Keywords

Introduction

The Bayer process uses high-pressure digestion conditions to extract alumina minerals from bauxite [1]. During the process, a variety of impurities often dissolve and eventually impact the liquor characteristics, alumina yield, and quality, often making the alumina unsuitable for smelter operations [2]. Traditionally mined Jamaican bauxites are gibbsitic, red in colour, and contain hematite as the main iron mineral. These ores, generally found at low altitudes, contain low concentrations of phosphorus and metal impurities. As supplies of these ores decline, new mining areas with goethitic bauxite having lower available alumina and higher concentrations of chromium, zinc, and phosphorus are being explored [3]. These yellow bauxites are mainly found in hillside locations that are dominated by pocket deposits with highly variable mineral compositions [4, 5].

The local Bayer plants have well-established methods for processing hematite-rich ores containing 0.1–0.4% P2O5. However, as the use of goethitic bauxites becomes imminent, plants require new strategies to manage bauxite blends with higher concentrations of phosphorus (1–2.5% P2O5), zinc (0.02–0.06% ZnO), and chromium (0.06–0.20% Cr2O3). High P2O5 concentrations in bauxite have significant implications for soluble phosphorus control and lime requirements; liquor viscosity and alumina yield and purity may also be impacted [3]. Poorly controlled soluble phosphorus may reduce mud settling rates, lower filtration efficiency, cause scale formation, and increase fines production during precipitation [6].

The main phosphorus mineral in goethitic bauxite is crandallite (CaAl3(PO4)2(OH)5·H2O), however apatite (Ca5(PO4,CO3)3 (F, OH, Cl), wavellite (Al3(PO4)2(OH)3(H2O)5, and variscite (AlPO4·2H2O) may also be present. These minerals dissolve differently in caustic, for example, apatites are sparingly soluble while crandallite dissolves quite readily. Hence, unless the phosphorus mineralogy is known, it is difficult to establish the lime requirements needed for phosphorus control during predesilication and digestion [7, 8]. Managing the metal impurities (Cr2O3, ZnO, and MnO) is also challenging as their low concentrations in the ore make it difficult to identify their source minerals using traditional X-ray diffraction (XRD) techniques. It is also difficult to predict their caustic soluble concentrations and to design protocols for their control [7, 9].

Jamaican bauxites are mainly fine particles but these may aggregate to form larger units. The presence of sand, limestone, and a variety of other minerals often results in crushed bauxites having a fairly wide range of particle size.

This paper compares the characteristics of traditionally mined, red, hematite-rich Jamaican bauxite with samples of yellow, goethitic ores that are slated for mining in the near future. These new bauxites have high phosphate and metal impurity concentrations. Crushed exploration samples from both mines were wet-sieved and fractioned into three typical categories: >75 μm (+200 mesh), 45–75 μm (−200/+325 mesh), and <45 μm (−352 mesh). Selected fractions from each mining area were analysed using XRF to investigate metal oxide concentrations. XRD was used to explore potential differences in mineralogy and following digestion, caustic soluble P2O5, Cr2O3, and ZnO were measured to assess dissolved impurity concentrations in the process liquors. The study seeks to determine whether there are distinct differences in the concentrations of bauxite impurities in the different fractions of bauxite samples taken from the current and future mining areas. Since the processing of goethitic bauxites has been associated with slow mud settling and low alumina productivity, the study will also explore process challenges that might impact use of the new bauxites to produce smelter-grade alumina [3, 10].

Materials and Methods

Sample Selection

Composite exploration samples of traditionally mined bauxites were studied alongside the ore samples that dominate the future mining area. Samples having comparable Al2O3, Fe2O3, SiO2, and TiO2 concentrations were selected from both mines. Samples were selected to span the concentration ranges of P2O5 (0.11–0.95%), ZnO (0.003–0.140%), and Cr2O3 (0.09–0.19%) within the traditional mining area. In selecting samples from the new mining area, the concentrations of impurities within the different ore bodies (P2O5, 0.32–24.6%; ZnO, 0.018–0.308%; and Cr2O3, 0.06–0.37%) were also taken into consideration [11]. The samples selected included processable bauxites and also a few with unusually high or low impurity concentrations. The extreme samples were included to enhance detection of differences in the mineral concentrations of each bauxite fraction.

Six bauxite samples were selected from the traditional mining area and six from the new mines. The metal oxide concentrations of exploration samples from each mining area are shown in Tables 1 and 2. These XRF concentrations were not confirmed in this study and were only used as guides for sample selection. Data in the tables are organized based on the % P2O5 concentrations.

Table 1 XRF data of unsieved exploration samples from the traditional mining area
Full size table
Table 2 XRF data of unsieved exploration samples from the new mining area
Full size table

Bauxite Analysis

Bauxite samples (100 g) were spread out over the surface of a #200 mesh sieve (Fisher Scientific). The mesh and sample were then partially submerged in distilled water and the sieve shaken for four minutes to allow particles capable of exiting the mesh to escape. Particles retained on the sieve were transferred, dried at 100 °C, weighed, and subsequently analysed. The filtrate suspension was further wet-sieved using a #325 mesh and particles retained on this sieve were treated as described above. The fine bauxite particles that exited the #325 sieve were allowed to settle overnight and were then filtered using Whatman #1, #40, and #42 filter papers. The #42 filter paper (2.5 µm pore size) was fitted in the Büchner funnel first, followed by the #40 (8 µm pore size) while the #1 filter paper (24 cm) was cupped within the filter funnel to ensure that all the −325 fraction particles were captured. After filtration, the sample was dried as previously described.

Pressed pellets were produced from each bauxite fraction and elemental concentrations were determined using a PANalytical Magix Pro XRF operating at 40 kV and 60 mA. Mineralogy of selected samples was examined using a Bruker Analytical D5005 X-ray diffractometer operated at 45 kV and 40 mA. Scans were collected between 4 and 75° 2θ [11].

The XRF elemental data and the measured loss of mass (LOM) values were combined with data on peak intensities from the XRD scans using the X-ray database programme (XDB Powder Diffraction Phase Analytical System version 3.107) [12, 13]. The XDB software creates a full-profile fit of the XRD and quantifies the minerals identified based on the XRF and LOM measurements. Minerals that are undetected in the XRD scans are sometimes identified using the XDB software and their concentrations are estimated [14].

The available alumina concentrations were determined on each bauxite fraction using a proprietary FTIR analytical method.

Caustic Soluble Phosphorus

Separate fractions of three of the bauxites samples from the new mining area were digested in 102 g/L NaOH (ACS grade; BDH Chemicals) at 145 °C for 30 min. Slurries were centrifuged, filtered, diluted, and were then acidified to pH 2 with 5 M H2SO4. Soluble phosphorus was measured at 880 nm using the molybdenum blue ascorbic acid procedure [11]. Fractions of currently mined ores were not studied similarly as they generally have low % P2O5 (typically 0.10–0.20) and there are established plant protocols for managing soluble P in these bauxites.

Caustic Soluble Zn and Cr

Bauxite samples were digested, and the slurries cooled, centrifuged, filtered, and then 1.0 mL portions of the clarified liquors were separately combined with 2.5 mL ultra-pure concentrated HNO3 and diluted quantitatively. The dilute samples were analyzed alongside suitable standard solutions via flame atomic absorption spectrophotometry (FAAS) at 231.9 nm for Zn and 357.9 nm for Cr.

Results and Discussion

Bauxite samples from a currently mined hematite-rich area and from a new potential replacement mine were studied to compare their characteristics and to evaluate their low-temperature processabilities. Bauxites were compared in terms of their bulk particle size, elemental concentrations, mineralogy, and the caustic solubilities of key contaminants (P2O5, Cr2O3, and ZnO).

Size Fraction

Table 3 compares the size fractions and % available alumina in the two types of bauxite samples. A reproducibility check on the sieving procedure showed agreement to within ±1.2% while the error associated with the % available alumina was ±1.3%. The samples in the table are arranged based on the P2O5 concentrations in the original exploration samples.

Table 3 Size fraction and % available alumina of bauxite samples from the traditional and new mining areas
Full size table

Four of the six bauxites from the traditional mining area had >60% of bulk particles in the +200 fraction. The remaining two bauxites (DCS10B and DCS14B) were dominated by fines (~50% −325 fraction). For the new mining area, only two bauxites had >50% of particles in the +200 fraction. On average, 45.8% of particles from the new mines were in the −325 fraction in comparison to only 32% of particles from the traditional mines. For both bauxite mines, less than 10% of particles were of intermediate size (−200/+325 fraction).

Data in Table 3 shows that the % available alumina was slightly higher for samples from the traditional mines. Across both mining areas, the larger sized fractions of each bauxite sample (+200 and −200/+325 fractions) had similar % available alumina; these were consistently higher than the alumina content of the −325 bauxite fractions.

XRD Analysis

Figure 1 shows the overlaid diffractograms of the unsieved DCN21A bauxite sample and its +325 and −325 fractions. DCN21A is from the new mining area. Scans of the sieved and unsieved samples are similar which suggests that their mineralogies are also similar. The broad peak at values >15° 2θ concealed the boehmite peaks. This broad peak was characteristic of all the scans recorded on the instrument during the time and resulted from interferences due to the sample holder. Fractions of the traditionally mined bauxites were not examined via XRD.

Fig. 1
figure 1

Overlaid X-ray diffractograms of DCN21A, a bauxite from the new mining area: scans are of unsieved DCN21A and the +325 and −325 fractions

Full size image

XRF Analysis

Traditional Bauxite Mines

The XRF data for bauxites from the traditional mines are in Table 4 and those for the new mining area are in Table 5. As with previous tables, samples are organized based on the XRF concentrations of P2O5 in the original unsieved samples. Data for the traditionally mined ores show that the concentrations of Al2O3, SiO2, CaO, P2O5, and ZnO are roughly similar within the three size fractions for each bauxite (Table 4). Figure 2 shows that the MnO concentrations increase with particle size for all but one sample (DCS11A). Significant differences in MnO concentrations are observed for the fractions of samples DCS21B and DCS14 and it would be worthwhile to probe these fractions to explore whether there is improved opportunity to identify the Mn minerals in each. Similar differentiation was not observed for any other element within the traditional bauxite mines.

Table 4 XRF data for fractions of bauxite samples from the traditional mining area
Full size table
Table 5 XRF data for fractions of bauxite samples from the new mining area
Full size table
Fig. 2
figure 2

% MnO in fractions of bauxite samples from the traditional hematite-rich mining area. (DCS precedes the name of each sample in the chart)

Full size image

The New Bauxite Mines

The % Fe2O3 and % Cr2O3 in goethitic bauxites from the new mining area occur in higher concentrations in the +200 bauxite fractions (Table 5). The % Fe2O3 were lowest in the −325 bauxite fractions except maybe for sample DCN21A. The fine fractions tended to have higher % P2O5, however the trends were not always consistent (Fig. 3). The % P2O5 in fractions of DCN29 from the new mining area (Fig. 3) are well differentiated and are potentially useful for identifying the phosphorus minerals in each. Studies of their caustic soluble concentrations could also be pursued. No defined trends were observed between the % Al2O3, SiO2, CaO, MnO, and ZnO in relation to the particle size of samples from the new mining area.

Fig. 3
figure 3

% P2O5 in fractions of bauxite samples from the new mining area (DCN precedes the name of each sample in the chart)

Full size image

An assessment of the XRF data for bauxite samples from both mining areas shows that some metal species concentrate in the coarse or fine bauxite particles but this is not widespread. It is probable that the characteristics of bauxite fractions of each sample are mainly determined by small constituent bauxite particles that aggregate together. The data so far are not conclusive however they suggest that fractioning bauxite samples followed by use of conventional XRF, XRD or other traditional techniques will not concentrate mineral species enough to provide improved mineralogy information. The approach seems promising for samples with unusually high impurity concentrations such as DCN29 however.

Elemental Correlations

The mineral impurities in bauxites are typically at low concentrations and are difficult to identify using traditional XRD. Simple plots of the elemental concentrations of species in the ore may give clues about their mineral associations [15]. This approach is useful where distinct minerals of the low-concentration species do not exist or where the mineral concentrations are extremely low.

The plots below are for fractions of the bauxite samples from each mining area. Figures 4a, c, and e are for currently mined bauxites while Figs. 4b, d, and f are for the new ores. A plot of % MnO versus % ZnO for bauxite fractions from the traditional mine (Fig. 4a) shows a strong correlation with only one bauxite (DCS16A) deviating significantly from the line. Feret and See identified lithiophorite ((Al,Li)(Mn4+,Mn3+)O2(OH)2) as the main manganese mineral in Jamaican bauxite. They proposed that Zn2+ substitutes for Li+ and results in a strong correlation between the Zn and Mn [16]. They proposed that zinchophorite, a zinc-enriched lithiophorite, is the zinc-contaminating mineral in Jamaican bauxite.

Fig. 4
figure 4

Correlation of selected elemental oxides in bauxites from the traditional (a, c and e) and new goethitic mining areas (b, d and f)

Full size image

Although the number of samples in this study is small, the strong correlation observed between % MnO and % ZnO in the hematitic bauxites suggests occurrence of a single mineral that contains both metals. A plot of % MnO versus % ZnO for the goethitic bauxites suggests that the mineral associations of zinc and manganese are much more complex. Zinc concentrations in these ores are also consistently higher and more than one mineral sources might be present. The higher zinc concentrations and possibly the presence of uncharacteristic zinc-bearing contaminants might have implications for the management of soluble Zn in the new bauxite.

Figure 4c shows a very weak correlation between % P2O5 and % CaO. This contrasts with the fairly strong relationship observed in Fig. 4d which shows the plot for the new bauxite mine. Previous work has identified crandallite as the main P mineral in the new mine [6, 11]. Bauxites from both mines will likely have very different caustic soluble P2O5 concentrations.

Figure 4e plots % Cr2O3 against % Fe2O3 for the currently mined ore while Fig. 4f gives a similar plot for the new mine. Clear differences in the mineral associations are again observed. This suggests need for careful study of caustic soluble chromium in the new bauxites. Based on the strong correlation in Fig. 4f, the chromium impurity is probably associated with the iron mineral in the new mine.

XDB Analysis

The XDB software was applied to fractions of the bauxite samples to predict their mineral constituents by relating their measured XRF and LOM data with the XRD scan obtained in each case. The programme was specially developed for bauxite samples and has an extensive database on the minerals found in bauxites across the world. For a given bauxite, the programme apportions the metal species to their natural host minerals such that they are distributed to match the XRD profile while achieving a balance with the XRF concentrations [14].

Only one bauxite sample was studied via XDB in this preliminary work. Fractions of the unsieved and sieved portions of sample DCN21A (+200, −200/+325, and −325) were examined. The XDB output (Table 6) predicts the distribution of the metal oxides among the various bauxite minerals in the sieved and unsieved fractions of DCN21A. The programme identified gibbsite as the main alumina mineral in all the samples. All samples had additional alumina in boehmite, crandallite, and aluminous goethite; however, nordstrandite was only detected in the −325 fraction. Based on the XDB analysis, crandallite was the only P mineral detected in the unsieved DCN21A. Analysis of the +200 mesh DCN21A fraction identified both crandallite and hydroxyapatite however.

Table 6 XDB predictions of the mineral species in the unsieved and sieved fractions of sample DCN21A (new mining area)
Full size table

While the predictive ability of XDB relies heavily on the quality of the XRD scans and the accuracy of the XRF data, the results obtained for DCN21A demonstrate the potential of using XDB to study fractions of bauxite samples, especially those with unusually high concentrations of mineral impurities. Further work is required to determine whether these approaches can eventually identify the presence of low-concentration mineral species, especially those associated with Cr or Zn.

Caustic Soluble Concentrations

The concentrations of caustic soluble P2O5, Cr2O3, and ZnO were studied for three bauxite samples from the new mining area (Table 7). DCN30A is a typical low P bauxite but its low % CaO suggests presence of a non-crandallite P mineral, possibly variscite (Table 5). DCN21A has excess % CaO beyond that due to crandallite alone; however, for sample DCN16A, the %CaO:%P2O5 ratio suggests that crandallite is likely the only P mineral that is present (Table 5) [11]. The caustic soluble concentrations of bauxite fractions from the currently mined areas were not studied as their soluble concentrations in plant liquors are typically low and are well known. Strategies to manage their build-up in the traditional low-temperature process are well established. On the other hand, the solubility characteristics of P2O5, Cr2O3, and ZnO from the new bauxite mines are unclear and the extent to which they dissolve from each bauxite fraction has not been reported.

Table 7 % caustic soluble P2O5, Cr2O3, and ZnO in different size fractions of bauxite samples from the new mining area
Full size table

Based on data from the three bauxites, there are no major differences in the extraction of phosphorus from the fractions of each ore (Table 7). There were major differences however between each sample. Several factors could be responsible for the differences, for example, the nature of the P mineral (e.g., in DCN30A) or the presence of excess CaO in the bauxite (e.g., in DCN21A), among others [11]. On average, about 70% of the P2O5 that enters with the bauxite is expected to end up in the liquor [11]. The results in this study, however, show a much wider range and illustrate the difficulties that will be encountered in managing process impurities from these ores.

As seen for soluble phosphorus (Table 7), there is no significant differentiation in the soluble Zn concentrations among the different bauxite fractions. Cr2O3 was more extractable from the finer fractions of DCN16A and DCN30A, however there is insufficient data from which to draw conclusions.

Management of Caustic Soluble P2O5, ZnO, and Cr2O3 from the New Mining Area

Of the three impurities studied in this work, the management of P2O5 will likely require the most significant effort and possibly will incur the greatest cost. The solubilities of the P2O5 minerals in this work (52.5–100%) are much higher than those observed for the Zn (1.2–12.1%) and Cr impurities (10.0–38.5%) (see Table 7). Bauxite blending has long been used as an option for managing high concentrations of bauxite impurities, hence this section explores the use of blending to manage the high % P2O5 in the new bauxite. Samples of currently mined bauxites were digested by themselves and in blends with the goethitic ores. The mg/L caustic soluble P2O5 was measured in each case and is reported in Table 8.

Table 8 Caustic soluble P2O5 based on digests of a traditional hematite-rich bauxite, digests of a goethitic bauxite, and digests of blends from both ores
Full size table

When bauxite from the traditional mines (0.22% P2O5) was digested under low-temperature conditions, the average P2O5 concentration was 169 mg/L. In contrast, separate digestions of unsieved ores from the new mines containing 1.0%, 2.0%, and 3.0% P2O5 gave soluble P2O5 concentrations of 981 g/L (75% caustic soluble P2O5), 2287 mg/L (85.5% caustic soluble P2O5), and 3019 mg/L P2O5 (74.9% caustic soluble P2O5), respectively (Table 8). Separate blends comprising 25, 50, and 75% ore from the new mines (1, 2, and 3% P2O5) were combined with the matching proportions of the traditionally processed ore (i.e., 75, 50, and 25% of traditional ore; 0.22% P2O5).

The soluble phosphorus concentrations (mg/L P2O5) of digest liquors from each goethitic bauxite sample and its blends with the low P ore (from current mines) were plotted against the % P2O5 in the bauxite. The plots gave linear relationships with R2 > 0.995 (Table 8). As the total % P2O5 in the blends increased, higher soluble P2O5 concentrations were observed. This means that blends of the new bauxites and the traditional ores can be successfully processed and the resulting soluble P2O5 concentrations can be predicted (Table 8). Lime requirements for phosphorus control can therefore be estimated. Accuracy of the estimates will be impacted by the phosphorus mineralogy of the ores and the calcite concentrations. When goethitic bauxite with 1.0% P2O5 was blended 50/50 with traditional bauxite containing 0.22% P2O5, the resulting soluble P2O5 in the digestion liquor was 615 mg/L. This is almost four times higher than the 165 mg/L P2O5 that the plant would be accustomed to if the traditional ore (0.22 % P2O5) was processed. Much higher soluble P2O5 concentrations occur with bauxite blends that contain higher proportions of the new bauxite.

Blending bauxite from the new mines with low P2O5 ores is the best option currently available for managing the high phosphorus concentrations in the new bauxite mines. Even with blending, higher soluble P2O5 concentrations will occur. This will require more CaO to precipitate the impurity and will result in a higher mud load. New processes for managing soluble P2O5 are therefore required if these high phosphorus bauxites are to be economically processed to yield high-quality smelter-grade alumina [17].

The processing of goethitic bauxites is usually associated with slow mud settling and lower alumina yields. Increased soda and alumina losses, higher energy costs, and lower quality alumina also occur [3, 18]. The processing of ores from the new bauxite mines will likely experience some of these challenges; however, in addition, the management of P2O5, ZnO, and Cr2O3 impurities in the bauxite will further complicate alumina production from these replacement ores [19, 20].

Conclusions

On average, bauxite from the traditional mining areas had larger +200 fractions while the new bauxites had larger −325 fractions. For the traditional mines, concentrations of Al2O3, SiO2, CaO, P2O5, and ZnO were similar for all three fractions of each bauxite; however, the % MnO values were higher in the +200 size fractions. In the new mining areas, there were higher % Fe2O3 and Cr2O3 in the +200 fractions but Al2O3, SiO2, CaO, MnO, and ZnO were similar for the three fractions of each bauxite sample. Nevertheless, bauxites with high mineral concentrations (MnO, Cr2O3, or P2O5) tended to show enrichment in one of the fractions. Study of these fractions using XDB or other sophisticated techniques may show enhanced potential to identify minerals that were not observed during analysis of the bulk, unsieved bauxites.

The total impurity concentrations (% P2O5, % Cr2O3, % ZnO, and % MnO) in the new bauxites are sometimes more than five times higher than in the traditionally processed ores. Processing these ores will therefore require strategies to manage these impurities. There are options to manage soluble zinc impurities but these are traditionally associated with high-temperature bauxite plants. Pilot-scale recirculating experiments are probably needed to assess their relevance for use in low-temperature processing of the new bauxite feed. Managing soluble Cr2O3 may also be challenging and more data are therefore needed to better understand the behaviour of these impurities.

Whether the new bauxites are processed by themselves or in blends with traditional hematitic ores, much higher caustic soluble impurities will result. Managing these will likely impact production costs and also alumina yield and purity. If the process slurries settle slowly, this will further complicate the processing operations.