M.M. Reyhani1, A. Dwyer1, G.M. Parkinson1, S.P. Rosenberg2, S.J. Healy2
L. Armstrong3, A. Soirat3 and S. Rowe3

1)A.J. Parker CRC for Hydrometallurgy, School of Applied Chemistry, Curtin University of Technology, GPO Box U1987, Perth WA 6845
2)Worsley Alumina Pty. Ltd, P. O. Box 344 Collie WA 6225
3)Queensland Alumina Ltd, Parsons Point Gladstone, Queensland 4680


Sodium oxalate is one of the most important impurities in an alumina refinery, due to its limited solubility in Bayer Process liquor streams. If allowed to build to a critical supersaturation, sodium oxalate will co-precipitate with gibbsite in Bayer precipitator circuits. The occurrence of this in the first stages of gibbsite precipitation can lead to gibbsite fines "showers" and hence to excessively fine alumina. This has variously been ascribed to oxalate interfering with gibbsite agglomeration, or to enhanced gibbsite nucleation. Despite the serious and costly impact that oxalate can have on the particle size distribution of gibbsite, little work appears to have been published which examines this relationship.

In this study, the nucleation of gibbsite at oxalate surfaces is examined in both plant and synthetic liquors. Scanning and transmission electron microscopy (SEM and TEM) of these structures has revealed interesting characteristics of these crystal intergrowths. The relationship between this interaction and gibbsite particle size characteristics in Bayer precipitation circuits is also discussed.


sodium oxalate; nucleation; precipitation; intergrowth; gibbsite


M.M. Reyhani, A. Dwyer, G.M. Parkinson, S. P. Rosenberg, S. J. Healy

L. Armstrong, A. Soirat and S. Rowe


Sodium oxalate is produced by the decomposition of humic substances in caustic solutions. Bayer liquors serve to extract these humic substances from bauxites, and sodium oxalate (along with a variety of other species) is generated as a product of their degradation. If the concentration of this sodium oxalate rises to a critical supersaturation, the oxalate will precipitate, along with gibbsite in the crystallisers.

The presence of this solid phase oxalate (SPO) has long been associated with the tendency towards finer gibbsite product from precipitation circuits. There are numerous references in the literature regarding the effect of oxalate on the precipitation of gibbsite in Bayer refineries (Lever, 1978; Sato,1971; Brown,1980; Sang,1988; The, 1987; Calalo, 1993). However, the discussion is generally restricted to a comment that when solid phase oxalate is present, precipitators tend to produce finer gibbsite product. The cause of this phenomenon is ascribed to either nucleation, inhibition of agglomeration, or both.

Minai et al. (Minai, 1978) determined experimentally that increasing concentrations of solid phase oxalate in gibbsite seed increased the fineness of gibbsite precipitated from a seeded liquor. Grocott and Rosenberg (Grocott, 1988) discussed the inclusion of oxalate into gibbsite agglomerates and presented an optical micrograph showing oxalate needles surrounded by regularly sized gibbsite crystals similar to some of the cases illustrated in this paper. Also discussed is the relationship between oxalate and finer alumina product. They proposed that the inclusion of oxalate crystals in the gibbsite agglomerates results in particle degradation in calcination.

Power and Tichbon (Power,1990) demonstrated experimentally that excessive nucleation of gibbsite occurred in plant liquor when seeded with unwashed plant oxalate (1.5% SPO). Similar experiments in which A.R. oxalate was added to liquors made from clean gibbsite seed did not have the same effect. It was concluded that the oxalate surfaces had to be contaminated in some way, presumably with organic adsorbents, for this effect to occur. Power (Power,1991) published a micrograph of "single crystal nucleation on plant oxalate", and discussed the potential effects of this nucleation phenomena, such as the influence of oxalate on the formation of scale. It was proposed that when oxalate precipitates onto metal surfaces, it can act as a nucleation site for gibbsite. Also discussed was the possibility of the formation of open oxalate/gibbsite structures that, due to their hydrodynamic behaviour, disrupt classification.

All of the above studies deal with systems where liquors are seeded with gibbsite and there is oxalate and often organics present. Power (Power, 1991) appears to have made the most detailed study of this behaviour, however, no mechanism was proposed. Published studies where the behaviour of supersaturated sodium aluminate solutions is examined in the presence of sodium oxalate crystals alone are rare or nonexistent. A study under these conditions is a prerequisite to a full understanding of the interaction between oxalate and gibbsite, without the complicating factors of nucleation, agglomeration and growth introduced by the presence of gibbsite seed.

The object of this work was to use electron microscopy to examine nucleation in synthetic and plant liquors using plant and pure oxalate. This would indicate whether or not there is a direct interaction between oxalate and sodium aluminate solutions, or whether the nucleation phenomenon was provoked by another species, or a particular set of conditions that exist only in refinery situations. Examination of the resultant crystals was also focussed on defining a mechanism for the interaction.


Gibbsite nucleation from supersaturated aluminate solutions was studied in the presence of pure and plant sodium oxalate, in both plant and synthetic liquors. Experiments were performed as isothermal precipitation tests in a rolling water bath at 80C in polypropylene bottles. Bottles were seeded with solid phase oxalate and allowed to precipitate for various times up to 24 hours. Samples were taken at different stages of the precipitation process to observe the nucleation and subsequent growth of the gibbsite crystals. Special sample preparation techniques were developed to minimise the artefacts caused by residual liquor. Liquors were analysed for A, C, S to determine the extent of precipitation. The samples were examined using electron microscopy (SEM & TEM).

Figure 1a: Gibbsite free plant oxalate


Figure 1b: A.R. sodium oxalate (BDH )

Figure 1c: Plant synthetic oxalate


Figure 1d: Pure synthetic oxalate



Plant green liquor to precipitation (LTP) was used in the plant liquor tests. Plant and AR sodium oxalate were used as seed materials (Figure 1a & 1b). All traces of fine gibbsite were removed from plant oxalate by digesting the oxalate cake in a hot caustic solution saturated with respect to oxalate, then displacement washing with alcohol to remove residual liquor and surface contaminants. The plant oxalate was tested for the presence of gibbsite by XRD analysis and by SEM inspection. ‘BDH Analar’ sodium oxalate was used for the pure oxalate seed (hereafter called "A.R. oxalate"). Figure 1c shows an SEM image of fine oxalate needles prepared using plant ingredients and Figure 1d shows the synthetic sodium oxalate produced by precipitation from caustic/sodium oxalate solutions prepared from pure ingredients. All three types of solid oxalate needles were used as seed materials in precipitation experiments.



In plant liquors, substantial nucleation of gibbsite occurs along the length of the oxalate needles. Typical structures are shown in Figure 2a. Clearly visible are the numerous small tabular gibbsite crystals distributed along the length of each needle. These crystals are very regular, and appear mono-dispersed with a diameter of approximately 10 mm. Due to their appearance, we refer to these gibbsite-coated oxalate needles as "kebab" structures. Some of the intergrowths appear to have agglomerated with other similar structures. The nature of these structures is more clearly seen by referring to Figures 2b and 2c. The regularity of the gibbsite crystals is readily apparent, with each particle almost identical in shape and size. This mono-disperse size distribution implies that these crystals all nucleated at the same time, and grew at the same rate. Since the gibbsite crystals are immobilised on the oxalate needle, they are prevented from agglomerating with each other to the extent which would otherwise normally occur under these conditions.

A notable characteristic of the gibbsite is the seemingly random orientation of the platelets with respect to the axis of the oxalate needle. If some characteristic of the lattice structure of the oxalate needle were acting as a template for gibbsite nucleation, this randomness would not be expected. This observation suggests that either the interaction is not epitaxial or there are a number of lattice matches between gibbsite and sodium oxalate that have similar interfacial energies. High resolution transmission electron microscopy and diffraction studies are underway to distinguish between these alternatives. The shape of the gibbsite crystals themselves is unusual. Rather than assuming the hexagonal platelet shape expected of gibbsite nucleating and growing in caustic solutions, they appear to have formed as near-circular discs. In other crystalline systems, this type of morphology change is generally caused by the presence of a "crystal modifier", a species in solution which disturbs the crystal morphology during its growth. Organic species are trapped within and/or on the plant oxalate, deriving from the liquor from which it is crystallised. Therefore, it is expected that plant liquors will contain significant concentrations of these organics. It is possible that these organics are responsible for the behaviour seen here.

The product formed by the seeding of LTP with AR oxalate, shown in Figure 2d, consists of extensive fine gibbsite crystals overlying the oxalate seed crystals; the particles retain the overall form of the underlying seeds. This suggests that fairly widespread nucleation has occurred over the oxalate surface. Some of the oxalate needles have gibbsite crystals growing on the end (002) faces, and relatively few on their prismatic (110) and (200) faces, suggesting that preferential nucleation occurs on the former. There is some confusion in the literature over the indexing of crystal faces exhibited by sodium oxalate, due to the different assignment of axes in two reported crystal structures. In this paper, we use the more recent work of Strom et al (Strom, 1995), rather than those of Reed and Olmstead (Reed, 1981). Thus sodium oxalate has a monoclinic crystal structure with space group P21/a, a = 10.375, b = 5.243, C = 3.449 oA, b = 92.66o (11). Figure 2e & 2f show the growth of gibbsite crystals on the fine long oxalate needles prepared in the laboratory from plant liquor (Figure 1c), and which therefore contain some plant organics. At points, some crystalline outgrowths have appeared on the surface, with the normal hexagonal crystal habit expected of gibbsite, with less well defined gibbsite beneath and between them.

Figure 2a: Formation of gibbsite crystals on plant sodium oxalate

Figure 2b: Formation of gibbsite crystals on plant sodium oxalate

Figure 2c:Formation of gibbsite crystals on plant sodium oxalate

Figure 2d Formation of gibbsite crystals on AR sodium oxalate

Figure 2e: Formation of gibbsite crystals on fine oxalate needles prepared from plant ingredients

Figure 2f: Formation of gibbsite crystals on fine oxalate needles prepared from plant ingredients


Synthetic liquors were prepared to the same aluminate supersaturation targets as the LTP. Supersaturation was calculated using the Rosenberg/Healy solubility model (Rosenberg, 1996).

The oxalate needles used for seeding were prepared by precipitation from caustic/sodium oxalate solutions prepared from pure ingredients. Oxalate supersaturation and precipitation conditions were manipulated to precipitate oxalate needles with a similar aspect ratio to that of plant needles. Apart from variations in aspect ratio, these pure needles differed from the plant oxalate in being generally well defined hexagonal prisms, as opposed to the ‘splinter’ forms of plant oxalate, a difference seen by comparing Figure 1d with Figure 1a.

The nucleation of gibbsite on synthetic oxalate needles was analogous to that observed with plant liquor seeded by plant oxalate. The ‘kebab’ structure was observed again. The gibbsite crystals were distributed in a similar way and were of a similar size to those observed in the plant experiments. The gibbsite crystals displayed a similar habit to those of the plant experiments but were not rounded in the way that was observed under those conditions. Figure 3a shows the formation of the gibbsite crystals produced in plant liquor on plant oxalate crystals, which may be compared with the similar "kebab" structure observed in the pure synthetic system in Figure 3b. A systematic study of the formation of gibbsite crystals looking at the early stages of their nucleation and growth is shown in Figure 4. Figures 4 a to c show the defined hexagonal morphology of the gibbsite crystals nucleated onto oxalate needles at the early stages of their formation, which exhibit a more defined structure when compared to those produced in the presence of plant ingredients (see, for example Figure 2c). Figures 4d to 4 f are SEM images of the crystals formed after 4, 7 and 24 hours of gibbsite growth, respectively, showing the sequence by which initial nucleation and subsequent growth of gibbsite leads to the final product. Studies of the sites of initial gibbsite nucleation suggest that it occurs preferentially on steps on the needle sides and on the needle ends, both involving the (002) face.


Figure 3a: Formation of gibbsite on plant oxalate

Figure 3b: Formation of gibbsite on synthetic oxalate needles

Figure 4a: Early stages of the formation of gibbsite onto synthetic oxalate needles

Figure 4b: Early stages of the formation of gibbsite onto synthetic oxalate needles

Figure 4c: Early stages of the formation of gibbsite onto synthetic oxalate needles

Figure 4d: Formation of gibbsite onto synthetic oxalate needles after 4 hours.

Figure 4e: Formation of gibbsite onto synthetic oxalate needles after 7 hours.

Figure 4f: Formation of gibbsite onto synthetic oxalate needles after 24 hours.

Figures 5a to d show high resolution TEM images taken from the early stages of the formation of gibbsite crystals onto the side faces of oxalate needles in a pure synthetic system. The oxalate needles appear as black areas on the micrographs, because they are too thick to allow the penetration of the electron beam, unlike the gibbsite crystals which are comparatively thin. Interestingly, gibbsite crystals as small as <100 nm show well defined hexagonal morphologies, indicating that gibbsite nucleates onto sodium oxalate as highly crystalline material, even at such small sizes. The crystallographic relationship between the two crystals is being currently studied by TEM and electron diffraction, and this will be reported in future work (Reyhani, 1999a). However, a preliminary interpretation of the observed orientation of the gibbsite nuclei with respect to the sodium oxalate needle crystals is that the gibbsite preferentially nucleates with its prismatic rather than basal faces aligned close to the oxalate needle axis.


Figure 5

TEM Images of the early stages of the formation of synthetic gibbsite on a synthetic oxalate needle



In the Bayer process, solid sodium oxalate and gibbsite will co-exist at various points within the refinery, the location dependent upon the process used. Refinery experience has shown that the presence of sodium oxalate needles with gibbsite seed can result in a fines "shower", which is variously reported as either a surplus of nuclei or a deficit of agglomeration. Laboratory studies (Power,1990; Power,1991; Reyhani, 1999a) suggest that the presence of organics in the refinery liquor stream can greatly exacerbate this effect.

In the present study, the nucleation and subsequent growth of gibbsite crystals appeared to be favoured at the end faces (002) of oxalate needles, and at various points along the needle axis. We have observed that oxalate grown in caustic solutions undergoes considerable branching, creating steps along the length of the crystal that may result in exposed (002) faces. The empirical observations in these tests suggest that the gibbsite particles preferentially nucleate at these discontinuities. The orientation of the gibbsite crystals at the oxalate surface appears to be random or at least not unique, implying that the growth is not epitaxial. However, if the exposed (002) faces of the oxalate branches are themselves irregular or distorted, epitaxy cannot be ruled out, as the orientation of the gibbsite crystals may simply reflect the disorder of the host oxalate crystal. Moreover, although the overgrowths of gibbsite onto sodium oxalate produced in plant and synthetic liquors appear to be very similar on the microscopic scale, the other organic species in the plant liquor may mediate a different interaction at the molecular level in the two cases. Further studies of these aspects of the interaction of gibbsite with sodium oxalate (Reyhani, 1999b) will help clarify this issue.

The experiments indicate that sodium oxalate acts as a seed surface for gibbsite nucleation both with and without the presence of plant impurities. Indeed, the rate of gibbsite nucleation appears to be comparable in both pure synthetic and plant liquors. This result is in direct contrast to the previous studies (Power,1990; Power,1991; Reed, 1981) and refinery experience, and raises the question of the role that organics apparently play in nucleation and/or agglomeration. If plant organics were directly involved in the nucleation mechanism, it would be reasonable to expect that the nucleation rate of gibbsite on oxalate in plant liquors should be greater than in synthetic liquors, if the latter occurred at all. No such disparity was observed. Certainly, in this study, the presence of organics was seen to influence the morphology of both gibbsite and sodium oxalate. Gibbsite crystals formed in some plant liquors possessed an unusual rounded tabular habit, while those grown in pure system exhibited the expected hexagonal platelet form. Oxalate needles grown in plant liquors were thinner than those grown in pure liquors under comparable conditions (same caustic), and contained more imperfections.

The current work, performed in the absence of seed gibbsite, clearly demonstrates that a third organic component is not necessary for gibbsite to nucleate onto solid sodium oxalate. This is in contrast to previously published work (Power, 1990) in which it was concluded that contamination of oxalate surfaces with organics was a prerequisite. It should be noted, however, that these studies were conducted with both sodium oxalate and seed gibbsite crystals present. The presence of gibbsite may be influential in determining the role of organics. In synthetic systems, it is probable that clean gibbsite crystals offer a more effective surface for gibbsite nucleation than sodium oxalate. In plant liquors, however, organics adsorbed at active sites on the gibbsite crystal may make the nucleation on oxalate more energetically favourable. Consequently, dirty seed may result in a greater degree of nucleation than an equivalent pure system containing equivalent surface areas of oxalate and gibbsite. The role of plant organics will be discussed in greater detail in a forthcoming paper (Reyhani, 1999b).

The observations made in this study permit a mechanism to be proposed for the role of oxalate in fines "showers". In plant operation, if gibbsite contaminated with oxalate enters the agglomerators, part of the supersaturation will be relieved by gibbsite nucleating on the oxalate needles. Particles such as these will not agglomerate effectively. If the gibbsite-laden oxalate crystals are subsequently washed, the gibbsite nuclei will pass into the seed circuit, creating a surplus of very fine seed particles. As these particles are returned to precipitation, they may cause a seed imbalance. Fragile agglomerates may also form. The fine particles themselves will take several cycles to be effectively agglomerated and for the circuit to thus return to normality.

On the other hand, if the gibbsite seeds have been effectively washed (so that they are largely free of adsorbed organics), the presence of a small quantity of sodium oxalate crystals should have relatively little impact on the nucleation and agglomeration behaviour of the circuit. It is possible that this mechanism applies in most alumina refineries to some degree, but agglomeration conditions are tuned to cope with the amount of fines generated by this and other gibbsite nucleation mechanisms. Where nucleation from this mechanism is significant, agglomeration phases will be necessarily aggressive, leading to a precipitation product that will attrite easily, has a wide particle size distribution and higher product soda levels. All of these characteristics are undesirable.

This work raises many questions regarding oxalate, gibbsite, organics and their interactions which are being addressed (Reyhani, 1999a; Reyhani, 1999b). This may lead to a means of improved precipitation control.


G. Lever, " Identification of Organics in Bayer Liquor.", Light Metals, Vol 2, 1978, pp.71- 83

C. Sato and S. Kazama, "Behaviour of Organic Matter in Aluminate Solution", Light Metals, 1971, pp.63-74.

E. Brown and T.J. Cole, "The Behaviour of Sodium Oxalate in Bayer Alumina Plant.", Light Metals 1980 p 102

J.V.Sang, "Factors Affecting Residual Na20 in Precipitation Products.", Light Metals 1988, pp. 147-155.

P.J. The and J.F. Bush, "Solubility of Sodium Oxalate in Bayer Liquor and a Method of Control" Light Metals 1987, pp. 5-10.

R. Calalo and T. Tran, "Effects of Sodium Oxalate on the Precipitation of Alumina Trihydrate from Synthetic Sodium Aluminate Liquors." pp. 125-133, Light Metals 1993

Minai et al. "Sandy Alumina Conversion", Light Metals 1978, pp. 95-110.

S.C. Grocott and S.P. Rosenberg, "Soda in Alumina. Possible Mechanisms for Inclusion.", International Alumina Quality Workshop, Sept. 1988, pp. 271-287.

G.P. Power and W. Tichbon, "Sodium Oxalate in the Bayer Process.. Its Origins and Effects" Proceedings of 2nd International Alumina Quality Workshop, Perth, 1990, pp.99-115.

G.P. Power, "The Impact and Control of Organic Compounds in the Extraction of Alumina from Bauxite.", Extractive Metallurgy Conference, Perth Oct 1991.

C.S. Strom, R. F. P. Grimbergen, I.D.K. Hiralal, B.G. Koenders and P. Bennema, "Mophology of sodium oxalate: Theory and Experiment", J. Cryst. Growth 149 (1995), 107-112.

D.A. Reed and M.M. Olmstead, Acta Cryst., B37 (1981), 938

S.P. Rosenberg and S. J. Healy, Fourth International Alumina Quality Workshop, Darwin, 2-7 June 1996 pp. 301-310

M.M. Reyhani, A. Dwyer, G.M. Parkinson, S.P. Rosenberg, S.J. Healy L. Armstrong, A. Soirat and S. Rowe, submitted to 14th International Symposium on Industrial Crystallisation - Cambridge 12-16 September 1999 (a).

M.M. Reyhani, A. Dwyer, G.M. Parkinson, S.P. Rosenberg, S.J. Healy L. Armstrong, A. Soirat and S. Rowe, submitted to 14th International Symposium on Industrial Crystallisation - Cambridge 12-16 September 1999 (b).