Minerals Engineering 101 (2017) 3039

Contents lists available at ScienceDirect

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

Towards cleaner production Using flotation to recover monazite from a

heavy mineral sands zircon waste stream
E. Tranvik a,b,, M. Becker b,c, B.I. Plsson a, J.-P. Franzidis b, D. Bradshaw b
a
Minerals and Metallurgical Engineering, Lule University of Technology, Sweden
b
Minerals to Metals Initiative, Department of Chemical Engineering, University of Cape Town, Rondebosch, South Africa
c
Centre for Minerals Research, Department of Chemical Engineering, University of Cape Town, Rondebosch, South Africa

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

Article history: In line with the principles of cleaner production, the removal of monazite via reverse flotation was inves-
Received 1 March 2016 tigated with a view to reducing the radioactivity of a heavy mineral sands waste stream. Another benefit
Revised 8 October 2016 was to create a potential REE by-product from the Namakwa Sands operation in South Africa.
Accepted 15 October 2016
Understanding the mineralogy of the zircon waste stream was essential owing to the cemented nature
Available online 14 November 2016
of the deposit and the potential impact of surface coatings on the flotation performance. SEM,
QEMSCAN and optical microscopy showed that amorphous SiO2 was the most abundant surface coating
Keywords:
associated with both monazite and zircon, which is likely to constitute a major challenge in achieving
Mineralogy
Surface coatings
flotation selectivity. A D-optimal statistical screening design was applied to find the most relevant flota-
pH tion parameters and a full factorial design to find the optimal flotation conditions. The most promising
Mineral sands results showed that monazite could be successfully removed from the zircon waste with an oleate collec-
Design of experiment tor at pH 10. The selectivity was found to be highly dependent on pH, with no selectivity at pH 9 and no
Statistical analysis mineral flotation at pH 11. Further work is recommended to confirm and optimise these conditions and
Rare earth elements test them on a larger scale.
2016 Elsevier Ltd. All rights reserved.

1. Introduction minerals with gravity concentration, removing gangue minerals

The mining industry is continually challenged to achieve more
sustainable production with less negative environmental impact.
The traditional and most commonly used method of tailings dis-
posal is to store waste produced by mineral processing in large
unsightly dams. Some of the challenges linked to maintaining tail-
ing dams include the handling and storage of these large volumes,
as well as the slow settling of fine particles that can hinder the
recycling of water (Edraki et al., 2014; Wang et al., 2014). Hazar-
dous processing chemical residues and acid conditions caused by
oxidising sulphide minerals further complicate water recycling
and are a risk for the environment if released without treatment.
Alternatives to conventional tailing storage facilities are paste
and thickened tailings that can be deposited on the surface or as
backfill underground. It is also possible to implement the reuse,
recycling and reprocessing of tailings. The review paper by Edraki
et al. (2014) provides examples of several attempts to reprocess
tailings, such as using waste in brick making, separating valuable
Corresponding author at: Minerals and Metallurgical Engineering, Lule
University of Technology, Sweden.
E-mail address: elin.tranvik@gmail.com (E. Tranvik).
by means of flotation and using leaching or bioleaching of tailing
deposits to extract metals. Reducing mineral processing waste
would improve sustainability in the mining industry; thus, a
greater focus on sustainable development in respect to the reuse
of mining and mineral processing waste should be considered
(Bian et al., 2012). Waste reprocessing is in line with cleaner pro-
duction, which is an initiative to increase product efficiency and
minimise waste and pollution (not to be confused with the conven-
tional use of cleaner circuits in the context of flotation). Cleaner
production is used in many industries including the mining indus-
try (Hilson, 2000, 2003; Silvestre and Silva Neto, 2014) and is part
of the United Nations Environment Programme.
In the processing of heavy mineral sands, handling and disposal
of radionuclides is a concern (World Nuclear Association, 2015). In
particular, the thorium and uranium containing mineral monazite
is one of the main contributors to radioactivity and can be haz-
ardous when concentrated in the products and wastes of heavy
mineral sands processing. Monazite ([Ce,La,Nd,Th]PO4), however,
is also well known to be one of the main host minerals of the light
rare earth elements (LREE), for which there is currently a signifi-
cant demand for use in cell phones, laptops, catalysts for fuel pro-
duction and green technology in wind power turbines (Hurst,

http://dx.doi.org/10.1016/j.mineng.2016.10.011
0892-6875/ 2016 Elsevier Ltd. All rights reserved.
 E. Tranvik et al. / Minerals Engineering 101 (2017) 3039
2010; Jordens et al., 2013). Therefore, opportunity exists to recover
the valuable rare earth elements (REE) as a by-product, while
simultaneously reducing the environmental risk of the tailings.
Valuable heavy minerals such as zircon (ZrSiO4), rutile (TiO2)
and ilmenite (FeTiO3) are generally separated from the gangue
minerals in heavy mineral sands deposits using a combination of
physical separation techniques. Ilmenite is separated based on its
magnetic properties, and thereafter zircon, rutile and monazite
are separated based on their conducting and magnetic properties
(Jones, n.d). Zircon and monazite are both non-conductors and
must be separated using their magnetic properties. One of the chal-
lenges in a heavy mineral separation plant is to produce a clean zir-
con concentrate without the monazite that has reported to the
cleaning circuit. This frequently means sacrificing zircon recovery
by producing a waste stream with monazite and high zircon grade.
This is particularly so in the fine sizes: wet gravity and electrostatic
separation methods are efficient for coarse particle sizes, but gen-
erally fail to recover particles <100 lm (Wills and Napier-Munn,
2006). Consequently, waste streams from cleaner electrostatic sep-
aration circuits may have a fine size fraction containing valuable
minerals such as zircon, rutile and monazite. One of the few min-
eral separation techniques that have an operating window which is
ideal for particles in this size range is flotation.
The successful separation of monazite from zircon using flota-
tion has been reported by several authors (Bruckard et al., 1999;
Cheng et al., 1993; Cuthbertson, 1951; Jordens et al., 2015; Pavez
and Peres, 1993; Ren et al., 2000). Cuthbertson (1951) secured a
patent for monazite flotation from heavy minerals such as wol-
framite, rutile and cassiterite using boiled starch as the depressant
and an amine collector. Thereafter, Abeidu (1972) studied the sep-
aration of monazite from zircon using experiments in a Hallimond
tube with an oleic acid collector and Na2S as a monazite activator.
Pavez and Peres (1993) more specifically studied the monazite-
zircon-rutile flotation system. Sodium oleate and hydroxamate col-
lectors were used, with metasilicate to depress zircon and rutile.
Both collectors showed promising results in the Hallimond tube
experiments. The best reagent regimes were subsequently
repeated using bench scale flotation by Pavez and Peres (1994)
resulting in both high monazite recovery and good depression of
rutile and zircon. The flotation of monazite from a zircon flotation
tail was performed by Bruckard et al. (1999), and reverse flotation
of a titanium-rich product was studied by Bruckard et al. (2001) by
floating monazite and zircon. However, few attempts to separate
monazite from zircon were made in either of the studies by Bruck-
ard et al. A recent paper by Kumari et al. (2015) noted the lack of
any common view of a standard reagent regime for monazite flota-
tion and highlighted the need for a better understanding of the
physio-chemical properties of monazite.
Apart from understanding the physio-chemical properties of
monazite, the ability to investigate selected factors influencing
the flotation system is also needed. Of interest in this study is
the role of process mineralogy and statistical design. The under-
standing of mineralogy is important to gain knowledge about the
valuable and deleterious element deportment, bulk mineralogy,
mineral liberation and association, grain size distribution and char-
acteristics of any well-developed coatings on the mineral surfaces.
Long standing techniques such as optical microscopy (Jones, 1987),
as well as the more modern techniques which are based on the
Scanning Electron Microscope (SEM) coupled with energy disper-
sive X-ray spectroscopy, provide a useful platform to investigate
these features in ores and processing streams (Fandrich et al.,
2007; Lamberg and Rosenkranz, 2014; Restarick and Gottlieb,
1991).
Similarly, the design of experiment (DoE) approach is essential
to control and evaluate the factors that influence flotation perfor-
31
mance. The use of DoE allows for objective conclusions to be drawn
due to the statistical methods that are used in the experimental
methodology (Antony, 2003). The full factorial design is an exper-
imental method where tests with every possible combination of
the chosen factors and levels are carried out. The advantages of fac-
torial design are many (Napier-Munn, 2014): it allows for the
determination of the linear interactions between variables, it esti-
mates the main effects efficiently as every test includes informa-
tion about all factors, and it provides a good estimation of error.
The objective of this paper is to investigate the mineralogy and
the potential to separate monazite from a terminal zircon waste
stream by flotation with a case study from the Namakwa Sands
heavy mineral deposit in South Africa. The attempt to produce a
monazite product enriched in La, Ce, Nd and radioactive compo-
nents, and a zircon product that potentially meets the specifica-
tions of an additional low grade zircon product (<0.5% TiO2,
1000 ppm U and Th) brings the heavy mineral sands industry
one step closer to achieving cleaner production.
2. Material and methods
2.1. Namakwa Sands
The Namakwa Sands operation, owned by Tronox Limited, cur-
rently upgrades the heavy mineral sands deposit to produce rutile
(TiO2), ilmenite (FeTiO3) and zircon (ZrSiO4) concentrates that are
processed and separated based on density, magnetic and conduct-
ing properties using wet tables and spirals, and magnetic and elec-
trostatic separation. The annual production capacity is 21 Mt run-
of-mine ore (Philander and Rozendaal, 2013).
The deposit consists of 8 wt.% heavy minerals (Philander and
Rozendaal, 2013) comprising ilmenite, rutile, zircon, leucoxene,
garnet and pyroxene as the dominant heavy minerals. The deposit
is divided into two major ore bodies, the Graauwduinen West and
Graauwduinen East deposits. The East deposit is a typical, loosely
consolidated minerals sands deposit, whereas the atypical West
deposit consists of heavy minerals in a cemented matrix (known
as cemented hard layers; Philander and Rozendaal, 2013). Conse-
quently, ore from the West deposit needs comminution to liberate
the valuable minerals, as well as attritioning and hot acid leaching
to remove surface coatings in order to achieve better separation.
The most common surface coatings in the operation are clay col-
loidal coatings: sepiolite, sepiolite with calcite, calcite, apatite
and opaline silica. These surface coatings, which affect mineral sur-
face properties, are likely to have an impact on flotation
performance.
Monazite is one of the gangue minerals and varies between
0.08% and 0.36% of the total heavy mineral content of the deposits.
Its composition is presented in Table 1. The monazite at Namakwa
Sands is enriched in the LREE, lanthanum, cerium and neodymium.
The mineral chemistry also indicates that monazite is a significant
Table 1
Average major oxide chemistry (wt.%) of mon-
azite in the Namakwa Sands deposit (Philander
and Rozendaal, 2009).
Monazite
SiO2 8.72
CaO 2.66
P2O5 32.03
La2O3 15.15
Ce2O3 32.20
Nd2O3 9.10
ThO2 2.07
32 E. Tranvik et al. / Minerals Engineering 101 (2017) 3039

carrier of thorium, and therefore any stream concentrated in mon-
azite will have a degree of radioactivity.
2.2. Methodology
During a dedicated sampling campaign at Namakwa Sands, an
8000 ton terminal zircon waste stream stockpile was sampled. Fol-
lowing a series of subsampling, 425 g representative samples were
prepared for characterisation and flotation test work and an addi-
tional subsample was screened to determine the particle size dis-
tribution. All flotation test work was conducted on-site.
This section presents the flotation procedure and the factor set-
tings used in the statistical designs (carried out in three phases)
and the procedure for mineralogical evaluation. The terminal zir-
con waste stream is presented as flotation feed from this point
forward.
2.2.1. Flotation procedure
The effects of physical and chemical modifications on the flota-
tion of monazite were investigated using a statistical designs.
MODDE version 10 (www.umetrics.com) is the statistical software
that was used for the design of experiments and statistical analysis.
In the three experimental phases, three centre points were used for
reproducibility and correlations. The physical modification of the
pulp was focused on trying to remove the surface coatings from
the flotation feed to allow for maximum selectivity between min-
erals based on fresh surfaces. Initially ultrasonication was used,
followed by attritioning in a later stage. Phase 1 investigated two
collector types, collector dosage, two depressant types, depressant
dosage, pH and ultrasonication in a D-optimal screening design.
Sodium oleate and hydroxamate collectors and dosages were cho-
sen based on the findings of Pavez and Peres (1993, 1994) as well
as the choice of metasilicate as a depressant. Cyquest4000 was
used as a second depressant on the recommendation of Cytec for
its properties similar to metasilicate. The lower limit of depressant
dosage was chosen based on recommended Cyquest4000 dosage
by Cytec (C. Lotter, Pers. Comm, 2014) and the upper depressant
dosage was based on the findings by Pavez and Peres (1993,
1994). The pH range was chosen based on the neutral pulp pH
(lower) and the optimal conditions for separating monazite from
zircon and rutile found by Pavez and Peres (1993, 1994) (higher).
The use of ultrasonication as a factor in the screening design was
based on prior knowledge of the presence of surface coatings fol-
lowing the consolidated state of the ore deposit (Philander and
Rozendaal, 2013). Other factors such as conditioning time, reagent
addition order and air rate were not part of the screening design as
these factors were suspected of having less impact than the factors
chosen for the design. Literature has reported slow adsorption
kinetics of hydroxamate collectors (Sreenivas and Padmanabhan,
2002), however the conditioning times applied in this project were
based on preliminary testwork on the same mineral deposit
(Naud and Liu, 2012; Tilbury and Harley, 2013) and communica-
tion with the reagent supplier. In Phase 2:1, the most successful
conditions from Phase 1 were used as the centre point in a full fac-
torial design investigating pH, collector dosage and depressant
dosage. Phase 2:2 continued to investigate collector dosage and
attritioning intensity in another full factorial design. The condi-
tions and factor levels are presented in Table 2. Local tap water
was used in all tests.
A top stirred modified Leeds 3L cell was used in all experiments.
Ultrasonication or attritioning were carried out in the cell, before
pH adjustment (neutral pH was 5.5) and collector, depressant
(and frother) were added to the pulp. The conditioning times were
5 min for collector and depressant and 1 min for frother. The pH
was adjusted again after reagent addition, before the air was
turned on. Four concentrates were collected at 1, 3, 7 and 13 min
by scraping the froth into collecting pans every 15 s. All reagents,
except metasilicate (Fluka Chemie AG), were supplied by Cytec
Chemicals.
Two types of equipment were used for surface cleaning: an
ultrasonic probe in Phase 1 and a mini-pin mill for attritioning in
Phases 2:1 and 2:2. The ultrasonic probe was a Sonic Rupture
400 and was used at 80% of its 300 W capacity. The mini-pin mill
was manufactured at the University of Cape Town, and consisted
of a 1.1 L stainless steel sampling cup and a mechanical stirrer that
was fitted with six pins. Zirconia beads (1.5 mm diameter) were
used for attritioning.
During the flotation tests, water and mass recoveries were
recorded. In Phase 1, the efficiency was measured with chemical
assays: the ZrO2 and P2O5 contents were used to approximate
the zircon and monazite grade and recoveries, based on ZrO2 being
the main component of zircon (ZrSiO4) and P2O5 as one of the main
components of monazite (Table 1). In Phase 2, the separation effi-
ciency was measured by optical microscopy grain-counting (insuf-
ficient mass was available for chemical analysis), and thus results
from Phase 2:1 and 2:2 represent real zircon and monazite
grades and recoveries.
2.2.2. Mineralogical characterisation
A combination of analytical methods was used to characterise
the flotation samples in this study: namely, chemical assays using
X-ray fluorescence spectrometry (XRF), grain counting using opti-
cal microscopy, mineral-morphological characterisation using
Scanning Electron Microscopy coupled with Energy Dispersive

Table 2
Flotation factors, levels and conditions for Phase 1, Phase 2:1 and Phase 2:2 factorial designs. The factor levels are written in bold and italic, whereas the other factors are constant.

Factor Unit Phase 1 Phase 2:1 Phase 2:2

Collector Alkyl hydroxamate (acid), Sodium oleate Sodium oleate Sodium oleate
Collector dosage g/ton 180500 180250 215315
Depressant Metasilicate, Cyquest4000 Cyquest4000 Cyquest4000
Depressant dosage g/ton 1001200 8001200 800
pH 5.510 911 10
Ultrasonication Yes, No
Attritioning intensity RPM 400 400600
Air flowrate L/min 5 4 4
Conditioning impeller speed RPM 1800 1400 1400
Flotation impeller speed RPM 1800 1800 1800
Conditioning pulp density wt.% solids 13 35 35
Flotation pulp density wt.% solids 13 13 13
Froth height cm 2 2 1
Collector addition Single stage Multi-stage Multi-stage
Ultrasonication time seconds 40
Attritioning time minutes 20 20
 E. Tranvik et al. / Minerals Engineering 101 (2017) 3039 33

Spectrometry (SEM-EDS), and quantitative mineralogical charac-
terisation using Quantitative Evaluation of Minerals by Scanning
Electron Microscopy (QEMSCAN). XRF (ALR 900 XRF spectrometer)
analysis and manual grain counting (363 grains, i.e. 3 fields, per
sample) were done in-house at Namakwa sands. An FEI Nova
NanoSEM 230, with Oxford X-Max detector and INCA software,
was used for SEM-EDS analysis at the University of Cape Town.
QEMSCAN analysis was carried out by EXXARO Services using a
ZEISS Evo 50 SEM platform with four light element Bruker Xflash
Silicon Drift energy dispersive X-ray detectors. The iDiscover soft-
ware package was used for QEMSCAN data processing to further
characterise and quantify the mineral surface coatings.
Rutile and ilmenite, both valuable minerals, were the second and
third biggest components. The QEMSCAN analysis identified three
phosphorous-bearing minerals (monazite, florencite and apatite
respectively) from which the P deportment could be calculated
indicating that 95% of the P is hosted by monazite. The visual
characteristics of the sample are illustrated by the optical micro-
scope photograph in Fig. 1a, and the SEM photomicrograph in
Fig. 1b, which shows the coatings on the zircon particles. EDS anal-
ysis of the zircon surface coatings indicated their composition was
SiO2, which is likely to be amorphous opaline material rather than
crystalline quartz. The d50 and d80 of the flotation feed were
106 lm and 133 lm respectively.

3. Results

The results are divided into four sections: characterisation of 3.2. Screening of factors influencing monazite selectivity (Phase 1)
the flotation feed, screening of the factors affecting monazite flota-
tion selectivity (Phase 1), monazite flotation optimisation (Phase 2) The following factors were investigated in this screening stage:
and finally mineralogical characterisation of the flotation products. collector type, collector dosage, depressant type, depressant
Outcomes from the statistical evaluation as well as flotation results dosage, ultrasonication and pH. The summary of experimental con-
are presented for each phase in the dedicated section. ditions and flotation results is given in Table 4.
The screening flotation tests in Phase 1 yielded a broad range of
results with mass recovery varying from being almost negligible up
3.1. Characterisation of the flotation feed
to 80% recovery. The statistical evaluation showed that the most
significant factors (main effects) for P2O5 recovery and grade
The main component of the flotation feed was zircon, as con-
(P2O5 used as a proxy for monazite recovery) were collector type,
firmed by the QEMSCAN mineralogical characterisation (Table 3).
collector dosage and pH (Fig. 2) (further details on the statistical
Table 3 evaluation can be found in Tranvik, 2014). Several of the flotation
Mineralogical composition of the flotation feed tests with almost negligible mass recovery delivered a yellow pro-
(in wt.%) comprising the flotation feed as
duct that differed in colour from the otherwise brown-coloured
determined by QEMSCAN. Others comprises
gangue minerals such as other TiO2-minerals,
concentrates. Subsequent optical microscopy of these yellowish
kaolinite and quartz. concentrates indicated they were enriched in monazite, providing
the first real indication of selective separation of monazite (insuf-
Mineral Wt.%
ficient concentrate mass was obtained for XRF analysis). The com-
Zircon 82.5 mon factors in these tests with promising selectivity (highlighted
Rutile 6.4
Monazite 3.2
in grey in Table 4) were the oleate collector at a dosage of
Anatase 2.3 180 g/ton and a pH of 10. These conditions, which were more
Garnet 1.6 based on visual observations than on the outcome of the statistical
Ilmenite 0.8 evaluation became the basis for future tests (Phase 2:1, 2:2). It
Pyroxene 0.7
should be noted that the statistical evaluation was ineffective in
Leucoxene 0.4
Fe-oxides 0.1 finding the best monazite selectivity, P2O5 recovery and grade
Others 2.0 could not be determined, as the inputs for P2O5 recovery and grade
in the negligible mass tests was effectively zero.

Fig. 1. (a) Optical microscope photograph showing typical heavy minerals and their characteristics in the zircon waste sample (flotation feed). Monazite occurs as well
rounded, slightly opaque grains with a yellowish colour. (b) SEM photomicrograph of the zircon waste sample (flotation feed) with patchy, fluffy coverage of bright white
zircon with dark grey SiO2 (amorphous opaline silica). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this
article.)
34 E. Tranvik et al. / Minerals Engineering 101 (2017) 3039

Table 4
Summary of the experimental conditions in Phase 1 and responses for mass recovery. The conditions found to potentially be selective for monazite are written in italics. Indicates
tests used as centre points in the statistical design and used to estimate error.

Order Collector Collector dosage (g/ton) Depressant Depressant dosage (g/ton) pH Ultrasonication Mass recovery (%)
1 Oleate 180 Cyquest4000 100 5.5 Yes 28.6
2 Oleate 500 Metasilicate 1200 10 No 51.6
3 Hydroxamate 180 Metasilicate 100 10 Yes 70.3
4 Oleate 180 Metasilicate 1200 10 Yes 0.1
5 Hydroxamate 180 Metasilicate 1200 10 No 68.9
6 Hydroxamate 180 Cyquest4000 1200 5.5 Yes 39.1
7 Hydroxamate 340 Cyquest4000 650 7.75 Yes 67.9
8 Hydroxamate 500 Metasilicate 100 10 No 52.4
9 Oleate 180 Metasilicate 100 5.5 No 44.3
10 Oleate 500 Cyquest4000 100 5.5 No 86.4
11 Oleate 180 Metasilicate 1200 5.5 Yes 38.5
12 Oleate 500 Metasilicate 1200 5.5 No 42.9
13 Hydroxamate 340 Cyquest4000 650 7.75 Yes 67.5
14 Hydroxamate 500 Cyquest4000 1200 10 No 63.1
15 Hydroxamate 180 Cyquest4000 100 5.5 No 24.3
16 Oleate 180 Cyquest4000 1200 10 No 0.5
17 Oleate 180 Cyquest4000 100 10 Yes <0.1
18 Oleate 180 Cyquest4000 1200 5.5 No <0.1
19 Hydroxamate 180 Metasilicate 1200 5.5 No 32
20 Hydroxamate 500 Metasilicate 1200 10 Yes 63.1
21 Hydroxamate 180 Metasilicate 100 5.5 Yes 34.8
22 Oleate 500 Metasilicate 100 5.5 Yes 22.8
23 Hydroxamate 180 Cyquest4000 1200 10 Yes 68.7
24 Hydroxamate 500 Cyquest4000 100 10 Yes 58.7
25 Hydroxamate 500 Metasilicate 100 5.5 No 37.6
26 Oleate 180 Metasilicate 100 10 No <0.1
27 Oleate 500 Cyquest4000 1200 10 Yes 76.3
28 Oleate 500 Metasilicate 100 10 Yes 49.9
29 Hydroxamate 180 Cyquest4000 100 10 No 61.7
30 Oleate 500 Cyquest4000 100 10 No 54.1
31 Hydroxamate 500 Cyquest4000 100 5.5 Yes 39.2
32 Oleate 500 Cyquest4000 1200 5.5 Yes 29.8
33 Hydroxamate 340 Cyquest4000 650 7.75 Yes 69.3
34 Hydroxamate 500 Cyquest4000 1200 5.5 No 38.9
35 Hydroxamate 500 Metasilicate 1200 5.5 Yes 49.8

Fig. 2. Effects plot from statistical evaluation in MODDE, presenting the effects in the reduced MLR (Multiple Linear Regression) model. CDo = Collector dosage,
Col = Collector, SO = Sodium Oleate, Dep = Depressant, DDo = Depressant dosage.

3.3. Optimising flotation conditions for monazite selectivity (Phase 2)
More intense attritioning was carried out in Phase 2, since no
effect of ultrasonication was noted in the statistical evaluation of
Phase 1. The intention was to polish the mineral surfaces to
remove the amorphous SiO2 coatings illustrated in Fig. 1b, result-
ing from the consolidated nature of the heavy mineral deposit. A
narrower pH interval (pH 911), one type of collector (oleate)
and one type of depressant (Cyquest4000) were used during Phase
2. The aim was to selectivity separate monazite, with the most
promising conditions in Phase 1. The mineral grades and recoveries
reported in this section are calculated from optical microscopy
grain counting.
3.3.1. Finding the optimum pH for monazite selectivity (Phase 2.1)
The pH had a strong effect on flotation performance, especially
on the appearance of the froth (Fig. 3). At pH 9, the froth was
brown-coloured, heavily loaded, very stable and mobile. At pH
 E. Tranvik et al. / Minerals Engineering 101 (2017) 3039
10, the froth was of a brighter yellow colour and less loaded.
Finally, at pH 11, a very mobile foam was obtained. The mass
recoveries varied from close to zero at pH 11 to 6070% at pH 9
(Table 5). The golden brown colour that was obtained in Phase 1
at pH 10 was again repeated in Phase 2:1; optical microscopy con-
firmed that monazite selectivity was high in this centre point
(grades as high as 66% were obtained).
No significant improvement in monazite grade could be seen
when varying depressant dosage (Fig. 4), which may indicate that
the reagent was not acting as a depressant. According to the
supplier, Cyquest4000 can also be used as a dispersant to control
colloidal and fine particles during flotation. As no significant effect
of depressant dosage was observed, a fixed dosage was used in
Phase 2:2.
3.3.2. Optimising collector dosage and investigation of attritioning
intensity (Phase 2:2)
All tests during Phase 2:2 were performed at pH 10 with the
intention to increase monazite recovery by varying collector
dosage and attritioning intensity. From the findings of Phase 2:1,
the depressant dosage was fixed at 800 g/ton. Monazite was sepa-
rated selectively by flotation in all tests (Fig. 5), with up to a 54%
recovery (Table 6). The mass and water recoveries generally
increased with a higher collector dosage, however, the effect of
attritioning intensity was unclear. The mass recovery increased
with attritioning intensity at a high collector dosage, whereas, no
difference could be seen at low collector dosage (Fig. 6).
3.4. Characterisation of flotation products
The bulk mineralogy and effect of attritioning on the surface
coatings was studied with SEM and quantified by QEMSCAN anal-
ysis of the feed sample, three concentrates, and six tailing products
from Phase 2. The bulk mineralogy determined by QEMSCAN lar-
gely coincided with earlier results based on optical microscopy
grain counting. The SEM images gave an indication of the nature
of the surface coatings, which were shown by point EDS analysis
to consist of SiO2 (Fig. 1b). The QEMSCAN coating data is given
for premium zircon (the most common quality of zircon in the
samples) in Fig. 7. The coating data showed that more than 80%
of the zircon was clean with only a minimum coating. As expected,
the feed sample without prior attritioning was the most coated,
although the variation between the samples is small (Fig. 7). The
Fig. 3. Froth photos from the three pH levels during Phase 2:1 varying through no selectivity at pH 9, monazite selectivity at pH 10, and only foam at pH 11.
35
QEMSCAN analysis showed that quartz was the main coating on
the particles, confirming what was shown in the EDS result. Based
on previous knowledge of the deposit, this coating is more likely to
be amorphous opaline silica (QEMSCAN is unable to determine
crystallinity).
The most common minerals associated with monazite were
shown to be zircon, kaolinite and quartz (Fig. 8). The QEMSCAN
results shown in Fig. 8 also illustrates that the degree of liberation
for monazite is higher in the flotation concentrates than in the tail
samples, which is expected as the liberated monazite particles
should be more easily recovered.
4. Discussion
As reported in literature (Pavez and Peres, 1993, 1994), it is pos-
sible to separate monazite from zircon using an oleate collector.
Anionic and cationic collectors selectively adsorb onto charged
mineral surfaces and since pH has the largest influence on mineral
surface charge, it is clear that determining the optimum pH for
selective flotation is paramount. The narrow pH interval that
favoured monazite selectivity in this test work indicates that the
mechanism for flotation separation is likely a difference in mineral
surface charge.
The oleate collector used was anionic and therefore adsorbed
well onto minerals with positive surface charge. The sharp drop
in zircon recovery between pH 9 and 10 suggests that the point
of zero charge for zircon occurs in this range. The same observation
for monazite was seen between pH 10 and 11, which indicates that
the point of zero charge for monazite is in this range. The drop in
recovery could be explained by the change in mineral surface
charge where at pH 9 both monazite and zircon surfaces are posi-
tively charged and no selectivity is obtained. Flotation carried out
at a pH of 10 is selective to monazite which might be a result of
positive monazite mineral surfaces, whereas zircon has passed its
point of zero charge. At pH 11, neither mineral is collected, indicat-
ing that mineral surfaces have negative charge. Although the mon-
azite selectivity can be explained by charge-induced flotation,
more test work is necessary to prove this hypothesis. Oleate collec-
tors are known to chemisorb onto mineral surfaces (Chelgani et al.,
2015), even though physical adsorption of oleate collectors onto
monazite surfaces has been discussed in the literature (Pavez
et al., 1996). According to Cheng (2000), who summarised what
has been reported in literature, monazite has a point of zero charge
36 E. Tranvik et al. / Minerals Engineering 101 (2017) 3039

Table 5
Flotation conditions and results for Phase 2.1. Collector was Oleate and the depressant was Cyquest4000. Monazite recovery and grade were determined by optical microscopy
grain counting. Indicates tests used as centre points in the statistical design and used to estimate error.

Run order Collector dosage (g/ton) Depressant dosage (g/ton) pH Mass recovery (%) Monazite recovery (%) Monazite grade (%)
P21 180 1200 9 10.3 7 4
P22 250 1200 11 0.1 <0.1 4
P23 180 800 11 <0.1 <0.1 5
P24 250 800 9 66.6 31 3
P25 215 1000 10 2.5 28 66
P26 250 800 11 0.4 1 5
P27 180 800 9 70.0 45 4
P28 215 1000 10 3.4 32 57
P29 180 1200 11 0.1 <0.1 4
P210 250 1200 9 64.4 46 4
P211 215 1000 10 2.8 20 43

Fig. 4. Mass and water recoveries for flotation tests in Phase 2:1. Confidence limits were calculated from the centre points, using an alpha of 0.1.

the point of zero charge of a mineral. In this particular deposit,

Fig. 5. Graph of zircon versus monazite recovery used to illustrate the selectivity in
the flotation tests in Phase 2:2. The shaded area represents the desired operating
window with good monazite selectivity and high recovery. The line y = x indicates
zero selectivity. The monazite and zircon recoveries were determined by optical
microscopy grain counting.
between pH 1 and 9. The point of zero charge for zircon is around
pH 6 (Mao et al., 1994). However, since every ore body is different,
surface coatings or the unique mineral composition might affect
the presence of surface coatings such as amorphous silica are
thought to be the reason for the change in surface charge from
what has been presented in the literature.
Mineralogy was used throughout the test work to analyse the
grade of samples due to the small sample size and nature of the
mineral. SEM and QEMSCAN analysis indicated that there were
surface coatings present on monazite which would have a negative
impact on recovery. The method is believed to have the potential of
being further optimised to increase monazite recovery without
sacrificing selectivity, and move into the desired operating win-
dow in Fig. 5. Attempts at sample attritioning prior to flotation
were fairly unsuccessful as there was only a small decrease in sur-
face coating evident in the QEMSCAN analyses. If more rigorous
attritioning could be carried out with a more suitable mill, the
degree of surface coating could be decreased further. With better
exposed fresh mineral surfaces, a higher monazite recovery may
be possible, but the significance and methodology of surface coat-
ing removal remains unknown. The use of additional surface speci-
fic analytical techniques such as XPS or ToF-SIMS are
recommended to confirm this, and to further understand the role
of attritioning. Mineralogy is therefore an essential tool in deter-
mining the nature and amount of surface coatings on minerals in
order to potentially enhance monazite recovery.
 E. Tranvik et al. / Minerals Engineering 101 (2017) 3039 37

Table 6
Flotation conditions and results from Phase 2:2. Collector was Oleate and the depressant was Cyquest4000 at 800 g/t. Indicates tests used as centre points in the statistical design
and used to estimate error. All mineral grades and recoveries have been determined by optical microscopy grain counting.

Run order Collector dosage (g/ton) Attrition intensity (RPM) Mass recovery (%) Monazite recovery (%) Monazite grade (%) Zircon recovery (%)
P221 215 400 2.3 24 62 0
P222 215 600 2.4 28 71 0
P223 315 400 5.7 51 54 0
P224 265 500 5.0 42 51 0
P225 265 500 4.6 35 48 0
P226 315 600 8.6 54 39 2
P227 265 500 5.4 48 54 1

Fig. 6. Mass and water recoveries for flotation tests in Phase 2:2. Confidence limits were calculated from the centre points, using an alpha of 0.1.

Fig. 7. Coating data for premium zircon (the most common zircon type). Columns are sorted by increasing attritioning intensity and type of sample. From left: feed, P21
conc, P223 conc, P226 conc, P222 tail, P226 tail. QEMSCAN categories as follows - uncoated:<5% coated surface area, minimum coated: >5 and 620% coated surface area,
moderately coated: >20 and 640% coated surface area, highly coated: >40% coated surface area.

5. Conclusions is likely to be dependent on surface charge since the variation of

mineral floatability has been shown to vary profoundly with pH.
Monazite can be selectively separated from zircon using an ole- SiO2 coatings were detected in the flotation feed, most likely as
ate collector for flotation at a pH of 10. The separating mechanism opaline silica that originated from the cemented nature of the
38 E. Tranvik et al. / Minerals Engineering 101 (2017) 3039
Fig. 8. Monazite liberation and surface association as determined by QEMSCAN (liberation defined as % of exposed 2D particle perimeter) for the flotation feed, concentrate
and tail samples. Association to quartz most likely as opaline silica. Association with others represents minerals with <3% surface association to monazite.
deposit. It could not be shown with statistical significance that the
surface coatings did have an effect on monazite recovery in this
study. Nevertheless, an understanding of the ore mineralogy is
important since the coatings do affect the surface properties of
the minerals, and are therefore likely to influence flotation
performance.
The use of a statistical design approach assisted in concluding
which the most promising conditions were, and proving the ability
to evaluate these by presenting statistically sound results with
confidence.
Selective monazite flotation is a promising step towards reduc-
ing hazardous waste at Namakwa Sands by producing two possible
by-products - a potential REE feedstock, and a low quality zircon
product - demonstrating the principles of cleaner production. Fur-
ther work is recommended to confirm and optimise these condi-
tions and test them on a larger scale alongside detailed
quantitative mineralogy using auto-SEM-EDS (e.g. QEMSCAN).
Acknowledgements
Grateful thanks go to staff at Namakwa Sands for their support
of this research. Thanks to Professor Aubrey Mainza for his assis-
tance in manufacturing the pin mill and to Cytec for the providing
of flotation reagents. This work is based on research supported in
part by the South African Research Chairs Initiative (SARChI Chair
in Minerals Beneficiation) of the Department of Science and Tech-
nology and National Research Foundation of South Africa. Any
opinion, finding, conclusion or recommendation expressed in this
material is that of the authors and the NRF does not accept any lia-
bility in this regard.
References
Abeidu, A.M., 1972. The separation of monazite from zircon by flotation. J. Less-
Common Metals 29, 113119.
Antony, J., 2003. Design of Experiments for Engineers and Scientists. Butterworth-
Heinemann, Jordan Hill.
Bian, Z., Miao, X., Lei, S., Chen, S., Wang, W., Struthers, S., 2012. The challenges of
reusing mining and mineral-processing wastes. Science 337, 702.
Bruckard, W.J., Creed, M.D., Guy, P.J., Heyes, G.W., 1999. Beneficiation of Australian
mineral sands deposits using flotation. In: Proceedings of the Murray Basin
Mineral Sands Conference. Mildura, Australia, 2123 of April 1999, pp. 111
115.
Bruckard, W.J., Heyes, G.W., Guy, P.J., 2001. Flotation and screening recovery of
titanium minerals from monazite mineral sands circuit. Aust. Inst. Min.
Metallurgy Bulletin 306 (1), 1121.
Cheng, T.W., Holtham, P.N., Tran, T., 1993. Froth flotation of monazite and xenotime.
Miner. Eng. 6 (4), 341351.
Cheng, T.W., 2000. Technical note: the point of zero charge of monazite and
xenotime. Miner. Eng. 13 (1), 105109.
Chelgani, S.C., Rudolph, T., Leistner, T., Gutzmer, J., Peuker Urs, A., 2015. A review of
rare earth minerals flotation: monazite and xenotime. Int. J. Min. Sci. Technol.
25, 877883.
Cuthbertson, R.E., 1951. Froth flotation of monazite from heavy gravity minerals. US
Patent Number 2,610,738.
Edraki, M., Baumgartl, T., Manlapig, E., Bradshaw, D., Franks, D.M., Moran, C.J.,
2014. Designing mine tailings for better environmental, social and
economic outcomes: a review of alternative approaches. J. Clean. Prod.
84, 411420.
Fandrich, R., Gu, Y., Burrows, D., Moeller, K., 2007. Modern SEM-based mineral
liberation analysis. Int. J. Miner. Process. 84 (14), 310320.
Hilson, G., 2000. Barriers to implementing cleaner technologies and cleaner
production (CP) practices in the mining industry: a case study of the
Americas. Miner. Eng. 13 (7), 699717.
Hilson, G., 2003. Defining cleaner production and pollution prevention in the
mining context. Miner. Eng. 16, 305321.
Hurst, C., 2010. Chinas Rare Earth Elements Industry: What Can the West Learn?
Institute for the Analysis of Global Security (IAGS).
Jones, G. (n.d). <http://www.iluka.com/docs/company-presentations/mineral-
sandsan-overview-of-the-industry-by-greg-jones-manager-development-
geology.pdf>. Accessed 2016-09-24.
Jones, M.P., 1987. Applied Mineralogy: A Quantitative Approach. Graham and
Trotman, London.
Jordens, A., Cheng, Y.P., Waters, C., 2013. A review of the beneficiation of rare earth
element bearing minerals. Miner. Eng. 41, 97114.
Jordens, A., Marion, C., Grammatikopoulos, T., Waters, C., 2015. Beneficiation of the
Nechalacho rare earth deposit: froth flotation. In: proceedings MEI Flotation 15,
CapeTown, South Africa, 2015.
Kumari, A., Panda, R., Jha, M.K., Kumar, J.R., Lee, J.Y., 2015. Process development to
recover rare earth metals from monazite mineral: a review. Miner. Eng. 79,
102115.
Lamberg, P., Rosenkranz, J., 2014. Systematic diagnosis of flotation circuit
performance based on process mineralogical methods. In: Proceedings of 14th
International Mineral Processing Symposium Kusadas, Turkey, 1517
October, 2014.
Mao, M., Fornasiero, D., Ralston, J., Smart, R.S.C., Sobieraj, S., 1994. Electrochemistry
of the zircon-water interface. Colloids Surf. A: Physiochem. Eng. Aspects 85, 37
49.
Napier-Munn, T.J., 2014. Statistical Methods for Mineral Engineers: How to Design
Experiments and Analyse Data. Julius Kruttschnitt Mineral Research Centre,
Queensland, Australia.
Naud, N., Liu, B., 2012. Heavy Mineral Flotation: Namakwa Sands Undergraduate
Research Project. Department of Chemical Engineering, University of Cape
Town.
Pavez, O., Peres, A.E.C., 1993. Effect of sodium metasilicate and sodium sulphide on
the floatability of monazite-zircon-rutile with oleate and hydroxamates. Miner.
Eng. 6 (1), 6978.
Pavez, O., Peres, A.E.C., 1994. Technical note: bench scale flotation of a Brazilian
monazite ore. Miner. Eng. 7 (12), 15611564.
Pavez, O., Brandao, P.R.G., Peres, A.E.C., 1996. Technical note: adsorption of
oleate and octyl-hydroxamate onto rare-earths minerals. Miner. Eng. 9 (3),
357366.
Philander, C., Rozendaal, A., 2009. Mineral intricacies of the Namakwa Sands
mineral resources. In: Proceedings of the 7th International Heavy Minerals
Conference Whats next, The Southern African Institute of Mining and
Metallurgy, South Africa, 2023 September 2009, Drakensberg, South Africa.
 E. Tranvik et al. / Minerals Engineering 101 (2017) 3039
Philander, C., Rozendaal, A., 2013. The application of a novel geometallurgical
template model to characterise the Namakwa Sands heavy mineral deposit,
West Coast of South Africa. Miner. Eng. 52, 8294.
Ren, J., Song, S., Lopez-Valdivieso, A., Lu, S., 2000. Selective flotation of bastnaesite
from monazite in rare earth concentrates using potassium alum as depressant.
Int. J. Miner. Process. 59, 237245.
Restarick, C.J., Gottlieb, P., 1991. QEMSEM image analysis of cyclosizer
classification. Miner. Eng. 4 (3/4), 497502.
Silvestre, B.S., Silva Neto, R., 2014. Are cleaner production innovations the solution
for small mining operations in poor regions? The case of Padua in Brazil. J.
Clean. Prod., 19
Sreenivas, T., Padmanabhan, N.P.H., 2002. Surface chemistry and flotation of
cassiterite with alkyl hydroxamates. Colloids Surf. A 205, 4759.
Tilbury, D., Harley, R., 2013. Namakwa Sands: Heavy Mineral Flotation
Undergraduate Research Project. Department of Chemical Engineering,
University of Cape Town.
39
Tranvik, E., 2014. Using Flotation to Recover Monazite from a Zircon Reject Stream
at Namakwa Sands Masters Thesis. Lule University of Technology, Sweden.
http://pure.ltu.se/portal/sv/studentthesis/utvinning-av-monazit-fraan-ett-
zirconavfall-genom-flotation-vid-namakwa-sands-sydafrika(37554dcd-359d-
45f9-a323-69e4b0d926b6).html 08-07-2016.
United Nations Environment Programme, <http://www.unep.org/
resourceefficiency/Business/CleanerSaferProduction/
ResourceEfficientCleanerProduction/UnderstandingRECP/RECPPolicies/tabid/
78841/Default.aspx> 03-07-2016.
Wang, C., Harbottle, D., Liu, Q., Xu, Z., 2014. Current state of fine mineral tailings
treatment: a critical review on theory and practice. Miner. Eng. 58, 113131.
Wills, B.A., Napier-Munn, T., 2006. Mineral Processing Technology (7th Ed): An
Introduction to the Practical Aspects of Ore Treatment and Mineral Recovery.
Elsevier. ISBN: 978-0-7506-4450-1.
World Nuclear Association, <http://www.world-nuclear.org/info/inf30a.html>
29-09-2015.