Chalcopyrite dissolution rate laws

Applied Geochemistry 25 (2010) 972–983

Contents lists available at ScienceDirect

Applied Geochemistry journal homepage: www.elsevier.com/locate/apgeochem

Chalcopyrite dissolution rate laws Bryn E. Kimball a,*, J. Donald Rimstidt b, Susan L. Brantley a a b

Department of Geosciences, Pennsylvania State University, University Park, PA 16802, United States Department of Geosciences, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, United States

a r t i c l e

i n f o

a b s t r a c t

Article history: Received 9 November 2009 Accepted 30 March 2010 Available online 11 April 2010

Meta-analysis of 173 rate measurements from 21 publications was used to develop rate laws for chalcopyrite dissolution under environmentally relevant conditions. Multiple linear regression analysis of 28 data for nonoxidative chalcopyrite dissolution in the presence of O2 and Cl produced the following rate law:

Editorial handling by R. Seal

r ¼ 101:52 e28200=RT ½Hþ 

1:68

Here, r is the rate of chalcopyrite dissolution in units of mol m2 s1 where the surface area is expressed on a geometric basis. Multiple linear regression analysis of 36 data for chalcopyrite dissolution caused by reaction with Fe(III) in the presence and absence of O2 and Cl produced the following rate law: 0:8

r ¼ 101:88 e48100=RT ½Hþ  ½FeðIIIÞ

0:42

Some data were excluded from these rate law models because they were inconsistent with the overall dataset and/or were relatively unconstrained. There were no published rate data that could be clearly identified as representing chalcopyrite dissolution caused by O2 oxidation alone. Although there are numerous reports that suggest that chalcopyrite dissolution rates are increased by the presence of Cl in solution, the regression models documented that the effect of Cl on dissolution was insignificant for this dataset. The rate laws developed in this work are most appropriate for characterizing chalcopyrite dissolution at low pH (63), and will ultimately allow better modeling of acid, sulfate, Fe, and Cu release to the environment. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction Chalcopyrite (CuFeS2) is the most common Cu-bearing mineral on earth (Nesse, 2000). This mineral often precipitates from hydrothermal fluids in association with other sulfides, but also occurs as an accessory mineral in mafic igneous rocks following exsolution within cooling rock (Sugaki et al., 1975). Solutions reacting with chalcopyrite tend to evolve towards low pH and high SO4, Fe, and Cu concentrations, thereby contributing to the problem of acid rock drainage (ARD). There are two situations where the release rate of Cu from chalcopyrite is particularly important. First, chalcopyrite is often the most abundant Cu-bearing mineral in mine wastes and Cu released by its dissolution is responsible for aquatic toxicity in nearby receiving waters (Schumbauer-Berigan et al., 1993) and significant phytotoxicity in soils (Alva et al., 2000). In this case, researchers often seek to minimize the natural rate of chalcopyrite dissolution, so experiments are designed to closely match the pH, Eh, and temperature conditions typical of surface weathering environments. Second, chalcopyrite is often the most abundant Cu-bearing mineral in sulfide ores, and in this case, * Corresponding author. Fax: +1 703 648 6252. E-mail address: [email protected] (B.E. Kimball). 0883-2927/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.apgeochem.2010.03.010

researchers seek to maximize its dissolution rate during hydrometallurgical recovery of Cu (Bartlett, 1997). Chalcopyrite dissolution is maximized in the presence of high oxidant concentrations (e.g., Palmer et al., 1981; Lin et al., 1986), strong oxidants (e.g., Antonijevic´ et al., 2004; Aydogan et al., 2006), and high temperatures (e.g., Yu et al., 1973; Habashi and Toor, 1979; Hackl et al., 1995; Qui et al., 2007). In experiments under such conditions, much of the mineral can be reacted away and grain surfaces may develop product layers (e.g., elemental sulfur [S0] or jarosite [KFe3(SO4)2(OH)6]). In those cases, the thickness and density of product layers can act as a barrier that causes the rates to be controlled by the rate of diffusive transport of reactants to or products away from the chalcopyrite surface (Berner, 1978). For example, Majima et al. (1985) leached chalcopyrite in acidic Fe2(SO4)3 and FeCl3 solutions and found that S0 formed on chalcopyrite grain surfaces in both experiments, but the S0 formed in Fe2(SO4)3 was denser than that formed in FeCl3. This likely contributed to the slower oxidation rate determined for Fe2(SO4)3 leaching compared to that for FeCl3 leaching (Majima et al., 1985). In this study, the goal was to develop empirical rate laws for chalcopyrite dissolution that can be applied over a range of environmentally relevant conditions. Published kinetic data were compiled in order to conduct a meta-analysis of the most important

973

B.E. Kimball et al. / Applied Geochemistry 25 (2010) 972–983

variables that contribute to chalcopyrite dissolution using multiple linear regression. To the authors’ knowledge, collection and metaanalysis of chalcopyrite dissolution rate data have not been attempted previously. While published chalcopyrite dissolution rate laws exist, they stem from individual studies. The authors anticipate that rate laws developed from multiple studies will be more robust and could be more useful for predicting acid, SO4, Fe, and Cu release to the environment. Importantly, this compilation revealed the conditions where rate data are lacking, and where future experimentation should focus. In order to predict the rate of chemical reaction at the chalcopyrite surface, and to reduce the potential for including rates affected by surface product layers, data from experiments conducted at high temperatures and oxidant concentrations were not included. Specifically, only data from experiments conducted at T < 100 °C, P O2 6 1:7 atm, and [Fe(III)] 6 1 M were considered. These limits are not arbitrary, but instead reflect the available data within the range of conditions for Earth surface environments. The kinetic data used to derive empirical rate laws were critically selected in order to maintain statistical validity and chemical reality. Based on the available data, the rate laws developed in this work will be most appropriate for characterizing chalcopyrite dissolution at low pH (63). Although microbial bioleaching of sulfides is important, this study was restricted to abiotic experiments because of the limited number of quantitative studies of microbiological rates. 2. Background 2.1. Chalcopyrite dissolution reactions The difficulty in understanding and analyzing published rate data is in the determination of which of several possible reactions is responsible for the observed rate. Data compilation and analysis were limited to consider the three reactions that are most likely to be important at ambient conditions. A relatively simple reaction that can liberate Cu from chalcopyrite is nonoxidative dissolution:

pected to depend only upon the chalcopyrite surface area and the solution pH. In the environment, dissolved oxygen (DO) is a common electron acceptor so chalcopyrite is expected to oxidize and release Cu by a reaction that involves DO: 2 2þ CuFeS2ðsÞ þ 4O2ðaqÞ ¼ Cu2þ ðaqÞ þ FeðaqÞ þ 2SO4ðaqÞ

ð3Þ

Because DO concentrations are relatively low (<10 mg L1) in ARD and heap-leach liquor, and because the reduction of O2 is a complex multi-step reaction, the rate of this reaction is expected to be relatively slow. Furthermore, based on the stoichiometry of the reaction, little or no pH dependence of the rate is expected, so the rate should depend mostly on the chalcopyrite surface area and the DO concentration. Usually, the most important reaction in ARD and heap leaching is the oxidation of chalcopyrite by Fe(III): 2þ 2 2þ þ CuFeS2ðsÞ þ 16Fe3þ ðaqÞ þ 8H2 O ¼ CuðaqÞ þ 17FeðaqÞ þ 2SO4ðaqÞ þ 16HðaqÞ

ð4Þ Because Fe(III) is relatively soluble in the low pH solutions that are characteristic of ARD and heap-leach liquor, and because it is rapidly regenerated from Fe(II) oxidation by microorganisms at low pH (Schwertmann and Fitzpatrick, 1992; Nordstrom and Southam, 1997), dissolved Fe(III) concentrations can be high in these settings. In addition, like pyrite oxidation, the rate-determining step for reaction (4) is likely the simple and fast transfer of one electron from the chalcopyrite to an adsorbed Fe(III) (Rimstidt and Vaughan, 2003). As a result, reaction (4) is expected to dominate chalcopyrite dissolution in most low pH settings. Based on the stoichiometry of the reaction, it is expected that the rate law contains a term for Fe(III) concentration. Because Fe(III) (oxyhydr)oxide solubility, which is very pH dependent, tends to control Fe(III) concentrations, the rate of reaction (4) is expected to be relatively high at low pH and to decrease rapidly with increasing pH. 2.2. Chemical reactors

2þ CuFeS2ðsÞ þ 4HþðaqÞ ¼ Cu2þ ðaqÞ þ FeðaqÞ þ 2H2 SðaqÞ

ð1Þ

The term ‘‘nonoxidative” dissolution is used to describe this reaction because the valence state of S does not change and there is no net transfer of electrons from the chalcopyrite to solution species. The valences of Cu and Fe in chalcopyrite are +1 and +3, respectively (Boekema et al., 2004; Goh et al., 2006; Pearce et al., 2006; Wincott and Vaughan, 2006). Because sulfide surfaces readily oxidize upon exposure to air and water (Yin et al., 1995; Rosso and Vaughan, 2006), however, Cu on the chalcopyrite surface is likely to be a mixture of Cu(I) and Cu(II). O’Malley and Liddell (1986) measured the oxidation state of Cu during chalcopyrite dissolution in batch experiments and found that initially, the released Cu was predominantly Cu(II). After about 35% of the total Cu had dissolved, however, the reaction rate increased and nearly all released Cu was Cu(I). This experiment is consistent with release of Cu(II) from oxidized chalcopyrite grain surfaces, followed by release of Cu(I) from unoxidized chalcopyrite surfaces after a significant extent of reaction. Regardless of the valences of Cu and Fe in the mineral, oxidation of Cu(I) by Fe(III) in solution is rapid under acidic conditions (Orth and Liddell, 1990): 2þ 2þ CuþðaqÞ þ Fe3þ ðaqÞ ¼ CuðaqÞ þ FeðaqÞ

ð2Þ

so we expect Cu(II) and Fe(II) to be the dominant aqueous products even if Cu(I) and Fe(III) are the species that are initially liberated from the chalcopyrite. Because reaction (1) involves only the interaction of H+ ions and chalcopyrite, the reaction rate would be ex-

Many rates of chalcopyrite dissolution have been reported for experiments under controlled temperature, pH, and solution composition. The most common experimental set-up is the batch reactor (BR), which is a flask containing mineral and solution, open or closed to the atmosphere, that is stirred or agitated without flow (Brantley, 2003; Lengke et al., 2009). In most cases, the mineral grain size is prepared by mechanical grinding with or without sieving. The reaction is monitored by measuring the change in product concentrations over time. Care must be taken to correct for sample removal over time, if applicable, and to account for possible backreactions such as precipitation of secondary phases (Brantley, 2003). The second most common reactor used in this compilation was the mixed flow-through reactor (MFR). In a MFR, a mineral sample is placed within a reactor of a given volume, then solution is pumped through the reactor at a known rate while the solution within the reactor is stirred. One study utilized a flow reactor (FR) without any mixing; it was analyzed as if it were a MFR. Determinations of rates from well-stirred flow-through reactors are calculated by comparing the inlet concentration (Ci) to the steadystate outlet concentration (Co) of a component of the mineral sample (Brantley and Chen, 1995):

r¼

Q ðC o  C i Þ v i Am

ð5Þ 2

1

Here r is the rate of dissolution (mol m s ), Q is the flow rate (L s1), vi is the stoichiometric coefficient of component i in the

974

B.E. Kimball et al. / Applied Geochemistry 25 (2010) 972–983

mineral, A is the specific surface area of the mineral sample (m2 g1), and m is the initial mineral mass (g). The third type of reactor used by many investigators is the plugflow reactor (PFR). Plug-flow reactors are tubes or columns packed with mineral sample through which fluid is pumped or drained. In an ideal PFR, every packet of fluid is assumed to have had the same contact time with the mineral sample (Hill, 1977). The calculation of rate from the PFR is more complex than that for flow-through reactors (see Brantley, 2003), and the set-up is more time-consuming. However, the results can be more representative of water–rock interaction in the environment (Brantley and Conrad, 2008). 2.3. Data quality The dissolution data considered in this paper were produced by several different experiments, none of which controlled all of the variables that could possibly influence the rate. Several uncontrolled variables commonly exist in kinetic data sets. For example, the chalcopyrite samples were from different sources so they are expected to have different trace element contents. Likewise, some experiments used rapid stirring, and some were stirred less vigorously. Because these factors were not always quantified in the sources, there was no way to directly assess their effects on the reported rates. Prosser (1996) identified over 30 variables and phenomena that affect the leaching rate of sulfide minerals, each of which has an associated uncertainty. Among the variables included in this compilation, time and temperature are likely known with the most certainty. Solution pH or log [H+] is known with reasonable certainty at the beginning of experiments, but is subject to change over the course of the reaction. Few studies monitored changes in pH over time. Likewise, when Fe(III) was added as an oxidant, the initial [Fe(III)] was known with some certainty, but only one study observed changes in this variable with time. The PO2 was relatively constant compared to [Fe(III)] in oxidation experiments. In sulfide mineral dissolution experiments, the extent of reaction is commonly reported as a relative value by calculating the total moles of element i released to solution divided by the total moles of element i in the starting mineral (ai = molesi solution/molesi total). The reaction progress variable (RPV) most often used to describe chalcopyrite leaching is aCu. While the concentration of Cu, [Cu], and thus moles of Cu, can be measured with reasonable certainty, calculated aCu values can have large errors due to the uncertain composition of the solid. Sulfide mineral specimens are rarely pure, and commonly contain minor quantities of other sulfides or minerals (Prosser, 1996). Most studies report the bulk composition of the solids used in experiments based on mineral digestion and spectrometry or quantitative X-ray diffraction (XRD) analysis. When possible, the results from such techniques were used to estimate the total Cu in an experimental system in order to calculate aCu as accurately as possible. The calculated aCu values, and rates based on these values, are likely the most uncertain parameters included in this analysis. It is assumed that the uncertainty associated with aCu does not significantly affect the general trends observed when considering multiple studies simultaneously, and therefore does not detract from identifying the influence that certain variables have on dissolution rates. 3. Methods 3.1. Data conversion and analysis Chalcopyrite dissolution rate data were obtained from experiments where dissolution was identified as being surface-controlled based on criteria developed by Berner (1978): (1) no increase in

rates with increased stirring speeds and/or (2) activation energies that exceed 20 kJ mol1. Experiments where surface or diffusion control were not tested were also included in the compilation, whereas reactions identified as being diffusion controlled were avoided. In total, 173 rates for low-temperature chalcopyrite dissolution from 21 publications were compiled (Fig. 1; Table S1, Supplementary Material). The physiochemical conditions (i.e., pH, temperature, and solution composition), grain size and/or grain surface area, and reactor type for experiments included in this compilation are listed in Table 1. Although some experiments were conducted under anoxic conditions, and others were carried out under controlled atmospheres, many experiments were open to the atmosphere, so a P O2 ¼ 0:21 atm was assumed if not otherwise stated. Only the initial Fe(III) concentration in leach solutions was considered in the compilation, as most studies did not measure changes in Fe(III) during the reaction. Some data sets included initial pH, and others only the initial acid concentration. To be consistent, the following was assumed:

 pH ¼ log½Hþ 

ð6Þ

log½acid ¼ log½Hþ 

ð7Þ

The normality of acid was used to represent log [acid]. Hydrochloric acid and HNO3 were assumed to dissociate completely, and H2SO4 was assumed to dissociate to H+ and HSO 4 at pH < 2 and to 2H+ and SO2 4 at pH P 2. To normalize dissolution rates by surface area, available data were collected describing the initial specific surface area of the chalcopyrite grains. Only 7 of the 21 studies included in this compilation report initial specific surface area measured by BET sorption analysis (ABET), so the initial specific geometric surface area (Ageo) for all experiments was calculated and used to normalize rate calculations. The Ageo represents the sum of surface areas of different grain sizes for a given sample mass. This parameter was calculated with the following relationship:

Ageo ¼

6V m W m de

ð8Þ

where Vm is the molar volume of chalcopyrite (4.39  105 m3 mol1), Wm is the molar mass of chalcopyrite (183.5 g mol1), and de is the effective grain diameter of particles in a given sieved size fraction (m). Most studies reported a range of grain sizes obtained from sieving. Given the maximum grain size (dmax) and minimum grain size (dmin), de was approximated (Tester et al., 1994) by:

de ¼

dmax  dmin   ln ddmax min

ð9Þ

A few studies only reported dmax, so it was assumed that smaller particles were at least the size of colloidal material with a dmin of 1 nm in order to use Eq. (9). In the few cases where both ABET and Ageo were reported, it was observed that ABET > Ageo. Therefore, the geometric surface area-normalized rates are expected to be faster than BET-normalized rates. Several experiments were conducted on multi-mineral samples where chalcopyrite was a minor constituent. In this case, the specific geometric surface area of the whole sample was multiplied by the fraction of chalcopyrite in the sample in order to estimate the specific geometric surface area of chalcopyrite. All but one study reported Cu release from chalcopyrite over time. A few studies used Fe release from chalcopyrite as the RPV (Acero et al., 2007, 2009) or Fe(III) consumption as the RPV (Rimstidt et al., 1994). Acero et al. (2007, 2009) noted the incongruent dissolution of chalcopyrite (except at pH 1), and chose to use Fe release as the RPV because of the faster dissolution of Fe compared to Cu (the steady-state Fe:Cu in solution was as high as 3:1). While Cu

975

B.E. Kimball et al. / Applied Geochemistry 25 (2010) 972–983

Fe(III) originally present No Fe(III) originally present

a

log r

-4 -6

-6

-8

-8

-10

-10

-12

-12 0

b

-4

20

40

60

80

100

-1

0

1

2

Temperature (ºC)

log r

-4

c

-6

-8

-8

-10

-10

-12

-12 -4

-3

-2

-1

4

5

-1

0

1

d

-4

-6

-5

3

pH

0

1

-5

-4

-3

-2 -

log [Fe(III)] or log P

log [Cl ]

O2

Fig. 1. Reported and calculated rate measurements (log r) are plotted against temperature (a), pH (b), oxidant concentration (log [Fe(III)] or log P O2 ; (c), and Cl concentration (log [Cl]; (d). Experiments where Fe(III) was originally present in the leach solutions are symbolized by closed circles, and experiments where Fe(III) was not originally present are symbolized by open squares. Note that the rates vary over approximately eight orders of magnitude. All data are given in Table S1 (Supplementary Material).

may or may not be the most appropriate RPV for chalcopyrite dissolution, due to incongruent dissolution and impure starting minerals, it is the most common RPV used to monitor chalcopyrite dissolution. In order to maximize the number of studies and data points in the regression models, the Acero et al. (2007, 2009) reaction rates based on Fe release were replaced with rates based on their reported Cu release as the RPV. Like all previous studies, chalcopyrite dissolution rates were calculated based on Cu release assuming that every mole of Cu released equaled one mole of chalcopyrite dissolved. Chalcopyrite dissolution rates are presented in terms of mol m2 s1 (Table S1). Chalcopyrite dissolution rates from MFR experiments were calculated using Eq. (5) and the steady-state outlet Cu concentration. This method of calculating rates is referred to as the steady-state method (SSM) in Table 1. When chalcopyrite dissolution rates were not tabulated for experiments in BR, rate data were extracted from graphs using digitizing software (B. Tummers, Data Thief III v. 1.1). With these data, rates were calculated using both the standard initial rate method (IRM) (Rimstidt and Newcomb, 1993) and the shrinking particle model (SPM, see the recent summarized derivation in Kimball (2009)). Rates from BR experiments on multi-mineral samples were calculated with the IRM only. To calculate rates using the IRM (Rimstidt and Newcomb, 1993) Cu release (in moles) was plotted versus time (s) for the initial linear portion of the experiment, then the best-fit linear function was found:

moles Cu ¼ a1 þ b1 t

ð10Þ

where a1 and b1 are constants. The derivative of (10) is:

r0Cu ¼

dmoles Cu ¼ b1 dt

ð11Þ

where r0Cu is the apparent rate of production of Cu (in mol s1), which can be converted to the surface area-normalized rate of production (rCu; mol m2 s1):

rCu ¼

r 0Cu Ageo m

ð12Þ

Here Ageo is the specific geometric surface area of the reacting mineral (m2 g1), and m is the initial mineral mass (g). With the SPM, the dissolution rate was calculated from the shrinking particle rate constant, kp (s1), by assuming the reaction is pseudo-zeroth order. The SPM is most appropriate for experiments where the specific surface area and extent of reaction changed significantly over time. Among the studies included in this compilation, aCu values, or the extents of reaction, range from 4  104 to 1.0. The SPM was applied to all calculated aCu and it was found that for most experiments, rates calculated using the initial rate method and the shrinking particle model were within ± 0.5 log units. Unless otherwise stated (Table 1), for those rates that were calculated using both SPM and IRM, SPM rates are reported (Table S1).

976

Table 1 Identification, experimental conditions and design, and rate calculation method used for studies included in this compilation. Identifier – Reference

Reactor type

pH or [H+] (acid) (M)

Temperature (K)

PO2 (atm)

[Fe(III)]initial (M)

[Cl] (M)

Ageo (m2 g1)

ABET (m2 g1)

AbO4 – Abraitis et al. (2004) Ac07 – Acero et al. (2007) Ac09 – Acero et al. (2009) A108 – Al-Harahsheh et al. (2008) An04 – Antonijevic´ and Bogdanovic´ (2004) Ca05 – Cama and Acero (2005) Co09 – Córdoba et al. (2009) Do02 – Domènech et al. (2002)

BR MFR MFR BR

2.5 (HC1) 3 (HC1, H2SO4) 1, 2, 3(HC1, H2SO4) [H+] = 0.5 M (HCl)

298 298 298 343, 353, 363

0.21 0.005, 0.05, 0.21 0.21 0.21

0 0 0 0.5

0.1 0, 0.001 0, 0.01, 0.001 2

0.02 0.04 0.04 0.04, 1.4

– 0.71 0.71 –

PFR

0.5, 0.7, 1.0, 1.3, 2.0(H2SO4)

298

0.21

0

0

0.0001

–

FR BR FR

3 (HC1) 0.5, 1.0, 1.5, 2.0(H2SO4) 2.5–4.8 (HCl)

298 341 295

0.21 0, 0.21 0.00074, 0.0044, 0.09, 0.20

0 0.02 0

0.001 0 0.000072, 0.0024

1.2 0.06 0.008

Du81 – Dutrizac (1981) Fe75 – Ferreira and Burkin (1975) Ha95 – Havlík et al. (1995) HiOl – Hiroyoshi et al. (2001) Li76 – Linge (1976) Li86 – Lin et al. (1986) LuOO – Lu et al. (2000) Ma85 – Majima et al. (1985) Mu71 – Muñoz-Ribadeneira and Gomberg (1971) Om86 – O’Malley and Liddell (1986)

BR BR

[H+] = 0.3 M (H2SO4) 1 (H2SO4)

363 323, 338, 353, 368

0.21 0

0.1 0.01, 0.03

0 0

– 0.18 0.0007 0.01, 0.04, 0.06, 0.08, 0.12 0.011, 0.016, 0.023

BR BR, PFR BR BR BR BR BR

0.25, 0.5, 0.75 M (HCl) [H+] = 0.1 M (H, SO4) [H+] = 0.01 M (HNO3) 1.78 (H2SO4) [H+] = 0.8 M (H2SO4) [H+] = 0.2 M (HC1, H2SO4) [H+] = 0.1 M (H2SO4)

276.5, 293, 313, 323, 333, 343, 353 303 298 363 303 343 298

0 0 0 1.7 0.21 0 0.21

0.5, 1.0 0.25, 0.03 0.0002 0 0 1 0

1.5, 3.0 0 0 0 0 0, 3.2 0, 0.05, 0.10, 0.25, 0.5, 0.75, 1

0.006 0.01, 0.48 0.0003 0.0005 0.10, 0.36 – 0.02

– – – – – – –

BR

[H+] = 0.4. 3 M (HCl)

368

0

0.1

3.7

0.03

–

Pa81 – Palmer et al. (1981)

BR

[H+] = 0.2, 1 M (HCl)

–

MFR BR

1.86–1.93 (HCl) 2–3.5 (H2SO4, HNO3)

0.8, 1.2, 1.8, 2.8, 3.8, 4.06, 4.15, 4.3, 4.75, 5.5 0.013, 0.014 0

0.03, 0.04

Ri94 – Rimstidt et al. (1994) Sa06 – Salmon and Malmström (2006)

0.02, 0.05, 0.1, 0.2, 0.25 0.5 1.5  l05 0

0.0073 0.002, 0.0002

0.049 0.39, 0.01

0 0.21 0.21

Identifier – Reference

Grain size (lm)

Measured element release

Surface or diffusion controlled?

Reason

Rate calculation method

Data source

AbO4 – Abraitis et al. (2004) Ac07 – Acero et al. (2007)

45–150 10–100

Cu, Fe, S Cu, Fe, S

Unclear Surface

– Ea > 20 kJ mol1

SPM SSM

Graph Table

1

SSM

Table

Comments

Cu- and S-enriched phases observed on reacted surfaces

SPM

Graph

Unclear

Ea > 20 kJ mol Ea > 20 kJ mol1; no increase In rate with increased Stirring speed –

SSM

Graph

Mixed ore sample; 1.7 wt.% cpy

Cu, Fe, S, Zn, Pb

unclear

–

Reported

table

<70

Cu, Fe

Unclear

–

SPM = IRM

Graph

Do02 – Domènech et al. (2002)

Med. = 12

Cu, Fe, S, Zn, Pb, As, Sb, Tl

Unclear

–

Reported

Graph, table

Du81 – Dutrizac (1981)

10–152

Cu

Surface

Ea > 20 kJ mol1

SPM = IRM

Graph

Pyritic sludge (0.6 wt.% cpy) and mineral separate S0 and jarosite observed on reacted cpy surfaces Pyritic sludge; 0.6 wt.% cpy Fe(II) was initially present in some leach solutions; jarosite was Observed on reacted cpy surfaces

Surface Surface

1

SPM = IRM SPM = IRM

Graph Graph

Ac09 – Acero et al. (2009)

10–100

Cu, Fe, S

Surface

A108 – Al-Harahsheh et al. (2008)

<25, < 38

Cu, SO4

Surface

An04 – Antonijevic´ and Bogdanovic´ (2004) Ca05 – Cama and Acero (2005)

<5000

Cu, Fe, Al, Si

10–100

Co09 – Córdoba et al. (2009)

Fe75 – Ferreira and Burkin (1975) Ha95 – Havlík et al. (1995)

53–153 200–315

Cu Cu

Ea > 20 kJ mol Ea > 20 kJ mol1

S0 was observed on reacted cpy surfaces

B.E. Kimball et al. / Applied Geochemistry 25 (2010) 972–983

348, 353, 355.5, 359.5, 363, 365.5, 369 313, 333 296

– –

B.E. Kimball et al. / Applied Geochemistry 25 (2010) 972–983

977

Steady-state [Cu] was not given Mill tailings; 4, 0.8 wt.% cpy

log r ¼ log A 

Graph Graph Table Table BR = batch reactor, MFR = mixed flow reactor, PFR = plug-flow reactor. SPM = shrinking particle model, IRM = initial rate method, SSM = steady-state method, dashes indicate where measurements were not given.

Cu Cu Fe Cu 45–53 37–53 152–251 Med. = 29, 63

ð13Þ

Here A represents the pre-exponential in the Arrhenius rate law, Ea is the activation energy (J mol1), R is the gas constant (J mol1 K1), T is temperature (Kelvin), mi represents the concentration or partial pressure of appropriately chosen reactants in molarity (i = H+, Fe(III), and Cl) or atmospheres (i = O2(g)), respectively, and ni is the rate order with respect to that reactant. Eq. (13) can be log transformed to:

SPM = IRM SPM = IRM Reported Reported – Ea > 20 kJ mol1 Ea > 20 kJ mol1 –

Graph Table Reported SPM = IRM Surface Unclear Cu Cu Not reported 74–105

Ma85 – Majima et al. (1985) Mu71 – Muñoz-Ribadeneira and Gomberg (1971) Om86-O’Malley and Liddell (1986) Pa81 – Palmer et al. (1981) Ri94 – Rimstidt et al. (1994) Sa06 – Salmon and Malmström (2006)

Surface Cu Med. = 4, 15 LuOO – Lu et al. (2000)

Unclear Surface Surface Unclear

S0 was observed on reacted cpy surfaces Graph SPM = IRM

Mixed ore sample; 2 % cpy Graph Graph Unclear Unclear 3.9–4.6xlO 53–75 Li76 – Linge (1976) Li86 – Lin et al. (1986)

5

Cu <75

Cu Cu, SO4

  Y ni mi r ¼ A eEa =RT

HiOl – Hiroyoshi et al. (2001)

Unclear

– – Ea > 20 kJ mol1; no increase In rate with increased Stirring speed Ea > 20 kJ mol1 –

SPM = IRM (BR); SPM (PFR) SPM = IRM IRM

Graph

Fe(II) and Cu(II) were initially Present in some leach solutions

Dissolution rates were approximated by an empirical dissolution rate law of the form:

  X Ea 1 þ ni ðlog mi Þ 2:303R T

ð14Þ

Values for A, Ea, and ni were found with multiple linear regression models using JMPÒ (SAS v. 8.0) software. Multiple regression involves fitting a dependent variable with a linear combination of independent variables using least squares analysis. Log [H+], log P O2 , log [Fe(III)], log [Cl] and 1/T were considered as the regressors (or independent variables) and log r as the response (or dependent variable). Regressors had a significant effect on the response when the significance level, or p-value, was <0.0001. When the p-value for a certain regressor was >0.0001, it was excluded from the regression. Thus, the values for A, Ea, and ni were determined iteratively. Because the reported and calculated chalcopyrite dissolution rates vary over approximately eight orders of magnitude and because obvious outliers exist, the data used in multiple regression fitting were critically filtered. The first filter was based on the solid samples used in leach experiments. Those studies conducted on multi-mineral samples where chalcopyrite was a minor constituent (see Table 1) were excluded. The remaining studies used chalcopyrite mineral separates, where chalcopyrite was the only or major mineral present. The second filter relates to the observation of incongruent chalcopyrite dissolution. Comparison of chalcopyrite dissolution rates based on the same RPV may reduce any biases introduced by incongruent dissolution. Thus, the second filter was to include only those experiments where Cu was the RPV. Only one study (Rimstidt et al., 1994) did not provide Cu release data, so it was excluded. As will be seen below, chalcopyrite can dissolve oxidatively and nonoxidatively, depending upon the conditions. Based on previous experiments with chalcopyrite and other sulfides, some combination of pH, P O2 , [Fe(III)], [Cl], and temperature will act as rate-controlling variables depending upon which reaction is fastest at a given set of conditions. An attempt was made to identify which reaction dominated at each set of conditions, and include the data that reflected the given reaction type and conditions in the regression. Using the mathematical approach suggested by Bandstra and Brantley (2008) for fitting kinetic data, it was inferred that when inclusion of specific datasets in the regression revealed clear outliers and/or caused violations of statistical validity, those data were not consistent with the overall dataset. As a result, they were excluded from the regression model. Specific datasets were also excluded from regression models when their inclusion caused unrealistic chemical responses, such as a decreasing reaction rate with increasing temperature or increasing oxidant concentration. When possible, problems with the excluded data were identified in their respective sources. 4. Results 4.1. Nonoxidative chalcopyrite dissolution Seventy-four rate measurements were compiled for chalcopyrite dissolution in the presence of O2, with or without Cl present,

978

B.E. Kimball et al. / Applied Geochemistry 25 (2010) 972–983

of which 28 were used to determine the rate law (Fig. 2). Multiple linear regression yielded:

log r ¼ 1:52ð0:65Þ  1475ð226Þ=T þ 1:68ð0:06Þ log½Hþ 

ð15Þ

where the numbers in parentheses are 1 standard error of the regression coefficient. The R2 for this model is 0.99, and the empirical rate law is:

r ¼ 101:52 e28200=RT ½Hþ 1:68

ð16Þ 1

The activation energy for this reaction is 29 ± 4 kJ mol . A best-fit model was obtained by excluding the rates of Ac09, An04, Ca05, Co09, Do02, Li86, and Sa06 (these abbreviations refer to published studies as referenced in Table 1) from the regression. The solid samples used in the An04, Do02, Li86, Sa06, and one of the Ca05 experiments were multi-mineral samples, with chalcopyrite making up a small percentage of the composition. Thus, these experiments did not pass the first filter. The ore sample used in the An04 study contained other sulfides and oxides, with a chalcopyrite content of 1.7 wt.%. Additionally, the effective grain diameter (de) of the particles used in the An04 experiment was larger than that for all other experiments conducted in the presence of O2. The larger grain size in the An04 experiments may have also contributed to the slower rates, given that smaller particles tend to dissolve quicker (e.g., Dutrizac, 1981; Palmer et al., 1981). In addition to chalcopyrite, the solids used in the Do02 experiments contained other sulfides, quartz, gypsum and clays, with chalcopyrite making up 0.6 wt.%. The Li86 experiments were conducted on an ore sample that contained 2% chalcopyrite and pyrite, as well as quartz, feldspar, and biotite. Additionally, the Li86 experiments

differed from others in that the P O2 was 1.7 atm, whereas most of the other experiments were conducted under P O2 of 0.21 atm. The Sa06 rates were from experiments conducted on fine and coarse mill tailings with mixed mineralogy and 4 (fine) and 0.8 (coarse) wt.% chalcopyrite. One Ca05 rate measurement was from an experiment with pyritic sludge containing 0.6 wt.% chalcopyrite, as well as other sulfides, sulfates and silicate minerals. The remaining rate data that were not included were excluded for the following reasons. The Ac09 rates determined from experiments conducted at pH 3 are similar to those conducted under similar conditions in earlier experiments (Ac07). The Ac09 rates at pH 2 and 1, however, are slower than the rates determined by other researchers under similar conditions, by as much as three orders of magnitude. These data appear to follow a linear trend that diverges from the rest of the data. Because the statistical validity of the regression model was compromised when including these data, the data were excluded. Two of the Co09 rates are orders of magnitude slower than others at similar conditions. In these experiments, S0 and jarosite were detected on reacted chalcopyrite grains (Córdoba et al., 2009). Diffusive transport of O2 and/or H+ from solution to the chalcopyrite surface through these product layers could account for the slower rates. The included rates were nearly all based on experiments carried out in Cl-bearing solutions. For consistency, the Lu00 and Ac07 rates that were from experiments without Cl (three in total) were also excluded. In a separate analysis, all the rates for experiments conducted in the presence of O2 were regressed and it was found that temperature was the only variable that significantly affected the rate. The rate did not depend on PO2 , likely because most measurements were from experiments at P O2 of 0.21 atm. Over half of

Fig. 2. The whole-model leverage plot for the multiple linear regression of chalcopyrite dissolution in the presence of O2 and Cl (a). A leverage plot is a graphical display of how each data point contributes to the significance of a model (a) or variable within the model (b and c). The 45° line (or line of fit) is where the actual response and predicted response are equal, and the accompanying dashed lines represent the 5% and 95% confidence curves. The vertical distance from data points to the line of fit represents the residual error. The horizontal line represents the actual mean log r value. The distance from each data point to the actual mean represents what the residual would be if the variable (1/T in b and log [H+] in c) were absent from the model. The references correspond to those in Table 1.

B.E. Kimball et al. / Applied Geochemistry 25 (2010) 972–983

these rates were calculated for experiments conducted at 25 °C. At this temperature the rates varied over three orders of magnitude, suggesting that an additional variable (not determined in this analysis) affected the rates for chalcopyrite dissolution in the presence of O2. 4.2. Chalcopyrite oxidation by Fe(III) Ninety-nine rate measurements were compiled for chalcopyrite oxidation by Fe(III), of which 36 were used to determine the rate law (Fig. 3). Multiple linear regression yielded:

log r ¼ 1:88ð1:56Þ  2514ð548Þ=T þ 0:80ð0:08Þ log½Hþ  þ 0:42ð0:07Þ log½FeðIIIÞ

ð17Þ

2

The R for this model is 0.91, and the empirical rate law is: 1:88 48100=RT

r ¼ 10

e

þ 0:8

½H  ½FeðIIIÞ

0:42

ð18Þ

979

observed before in Fe2(SO4)3 solutions like those used in the Du81 study (e.g., Córdoba et al., 2008, 2009). Regardless, inclusion of the Du81 data caused the rate law to be independent of temperature. The Ha95 rates are generally faster than most others, despite observed S0 coatings on the chalcopyrite grains (Havlík et al., 1995). When the Ha95 rates are included, the rate law shows no dependence on temperature. The Hi01 data were excluded because one rate is much slower than the others, and some experiments included Fe(II) and Cu(II) in the leach solution, unlike other studies. The Li76 rates are much faster than all other rates, which likely resulted from leaching in HNO3 solutions. Nitrate might have acted as an additional oxidant. The Pa81 rates vary over an order of magnitude, and inclusion of these data in the regression model caused temperature to have no significant effect on the rate. 5. Discussion 5.1. Identifying the predominant reaction



Preliminary models showed that log [Cl ] was not a significant predictor of rate, so data for chalcopyrite oxidation by Fe(III) in the presence and absence of Cl were combined to obtain the rate law. According to this rate law, the activation energy for chalcopyrite oxidation by Fe(III) is 48 ± 10 kJ mol1. The An04 rates were excluded from this model because a mixed ore sample was used. The Ri94 rates were excluded because the RPV used in this study was disappearance of Fe(III), and chalcopyrite dissolution rates based on Cu release were unavailable. To obtain the best model fit, the Du81, Ha95, Hi01, Li76, and Pa81 (Table 1) rates were excluded from the regression. The following are considered as valid reasons for excluding those data. A few of the Du81 rates are much slower than all other rates. This may have resulted from formation of jarosite on chalcopyrite surfaces, which has been

The most difficult task in this study was identifying the reaction that was the main cause of the reported rate of chalcopyrite dissolution. Most experiments contained more than one species that was capable of reacting to cause chalcopyrite dissolution. For example, most of the reactions involving Fe(III) contained relatively high H+ activities and were carried out in contact with air so they contained some DO. The following guidelines were developed to help identify which of these possible reactants caused the predominant (fastest) reaction: (1) if an experiment containing a possible reactive species showed reaction rates that were comparable to an experiment without that reactive species, it was concluded that the possible reactive species did not contribute to the rate and (2) if the regression coefficient for a possible reactive

Fig. 3. The whole-model leverage plot for the multiple linear regression of chalcopyrite oxidation by Fe(III) in the presence or absence of O2 and Cl (a). See Fig. 2 caption for description of leverage plots and plot features. In this multiple regression, log r was regressed against 1/T (b), log [H+] (c), and log [Fe(III)] (d). The references correspond to those in Table 1.

980

B.E. Kimball et al. / Applied Geochemistry 25 (2010) 972–983

species did not appear to be significant in the regression models, it was concluded that the species did not contribute to the rate. As a result of these guidelines, it was determined that none of the experiments considered here gave rates that could be assigned to chalcopyrite dissolution resulting from oxidation by DO alone. 5.2. Nonoxidative chalcopyrite dissolution The rate law calculated for nonoxidative dissolution of chalcopyrite (Eqs. (15) and (16)) shows that the rate depends only on temperature and [H+] (Fig. 2). Even though all of the experiments were performed with solutions that contained DO, the regression model showed that the effect of increased P O2 on the rate was not significant, so this variable therefore does not appear in the rate law. Lin et al. (1986) showed that leaching rates increased when P O2 was raised from 1.7 to 30.6 atm at 90 °C, but at 25 °C, chalcopyrite dissolution rates were the same at PO2 of 0.21, 0.05, and 0.005 atm (Acero et al., 2007). Likewise, leaching of chalcopyrite-bearing pyritic sludge showed no change in dissolution rate under PO2 atmospheres ranging from 7.4  104 to 0.20 atm at 22 °C (Domènech et al., 2002). The positive dependence of the dissolution rate on pH (Fig. 2c) is consistent with nonoxidative dissolution (reaction (1)). 5.3. Chalcopyrite oxidation by Fe(III) Most of the rate measurements for chalcopyrite oxidation by Fe(III) were based on experiments conducted at relatively high [Fe(III)] and [H+] (Fig. 1b and c). The rate law calculated for chalcopyrite oxidation by Fe(III) (Eqs. (17) and (18)) indicates that the rate depends significantly upon temperature, [H+], and [Fe(III)] (Fig. 3). If the rates calculated by Eq. (18) represent the total reaction rate (rtotal), due to both nonoxidative dissolution and oxidation by Fe(III), and the rates calculated by Eq. (16) represent nonoxidative rates (rH+) only, the rate of oxidation by Fe(III) only (rFe(III)) can be estimated from the equation rFe(III) = rtotal  rH+ if rtotal represents the combined rH+ and rFe(III) rates. According to this calculation, at a given temperature, pH, and [Fe(III)], rFe(III) is essentially the same as rtotal because rH+ is two orders of magnitude slower than rtotal. Further work is needed to determine if rH+ is a negligible component of rtotal at higher pH values as well. X-ray photoelectron spectroscopy (XPS) and scanning electron microscopy (SEM) show that when chalcopyrite grains react with solutions containing Fe(III), several products can be observed. Reacted chalcopyrite grains may exhibit leach layers enriched in sulfate (a jarosite-like phase) (Parker et al., 2003; Klauber, 2008), phase) (Klauber et al., 2001; Parker et al., disulfide (a metal-S2 2 2003), and/or S0 (Rimstidt et al., 1994; Klauber et al., 2001; Parker et al., 2003; Klauber, 2008). Thus, in addition to reaction (4), chalcopyrite oxidation by Fe(III) presumably also occurs as follows: 2þ 0 2þ CuFeS2ðsÞ þ 4Fe3þ ðaqÞ ¼ CuðaqÞ þ 5FeðaqÞ þ 2SðsÞ

ð19Þ

Because S0 is observed to coat chalcopyrite surfaces, the dissolution reaction could eventually become limited by the rate of diffusion of reactants or products through the surface layer. Acero et al. (2007) did not observe such a passivating effect due to the formation of various S phases on chalcopyrite surfaces in longterm leach experiments, however. In this context, ‘‘passivation” refers to a decrease in chalcopyrite dissolution over time due to the formation of an armoring product layer (Klauber, 2008). In the experiments where S0 and/or jarosite were observed to have formed on reacted chalcopyrite grains, the experiments were designed to include Fe(III) in the initial leach solution (Co09, Du81, Ha95, Ma85). Most of these studies provided evidence for a surface-controlled reaction (Table 1). Which product will form, and

whether or not its formation will inhibit dissolution rates, depends more on the experimental conditions than on qualities inherent to chalcopyrite (Klauber, 2008). Based on the evidence that the reactions that occurred in the Co09, Du81, Ha95, Ma85 studies were surface-controlled, it was chosen to include these experiments in the regression models, despite the observation of product phases on reacted chalcopyrite grains. 5.4. Effect of Cl There are numerous reports of enhanced chalcopyrite dissolution rates with increasing Cl concentrations (e.g., Muñoz-Ribadeneira and Gomberg, 1971; Palmer et al., 1981; Majima et al., 1985; Lu et al., 2000; Al-Harahsheh et al., 2008). In this compilation, the studies that included Cl-bearing solutions show roughly constant chalcopyrite dissolution rates between about 105 and 102 M Cl, and increasing rates with increasing [Cl] (Fig. 1d). Improved leaching in Cl-bearing, oxygenated, acid solutions is believed to result from the formation of intermediate mixed-ligand (e.g., Cu(OH)Cl, Cu2(Cl(OH3)) complexes on mineral surfaces (Senanayake, 2009), which are conducive to ligand-promoted dissolution. While increased [Cl] in leaching solutions has been shown to increase chalcopyrite dissolution rates in individual studies, this variable was found to have an insignificant effect on the rate for the compiled data considered in this study. This may have resulted from using rate measurements that were from experiments conducted at pH 6 3. At low pH, nonoxidative dissolution may predominate over ligand-promoted dissolution. Additional experiments with chalcopyrite leaching by Cl-bearing solutions, conducted at higher pH and/or higher [Cl], may reveal more of a rate dependence on Cl. 5.5. Comparison of chalcopyrite and pyrite dissolution rates The chalcopyrite dissolution rate laws determined in this study can be compared to those for pyrite oxidation by O2 only and by Fe(III) in the presence of O2 (Williamson and Rimstidt, 1994) (Fig. 4a). At pH 6 4, pyrite oxidation by Fe(III) in the presence of O2 is four orders of magnitude faster than Fe(III)-oxidative chalcopyrite dissolution (Fig. 4a). During chalcopyrite and pyrite dissolution in the environment, Fe is released to solution and is cycled between the reduced and the oxidized states by microbial Fe(II)-oxidation and consumption of the Fe(III) oxidant during sulfide oxidation (e.g., reactions (4) and (19)). Given the characteristic presence of Fe(II)/Fe(III) in acidic leaching solutions near sulfide deposits, Fe(III)-oxidative sulfide dissolution is likely to be more important than sulfide oxidation by DO at low pH. This is reflected in the slower calculated rates for nonoxidative chalcopyrite dissolution where Fe(III) was not initially present in leach solutions relative to Fe(III)-oxidative chalcopyrite dissolution (Fig. 4a). With increasing pH, Fe(III) becomes less soluble, and tends to precipitate into various Fe(III) (oxyhydr)oxide minerals. As a result, sulfide oxidation by DO becomes more important at higher pH, as shown by the increasing pyrite oxidation by DO with increasing pH (Fig. 4a). Assuming the rate law for pyrite oxidation by Fe(III) applies to pH values beyond the experimental maximum of about 3, and that the activity of Fe(III) is controlled by ferrihydrite solubility [log Keq (25 °C) = 4.89, Drever, 1997], the rate of oxidation of pyrite by DO overtakes that for oxidation of pyrite by Fe(III) between pH 4 and 5 at 25 °C (Fig. 4a). The rate of oxidation of chalcopyrite by DO is likely to be faster at higher pH as well. Future work is needed that focuses on chalcopyrite dissolution in the presence of O2 at pH P 3. Such experiments would provide the data needed to determine the dependence of the rate of chalcopyrite dissolution on DO.

981

B.E. Kimball et al. / Applied Geochemistry 25 (2010) 972–983

0

-6

a

cpy = chalcopyrite py = pyrite py

cpy | H+ An04 Ca05 Do02 Sa06

-8

|F )

III

e(

-5

b

-7

y cp

log r

e(

log r

|F )

III y cp

-10

+

|H

py | O2

-9

-10

-11 -12

-15

-13 0

2

4

6

8

10

0

1

2

3

4

5

pH

pH

Fig. 4. A plot of log rate versus pH for chalcopyrite and pyrite dissolution (a). The rate laws used to estimate log rate for chalcopyrite are from this study, and the rate laws for estimating log rate for pyrite are from Williamson and Rimstidt (1994). The nonoxidative rate law from this study is also compared to chalcopyrite dissolution rates calculated from experiments where chalcopyrite made up a minor fraction of the solids (b). References in (b) correspond to those in Table 1. For the pyrite dissolution rate that depends on O2, a DO concentration of 0.3 mM (i.e., equilibrium with P O2 ¼ 0:21 atm) was assumed. For those rates that depend on Fe(III), the activity of Fe(III) was assumed to be controlled by ferrihydrite solubility using log Keq (25 °C) = 4.89 (Drever, 1997). A temperature of 25 °C was assumed for all rate calculations. Heavy solid lines cover the pH range over which the rate laws were calculated. The lighter dashed line for pyrite oxidation by Fe(III) (py|Fe(III)) extends beyond the experimental pH range in order to show where the rate is equal to that for pyrite oxidation by O2 (py|O2) at higher pH. Note that the rates for nonoxidative chalcopyrite dissolution (cpy|H+) and Fe(III)oxidative chalcopyrite dissolution (cpy|Fe(III)) are faster at lower pH, but are likely overtaken by chalcopyrite oxidation by O2 at higher pH. The nonoxidative rate law predicts experimental chalcopyrite dissolution rates fairly well between pH 1 and 2.5 (b).

The activation energy estimates for nonoxidative chalcopyrite dissolution (29 ± 4 kJ mol1) and Fe(III)-oxidative chalcopyrite dissolution (48 ± 10 kJ mol1), which were determined from refined datasets, are within the range of activation energies previously reported for chalcopyrite dissolution (15–95 kJ mol1; Table 2). The wide range of reported activation energies (Table 2) likely results from differences in experimental design, leach solutions, preparation of starting minerals, and starting mineral compositions among the different studies. The chalcopyrite dissolution rates included in this compilation come from experiments conducted at temperatures ranging from 3.5 to 95 °C, with most experiments conducted at 25 °C or between 65 and 95 °C (Fig. 1a). The localized distribution of the data over the given temperature range likely contributes to the relatively large uncertainty associated with the activation energy estimates. 5.6. Application of rate laws Comparing the rate laws developed in this work to chalcopyrite dissolution rates determined from laboratory experiments with multi-mineral samples collected from the field is a step towards testing the applicability of these rate laws to predict chalcopyrite dissolution in the environment. The studies included in this compilation that used multi-mineral samples to determine chalcopyrite dissolution rates were not included in the multiple regression models. These studies did not include Fe(III) in the original leach solution, so the nonoxidative rate law is most appropriate for comparison with these rates. Four of the studies (An04, Ca05, Do02, and Sa06) were conducted under PO2 of 0.21 atm, and one study (Li86) was conducted under P O2 of 1.7 atm. For consistency, only the An04, Ca05, Do02, and Sa06 rates are compared to predicted nonoxidative rates. The nonoxidative rate law predicts chalcopyrite dissolution rates produced from these studies fairly well between pH 1 and 2.5 (Fig. 4b). Below pH 1, the empirical rate law overestimates experimental rates, and above pH 2.5, the rate law underestimates experimental rates. Above pH 2.5, the rate of chalcopyrite oxidation by DO is likely faster than the nonoxidative rate (see Section 5.5), hence the faster experimental rates between

pH 2.5 and 4.8 are expected given that O2 was present (Fig. 4b). Another step towards testing the applicability of the chalcopyrite dissolution rate laws would be to incorporate them into a geochemical reaction model of a given environment. Ideally, the rate laws would provide accurate estimates of the dissolved Cu concentrations measured in environmental samples. The rate law could then be used to predict future Cu release from weathering ore deposits that contain chalcopyrite. 5.7. Future work This compilation of chalcopyrite dissolution rates reveals that data gaps exist within the range of environmentally relevant conditions. For example, when calculated rate measurements (log r) are plotted against log [Fe(III)] (Fig. 1c), it is clear that no data exist between about 104 and 102 M Fe(III), and most experiments contained relatively high Fe(III) concentrations. Dissolution experiments were mostly conducted at pH 6 3 (Fig. 1b); thus, the rate laws derived in this work will apply most appropriately to acidic conditions. Future experiments should focus on chalcopyrite dissolution at higher pH in order to generate the kinetic data needed to derive rate laws for dissolution under such conditions. This work would benefit those attempting to model chalcopyrite dissolution in environments where pH values are greater than 3, where ARD has been diluted or neutralized by basic host rocks. Given that [Fe(III)] tends to decrease with increasing pH, such experiments may also provide better information on the rate dependence of chalcopyrite oxidation by Fe(III). There are several uncontrolled variables in this dataset, such as starting mineral composition and stirring speed. The practical result of these uncontrolled variables is displayed in the scatter of the data about the regression lines in the residual plots (Figs. 2 and 3). This scatter, which is typical for most dissolution rate data, amounts to ±0.5 log units (±300%), even though a scatter of <±20% would be expected based on propagation of the uncertainties in measured surface area, species concentration, and flow rate (or time) determinations. We are limited to this level of resolution of dissolution rates until experimental designs for dissolution

982

B.E. Kimball et al. / Applied Geochemistry 25 (2010) 972–983

Table 2 Reported activation energies for chalcopyrite dissolution in acidic, Cl- and/or SO4bearing solutions. Reference

Solution Description

T (°C)

Ea (kJ mol1)

Saxena and Mandre (1992) Aydogan et al. (2006)

Iron(III) chloride

30–90

15–28

Potassium dichromate in sulfuric acid Hydrochloric acid pH 3 Sulfuric acid pH 1

50–97

24

25–70 70–110

32 ± 5 30

Iron(III) chloride

25–75

38 ± 4

Iron(III) sulfate Hydrogen peroxide in sulfuric acid Iron(III) chloride + hydrochloric acid Iron(III) chloride Iron(III) chloride Iron(III) sulfate

30–95 30–80

38–63 39

65–110

38

30–100 30–100 30–100

44 ± 3 42–46 42

Mixed hydrochloric–sulfuric acid Iron(III) chloride Iron(III) chloride Hydrogen peroxide in sulfuric acid Iron(III) chloride + hydrochloric acid Hydrochloric acid pH 2 Iron(III) chloride + hydrochloric acid + NaCl Iron(III) chloride Iron(III) chloride

60–95

48

60–106 3.5–80 25–50

50 55 ± 5 60

80–100

62

75–97 60–90

64 68

55–85 45–80

69 69

Iron(III) chloride + hydrochloric acid Iron(III) chloride + hydrochloric acid + NaCl Iron(III) sulfate Iron(III) nitrate No Iron(III) originally present Iron(III) originally present

52–85

69

82.5–96

83

20–112 25–40 25–95 40–95

83.7 95 29 ± 4 48 ± 10

Acero et al. (2007) Le Houillier and Ghali (1982) Ammou-Chokroum et al. (1977) Dutrizac (1978) Adebayo et al. (2003) Ngoc et al. (1990) Dutrizac (1978) Dutrizac (1981) Ferreira and Burkin (1975) Lu et al. (2000) Ermilov et al. (1969) Havlík et al. (1995) Antonijevic´ et al. (2004) Palmer et al. (1981) Lin and Sohn (1987) Maurice and Hawk (1998) Majima et al. (1985) Havlík and Kammel (1995) Hirato et al. (1986) Palmer et al. (1981)

Munoz et al. (1979) Linge (1976) This study This study

mine a rate law for chalcopyrite oxidation by DO. Interestingly, it was found that [Cl] (ranging from 0 to 5.5 M) also does not affect the rate at pH 6 3, despite the conventional practice of leaching with acidic Cl solutions as a means of increasing metal extraction efficiency (e.g., Muñoz-Ribadeneira and Gomberg, 1971). Future experiments with chalcopyrite dissolution in oxic, Cl-bearing solutions at higher pH may reveal a rate dependence on Cl concentrations. The regression model for experiments that contained Fe(III) showed that the rates were influenced by temperature, [H+], and [Fe(III)]. Although these results give strong support for the reaction stoichiometry given in Eq. (4), the cause of the positive correlation between rate and H+ activity is at present unclear. Fig. 4a shows that predicted chalcopyrite dissolution rates are about four orders of magnitude slower than pyrite rates in ARD settings where Fe(III) is the predominant oxidant. This chalcopyrite dissolution rate synthesis and meta-analysis has provided insights into the mechanisms by which chalcopyrite dissolves in low pH, O2-, and/or Fe(III)-bearing solutions. The determined rate laws will benefit those concerned with predicting Cu concentrations during ARD and during hydrometallurgical leaching. This compilation has also highlighted gaps in the chalcopyrite dissolution rate database. Future experimentation with chalcopyrite dissolution at pH > 3 and at lower oxidant concentrations will help close these gaps and complete our understanding of chalcopyrite dissolution over a wider range of conditions. Acknowledgements Funding was provided by the Penn State Center for Environmental Kinetics Analysis (CEKA) (NSF Grant CHE-0431328). B.E.K. thanks J. Bandstra and J. Moore for helpful discussions. We thank R. Seal, M. McKibben, and M. Stanton for their thorough reviews and invaluable comments on the original manuscript. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.apgeochem.2010.03.010. References

experiments improve sufficiently to account for these often uncontrolled variables.

6. Summary Using published kinetic data for chalcopyrite dissolution, rates of reaction were compiled, then a meta-analysis was conducted in order to determine rate laws for chalcopyrite dissolution. Chalcopyrite dissolution rates were taken directly from the literature or were calculated using the initial rate method and the shrinking particle model for batch reactors, and the steady-state method (Eq. (5)) for mixed flow reactors. Multiple linear regression analysis was performed to determine the dependence of log rate on the independent variables 1/T, log [H+], log P O2 , log [Fe(III)], and log [Cl]. The regression models for experiments that did not originally contain Fe(III) in leach solutions showed that chalcopyrite dissolution at pH < 3.5 is influenced only by temperature and [H+]. From this it is concluded that the experiments considered in the model produced data for the nonoxidative dissolution reaction (1), even though most studies were carried out at O2 partial pressures of 0.21 atm. Future measurements of the rate of chalcopyrite dissolution in oxic solutions at higher pH are necessary in order to deter-

Abraitis, P.K., Pattrick, R.A.D., Kelsall, G.H., Vaughan, D.J., 2004. Acid leaching and dissolution of major sulphide ore minerals: process and galvanic effects in complex systems. Mineral. Mag. 68, 343–351. Acero, P., Cama, J., Ayora, C., 2007. Kinetics of chalcopyrite dissolution at pH 3. Eur. J. Mineral. 19, 173–182. Acero, P., Cama, J., Ayora, C., Asta, M.P., 2009. Chalcopyrite dissolution rate law from pH 1 to 3. Geologica Acta 7, 389–397. Adebayo, A., Ipinmoroti, K., Ajayi, O., 2003. Dissolution kinetics of chalcopyrite with hydrogen peroxide in sulphuric acid medium. Chem. Biochem. Eng. Quart. 17, 213–218. Al-Harahsheh, M., Kingman, S., Al-Harahsheh, A., 2008. Ferric chloride leaching of chalcopyrite: synergistic effect of CuCl2. Hydrometallurgy 91, 89–97. Alva, A.K., Huang, B., Paramasivam, S., 2000. Soil pH affects copper fractionation and phytotoxicity. Soil Sci. Soc. Am. J. 64, 955–962. Ammou-Chokroum, M., Cambazoglu, M., Steinmetz, D., 1977. Oxidation menagee de la chalcopyrite en solution acide: analyse cinetique des reactions. Bull. Soc. Fr. Mineral. Cristallophr. 100, 149–161. Antonijevic´, M.M., Bogdanovic´, G.D., 2004. Investigation of the leaching of chalcopyrite ore in acidic solutions. Hydrometallurgy 73, 245–256. Antonijevic´, M.M., Jankovic´, Z.D., Dimitrijevic´, M.D., 2004. Kintetics of chalcopyrite dissolution by hydrogen peroxide in sulphuric acid. Hydrometallurgy 71, 329– 334. Aydogan, S., Ucar, G., Canbazoglu, M., 2006. Dissolution kinetics of chalcopyrite in acidic potassium dichromate solution. Hydrometallurgy 81, 45–51. Bandstra, J.Z., Brantley, S.L., 2008. Data fitting techniques with applications to mineral dissolution kinetics. In: Brantley, S.L., Kubicki, J.D., White, A.F. (Eds.), Kinetics of Water–Rock Interaction. Springer, New York, pp. 211–257. Bartlett, R.W., 1997. Metal extraction from ores by heap leaching. Metall. Mater. Trans. B 28, 529–545. Berner, R.A., 1978. Rate control of mineral dissolution under earth surface conditions. Am. J. Sci. 278, 1235–1252.

B.E. Kimball et al. / Applied Geochemistry 25 (2010) 972–983 Boekema, C., Krupski, A.M., Varasteh, M., Parvin, K., van Til, F., van der Woude, F., Sawatzky, G.A., 2004. Cu and Fe valence states in CuFeS2. J. Magn. Magn. Mater. 272–276, 559–561. Brantley, S.L., 2003. Reaction kinetics of primary rock-forming minerals under ambient conditions. In: Drever, J.I. (Ed.), Treatise on Geochemistry, vol. 5 (Holland, H.D., Turekian, K.K., Exec. Eds.). Elsevier, pp. 73–117. Brantley, S.L., Chen, Y., 1995. Chemical weathering rates of pyroxenes and amphiboles. In: White, A.F., Brantley, S.L. (Eds.). Rev. Mineral. Geochem., 119– 172. Brantley, S.L., Conrad, C.F., 2008. Analysis of rates of geochemical reactions. In: Brantley, S.L., Kubicki, J.D., White, A.F. (Eds.), Kinetics of Water–Rock Interaction. Springer, New York, pp. 1–37. Cama, J., Acero, P., 2005. Dissolution of minor sulphides present in a pyritic sludge at pH 3 and 25 °C. Geologica Acta 3, 15–26. Córdoba, E.M., Muñoz, J.A., Blázquez, M.L., González, F., Ballester, A., 2008. Leaching of chalcopyrite with ferric ion. Part II. Effect of redox potential. Hydrometallurgy 93, 88–96. Córdoba, E.M., Muñoz, J.A., Blázquez, M.L., González, F., Ballester, A., 2009. Passivation of chalcopyrite during its chemical leaching with ferric ion at 68 °C. Miner. Eng. 22, 229–235. Domènech, C., de Pablo, J., Ayora, C., 2002. Oxidative dissolution of pyritic sludge from the Aznalcóllar mine (SW Spain). Chem. Geol. 190, 339–353. Drever, J.I., 1997. The Geochemistry of Natural Waters: Surface and Groundwater Environments. Prentice Hall, Upper Saddle River. Dutrizac, J.E., 1978. The kinetics of dissolution of chalcopyrite in ferric ion media. Metall. Trans. B 9, 431–439. Dutrizac, J.E., 1981. The dissolution of chalcopyrite in ferric sulfate and ferric chloride media. Metall. Trans. B 12, 371–378. Ermilov, V.V., Tkachenke, O.B., Tseft, A.L., 1969. Kinetics of the dissolution of chalcopyrite in ferric chloride, vol. 30. Trudy Instituta Metallov Obogashch. pp. 3–14. Ferreira, R.C.H., Burkin, A.R., 1975. Acid leaching of chalcopyrite. In: Burkin, A.R. (Ed.), Leaching and Reduction in Hydrometallurgy. Institution of Mining and Metallurgy, pp. 54–66. Goh, S.W., Buckley, A.N., Lamb, R.N., Rosenberg, R.A., Moran, D., 2006. The oxidation states of copper and iron in mineral sulfides, and the oxides formed on initial exposure of chalcopyrite and bornite to air. Geochim. Cosmochim. Acta 70, 2210–2228. Habashi, F., Toor, T., 1979. Aqueous oxidation of chalcopyrite in hydrochloric acid. Metall. Trans. B 10, 49–56. Hackl, R.P., Dreisinger, D.B., Peters, E., King, J.A., 1995. Passivation of chalcopyrite during oxidative leaching in sulfate media. Hydrometallurgy 39, 25–48. Havlík, T., Kammel, R., 1995. Leaching of chalcopyrite with acidified ferric chloride and carbon–tetrachloride addition. Miner. Eng. 8, 1125–1134. Havlík, T., Škrobian, M., Balázˇ, P., Kammel, R., 1995. Leaching of chalcopyrite concentrate with ferric chloride. Int. J. Miner. Process. 43, 61–72. Hill, C.G., 1977. An Introduction to Chemical Engineering Kinetics and Reactor Design. John Wiley & Sons, Inc., New York. Hirato, T., Kinoshita, M., Awakura, Y., Majima, H., 1986. The leaching of chalcopyrite with ferric-chloride. Metall. Trans. B 17, 19–28. Hiroyoshi, N., Miki, H., Hirajima, T., Tsunekawa, M., 2001. Enhancement of chalcopyrite leaching by ferrous ions in acidic ferric sulfate solutions. Hydrometallurgy 60, 185–197. Kimball, B.E., 2009. Biogeochemical cycling of copper in acid mine drainage. In: Geosciences. Pennsylvania State University, University Park. Klauber, C., 2008. A critical review of the surface chemistry of acidic ferric sulphate dissolution of chalcopyrite with regards to hindered dissolution. Int. J. Miner. Process. 86, 1–17. Klauber, C., Parker, A., van Bronswijk, W., Watling, H., 2001. Sulphur speciation of leached chalcopyrite surfaces as determined by X-ray photoelectron spectroscopy. Int. J. Miner. Process. 62, 65–94. Le Houillier, R., Ghali, E., 1982. Contribution to the study of the leaching of chalcopyrite. Hydrometallurgy 9, 169–194. Lengke, M.F., Sanpawanitchakit, C., Tempel, R.N., 2009. The oxidation and dissolution of arsenic-bearing sulfides. Can. Mineral. 47, 593–613. Lin, H., Sohn, H., 1987. Mixed-control kinetics of oxygen leaching of chalcopyrite and pyrite from porous primary ore materials. Metall. Trans. B 18, 497–503. Lin, H.K., Sohn, H.Y., Wadsworth, M.E., 1986. The kinetics of leaching of chalcopyrite and pyrite grains in primary copper ore by dissolved oxygen. In: Bautista, R.G., Wesely, R.J., Warren, G.W. (Eds.), Hydrometallurgical Reactor Design and Kinetics. The Metallurgical Society, Inc., Warrendale, PA, pp. 149–168. Linge, H.G., 1976. A study of chalcopyrite dissolution in acidic ferric nitrate by potentiometric titration. Hydrometallurgy 2, 51–64. Lu, Z.Y., Jeffrey, M.I., Lawson, F., 2000. The effect of chloride ions on the dissolution of chalcopyrite in acidic solutions. Hydrometallurgy 56, 189–202. Majima, H., Awakura, Y., Hirato, T., Tanaka, T., 1985. The leaching of chalcopyrite in ferric chloride and ferric sulfate solutions. Can. Metall. Quart. 24, 283–291.

983

Maurice, D., Hawk, J.A., 1998. Ferric chloride leaching of mechanically activated chalcopyrite. Hydrometallurgy 49, 103–123. Munoz, P., Miller, J., Wadsworth, M., 1979. Reaction-mechanism for the acid ferric sulfate leaching of chalcopyrite. Metall. Trans. B 10, 149–158. Muñoz-Ribadeneira, F.J., Gomberg, H.J., 1971. Leaching of chalcopyrite (CuFeS2) with sodium chloride sulfuric acid solutions. Nucl. Technol. 11, 367– 371. Nesse, W.D., 2000. Introduction to Mineralogy. Oxford University Press, Oxford. Ngoc, N.V., Shamsuddin, M., Prasad, P.M., 1990. Oxidative leaching of an offgrade/ complex copper concentrate in chloride lixiviants. Metall. Trans. B 21, 611– 619. Nordstrom, D.K., Southam, G., 1997. Geomicrobiology of sulfide mineral oxidation. Rev. Mineral. Geochem. 35, 361–390. O’Malley, M.L., Liddell, K.C., 1986. A rate equation for the initial stage of the leaching of CuFeS2 by aqueous FeCl3, HCl and NaCl. In: Bautista, R.G., Wesely, R.J., Warren, G.W. (Eds.), Hydrometallurgical Reactor Design and Kinetics. The Metallurgical Society, Inc., Warrendale, PA, pp. 67–73. Orth, R.J., Liddell, K.C., 1990. Rate law and mechanism for the oxidation of copper(I) by iron(III) in hydrochloric acid solutions. Ind. Eng. Chem. Res. 29, 1178– 1183. Palmer, B.R., Nebo, C.O., Rau, M.F., Fuerstenau, M.C., 1981. The phenomena involved in the dissolution of chalcopyrite in chloride-bearing lixiviants. Metall. Trans. B 12, 595–601. Parker, A., Klauber, C., Kougianos, A., Watling, H., van Bronswijk, W., 2003. An X-ray photoelectron spectroscopy study of the mechanism of oxidative dissolution of chalcopyrite. Hydrometallurgy 71, 265–276. Pearce, C.I., Pattrick, R.A.D., Vaughan, D.J., Henderson, C.M.B., van der Laan, G., 2006. Copper oxidation state in chalcopyrite: mixed Cu d9 and d10 characteristics. Geochim. Cosmochim. Acta 70, 4635–4642. Prosser, A.P., 1996. Review of uncertainty in the collection and interpretation of leaching data. Hydrometallurgy 41, 119–153. Qui, T., Nie, G., Wang, J., Cui, L., 2007. Kinetic process of oxidative leaching of chalcopyrite under low oxygen pressure and low temperature. Trans. Nonferr. Metals Soc. China 17, 418–422. Rimstidt, J.D., Newcomb, W.D., 1993. Measurement and analysis of rate data: the rate of reaction of ferric iron with pyrite. Geochim. Cosmochim. Acta 57, 1919– 1934. Rimstidt, J.D., Vaughan, D.J., 2003. Pyrite oxidation: a state-of-the-art assessment of the reaction mechanism. Geochim. Cosmochim. Acta 67, 873–880. Rimstidt, J.D., Chermak, J.A., Gagen, P.M., 1994. Rates of reaction of galena, sphalerite, chalcopyrite, and arsenopyrite with Fe(III) in acidic solutions. In: Alpers, C.N., Blowes, D.W. (Eds), Environmental Geochemistry of Sulfide Oxidation. Am. Chem. Soc. Symp. Series 550, pp. 2–13. Rosso, K.M., Vaughan, D.J., 2006. Reactivity of sulfide mineral surfaces. Rev. Mineral. Geochem. 61, 557–607. Salmon, S.U., Malmström, M.E., 2006. Quantification of mineral dissolution rates and applicability of rate laws: laboratory studies of mill tailings. Appl. Geochem. 21, 269–288. Saxena, N.N., Mandre, N.R., 1992. Mixed control kinetics of copper dissolution for copper ore using ferric chloride. Hydrometallurgy 28, 111–117. Schumbauer-Berigan, M.K., Dierkes, J.R., Monson, P.D., Ankley, G.T., 1993. PHdependent toxicity of Cd, Cu, Ni, Pb, and Zn to Ceriodaphnia dubia, Pimephales promelas, Hayalella azteca and Lumbriculus variegatus. Environ. Toxicol. Chem. 12, 1261–1266. Schwertmann, U., Fitzpatrick, R., 1992. Iron minerals in surface environments. In: Skinner, H., Fitzpatrick, R. (Eds.), Biomineralization: Processes of Iron and Manganese: Modern and Ancient Environments. Catena Verlag, Cremlingen, Germany, pp. 7–30. Senanayake, G., 2009. A review of chloride assisted copper sulfide leaching by oxygenated sulfuric acid and mechanistic considerations. Hydrometallurgy 98, 21–32. Sugaki, A., Shima, H., Kitakaze, A., Harada, H., 1975. Isothermal phase relations in system Cu–Fe–S under hydrothermal conditions at 350 degrees C and 300 degrees C. Econ. Geol. 70, 806–823. Tester, J.W., Worley, G., Robinson, B.A., Grisby, C.O., Feerer, J.L., 1994. Correlating quartz dissolution kinetics in pure water from 25 to 625 °C. Geochim. Cosmochim. Acta 58, 2407–2420. Williamson, M.A., Rimstidt, J.D., 1994. The kinetics and electrochemical ratedetermining step of aqueous pyrite oxidation. Geochim. Cosmochim. Acta 58, 5443–5454. Wincott, P.L., Vaughan, D.J., 2006. Spectroscopic studies of sulfides. Rev. Mineral. Geochem. 61, 181–229. Yin, Q., Kelsall, G.H., Vaughan, D.J., England, K.E.R., 1995. Atmospheric and electrochemical oxidation of the surface of chalcopyrite (CuFeS2). Geochim. Cosmochim. Acta 59, 1091–1100. Yu, P.H., Hansen, C.K., Wadsworth, M.E., 1973. A kinetic study of the leaching of chalcopyrite at elevated temperatures. Metall. Trans. 4, 2137–2144.