Characterisation of phosphorus and other impurities in goethite-rich iron ores – Possible P incorporation mechanisms

Minerals Engineering 143 (2019) 106022

Contents lists available at ScienceDirect

Minerals Engineering journal homepage: www.elsevier.com/locate/mineng

Characterisation of phosphorus and other impurities in goethite-rich iron ores – Possible P incorporation mechanisms

T

⁎

Mark I. Powncebya, , Sarath Hapugodab, James Manuelb, Nathan A.S. Webstera, Colin M. MacRaea a b

CSIRO Mineral Resources, Private Bag 10, Clayton South 3169, Victoria, Australia CSIRO Mineral Resources, Kenmore 4069, Queensland, Australia

A R T I C LE I N FO

A B S T R A C T

Keywords: Goethite Iron ore Phosphorus Impurities

Phosphorus is one of the most deleterious elements in iron ore as it follows iron during downstream reduction processes, forming iron phosphides that make steel brittle. Excess phosphorus increases the cost of steelmaking and the steel industry has placed an upper limit of 0.07–0.08 wt-% P on the iron ore feed. Goethite grains containing high levels of phosphorus are abundant in many iron ores and can be difficult to remove without also discarding valuable iron-containing units. The goethite forms during supergene metasomatic enrichment of BIF-derived ores and the phosphorus is typically associated in goethite with other impurity elements such as Si and Al. The current study focusses on determining the distribution and association of phosphorus within goethite present in a high-P Brockman type iron ore from the Pilbara region of Western Australia. Detailed characterisation of the chemistry and mineralogy of the goethite-rich ore was conducted using XRF, optical microscopy and EPMA to determine the distribution of phosphorus and other impurity elements. Using this knowledge, we speculate on the possible P substitution mechanisms in goethite. The latter has important implications in designing strategies for beneficiating high-P goethitic iron ores.

1. Introduction Australia’s iron ore industry is undergoing a major transformation. Reserves of traditional high-grade ores are becoming depleted while the replacement ore types are lower in grade and sometimes require beneficiation to remove or minimise contaminants such as alumina, silica and phosphorus (Dukino et al., 2000). The phosphorus content of Australian iron ores is a particularly serious problem. There are extensive (~7 billion tonnes) amounts of high phosphorus-containing ore in Western Australia deposits that are close to existing infrastructure. These deposits have acceptable iron grade, but the phosphorus levels (> 0.1 wt-% P) incur a price penalty because of the adverse effects of phosphorus on the quality of the end-product steel (Mintz, 1999; Dub et al., 2006) and the costs associated with dephosphorisation during steelmaking. Phosphorus has an embrittling effect on steel, the degree of which can be enhanced depending on the presence of other alloying elements, particularly chromium and manganese (Briant and Mesmer, 1982). Current specifications for phosphorus in iron ore require ores averaging less than 0.07–0.08 wt-% P (Cheng et al., 1999; Thorne et al., 2008), with penalties for every 0.001 wt-% increase in P above the ⁎

acceptable limit (Cheng et al., 1999). Goethite is a naturally occurring iron oxy-hydroxide mineral found in association with several types of mineral deposits including iron, manganese and bauxite ores. In iron ore, goethite forms during supergene metasomatic replacement of fine-grained chert and other gangue minerals (Morris, 1985; Dukino et al., 2000) with the formation initially of ferrihydrite, FeO[OH]3·nH2O, which is subsequently dehydrated and recrystallised to form goethite (Dukino et al., 2000). It contains variable quantities of oxide impurities such as P, Al, Mn, Si and Ca, which can reach a collective level of 5 wt-% (Klein and Hurlbut, 1985). An association of phosphorus with goethite was demonstrated by Graham (1973) using electron probe microanalysis techniques and by Ward et al. (1975), who observed that phosphorus contents were lower in drill cuttings with lower goethite content. At present there is an incomplete understanding of the mechanism(s) for incorporation of phosphorus and other impurities within iron ores; however, such an understanding may be important in identifying the best technique for their removal. To better understand the occurrence of phosphorus and associated impurities within goethite-rich iron ores, a detailed characterisation

Corresponding author. E-mail address: [email protected] (M.I. Pownceby).

https://doi.org/10.1016/j.mineng.2019.106022 Received 29 August 2019; Accepted 9 September 2019 Available online 16 September 2019 0892-6875/ Crown Copyright © 2019 Published by Elsevier Ltd. All rights reserved.

Minerals Engineering 143 (2019) 106022

M.I. Pownceby, et al.

Table 1 List of the most common iron-bearing minerals in Australian iron ore deposits. Ore Mineral*

Chemical formula

Description

Hematite Magnetite Goethite

Fe2O3 Fe3O4 FeOOH

‘Kenomagnetite’ ‘Martite’ ‘Hydrohematite’

Fe2−x()xO4 Fe2O3 Fe(2−x)/3 (OH)xO3−x

Iron oxide with iron fully oxidised to Fe3+ Primary iron oxide with iron present in Fe2+ and Fe3+ oxidation states Most abundant iron oxyhydroxide with 3 subtypes: ochreous (yellow) is highly microporous, brown (light to dark brown) is dense and typically close to stoichiometric composition, whilst vitreous (dark red-brown to black) is glassy and contains SiO2 and Al2O3 as impurities and/or localised quartz/aluminosilicate submicron inclusions. Intermediate metal-deficient phase between magnetite and maghemite Hematite which has replaced primary magnetite A defect solid solution where OH− ions replace oxygen atoms

* Mineral names parenthesised are not officially recognised but are commonly used in the industry.

generating fines or ultrafines due to its friable nature. It is microporous (with a very high moisture saturation point) and has low specific gravity and high loss on ignition (LOI). It is the second most abundant type of goethite and increases with depth over hard brown goethite. Ochreous goethite also displays variably amorphous to uniform, fibrous textures and contains pseudomorphs after silicates/carbonates. Vitreous goethite is macroscopically dark brown, lustrous, has a conchoidal fracture and can be found as hard lump to coarse fines. It is hard, has low porosity, intermediate specific gravity and intermediate LOI. Vitreous goethite is the most common form of goethite in the hardcap zone, but overall it accounts for typically less than 2% of the total goethite in any iron ore product. Earthy goethite represents an intermediate, with mixtures of discrete vitreous/brown/ochreous components and intermediate porosity/physical properties. Previous work suggests that phosphorus is mainly located in ultrafine goethite crystallites, being especially high in secondary or late stage goethite that had not undergone any recrystallisation (Dukino et al., 2000). Ochreous goethite was noted to be typically associated with high levels of other impurity elements such as silicon and aluminium (Dukino et al., 2000). More recent work by Manuel and Clout (2017) concluded that phosphorus is consistently higher (mean of 0.112 wt-% P and up to 0.296 wt-% P) in ochreous goethite compared to brown goethite (mean of 0.011 wt-% P) or lower grade vitreous goethite (mean of 0.031 wt-% P). The highly porous (up to 75% total porosity) microstructure of ochreous goethite is thought to have a much higher surface area, which may aid in preferential absorption of phosphorus onto the ochreous goethite mineral surface.

study was undertaken on a high-P Brockman iron ore from the Pilbara region of Western Australia. This paper is divided into a number of parts; (a) a brief review of phosphorus deportment within iron ores, (b) characterisation of high-P iron ores to determine the association between goethite, phosphorus and other impurity elements using combined XRF, optical microscopy, and EPMA techniques and (c) the development of a plausible mechanism for phosphorus incorporation in goethite. The implications of the findings for phosphorus removal from high-P iron ores are also discussed.

2. Phosphorus in iron ore deposits 2.1. The goethite-phosphorus association A list of the iron ore minerals commonly observed in Australian iron ore deposits is provided in Table 1. In the Pilbara region of northwest Western Australia, the high-grade (i.e., > 62 wt-% Fe) bedded iron deposits (BID) are characterised by martite-microplaty hematite (M-mpl H) and/or martite-goethite (M-G) ore types (Morris, 1980, 1985; Thorne et al., 2008; Ramanaidou and Wells, 2014). In the process of enrichment to high-grade mineralisation, primary apatite in the original banded iron formation decomposes under supergene conditions at slightly acid pH and is remobilised (Morris, 1985; Clout, 2005). Part of the leached phosphorus precipitates in pore spaces as secondary phosphates such as hydroxyapatite, Ca5[PO4]3(OH) (Morris, 1973), vivianite, Fe32+[PO4]2·8H2O and wavellite, Al3[PO4]2(OH,F)3 (Ostwald, 1981), resulting in variable and locally very high phosphorus contents of up to 6.75 wt-% P (Morris, 2002). While these secondary minerals represent potential sources for phosphorus in iron ore, the majority of phosphorus in iron ore, where the average levels are of the order of 0.1 wt-% or more, occurs not as discrete mineral phases, but in association with goethite (Graham, 1973; Dukino and England, 1997; Dukino et al., 2000). In contrast, levels of phosphorus associated with hematite are much lower, rarely > 0.1 wt-% P and typically < 0.05 wt% P (Morris, 1985; Dukino et al., 2000). For example, high-grade iron ore mineralisation of the style typified by the Mount Whaleback and Mount Tom Price deposits (i.e., M-mpl H), have on average a low phosphorus content of < 0.06–0.07 wt-% P (Morris, 1985; Dukino et al., 2000; Thorne et al., 2008). Supergene M-G ores, such as those hosted by the Brockman Iron Formation have a high phosphorus content > 0.07 wt-% P, and typically > 0.1 wt-% P (Morris, 1985; Thorne et al., 2008). Studies of different iron ore types from the Pilbara region have shown it is possible to categorise the contained goethite into a number of morphological and textural/microstructural types (Table 1). These include: brown goethite, vitreous goethite and yellow ochreous goethite, with brown goethite being overall the most abundant type. Brown goethite typically surrounds martite in the upper hydrated zone of supergene M-G deposits and is interpreted to have replaced primary yellow ochreous goethite as the paleo-water table fell during progressive metasomatic alteration. Ochreous goethite is macroscopically yellow with a dull/chalky appearance and easily breaks down

2.2. P in goethite incorporation mechanisms On the basis of electron microprobe and transmission electron microscope (TEM) examination of a range of M-G ore types Dukino et al. (2000) demonstrated a preferred P-goethite association. These findings were consistent with earlier work, where a direct P-goethite association with goethite of up to 1 wt-% P was shown (e.g., Graham, 1973; Morris, 1985). Graham (1973) favoured the view that elemental phosphorus is in solid solution with goethite. Significant interstitial solid solution however is likely only if the substituting cation is of a similar or smaller size than the lattice bound cation (in this case, Fe3+) and its charge is the same. On the basis of ionic size considerations, the pentavalent P5+ cation (0.38 Å) is considerably smaller than the Fe3+ ionic radius (Shannon, 1976) and would be expected to easily replace the iron in the goethite structure. However, size considerations aside, the formal valence state of phosphorus is 5+ whereas that of iron in goethite is 3+ Based on these facts, a direct solid solution between P5+ and Fe3+ in goethite is therefore unlikely. Fixing, or surface adsorption, of phosphorus to hydrated iron oxides (e.g. ferrihydrite) is extensively documented (e.g. see the summary in Cornell and Schwertmann, 1996). These observations, and those reported earlier in the literature have been interpreted as indicating that phosphorus occurs as an adsorbed complex at the goethite surface. 2

Minerals Engineering 143 (2019) 106022

M.I. Pownceby, et al.

Phosphorus fixing occurs by the specific replacement of surface hydroxyl groups by an adsorbing ligand (anion), thereby involving direct coordination to a surface Fe atom. Surface complexation onto nucleating sites of gel-like precursors enables phosphorus to be ultimately incorporated within the growing goethite crystal (e.g., Ostwald, 1981; Dukino et al., 2000; Thorne et al., 2008). However, in these studies no direct evidence for the replacement of iron by phosphorus within goethite has been demonstrated. Intra-crystal phosphorus such as proposed by this mechanism would appear to be in solid solution under electron probe microanalysis, accounting for the observations made by Graham (1973).

Queensland, using a JEOL Super Probe (JEOL-JXA 8200) equipped with wavelength dispersive (WD) spectrometers. The microprobe was operated with an accelerating voltage of 15 kV, a beam current of 15 nA, and a spot size of 1–5 µm. Counting times were 30 s. for major elements and 40 s. for trace elements. The instrument was calibrated with mineral and oxide standards. Bulk Ores 1 and 2: Approximately 200 (Bulk Ore 1) and 500 (Bulk Ore 2) randomly targeted goethite grains were analysed quantitatively using a JEOL JXA-8500F field emission gun electron probe microanalyser (FEG-EPMA) at CSIRO. No distinction was made between goethite textural types in the two samples. Quantitative analyses were performed in WD mode at an accelerating voltage of 15 kV and a beam current of 20 nA. The electron beam was defocused to 2 μm and the following suite of elements measured: Na, P, Fe, Ca, Mg, Al, Mn and Si. Oxygen was calculated by difference, based on valence. The following standards were used: spinel (“Magalox”, MgAl2O4), hematite (Fe2O3), albite (NaAlSi3O8), pyroxmangite (MnSiO3), apatite (Ca5[PO4]3(OH,F)), and wollastonite (CaSiO3).

2.3. Other impurities in goethite Since goethite has the same structural topology as several other oxides and oxyhydroxide minerals, it forms limited to extensive solid solutions with many of these (Waychunas, 1991). For example, there is considerable solubility towards diaspore, the direct aluminium analogue of goethite, and up to a third of the iron in goethite may be substituted by aluminium (Cornell and Schwertmann, 1996), with aluminium substitution of as much as 33 wt-% reported by Fysh and Clark (1982), resulting in changes in unit cell dimensions and inferred structural defects (Schulze, 1984; Schulze and Schwertmann, 1984). Silicon (as Si4+) also appears to substitute into goethite but the relationship is less well characterised (Waychunas, 1991). In contrast to phosphorus, XRF and electron probe microanalysis results indicate vitreous goethite to be significantly higher in silicon (up to 18.7 wt-% Si), aluminium (up to 3.5 wt-% Al) and sulfur (up to 0.14 wt-% S) compared to the generally very low (< 1 wt-% for Si and Al) levels of these elements in ochreous goethite and brown goethite. The data dispels the common misconception that friable earthy or yellow ochreous goethite is high in aluminium and silicon and other chemical impurities (Manuel and Clout, 2017). 3. Experimental

3.2.3. High resolution EPMA mapping and quantitative analysis To identify high-P goethite grains, a 2 × 2 mm region was mapped using an accelerating voltage of 15 kV, a beam current of 100 nA and a step size (in x and y) of 2 µm. The coordinates of all high-P grains were recorded for subsequent high resolution mapping and microanalysis and high resolution maps were then collected using an accelerating voltage of 7 kV, a beam current of 70 nA, and step sizes of 100 nm. In addition to phosphorus, the distributions of Al, Si, Fe and Ca were also mapped. Once high-P goethite grains were identified and mapped, quantitative microanalyses (linescans) were performed using an accelerating voltage of 7 kV and a beam current of 13 nA, with P, Al, Si, Fe being measured by WD spectrometry and oxygen calculated by stoichiometry. Standards used were; berlinite (AlPO4) for phosphorus, spinel (MgAl2O4) for aluminium, wollastonite (CaSiO3) for silicon, and hematite (Fe2O3) for iron.

3.1. Samples

4. Results

To examine the distribution of phosphorus, aluminium and silicon between the various iron ore phases, a range of Brockman high-P type ores were examined using Quantitative Electron Probe Microanalysis (EPMA). Following this initial investigation, two representative samples of Brockman P-rich iron ore (labelled as Bulk Ore 1 and Bulk Ore 2) were selected for further detailed analysis in an attempt to understand the association, if any, between the impurity elements in goethite.

4.1. Impurity variation with mineralogy in high-P Brockman iron ores All Brockman high-P iron ores were dense to moderately microporous and hematitic. Dense and/or microporous hematite/martite, microporous microplaty hematite and dense hematite/martite with goethite infill were the most abundant particle types. Some dense to microporous goethite, moderately porous earthy goethite and moderately porous martite was also present. Representative images are shown in Fig. 1. Phosphorus was mainly located within the various goethite types (Table 2) with hydrohematite and clays containing minor amounts. Rare apatite associated with microplaty hematite and complex REE minerals as inclusions within goethite were also found. Kenomagnetite, martite and microplaty hematite were essentially phosphorus free. Analysis of the various types of fully liberated and/or combined ore particles consisting of hematite/martite and goethite (dense, earthy and ochreous) indicated that phosphorus can be located in any type of goethite. The highest levels of phosphorus were found in brown/vitreous goethite particles although even in these particles the phosphorus content was variable e.g. some brown goethite particles contained 1.6 wt-% P while some from the same sample contained less than 0.2 wt-% P or were virtually phosphorus-free. This may be due to the fact that the high-P goethite particles were closer to the phosphorus releasing source (original apatite or REE minerals) and the low-P goethite particles are from regions more distant from the phosphorus source. The phosphorus and iron contents of individual iron-bearing mineral phases in the high-P Brockman ores is compared in Fig. 2. The

3.2. Sample preparation 3.2.1. X-Ray fluorescence spectroscopy (XRF) Chemical analysis of the two high-P bulk samples was carried out using XRF. This involved accurately weighing approximately 0.4 g of each of the finely ground, oven dried powders into a 95% Pt/Au crucible with approximately 4 g (also accurately weighed) of 12:22 lithium tetraborate: metaborate flux. The mixture was then fused into a homogeneous glass disc over an oxy-propane flame at a temperature of approximately 1050 °C and the molten material was poured into a 32 mm diameter 95% Pt/Au mould pre-heated to a similar temperature. The melt was then cooled by air jets for approximately 60 s. The resulting glass discs were analysed on a Philips PW2404 XRF system using an ilmenite-specific control program developed by Philips and algorithms developed in-house by CSIRO. 3.2.2. Quantitative EPMA Brockman high-P ores: For the initial analyses examining a range of Brockman high-P iron ores, EPMA analyses were carried out at the Centre for Microscopy and Microanalysis (CMM) at the University of 3

Minerals Engineering 143 (2019) 106022

M.I. Pownceby, et al.

H 250 μm

P

P

Dense hematite

Porous martite

Microplaty hematite

H

Porous microplaty hematite

P

H

P

H M

Mt

G

Fine microplaty hematite

Hydrohematite

M

G

M

Martite-kenomagnetite-goethite

Martite-goethite

G M

M

H G

oG

G P

M Dense brown goethite-martite

Microplaty hematite-goethite

Brown goethite-martite

M Ochreous goethite-martite

P

vG

P oG Dense brown goethite

Earthy goethite

Ochreous goethite

Vitreous-ochreous goethite

Fig. 1. Main iron ore particle types identified in Brockman high-P iron ores (plane-polarised, reflected light phtomicrographs, all at the same magnification; see 250 µm scale bar at top left). H = hematite (microplaty), M = martite, Mt = kenomagnetite, G = brown goethite, oG = ochreous goethite, vG = vitreous goethite, P = pore.

data confirm that phosphorus was higher in the goethite phases with the brown/vitreous goethite being particularly P-rich, with contents from 0.01 to 1.5 wt-% P (average of 0.322 wt-% P). The distribution of phosphorous contents within the earthy and ochreous goethite types was lower.

Table 2 EPMA data showing phosphorus contents in the various mineral phases present in Brockman high-P iron ores. Besides P-rich minerals such as apatite and REEbearing phosphate minerals, P is concentrated primarily in the different forms of goethite. Mineral phase

Earthy goethite Brown/vitreous goethite Ochreous goethite Microplaty hematite Martite Hydrohematite Kenomagnetite REE minerals Apatite Kaolinite Gibbsite Quartz

Phosphorus (P %)

No. of analytical points

Min.

Max.

Mean

0.004 0.010 0.000 0.000 0.000 0.003 0.000 6.20 16.50 0.003 0.000 0.000

1.44 1.60 0.920 0.083 0.371 0.360 0.050 12.76 18.62 0.261 0.351 0.000

0.214 0.322 0.186 0.011 0.047 0.098 0.020 9.50 17.20 0.097 0.068 0.000

4.2. Quantitative analysis of Bulk Ores 1 and 2 Results from quantitative FEG-EPMA analyses of goethite grains in Bulk Ores 1 and 2 are summarised in Table 3, which also includes data from bulk XRF analyses for comparison. For Bulk Ore 1, the results indicate that the Al and Si impurities were present in goethite at significantly lower levels than measured in the bulk sample. This suggests an additional source(s) for Al and Si impurities in the ore which, confirmed from EPMA analyses, include aluminosilicates (kaolinitic shale), quartz and trace gibbsite. In comparison, goethite appears to be the main repository for phosphorus, containing just over double the amount of P than measured for the bulk sample. For Bulk Ore 2, Al and Si impurities within the bulk and within the goethite were almost identical implying low levels of gangue accessory phases in the ore. As for Bulk Ore 1, goethite was confirmed as the main P-bearing phase, containing over three times the amount of phosphorus measured in the

465 996 337 153 72 97 18 22 20 31 14 8

4

Minerals Engineering 143 (2019) 106022

M.I. Pownceby, et al. 1.6

1.4

Earthy Goethite

Hydrohematite

Martite

Microplaty Hematite

Ochreous Goethite

Brown/Vitreous Goethite

1.2

P (wt-%)

1.0

0.8

0.6

0.4

0.2

0.0 30

35

40

45

50

55

60

65

70

Fe (wt-%) Fig. 2. Phosphorus versus iron content in the various mineral phases in Brockman high-P iron ores.

appears to be associated with low phosphorus; and a lower layer which shows significantly higher porosity and more phosphorus. A high resolution map collected over this grain, shown in Fig. 5b, indicates a denser, phosphorus-enriched zone along the top of the grain possibly indicating late stage precipitation. This appears to be secondary, precipitated goethite, with colloform texture. In comparison, the more porous lower parts of the grain have significantly less phosphorus but contain elevated Si levels. This low phosphorus porous region is interspersed with high Al and P inclusions (yellow grains in Fig. 5a), possibly filling voids within the original structure. In order to better understand the mechanism of phosphorus and other impurity element incorporation within the goethite, a quantitative microanalysis line traverse was taken to show the variation in Al and Si levels with phosphorus levels. The elemental and spatial relationships between Al vs P and Si vs P can be seen in the accompanying elemental scatter plots in Fig. 6. The results show that low-P regions are associated with low-Al and high-Si primary goethite, whereas the highP regions are associated with high-Al and low-Si secondary goethite.

bulk sample. The quantitative EPMA data are shown in Fig. 3 as a series of x-y scatter plots to examine possible trends between Al and P, Si and P, and Al and Si association. For both ores there appears to be a trend of increasing P with increasing Al. This trend is more obvious in the data for Bulk Ore 2. Close inspection of the data, however, suggests the possibility of two distinct Al/P trends with one group characterised by much higher Al and lower P contents (typically < 0.5 wt-% P) and the other group having high P and lower Al (typically < 2.0 wt-% P). Further work is required to verify any association between Al and P although we note that previous workers (e.g. Mohapatra et al., 2008) have also observed a relationship between Al and P in goethite. Scatter plots showing Si versus P reveal that for both ores, high P levels correspond to low Si levels and vice versa. Plots for Al versus Si do not indicate any clear correlation. 4.3. FEG-EPMA mapping and high resolution quantitative analyses of Bulk Ores 1 and 2

4.3.2. Bulk Ore 2 A second high-P goethite grain was examined from Bulk Ore 2. In this example, an area where a Fe-oxide (hematite) grain was surrounded by goethite was targeted. The resultant high resolution element distribution maps are shown in Fig. 7. The elemental maps show at least two different impurity element associations within the goethite. The first causes phosphorus (pink) enrichment along the hematitegoethite grain boundary-interface, while the second involves Al incorporation (green-yellow) within the goethitic region. Inclusions of an

4.3.1. Bulk Ore 1 A large area, low-resolution map from Bulk Ore 1 shows a wide range of grain sizes and textures (Fig. 4). The Fe, Al, and P elemental map shows that phosphorus exists as discrete P-rich grains which are either apatite or rare earth (REE) containing phosphates. The region highlighted at the top left of Fig. 4 shows phosphorus enrichment around the edges of a goethite grain. A magnified, backscattered electron (BSE) image of this grain is provided in Fig. 5a, which shows two distinct morphologies within the grain: a dense upper layer which

Table 3 Summary table showing average EPMA data from all goethite types, excluding oxygen, as well as XRF results for the head samples for the two bulk ores examined in this study. Sample

Bulk Ore 1 XRF (Bulk) Bulk Ore 2 XRF (Bulk) † ‡

Element (wt. %)

# analyses

Na

P

Fe

Ca

Mg

Al

Mn

Si

Total

0.043 n.d.† 0.011 n.d.

0.298 0.146 0.340 0.100

56.69 62.1 57.16 63.6

0.022 0.021 0.006 b.d.

0.051 b.d.‡ 0.074 b.d.

0.701 1.20 0.909 0.951

0.061 0.068 0.226 0.081

0.903 1.47 1.31 1.23

85.22

261

87.45

532

n.d. = not determined. b.d. = below detection limit. 5

Minerals Engineering 143 (2019) 106022

M.I. Pownceby, et al.

Bulk Ore 2 5.0

4.0

4.0

Al (wt.%)

Al (wt.%)

Bulk Ore 1 5.0

3.0 2.0 1.0

3.0 2.0 1.0

0.0

0.0

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0.0

0.5

1.0

5.0

4.0

4.0

3.0 2.0 1.0

2.0

1.5

2.0

3.0 2.0 1.0

0.0

0.0 0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

0.0

0.5

1.0

P (wt.% )

P (wt.% )

5.0

5.0

4.0

4.0

Al (wt.%)

Al (wt.%)

1.5

P (wt.%)

5.0

Si (wt.%)

Si (wt.%)

P (wt.% )

3.0 2.0 1.0

3.0 2.0 1.0

0.0

0.0 0.0

1.0

2.0

3.0

4.0

5.0

0.0

1.0

Si (wt.% )

2.0

3.0

4.0

5.0

Si (wt.% )

Fig. 3. Quantitative EPMA scatter plot data for the elements P, Al and Si in goethite for the two Brockman high-P type iron ore samples.

oxide” we note that compositionally it is a form of goethite. The reduction of Al + P with increasing Si was confirmed by analyses extended into the surrounding goethite phase. In this region, high levels of Si (between 0.8 and 1.8 wt-% Si) appear to prohibit Al + P substitution beyond a certain value (~0.3–0.4 wt-% Al + P combined). Similar to the results from Bulk Ore 1, at low Si contents, Al + P are able to be incorporated in goethite, however, beyond a certain Si substitution level (typically > 1.0 wt-% Si) the goethite cannot accommodate any more Al or P within its structure.

Al-rich phase present in the goethite and the Al levels is highest around the rims of the inclusions and lower (although still elevated) inside, also with minor Si content. This suggests leaching of Al from the grains, with concentration around the edges at the diffusion interface with goethite i.e. solid state diffusion of Al into surrounding goethite is the ratelimiting step and the inclusions were already present when the goethite was precipitated. Larger micropores are also present in goethite and these are un-filled which also suggests that Al was leaching out rather than inwards. The phosphorus appears to have penetrated the hematite, with high phosphorus levels being involved in the initial incorporation/ conversion of the Fe-oxide into goethite followed by the two distinct high-P and high-Al forms of goethite. The Al element distribution map shown in Fig. 7 indicates Al is also associated with P incorporation in the P-rich goethite, however, it appears to be at significantly lower levels than in the Al-rich form of goethite (which has a patchy distribution). A scatter plot of Al + P versus Si analyses from a line traversing from the Fe-oxide through the phosphorus-rich interface and into the goethite region shows three distinct associations, one for each of; goethite, Fe-oxide and replaced Fe-oxide (Fig. 8). The Fe-oxide does not favour incorporation of Al and/or P at all, however, Si is easily accommodated at low levels (up to 1.2 wt-% Si). In comparison, within the replaced Fe-oxide there is a clear association between Al + P and Si, with results suggesting that as the Si content of the replaced Fe-oxide increases, the amount of Al and P able to be incorporated into the structure is reduced. Although we have termed this phase ‘replaced Fe-

4.4. A plausible P-substitution mechanism in goethite? The quantitative EPMA data indicate a strong positive correlation between Al and P (Fig. 6a) and an equally strong negative correlation between P and Si (Fig. 6b) and Si and Al (Fig. 6c). The latter two translate to high total Al + P contents at low Si and vice versa (Fig. 6d). We believe the EPMA results support a phosphorus incorporation mechanism for goethite that suggests a coupled substitution reaction mechanism according to:

2Si4 + = 1P5 + + 1Al3 +

(1)

A similar mechanism has previously been observed in the mineral berlinite (Thomas and Webster, 1999). In this reaction mechanism, the presence of the trivalent Al and quadrivalent Si are both vital in stabilising P within goethite. This may be the reason that previous attempts to produce P-substituted goethites have failed – the presence of Al and Si are both vital. Reaction (1) also explains the positive 6

Minerals Engineering 143 (2019) 106022

M.I. Pownceby, et al.

Fe Al P

Fig. 4. Low resolution Fe/Al/P distribution FEG-EPMA map for Bulk Ore 1. The map reveals a number of very high phosphorus grains (yellow grains circled), which are apatite or REE phosphates, and a separate goethite grain (highlighted in the rectangular region) which appears to have elevated phosphorus levels. Green particles are kaolinite. Figure modified from MacRae et al. (2011). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

correlation between Al and P in goethite and also the observation that P-containing goethites tend to have lower levels of Si. Quantum mechanical modelling (Wilson, 2013) examined potential sites for incorporation of phosphorus into the goethite by placing a single phosphorus atom at different sites within a goethite supercell. The most stable configuration was found when phosphorus was coupled with an iron vacancy. Coupled aluminium and phosphorus incorporation was also investigated with the most stable configuration found when aluminium substituted for iron next to the phosphorus-vacancy defect. This preliminary result appears to support the coupled substitution model suggested above. To further check the possibility of the coupled substitution, in Fig. 9 we have plotted all the EPMA data from the traverse in Fig. 5b in a plot of Al + P versus Si, and Al + S i versus 2Si for the high-P data only. For the regions where high P and high Al levels were measured, the slope of the graph is approximately 0.5 when Al3+ + P5+ are plotted against Si4+ (and 1.0 when plotted against 2Si4+, as shown in the insert) which supports the mechanism outlined in reaction (1). The results in Fig. 9 also confirm that the coupled substitution is only valid for low-Si containing goethite. As the silicon concentration in the goethite increases, the total Al + P decreases until the silicon level in the goethite reaches greater than about 1.0 wt-%. Thereafter, both Al and P remain constant at ~0.4 wt-% (combined). This suggests an inhibiting effect of silicon on Al and P incorporation. Such an effect has previously been demonstrated by Swedlund et al. (2010) who suggested that siliconbased monomeric and trimeric ligands produce bonding Si/Fe complexes at the surface ferrihydrite. While the FEG-EPMA analyses offer preliminary conclusions to be reached regarding the possible mechanism(s) of phosphorus incorporation in goethite, the results are not conclusive as the chemical correlations could equally be explained by the presence of sub-100 nm inclusions of P-, Al- and Si-rich phases or perhaps adsorbed species such as [AlPO3]3+. If present, these would be below the ~150 nm resolution of the FEG-EPMA technique but when probed, would appear to be present as solid solution components. To determine conclusively the mechanism of P, Al and Si incorporation requires examination of

50 μm

Fig. 5. (a) Backscattered electron image of the phosphorus-rich grain outlined in Fig. 4. (b) High resolution Fe/Al/P FEG-EPMA map of the goethite grain highlighted in (a). A quantitative microanalysis line traverse was conducted along the line shown in (5b). Figure modified from MacRae et al. (2011).

goethite-rich regions that are known to contain these impurities, via a technique such as transmission electron microscopy (TEM). TEM is capable of providing chemical and structural information, down to atomic levels, provided the impurity-rich goethite can be successfully located, prepared as a thin film, and then analysed. Each of these steps provides a considerable technical challenge. In previous publications we describe results from initial TEM investigations on thick electron-transparent foils prepared from the part of the sample shown in Fig. 7 (see MacRae et al., 2010, 2011). The thickness of this layer was approximately 20 nm and confirmed the presence of elevated P, Al and Si as measured by energy dispersive spectroscopy (EDS) at the interface between hematite and goethite crystals. The TEM probe was then moved approximately 50 nm further into the goethite and only Al and P were observed by EDS. This suggests that while silicon is a common impurity in hematite, its role in goethite may be less important and that it may only play a role in the substitution of Al and P into the goethite structure. Based on these preliminary TEM results, however, the mechanism of replacement is not clear, although we propose that an interfacial structure rich in Al, Si and P initially forms, allowing the incorporation of Al and P into goethite. Further TEM studies are planned to probe the structure of the interfacial layer between hematite and goethite.

7

Minerals Engineering 143 (2019) 106022

1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0

Si (wt%)

Al (wt%)

M.I. Pownceby, et al.

0.0

0.2

0.4

0.6

0.8

2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0.0

1.0

0.2

0.4

Si (wt%)

Si (wt%)

2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0.0

0.2

0.4

0.6

0.8

1.0

P (wt%)

P (wt%)

0.6

0.8

1.0

2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0.0

Al (wt%)

0.5

1.0

1.5

2.0

Al+P (wt%)

Fig. 6. Element scatter plots from EPMA data along the traverse shown in Fig. 5b.

5. Implications for beneficiation of high-P goethite ores

considered to accelerate the transformation of goethite to hematite (αFe2O3), although a mechanism was not provided by Fisher-White et al. (2009, 2012a, 2012b). The method has also been demonstrated to be effective in reducing other impurities in the ore such as aluminium and silicon. This method of phosphorus and impurity level reduction, however, has yet to implemented on a commercial scale and currently in Australia, the majority of iron ore feedstock production is a blend of low-P (< 0.05 wt-% P) hematite-rich ores and high-P (> 0.10 wt-% P) goethite-rich ores.

The distribution of phosphorus within goethite grains at a micro- to nanometre crystalline level means that for mechanical beneficiation techniques to remove all the phosphorus, grinding would have to proceed down to the nanometer scale. As observed by Dukino and England (1997) this is clearly unrealistic in terms of energy usage as well as potential for iron unit losses and has been borne out by previous studies which have had limited success in removing phosphorus by mechanical means (e.g. Bensely and Rogers, 1987; Peixoto, 1991). Since it has been demonstrated in this study that phosphorus is somehow contained within the goethite structure it is likely that any method to substantially reduce the phosphorus content of the goethitic ore will require some sort of heat treatment and subsequent leaching in acid or caustic solution with minimal loss of iron units (e.g. Gooden et al., 1974; Bensely and Rogers, 1987; Cheng et al., 1999). The heat treatment dehydroxylates the goethite, opens the structure and converts it to hematite thereby allowing the leachant to access the impurities. Previous work has investigated roasting of high-P, goethite-rich ores followed by application of various acid leachates (e.g., sulphuric, nitric and hydrochloric acid) as a means of reducing the P content (e.g., Cheng et al., 1999). These earlier studies typically roasted ores at temperatures in the range 500–1300 °C, with the most successful methods requiring heating at temperatures > 850–900 °C (Fisher-White et al., 2009). However, such high temperatures are too energy intensive to be considered economic for commercial application (Fisher-White et al., 2009). Recently, Fisher-White et al. (2009, 2012a, 2012b) reported a modified heating-extraction method, which successfully reduced the P content of a high-P Brockman (M-G) iron ore by approximately 50% from an initial P content of ~0.15 wt-% P. The addition of small amounts of powdered sodium hydroxide to the ore, followed by heating the mixture in the range 300–350 °C before using a simple water leach, was found to be as successful in reducing the P content as the high temperature roast/alkaline leach method. The addition of NaOH was

6. Summary The importance of goethite to Australian iron ore production is steadily increasing as the proportion of goethite-rich and martite-goethite ore increases due to the declining production from high grade microplaty hematite deposits. The increase in goethite content can result in lower grade through the introduction of variable quantities of oxide impurities such as P, Al, and Si (although this depends on the type of goethite present) and also through the effect of higher LOI. Of the impurities in goethite, phosphorus is the most critical as high phosphorus level (> 0.1 wt-% P) invokes a price penalty because of the adverse effects on the quality of the end-product steel and increases production costs and CO2 emissions if dephosphorisation is carried out during steelmaking. Examination of a high-P type iron ore from the Pilbara region of Western Australia showed that three types of goethite could be recognised, with brown goethite, yellow ochreous goethite and vitreous goethite having different physical properties and visual characteristics. Results showed that phosphorus was higher in the goethite phases with the brown/vitreous goethite being particularly P-rich. The distribution of phosphorous contents within the earthy and ochreous goethite types was lower. The characterisation of high-P iron ores by field emission gun electron probe microanalyser (FEG-EPMA) was carried out to determine elemental associations and, using this knowledge, to speculate on 8

Minerals Engineering 143 (2019) 106022

M.I. Pownceby, et al.

1.3

Al

Fe Al P

0.2 10 μm

3.6

Si

1.1

10 μm

10 μm

Fig. 7. Elemental maps from Bulk Ore 2 showing the infiltration of Al, Si and P into Fe-oxide. Goethite is represented by the pink/mauve material interstitial to the hematite (blue). The areas containing elevated aluminium (green regions), are a goethite phase containing up to several weight percent aluminium. The line on the right hand Fe/Al/P map shows the position of the quantitative line traverse. Figure modified from MacRae et al. (2011). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) 0.09

Goethite (matrix) Fe-oxide Replaced Fe-oxide

0.8 0.7

0.08

1.8

0.07 0.06

2xSi (atom.)

0.9

1.6

0.05

1.4

0.5

1.2

0.03 0.02

Al + P (wt%)

Al + P (wt%)

0.04

0.6

0.4 0.3 0.2

0.01 0 0.02

1.0

0.03

0.04

0.05

0.06

0.07

0.08

0.09

0.1

Al+P (atom.)

0.8 0.6 0.4

0.1

0.2

0 0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

0.0 0.0

1.8

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

Si (wt%)

Si (wt%)

Fig. 8. Scatter plot of Al plus P versus Si quantitative microanalyses performed on a line traverse from hematite through the phosphorus-rich interface and into the goethite region from the area shown in Fig. 7. The scatter plot shows three distinct associations; Goethite (matrix), Fe-oxide and Replaced Fe-oxide.

Fig. 9. Plot of Al + P versus Si from the EPMA line traverse shown in Fig. 5b. Data at Si < 1.0 wt-% are from the high-P, high-Al region of Fig. 5, while data at Si > 1.0 wt-% are from the low-P, low-Al region. The inset graph shows the high-P, high-Al data plotted as 2Si4+ versus Al3+ + P5+ (in atom %).

possible P incorporation mechanisms. Quantitative data was collected from a range of Brockman high-P ores followed by a more detailed study of two goethite-rich bulk ores. Brown goethite was rare in the samples however the results confirmed goethite was the main repository for phosphorus, with vitreous goethite in particular containing the highest levels of phosphorus. The P-rich goethite(s) also contained elevated levels of both aluminium and silicon.

High resolution element distribution maps on Bulk Ores 1 and 2 showed that low-P regions were typically associated with low-Al and high-Si, whereas high-P regions were associated with high-Al and lowSi. A coupled substitution mechanism for phosphorus incorporation within goethite was proposed: 2Si4+ = 1P5+ + 1Al3+. Additional quantitative analyses on further examples indicated the mechanism was only valid for Si contents in goethite of < 1 wt-% Si. Above this level, P 9

Minerals Engineering 143 (2019) 106022

M.I. Pownceby, et al.

and Al substitution is inhibited. While a coupled substitution mechanism is plausible (at least for low-Si goethites), the association between impurity elements may also be explained by the presence of sub 100 nm inclusions of P-, Al- and Si-rich phases which would give the appearance of solid solution components. The presence of P, Al and Si impurities in the goethite mean that limited removal of phosphorus by mechanical separation techniques is possible but will most likely result in significant loss of iron units. Chemical leaching is also possible however this approach is only successful after thermal pre-treatment to re-crystallise the goethite as Fe2O3 and consequent concentration of impurities at grain boundaries.

8, 180–187. Gooden, J.E., Walker, W.M., Allen, R.J., 1974. Amdephos – A chemical process for dephosphorisation of iron ore. In: Natl Chem. Eng Conf 2nd Proc., pp 38-49 (Surfers Paradise, QLD, July 10-12, Univ. of QLD, St Lucia. Graham, J., 1973. Phosphorus in iron ore from the Hamersley Iron Formations. In: Proc. Aust. Inst Min. Metall. 246. pp. 41–42 June. Klein, C., Hurlbut Jr., C.S., 1985. Manual of Mineralogy, 21st ed. John Wiley and Sons, New York, pp. 391. MacRae, C.M., Wilson, N.C., Pownceby, M.I., Miller, P.R., 2010. Phosphorus and other impurities in Australian iron ores. Microsc. Microanal. 16, 896–897. MacRae, C.M., Wilson, N.C., Pownceby, M.I., Miller, P.R., 2011. The occurrence of phosphorus and other impurities in Australian iron ores. In: Iron Ore 2011 Conference, 11–13 July. The Australasian Institute of Mining and Metallurgy, Melbourne, Perth Western Australia, pp. 281–289. Manuel, J.R., Clout, J.M.F., 2017. Goethite classification, distribution and properties with reference to Australian iron deposits. In: Iron Ore 2017 Conference, 24–26 July. The Australasian Institute of Mining and Metallurgy, Melbourne, Perth Western Australia, pp. 567–574 Publication series 3/2017. Mintz, B., 1999. The influence of the composition on the hot ductility of steels and to the problem of transverse cracking. ISIJ Int. 39, 833–855. Mohapatra, B.K., Jena, S., Mahanta, K., Mishra, P., 2008. Goethite morphology and composition in banded iron formation, Orissa, India. Res. Geol. 58, 325–332. Morris, R.C., 2002. Genesis of high-grade hematitic orebodies of the Hamersley Province, Western Australia – a discussion. Econ. Geol. 97, 177–181. Morris, R.C., 1973. A pilot study of phosphorus distribution in parts of the Brockman Iron Formation, Hamersley Group, Western Australia, Western Australia Department of Mines Annual Report 1973, pp. 75–81. Morris, R.C., 1973b. A textural and mineralogical study of the relationship of iron ore to banded iron formation in the Hamersley Province of Western Australia. Econ. Geol. 75, 184–209. Morris, R.C., 1980. A textural and mineralogical study of the relationship of iron ore to banded iron-formation in the Hamersley iron ore Province of Western Australia. Econ. Geol. 75, 184–209. Morris, R.C., 1985. Genesis of iron ore in banded iron formation by supergene-metamorphic processes – a conceptual model. In: Wolf, K.H. (Ed.), Handbook of Stratabound and Stratiform Ore Deposits. Elsevier Science, The Netherlands, pp. 72–235. Ostwald, J., 1981. Mineralogy of Australian iron ores. BHP Technical Bull. 25, 4–12. Peixoto, G., 1991. Improvement of the reduction process in P content and other gangues in iron ore and its agglomerates. USA Patent Filed Nov. 14, Inter. Appl. No. PCT/ BR9/00030. Ramanaidou, E.R., Wells, M.A., 2014. 13.13 - Sedimentary hosted iron ores. In: Holland, H.D., Turekian, K.K. (Eds.), Treatise on Geochemistry, 2nd ed. Elsevier, Oxford, pp. 313–355. Schulze, D.G., 1984. The influence of aluminium on iron oxides VIII. Unit-cell dimension of Al-substituted goethites and estimation of Al from them. Clays Clay Miner. 32, 36–44. Schulze, D.G., Schwertmann, U., 1984. The influence of aluminium on iron oxides XIII. Properties of Al-substituted goethites. Clay Miner. 19, 521–539. Shannon, R.D., 1976. Revised effective ionic radii and systematic studies of inter-atomic distances in halides and chalcogenides. Acta Crystallogr. A A32, 751–767. Swedlund, P.J., Miskelly, G.M., McQuillan, A.J., 2010. Silicic acid adsorption and oligomerization at the ferrihydrite-water interface: interpretation of ATR-IR spectra based on a model surface structure. Langmuir 26, 3394–3401. Thomas, R., Webster, J.D., 1999. Characteristics of berlinite from the Ehrenfriersdorf pegmatite, Erzgebirge, Germany. Z. Geol. Wiss. 27, 443–454. Thorne, W., Hagemann, S., Webb, A., Clout, J., 2008. Banded iron formation-related iron ore deposits of the Hamersley Province, Western Australia. SEG Rev. 15, 197–221. Ward, D.F., Coles, I.G., Carr, W.M.B., 1975. Jimblebar and Western Ridge Iron Ore deposits, Hamersley Iron Province. In: Econ. Geol. Aust. Papua New Guinea. The Australasian Institute of Mining and Metallurgy, Melbourne, pp. 916–924. Waychunas, G.A., 1991. Crystal chemistry of oxides and oxyhydroxides, In: Lindsley, D.H. (Ed.). Oxide Minerals: Petrologic and Magnetic Significance, Rev. Mineral. Vol. 25, pp. 11–68. Wilson, N.C., 2013. CSIRO Mineral Resources, unpublished results.

Acknowledgements This work has been conducted over a number of years and involved discussions with a number of current and former CSIRO colleagues. In particular we would like to thank Erick Ramanaidou, Martin Wells, Nick Wilson and Roy Lovel for their input. Graham Sparrow and Warren Bruckard (CSIRO) are thanked for providing constructing reviews. Parts of this work were presented at the IronOre2011 conference (these have been acknowledged in the text) but the work therein has been significantly expanded upon. References Bensely, C., Rogers, H., 1987. Phosphorus removal before the blast furnace, In: BHP Internal Report (cited in Dukino and England, 1997). Briant, C.L., Mesmer, R.P., 1982. An electronic model for the effect of alloying elements on the phosphorus induced grain boundary embrittlement of steel. Acta Metall. 30, 1811–1818. Cheng, C.Y., Misra, V.N., Clough, J., Mun, R., 1999. Dephosphorisation of Western Australian iron ore by hydrometallurgical process. Miner. Eng. 12, 1083–1092. Clout, J.M.F., 2005. Iron formation-hosted iron ores in the Hamersley Province of Western Australia. In: Iron Ore 2005 Conference, 19-21 September. The Australasian Institute of Mining and Metallurgy, Melbourne, Perth, Western Australia, pp. 9–19 Publication series 8/2005. Cornell, R.M., Schwertmann, U., 1996. The Iron Oxides. VCH Verlagsgesellschaft, Weinheim, pp. 571. Dub, V.S., Dub, A.V., Makarycheva, E.V., 2006. Role of impurity and process elements in the formation of structure and properties of structural steels. Met. Sci. Heat Treat. 48, 279–286. Dukino, R.D., England, B.M., 1997. Phosphorus in Hamersley Range iron ores, In: Proceedings Ironmaking Resources and Reserves Estimation, Perth, 25-26 September, pp. 197–202. Dukino, R.D., England, B.M., Kneeshaw, M., 2000. Phosphorus distribution in BIF-derived iron ores of Hamersley Province, Western Australia. Trans Inst. Min. Metall., B B108, 168–176. Fisher-White, M.J., Lovel, R.R., Sparrow, G.J., 2009. Phosphorus removal from iron ore with a low temperature heat treatment. Publication series 15/2009 In: Iron Ore 2009 Conference, 27-29 July. The Australasian Institute of Mining and Metallurgy, Melbourne, Perth, Western Australia, pp. 249–254. Fisher-White, M.J., Lovel, R.R., Sparrow, G.J., 2012a. Phosphorus removal from goethitic iron ore with a low temperature heat treatment and a caustic leach. ISIJ Int. 52, 797–803. Fisher-White, M.J., Lovel, R.R., Sparrow, G.J., 2012b. Heat and acid leach treatments to lower phosphorus levels in goethitic iron ores. ISIJ Int. 52, 1794–1800. Fysh, S.A., Clark, P.E., 1982. Aluminous goethite: a Mössbauer study. Phys. Chem. Miner.

10