PAPER 28

On The Floatability Of Gold Grains

Stephen L. Chryssoulis, Daniela Venter, Stamen Dimov

AMTEL
Suite 205, 100 Collip Circle
London, Ontario
N6G 4X8

Key words: gold, floatability, collectors, activation

Proceedings The Westin

35th Annual Meeting of the January 21-23, 2003
Canadian Mineral Processors Otawa, Ontario, Canada

455
ABSTRACT

Comparative surface analysis of floated and rejected free gold grains from a number of full scale
benefication plants and from pilot plant testwork, has shown silver (the most common impurity
in gold) and sulphur invariably in greater concentration on floated gold grains, implying that they
assisted gold flotation (gold activators). Collector loading measurements on individual gold
grains from concentrates of staged rougher flotation tests on a cyclone U/F product of a pyritic
gold ore, revealed first progressively lower concentrations of the collector monomer on gold
from each follow-up concentrate and secondly a direct proportionality between the surface
concentration of silver and the collector monomer. These two observations leave no doubt that
silver activates gold and that collector monomers play a significant, previously not fully
recognized role in gold flotation.

INTRODUCTION

Gold to some degree is naturally floatable in industrial systems, which means that it can be
recovered without collector addition (Marsden and House, 1992). This is due to the adsorption
of hydrocarbons and deposition of sulphur, which are dependent on the metallic properties of
gold, particularly its high electrical conductivity, which allows surface electrochemical reactions
to occur catalytically and selectively. Cyclic voltammetry has shown that gold can be rendered
hydrophobic by deposition of a surface layer of sulphur (Sº) from S2- and HS- solutions (Walker
et al., 1984). Twenty monolayers of sulphur were required to produce a hydrophobic gold
surface.

However, no sulphide flotation plant relies solely on the natural floatability of gold for its
recovery. Gold hydrophobicity is enhanced by the addition of flotation collectors such as
xanthates (X), dithiophosphates (DTP), dithiophosphinates (DTPI), mercaptobenzothiolates
(MBC) etc., as used in sulphide mineral flotation (Dekok, 1973: Adams et al., 1986). The
mechanism invoked in enhancing the hydrophobicity of pure gold with the aforementioned four
groups of collectors is similar to that of pyrite, whereby collector anions are oxidized at the gold
surface to form dimers (Groot, 1987; Marsden and House, 1992; Basilio et al., 1992) As an
example, xanthate ions are oxidized at the gold surface to form the neutral dimer dixanthogen (X2):
2X- ' X2 + 2e- (1)
- -
O2 + 2H2O+ 2e ' 4OH (2)
The non-polar liquid, dixanthogen, forms an oily surface coating on gold surfaces rendering
them hydrophobic. The conducting surface of gold facilitates the transfer of electrons, sustaining
the redox couple described by equations (1) and (2), thereby catalyzing the otherwise slow
oxidation of xanthate ions in bulk solution (Fuerstenau, 1982). As this catalytic oxidation of
xanthate takes place also on pyrite surfaces, these redox reactions offer no selectivity of gold
flotation against pyrite.The reaction models for collector behaviour at gold surfaces are based on
cyclic volammetry measurements on pure gold coupled with FTIR spectroscopy for the
identification of the surface species and contact angle data for changes in hydrophobicity.

The positive effect silver has on gold floatability, was first recognized in experiments using
plates of pure gold, silver and of gold-silver alloys. In the presence of silver, adsorption of the

456
di- and mono-thiophosphinate (DTPI and MTPI) was indicated by the formation of the
corresponding silver complexes (Ag-DTPI and Ag-MTPI), determined by IR spectroscopy,
(Basilio et al, 1992). On the gold electrode only the dimer of the di-thiophosphinate (DTPI)2
could be detected, also observed at higher potentials on the gold-silver alloy electrodes. Albeit
passivation of the gold electrode observed by voltammetry, no dimer or the monomer of MTPI
was detectable by IR spectroscopy (Basilio et al., 1992) suggesting that they either were removed
during sample preparation or that they are present in concentration levels below the limit of
detection of IR spectroscopy. (Nagaraj et al., 1992) using the same approach, showed that silver
assists adsorption of dicresyl monothiophosphate (DCMTP) on gold electrodes. Based on these
findings, it was suggested that the increase in gold recovery observed in practice when using
thiophosphinates and DCMTP, is related to their preferred interaction with silver.

The scope of this investigation has been to obtain a good understanding of the underlying factors
that control flotation kinetics of free gold particles, in order to determine if and how such
kinetics can be selectively enhanced. In contrast to previously published findings, this work is
based on the study of gold particles from flotation test products of a pyritic gold ore with reagent
addition at plant concentration levels; thereby demonstrating the applicability of this approach to
the study of plant samples.

EXPERIMENTAL

Materials

A cyclone underflow sample from an operating gold mine, assaying 155g Au/t with 35% pyrite,
was used as feed to the flotation tests. Four distinct types of free gold particles were observed
(Plate 1). These were differentiated based on size, shape and colour (due to some coating),
which are three of the six parameters identified to affect free gold floatability (Table 1). The
smaller more equidimensional gold particles can be shiny or with a dull gray tarnish. They range
in size from 10 to 100µm with an average of 50µm. The large gold flakes have a distinctly platy
shape, range in size from 150-400µm (average 250µm) and are also shiny or tarnished. The
composition of the gold particles was determined by electron probe microanalysis (EPMA).
Their silver content is low, 2.5 wt% Ag on average, but they also contain appreciable mercury
(avg. 9.2 wt% Hg). Tarnished gold has a markedly higher mercury content (12.5wt% Hg)
compared to shiny gold (4.7% Hg), while gold grains tends to be more silver-rich than flakes
(Fig. 1).

Table 1: Parameters Affecting Free Gold Grain Floatability

• particle size
• flakiness
• composition
• surface contaminants
• collector type and loading
• type of flotation cell

457
Plate 1: Shiny (right column) and tarnished (left column) gold grains and large flakes. The tarnish
is due to a mercury thiochloride coating (HgS·HgCl2?) determined by TOF-LIMS surface
microanalysis.

The collectors used were all commercial grade, supplied by the manufacturer, without any prior
purification.

ANALYTICAL

The composition of the gold particles was determined by electron probe microanalysis (EPMA)
in the wavelength dispersive X-ray (WDX) mode at 20 KV and 20nA, using the La lines of gold,
silver and mercury. The standards used were pure metals for gold and silver and cinnabar (HgS)
for mercury. Twenty gold particles of each type were analyzed. Results are depicted in Fig. 1
and discussed in the section on materials.

458
 S h in y g ra in s Au
100%
S h in y fla k e s
3
Ta rn is h e d g ra in s 97

Ta rn is h e d fla k e s
94 6

9
91

S il
)
12

t. %

ve
88

r (w
(w
85 15

ld

t. %
Go
18

)
82

21
79

76 24

73 27

Ag %
30
70 H 15
27 24 21 18 12 9 6 3
30 g
%

M e rc u ry (w t.% )

Figure 1: Composition of free gold particles

The surfaces of gold particles from each sequential rougher concentrate were analyzed by TOF-
LIMS to determine differences in inorganic surface composition. Using this technique it is
possible to analyze in a matter of a few minutes several spots on the surface of a single gold
particle 20 µm or less in diameter with a sampling depth of 0.01-0.05µm. Statistically
significant differences at the 2σ level were established by T-test analysis of data from 100
surface spectra acquired from 25 randomly picked gold particles from each concentrate.

Collector measurements on individual gold particles were accomplished by TOF-MS. Collector

identification is based on detecting the unfragmented collector molecular ion in the TOF-SIMS
surface spectra (Chryssoulis et al, 1995). Evaluations of collector loadings are based on peak
area measurements for the unfragmented collector molecules in the spectra. Typically ten to
fifteen TOF-MS spectra are acquired from an equal number of gold particles of each type from
every concentrate. Because of differences in their flotation kinetics not all types are present in a
single concentrate.

RESULTS AND DISCUSSION

The cyclone U/F sample used was ground for 12 minutes to 80% passing 170µm. A total of nine
tests were run, the first five to optimize conditions for evaluating factors that controlled free gold
flotation kinetics and the latter four to acquire data for analysis. Tests were deliberately longer
than normal lasting from 24 to 30 minutes in order to deconvolute those factors that affected
kinetics. Potassium amyl xanthate was added in the mill at 12g/t and then added in one (or two)
stages (@ 6 g/t). The secondary collector, DIBDTP, was added together with PAX in the mill

459
(@ 1 g/t) and during flotation (@ 0.5 g/t) in one or two places. These collectors were selected
for their commonality more than anything else. The envisaged primary function for PAX was to
adsorb on pyrite thus focusing DIBDTP more on the gold particles. However, with the addition
schedule selected this most likely did not happen.

Collectorless flotation and flotation with PAX alone

Test T7 was run with only MIBC (no collector) added in the mill, to determine the fraction of
free gold floating collectorless. A low weight 0.8% concentrate was obtained containing some
free gold, molybdenite and pyrite. The free gold recovered was only 1.1% of the total free gold
in the feed. Then a second concentrate was pulled from the collectorless flotation tails after
conditioning with PAX (@ 12g/t) and MIBC for a couple of minutes. Free gold recovery in the
second concentrate, was 59%. Most of the gold particles were of the shiny grain type.

Flotation with PAX and DIBDTP

The objective of test T6 (and to a lesser extent T8) was to assess factors affecting free gold
flotation kinetics individually and in relation to each other. The effect on grade/recovery was
evaluated using the results of test T8, while test T9 was used as a first trial to enhance gold
flotation kinetics.

Factors controlling free gold flotation kinetics

Under the operating parameters of test T6 (Table 2), the systematically descending concentration
of silver on the surface of gold particles (irrespective of type) floated in the sequential
concentrates C1 to C6 (Fig. 2) suggests that silver accelerated free gold flotation. On the
contrary, the ascending concentration of chloride ions would imply that chloride retards gold
flotation (Fig. 2). Based on cumulative results of t-test analysis of surface data from gold of
sequential pairs of concentrates (e.g. C1/C2, C2/C3, etc.) activators and depressants of gold
flotation were identified (Fig. 3). T-test analysis (Fig. 3) established silver to be the only
element that activated free gold flotation and chloride ions as the single most important
depressant.

Table 2: Test Conditions

K80: 90µm
pH: natural
frother: MIBC (35-50 g/t)
collectors: PAX 12 g/t in mill and 6 g/t in 2nd/3rd
point of addition
DIBDPT 1 g/t in mill and 0.5 g/t in
2nd/3rd point of addition
conditioning: 4 minutes after collector addition; 1
minute in between collection of
concentrates
collection: typically 1 minute, except for last conc.
(2 or 3 minutes)

460
 16
Cl
14
12

Ag, Cl (arb. units)

10
8 Ag
6
4
2
0
C1 C2 C3 C4 C5 C6 C7
Concentrate No.

Figure 2: Silver and chloride content of the surface layer of free gold particles and flakes in
sequential concentrates C1-C7 of test T8, in arbitrary units.

+
activators
Ag

enrichment

S Hg

FeOx
depletion

depressants
Cl

Figure 3: Ranking of inorganic surface modifiers on free gold which affected gold flotation kinetics.
The ranking is based on averaged data from comparative t-test analysis of surface “contaminants”
on gold grains (and flakes) floated in sequential concentrates (eg. C1/C2, C2/C3, etc. of test T8).

461
The correlation of the surface concentration of silver on sequentially floated gold particles with
collector loadings (PAX and DIBDTB) is illustrated in Fig. 4. The same scatterplot shows that
the DIBDTP loading is more sensitive (than PAX) to changes in surface silver. The parabolic
correlation of surface silver and DIBDTP loadings with flotation time is illustrated in Fig. 5.
Gold particles that took 19 minutes to float had the least surface silver and DIBDTP. The effect
of mercury thiochloride (Hg3S2Cl2?) coatings and particle size/shape on gold floatability is
illustrated in Fig. 6; tarnished gold grains floated in the same concentrate as shiny gold, had the
same silver content but required more DIBDTP (Fig. 6). On the other hand, the more awkward to
float (because of their weight and shape) large (shiny) gold flakes necessitated even greater
collector loadings and did not begin to float until the third concentrate (C3).

400

DIBDTP
300
Collector loading on Au grains

200
(arb. units)

PAX
100

0
1 3 5 7
Ag on surface of Au grains (arb. units)

Figure 4: Correlation of DIBDTP (z) and PAX (€) loadings with surface concentrations of silver
(all in arb. units) on shiny gold grains from sequential concentrates C1- C5 of staged rougher test
T6. DIBDTP (1 g/t) and PAX (12 g/t) were added in the mill.

462
 10 400
DIBDTP

Silver concentration on

Au grains (arb. units)

DIBDTP loading on
8

Au grains (arb. units)

300
6
200
4 Ag
100
2

0 0
2 6 12 16 20 24
Flotation time ( min)

Figure 5: DIBDTP loadings (z) and silver concentration (+) on surface of sequentially floated shiny
gold grains. Same surface silver but higher DIBDTP loadings were measured on tarnished gold
grains (not shown, Test T6)

C3
1400

1200
DIBDTP loading on Au grains (arb. units)

flakes

C1
1000

800

C4
600 C2
grains
C3
400
C5
C4
C1
C2
200
C5 C3
C4
C5
0
0 2 4 6 8

Ag on surface of Au particles (arb. units)

Figure 6: Comparative DIBDTP loadings on shiny (z), tarnished (S) gold grains and large shiny
gold flakes ( ) from sequential concentrates (C1-C5) of the staged rougher test T6. DIBDTP was
added in the mill (1g/t).

463
The impact DIBDTP and PAX loadings have on the flotation kinetics of shiny and tarnished gold
grains is illustrated respectively in Fig. 7 and 8. The first observation is that shiny gold required
less of both collectors to become floatable. Not surprisingly, the impact of collector loading on
flotation kinetics is more obvious for the harder to float tarnished gold. A six-fold drop in
DIBDT (and eight-fold drop in PAX) loadings was measured between gold floated after 5
minutes (Conc. C1) and after 13 minutes (Conc. C5). Gold grains not floated in conc. C5
required an additional collector dosage (equal to one half the original) and conditioning. The
two-fold (for the shiny gold) to four-fold (for the tarnished gold) increase in PAX loading after
the second collector addition point (first was in the mill feed) is because the competing-for-PAX-
pyrite had been removed by that point. Contrary, DIBDTP loadings on gold floated in
concentrate C6 only matched that of the gold floated in the first concentrate (C1). Again gold
grains floated in the next concentrate (C7) had significantly less collector. In the last three
concentrates C8- C10 no free gold grains were detected, leading to the conclusion that these two
types of gold (shiny and tarnished gold grains) had been all floated.

2000
1800 DIBDTP
A B
DIBDTP loading on Au grains (arb. units)

1600
1400
1200
Tarnished
1000
800
600 Shiny
400
200
C1 C2 C3 C4 C5 C6 C7
0
0 10 20 30
Flotation Time (min)

Figure 7: Effect of surface composition on DIBDTP loading on gold grains floated sequentially in
staged rougher test T8. Shiny grains gold grains have silver sulphide coatings while tarnished gold
grains have silver/mercury thiochloride coatings . A, B: addition points of collector.

464
The combined effect of particle size and shape on flotation kinetics of shiny gold is illustrated in
Fig. 9 where the large shiny flakes do not show up until the sixth concentrate (after the second
point of collector addition) and continue to appear after 30 minutes of flotation (conc. C10) and a
third point of collector addition.
Given the dosage ratio was 20g/t PAX to 1.5 g/t DIBDTP, in test T8, comparison of their loading
on (shiny) gold grains (and large flakes) from sequential concentrates shows twice as much
DIBDTP loaded on gold from every concentrate except for the last one. Once pyrite is removed
from the system (by conc. 5) PAX loading overtakes DIBDTP (Fig. 10).

2000
PAX
1800

B
PAX loading on Au grains (arb. units)

1600

1400

1200

1000

800

600
A Tarnished
400

200 Shiny
C1 C5 C6 C7
0
0 10 20 30

Flotation tme (min.)

Figure 8: Effect of surface composition on PAX loading on gold grains floated sequentially in staged
rougher test T8. Shiny grains gold grains have silver sulphide coatings while tarnished gold grains
have silver/mercury thiochloride coatings. A, B: addition points of collector.

465
 B
A

DIBDTP loading on Au grains (arb. units)

400

C
Shiny flake s
300

200
Shiny grains

100

C1 C2 C3 C4 C5 C6 C7 C8 C9 C10
0
10 20 30

Flotation time (min.)

Figure 9: Effect of grain size/shape and collector loading on floatability of shiny gold grains (z) and
large flakes (‹). DIBDTP was added in the mill (1 g/t) after concentrate C5 (0.5 g/t) and after
conc.C8 (0.5g/t). A, B,C: addition points of collector.

500

B
Collector Loading on Au grains

400 A
(arb. units)

300

200

100

0
10 20 30
Flotation time (min.)

Figure 10: Comparison of the PAX and DIBDTP loadings on shiny gold grains and large flakes
from sequential concentrates of the staged rougher test T8. Collector addition points in mill (12 g/t
PAX; 1 g/t DIBDTP); and after conc. C5 (6 g/t PAX; 0.5 g/t DIBDTP.) For the tarnished gold grains
the measured DIBDTP/PAX loading ratio is higher. A, B: addition points of collector.

466
Free gold floatability and grade/recovery

The effect of free gold floatability on the gold grade/recovery relation for test T8 is shown in Fig. 11.
From the three-grade/recovery relations shown, it is apparent that the associated (with pyrite)
gold (at K80=170µm) forms the backbone of the overall grade/recovery relation. All pyrite
fractions that floated assayed about 350 g/t, accounting for 63% of the recovered gold. Free gold
accounted for an additional 35% in recovery, one half of which was floated in the first five
concentrates and the remainder in the last two concentrates, with free gold in conc. C6 improving
final concentrate grade by 100 g Au/t.

The deportment of floated gold in each sequential concentrate of test T8 (Fig. 12) shows that
associated gold floated mostly in the first four concentrates. Free gold floated in all concentrates
but peaked after the second point of collector addition (conc. C6). Analysis of the floated free
gold based on particle size/shape and coatings (Fig. 13) shows that smaller than 100µm gold
particles with relatively clean surfaces (i.e., shiny grains of plate 1) floated first, followed closely
by gold particles of comparable size but with coatings (tarnished gold grains). The large
(>100µm) flakes floated in conc. C6, account for the significant increase in the final concentrate
grade and a boost in overall recovery of 16%. Given the need for heavier collector loadings to
render floatable large gold flakes (Fig. 9) one way tested for accelerating free gold kinetics was
by advancing the second point of collector addition ahead of the third concentrate. This resulted
in gold flakes showing up earlier (Fig. 14) thus minimizing the chances of loosing free gold in
the final tails.

100
Test 8

80

assayed gold
Gold Recovery (%)

associated gold
60

40

20
free gold

0
100 200 300 400 500 600 700

Cumulative Grade (g/t)

Figure 11: Grade/recovery curves for the free (dashed line), associated (dotted line) and total
(assayed) gold (solid line) of test T8.

467
 9% 100%

100
Test 8 50% 50%
91%

56% 44%
80 67% 33%
Gold Recovery (%)

20%

60 80%

18%

40
82%
free
associated
20
400 500 600

Cumulative Grade (g/t)

Figure 12: Grade /recovery curve for the staged rougher test T8. Pie charts depict the deportment of gold in the sequential
concentrates. Free gold percent shown by darker segment and associated gold by lighter segment.

468
 8% 30% 12%

Test 8
88%
100
11%
51%

80
15%
38% 8%
Free Gold Recovery (%)

1% 4% 39%
40%
60
53% 55%
47%

62%
40 38%

43%
shiny grains shiny flakes
20
tarnished grains tarnished flakes

57%
0
50 100 150 200 250

Free Gold Cumulative Grade (g/t)

Figure 13: Grade/Recovery curve for the free gold in test T8. The pie charts show the breakdown of floated free gold based on particle
size/shape and coatings.

469
 100% 100%
10%
85% 5%

100
Test 9 25%
27%

18%
30%
21%
28%
80 22%
Free Gold Recovery (%)

29%

61% 39%
60

40 41% 59%

shiny grains shiny flakes

tarnished grains tarnished flakes

20
50 100 150 200 250 300 350

Free Gold Cumulative Grade (g/t)

Figure 14: Grade/Recovery curve for the staged rougher test T9 with pie charts illustrating the gold particle types being floated.

470
CONCLUSIONS

Advancements in surface microbeam technology have enabled the measurement of surface

modifiers of collector loadings on individual particles. The identification and quantitation of
collectors is based on the detection of unfragmented collector monomers (Fig. 15). In this study
the dimers was restricted by their volatility at room temperature under vacuum. The established
practical detection limit for collectors is in the range 0.1 g/t.

Figure 15: TOF-MS spectrum of gold particle showing collectors.

The present investigation has shown that routine comprehensive characterization of the factors
controlling the floatability of free gold from flotation tests using reagents at plant concentration
levels is possible. Hence it should be no problem in doing so with plant samples. Singling out
the factor that is primarily responsible for free gold floating slower, enables to target this factor
more specifically. This study is by no means all-inclusive or exhaustive but demonstrates the
potential of integrating mineralogical with surface analytical and metallurgical test data in an
attempt to optimize gold floatability.

Based on these findings it was determined that the faster floating gold grains have always more
surface silver thus leaving no doubt that silver assisted gold flotation, a conclusion also derived
from studying gold/silver electrodes. It was further documented that collector loading is
proportional to the amount of silver on the gold surfaces, confirming that silver activates gold
flotation. This is consistent with IR spectroscopy data of gold/silver electrodes showing Ag-
collector complexes. Of the two collectors tested (PAX and DIBDTP), di-isobutyl
dithiophosphate loads preferentially on gold and shows a stronger dependence on the amount of
surface silver. Not surprisingly, there is a direct correlation between collector loading and
flotation rate. Large flakes required at least four times the collector loadings of smaller gold

471
grains in order to float, thus explaining the occurrence of flakes observed in the final tails of
many plants. Coatings on gold grains, in this case of mercury thiochloride, reduced flotation
rate and required higher collector loadings compared to shiny gold in order to float. Advancing
the staged addition of collectors resulted in faster floating of the large gold flakes. In this
example pyrite competed with gold for PAX. Once pyrite was removed, PAX loaded on to gold.

REFERENCES

Adams et al., “ Mining chemicals handbook”; Cyanamid Mineral dressing notes No. 26, pp. 178,
1986.

Basilio, C. I., Kim, D.S. and Yoon, R.H., “Interaction of thiophosphinate collectors with precious
metals”; SMS annual meeting, Phoenix, Arizona preprint 92-174, pp. 7, 1992.

Chryssoulis, S.L., Stowe, K.G., Niehuis, E., Cramer, H.G., Bendel, C., and Kim, J.Y., “Detection
of collectors on concentration mineral grains by TOF-MS”,.Trans. Instn. Min. Metall. (Sect. C;
Mineral Process. Extr. Metall., v. 104, c141-c150, 1995.

DeKok, S.K., “Gold concentration by flotation”; J.S. Afr. Inst. Min. Metall., Special Issue, pp.
139-141, 1973.

Fuerstenau, M.C, “Chemistry of collectors in solution” in Principles of flotation ed. R.P. King;
published by the S. A. Inst. Min. Metall.Monograph Series No 3., pp. 1-16.

Groot, D. R., “ The reactions of some thiol collectors at noble metals and pyrite electrodes”;
Mintec Report No. M 12, pp. 15, 1987.

Lloyd, P.J.D., “The flotation of gold, uranium and pyrite from Witwatersrand ores”; J.S. Afr.
Inst. Min. Metall. , pp. 41-47, 1981.

Marsden, J. and House, I., “The chemistry of gold extraction”; published by Ellis Horwood, pp.
597, 1992.

Nagaraj, D. R., Brinen, J.S., Farinato, R.S. and Lee, J., “A study of the interaction of dicresyl
monothiophosphate with noble metals using electrochemical, wetting and spectroscopic
methods”; published by Langmuir, vol. 8, pp. 1943-49, 1992.

Walker, G.W., Walters, C.P. and Richardson, P.E., “ Correlation of the electrosorption of sulphur
and thiol collectors with contact angle and flotation“; Electrochemistry in Mineral and Metals
Processing, pp. 202-217, 1984.

472