Minerals Engineering xxx (2013) xxx–xxx

Contents lists available at SciVerse ScienceDirect

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

Rare Earth Element deportment studies utilising QEMSCAN technology

Duncan M. Smythe ⇑, Annegret Lombard, Louis L. Coetzee
SGS South Africa, Advanced Mineralogical Facility, 58 Melvill Street, Booysens 2091, South Africa

a r t i c l e i n f o a b s t r a c t

Article history: Due to the variability and complex nature of REE-containing ores, it is vital to understand the mineral-
Available online xxxx ogical characteristics, before embarking on any metallurgical testwork campaign. REE’s are present in a
variety of phases/minerals, all of which may react differently during processing. It is therefore important
Keywords: to identify and quantify all the REE-phases present in the ore, their mineral associations, grain size dis-
Precious metal ores tributions as well as their liberation characteristics. In order to quantify the elemental deportment of the
Liberation analysis different REE’s into the different REE-bearing phases, it is necessary to determine the mineral chemical
(REE)-Ore mineralogy
compositions by Electron Microprobe. By assigning the average measured REE-elemental compositions
to the different REE-phases, it becomes possible to determine the amount of each REE per REE-phase. Pre-
dictions about the best theoretically achievable grades and recoveries can be made. The data obtained are
used to design a metallurgical testwork program suitable for the specific ore-type.
Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction containing phases were therefore possibly overlooked by optical

techniques. Thanks to modern Scanning Electron Microscope
An elemental deportment involves the quantitative investiga- (SEM) technology, we are able to detect REE’s to within 1% min-
tion of the relationship between one or more elements and the eral mass percent with standard QEMSCAN Energy Dispersive
hosts, in which these elements occur in significant concentrations. Spectroscopy (EDS) and as low as 0.01% by Wavelength Disper-
These hosts generally consist of an array of minerals of which the sive Spectroscopy (WDS).
proportions vary significantly from one rock type to another. The
average concentration of each element of interest is calculated
per mineral phase, using various techniques such as Electron 2. REE background information
Microprobe (EMP) analyses. As the mineral content of the sample
is quantitatively measured, by Quantitative Evaluation of Minerals The term ‘‘Rare Earth Elements’’ refers to a group of 17 elements
by Scanning Microscopy (QEMSCAN) analyses, the proportion of (Fig. 1), that include scandium (atomic number 21), yttrium, due to
the elements of interest per phase can be calculated. This gives a its similar chemical characteristics (atomic number 39) (Verplanck
good indication of the likely elemental recoveries (also referred and Van Gosen, 2011), and a group of 15 lanthanides (atomic num-
to as an elemental deportment), as well as an indication of the pos- bers 57–71) (Kerr, 2011). The combination of all REE’s, in this
sible processing methods for concentration. A single or select few paper, are referred to as Total Rare Earth Elements (TREE), Light
host minerals may be targeted according to their physical proper- Rare Earth Elements (LREE) and Heavy Rare Earth Elements (HREE).
ties with regards to size, density or flotation potential (relative to Europium forms the boundary phase of the LREE’s to HREE’s with
the gangue mineral assemblage). Furthermore, the textural rela- La, Ce, Pr, Nd, Pm, Sm and Eu being classed as LREE’s, while Gd,
tionships between the host minerals and the gangue assemblage Tb, Dy, Ho, Er, Tm, Yb and Lu are classed as HREE’s. Promethium
can be investigated. only occurs in trace amounts in nature. It is generally unstable
Rare Earth Element (REE) deportments are notoriously difficult and rapidly breaks down to other isotopes (British Geological Sur-
due to the complexity and large degree of variability in chemical vey, 2010). Current market values indicate that, at present, the
composition of the REE phases. In the past optical mineralogy HREE’s are more valuable than the LREE’s (British Geological Sur-
was used to identify the REE host minerals. However, REE’s occur vey, 2010).
in small concentrations in a vast array of minerals and only a select Rare Earth elements have similar atomic characteristics allow-
few contain large concentrations. Significant amounts of REE ing them to substitute effectively for one another (Rollinson,
1993). This invariably means that minerals containing REE’s will
⇑ Corresponding author. Tel./fax: +27 (0) 11 681 2500. contain a wide selection of REE’s in varying concentrations. Large
E-mail addresses: [email protected], [email protected] processing costs are therefore incurred attempting to separate
(D.M. Smythe). individual REE’s into ‘‘clean’’ metals and alloys for industry (Gupta

0892-6875/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.mineng.2013.03.010

Please cite this article in press as: Smythe, D.M., et al. Rare Earth Element deportment studies utilising QEMSCAN technology. Miner. Eng. (2013), http://
dx.doi.org/10.1016/j.mineng.2013.03.010
2 D.M. Smythe et al. / Minerals Engineering xxx (2013) xxx–xxx

Fig. 1. Periodic table of the elements (http://www.elementsdatabase.com).

Table 1 then be made as to the economic potential of the ore resource in

Common REE-containing minerals and corresponding approximate REO contents question as well as possible beneficiation methods. The following
(after British Geological Survey, 2010).
methodologies were employed when conducting the REE
Mineral name Mineral formula Approximate REO% deportment:
Aeschynite (Ce,Ca,Fe,Th)(Ti,Nb)2(O,OH)6 32
Allanite (Ce,Ca,Y)2(Al,Fe)3(SiO4)3(OH) 38 3.1. Crushing and milling
Apatite Ca5(PO4)3(OH,F,Cl) 19
Bastnaesite (Ce,La)(CO3)F 75
Brannerite (U,Ca,Ce)(Ti,Fe)2O6 9
The REE ore minerals as well as the gangue mineral assemblage
Britholite (Ce,Ca,Th,La,Nd)5(SiO4,PO4)3(OH,F) 32 have an important influence on the grindability of the ore. The
Eudialyte Na4(Ca,Ce)2(Fe,Mn,Y)ZrSi8O22(OH,Cl)2 9 grain size distribution of the minerals in question determines the
Euxenite (Y,Ca,Ce)(Nb,Ta,Ti)2O6 24 optimum grind for exposure and association of the ore minerals.
Fergusonite (Ce,La,Nd)NbO4 53
The hardness of the gangue and ore mineral assemblage further
Gadolinite Y2Fe2Be2Si2O10 60
Kainosite Ca2(Y,Ce)2Si4O12(CO3)(H2O) 38 dictates the degree to which the ore may be economically milled.
Loparite (Ce,Na,Ca)2(Ti,Nb)2O6 30 In general a grind size of 80% passing 75 lm is used for laboratory
Monazite (Ce,La,Nd,Th)(PO4) 65 scale testwork. This grind size is commonly used in comminution
Parisite Ca(Ce,La)2(CO3)3F2 61 mills in larger scale processing plants.
Xenotime YbPO4 61
Yttrocerite CaF2(Y,Ce)F3 53
Deposits which are relatively homogeneous, such as REE-rich
Huanghoite BaCe(CO3)2F 39 alluvial sands or monazite beach sands, may be treated as bulk
Cebaite Ba3Ce2(CO3)5F2 32 samples in which a large amount of sample material is crushed,
Florencite (La,Ce)Al3(PO4)2(OH)6 32 milled and blended, creating one representative sample.
Synchysite CaCe(CO3)2F 51
Deposits which are heterogeneous may be characterized by
Samarskite (Y,Fe,U)(Nb,Ta)5O4 24
analysing a number of different samples separately, in order to
ascertain variation within the ore body itself. This method is gen-
erally used when small amounts of sample are available from very
and Krishnamurthy, 2005). Furthermore, REE’s are not found in selective locations. This provides detailed analyses of the samples
their elemental state in nature. REE’s are therefore only recover- themselves, but does not necessarily reflect the entire ore body.
able from minerals containing economic grades of REE’s. The fol-
lowing minerals act as predominant REE hosts within the natural
3.2. Head chemical assay
environment (Table 1).
These minerals are found in a wide selection of geological set-
Aliquots from representative sample material are split and
tings such as: Iron-REE deposits, carbonatite deposits, laterite
analysed by various quantitative methodologies, most commonly:
deposits, placer deposits, HREE deposits in peralkaline igneous
X-ray Fluorescence (XRF) for major element quantification, Induc-
rocks and vein deposits (Castor and Hedrick, 2006). It is important
tively Coupled Plasma Mass Spectroscopy (ICP-MS) for REE quanti-
to note that REE’s may be found in a far larger assemblage of min-
fication, and Leco for total sulphur quantification.
erals than listed above, including common minerals such as clays
and micas.
3.3. Mineralogical characterization

3. REE-deportment methodology Transverse and normal polished blocks are created from milled
representative sample aliquots. Polished thin sections are also cut
Although the processing of REE’s can be discussed in great from core sections where needed. These polished blocks and
detail, this paper aims to provide an understanding of the detection polished thin sections are subjected to QEMSCAN analyses in
and quantification of REE-containing minerals, in order to gain which the surface of each block is mapped according to various
an understanding of the likely REE recoveries. Decisions may methodologies.

Please cite this article in press as: Smythe, D.M., et al. Rare Earth Element deportment studies utilising QEMSCAN technology. Miner. Eng. (2013), http://
dx.doi.org/10.1016/j.mineng.2013.03.010
 D.M. Smythe et al. / Minerals Engineering xxx (2013) xxx–xxx 3

Initially a QEMSCAN Bulk Modal Analysis (BMA) is performed 4.1. Chemical assay
on a transverse cut polished block. The block is cut in a transverse
manner in order to eliminate the effect of densimetric separation The results of the ICP-MS analyses on the TREE of the samples
of minerals during the curing of the resin in which they are shows that the samples are dominated, in terms of their TREE com-
mounted. The transverse cut blocks are mapped by a line scan position, by Ce, La, Nd, Pr, Sm and Y (Table 2).
methodology in which the entire face of the block is scanned in The chondrite normalized plots of the TREE’s of all the samples
the X-direction while fixed line spacing is maintained in the are presented in Fig. 2. The general trend of the graph is a good
Y-direction. High counting statistics are produced decreasing mea- indication as to which REE’s are concentrated and which are
surement and orientation error. Furthermore, a calculated chemi- depleted. In the example below, the inclination of the data indi-
cal composition is supplied by the QEMSCAN iDiscover software. cates that LREE’s are concentrated while HREE’s are depleted. A
This theoretical chemical composition is compared to that of the typical Eu anomaly is present in the data. Europium occurs in
measured chemical composition as a quality assurance and quality two states, namely Eu2+ and Eu3+. The concentration or depletion
control (QA/QC) process. If the difference between the measured of Eu is primarily controlled by feldspars as Eu2+ is easily substi-
bulk chemical composition (XRF) and the calculated bulk chemis- tuted for Ca2+. This is therefore indicative of intrusive bodies in
try (QEMSCAN) is outside of acceptable limits (>2% deviation), which fractionation mobilized feldspar from the cooling fluids, as
the results are rejected. well as deposits formed from fluid generated by partial melting
Normal polished sections are created for analyses by QEMSCAN in which the feldspars remained in situ (Rollinson, 1993). The
Specific Mineral Search (SMS). This orientation allows for densi- chondrite normalized plots therefore provide invaluable informa-
metric separation as heavier minerals settle initially at the face tion on the deposit type and petrogenesis of the sample.
of the block followed by lighter smaller particles towards the rear. The major element analyses by X-ray Fluorescence (XRF), borate
This effect is desirable as it is generally the denser minerals that fusion are presented in Table 3.
contain the elements of interest. The likelihood of encountering a The samples show a large degree of variation specifically with
mineral containing the target element (REE) is therefore increased. regards to their SiO2, P2O5, Al2O3, CaO and Fe2O3 contents. Samples
False colour particle maps are created of the minerals of interest, B and C are dominated by SiO2 indicating a large amount of silicate
based on a point spacing surface map. Entire particles are mapped minerals present in these samples. Samples A and D are dominated
in this manner. by P2O5 indicating they are dominated by phosphates. The low
If needed, thin sections are used for petrographic descriptions of total values are indicative of elements present, which have not
the in situ minerals, as well as the relationships of ore minerals been measured and accounted for. Taking into account the ICP-
with the surrounding gangue mineralogy. Furthermore, a QEM- MS data, this deficit is mostly accounted for by extremely high lev-
SCAN Field Map (FM) analyses may be conducted on the polished els of La, Ce, Nd, Sm and Y.
thin sections. This methodology maps the entire face (or a delin-
eated portion) of the thin section in 2-dimensional sections. The 4.2. Mineral composition
sections are stitched together using the iDiscover software to gen-
erate a large false colour composite map which can be configured The XRD analyses indicate that quartz, mica and illite are the
to display changes in mineralogy, elemental composition, density major gangue minerals present, while monazite, allanite and
variations, in situ mineral associations, liberations and potential apatite appear to be the main potential hosts for REE’s. The over-
exposures. all mineralogy does not vary much between samples, however,
the abundance of the various crystalline phases in the samples
does vary significantly (Table 4). A function of the peak height
3.4. REE deportment can be used to infer, qualitatively, the amount of the specific
identified mineral. When regarding radioactive minerals, in this
An Electron Microprobe (EMP) investigation forms an integral case high Th-monazite, large absorption coefficients affect the
part of the deportment analyses of the samples. As REE’s generally peak intensity of the monazite grains. Although corrections are
occur in moderate yet variable concentrations within REE-contain-
ing mineral phases, it becomes important to achieve a good statis-
tical analysis of the average REE-content of each REE-containing
mineral phase. By measuring a large number of REE-containing Table 2
phases, very accurate TREE content can be evaluated. The calcu- Inductively Coupled Plasma Mass Spectroscopy (ICP-MS) results for TREE’s in ppm.
lated average TREE concentrations are assigned to the correspond- REE (ppm) Sample A Sample B Sample C Sample D
ing minerals in the mineral characterization process. As the modal
La 71,500 21,600 19,300 63,700
mineralogy of the sample is accurately known from the BMA inves- Ce 157,239 47,200 41,700 139,132
tigation, and the average TREE-concentration of each mineral Pr 17,500 5420 4700 15,000
phase is know from the EMP investigation, a very accurate REE- Nd 62,800 19,000 16,400 54,400
Sm 10,500 3080 2710 8970
elemental deportment can be derived from the data. This gives a
Eu 214 65 57 164
good indication which minerals are important to target for their Gd 7540 2300 2010 6870
REE-content, as well as determines the maximum possible recov- Tb 814 254 226 755
ery from the samples, once the relevant minerals have been Dy 3360 1060 957 3250
concentrated. Ho 511 168 153 520
Er 962 331 300 1060
Tm 82 30 28 101
Lu 29 13 12 47
4. REE deposit case study Yb 348 144 132 499
Y 12,100 4160 3810 13,200
Sc 6 5 5 5
The following set of data was taken from a recent REE-deport-
ment study. The results and methodologies are presented as well 16 REE total (%) 34.55 10.48 9.25 30.77
Th 46,900 15,500 13,500 44,500
as a brief discussion of the conclusions attained during the
U 758 305 357 906
investigation.

Please cite this article in press as: Smythe, D.M., et al. Rare Earth Element deportment studies utilising QEMSCAN technology. Miner. Eng. (2013), http://
dx.doi.org/10.1016/j.mineng.2013.03.010
4 D.M. Smythe et al. / Minerals Engineering xxx (2013) xxx–xxx

Fig. 2. Chondrite normalized TREE plot obtained by normalizing the chemical analyses obtained in the present study to the chondritic composition of Boynton (1984) using
the method recommended by Rollinson (1993).

Table 3 content. Furthermore, it contains a high P2O5 content, and con-

Major element composition, reported as oxide%, as determined by XRF borate fusion. sidering the BMA analyses, Sample A therefore contains the larg-
Element Sample A Sample B Sample C Sample D est amount of monazite of the samples studied. A direct
SiO2 7.81 33.00 50.80 2.33 correlation can be implied between the monazite content and
Al2O3 2.30 13.30 9.89 0.80 the REE-grade. Sample A should then have the highest grade fol-
CaO 2.86 4.86 6.33 11.70 lowed by Sample D. As allanite is also a REE-containing mineral,
MgO 0.30 1.07 0.40 0.08 an increase in allanite would decrease the recoverable REE’s
Fe2O3 12.10 12.40 5.48 7.55
from monazite, as may be the case in Sample A as it has a high-
K2O 0.20 1.99 2.40 0.03
MnO <0.01 0.07 0.03 <0.01 er allanite content compared to Sample D. The REE-recovery
Na2O <0.05 1.71 0.20 <0.05 may therefore potentially be significantly affected by the types
P2O5 17.20 6.02 6.55 22.70 of mineral hosting the REE’s. Samples B and C contain high
TiO2 1.30 0.55 0.24 0.51
TREE-contents however, the monazite and corresponding P2O5
Cr2O3 0.17 0.08 0.06 0.07
V2O5 0.03 0.02 <0.01 <0.01
contents are very low. The REE-deportment is extremely impor-
LOI 1.76 3.38 1.79 4.21 tant in evaluating these two samples, as the REE’s may also be
Total 51.00 83.80 86.10 56.40 present in allanite and apatite. Table 5 shows the calculated
mineral composition of the samples based on a Bulk Modal
Analyses (BMA) investigation of representative splits of each
fraction.
applied to the patterns, underestimation of the monazite may be
unavoidable if the surrounding mineral peaks have very high 4.3. QEMSCAN QA/QC
intensities. Fig. 3 shows the individual X-ray diffractograms of
each sample. The major mineral identification peaks are labelled In order to evaluate the accuracy and consistency of the derived
in the diffractogram. data, calculated QEMSCAN bulk chemistry is compared to that of
The BMA analyses (Table 5) concur with the XRD results with the whole rock bulk chemistry, conducted by XRF. The values of
regards to the bulk mineralogical composition. Specific zones of the comparison are provided in Table 6. An R2 value is presented
high/low grade mineralization may already be inferred from the for each sample. This value describes the deviation of the results
BMA and chemical analyses. Sample A contains the highest TREE from a perfectly linear relationship (R2 = 1).

Table 4
Approximate mineral abundance as determined by XRD qualitative analyses.

Mineral Approximate formula Approximate abundance

Sample A (%) Sample B (%) Sample C (%) Sample D (%)
Quartz SiO2 3–10 20–50 20–50 <3
Chlorite (Mg,Fe)6(Si,Al)4O10(OH)8 10–20 3–10 3–10 3–10
Plagioclase (Na,Ca)(Al,Si)4O8 B.Q.L. 3–10 3–10 B.Q.L.
Allanite Ca–Mn–REE–Fe–Ti–Al–O–OH–Si 10–20 3–10 3–10 10–20
Illite KAl2(Si3AlO10)(OH)2 10–20 10–20 20–50 3–10
Mica K–Mg–Fe–Al–Ti–Si–O <3 10–20 3–10 3–10
Monazite (Ce, La, Th)PO4 20–50 10–20 10–20 20–50
Apatite Ca(PO4)3(OH,F,Cl) B.Q.L. 3–10 3–10 10–20
Magnetite Fe3O4 <3 <3 B.Q.L. <3
Ilmenite FeTiO3 <3 <3 <3 3–10
Pyrite FeS2 B.Q.L. <3 <3 3–10
Chalcopyrite CuFeS2 <3 3–10 B.Q.L. <3

B.Q.L. = below quantification limit (<1% modal abundance).

Please cite this article in press as: Smythe, D.M., et al. Rare Earth Element deportment studies utilising QEMSCAN technology. Miner. Eng. (2013), http://
dx.doi.org/10.1016/j.mineng.2013.03.010
 D.M. Smythe et al. / Minerals Engineering xxx (2013) xxx–xxx 5

Fig. 3. Overlain XRD diffractograms of each sample.

Table 5
Quantitative mineral composition as determined by QEMSCAN BMA.

Mineral Approximate mineral formula Mineral mass (%)

Sample A Sample B Sample C Sample D
Quartz SiO2 4.35 12.64 34.26 0.95
Plagioclase (Na,Ca)(Al,Si)4O8 0.05 3.54 1.68 0.01
Allanite Ca–Mn–REE–Fe–Ti–Al–O–OH–Si 9.01 25.16 18.57 1.17
Chlorite (Mg,Fe)6(Si,Al)4O10(OH)8 5.81 8.12 2.26 2.38
Zircon Zr(SiO4) 0.84 0.50 0.54 0.29
Mica/Clay KAl2(Si3AlO10)(OH)2 4.05 26.45 26.65 0.36
Silicates – 24.10 76.40 83.96 5.16
Pyrite FeS2 0.06 0.88 0.04 8.92
Chalcopyrite CuFeS2 0.04 5.83 0.01 2.49
Galena PbS 0.05 0.01 0.03 0.02
Sulphides – 0.16 6.73 0.08 11.43
Fe-oxide Fe3O4 9.99 1.08 0.06 1.24
Ilmenite FeTiO3 3.07 0.65 0.18 1.32
Oxides – 13.06 1.72 0.23 2.56
Calcite CaCO3 0.17 0.02 0.15 0.32
Carbonates – 0.17 0.02 0.15 0.30
Monazite (Ce,La,Th)PO4 59.13 11.86 8.87 52.25
Apatite Ca(PO4)3(OH,F,Cl) 3.38 3.27 6.71 28.31
Phosphates – 62.51 15.14 15.58 80.56
Other – 0.00 0.00 0.00 0.00
Total – 100.00 100.00 100.00 100.00

Table 6
Calculated QEMSCAN chemistry versus measured XRF chemistry.

Element Sample A Sample B Sample C Sample D

Calculated Measured Calculated Measured Calculated Measured Calculated Measured
Ce 14.41 15.72 4.49 4.72 3.34 4.17 12.25 13.91
La 6.62 7.15 2.03 2.16 1.51 1.93 5.63 6.37
Al 2.00 1.22 6.93 7.04 5.36 5.24 0.33 0.42
Ca 2.60 2.04 4.88 3.47 4.96 4.52 8.36 8.36
Cu 0.12 0.05 1.45 2.04 0.01 0.05 1.14 0.49
Fe 8.57 9.41 8.03 9.64 3.42 4.26 6.18 5.87
K 0.32 0.17 2.40 1.65 1.48 1.99 0.05 0.02
Mg 0.98 0.18 0.43 0.65 0.85 0.24 0.26 0.05
Mn 0.14 0.01 0.17 0.05 0.02 0.02 0.01 0.01
Na 0.06 0.04 0.95 1.27 0.48 0.15 0.01 0.04
P 8.31 7.51 3.63 2.63 3.46 2.86 10.32 9.91
Pb 0.24 0.05 0.08 0.05 0.09 0.05 0.20 0.05
S 0.11 0.08 3.43 2.88 0.11 0.06 4.82 5.21
Si 3.68 3.65 16.57 15.43 23.88 23.75 1.92 1.09
Th 4.98 4.27 1.55 1.39 1.29 1.26 4.41 3.72
Ti 0.80 0.78 0.56 0.33 0.15 0.14 0.78 0.31
R2 0.988 0.969 0.994 0.984

Please cite this article in press as: Smythe, D.M., et al. Rare Earth Element deportment studies utilising QEMSCAN technology. Miner. Eng. (2013), http://
dx.doi.org/10.1016/j.mineng.2013.03.010
6 D.M. Smythe et al. / Minerals Engineering xxx (2013) xxx–xxx

Table 7 degree into allanite and Fe-oxide compared to La and Ce. As alla-
Ce, La and Dy deportment into associated REE-containing minerals. nite forms as a secondary alteration mineral in the samples, it is
Mineral Allanite Fe-oxide Monazite assumed that the HREE’s are more mobile during high temperature
EMP assigned Ce-content (%) 6.84 0.26 23.28 metasomatic events, when compared with the LREE’s.
Ce-deportment Sample A 4.28 0.18 95.54
Sample B 38.37 0.06 61.57
Sample C 38.07 0.00 61.92
4.5. Ore mineral association
Sample D 0.66 0.03 99.32
EMP assigned La-content (%) 3.00 0.15 10.71
A QEMSCAN FM analysis is used to create false colour composite
images of in situ minerals in the prepared polished thin sections.
La-deportment Sample A 4.08 0.23 95.69
Sample B 37.24 0.08 62.68
This technique is used to more accurately investigate textural rela-
Sample C 36.95 0.01 63.04 tionships in the ore. Figs. 4–7 illustrate the associations and tex-
Sample D 0.62 0.03 99.34 tural intergrowths of monazite in the samples. Table 8 shows
EMP assigned dy-content (%) 0.24 0.04 0.43 that monazite is mainly associated with Fe-oxide in Sample A,
Dy-deportment Sample A 7.72 1.43 90.85
chlorite in Sample B, mica/clay in Sample C and apatite in Sample
Sample B 54.00 0.38 45.62 D. Monazite appears to be relatively euhedral in habit and gener-
Sample C 53.86 0.03 46.11 ally coarse (Figs. 4–7).
Sample D 1.24 0.22 98.55 The mineral associations are derived from the Bulk Modal Anal-
ysis line scan analyses in which the numbers of adjacent pixels of
minerals are compared. If a pixel of one mineral lies exactly
adjacent to a pixel of another mineral, the two minerals are termed
4.4. REE deportment to be ‘‘associated’’.

The TREE-elemental deportment is calculated by evaluating

EMP data and relating the measured TREE concentrations in 4.6. Ore mineral exposure
discrete REE-containing minerals, to the modal abundance of the
minerals in the samples. Table 7 shows the results of the Ce, La The results of the exposure and association properties of dis-
and Dy elemental deportment. Ce and La are expected to be repre- crete monazite grains are presented in Table 9. The level of expo-
sentative of the LREE’s while Dy is expected to be representative of sure is a function of the surface area of a particular mineral
the HREE’s. which may be exposed to a solution. Liberation is an area percent-
The Ce, La and Dy deportment indicates that the majority of the age function by which the total area of a particle is compared to the
REE’s are locked within monazite, allanite (especially Samples B area which is occupied by a discrete mineral. Furthermore, miner-
and C), and to a lesser extent Fe-oxide. A pattern emerges in the als which are well exposed may be poorly liberated (e.g. alteration
data indicating that Ce and La deport in similar concentrations to rims) and conversely well liberated minerals may show a poor
specific REE-containing minerals. Dy however, deports to a greater exposure.

Fig. 4. QEMSCAN Field Map of in situ minerals in Sample A. The image on the left is a false colour composite image of different minerals present. The central image shows the
elemental distribution of La within the minerals in question. The image on the right is an elemental distribution map of Ce. (For interpretation of the references to colour in
this figure legend, the reader is referred to the web version of this article.)

Please cite this article in press as: Smythe, D.M., et al. Rare Earth Element deportment studies utilising QEMSCAN technology. Miner. Eng. (2013), http://
dx.doi.org/10.1016/j.mineng.2013.03.010
 D.M. Smythe et al. / Minerals Engineering xxx (2013) xxx–xxx 7

Fig. 5. QEMSCAN Field Map of in situ minerals in Sample B. The image on the left is a false colour composite image of different minerals present. The central image shows the
elemental distribution of La within the minerals in question. The image on the right is an elemental distribution map of Ce. (For interpretation of the references to colour in
this figure legend, the reader is referred to the web version of this article.)

Fig. 6. QEMSCAN Field Map of in situ minerals in Sample C. The image on the left is a false colour composite image of different minerals present. The central image shows the
elemental distribution of La within the minerals in question. The image on the right is an elemental distribution map of Ce. (For interpretation of the references to colour in
this figure legend, the reader is referred to the web version of this article.)

4.7. Ore mineral liberation well liberated monazite particles. Sample C contains the poorest
liberated monazite grains. Although poor, 80% of the monazite
The liberation analysis reveals that monazite grains are more grains still fall into the liberated and middlings liberation classes
liberated in Samples B and D. Sample A contains slightly less (30–100% liberation). The sample could be milled slightly finer

Please cite this article in press as: Smythe, D.M., et al. Rare Earth Element deportment studies utilising QEMSCAN technology. Miner. Eng. (2013), http://
dx.doi.org/10.1016/j.mineng.2013.03.010
8 D.M. Smythe et al. / Minerals Engineering xxx (2013) xxx–xxx

Fig. 7. QEMSCAN Field Map of in situ minerals in Sample D. The image on the left is a false colour composite image of different minerals present. The central image shows the
elemental distribution of La within the minerals in question. The image on the right is an elemental distribution map of Ce. (For interpretation of the references to colour in
this figure legend, the reader is referred to the web version of this article.)

Table 8
Normalized mineral association as determined by QEMSCAN pixel associations.

Minerals Sample A Sample B Sample C Sample D

Allanite Fe-oxide Monazite Allanite Fe-oxide Monazite Allanite Fe-oxide Monazite Allanite Fe-oxide Monazite
Quartz 3.45 15.24 5.23 23.73 9.54 13.61 19.83 13.98 25.17 9.04 1.45 5.67
Plagioclase 0.23 0.02 0.04 7.42 0.30 0.84 5.05 0.54 0.68 0.81 0.00 0.04
Allanite 0.00 13.08 15.45 0.56 0.67 0.18 0.05 0.00 0.00 0.00 0.00 0.07
Chlorite 21.10 13.75 2.48 0.00 37.70 31.70 0.00 50.00 17.44 0.00 26.09 13.81
Zircon 4.27 0.42 2.44 1.36 0.30 2.47 0.45 0.00 1.85 1.25 0.57 5.11
Mica/clay 32.05 4.69 6.21 35.93 13.49 23.43 51.23 15.59 37.62 3.23 0.73 3.55
Pyrite 0.04 0.14 0.57 3.10 5.22 0.62 0.03 0.00 0.14 4.11 6.06 8.62
Chalcopyrite 0.13 0.06 0.66 8.09 5.66 5.24 0.00 0.00 0.00 1.03 3.39 6.94
Galena 0.02 0.02 0.33 0.01 0.00 0.09 0.03 0.00 0.07 0.00 0.00 0.00
Fe-oxide 12.70 0.00 36.35 4.19 0.00 7.62 1.40 0.00 0.48 23.73 0.00 12.80
Ilmenite 3.95 15.63 7.29 2.49 10.51 0.57 11.49 8.60 0.34 2.72 13.17 1.57
Calcite 3.57 1.83 11.18 0.99 0.37 4.93 1.58 1.61 6.98 11.02 5.01 19.93
Monazite 13.98 33.88 0.00 5.97 12.89 0.00 3.83 3.76 0.00 27.19 27.71 0.00
Apatite 4.11 1.19 11.38 6.09 3.28 8.37 4.99 5.91 8.55 15.72 15.83 21.20
Other 0.41 0.06 0.39 0.07 0.07 0.35 0.05 0.00 0.68 0.15 0.00 0.67
Total 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00

to increase the liberation to match that of the remaining order of tens of percents) of TREE. The samples were highly
samples. Overall the samples show exceptionally good liberation enriched in LREE and slightly depleted in HREE. The primary REE’s
as they contain, on average 95% particles with 30–100% in the samples are Ce, La, Pr, Nd, Sm and ±Y.
liberation. The mineral abundances vary considerably between the four
Fig. 8 shows the results of the liberation analyses of REE- samples. However, the samples are generally high in monazite,
containing monazite grains in the samples. The liberation classes allanite, quartz and mica/clay. A chemical correlation occurs
are illustrated in Fig. 9. between the TREE content, P2O5 content and monazite content of
the samples. Where these elements are high in concentration,
4.8. Case study observations and interpretation monazite generally dominates the mineral composition. In the case
of the sample containing high SiO2 content as well as high TREE
The results of the case study investigation indicate that the and P2O5 contents, monazite, allanite and ± quartz dominate the
samples in question contained very high concentrations (in the mineral assemblage.

Please cite this article in press as: Smythe, D.M., et al. Rare Earth Element deportment studies utilising QEMSCAN technology. Miner. Eng. (2013), http://
dx.doi.org/10.1016/j.mineng.2013.03.010
 D.M. Smythe et al. / Minerals Engineering xxx (2013) xxx–xxx 9

Table 9
Exposure and association of the primary REE containing mineral phase – monazite.

Exposure and association categories Sample A Sample B Sample C Sample D

(monazite)
Total area Mass% Total area Mass% Total area Mass% Total area Mass%
(pixels) (pixels) (pixels) (pixels)
Exposed (P80%) 33,185 24.67 11,279 30.47 3815 30.26 45,911 50.63
Exposed (P10 < 80%) 101,071 75.15 25,455 68.78 8696 68.97 44,202 48.74
Exposed (<10%) 51 0.04 179 0.48 41 0.33 423 0.47
Total exposed (>10%) 13,4256 99.83 36,734 99.25 12,511 99.23 90,113 99.37
Locked in sulphide (100%) 0 0.00 0 0.00 0 0.00 10 0.01
Locked in silicate (100%) 24 0.02 37 0.10 19 0.15 75 0.08
Locked in oxide (100%) 8 0.01 0 0.00 0 0.00 1 0.00
Locked in carbonates (100%) 0 0.00 0 0.00 0 0.00 0 0.00
Locked in phosphates (100%) 9 0.01 0 0.00 0 0.00 6 0.01
Locked on sulphide–silicate boundary 0 0.00 7 0.02 2 0.02 0 0.00
Locked on sulphide–oxide boundary 0 0.00 0 0.00 0 0.00 0 0.00
Locked on sulphide–carbonate boundary 0 0.00 0 0.00 0 0.00 0 0.00
Locked on silicate–oxide boundary 60 0.04 1 0.00 0 0.00 42 0.05
Locked on silicate–carbonate boundary 7 0.01 3 0.01 29 0.23 0 0.00
Locked on poly-mineral boundary 75 0.06 50 0.14 6 0.05 15 0.02
Total locked 183 0.14 98 0.26 56 0.44 149 0.16
Total 268,746 100.00 73,745 100.00 25,119 100.00 180,798 100.00

Fig. 8. Liberation of the primary REE-containing mineral phase – Monazite.

The REE-deportment and elemental field maps indicate that the would have to be concentrated separately from allanite to allow
majority of the REE’s are contained within monazite, allanite and for a high grade product.
(to a lesser extent) ilmenite and apatite. The deportment reveals The liberation of monazite indicates that it is, generally very
that Ce and La behave in a similar manner. It may therefore be well liberated at a relatively coarse grind size of 100% passing
extrapolated that the LREE’s behave and propagate differently to 1 mm. Monazite is furthermore extremely well exposed at this
the HREE’s. Dysprosium may be used as an analogy or tracer for grind size. The monazite grains show no evidence of zonation
HREE’s in this case. A concentrate of monazite and allanite would at the rims or centre of the grains. They appear generally
be preferential for concentrating the majority of the REE’s into a homogenous, apart from alteration textures present in some of
saleable product. However, the two minerals show very different the grains. Possible flotation as well as densimetric separation
properties. Allanite is fine grained, with a relatively low SG and would therefore be viable possibilities for monazite concentra-
is easily weathered, while monazite is relatively hard, has a higher tion. If 100% of the monazite grains are recovered, this monazite
SG and is coarse grained, which would make total rare earth min- concentrate would contain in the order of 62–98% of the TREE’s
eral concentration highly problematic. Realistically monazite in the samples.

Please cite this article in press as: Smythe, D.M., et al. Rare Earth Element deportment studies utilising QEMSCAN technology. Miner. Eng. (2013), http://
dx.doi.org/10.1016/j.mineng.2013.03.010
10 D.M. Smythe et al. / Minerals Engineering xxx (2013) xxx–xxx

Fig. 9. Scanning Electron Microscope (SEM) electron micrograph, accompanied by a false colour composite image of the mapped particle showing, (A) liberated and well
exposed monazite, (B) middlings liberated monazite, and (C) locked, unexposed monazite. (For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)

5. Conclusion References

The QEMSCAN analysis forms an integral part of the REE- Anonymous, 2010. Rare Earth Elements. British Geological Survey, Natural
Environment Research Council.
deportment process. As demonstrated in the case study, the QEM- Boynton, W.V., 1984. Geochemistry of the rare earth elements: meteorite studies.
SCAN identified monazite, as well as previously overlooked allanite In: Henderson, P. (Ed.), Rare Earth Element Geochemistry. Elsevier, pp. 63–114.
as the main REE-bearing phases present in the studied samples. Castor, S.B., Hedrick, J.B., 2006. Rare Earth Elements, Industrial Minerals and Rock
Commodities, Markets and Uses, seventh ed. Society for Mining, Metallurgy
The identification of allanite influenced the outcome of the REE- and Exploration, Littleton, Colorado.
deportment in the order of 30%. In this case these results would Gupta, C.K., Krishnamurthy, N., 2005. Extractive metallurgy of rare earths. CRC
justify a lower than expected theoretical TREE recovery achieved Press.
Kerr, A., Rare-earth-element (REE) mineralization in labrador: a review of known
by metallurgical concentration. environments and the geological context of current exploration activity.
Mineralogy may often be overlooked in favour for extensive Newfoundland and Labrador Department of Natural Resources Geological
metallurgical testing. This often proves a costly affair as plant Survey, 2011, Report 11–1, pp. 109–143.
Rollinson, H., 1993. Using Geochemical Data: Evaluation, Presentation,
optimization may never be achieved without this essential
Interpretation. Pearson Education Limited.
information. QEMSCAN mineralogical investigations have proven Verplanck, P.L., Van Gosen, B.S., 2011. Carbonatite and alkaline intrusion-related
themselves to be cost effective and reliable by incorporating and rare earth element deposits – a deposit model. U.S Geological Survey.
integrating a broad range of data from numerous analytical tech-
niques, to provide a detailed understanding of the implications of
mineralogical characteristics on the processing of ore.

Please cite this article in press as: Smythe, D.M., et al. Rare Earth Element deportment studies utilising QEMSCAN technology. Miner. Eng. (2013), http://
dx.doi.org/10.1016/j.mineng.2013.03.010