REMOVAL OF IRON-BEARING MINERALS FROM GIBBSITIC BAUXITE BY DIRECT FROTH FLOTATION

The refractory bauxite needs to present less than 2.5% of Fe2O3 to be applied in the ceramics industry. The depletion of high Al2O3 grade deposits has stimulated the improvement of bauxite concentration methods in order to remove iron-bearing minerals. The objective of this study was to evaluate the influence of collector dosage, pH and milling time on the gibbsite flotation performance. Firstly, the sample mineralogical composition was determined by means of X-ray diffraction (XRD) and binocular loupe analysis. X-ray fluorescence (XRF) analysis was used to determine the sample chemical composition. Flotation was then accomplished by using hydroxamate as gibbsite collector, sodium silicate as silicate depressant and starch as iron-bearing minerals depressant. The bauxite Fe2O3 content was reduced from 7.66% to 4.81-5.03%. In addition, the flotation performance decreased by diminishing the pH from 9.5 to 8.5 or increasing the pH to 10.5. The milling time influence on the flotation indicates that the presence of slime can significantly affect the gibbsite concentration.

Brazil is the world's third largest producer of bauxite (12.7%), after Australia (29.9%) and China (18.2%).Most of the Brazilian bauxite beneficiation plants comprise the operations of comminution, classification and solid-liquid separation.Very few beneficiation processes include gravimetric concentration in order to remove silicates or magnetic separation to decrease the amount of iron-bearing minerals [4].In this context, the depletion of high Al 2 O 3 grade deposits has stimulated the improvement of bauxite concentration methods.Flotation has been pointed to as a promising technique to be applied in order to suit the bauxite to the market specifications [1][2][3][4]6,7].
Regardless of the bauxite type, direct and reverse flotation routes have been applied in order to concentrate the aluminum hydroxide (gibbsite or diaspore) mineral.In the reverse flotation route the silicates are floated with cationic collectors at a pH under 6 and gibbsite/diaspore is depressed.Common reagents used in the cationic reverse flotation have been fatty amines as collectors, sodium silicate and starch as dispersant/depressants, and sulfuric acid and sodium carbonate as pH regulators.In the case of diasporic bauxite, hexametaphosphate is also applied as a diaspore depressant [2,3,6,7,10,18].
On the other hand, in direct flotation gibbsite/diaspore is floated with anionic collectors at pH greater than 9, while silicates and iron-bearing minerals are depressed with sodium silicate and starch, respectively [5][6][7].Marino et al. [6] tested the direct and reverse flotation routes in order to concentrate the bauxite from Paragominas-PA, Brazil, which contains 47.7% of Al 2 O 3 and 4.0% of reactive silica, besides kaolinite and iron-bearing minerals.The reverse flotation was accomplished at pH under 6 by using a mixture of fatty acids and amine as kaolinite collector, sodium hexametaphosphate as gibbsite depressant and corn starch as iron-bearing minerals depressant.The gibbsite concentrate comprised 55. 9 Bittencourt [5] were able to concentrate Brazilian refractory gibbsitic bauxite (50% gibbsite, 15% kaolinite and 35% quartz) by direct flotation using a quaternary amine salt as kaolinite and gibbsite collector at pH 6.The gibbsitic concentrate presented Al 2 O 3 content of 97.4% and 90% metallurgical recovery.Barbosa et al. [18] and Barbosa et al. [19] carried out direct flotation experiments in order to remove iron-bearing minerals from the Barro Alto (Goiás state, Brazil) gibbsitic bauxite.S9849 hydroxamate (Cytec) was used as gibbsite collector, starch as iron-bearing minerals depressant and sodium silicate as silicates depressant.The gibbsite Fe 2 O 3 content decreased from 7.57% to 3.44% and the Al 2 O 3 grade increased from 84.55% to 93.80% with an Al 2 O 3 recovery of 45.76%.In addition, the concentrate was submitted to magnetic separation and the non-magnetic product showed 1.91% of Fe 2 O 3 .
The purpose of this study was to investigate the most appropriate conditions (collector dosage, pH and milling time) for the gibbsite concentration by direct flotation with the Barro Alto bauxite.The main objective was to reduce the Fe 2 O 3 content in order to make the bauxite suitable for ceramics production.

Sample Preparation and Characterization
The bauxite sample from Barro Alto, located in the north of Goiás state, Brazil, was crushed (in a Contenco jaw crusher) until 100% of the particles were smaller than 1.70 mm.The crushed product was then homogenized and divided in 1 kg samples.Each of these samples was then milled during 6, 8 or 10 minutes just before the flotation experiments in order to avoid the gibbsite surface oxidation.The samples were milled in rod mill under 50 % of solids.Different milling times were used in order to investigate the particle size influence on the gibbsite flotation performance.For the milling time of 10 minutes 99% of the particles were lesser than 0,119 mm.
The size distribution was determined after each milling time by screening in 0.354, 0.251, 0.178, 0.125, 0.089, 0.066 and 0.044 mm sieves.
The mineralogical composition was determined by means of X-ray diffraction (XRD) and binocular loupe analysis.X-ray fluorescence analysis (XRF) was used to determine the content of silica (SiO 2 ), alumina (Al 2 O 3 ) and iron oxide (Fe 2 O 3 ).The XRD and XRF analysis were carried out in the Curimbaba Company characterization Laboratory.In addition, the sample Liberation Level was evaluated by using the grain-counting technique.

Flotation Experiments
Direct flotation experiments (flotation cell: CFB -1000N, Engendrar) were carried out with a rougher and two cleaner steps in order to investigate the effect of collector dosage (264, 395 and 528 g/t), flotation pH (8.5, 9.5 and 10.0) and milling time (6, 8 and 10 min) on the gibbsite flotation performance.Conditioning was performed under 60% of solids and flotation under 40%.All experiments were conducted with distilled water.
Hydroxamate (AERO 6493, Cytec) was used as gibbsite collector (2 min of conditioning), sodium silicate (Akzo Nobel) as silicates depressant (2 min conditioning) and corn starch as iron-bearing minerals depressant (5 min of conditioning in the rougher step and 2 min in the cleaner steps).The depressant dosages were kept constant along the flotation experiments.It was used 400 g/t in the rougher step and 200 g/t in the cleaner steps for all experiments.The collector dosage was performed only in the rougher step.

Characterization
Table 1 shows the chemical analysis (XRF) for different particle sizes.While the Al 2 O 3 content is higher in the rough fraction, SiO 2 and Fe 2 O 3 contents increased with decreasing particle size.It indicates that the impurities are concentrated in the finer sizes.
The head sample grades are 84% of Al 2 O 3 , 6.6% of SiO 2 and 7.6% of Fe 2 O 3 .While the Al 2 O 3 grade is considered adequate for the ceramics industry, the Fe 2 O 3 is higher than the market specification (2.5%).Accordingly, the bauxite ore must be submitted to concentration operations in order to remove iron-bearing minerals before feeding the ceramics production.
As observed in the photomicrographs of Figure 1, aluminum-bearing minerals are present as: (A) small tabular clay particles with iron hydroxide impregnation/association; (B) Prismatic gibbsite crystals, sometimes containing inclusions or associations of iron oxides and hydroxides such as clays; (C) Slim granular clay aggregates.It can be observed in Figure 1d that lepidocrocite grains are interspersed with goethite grains and limonitic aggregates, causing pigmentation in this mineral.
According to the liberation analysis the gibbsite particles in the size range of -0.105+0.074mm are 85.62% liberated with a Beribé Coefficient of 91.15%.The Beribé coefficient is an average percent of the interest mineral area (aluminum oxides and hydroxides).There is less literature information about the ideal gibbsite liberation for flotation.Marino (2012) flotation work, for instance, was accomplished with a more than 50% liberated gibbsitic bauxite.Therefore, the obtained value of 85,62% of liberation was considered satisfactory.
Finally, the granulometric distribution variation with the milling time is shown in Table 2.It can be observed that the slime amount significantly increases by increasing the milling time.The content of particles smaller than 44 µm increased from 6.1% to 11.8% when the milling time was raised from 6 to 10 min.

Flotation
The direct gibbsite flotation results will be analyzed in terms of the Al 2 O 3 , Fe 2 O 3 and SiO 2 contents (%) in the concentrates, besides the Al 2 O 3 recovery (%).The influence of the collector dosage on the flotation performance is shown in Table 3 and Figure 2. Regardless of the collector dosage, the direct gibbsite flotation resulted in a Fe 2 O 3 reduction from 7.66% to 4.81-5.03%.In addition, the Fe 2 O 3 decrease was followed by the SiO 2 content decrease due to the fact that iron is present mainly as silicates.
Comparing the different hydroxamate dosages, the SiO 2 content presented a reduction from 4.95% to 2.71% when the collector dosage increased from 264 g/t to 528 g/t.Nevertheless, the hydroxamate dosage seems not to have influenced the Fe 2 O 3 grade.Regarding the Al 2 O 3 content, the highest value was obtained with the collector dosage of 395 g/t (Table 3 and Figure 2).Therefore, this dosage was selected to be applied in the experiments conducted under different pH values and milling times.10, its surface presents an excess of negative charges (Ψ<0) at pH 10.5 [23,24].Considering that the polar hydroxamic group also has a negative charge (CONHO -), there can be an electrostatic repulsion between the gibbsite surface and collector molecules.This effect may have contributed to the decrease in the gibbsite flotation performance at pH 10.5.In order to confirm this possibility microflotation experiments with pure gibbsite must be carried out.The third parameter evaluated was the milling time (6, 8, and 10 min), which is directly related to the flotation feed particle size.Table 5 and Figure 4 illustrate that the milling time reduction from 8 to 6 min resulted in lower flotation performance since the Al 2 O 3 content decreased from 91.10% to 88.9% and the Fe 2 O 3 content increased from 5.03% to 6.75%.This outcome can be explained by a possible decrease in the liberation level as the sample amount in the particle size range corresponding to 85.62% of liberation reduces from 85.5% to 67.5%, as observed in the size distribution after 6 min of milling, as shown in Table 2.
Similarly, the milling time increase from 8 to 10 min impaired the gibbsite flotation performance since the The pH decrease from 9.5 to 8.5, as illustrated in Table 4 and Figure 3, resulted in the Fe 2 O 3 content enhancement from 5.03% to 5.38%, followed by a reduction in the Al 2 O 3 content from 91.10% to 89.60% and consequent Al 2 O 3 distribution increase from 24.36% to 25.63%.The flotation performance reduction can be explained by a reduction in the hydroxamate collector activity at pH 8.5.As the hydroxamic acid pKa (solubility product) is approximately 9.0, there is a predominance of protonated species [20] below this pH.Considering that the collector activity of the protonated species (R-CONHOH) is lower than that of the hydroxamic ions (R-CONHO -), the hydroxamic acid concentration increase, instead of the hydroxamic ion, could have reduced the gibbsite collection and consequent flotation.
In addition, an increase in the flotation pH from 9.5 to 10.5 resulted in a significant Al 2 O 3 content reduction (91.10% to 81.00%).Under this condition there was no iron-bearing minerals removal from the bauxite since the SiO 2 and Fe 2 O 3 contents (9.75% and 7.72%, respectively) are higher than those of the feed (6.66% and 7.66%, respectively).The reduced flotation performance observed at pH 10.5 could be related to the contribution of electrostatic forces in the interaction between collector molecules and mineral surface.
The collector adsorption on the solid/liquid mineral interface is affected by parameters such as electrostatic forces and chemical interaction [21,22].Regarding the hydroxamate adsorption on aluminum-bearing minerals, while chemical interaction seems to be the main driving force for the collector interaction with the diaspore surface [12,13], no literature work has reported its interaction on gibbsite particles.On the other hand, the electrostatic contribution to the collector adsorption depends on the mineral zeta potential (Ψ) and the collector charge.As the gibbsite isoelectric point (Ψ = 0) is By increasing the milling time, the fine particles (< 0.066 mm) amount increased from 6.1% to 11.8%, as shown in Table 2.This result indicates that the gibbsite direct flotation could be sensitive to the presence of slime.Many literature works have already reported the harmful effect of fine particles in the reverse diaspore flotation [8][9][10][11]25].

CONCLUSION
It was possible to reduce the bauxite Fe 2 O 3 content from 7.66% to 4.81-5.03%with reverse gibbsite cationic flotation.However the market specification (2.5%) has not yet been reached.In addition, all Al2O3 recoveries are lesser than what is considered satisfactory for an industrial plant.This way it is necessary to improve the concentration strategy by introducing cleaner steps, varying the hydrodynamic conditions and adding other concentration techniques such as magnetic separation.
The most adequate flotation pH for direct gibbsite concentration is 9.5.The flotation performance decreases by decreasing the pH to 8.5 or increasing the pH to 10.5.The first effect could be related to the decrease in the hydroxamic ions availability while the second may be due to the contribution of electrostatic repulsion forces between gibbsite surface and collector molecules.Finally, the investigation of the milling time effect on gibbsite direct flotation indicates that the concentration can be harmed by the presence of slimes.

Figure 2 .
Figure 2. Influence of hydroxamate dosage on the gibbsite flotation performance.

Figure 4 .
Figure 4. Influence of the milling time on the gibbsite flotation performance.
-70.9% of Al 2 O 3 and 7.41-9.83% of Al 2 O 3 /SiO 2 .The direct gibbsite flotation was carried out at pH 10.5 with a mixture of alkyl hydroxamate and fatty acid as gibbsite collector and sodium carbonate plus sodium silicate as iron-bearing minerals and kaolinite depressant.The resulting concentrate showed 53.4-58.3% of Al 2 O 3 and 8.44-10.58% of Al 2 O 3 /SiO 2 .

Table 1 .
XRF Analysis in different size ranges

Table 2 .
Size distribution after different milling times

Table 3 .
Influence of the hydroxamate dosage on the gibbsite flotation performance (pH = 9.5, starch and sodium silicate dosages in the rougher step = 400 g/t, starch and sodium silicate dosages in the cleaner steps = 200 g/t, milling time = 8 min)

Table 5 .
Influence of the milling time on the gibbsite flotation performance (pH = 9.5, hydroxamate dosage = 395 g/t, starch and sodium silicate dosages in the rougher step = 400 g/t, starch and sodium silicate dosages in the cleaner steps = 200 g/t)

Table 4 .
Influence Al 2 O 3 content decreased and the Fe 2 O 3 grade increased.