Applied Surface Science 137 1999.

207223

The mechanism of copper activation of sphalerite

a,)
Andrea R. Gerson , Angela G. Lange a , Kathryn E. Prince b, Roger St.C. Smart a

a
Ian Wark Research Institute, Uniersity of South Australia, South Australia, 5095, Australia
b
Australian Nuclear Science and Technology Organisation, PMB1, Menai, New South Wales, 2234, Australia
Received 13 May 1998; accepted 26 August 1998

Abstract

On the basis of recent SIMS and XAFS measurements in conjunction with already published XPS results, a mechanism
for the adsorptionrabsorption of Cu onto sphalerite is proposed. Under conditions of high pH and high nominal surface
coverage of the sphalerite by the Cu, CuOH. 2 colloidal particles are observed on the sphalerite surfaces using SIMS. Under
other conditions, SIMS measurements have indicated that adsorption of the Cu is essentially uniform over the sphalerite
surface and is not related to low coordination sites on the surface of the sphalerite. Depth profiling of sphalerite surfaces
with Cu adsorbed under conditions that do not result in CuOH. 2 colloidal particles show that the Cu adsorbedrabsorbed on
the sphalerite surface is largely in the first few atomic layers. XAFS analysis of Cu activated sphalerite has indicated that the
Cu occupies a distorted trigonal planar geometry, coordinated to three S atoms, in both surface and bulk sites. In addition
Cu1s., absorption edges in XAFS show that both bulk and surface adsorbed copper have an oxidation state less than q1
with the surface Cu being slightly more oxidised than the bulk absorbed Cu. On the basis of the combined XPS, SIMS,
XAFS and solution studies, a model is proposed that, on surface adsorption of Cu, the surface ZnII. atoms are replaced by
CuII. atoms which are then reduced in situ to CuI.. This reduction is accompanied by the oxidation of the three
neighbouring S atoms to an oxidation state of approximately y1.5. On bulk absorption of Cu atoms into the sphalerite
lattice a distorted trigonal planar configuration is achieved through the breakage of a formerly tetrahedral ZnS bond. The
breakage of this bond results in a 3-fold coordinated Cu plus one S 3-fold coordinated to Zn atoms. The breakage of this
bond leads to a greater reduction of the Cu than on surface absorption and also oxidation of the 3-fold coordinated S atom to
an approximately y0.5 oxidation state. This model does not invoke any polysulfite or SS bonded species to explain the
higher binding energy components of the S2p. XPS spectra. q 1999 Elsevier Science B.V. All rights reserved.

Keywords: Copper activation; Sphalerite; Adsorption; Mineral flotation

1. Introduction than three decades we.g., w14xx. Much is understood

The activation mechanism in which sphalerite
ZnS. is conditioned with CuII. normally as nitrate
or sulphate., prior to adsorption of xanthate and
flotation, has been extensively studied over more
)
Corresponding author. Tel.: q61-8-8302-3044; Fax: q61-8-
8302-3683; E-mail: andrea.gerson@unisa.edu.au
concerning the adsorption of CuII. from solution
and the release of ZnII. to solution. The nature of
the copper species formed on the surface of spha-
lerite, however, continues to be the subject of some
controversy. In particular, the structural nature of the
reaction and incorporation of the Cu atoms into the
sphalerite surface is not well understood. This paper
proposes a mechanism, based on reliable literature

0169-4332r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 9 - 4 3 3 2 9 8 . 0 0 4 9 9 - 1
208 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223
and new X-ray absorption fine structure XAFS.
spectroscopy and secondary ion mass spectrometry
SIMS. results, for this reaction and incorporation.
The following results seem to be well established
from several reliable studies.
At low nominal surface coverages, i.e., less
than two monolayers, at both low e.g., 5.5. and high
e.g., 9. pH, the initial uptake of CuII. by the
sphalerite surface i.e., loss from solution. results in
a 1:1 release of ZnII. w2,3,5,6x implying an ion
exchange mechanism.
At low pH e.g., 5.5. and higher nominal
surface coverages, CuII. uptake occurs in two stages:
the first rapid 15 min. w5,6x and the second slow
w7x eventually becoming apparently limited by diffu-
sion through the surface layers w46x. Gaudin w1x
showed, however, that copper uptake continued be-
yond 63 h, albeit very slowly, implying continuing
adsorption, reaction and exchange.
At high pH e.g., 8.5. and higher nominal
surface coverages, CuII. is adsorbed as colloidal
CuOH. 2 species predicted to precipitate near pH
7.07.5 at 10y4 M. with subsequent partial only.
reaction with ZnII., i.e., exchange is not 1:1 w2,3x.
Electrochemical studies show that a conducting
layer is produced with reduction currents at poten-
tials - 200 mV SHE. in the initial 3050 s of
activation. This is followed by a second stage in
which the electrochemical characteristics are re-
ported to resemble those of copper sulphides. Copper
uptake is strongly dependent on electrochemical po-
tential; more is adsorbed at potentials below the rest
potential i.e., 200210 mV SHE. and less above
this potential w8x.
X-ray photoelectron spectroscopy XPS. identi-
fies copper in the sphalerite surface layers as CuI.
w912x in all conditions except at relatively high
concentrations i.e., ) 10y4 M. and high pH ) 8.
when more than 10 equivalent monolayers of CuII.
were added to solution w11x. In the latter case, XPS
measures both CuI. and CuII. strongly associated
with adsorbed hydroxide species.
This observation is not due to photoreduction
of CuII. to CuI. in the XPS measurement; this
effect is only significant for CuOH. 2 after dehydra-
tion and long i.e., hours. X-ray exposure w13x.
XPS has directly verified that the increase in
CuI. in the ZnS surface at pH 9.2 is matched by
displacement of zinc ions showing that the reaction
mechanism involves exchange not simply adsorption
w14x.
Surface analysis w14x also shows that the CuI.
incorporation in the first minute reaches 25% of its
value after 9 h and that the second stage has a
logarithmic time dependence similar to that found
from solution analysis w5,6x.
XPS evidence indicates a mechanism in which,
as CuII. is reduced to CuI., corresponding oxida-
tion of S 2y occurs to give oxidation states near y1
e.g., S 2y
2
. w14x. Further oxidation produces polysul-
phide-like states S 2yn
. with oxidation states above
y1 w10,11x.
Hence, it is generally agreed that, leaving aside
adsorbed colloidal CuOH. 2 at high pH, CuII. re-
acts with the ZnS lattice to produce CuI. and re-
leases ZnII. to solution probably accompanied by
oxidation of sulphide ions. This leaves the composi-
tion and structure of the reaction products to be
defined and many different possibilities have been
proposed. Some authors have suggested that specific
copper sulphide phases, such as covellite, CuS
w2,3,15,16x, or chalcocite, Cu 2 S w12,17x, form on the
sphalerite surface in activation. Substitution of Zn by
Cu in the sphalerite lattice, resulting in a layer of
sphalerite extensively doped with copper, i.e.,
Cu x Zn 1yx S with x depending on the extent of
activation, has been proposed w10,11x.
There are significant difficulties with these pro-
posals. The structures of sphalerite and covellite are
incompatible and the stoichiometry of Cu 2 S cannot
be accommodated. A simple ion exchange of CuI.
into the sphalerite lattice is not expected due to the
resulting charge imbalance and the larger ionic ra-
dius of CuI. compared to ZnII..
The final structural and compositional mechanism
will need to recognise: the adsorption of CuII.
possibly associated with hydroxide ions, e.g., as
CuOHq and dissolved CuOH. 2 since CuI. forms
are not stable in solution under these pH and Eh
conditions; surface reaction to CuI. and release of
ZnII. from the sphalerite lattice; oxidation of sul-
phide to higher oxidation states; continued reaction
involving CuI. incorporation and ZnII. release im-
plying bulk processes. In this context, the possibility
of copper sulphide phases and mixed CuI. sul-
phiderhydroxide products will need to be examined.
 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223
Experimentally, further examination of the mech-
anism of copper activated sphalerite has been under-
taken using static secondary ion mass spectrometry
SIMS. in the imaging mode and X-ray absorption
fine structure XAFS. spectroscopy.
The specific aims of the experimentation were:
1. to determine, using SIMS, the lateral distribution
of copper on individual sphalerite particle surfaces
and, in particular, to determine whether this is rela-
tively uniform, patchy or particulate i.e., colloidal.
in form, 2. to determine the depth of penetration of
copper into the sphalerite lattice, 3. to determine,
using XAFS, the oxidation state of Cu adsorbedrab-
sorbed onto sphalerite through the measurement of
the Cu1s. binding energy relative to a number of
standard Cu containing compounds, and 4. to deter-
mine, using XAFS, the local elemental coordination
and geometry of the Cu adsorbedrabsorbed on the
sphalerite.
Using these data and the literature, we have theo-
retically modelled the structure of CuI. incorporated
in the surface and bulk layers of the sphalerite
lattice.
2. Methodology
2.1. Terminology
X-ray photoelectron spectroscopy measures the
energy required to excite the Cu2p 3r2 . electron to
the Fermi level. This occurs at approximately 934
eV. This energy, as measured in X-ray photoelectron
studies of Cu adsorbed on sphalerite, is referred to as
the Cu2p 3r2 . binding energy. In the XAFS mea-
surements presented herein, the Cu K a absorption
edge was examined. This occurs at an energy re-
quired to excite the Cu1s. electron to the vacuum
level, at approximately 8985 eV and will henceforth
be referred to as the Cu1s. binding energy.
2.2. SIMS imaging
Natural Balmat Tennessee. sphalerite Wards
Scientific, S 32.1 wt.%, Zn 66.9%, Fe 0.49%, Pb
0.04%, Cu 0.06%, SiO 2 0.13% and Ag 7 ppm. was
examined in the size fraction 3875 mm for most
experiments. The sphalerite particles were pressed
209
into soft conductive foil for charge minimisation and
mounted on the sample stub. A pure polycrystalline
sphalerite polished flat 2 = 2 mm was used for two
experiments: one conditioned at pH 5.5 with 37
nominal monolayers of CuII. i.e., 5 = 10y7 M, 20
ml. for 18 h; the other conditioned at pH 8.5 with
100 nominal monolayers of CuII. i.e., 5 = 10y4 M,
0.1 ml. for 15 min.
CuNO 3 . 2 was used in this study, as opposed to
CuSO4 which is added in plant practice, to avoid the
additional complication of the specific adsorption of
SO42y on the sphalerite surface. At either pH 5.5 or
8.5, the sample was conditioned under N2 for 20 min
followed by activation of the sample by 10y5 M
CuII. nitrate for 15 min. This concentration corre-
sponds to a nominal initial 7.4 equivalent monolay-
ers assuming complete adsorption of all Cu in solu-
tion. This coverage is calculated from the solution
volume and sample surface area as measured by Kr
adsorption BET assuming each Cu occupies 20.8 A 2
w18x. The adsorption isotherm, measured by copper
lost from solution, showed that, at pH 5.5, 1.8
monolayers was actually adsorbed. In another work
w19x, a procedure has been developed and tested in
which EDTA extraction is used to dissolve and
measure oxidised and adsorbed metal species. The
EDTA concentration and time 5 min. of extraction
were determined to maximise these species without
significant dissolution of the sulphide mineral. At pH
5.5, 0.9 monolayers were unextractable by the EDTA
complexation. At pH 8.5, 5.0 monolayers were ad-
sorbed with 1.4 monolayers unextractable w19x.
All SIMS spectra and images were recorded on
the Cameca 5f instrument at ANSTO, Lucas Heights
Laboratories, Sydney.
Table 1 lists the instrument parameters for the
SIMS imaging conditions. An area of 250 = 250
mm, encompassing 1030 particles, was normally
imaged with depth profiles taken on areas 125 = 125
mm over approximately three larger particles. Smaller
areas could not be reliably imaged or depth profiled
due to instability resulting from particle charging.
The beam diameter was estimated visually from the
screen to be f 10 mm with the majority of the
intensity in a 5 mm diameter. Ion etch rates were
estimated from profilometry measurement of crater
depths for similar beam conditions on ZrO 2 and Si
giving 1 = 10y3 nm sy1 and 5 = 10y3 nm sy1 ,
210 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223

Table 1 20 min to allow equilibration of the surface species

SIMS imaging conditions to occur. A smaller sphalerite size fraction was used
Instrument parameters than for the SIMS measurements to increase the
Primary ions Oq2 surface area available for analysis. Activation by
Primary ion energy 12.5 keV CuII. was performed by adding 0.001 dm3 of a
Primary current 4 nA
stock solution of cupric nitrate CuNO 3 . 2 . to the
Secondary ion energy q4.5 keV
Energy window 130 eV pre-conditioned mineral and stirring for a further 15
Contrast aperture 150 mm min or 30 min.. The stirring was then stopped and
Field aperture 1800 mm the sample settled generally about 5 min.. After
Image field 250=250 mm settling, the supernatant was decanted and the re-
Raster area 125=125 mm
maining 10 ml was transferred by syringe to a
Mass resolution 300
centrifuge tube.
Samples were prepared on-site at the Photon Fac-
respectively. An etch rate for these beam conditions tory for immediate analysis to eliminate the possibil-
was also estimated from the profile of gold-im- ity of aging effects. The samples were taken to the
planted arsenopyrite to be 6 = 10y3 nm sy1 . We beamline where they were again decanted and trans-
have therefore used the value of 5 = 10y3 nm sy1 in ferred to the sample holder by spatula and secured
estimating depths in this work. The SIMS signals with Kaptone tape in a spherical aperture 1.5 cm
corresponding to Oq mass 16., Cuq mass 63. and diameter. through a 1 mm thick aluminium holder.
Znq mass 64. were examined. Mass 32 was also Samples were prepared in two cupric nitrate solu-
examined comprising a composite of Sq and Oq tion concentrations, 10y5 and 10y4 M. For particles
2
ions. with diameter less than 20 mm, this corresponds to
0.37 and 3.7 monolayers coverage, respectively, as
2.3. XAFS calculated from the average particle surface area of
1.68 m2 gy1 . The samples generated are listed in
Cu activated sphalerite samples for XAFS analy- Table 2. Results from adsorption isotherms and
sis were prepared in the following manner. A total of EDTA extraction 5 min. are also shown.
500 ml of 10y3 M KNO 3 background electrolyte The XAFS data were collected using the 3 GeV
was N2 purged for 20 min at pH 5.5 or 8.5 in a Photon Factory Tsukuba, Japan. synchrotron X-ray
closed conditioning vessel. One gram of the dried radiation source. The experimental configuration used
- 20 mm sphalerite Wards Scientific. was weighed for both transmission and fluorescence modes has
and transferred to the conditioning vessel on comple- been described fully by Maclean et al. w20x. A de-
tion of electrolyte pre-purging and conditioned for tuned Si111. channel-cut monochromator was used

Table 2
Cu activated sphalerite sample preparation conditions
pH wCuNO 3 . 2 x M Particle size Activation time Monolayers Monolayers unextractable
mm. min. adsorbed EDTA.
5.5 10y5 3875 15 1.8 0.9
8.5 10y5 3875 15 5.0 1.4
5.5 10y4 - 20 15 3.7 1.6
8.5 10y4 - 20 15 3.7 1.3
5.5 10y5 - 20 15 0.37 0.34
8.5 10y5 - 20 15 0.37 0.33
5.5 10y4 - 20 30

The top two rows refer to samples used for the SIMS measurements. The other rows describe the conditions used to formulate the XAFS
samples.
 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223
in order to filter out high order X-ray harmonics. For
both fluorescence and transmission mode measure-
ments, the incident beam I0 . was monitored using a
15 cm long ion chamber Oken, Japan. filled with
N2 . The detector used to measure the transmitted
X-ray intensity I . was a 30 cm long ion chamber
Oken, Japan. filled with a 1:1 mixture of Ar and
N2 .
Standards of Cu 2 S, CuS, CuNO 3 . 2 , CuSO4 , CuO,
Cu 2 O and Cu foil were used to determine the Cu1s.
binding energies for various Cu oxidation states and
XAFS spectra for various local Cu geometries. XAFS
data collection for the Cu containing standards was
run in transmission mode, using X-ray transparent
boron nitride as the diluent to give sample X-ray
absorptions of about 55%. The data for each sample
was recorded 3 times from 8770 to 10,000 eV. The
derivative of this spectrum with respect to energy.
gives a peak minimum at the point of inflection, i.e.,
the absorption edge. enabling the edge positions to
be accurately determined.
The XAFS data collection from the adsorbed
copper on sphalerite were carried out in fluorescence
mode w20x to enhance surface sensitivity since the
copper was primarily on the surface of the particles.
Prior to detection, the backscattered and fluorescence
generated X-rays were passed through a nickel foil
filter to reduce the signal contamination by the
backscattered X-ray component. An ion fluorescence
Lytle detector was used to measure the resulting
fluorescent X-ray intensity generated I .. Three data
sets were collected from 8700 to 9600 eV for each
sample. Data for the Cu adsorbed onto sphalerite
samples was not collected at higher energies due to
the interference of the Zn1s. edge.
Repeat data sets were averaged and calibrated
relative to any energy scale error determined from
the Cu foil Cu1s. binding energy. Pre and post-edge
background subtraction was then carried out. Pre-
edge data substraction was accomplished by subtract-
ing a polynomial fitted to the pre-edge region from
the entire data set. The post-edge data was fitted to a
three region spline that was subsequently subtracted.
Analysis of the averaged and background sub-
tracted files was carried out using XFIT, an interac-
tive XAFS analysis program w21x. This program uses
non-linear least squares to refine a model against the
measured data. For this purpose, the data is trans-
211
formed to k space, k s "y1 w2 m e E y E0 .x1r2 where
E0 is the threshold binding energy and k is the
photoelectron momentum. and is weighted as a func-
tion of k 3.
The XAFS spectrum simulated from the model
structure was calculated using both ab intio single
generated by the computer program FEFF 4.06. w22x
and multiple scattering models generated by the
computer program FEFF 6.01. w23x. Multiple scatter-
ing analysis provided no significant benefit in terms
of the agreement between the simulated and mea-
sured data as compared to single scattering analysis
for a comparable number of refinable parameters.
Hence, all the results reported here are based on
single scattering analysis. The parameters optimised
were the DebyeWaller factors s 2 , A 2 ., the shell
occupancy n., the threshold energy E0 ., the overall
scale factor S02 . and the distance of the shell from
the central emitting atom x, A .. In all cases, E0 , S02 ,
2
x, and s were optimised first while n was held at
3. When optimisation was satisfactorily completed
for these variables all the variables including n were
then reoptimised.
The fit R . of the experimentally observed XAFS
data to that generated by the model is defined as:
Rs ( 2
x EXAFS rX 02
2
where x EXAFS is the quantity being minimised and
X 0 is when the calculated XAFS is uniformly 0.
2
X 2sS xcalc k . y xobs k . . rwobs k .
where k is the photoelectron momentum A ., xcalc k .
and xobs k are the calculated and observed XAFS
.
values and wobs k . is the root mean square of the
noise.
3. Results
3.1. SIMS
Six regions were examined for each sample.
Fig. 1a and d show typical images of the lateral
distribution of copper and zinc, respectively, in a 250
mm2 area after conditioning at pH 5.5 with 10y5 M
CuNO 3 . 2 , i.e., nominal 7.4 monolayers added; 1.8
monolayers adsorbed. It is clear that the initial distri-
bution of Zn and Cu is closely similar. On larger
212 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223

Fig. 1. a. and d. show the lateral distributions of copper and zinc, respectively, in a 250 = 250 mm area after conditioning at pH 5.5 with
10y5 M copper nitrate. An illustration of the removal of copper from the sphalerite surfaces during these depth profiles can be seen in b.
where the image is taken from the same area 250 = 250 mm. as a. after removal of f 2 nm corresponding to approximately six
monolayers of the sphalerite lattice.. c. and f. show the lateral distributions of Zn and Cu obtained after conditioning a sphalerite flat at pH
8.5 with addition of nominally 100 monolayers. Unlike a., the Cu distribution is not even and colloidal CuOH. 2 particles 630 mm. are
clearly observable.

particles, in the centre of the area, the copper distri-
bution is relatively uniform and directly matched to
the distribution of the zinc. There are, however, a
number of smaller particles identified in the Zn
image that are not observed with the same intensity
in the Cu image although, where smaller particles are
observed in the Cu image, the distribution of copper
appears to be relatively uniform across their surfaces.
The observation that the Cu signal from smaller
particles is less intense than that from larger particles
is quite general.
Images obtained for pH 8.5 conditioning i.e.,
nominal 7.4 monolayers added; 5.0 monolayers ad-
sorbed. were closely similar. The copper distribution
is obviously not patchy on the ) 10 mm scale. No
evidence for colloidal copper species larger than the
ion beam i.e., f 10 mm. was detected for either pH
5.5 or 8.5 under these conditions. That colloidal
CuOH. 2 particles ) 10 mm would have been ob-
served is illustrated in Fig. 1c. In this case, a spha-
lerite flat conditioned with 100 nominal monolayers
i.e., 5 = 10y4 M. at pH 8.5 for 15 min shows clear
evidence of Cu-rich colloidal particles 630 mm
obscuring the underlying ZnS surface. These col-
loidal particles only become apparent at higher nom-
inal coverages i.e., ) 10 monolayers..
Depth profiles were carried out on a large number
of different areas containing between three and 30
 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223 213

particles. The ion etch rate of the beam was very
slow, corresponding to 5 = 10y3 nm sy1 , so that the
depth profile and imaging could be followed over
1015 min. An image after the removal of copper
from the sphalerite surfaces during these depth pro-
files can be seen in Fig. 1b. The image is taken from
the same area 250 mm. as Fig. 1a after removal of
f 2 nm corresponding to approximately six mono-
layers of the sphalerite lattice.. The removal of
copper from the majority of the particle surfaces is
almost complete although some intensity remains at
isolated positions that may represent thicker layers.
In this instance adsorption isotherms indicate that 1.8
equivalent monolayers of Cu were adsorbed. The
SIMS results at pH 5.5 directly support the adsorp-
tion isotherm data, i.e., 12 equivalent monolayers
are incorporated over f 6 monolayers depth after 15
min conditioning at 10y5 M. Full exchange of Zn by
Cu does not take place in any layer; even the outer-
most layer has probably - 40% exchange of Cu for
Zn.
Illustrations of depth profile data from different
areas are shown in Fig. 2. The ion etch rate for these
profiles was 5 = 10y3 nm sy1. Fig. 2a is taken from

Fig. 2. SIMS depth profiles from 125 = 125 mm areas i.e., three particles. of the same sample as Fig. 1a.. The etch rate for the profile was
5 = 10y3 nm sy1 . a. shows the depth profile of the central area of Fig. 1. b. shows the depth profile of a ZnS flat conditioned at pH 5.5
for 18 h with 37 nominal monolayers.
214 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223

a 125 = 125 mm area three particles. showing simi-
lar intensity characteristics to that in Fig. 1a. Smaller
areas i.e., single particles. could not be measured
due to excessive surface charging. The signal for
Cuq mass 63. shows an initial increase, probably
due to adsorbed copper ions, followed by a steadily
decreasing concentration profile over the first ap-
proximately 1 nm depth. The majority of the Cu is
removed in the first 0.7 nm. This confirms that the
Cu is largely confined to the top two monolayers.
The reason for the continuation of the Cu signal
beyond 2 nm is probably the exposure of new sur-
faces at the edges of the three particles during profil-
ing.
For comparison, the depth profile from the ZnS
flat conditioned at pH 5.5 for 18 h with 37 nominal
monolayers i.e., 5 = 10y7 M. is shown as Fig. 2b.
The longer conditioning has clearly exchanged more
Zn in the surface layers and the depth of penetration
has increased to ; 15 nm i.e., f 45 monolayers. as
found by Gaudin w1x. Exchange in the outermost
layer apparently has increased to 7075% of the Zn.
3.2. XAFS
In order to properly interpret the XAFS analysis
of the Cu adsorbed onto sphalerite we firstly under-
took an examination of several Cu-containing stan-
dard compounds. This was done so that an estima-
tion of the errors associated with the determined
bond lengths and shell occupancies could be carried
out. Table 3 shows CuS and CuO bond lengths

Table 3
Cu to nearest neighbour distances determined from single crystal X-ray diffraction
Compound Bond Bond Occupancy Geometry of the Cu;
.
length A R, fit of data to model
CuS, Covellite w25x CuS 2.19 3 Trigonal planar
3.31 3 Tetrahedral
2.35 1
2.17 3 8.14%
2.30 7
Cu 2 S, CuS 2.32 average. 3 Triangular
Low chalcocite w26x 2.48 average. 4 Tetrahedral
each Cu also has from 2 to 8 Cu both distorted and highly variable.
.
neighbours at less than 3 A
2.29 6 17.8%
CuO, CuO 1.96 4 Tetrahedral
Cupric oxide w27x, CuO 2.78 2
CuCu 2.90 4
1.94 5 17.0%
2.91 4
Cu 2 O, CuO 1.85 2 Linear
Cuprite w28x CuCu 3.02 12
1.84 2 18.7%
3.00 13
CuSO4 CuO 1.87 2 Distorted
Cupric sulphate w29x 2.03 2 Octahedral
2.38 2
1.95 3 7.58%
2.36 45
CuOH. 2 , CuO 1.93 2 Square planar
Copper hydroxide w30x 1.94 2

The bond lengths and shell occupancies derived from the XAFS data are given in italics.
 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223 215

Table 4
2 ., x s bond length A .,
XAFS results from Cu adsorbed onto sphalerite: n s occupancy, s 2 s DebyeWaller temperature factor A
R s goodness of fit
Conditions CuNO 3 . 2 concentration M., pH, S O
CuNO 3 . 2 conditioning time min. R Geometry R Geometry
y4
10 8.7 n s 3.0 25.2 n s 3.0
5.5 s 2 s 0.007 s 2 s 0.001
15 x s 2.26 x s 2.25
10y4 11.7 n s 3.0 27.8 n s 3.0
8.5 s 2 s 0.008 s 2 s 0.002
15 x s 2.27 x s 2.25
10y5 19.3 n s 3.0 20.3 n s 2.3
5.5 s 2 s 0.008 s 2 s 0.003
15 x s 2.26 x s 2.25
10y5 15.5 n s 2.9 34.2 n s 2.7
8.5 s 2 s 0.002 s 2 s 0.001
15 x s 2.30 x s 2.29
10y4 14.3 n s 3.0 22.2 n s 3.0
5.5 s 2 s 0.008 s 2 s 0.002
30 x s 2.27 x s 2.26

and shell occupancies determined from the single
crystal structures of the standard compounds exam-
ined as compared to those values derived from XAFS
analysis. In all cases, the analysis was carried out
prior to determination of bond lengths from the
single crystal X-ray structures. It is apparent from
Table 3 that XAFS-derived occupancies are inaccu-
rate and will be essentially discarded in the follow-
ing discussion. Bond lengths are very accurate, par-
ticularly for the closest shell, generally in the order
of "0.2 A as compared to the bond lengths derived
from single crystal X-ray data. Discrepancies in bond
lengths tend to occur only where two bond lengths
are reasonably similar. For instance, for CuSO4 the
CuO bond lengths of 1.87 and 2.03 A have been
fitted by a single oxygen containing shell at 1.95 A.
Table 4 shows the bond lengths and occupancies
derived from the data recorded from the Cu adsorbed
onto sphalerite samples. The models fitted contained
either one S-containing shell or one O-containing
shell. When both O and S shells were fitted, at least
one of the DebyeWaller factors became negative.
The fit of the data set to each of these models is also
given in Table 4. In all cases, a nearest neighbour of
S gives a better fit than a nearest neighbour of O.
The data derived from the 10y5 M CuNO 3 . 2 solu-
tion samples give rise to the poorest fits of the
simulated data due to the noise of the experimental
signal. One shell of S gives sufficiently reasonable
fits that the addition of further shells is not justified.
As an example of these calculations, a compari-
son of the background-subtracted XAFS data in k-
space for Cu adsorbed onto sphalerite conditioned in
a 10y4 M CuNO 3 . 2 solution for 15 min at pH 5.5
with the simulated data is given in Fig. 3. The
average CuS bond length derived is 2.27 A with
standard deviation of 0.02 A Table 4.. This bond
length is not affected by the concentration of Cu
added or the pH or the length of conditioning time.
Fig. 3. The simulated XAFS spectra for one S shell at a CuS
as compared to the experimental data back-
distance of 2.26 A
ground subtracted. for the sphalerite sample conditioned in 10y4
M CuNO 3 . 2 for 15 min at pH 5.5. x indicates measured data,
simulated data.
216 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223

Table 5
Cu1s. binding energies of the Cu-containing standards and Cu activated sphalerite
Compound Cu1s. binding energy eV. Formal oxidation state
Cu metal 8979 0
Cu activated ZnS 8985.7 1
pH 5.5, 10y4 M CuNO 3 . 2 15 min
Cu activated ZnS 8985.7 1
pH 8.5, 10y4 M CuNO 3 . 2 , 15 min
Cu activated ZnS 8985.7 1
pH 5.5, 10y4 M CuNO 3 . 2 30 min
Cu activated ZnS 8986.8 1
pH 5.5, 10y5 M CuNO 3 . 2 415 min
Cu activated ZnS 8986.8 1
pH 8.5, 10y5 M CuNO 3 . 2 15 min
Cu 2 O 8987.0 1
Cu 2 S 8987.5 1
CuS 8989.5 2
CuO 8991.0 2
CuSO4 8994.0 2
CuNO 3 . 2 8995.0 2

Table 5 shows the Cu1s. binding energy of both
the Cu standards and the Cu adsorbed onto spha-
lerite. Also shown is the formal oxidation state of the
samples. The derivative of the emitted fluorescence
X-ray intensities across the Cu1s. binding energy
for the different samples of Cu adsorbed onto spha-
lerite are shown in Fig. 4. This figure indicates that
the samples display two Cu1s. binding energies
depending on the concentration of the initial condi-
tioning solution. The samples derived from the less
concentrated Cu solution, i.e., 10y5 M CuNO 3 . 2
have a slightly higher energy edge position, i.e.,

Fig. 4. Cu1s. binding energies of Cu adsorbed onto sphalerite from solutions containing 10y4 and 10y5 M CuNO 3 . 2 are indicated. The
binding energies for the samples from the 10y4 M CuNO 3 . 2 solutions are approximately 1 eV lower than the samples formed in the 10y5
M CuNO 3 . 2 solutions. In addition, the profile of the derivative of the Cu1s. edge for the Cu adsorbed onto sphalerite from the 10y4 M
CuNO 3 . 2 solution is not symmetrical. This may occur as this profile is made up of components of Cu within the bulk of the sphalerite and
the surface of the sphalerite, whereas samples derived from 10y5 M CuNO 3 . 2 solutions only contain Cu adsorbed onto the surface of the
sphalerite.
 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223 217

Fig. 5. The Cu1s. binding edge positions of the Cu adsorbed onto sphalerite samples is shown as compared to a number of Cu containing
standard compounds. This figure illustrates that there is no fixed Cu1s. binding energy for a given nominal oxidation state, e.g., the edge
positions for CuO and CuS both nominally CuII. do not fall at the same energy. Cu adsorbed onto sphalerite both from 10y4 and 10y5 M
CuNO 3 . 2 solutions. occurs at a very low CuI. position, i.e., at a lower energy than either Cu 2 O or Cu 2 S.

8986.8 eV., indicating a higher copper oxidation
state, than the more concentrated solution i.e.,
8985.7 eV..
Fig. 5 shows the derivatives of the Cu containing
standards Cu1s. binding edges. It is apparent from
Table 5 and Fig. 5 that there are a considerable range
of Cu1s. binding energies for a formal oxidation
state, e.g., for CuII. the edge positions range from
8989.5 eV to 8994.0 eV for CuS and CuSO4 , respec-
tively. Cu 2 S shows a distinctly asymmetric edge
differential. In low chalcocite, there are many differ-
ent local Cu geometries. The observed asymmetry is
likely to be a reflection of the slightly differing
oxidation states of these Cu atoms.
4. Discussion
4.1. SIMS
In terms of the specific aims of the experimenta-
tion, the results have shown that the distribution of
copper on individual sphalerite particle surfaces is
relatively uniform above the resolution of the ion
beam, i.e., ) 10 mm. at nominal additions less than
10 monolayers for low or high pH adsorption. There
is no evidence at this scale for patchy or particulate
i.e., colloidal. adsorption. There are, however, sug-
gestions from the Cu images that smaller particles
give less intense signals and, from the depth profiles,
that there may be some regions in which the depth of
copper incorporation is not uniform. The depth pro-
files are in strong agreement with the adsorption
isotherm data in showing that 12 equivalent mono-
layers of copper are incorporated mostly in the first
nm of the sphalerite surface at pH 5.5 or 8.5 after 15
min conditioning at 10y5 M CuII. concentration.
Only at relatively high nominal monolayer additions
and high pH i.e., above 7.5. are colloidal CuOH. 2
particles found in SIMS images.
The SIMS and XPS results support a mechanism
in which, at low pH, CuII. ions arriving at the ZnS
surface are immediately reacted to CuI. and incor-
porated into any Zn site on the surface with release
of the ZnII. ion to solution. At higher copper con-
centrations, further reaction to the second layers and
subsequent layers of the ZnS lattice takes place with
similar substitution of CuI. for ZnII. in the lattice.
We can speculate that this diffusion may take place
either by site exchange between the CuI. in the
surface layer and ZnII. in the second layer or by
interstitial movement of the smaller CuII. ion
through the lattice. In either case, the adsorption
isotherms and EDTA extraction data w19x suggest
that, once the CuI. is bonded into the sulphide sites,
it is not easily extracted compared with adsorbed
218 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223
CuII. ions and hydroxide species, e.g., CuOHq,
CuOH. 2 . The CuS-like activation product has been
shown by both XPS w14x and electrochemical studies
w24x to be oxidised in preference to ZnS suggesting
that the CuI. site is more reactive than the Zn II.
site.
4.2. XAFS
The average CuS bond length of 2.27 A deter-
mined from the XAFS data analysis is too long for a
CuO bond except in the case of CuSO4 Table 3..
However, in the case of CuSO4 , besides the two 2.38
A bonds, there are four other much shorter accompa-
nying bonds, attached to the octahedrally coordinated
Cu. No evidence for bonds shorter than 2.27 A can
be derived from the XAFS spectra. Therefore, nei-
ther CuOH.q species nor CuOH. 2 colloids are
present in significant concentrations on the sphalerite
surface even at pH 8.5 where they may be expected.
Hence, further discussion will focus only on the
possible interactions of Cu with sphalerite and will
not include the analysis of any effects involving
competitive reactions with either O 2 or OHy.
The ZnS bond length in ZnS is 2.34 A. Assum-
ing a direct replacement of Zn by Cu in bulk spha-
lerite the resulting CuS bond length would be
comparable in dimension to the CuS bond length
for tetrahedral Cu geometries in covellite. However,
2.34 A is significantly longer than the CuS bond
length derived from the XAFS data i.e., 2.27 A ..
The observed CuS bond length is intermediate be-
tween the trigonal planar Cu geometry observed in
covellite i.e., 2.19 A . and the tetrahedral Cu geo-
metry also found in covellite. Should the Cu replace
Zn within the sphalerite lattice and then relax to a
distorted tetrahedral geometry with 3 CuS bond
lengths of 2.27 A, then the fourth CuS bond length
would extend to 2.59 A effectively breaking this
bond, leaving the Cu in a distorted trigonal planar
geometry bonded to 3 S atoms. However, no evi-
dence of this longer interaction length is observed in
the XAFS data. Three possible explanations for this
exist which may all contribute Firstly the calculation
that has given rise to this distance of 2.59 A assumes
no movement of the S atom with reduced coordina-
tion, i.e., no retraction of its position from what was
formerly tetrahedral coordination to a trigonal planar
coordination. Were this contraction to occur, the Cu
to S separation distance would be closer to 3.94 A
and would thus be close to the upper limit of atomic
separation to which XAFS is sensitive. Secondly,
assuming 3.7 Cu monolayers are adsorbedrab-
sorbed. This would give rise to sulfur atoms totalling
11.1 monolayers in near-trigonal planar coordination
around the copper atoms. Sulfur atoms totalling only
2.7 monolayers would be at the extended Cu to S
interaction distance of 2.59 A or greater.. The XAFS
data may not be sufficiently sensitive for this more
minor contribution to play a major role in model
optimisation. Thirdly, data was only collected out to
9600 eV due to interference from the Zn1s. edge,
therefore, it is possible that data collected over a
larger energy range were this to be possible. may be
more sensitive to the longer Zn to S interaction
distance.
A similar argument can be invoked for Cu ad-
sorbed onto the ZnS surface. In this instance the Zn
atoms which the Cu replaces already lie in a trigonal
planar or distorted trigonal planar geometry with
respect to the surrounding S atoms as one ZnS
bond is lacking due to surface termination. In this
case, Cu can replace Zn and relax into an optimal
geometry without the necessity to disrupt what was
formerly a tetrahedral geometry surrounding the bulk
Zn atom. Schematics of the geometry surrounding
both surface and bulk copper atoms in the sphalerite
lattice are shown in Figs. 6 and 7, respectively. The
CuS bond lengths in these figures have been ad-
It is apparent that no significant lat-
justed to 2.27 A.
tice distortion is required to accommodate the Cu in
this geometry.
The CuS bond lengths observed in low chal-
cocite have not been invoked in this discussion due
to the complicating factor of significant CuCu in-
teractions.
4.3. Oxidation states of Cu, S and Zn
The S2p. binding energies of 161.3 eV 162.1 eV
and 163.1 eV were simultaneously observed on anal-
ysis of Cu activated ZnS and were assigned to S 2y,
S1y or S 2y
2
. and S 0 species, respectively w31x.
These XPS measurements were carried out using a
 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223
Fig. 6. The position of the Cu on the surface of the sphalerite
lattice is shown as indicated by the XAFS results, i.e., the Cu has
replaced one Zn and is in a distorted trigonal planar position
between three S with a CuS bond length of 2.27 A. The ZnS
surface structure shown in this figure is of the 110. surface the
major sphalerite cleavage plane. and assumes no surface relax-
ation has taken place. The Cu is indicated by the light grey sphere,
the surrounding sulphurs by darker grey spheres or grey lines and
the zinc atoms by black lines.
synchrotron radiation X-ray source with samples that
had been activated by Cu under either air or N2 . The
binding energy of 161.3 eV is also observed on
analysis of unactivated ZnS. The binding energies at
162.1 and 163.1 eV cannot be attributed to further
oxidation of the activation product by O 2 as they are
also observed in significant quantities on activation
under N2 w31x.
The binding energy observed at 163.1 eV is still
significantly lower than that observed for elemental
sulphur. For S 8 , the binding energies of 164.2 w32x
and 163.8 w33x have been observed. Approximating
the oxidation state of the two S species observed in
Cu activated sphalerite by linear interpolation be-
tween the binding energies of S 0 164.2 eV. and S 2y
161.3 eV. indicates oxidation states of approxi-
mately y0.5 and y1.5 for the binding energies of
163.1 eV and 162.1 eV, respectively.
The observation of binding energies greater than
those for Sy1 is generally attributed to either ele-
mental sulfur species we.g., w31xx or the formation of
219
polysulfide species S 2y
n which are described as long
chains of S, the middle S atoms having a near-neu-
tral charge and the two end S having a charge near
y1 w34x. It is difficult to postulate a mechanism by
which these chains may form in ZnS as the shortest
SS distance is 3.82 A, considerably larger than the
longest possible SS bond. The formation of a poly-
sulfide chain would therefore require a considerable
rearrangement of the activated ZnS structure. Due to
the high degree of bond breaking that would be
required to form a polysulfide chain within or on the
sphalerite lattice a high activation energy for this
reaction might reasonably be expected. As Cu re-
placement of Zn on the sphalerite lattice occurs
readily at room temperature, the mechanism involved
does not, apparently, require a high activation en-
ergy, a result which is unlikely to be consistent with
the surface reconstruction required to forma S 2y n
species.
The Cu1s. binding energy of a number of com-
pounds including the range of copper activated spha-
lerite samples shown in Table 5 has been investi-
Fig. 7. The position of a Cu absorbed into the sphalerite lattice is
shown as indicated by the XAFS results. As for Fig. 6, the Cu
occupies a distorted trigonal planar position between three sul-
phurs and has replaced one zinc atom. The CuS bond length is
The retraction of the Cu from the tetrahedral position
2.27 A.
formerly occupied by the zinc atom, has caused the breakage of
one possible CuS bond. The distance separating these atoms is
now 2.59 A.
220 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223
gated using XAFS. Fig. 4 shows the derivative of the
X-ray fluorescence intensity as a function of eV
across the Cu1s. binding energy for the Cu acti-
vated sphalerite samples. The maximum slope of a
plot of the fluorescence emission intensity vs. inci-
dent X-ray energy defines the binding energy. The
Cu1s. binding energy of all the Cu activated spha-
lerite samples prepared in a 10y4 M CuNO 3 . 2
solution is 8985.7 " 0.2 eV. This energy for the
samples prepared in a 10y5 M CuNO 3 . 2 solution is
8986.8 " 0.2 eV. These two energies are signifi-
cantly different and indicate that the Cu stemming
from the less concentrated solution are more oxi-
dised than those from the more concentrated solu-
tion. It is a reasonable assumption that the copper
from the less concentrated solution is located largely
on the surface of the sphalerite equivalent of 0.37
monolayers. and hence, surface adsorbed copper is
more oxidised than bulk absorbed copper.
Fig. 5 compares the Cu1s. binding energy posi-
tions of Cu on sphalerite samples with those from
the copper oxide and sulfide standards. It is apparent
Fig. 5. that a specific binding energy does not exist
for a given formal oxidation state although there is a
clear trend with increasing oxidation state. Were this
to be the case, then the Cu1s. binding energies of
Cu 2 O and Cu 2 S would be the same as would the
binding energies of CuS, CuO, CuSO4 and
CuNO 3 . 2 . In reality, the Cu1s. binding energy is in
part a function of the strength of bonding to the Cu
which in turn is a function of the electronegativity
and electron affinity of the surrounding atoms as
compared to the copper. As these properties are not
integer values there is no reason for the oxidation
state of the Cu to be an integer value.
The binding energy of the Cu1s. on activated
sphalerite falls at intermediate energies 8985.7 eV,
bulk and 8986.8 eV, surface. between Cu 2 O 8987.0
eV. or Cu 2 S 8987.5 eV. and Cu metal at 8979 eV.
This indicates that the oxidation state of Cu is actu-
ally intermediate between q1 and 0 with the oxida-
tion state of the surface Cu being very close to that
of Cu 2 O, i.e., nominally q1 and the oxidation state
of the bulk copper being slightly lower than that of
the surface copper but still falling between 0 and
q1.
In addition, on activation of sphalerite by Cu a
significant reduction of Zn takes place from a Zn2p.
binding energy of 1022.1 eV to 1021.4 eV, i.e., the
oxidation state is decreased on sphalerite activation
w31x.
4.4. A mechanistic model for the substitution of
surface Zn by Cu
The replacement of Zn by Cu in sphalerite, as-
suming a resulting Cu1q oxidation state can be
written as:
0
Cu2qq Zn2q S 2y Cu1q S1y . q Zn2q. 1.
A structural examination of a possible sphalerite
crystal 110. cleavage face as in Fig. 6. indicates
that surface absorbed Cu would be coordinated to
three sulphur atoms due to the termination of the
bulk lattice at the crystal surface. The replacement of
a surface Zn by a Cu may therefore be written using
oxidation states approximated from the XAFS data
reported here and literature XPS results already dis-
cussed as:
4y 4y
Cu2qq Zn2q S 2y
3 . Cu0.9 Sy1.63
3 . q Zn2q.
2.
It is apparent from Eq. 2. that for Cu to achieve
approximately the oxidation states predicted from the
observed Cu1s. binding energies, the surrounding S
atoms must be oxidised to approximately y1.6.
Further surface Cu substitution for neighbouring Zn
atoms would require the further oxidation of the S
atoms. Eq. 3. indicates the progressive oxidation of
the surface S atoms that may take place as further Cu
substitution in the surface layer takes place.
6y
S1.63y
2 Cu0.9qy S1.63yy Zn2q S 2y
2 .
6y
S1.63y
2 Cu0.9qy S1.28yy Cu0.9q S1.63y
2 . .
3.
The distribution of oxidation states of the S atoms
on the surface will therefore be dependent on the
concentration of Cu atoms on the surface and hence
the number of neighbouring Zn sites that have been
substituted by Cu. It is not surprising therefore that a
number of S2p. binding energies have been ob-
served for Cu activated sphalerite.
 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223
4.5. A mechanistic model for the bulk substitution of
Zn by Cu
In the bulk, if the Cu were to directly replace a Zn
atom each Cu would have 4-fold tetrahedral S coor-
dination. XAFS data analysis clearly indicates that as
with the surface adsorbed Cu the bulk absorbed Cu
also has a 3-fold distorted trigonal coordination to S
atoms Fig. 7.. Hence, any analysis that does not
include the breakage of one former ZnS bulk is
unlikely to be correct.
Surface replacement of Zn by Cu results in an
oxidation state of the Cu that is slightly greater than
for bulk absorbed Cu. The reduction of the surface
Cu from CuII. may stem from a weakening in
coordination bond strength as compared to the for-
mer ZnS bonds accompanied by the donation of
electron density from the S to the Cu atoms while
maintaining sufficient shared electron density to
maintain the three CuS bonds. In the bulk case, a
ZnS bond is broken releasing the electrons for-
merly in this bond to associate with the most elec-
tronegative neighbouring atoms. Hence, more elec-
trons are available for localisation on the Cu atom
and the reduction of the bulk Cu atoms as compared
to surface Cu atoms is therefore more pronounced.
The following type of electron exchange reaction
may occur, in the bulk on the formation of the
proposed distorted trigonal planar Cu conformation.
0 Cu1s. binding energies of the absorbed and ad-
S 2y 2q
3 Zn y S
2y
Zn2q
3 q Cu2q
4.09y
S1.63y
3 Cu0.8q
4.09q
q S 0.8y Zn1.63q
3 q Zn2q.
In this instance, due to the breakage of former
ZnS bond, a S oxidation state similar to that ex-
pected in a polysulfide may be achieved in a single
step. This is accompanied by oxidation of the other
neighbouring sulphides to oxidation states similar to
surface Cu activation. In addition, as observed exper-
imentally w31x, the neighbouring Zn atoms have been
reduced. On further local Cu substitutions it is likely
that S atoms with even less negative oxidation states
will be present. Thus, the exact distribution of S
oxidation states, as for surface Zn substitution be Cu,
will be determined by the extent and concentration
of the substitution.
221
5. Conclusions
SIMS images from sphalerite surfaces conditioned
with less than 10 normal monolayers of copper II.
at both pH 5.5 and 8.5 have shown relatively uni-
form distribution of copper species on individual
particles and strongly correlated copperrzinc distri-
bution images. The copper distribution does not ap-
pear to be patchy or particulate in nature on any of
the individual particle surfaces. It appears from this
result that Cu adsorption is not dependent on the
presence of corner, edge or other defect sites on the
sphalerite lattice and that therefore, any mechanism
proposed for Cu adsorption must not be dependent
on Zn sites that are of lower coordination than those
expected for normal sphalerite 110. faces. SIMS
results also indicate that the Cu is present only in the
surface layers of the sphalerite, for instance on etch-
ing 1.8 equivalent monolayers of adsorbed Cu are
almost entirely removed in the first 0.7 nm or 2
sphalerite atomic planes.
XAFS results indicate that the Cu is present in
both bulk and surface sites in a distorted trigonal
planar geometry associated with 3 S atoms. The
CuS bond length has been derived, from the XAFS
data, to be 2.27 " 0.02 A. This bond length is not
analogous to either the tetrahedral or trigonal Cu
covellite sites. Nor can it be attributed to the forma-
tion of Cu hydroxy species, i.e., a CuO bond. The
sorbed Cu indicate a Cu oxidation state between 0
and 1. The slightly higher binding energy, by 1 eV,
of the surface adsorbed Cu 0.37 nominal monolay-
ers. as compared to the bulk absorbed Cu 3.7
4.
nominal monolayers. indicates that the surface Cu
may be slightly more oxidised than the bulk Cu.
One S2p. binding energy is observed prior to Cu
activation of sphalerite. Two further S binding ener-
gies are observed on Cu activation w31x. A mecha-
nism has been proposed which may explain both
these two extra S2p. binding energies and also the
apparently similar Cu coordination in both bulk and
surface sphalerite sites.
The S2p. binding energy of 162.1 eV indicating
a S oxidation state between y2 and y1. appears to
be associated with surface Cu adsorption. To obtain
a distorted trigonal planar Cu coordination each for-
mer ZnS bond is replaced by a CuS bond. The
222 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223
reduction of the Cu occurs by the localisation of
some of the electron density from the CuS bonds
onto the Cu and is due to the higher electronegativity
of Cu as compared to Zn. Hence, the Cu is reduced
from Cu2q to approximately Cu0.9q. In addition, the
S atoms surrounding the Cu are oxidised from S 2y
to an oxidation state between y2 and y1.
The S2p. binding energy of 163.1 eV indicating
a S oxidation state between 0 and y1, i.e., polysul-
fide like., seems to be associated with bulk Cu
absorption. If, on replacement of a bulk Zn with a
Cu, one of the ZnS bonds is not replaced by a
CuS bond a distorted trigonal planar CuS geome-
try is formed. This is the same Cu geometry as
formed on surface replacement of a Zn by a Cu. The
coordination of one of the S atoms formerly associ-
ated with the replaced Zn atom is also reduced from
tetrahedral to distorted trigonal planar. In this in-
stance, the main electron donating species to the Cu
is likely to be this S with reduced coordination. This
is in contrast to the surface adsorption Cu mecha-
nism where the Cu is reduced by its three bonded S.
In the bulk Cu case, the S of reduced coordination
may be oxidised from S 2y to S 0.8y while reducing
the Cu from Cu2q to Cu0.8q.
The difference in mechanism between the replace-
ment of a bulk Zn by a Cu as compared to the
replacement of a surface Zn offers a possible expla-
nation for the two step activation kinetics observed
w2,3,5,6x. Replacement of a surface Zn by a Cu atom,
as proposed here, requires no additional bond break-
age, the three former ZnS bonds are replaced with
three CuS bonds albeit of different electron density.
It is predicted therefore the kinetics of the surface
reaction may be rapid, involving a low activation
energy. In contrast, bulk replacement of Zn by Cu
requires a larger electron relocalisation as four for-
mer ZnS bonds are placed by three CuS bonds. It
is expected therefore that this reaction would require
a larger activation energy and would proceed more
slowly than the surface replacement of Zn by Cu at
the same temperature.
Acknowledgements
We gratefully acknowledge the support of the
Australian Institute of Nuclear Science and Engineer-
ing for the support in providing access to the SIMS
equipment. In addition, we acknowledge the contri-
bution of the Australian National Beamline Facility
for the opportunity to carry out the XAFS measure-
ments, in particular Dr. Garry Foran for making
available his expertise in this area. One of us Roger
St.C. Smart. gratefully acknowledges the award of
an ARC Senior Research Fellowship.
References
w1x A.M. Gaudin, Flotation, Chap. 1, 2nd edn., McGraw-Hill,
New York, 1957.
w2x N.P. Finkelstein, S.A. Allison, in: M.C. Fuerstenau Ed..,
Flotation, A.M. Gaudin Memorial Volume, Chap. 14, Vol. I,
American Institute of Mining and Petroleum Engineers, New
York, 1976.
w3x N.P. Finkelstein, The activation of sulfite minerals for flota-
tion: a review, Int. J. Miner. Process., 1998, in press.
w4x G. Siva Reddy, C. Konda Reddy, The chemistry of activation
of sphaleritea review, Miner. Proc. Extract. Metall. Rev. 4
1988. 1.
w5x J. Ralston, T.W. Healy, Int. J. Miner. Process. 7 1980. 175.
w6x J. Ralston, T.W. Healy, Int. J. Miner. Process. 7 1980. 203.
w7x M. Wada, S. Okada, Nippon Kogyo Kaishi 79 1963. 23.
w8x R.-H. Yoon, Z. Chen, N.P. Finkelstein, P.E. Richardson,
Proc. XIX International Mineral Processing Congress, Vol. 3,
Society for Mining, Metallurgy and Exploration, Littleton,
CO, 1995, p. 297.
w9x R.K. Clifford, K.C. Purdy, J.D. Miller, Am. Inst. Chem. Eng.
Symp. Ser. 71 1975. 138.
w10x A.N. Buckley, R. Woods, H.J. Wouterlood, Int. J. Miner.
Process. 26 1989. 29.
w11x C.A. Prestidge, W.M. Skinner, J. Ralston, R.St.C. Smart,
Appl. Surf. Sci. 108 1997. 333.
w12x V.I. Nefedov, Ya.V. Salyn, P.M. Solozhenkin, G.Yu. Pula-
tov, Surf. Interface Anal. 2 1980. 170.
w13x W.M. Skinner, C.A. Prestidge, R.St.C. Smart, Surf. Interf.
Anal. 24 1996. 620.
w14x I.J. Kartio, C.I. Basilio, R.H. Yoon, Proc. Electrochem. Soc.
96 1996. 25.
w15x J.L. Cecile, in: K.S.E. Forssberg Ed.., Developments in
Mineral Processing 6, Flotation of Sulfide Minerals, Elsevier,
Amsterdam, 1985, p. 61.
w16x J.S. Laskowski, Q. Liu, Y. Zhan, Min. Eng. 10 1997. 787.
w17x D.L. Perry, L. Tao, J.A. Taylor, in: P.E. Richardson, S.
Srinivasan, R. Woods Eds.., Proc. of the International Sym-
posium on Electrochemistry in Mineral and Metal Process-
ing, The Electrochemical Society, Pennington, NJ, 1985, p.
169.
w18x A.M. Gaudin, G.S. Preller, Trans. Am. Inst. Man. MeE. Eng.
169 1946. 248.
 A.R. Gerson et al.r Applied Surface Science 137 (1999) 207223
w19x A. Lange, W.M. Skinner, R.St.C. Smart, paper in prepara-
tion.
w20x A.L. Maclean, G.J. Foran, B.J. Kennedy, Inorg. Chem. Acta
268 1998. 231.
w21x P.J. Ellis, H.C. Freeman, J. Synchrotron Rad. 2 1995. 190.
w22x J. Mustre de Leon, J.J. Rehr, S.I. Zabinsky, R.C. Albers,
Phys. Rev. Condens. Matter 44 1991. 4146.
w23x J.J. Rehr, R.C. Albers, S.I. Zabinsky, Phys. Rev. Lett. 69
1992. 3397.
w24x R.-H. Yoon, Z. Chen, in: R. Woods, F.M. Doyle, P. Richard-
son Eds.., Proc. of the 4th International Symposium on
Electrochemistry in Mineral and Metal Processing, The Elec-
trochemical Society, Pennington, NJ, 1996, p. 38.
w25x H.T. Evans, J.A. Konnert, Am. Mineral. 61 1976. 996.
w26x H.T. Evans, Z. Kristallogr. 150 1979. 299.
223
w27x S. Asbrink, L. Norrby, Acta Cryst. B 26 1970. 8.
w28x R.W.G. Wyckoff, Crystal Structures, Chap. IV, Vol. 1, Inter-
science Publishers, New York, 1948, p. 331.
w29x R.W.G. Wyckoff, Crystal Structures, Chap. VIIIA, Vol. 1,
Interscience Publishers, New York, 1948, p. 40.
w30x J. Von Jaggi, H.R. Oswald, Acta Cryst. 14 1961. 1041.
w31x I.J. Kartio, C.I. Basilio, R.H. Yoon, Langmuir, 1998, in
press.
w32x T.A. Carlson, Photoelectron and Auger Spectroscopy,
Plenum, 1975.
w33x C.F. Jones, S. Lecount, R.St.C. Smart, T. White, Appl. Surf.
Sci. 55 1992. 65.
w34x J.R. Mycroft, G.M. Bancroft, N.S. McIntyre, J.W. Lorimer,
I.R. Hill, J. Electroanal. Chem. 292 1990. 139.