John Farrow1, Franca Jones2 and Bill van Bronswijk2

AJ Parker Cooperative Research Centre for Hydrometallurgy

    (CSIRO Minerals, PO Box 90, Bentley, Western Australia 6982)

AJ Parker Cooperative Research Centre for Hydrometallurgy

(Curtin University of Technology, GPO Box U1987, Perth, Western Australia 6845)

Corresponding Author:

John Farrow, CSIRO Minerals, PO Box 90, Bentley, Western Australia 6982

Email: John.Farrow@minerals.csiro.au


Despite its importance to the Bayer process, little is known about the mechanism of residue solids flocculation in the primary settlers and how flocculation is influenced by liquor impurities. Towards this end, an investigation has been made of the flocculation of pure hematite by a commercial polyacrylate flocculant in synthetic Bayer liquors.

The effect that the major liquor impurities, namely carbonate, sulphate, chloride, phosphate, silicate and calcium, have on flocculation performance was determined from standard cylinder settling measurements. These results have been correlated to changes in the extent of flocculant adsorption determined from analysis of the residual flocculant concentration in the synthetic Bayer liquor under different conditions.

Contrary to a widespread belief, the polyacrylate flocculant was found to be effective in flocculating the hematite in the absence of calcium. The flocculant adsorption density on the hematite surface remained unchanged when calcium was added. Flocculation performance was unaffected by the presence of sulphate, chloride and phosphate. Both carbonate and silicate had a major influence on flocculation performance, with the flocculant adsorption density on the hematite surface decreasing significantly with increasing concentration of these impurities in the liquor. It appears that although the flocculant has a higher affinity for the hematite surface than carbonate, the rate at which flocculant can displace carbonate pre-adsorbed on the hematite surface limits its performance.


hematite, Bayer, flocculation, impurities, adsorption


John Farrow, Franca Jones and Bill van Bronswijk

1.0 introduction

In the Bayer process, the residue after extraction of the aluminous minerals from bauxite is separated in the primary settlers from the aluminium rich, highly caustic liquor by flocculation at ~100C. As with all gravity settlers, strong interaction between the flocculant and the residue surface is crucial to the efficiency of solid-liquid separation. While the nature of the interaction will clearly depend upon the type of flocculant used and the mineralogy of the residue, it also depends strongly on the liquor composition and the presence of impurities.

Previous investigations of the chemical factors affecting the efficiency of Bayer residue flocculation, and hence settler performance, have used refinery residue suspensions. The variability of these samples in terms of mineralogy, solids concentration, particle size and liquor composition, together with the instability of the liquor, has only allowed general trends to be identified.

Previous researchers have drawn different conclusions as to the effect that liquor impurities have on residue flocculation. Most have focussed on the role of calcium, with general agreement that it is beneficial. However, the mechanism remains a point of conjecture, with suggestions that it can:

i) aid the transformation of goethite to hematite (Yamada et al., 1980; Spitzer et al., 1991; O’Donnell and Martin, 1976);

ii) decrease the solubility of iron in aluminate solutions, promoting precipitation of fresh iron oxide surfaces onto which the flocculant can adsorb (Basu et al., 1986);

iii) remove phosphate as "carbonate apatite" (Sankey and Schwarz, 1984; O’Donnell and Martin, 1976);

iv) form calcium "bridges" for flocculant adsorption (Rothenberg et al., 1989);

v) reduce the level of carbonate in the liquor (Hunter et al., 1990);

vi) precipitate as either tricalcium aluminate, calcium carbonate or calcium hydroxide to form a fresh surface on which flocculant attachs (The and Sivakumar, 1985).

Little attention has been paid to other liquor impurities. Basu et al. (1986) showed that sulphate increased the settling rate. No reasons as to why sulphate was beneficial were given. Grubbs et al. (1980) found phosphate was detrimental to the settling characteristics of residue from goethite rich Jamaican bauxites, and proposed flocculation efficiency was controlled by the rate of phosphate desorption.

Organics are generally considered to have a detrimental effect on residue flocculation, presumably due to a competitive adsorption mechanism. Humics and oxalate have been found to decrease the settling rate of residue solids (Yamada et al., 1980). Sankey and Schwarz (1984) showed that Jamaican bauxites settled much faster in synthetic liquors than in refinery liquors containing a high concentration of soluble organics. In neither case was the mechanism identified.

Our long term aim is to develop a mechanistic understanding of residue flocculation and the effect of impurities. Towards this end we present an investigation of the influence that common inorganic impurities in Bayer liquors have upon the flocculation of hematite in synthetic Bayer liquor. This has been achieved through correlation of the flocculation performance (as assessed by settling rate and residual turbidity) with the flocculant adsorption density under different conditions.

2.0 Experimental

2.1 Materials

The hematite used in this study was a hand picked natural sample from Mt. Newman, Western Australia. It was crushed, ground and stored in deionised water prior to use. The sample had a surface area of 2.6 m2 g-1 and a particle size distribution having d10 = 0.43 m, d50 = 3.8 m and d90 = 12.4 m (Malvern Mastersizer). The XRD pattern fully matched that of hematite, JCPDS-PDF #33,664. Impurities were SiO2 (1.0 wt%) and TiO2 (0.02 wt%). XPS analysis indicated no preferential surface composition.

Aluminium trihydrate ("C31") was from Alcoa Chemicals, Arkansas. All other reagents were analytical grade. The methods used to prepare and analyse the synthetic Bayer liquor samples and the methods used to prepare flocculant solutions (1 mg g-1) have been previously described (Jones et al., 1998a). Flocculant dosages are expressed in terms of
g of neat flocculant per g of dry hematite (i.e. grams per tonne).

The dry powder polyacrylate flocculant (SNF Floerger) had an average molecular weight of 1.4 0.1 107 Daltons, determined by multi-angle laser light scattering (Scott et al., 1996). 13C NMR confirmed it was a 100% polyacrylate flocculant.

2.2 Batch settling procedure

Batch settling tests (Michaels and Bolger, 1962; Farrow and Swift, 1996) were used to assess the properties of flocculated suspensions. Measurements were made at 95C in 250 mL cylinders. A pneumatically driven plunger system provided repeatable mixing during flocculation. Hematite suspensions (60 g L-1) were fully dispersed with ultrasonics prior to flocculant addition. The hindered settling rate was determined from the linear section of the mudline height versus time plot. After 1 h, a syringe was used to take a sample 5 cm below the liquor surface for measurement of the residual turbidity and for determination of the concentration of unadsorbed flocculant (Section 2.3). Turbidity was measured using a Hach Turbidimeter (0-2000 NTU) calibrated regularly against standards.

2.3 Hyamine method for flocculant analysis

Residual flocculant concentrations in the synthetic Bayer liquor were determined by measuring the turbidity resulting from precipitation of the polyacrylate flocculant by Hyamine 1622 (Jones et al., 1998a). This technique gives an excellent linear response between flocculant concentration in the synthetic Bayer liquor and the turbidity of the resultant Hyamine-flocculant precipitate and is very sensitive; flocculant concentrations of 0.5 g g-1 are readily detectable, with good reproducibility (Jones et al., 1998a). The flocculant adsorption density (i.e. the surface excess, g m-2) was calculated from the difference between the flocculant concentration before and after contact with the hematite and the total surface area of the solids.

3.0 Results and discussion

3.1 Flocculation of hematite in synthetic Bayer liquor

We have recently published a detailed account of the flocculation of hematite in synthetic Bayer liquor (Jones et al., 1998a). It was found that a 100% polyacrylate flocculant adsorbs strongly onto the surface of hematite in synthetic Bayer liquor (1.89 M NaOH, 0.294 M Al2O3, 0.094 Na2CO3) in the absence of calcium (<1 g g-1).

Adsorption isotherm measurements showed that the flocculant reached a plateau coverage at ~165 g m-2 (see Figure 1). Optimum performance, as assessed by the maximum hindered settling rate, occurred at an adsorption density of ~35 g m-2. If the plateau adsorption density of ~165 g m-2 is assumed to correspond to "monolayer" coverage, then the maximum settling rate occurred at ~20% of a "monolayer", in contrast to the widely considered concept that optimum performance occurs at 50% coverage. The observation that optimum performance occurred at the lower coverage was attributed to the hematite being coagulated under the high ionic strength conditions of the synthetic Bayer liquor. The observation that hematite is coagulated under these conditions contrasts with the recently reported "unusual colloid stability of gibbsite" in similar liquors (Addai-Mensah et al., 1998).

Figure 1

Adsorption isotherm of polyacrylate flocculant on hematite in synthetic Bayer liquor

The influence of the aluminate ion on hematite flocculation was determined in experiments conducted with liquors having 1.89 M NaOH, 0.094 M Na2CO3 and varying A/TC ratios (0, 0.2, 0.7). The results (Fig. 2) show no discernible trend in the settling rate obtained at different flocculant dosages for the different solutions. This implies that either the aluminate ion does not interact with the hematite surface in the synthetic Bayer liquor or that it interacts so weakly as to not affect flocculant adsorption. In view of these observations, all subsequent measurements were conducted in low A/C liquors (0.3) to avoid complications due to auto-precipitation of gibbsite (which provides an alternative surface for flocculant adsorption).

Figure 2

Sensitivity of hematite flocculation on the A/TC of synthetic Bayer liquor

3.2 Effect of impurities on hematite flocculation in synthetic Bayer liquor

3.2.1 Carbonate

The effect of carbonate on the flocculation of hematite was determined using liquors containing 1.89 M NaOH, 0.29 M Al2O3 and carbonate levels ranging from 2 to 70 g L-1 Na2CO3, with a fixed flocculant dosage of 15 g g-1. This work (discussed more fully in a recently submitted journal paper, Jones et al., 1998b) has shown that carbonate interferes with the adsorption of polyacrylate flocculant onto the hematite surface. This is seen from Figure 3, where the adsorption density of the flocculant falls with increasing carbonate in the liquor, from about ~4.5 g m-2 to ~1.0 g m-2. However, it is notable that the flocculant adsorption density is only affected when the carbonate level is above ~10 g L-1, indicating carbonate does not compete strongly. There was only a relatively small effect on the settling rate, which fell from ~23 m h-1 for liquors containing 2 g L-1 Na2CO3 to 15 m h-1 in liquors containing 70 g L-1 Na2CO3, despite the significant drop in the flocculant adsorption density. The residual turbidity increased from 85 NTU at 2 g L-1 Na2CO3 to 145 NTU at 70 g L-1 Na2CO3, due to the decrease in flocculant adsorption density at the higher carbonate concentration.

Figure 3

Flocculant adsorption density as a function of carbonate concentration (carbonate added before flocculant)

Figure 4

Flocculant adsorption density as a function of carbonate concentration (carbonate added after flocculant)

When flocculation was conducted in the absence of carbonate, and then carbonate added to the liquor, the flocculant adsorption isotherm was unaffected (Fig. 4). This confirms that the flocculant has a greater affinity for the hematite surface than carbonate. The influence of carbonate seen in Figure 3 therefore appears to be a kinetic effect, arising from the slow rate at which the flocculant can displace pre-adsorbed carbonate from the surface. These measurements were made after 1 h; it would be expected that higher adsorption densities would have been observed with longer equilibration times. Measurements of the adsorption density of carbonate on hematite under these conditions would also have been valuable, but this was not done due to the analytical difficulties of accurately measuring small changes in high carbonate concentrations.

3.2.2 Calcium

Calcium chloride was used as the source of calcium, since chloride does not interfere with flocculation (see 3.2.4 below). Pure caustic solutions (1.89 M NaOH) rather than synthetic Bayer liquors were used because C31 gibbsite contains entrained calcium (resulting in up to ~10 g g-1 dissolved Ca2+). As shown in Section 3.1, the A/TC ratio does not affect flocculation performance, and therefore the results with pure caustic solution should be applicable to Bayer systems.

The adsorption isotherm of calcium on hematite in 1.89 M NaOH (Fig. 5) shows Langmuir behaviour reaching a plateau at ~3 g g-1 Ca2+ in solution with a monolayer of ~80 g m-2. This equates to ~2 M calcium per m2 of hematite. If calcium was being precipitated (as the hydroxide) the isotherm would be expected to exhibit a further sharp increase in adsorption density at higher calcium concentrations. Thus, precipitation of a fresh surface, as proposed by The and Sivakumar (1985), can be excluded as a possibility.

Figure 5

Adsorption isotherm of calcium on hematite in 1.89 M NaOH

Flocculation tests were conducted with different calcium dosages at a fixed flocculant dosage of 15 g g-1. The settling rate showed no significant dependence upon the calcium adsorption density (LHS, Fig. 6). Although it may be argued there was a slight increase in settling rate above ~50 m m-2, this was barely beyond experimental error. In contrast, the turbidity shows a significant dependence upon the calcium adsorption density reducing significantly with increasing calcium adsorption (RHS, Fig. 6).

Figure 6

Flocculation performance of hematite in 1.89 M NaOH as a function of the calcium adsorption density on hematite

The adsorption density of the flocculant on the hematite surface as a function of the calcium adsorption density is shown in Figure 7. Notably the flocculant adsorption density is constant at ~3.8 g m-2, unaffected by the presence of calcium which reaches "monolayer" coverage at ~80 g m-2 (Fig. 5) in this 1.89 M NaOH solution. This is a clear indication that calcium does not enhance the adsorption density on hematite of this 100% polyacrylate flocculant. Although the solution used in these experiments was pure caustic (1.89 M NaOH) the adsorption density of the flocculant on hematite under these conditions (~3.8 g m-2) is very similar to that observed in low carbonate synthetic Bayer liquor (~4.5 g m-2, see Fig. 3) at the same flocculant dosage.

Figure 7

Flocculant adsorption density on hematite in 1.89 M NaOH as a function of calcium adsorption density

The results in Figure 6 do not support the literature reports that calcium improves the settling rate (Yamada et al., 1980; Spitzer et al., 1991; Rothenberg et al, 1989; Hunter et al., 1990). However, it cannot be concluded that there is any conflict. All previous data available on the effect of calcium has been obtained for residue solids that are of mixed mineral composition. It is quite possible that calcium does not improve hematite settling, but proves beneficial with other minerals, for example geothite, silicate or a DSP phase. The turbidity data in Figure 6 shows a distinct improvement with increasing flocculant dosage, despite the flocculant adsorption density remaining constant (Fig. 7) under these conditions. A possible explanation is that the turbidity arises mainly from minor impurities within the hematite sample (probably silicates), which are more efficiently flocculated in the presence of calcium. Unfortunately this was not confirmed by mineralogical analysis of the non-settling solids.

3.2.3 Silicate

The adsorption isotherm of silicate on hematite, determined by dosing sodium metasilicate into the synthetic Bayer liquor (1.89 M NaOH, 0.094 M Na2CO3, 0.29 M Al2O3), is given in Figure 8.

Figure 8

Adsorption density of SiO2 on hematite in synthetic Bayer liquor


The results show that the silicate adsorbs on the hematite surface, but unlike calcium does not reach monolayer coverage even at a residual solution concentration (as SiO2) of 600 g g-1. Thus, it appears that silicate adsorbs weakly, but at high density, on the hematite surface. Due to the high aluminium solution concentration it was not possible to confirm whether the adsorbant was a pure silicate (which DRIFT measurements indicated) or an aluminosilicate species (which would involve only very slight loss of aluminium from solution). However, the high adsorption density and the profile of the adsorption isotherm would suggest partial precipitation, rather than pure adsorption, of silicate on the hematite.

Flocculation tests were conducted using different silicate dosages at a fixed flocculant dosage of 15 g g-1. The settling rate decreased significantly and the turbidity increased slightly with increasing silicate adsorption density (Fig. 9). For example, the settling rate decreased from ~38 m h-1 to 12 m h-1 at a SiO2 adsorption density of 1200 g m-2 (corresponding to 600 g g-1 SiO2 in solution, see Fig. 8).

Figure 9

Flocculation performance of hematite in synthetic Bayer liquor as a function of SiO2 adsorption density

Figure 10

Flocculant adsorption density as a function of the SiO2 adsorption density on hematite in synthetic Bayer liquor

The adsorption of the polyacrylate flocculant decreased markedly with increasing silicate adsorption (Fig. 10) from ~5.5 g m-2 to ~2 g m-2 at 1200 g m-2 SiO2 on hematite. This confirms that silicate, like carbonate, restricts flocculant adsorption on the hematite surface. Of significance is that silicate interferes with flocculant adsorption at much lower concentrations than carbonate. For example, carbonate concentrations above ~10 g L-1 are required to have any effect on the flocculant adsorption density (see Fig. 3) whereas silicate solution concentrations above ~200 g g-1 (i.e. ~0.25 g L-1) had a significant effect (see Figs. 8, 9 & 10). This suggests silicate adsorbs more strongly than carbonate on key flocculant adsorption sites on the hematite surface.

The reduction in flocculant adsorption in the presence of silicate could be simply due to a reduction in the number of unoccupied surface adsorption sites available to the flocculant. Alternatively, it may be due to a slow rate at which the flocculant can displace adsorbed silicate from the hematite surface (i.e a kinetic effect as found with carbonate, see Fig. 4). This seems to be the more likely mechanism since flocculant adsorption was not entirely prevented at high silicate adsorption densities (Fig. 10).


3.2.4 Sulphate, chloride and phosphate


The effect of these species on the flocculation of hematite was determined from flocculation performance measurements (settling rate and residual turbidity) and from measurement of the flocculant adsorption density using a fixed flocculant dosage of 15 g m-2. The synthetic liquor comprised 1.89 M NaOH, 0.094 M Na2CO3, 0.29 M Al2O3 plus the inorganic impurity.

Sulphate concentrations up to 50 g L-1 had no effect on the flocculant adsorption density as shown in Figure 11. The surface excess remained constant at ~4.3 g m-2, indicating that sulphate does not inhibit flocculant adsorption. These observations were supported by the flocculation performance measurements that showed both the settling rate (30 m h-1) and residual turbidity (130 NTU) remained unchanged over this range of sulphate concentrations. This is in contrast to Basu et al. (1986) who stated that sulphate was beneficial to the settling of Bayer residue. As hematite was the only mineral phase investigated here, this may indicate sulphate has a beneficial interaction with other mineral phases.

Figure 11

Flocculant adsorption density on hematite as a function of sulphate concentration

Chloride behaved similarly to sulphate with no effect on flocculation performance (settling rate, residual turbidity) or flocculant adsorption density at concentrations up to 40 g L-1.

The presence of phosphate at concentrations up to 1.2 g L-1 also had no effect on hematite flocculation, as indicated by flocculation performance and flocculant adsorption density measurements. The high levels of sulphate and chloride in the synthetic liquors prevented assessment of their adsorption characteristics on the hematite surface (due to the small difference adsorption would have on the solution concentration). However, with the lower phosphate concentrations it was possible to show that phosphate did not adsorb on the hematite surface under these conditions. These observations support O’Donnell and Martin (1976) who stated that phosphate consumed lime, which is beneficial to flocculation performance but otherwise did not affect residue aggregation.

Our results contrast with the observations of Grubbs et al. (1980) who proposed, but gave no supporting experimental evidence, that phosphate inhibits flocculation of goethite rich residue solids by blocking surface adsorption sites. Clearly the influence that phosphate (and other impurities) have on flocculation depends upon the residue’s mineralogy.

4.0 Conclusions


The effectiveness of a polyacrylate flocculant to flocculate hematite in synthetic Bayer liquor can be significantly affected by the presence of some common Bayer liquor impurities.

Calcium does not increase the amount of flocculant adsorbed on hematite in 1.89 M NaOH, and consequently does not increase the settling rate. The residual turbidity decreased significantly with increasing calcium, but this is likely a reflection of the calcium effect on minor impurities within the hematite, not the hematite itself.

Both carbonate and silicate interfere with flocculant adsorption on the hematite surface. Silicate impacts at low levels (~0.25 g L-1) while carbonate only impacts at much higher levels (~10 g L-1). Above these threshold concentrations the adsorption density of the flocculant on the hematite falls in the presence of these species, resulting in slower settling rates and higher residual turbidities. In the case of carbonate (and probably silicate as well) the effect appears to be due to the slow rate at which the flocculant can displace the carbonate from the surface.

Sulphate (up to 50 g L-1), chloride (up to 40 g L-1) and phosphate (up to 1.2 g L-1) did not affect the adsorption of the polyacrylate flocculant on hematite or flocculation performance.



This research has been supported under the Australian Government's Cooperative Research Centre (CRC) Program, through the AJ Parker CRC for Hydrometallurgy. This support is gratefully acknowledged. F. Jones also acknowledges the support of the ARC by way of an Australian Postgraduate Award through Curtin University of Technology.


Addai-Mensah, J., Dawe, J., Hayes, R., Prestidge, C. and Ralston, J. (1998), "The unusual colloidal stability of gibbsite at high pH", J. Colloid Interface Sci., 203, 115-121.

Basu, P., Nitowski G. A. and The, P. J. (1986), "Chemical interactions of iron minerals during Bayer digest and clarification", in: Iron Control in Hydrometallurgy, J.E. Dutrizac and A.J. Monhemius (eds), Ellis Horwood, Chichester, 223-244.

Farrow, J. B. and Swift, J. D. (1996), "A new procedure for assessing the performance of flocculants", Int. J. Min. Process., 46, 263-275.

Grubbs, D. K., Rodenburg, J. K. and Wefers, K. A. (1980), "The geology, mineralogy and clarification properties of red and yellow Jamaican bauxites", Proceedings of Bauxite Symposium IV, 176-186.

Hunter, T. K., Moody, G. M. and Tran, C. A. (1990), "Advances in liquor clarification and mud flocculation in the Bayer process alumina industry", International Alumina Quality Workshop, Perth, Australia, 395-404.

Jones, F., Farrow J. B. and van Bronswijk, W. (1998a), "Flocculation of hematite in synthetic Bayer liquors", Colloids Surfaces A – Physicochem. Eng. Aspects, 135, 183-192.

Jones, F., Farrow J. B. and van Bronswijk, W. (1998b), "Effect of caustic and carbonate concentration on the flocculation of hematite in synthetic Bayer liquors", Colloids Surfaces A – Physicochem. Eng. Aspects, (In Press).

Michaels, A. S. and Bolger, J. C. (1962), "Settling rates and sediment volumes of flocculated kaolin suspensions", I & EC Fundam., 1, 24-33.

O'Donnell, N. B. and Martin, W. (1976), "The commercial processing of goethitic bauxites from Western Jamaica", Light Metals, 135-146.

Rothenberg, A. S., Spitzer, D. P., Lewellyn, M. E. and Heitner, H. I. (1989), "New reagents for alumina processing", Light Metals, 91-96.

Sankey, S. E. and Schwarz, R. J. (1984), "The use of synthetic flocculant polymers in settling red muds derived from high goethitic bauxite ores", Light Metals, 1653-1667.

Scott, J. P., Fawell, P. D., Ralph, D. E. and Farrow, J. B. (1996), "The shear degradation of high-molecular-weight flocculant solutions", J. Appl. Polym. Sci., 62, 2097-2106.

Spitzer, D. P., Rothenberg, A. S., Heitner, H. I., Lewellyn, M. E., Laviolette, L. H., Foster, T. and Avotins, P. V. (1991) "Development of new Bayer process flocculants", Light Metals, 167-171.

The, P. J. and Sivakumar, T. J. (1985), "The effect of impurities on calcium in Bayer liquor", Light Metals, 209-222.

Yamada, K., Harato, T. and Shiozaki, Y. (1980), "Flocculation and sedimentation of red mud", Light Metals, 39-50.