Minerals Engineering 18 (2005) 1010–1019

This article is also available online at:

www.elsevier.com/locate/mineng

Applications of PIXE and diagnostic leaching

in the characterisation of complex gold ores
a,*
W.R. Goodall , P.J. Scales a, C.G. Ryan b

a
Particulate Fluids Processing Centre, Department of Chemical and Biomolecular Engineering, University of Melbourne, Victoria 3010, Australia
b
CSIRO Exploration and Mining, North Ryde, NSW 2113, Australia

Received 16 December 2004; accepted 16 January 2005

Available online 5 March 2005

Abstract

A new method using a combination of Proton Induced X-Ray Emission (PIXE) and diagnostic leaching has been used to char-
acterise the associations and distribution of gold in a variety of ores from the Pilbara region of Western Australia. The different ore
types studied highlighted the versatility of the approach and has demonstrated its applicability to any type of ore. Characterisation
of an oxide ore deposit has shown how the technique can be used to gain a good initial understanding of potential metallurgical
problems for a new ore. From the application of the approach to the study of two other types of refractory ores it was possible
to demonstrate that in these ores the gold occurs locked within the sulphide matrix as ‘‘invisible gold’’ and that the form of its occur-
rence affects its recovery.

2005 Elsevier Ltd. All rights reserved.

Keywords: Sulphide ores; Liberation analysis

1. Introduction deposits to sub-micron sized particles or even gold in so-

lid solution with the ore minerals. Ore deposits with pre-
In recent years, the gold mining industry has faced a dominantly free gold that is easily amenable to cyanide
depletion of readily extractable oxide ores and an in- extraction have previously been defined as free-milling
crease of more problematic sulphide ores, which gener- ores and deposits where gold is not amenable to a direct
ally underlie the oxide deposits. This has made it ever cyanide leach and requires a pre-treatment stage to lib-
more important to understand the metallurgical prob- erate the gold are termed as refractory (Marsden and
lems associated with these sulphide ores and develop House, 1992).
economically viable extraction techniques. One thing The refractory nature of gold deposits can be attrib-
that is vital in the development of an optimised process- uted to a number of different factors. The most common
ing route is a good understanding of the distribution and cause of metallurgical problems is the encapsulation of
associations of gold. gold by sulphide minerals (Marsden and House, 1992).
Gold can be associated with a variety of minerals This problem can often be overcome by fine grinding
within an ore matrix in many different ways. Knowing of the ore to liberate macroscopic gold, however, in
these associations can be vital in optimisation of the some cases the particles of gold may be too small to
extraction route. The occurrence of gold can range from be effectively liberated by fine grinding or may even be
large nuggets of free gold generally found in alluvial in solid solution. When this occurs gold is termed as
‘‘invisible gold’’, which has been defined by Boyle
*
Corresponding author. Tel.: +61 3 8344 6654; fax: +61 3 8344 6233. (1979) as colloidal or chemically bound gold that is
E-mail address: w.goodall@pgrad.unimelb.edu.au (W.R. Goodall). undetectable by microscopic techniques. ‘‘Invisible

0892-6875/$ - see front matter
2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.mineng.2005.01.011
 W.R. Goodall et al. / Minerals Engineering 18 (2005) 1010–1019 1011

gold’’ is generally seen in arsenopyrite or arsenic rich ysis (EMPA), which has an achievable MDL of 50–
pyrite. A number of studies have determined the mech- 100 ppm when great care is taken for sample prepara-
anism of occurrence of ‘‘invisible gold’’. Cook and tion but is generally >500 ppm (Sie et al., 1991). In addi-
Chryssoulis (1990) showed that ‘‘invisible gold’’ will tion to lower detection limits, the use of protons in
only occur in pyrite once a certain bulk concentration PIXE rather than electrons to form the beam allows
of arsenic has been reached, while Vaughan (1995) the beam to penetrate up to 30 lm into the surface
showed that gold will be concentrated in arsenic rich (Sie et al., 1991). This allows the detection of elements,
rims of pyrite. It has been suggested that in arsenopyrite, which may be buried in the sample or are covered by
depletion of Fe with gold concentration suggests Au an oxidised layer. This feature has also been used to ana-
substitutes for Fe in the sulphide matrix (Genkin lyse fluid inclusions in quartz in-situ to give information
et al., 1998) and in pyrite the incorporation of As into on the rock forming processes (Ryan et al., 2001b).
the matrix could lead to distortion of the lattice allowing As with any X-ray spectrometry care must be taken
the incorporation of gold (Tarnocai et al., 1997). with PIXE to allow for interferences caused by overlap
Knowing the distribution and association of gold of the X-ray line (Annegarn and Bauman, 1990). This
within ores can help not only in identifying a processing is especially relevant for gold analysis as the tails of
route but also in the optimisation of an accurate assay the arsenic and tungsten lines overlap the gold line mak-
technique. It is often assumed that a simple fire assay ing analysis uncertain in matrices high in these elements.
or aqua regia digestion will give an accurate head grade In recent years, PIXE has been used for a wide range
for any ore. This is not always true and optimisation of of investigations within the earth sciences. A number of
the fire assay analysis is often required, which is difficult reviews of some of these applications have been com-
if the true head grade is unknown. These problems are pleted by Ryan (1995), Annegarn and Bauman (1990)
most often encountered where gold is in solid solution and Sie et al. (1991). PIXE has also been used exten-
within a sulphide or in a gold-telluride mineral and it sively for analysis of distribution of gold within sulphide
is for this reason that a good understanding of the asso- ores. One example of this application was a study by
ciations and distribution of gold is essential. This infor- Foya et al. (1999) who demonstrated ‘‘invisible gold’’
mation can be used to identify if gold occurs in multiple associated with zones of high As in the Kimberly Reefs
forms and if all of those forms are amenable to the assay of the Witwatersrand Basin.
technique being utilised. If they are not, then a new as-
say technique can be developed to ensure accurate head
grade analysis. 3. Diagnostic leaching
The present investigation seeks to show how a combi-
nation of analytical (Proton induced X-ray emission) Diagnostic leaching was developed by Anglo Ameri-
and chemical (Diagnostic Leaching) techniques can be can Research Laboratories in the late 1980s to provide
used to identify the distribution of gold within a given a way to distinguish the deportment of gold within ore.
ore and hence shed light on potential metallurgical This is achieved by selective destruction by oxidative
and assay problems. acid leach of minerals, with each stage followed by a
cyanide leach to recover gold liberated. Generally it is
expected that in addition to minerals destroyed at each
2. Proton induced X-ray emission (PIXE) stage approximately 10% of the next most stable min-
eral will also be destroyed (Lorenzen and Tumilty,
Proton induced X-ray emission (PIXE) is a non- 1992).
destructive, simultaneous trace multi-element analytical As an analytical technique, diagnostic leaching is very
technique. Protons, focussed into a beam spot on the useful in establishing the ratio of gold associated with
sample in a nuclear microprobe (NMP), ionise atomic different mineral phases within a given ore. It can be
electrons producing inner-shell vacancies. De-excitation applied both to characterisation of new ores and to
of these vacancies causes the emission of characteristic problem solving in well-characterised ores, where metal-
X-rays for that element. A spectrum of characteristic lurgical problems are experienced. Studies of this nature
X-rays is then developed, with each event labelled with have been completed by Teague et al. (1998) who used
the coordinates of the scanning beam to permit images diagnostic leaching to show the behaviour of gold and
of the elemental composition of the target mineral grain gold minerals in froth flotation. Lorenzen and Tumilty
to be constructed (Johansson et al., 1995). (1992) used diagnostic leaching to show the effect of
As a trace elemental analytical technique PIXE is the addition of reagents to the performance of a gold
very powerful with minimum detection limits (MDL) plant. The technique has also been used to identify the
between 0.1 and 50 ppm depending on the element and refractory nature of gold ores and determine possible
host matrix (Johansson et al., 1995). This is in contrast extraction routes (Lorenzen and van Deventer, 1992;
to similar techniques such as electron microprobe anal- Lorenzen and van Deventer, 1993).
1012 W.R. Goodall et al. / Minerals Engineering 18 (2005) 1010–1019

When combined with PIXE analysis diagnostic leach- a combination of PIXE and diagnostic leaching to
ing can be used to establish the bulk ratio of gold asso- identify the association of gold within sulphide ores
ciated with each mineral phase. This information can be and to allow the selection of an effective pre-treatment
used to verify if gold observed in grains by PIXE is rep- stage.
resentative of the bulk sample or simply an abnormality. Previous metallurgical and processing trials using this
ore have identified a number of problems. Recovery of
gold by direct cyanide leach was observed to account
4. Materials for only 75% of the fire assay head grade with high re-
agent consumption. To help overcome these problems
Three ores from the Pilbara region of Western Aus- it was hoped that diagnostic leaching could be used to
tralia were selected for this investigation. A variety of identify the distribution of gold in the bulk phases along
ore types were chosen to show the versatility of using with PIXE to show how gold was associated within each
PIXE and diagnostic leaching to characterise the of these phases.
extractability and assay amenability of different types
of gold ores. Ore 1 was an oxide ore, which has been 4.3. Ore 3
found to have no metallurgical processing or assay
problems. Ore 2 was a sulphide ore that has shown typ- Ore 3 is a complex sulphide ore, which historically
ical characteristics of a refractory ore, with high cyanide has shown numerous metallurgical problems. It was
consumption and low recoveries without pre-treatment. chosen to show how the PIXE and diagnostic leaching
Ore 3 was selected as a complex sulphide ore for which techniques could be used to show the distribution of
there has been a history of metallurgical problems and gold within a complex ore and establish a processing
was known to respond poorly to traditional assay tech- route based on this distribution.
niques such as fire and aqua regia assay methods. This ore has been shown to respond poorly to tradi-
Preliminary head grade analysis and percentage tional assay techniques such as aqua regia dissolution
extraction by cyanide of each ore are shown in Table and fire assay. These techniques have consistently shown
1. Head grade analysis was by commercial fire assay a wide variability in results suggesting that poor extrac-
and cyanide amenability was based on a 24-h agitated tion/collection is occurring. This can be seen from
leach in 1000 ppm NaCN. Tables 1 and 2, which show very poor extraction by cya-
nide and a large standard deviation associated with
4.1. Ore 1 other assay techniques.
The poor response of the ore to cyanidation (26%)
The deposit from which this ore was selected was a suggests that the high error associated with assay is un-
recent discovery that had undergone very little metallur- likely to be simply the result of a ÔnuggetÕ effect associ-
gical characterisation. It was chosen to demonstrate how ated with coarse free gold.
the combination of PIXE and diagnostic leaching could This observation along with the poor response to tra-
be used early in the characterisation of a new ore to ditional assay techniques suggests that there is a form of
establish if metallurgical problems were likely to be gold present in Ore 3 that is not amenable to direct cyan-
experienced. It was not expected that metallurgical idation or consistent assay.
problems would be identified, as it is essentially an oxide By using the combination of PIXE and diagnostic
ore consisting predominantly of quartz and hematite. leaching it was hoped that the distribution of gold with-
This mineralogy suggests that the gold should occur as in this ore could be established and an accurate head
free grains and that the ore should be free milling. grade determined. This information could then be used
to establish a reliable assay technique for use with ores
4.2. Ore 2 of this type and an optimised processing route
determined.
Ore 2 was chosen as a typical refractory sulphide ore.
It was selected to demonstrate the effectiveness of using

Table 2
Table 1 Assay error associated with traditional assay techniques for Ore 3
Fire Assay head grade and cyanide extraction of ores investigated
Mean grade Standard
Head grade (g/t Au) Cyanide extraction (%) (g/t Au) deviation (%)
Ore 1 4.69 85 Fire assay (50 g sample) 5.11 27
Ore 2 4.97 75 Aqua regia (25 g sample) 3.857 25
Ore 3 5.11 26
Error calculated from standard deviation from mean grade, based on a
Percentage cyanide extraction based on fire assay head grade. series of identical assays.
 W.R. Goodall et al. / Minerals Engineering 18 (2005) 1010–1019 1013

5. Sample preparation Table 3

Elements detected by EMPA
5.1. PIXE analysis Oxide Sulphide
Na, K, Al, Mg, Si, Ti S, Fe, As, Zn, Pb, Ni
Thin-sections of each ore type were prepared from Ca, Fe, Cr, Zn, Mn, O
hand specimens randomly chosen from ore stockpiles.
Well mineralised samples were preferentially selected.
To allow for the greater penetration depth of the proton It should be noted that EMPA did not analyse for Au
beam thin-sections were cut to a thickness of 100 lm. and Ag, as the detection limit for these elements is too
Areas of interest in the thin-sections were identified high. For gold it is possible to achieve a detection limit
by optical microscopy and digitally logged for initial of 200 ppm with increased counting times, which is still
analysis by Electron Microprobe Analysis (EMPA). much higher than the average concentration in the ore
This was performed to identify areas of interest for anal- (about 5 ppm). For this reason EMPA was only used
ysis by PIXE. It was, however, also necessary to identify to identify the phases present and areas of interest for
the major elements present using EMPA because of the Nuclear Microprobe (NMP), not for trace elemental
difficulties faced in PIXE of identifying lighter elements analysis.
(Z < 20). This allows PIXE to be used exclusively for The results of phase identification based on elemental
trace elemental analysis and gives a more accurate over- ratios established on EMPA are shown in Table 4. A
all understanding of the composition of the ore. number of minerals were identified that were not seen
by quantitative XRD. This is consistent with low bulk
5.2. Diagnostic leaching concentrations of these minerals; however, they are still
significant if they carry gold.
Samples for diagnostic leaching were taken from a It was noted that for both Ores 2 and 3 appreciable
bulk sample of each ore. This bulk sample had been arsenic (1– 4%), was associated with some pyrite grains.
hand picked from ore stockpiles and crushed to
850 lm using a laboratory jaw crusher and laboratory 6.2. X-ray diffraction (XRD)
hammer mill. The crushed material was then homogen-
ised using a riffle splitter. A 2 kg portion of this bulk X-ray diffraction (XRD) was performed using a Phi-
sample was taken for the diagnostic leaching trial and lips PW 1800 X-ray diffractometer. Copper Ka radia-
ground to a P80 of 45 lm using a laboratory ring mill. tion, a graphite monochromator and a proportional
The sample was then split into two 1 kg portions in a detector were used to collect the diffractogram.
rotary splitter. One portion was used for the diagnostic Quantitative analysis of each of the ore samples was
leaching trial and the second for mineralogical performed using the ‘‘siroquant’’ software package
evaluation. (Table 5).
The mineralogical evaluation consisted of quantita-
tive X-ray diffraction (XRD) and X-ray fluorescence Table 4
(XRF) analysis, along with a comprehensive elemental Phases identified by EMPA
and trace elemental analysis by acid digestion. It is nec- Ore 1 Ore 2 Ore 3
essary to have a good understanding of the mineralogy
Quartz Quartz Quartz
and elemental composition when undertaking diagnostic Hematite Pyrite Pyrite
leaching because the mineral phases present dictate Chalcopyrite Gersdorffite
which oxidative acid leaches are necessary. Dolomite
Magnesite

6. Methods
Table 5
6.1. Electron microprobe analysis (EMPA) Mineral Phases identified by Quantitative XRD analysis
Ore 1 Ore 2 Ore 3
Electron microprobe analysis (EMPA) was per-
Phase Fraction Phase Fraction Phase Fraction
formed using a fully automated Cameca SX50 electron (%) (%) (%)
microprobe equipped with four wavelength dispersive
Quartz 84 Quartz 43 Quartz 27
spectrometers (WDS) and an energy dispersive spec- Hematite 5 Muscovite 38 Magnesite 37
trometer (EDS). For the sulphide samples, a beam of Goethite 8 Ankerite 5 Dolomite 22
20 keV and 20 nA was used and for oxide samples a Chlorite 7 Chlorite 7
beam of 15 keV and 25 nA was used. The elements ana- Pyrite 5 Anorthite 5
Calcite 2 Talc 2
lysed are presented in Table 3.
1014 W.R. Goodall et al. / Minerals Engineering 18 (2005) 1010–1019

No sulphide mineral phases were identified in Ore 3 Table 6

using this technique. This is probably caused by the total Oxidative leach stages used for diagnostic leaching of Ores 2 and 3
concentration of sulphides being below the detection Step Leach Mineral leached
limit. 1 Mineralogical –
examination
6.3. Proton induced X-ray emission (PIXE) 2 Sample –
preparation
3 Cyanide Free Gold
PIXE analysis for this study was carried out on the 4a HCl Pyrrhotite, Calcite, Dolomite, Galena,
CSIRO-GEMOC Nuclear Microprobe. This instrument Goethite, Calcium Carbonate
utilises a quadrupole quintuplet lens configuration to 4b Cyanide Free Gold
obtain high resolution over a short length. This lens con- 5a H2SO4 Sphalerite, Labile Copper Sulphides,
Labile Base metal Sulphides, Labile Pyrite
figuration allows a spot size of between 1 lm at 5b Cyanide Free Gold
100 pA growing to 3 lm at 20 nA (Ryan et al., 6a HNO3 Pyrite, Arsenopyrite, Marcasite
1999). For this analysis, a beam current of 5 nA was 6b Cyanide Free Gold
used with a spot size of 2 lm. A 300 lm Al filter was 7a HF Silicates
used to attenuate major elements to the hyper pure Ge 7b Cyanide Free Gold
detector. Dynamic analysis using the GeoPIXE II soft-
ware package was used to decompose the PIXE spec- lation between As and Fe suggests that As is associated
trum into its elemental components to give with the iron oxide and the low concentration (avg.
quantitative concentration information in the form of 0.8%) infers that the As is present in solid solution.
images specific to each element (Ryan, 2001). A sum- The gold hotspot is correlated with Cu, Ni and Ag
mary of the features of the CSIRO-GEMOC Nuclear and has dimensions of 5 l m · 12 lm (plan in view),
Microprobe can be found elsewhere (Ryan et al., 1999, although the thickness is difficult to determine because
2001a). the exact composition of the inclusion is not known.
This is contrary to the postulate that gold in this ore
6.4. Diagnostic leaching should occur as free particles and suggests that it occurs
as electrum associated with a Cu–Ni sulphide or as a
A set of guidelines for the design of a diagnostic more complex mineral (Table 7).
leaching experiment has been demonstrated by Loren- The very simple mineralogical composition of Ore 1
zen (1995). These have been used in this study to estab- meant that diagnostic leaching was not considered
lish an appropriate experimental regime. The key to the necessary for complete characterisation. Future tests
design of a successful diagnostic leaching experiment is will be conducted on the sulphide ore from this deposit.
to have a comprehensive understanding of the mineral-
ogy of the ore to be studied. This allows the selection 7.2. Ore 2
of the minimum number of oxidative acid leaches. Mini-
mising the number of stages reduces experimental error. PIXE analysis of Ore 2 was completed and typical re-
The mineralogy of each ore to be studied in this investi- sults are shown in Fig. 2(a) and (b). A number of grains
gation was determined by quantitative XRD and appro- were analysed. In the image shown in Fig. 2(a), it can be
priate oxidative leach stages determined from guidelines seen that there are three separate grains. One grain is
(Lorenzen, 1995) depending on the complexity of the high in Cu and Fe, with the relative ratios of each of
ore. A typical set of steps for a sulphide ore are summa- these elements consistent with it being chalcopyrite.
rised in Table 6. The second grain, on the left of the image, is most prob-
The oxidative acid leach stages listed in Table 6 were ably As rich pyrite. The final grain is rich in Zn, As and
used for Ores 2 and 3. Both have similar mineralogy. Sb and shows considerable Ag, the composition of this
Diagnostic leaching for Ore 1 was not performed be- grain is consistent with tetrahedrite ((Cu,Fe,Zn,Ag)12
cause the simple mineralogy of the ore and good amena- Sb4S13).
bility to cyanidation suggested that it was not necessary. Gold is weakly distributed throughout the arsenian
pyrite grain. The low concentration of gold (average.
50 ± 17 ppm) is consistent with solid solution or colloi-
7. Results dal gold, and is just above the detection limit (34 ppm)
for this small area. In either case it is consistent with
7.1. Ore 1 ‘‘invisible gold’’ occurring in this ore. It is also interest-
ing to note that no gold is associated with Ag or Cu
A typical grain analysis by PIXE for Ore 1 is shown in these grains. Ag occurs in the tetrahedrite grain
in Fig. 1. As expected the grain is predominantly Fe, and because of the low concentrations determined
consistent with it being hematite or goethite. The corre- (max. 0.27%) appears to be in solid solution. Cu is
 W.R. Goodall et al. / Minerals Engineering 18 (2005) 1010–1019 1015

Fig. 1. PIXE images of elemental distribution in hematite grain from Ore 1. Image area is 1.0 · 1.0 (mm).

Table 7 associated with amorphous base metal sulphides or

Maximum concentrations of elements in PIXE image (Fig. 1) amorphous pyrite. When combined with the results of
Fe 136% PIXE it is strongly suggested that the gold is associated
Ni 3.31% with amorphous pyrite.
Cu 7.35%
Ag 0.50%
As 1.79% 7.3. Ore 3
Au 0.17%
The PIXE images of some typical sulphide grains for
Ore 3 are shown in Fig. 3. The images demonstrate a
concentrated mainly in the chalcopyrite grain and shows clear association of Ag with As, Ni and Sb. Unfortu-
no association with gold. nately, Au could not be distinguished in the presence
Fig. 2(b) shows a pyrite grain with As concentrated of the high As. This As–Ni–Sb sulphide matrix appears
around the rim but not within the grain. This is an inter- to be concentrated in defects in the pyrite grain suggest-
esting feature because statistically significant levels of ing that it was deposited in the ore at a later paragenetic
gold are not seen in the image for this grain, again con- stage. As observed in Ore 2, Ag occurs only in low con-
sistent with ‘‘invisible gold’’ being associated only with centrations consistent with it, and probably gold, being
As rich pyrite or arsenopyrite. The average gold content in solid solution with the sulphide matrix or present in
within the area of the image is 20 ppm (Table 8). colloidal form. This would be consistent with the postu-
The results of diagnostic leaching for Ore 2 are shown late that molecular gold is the cause of assay problems
in Table 9. It should immediately be noted that the cal- experienced with this ore and would suggest that it is
culated gold grade (7.85 g/t Au) recovered is greater held within a Ni–As–Sb matrix.
than the fire assay head grade (4.97 g/t Au) for this Although coarse free gold is known to occur in this
ore. This is most probably attributable to the use of a ore it was not detected by PIXE analysis. This is almost
poorly optimised fire flux used for the assay test. The certainly a function of the small sample size analysed.
distribution of gold suggests that over half is not amena- (Table 10).
ble to direct cyanide leaching, demonstrating that this Diagnostic leaching for Ore 3 demonstrated a very
ore is clearly refractory. The remainder of the gold is low recovery of gold by direct cyanide leach and showed
distributed amongst the sulphide minerals. Little gold that the majority of gold is locked within a sulphide ma-
was associated with crystalline pyrite or arsenopyrite, trix. The results are shown in Table 11. The recovery
which would have been destroyed by the nitric acid oxi- demonstrated in the HCl oxidative leach infers that
dative leach and showed up in this step of the diagnostic gold is associated with galena but the elevated recovery
leaching test. This infers that the majority of gold is from the HNO3 oxidative leach demonstrates that the
1016 W.R. Goodall et al. / Minerals Engineering 18 (2005) 1010–1019

Fig. 2. (a) Typical PIXE images of elemental distribution in grain from Ore 2. Image area is 1.0 · 1.0 (mm). (b) PIXE image from analysis of Ore 2.
Image area is 1.0 · 1.0 (mm).

Table 8 majority of gold is bound within a crystalline sulphide.

Maximum concentrations of elements in typical PIXE image (Fig. 2a)
This is consistent with the postulate that gold is associ-
for Ore 2
ated with a Ni–As–Sb matrix, based on PIXE results.
Fe 62.2% The calculated gold grade (6.62 g/t Au) is higher than
Cu 40.4%
As 2.42%
the fire assay head grade (5.11 g/t Au) but is only just
Sb 13.5% outside the error range (27% std. dev.). The high recov-
Zn 3.88% ery of gold from crystalline sulphides indicates a need
Ag 0.27% to optimise the fire assay technique for this ore. It would
Au 581 ppm be expected that a fire assay method that had not been
 W.R. Goodall et al. / Minerals Engineering 18 (2005) 1010–1019 1017

Table 9 explains why recovery by direct cyanidation is only

Diagnostic Leaching results for Ore 2 85%. However, the high extraction with cyanide sug-
Stage Grade Cumulative grade % Recovery Cumulative % gests that the majority of gold is free and that gold iden-
recovered recovered (calculated recovery tified by PIXE analysis is representative of only a small
(g/t Au) (g/t Au) grade) (calculated
grade)
proportion of the total gold. Although these results
donÕt highlight any obvious metallurgical problems, they
NaCN 3.74 3.74 47.6 47.6
HCl 1.53 5.27 19.5 67.1
demonstrate how the combination of PIXE and diag-
H2SO4 2.08 7.35 26.5 93.6 nostic leaching can be used on a new ore to immediately
HNO3 0.44 7.79 5.6 99.2 show the distribution of gold and whether metallurgical
HF 0.07 7.85 0.8 100.0 problems should be expected. This information is vital
Total 7.85 100.0 for establishing an efficient characterisation regime and
eventually in developing an effective processing route.
The results seen for Ore 2 demonstrate the existence
optimised for this sulphide ore would not liberate gold of two forms of gold. The first form is free gold that is
efficiently from this matrix because of poor slag forma- readily amenable to cyanide leaching and the second is
tion and separation. associated with the sulphide matrix and is thought to
be ‘‘invisible gold’’. The occurrence of gold in As rich
pyrite grains as demonstrated by the PIXE images are
8. Discussion consistent with the observations of Cook and Chryssou-
lis (1990) pointing out that there is a correlation between
The use of PIXE analysis to show the association of bulk arsenic content and ‘‘invisible gold’’ in pyrite. The
Au with its host minerals and diagnostic leaching to results of the diagnostic leaching suggest that this ‘‘invis-
show its bulk distribution in the three ore types under ible gold’’ is concentrated in amorphous pyrite rather
investigation demonstrates the wide range of gold asso- than crystalline pyrite. This is consistent with the obser-
ciations and the importance of its distribution to leach- vations of Wilson and Rucklidge (1987) who showed
ing and recovery. This highlights the importance of gold concentrations to be higher in fine-grained porous
understanding these factors to help develop an effective arsenian pyrite at the Owl Creek deposit, Ontario, but
assay and processing strategy for any given ore. in contradiction to the results of Cook and Chryssoulis
It was seen for Ore 1 that it was free milling although (1990) who observed that ‘‘invisible gold’’ was found in
not all gold occurred as free particles because PIXE pyrite independent of textural type.
analysis identified discrete particles of gold associated One drawback of using a combination of PIXE and
with Cu, Ni and Ag. This shows that some metallurgical diagnostic leaching is the inability of the technique to
problems may be encountered with this oxide ore and determine if ‘‘invisible gold’’ occurs as inclusions of

Fig. 3. PIXE image of typical elemental distribution for Ore 3. Image area is 1.0 · 1.0 (mm).
1018 W.R. Goodall et al. / Minerals Engineering 18 (2005) 1010–1019

Table 10 The results for Ore 3 highlight the effectiveness of

Maximum concentrations of elements in PIXE image (Fig. 3) using a combination of PIXE and diagnostic leaching
Fe 71.60% to identify the associations and distribution of gold.
As 57.10% The use of PIXE alone to show the association of gold
Ni 39.90%
Sb 5.71%
in this ore was not satisfactory because of the high con-
Ag 1.37% centration of As masking the gold content. However,
Au 0.31% when combined with diagnostic leaching it was possible
to show that gold was associated with a sulphide matrix
and when the results of the two techniques were com-
Table 11 bined show the definite association of gold.
Diagnostic leaching results for Ore 3 For both sulphide ores, where ‘‘invisible gold’’ was
Stage Grade Cumulative % Recovery Cumulative % identified, it was noted that the calculated gold grade
recovered grade (calculated recovery from diagnostic leaching was higher than the fire assay
(g/t Au) recovered grade) (calculated
head grade. This highlights the need to be wary of using
(g/t Au) grade)
techniques such as fire assay and aqua regia digestion
NaCN 1.33 1.33 20.03 20.03
with no validation or optimisation reference. The result
HCl 1.62 2.95 24.49 44.52
H2SO4 0.78 3.72 11.73 56.24 suggests that the fire assay technique used did not
HNO3 2.83 6.55 42.74 98.98 extract ‘‘invisible gold’’ efficiently and that the high
HF 0.07 6.62 1.01 100.00 variability noted for Ore 3 is due to unpredictable collec-
Total 6.62 100.00 tion from the Ni–As–Sb sulphide matrix. This need to
optimise fire assay methods for sulphide ores is a well
known phenomenon, however, the use of the PIXE
metallic colloidal gold or in solid solution with the sul- and diagnostic leaching techniques prior to this optimi-
phide matrix. This is an important factor to consider sation provides information about the total gold content
since if gold occurs as colloidal particles, then ultra fine of the ore and the distribution of gold. This information
grinding may be used to initiate liberation, however, if can then be used to establish the ideal flux ratio for fire
the gold is in solid solution then the sulphide matrix assay without the trial and error that is usually required.
must be destroyed before any liberation occurs. A study This process will be implemented for these two ores and
of the nature of ‘‘invisible gold’’ occurrence in pyrite the sulphide ore associated with Ore 1.
carried out by Wagner et al., 1986) used 197Au Möss-
bauer Spectroscopy to show that gold is structurally
bound and bonded covalently. This method while pow- 9. Conclusion
erful has a MDL of 100 ppm, requiring concentrated
samples. In another study, channelling of an MeV ion The associations and distribution of gold in three ores
beam was used to demonstrate that a fraction of gold from the Pilbara region of Western Australia have been
in arsenian pyrite from the Emperor Mine in Fiji is shown. For the first ore, it was demonstrated that some
structurally bound (Den Besten et al., 1999). gold occurred associated with Ag, Cu and Ni, but the
As with Ore 2, a number of different forms of gold majority occurred as free particles. This was consistent
were identified in Ore 3. Coarse free gold is present with the postulate that the ore should be free milling
although it comprises only a small portion (20%) of the but highlighted areas that may cause metallurgical prob-
total gold. The second form is fine gold probably associ- lems. By using the combination of PIXE and diagnostic
ated with galena. This was liberated by the HCl oxidative leaching, to characterise this newly discovered deposit it
leach and would be expected under optimised conditions was demonstrated how these techniques could benefit a
to be amenable to cyanide leaching. The final form of characterisation study by immediately showing the asso-
gold seen to occur in Ore 3, comprising 43% of the total ciations and distribution of gold.
gold, appears to be in solid solution with a Ni–As–Sb sul- The second ore studied was a refractory sulphide ore.
phide matrix and has the same characteristics as ‘‘invisi- It was shown that two forms of gold were present. Free
ble gold’’. The composition of the matrix is consistent gold and gold locked within arsenian pyrite. This gold
with that of gersdorffite (NiAsS), however, the inclusion was identified as ‘‘invisible gold’’ and was postulated
of Sb and Ag suggests some tetrahedrite ((Cu,Fe, to be the cause of unreliable assay results by fire assay.
Zn,Ag)12Sb4S13) may also be present. It has been sug- The final ore studied was a complex refractory ore
gested that gersdorffite may contain ‘‘invisible gold’’ by with a history of metallurgical and assay problems.
Cook and Chryssoulis (1990). This is an interesting result Three forms of gold were identified. Free gold was
as the pyrite and arsenopyrite in this sample, which shown to make up only a small portion of the total gold,
would be expected to offer a source ‘‘invisible gold’’, were with gold also present associated with galena and locked
seen to be deficient in gold by the PIXE analysis. within a Ni–As–Sb sulphide matrix postulated to be
 W.R. Goodall et al. / Minerals Engineering 18 (2005) 1010–1019 1019

gersdorffite or tetrahedrite. It was demonstrated that Lorenzen, L., van Deventer, J.S.J., 1993. The identification of
this last form of gold is ‘‘invisible gold’’ and as with refractoriness in gold ores by the selective destruction of minerals.
Minerals Engineering 6 (8–10), 1013–1023.
the second ore is postulated to be the cause of assay Lorenzen, L., van Deventer, J.S.J., 1992. The mechanism of leaching of
unreliability. gold from refractory ores. Minerals Engineering 5 (10–12), 1377–
In conclusion, this investigation has shown the effec- 1387.
tiveness of using a combination of PIXE and diagnostic Marsden, J., House, I., 1992. The Chemistry of Gold Extraction. Ellis
leaching to identify the associations of gold and its bulk Horwood, New York.
Ryan, C.G., 1995. The nuclear microprobe as a probe of earth
distribution within a given ore. This information can be structure and geological processes. Nuclear Instruments and
used to identify the cause of metallurgical and assay Methods in Physics Research Section B: Beam Interactions with
problems and help develop an effective processing or Materials and Atoms 104 (1–4), 377–394.
assay route. Ryan, C.G., 2001. Quantitative trace element imaging using PIXE and
the nuclear microprobe. International Journal of Imaging Systems
and Technology. Special issue on Advances in Quantitative Image
Analysis 11, 219–230.
Acknowledgments
Ryan, C.G., Jamieson, D.N., Griffin, W.L., Cripps, G., 1999. The
CSIRO-GEMOC nuclear microprobe: A high-performance system
The authors would like to thank the Directors of Ela- based on a new closely integrated design. Nuclear Instruments and
zac Mining Pty Ltd and Haoma Mining NL for the sup- Methods in Physics Research Section B: Beam Interactions with
port of this project; and HaomaÕs Consultant, Mr Peter Materials and Atoms 158 (1–4), 18–23.
Ryan, C.G., Jamieson, D.N., Griffin, W.L., Cripps, G., Szymanski, R.,
Cole, who helped in collection of the samples.
2001a. The new CSIRO-GEMOC nuclear microprobe: First
results, performance and recent applications. Nuclear Instruments
and Methods in Physics Research Section B: Beam Interactions
References with Materials and Atoms 181 (1–4), 12–19.
Ryan, C.G., McInnes, B.M., Williams, P.J., Dong, G., Win, T.T.,
Annegarn, H.J., Bauman, S., 1990. Geological and mineralogical Yeats, C.J., 2001b. Imaging fluid inclusion content using the new
applications of PIXE: A review. Nuclear Instruments and Methods CSIRO-GEMOC nuclear microprobe. Nuclear Instruments and
in Physics Research Section B: Beam Interactions with Materials Methods in Physics Research Section B: Beam Interactions with
and Atoms B 49, 264–270. Materials and Atoms 181, 570–577.
Boyle, R.W., 1979. The Geochemistry of Gold and its Deposits. Sie, S.H., Griffin, W.L., Ryan, C.G., Suter, G.F., Cousens, D.R., 1991.
Geological Survey of Canada Bulletin, 280. The proton microprobe: A revolution on mineral analysis. Nuclear
Cook, N.J., Chryssoulis, S.L., 1990. Concentrations of invisible gold Instruments and Methods in Physics Research Section B: Beam
in common sulphides. The Canadian Mineralogist 28 (1), 1–16. Interactions with Materials and Atoms B 54, 284–291.
Den Besten, J., Jamieson, D.N., Ryan, C.G., 1999. Lattice location of Tarnocai, C.A., Hattori, K., L.J., C., 1997. Invisible Gold in Sulphides
gold in pyrite in natural pyrite crystals. Nuclear Instruments and from the Campbell Mine, Red Lake Greenstone Belt, Ontario:
Methods in Physics Research Section B: Beam Interactions with Evidence for Mineralization During the Peak of Metamorphism.
Materials and Atoms B 152, 135–144. The Canadian Mineralogist 35 (4), 805–815.
Foya, S.N., Reimold, W.U., Przybylowicz, W.J., Gibson, R.L., 1999. Teague, A.J., Swaminathan, C., Van Deventer, J.S.J., 1998. The
PIXE Microanalysis of Gold–Pyrite Associations from the Kim- behaviour of gold bearing minerals during froth flotation as
berley Reefs, Witwatersrand Basin, South Africa. Nuclear Instru- determined by diagnostic leaching. Minerals Engineering 11 (6),
ments and Methods in Physics Research Section B: Beam 523–533.
Interactions with Materials and Atoms 158, 588–592. Vaughan, J.P., 1995. Gold and sulphide systematics of archaean gold
Genkin, A.D., Bortnikov, N.S., Cabri, L.J, Wagner, F.E., Stanley, ores—processing implications. In: Hagni, R.D. (Ed.), Process
C.J., Safanov, Y.G., McMahon, G., Friedi, J., Kerzin, A.L., Mineralogy XIII: Applications to Beneficiation Problems, Pyro-
Gamyanin, G.N., 1998. A Multidisciplinary Study of invisible gold metallurgical Products, Advanced Mineralogical Techniques, Pre-
in Arsenopyrite from Four Mesothermal Gold Deposits in Siberia, cious Metals, Environmental Concerns, Ceramic Materials,
Russian Federation. Economic Geology 93, 463–487. Hydrometallurgy and Minerals Exploration. The Minerals Metals
Johansson, S.A.E., Campbell, J.L., Malmqvist, K.G., 1995. Particle- and Materials Society.
induced X-ray Emission Spectrometry (PIXE). New York, Wiley. Wagner, F.E., Marion, P.H., Regnard, J.R., 1986. Mossbauer Study of
Lorenzen, L., 1995. Some guidelines to the design of a diagnostic Chemical States of Gold in Gold Ores. In: Gold 100. Proceedings
leaching experiment. Minerals Engineering 8 (3), 247–256. of the International Conference on Gold, vol 2: Extractive
Lorenzen, L., Tumilty, J.A., 1992. Diagnostic leaching as an analytical Metallurgy of Gold. Johannesburg, SAIMM. 2, 435–443.
tool for evaluating the effect of reagents on the performance of a Wilson, G.C., Rucklidge, J.C., 1987. Mineralogy and microstructures
gold plant. Minerals Engineering 5 (3–5), 503–512. of carbonaceous gold ores. Mineral. Petrology 36, 219–239.