The Mechanism of Elution

of Gold Cyanide from Activated Carbon

J.S.J. VAN DEVENTER and P.F. VAN DER MERWE

Numerous articles have appeared on the mechanism of the adsorption of gold cyanide onto
activated carbon. In contrast, little information is available on the mechanism of elution of the
adsorbed gold. It is the objective of this article to formulate such a mechanism on the basis of
batch and column elution tests without analyzing adsorbed species on the carbon directly. The
presence of spectator cations (M"+) enhances the formation of M"+{Au(CN)~}, ion pairs on the
carbon, which in turn suppress the elution of gold cyanide. The dynamics of removal of these
cations determine the horizontal position of the gold peak in an elution profile. When the con-
centration of cations in the eluant is high and no cyanide is present in the solution or on the
carbon, very little desorption of gold is observed. The quantitative effect of the concentration
of spectator cations on the equilibrium for desorption of aurocyanide can be estimated from the
elution profiles for gold and cations. Free cyanide in the eluant, which causes some competitive
adsorption of cyanide with aurocyanide, therefore plays a minor role at the elevated temperatures
used in industry. A more important effect of cyanide is its reaction with functional groups on
the carbon, the products of which passivate the surface for adsorption of aurocyanide, and thereby
cyanide promotes the elution of aurocyanide. The degree of passivation, which is determined
to a large extent by the temperature of pretreatment, also affects the elution of cations and the
degradation/adsorption of cyanide itself. Reactivation of the carbon surface occurs when the
adsorbed/decomposed cyanide is removed by the eluant. At high temperatures of pretreatment,
such as used in practice, it is not necessary to include a reactivation term in the mathematical
model for elution.

I. INTRODUCTION experiments, without directly analyzing the species ad-

THE carbon-in-pulp (CIP) process is the preferred
method used internationally for extracting gold from
leached pulps. Although the kinetics and mechanism of
adsorption of gold cyanide onto activated carbon have
been studied thoroughly, the elution of gold from the
carbon has not been researched methodically to the same
extent. In two recent articles, we have made first at-
tempts to quantify the complex inter-relationship be-
tween variables affecting the equilibrium of desorption t~]
and to model the dynamic behavior of the elution process
on a general basis, lz} However, little research has been
conducted on the mechanism of the elution process, aside
from the brief discussion presented by Adams and
Fleming. t3] The ongoing debate in industry whether or
not cyanide is required in elution is mainly the result of
a lack of understanding the mechanism of elution of
aurocyanide. In neither of the existing articles on the elu-
tion process has any attempt been made to isolate some
of the complex phenomena affecting elution. Some vari-
ables, such as the concentrations of cyanide and cations,
can have opposing effectstq but are usually difficult to
decouple in a normal elution run.
Consequently, there is a need for a systematic study
on the mechanism of elution, which will require the de-
sign of special experiments. It is the objective of this
article to formulate such a mechanism of elution from
J.S.J. VAN DEVENTER, Professor, and P.F. VAN DER MERWE,
Ph.D. Graduate, axe with the Department of Chemical Engineering,
University of Stellenbosch, Stellenbosch, 7600, Republic of South
Africa.
Manuscript submitted November 29, 1993.
sorbed on the surface of the carbon.
Two main techniques of elution are used in industry:t4]
(1) The Zadra process, in which warm cyanide solution
is circulated through an elution column and electro-
winning cell (a United States Bureau of Mines inven-
tion); and (2) The Anglo American Research Laboratory
(AARL) method, consisting of a pretreatment step with
hot caustic cyanide solution, followed by elution with
hot deionized water (a South African invention). An
Australian invention, t5,6} which involves the use of or-
ganic solvents, is also used in a few industrial applica-
tions. In general, the AARL method is preferred because
of lower operating costs, especially with the develop-
ment of continuous elution. Therefore, the AARL pro-
cedure was used as a basis for the present investigation.
According to Adams and Nicol, tT] temperature, cya-
nide and hydroxide concentrations, and the ionic strength
of the eluant constitute the most significant effects on
elution. In cases where the loaded carbon contains high
levels of calcium carbonate, the temperature and acid
washing of the carbon are considered to be most im-
portant for efficient elution.[8] Cyanide and hydroxide salts
are added during both the AARL and Zadra elution pro-
cesses to promote the elution of the gold cyanide. The
equilibrium between the adsorbed gold and the gold cy-
anide in solution is affected by the concentrations of these
additives, which change continuously.[9} Both the equi-
librium of adsorption/desorption as well as the decom-
position of the cyanide are affected significantly by
temperature, tz~ High temperatures and intense cyanide
pretreatment (i.e., conditions favorable for elution) de-
crease the sensitivity of aurocyanide elution to flow rate

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 25B, DECEMBER 1994--829

and the radial distribution of the gold through the carbon
particles, c~]This implies that the resistance to mass trans-
fer is less profound under strong elution conditions. An
elution model, which assumes equilibrium of auro-
cyanide between the solution and carbon phases at any
point in an elution column, can then be formulated, tzl It
has been shown that such a model simulates AARL elu-
tion runs at different temperatures and pretreatment con-
ditions using the same set of parameters, tal
Adams and Nicol [TJ surmised that the desorption of gold
involves the competitive adsorption of cyanide and auro-
cyanide on activated carbon. Likewise, Tsuchida tul as-
sumed that the desorption of gold from carbon can be
expressed by the following equilibria:
{AuCN}r + CN- r {Au(CN)2-}~o,,,,~on
{Au(CN)2}ca~n + CN- 4:> { C N - } ~
+ {Au(CN)~}solution
The ratio of occurrence of AuCN to Au(CN)~- on the
carbon in the elution column is dependent on the type
of activated carbon and the conditions of acid wash-
ing. I12] It is important to know this ratio and hence the
reversibility of adsorption, because it will determine the
need for cyanide in the elution step. If the conditions of
acid washing are not too severe, most of the gold will
be present as Au(CN)~-, so that elution without cyanide
is sometimes possible in practice.
Measuremenfl ul of the activity coefficients of CN- and
Au(CN)~- revealed that in an organic-rich environment,
the activities of CN- and Au(CN)~ were, respectively,
102 to 104 and 20 times higher than in water. It was
claimed that in organic solvent~water mixtures, the much
higher increase in activity of the C N - than the
Au(CN)~- will shift the preceding equilibria to favor the
desorption of the gold. Tsuchida et al. I~3I attributed
the decreased kinetic activity of eluted carbon to - C N
being a poor leaving group, as well as to the deactivation
of the active sites by the oxidation of cyanide by chemi-
sorbed oxygen to CO~- and NH~. In testing the effect
o f various anions on the elution o f gold, it was found
that anions with higher nucleophilicities were more ca-
pable of desorbing gold. This was interpreted as an in-
dication that the active sites are able to undergo
nucleophilic substitution reactions.
Adams et al. U4] and Adams and Fleming TM explained
the desorption of Au(CN); from activated carbon at high
pH on a similar basis as a result of the elution from a
polymeric adsorbent with phenolic hydroxyl functional
groups. They proposed that the weakly acidic functional
groups, such as phenolic hydroxyl, are deprotonated by
OH-, leaving the surface more negatively charged and
hydrophilic:
R-OH + NaOH ~ R-O-Na + + H20
Such a surface will be less compatible with the auro-
cyanide ion pair, thus favoring desorption. This mech-
anism is supported by the observation of Cho and Pitt t~51
that an increase in pH results in a more negative zeta
potential of the activated carbon.
The preceding findings indicate that cyanide can pro-
mote the elution of gold by three different mechanisms:
8 3 0 - - V O L U M E 25B, DECEMBER 1994
(1) reaction with the adsorbed gold cyanide species,
(2) competitive adsorption (which is regarded here as being
similar to ion exchange), and (3) reaction with the car-
bon functional groups. It is the objective of this article
to investigate the relative importance of mechanisms (2)
and (3) and to distinguish between the depressing effect
of spectator cations [~] and the enhancing effect of cya-
nide on the elution of aurocyanide.
II. EXPERIMENTAL
The main adsorbate used in experiments was pow-
dered potassium aurocyanide KAu(CN)2 dissolved in
distilled water. Small amounts of HC1 or KOH were added
to the process solution in order to control the pH. Eluted
[1] coconut-shell-based activated carbon from the Beatrix
gold mine near Virginia in South Africa was acid washed
and rinsed with deionized water to remove as many im-
[2] purities as possible prior to use. The carbon was dried
at 120 ~ for 3 days before being weighed, soaked in
distilled water, and used in the experiments. By con-
ducting a series of tests as described elsewhere, ~12]it was
found that no AuCN was formed on the carbon during
adsorption from alkaline solutions at room temperature.
The mean particle size was 1.42 mm and provided a void
fraction of 0.292 in a packed column; the apparent den-
sity was 840 kg" m-a; and the Brunauer-Emmett-Teller
(BET) surface area was determined as 790 m 2. g-l. The
pore volume of the carbon was determined as 6.35 x
1 0 - 4 m 3 - kg -1 by measuring the mass loss upon oven
drying of a sample of carbon that was saturated with
water.
The objective of this study was to investigate the
mechanism of elution and not to obtain the highest pos-
sible elution efficiencies. Hence, a convenient temper-
ature of 70 ~ was selected for most of the elution runs.
Besides some batch elution tests conducted in 1.0-L so-
lution (or less) at a stirring speed of 700 rpm, all elution
runs were conducted in a glass column with a temperature-
controlled water jacket. The downward flow of eluant
through the column was controlled manually. A bed vol-
ume (BV) is defined here as the empty volume of the
column that is occupied by the packed bed of carbon
(i.e., 17.16 cm3). The bed height used was 14.3 cm, the
height/diameter ratio was 11.57, and the flow area was
1.20 cm 2. One bed volume of the glass column con-
tained approximately 9.4 g of dry carbon.
The pretreatrnent step was conducted outside the col-
umn in glass beakers. Unless otherwise specified, pre-
treatment was conducted for 30 minutes in 20 mL of a
20 g K C N / L solution at 20 ~ After the pretreatment,
the carbon was separated from the solution with a strainer
and excessive solution was removed by blotting with fil-
ter paper. The carbon was then dropped into the glass
[3] column containing half a bed volume of eluant at the
elution temperature. The high concentration of cyanide
in the pretreatment ensured sufficiently high pH values
in the pretreatment step, as well as during the subsequent
elutions. The starting time for the elution was taken as
the moment when flow of eluant was introduced. Al-
though distilled water was used for elution in most cases,
some experiments required the addition of potassium
chloride to the eluant.
METALLURGICAL AND MATERIALS TRANSACTIONS B
 III. SUBPROCESSES OF ELUTION
Mainly, three important subprocesses that occur si-
multaneously govern the continuous change in condi-
tions during an AARL elution cycle. The removal of the
pretreatment reagents from the carbon bed constitutes,
first, the diffusion of cations from the pores of the car-
bon and, second, the decomposition and diffusion of the
cyanide from the carbon pores. These subprocesses cause
the desorption of gold, followed by the diffusion of gold
cyanide out of the carbon into the interparticle solution.
A typical set of elution profiles obtained at 70 ~ is
depicted in Figure 1. The elution profiles of gold, cya-
nide, and potassium, as well as the pH, are normalized
for comparative purposes. A sharp decline in the cyanide
and potassium concentrations occurred within the first
two bed volumes, followed by an elongated tail as a re-
sult of the slow removal of reagents being trapped in the
pores of the carbon. The peak in the gold profile only
appeared once the bulk of the pretreatment reagents (i.e.,
the cyanide and spectator cations) had been removed.
Despite the fact that the pH of the water supply was not
adjusted, the pH remained high for the whole duration
of the run. It was shown in a previous article TM that all
these factors affect the gold-cyanide equilibrium, which
consequently changes continuously throughout an elu-
tion run. A mathematical model for this shifting iso-
therm is required in order to simulate a complete elution
profile. Whereas Banini and Stange t161 considered only
the concentration of hydroxyl ions in this isotherm (by
eliminating the effects of cyanide pretreatment), it is clear
that both cyanide and spectator cations should be taken
into account. 1],2]
From the profiles in Figure 1, it is evident that the
presence of potassium as the spectator cation depresses
the elution of gold because of the strong adsorption of
ion pairs. However, the role of cyanide is not apparent
from these profiles. It is well known 17,nA3! that the pres-
ence of cyanide depresses the adsorption of gold, and it
is known from industrial practice that cyanide enhances
1 adsorbed with cyanide and does not adsorb on its own,
0.8
-~ -~1(- -X-
0.6
0.4
I I
0.2 I-~- pH I
O-
2 4 6 8 10 12 14 16
Bed volumes
Fig. 1 - - N o r m a l i z e d AARL elution profiles in a column at 70 ~ and
a flow rate of distilled water as eluant of 49.2 m L / h . [Au] =
130- Y mg A u / L ; [K*] = 4000- Y mg K+/L; [CN-I = 2200. Y mg
C N - / L ; pH = 15- Y. The 9.41 g of carbon contained 4.85 g A u / k g
after 22 h of loading and was pretreated for 30 rain in 20 mL solution
of 20 g K C N / L at 20 ~
METALLURGICAL AND MATERIALS TRANSACTIONS B
the elution of gold. Nevertheless, most desorption of gold
(Figure 1) occurred when most of the free cyanide had
already been washed out of the carbon bed. No AuCN
was formed on this specific carbon, so that the effect of
cyanide could not be linked to the dissolution of an ir-
reversibly adsorbed gold species, tul Furthermore, if CN-
had displaced Au(CN)2 from loaded activated carbon by
a mechanism of competitive adsorption, it could have
been expected that more desorption of gold would have
occurred during the pretreatment step. In fact, almost no
gold desorbed during the pretreatment step in any of the
experiments. This indicates that the relative effects of
spectator cations and cyanide are not clear and require
further investigation.
IV. E L U T I O N O F S P E C T A T O R CATIONS
The concentration of potassium in the pretreatment so-
lution for the elution run depicted in Figure 1 decreased
from 12,300 to 7380 mg K + / L after it was added to the
gold-loaded carbon. The carbon was only blotted with
filter paper after it was removed from the adsorption so-
lution, so that the residual water in the pores could have
been responsible for the dilution of the pretreatment re-
agents. An experiment was then devised to determine the
effect of the type of anion on the possible adsorption of
cations.
A sample of dry carbon was added to a KCI solution
with the same initial potassium concentration as de-
scribed previously. This led to a slight increase in the
concentration of potassium in the solution, which was
attributed to the fact that the carbon must have contained
some adsorbed potassium that was desorbed when it was
brought into contact with water. When the pretreatment
step prior to the elution run of Figure 1 was repeated
with dry carbon (so that dilution was eliminated), it was
found that the potassium concentration still decreased from
12,300 to 10,200 mg K + / L and the cyanide concentra-
tion from 7435 to 5841 mg C N - / L .
The observation that more potassium is adsorbed from
a KCN than a KC1 solution indicates that K § is co-
because carbon does not have a high affinity for chlo-
ride. tln The molar ratio of potassium to cyanide con-
-
sumed in the elution run with dry carbon at 20 ~
described previously was calculated as 1 : 1.03, which is
almost stoichiometric. When the pretreatment was con-
ducted at 70 ~ this ratio decreased to 1:1:37, which
agrees with the finding that more decomposition and less
adsorption of cyanide occurs at higher temperatures) 4~
Van der Merwe t4~ showed that the elution of potassium
is independent of temperature and flow rates up to
37 bed volumes per hour. However, the elution of po-
tassium is affected by the nature of the anion in the pre-
18 treatment reagents. This is supported by the results of
Figure 2, where the elution profiles of K + after a KC1
and KCN pretreatment were compared. The concentra-
tion of K + in the eluate was higher for KCN than for
KCI for most of the run, as a result of the significantly
lower adsorption of K + from KCl than KCN solutions.
Although the preceding experiment indicated that
spectator cations, such as potassium, adsorb mainly in
VOLUME 25B, DECEMBER 1994--831
 elution step of this experiment consisted of two stages.
During the first stage, an eluant containing 1972 mg K+/L
as KCI was pumped through the carbon bed. The high
cation concentration suppressed the elution of the gold
and provided four bed volumes of eluant to rinse the car-
bon of cyanide that was carried over from the pre-
treatment step. For the second stage of the elution, the
__~ 100 potassium-rich eluant was replaced with distilled water.
r Although the cyanide concentration in the solution was
practically zero at that stage, a peak in the gold concen-
10 tration appeared almost instantaneously. Moreover, the
maximum gold concentration was more than 60 pct higher
than that measured in a similar elution without the initial
high cation stage, as depicted in Figure 1. This can be
1
ascribed, however, to the more ~intense" pretreatment
0 2 4 6 8 10 12 14 that was carried out here. The latter effect will be in-
vestigated in Section V.
Bed volumes
From the preceding experiment, it follows that the free
Fig. 2 - - T h e relative effects o f pretreatment with K C N and KCI on cyanide that is present during the elution step of the AARL
the elution o f potassium in a column at 20 ~ and a flow rate of procedure is of secondary importance to that in the pre-
distilled water as cluant of 39.8 m L / h . The 9.41 g of carbon was
treatment. This means that it is the effect of the cyanide
pretrcated for 30 min at 20 ~ in 20 m L of solution initially containing
[K § = 12,300 m g K + / L as K C N or KCI. At the end of pretreatment, that is present in the pretreatment step and not the carry-
[K +] = 7380 m g K+/L for the K C N and [K+I = 12,240 m g K + / L over of pretreatment reagents to the eluant that enhances
for the KCI. the elution of the gold. Because of the high concentra-
tion of spectator cations in the pretreatment step, the bulk
of the gold desorption occurs only during the elution stage
association with cyanide, it is still necessary to investi- once the cation concentration is lowered. The horizontal
gate the relative importance of spectator cations v s cy- position of the gold peak will therefore be a function of
anide during elution. Figure 3, depicts an experiment to the dynamics of the removal of the cations carried over
determine the effect of potassium in the absence of cy- from the pretreatment stage.
anide. After pretreatment in a solution that contained ini-
tially 7886 mg C N - / L as free cyanide, the carbon was
boiled for 1 hour in the pretreatment reagent to decom- V. PASSIVATION OF THE CARBON SURFACE
pose most of the cyanide. This was done to ensure the
lowest possible carryover of cyanide from the pretreat- Davidson and Duncansont~yj showed that more than
ment to the elution stage, and resulted in a residual 99 pct of the gold loaded onto a coconut-shell-based car-
free-cyanide concentration of only 80 mg CN-/L. The bon could be recovered with deionized water after a pre-
treatment with potassium carbonate. A nearly identical
gold elution profile was obtained with an equimolar
2500 sodium-carbonate pretreatment. The carbonate pre-
treatment led, however, to poor elution of silver and other
base metals, as well as to the buildup of calcium car-
2000 .t IAul* I0 bonate on the carbon. The poor elution of the metals
other than gold was blamed on the instability of these
complexes under high pH pretreatment conditions. Fur-
~ 1500 ther tests indicated that it was the higher hydroxide con-
centration associated with the addition of carbonate, and
I.~ 1000 not the carbonate itself, that was responsible for the ef-
ficient elution of gold. The carbonate in the pretreatment
was then replaced with sodium hydroxide and sodium
51111 cyanide. The sodium cyanide was added in order to in-
crease the stability of the metal-cyanide complexes. Fur-
ther investigations revealed that the elution of all the
0 I I I metal-cyanide species could be improved by adding less
0 5 10 15 20 hydroxide to the pretreatment. In practice, hydroxide is
still added to the pretreatment, but mainly to stabilize
Bed volumes
the cyanide.
Fig. 3 - - A n A A R L column elution subsequent to a step change in The relative insignificance of hydroxide compared to
potassium concentration in the eluant at a flow rate of 49 m L / h and cyanide in the pretreatment is illustrated in Figure 4.
70 ~ The eluant initially contained 1972 m g K § as KCI and was
replaced by distilled water after 2.5 bed volumes. The 9.41 g o f car-
Identical samples of gold-loaded carbon were pretreated
bon initially contained 4.98 g A u / k g . It was pretreated at 35 ~ for at the same pH and the same initial concentration of po-
44 h in 20 m L of solution o f 20 g K C N / L and then boiled for 1 h, tassium. The only difference was that one sample was
which decreased the [CN-] to 80 mg C N - / L . pretreated with a KCN solution and the other with a KOH

832--VOLUME 25B, DECEMBER 1994 METALLURGICAL AND MATERIALS TRANSACTIONS B

solution. As in the case of Figure 2, more potassium
adsorbed in the presence of cyanide, so that the residual
concentrations were 7380 and 8712 mg K+/L, respec-
tively, for the KCN and KOH solutions. Despite the higher
loading of potassium in the case of cyanide pretreatment,
the maximum gold concentration in the elution was more
than 10 times higher than that for the hydroxide pre-
treatment. It has been shown that the free cyanide in
solution is not the dominant effect, so that this experi-
ment suggests that the presence of cyanide during pre-
treatment causes passivation of the carbon surface, which
enhances the elution of gold.
This passivation of activated carbon by cyanide was
investigated by comparing the ability of fresh carbon to
adsorb gold to that of carbon that was pretreated in a
strong cyanide solution. When this test was conducted
initially, the cyanide-treated carbon was boiled in dis-
tilled water before the adsorption test to decompose all
the free cyanide occluded in its pores. No difference could
be detected in the subsequent affinity of the two samples
toward gold cyanide, as determined by adsorption in a
batch-stirred tank reactor. Hence, any possible passi-
vating effect was eliminated by the boiling, which de-
stroyed all residual free cyanide. This is in contrast with
the experiment represented in Figure 3, which preserved
the passivating effect of pretreatment by leaving a resid-
ual 80 mg CN-/L as free cyanide. Although high tem-
peratures could convert Au(CN)2 to AuCN, this was not
the case here, otherwise the same adsorption profiles could
not have been obtained, t1'~21
carbon, which proves that the surface of the carbon was
modified in such a way that its affinity for the auro-
cyanide complex was reduced. However, this passiva-
tion can be reversed easily by exposure of the carbon to
high temperature in the absence of free cyanide.
A further experiment was devised to determine the rel-
ative importance of the competitive adsorption and the
passivating effect of cyanide during pretreatment.
Figure 6 depicts the gold elution profiles of two exper-
iments in which the carbon was pretreated at different
temperatures, but with the same "loading" of cyanide of
7.0 g CN-/kg after pretreatment. The loading is defined
as the amount of cyanide that is removed from the pre-
treatment solution as a result of either competitive ad-
sorption or decomposition on the carbon surface. The
1
0.8
0.6
0.4
0.2
0 i I I I I I I

Instead of boiling the cyanide-treated carbon, the test 0 5 10 15 20 25 30

was then repeated by eluting all the remaining cyanide
with 190 bed volumes of distilled water in a glass col-
umn at room temperature. The carbon was then removed
from the column, and the adsorption of gold cyanide onto
the pretreated carbon was compared to the adsorption
onto a similar sample of fresh carbon in a batch-stirred
tank reactor9 Figure 5 shows that the carbon pretreated
in KCN had a significantly lower adsorptivity than fresh
Time (h)
Fig. 5 - - T h e deactivating effect of cyanide pretreatment on the batch
adsorption of gold in a batch reactor. Initial [Au] = 24.5 m g A u / L ;
pH = 9.5; volume of solution = 1.0 L; mass of carbon = 1.0 g; The
carbon was pretreated in 10 m L solution of 20 g K C N / L for 40 h at
20 ~ and rinsed with 190 bed volumes of distilled water at a flow
rate of 35 m L min. For untreated carbon, only distilled water was
used.

140 200

120 * 100 d e g C~

100 150
_l
:3
< 80
E 60
tO 40
20
0
0 ! . : : : : : l !
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18

Bed volumes Bed volumes

Fig. 4 - - T h e relative effects of pretreatment with either K C N or K O H
on gold-elution profiles in a column at 70 ~ and a flow rate of dis-
tilled water as eluant of 49.5 m L / h . The 9.41 g of carbon initially
contained 4.9 g A u / k g and was pretreated in 20 m L of solution ini-
tially containing 12,300 m g K + / L (as either KCN, or K O H and KCl)
at a pH o f 12.3 and 20 ~ for 30 min.
Fig. 6 - - The effect of the temperature of cyanide pretreatment on the
elution of gold cyanide in a column at 70 ~ and a flow rate of dis-
tilled water as eluant of 48.5 m L / h . The 9.41 g of carbon initially
contained 4.83 g A u / k g and was pretreated for 30 min in 20 m L (at
20 ~ or l0 m L (at 100 ~ solution of 20 g K C N / L . At both 20 ~
and 100 ~ the cyanide loading was 7.0 g C N - / k g after pretreatment.

METALLURGICAL AND MATERIALS TRANSACTIONS B VOLUME 25B, DECEMBER 1994--833

extent of adsorption decreases with an increase in tem-
perature, so that the carbon pretreated at 100 oC could
be expected to have a significantly higher ratio of de-
composed to adsorbed cyanide than that at 20 ~ 1~SlUn-
fortunately, no analytical method exists whereby this ratio
can be determined.
Figure 6 reveals that significantly improved elution of
gold was obtained after pretreatment at 100 ~ where
very little adsorption of cyanide would have occurred.
A mechanism of displacement of aurocyanide by cya-
nide during cyanide pretreatment at industrial elution
temperatures can thus be ruled out to a large degree. As
it was proved that the cyanide in the pretreatment step
does not displace or modify the gold-cyanide species, t41
it is postulated that the cyanide that is decomposed on
the surface of the carbon during the pretreatment stage
changes the functional groups on the surface in such a
way that the surface becomes less receptive for adsorp-
tion. Adams and Fleming TM presented the following ex-
amples of possible reac.tions that can occur between
cyanide and the functional groups present on the surface
of the carbon:
O O-
It K I
R - - C - - R + CN- r R--C--R
[ representing the removal of free cyanide from solution,
CN
(K -~ 11,000)
I pected. Figure 7 shows a plot of the firt-order rate con-
--C--C-- + CN- r --C=C-- + HCN
I II ] [5]
H O O-
Under industrial alkaline conditions of gold elution, cy-
anide is decomposed via several reactions, tl91 of which
the following are considered to be the most important.
Hydrolysis
CN- + 3H20 ~ {HCOONH4} + OH-
1
{HCOONH4} + 202 ~ H C O ; + NH~"
HCO3 + NH + + 2OH- ~ (pH 10.5)
NH3+CO~ +2H20
Oxidation
1
C N - + 2 02 ~ {CNO-}
{CNO-} + 2H20 ~ CO~- + NH4
These reactions would increase the negative-charge den-
sity on the surface and render it less receptive for
Au(CN)~. Although this modification of the carbon sur-
face would occur already during the pretreatment step,
the high concentration of cations (M n+) would favor the
formation of M"+{Au(CN)2}, ion pairs on the carbon sur-
face and thereby restrict the desorption of gold during
this stage. Once the concentration of cations is lowered
during the elution stage, the Au(CN); will be desorbed
from the deactivated carbon surface.
834--VOLUME 25B, DECEMBER 1994
In the AARL elution process, most of the adsorption
and reaction of cyanide occur in the pretreatment step,
whereas in the Zadra process, the cyanide is decomposed
simultaneously with the elution of the gold. Yet in both
processes, the presence of cyanide in solution as well as
the reactions of cyanide on the carbon surface enhance
the desorption of gold. Therefore, it is important to in-
vestigate the possible effect of cyanide loading on the
decomposition of cyanide itself.
The oxidation reactions, Eqs. [9) and [10}, are cata-
lyzed by the presence of activated carbon and are more
significant at low temperatures. However, the hydrolysis
reactions, Eqs. [6] through [8] dominate at high tem-
peratures and are the main cause for the loss of cyanide
during Zadra or AARL elution/2~ Measurement of the
decomposition rate of cyanide revealed first-order ki-
netics for both the hydrolysis and oxidation reac-
tions. [4,t91 In the present study, no distinction is made
between adsorption, hydrolysis, or oxidation of cyanide
during measurement of the kinetics of removal of cya-
nide from solution. An experiment was conducted in
which one sample of carbon was repeatedly contacted
with fresh solutions of cyanide in a batch-stirred tank
reactor for different periods of time. During such a batch
test, small samples of solution were taken and analyzed
[4] for free cyanide. The loading of cyanide on the carbon,
was then calculated from a mass balance at the end of
each period. The decrease in free-cyanide concentration
during each period revealed first-order kinetics, as ex-
stant k(CN) for each period of contact as a function of
the calculated cyanide loading at the end of the preced-
ing period. Clearly, the rate of decomposition of the cy-
anide decreased with an increase in the cyanide loading
of the carbon. This relationship appears to be strongly
temperature dependent, laj Therefore, the passivation of
the carbon surface caused by the decomposition of cy-
anide inhibits not only the adsorption of gold cyanide,
[6]
[7] 12
10
[8] 8
[9]
[10]
2
0
0 l 2 3 4 5 6
Cyanide "loading" [gCN/kg]
Fig. 7--The decrease in the first-order rate constant for the decom-
position of cyanide with an increase in the cyanide loading. The
carbon, with a mass of 21 g, was contacted repeatedly for different
periods with 350 mL of solution containing 0.525 g CN-/L at
18 ~
METALLURGICAL AND MATERIALS TRANSACTIONS B
but also most other reactions on the carbon surface. Van
der Merwe 14] confirmed this observation by showing that
the conversion of cyanide during a Zadra elution run de-
clined with time as the carbon surface became passivated.
VI. REACTIVATION OF THE
CARBON SURFACE
It is well known that elution under industrial condi-
tions restores the activity of the carbon completely if it
is not fouled with undesirable organic and inorganic spe-
cies. This means that any passivation of the carbon sur-
face should also be reversed. The possible reactivation
of the carbon surface after a cyanide pretreatment was
investigated by monitoring the elution of gold from car-
bon that had been loaded over a period of 3 weeks. Suf-
ficient time thus was allowed during the adsorption stage
to ensure a fairly homogeneous distribution of gold
throughout the carbon particles. A short period of ad-
sorption would have resulted in a loading gradient of ad-
sorbed gold through the carbon particles, t2~] Under these
circumstances, further diffusion into the finer pores of
the carbon during the elution step could have led to the
readsorption of gold that had been eluted from the sur-
face of the particles during the initial stage of the
experiment.
Figure 8 depicts the batch elution profiles for two
samples of carbon loaded to different levels after a cy-
anide pretreatment in a separate container. Desorption
occurred for the initial 3 hours of the experiment, where-
after the concentration of gold in solution started to de-
crease gradually. In this case, the readsorption of gold
could be ascribed only to the deterioration of the effect
of the cyanide pretreatment during the elution step. This
reactivation of the carbon makes it difficult to accurately
40
30
<
E
20 I
-- 6.3 gAu/kg
7.0 gAu/kg
O
10
I I I I I I
0 5 10 15 20 25 30
T i m e (h)
Fig. 8 - - R e a c t i v a t i o n of carbon during the batch elution of gold cy-
anide at 60 ~ in 500 m L of K O H solution at a pH of 10.7. The
3.0 g of carbon was loaded with gold for 21 days and then pretreated
for 30 min at 20 ~ in 5 m L solution of 7.3 g CN / L .
METALLURGICAL AND MATERIALS TRANSACTIONS B
quantify the effect of the cyanide pretreatment, or the
cyanide loading, on the gold cyanide equilibrium. When
sufficient time has elapsed to reach equilibrium, the ef-
fect of the passivation has already been destroyed through
reactivation.
It was shown in a previous article t~j that the parame-
ters in the functional relationships between elution con-
ditions and adsorption/desorption isotherms are strongly
dependent on the conditions of pretreatment. Further-
more, because reactivation of the carbon surface pro-
ceeds with time, it is necessary to devise a method to
estimate the equilibrium curve after a cyanide pre-
treatment, but during the subsequent elution stage. The
method described subsequently was used to estimate the
change in the equilibrium isotherm parameter A during
an AARL elution in a packed bed of carbon if the mod-
ified Freundlich isotherm tlJ is defined as follows, with
Qe and Ce being the equilibrium loading and solution
concentration, respectively:
Qe = A" Ce -0"002688"A+0"2902 [1 1]
If it is assumed that resistance to mass transfer becomes
negligible after the pretreatment step, t2j equilibrium be-
tween the gold cyanide on the carbon surface and that
in the interparticle solution will pertain during the elu-
tion step at every position in the column, including the
outlet. As the concentration of gold in solution is mea-
sured at the outlet, the equilibrium at the outlet will be
known if the loading on the carbon is known. The latter
can be estimated by assuming that at any stage, the load-
ing at the outlet will be lower than or equal to the initial
gold loading and higher than the average loading in the
column. The average loading at time t can be calculated
from a gold balance by integration of the area under the
gold elution profile from t = 0 to t = t. With the maxi-
mum and minimum loadings at the outlet known, the
minimum and maximum values of A at the outlet thus
can be calculated from Eq. [11]
Figure 9 shows the result of applying the preceding
method to the elution data of Figure 1, with the values
of A varying between 0.8 and 2.1. While the initial gold
loading was a constant value, the average loading in the
2.5
2
A (max.)]
~ A (min.) [
1.5
<
0.5 I ! I I !
0 5 10 15 20
Bed volumes
Fig. 9 - - E s t i m a t i o n of the m i n i m u m and m a x i m u m A values in the
shifting isotherm during the A A R L elution depicted in Fig. 1.
VOLUME 25B, DECEMBER 1994--835
column decreased with an increase in the volume of eluant The observed reactivation is probably caused by the
that passed through the column. The difference between removal of the decomposition products of cyanide from
the maximum and minimum Qe values, and conse- the carbon surface as eluant flows through the carbon
quently the difference between the maximum and mini- pores. However, this postulated phenomenon would be
mum A values, increased with time (or bed volumes). very difficult, if not impossible, to quantify or to inves-
However, during the initial two to three bed volumes tigate as an isolated process. The volume of eluant with
when little gold had been eluted, the values of A were which the carbon has been rinsed thus can be taken as
more exact. The latter A values were plotted in Figure 10 being representative of the degree of reactivation of the
as a function of the potassium concentration. The power carbon surface. Hence, the reactivation in Figure 11 is
relationship between A and the potassium concentration accounted for empirically by expressing the value of A
in Figure 10 serves as further proof of the validity of an as a function of the number of bed volumes:
equilibrium approach, as it was shown previously t*~ that
such a relationship exists at true equilibrium. The ex- A = 2.16 (Vs + 1)~ [~2]
ponent in this relationship is 0.24, as determined from
the slope in Figure 10, which corresponds reasonably Figure 12 depicts the gold profile for the experiment
with the value of 0.20 determined for true equilibrium. 141 considered in Figure 11 and indeed shows some degree
Therefore, this is a new method to estimate the depen- of readsorption of gold at the constant level of potassium
dence of the equilibrium curve on the concentration of
spectator cations.
Consequently, if the concentration of spectator cations 3
is kept constant during an AARL elution, the value of 2.9
A should also be constant if all other factors remain the 2.8
same. This was examined in Figure 1 l, where the cal-
2.7
culated minimum and maximum A values were plotted
as a function of the number of bed volumes that passed 2.6
through the column with a constant level of potassium 2.5
in the eluant. As a result of the high potassium concen- < 2.4 --@-- A (max.)
tration, little elution of gold, if any, occurred, leading 2.3
to a very small difference between the minimum and 2.2
maximum estimates of A. However, Figure 11 shows
2.1
that the A values did not remain constant, but increased
2 1 ! t i
with time. The carbon had been loaded over a period of
3 days, so that a loading gradient of adsorbed gold 0 5 10 15
throughout the carbon particles was unlikely, t2~] There-
fore, this increase in A could not be ascribed to re- Bed volumes
adsorption caused by a favorable loading gradient. Fig. 11 - - E s t i m a t i o n of the m i n i m u m and m a x i m u m A values for a
Consequently, reactivation of the carbon surface toward constant concentration o f 12.0 g KC1/L in the eluant with a flow rate
gold cyanide after a cyanide pretreatment is the only other of 47.6 m L / h during an A A R L column elution at 70 ~ The 9.41 g
of carbon initially contained 4.81 g A u / k g and was pretreated for
mechanism responsible for this readsorption and pro- 30 min at 20 ~ in 20 m L of solution of 20 g K C N / L .
vides further support for the reactivation postulated in
Figure 8.

18 I
16 . . . . . . . . . . . . . . . . . . .

14
0.8
ill 9 Experimental
"~ 12 , Model with
"~1 reactivation
0.6 _,r lo

0.4 8

9 Exp
0.2
4 I t I I I I : : '

0 2 4 6 8 10 12 14 16 18
0
5 5.5 6 6.5 7 7.5 8 8.5 Bed volumes

In ( K + 5 0 ) Fig. 1 2 - - T h e use of an equilibrium model to simulate an A A R L col-

u m n elution at 70 ~ with a constant potassium concentration of
Fig_ 10--Quantification of the effect of potassium on the equilibrium 12.0 g K C I / L in the eluant at a flow rate of 47.6 m L / h . The 9.41 g
of gold elution after a cyanide pretreatment, based on the average A of carbon initially contained 4.81 g A u / k g and was pretreated at 20 ~
values in Fig. 9, for 30 min in 20 m L solution o f 20 g K C N / L

836--VOLUME 25B, DECEMBER 1994 METALLURGICAL AND MATERIALS TRANSACTIONS B

 in the eluant. When an equilibrium-based elution model,
such as that proposed in a previous article, [21 is applied
to the experimental results of Figure 12 without consid-
ering reactivation, no desorption or adsorption is pre-
dicted because of the high level of potassium. However,
if a reactivation term, such as that in Eq. [12], is in-
corporated in the model, a satisfactory fit of the experi-
mental data is obtained. This provides further evidence
that reactivation of the carbon surface is one of the sub-
processes constituting the elution of gold cyanide.
Nevertheless, Van der Merwe [41 showed that the
equilibrium-based elution model becomes less dependent
on reactivation at temperatures of about 130 ~
VH. CONCLUSIONS AND SIGNIFICANCE
The most important subprocesses constituting the elu-
tion of aurocyanide are the decomposition or adsorption
of cyanide on the surface of the carbon and the elution
of cyanide and the concomitant removal of spectator cat-
ions, as well as the subsequent elution of aurocyanide.
Spectator cations (M~§ depress the desorption of auro-
cyanide by the formation of the neutral M'§
species on the carbon. The formation of these species is
promoted by the presence of high concentrations of cat-
ions in the solution, as found in the pretreatment step of
an AARL elution. The horizontal position of the gold
peak in an AARL elution profile is therefore a function
- of the dynamics of removal of the cations present in the
carbon bed after the pretreatment step. Consequently, the
aurocyanide experiences a shifting isotherm during an
elution run.
When the concentration of cations in the eluant is high
and no cyanide is present in solution or on the carbon,
very little desorption of gold is observed. However, if
such carbon has been pretreated with cyanide and the
concentration of cations in the eluant is reduced signif-
icantly, the aurocyanide desorbs rapidly. Therefore, the
free cyanide present in the wash cycle of an AARL elu-
tion is of minor importance in comparison with the cy-
anide in the pretreatment step. During pretreatment,
cyanide reacts with the functional groups on the carbon
surface and thereby passivates the carbon for the ad-
sorption of anionic metal cyanides. It is this passivation
and not the free cyanide carded over from the pre-
treatment step that causes the elution of the gold in the
washing step of an AARL elution. Similarly, passivation
occurs in a Zadra elution, except that it occurs during
the elution. The height of the gold peak in an AARL
elution is determined by the temperature and severity of
pretreatment and, hence, the degree to which the equi-
librium of gold loading has been lowered.
The degree of passivation, which is determined to a
large extent by the temperature of pretreatment, also af-
fects the elution of cations and the degradation/adsorption
of cyanide itself. Little more cyanide will decompose on
the carbon during the washing step of an AARL elution,
so that the elution of gold reveals a low sensitivity to-
ward variations in the cyanide concentration in the wash-
ing step. This agrees with the observation by Davidson[221
that an increase in the cyanide concentration from 3 to
10 pet in the pretreatment step had little effect on the
METALLURGICAL AND MATERIALSTRANSACTIONS B
efficiencyof gold elution. At cyanide concentrations below
3 pet, the gold-elution profile was much more sensitive
to the addition of cyanide in the pretreatment step. The
carbon, therefore, had reached its maximum passivation
at a cyanide concentration of 3 pct in the pretreatment
and was unaffected by higher cyanide concentrations. A
similar result was obtained by Laxen et al.,[23] who ob-
served that a reduction in pretreatment time from 8 to
2 hours and a four times weaker pretreatment solution
had no detrimental effect on the elution.
A new method is proposed in which the dependency
of the adsorption/desorption isotherm for aurocyanide
on the concentration of spectator cations can be esti-
mated from the elution profiles for gold and cations.
Reactivation of the carbon surface occurs when the
adsorbed/decomposed cyanide is removed by the eluant.
This reactivation can be observed more accurately when
a high concentration of cations in the eluant depresses
the desorption of aurocyanide. At high temperatures of
pretreatment, such as those used in practice, it is not
necessary to include a reactivation term in the mathe-
matical model for elution.
REFERENCES
1. J.S.J. Vail Deventer and P.F. Van tier Merwe: Miner. Eng., 1994,
vol. 7 (1), pp. 71-86.
2. J.S.J. Van Deventer, L. Lorenzen, P.F. Van der Merwe, D.W.
Morrison, and J.P.W. Van der Westhuysen: Miner. Eng., 1994,
vol. 7 (2-3), pp. 265-78.
3. M.D. Adams and C.A. Fleming: Metall. Trans. B, 1989, voi. 20B,
pp. 315-25.
4. P.F. Van der Merwe: Ph.D. Thesis, University of Stellenbosch,
Stellenbosch, South Africa, 1991, p. 453.
5. D.M. Muir, W.D. Hinchliffe, N. Tsuchida, and M. Ruane:
Hydrometallurgy, 1985, voi. 14, pp. 47-65.
6. D.M. Muir, W.D. Hinchliffe, and A. Griffin: Hydrometallurgy,
1985, vol. 14, pp. 151-69.
7. M.D. Adams and M.J. Nicol: Gold 100: Proc. Int. Conf. on
Gold, C.E. Fivaz, and R.P. King, eds., South African Institute
of Mining and Metallurgy, Johannesburg, South Africa, 1986,
vol. 2, pp. 111-21.
8. R.J. Davidson and V. Veronese: J. S. Afr. Inst. Min. Metall.,
1979, vol. 79, pp. 43745.
9. P.F. Van der Merwe and J.S.J. Van Deventer: Proc. 14th Congress
of CMMI, Institution of Mining and Metallurgy, London, 1990,
pp. 161-71.
10. J.S.J. Van Deventer and P.F. Van der Merwe: Thermochim. Acta,
1993, voi. 221, pp. 99-113.
11. N. Tsuchida: Ph.D. Thesis, Murdoch University, Perth, Western
Australia, 1984, p. 212.
12. J.S.J. Van Deventer and P.F. Van der Merwe: Metall. Trans. B,
1993, vol. 24B, pp. 433-40.
13. N. Tsuchida, M. Ruane, and D.M. Muir: Proc. MINTEK50: Int.
Conf. on Miner. Sci. and Tech., L.F. Haughton, ed., Mintek,
Randburg, South Africa, 1985, vol. 2, pp. 647-56.
14. M.D. Adams, G.J. McDougall, and R.D. Hancock:
Hydrometallurgy, 1987, vol. 19, pp. 95-115.
15. E.H. Cho and C.H. Pitt: Metall. Trans. B, 1979, vol. lOB,
pp. 159-64.
16. G. Banini and W. Stange: Miner. Eng., 1994, vol. 7 (5-6),
pp. 591-604.
17. R.J. Davidson and D. Duncanson: J. S. Aft. Inst. Min. Metall.,
1977, voi. 77 (12), pp. 254-61.
18. M.D. Adams: J.S. Afr. Inst. Min. Metall., 1990, vol. 90 (2),
pp. 37-44.
19. W. Hoecker and D.M. Muir: Proc. Syrup. on Research and
Development in Extractive Metallurgy, Australasian Institute of
Mining and Metallurgy, Adelaide, 1987, pp. 29-37.
VOLUME 25B, DECEMBER 1994--837
20. D.M. Muir, A. Aziz, and W. Hoecker: Proc. 1st Int.
Hydrometallurgy Conf., Beijing, Z. Yulian and X. Jiazhong, eds.,
International Academic Publishers, Beijing, Oct. 1988, pp. 461-65.
21. J.S.J. Van Deventer: Chem. Eng. Comm., 1986, vol. 44,
pp. 257474.
838--VOLUME 25B, DECEMBER 1994
22. R.J. Davidson: in Carbon-In-Pulp School, South African Institute
of Mining and Metallurgy, Johannesburg, 1986.
23. P.A. Laxen, C.A. Fleming, D.A. Holtum, and R. Rubin: Proc.
12th CMMI Congress, H.W. Glen, ed., South African Institute
of Mining and Metallurgy, Johannesburg, 1982, pp. 551-60.
METALLURGICAL AND MATERIALS TRANSACTIONS B