Vous êtes sur la page 1sur 9

Hydrometallurgy 169 (2017) 103111

Contents lists available at ScienceDirect

Hydrometallurgy

journal homepage: www.elsevier.com/locate/hydromet

Mechanism and kinetics of gold leaching by cupric chloride


Matti Lampinen a,, Sipi Seisko b, Otto Forsstrm b, Arto Laari a, Jari Aromaa b,
Mari Lundstrm b, Tuomas Koiranen a
a
Lappeenranta University of Technology, LUT School of Engineering Science, P.O. Box 20, FI-53851 Lappeenranta, Finland
b
Aalto University, School of Chemical Technology, Department of Materials Science and Engineering, P.O. Box 16200, 00076 Aalto, Helsinki, Finland

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

Article history: A non-cyanide lixiviant for gold dissolution, cupric chloride, was studied. The work used a rotating disk electrode
Received 10 May 2016 technique to investigate the dissolution reaction mechanism and kinetics of gold dissolution in concentrated
Received in revised form 6 December 2016 chloride (3 M) solution with cupric ion concentration in the range 0.021.0 M and temperature 6595 C. A re-
Accepted 15 December 2016
action mechanism including a disproportionation step of Au(I) is proposed. Increase in the gold dissolution rate
Available online 18 December 2016
was found to be proportional to temperature and cupric ion concentration up to 0.5 M. The dissolution rate was
Keywords:
observed to decrease with cupric ion concentration of 1.0 M. A mechanistic kinetic model was developed to study
Leaching of gold the rate-controlling steps in the dissolution process. Uncertainties in the model parameters of the mechanistic
Cupric chloride kinetic model were studied with Markov chain Monte Carlo (MCMC) methods. The modeling results showed
Reaction mechanism that the gold dissolution is mainly under mixed control, where both the intrinsic surface reaction and mass trans-
Kinetics fer have an effect on the overall dissolution rate.
Mechanistic modeling 2016 Elsevier B.V. All rights reserved.

1. Introduction process is that the cupric ions can be considered a natural and self-prop-
agating oxidant for the leaching process. Thus, chloride leaching with
Cyanidation is currently the prevailing technology in gold produc- Cu2 + as the oxidant would appear to be a promising alternative to
tion. However, many factors are forcing the precious metal industry to cyanidation. However, only a limited amount of published data is avail-
reconsider the use of cyanide as a lixiviant. Interest in the use of non-cy- able about gold dissolution behavior and kinetics in chloride solutions
anide methods for gold dissolution arises primarily from concerns re- with Cu2+ as the oxidant.
garding the toxicity of cyanide and the inability of cyanide solution to Nikoloski and Stockton (2000) reviewed oxidation of gold and its
effectively leach gold from complex ores. A number of alternatives to cy- complexation in acidic aqueous chloride solutions and based on previ-
anide have been presented (Hilson and Monhemius, 2006) such as thio- ous studies (Gallego et al., 1975; Nicol, 1980; Marsden and House,
sulfate, halides, thiourea and thiocyanate. In recent years, thiosulfate 1992; Schmid, 1985; Lide and Frederikse, 1997) presented that both
(Feng and van Deventer, 2010; Lampinen et al., 2015a; Xu et al., Au(I) and Au(III) species are formed by anodic half reactions given as:
2016), chloride (Hasab et al., 2013, 2014; Nam et al., 2008) and thiourea

(Jing-ying et al., 2012; Gnen et al., 2007) have received the greatest AuCl2 eAu 2Cl ; E0 1:113 V 1
attention.

Chloride leaching can be considered to be a promising alternative to AuCl 3eAu 4Cl ; E0 0:994 V 2
cyanide leaching that overcomes problems regarding toxicity and offers
a possibility for direct refractory ore leaching (Lundstrm et al., 2015). Nicol (1980) presented that electrochemical dissolution of gold in
Gold production with chloride leaching has been investigated with acidic chloride solutions takes place at potentials below 1.2 V with the
both primary (Hyvrinen et al., 2004; Pangum and Browner, 1996) formation of both Au(I) and Au(III) species. It has been reported that
and secondary raw material (Yazici and Deveci, 2015; Nam et al., at potentials below 0.8 V gold dissolves as Au(I) species, while at poten-
2008). Cupric chloride leaching of gold has also been investigated as a tials above 1.1 V it forms Au(III) species (Frankenthal and Siconol,
following step for copper sulde leaching processes such as Intec 1982).
(Moyes, 2002), HydroCopper (Haavanlammi et al., 2007), and Nippon Several reaction mechanisms have been proposed (Diaz et al., 1993;
N-Chlo/NIKKO (Ishiguro et al., 2008). The advantage of this kind of Kozin and Melekhin, 2004; von Bonsdorff et al., 2007; Putnam, 1944;
Nicol, 1980) for gold dissolution in aqueous chloride solutions. The elec-
trochemistry of gold in acid chloride solutions has been studied to in-
Corresponding author. vestigate corrosion behavior, to dene the reactions and mechanisms
E-mail address: matti.lampinen@lut. (M. Lampinen). involved in the electrorening process, and to examine the leaching of

http://dx.doi.org/10.1016/j.hydromet.2016.12.008
0304-386X/ 2016 Elsevier B.V. All rights reserved.
104 M. Lampinen et al. / Hydrometallurgy 169 (2017) 103111

gold. Studies have been carried out under many different conditions and be considered to be negligible, since the average redox potential for cu-
with many different solutions. Cupric chloride leaching for gold produc- pric concentration range from 0.02 to 1.0 M was 0.449 0.003 V with
tion has been studied by von Bonsdorff et al. (2005, 2006, 2007) and by air sparge and 0.446 0.004 V with nitrogen sparge at T = 25 C. Fur-
McDonald et al. (1987). Von Bonsdorff et al. (2005) conducted electro- thermore, at T = 95 C, the average redox potential for cupric concen-
chemical measurements to obtain the dissolution rate of gold in Outotec tration range from 0.1 to 1.0 M was 0.595 0.013 V with air sparge
HydroCopper process solution and presented that the dissolution rate and 0.610 0.009 V with nitrogen sparge. Hence, it can be concluded
of gold increases with increasing cupric ion concentration, increasing that dissolved oxygen does not affect gold dissolution in the system.
chloride concentration and increasing temperature. Von Bonsdorff et The chemicals used were NaCl and CuCl2 2H2O, supplied by VWR
al. (2007) proposed a simplied reaction mechanism for gold oxidation, and of technical grade.
in which gold is dissolved as trivalent auric ions. McDonald et al. (1987) The RDE measurements were performed in a water-jacketed three
and von Bonsdorff et al. (2007) presented that the effect of pH on the electrode cell with volume of 200 mL (a water bath, Lauda M3, was
dissolution rate of gold is negligible as long as the solution is sufciently used for heating). The volume of solution used was 110 mL. A gold rotat-
acidic (pH b 3). ing electrode (Pine Research Instrumentation Inc.) was used as the
While the mechanism of gold dissolution is important, it is the kinet- working electrode, and a pure platinum plate as the counter electrode.
ics that determines whether the dissolution can be carried out in a time The diameter of the gold disc was 4.08 mm, and the disc was attached
frame acceptable for industrial application. Marsden and House (1992) into a polytetrauoroethylene (PTFE) sheath. An Ag/AgCl electrode
presented that the mass transport of chloride ions to the surface is the from SI Analytics with a potential of 197 mV vs. standard hydrogen elec-
rate-determining step, and that the chemical reaction step is fast in trode (SHE) was used as the reference electrode (Bard and Faulkner,
gold dissolution in chloride solutions. Nikoloski and Stockton (2000) ar- 1980).
gued that gold dissolution in chloride solutions is controlled by diffusion Linear polarization resistance (LPR) can be determined from RDE
of either the chlorine (oxidant) or chloride (complexant) ion through measurement. LPR is inversely related to the corrosion current density
the Nernst boundary layer to the gold surface. Jeffrey et al. (2001) pre- of the working electrode. The LPR measurements were executed with
sented that the chloride concentration used in chloride leaching is usu- ACM Instrument's Gill AC potentiostat using an LPR sweep program
ally very high (~100 g/L), and hence the dissolution rate is ultimately with Gill AC Sequencer software. The LPR measurements were per-
limited by the diffusion of the oxidant. However, it was argued by von formed from 10 mV below the rest potential to 10 mV above the rest po-
Bonsdorff et al. (2005, 2007) that the reaction rate is under mixed con- tential. Three replicate measurements were conducted for every
trol and, both charge transfer and mass transfer inuence the rate of condition in order to avoid random errors. In addition, several imped-
gold dissolution. ance measurements were performed to ensure that uncompensated
It is evident that the mechanism and kinetics of gold chloride ohmic resistance did not cause errors in LPR determination.
leaching are not yet fully-understood and there is thus a need for further The LPR value can be determined as the slope of the potential-cur-
studies. In this work, experimental data from rotating disk electrode rent density diagram. When the LPR value increases, the current density
(RDE) experiments are used to study the gold leaching mechanism decreases and, therefore, the dissolution rate also decreases. Eq. (3)
and gold leaching kinetics in chloride solution in the presence of shows the proportionality between LPR and the corrosion current den-
Cu2+. An electrochemical technique was chosen to study the kinetics sity, jcorr, i.e. the current density at open circuit potential (OCP) (Stern
and gold leaching mechanism because such approaches enable study and Geary, 1957):
without undue interference from factors such as the matrix of the
leaching solution and the ore/concentrate mineralogy. As presented ba bc 1 B
jcorr  3
above, it would appear that the reaction rate is under mixed control. 2:303ba bc Rp Rp
In the case of mixed control in which the leaching rate has both intrinsic
surface reaction rate and mass transfer limitations, it is necessary to sep- where ba is the Tafel slope coefcient of the anodic side (mV/decade), bc
arate these effects. For this purpose, a mechanistic kinetic model was the coefcient of the cathodic side (mV/decade), B the systematic coef-
developed to evaluate the intrinsic gold reaction rate at the gold surface cient (mV) and Rp the LPR ( cm2). Since the sweep range is very short
and the mass transfer limitations. The developed model includes kinetic in the LPR measurements, the Tafel method cannot be used to interpret
parameters that need to be estimated from measured data. Parameter the LPR data. However, the value of B was calculated from separate Tafel
estimates are often uncertain since they are estimated from incomplete measurements and its value was found to be 13.5 0.5 mV. From these
and noisy measurements. Hence, the uncertainty of the model parame- RDE measurements, the obtained values of the corrosion current densi-
ters of the developed mechanistic model was evaluated with Markov ty were converted into dissolution rates with a unit g/s/m2.
chain Monte Carlo (MCMC) methods.
3. Results and discussion
2. Materials and procedures
3.1. Reaction mechanism of gold dissolution
RDE experiments were carried out to study gold dissolution phe-
nomena and the rate controlling steps. A total chloride concentration Many studies (Nicol, 1980; Frankenthal and Siconol, 1982; Diaz et
of 3 mol/L, which can be considered to be very high, was used in all ex- al., 1993) have presented that gold forms aurous Au(I) and/or auric
periments, thus guaranteeing that the leach rate is ultimately limited by Au(III) ion complexes with chloride ions depending on the chloride con-
diffusion of the oxidant (Jeffrey et al., 2001). The solution pH was adjust- centration and electrode potential. According to Diaz et al. (1993), gold
ed to 1.0 in every experiment by adding HCl (1.0 M). The cupric ion con- dissolves as an [AuCl2] complex with oxidation state +1, when the po-
centrations used were 0.02, 0.1, 0.5 and 1.0 M. The calculated tential is 0.8 V vs. SCE and also as [AuCl4] complex with oxidation state
concentration of the AuCl 2 complex varied from 1.6 10
8
M to + 3 at higher potentials. Frankenthal and Siconol (1982) suggested
7
8.3 10 M, which indicates that the concentration of cuprous ions that gold dissolves as aurous ions when the potential is below 0.8 V
formed in the leaching process (Eq. (8)) is relatively small compared vs. SCE, but as auric Au(III) ions when the potential is above 1.1 V vs.
to the cupric ions. Hence, it is justied to assume that the effect of cu- SCE. Furthermore, Nicol (1976b) proposed that [AuCl2] complexes ox-
prous ions in the current conditions is negligible. The temperatures idise into [AuCl4] complexes in a disproportionation reaction.
used were 65, 75, 80, 85, 88, 92 and 95 C. Gas was not added to the sys- Corrosion potentials and a plot of logarithmic corrosion current den-
tem during the experiments since it would have disrupted the RDE sity as a function of logarithmic cupric concentration were used in the
measurements. However, the possible effect of dissolved oxygen can current work to determine the oxidation state of the gold. With cupric
M. Lampinen et al. / Hydrometallurgy 169 (2017) 103111 105

ion concentrations 0.02 and 0.1 M, the measured corrosion potentials


varied from 708 to 787 mV vs. SCE, which indicates in the opinion of
Diaz et al. (1993) that the oxidation state is +1. With cupric concentra-
tions 0.5 and 1 M corrosion potentials varied from 839 to 859 mV vs.
SCE, which indicates in the opinion of Diaz et al. (1993), that the oxida-
tion state + 3 is also possible. The equilibrium potentials were deter-
mined in the current work in chloride-gold solution, with [Cl] =
3 M, and the gold concentration in the diluted solution was estimated
to be 106 M. The equilibrium potentials were calculated with the
Nernst equation (Eq. (4)) for Au/AuCl
2 and Au/AuCl4 and they are pre-
sented in Fig. 1:

RT AuClx 
E E0 ln x 4
zF Cl 

where E0 is the standard potential (V) determined by HSC software


(2015), z the number of transferred electrons during the reaction, T is
the temperature (K), [AuClx] the concentration of AuCl
2 or AuCl4 , and Fig. 2. Equilibrium potentials (E vs. SCE) of AuCl
2 and AuCl4 and the measured corrosion

[Cl ] the concentration of chlorides. Fig. 1 illustrates the measured cor- potentials (E vs. SCE) as a function of cupric ion concentration (M).
rosion potentials (V vs. SCE) in the RDE measurements with [Cl] = 3 M
at temperatures between 65 and 95 C compared to the calculated
values (HSC (2015) and Nerst equation) equilibrium potentials of as auric ions at higher Cu2+ concentrations. Nicol (1980) presented that
AuCl
2 and AuCl4 for temperatures between 20 and 100 C. The equilib- the mechanism for the dissolution of gold in chloride solutions involves
rium potentials (V vs. SCE) of Au/AuCl 4 did not vary signicantly de- the formation of Au(I) as an intermediate species that can either be fur-
spite the increased temperature. However, the equilibrium potentials ther oxidised to Au(III) in a rate determining two-electron step or dif-
of Au/AuCl 2 decreased as temperature increased. fuse into the solution.
Fig. 2 presents the corrosion and equilibrium potential as a function Gold dissolution (Nicol, 1976a) and the disproportionation of Au(I)
of cupric ion concentration. The measured data show the corrosion po- (Nicol, 1976b) species has been studied in acid chloride media, although
tential with [Cu2+] = 0.011.0 M, [Cl] = 0.5 M, T = 95 C and pH = 1, in relatively concentrated gold solutions typical of the electrolytes used
and the lines for AuCl
2 and AuCl4 present the calculated equilibrium in electrorening. Gold dissolution (Nicol, 1976a) in chloride electro-
potential of the complex at the same conditions. lytes was found always to produce some fraction x of solubilized gold
The measured potential values presented in Figs. 1 and 2 indicate as Au(I) species and a mechanism was proposed based on Eqs. (5)(7):
that gold can dissolve as AuCl 2 in the leaching environment used in
this work, since the corrosion potentials are higher than the equilibrium AuAuIads e 5
potentials of AuCl2 but lower than the equilibrium potentials of AuCl4

regardless of temperature and cupric concentration. Von Bonsdorff xAuIads xAuIIIaq 2xe 6
(2006) determined the value of the slope (the logarithm of dissolution
rate of gold versus the logarithm of copper concentration) as 1xAuIads 1xAuIaq 7
being around 0.33 for gold leaching at T = 90 C, [Cu2 +] = 030 g/L
(00.47 mol/L), [NaCl] = 280 g/L, pH = 2 and solution ow of
Nicol (1976a) presented that increased agitation, high temperatures,
200 mL/min. It was suggested that the reaction order value is close to high concentrations of chloride and low current densities are factors
1 at low Cu2 + concentrations whereas the reaction order is lower
that will result in increased production of Au(I) in the solution. Au(I)
(0.315) at high Cu2+ concentrations. Von Bonsdorff (2006) suggested species were shown to disproportionate, which was identied as the
that three cupric ions were required to dissolve one gold atom and pro-
source of rough deposits in the electrorening process (Nicol, 1976b).
posed that gold dissolves as aurous ion when [Cu2+] is 0.1 M or less and The rate of disproportionation of Au(I) in chloride solutions was found
to be very low, but enhanced in the presence of elemental gold, which
would provide low energy sites on which gold deposits could grow.
The presence of oxygen was also shown by Nicol (1976b) to increase
the rate of Au(I) disproportionation, although in the presence of ele-
mental gold.
Fig. 3 presents the obtained logarithm of corrosion current density as
a function of the logarithmic of cupric ion concentration. The reaction
order can be determined from the slope of Fig. 3, which varies from
0.27 to 0.34. A possible reaction mechanism can be presented as follows.
Gold dissolves as AuCl 2 according to Eqs. (5) and (7). Nicol (1976b)
presented that gold(I) is unstable in aqueous chloride solutions with re-
spect to disproportionation into Au(III) and metallic gold. Hence, it is
proposed here that Eq. (7) is reversible and that Au(I) is
disproportionated through Eq. (6), as the homogeneous disproportion-
ation of gold(I) is argued by Nicol (1976b) to be very slow. The reaction
mechanism can be written as follows:


Fig. 1. Measured corrosion potentials (V vs. SCE) in RDE measurements with [Cl] = 3 M Au0 Cu2 2 Cl AuCl2 aq Cu 8
at temperatures between 65 and 95 C compared to equilibrium potentials (V vs. SCE) of
AuCl
2 or AuCl4 in 3 M chloride and 10
6
M gold solution at temperatures between 20 and
100 C.
AuCl2 aq AuCl2 ads 9
106 M. Lampinen et al. / Hydrometallurgy 169 (2017) 103111

solution temperatures are presented in Fig. 4. In the gure, it can be ob-


served that the LPR decreases as the temperature and rotational speed
increase. The change in LPR is most signicant between 100 and
500 rpm. This indicates that mass transfer affects the gold dissolution
rate most at low rotational speed. Rotational speeds higher than 500
or 1000 rpm did not remarkably affect the LPR values. Additionally,
the kinetics of gold dissolution increased most when the temperature
was increased from 65 to 75 C.
The dissolution rates of gold can be calculated from the LPR values.
The dissolution rates of gold at different temperatures are presented
in Fig. 5. In the gure, it can be seen that the gold dissolution rate in-
creases when temperature increases, as expected. McDonald et al.
(1987) and von Bonsdorff et al. (2005) presented that the dissolution
rate of gold increases linearly with increasing temperature in cupric
chloride solutions. Von Bonsdorff et al. (2007) presented that dissolu-
tion rate of gold varies in the range 12 m/h, which is comparable to
the values (02 m/h) measured in the current work.

Fig. 3. Logarithmic corrosion current density as a function of logarithmic cupric


concentration in a solution with [Cu2+] = 0.021.0 M, T = 95 C, [Cl] = 3.0 M and 3.3. Effect of Cu2+ concentration
pH = 1.0 (adjusted with 1.0 M HCl).
The effect of cupric ion concentration on LPR was studied in chloride
solutions (3 M) at T = 95 C and cupric ion concentrations of 0.02, 0.1,
0.5 and 1.0 M. The results of these measurements for CuCl2 solutions are

3 AuCl2 ads AuCl4 2Au0 2 Cl 10 illustrated in Fig. 6. In the gure, it can be seen that the linear polarization
resistance decreases when the cupric ion concentration is increased from
The proposed reaction mechanism explains the conict between the 0.02 to 0.5 M. However, linear polarization resistance increases when the
relatively low corrosion potential, indicating Au(I) as the reaction prod- cupric ion concentration is increased from 0.5 to 1.0 M.
uct and the high oxidant consumption, indicating Au(III) as the reaction Fig. 7 describes the relation between the dissolution rate and the oxi-
product. The measured polarization resistances and calculated current dant concentration. From Fig. 7, it can be observed that the dissolution
densities describe gold dissolution as the Au(I) complex, but the actual rates increase as the rotational speed is increased. When the copper con-
gold dissolution rate is only one third of that because of back precipita- centration is increased, the dissolution rate of gold increases up to 0.5 M.
tion of gold, as shown in Eq.(10). When the copper concentration is further increased from 0.5 M to 1.0 M
the dissolution rate of gold decreases. Von Bonsdorff et al. (2005) present-
3.2. Effect of temperature ed that the dissolution rate of gold increases linearly with increasing cu-
pric chloride concentration. However, the highest cupric chloride
The effect of temperature and rotational speed on gold dissolution concentration studied by von Bonsdorff et al. (2005) was 40 g/L
was studied for the temperature range 6595 C and rotational speed (0.63 M). McDonald et al. (1987) presented that the dissolution of gold in-
1002500 rpm in solution where the cupric concentration was 0.5 M creases when cupric chloride concentration is increased up to 0.6 M cu-
and chloride concentration 3 M. The impedance measurements indicat- pric chloride. McDonald et al. (1987) noticed that with initial cupric
ed that the effect of solution resistance was minor. The ohmic resistance chloride concentration over 0.6 M only a small increase in gold dissolution
varied in the range of 0.92.7 cm2, which corresponds to approxi- is observed and presented that this is because the rise in cuprous concen-
mately 23% of the polarization resistance. Additionally, the charge tration begins to limit the process. However, the concentration of cuprous
transfer resistance from the impedance measurements was shown to ions can be considered to be relatively low in the current study. Hence,
be at the same level as the polarization resistances from the RDE mea- with the data available it cannot be directly concluded if cuprous ions
surements. The LPR values as a function of rotation rate for different cause the lower dissolution rates with the initial cupric ion concentration
of 1.0 M. Consequently, further research is required in order to reveal

Fig. 4. The results for the LPR values for gold electrode (RDE) in solution with [Cu2+] = Fig. 5. Dependence of gold dissolution rate on rotational speed and temperature.
0.5 M, T = 6595 C, [Cl] = 3.0 M and pH = 1.0 (adjusted with 1.0 M HCl). Measurements were performed at [Cu2+] = 0.5 M and [Cl] = 3.0 M.
M. Lampinen et al. / Hydrometallurgy 169 (2017) 103111 107

The temperature dependence of the rate constant is taken into ac-


count by the Arrhenius equation given in a parameterised form as:
  
1 1
k kmean E 12
T T mean

where Tmean is a reference temperature, kmean is the rate constant at the


reference temperature, and E is the activation energy.
It is expected that the dissolution rate can be described by a mixed-
control mechanism where both the surface reaction and diffusion affect
the gold dissolution rate. The oxidant Cu2+ diffuses through a boundary
layer to the disc surface and the mass transfer of the oxidant through the
boundary layer can be described by the equation:

r kL ccs 13

where kL is the mass transfer coefcient through the boundary layer and
c is the concentration of oxidant in the liquid bulk phase. At steady state,
Fig. 6. LPR of gold RDE in CuCl2 solution as a function of rotational speed. Measurements rs = r and the surface concentration of the oxidant cs can be solved it-
were performed at [Cu2+] = 0.021.0 M, [Cl] = 3.0 M, pH = 1.0 and T = 95 C. eratively from Eqs. (11) and (13). is the stoichiometric coefcient of
Cu2 + to oxidise 1 mol of gold. Based on previous discussion about
gold leaching chemistry, a value of 3 is used for .
cause of the lower dissolution rates. It can, however, be concluded that
Mass transfer to the surface of a rotating disc has been studied by
the optimal initial cupric ion concentration is around 0.5 M.
several authors (Sulaymon and Abbar, 2012; Petrescu et al., 2009; Dib
and Makhlou, 2007). The mass transfer correlation has the general
form:
4. Kinetic model for gold leaching
Sh a1 Rea2 Sc1=3 14
As pointed out above, it is important to determine whether the dis-
solution can be carried out in a nite time and in a time frame accept- where the Sherwood number is dened as:
able for industrial application. In practice, this means nding optimal
leaching conditions and determining the rate limiting steps at these kL d
Sh 15
conditions. The results presented above show that the leaching rate de- D
creases when cupric concentration is higher than 0.5 M. Hence, leaching
is modelled at cupric concentrations of 0.02 to 0.5 M and temperatures where d is the rotating disc diameter, and D is the diffusion coefcient.
of 6595 C, which can be considered the optimal leaching range. The Reynolds number is dened as:
Gold is dissolved from the rotating gold disc electrode as results of
2
the oxidation reaction of gold in the chloride solution by the oxidant Nd
Re 16
Cu2+ at the surface of the disc. It is assumed that the surface reaction
rate can be described by a simple rate equation:
where N is the rotational speed of the electrode, and is the kinematic
viscosity.
n
r s kcs 11 The Schmidt number is dened as:


Sc 17
D
where k is the rate constant, cs is the concentration of oxidant at the disc
surface, and n is the reaction order for the oxidant.
Both the diffusion coefcient and the kinematic viscosity depend on
temperature, which needs to be taken into account when calculating
the mass transfer coefcient.
Lundstrm et al. (2007) have determined Cu2+ diffusion coefcients
in chloride solutions. The diffusion coefcient was found to be
8.5 109 m2/s at 90 C (pH = 2, [Cu2+] = 30 g/L, [NaCl] = 280 g/L).
Correction to other temperatures can be done using the Stokes-Einstein
relation:
   
D D
18
T T1 T T2

where D is the diffusion coefcient of Cu2+ and is the dynamic viscosity.


The diffusion coefcients of Cu2+ were calculated at different tem-
peratures using data given by Lundstrm et al. (2007) and Eq. (18),
and are shown in Table 1. The effect of temperature on the calculated
mass transfer coefcients is shown in Fig. 9.
The gold dissolution rate model composed of Eqs. (11) to (15) has
Fig. 7. Dependence of gold dissolution rate on cupric ion concentration at 0.02, 0.1, 0.5 and
ve parameters: k1,mean, E1, n1, a1, and a2. These parameters were esti-
1.0 M. Measurements were performed at T = 95 C and [Cl] = 3.0 M with rotational mated by comparing the calculated gold dissolution rates to the mea-
speeds of 1002500 rpm. sured rates from the RDE experiments. In the solution of the model,
108 M. Lampinen et al. / Hydrometallurgy 169 (2017) 103111

Table 1 a probability distribution of solutions. MCMC methods have recently


Temperature dependence of water viscosity and diffusion coefcients. been successfully applied in a number of modeling cases to study pa-
T, C , m Pa s DCu2+, m2/s rameter reliability (Vahteristo et al., 2009; Vahteristo et al., 2013;
65 0.433 5.76 109
Lampinen et al., 2015b; Zhukov et al., 2015).
75 0.378 6.79 109 Moreover, the question of the reliability of the model predictions re-
85 0.334 7.91 109 mains unaddressed, i.e., how the uncertainty in the model parameters is
90 0.315 8.5 109 reected in the model response. According to the Bayesian paradigm, all
95 0.298 9.11 109
the parametrizations of the model that statistically t the data are
equally well determined. A distribution of the unknown parameters
can be generated using available prior information (e.g., results obtain-
the surface concentration of the oxidant cs was solved iteratively from ed from previous studies or bound constraints for the parameters) and
the balance between Eqs. (11) and (13). The model parameters were es- statistical knowledge of the observation noise. Computationally, the dis-
timated using the Modest software (Haario, 2002). tribution was generated using the MCMC sampling approach. The
The model parameters were rst estimated with standard least length of the calculated chain for the distribution was 200,000 samples.
squares tting by minimizing the squared difference between the mea- Simple at, uninformative priors with minimum and maximum bounds
sured and the calculated gold leaching rates. The goodness of the t was set for each parameter were used in the calculation of the chain. Up-to-
determined using the R2 value and the standard errors from the estima- date adaptive computational schemes were employed in order to make
tion. However, to evaluate thoroughly the accuracy and reliability of es- the simulations as effective as possible (Haario et al., 2001; Laine, 2008).
timated parameters in a nonlinear multiparameter model, it is In this study, a FORTRAN 90 software package, MODEST 15 (Haario,
important also to consider possible cross-correlation and the 2002), was used for both the least-squares and the MCMC estimation.
identiability of the parameters. Classical statistical analysis that gives The two methods are also implemented in a MATLAB package (Laine,
optimal parameter values, their error estimates, and correlations be- 2008; Laine, 2013).
tween them is based on linearization of the model and is therefore ap-
proximate. Furthermore, it may sometimes even be quite misleading, 5. Modeling results and discussion
especially if the available data are limited and the parameters are poorly
identied. The reliability of the models and their parameters was inves- Comparison of the measured and calculated gold dissolution rates is
tigated in this study using the Markov chain Monte Carlo (MCMC) shown in Fig. 8. The calculated dissolution rate follows the measured
method. The MCMC method is based on Bayesian inference and gives points quite closely. The coefcient of regression for the model was

Fig. 8. Comparison of the measured and predicted gold dissolution rates. The total chloride concentration in all experiments is 3.0 mol/L.
M. Lampinen et al. / Hydrometallurgy 169 (2017) 103111 109

Fig. 9. Ratio of the intrinsic surface reaction rate to reaction rate at bulk oxidant concentration. Total chloride concentration all experiments is 3.0 mol/L.

94.96%. The main discrepancy between the measurements and the sim- 2001; Marsden and House, 1992; Nikoloski and Stockton, 2000; von
ulations can be seen at higher rotational speeds, where the measured Bonsdroff et al., 2005, 2007) presenting that the leaching of gold in chlo-
dissolution rate in some experiments is higher than the result obtained ride systems is under diffusion or mixed control.
by the model. Determining whether a reaction is under diffusion, chemical or
The existence and effect of mass transfer limitations can be studied mixed control is the rst major step to understanding a hydrometallur-
with the established model using the simulated curves presented in gical process. Once the rate controlling step or steps have been deter-
Fig. 9, which show the ratio of the actual leaching rate over the leaching mined, it is important to establish how different parameters affect
rate without diffusion limitations (leaching agent concentration at the them. This is important from the process development and design
surface equals the bulk leaching agent concentration calculated from point of view. As shown above, gold leaching is mainly under mixed
the model as presented). If the ratio presented in Fig. 9 is low ( 1), control. Hence, it is important to know the effect of temperature and
the overall dissolution rate is limited by mass transfer of the oxidant
from the bulk phase to the gold surface. If, on the other hand, the ratio
is close to unity, the dissolution is mainly controlled by the intrinsic sur-
face reaction. Between these two cases, the dissolution rate is affected
by both the reaction and mass transfer steps. The model predicts mainly
a mixed-control system for gold dissolution where both the surface
reaction and the mass transfer limit the dissolution rate. The relative
magnitude of these limitations depends on temperature, oxidant con-
centration and rotational speed. As expected, the obtained results
show that at low rotational speeds and with low oxidant concentrations
(0.02 and 0.1 mol Cu2+/L) the dissolution is almost entirely controlled
by mass transfer. With low oxidant concentrations the concentration
difference between the bulk liquid phase and the disc surface is so low
that the generated mass transfer rate is not enough to maintain high ox-
idant surface concentration even at high stirring speeds. Therefore, the
dissolution rate remains low. At higher rotational speeds the concentra-
tion ratio levels out to 0.5, indicating mixed control in most of the exper-
iments when the rotational speed is higher than 1000 rpm, as shown in
Fig. 9. The results are in agreement with various studies (Jeffrey et al., Fig. 10. Effect of temperature and rotational speed on mass transfer to a rotating gold disc.
110 M. Lampinen et al. / Hydrometallurgy 169 (2017) 103111

Table 2 6. Conclusions
Estimated parameters for the gold dissolution model in the RDE experiments.

Parameter R2 = 94.96% In the current work, the mechanism and kinetics of gold dissolution
Value Std. error, %
in cupric chloride solutions were studied for cupric ion concentration in
the range 0.021.0 M and temperature 6595 C in concentrated chlo-
k1, mean, (m3/kmol)n1 m/s 2.68 105 7.3
ride (3 M) solution. A rotating disk electrode technique was used in
E1, J/mol 7.89 10+4 5.2
n1, - 0.377 21.7 the study to measure dissolution rate. Comparison of the equilibrium
a1 , - 0.442 13.1 and corrosion potentials indicated that gold dissolves as Au(I) rather
a2 , - 1.29 9.2 than Au(III) in the conditions used in this study regardless of tempera-
ture and cupric concentration. A reaction mechanism including a dis-
proportionation step of Au(I) is proposed in the present work. The
rotational speed on the mass transfer rate constant, which can be simu- increase in gold dissolution rate was observed to be proportional to
lated with the developed model and is presented in Fig. 10. In the gure, the temperature and cupric ion concentration up to 0.5 M. At higher cu-
it can be seen that at low rotational speeds (b 1000 rpm) temperature pric ion concentration of 1.0 M, the dissolution rate was found to de-
increase does not have much effect on mass transfer. However, the ef- crease; the precise cause of this remains unknown.
fect of temperature increases with higher rotational speeds A mechanistic kinetic model was developed to study the rate-con-
(N1000 rpm). trolling steps in the dissolution process. The uncertainties in the
The estimated parameters are shown in Table 2. From the table, it model parameters of the mechanistic kinetic model were studied with
can be seen that the activation energy (E1) for the intrinsic surface reac- Markov chain Monte Carlo (MCMC)methods. The modeling results
tion is 78.9 kJ/mol and the reaction order for Cu2+ is 0.38. It may be no- showed that the dissolution of gold is mainly under mixed control,
ticed from the parameter estimation results that the tted exponent for where both the intrinsic surface reaction and mass transfer affect the
the Reynolds number (a2) is much higher than the theoretical value of overall dissolution rate. All the estimated model parameters showed
0.5 in the Levich correlation (Dib and Makhlou, 2007). The exact rea- good reliability. The relative importance of surface reaction and mass
son for this discrepancy is not known, but it can be speculated that transfer limitations depends on temperature, reactant concentration
since the Levich equation was derived for an ideal case, nonidealities and mass transfer conditions. When the rotational speed is lower than
such as surface nonuniformities, vibrations or the existence of vapour 1000 rpm and when the oxidant concentration is low ( 0.1 mol/L),
bubbles probably disturb the boundary layer at the disc surface generat- the overall leaching rate is mostly controlled by mass transfer, which
ing higher mass transfer rates than expected compared to the ideal case. is as expected based on previously reported ndings in the literature.
The reliability of the estimated parameters studied by the MCMC The results clearly show that mechanistic modeling offers a good
method is depicted in Fig. 11 showing the marginal posterior distribu- method to evaluate the importance of rate controlling steps. This is an
tions. It can be seen that the reliability of the parameters is good. All important feature especially from the process development point of
the distributions are well-centred around the most probable point. view, since determining whether the dissolution is under diffusion,

Fig. 11. 2D and 1D marginal posterior distributions for the parameters of the mechanistic model. The lines represent 95% and 50% condence regions and one-dimensional density
estimates.
M. Lampinen et al. / Hydrometallurgy 169 (2017) 103111 111

chemical reaction or mixed control is a rst major step to understanding Haario, H., 2002. Modest User's Guide. ProfMath Oy, Helsinki, Finland.
Haavanlammi, L., Hyvrinen, O., Tiihonen, M., Tontti, R., 2007. Method for Recovering
of a hydrometallurgical process. Gold, Worldwide Patent 093666 A1.
Hilson, G., Monhemius, A.J., 2006. Alternatives to cyanide in the gold mining industry:
List of symbols prospects for the future? J. Clean. Prod. 14, 11581167.
HSC 8.1, 2015, Outotec.
Hyvrinen, O., Hmlinen, M., Lamberg, P., Liipo, J., 2004. Recovering gold from copper
concentrate via the HydroCopper process. JOM 56 (8), 5759.
ba Tafel slope coefcient of the anodic side (mV/decade) Ishiguro, Y., Hosaka, H., Ylshimura, A., Takabayashi, K., Abe, Y., 2008. Process of Leaching
Gold, AU Patent 202814B8.
bc Tafel slope coefcient of the cathodic side (mV/decade) Jeffrey, M.I., Breuer, P.L., Choo, W.L., 2001. A kinetic study that compares the leaching of
B systematic coefcient (mV) gold in the cyanide, thiosulfate, and chloride systems. Metall. Mater. Trans. B Process
c concentration of oxidant in the liquid bulk phase (mol/L) Metall. Mater. Process. Sci. 32 (2), 979986.
Jing-ying, L., Xiu-li, X., Wen-quan, L., 2012. Thiourea leaching gold and silver from the
cs concentration of oxidant at the disc surface (mol/L)
printed circuit boards of waste mobile phones. Waste Manag. 32, 12091212.
d rotating disc diameter (m) Kozin, L.F., Melekhin, V.T., 2004. Extraction of gold from ores and concentrates by leaching
D diffusion coefcient (m2/s) with the use of cyanides and alternative reagents. Russ. J. Appl. Chem. 77 (10),
E activation energy (J/mol) 15731592.
Laine, M., 2013. MCMC toolbox for Matlab, 2013. http://helios.fmi./~lainema/mcmc/
E0 standard potential (V) (accessed April 18, 2016).
jcorr corrosion current density (A/cm2) Laine, M., 2008. Adaptive MCMC Methods with Applications in Environmental and Geo-
k reaction rate constant physical Model. (Ph.D. Thesis). Finnish Meteorological Institute, Helsinki, Finland.
Lampinen, M., Laari, A., Turunen, I., 2015a. Ammoniacal thiosulfate leaching of pressure
kL mass transfer coefcient through the boundary layer (m/s) oxidized sulphide gold concentrate with low reagent consumption. Hydrometallurgy
kmean rate constant at the reference temperature ((m3/kmol)n11 m/s) 151, 19.
n reaction order for the oxidant Lampinen, M., Laari, A., Turunen, I., 2015b. Kinetic model for direct leaching of zinc sul-
phide concentrates at high and solute concentrations. Hydrometallurgy 153,
N rotational speed of the electrode (rpm) 160169.
r mass transfer of the oxidant through the boundary layer Lundstrm, M., Ahtiainen, R., Laihonen, P., Lindgren, M., 2015. Possibilities and challenges
(mol/m2 s) in gold chloride processing, in Proceeding ALTA 2015. pp. 269280.
Lundstrm, M., Aromaa, J., Forsn, O., Hyvrinen, O., Barker, M.H., 2007. Cathodic reac-
rs surface reaction rate (mol/m2 s) tions of Cu2+ in cupric chloride solution. Hydrometallurgy 85, 916.
R gas constant (J/mol K) Lide, D.R., Frederikse, H.P.R., 1997. CRC Handbook of Chemistry and Physics. 76th ed.
Re Reynolds number Cleveland, CRC Press.
Marsden, I., House, I., 1992. Chemistry of Gold Extraction. Hollingsworth, London.
Rp linear polarization resistance (LPR) ( cm2)
McDonald, G.W., Saud, A., Barger, M., Koutsky, J.A., Langer, S.H., 1987. The fate of gold in
Sc Schmidt number cupric chloride hydrometallurgy. Hydrometallurgy 18 (3), 321336.
Sh Sherwood number Moyes, A.J., 2002. The Intec Copper Process, Superior and Sustainable Copper Production,
T temperature (K) S.1. Intec Copper, Australia.
Nam, K.S., Jung, B.H., An, J.W., Ha, T.J., Tran, T., Kim, M.J., 2008. Use of chloride-hypochlo-
Tmean reference temperature (K) rite leachants to recover gold from tailing. Int. J. Miner. Process. 86, 131140.
z the number of transferred electrons during reaction Nicol, M.J., 1980. The anodic dissolution of gold. Part I. Oxidation in acidic solutions. Gold
Bull. 13, 4655.
Nicol, M.J., 1976a. An Electrochemical and Kinetic Investigation of the Behavior of Gold in
Greek symbols Chloride Solutions. II. The Anodic Dissolution of Gold, NIM Rep. 1844, July 7. pp. 120.
Nicol, M.J., 1976b. An Electrochemical and Kinetic Investigation of the Behavior of Gold in
Chloride Solutions. III. The Gold(III)-Gold(I) Reaction on Platinum and the Dispropor-
tionation of Gold(I), NIM Rep. 1846, July 7. pp. 120.
stoichiometric coefcient of Cu2+ to oxidise 1 mol of gold Nikoloski, A., Stockton, N., 2000. Application of alternative lixiviants for secondary heap
solution dynamic viscosity (m Pa s) leaching of gold. Proceedings of 7th Mill Operators Conference, Kalgoorlie, Western
solution kinematic viscosity (cm2/s) Australia.
Pangum, L.S., Browner, R.E., 1996. Pressure chloride leaching of a refractory gold ore. Min.
Eng. 9 (5), 547556.
Acknowledgements Petrescu, S., Secula, M.R., Nemof, G., Greescu, I., 2009. Study on metal anodic dissolution.
Rev. Chim. (Bucureti) 60, 462467.
Putnam, G.L., 1944. Chlorine as a solvent in gold hydrometallurgy. Eng. Min. J. 145 (3),
The Finnish Funding Agency for Innovation (Tekes) and Outotec Oyj
7075.
are gratefully acknowledged for the nancial support (AuChloride Schmid, G.M., 1985. In: Bard, A.J., Parsons, R., Jordan, K., Dekker, M. (Eds.), Gold, in Stan-
project). dard Potential in Aqueous Solution, pp. 313320.
Stern, M., Geary, A.L., 1957. Electrochemical polarization. I. A theoretical analysis of the
shape of polarization curves. J. Electrochem. Soc. 104 (1), 5663.
References Sulaymon, A.H., Abbar, A.H., 2012. Mass transfer to amalgamated copper rotating disc
electrode. J. Electrochem. Sci. Technol. 3, 165171.
Bard, A.J., Faulkner, L.R., 1980. Electrochemical Methods Fundamentals and Applications. Vahteristo, K., Maury, S., Laari, A., Solonen, A., Haario, H., Koskimies, S., 2009. Kinetics of
rst ed. John Wiley & Sons. neopentyl glycol esterication with different carboxylic acids. Ind. Eng. Chem. Res.
Dib, A., Makhlou, L., 2007. Mass transfer correlation of simultaneous removal by cemen- 48, 62376247.
tation of nickel and cobalt from sulphate industrial solution containing copper part 1: Vahteristo, K., Laari, A., Solonen, A., 2013. DielsAlder reaction kinetics for production of
onto rotating zinc electrode disc. Chem. Eng. J. 130, 3944. norbornene monomers: evaluation of parameters and model reliability using Markov
Diaz, M.A., Kelsall, G.H., Welham, H.J., 1993. Electrowinning coupled to gold leaching by chain Monte Carlo methods. Ind. Eng. Chem. Res. 52, 63576365.
electrogenerated chloride: I. Au(III) Au(I)/Au kinetics in aqueous Cl2/Cl electrolytes. von Bonsdorff, R., 2006. The Use of a Quartz Crystal Microbalance in Determining the Dis-
J. Electroanal. Chem. 361, 2538. solution Rate of Gold in Cupric Chloride Solution. (Lic. Thesis). Helsinki University of
Feng, D., van Deventer, J.S.J., 2010. Oxidative pre-treatment in thiosulphate leaching of Technology, Espoo, Finland.
sulphide gold ores. Int. J. Miner. Process. 94, 2834. von Bonsdorff, R., Aromaa, J., Forsn, O., 2005. Dissolution of gold in cupric chloride solu-
Frankenthal, R.P., Siconol, D.J., 1982. The anodic corrosion of gold in concentrated chlo- tions. Proceedings of EMC, Dresden, Germany, pp. 237245.
ride solution. J. Electrochem. Soc. 129, 11921196. von Bonsdorff, R., Aromaa, J., Forsn, O., 2007. The rate of gold dissolution in concentrated
Gallego, J.H., Castellano, C.E., Calandra, A.J., Arvia, A.J., 1975. The electrochemistry of gold cupric chloride solutions. Proceedings of the Sixth International Copper-Cobre Con-
in aqueous solutions containing chloride ions. J. Electroanal. Chem. 66, 207230. ference 4(2), Toronto, Ontario, Canada, pp. 121131.
Gnen, N., Krpe, E., Yilidirim, M.E., Selengil, U., 2007. Leaching and CIL processes in gold Xu, B., Yang, Y., Li, Q., Jiang, T., Liu, S., Li, G., 2016. The development of an environmentally
recovery from refractory ore with thiourea solutions. Min. Eng. 20, 559565. friendly leaching process of a high C, As, and Sb bearing sulde gold concentrate.
Hasab, M.G., Rashchi, F., Raygan, S., 2014. Chloride-hypochlorite leaching of gold and hy- Miner. Eng. 89, 138147.
drochloric acid washing in multi-stages for extraction of gold from a refractory con- Yazici, E.Y., Deveci, H., 2015. Cupric chloride leaching (HCl-CuCl2-NaCl) of metals from
centrate. Hydrometallurgy 142, 5659. waste printed circuits boards (WPCBs). Int. J. Miner. Process. 134, 8996.
Hasab, M.G., Raygan, S., Rashchi, F., 2013. Chloride-hypochlorite leaching of gold from a Zhukov, V.V., Laari, A., Koiranen, T., 2015. Kinetic modeling of high-pressure pyrite oxida-
mechanically activated refractory sulde concentrate. Hydrometallurgy 138, 5964. tion with parameter estimation and reliability analysis using the Markov chain Monte
Haario, H., Saksman, E., Tamminen, J., 2001. An adaptive Metropolis algorithm. Bernoulli Carlo method. Ind. Eng. Chem. Res. 54, 99209930.
7, 223242.