PGM Recovery

Minerals Engineering 17 (2004) 961–979This article is also available online at:

www.elsevier.com/locate/mineng

Characterizing and recovering the platinum groupminerals––a review

Z. Xiao *, A.R. Laplante

Mining, Metals and Materials Engineering Department, McGill University, Wong Building, 3610 University Street,

Montreal, Qu�ebec, Canada H3A 2B2

Received 16 February 2004; accepted 1 April 2004

Abstract

Methods of characterizing and recovering the platinum group minerals are reviewed in this paper. First, a classification of

platinum group minerals (PGMs) ore types is briefly introduced, followed by the introduction of some representative platinum

group minerals. Second, the sample preparation techniques for mineralogy studies are presented, followed by a brief introduction of

instruments used for characterizing PGMs. Third, the mineralogy of specific ores amenable to gravity, flotation and the flowsheet

for recovering platinum group minerals in several mills of interest are discussed in details. Finally, new research trends of recovering

PGMs and conclusions are briefly presented.

� 2004 Elsevier Ltd. All rights reserved.

Abbreviations: BMS, base metal sulphides; BSE, backscattered electrons; CMC, carboxymethyl cellulose; EDX, energy dispersive X-ray analyser;

ESEM, environmental scanning electron microscopy; GRG, gravity recoverable gold; GRPGMs, gravity recoverable platinum group minerals;

LIMS, laser ionization mass spectrometer microscopy; MLA, mineral liberation analyser; PGEs, platinum group elements; PGMs, platinum group

minerals; QEMScan, quantitative evaluation of materials by scanning electron; ROM, run of mine; SE, secondary electrons; SEM, scanning electron

microscopy; SIMS, secondary ion mass spectrometer; VP-SEM, variable pressure scanning electron microscopy

Keywords: Precious metal ores; Ore mineralogy; Gravity concentration; Flotation

1. Introduction

The six platinum group elements (PGEs); ruthenium(Ru), rhodium (Rh), palladium (Pd), osmium (Os),

iridium (Ir) and platinum (Pt), together with gold and

silver have been considered to be ‘‘precious’’ metals.

Platinum was first discovered in the 16th century in the

Choco district of Columbia (McDonald, 1960). Palla-

dium, rhodium, osmium and iridium were all discovered

in 1803, some 300 years after platinum. The last plati-

num group element discovered was ruthenium.All six of the platinum group metals are silvery white

lustrous metals, although osmium has a bluish tinge.

They are all sufficiently ductile and malleable to be

drawn into wire, rolled into sheet or formed by spinning

and stamping. These six elements can be classified into

*Corresponding author.

E-mail address: [email protected] (Z. Xiao).

0892-6875/$ - see front matter � 2004 Elsevier Ltd. All rights reserved.

doi:10.1016/j.mineng.2004.04.001

two groups compared to the specific gravity of gold.

These elements in the group lighter than gold are

ruthenium, rhodium, and palladium, with specificgravities around 12.0–12.4. Those elements in the group

heavier than gold are osmium, iridium, and platinum

with the specific gravity in the range of 21–22.5. The

elements of the later group also have a higher atomic

number of 76, 77, and 78 respectively.

Valuable for their resistance to corrosion and oxida-

tion, high melting points, electrical conductivity, and

catalytic activity, these elements have wide industrialapplications. The major uses are found in the chemical,

electrical, electronic, glass, and automotive industries.

However, the application of platinum group elements in

the automotive industry is fairly recent, resulting from

emission-control legislation in the USA. The exhaust

gases are passed over a catalyst that contains Pt, Pd

and Rh in the ratio 67:26:7, which converts the hydro-

carbons, carbon monoxide, and nitrous oxide to harm-less emissions. The approximate quantity of platinum

group metals per automobile is 2.4 g. With the emission

mail to: [email protected]

Fig. 1. PGEs ore type classification (Cole and Ferron, 2002).

962 Z. Xiao, A.R. Laplante / Minerals Engineering 17 (2004) 961–979

control legislation widely passed and Kyoto protocol

accepted in more countries, the demand for PGEs will

significantly increase. In the glass industry, the high

melting points of the PGEs and their resistance to the

abrasive nature of molten glass are utilized for produc-

ing high-quality optical glasses. Due to their rarity,

platinum and palladium are widely used as jewellery inthe world. In the chemical industry, PGEs are used

extensively as catalysts, as well as in chemical and in

laboratory equipment such as crucibles, forceps, com-

bustion vessels and filters. Palladium is predominant in

the electro-mechanical industry, relying on its resistance

to corrosion for its application in connectors, sensors,

and relays. On the medical side, platinum has found

application in the prophylactic and therapeutic aspectsof both human and veterinary science. The PGEs can

also be used as coating materials on computer disks and

the polished samples for scanning electron microscopy

(SEM).

Major PGEs’ reserves and production are in South

African, with Russia taking the second place and Can-

ada the third. South Africa production centers on the

Bushveld Complex, the platinum group minerals bear-ing ores being primarily mined for the recovery of these

metals. Canada’s PGEs are by-products of nickel–cop-

per mining, primarily from Falconbridge’s and Inco’s

deposits in the Sudbury area.

Historically, little information relating to the mining

and processing of the PGM ore can be obtained due to

the combined geographic and academic isolation of the

world’s primary PGE producers. Further, the PGEsindustry’s strict corporate-secrecy policies blocked the

research cooperation with academic research group.

This veil of secrecy is now slowly lifting. Recently, the

stronger demand for PGEs (Johnson Matthey, 2002),

the economic and strategic importance of PGEs is

fuelling the exploration and development efforts in the

precious metals mining sector. More geologists began

looking for PGEs for the first time in their careers(Freeman, 2003), more PGMs companies are now will-

ing to put more effort and cooperate with academic

groups on the research of improving the recovery of

PGMs.

Much work has been done on the flotation recovery

of PGMs from primary ores (Cole and Ferron, 2002).

Recently, more concerted efforts are made to study the

mineralogy of PGEs and use gravity or flash flotation toimprove PGE recoveries in the process of nickel–copper

dominant ores. This is strongly desirable for Canada’s

ore types because virtually all PGEs are produced as

by-products of nickel–copper dominant ores. Due to

mineralogical studies playing an important role in

optimizing and improving the PGEs recovery in the

nickel–copper ores. This paper will review the methods

for characterizing and recovering PGEs in various oretypes in the world.

2. Classification of PGEs ores

Various methods (such as the mineralogical data,

chromite content, grade of PGEs, and sulphur content,

etc.) are used to classify PGEs ores. A combination of

PGEs content and mode of geological occurrence to

classify ore types has been adopted from the classifica-tion presented in CIM Special Volume 23: Platinum

Group Elements––Mineralogy, Geology, Recovery

(Edited by Louis Cabri) (Cole and Ferran, 2002). The

information presented in this section was largely

assembled from Chapters 10 and 11 of that volume.

Generally, PGEs ores are grouped into three primary

classes according to this method (Fig. 1):

1. PGEs dominant ores––those ores are exploited pri-

marily for their PGEs content, other associated met-

als, such as Cu, Ni and Co, produced as by-products.

The economic values of the PGEs are, in general,

major in comparison to the by-product values.

2. Ni–Cu dominant ores––those ores mined primarily

for the value of Ni and Cu. The PGEs are produced

as by-products. Usually, the economic importance ofPGEs in these ores is minor. However, they can, in

some cases, become the very important ‘‘deciding fac-

tors’’ for project economics.

3. Miscellaneous ores––these ores contain very low

PGEs concentration compared to the previous two

types of ores. The value of PGEs is little or no eco-

nomic advantage compared to the primary product.

Typically, little is known about the distribution andrecovery of the PGEs in these ores (Fig. 1).

Z. Xiao, A.R. Laplante / Minerals Engineering 17 (2004) 961–979 963

2.1. PGEs dominant ores

2.1.1. Merensky type

Dr. Hans Merensky discovered the platinum-bearing

horizon in South Africa that was to bear his name. In

general the Merensky type deposits are layered with

disseminated sulphides. The total sulphides content isfairly low. In some cases, the PGEs are associated with

chromites as well as sulphides. Some examples of this

type of deposits are the Merensky ‘‘reef’’ and the Great

Dyke in Zimbabwe (Cole and Ferron, 2002). The Still-

water complex is thought to be similar in nature to the

Merensky type deposit (Zientek et al., 2002).

2.1.2. Chromite type

There is a correlation between chromite and PGEs

mineralization. Two types of chromite deposits are

economically or potentially economically significant.

Stratiform chromite deposits occur in the Bushveld

complex (UG2) reef, the Great Dyke, the Stillwater

complex (Lower Chromitites), and the Muskox intru-

sion (Northwest Territories). All are known to contain

PGE concentrations.

2.1.3. Placer type

These deposits include alluvial, eluvial and Alaskan

derived type deposits. The alluvial and eluvial deposits

are typically derived from weathered ultramafic rocks.

Typically, the PGEs occur as Pt rich alloys in the form

of loose grains or nuggets.

2.1.4. Dunite pipes

Dunite pipes occur in the eastern and western sections

of the Bushveld complex. These pipes have had reported

grades of up to 2000 g/t in some sections. The mineral-

ogy of the pipes differs greatly from the Merensky and

UG2 reefs in that the sulphides of the PGEs are rare.

Fifty percent of the PGEs are presented as Pt–Fe alloys;

a further 30% of the PGEs are presented as sperryliteand greversite. The PGEs were mined out and recovered

using gravity recovery methods.

2.2. Ni–Cu dominant ore

Usually, PGEs are recovered as by-products and play

a ‘‘lesser’’ role in this ore type. Naldrett and Macdonald

(1980) classified the Ni–Cu dominant sulphides ore interms of their petro-tectonic setting. The following four

settings or classes account for more than 95% of known

Ni–Cu ores:

1. Class I: Nortitic rocks associated with an astrobleme

(scar resulting form meteorite impact). The only

known example of this type is the Sudbury mining

camp in Canada.

2. Class II: Intrusive equivalents of flood basalts associ-

ated with intracontinental rifting. The most impor-

tant example of this ore type is the Norilsk deposit

in Russia. The Duluth complex in Minnesota is an-

other example of this class of ore.

3. Class III: Magmatic activity accompanying the early

stages of formation of Precambrian greenstone belts.Class III can be subdivided into two further classes.

Examples of this Class deposits include the Kola Pen-

insula, Lyn Lake, and Thompson and the Northern

tip of the Ungava Peninsula in Canada.

4. Class IV: Tholeiitic intrusions, generally synchronous

with orogenesis in Phanerozoic orogenic belts.

The PGEs in the ore types mentioned above occuras discrete PGMs and in solid solution with metal

sulphides and to a lesser extent with gangue minerals.

PGEs recovery is not the prime driving force behind the

flowsheet development and optimization. However, in

recent years a concerted effort has been made to study

the mode of loss for the PGEs and to attempt to recover

more PGEs into the Cu and Ni concentrates.

2.3. Miscellaneous ores

These are ores in which the PGEs are difficult to re-

cover due to low concentration and PGEs present

mostly in solid solution with other minerals. The PGEsare considered ‘‘accessory’’ metals and the deposits are

not mined for the sake of the PGEs and in some cases

the PGEs are not recovered as a by-product.

3. Platinum-group minerals

Unlike gold and major base metals, which form afairly small number of minerals, there are 109 PGM

species recognized by the International Mineralogical

Association (IMA), ranging from sulphides (i.e. bragg-

ite, (Pt,Pd)S) to tellurides (i.e. maslovite, PtBiTe), anti-

monides (i.e. sudburyite, PdSb) to arsenides (i.e.

sperrylite, PtAs2), and alloys (i.e. ferroplatinum alloy)

to native species (i.e. native Pt nuggets). Table 1 lists

some of the PGMs extracted from Cabri’s two books:‘‘Platinum-Group Elements: Mineralogy, Geology,

Recovery’’ (1981) and ‘‘The Geology, Geochemistry,

Mineralogy and Mineral Beneficiation of Platinum-

Group Elements’’ (2002) which gives more details about

the PGMs. Apart from the multitude of PGMs, their

associations are also diverse. The three main minerals

associated with PGMs are pyrrhotite, chalcopyrite, and

pentlandite, such as in the South African Merensky reefdeposit. The UG-2 reef contains low concentrations of

copper- and nickel-bearing sulphides and a large

amount of chromite (FeCr2O4). In the Stillwater J-M

reef ore body (USA), the principal sulphide minerals are

Table 1

Some PGMs minerals properties (Cabri, 2002)

Name Ideal formula General appearance Density (g/cm3)

(calculated)

Hardness (mohs)

Borovskite Pd3SbTe Isolated grains up to 0.2 mm included in pyrrhotite

and chalcopyrite

8.25 N/A

Braggite (Pt,Pd)S Fractured grains up to 8 mm long 9.36 5

Cherepanovite RhAs Grains and their aggregates up to 0.1 mm 9.72 Brittle

Cooperite PtS Euhedral to anhedral micrometer to 1.5 mm grains 10.10 4–4.5

Foodite PdBi2 Grains up to about 1 mm 11.62 Brittle

Genkinite (Pt,Pd)4Sb3 Irregular grains from <0.005 up to about 0.165

mm� 0.165 mm in size

9.26 5.5

Geversite PtSb2 As small drop like inclusion 10.91 4.5–5

Insizwaite PtBi2 As small rounded grains up to 120 lm 12.86–13.59 5

Isoferroplatinum Pt3Fe In placer deposits occurs as various size nuggets,

usually containing chromite and many other mineral

inclusion, varying from flakes to nugget size

18.23 N/A

Kotulskite PdTe Occurs as small rounded to anhedral inclusions in

sulphides or intergrown with other PGM

9.18–10.54 N/A

Laurite RuS2 Grains and crystals from 0.0001 to few mm 6.39-6.43 7

Maslovite PtBiTe Grain up to 0.120 mm in size 11.23 4–5

Merenskyite PdTe2 As minute grains, intimately intergrown with other

PGM, or as single-phase inclusions

8.30 3.5–4

Michenerite PdBiTe Grains from 0.0001 to 2 mm 9.81 4–4.5

Moncheite PtTe2 Crystals up to 1 mm and minute grains 10.24 3.5

Palladoarsenide Pd2As Long and irregular grains 10.59 4.5

Sperrylite PtAs2 Micrometer to centimeter size crystals and rounded

grains

10.8 6–7 Brittle

Stillwaterite Pd8As3 Grains up to 0.12 mm� 0.265 mm 10.95 4.5–5

Sudburyite PdSb Elongated inclusions up to 55 lm� 120 lm 9.41 4–4.5

Tetraferroplatinum PtFe Irregular grains, rims on other Pt–Fe grains 15.81 N/A

Vysotskite PdS As intergrowths and lensoid inclusions in other

sulphides

6.74 N/A

Ruthenium Ru The type material occurred as tabular grains

(35 lm� 7 lm)12.2 6.5

Rhodium Rh A grain from 0.175 to 0.195 mm in cross-section 16.5 3.5

Palladium Pd Commonly as loose grains, sometimes with a radial

fibrous texture

12.0 4.5–5

Osmium Os Occurs as inclusions in Pt–Fe alloys 22.1 7

Iridium Ir Occurs as exsolutions in Pt–Fe alloys 22.2 6–6.5

Platinum Pt In placer deposits, it occurs as various size nuggets 19.1 4–4.5


chalcopyrite, CuFeS2, and pentlandite, (Ni,Fe)9S8. Most

platinum minerals are associated with the copper sul-

fides and palladium with the nickel sulfides. The ore also

contains less than 1% quartz but significant amounts oftalc and serpentine (MgO). Platinum was long known to

exist in the arsenide form (sperrylite, PtAs2) in nickel–

copper sulfides in the Sudbury area.

Pentlandite and chalcopyrite are generally well

recovered (depending on their particle size and degree of

liberation), as are any PGEs in their lattice (i.e. in solid

solution) or present in PGM blebs they may contain.

PGEs present as blebs or in solid solution in pyrrhotitemay or may not be recovered. For example, in the

Sudbury basin, pyrrhotite is now largely rejected to

minimize smelting costs and environmental pollution.

The pyrrhotite contains significant amounts of nickel

(anywhere from 0.4% to 0.8%) and accounts for most of

the nickel losses in the Sudbury area.

The extremely low concentration of PGMs, their fine

size distribution, the difficulty in their detection and

identification, and sample representativity are typical

problems when carrying out the mineralogical studyon PGMs. In a word, the range of minerals present,

their relative densities, shape, particle size, and associ-

ations present a challenge to the metallurgist in design-

ing and optimizing the extraction process. The various

ore types (end members) of typical ore bodies each with

its own metallurgical response will heighten this chal-

lenge.

All of the PGM species in Table 1 are listed with theirideal formula, although composition may vary locally.

The general occurrence is depicted based on where the

individual PGM has been first reported (Cabri, 2002).

The calculated density and hardness listed here are

critical factors regarding gravity recovery, grinding and

classification behavior.


4. Applied/process mineralogy and sampling preparation

techniques using pre-concentration of the PGMs

Petruk (2000) defined applied mineralogy as the

application of mineralogical information to understand

and solve problems encountered during processing of

ores and concentrates. It involves characterizing min-erals and interpreting the data with respect to mineral

processing. When processing problems are due to the

mineralogical characteristics of the ore and/or process

products, mineralogical data should be generated to

solve this problem.

Henley (1983) reviewed process mineralogy as an

integration of mineral processing and mineralogy. In his

review, a flowsheet starting from orebody explorationthrough to optimization of plant operation was pro-

posed. An actual plant flowsheet was developed based

on mineralogical information, laboratory, and pilot

plant testing. After commissioning of the plant, an

optimization program was continued for second and

subsequent iterations of this study. These interactions

extended beyond new samples of drill core and into

samples taken from the operating plant.There is relationship between ore mineralogy and

metallurgical performance in a plant (Petruk and

Hughson, 1977; Cabri, 1981; Henley, 1983). Not only

does the ore mineralogy play a critical role on the

recovery method chosen but also dictate the process

flowsheet for different PGEs ore types and for plant

flowsheet optimization.

How can the mineralogical information obtained beas precise as possible? Many techniques for determining

mineral characteristics have been developed in the last

three decades. They were used to acquire the necessary

mineralogical information and these include scanning

electron microscope equipped with an energy dispersive

X-ray analyzer (SEM/EDX) or wavelength dispersive

X-ray analyzer (SEM/WDX), variable-pressure (or

low-vacuum) scanning electron microscope (VP-SEM),electron probe micro-analyzer (EPMA), X-ray diffrac-

tion (XRD), quantitative evaluating material by scan-

ning electron microscope (QEM*SEM), and mineral

liberation analyzer (MLA). These techniques will be

discussed briefly in the following chapter.

Maximum recovery with acceptable grade is a con-

stant goal for operators and producers. Knowledge of

ore characteristics that affect metal recoveries can helpin achieving this goal. Although ore characteristics can

be determined by mineralogical technologies as men-

tioned above, there is a distinct problem when charac-

terizing the PGMs in an ore, which is from their

extremely low grade (often less than 1–2 g/t). The

magnitude of finding PGMs in a polished section may

be compared to find a needle in a haystack (Cabri,

1981). Fortunately, this problem can be solved or re-duced by using some pre-concentration techniques such

as gravity methods, magnetic or flotation or these

methods combined.

Upgrading of PGMs is effected mainly by gravity

methods due to their high specific gravity, such as jig-

ging, panning, and heavy liquid separation (Cabri,

1981).

Zhou and Zhang (1975) gave a detailed account oftheir comminution and separation techniques for a

PGMs-bearing chromite ore. Their method involved a

large sample size of 1060 kg. The original sample was

processed as follows: coarse, medium, and fine crushing;

splitting out a 1000-kg sample for grinding and sieving

(grinding was used to liberate enough minerals for

subsequent concentration), with the balance split further

for chemical analyses and storage. The final product,reported to contain only 5% unliberated PGMs, was

distributed as follows: 37% +150 lm, 17% )150+ 106lm, 4.2% )106+ 100 lm, 2.8% )100+ 74 lm, and 39%)74 lm. Then the ground and sized samples were pro-cessed with a high frequency superpanner (320–800

vibration/min). The first concentrates were considered to

be the PGMs concentrates. The second concentrates and

middlings were ground further and the superpanner wasused to recover more. The PGMs concentrates were

combined with the first concentrates. Chromite con-

centrates, middling, and tailings were also obtained

from this operation. They reported a 50% recovery of

PGMs with the superpanner, increasing to more than

70% for high-grade samples. The PGMs and chromite

concentrates were then subjected to magnetic separa-

tion, first with a weak current to remove any stronglymagnetic minerals (including some PGEs-alloys) and

subsequently with increasing current. Finally, a selective

dissolution and heavy liquid separations were used for

further separation to obtain the final PGMs and chro-

mite concentrates.

Although this method used a large sample to achieve

a higher degree of representation, the reported recovery

was only 50% (up to 70% for some higher grade ore).The size distribution of PGMs, and some information

about source of PGMs, such as PGE-alloys, was lost

due to the selective dissolution process. This procedure

is also extremely time consuming.

Cabri and Laflamme (1976) and Cabri (1981) devel-

oped a different mineral separation technique––heavy

media elutriation to separate small samples (less than

500 g) of unconsolidated rock or mill samples. Theytested this new recovery method on a sized synthetic

samples of sperrylite mixed with purified and sized

pyrrhotite, chalcopyrite and pentlandite. They reported

that excellent recoveries were obtained down to about

53 lm, but recoveries fell off steeply in fractions finerthan 53 lm.The inability to recover sperrylite below 53 lm, as

well as the relatively small masses processed, must beconsidered significant handicaps for this approach.


Another sample preparation method was developed

by Williamson and Savage (1965). The Witwatersrand

Au-U fossil placers contain small quantities of by-

product PGEs, and the PGMs are considerably finer

than those of usual placers, partly due to the grinding of

the ore to 80% less than 74 lm (Reimer, 1979). There-

fore, Williamson and Savage used flotation on 25-lbs(11.33 kg) lots of gravity concentrates. Final cleaning

operations were carried out in an 800-gm Fagergren cell.

They reported that the Ir–Os–Ru alloys float indiffer-

ently at a neutral or alkaline pH, floating readily at a pH

of 1.5–3.5. Their recovery in PGEs was reported to be

close to 100%, as determined by adding known quanti-

ties of irradiated ‘‘osmiridium’’. The final pyrite-rich

flotation concentrate was still too large (2–4% of origi-nal weight) for their purposes, so they used pyrometal-

lurgical techniques to remove the pyrite. Although their

method reported 100% flotation recovery, the response

of the fine PGEs to gravity was not reported or char-

acterized.

Sizgoric (1985) developed a method of locating and

identifying PGMs in fine-grained crushed samples. This

method consists of systematic searching of the samplesusing the back-scattered electron detector of a scanning

electron microscope. Usually, the ores containing PGEs

are pre-concentrated by using gravity separation.

However, most of Inco’s mill samples are much too fine-

grained to respond to regular gravity separation facility

as cited in his paper; in many cases, over 80% of the

platinum in the samples occurs in particles smaller that

38 lm (400 mesh). Because the milling has not com-pletely separated the PGMs form the other minerals,

Sizgoric (1985) thought the pre-concentration of the

combination of gravity and heavy liquid was not feasible

in this case.

In his method, the whole sample was examined with

SEM without pre-concentration. The procedures are

briefly described as following: the sample is thoroughly

mixed and a representative portion is sent for Pt, Pd andAu analyses. The rest of the sample is wet screened, and

a representative portion of each size fraction is sent for

Pt, Pd and Au assays. A number of epoxy polished

section mounts are made of each size fraction for the

SEM work. This number varies from two to eighteen

depending on the size of the particles and the PGEs

content of the size fraction. All polished sections are

Table 2

The weight of SF and HC of the primary ore and tailings (Knauf and Kozy

Sample )250+ 90 lm )

Primary ore SF¼ 244 g, S

HC¼ 13 mg H

Tailings SF¼ 56 g, S

HC¼ 7 mg H

scanned on the SEM in their entirety. The number of

particles inspected range from approximately 20,000 for

the coarsest (+100 lm) size fraction, to 1.5 million forthe finest (<10 lm) size fraction in each polished section.This method was used to examine the tailings from

the Copper Cliff Mill. The encouraging results achieved

with these samples prompted the examination of mate-rials from other streams in order to determine the origin

of the PGMs losses found in the tailings. Although this

method can determine the PGMs identity, size distri-

bution, and mineral associations, it is time consuming

and expensive, the results are only semi-quantitative.

With the development of gravity technology, such as

Knelson concentrator, the PGMs in finer size fractions

can be effectively recovered into concentrate.Knauf and Kozyrev (2002) used the Knelson con-

centrator to study the mineralogy of Cu–Ni ore (both

the ore fed to KC and the tailings of KC operation).

They reported that when a primary ore about 40 kg with

a grain size class of 0.5 mm was fed to the KC, 80 g of

concentrate with the reduction coefficient (RC) equal to

530 was obtained.

The quarters of the primary ore and the KC tailingswere investigated using the ‘‘ppm-mineralogy’’ technol-

ogy: primary ore and tailings were separated by size

fraction (SF) of 250–90 lm, 90–37 lm and )37 lm.Using a hydroseparator NATI with the reduction

coefficient of 18,000–72,000 for primary ore and 8000–

13,000 for tailings, 6 heavy concentrates (HC) were

extracted. The weights of SF and HC of the primary ore

and tailings are reproduced in Table 2:Then, microprobe samples SF were prepared from

the heavy concentrates. The PGMs were identified and

their volumetric relations were measured.

They concluded that the large grains of PGMs are

effectively extracted: the grains larger than 70 lm are

extracted completely, the grains with in a size of 30–70

lm are extracted partially and the grains smaller than 30

lm are totally lost. They also concluded that the dis-tribution of the PGMs by grain size yields all necessary

information for the choice of processing flowsheet for

PGMs extraction. They argued that pre-concentration is

very helpful for mineralogical investigation, while min-

eralogical data is one of the most important factors to

determine the processing route. Their approach yields

significant upgrading ratios, which would contribute to

rev, 2002)

90+ 37 lm )37 lm

F¼ 146 g, SF¼ 72 g,C¼ 8 mg HC¼ 1 mg

F¼ 37 g, SF¼ 13 g,C¼ 6 mg HC¼ 0.4 mg


more reliable mineralogical examinations. However, the

partial PGM recovery below 75 lm and total lack of

recovery below 30 lm remains a serious problem.

Kojonen et al. (2003) studied the mineralogy of ore at

the Proterozoric Hitura Ni–Cu–PGEs deposit in Wes-

tern Finland. The Hitura Ni–Cu–PGEs ore is hosted by

serpentinized ultramafic intrusions. The ore types (endmembers) include (1) fine grained sulphides dissemi-

nated in the serpentinite core, (2) medium-grained dis-

semination in the serpentinite and amphibole rock and

(3) high grade interstitial disseminated sulphides and

massive accumulations in the amphibole rock of the

contact zones.

Before doing mineralogical analysis, the PGMs were

enriched in laboratory conditions by a gravitationalhydro-separation method with a concentration factor

RC of 50,000:1. To avoid losses of PGMs grains, two

fractions of sulphides concentrate were separated by the

NATI method: grain size 71–32 lm weighting 85 g and

grain size )32 lm weighting 50 g, yielding PGMs con-

centrate 3 and 1 mg, respectively.

The gravitational enrichment of phases with density

more than 5–6 g/cm3 with a grain size 10–70 lm and RCfactor between 103 and 106 for many reasons excludes

the application of traditional enrichment techniques.

Therefore, a gravitational hydro-separator of original

engineering was used in ppm-mineralogy technique

(patented by ‘‘NATI’’ Research JSC., 1996, Russia).

These hydro-separators are reported to allow the

enrichment of ‘‘heavy’’ phases in a continuous mode

with RC factor up to 1000–5000, as well as in mode ofsingle batch separation of a sample up to 500 g with RC

factor up to 100,000.

The samples were then prepared for the heavy con-

centrates for SEM and EPMA study. The SEM is

mainly for the identity of PGMs minerals and the

EPMA for the quantitative analysis. Finally, they report

that PGMs discovered in the PGMs concentrate samples

are sperrylite, PtAs2 (81.1%), michenerite, PdBiTe(6.6%), irarasite, IrAsS (1.9%), froodite, PdBi2 (6.9%),

hollingworthite, RhAsS (1.9%), and an undefined Rh,

Co, Ni sulpharsenide and an undefined Rh, Re mineral

(1.6%). Trace amounts of Pd averaging 37, 116 and 102

ppm were analyzed from pentlandite, niccolite and

gersdorffite, respectively. However, the information

about the PGMs liberation, the grain size of different

PGMs and the potential response to gravity recovery isnot available in their paper.

A test has been developed at McGill University to

characterize the gravity recoverable gold (GRG) of an

ore in terms of its total amount, size distribution, at

grind size at which it becomes liberated (Woodcock and

Laplante, 1993; Laplante, 2001). The procedure involves

a 3-stage sequential liberation and recovery of gold with

a 3-inch Knelson concentrator. At the final grind of 80%)75 lm, as little as 3% and as much as 97% of the gold

has been recovered. All results are presented in a sized

basis (from )20 up to 850 lm). Examination of GRGconcentrates using an automated SEM technique, the

mineral liberation analyzer (MLA), has shown that be-

low 75 lm, gold particles recovered in stage 1 of the testwere very well liberated (Guerney et al., 2003). Particles

above 150 lm for stage 1 can have various degree ofliberation, concentrates from stages 2 and 3 were also

found to have a high degree of liberation. This test has

successfully characterized over 140 different ores the

GRG contents, which is one of the most important

parameters to justify the installation of a gravity circuit.

The GRG test can be used not only a measurement of a

potential response for gravity recovery but also a pre-

concentration method to study the gold mineralogy as itundergoes liberation.

The three-stage GRG procedure may apply to the

platinum group minerals owing to the specific gravity of

some platinum group minerals. However, the test needs

to be modified to account for the fact that most of the

PGMs tend to report to fines size classes (Cabri, 1981;

Thurman and Allen, 1994) and have a lower specific

gravity than native gold. the modified procedure in-cludes the standard GRG procedure, but a fourth

stage is added at higher rotating velocity (a fifth cycle

might be added to examine what’s lost in stage four).

The gravity recoverable platinum group minerals

(GRPGMs) test can be used not only for the justifica-

tion of installation of gravity circuit but also for the

preconcentration technique of mineralogical study

(Xiao and Laplante, 2003).

5. Instruments for characterizing PGMs in applied/

process mineralogy studies

There are a wide variety of instruments used for ap-

plied/process mineralogy: The optical microscope is used

to identify many minerals, to observe mineral textures,and to determine mineral quantities by point counting.

X-ray diffractometer (XRD) is used to identify many

minerals with a high degree of certainty, and to quali-

tatively determine mineral contents in powdered mate-

rials. The development of the microprobe (MP) was a

giant progress in applied mineralogy. Not only can it

determine the major, minor, and trace contents of

minerals in polished sections, but also it can keep theoriginality of the mineral grain. The developments of

SEM together with the energy dispersive X-ray analyzer

(EDX) enable the mineralogist to nearly instantly

identify mineral grains based on backscattered electron

images. The development of VP-SEM makes it easy to

do the analysis without the need to polish or coat the

polished sections. QEM*SEM developed by CSIRO and

MLA developed by JKMRC in Australia can per-form fully automatic mineralogical analyses. Other

Fig. 2. Backscattered electron images of Cu–Ni–PGEs ore. It shows

maslovite (Pt–Bi–Te) (white). This BSE image was produced with

15 keV accelerating voltage, and composition mode, the bar length is

20 lm.


instruments can be used in applied mineralogy include

proton induced X-ray analyzer (PIXE), secondary ion

mass spectrometer (SIMS), and laser ionization mass

spectrometer (LIMS). In this literature review, only

SEM, VP-SEM, QEM*SEM and MLA are briefly pre-

sented as an introduction to the types of analyses used

worldwide recently.

5.1. Scanning electron microscope with energy dispersive

X-ray analyzer

The scanning electron microscope (SEM) is one of

the most versatile and widely used tools of modern sci-

ence as it allows the study of both the morphology and

composition of materials. Equipped with an energydispersive X-ray analyzer (EDX), it is used in applied/

process mineralogy to analyse the polished and/or thin

sections of samples, as well as unmounted pieces of

material (Petruk, 2000). By scanning an electron probe

across a specimen, high-resolution images of the mor-

phology or topography of a specimen, with great depth

of field, at very low or very high magnifications, can be

obtained. Therefore, the identities of most minerals, thesize and the relationships of mineral grains, and the

X-ray spectra that show the distributions of elements in

minerals can be obtained.

The SEM functions exactly as its optical counterparts

except that it uses a focused beam of electrons instead of

light to ‘‘image’’ the specimen and gain the necessary

information. The gun on the top of SEM column pro-

duces an electron beam under high vacuum. The elec-tron beam, which is confined and focused by apertures

and magnetic lenses, is either scanned over the entire

sample, or is focused on a grain in the sample. The

sample, which is coated to prevent surface charging,

interact with the electron beam and further to produce

backscattered electrons (BSE), secondary electrons (SE),

X-rays and other signals. The SEM is generally equip-

ped with BSE, SE and EDX detectors to detect thesesignals.

Backscattered electrons are incident electrons which

are scattered ‘‘backward’’ 180� when colliding with anatom in the specimen. The production of BSE varies

directly with the mineral’s average atomic number. This

differential production rate causes higher average atomic

number minerals to appear brighter than lower average

atomic number ones. For example, most silicate miner-als have lower average atomic number and appear dark

grey in BSE image. In contrast, some PGMs, i.e. sper-

rylite (PtAs2), michenerite (PdBiTe), maslovite (PtBiTe)

have a higher average atomic number and appear in

shades of light to white in BSE image. These features

can be used as a technique to search for PGMs minerals

in the BSE images of conventional SEM. Fig. 2 shows

the maslovite mineral BSE composition image; it is clearthat this PGMs particle appears brighter than other

minerals around it. However, the difference in the shades

of grey between the minerals can be either enhanced or

reduced by changing the contrast, brightness, voltage

and probe current on the SEM.Secondary electrons are ionized electrons which leave

the atom with a very small energy (5 eV) when a incident

electron strikes an atom. Due to their low energy, only

secondary electrons that are very near the surface can

exit the sample and be examined. The SE images can

show a very good topographic effect of particle and

display details of surface irregularities much better.

However, the SE image is not as useful as the BSE imagefor showing mineral distributions. The SE image can be

produced at a much lower current and voltage than is

required for the BSE image.

X-rays are caused by the de-energization of the

specimen atoms after secondary electrons are produced.

Since a lower energy electron was emitted from the atom

during the secondary electron process, a lower energy

shell now has a vacancy. A higher energy electron can‘‘fall’’ into the lower energy shell, filling the vacancy. As

the electron ‘‘fall’’, it emits X-ray. Therefore, X-rays

have a characteristic energy that is unique to the element

from which it originated. These signals are detected with

the EDX detector. The minerals are identified by col-

lecting the X-ray signal. In other words, the EDX

detector sends the X-ray signal to the EDX analyser that

sorts the signal into the different elements present inthe particle, and into X-ray counts for each element. The

X-ray counts are recorded, and displayed as peaks.

Usually, the working distance is optimized to 15 mm

when doing X-ray analysis.

5.2. Variable pressure (low-vacuum) SEM

Variable pressure (low-vacuum) scanning electronmicroscopy, generally more frequently called environ-

mental scanning electron microscopy (ESEM), is one of

Fig. 3. Backscattered electron image of sperrylite (white) particle with

low-vacuum SEM at working distance of 15 mm and 20 keV acceler-

ating voltage.

Fig. 4. Sperrylite (white) crystal obtained by gravity concentration,

BSE image with low-vacuum SEM at working distance of 15 mm and

30 keV accelerating voltage.


the latest developmental stages of electron beam

microscopy. It retains most of the performance advan-

tages of a conventional SEM, but removes the high

vacuum constraint on the sample environment. Wet,

oily, dirty, insulated samples may be examined in their

natural state without modification or preparation. This

high-pressure (low-vacuum) technique can stabilize theinsulator sample surface potential close to ground po-

tential during imaging even when high beam voltages up

to 20 kV or more are employed (Farley and Shah, 1988).

The basic difference between the conventional SEM and

ESEM is the pressure in the specimen chamber. For

conventional SEM (the column and specimen chamber

share the same vacuum) it amounts to 10�3 Pa and for

ESEM it can be as low as 10�3 Pa and as high as 103 Pa(Danilatos, 1991).

When the primary electron beam passes through the

gaseous medium of the specimen chamber, collisions of

the electrons with atoms and molecules of the gas occur.

As a result of these collisions, the electrons can loose

some portion of their energy and can change the direc-

tion of their propagation. A portion of the scattered

electrons, the so-called ‘‘skirt’’, does not contribute tothe image signal. It produces only the background

(noise) in the image. However, there exists some portion

of the electrons of the primary beam, which does not

scattered and is incident on the specimen in the spot.

This portion of electrons creates a useful image signal

with a sufficiently high resolution. The air in the sample

chamber is ionized by the primary electron beam, con-

ducting electricity sufficiently to allow the electrons ab-sorbed by the sample to leak through the air to a ground

contract, so that no coating is needed, even at high

accelerating voltages (Moncrieff et al., 1978).

The same signals as those in the conventional SEM

can be detected. As mentioned above, wet sample can be

analysed so that a sample from slurry in a process

stream can be analysed within a very short time after it is

collected. The low-vacuum SEM system can easilychange the pressure and accelerating voltage settings to

obtain good BSE images.

Robinson (1998) pointed out that the low-vacuum

SEM is as fast and as easy to use as an optical micro-

scope for an experienced mineralogist. However, the

conventional SEM requires a coating, such as carbon,

gold and platinum, on the surface of polished section,

which is not always possible or desirable (Farley andShah, 1991). If a PGMs mineral is under study, then it

can be only coated with carbon. Otherwise, it will give

some misunderstanding information about the PGMs.

A coating is not required when using the low-vacuum

SEM.

In practice, to ensure that the losses in the number of

primary beam electrons due to their collisions with the

molecules of gases are as low as possible, the workingdistance must be optimized; usually it is set to 15 mm.

Fig. 3 is the BSE image of a sperrylite particle found in

the operation condition of 15 mm working distance and

20 keV. Fig. 4 is a well-developed single sperrylite crystal

(not a particle) obtained with 15 mm working distance

and 30 keV.

5.3. Quantitative evaluation of minerals by scanning

electron microscope (QEM*SEM)

The names QEMScan or QEM*SEM are usually

exchangeable. The QEMScan was developed by CSIRO

in Australia to provide an automatic, off-line, size-

by-size and particle-by-particle mineralogical analysis of

metallurgical products and exploration samples. It is

used widely in mineral processing to analyze ores and

mill products to obtain quantitative information aboutthe distribution of minerals in plant and test prod-

ucts. Currently, the QEMScan is used for ore charac-

terization, comminution, liberation analysis, process


optimization (efficient fine grained ore processing, par-

ticularly improved feed preparation and grinding opti-

mization), process modeling and plant problem solving.

This system actually uses a combination of back-

scattered electron (BSE) images and EDX analysis to

create an image of a sample based on chemical com-

position, and then X-ray is acquired based on the ima-ges, finally, the mineral composition is obtained by

comparison with the database in the computer system.

The electron gun is steered within the image frame to

scan each particle and to obtain X-ray counts for 16

elements at designated pixel points within the particle.

The system uses four EDX detectors. The X-ray counts

for each element at each pixel position are sent to

computers. The computers compare the X-ray counts toa reference mineral file to identify the mineral at each

pixel point. Each identified point is recorded in a file and

displayed on the CRT screen by a colour which repre-

sents the mineral. Calculations of the data with respect

to mineral processing are performed automatically, and

no further image analysis is performed (Pignolet-Bran-

dom and Reid, 1988; Reid and Pignolet-Brandom,

1988). It is reported that the QEMScan has three basicmodes of operation:

(1) Point scan, this is the most basic mode of QEMScan

operation, and is similar to a mineralogical point

count. EDX analyses are performed on a grid pat-

tern with equidistant points. Only modal abundance

information can be determined from this image.

(2) Line scan, the scan grid is set up so that points areclosely spaced in the X -direction and widely spacedin the Y -direction.

(3) Area scan, points are closely spaced in both X and Ydirections, this mode is used to determine grind size

for liberation in feed samples, diluents in concen-

trates and losses in tailing samples.

QEMScan is highly automated to ensure reliable andrepeatable results. It is gradually accepted and operating

in many countries worldwide. It can perform quantita-

tive analysis more easily and automatically. However,

this system is much more expensive than SEM and

ESEM.

5.4. Mineral liberation analyzer

The mineral liberation analyzer was developed to

provide an automatic, off-line, size-by-size good quan-

titative mineralogy and liberation data for mineralogists

and processors to fully assess the orebody, improve the

plant recovery, and maintain the quality of product.MLA is equipped with backscatter electron imaging

and EDX system, and combined with the MLA software

package, which enables liberation measurement, data

analysis and presentation. High speed and high-resolu-

tion BSE imaging, EDX analysis and image analysis

result in accurate mineral identification for particle

ranging from 2 to 600 lm. In briefest terms, the systemtakes advantage of the capability of SEM to provide

consistent grey-levels for each mineral in BSE image of a

sample, different minerals with various average atomic

numbers will have different grey levels. The software‘‘segments’’ the image into minerals, taking into account

cracks, other surface imperfections and ‘‘edge effect’’

around particles. The X-rays are used to confirm the

identity of each mineral.

The measurement system is automatic and capable of

measuring up to 14 samples overnight. Image analysis

occurs partly off-line and produced a database of min-

eralogical results and a set of colored mineral maps.When doing liberation analysis, a database of quanti-

tative mineralogical information and software to pro-

duce tables and graphs such as mineral abundance as

volume and weight percent, calculated elemental assay,

and particle and mineral size distribution. When doing

rare phase searching, such as gold or a platinum group

mineral, each occurrence, to less than 1 mm in size, is

imaged at high-resolution and the associated mineralsare identified.

Most minerals can be uniquely identified from their

BSE signal strength. However, for minerals with simi-

lar BSE levels, the EDX is employed and the X-ray

spectra of just these minerals are stored with the BSE

data for off-line image segmentation. In all cases, once

subsequent image processing is complete, the SEM

can be driven back to view any desired particles bymouse click in order to check mineral identification or to

obtain a photographic record of the original particle of

interest.

Because the output of mineral liberation statistics

obtained from the measurement can assist ore source

evaluation, circuit design and optimization, MLA is

accepted gradually by the mining community. Anglo

platinum officially announced the acquisition of theirthird MLA in 2003.

6. Relationship between mineralogy and the recovery of

platinum-group elements from ores

Different platinum-group element deposits or ores

should be treated with different recovery methods

according to their mineralogical features and other

factors. The following briefly discusses several different

ores: those amenable to gravity separation, those ame-

nable to flotation, and those where platinum-group

elements are by-products of base metal sulphides. Mostof these following information are from the special

volume of ‘‘Platinum-Group Elements: Mineralogy,

Geology, Recovery’’ edited by Cabri (1981) and other

sources.


6.1. Ores amenable to gravity recovery

The mineralogical features of most importance in

those ores amenable to gravity separation are that the

platinum-group elements occur as minerals of high

density, that they are free or free milling, and that the

grain size distribution falls in the region where gravitytechniques can be applied successfully (Cabri, 1981b).

Ores in this category include magmatic Dunite or

Alaskan-type deposits, and alluvial, eluvial and fossil

placer deposits. With new gravity equipment developed,

more and more deposits can be classified into this cat-

egory.

Mineralogy: Primary PGEs deposits in Alaska-type

ultramafic rocks were found first in the Nizhnii–Tagildistrict of the Ural Mountains in 1890 (Mertie, 1969).

The platinum-group minerals in this ore type are similar.

The principal PGMs is Pt–Fe alloys, mostly isoferro-

platinum (Pt3Fe) and the next most common PGM is

platiniridium (Ir,Pt) according to Razin (1976). This

type of deposit also includes rare and very rare PGMs:

osmiridium (Ir,Os), iridosmine (Os,Ir), cooperite (PtS),

other Pt alloys, tulameenite (Pt2FeCu), laurite (RuS2)tetraferroplatinum (PtFe) and irarasite (IrAsS). Most of

the Pt-Fe alloy grains in the Gusevogorskiy deposit are

smaller than 0.1 mm; however, grains as large as 3 mm

occur sporadically (Begizov et al., 1975).

Recovery method: The mineralized rock at this de-

posit is ground to )0.2 mm and then subjected to

magnetic separation followed by tabling (SK-1 con-

centration tables). The table concentrate is processedagain to produce a gravity concentrate in a hydro-

separator.

Prior to the 20th century, all PGEs were obtained

from alluvial deposits. The PGEs occur in these deposits

as alloys, usually Pt-rich, in the form of loose grains and

nuggets. There are virtually no published data on the

recovery of PGEs from placer mining due to lack of

mineralogical data.Norilsk Mining Company is one of the biggest PGEs

producers in the world. According to a typical economic

classification, ores of the Norilsk deposits belong to a

group of sulphide copper–nickel ores with associated

PGEs mineralization (Blagodatin et al., 2000). This de-

posit is subdivided into three groups, namely, massive

(rich) ore, disseminated ore occurring in host rocks, and

stringer-disseminated ore occurring in intrusive hostrocks (Kozyrev et al., 2002). Another classification

method is adopted by Blagodatin et al. (2000): the ore is

divided into two groups, one sulphides copper–nickel

ores, and another platinum ores with associated non-

ferrous and rare metals, including all kinds of dissemi-

nated ores.

Mineralogy: PGEs content in major sulphides (pyr-

rhotite, pentlandite, and chalcopyrite) and the PGEsoccurrence in many varieties of disseminated ore have

been studied in Research Center of Norilsk Company.

The studies revealed that the bulk of platinum (90% or

more) is in the mineral form of cooperite (PtS), Pt–Fe

alloys, sperrylite (PtAs2), Pd–rustenbergite (Pt,Pd)3Sn,

etc., while minor quantities are dissolved in pyrrhotite,

with a maximum concentration of 0.8 ppm. As to pal-

ladium, 27.3% of the metal is in mineral form. Few ofthe PGEs metals are represented by solid-state solutions

in pyrrhotite and pentlandite. They are largely hosted by

pyrrhotite, which reaches 2.4–4.0% concentration in the

ores. The 250 lm size fraction contains only occasional

grains of platinum minerals, whereas size fractions of

50–250 lm, and especially 50–250 lm, appear to be mostenriched in these minerals. In the )50 lm size fraction

their share amounts to 32–34%. However, palladiumminerals occur as very fine grains.

Recovery method: The nature of PGEs concentration

and their form of occurrence in platinum ores dictate the

use of modern high-performance techniques of gravity

separation to produce high-grade PGEs concentrates.

This decreases the loss of PGEs in final tailings and

makes it possible to process PGEs gravity concentrates

without involving the pyro- and hydrometallurgicalcopper and nickel production cycle (Blagodatin et al.,

2000).

The Knelson concentrator’s installation points were

selected on the basis of the mineralogical investigation

of concentration products obtained by the conventional

flotation technology. Based on the study of the size

distribution in flotation concentrate and tailings, it was

found that the bulk of the platinum and palladiummineral grains in the concentrates was below 70 lm, thehighest distribution was in 30–40 lm size fraction and

no grain found above 300 lm. For the final flota-

tion tailing, there were no grains coarser than 100 lm in

non-magnetic fraction of tailings, the highest distribu-

tion was in 20–30 lm size fraction. In magnetic fraction,

there were quiet amount of Pt and Pd bulk grains dis-

tributed in 100–200 lm size fraction. Such a size distri-bution indicates that the flotation is not good method

to completely recover the PGMs and the over-grinding

may appear. Therefore, the company decided that

the gravity separator should be installed at the first

grinding stage, thus removing the bulk of PGEs miner-

als from the flotation concentration cycle. Since June

1998, in both grinding and tailings cycle, Knelson con-

centrator KS-SD 48 was used to processed the dissem-inated ore. The recovery of 50–60% Pt, 10–13% Pd,

and 17–20% Au was achieved with 400–500 g/t total

content in concentrates. These results were obtained

with ore assaying 1.1–1.3 g/t Pt, 3.4–4.6 g/t Pd, and

0.16–0.2 g/t Au. The positive results made the company

install two additional 48-inch concentrators with modi-

fication of bowl cone profile, the total PGE content rose

to 1000 g/t and the recovery grew by 25% (Blagodatinet al., 2000).

Fig. 5. The Northam flowsheet.


6.2. Ores amenable to flotation recovery

Flotation might be the most used method in recovery

PGEs. As it is now well known, the PGMs usually occur

within base metal sulphides, and often prefer the grain

boundary between sulphides and silicate (Cabri, 1981).

Most PGEs occur either as discrete minerals or as solidsolutions in major sulphides.

In those magmatic deposits where platinum-group

elements are intimately associated with the base metal

sulphides, flotation is used as the first recovery process.

Although high recovery, usually between 80% and 95%,

can be obtained, the grade of concentrates so obtained is

still quite low, which need further upgrade.

6.2.1. Merensky reef

Mineralogy: Extensive mineralogical studies show

that seven PGMs are important in the Merensky reef

although varieties occur in different areas and ore types

(Vermaak and Hendriks, 1976; Brynard et al., 1976;

Schwellus et al., 1976). Three of the PGMs are sul-

phides: braggite (Pt,Pd)S, cooperite, PtS, laurite RuS2,

one is a Pt–Fe alloy, one an arsenide (sperrylite, PtAs2)and two tellurides (moncheite, PtTe2, kotulskite PdTe).

The presence and relatively large percentage of platinum

minerals and sulphides is unique to this type deposits (it

is divided into two groups, one is a silicate ore, the other

chromite ore).

The principal opaque minerals are base metal sul-

phides (BMS), pyrrhotite, pentlandite and chalcopyrite,

in order of decreasing abundance (Vermaak and Hend-riks, 1976; Brynard et al., 1976). Minor sulphides

include cubanite, mackinawite and pyrite––the latter

being more abundant in chromite-rich bands (Brynard

et al., 1976).

The majority of the principal PGMs tends to occur as

idiomorphic inclusions (braggite, cooperite, sperrylite

and laurite) but a Pt–Fe alloy, especially in those areas

where it is an important constituent, commonly occursfinely intergrown with base metal sulphides or, more

rarely, with cooperite (Vermaak and Hendriks, 1976).

These authors also pointed out that most PGMs occur

at the BMS-gangue contact in the silicate ore, whereas in

the chromite ore the PGMs occur mostly in the BMS or

in the gangue. The exception to this is braggite, which

shows a strong preference for pentlandite.

The sperrylite, braggite and cooperite occur as ‘‘fairlycoarse’’ in size and moncheite as being ‘‘much smaller’’

(Brynard et al., 1976).

Recovery method: Although the basic process route

used for the Merensky and UG-2 ore is the same, the

different mineralogical features dictate the many subtle

differences in each flowsheet. The conventional commi-

nution circuit, including crushing, rod milling, and ball

milling or by autogenous milling is used for Merenskyore, however, UG-2 is not amenable to autogenous

milling. The mill product is classified in hydrocyclones

for both ores. There is also a big difference in the mill

circuit of processing these two types of ores. For

example, preconcentration methods, such as the cordu-

roy strakes, James table and flash flotation, have been

used for Merensky ore due to the PGMs enrichment in

the cyclone underflow. However, the UG-2 ore is not asamenable as Merensky ore to preconcentration in the

milling circuit because a much finer grind is needed to

liberate the PGMs and because the chromite present can

lead to poor separation.

For Merensky and UG-2 ores, the flotation process is

a bulk sulphides float that recovers the base-metal sulp-

hides containing any liberated PGMs sulphides minerals.

Flotation takes place at natural pH (7.5–9) with a xan-thate collector such as isobutyl xanthate or normal

propyl xanthate. Some mines also add another collector,

usually Cyanamid 3477 (a dithiophosphate), which is

mixed with the xanthate at a ratio of up to 7:3. Copper

sulphate is added as an activator. The flotation circuit for

Merensky ore usually consists of a rougher stage and two

cleaner stages, in which closed-circuit cleaners are used.

To depress the talc, a depressent, such as Detrin orCarboxymethyl cellulose (CMC), is added to the rougher

and cleaner stages. The flotation of UG-2 ore is similar in

general to that of Merensky ore, but differs in a number

of aspects which is of no interest to this paper.

The Northam flowsheet (Fig. 5) is of interest here

because of flash flotation used in grinding circuit. The

Northam Merensky concentrator was designed to treat

270 t/h of run-of-mine (ROM) ore. The SAG mill dis-charge is fed to a ball mill operated in a closed circuit.

Flash flotation is utilized on the cyclone underflow. The

flash flotation cell recovers over 60% of the PGEs. The

cyclone overflow goes to flotation after it is conditioned.

Primary flotation is undertaken in a single rougher/

scavenger flotation bank. The concentrate of the first

cell in the rougher bank is collected as a part of the final

Fig. 6. The Stillwater flowsheet.


concentrate. The other rougher concentrate is further

processed with the cleaning circuit (no regrinding) that

consists of three column cells operating in series. The

cleaner concentrates are the main final concentrates. The

cleaner tail is recycled back to the head of the cleaner

circuit. The scavenger concentrate can be routed to the

head of the rough bank or to the cleaner circuit. Nogrinding is performed on the scavenger concentrate

(Cole and Ferran, 2002).

6.2.2. The Stillwater complex, Montana

The ore body at stillwater is referred to as the J-M

reef and is nearly 30 miles in length. The PGEs miner-

alization in the J-M reef is associated with disseminated

base-metal sulphides (BMS) minerals that from fine- tocoarse-grained aggregates that are moulded around and

are interstitial to the cumulus or earlier formed silicates.

Mineralogy: Zientek et al. (2002) reports that the

PGMs include palladium, platinum and ruthenium

sulphides; Pt and Pd tellurides and arsenides; and Pt–Fe,

Pt–Pd–Sn, Pd–Pb, Pd–Hg, Au–Pt–Pd and Rh–Pt alloys.

The dominant PGMs are braggite, cooperite, monche-

ite, vysotskite and isoferroplatinum. Platinum occurslargely as discrete PGMs: 67% as sulphides minerals

(braggite, cooperite); 25% as metal alloy (isoferroplati-

num); and 8% as telluride (moncheite). Palladium lar-

gely occurs in solid solution in pentlandite; 15% of the

palladium occurs in other sulphides minerals (vysotsk-

ite, braggite, cooperite) and 5% is associated with tel-

luride minerals (moncheite). There are some rare PGMs,

which include rustenbergite (Pd3Sn), hollingworthite(RhAsS), etc. The grain size of the PGMs is variable,

ranging from micron-size to grains with one side about

200 lm. Of the main PGMs, braggite and vysotskite arethe coarsest grained; the Pt–Fe alloy is finer.

The principal sulphides minerals are chalcopyrite

(CuFeS2), and pentlandite ((Ni,Fe)9S8). Most of the

platinum minerals are associated with the copper sul-

phide and the palladium with the nickel sulphides(Thurman and Allen, 1994). The ore contains approxi-

mately 3.5 times more palladium than platinum. Small

amount of gold and rhodium along with the copper and

nickel are also found. The ore contains less than 1% free

silica, but significant amounts of MgO bearing minerals

such as talc and serpentine. The ore is basic with a natural

pH of approximately 9.0 (Thurman and Allen, 1994).

Recovery method: Thurman and Allen (1994) andTurk (2001) provide a detailed summary of operations

and reagent scheme at the Stillwater Nye concentrator.

A SAG mill/ball mill grinding circuit with hydro-

cyclones for classification is used to process ROM ore in

stillwater. A flash flotation cell has been used in the

cyclone underflow to produce a final grade concentrate.

The cyclone overflow is fed to primary flotation which is

undertaken in three stages: rougher flotation, middlingflotation and scavenger flotation. There is a tertiary

milling stage between the rougher and middling flotation

circuit. The middling flotation concentrate is recycled

back to the head of the rougher circuit while the scav-

enger flotation concentrate is recycled to feed the mid-

dling flotation circuit. Cleaning is achieved in three

stages operating in a counter-current configuration. The

concentrate from the first stage of the first cleaner by-passes the second stages of cleaning and feeds the third

cleaner. The tailings from the first part of the first cleaner

are reground prior to further cleaning. The cleaner cir-

cuit tails recycle to feed the rougher flotation circuit.

Fig. 6 shows a schematic of the Stillwater flowsheet.

Thurman and Allen (1994) reported that the reagent

scheme is 0.09 lb/ton of potassium amyl xanthate and

0.08 lb/ton of dithiophosphate and 0.9 lb/ton of Carb-oxymethyl cellulose (CMC) used fro talc depression.

MIBC is added as froth conditions warrant. Sodium

hydrosulfide (NaHS) is added as a sulfidizer and pH is

controlled with sulfuric acid.

A bulk sulfide concentrates is produced averaging 60–

70 oz/ton of Pt +Pd from ore averaging 0.8 oz/ton.

Initially the concentrator operated at 500 tpd. It has

been steadily increased since 1987 start-up; concentratorthroughput for 1994 was around 1050 tpd increasing to

3000 tpd in 2001.

The overall platinum recovery after the flash cell

installation improved by 1.5% (Thurman and Allen,

1994). This pushes overall Pt recovery to 95%. This

improvement in recovery is due to the reduction in over

grinding of the Pt minerals. However, because of the

extremely short retention time in the flash cell (2–5 min),Pd recoveries were not affected.

6.2.3. Lac des Iles Ore

North American Palladium’s Lac des Iles mine began

commercial production in December 1993. It has since

commissioned a new 15,000 tonnes per day milling and

flotation circuit. The open-pit mine and milling opera-

tion is one of only two primary platinum group element


producers in North America. The mill treats an ore

containing 2g/t palladium (and 0.3 g/t total platinum,

gold and rhodium) producing a concentrate assaying

250g/t palladium, at a recovery of roughly 75% (Martin

and McKay, 2003).

Mineralogy: The major opaque minerals of the ore

are pentlandite, pyrite, chalcopyrite and pyrrhotite.Galena, magnetite and sphalerite are common but

minor minerals. Cabri and Laflamme (1979), on the

basis of the sample they studied, reported that the

principal PGMs are braggite-series minerals (braggite,

(Pt,Pd)S+ vysotskite, PdS), kotulskite, PdTe, isomerti-

ete, Pd11(As,Sb)4, merenskyite, PdTe2, sperrylite PtAs2,

moncheite, PtTe2; minor PGMs include stillwaterite,

Pd8As3, and palladoarsenide, Pd2As. Dunning (1979)reported that vysotskite is the most abundant PGMs in

the ‘‘Roby zone’’, where it frequently occurs with nickel

minerals, especially pentlandite.

Mineralogical studies were also performed by Martin

on five mill feed samples taken at different times during

the year of 2002. Sample assays ranged from 1.56 to 3.0

g/t Pd+Pt. By using LEO440 QEMScan system, a total

of 444 individual PGMs grains comprising 12 distinctPGMs species were identified. These are dominated by

tellurides, with lesser amounts of arsenides/antimonides,

sulphides and alloys. Martin and McKay (2003) also

reported that the kotulskite–telluropalladinite (Pd(Te,

Bi)–Pd9Te4) comprised roughly two-thirds of the PGMs

grain population. Palladoarsenide (Pd2As) was the next

most abundant PGMs species, comprising 20% of the

PGMS. Ten other PGMS species were observed. PGMsparticle size ranged from less than 1–15.7 lm (recalcu-

lated circular diameter). Some 45% of the PGMs oc-

curred as liberated grains, or inclusions within and

attachments to sulphides. The remainder occurred as

fine inclusions within, and attachments to silicate min-

erals (Fig. 7).

Recovery method: Flotation is the main method for

recovering the PGMs in Lac des Iles due to the fine sizedistribution of PGMs. One SAG mill feeds two ball

Fig. 7. The Lac des Iles flowsheet.

mills at a rate of 15,000 tonnes per day. A portion of the

SAG mill feed is crushed to )25 mm to increase

throughput. The product from the SAG mill circuit is

further ground in ball mills, before being delivered to the

flotation circuit. Two banks of roughers and scavengers

perform the primary flotation. The rougher concentrate

is reground and cleaned in a single stage column. Therougher cleaner tail joins the scavenger concentrate and

is reground in three verti mills. Cleaner flotation uses

two stages of cleaning with mechanical cells, and a third

stage using flotation columns, the column tails being

scavenged using a bank of mechanical cells. The flota-

tion reagent system consists of amyl xanthate collector,

dithiophosphate promoter, MIBC frother and a poly-

meric talc depressant, usually a form of carboxy methylcellulose.

The flotation circuit includes a fine grinding mill to

regrind the ore for further liberation due to the fine

distribution of PGMs in the ore. The expected feed

grade to the plant is 2 g/t Pd with targeted recovery in

excess of 80%. However, it is reported that the recovery

is around 75% due to the problem of grinding circuit

and other factors.

6.3. Cu–Ni sulphide deposits with by-product PGEs

The by-product PGEs of the Sudbury Cu–Ni deposits

was the principal source of PGEs prior to the discovery

and subsequent exploration of the PGEs-dominant

Merensky reef deposits. The production of Merensky

reef and, later, the Cu–Ni deposits of the Noril’sk areareduced Sudbury’s share of world PGEs production.

However, the by-product of PGEs mainly from Cu–Ni

sulphide deposits is still a very important source in the

world. Here, the Sudbury area and Noril’sk–Talnakh

area will now be discussed.

6.3.1. The sudbury area, Ontario

Mineralogy: Although several researchers have stud-ied the mineralogy of the Sudbury ores, published data

are very sparse on the PGEs contents. The principal

sulphides are pyrrhotite, chalcopyrite, and pentlandite;

several other minerals, such as cobaltite, pyrite, miller-

ite, cubanite, galena, sphalerite, magnetite, etc., occur in

minor and variable quantities in the ore.

Michenerite, PdBiTe, is the principal palladium min-

eral and sperrylite, PtAs2, is by far the most commonplatinum mineral for many deposits (South Range and

offset) in Sudbury area (Cabri, 1981b). Moncheite, PtTe2,

is the principal platinum mineral in deposits with essen-

tially no sperrylite (Levack West). Froodite, PdBi2, insi-

zwaite, PtBi2, and sudburyite, PdSb, are less common.

There are also some rare PGMs. According to Cabri

(1981), the PGMs occur as micron-sized inclusions, usu-

ally less than about 150 m in size, but millimeter-sizegrains (sperrylite) are also present. Recently studies by


Xiao and Laplante (2003) confirmed that the sperrylite

presents below 150 lm and is well liberated in the Clar-

abelle ore. Although sperrylite characteristically occurs as

monocrystalline inclusions or coarse grains, the other

PGMs are present more frequently as complex multimi-

neralic intergrowth, often with Bi and Ag tellurides.

Cabri and Laflamme (1976) made a detailed investi-gation of 150–300 lm sink fractions. They found that

although significant percentages of the PGMs are liber-

ated in this size class, especially for sperrylite, the

majority of PGMs are locked at that size in sink products.

About 5–15% of the michenerite, moncheite, sperrylite

and sudburyite grains found in the 150–300 lm sink

fraction occurred as inclusions in pyrrhotite or magnetite.

They concluded that the PGEs values might be ac-counted for by discrete PGMs and by PGEs in solid

solution in arsenides and sulpharsenides. They felt that

dilute PGEs solution in arsenides and sulpharsenides, if

present, would naturally be very important in view of

the large tonnages involved, but that conclusive evidence

was still required. Detailed mineralogical studies of a

second-stage mill sample and metal balance calculations

showed that, for the particular sample studied, from 1%to 4.5% Pt, 22.3% to 23.3% Pd and 31.2% to 41.2% Rh

are present as solid solutions in cobaltite and gersdorffite

(Cabri, 1981a).

The PGEs in Sudbury area is a by-product of

smelting and refining Ni–Cu concentrates. Recovery

values for PGEs are not available.

Cabri (1981a) reported that 66% of the sperrylite

found in a mill tailing was either liberated or attached tosulpharsenides. Because the sperrylite is the principal Pt

mineral in South Range mines and its close textural

association with sulpharsenides requires that both be

recovered, he suggested that it was worth investigating

the flotation characteristics of sperrylite.

6.3.2. The Noril’sk–Talnakh area, USSR

Kozyrev et al. (2002) reported that the Noril’sk–Talnakh PGEs–Cu–Ni sulphide ore was divided into

three economic types, namely, massive (rich) ore, dis-

seminated ore occurring in host rock, and stringer-dis-

seminated ore occurring in intrusive host rocks. The

mineralogy of these three ore types is different from each

other. Therefore, the corresponding concentrate meth-

ods are also different.

6.3.2.1. Massive ore mineralogy and recovery method.

The massive ores deposit refers to the Kharaelakh ore-

body, which occur at the base of layered intrusion and

contain more than 70 vol.% sulphides. The orebody is

subdivided into three groups, namely, pyrrhotite ore,

cubanite ore, and chalcopyrite ore. Pyrrhotite ore in-

clude pyrrhotite, chalcopyrite–pyrrhotite and cubanite–

chalcopyrite–pyrrhotite varieties, which make up 85vol.% of the Kharaelakh orebody. Cubanite ore makes

up only 7–8 vol.%, and chalcopyrite ore also 7–8 vol.%

of the Kharaelakh orebody.

Mineralogy: Variability in the chemical composition of

massive ore manifests itself in a gradual increase of

PGEs content from 2.3–11.9 ppm up to 20.2–111.7 ppm

(Kozyrev et al., 2002). The pyrrhotite ore has few PGMs

in the )45 lm (75 wt.%) and 45–90 lm (25 wt.%) sizefractions. These include isoferroplatinum (IFP), sper-

rylite, cooperite, rustenbergite, Kotulskite, merenskyite

and native gold. The dominant PGMs is IFP, which

constitutes 99 wet% of the total PGMs assemblage

((Kozyrev et al., 2002).

Kozyrev et al. (2002) also characterized mineralogy

of PGMs in the feed and concentrates of chalcopyrite–

pyrrhotite and cubanite–chalcopyrite ores at laboratory-scale. It was found that chalcopyrite–pyrrhotite ore and

process products contain 26 PGMs species. Palladium

minerals are dominant (16 species), Pt constitutes nine

minerals while Rh occurs in only one species. Most of

the PGMs occur in the )45 lm size fraction (67–91

wt.%) of the primary ore as well as of the process

products. The 45–90 lm size fraction of the nickel

concentrate contains 66 wt.% of the PGMs. PGMs havenot been detected in the tailings.

The major PGMs are amenable to flotation: sperrylite

and cooperite are concentrated in copper concentrate

(69% and 90% of the total PGMs, respectively) and in

the nickel concentrate (29% and 8%); stibiopalladinite

(74%) favours the nickel concentrate and the pyrrhotite

concentrate hosts 70% of isoferroplatinum.

Kozyrev et al. (2002) also reported that the cubanite–chalcopyrite–pyrrhotite ore and concentrates contain

the most PGMs in the 20–45 lm size fraction (42–84

wt.%). Twenty-nine PGMs have been identified and

these include 18 Pd species and 11 Pt species.

The major PGMs in the flotation feed are cooperite

(28 wt.%), atokite–rustenburgite (24 wt.%) and sperry-

lite (20 wt.%); copper concentrate has sperrylite (42%),

cooperite (24 wt.%) and isoferroplatinum (14 wt.%);nickel concentrate contains cooperite (36 wt.%), sper-

rylite (15 wt.%), atokite–rustenburgite (23 wt.%) and

sobolevskite (11 wt.%); the pyrrhotite concentrate has

isoferroplatinum (41 wt.%) and cooperite (20 wt.%)

while the tailings have atokite–rustenburgite (54 wt.%)

and sperrylite (17 wt.%). Most of the major PGMs

grains in the concentrate are liberated while some occur

as complex intergrowths with each other, with sulphidesand with silicates.

Recovery method: In the actual plant, massive ores,

ground at 85% passing 45 lm, are being processed

according to a direct selective flotation flowsheet (Fig. 8)

resulting in an output of copper, nickel and pyrrhotite

concentrates and tailings. In copper flotation, pine oil is

used as a frothing agent and butyl dithiophosphate as a

collector; in nickel flotation, T-80 serves as the frother,potassium butyl xanthate as the collector with sodium

Ore

Grinding (80% - 45 m)

Copper flotation

Nickel flotationCopper Conc.

Pyrrhotite flotationNickel Conc.

Pyrrhotite Conc. Tailings

µ

Fig. 8. Selective flotation flowsheet for massive ore.


dimethyl dithiocarbamate as a depressant, NaHSO3 as a

modifier and CaO is used to regulate alkalinity.

6.3.2.2. Disseminated ore mineralogy and recovery

method. Three varieties of disseminated ores have been

distinguished: (1) pyrrhotite ore, the most abundant, (2)

cubanite, and (3) chalcopyrite ores, which are less

abundant. The major sulphide minerals in pyrrhotite oreare the pyrrhotite-group minerals, chalcopyrite, pent-

landite, with some minor and accessory minerals. Major

sulphide minerals in the cubanite ore are cubanite,

chalcopyrite, pyrrhotite and pentlandite; minor sulp-

hides includes mackinawite, sphalerite, galena etc. The

chalcopyrite ore includes the following major sulphides:

chalcopyrite-group minerals, pentlandite, pyrrhotite-

group minerals and cubanite. Some minor sulphideminerals also appear in this type of ore.

With new technology developed in the past decade,

gravity–flotation techniques have been developed to in-

crease the PGMs recovery, such as recovering PGMs

from concentrates, as well as scavenging of the Noril’sk

mill tailing using gravity concentration methods (Ko-

zyrev et al., 2002).

The Knelson concentrator has been used by Kozyrevet al. (2002) to study the PGMs distribution in the

Noril’sk mill gravity products. These include:

(1) Gravity concentrate of the 12 in. Knelson concen-

trator (laboratory-scale test);

Table 3

PGMs distribution (vol.%) by size fraction in disseminated ore, Noril’sk I o

Size fraction (lm) Conc. of the 12 in. KC Feed of the 48

+250

)250+ 90 7.8

)90+ 45 40.8

+74 0.2

)74+ 45 32.1

)45+ 20 67.7 50.9

)20 0.5

(2) Feed of the 48 in. Knelson concentrator;


trator; and


trator (processed concentrate of the 48 in. Knelson

concentrator).

Mineralogy: Kozyrev et al. (2002) reported that 26

PGMs species plus alloys of gold and silver have been

found in the gravity concentrates of Knelson 12 in. con-

centrator and include 19 Pd minerals, 6 Pt minerals and a

Rh minerals. Most of PGMs recovered is in minus 45 lmand 45–75 lm size fractions. In the latter size fraction, the

most abundant (80 wt.%) of platinum minerals are iso-

ferroplatinum, rustenbergite, and sperrylite; in the for-mer, Pd species of the atokite–rustenburgite series

dominate, with a twofold decrease of Pt–Fe alloy. Most

PGMs grains and intergrowths (90%) in concentrates are

liberated while the rest (10%) are attached to sulphides.

The PGMs distribution is shown in Table 2.

Table 3 shows that PGMs in the feed of 48-in. KC are

mainly distributed in the 45–90 and 20–45 lm size

fraction, 41 and 51 vol.% respectively. Thirty-ninePGMs species including 26 Pd minerals, 12 Pt minerals

and 1 Rh mineral, have been found. The major PGMs

include atokite (28%), sperrylite (16%) and isoferro-

platinum (11%) plus other minor PGMs. The main

platinum carriers are isoferroplatinum, atokite, sperry-

lite and tetraferroplatinum, and palladium minerals

comprises mostly atokite and to a lesser degree, paolo-

vite and taimyrite. Most of the major PGMs grains arefree while the rest occur as intergrowths with other

PGMs, Au–Ag alloys, sulphides and silicates.

The PGMs in the concentrates of 48-in. Knelson

concentrator are mainly distributed in the 90–250 lm(25%), 45–90 lm (29%), and 20–45 lm (45%) fractions.

A total 28 PGMs species, plus Au and Ag minerals, has

been found. The major PGMs include isoferroplatinum

(41%), tetraferroplatinum (18%) and atokite (15%) plus(3–9%) sperrylite rustenbergite and taimyrite. More than

half of the Pt is contained in isoferroplatinum, as well as

tetraferroplatinum, sperrylite and rustenburgite.

It is not clear why the authors used the 20-in. Knelson

concentrator to process the concentrates from the 48-in.

Knelson concentrator.

rebody (Kozyrev et al., 2002)

in. KC Conc. of the 48 in. KC Conc. of the 20 in. KC

6.7

18.6 34.8

29.5 25.4

45.0 38.6

0.2 1.2

Ore Gravity conc.

KC Ball millSD-48

Rougher TailScavenger

1st cleaner

2nd cleaner

Final Conc.

Fig. 9. Noril’sk flowsheet for processing the disseminated ore.


The major PGMs consist of isoferroplatinum (35%),

tetraferroplatinum (20%) and atokite (20%) plus minor

sperrylite, rustenburgite and taimyrite (3–9%). Based on

the information provided, it is clear that the main Pt and

Pd carriers in this product are much the same as those in

the 48 in. Knelson product.Recovery method: In actual practice, disseminated

ores are preconcentrated by gravity methods and are

then processed a selective bulk flotation (Fig. 9). The

fineness of grinding is about 55% )74 lm. Gravityconcentration uses Knelson concentrators of various

capacities to separate the noble-metal minerals. Bulk

flotation produces a concentrate that undergoes further

grinding to further liberate the minerals. Cleaning stagesare used to produce higher-grade concentrates, which

are processed further to produce copper and nickel

concentrates. For flotation, potassium butyl xanthate

and sodium butyl dithiophosphate serve as collectors,

T-80 as a frother and CaO as a depressant.

Stringer-disseminated ore mineralogy and recovery

method is not discussed in this article due to the minor

amount of PGMs in the ore.

7. New trends in research and development of PGMs

recovery

7.1. Intensive mineralogy research and development

The strong correlation between the mineralogy andplant performance discussed above drives the mining

industry put more efforts to the applied/process min-

eralogy studies. Freeman (2003) advocated that min-

eralogical identification should be conducted early and

often, during exploration. Such effort would lead to a

maximum benefit of a project. For example, early

determination of whether the PGMs were sulphides or

alloys and whether there is a presence or absence ofbismuth, tin, tellurium or arsenic-bearing PGMs can

benefits the overall recoveries during flotation, smelting

and refining operation. Identification of the range of

PGM grain sizes can help predict whether or not over-

grinding of the PGMs is likely or occur during grind-

ing. Determination of the host minerals is also very

critical, for example, during froth flotation, talc can

cause sliming or coating of PGM grains, making them

difficult or impossible to float. Some PGEs are in the

solid solution of pyrrhotite grains in copper–nickelores, when the operation try to depress or actively re-

ject pyrrhotite from the froth concentrate to recover Cu

and Ni, the operation may reject high unit-value

PGMs.

After the commission of plants, more efforts also are

putting to the mineralogy studies, which try to optimize

the performances. In South Africa, the platinum

industry is well known to use QEMScan or MLAimaging analysis technology intensively to improve both

final recovery and final concentrate grade. In Australia,

QEMScan or MLA are used intensively in liberation

analysis. In Canada, QEMScan at SGS Lakefield re-

search have been used to gain an understanding of the

mineralogical factors affecting plant performance, and

this has helped in the post-commissioning optimization

of the mill (Martin and McKay, 2003). It was used tostudy the role of process mineralogy in optimizing

mineral processing at North American palladium’s Lac

Des Iles mill. The process mineralogy developed at

Falconbridge Limited has been applied to the Raglan

mill to increase both the recovery and grade (Lotter

et al., 2002). However, a sound and cost-effective pre-

concentration method of sample preparation for the

mineralogy study is not universal and still underdevel-opment to obtain the precise mineralogical information.

7.2. Characterization of the gravity recoverable platinum

group minerals in ores

Not only can the characterization of the gravity

recoverable platinum group minerals in an ore measure

the potential response to gravity, but also can be a veryeffective preconcentration method for sample prepara-

tion for some PGMs mineralogical study.

Recently, a renewed interested in the use of gravity

concentration in recovering PGMs is being studied due

to the economic importance and new available gravity

concentration devices (Cole and Ferron, 2002; Kozyrev

et al., 2002). Further, the flash flotation is more attrac-

tive application due to the capability of quickly lowingcirculating load of gold and relatively low cost of install-

ing a flash unit (Laplante and Xiao, 2002). In conjunc-

tion with this application of gravity concentration is

the desire to predict or determine how much gravity

recoverable platinum group minerals (GRPGMs) is

present in an ore. GRPGMs refer to the portion of

platinum group minerals in an ore or stream that can be

recovered by gravity at a very low yield (<1%). It in-cludes PGMs that are totally liberated, as well as PGMs

Table 4

Summary of the method used in PGMs recovery plant

Name of mills Option used in grinding circuit Location Recovery

Northam Flash flotation Cyclone underflow 60% Pt

Noril’sk Knelson concentrator First crushing stage 50–60% Pt

10–13% Pd

17–20% Au

Stillwater Flash flotation Cyclone underflow 50–60% Pt

30–40% Pd

Lac des Iles N/a N/a


in particles that are not totally liberated but with such

density that they report to the gravity concentrates.

Conversely, it excludes any fine, completely liberated

PGMs that are not recovered into concentrates.

The GRPGMs test can be adopted from the Gravity

recoverable gold (GRG) test, which is a concept used to

characterize ores for their gravity recoverable gold

content. The amenability of an ore to gravity recovery isthe single most important parameter to justify the

installation of a gravity circuit (Laplante et al., 1993).

Therefore, the ore must be characterized for its gravity

recovery potential, as it is ground and progressively

liberated. This is the most common definition of GRG

and the GRG standard test is designed to address this

question (Xiao, 2001).

The transfer from GRG to GRPGMs is based on thespecific gravity range for some platinum group minerals

(between 10 and 22) is similar to that of gold (between

16 and 19). However, the PGMs tend to report to finer

size classes (Xiao and Laplante, 2003), the test needs to

be modified. The modified procedure includes the stan-

dard GRG protocol, but a fourth stage is added at a

higher rotating velocity. Gravity research group in

McGill University is developing this technique.

7.3. Use of flash flotation or centrifugal gravity concen-

trators

As shown in Table 4, the flash flotation has been used

in Northam mill in South Africa, Stillwater mill in

North America, and gravity recovery using Knelson

concentrator in Noril’sk, Russia. For gold, the bestprocessing methods for recovering gold are determined

by the mineralogy, particle size distribution and the

behavior of gold in grinding circuit. Both the gravity

recovery and flash flotation are being used as additional

recovery steps to recovery gold. Gravity recovery aims

at the coarser gold ahead of cyanidation or base metal

flotation, and flash flotation was extremely efficient in

dropping the circulating load of gravity recoverable goldat fine range, typically below 106 lm to below 212 lm(Laplante and Xiao, 2002). Although the flash flotation

is used for recovering platinum group minerals, the ra-

tional and the behavior of PGMs still needs to be sorted

out. Based on the PGMs mineralogical investigation and

size distribution in Noril’sk mine, the gravity recovery

by using KC SD-48 has been used to recover PGMs in

crushing stage. However, the behavior of PGMs in

crushing, grinding and classification units needs to be

further investigation.

8. Conclusions

In this review, documented information about the

classification of PGEs ores, the techniques of sample

preparation for mineralogy study was examined. It was

found that for different ore types the preconcentration

methodsdiffer.Asdo themineralogical features.Although

the preconcentrationmethod is complicated in some cases,

it is very effective for the mineralogical analysis.Mineralogical analysis is very important to choose

the flowsheet for recovering the PGMs. It is also critical

to optimize the actual plant for improving the perfor-

mance.

Flash flotation and gravity recovery using the cen-

trifugal equipment was used in some major PGMs

producer plant.

References

Begizov, V.D., Borisenko, L.F., Uskov, Y.D., 1975. Sulphides and

natural solid solutions of platinum metals from the ultrabasites of

the Gusevogorskiy Pluton. Urals. Doklady Akad. Nauk SSSR 225,

1408–1411.

Blagodatin, Y.U., Distler, V.V., Zakharov, B.A., Sluzhenikin, S.F.,

2000. Disseminated ores of the Noril’sk ore district as a potential

for increasing output of platinum metals at the Noril’sk Mining

Company, Obogashcheniye Rud, special issue, pp. 3–9.

Brynard, H.J., de Villiers, J.P.R., Viljoen, E.A., 1976. A mineralogical

investigation of the Merensky reef at the Western Platinum Mine,

near Marikana, South Africa. Econ. Geol. 71, 1299–1307.

Cabri, L.J., 2002. The platinum-group minerals. In: Cabri, L.J. (Ed.,)

The Geology, Geochemistry, Mineralogy and Mineral Benefication

of platinum-group elements. Canadian Institute of mining and

metallurgy, Special volume 54, 13–119.

Cabri, L.J., 1981a. In: Cabri, L.J. (Ed.), Platinum Group Elements:

Mineralogy, Geology, Recovery. Canadian Institute of Mining and

Metallurgy, Special Volume 23.

Cabri, L.J. 1981b. Relationship of mineralogy to the recovery of

platinum-group elements from ores. In: Cabri, L.J. (Ed.), Platinum

Group Elements: Mineralogy, Geology, Recovery. Canadian


Institute of Mining and Metallurgy, Special Volume 23. pp. 233–

252 (Chapter 11).

Cabri, L.J., Laflamme, J.H.G., 1976. The mineralogy of the platinum-

group elements from some copper-nickel deposits of the Sudbury

area. Econ. Geol. 71, 1159–1195.

Cabri, LJ., Laflamme, J.H.G., 1979. Mineralogy of samples from the

Lac des Iles area, Ontario. CANMET Report 79–27.

Cole, S., Ferron, C.J., 2002. A review of the benefication and extractive

metallurgy of the platinum-group elements, highlighting recent

process innovations. In: Cabri, L.J. (Ed.), The Geology, Geochem-

istry, Mineralogy and Mineral Benefication of platinum-group

elements. Canadian Institute of mining and metallurgy, Special

volume 54, pp. 811–818.

Danilatos, G.D., 1991. Review and outline of environmental SEM at

present. J. Microsc. 162 (Pt 3), 391–394.

Dunning, G.R., 1979. The geology and platinum group mineralization,

Lac des Iles Complex, Northwestern Ontario, MSc thesis, Carleton

University.

Farley, A.N., Shah, J.S., 1988. A new detection technique for high

pressure scanning electron microscopy. In: Goodhew, P.J., Dick-

ison, H.G. (Eds.), Proc. EUREM 88. In: Inst. of Physics Confer-

ence Series, No. 93, vol. 3. IOP Publishing, pp. 241–242.

Farley, A.N., Shah, J.S., 1991. High-pressure scanning electron

microscopy of insulating materials: A new approach. J. Microsc.

Part 2 164 (November), 107.

Freeman, C., 2003. Mineral beneficiation of the platinum group

elements: A view through the hand-lens. Eng. Mining J. 3, 241–245.

Guerney, P.J., Laplante, A.R., O’Leary, S., 2003. Gravity recoverable

gold and the mineral liberation analyzer. In: CMP Conference,

Ottawa, pp. 402–412.

Henley, K.J., 1983. Ore-dressing mineralogy: A review of techniques,

applications and recent develops. Geol. Soc. South Africa 7, 175–

200 (Special Publication).

Johnson Matthey, 2002. http://www.matthey.com.2003.

Knauf, V.V., Kozyrev, S.M., 2002. Knelson’s concentrator: Extraction

evaluation, website of www.natires.com.

Kojonen, K., Isomaki, O.P., Pulkkinen, K., Knauf, V., 2003. Mineral

processing of base metal sulphides and platinum group minerals

(PGMs) at the proterozoric hitura Ni–Cu–PGEs deposit, Western

Finland. Proc. Appl. Mineral., Finland.

Kozyrev, S.M., Komarova M.Z., Emelina, L.N., Olesnkevich, O.I.,

Yakovleva, O.A., Lyalinov, D.V., Maximov, V.I., 2002. The

mineralogy and behaviour of PGM during processing of the

Noril’sk-Talnakh PGE–Cu–Ni ores. In: Cabri, L.J. (Ed.),

The Geology, Geochemistry, Mineralogy and Mineral benefication

of platinum-group elements. Canadian Institute of mining and

metallurgy, Special volume 54, pp. 757–773.

Laplante, A.R., 2001. Report on the Characterization of Gravity

Recoverable Gold in a Sample of Kemess #1 Ore, Technical

Report, McGill University, January.

Laplante, A.R., Putz, A., Huang, L., 1993. Sampling and sample

processing for gold gravity circuits. In: Proceedings, Professional

Development Seminar on Gold Recovery by Gravity, McGill

University, Montreal, May.

Laplante, A.R., Xiao, Z., 2002. Introduction to gold recovery by

gravity. Presented at the Quinto simposium international del Oro

Lima, Peru.

Lotter, N.O., Whittaker, P.J., Kormos, L., Stickingand, J.S., Wilkie,

G.J., 2002. The development of process mineralogy at Falcon-

bridge Limited and application to the Raglan Mill.

Martin, C.J., McKay, N.A., 2003. The role of mineralogical studies in

optimizing mineral processing at North American Palladium’s Lac

Des Iles Mill. 2003 Mineralogy Conference, Finland.

McDonald, D., 1960. A History of Platinum. Johnson Matthey and

Co. Ltd, London.

Mertie, J.B. Jr., 1969. Economic geology of the platinum metals. US

Geol. Survey, Prof. Paper 938.

Moncrieff, D.A., Robison, V.N.E., Harris, J.B., 1978. Neutralization

of insulating surfaces in the scanning electron microscope. J. Phys.

D: Appl. Phys. 12, 2315–2325.

Naldrett, A.J., Macdonald, A.J., 1980. Tectonic settings of some Ni–

Cu sulfide ores: their importance in genesis and exploration. In:

The continental crust and its Mineral Deposits. Geol. Assoc.

Canada Special Paper No. 20, pp. 633–657.

Petruk, W., 2000. Applied Mineralogy in the Mining Industry. Elsevier

Publishers.

Petruk, W., Hughson, M.R., 1977. Image analysis evaluation of the

effect of grinding media on selective flotation of two zinc–lead–

copper ores. CIM Bull. 70, 782, pp. l28–135.

Pignolet-Brandom, S., Reid, K.J., 1988. Mineralogical characteriza-

tion by QEM*SEM. In: Carson, D.J.T., Vassoilou, A.H. (Eds.),

Process Mineralogy VIII, EDX. TMS, Warrendale, PA, pp. 337–

346.

Razin, L.V., 1976. Geologic and genetic features of forsterite dunites

and their platinum-group mineralization. Econ. Geol. 71, 1371–

1376.

Reid, K.J., Pignolet-Brandom, S., 1988. Application of QEM*SEM

for beneficiation studies of Minnesota Taconite. In: Carson, D.J.T.,

Vassoilou, A.H. (Eds.), Process Mineralogy VIII, EDX. Minerals,

Metals and Materials Society, Warrendale, PA, pp. 369–377.

Reimer, T.O., 1979. Platinoids in auriferous proterozoric conglomer-

ates of South Africa. Evaluation of existing data. Neues. Jb. Miner.

Abh. 135, 287–314.

Robinson, B.W., 1998. The ‘‘GEOSEM’’ (low-vacuum SEM): An

underutilized tool for mineralogy, In: Cabri, L.J., Vaughan, D.J.

(Eds.), Modern Approaches to Ore and Environmental Mineral-

ogy, EDX, Mineral. Association, Can., Short Course 27, pp. 139–

151.

Schwellus, J.S.I., Hiemstra, S.A., Gasparrini, E., 1976. The Merensky

reef at the Atok platinum mine and its environs. Econ. Geol. 71,

249–260.

Sizgoric, M.B., 1985. Tracking platinum-group minerals in the milling

of Cu–Ni ores at Sudbury, In: Proceedings of the Ninth Interna-

tional Precious Metals Institute Conference, New York, pp. 923–

932.

Thurman, W.M., Allen, A.R., 1994. Flash flotation at stillwater

mining Co. Paper presented at the Montana Mining Association

Annual Meeting and Convention, Butte, Montana.

Turk, D.J., 2001. Stillwater Mining Company Nye Concentrator

Operation, SME Annual Meeting.

Vermaak, C.F., Hendriks, L.P., 1976. A review of the mineralogy of

the Merensky reef with specific reference to new data on the

precious metal mineralogy. Econ. Geol. 71, 1244–1269.

Williamson, J.E., Savage, J.A., 1965. The determination of Osmirid-

ium in Witwatersrand Ores. J. South African Inst. Min. Metall.

(January), 343–356.

Woodcock, F., Laplante, A.R., 1993. A laboratory method to measure

free gold content. In: Proceedings, Randol Gold Seminar, Beaver

Creek, September, pp. 151–155.

Xiao, Z., 2001. Developing simple regressions for predicting gold

gravity recovery in grinding circuit. Master of Engineering thesis,

McGill University, Montreal, pp. 9–14.

Xiao, Z., Laplante, A.R., 2003. Characterization of gravity recover-

able platinum group elements in a drill ore sample from Ni Rim

South. Report, 9 June 2003.

Zhou, Z., Zhang, D., 1975. Heavy mineral separation for platinum-

group minerals in platinum-bearing chromite deposits. Acta Geol.

Sin. (2), 187–193, in Chinese with English abstract.

Zientek, M.L., Cooper, R.W., Corson, S.R., Geraghty, E.P., 2002.

Platinum-group element mineralization in the Stillwater Complex,

Montana. In: Cabri, L.J. (Ed.), The Geology, Geochemistry,

Mineralogy and Mineral Beneficiation of Platinum-Group Ele-

ments. Canadian Institute of Mining, Metallurgy and Petroleum,

CIM special vol. 54, pp. 459–482.

http://www.matthey.com.2003

http://www.natires.com

Documents

PGM Recovery