The organic soda,normally expressed in terms of total organic carbon (TOC) in unitsof g/L carbon, or sometimes as total organic soda (TOS) in units ofequivalent g/L sodium carbonate, may be further sub-divided into ox-alate carbon (expressed either as g/L sodium oxalate, oxalate carbon(OC) in g/L carbon, or oxalate soda (OS) in units of equivalent g/L so-dium carbonate) and non-oxalate organic carbon (NOOC, g/L carbon)(Lectard and Nicolas, 1983). The NOOC is made up of formate andacetate, aromatic compounds (mainly carboxylic acid anions), highmolecular weight (“humate”) anions and other organics, primarilylow to medium molecular weight carboxylic anions (Guthrie et al.,1984; Lever, 1978; Picard et al., 2002; Power and Loh, 2010; Poweret al., 2011a).

The only sodium ions that are useful to the Bayer process are thoseassociated with hydroxide and aluminate anions, i.e. the “caustic” orTC portion highlighted in Fig. 1. The remainder which is “tied up with”(i.e. provides charge balance for) carbonate, organic acid anions and in-organic anions, provides no value and is a technical and economic bur-den on the liquor.

2.2. Physical properties

Knowledge of the physical properties of the liquor is essential for thechemical engineering design and operation of hydrometallurgical opera-tions such as Bayer process refineries. For example, values of liquor

density and viscosity are fundamental to the design of the vessels andpipes, agitators, pumping systems, and solid/liquid separation systems in-cluding settlers,filters, cyclones and centrifuges that are used in refineries.The boiling point elevation and specific heat of the liquor are important tothe design and ongoing efficient operation of heating and cooling systemsrequired for digestion, flashing and evaporation, all of which are essentialunit processes in Bayer process refineries (Dewey, 1981).

In general, the parameters which have been found to have the mostinfluence on the physical properties of Bayer liquors are the total soda(TS) and alumina contents, with TS having the dominant impact. Teasand Kotte (1980) estimated that 10 g L−1 of any impurity increases theliquor density by approximately 1%, but noted that this does depend tosome extent on the nature of the impurity. Nicolas (1986) concludedthat knowledge of the ionic strength of plant liquors was sufficient to en-able prediction of the main effects of impurities on the properties of theliquor. He proposed that the contribution of organic compounds to theionic strengthmay be calculated using a “carbon number”, n, which is de-fined as the mean number of carbon atoms in the mixture of organic an-ions per sodium ion required for charge balance in the solution. The ionicstrength associatedwith the organics can then be calculated using the fol-lowing equation (note that there is a typographical error in the originalpaper which is corrected here):

I ¼ 0:5�m� 1=nþ N=nð Þ2n o

ð4Þ

Where:

I ionic strength of solutionm molality of carbon in solutionn average number of carbon atoms per sodium ion (“carbon

number”)N average number of carbon atoms per organic anion

This is consistent with Dewey's (1981) finding that the boiling pointrise of Bayer process solutions could be accurately estimated fromknowledge of the ionic strength of the solution. He also found thatthis enabled the heat of vaporisation of Bayer liquors to be calculatedby rigorous thermodynamic relations with the aid of an equation ofstate for water or the steam tables. Some uncertainty was assigned tothe effect of the organic compounds present, not in relation to the as-sumption that it was the ionic strength that was important, but in rela-tion to how to estimate their contribution to the ionic strength. The

127G. Power et al. / Hydrometallurgy 127-128 (2012) 125–149

difficulty lay in determining the number of acid groups per mole of car-bonwith the techniques for chemical analysis available at the time; thisis a straightforwardmatterwith the analytical techniques nowavailable(Power et al., 2011b). It is interesting to note that Dewey found a signif-icant difference between liquors from plants operating high tempera-ture (>200 °C) and low temperature (143 °C) digestion. The lowtemperature liquor was found to be much less oxidised (i.e. fewer acidgroups) than the high temperature liquor, which meant that a givenlevel of organics in the high temperature liquor had a greater effect onboiling point rise than the same level of organics in the low temperatureliquor (Dewey, 1981).

Nicolas (1986) proposed the following empirical correlation (inwhichconcentrations are expressed in g L−1, andD andDo are the solution den-sities, respectively with and without impurities, in g cm−3) for the effectof organic compounds on liquor density:

D ¼ Do þ 10−3 � ð½Na2CO3� þ ½NaCl� þ ½Na2SO4� þ 0:7 � ½TOC�Þ ð5Þ

This equation implies that the contribution of organics can be alsoreduced to a parameter related to ionic strength for the purpose of cal-culating their contribution to liquor density. The equation is also consis-tent with the earlier observations of Teas and Kotte (1980).

Königsberger et al. have recently published a series of papers reportingthe development of a thermodynamic model for calculating the effects ofimpurities on liquor properties (Königsberger et al., 2005a, 2005b, 2005c,2006). The model is based on Pitzer equations (Pitzer, 1991), and at thetime of writing included ten components (water plus sodium hydroxide,aluminate, carbonate, sulphate, chloride, fluoride, oxalate, formate andacetate). The ultimate aim of thiswork is to provide a thermodynamicallyconsistent means for calculating a wide range of solution properties, in-cluding heat capacities, enthalpies, entropies, densities, water activities,activity coefficients, vapour pressures and boiling point elevations, aswell as solubilities of various compounds of interest, over thewide rangesof temperature and concentration of interest (Königsberger, 2008).

The authors have concluded that water activity is the key quantitygoverning the thermodynamic behaviour of concentrated process solu-tions (Königsberger et al., 2009). Consistent with the conclusions ofNicolas (1986), the overall effect of the presence of organic compoundsis therefore qualitatively similar to that of other (inorganic) anions in re-lation to their effect on the physiochemical properties of the liquor. Inorder to calculate the effects quantitatively, measurements are requiredto establish the necessary coefficients for each component and propertyof interest over a range of conditions, and at elevated temperatures inparticular. Some work has been done along these lines. For example,the apparent molar volumes and heat capacities of three dicarboxylates,oxalate, malonate and succinate, have been studied at 25 °C in dilute so-lutions and extrapolated to high ionic strengths (Tromans et al., 2005).These compounds are important components in Bayer liquors which,according to the authors, can be used as model impurities for estimatingthe overall effects of organic compounds on the properties of the liquor.This suggests that Nicolas' approach of defining a “carbon number” forindividual liquors may, in combination withmodern methods of chem-ical analysis, be worth revisiting as a way of simplifying the estimationof the effect of organics on the physical properties of Bayer processliquors.

3. TOC concentration

The steady-state concentrations of organic compounds in Bayer liquorare the result of a dynamic balance of inputs and outputs, which are com-plex, notoriously difficult to model in detail, and only scantily reported inthe literature (Power and Loh, 2010). Nevertheless, considerable under-standing can be gained bymaking a few simplifying assumptions and ap-plying basic chemical kinetics principles, as follows.

The input of organic compounds to Bayer liquor inmodern refineriesis almost entirely from the bauxite, but process additives, in particularstarch, can also be a significant input if it is used as a flocculent(Solymár et al., 1996). Organic compoundsmay exit the refinery liquorby a number of mechanisms, which may be divided into those whichare an inevitable consequence of the basic refinery process, and thosewhich are deliberate sub-processes for removal of organics. The mainprocesses in the former category are direct loss of liquor with the bauxiteresidue or other leakagemechanisms, and conversion to carbonate by ox-idation, which occurs mainly in the digestion stage of the process. Delib-erate removal strategies include crystallisation and removal of oxalate(by any of a variety of methods), direct pyrolysis (“liquor burning”), andreaction with air or oxygen in a wet oxidiser. These and other methodswill be examined in detail in Part 3 of this review series.

The time dependence of the concentration of organic compounds inliquor may be modelled to a first approximation by assuming that theinput from bauxite is in direct proportion to the rate of alumina produc-tion, and that the volumetric rate of liquor loss is constant. These as-sumptions lead to the following equation for the rate of change ofconcentration with time:

d TOC½ �ð Þ=dt ¼ I−L⋅ TOC½ � ð6Þ

In this equation,

[TOC] concentration of total organic carbon in liquor (g/L)I rate of input of TOC (g/s)L volumetric rate of loss of liquor (L/s)

If I and L are both constant, the steady-state TOC concentration issimply given by:

TOC½ � ¼ I=L ð7Þ

Fluctuations in [TOC] can therefore be understood in terms ofchanges in I, L or both. On the other hand, if the TOC concentration isnot at steady-state and I and L can be considered as constants, thenthe rate of change of [TOC] should follow a first-order law. This wouldbe the case, for example, for the build-up of TOC in the liquor of a newrefinery, or for the reduction in liquor TOC in a refinery which adds anorganic removal process (such as liquor burning).

The former effect is well illustrated by the build-up of TOC in liquor atthe Worsley refinery from its start-up in 1984 to a reasonably stablesteady-state in 1990–91 (Power, 1991). Over this period, the TOC build-up appears to occur in twodistinct stages, as shown in Fig. 2. The TOC con-centration follows an initial first-order curve quite closely from start-upuntil mid-1986 (Stage 1), after which the data suggests that the input oforganic compounds increased by about 14% while the output rose by208%, resulting in a second first-order curve from mid-1986 to 1991(Stage 2). The increased input in Stage 2 may for example be related toa change in bauxite quality and/or to an increase in plant yield resultingin greater bauxite input and consequent residue volume per volume ofliquor, and/or an increase in the amount of starch used per volume of li-quor (starch was used as a flocculent until 1993 (Kahane and McRae,1996)). The increased output in Stage 2 indicates that the volumetricrate of liquor loss doubled in 1986 and then remained constant at thatthe new level for the remainder of the period.

4. Liquor productivity

The economics of the Bayer process is strongly influenced by the li-quor productivity, which is the degree towhich alumina can be broughtinto solution in digestion according to Eq. (1) and subsequently be re-moved from the solution by precipitation in the presence of gibbsiteseed (Eq. (8)) before it is returned to digestion to again extract alumina

0

10

20

30

TO

C (

g/L

) Stage 1 Stage 2

03-Jul-83 14-Nov-84 29-Mar-86 11-Aug-87 23-Dec-88 07-May-90 19-Sep-91

Fig. 2. Build-up of TOC concentration in Worsley refinery liquor from start-up (Power, 1991), showing the fit to a simple two-stage first-order model.


from fresh bauxite. Typically only about half of the alumina contained inthe green (pregnant) liquor is recovered in the precipitation step, thebalance remaining to circulate unproductively with the liquor. Hencethe precipitation yield, which is the amount of alumina precipitated(as gibbsite) per volume of liquor (at a specified sodium concentra-tion) (Pearson, 1955), is a key factor in the production costs of a re-finery (AMIRA, 2001; Teas and Kotte, 1980).

Al OHð Þ4� �− →

SEED

Al OHð Þ3Al OHð Þ3 þ OH− ð8Þ

It has been known since at least the 1930s that organic com-pounds have a negative influence on Bayer liquor productivity (Volfand Pudovkinoj, 1935; Yamada et al., 1973). There are three mainways in which this may occur. Firstly, the presence of organic acid an-ions (and the carbonate ions that are present as a result of the oxida-tive degradation of organic anions) requires an equivalent amount ofcations, in particular sodium ions, to maintain charge balance, whichreduces the number of sodium ions available to balance hydroxideand aluminate ions. Secondly, organic compounds may adsorb to seedsurfaces and affect the kinetics of gibbsite precipitation, which reducesthe precipitation yield (see below). Finally, adsorption effects and theinfluence of crystallisation of solid sodium oxalate can have a major ef-fect on the size distribution of the precipitated gibbsite. This has an in-direct effect on precipitation yield because of the measures requiredto maintain an acceptable particle size distribution.

4.1. Total soda—thermodynamic effect

In order to understand this effect it is necessary to examine the effectof the total caustic concentration, TC, on yield. The digestion stages inalumina refineries are generally designed to extract the soluble aluminaphases from bauxite to the point that the liquor is almost saturated inaluminate at the digestion temperature. The goal of the precipitationstage is then to recover as much as possible of that alumina in theform of solid gibbsite by cooling and diluting the liquor, and seeding itwith crystalline gibbsite. Themaximum possible yield in such a processcorresponds to the aluminate concentration being in equilibrium withthe solids at the new temperature. This concept is illustrated for agibbsitic bauxite in Fig. 3, which is adapted from the data of Kotte

(1981). The theoretical maximum yield, YT, is then the product of thechange in ratio, ΔA/TC, and TC:

YT ¼ ΔA=TCð Þ �TC g=L ð9Þ

The available ΔA/TC decreases slightly as TC rises, however this ef-fect is small in relation to the multiplicative effect (Eq. (9)), so thatthe theoretical yield is almost exactly proportional to TC, as shownin Fig. 4. For this reason, refineries operate at the maximum possibleTC, which is determined by the maximum total soda concentration,TS. The TS includes the soda associated with inorganic and organicimpurities as well as hydroxide, so any increase in the concentrationof impurities correspondingly reduces the operating TC, which direct-ly reduces the available yield. Because the influence of TC is related toequilibria, this is essentially a thermodynamic effect. The presence ofTOC can be thought of as decreasing the carrying capacity of the liquorfor aluminate ions.

As an order of magnitude example, the theoretical effect of 20 g/LTOC in liquor is shown in Fig. 4. (20 g/L was chosen as typical of Bayer li-quors derived fromAustralian and Caribbean bauxites, which vary in therange of 10 to 40 g/L (Arnswald et al., 1991; Lever, 1978; Power, 1991;Rosenberg and Healy, 1996; Solymár et al., 1996)). Assuming that 1 g/LTOC is equivalent to 2 g/L sodium carbonate (i.e. a TOS/TOC ratio of 2(Rosenberg and Healy, 1996)) the loss of theoretical yield caused by20 g/L TOC is about 9 g/L Al2O3, or 7%.

4.2. Seed poisoning—kinetic effects

4.2.1. Background: general effects of organic compoundsIt has been noted that about 90 million tonnes of aluminium hydrox-

ide is produced by precipitation fromBayer process liquors annually (Li etal., 2003), which is probably the greatest rate of production of all precip-itates in hydrometallurgy. It is a slow chemical reaction, requiring highseed densities (250 to 1000 g/L), long holding times (24 to 36 h) andlow temperatures (the liquor is cooled from about 80 °C to about 60 °Cduring precipitation) to maximise the yield of product (Newchurch andMoretto, 1990; Pearson, 1955). As such, precipitation is the “rate limiting”stage in the Bayer process, and requires significant investment in tanks,heat exchangers and associated equipment to provide the holding timesand temperature profiles necessary to achieve acceptable yields. The

0.200

0.300

0.400

0.500

0.600

0.700

0.800

0.900

60 80 100 120 140 160

A/T

C R

atio

TC=250g/L

TC=200g/L

TC=150g/LDIGESTION

PRECIPITATION

ΔA/TC

Temperature (oC)

Fig. 3. Theoretical A/TC change under equilibrium conditions for a gibbsitic bauxite digested at 150 °C followed by precipitation at 70 °C (adapted from Kotte (1981)).


precipitation reaction is therefore of considerable commercial impor-tance, and so has been the subject of extensive research in the past severaldecades (e.g. see Vernon et al. (2002)).

Lowering the temperature decreases the equilibrium solubility ofgibbsite, and so increases the yield provided the residence time is suf-ficient to overcome the competing effect of a slower reaction rate(Chaubal, 1990). Achieving the best yield outcome requires carefuloptimisation of conditions, in particular the temperature profile inprecipitation.

It has been knownsince at least the 1950s that impurities in the liquor,both inorganic and organic, can affect precipitation yield considerably(Hudson, 1987). Pearson (1955) reported reductions in precipitationrates in the presence of organic compounds, which he attributed to poi-soning of the seed surface. He listed “saponin, gum Arabic, cane sugarand numerous other organic substances containing hydroxyl groups”as effective poisons, but noted that oxalate (which has no hydroxylgroups) is not. Pearson made a number of other insightful observations

60

80

100

120

140

140 180

TC

Th

eore

tica

l Yie

ld (

g/L

Al 2O

3)

Fig. 4. Theoretical yield as a function of TC, showing the impact of 20 g/L

that have largely stood the test of time. Hedescribed the possiblemodesof action of poisons as broadly falling into three types: (a) general over-all slowing of precipitation rate; (b) temporary inhibition, which hassince become related to an “induction period”; and (c) permanent inhi-bition. He observed that type (a) generally appeared to be caused by anelevation of the equilibrium aluminate concentration, i.e. an increase ingibbsite solubility. To quantify the effect he defined a “poisoning factor”,whichwas found to decrease with temperature. This was interpreted asevidence that the poisoning was related to adsorption of impurities onthe gibbsite surface. It has been suggested that poisons enter the liquoron contact with bauxite during digestion, but do not generally build upin the liquor to observable concentrations because of a combination ofremoval processes, in particular continuous degradation to simplercompounds and adsorption to gibbsite (Anderson et al., 2001).

The (1980) reported the identification of numerous organic com-pounds that were associated (apparently by adsorption) with the finegibbsite seed at a particular refinery. Of these, glucoisosaccharinate was

220 260

(g/L)

Δyield = 7%

TOC = 20g/L ~ ΔTC = 40 g/L

TOC (assuming a TOS/TOC ratio of 2 (Rosenberg and Healy, 1996)).


singled out for special study. It was found that glucoisosaccharinate, adegradation product of sugars thatmay bepresent as a result of the deg-radation of bauxite organics and/or starch added as a flocculent, had asignificant inhibitory effect on precipitation. The (The, 1980) postulatedthat glucoisosaccharinate was present in trace amounts in the Bayerprocess liquor and was concentrated at the gibbsite surface by adsorp-tion, where it interfered with precipitation by interacting with alumi-nate in such a way as to increase the solubility of aluminate at theliquid/solid interface. These and other related results (Solymár et al.,1996) undoubtedly accelerated the change from starch to syntheticflocculants as the principal settling and fining agents in Bayer processpractice (Ryles and Avotins, 1996).

4.2.2. Precipitation mechanismsA deeper understanding of the role of organic compounds in precip-

itation requires knowledge of themechanism(s) of the crystallization ofgibbsite. Indeed, the effects of organic compounds can be used as aprobe to investigate precipitation, as has been demonstrated the workby Rossiter et al. (1999, 1996) and Watling et al. (2000) on the effectsof gluconate in solution during crystallization, and by Beckham et al.(2005) and Loh et al. (2010a) who used measurements of precipitationyield inhibition to estimate the active surface areas of gibbsite seed.

These observations on the actions of poisons highlight the impor-tance of surface chemical reactions in the crystallization of gibbsite.This is consistent with the observed kinetics of precipitation, whichshows that it is not a simple first order process, but instead the rategenerally depends on the square of the aluminate supersaturation.The first reports of the square law appear to be the equation givenby Pearson (1955). A slightly different equation was presented by King(1973), and other variants have subsequently been developed to im-prove the prediction of reaction rates over wider ranges of solution con-centrations, temperatures and seed additions (Cornell et al., 1999;Veesler and Boistelle, 1994; Vernon et al., 1999).

King (1973) interpreted the square law in terms of a proposedpre-equilibrium between the predominant [Al(OH)4]− species and adimeric species at the gibbsite surface; using a linear growth model,the slowness of the reaction was interpreted as due to a very lowequilibrium concentration of the dimer at the growing solid surface.This appears to be consistent with the hypothesis of a dimeric “growthunit” which has been proposed (Paulaime et al., 2003; Seyssiecq et al.,1999) based on molecular modelling and crystallographic considerations(the “growth unit” concept in crystallization hypothesises the formationin solution of a fundamental building block of the crystal which can read-ily be incorporated into the crystal surface (Li et al., 1999; Zhong et al.,1994)). However other authors have proposed different growth unitsfor gibbsite. In particular, Wu et al. (2005) cite the existence of a growthunit of one aluminate ion with two waters attached, and another (ener-getically preferred) unit with six aluminium ions, based on thermody-namic calculations and in situ Raman measurements. A number of thesepossibilities, in particular the six aluminium ion growth unit, had beenpreviously suggested by Zambo (1986) based on theoretical analysis ofthe bulk properties of the solutions.

However, it has also been shown that quadratic dependences of therate on supersaturation can result fromother growthmodels. At low su-persaturation, a quadratic dependence is expected as a result of a spiralgrowth mechanism dominating the crystallisation rate (Vernon et al.,2002), according to the classical Burton–Carrera–Frank (BCF) model(Burton et al., 1951; Kumar, 2009). In this model, growth occurs at dis-continuities (steps) on the crystal surface, and the rate of growth de-pends on the concentration of such active sites. A consequence of thisis that the growth rate is sensitive to any blocking of active sites, for ex-ample by adsorption of organic impurities. Because of this, controlledintroduction of impurities can be used as a probe to investigate crystalgrowth, a method which has been used to good effect in studyingthe crystallisation of aluminium hydroxide (as gibbsite) from alka-line solutions.

At higher supersaturation levels, precipitation by a birth-and-spreadmechanism becomes dominant (Brown, 1972a, 1972b; Lee andParkinson, 1999; Li et al., 2003). This process accelerates the rateof crystallization by secondary nucleation at the gibbsite surface andthe consequent production of new growth sites. At still higher supersat-uration, macroscopic seed particles may also grow by agglomeration(Ilievski and White, 1995; Stephenson and Kapraun, 1997). All ofthese processes are present in modern Bayer precipitation circuits(Tschamper, 1987). The precipitation rate varies as the square of thealuminate concentration irrespective of whether the macroscopic re-gime is nucleation, agglomeration or growth (albeit with widely differ-ing rate constants), which implies the existence of a common ratedetermining step at the molecular level.

A number of differentmodels have been suggested to explain the ki-netics and mechanism of crystallization of gibbsite. There seems to begeneral agreement that growth occurs primarily at kink sites on the sur-face, but there is less unanimity in relation to the chemical processeswhich occur there. In particular, consensus has not been reached onthe mechanism of the transition from tetrahedrally-coordinated alu-minium in the aluminate ions in solution to octahedrally-coordinatedaluminium in the solid gibbsite. The simplest model which seems tobe consistent with the data available to date is the electrostatic modelof Vernon et al. (2002), inwhich it is proposed that the gibbsite/solutioninterface behaves as a pair of electrical double-layers (Bard and Faulkner,1980) in which the negatively-charged gibbsite surface is connected tothe (negative) aluminate anions by a layer of (positive) sodium ions, asshown schematically in Fig. 5. In thismodel, growthmay take place at sur-face discontinuities such as kink sites, by the formation of a new bond be-tween a surface O atom and the Al atom of an aluminate ion in solution,with the simultaneous exit of a sodium ion from the electrical doublelayer into the solution; there is no requirement for the existence of a hy-pothetical “growth unit” at trace concentrations, the reaction being be-tween the dominant aluminate ion in solution and the gibbsite surface.While this model may not fully describe all the actual and possible pro-cesses at the interface, it nevertheless provides a useful rationalisationof many of the phenomena that have been observed.

Gibbsite grown from sodium hydroxide solutions may display avariety of shapes and sizes. The resultant shape and size distributionsare relatively insensitive to crystallization conditions and to traces ofinorganic impurities, and are therefore a result of properties inherentin the crystallization process itself. It has been shown that the effectswhich dominate in the production of this diversity include crystaltwinning at the nucleation stage, and accumulation of surface defectsand the continuous formation of nuclei during growth (Sweegers etal., 2001). This results in crystals which evolve from thin, roundedhexagons and faceted lozenges into faceted plates and blocks withwell-formed basal, prismatic and chamfered faces. Individual crystalsaremost commonly in the form of hexagonal prisms, which are generallyclustered together by either twinning and intergrowth, or agglomeration.The result is illustrated in Fig. 6, which shows a particle of gibbsite prod-uct from an alumina refinery. These results are consistent with the the-oreticalmodel developed by Fleming et al. (2000), who showed that thedominance of the basal (001) planemay be explained by the lower sur-face energy afforded by its 6-fold coordination of Al in comparison to5-fold coordination in the other faces. The growth rate dispersion be-tween crystal facets is a key determining factor in the final shape ofeach crystal, but for gibbsite precipitation the very mechanism ofcrystallisation may also be different for different facets: Lee andParkinson (1999) showed that, with supersaturation above 0.81, birthand spread becomes the dominantmechanismon the basal facewhereasspiral growth remains dominant on the prismatic faces.

Also of importance in the understanding of the modes of action of or-ganic impurities is the existence of an induction period in the crystalliza-tion of gibbsite. This is observable in batch precipitation tests as an initialperiod after the addition of seed to a supersaturated liquor, during whichthe precipitation rate, as observed by the rate of change of aluminate

Al

OH

HO

OH

OH

-

-

Al

OH

HO

OH

OH

Al

OH

HO

OH

OH

Al

OH

HO

OH

OH

Al

OH

HO

OH

OH

Al

OH

HO

OH

OH

Al

OH

HO

OH

OH

Al

OH

HO

OH

OH

Al

OH

HO

OH

OH

AlAl AlAl

--

-

Na+

Al

OH

HO

OH

OH

-

Al

OH

HO

OH

OH

-Al

OH

HO

OH

OH

-

Al

OH

HO

OH

OH

-

Al

OH

HO

OH

OH

Al

OH

HO

OH

OH

Na+

Na+

Na+

AlAl

Na+

-

Na+

-

Aluminium hydroxide (gibbsite) solid:Octahedral co-ordination

Aluminate solution:Tetrahedral co-ordination

Fig. 5. Schematic diagram illustrating the concept, proposed by Vernon et al. (2002), of a layer of aluminate ions held electrostatically to the gibbsite surface by an intervening layerof sodium ions. Growth occurs at a surface kink after escape of a sodium ion by diffusion through the charged layer, which allows an aluminate ion to approach and become incor-porated at the kink.


concentration in the solution, is very low but nevertheless finite (SmithandWoods, 1993). The end of the induction period ismarkedby a suddenchange in the precipitation ratewhich subsequently follows the expectedquadratic law (Vernon et al., 2002). The induction period is characterisedby the formation of secondary nuclei on the surfaces of the seed, de-scribed by Brown (1972b) as follows: “the (001) seed crystal faceserupt into a sea of potential secondary nuclei simultaneously with thecrystal growth process via a surface nucleationmechanism”. The forma-tion of secondary nuclei appears to result in a surface which is more ac-tive to crystallization (Smith et al., 1995), presumably due to theformation of kink sites which can initiate spiral growth which even-tually (i.e. at the end of the induction period) becomes the dominantcrystallizationmechanism. Induction periods are strongly influenced bythe presence of calcium ions (Brown, 1988; Gnyra and Brown, 1975;Raty et al., 2004), which are in turn influenced by the organics in solu-tion (see Section 6 below).

Fig. 6. A gibbsite particle typical of refinery precipitation product, showing intergrowthof hexagonal prisms (Reyhani et al., 2000).

In Bayer plant liquor, growth on the prismatic faces has been observedto be almost completely inhibited, while growth occurs on the basal faceby a birth and spreadmechanism. This has been interpreted as indicatingstrong adsorption of organic anions to the prismatic faces, but not to thebasal face (Freij and Parkinson, 2005).

4.3. Effects of specific types of organic compounds

The effects that organic anions have on precipitation depend on theirchemical structures and on the conditions of precipitation. For example,compounds with two or more adjacent hydroxyl groups tend to inhibitprecipitation by surface adsorption, whereas mono-hydroxy compoundstend to adsorb without affecting the precipitation rate. Some compoundsaffect both precipitation rates and product morphology, while othershave little or no effect on either. Various compounds and compoundtypes are discussed in the following sections, starting with gluconate,the best known andmost studied of precipitation inhibitors. Themajor-ity of the compounds considered are not in fact found in detectablequantities in Bayer process liquors, even though many are expected tobe formed during bauxite digestion. This is at least in part because ofthe very properties that are of interest, e.g. chemical reactivity, forma-tion of complexes with aluminate ions, and adsorption to gibbsitesurfaces.

4.3.1. GluconateCertain organic acid anions are known to be capable of stabilising

aluminate solutions to the extent of preventing precipitation of gibbsite,even when the solution is neutralised to pH 7. Tartrate and gluconateare particularly effective in this respect, and so formed the basis of theoriginalmethods for determining the alumina contents of Bayer processliquors (Watts and Utley, 1953, 1956). In these methods, the organiccompound was added in large excess (>10× on a molar basis, corre-sponding to 40 g/L in the case of sodium gluconate). The stabilisingeffect was attributed to the formation of complex ions in which the or-ganic anion replaces one of the hydroxyl groups in the aluminate ion.

At lower concentrations, gluconate has been shown to have an effecton precipitation which is much greater than can be explained by

Table 1Effects of gluconate on aspects of crystallization as a function of gluconate concentra-tion (from information given by Watling et al. (2000)).

[gluconate] (mM) 1 2 3 4 5 6

1° nucleation Prevented2° nucleation Promoted SuppressedInduction period IncreasedAgglomeration DelayedGrowth rate ×10−3 ×10−4 ~0


complexation of the aluminate in solution. Rossiter et al. (1996) dem-onstrated that the presence of 1 g/L sodium gluconate in a sodium alu-minate solution containing 110 g/kgAl2O3 resulted in a 50-fold increasein the induction period and an 85% drop in the crystallization growthrate, compared with an expected 0.6% drop in growth rate if the effectwere due solely to complexation of the aluminate ion in solution. Elec-tron microscopy of the crystals showed that significant surface roughen-ing occurred during the induction period, corresponding to the formationof newgrowth sites by secondarynucleation. Itwas shown that the rate ofsurface roughening was much less in the presence of gluconate, consis-tent with the observation of an extended induction period and a lowergrowth rate. It was concluded that the low growth rate in the presenceof gluconate was due to a reduction in the surface area available for crys-tallization, but not to a change in the activity of the surface, i.e. the mech-anism of inhibition was by simple blocking rather than a fundamentalchange in the chemistry of the gibbsite seed itself. Coyne et al. (1989,1994) showed that a concentration of gluconate sufficient to cover only3.5% of the total seed surface could be enough to reduce the crystallizationrate to almost zero, indicating that gluconate adsorbs preferentially to ac-tive growth sites.

These conclusionswere corroborated by Beckhamet al. (2005), whodemonstrated that gluconate interacts with the gibbsite surface by aphysical adsorption mechanism that is accurately described by an ad-sorption isotherm, and that chemisorption is not involved. By inferringthe adsorption isotherms from de-supersaturation curves, these au-thors were able to measure the degree of adsorption to active sites di-rectly, and to show that it has strong selectivity for kinks created afterthe birth of growth sites. Using a modified Langmuir isotherm, theydemonstrated that adsorbed gluconate effectively blocks about 50%more of the growth surface than it actually covers.

By studying gibbsite crystallization under different conditions ofsupersaturation and seed density, Watling et al. (2000) were able toseparately observe the effects of gluconate under conditions thatfavoured each of the threemacroscopically different regimes (nucleation,

0.40

0.50

0.60

0 2

Tim

A/T

C R

atio

Bayer liquor + gluconate

Bayer liquor

Synthetic liquor + gluconate

Synthetic liquor

Fig. 7. Desupersaturation curves for synthetic and plant Bayer liquor, showing the effects odashed lines). Redrawn from Coyne et al. (1994).

agglomeration and growth) in turn. The effects were found to be a func-tion of not only crystallization conditions, but also of gluconate concentra-tion in the range 1 to 6 mM (0.22 to 1.31 g/L sodium gluconate). Theyfound that in all cases, the precipitation yield after an extended period(200 h) from the solutions containing gluconate was within 10% of theyield from a pure solution. This is consistent with the observations ofBeckham et al. (2005) who interpreted it as evidence that the gluconateacts by a physical adsorption mechanism. At shorter times however, theeffects on crystallization were found to be varied and significant. The ef-fects of gluconate as a function of concentration for each aspect of crystal-lization are summarised in Table 1.

The promoting effect of gluconate on secondary nucleation at lowconcentrations may be due to partial blockage of growth sites reduc-ing the adherence of nuclei to the surface such that nuclei can becomedetached, leaving the site available for the formation of new nuclei. Athigher gluconate concentrations this effect is overshadowed by a morecomprehensive blocking of growth sites.

The significance of even a very small amount of a precipitation in-hibitor in the context of a Bayer plant liquor is demonstrated in Fig. 7.An addition of 0.07 g/L carbon (1 mM) in the form of gluconate to asynthetic Bayer liquor (initially containing no carbon) is similar to theeffect of all the impurities in a Bayer plant liquor containing about30 g/L carbon as well as carbonate and other inorganic impurities.This indicates that the majority of the carbon in the plant liquor is notassociated with inhibitors (Coyne et al., 1994), a conclusion which hassince been corroborated by detailed chemical analysis of Bayer plant li-quors. This is to be expected, because the action of inhibitors generallyrequires that they be adsorbed to the precipitating gibbsite, whichmeans that they are continuously removed and so do not accumulatein the solution. Consistent with this, gluconate has never been detectedin Bayer plant liquor to our knowledge, and only minor amounts of afew other polyhydroxy compounds have been found in some liquors,even though they are almost certainly formed during bauxite digestion(Power et al., 2011a).

4.3.2. Aliphatic poly-hydroxy compounds—the alditolsAlditols are straight-chain hydroxy-alkanes with a single hydroxyl

group on every carbon atom. They have the general formula:

CH2OH\½C⁎HOH�n\CH2OH

In this formula C⁎ denotes a chiral centre, i.e. this carbon atom hastwo possible stereo-isomers. The first nine in the series of alditols are

4 6

e (hours)

f 1 mM sodium gluconate (added at the end of the induction periods indicated by the

image of Fig.�7

Table 2The first nine alditols, with stereo-isomers bracketed. The chiral centres are labelled as cir-cles (data from Smith et al. (1996)).


shown in Table 2, in which the chiral centres are highlighted as circlesand the first two groups of stereo-isomers are bracketed.

The first mention of hydroxyl groups in relation to the inhibition ofprecipitation rate appears to have been by Pearson (1955), and The(1980) discovered that gluciosaccharinate (which containsmultiple hy-droxyl groups) is a powerful crystallization inhibitor. Consideration ofaluminium co-ordination chemistry led Grocott and Rosenberg (1988)topropose that organic compoundswith adjacenthydroxyl groups shouldhave a strong affinity with the gibbsite surface, and correctly predictedthat this would increase the incorporation of sodium into the precipitate.Confirmation of the adsorption of such compounds onto gibbsite surfaces,and the effect this has on precipitation rates, was first made by Coyne(1989) and Coyne et al. (1989, 1994) who established equilibrium ad-sorption isotherms for a number of them. The only compounds thatwere found to exhibit significant adsorption were those with at leastone pair of adjacent hydroxyl groups. Compounds which in additionhave one or more carboxylic acid groups were found to adsorb morestrongly than those without, and adsorption increased with the lengthof the carbon backbone. Hence, the order of adsorption strengths ofthe compunds studied is mannitol (6 carbons)>adonitol (5 carbons),and gluconate (6 carbons)>tartrate (4 carbons). For compounds withmultiple hydroxyl groups but no carboxylic acid groups, no adsorptionwas detected for compounds with less than five carbon atoms.

The degree of yield inhibition was found to depend linearly on thedegree of surface coverage irrespective of the identity of the adsorbent,as shown in Fig. 8. The results for mannitol correspond to solution con-centrations of 0.1, 0.5, 1.0 and 2.0 mM, for which the surface coverage

0

50

100

10

Surface C

Yie

ld In

hib

itio

n (

%)

y = 27.3x + 3.7R2 = 0.96

Fig. 8. Relationship between surface coverage and yield inhibition for two alditols (mannitoextracted from Coyne et al. (1989, 1994)).

was confirmed to be a linear function of solution concentration. The so-lution concentration of the other three compounds was 1 mM. These re-sults are consistent with a simple physical adsorption mechanism inwhich inhibition results fromblockage of active growth sites, as discussedabove for the case of gluconate and as indicated by the partial activation ofgibbsite seed by treatment with boiling water (Zeng et al., 2007).

Detailed studies have beenmade of the effects of alditols on gibbsitecrystallization, not because they are likely to be present in significantquantities in Bayer process liquors, but because they have proven usefulin probing themechanism of adsorption and hence of inhibition (Smithet al., 1996; Watling et al., 1996).

Smith et al. (1996) found that alditols generally inhibit gibbsitecrystallization, and that the degree of inhibition generally increaseswith carbon chain length. Diols are generally less effective inhibitors,and it was found that compounds with three adjacent hydroxyl groupstended to be the strongest inhibitors. It was also found that stereochem-istry is very important; specifically, compounds inwhich the position ofthe hydroxyl groups alternate from one side to the other of the carbonchain (the “threo-threo sequence” (Angyal et al., 1974)) aremuch stron-ger inhibitors than those in which all the hydroxyl groups are all on thesame side. This is because the spatial arrangement of the three hydroxylson the same side of themolecule facilitates strong co-ordinationwith thealuminium ion. Hence, xylitol is a much stronger inhibitor than its ste-reoisomer ribitol (Table 2).

Degree of inhibition was found to correlate with the relative degreeto which these compounds are able to form complexes with metal ionsin aqueous solution, measured in calcium acetate buffer solution, as il-lustrated in Fig. 9 (Angyal et al., 1974). UsingNMR and Raman spectros-copy, Smith et al. (1996) confirmed that complexes also exist betweenthe alditols and aluminium in aqueous solution at pH 14. Furthermore,the degree of complexation was found to be qualitatively in the sameorder as previously reported for (acidic) acetate solutions for the limit-ed number of compounds studied.

These results suggest that the mode of action of the alditols ininhibiting gibbsite crystallization may not be by poisoning of growthsites by direct adsorption of the organic compound, on the basis thatthe compounds that form stronger complexes would have the lowerfree concentration in solution, would therefore be expected to beless available for adsorption, and so would be the weaker inhibitors.As discussed above, the mode action cannot be to reduce (by com-plexation) the amount of aluminate available for crystallization, be-cause the inhibitors act at far too low a concentration for that to bethe case. These considerations suggest that it may be the aluminium–

432

overage (%)

Mannitol

Gluconate

Adonitol

Tartrate

l and adonitol) and two hydroxyl-carboxylic acids (gluconate and tartrate) (from data

y = 0.13x + 6.84R2 = 0.96

0

10

20

0 25 50 75 100

% Inhibition

Co

mp

lexi

ng

Str

eng

th (

% v

s ci

s -i

no

sito

l)

Fig. 9. Degree of inhibition of gibbsite crystallization by alditols at a constant specific dose rate of 0.5 mM m−2 as a function of their ability to form complexes with metal ions inacidic solution. Data from Smith et al. (1996) and Angyal et al. (1974).


alditol complex, not the alditol itself, which attaches to growth sites andcauses them to be deactivated.

For a gibbsite ion to transform to solid gibbsite, two key steps are re-quired: first, the ion must adsorb to the surface, and second, the alumin-ium species must transition from tetrahedral to octahedral co-ordinationin order to become properly incorporated into the solid.

A possible explanation for the powerful inhibiting effect of alditolsmay therefore be related to stabilization of tetrahedrally co-ordinatedaluminium by complexation, as follows: alditols form complexes withaluminium by replacing two or three of the hydroxide groups on theoriginal aluminate ions. This leaves a least one hydroxyl group whichcan be eliminated to allow the aluminium to bond to a growth site onthe surface. Transition to octahedral co-ordination would then beinhibited by the presence of the chelating alditol resulting in thecomplex blocking the site, as illustrated schematically in Fig. 10. In thisway the growth site may become irreversibly de-activated. It has alsobeen found that inhibition is directly related to the acidity of the hydrox-yl groups, which is also consistent with the formation of stronger com-plexes (Grocott and Rosenberg, 1988).

This is consistent with observations on the importance of thestereochemistry of the inhibitor, in particular that compounds withthree adjacent hydroxyl groups alternating either side of the carbonbackbone are the most potent inhibitors. It is also similar in principleto the proposal by Seyssiecq et al. (1999) based on binding of alcoholgroups to a dimeric aluminate “growth unit” which was predicted bymolecular modelling to exist at low concentrations in aluminate solu-tions. The inter-atomic distances between pairs of oxygen atoms in thegrowth unit were found to be quite similar to the distances between

Al

OH

HO

O

OH

Al

OH

HO

OH

OH

Al

OH

HO

OH

OH

AlAl

Al

O

O O

O H O H

Fig. 10. Schematic representation of an aluminium–alditol complex attached to a growthsite. The co-ordination of the aluminium ion by the alditol forces the Al to remain tetrahe-drally co-ordinated, preventing it from becoming fully incorporated into the gibbsitephase and irreversibly blocking the growth site (interpreted from Smith et al. (1996)).

alternate OH groups in polyols (polyhydroxy carboxylic acids andalditols), leading the authors to propose the general structure shownschematically in Fig. 11 for a growth unit to which a polyol has been at-tached with the elimination of two hydroxyl groups. The unit wouldthen dock onto a gibbsite growth site via remaining hydroxyl groups,leaving the organic ligand facing outwards and effectively deactivatingthe growth site.

The alternative explanation of direct adsorption of the organic inhib-itor to the solid surface is favoured by other authors, by analogywith thecase of gluconate and on the basis that octahedral aluminium complexeswhich form under acidic conditionsmay be dissociated and form tetrahe-dral aluminate ions in highly alkaline solutions (Motekaitis and Martell,1984). Under this model, the relative strengths of inhibitors are seen asresulting from the relationship of the structure and stereochemistryof an inhibitor to its ability to adsorb to the gibbsite surface (vanBronswijk et al., 1999; Watling, 2000), in which the aluminium ionsare octahedrally co-ordinated. However from the available informationit does not appear to be possible to distinguish between adsorption ofthe organic compounds themselves and adsorption of an organic/aluminate complex.

Experiments on a series of alditols by van Bronswijk et al. (1999) re-vealed clear relationships between structure, adsorption and precipita-tion yield. In particular, the compounds studied were found to fall intodistinct two groups based on the key stereochemical characteristic ofthreo (alternating hydroxyls) or erythro (hydroxyls all on same side),as shown in Fig. 12, in which the behaviour of a group of threo com-pounds is compared with that of a set of erythro compounds. Each setof compounds showed adsorption increasing monotonically with thenumber of carbon atoms, with corresponding monotonic decreases inthe precipitation yield (shown here relative to the yield in an identicalexperiment carried out in the absence of organic additives (the “con-trol”), which has been assigned a value of 20 arbitrary units for the pur-poses of comparison). The effectswere significantly greater for the threogroup. A more extreme example (not shown) was between the struc-tural isomers xylitol (threo) and ribitol (erythro),which showed adsorp-tions of 22 units and 2.7×10−8 mol g−1, and relative precipitationyields of 2.2 (11% of the control) and 17 units (85% of the control).

The presence of organic inhibitors affects not only the precipitationyield, but also the shapes and numbers of the particles produced. Theexact effects on particle shape are complex and depend on both the na-ture of the inhibitor and the conditions of crystallization, in particularwhether the conditions favour nucleation, agglomeration or growth, al-though the fundamental (molecular) mechanism of inhibition appears

Fig. 11. Schematic representation of a gibbsite growth unit with a polyhydroxy organiccompound attached (adapted from Seyssiecq et al. (1999)). It is proposed that dockingof this unit to a site effectively blocks further growth at that site.


to be independent of the macroscopic crystallization regime (Watling,2000). Under conditions of moderate to high supersaturation, whichfavour secondary nucleation and agglomeration, crystals grown frompure solutions are agglomerates of blocky hexagonal prismswith cham-fered faces as illustrated in Fig. 6. The shapes of the individual crystalswithin the agglomerates are consistent with theoretical predictions(Seyssiecq et al., 1999; Sweegers et al., 2001, 2002). Crystals becomeelongated in pure sodium aluminate solutions because the rate of growthon the hexagonal (001) basal plane is mainly by a relatively fast birth-and-spread mechanism, whereas the side faces are characterised byslower spiral growth. Inhibition of the birth-and-spread mechanismby polyhydroxy compounds which block secondary nucleation sitestherefore results in the development of flat hexagonal plates (Watling,2000). It can also cause the formation of dendrites and of individualsmall particles, as well as reducing the overall size of the main growthparticles.

4.3.3. CarboxylatesIn neutral or acid solutions many carboxylates have strong inter-

actions with gibbsite surfaces, for example by chemisorption via a re-action analogous to esterification, which is key to their importance inareas such as catalysis and chromatography (Karaman et al., 2001).Low molecular weight dicarboxylates including oxalate, malonate andsuccinate (which are all generally present in Bayer process liquors)show significant adsorption, but only below pH 9 (Rosenqvist et al.,

0

10

20

1 2 3 4

Number of ca

ERYTHRO

THREO

Adsorption x 108 (mol g-1)

Relative precipitation yield (control = 20)

Fig. 12. Effect of chain length and stereochemistry on adsorption and precipitation yield (vmeso-erythritol, D-mannitol) and a threo series (threitol, binitol, D-sorbitol) of compounds

2003). In strongly alkaline solutions however, it seems unlikely thatsimple aliphatic carboxylates (i.e. saturated hydrocarbons with carboxylgroups as the only functional groups) have strong interactions with thegibbsite surface. Accordingly, the presence of simple aliphatic carboxyl-ates appears to have little effect on gibbsite precipitation apart from theaggregate effect on the ionic strength of the liquor discussed inSection 4.1.

Thirty-eight carboxylates are commonly present at measurable quan-tities in Bayer process liquors (Power and Loh, 2010). Twenty-one aresimple aliphatics, two have one hydroxyl group and one, tartrate, has apair of (adjacent) hydroxyl groups. Of these, tartrate is the only one thathas been confirmed to be a crystallization inhibitor (Coyne et al., 1994;Nikolic et al., 2008; Paulaime et al., 2003; Power, 1991). Lactate, acetateand succinate have been reported to have significant effects on yieldand product particle size distribution under conditions of unseeded pre-cipitation by carbonation of the liquor, apparently due to effects on theinduction period (Brown et al., 1986); no such effect has been reportedin the context of seeded precipitation however.

The other fourteen compounds are benzene carboxylates, includingfour with a single (phenolic) hydroxide. The simplest of these, benzoate,is commonly found in Bayer process liquors (Power and Loh, 2010). Theeffects of benzoate and the three methyl benzoate isomers (o-, m- andp- methyl benzoate (or toluenidate)) have been investigated by Lv et al.(2009), who found that they all reduce precipitation yield and also affecttheparticle size distributionof theproduct. The observedorder of yield ef-fectwas: p-methyl benzoate>benzoate>m-methyl benzoate>o-methylbenzoate. This order was found to correlate with the order of net chargeon the carboxylic acid group, on which basis the effect was rationalisedin terms of an electrostatic model of the solution/gibbsite interface(see Fig. 5). The presence of p-methyl benzoate has been reported inBayer liquor at least once (Picard et al., 2002), but to date there do notappear to be any reports of the other isomers.

A number of aliphatic carboxylates with two or more adjacent hy-droxyl groups, which are therefore either known or likely to be inhibi-tors (Bouchard et al., 2005; Loh et al., 2010b; Nikolic et al., 2008;Paulaime et al., 2003; Power, 1991), have been found at low concentra-tions in at least one Bayer process liquor (Power et al., 2011a). System-atic studies of this class of compounds have shown that they generallyinhibit precipitation, and that the trends in inhibitor strength tend tocorrelate with the trends for the corresponding alditols (Watling, 2000).In order to assess the effect of the presence of the carboxylate group, itis necessary to compare pairs of otherwise identical compounds. Suchcomparisons indicate that the acid group generally enhances the

5 6 7

rbon atoms

ersus a control of 20 in arbitrary units) for an erythro series (2,3-butane diol, glycerol,(adapted from data of van Bronswijk et al. (1999)).


inhibition effect (Watling, 2000). The differences are sometimes dra-matic, as between threitol which is a mild inhibitor, and threonicacid, which is one of the strongest. In contrast, Paulaime et al. (2003) ob-served that the alditols weremore potent inhibitors than the acids basedon a comparison of group of acidswhich had amaximumof four carbonswith a group of alditols which had aminimum of five carbons. However,in this study, the effects of carbon number and stereochemistry appear tohave been greater than the effects of the addition of carboxylate groups.

A study of the adsorption behaviour and yield impacts of a group ofbenzene carboxylates and hydroxy-acids, catechol, and the hydrogena-tion and oxidation products of catechol (1,2-cyclohexane diol andmuconate, respectively) in comparison with gluconate has illustrateda number of interesting features of these compounds (Bouchard et al.,2005). All of thosewhich contain either carboxylate or hydroxyl groupswere found to adsorb to the gibbsite surface and to be precipitation in-hibitors. According to the Henry constants, the adsorption strengths ofphthalate, isophthalate and 2-hydroxybenzoate are almost as great asfor gluconate, however all are far less effective inhibitors than gluco-nate. Catechol (1,2-dihydroxybenzene), which has two adjacent hy-droxyl groups, also shows an adsorption strength similar to gluconate,but is a considerablyweaker inhibitor. The related compounds resorcinol(1,3-dihydroxybenzene) and hydroquinone (1,4-dihydroxybenzene), inwhich the two hydroxyl groups are not adjacent, are not yield inhibitors(Power, 1991), consistent with an expected lower adsorption due to thedifferent stereochemistry of the hydroxyl groups. The hydrogenationproduct of catechol, 1,2-cyclohexane diol, retains the two adjacent hy-droxyl groups but does not adsorb and is not an inhibitor, presumablybecause of the significant change in stereochemistry (from planar tothree-dimensional boat and chair conformations). Similarly, oxidationof catechol to muconic acid by opening the benzene ring and oxidiz-ing the hydroxyl groups to carboxylates renders the molecule non-adsorbing and non-inhibiting.

Most of the adsorption measurements reported in the literature arereported as adsorption isothermsmeasured by contacting test solutionswith gibbsite solids, either in a suspension or a column, at room temper-ature and measuring the uptake of individual organic compounds atequilibrium (Bouchard et al., 2005; Coyne et al., 1994). The Henry con-stants determined in this way are then compared with the yield from aprecipitation test carried out at elevated temperature with a differentseed. These experiments have generally demonstrated the importanceof adsorption in relation to yield inhibition; in particular, adsorption ap-pears to be a pre-requisite to yield inhibition for any given compound.However the converse is not true (i.e. not all compounds that adsorbare inhibitors), and it has not been possible to compare results quanti-tatively from one study to another. It has also been suggested that ad-sorption under precipitation conditions may not always follow thesame behaviour as at room temperature (Loh et al., 2010a).

Accordingly, Loh et al. (2010a) undertook a study of adsorption inwhich the degree of adsorption was measured on the same seed andunder conditions as close as possible to the conditions under whichyield inhibition was determined. To do this, adsorption was measuredby determining the uptake of each organic compound by a gibbsiteseed in slurry with sodium aluminate at its equilibrium concentrationat the desired temperature. Precipitation tests were then carried outwith a supersaturated sodium aluminate solution under the same condi-tions of seeding and temperature. This work showed that, not only isthere no single correlation between adsorption strength and the degreeof inhibition, but that adsorption can lead to either inhibition or enhance-ment of precipitation yield, or to no change at all, depending on the typeof compound involved. For compounds containing hydroxyl groups, acomparative study of a small number of aromatic and aliphatic com-pounds indicateddirect correlations betweenadsorption strength andde-gree of inhibition. The degree of inhibition per amount adsorbed wasgreater for aliphatic compounds than for aromatics, and in both casesthe effects increased with the number of hydroxyl and/or carboxylategroups present, as summarised in Fig. 13. In most cases the trends are

qualitatively similar to those determined previously from room temper-ature adsorptionmeasurements, but the results from the different stud-ies cannot be compared quantitatively because of the different seedmaterials used and the lack of comparative information on their specificsurface areas, for example. Even so, the results for salicylate (2-hydroxybenzoate) are clearly different. Bouchard et al. (2005) found that theHenry constant for salicylate was similar to that of gluconate, indicatingvery strong adsorption potential consistent with the observation byPicard et al. (2002) that salicylate appears to have quite a strong affinityfor gibbsite under Bayer plant conditions. By contrast, Loh et al. (2010a)found that salicylic acid did not adsorb at all under precipitation condi-tions and correspondingly did not inhibit precipitation (Fig. 13).

The inhibition effects of three benzene carboxylates were also inves-tigated, and all were found to increase yield. This effect was ascribed toincreases in the total surface area of the solids due to enhanced second-ary nucleation. No adsorptionmeasurementsweremade for these com-pounds in this study, but the results were correlated with the “surfaceaffinity” as defined by Picard et al. (2002), as shown in Fig. 14.

These results indicate that adsorption via paired hydroxyl groupsoccurs by a different mechanism than adsorption via carboxylic acidgroups. As discussed above, the former appears to be associatedwith di-rect interactions with aluminium ion, either in growth units or directlyat the surface itself, resulting in permanent deactivation of growth sites.On the other hand, the presence of carboxylic acid groups alone appearsto result in the formation of an increased number of secondary nucleithat do not adhere to the growth surface, causing an increase in thenumber of fine particles and hence a greater overall surface area forprecipitation.

Further insight into the effects of hydroxy benzoates may be gainedfrom Fig. 15, which has been constructed from data given by Armstrong(1993). This figure summarises data from a set of precipitation experi-ments carried out in two stages, the first under conditions that favour ag-glomeration, and the second favouring growth. Yield inhibition in theagglomeration stage is shown as a function of organic compoundadded; the results have been adjusted to represent the effects of equimo-lar additions of each compound. Comparing the degrees of inhibitioncaused by 1,2-dihydroxy benzene and 3,4-dihydroxy benzoate indicatesthat the addition of a carboxylate group not adjacent to the hydroxylsmore than doubles the inhibition, whereas the adjacent carboxylate in2,3-dihydroxy benzoate increases the inhibition by about 40%. For thehydroxy benzoates with three adjacent hydroxyl groups the inhibitionis independent of the positioning of the carboxylate group, and is almostas high as for 3,4-dihydroxy benzoate. These observations support thehypothesis that pairs of adjacent hydroxyls (shown circled) are themain cause of the inhibition effect, and that this effect is magnified bythe presence of carboxyl groups.

Further work on 3,4-dihydroxy benzoate demonstrated that it ad-versely affects all aspects of the precipitation process, by increasingthe induction time and decreasing the rates of nucleation, agglomer-ation and growth (Tran et al., 1996).

4.3.4. Oils, fatty acids and miscellaneous compoundsOils enter Bayer process liquor by a variety of routes, including as

partial degradation products of the organic matter in bauxite, as car-riers for process chemical additives and from leakage of lubricationoils from process machinery, for example (Power and Loh, 2010). Fattyacids are present as degradation products of more complex compoundsand from the saponification of oils. Any possible direct effects of thesecompounds on precipitation yield do not appear to have been investigat-ed in detail and have generally been assumed to be negligible, althoughthere are reports that certain fatty acids may be beneficial in ameliorat-ing the effects of inhibitors to some extent (Zeng et al., 2008).

The addition of a crown ether (15-crown-5) to a sodium aluminatesolution has been shown to destabilise it towards nucleation and to in-crease the rate of seeded precipitation (Zeng et al., 2007). This crownether is an effective complexant for sodium ions (An et al., 1992), so

0%

50%

100%

0.00 0.05 0.10

Adsorption (µmol/g)

Rel

ativ

e Y

ield

(%

)

Fig. 13. Precipitation yield relative to control as a function of adsorption of individual organic compounds, in two series: 1. aromatics: 2-hydroxy benzoate (salicylate),1,2-dihydroxy benzene (catechol) and 3,4-hydroxy benzoate (3,4-DHBA), and 2. aliphatics: tartrate and gluconate (adapted from Loh et al. (2010a)).


the effect appears to be related to a reduction in the availability of sodi-um ionswhich facilitate the formation of gibbsite solids (Li et al., 2000a,2000b).

5. Size control in precipitation: crystal growth modifiers (CGMs)

The control of theparticle size distributions of theprecipitator solids inalumina refineries is crucial for proper operation of the continuous pre-cipitation process tomaximise yield and achieve acceptable product qual-ity. Because of the complex nature of the precipitation process, theengineering solutions involved in its optimisation are complex and varied,and are outside the scope of this review.However the influence of organiccompounds on sizing and size control can be significant, primarily be-cause certain compounds can inhibit or enhance agglomeration and/ornucleation. Someof these effects have alreadybeenmentioned in the con-text of yield impacts, for example the increases infines production causedby the presence of benzene poly-carboxylates,which in turn increases thesurface area available for precipitation and so increases the precipitationyield (see Carboxylates section). This has a disadvantage, however, be-cause the generation of excessivefinematerial can lead to a loss of control

100%

150%

200%

0.0 0.5

Surfa

Rel

ativ

e yi

eld

(%

)

Fig. 14. Precipitation yield relative to control as a function of the “surface affinity” of compouyield enhancement effects (adapted from Loh et al. (2010a)).

of the solids due to elutriation, carryover and size classification issues. Thecountermeasures to this include reducing seed loadings and supersatura-tion, which limit the precipitation yield. Considerable research effort hastherefore been focussed on the development of process additives to facil-itate particle size control in precipitation while maximising yield param-eters. This has led to a range of formulations generally known as “crystalgrowth modifiers” (CGMs) (Roe et al., 1988).

The first CGM was an agglomeration aid based on an acrylate/methacrylate copolymer (Roe and Perisho, 1986). This was supersededby a more economical formulation consisting of an ionic surfactant(a fatty acid) dissolved in an oil to produce a cheaper product whichachieved similar results (Owen and Davis, 1988). Improving agglomera-tion has the effect of coarsening the particle size distribution and there-fore reducing the available area for precipitation, which of itself wouldhave the effect of reducing yield. However, it is claimed that the im-proved process control flexibility afforded by the ability to manipulateagglomeration can lead to increases in precipitation yield of the orderof 10% with CGM addition rates of less than 25 ppm (Roe et al., 1988).

Many different formulations have since been examined to improvethe technical and economic performance of CGMs. A mixture of

1.0

ce Affinity

nds to gibbsite as defined by Picard et al. (2002) for benzene polycarboxylates, showing

0

0.2

0.4

0.6

0.8

1

none(control)

Organic compound added

Fra

ctio

n o

f Id

eal Y

ield

(%

)

1,2-DHB tartrate 2,3-DHBA 2,3,4-THBA 3,4,5-THBA 3,4-DHBA

Fig. 15. Effect of hydroxy-benzene compounds on agglomeration and growth stages of precipitation in comparison to the effect of tartrate, shown in order of degree of inhibition ofagglomeration yield. DHBA = dihydroxy-benzoate, THB = trihydroxybenzoate, 1,2-DHB = dihydroxybenzene (catechol). The hydroxyl pairs which are likely to promote adsorp-tion are circled. (Adapted from data in Armstrong (1993)).


polyalkoxylated and non-polyalkoxylated surfactants has been shownto be effective and to allow the amount of carrier oil to be minimised(Welton andMcColl, 1999). Even polysaccharides have been suggested,because they have been found to act as agglomeration aids at low con-centrations (Farquharson, 1996). The benefit in improved agglomera-tion is claimed to be an advantage provided the concentration is notallowed to become too high, in which case the overall precipitationyield would be inhibited. Given the complexity of Bayer liquors gener-ally and the need for conservatism in operation, it seems unlikely thatCGM formulations containing polysaccharides would be popular withplant operators. This example highlights the importance to plant oper-ators of knowing the ingredients of chemical additives, about whichsuppliers are naturally circumspect for commercial reasons, generallypreferring not to identify the nature or relative amounts of the ingredi-ents in CGMs. Suppliers therefore tend to refer to the additives simply as“CGMs” with arbitrary product numbers, and to rely on the results ofempirical tests to indicate performance (Counter, 2006; Kouznetsov etal., 2008; Liu et al., 2007).

Of the many CGM formulations now available, the current preferenceappears to be towards any of a number of fatty acid surfactants dissolvedin a suitable oil, in emulsion with water (Kouznetsov et al., 2011). Suchmixtures allow flexibility of formulation to create cost-effective productstailored to specific situations. In addition, there is increasing interest inunderstanding the effects of chemical additives inmodified Bayer processplants treating refractory ores such as those found in China (Lu et al.,2009; Yin et al., 2006, 2009a, 2009b; Yu et al., 2007; Zhang et al., 2007).

6. Calcium in liquor

Lime is the second most important additive in Bayer process tech-nology after sodium hydroxide, and it has major benefits to the Bayerprocess (Roach, 2000; Rosenberg et al., 2001; Whittington, 1996), butthe consequent presence of calcium ions in the green liquor is unde-sirable because calcium co-precipitates with aluminate ions. This re-sults in increased induction periods in Bayer precipitation (Cornellet al., 1996), and in contamination of the alumina product. The con-centration of calcium in Bayer process green liquors must thereforebe controlled to low levels.

The calcium concentration in Bayer liquors is increased by the pres-ence of organic anions. For a given Bayer liquor, the increase in green li-quor calcium concentration is directly proportional to the total organiccarbon content of the liquor. Not all of the organic compounds

contribute to this effect; rather, it is due to the presence of a small subsetof the organic compounds present. Aswith the inhibition of precipitationyield, the compounds that have the greatest effect on calcium concentra-tions are those with multiple hydroxide groups. Consequently, green li-quor calcium concentrations tend to be lower in liquors from plantswith high temperature digestion than from those with low temperaturedigestion (The and Sivakumar, 1985). Tests with sodium gluconate haveindicated that the effect is likely to be due to the stabilization (by chela-tion) of a calcium aluminate hemihydrate species (Rosenberg et al.,2001).

7. Sodium oxalate

Sodium oxalate occupies a special place in Bayer process technol-ogy because of problems that its limited solubility can create in theoperation of alumina plants (Yamada et al., 1973). The many and vari-ous problems caused by oxalate are not always fully understood (Kimand Lee, 2003), but as we will see in this section, much progress hasbeen made.

Although moderately soluble in water (approximately 35 g/L at25 °C and 65 g/L at 100 °C (CRC, 2011; Dean, 1998; Stephen andStephen, 1963)), sodium oxalate is less soluble in Bayer liquors by afactor of 20 or more, primarily due to the common ion effect of the so-dium. A number of empirical correlations have been derived for thedependence of oxalate solubility on the concentrations of other ionspresent in Bayer liquors (Bouzat and Philopponneau, 1991; Brownand Cole, 1980; Yamada et al., 1973). More recently, thermodynamicmodels have been developed which take into account all the mainions in solution, and which can be calibrated to give good correlationsfor the equilibrium oxalate solubility in actual plant liquors (Beckhamet al., 2005; Königsberger et al., 2005b).

In actual plant liquors however, it is generally found that the mea-sured sodium oxalate solubility is significantly higher than predictedfrom the thermodynamic models. This apparent solubility (The andBush, 1987) is a function a number of factors, in particular the natureand concentrations of other (highmolecularweight) organic compoundsin the liquor. It therefore cannot be predicted thermodynamically andmust be determined empirically for each liquor and circumstance(Beckham et al., 2005). Apparent solubility does not follow the sametrends with concentration as the thermodynamic solubility; for exam-ple apparent solubilitymay increasewith TA, as illustrated for particularplant liquor in Fig. 16.

0.0

1.0

2.0

3.0

220 240 260 280

TA (g/L)

[NaO

x] (

g/L

)

Apparent solubility

Thermodynamic solubility

Fig. 16. Behaviour of apparent solubility in comparisonwith thermodynamic solubility, for a liquor from a plant treating Australian bauxite (redrawn from Beckham and Grocott (1993)).


7.1. Oxalate input

The oxalate present in Bayer process liquors is a result of degradationof organic compounds extracted from the bauxite. Estimates of the pro-portion of the organic content of bauxite which reports as oxalate in sin-gle laboratory digestions vary from about 3 to 10 percent (Brown andCole, 1980; Grocott and Rosenberg, 1988; Lever, 1983). The total oxalateproduction in a plant will generally be somewhat greater than this be-cause oxalate is also produced by the progressive degradation of organiccompounds as the plant liquor cycles through digestion, which mayaccount for around 20% of the ultimate oxalate input to the liquor(Armstrong and Healy, 2006; Power and Tichbon, 1990). Oxalateinput varies widely with bauxite source and digestion temperature,as can be appreciated from Fig. 17 in which the oxalate productionrates of a range of bauxite types is compared. The measurements weremade on samples from bulk standards prepared as broadly representa-tive of the bauxite deposits concerned. The three USSR bauxites are di-asporic, the two Jamaica and Wagina Island samples are from mixed

0

1

2

Gove

Weip

a

Wag

ina I

s

Darlin

g Ra

Jam

aica 2

Surinam

e

Ext

ract

able

Org

anic

Car

bo

n(g

C /

t b

aux)

Fig. 17. Oxalate generation rate as a function of bauxite source and d

gibbsitic/boehmitic karst bauxites, and the remainder represent lateriticbauxites. In all but one case (Darling Range) the oxalate production rateat the higher temperature is equal to or higher than at the lower tem-perature, markedly so for the Jamaica 1 and Wagina Island samples.Lever's comment that oxalate production is roughly double at the highertemperature presumably applied to Jamaican bauxites, but he also ob-served that for Australian bauxites the ratio may be up to a factor ofthree (Lever, 1983)). These discrepancies highlight the limitations ofsuch generalisations and the need for each bauxite to be assessedindividually.

7.2. Direct effects: sodium oxalate crystallization

It has been shown that the oxalate anion reduces the rate of growth ofgibbsite by adsorbing to growth sites (Kingsley andWilcox, 1986; Nikolicet al., 2004), but its effect is minor in comparison to that of other morestrongly adsorbed anions (Power, 1991). The dominant effects of oxalate

Arkan

sas

Kimber

ly

Jam

aica 1

USSR 2

USSR 3

USSR 1

143oC digest

230oC digest

igestion temperature (from data in Power and Tichbon (1990)).

1 Aluminium tri-hydroxide, or gibbsite, is commonly referred to in the industry as“hydrate” (an abbreviation of “alumina tri-hydrate”).


are caused by the crystallization of sodium oxalate, and the interactionsthat occur between sodium oxalate and gibbsite solids.

Uncontrolled oxalate crystallization can be highly detrimental to theoperation and economy of Bayer process plants (Power and Tichbon,1990). For most plants processing lateritic bauxites, the rate of inputof oxalate to the liquor is greater than the natural outputs via solublelosses (including to the bauxite residue) and entrainment of solid sodi-umoxalatewith gibbsite product or scale (The and Bush, 1987). The im-provements in bauxite residue washing efficiencies that have beenachieved in modern plants has had the downside effect of increasingthe rate of oxalate accumulation in the liquor. Some form of oxalate re-moval process is therefore generally operated in order tomaintain oxa-late concentrationswithin acceptable levels (Chinloy, 1990; NewchurchandMoretto, 1990; Yamada et al., 1973). Reported studies of oxalate be-haviour in the absence of an oxalate removal process are therefore rare.An exception to this is the investigation at the Burntisland plant in Scot-land (Brown and Cole, 1980). This was a small plant producing gibbsiteand calcined alumina for the chemical industry, which operated for longperiods on low oxalate bauxite from Ghana. During these times theplant ran without operating an oxalate removal process, making it pos-sible to observe the “natural” behaviour of oxalate in the Bayer circuit. Itwas found that the oxalate level in the liquor would rise steadily over aperiod of about three months to a peak which corresponded to the ap-pearance of sodiumoxalate solids co-precipitatedwith the gibbsite seedand product. Oxalatewould then exit the circuitwith the productwhichwas calcined to alumina, thus pyrolysing the oxalate and leading to adecline in the concentration of oxalate in the liquor. The liquor oxalateconcentration would eventually drop until sodium oxalate crystalli-zation ceased (i.e. the sodium oxalate concentration was equal tothe apparent solubility). The soluble oxalate concentration wouldsubsequently increase until it was again sufficiently supersaturatedto create the next round of spontaneous crystallization. The oxalatesupersaturation/de-supersaturation cycle in this case was found tohave a period of about six months. The presence of sodium oxalatesolids was detrimental to the quality of the gibbsite product by causinga reduction in particle strength, and increases in both the sodiumcontentand the proportion of material in the finer size fractions. A direct causeand effect relationship was inferred as these product quality parameterswere found to vary in parallel with the oxalate cycle. These phenomenahave since been observed in many refineries, and are now well knownthroughout the industry.

Lever (1983) interpreted the metastable behaviour of oxalate as aresult of the adsorption of organic “stabilisers” on the surfaces of oxalatecrystals, noting the stabilizing effect associated with high molecularweight compounds (“humates”) from the bauxite. He also observedthat “humates” tend to co-precipitate with the oxalate, creating an in-teractive effect in which crystallization of oxalate can result in a lower-ing of liquor stability by the co-removal of stabilisers. It is the presenceof co-precipitated “humates” that imparts the brown colour typical ofthe sodium oxalate solids found in Bayer process plants.

Sodium oxalate crystals exhibit a variety of morphologies, dependingon the conditions of their formation. When grown from neutral solutionsthey tend to form blocky prismatic shapes, but in sodiumhydroxide solu-tions the crystals are generally acicular (needle-like, see Figs. 18 to 20)due to stabilization of the [110] crystal faces in relation to the [001]faces (Reyhani et al., 2000). Acicular oxalate crystals may agglomeratein a variety of ways to form shapes resembling fans, bow-ties, or balls(spherulites). The most important single factor determining the shapesof the agglomerates appears to be supersaturation,which in turn dependson the presence of organic impurities or additives, such as surfactants, inthe solution (Grocott and Harrison, 1996; Power and Tichbon, 1990).

In order to quantify oxalate supersaturation, Lever (1983) intro-duced a method for measuring the “Critical Oxalate Concentration”(COC), which may be defined as the highest concentration of oxalatethat can be achieved in a given solution before spontaneous nucleationof sodium oxalate occurs. An automatedmethod for COC determination

which can beused to evaluate the effects on oxalate stability of chemicalsadded to the liquor, or for routine monitoring the COC of plant liquors,has been described by Hiralal et al. (1994).

An understanding of oxalate stability is important in the operation ofBayer process plants, because the effects of uncontrolled crystallizationof oxalate can be severe. This is particularly evident in plants processingbauxites with higher organic contents, in which major operating disrup-tions leading to significant losses in production and degradation of prod-uct quality can be caused, as demonstrated by the results of a plant trial inwhich a “humate removal” chemical, poly-diallyldimethylammoniumchloride (poly-DADMAC), was added to the plant liquor (Power andTichbon, 1990). Dosing poly-DADMAC was found to be effective inremoving high-molecular weight organic compounds (loosely re-ferred to as “humates”—see Part 1 of this review for a discussion ofthis term in the Bayer context (Power and Loh, 2010)) from the liquor,and after about six weeks the level of “humates”was estimated to havebeen reduced by about 30%. At this point, sodium oxalate solids beganto appear throughout the precipitation circuit, which led to a series ofundesirable effects. First, the elutriation behaviour of the solids in theprecipitators was affected as fine oxalate needles co-precipitated withgibbsite triggered production of finer product and the formation ofthree-dimensionalmatriceswhich caused aflotation effect and severelydisrupted the seed balance of the plant. Co-precipitation also occurredon the walls of tanks and pipes resulting in a dramatic increase in therate of scale formation and consequent loss of equipment utilisation.Product quality was also affected by an increase in soda impurity andfines content, and a reduction in particle strength. Removal of dosingallowed the “humate” concentration to return to previous levels andnormal plant operation to be re-established.

The material associated with the flotation phenomenon has becomeknown as “non-setting hydrate1”, or “NSH”. NSH is caused by the forma-tion of complex three-dimensional networks of gibbsite and sodium ox-alate crystals which form flocculated masses which settle very slowlywith little compaction, and which are readily entrained by flowingliquor. The NSH can be seen as a layer which forms on top of settledgibbsite if a bottled sample of precipitator slurry taken from a plantor from a laboratory precipitation test is left to stand, as shown inFig. 18(A). Fig. 18(B) is an electron microscope image showing thestructure of the NSH, in which gibbsite crystals can be seen to be incor-porated into a three-dimensional web of oxalate crystals. The photo-graphs in Figs. 18 to 20 were extracted (with the permission of thepublisher) from a detailed study by Reyhani et al. (2000) of oxalate/gibbsite co-precipitation.

Careful study of the complex interactions between sodium oxalateand gibbsite solids during crystallization has revealed that the surfacesof each solidmay provide nucleation sites for the other. Also, organic sub-stances can have significant effects in either promoting or inhibiting suchinteractions. These effects can have a dramatic influence on the shapesand particle size distributions of the product solids (Reyhani et al., 1999,2000), as well as on the crystallization yields achieved. Kingsley andWilcox (1986) reported that oxalate adsorbs to the surfaces of gibbsiteto a small extent only, but neverthelessmay cause ameasurable decreasein gibbsite precipitation yield due to preferential adsorption at gibbsitegrowth sites. Nikolic et al. (2004) also report a decrease in gibbsite growthrate due to oxalate adsorption. Reyhani et al. (2000) have shown thatgibbsite particles are not preferred nuclei for oxalate crystallization.While some oxalate nucleation was observed on gibbsite in pure so-dium aluminate solutions, it appeared to occur only at a small num-ber of sites. Growth was found to not be epitaxial and therefore theoxalate crystals were attached only weakly to the gibbsite. An exampleof competition between gibbsite and oxalate growth units for growthsites on the gibbsite surface can be inferred from the SEM image shown

Fig. 18. (A) “Non-settling hydrate (NSH)” (middle layer—light brown) sitting on settled gibbsite (lower layer—cream) after a laboratory precipitation test in plant liquor (supernatant—dark brown). (B) SEM image showing the fine structure of the solids generally associated with NSH (Reyhani et al., 2000).


in Fig. 19, in which an (acicular) oxalate crystal appears to disrupt thenormal growth of a (hexagonal) gibbsite crystal.

On the other hand, sodiumoxalate needles appear to have siteswhichare favourable for gibbsite nucleation. This results in the formation ofmultiple single crystals of gibbsite nucleated at kink sites on the oxalatesurfaces (Reyhani et al., 1999), which can lead to the generation of alarge population of very fine gibbsite crystals attached to oxalate in theplant environment, as shown in Fig. 3 of Power and Tichbon (1990). Ithas been observed that oxalate nucleates preferentially on the end [002]faces of acicular oxalate crystals, which led to the suggestion that thepreference for kink sites may be enhanced because they reveal [002]faces, possibly indicating some degree of epitaxy (Reyhani et al., 1999).Fig. 20(A) shows the initial stages of nucleation of gibbsite on oxalate,and after 24 h the oxalate needles are completely encased in gibbsite(Fig. 20(B)). In this way, the presence of oxalate needles stimulates theformation of gibbsite nuclei, and at the same time inhibits them from ag-glomerating. This combination of effects provides at least a partial expla-nation of the well-known phenomenon of fines generation in thepresence of solid-phase oxalate in the Bayer process (Brown and Cole,1980; Lever, 1983; Reyhani et al., 1999; The and Bush, 1987; Verghese,

1988), which as noted above can have major negative implications forrefinery operation and product quality (Chartouni, 1989; Lever, 1983;Minai et al., 1978; Power and Tichbon, 1990; Yamada et al., 1973).

7.3. Indirect effects: sodium oxalate control

In addition to the direct effects that oxalate crystallization can haveon the operation and economics of the Bayer process, the control of ox-alate to mitigate those effects introduces other constraints which inturn may carry significant costs. If the concentration of sodium oxalatein the precipitation section of a refinery is allowed to build up to theCOC, spontaneous oxalate crystallization will occur, with the conse-quences described above. Short-termmanagement of such an event re-quires actions such as reducing the overall concentration (TA) of theliquor and water-washing the recycled gibbsite seed (The and Bush,1987). These actions are expensive in raw materials as they requirethe addition of extrawater for dilution and energy for subsequent evap-oration. They also tend to increase soluble liquor losses and hence theneed for additional caustic soda. Finally, lowering the TA reduces pre-cipitation yield which is a major cost to the operation.

Fig. 19. Oxalate (needle) competing with gibbsite (hexagonal plate) for gibbsite growthsite (circled) (Reyhani et al., 2000).


Maintaining the concentration of oxalate in the liquor stream tobelow the COC requires balancing the oxalate output with the oxalateinput. In the processing of bauxites of low organic carbon content, thiscan often be achieved through the normal soluble losses. However for

Fig. 20. Gibbsite nuclei growing on oxalate needles in laboratory test: (A) after 4 h; (B) after24 h (Reyhani et al., 1999, 2000).

bauxites of moderate to high organic carbon content, maintaining asuitable liquor TA generally requires the continuous removal of sodiumoxalate, which necessitates the installation and operation of significantcapital equipment. The costs of oxalate removal must be balancedagainst the benefits in order to determine the optimum removal rateand plant operating conditions, in particular liquor TA and precipitationtemperature profile. Optimising oxalate removal requires knowledge ofboth the thermodynamics and rates of the crystallization of sodium ox-alate as a function of key process variables, in particular seed rate, TAand temperature (BeckhamandGrocott, 1993;McKinnonet al., 1999). In-creased operating flexibility may be possible through the use of chemicaladditives designed to increase the COC in the precipitation area specifical-ly, while not interfering with oxalate removal elsewhere in the process(Farquharson et al., 1995).

8. Soda in product

Sodium is themain impurity byweight in smelter grade alumina,withtypical concentrations in the range of 0.25 to 0.50 %Na2O (soda), which isin the order of ten times the amount of the next largest impurities, silicaand iron (Syltevik et al., 1996). Although sodium is a major componentof the electrolyte bath (sodium aluminium fluoride), the soda content ofthe alumina is critical to the economic performance of the smelterthrough its effect on the sodium/aluminium balance (Syltevik et al.,1996). It is also detrimental to fluoride adsorption in the dry scrubbingprocess (Coyne et al., 1989).

Early theories suggesting that sodamay be incorporated in precipitat-ed gibbsite through the inclusion of liquor in voids (Misra and White,1971; Ohkawa et al., 1985; Sato and Kazama, 1971) have largely beendisproven, and it is now generally accepted that the soda in alumina ispredominantly due to chemical incorporation of sodium in the growinggibbsite lattice (Vernon et al., 2005).

The soda content of alumina has been found to depend on thesquare of the supersaturation during gibbsite precipitation (Sang,1988). Armstrong (1993); Armstrong et al. (1996) confirmed thequadratic dependence on supersaturation, and also quantified depen-dencies on temperature and the TOC for the Queensland Alumina(QAL) plant, as follows:

dS=dA ¼ a � A=TC–A=TC∞ð Þ2 ð9Þ

where:

dS/dA incremental mass of soda incorporated/incremental mass ofalumina precipitated

A/TC A/TC ratio at time tA/TC∞ equilibrium A/C ratioa 0.000598∗TC−0.00036∗T+0.019568∗TOC/TC

The coefficients for the “a” term demonstrate that TOC has the sec-ond greatest influence (after supersaturation). Eq. (9) has the sameform as the equation for gibbsite growth rate (Vernon et al., 2002),which implies that soda incorporation is a linear function of growthrate.

The first attempt to explain soda incorporation on the basis of aprecipitation mechanism appears to be the suggestion by Grocottand Rosenberg (1988) that the sodium ions which are released fromsodium aluminate at the growing surface must diffuse away fromthe surface to avoid being trapped by the growth front. It was furtherhypothesised that the presence of adsorbed organic molecules couldinterfere with the escape of sodium ions, facilitating their incorpora-tion into the growing crystal. Vernon et al. (2005) have investigatedthe mechanism of soda incorporation using a statistical thermody-namic model which assumes that sodium ions are trapped at surfacedefects, along the lines of the theory developed by Hall (1953) forsemiconductor growth in the solid state, and which has been shown

image of Fig.�20

image of Fig.�19

0

0.5

1

0 4 8 12 16

80oC% N

a 2 O

67oC60oC

70oC

RG (g m-2 h-1)

Fig. 21. Soda incorporation as a function of gibbsite growth rate, fitted according to a defect model, redrawn from Vernon et al. (2005). The range of supersaturation typically en-countered in Bayer process precipitation is indicated by the ellipse.


to be applicable to growth from solution (Tsuchiyama et al., 1981).According to this model, soda incorporation is low until a criticalgrowth rate is reached, after which soda incorporation increases rap-idly and then flattens off towards a plateau value, as shown for arange of temperatures by the S-shaped curves in Fig. 21, which dem-onstrates an excellent fit of the model to experimental data (Vernonet al., 2005). The range of supersaturation generally encountered inpractice in the Bayer process is shown in the ellipse, which is consis-tent with the observation that the relationship between soda incor-poration and gibbsite growth rate is generally found to be linear inpractice. This model is based on the assumption that the rate of gen-eration of surface defects increases with supersaturation, which in thecurrent content may be interpreted as an increase of the degree ofbirth-and-spread versus spiral growth. This assumption has been qual-itatively verified by using SEM imaging to observe the appearance ofsurface nuclei as a function of supersaturation, as illustrated in Fig. 22.

A number of studies have established a link between organic com-pounds in solution and the incorporation of soda into precipitatedgibbsite (Armstrong et al., 1996; Clerin and Cristol, 1998; Grocott andRosenberg, 1988; van Bronswijk et al., 1999; Watling et al., 1996).

Fig. 22. Nuclei on gibbsite surface created by “birth-and-spread” mechanism, whichcorrespond to defect sites at which favour the incorporation of sodium ions, enhancedby the presence of tartrate anions (Vernon et al., 2005).

Grocott and Rosenberg (1988) demonstrated a linear relationship be-tween incorporated soda and the fraction of Bayer plant liquor in synthet-ic liquor/plant liquor mixtures. They also demonstrated a correlationbetween the strength of adsorption of organic molecules to gibbsite andthe degree to which they enhanced soda incorporation. Hence, com-pounds withmultiple hydroxyl groups were found to be themost potentpromoters of soda incorporation.

Armstrong extended the range of compounds studied and con-cluded that compounds with adjacent hydroxy groups on an aromaticcarboxylic backbone could enhance soda incorporation by up to 100%during the growth phase of gibbsite precipitation (Armstrong, 1993).He also noted that the same compounds generally tended to decreasethe gibbsite growth rate, so the increase in soda incorporation wasnot a result of increased overall growth rate. This implies that thefirst factor, “a”, in Eq. (9) is reduced by the presence of these organicmolecules.

This point was investigated by Vernon et al. (2005) using tartrate asamodel organic compound known to adsorb to gibbsite surfaces, inhibitprecipitation and increase soda incorporation (Armstrong, 1993). Asshown in Fig. 23, tartrate has a dramatic effect on soda incorporation,particularly in the range of growth rates encountered in Bayer processpractice. It was observed that the presence of tartrate resulted in signif-icant roughening of the gibbsite surface due to a large increase in thenumber of secondary nuclei produced. This was interpreted as anincrease in secondary nucleation due to poisoning of growth sitesby adsorbed tartrate, which effectively increased localised crystalliza-tion rates. Increased soda incorporation is therefore consistent with anincrease in specific reaction rate (i.e. rate per unit of area representedby active growth sites) even though the overall reaction rate is dimin-ished by the presence of tartrate.

In addition to the incorporation of soda by trapping at gibbsitegrowth sites, soda in product can be significantly increased by thepresence of co-precipitated sodium oxalate (Brown and Cole, 1980;Power and Tichbon, 1990; Sang, 1988).

9. Organic process additives

Awide range of chemical additives is used to enhance the efficiency oftheBayer process. Themost significant in termsof both amounts used and

image of Fig.�21

0

0.5

1

0 4 8 12 16

70oC + Tartrate

% N

a 2 O

RG (g m-2 h-1)

70oC

Fig. 23. Soda incorporation in precipitated gibbsite as a function of gibbsite growth rate showing the effect of the presence of 100 ppm of tartrate (grey) in relation to Control with-out tartrate (black) at 70 °C, redrawn from Vernon et al. (2005). The region of supersaturation normally encountered in Bayer process practice is designated by the ellipse.


process impact are the flocculants, clarifiers and filtration aids that arenowused universally in various aspects of solid/liquid separation. Indeed,the introduction of additives was crucial to the development of the lowcost, energy-efficient alumina refineries of today. The first use of an or-ganic chemical additive appears to be the addition of starch as a filteraid for the pressure (“Kelly”)filters thatwere used to separate the bauxiteresidue from the green liquor after digestion (Hudson, 1988). Under thewartime conditions of the 1940s, refineries in the USA were required toprocess lower grade bauxite. This meant much higher residue loads anddrove the main Kelly filters to be replaced by settling tanks for bulk resi-due separation, with filtration reserved for “polishing”, i.e. for removingresidual fine material from the clarifier overflow liquor. This change in-creased the need for starch, which was then used as a flocculant, clarifierand filtration aid. The development of cheaper, more efficient flocculantsin the 1960s and 70s led to the gradual replacement of starch (Pearse andSartowski, 1984). This change was accelerated by the realisation thatstarch degradation products were detrimental to precipitation (The,1980), although it was reported as late as 1996 that for one particular re-finery the organic carbon input from starchwas 1.0 kg/tonne of aluminaproduced, which was 18% of the total organic carbon input, while thecarbon input from flocculant was only 0.009 kg/tonne (Solymár et al.,1996). It has also been suggested that the use of synthetic flocculantsin place of starch results in a greater proportion of the higher molecularweight “humic” compounds being removed by adsorption onto mudparticles because the synthetic flocculants are effective at much lowersurface coverage than starch, which leaves active sites available for theadsorption (Pearse and Sartowski, 1984).

9.1. Flocculants

Thefirst syntheticflocculants to be introduced to the alumina industrywere polyacrylate/polyacrylamide co-polymers, the molecular weightsand functionalities of which could be optimised to suit particular bauxitesand process conditions, in particular the liquor concentration (Hudson,1987; Pearse and Sartowski, 1984). At least two different formulationsare generally used, one for the high concentration green liquor settler,and others for the lower concentration liquors in themudwashing circuit.

Plants in Jamaica were among the first to adopt synthetic flocculant tech-nology, due to the poor settling qualities of the residue from Jamaicanbauxites (Kahane et al., 2008). A new generation of flocculants based onhydroxamated polyacrylamide polymers (HxPAM) was developed inthe 1980s. They delivered improved performance over a range of pa-rameters, including improved thickener underflow densities and over-flow liquor clarities at lower dose rates, and increasing the flow ratesthrough the polishing filters (Ryles and Avotins, 1996). For example,the change to HxPAM flocculants was credited as a significant enablerin achieving a 65% improvement in production rate without major cap-ital costs at the Worsley refinery (Kahane and McRae, 1996). However,the higher unit costs of these new flocculants and ongoing improvementsin polyacrylate/polyacrylamide technology slowed their adoption.

Liquid flocculants are generally supplied in diluted form to aid dosingand dispersion. The solvent oils enter the liquor circuit andmake aminorcontribution to the overall impurity input, but virtually all of the floccu-lant added to themud settlers andwashers reports to the bauxite residuewhere its impact is minimal (Pearse and Sartowski, 1984). Any negativeside-effects effects to the main refinery process are usually attributableto incorrect application rather than inherent effect on the liquor. For ex-ample, over-dosing or inadequatemixing of flocculants can lead to exces-sive solids carryover in clarifier overflow liquors and blinding of polishingfilters (Pearse and Sartowski, 1984), overloading of clarifier rakes, and/orincreased scale formation rates in tanks and pipes. Proper dosing and ves-sel design are essential in order to minimise the usage and maximise theeffectiveness of flocculants (Kahane et al., 2008).

9.2. Dewatering aids

The use of organic dewatering aids to improve the filtration of gibbsiteproduct prior to calcination has significant advantages in reducing the en-ergy requirement for calcination, the volume of wash water required toremove liquor from the filter cake, and the amount of residual soda inthe cake (Mura, 1998; Pearse and Sartowski, 1984; Singh et al., 1999;Wainwright, 1985). Dewatering aids are surfactants which work by re-ducing surface tension by adsorption at the air/liquid interface, and/orby reducing the solid/liquid contact angle by adsorption at the solid/liquid

image of Fig.�23


interface. Surface tension reduction is achieved by the use of anionic sur-factants such as sodium dodecyl sulphate (SDS), whereas cationic surfac-tants such as cetyl trimethyl ammonium bromide (CTAB) are used toreduce the contact angle. Non-ionic surfactants such as coco alcoholethoxylates can affect both surface tension and contact angle, the relativeeffects being a function of structure, in particular the number ofethoxylate groups in the molecule (Puttock et al., 1984). The presenceof high molecular weight compounds from the Bayer liquor adsorbed tothe gibbsite can have a significant effect on the adsorption of dewateringaids, by enhancing the adsorption of anionic surfactants and decreasingthe adsorption of cationic surfactants (Hind et al., 1999; Wainwright etal., 1986). Overall, it appears that anionic surfactants are the most impor-tant constituents of a successful dewatering aid (Singh et al., 1999;Wainwright et al., 1986), although a variety of formulations have beenproposed to optimise performance. For example, the addition of asmall amount of cationic surfactant to an anionic surfactant has beenclaimed to have significant benefits (Puttock, 1985).

In addition to their distinct process benefits, there may be nega-tive effects associated with the use of dewatering aids. For example:ammonium-based cationic surfactants can hydrolyse to the corre-sponding amines in alkaline solution (Singh et al., 1999), which couldbe responsible for amine odours in the filtration area; the anionic sur-factants may contain sulphonates (Osgerby and Osgerby, 1987;Puttock et al., 1984) which may contribute to sulphur dioxide ormercaptan emissions when calcined; and residual surfactant retainedin thewashwatermay enter themain liquor streamwhere it could con-ceivably affect precipitation yield, foaming or oxalate filtration in otherareas of the refinery.

9.3. Scale promotion and inhibition

The formation of scale on the surfaces of tanks, pipes and otherprocess equipment is a major concern in the operation of alumina re-fineries. Some examples of the effects of oxalate and gibbsite scalesare given above. Scale formation can be promoted by the presenceof some organic compounds. For example, presence of gibbsite pre-cipitation inhibitors can result in a tendency towards plate-like crys-tals, which in turn can lead to an increased gibbsite scale formationon pipes and vessel walls (Watling, 2000).

One of the most widespread and difficult scales in alumina plantsis sodium aluminosilicate (desilication product, or DSP) scale, whichtypically forms on hot surfaces, particularly in heat exchanger tubes,digesters and flash vessels for example. Various chemical additiveshave been tested with the aim of inhibiting scale formation, generallythrough preferential adsorption of a polymer (Addai-Mensah et al.,2001). Most of these studies have met with limited success, howevera proprietary formulation which does appear to be capable ofpreventing scale formation has recently been developed (Spitzer etal., 2005). It is claimed that the additive acts by interfering with thegrowth of embryonic crystals in a catalytic fashion, so that it is effec-tive at very low dose rates, and that extensive testing has revealed nodetrimental side effects.

9.4. Miscellaneous additives

A variety of chemicals is added to the liquor which may impactyield by changing the particle size distribution of the solids in precip-itation, and many of which contain surfactants and oils. These may bechemicals added deliberately to influence gibbsite particle size orshape (“crystal growth modifiers” (“CGMs”), see Section 5 above) orto increase oxalate stability, or which enter the liquor as a result of,for example: addition for a different purpose such as improvingdewatering of solids in filtration; unintentional leakage such as of lu-brication oils from process machinery; or contamination of reagentssuch as vegetable oils in bulk caustic soda due to inadequate cleanupbetween cargos. A number of process chemicals, such as liquid

flocculants, contain oils as carriers (diluents) of the active compo-nent. Because Bayer process liquors are generally almost fullyrecirculated, any chemical which continuously enters the liquorstream has the potential to build up to unexpectedly high concentra-tions which may have significant effects on process efficiency and/orproduct quality. Recognition of this has led alumina producers to in-stitute controls on the addition of chemicals to the process, includingcomprehensive impact testing prior to approval of any chemical foruse. Process chemicals may contain ten or more individual compo-nents in formulations which are intended to target one particular as-pect of the operation, such as dewatering aids applied to filtration.Such formulations are generally proprietary and may contain compo-nents with the potential for unexpected, and possibly severe, side ef-fects elsewhere. As a result, effective use of process chemicalsrequires a good level of co-operation between supplier and customer,and rigorous testing of formulations before approving them for use inan alumina plant (Graham and Davies, 1990).

10. Volatile emissions

It has long been recognised that the digestion of bauxite is gener-ally associated with odour emissions which are due to the formationof volatile organic compounds (VOCs) in the reaction of organic com-pounds with the alkaline Bayer liquor (Forster and Grocott, 1996). Ina study at the Queensland Alumina plant, the non-condensible frac-tion of the digestion vapours (which account for approximately 2%of the total gaseous emissions from digestion, the remaining 98%being water vapour) were shown to account for 82% of the odouremissions associated with the plant (Graham et al., 2002). The pro-portion of the VOCs that could contribute to digestion odour formedonly 0.3% of the mass of gases emitted (nearly 90% was hydrogenand 0.9% was methane, the remainder being mostly nitrogen, carbonoxides and oxygen) (Graham et al., 2002). Nevertheless, odour hasbecome a key issue for some alumina plants (Coffey, 2002; Coffeyand Donoghue, 2006; Coffey and Ioppolo-Armanios, 2004; Forster etal., 2005; Galbally, 2004, 2008), and is an emerging issue overall.

The generation of VOCs and hydrogen can be understood by referenceto the base-catalysed redox reactions of organic compounds in digestiongiven in Part 1 of this review series (Power and Loh, 2010). Further insightinto the mechanisms of hydrogen production has been provided by therecent work of Costine et al. (2010, 2011a, 2011b, 2012). The oxidationof complex organic compounds by water leads to hydrogen as themajor reduction product, but can also lead to the generation of methane,propane and other volatile hydrocarbons as well as alcohols, aldehydesand ketones, as reduction products (Loh et al., 2010b). Consistent withthis, methanol, acetaldehyde, formaldehyde, acetone and methyl-ethylketone have all been identified in the vapours from bauxite slurry stor-age (desilication) tanks, digesters and/or evaporators (Galbally, 2004,2008; Galbally et al., 2012).

In addition, it is reasonable to suppose that precursor compoundscontaining nitrogen and sulphur groups would lead to the formationof ammonia, amines and mercaptans as reduction products, which,though present in trace amounts only, would undoubtedly contributeto the odours that characterise bauxite digestion.

Acknowledgements

The authors would like to acknowledge the support of the ParkerCentre for IntegratedHydrometallurgical Solutions for providing fundingfor the preparation of the manuscript, Dr. Peter Smith and Melissa Loanof CSIRO for valuable discussions and detailed review and input to themanuscript, and the Minerals and Energy Research Institute of WesternAustralia (MERIWA) for permission to reproduce photographs frompub-lished reports.

(continued)

Compounds HAO Yield PSD Soda References

1.3. Di-acid anionsPimelate – – – – No references found to specific

effects.Octanedioate – – – – No references found to specific

effects.Azelate – – – – No references found to specific

effects.Hexane dioate – – – – No references found to specific

effects.

1.4. Tri-acid anionsEthane‐1,1,2-tricarboxylate

– – – – No references found to specificeffects.

Tricaballate – – – – No references found to specificeffects.

1.5. Hydroxy-acid anionsLactate ? L – – Brown et al. (1986); Verghese

(1988)2-hydroxy butanoate – – – – No references found to specific

Appendix (continued)


Appendix. Specific effects on precipitation of gibbsite of organiccompounds generally found in Bayer process liquors

The compounds included here are those identified in Part 1 of this re-view series as being generally present in Bayer process liquors (Powerand Loh, 2010). It is noted that not all authors agree on the nature ofthe specific effects in some cases; where there are disagreements the as-sessments shown in this table are an attempt to represent the consensusview. The references cited should be consulted for further information. Itshould also be noted that only compounds reported in the open litera-ture as present in Bayer process liquors are included here, so this list ex-cludes many of the compounds mentioned in this review which couldpresent and potentially have effects at low concentrations.

Key:HAO = hydrate-active organicYield = precipitation yield effectPSD = particle size distribution effectSoda = product soda effectN = no; Y = yes; L = lowers; R = raises

Compounds HAO Yield PSD Soda References

1. Aliphatic acid anionsMost authors agree that, apart from oxalate, the non-hydroxylated aliphatic an-ions found in Bayer process liquors have no discernable specific effect on gibbsiteprecipitation, and that their effects on overall precipitation yield are due solely totheir impacts on ionic strength and the available hydroxide concentration. How-ever, some authors claim minor specific effects under certain circumstances.The specific effects of oxalate are related the formation of solid sodium oxalate.

1.1. Mono-acid anions: straight chainFormate N L N N Ang and West (1992); Brown

et al. (1986); Verghese (1988)Acetate N L N N Ang and West (1992);

Beckham et al. (2005); Brownet al. (1986); Verghese (1988);Wellington and Valcin (2007)

Propanoate – – – – No references found to specificeffects.

Butanoate – – – – No references found to specificeffects.

Valerate – – – – No references found to specificeffects.

Palmitate – – – – No references found to specificeffects.

Stearate – – – – No references found to specificeffects.

1.2. Mono-acid anions: branched chainIsobyryrate – – – – No references found to specific

effects.Isovlalerate – – – – No references found to specific

effects.

1.3. Di-acid anionsOxalate N L Y Y Brown et al. (1986); Brown

and Cole (1980); Calato andTran (1992); Kingsley andWilcox (1986); Lever (1978);Lever (1983); Nikolic et al.(2004); Nikolic et al. (2004);Power (1991); Power andTichbon (1990); Seyssiecq etal. (1999); Smeulders et al.;(2001); The (1980); The andBush (1987); Tucker et al.;(1986)

Malonate N N – – Power (1991)Succinate N N – Y Brown et al. (1986); Grocott

and Rosenberg (1988)Glutarate – – – – No references found to specific

effects.Adipate – – – – No references found to specific

effects.

effects.Tartrate Y L Y Y Armstrong (1993); Coyne et

al. (1994); Grocott andRosenberg (1988); Lee et al.;(1996); Nikolic et al. (2008);Paulaime et al. (2003); Power(1991); Seyssiecq et al.(1999); Watling et al. (1996)

1.6. OtherDimethyl succinate – – – – No references found to specific

effects.

2. Aromatic acid anionsUnlike the aliphatic carboxylates, certain aromatic carboxylates do appear to havespecific effects on gibbsite precipitation due to adsorption through the carboxylatemoieties. Those with two or more hydroxyl groups tend to adsorb more strongly (as isthe case with aliphatic compounds).

2.1. Non-hydroxy-acid anionsBenzoate Y L Y – Ang and West (1992); Lv et al.

(2009)Phthalate; iso- andtere-phthalate

Y R Y – Bouchard et al. (2005); Loh etal. (2010a); Picard et al.(2002)

Hemi- andtri-mellitate;trimesate

Y R Y – Loh et al. (2010a)

Pyromellitate – – – – No references found to specificeffects.

Benzenepenta-carboxylate


Mellitate – – – – No references found to specificeffects.

2.2. Hydroxy-acid anionsSalicilate andm-salicilate

Y L – – Bouchard et al. (2005)

4-hydroxy- and5-hydroxy-phthalate


References

Addai-Mensah, J., Gerson, A.R., Jones, R., Zbik, M., 2001. Reduction of sodium aluminosilicatescale in Bayer plant heat exchangers. In: Anjier, J.L. (Ed.), Light Metals 2001: LightMetals, pp. 13–18.

AMIRA, 2001. Alumina Technology Roadmap. AMIRA International.An, H.Y., Bradshaw, J.S., Izatt, R.M., 1992. Macropolycyclic polyethers (cages) and related

compounds. Chem. Rev. 92 (4), 543–572.Anderson, J., Armstrong, L., Thomas, D., 2001. Wet oxidation in QAL high temperature di-

gestion for plant productivity improvement. Light Metals 2001. TMS, New Orleans,LA, pp. 121–126.

Ang, H.M., West, M., 1992. Effect of impurities on the precipitation of aluminium hy-droxide. Chemeca 1992, pp. 182–189.

Angyal, S.J., Greeves, D., Mills, J.A., 1974. Complexes of carbohydrates with metal cations.3. Conformations of alditols in aqueous solution. Aust. J. Chem. 27 (7), 1447–1456.


Armstrong, L., 1993. Bound Soda Incorporation during Hydrate Precipitation. Third In-ternational Alumina Quality Workshop, Hunter Valley, Australia, pp. 283–292.

Armstrong, L.G., Healy, S.J., 2006. Degradation of liquor organics during Bayer digestioncycles. Travaux 33 (37), 291–301.

Armstrong, L., Hunter, J., McCormick, K., Warren, H., 1996. Bound soda incorporation duringhydrate precipitation—effects of caustic, temperature and organics. In: Hale, W. (Ed.),Light Metals 1996. The Minerals, Metals and Materials Society, pp. 37–40.

Arnswald, W., Kaltenberg, H.G., Guhl, E., 1991. Removal of organic carbon from Bayerliquor by wet oxidation in tube digesters. Light Metals 1991. TMS, pp. 23–27.

Bard, A.J., Faulkner, L.R., 1980. Electrochemical Methods. John Wiley & Sons.Beckham, K.R., Grocott, S.C., 1993. A thermodynamically based model for oxalate solu-

bility in Bayer liquor. Light Metals 1993. TMS, pp. 167–172.Beckham, K., Clarke, P., Cornell, R., 2005. Inhibition of gibbsite crystallization: adsorp-

tion of the gluconate ion. 7th International Alumina Quality Workshop. AQW Inc,pp. 108–111.

Bouchard, N.-A., Breault, R., Picard, F., Chouinard, Y., Menard, H., 2005. Dynamic ad-sorption isotherm for some hydrate active organics and selected degradationproducts with implication for gibbsite precipitation yields. Light Metals 2005.TMS, pp. 197–202.

Bouzat, G., Philopponneau, G., 1991. Physical chemistry models of oxalate and gibbsitesolubilities in Bayer solutions. Light Met. 1991, 97–102.

Brown, N., 1972a. Crystal growth and nucleation of aluminium trihydroxide from seededcaustic aluminate solutions. J. Cryst. Growth 12 (1), 39–45.

Brown, N., 1972b. Secondary nucleation of aluminium trihydroxide in seeded causticaluminate solutions. J. Cryst. Growth 16 (2), 163–169.

Brown, N., 1988. Effect of calcium ions on agglomeration of Bayer aluminum trihydroxide.J. Cryst. Growth 92 (1–2), 26–32.

Brown, N., Cole, T.J., 1980. The behaviour of sodium oxalate in a Bayer alumina plant.Light Met. 1980, 105–117.

Brown, E.R., Headley, A., Greenaway, A.M., Magnus, K.E., 1986. The effect of low-molecular weight organics on precipitation efficiency. Proceedings of Bauxite Sym-posium, VI. Journal of the Geological Society of Jamaica, pp. 158–165.

Burton, W.K., Cabrera, N., Frank, F.C., 1951. The growth of crystals and the equilibrium struc-ture of their surfaces. Philos. Trans. R. Soc. Lond. A Math. Phys. Sci. 243 (866), 299–358.

Calato, R., Tran, T., 1992. Effects of sodium oxalate on the precipitation of aluminatrihydrate from synthetic sodium aluminate liquors. Light Met. 1992, pp. 125–133.

Chartouni, J.A., 1989. Oxalate removal at Alcan Ouro Preto alumina plant. Travaux 19(22), 357–363.

Chaubal, M.V., 1990. Physical chemistry considerations in aluminium hydroxide pre-cipitation. Light Met. 1990, 85–94.

Chinloy, D.R., 1990. Development of an Industrial Oxalate Removal System. Second In-ternational Alumina Quality Workshop, pp. 130–140.

Clerin, P., Cristol, B., 1998. Contribution of liquor composition on bound soda. LightMet. 1998, 141–146.

Coffey, P.S., 2002. Assessing Odour Impacts of an Alumina Refinery by Source Measure-ment, Dispersion Modelling and Field Odour Surveying. 6th International AluminaQuality Workshop. AQW Inc, Brisbane, Australia, pp. 320–326.

Coffey, P., Donoghue, M., 2006. The Wagerup refinery—beyond the controversy. Chem.Eng. 778, 32–36.

Coffey, P.S., Ioppolo-Armanios, M., 2004. Identification of the odour and chemical com-position of alumina refinery air emissions. Water Sci. Technol. 50 (4), 39–47.

Cornell, R., Vernon, C., Pannett, D., 1996. Growth of gibbsite in Bayer liquors. Fourth In-ternational Alumina Quality Workshop Proceedings. Darwin, N.T., pp. 97–103.

Cornell, R.M., Pannett, D.S., Sullivan, N., Clarke, P.C., Bailey, C.M., 1999. Precipitation ofgibbsite: development of a new rate equation. Fifth International Alumina QualityWorkshop, pp. 153–161.

Costine, A., Schibeci, M., Loh, J., Power, G., McDonald, R., 2010. Hydrogen production in Bayerprocess digestion. Travaux: XVIII International Symposium of ICSOBA, Zhengzhou,China, pp. 369–375.

Costine, A., Loh, J.S.C., Power, G., 2011a. Understanding hydrogen in Bayer process emis-sions. 2. Hydrogen production during the degradation of unsaturated carboxylic acidsin sodium hydroxide solutions. Ind. Eng. Chem. Res. 50 (22), 12334–12342.

Costine, A., Loh, J.S.C., Power, G., Schibeci, M., McDonald, R.G., 2011b. Understandinghydrogen in Bayer process emissions. 1. Hydrogen production during the degrada-tion of hydroxycarboxylic acids in sodium hydroxide solutions. Ind. Eng. Chem.Res. 50 (22), 12324–12333.

Costine, A., Schibeci, M., Loh, J.S.C., Power, G., 2012. Understanding the origin andmechanism of hydrogen production in the Bayer process. 9th International AluminaQuality Workshop, Perth, Australia, pp. 208–213.

Counter, J.A., 2006. Crystal growth modifying reagents; nucleation control additives oragglomeration aids? Light Met. 2006, 131–137.

Coyne, J.F., 1989. The Role of Organic Impurity Adsorption on the Precipitation of Alu-mina Trihydrate. The University of New South Wales, Sydney. (122 pp.).

Coyne, J.F., Wong, P.J., Wainwright, M.S., Brungs, M.P., 1989. Factors influencing hydro-gen fluoride adsorption on alumina. Light Met. 1989, 113–118.

Coyne, J.F., Wainwright, M.S., Cant, N.W., Grocott, S.C., 1994. Adsorption of hydroxy or-ganic compounds on alumina trihydrate. Light Metals 1994. TMS, pp. 39–45.

CRC, 2011. Properties of the elements and inorganic compounds. In: Haynes, W.M.(Ed.), Handbook of Chemistry and Physics, 92nd Edition, Internet Version 2012.CRC Press (http://www.hbcpnetbase.com/).

Dean, J.A., 1998. Lange's Handbook of Chemistry, 15th Edition. McGraw Hill.Dewey, J.L., 1981. Boiling point rise of Bayer plant liquors. Light Metals 1981. TMS,

pp. 185–197.Farquharson, G.J., 1996. Modifying crystal growth during precipitation of alumina in

Bayer process—by addition of polysaccharide(s) or polysaccharide graft polymers.Patent No.: AU633726-B2. Nalco Aust Pty Ltd; Catoleum Aust Pty Ltd.

Farquharson, G.J., Gotsis, S., Kildea, J.D., Grocott, S.C., Gross, A.E., 1995. Development ofan effective liquor oxalate stabilizer. Light Met. 1995, 95–101.

Fleming, S., Rohl, A., Lee, M.Y., Gale, J., Parkinson, G., 2000. Atomistic modelling ofgibbsite: surface structure and morphology. J. Cryst. Growth 209 (1), 159–166.

Forster, P., Grocott, S.C., 1996. Alumina refinery odour. Fourth Alumina Quality Work-shop, pp. 478–487.

Forster, P.G., Woodford, N., Schulz, T., 2005. Advances in assessments of odours fromalumina refining. 7th International Alumina Quality Workshop. AQW Inc, Perth,Western Australia, pp. 246–253.

Freij, S.J., Parkinson, G.M., 2005. Surface morphology and crystal growth mechanism ofgibbsite in industrial Bayer liquors. Hydrometallurgy 78 (3–4), 246–255.

Galbally, I., 2004. Wagerup air quality review. Report C/0936. CSIRO Atmospheric Re-search, PMB 1, Aspendale, Victoria 3195, Australia (http://www.epa.wa.gov.au/docs/1215/Response/AppC.pdf).

Galbally, I., 2008. A study of VOCs during winter 2006 at Wagerup, Western Australia.CSIRO Marine and Atmospheric Research, Private Bag 1, Aspendale, Victoria 3195,Australia. (Retrieved September 28, 2011, from http://www.alcoa.com/australia/en/environment/Wagerup%20Winter%202006%20Report_Final.pdf).

Galbally, I., Lawson, S., Brodie, G., Loh, J., Cheng, M., Hibberd, M., Power, G., 2012. Alu-mina refining and air quality: VOCs and odour. Ninth Alumina Quality Workshop,Perth, Australia, pp. 13–19.

Gnyra, B., Brown, N., 1975. Coarsening of Bayer alumina trihydrate by means of crystal-lization modifiers. Powder Technol. 11 (2), 101–105.

Graham, G.A., Davies, T.J., 1990. Chemical additives for the Bayer process: “buyerbeware”. International Alumina Quality Workshop, Perth, Western Australia,pp. 142–152.

Graham, G., Capil, R., Davies, R., 2002. Odour destruction for digestion vent gases. 6th In-ternational Alumina Quality Workshop. AQW Inc, Brisbane, Australia, pp. 316–319.

Grocott, S.C., Harrison, I.R., 1996. Two new oxalate removal processes. Fourth Interna-tional Alumina Quality Workshop, Darwin, Australia, pp. 423–435.

Grocott, S.C., Rosenberg, S.P., 1988. Soda in alumina: possible mechanisms for soda in-corporation. International Alumina Quality Workshop 1988, pp. 271–287.

Guthrie, J.D., The, P.J., Imbrogno, W.D., 1984. Characterization of organics in Bayer liquors.Light Met. 1984, 127–146.

Hall, R.N., 1953. Segregation of impurities during the growth of germanium and siliconcrystals. J. Phys. Chem. 57 (8), 836–839.

Hind, A.R., Bhargava, S.K., Grocott, S.C., 1999. The surface chemistry of Bayer processsolids: a review. Colloids Surf., A 146 (1–3), 359–374.

Hiralal, I.D.K., Holewijn, J.E., Moretto, K., 1994. Critical oxalate concentration (COC) de-termination in Bayer plant liquors using a turbidity method. Light Met. 1994,53–57.

Hudson, L.K., 1987. Alumina Production. The Society for Chemical Industry, Salisbury,UK.

Hudson, L.K., 1988. Evolution of Bayer process practice in the United States. LightMetals 1988. TMS, pp. 31–36.

Ilievski, D., White, E.T., 1995. Modelling Bayer precipitation with agglomeration. LightMet. 1995, 55–61.

Kahane, R.B., McRae, C., 1996. Improved bauxite residue settler performance atWorsley Alu-mina. Fourth International Alumina Quality Workshop, Darwin, Australia, pp. 227–237.

Kahane, R.B., Kumar, A.S., Mooney, A., 2008. Doubling the capacity of 50 year old CCDcircuits at low cost in Jamaica. Eighth International Alumina Quality Workshop.AQW Inc, Darwin, Australia, pp. 295–300.

Karaman, M.E., Antelmi, D.A., Pashley, R.M., 2001. The production of stable hydropho-bic surfaces by the adsorption of hydrocarbon and fluorocarbon carboxylic acidsonto alumina substrates. Colloids Surf., A 182 (1–3), 285–298.

Kim, M.J., Lee, S.O., 2003. Overview of the behaviour of sodium oxalate in Bayer liquorand its effect of the process. In: Crepeau, P.N. (Ed.), Light Metals 2003: Light Metals,pp. 19–24.

King, W.R., 1973. Some studies in alumina trihydrate precipitation kinetics. LightMetals. TMS, pp. 551–563.

Kingsley, J.P., Wilcox, W.R., 1986. Adsorption of oxalate on gibbsite. Hydrometallurgy16 (1), 105–107.

Königsberger, E., 2008. Thermodynamic simulation of the Bayer process. Int. J. Mater.Res. 99 (2), 197–202.

Königsberger, E., Bevis, S., Hefter, G., May, P.M., 2005a. Comprehensive model of syn-thetic Bayer liquors. Part 2. Densities of alkaline aluminate solutions to 90 degreesC. J. Chem. Eng. Data 50 (4), 1270–1276.

Königsberger, E., Eriksson, G., May, P.M., Hefter, G., 2005b. Comprehensive model ofsynthetic Bayer liquors. Part 1. Overview. Ind. Eng. Chem. Res. 44 (15), 5805–5814.

Königsberger, E., May, P.M., Hefter, G.T., 2005c. A comprehensive physicochemicalmodel of synthetic Bayer liquors. 7th International Alumina Quality Workshop.AQW Inc, Perth, Australia, pp. 74–77.

Königsberger, E., May, P.M., Hefter, G., 2006. Comprehensive model of synthetic Bayerliquors. Part 3. Sodium aluminate solutions and the solubility of gibbsite andboehmite. Monatsh. Chem. 137 (9), 1139–1149.

Königsberger, E., Hefter, G., May, P.M., 2009. Solubility and related properties in hydro-metallurgy. Pure Appl. Chem. 81 (9), 1537–1545.

Kotte, J.J., 1981. Bayer digestion and predesilication reactor design. Light Met. 1981,45–81.

Kouznetsov, D., Liu, J.J., O'Brien, K., Counter, J., Kildea, J., 2008. New crystal growth mod-ifiers for the Bayer process. Light Met. 2008, 121–125.

Kouznetsov, D.L., Liu, J., Coleman, K.R., Chester, R.T., Kildea, J.D., 2011. Composition forenhancing the production of crystal agglomerates from precipitation liquor crys-tallization e.g. in Bayer process comprises surfactant or its blend, an oil andwater incorporated together as an emulsion. Patent No.: WO2011002952-A2.Nalco Co.

http://www.hbcpnetbase.com/

http://www.epa.wa.gov.au/docs/1215/Response/AppC.pdf

http://www.epa.wa.gov.au/docs/1215/Response/AppC.pdf

http://www.alcoa.com/australia/en/environment/Wagerup%20Winter%202006%20Report_Final.pdf

http://www.alcoa.com/australia/en/environment/Wagerup%20Winter%202006%20Report_Final.pdf


Kumar, K.V., 2009. Simple kinetic expressions to study the transport process during thegrowth of crystals in solution. Ind. Eng. Chem. Res. 48 (24), 11236–11240.

Lectard, A., Nicolas, F., 1983. Influence of digestion conditions on the metabolism andsolubilization of organic impurities. Travaux 13 (18), 345–351.

Lee, M., Lincoln, F., Parkinson, G., Smith, P., 1996. Microscopic and macroscopic obser-vations of variabilities in the growth of gibbsite crystals. Fourth InternationalAlumina Quality Workshop, Darwin N.T., pp. 217–226.

Lee, M.Y., Parkinson, G.M., 1999. Growth rates of gibbsite single crystals determinedusing in situ optical microscopy. J. Cryst. Growth 198, 270–274.

Lever, G., 1978. Identification of organics in Bayer liquor. Light Met. 1978, 71–83.Lever, G., 1983. Some aspects of the chemistry of bauxite organic matter on the Bayer

process: the sodium oxalate–humate interaction: Travaux du Comite Interna-tional pour l'Etude des Bauxites, d'Alumine et d'Aluminium (ICSOBA), 13 (18),pp. 335–347.

Li, W.J., Shi, E.W., Zhong, W.Z., Yin, Z.W., 1999. Growth mechanism and growth habit ofoxide crystals. J. Cryst. Growth 203 (1–2), 186–196.

Li, J., Prestidge, C.A., Addai-Mensah, J., 2000a. The influence of alkali metal ions on ho-mogeneous nucleation of Al(OH)(3) crystals from supersaturated caustic alumi-nate solutions. J. Colloid Interface Sci. 224 (2), 317–324.

Li, J., Prestidge, C.A., Addai-Mensah, J., 2000b. Secondary nucleation of gibbsite crystalsfrom synthetic Bayer liquors: effect of alkali metal ions. J. Cryst. Growth 219 (4),451–464.

Li, T.S., Livk, I., Ilievski, D., 2003. Supersaturation and temperature dependency of gibbsitegrowth in laminar and turbulent flows. J. Cryst. Growth 258 (3–4), 409–419.

Liu, J.J., O'Brien, K., Kouznetsov, D., Counter, J., 2007. Performance of new crystal growthmodifiers in Bayer liquor. Light Met. 2007, 139–144.

Loh, J., Brodie, G., Naim, F., 2010a. The roles of adsorption in hydrate precipitation. LightMet. 2010, 215–220.

Loh, J.S.C., Brodie, G.M., Power, G., Vernon, C.F., 2010b. Wet oxidation of precipitationyield inhibitors in sodium aluminate solutions: effects and proposed degradationmechanisms. Hydrometallurgy 104 (2), 278–289.

Lu, B.L., Chen, Q.Y., Yin, Z.L., Hu, H.P., 2009. Combined effect of amino and carboxyl group inalpha-alanine on seeded precipitation of sodium aluminate solution. Trans. NonferrousMet. Soc. China 19 (3), 745–750.

Lv, B., Chen, Q., Yin, Z., Hu, H., 2009. Effects of four aromatic carboxylic acids as inhib-itors on the seeded precipitation ratios of sodium aluminate solutions and the ag-glomeration efficiency of gibbsite. Light Met. 2009, 189–192.

McKinnon, A., Parkinson, G., Beckham, K., 1999. A rate equation for crystallization ofsodium oxalate under Bayer conditions. Fifth International Alumina Quality Work-shop. AQW Inc, Bunbury, Western Australia, pp. 193–200.

Minai, S., Kainuma, A., Fujiike, M., Yoshida, O., 1978. Sandy alumina conversion. LightMet. 1978, 95–110.

Misra, C., White, E.T., 1971. Crystallisation of Bayer aluminium trihydroxide. J. Cryst.Growth 8 (2), 172–178.

Motekaitis, R.J., Martell, A.E., 1984. Complexes of aluminum(III) with hydroxy carboxylicacids. Inorg. Chem. 23 (1), 18–23.

Mura, I., 1998. Improving the filtration of alumina trihydrate at Eurallumina. Light Met.1988, 173–178.

Newchurch, F.N., Moretto, K.E., 1990. Modern refinery design for quality alumina: aWorsley case study. Second International Alumina Quality Workshop, Perth, Australia,pp. 1–13.

Nicolas, F., 1986. Solubility equilibrium of gibbsite and boehmite in sodium aluminateliquors. 5th Yugoslav International Symposium on Aluminium, Mostar, Yugoslavia,pp. 67–81.

Nikolic, I., Blecic, D., Blagojevic, N., Radmilovic, V., Kovacevic, K., 2004. Influence ofoxalic acid on the kinetics of AI(OH)(3) growth from caustic soda solutions. Hydro-metallurgy 74 (1–2), 1–9.

Nikolic, I., Blecic, D., Blagojevic, N., 2008. The influence of tartaric acid on the phenomenaof Al(OH)3 crystallization from the caustic soda solution. Chem. Ind. Chem. Eng. Q. 14(1), 39–45.

Ohkawa, J., Tsuneizumi, T., Hirao, T., 1985. Technology of controlling soda pick-up inalumina trihydrate precipitaiton. Light Met. 1985, 345–366.

Osgerby, D.J., Osgerby, K.J., 1987. Drainage aid in filtration processes, esp., the Bayerprocess—comprising an alkyl benzene sulphonic acid, opt. partially neutralised.Patent No.: AU8666927-A.

Owen, D.O., Davis, D.C., 1988. Alumina trihydrate production by Bayer process—usingmutually soluble oil and surfactant addition prior to crystallisation to give largerproportion of coarse crystals. Patent No.: US4737352-A. Nalco Chemical Company.

Paulaime, A.M., Seyssiecq, I., Veesler, S., 2003. The influence of organic additives onthe crystallization and agglomeration of gibbsite. Powder Technol. 130 (1–3),345–351.

Pearse, M.J., Sartowski, Z., 1984. Application of special chemicals (flocculants anddewatering aids) for red mud separation and hydrate filtration. In: Jacob Jr., L.(Ed.), Bauxite Symposium. American Institute of Mining, Metallurgical and Petro-leum Engineers, Los Angeles, CA, pp. 788–810.

Pearson, T.G., 1955. The Chemical Background of the Aluminium Industry. The RoyalInstitute of Chemistry, London. (103 pp.).

Picard, F., Audet, D., Boily, H., Larocque, J., 2002. Identification of hydrate active or-ganics (HAO) present in spent Bayer liquors by state-of-the-art analytical methods.6th International Alumina QualityWorkshop. AQW Inc, Australia, Brisbane, Australia,pp. 46–53.

Pitzer, K.S., 1991. In: Pitzer, K.S. (Ed.), Activity Coefficients in Electrolyte Solutions. CRCPress, Boca Raton, Florida, USA, p. 75.

Power, G.P., 1991. The impact and control of organic compounds in the extraction ofalumina from bauxite. 5th AusIMM Extractive Metallurgy Conference. AustralianInstitute of Mining and Metallurgy, Perth, Western Australia, pp. 337–345.

Power, G., Loh, J., 2010. Organic compounds in the processing of lateritic bauxites toalumina. Part 1: Origins and chemistry of organics in the Bayer process. Hydromet-allurgy 105, 1–29.

Power, G.P., Tichbon,W., 1990. Sodium oxalate in the Bayer process: Its origin and effects.2nd International Alumina Quality Workshop, Perth, Western Australia, pp. 99–115.

Power, G., Loh, J.S.C., Niemela, K., 2011a. Organic compounds in the processing of later-itic bauxites to alumina Addendum to Part 1: Origins and chemistry of organics inthe Bayer process. Hydrometallurgy 108 (1–2), 149–151.

Power, G., Loh, J.S.C., Wajon, J.E., Busetti, F., Joll, C., 2011b. A review of the determina-tion of organic compounds in Bayer process liquors. Anal. Chim. Acta 689, 8–21.

Puttock, S.J., 1985. The role of particle properties and interfacial tension in thedewatering of alumina trihydrate. Alumina Technology: Proceedings of Workshop.University of New South Wales, Sydney, Australia, pp. 330–347.

Puttock, S.J., Fane, A.G., Fell, C.J.D., Robins, R.G., Wainwright, M.S., 1984. Improveddewatering of alumina trihydrate. Chemeca, Melbourne, Australia, 84, pp. 277–283.

Raty, M., Stanton, K.T., Hodnett, B.K., Loan, M., 2004. Calcite in the mud washing circuit ofthe Bayer process: experimental solubility and adsorption. LightMet. 2004, 105–108.

Reyhani, M.M., et al., 1999. Gibbsite nucleation at sodium oxalate surfaces. Fifth Inter-national Alumina Quality Workshop, pp. 181–191.

Reyhani, M.M., et al., 2000. Investigations at the atomic level of interactions betweengibbsite and sodium oxalate in the Bayer process. Report No 210. Minerals and En-ergy Research Institute of Western Australia (MERIWA) (73 pp.).

Roach, G.I.D., 2000. Equilibrium approach to causticisation for optimising liquorcausticicity. Light Metals Minerals. Metals and Materials Society/AIME, Nashville,TN, pp. 97–103.

Roe, W.J., Perisho, J.L., 1986. Acrylic acid polymer addition to Bayer process liquors - toincrease diameter of alumina trihydrate crystals and agglomerate contaminatingsodium oxalate crystals. Patent. Nalco Chemical Company.

Roe, W.J., Owen, D.I., Jankowski, J.A., 1988. Crystal growth modification: practical andtheoretical considerations for the Bayer process. International Alumina QualityWorkshop. Gladstone, Australia, pp. 288–298.

Rosenberg, S.P., Healy, S., 1996. A thermodynamic model for gibbsite solubility in Bayer li-quors. Fourth International Alumina Quality Workshop, Darwin, Australia, pp. 301–310.

Rosenberg, S.P., Wilson, D.J., Heath, C.A., 2001. Some aspects of calcium chemistry inthe Bayer process. In: Anjier, J.L. (Ed.), Light Metals 2001: Light Metals, pp. 19–25.

Rosenqvist, J., Axe, K., Sjoberg, S., Persson, P., 2003. Adsorption of dicarboxylates onnano-sized gibbsite particles: effects of ligand structure on bonding mechanisms.Colloids Surf., A 220 (1–3), 91–104.

Rossiter, D.S., Ilievski, D., Smith, P.G., Parkinson, G.M., 1996. The mechanism of sodi-um gluconate poisoning of gibbsite precipitation. Chem. Eng. Res. Des. 74 (A7),828–834.

Rossiter, D.S., Ilievski, D., Parkinson, G.M., 1999. Sodium gulconate poisoning of prima-ry and secondary nucleation of gibbsite from synthetic Bayer liquors. Fifth Interna-tional Alumina Quality Workshop, pp. 170–181.

Ryles, R.G., Avotins, P.V., 1996. A new technology for the alumina industry. Fourth Alu-mina Quality Workshop, Darwin, Australia, pp. 206–215.

Sang, J.V., 1988. Factors affecting residual Na2O in precipitation products. Light Met.1988, 147–156.

Sato, C., Kazama, S., 1971. Behaviour if organic matter in aluminate solution. Light Met.1971, 63–74.

Seyssiecq, I., Veesler, S., Pepe, G., Boistelle, R., 1999. The influence of additives on thecrystal habit of gibbsite. J. Cryst. Growth 196 (1), 174–180.

Singh, B.P., Besra, L., Prasad, A.R., 1999. Role of surface effects in improved dewateringof alumina trihydrate. Light Met. 1999, C27–C31.

Smeulders, D.E., Wilson, M.A., Armstrong, L., 2001. Insoluble organic compounds in theBayer process. Ind. Eng. Chem. Res. 40 (10), 2243–2251.

Smith, P., Woods, G., 1993. The measurement of very slow growth rates during the in-duction period in aluminum trihydroxide growth from Bayer liquors. Light Met.1993, 113–117.

Smith, P., Austin, P., Ilievski, D., 1995. Modelling the induction period in gibbsite pre-cipitation. Light Met. 1995, 45–49.

Smith, P.G., 2011. Bayer process notes: personal communication.Smith, P.G., Watling, H.R., Crew, P., 1996. The effects of model organic compounds on

gibbsite crystallization from alkaline aluminate solutions: polyols. Colloids Surf.,A Physicochem. Eng. Asp. 111 (1–2), 119–130.

Solymár, K., Gimpel-Kazár, M., Molnár, E., 1996. Determination and evaluation of or-ganic balances of alumina refineries. Light Met. 1996, 29–35.

Spitzer, D., et al., 2005. Reagents for the elimination of sodalite scaling. Light Metals2005. TMS, pp. 183–188.

Stephen, H., Stephen, T. (Eds.), 1963. Solubilities of Inorganic and Organic Compounds,vol. 1, Parts 1 & 2. Pergamon Press, Oxford.

Stephenson, J.L., Kapraun, C., 1997. Dynamic modelling of yield and particle size distri-bution in continuous Bayer precipitation. Light Met. 1997, 89–95.

Sweegers, C., de Coninck, H.C., Meekes, H., van Enckevort, W.J.P., Hiralal, I.D.K., Rijkeboer,A., 2001. Morphology, evolution and other characteristics of gibbsite crystals grownfrom pure and impure aqueous sodium aluminate solutions. J. Cryst. Growth 233(3), 567–582.

Sweegers, C., Boerrigter, S.X.M., Grimbergen, R.F.P., Meekes, H., Fleming, S., Hiralal,I.D.K., Rijkeboer, A., 2002. Morphology prediction of gibbsite crystals—an explana-tion for the lozenge-shaped growth morphology. J. Phys. Chem. B 106 (5),1004–1012.

Syltevik, D., Lillebuen, B., Larsen, J.A., 1996. Alumina quality—user perspective. FourthAlumina Quality Workshop, pp. 21–31.

Teas, E.B., Kotte, J.J., 1980. The effect of impurities on process efficiency and methodsfor impurity control and removal. Bauxite Symposium, IV. Journal of the GeologicalSociety of Jamaica, Kingston, Jamaica, pp. 100–129.


The, P.J., 1980. The effect of glucoisosaccharinate on the Bayer precipitation of aluminatrihydrate. Light Met. 1980, 119–130.

The, P.J., Bush, J.F., 1987. Solubility of sodium oxalate in Bayer liquor and a method ofcontrol. Light Metals 1987. TMS, pp. 5–10.

The, P.J., Sivakumar, T.J., 1985. The effect of impurities on calcium in Bayer liquor. LightMet. 1985, 209–222.

Tran, T., Kim, M.J., Wong, P.L.M., 1996. The effect of 3,4 dihydroxy benzoic acid (3,4DHBA) on the precipitation of alumina trihydrate. Light Met. 1996, 557–572.

Tromans, A., Konigsberger, E., May, P.M., Hefter, G., 2005. Heat capacities and volumesof aqueous dicarboxylate salt solutions of relevance to the Bayer process. J. Chem.Eng. Data 50 (6), 2019–2025.

Tschamper, O., 1987. Precipitation in the Bayer process. Light Met. 1987, 38–41.Tsuchiyama, A., Kitamura, M., Sunagawa, I., 1981. Distribution of elements in growth of

(Ba, Pb)(NO3) crystals from the aqueous solution. J. Cryst. Growth 55 (3), 510–516.Tucker, D.M., Greenaway, A.M., D.C.G., C., 1986. The effect of Bayer liquor composition

on the precipitation of alumina trihydrate. Proceedings of Bauxite Symposium, VI.Journal of the Geological Society of Jamaica, 145–165.

van Bronswijk, W., Watling, H.R., Yu, Z., 1999. A study of the adsorption of acyclicpolyols on hydrated alumina. Colloids Surf., A 157 (1–3), 85–94.

Veesler, S., Boistelle, R., 1994. Growth kinetics of hydragillite Al(OH)3 from caustic sodasolutions. J. Cryst. Growth 142 (1–2), 177–183.

Verghese, K.I., 1988. The impact of impurities on the Bayer process. Light Metals 1988.TMS, pp. 42–46.

Vernon, C., Parkinson, G.M., Lau, D., 1999. Towards a fundamental rate equation forgibbsite growth in Bayer liquors. Fifth International Alumina Quality Workshop,pp. 129–139.

Vernon, C., Brown, M.J., Lau, D., Zieba, M.P., 2002. Mechanistic investigations of gibbsitegrowth. 6th International Alumina Quality Workshop. AQW Inc, pp. 33–39.

Vernon, C., Loh, J., Lau, D., Stanley, A., 2005. Soda incorporation during hydrate precip-itation. Light Met. 2005, 191–196.

Volf, F.F., Pudovkinoj, O.I., 1935. Bayer Method in the Processing of Ural bauxite [inRussian]. Uralian Scientific Chemical Institute Unikhim, Sverdlovsk. (115 pp.).

Wainwright, M.S., 1985. Organic chemicals in Bayer processing of alumina. Chem. Aust.52 (11), 430–431.

Wainwright, M.S., Fane, A.G., Fell, C.J.D., Robins, R.G., Fox, M.H., 1986. The role of surfac-tant adsorption in the improved dewatering of alumina trihydrate. Light Met.1986, 231–238.

Watling, H., 2000. Gibbsite crystallization inhibition—2. Comparative effects of selectedalditols and hydroxycarboxylic acids. Hydrometallurgy 55 (3), 289–309.

Watling, H.R., Smith, P.G., Loh, J., Crew, P., Shaw, M., 1996. Comparative effects of modelorganic compounds on gibbsite crystallization. Fourth International Alumina QualityWorkshop, pp. 553–555.

Watling, H., Loh, J., Gatter, H., 2000. Gibbsite crystallization inhibition—1. Effects of sodiumgluconate on nucleation, agglomeration and growth. Hydrometallurgy 55 (3), 275–288.

Watts, H.L., Utley, D.W., 1953. Volumetric analysis of sodium aluminate solutions. Anal.Chem. 25, 4.

Watts, H.L., Utley, D.W., 1956. Sodium gluconate as a complexing agent in the volumetricanalysis of aluminum compounds. Anal. Chem. 28, 6.

Welton,R.,McColl, P., 1999. Crystallizationmodifier composition for Bayer process pregnantliquor. Patent No.: WO2008076783–A2. Ciba Specialty ChemWater Treatments Ltd.

Wellington, M., Valcin, F., 2007. Impact of Bayer process liquor impurities oncausticization. Ind. Eng. Chem. Res. 46 (15), 5094–5099.

Whittington, B.I., 1996. The chemistry of CaO and Ca(OH)(2) relating to the Bayer pro-cess. Hydrometallurgy 43 (1–3), 13–35.

Wu, Z.P., Chen, Q.Y., Yin, Z.L., 2005. DFT and ab initio calculation on thermochemistry ofAl(6)(OH)(18)(H(2)O)(x)(x=0-6), Al(OH)(6)(−3) and Al(OH)(4)(H(2)O)(2)(−).In: Kvande, H. (Ed.), Light Metals 2005: Light Metals, pp. 229–234.

Yamada, K., Hashimoto, T., Nakano, K., 1973. Behavior of organic substances in theBayer process. Light Met. 1973, 745–754.

Yin, J.G., Li, J., Zhang, Y.L., Chen, Q.Y., Yin, Z.L., 2006. Effects of monohydroxy-alcohol ad-ditives on the seeded agglomeration of sodium aluminate liquors. In: Galloway, T.J.(Ed.), Light Metals 2006: Vol 1: Alumina & Bauxite. Light Metals, pp. 153–157.

Yin, J.G., Li, W.X., Chen, Q.Y., Yin, Z.L., 2009a. Effect of cationic polyacrylamide on theseeded agglomeration of sodium aluminate liquors. Light Met. 2009, 201–206.

Yin, Z.L., Zeng, J.S., Chen, Q.Y., 2009b. Effect of oleic acid on gibbsite precipitation fromseeded sodium aluminate liquors. Int. J. Miner. Process. 92 (3–4), 184–189.

Yu, H., Bi, S., Zhou, H.L., Cheng, T., Miao, Y., 2007. Effect of long chain fatty acid collectoron seed precipitation from sodium aluminate liquor. Light Met. 2007, 163–165.

Zambo, J., 1986. Structure of sodium aluminate liquors; molecular model of the mech-anism of their decomposition. Light Met. 199–215.

Zeng, J.S., Yin, Z.L., Chen, Q.Y., 2007. Intensification of precipitation of gibbsite fromseeded caustic sodium aluminate liquor by seed activation and addition of crownether. Hydrometallurgy 89 (1–2), 107–116.

Zeng, J.S., Yin, Z.L., Zhang, A.M., Qi, G.W., Chen, Q.Y., 2008. Unusual effect of 1,2-octanediol on sodium aluminate solutions leading to inhibition of gibbsite crystal-lization. Hydrometallurgy 90 (2–4), 154–160.

Zhang, B., Li, J., Chen, Q., 2007. Effect of modified additives on process of Bayer seedsprecipitation from caustic aluminate solutions. Light Met. 2007, 177–181.

Zhong, W.Z., Liu, G.Z., Shi, E.W., Hua, S.K., Tang, D.Y., Zhao, Q.L., 1994. Growth units andformation mechanisms of the crystals under hydrothermal conditions. Sci. ChinaSer. B Chem. Life Sci. Earth Sci. 37 (11), 1288–1297.

organic compounds in the processing of lateritic bauxites

Documents