Interaction of cuprite with dialkyl dithiophosphates

Int. J. Miner. Process. 93 (2009) 155–164

Contents lists available at ScienceDirect

Int. J. Miner. Process. j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / i j m i n p r o

Interaction of cuprite with dialkyl dithiophosphates Alan N. Buckley a,⁎, Siew Wei Goh a, William M. Skinner b, Ronald Woods c, Liang-Jen Fan d a

School of Chemistry, The University of New South Wales, Sydney, NSW 2052, Australia Ian Wark Research Institute, ARC Special Research Centre for Particle and Material Interfaces, University of South Australia, Mawson Lakes, SA 5095, Australia c School of Biomolecular and Physical Sciences, Griffith University, Nathan, Qld 4111, Australia d National Synchrotron Radiation Research Center, Hsinchu 30076, Taiwan b

a r t i c l e

i n f o

Article history: Received 16 December 2008 Received in revised form 23 July 2009 Accepted 26 July 2009 Available online 5 August 2009 Keywords: Cuprite Surface adsorption Dithiophosphate collectors X-ray photoelectron spectroscopy

a b s t r a c t The interaction of freshly abraded surfaces of cuprite, Cu2O, with neutral or mildly alkaline aqueous solutions of diethyl or di-n-butyl dithiophosphate (DTP) has been investigated by means of conventional and synchrotron X-ray photoelectron spectroscopy and near-edge X-ray absorption fine structure spectroscopy. It was confirmed that DTP adsorbs readily on Cu atoms in the surface layer of the mineral treated with solutions of the collector at pH values near 7 and 9 in the presence of air, and renders the surface hydrophobic. When cuprite is treated with relatively high concentrations of DTP for sufficiently long periods, collector can also be adsorbed as CuDTP, but the coverage does not exceed a thin layer of CuDTP on the adsorbed DTP monolayer, unlike the situation with Cu metal or chalcocite where a thick multilayer can be formed. © 2009 Elsevier B.V. All rights reserved.

1. Introduction Cuprite, nominally Cu2O, is one of the well-defined minerals found in the oxidised zones of copper sulfide deposits. Its behaviour as fine particles during the concentration by flotation of a mixed copper sulfide and oxide ore from near an oxidised zone is consequently of commercial importance. Therefore, in research designed to optimise the flotation of these minerals, the interaction of cuprite with thiol flotation collectors, such as ethyl xanthate (EX), diethyl or di-n-butyl dithiophosphate (EDTP or BDTP) and 2-mercaptobenzothiazole (MBT), is of considerable interest yet not thoroughly investigated. In practice, most oxide copper minerals do not respond well to the thiol collectors normally used in copper sulfide flotation without prior sulfidisation with sodium or ammonium sulfide (Hanson and Fuerstenau, 1991; Lee et al., 1998). Typically, mixed copper sulfide and oxide ores are treated by flotation of the sulfide minerals before sulfidisation and subsequent flotation of the oxides (Kelsall, 1961). However, cuprite has been shown in laboratory experiments to be readily floatable with EX at pH 11 within the pulp potential range of −0.4 to 0.29 V (vs SHE) (Heyes and Trahar, 1979; Senior et al., 2006). Only low concentrations of EX were required, similar to those used for chalcocite (Cu2S). At potentials higher than 0.29 V, flotation recovery of cuprite was found to be inhibited as it was for chalcocite (Cu2S) and, for both minerals, this behaviour was assigned to oxidation of copper xanthate to CuO and dixanthogen. It was pointed out by Heyes and Trahar (1979) that, at pH 11, the cuprite surface would have been reduced to copper metal below −0.18 V. Copper metal is known to interact with xanthate at such potentials to form a ⁎ Corresponding author. E-mail address: [email protected] (A.N. Buckley). 0301-7516/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.minpro.2009.07.011

chemisorbed xanthate monolayer (Woods, 1996), and this would explain the observed floatability of cuprite at low potentials and the apparent absence of a lower potential edge for flotation of this mineral. Curiously, at a pulp potential near zero, flotation recovery of cuprite with EX was low at pH values of ≤9 (Heyes and Trahar, 1979). Those authors concluded that their data were “insufficient to permit any useful conclusions to be drawn” regarding the influence of pH, but pointed out that “cuprite only floated at pH values above the zero point of charge” which they assumed to be near pH 9 on the basis of the work of Attia (1975). By contrast, it had been reported that whereas sulfides, cuprite and native copper all float with xanthates without sulfidisation, tenorite (CuO) floats only poorly (Rey, 1954), notwithstanding the fact that CuO is also a p-type semiconductor with a band gap smaller than that of Cu2O (Ghijsen et al., 1988; Pierson et al., 2005). Cuprous oxide is thermodynamically unstable relative to cupric oxide in the presence of oxygen and water, consequently a surface layer of a CuII–O species might be expected to form when a fresh cuprite surface is exposed to air under ambient conditions or in aqueous environments. Indeed, an investigation of mineral cuprite fracture surfaces by means of synchrotron X-ray photoelectron spectroscopy (SXPS) and near-edge X-ray absorption fine structure (NEXAFS) spectroscopy established that about a monolayer of cupric hydroxide was formed at the surface after exposure to air for 1 h (Harmer et al., 2009). Thus when cuprite particles are treated with a flotation collector, the initial interaction is expected to occur at a surface altered by oxidation and hydroxylation, and hence to also be of relevance to the flotation of CuII oxide minerals. Moreover, the way in which thiol collectors interact with cuprite is expected to have a strong bearing on the way in which those collectors interact with an oxidised copper sulfide mineral surface.

156

A.N. Buckley et al. / Int. J. Miner. Process. 93 (2009) 155–164

It is generally accepted that cuprite is normally copper-deficient to some extent (Shuey, 1975; Yoshimura et al., 1976; Ohyama et al., 1997), and it has been argued that the cation vacancies facilitate copper mobility (North and Pryor, 1970). Cuprite is a p-type semiconductor of band gap ~2 eV at ambient temperature (Weichman, 1960), and thus exhibits moderate electronic conductivity; reported values of ambient temperature resistivity for mineral cuprite have been 101–104 Ω m (Harvey, 1928; Shuey, 1975). Accordingly, it is possible that thiol collectors might chemisorb to the copper atoms in the outermost layer of the cuprite surface (which may be oxidised as noted above) in a manner analogous to the situation with chalcocite (Mielczarski and Minni, 1984; Richardson et al., 1984; Buckley and Woods, 1993) and copper metal. Furthermore, for DTP and copper metal or chalcocite, it has been established that at sufficiently high potentials and collector concentrations, multilayer copper dithiophosphate, CuDTP, can adsorb on top of the chemisorbed DTP monolayer (e.g., Chander and Fuerstenau, 1975; Mielczarski and Minni, 1984; Buckley and Woods, 1993; Woods et al., 1993). Given the cation mobility and moderately high electronic conductivity of cuprite, if multilayer CuDTP were able to form as easily on that mineral as on chalcocite, then the presence of Cu oxides in oxidised sulfide ores could lead to higher than expected consumption of collector. Such ores often contain CuII–oxide minerals (e.g., malachite), but there are chalcocite-rich ores in which cuprite is the principal oxide component. In the research reported here, cuprite surfaces prepared by abrasion before and after treatment with DTP solutions at pH values of either 7 or 9 have been characterised by means of conventional XPS, SXPS and partial electron yield (PEY) NEXAFS spectroscopy to establish the extent of collector adsorption on the mineral surface. The emphasis was on relatively high DTP concentrations (10− 2 or 10− 3 mol dm− 3) and extended interaction times to establish whether multilayer adsorption occurred, but short interaction times were included to assess the rate of initial adsorption. No attempt was made to establish the precise conditions that would result in full monolayer coverage without any multilayer formation. Abraded rather than fracture surfaces were considered more representative of those for cuprite particles in a flotation pulp. The most common cleavage plane for cuprite is (111) therefore the outermost layer of a surface prepared by fracture would consist predominantly of O atoms, but a layer of incompletelycoordinated Cu atoms would be very close to the surface in a plane slightly below the plane containing the outermost O atoms. It is expected that other crystal planes such as {001} would also be exposed at the surface of cuprite that had been crushed during comminution. Some analogous DTP adsorption experiments on Cu metal were undertaken as a reference system for the formation of multilayer CuDTP. The abraded mineral and metal surfaces were sufficiently smooth to reveal hydrophobicity on emersion of the specimens from DTP solutions by the obvious non-wettability of the surface after, but not before, treatment with collector solution. This is one of the several advantages of using single piece mineral specimens in such research. 2. Experimental details 2.1. Cuprite mineral and DTP solutions The cuprite used in the investigation was a massive, crystalline specimen from the Red Dome Mine, Chillagoe, Queensland, Australia. Bulk analysis of the mineral by ICP-MS confirmed the high purity of the sample investigated; no minor element was present at a concentration greater than 0.6 mg kg− 1 and only a few had a concentration above 0.1 mg kg− 1. The DTP used in the SXPS and X-ray absorption studies was potassium di-n-butyl DTP (BDTP), whereas that used in the conventional XPS measurements was potassium or ammonium diethyl DTP (EDTP). Establishing a pH of 9 was effected by dissolving the DTP in 0.05 mol dm− 3 sodium tetraborate solution. No residual borate at

surfaces treated with DTP at pH 9 and subsequently rinsed with water was detected by XPS. Because the possible formation of Cu thiolate multilayers was under investigation, treatment times were from 1 to 90 min in DTP solutions of concentration up to 10− 2 mol dm− 3. 2.2. Spectroscopic measurements The SXPS and NEXAFS spectroscopic measurements were carried out in the Australian Synchrotron Research Program (ASRP) soft X-ray spectroscopy end-station while connected to beam-line 24A at the NSRRC in Hsinchu or 14ID at the Australian Synchrotron (AS). BL24A is a bending magnet, wide range beam-line incorporating a grating monochromator. The synchrotron was operated in continuous top-up mode at a stored current of 300 mA. BL14ID is also a wide range, grating monochromator beam-line, but it has an elliptically polarising undulator source. The AS was operated in decay mode at a maximum stored current of 200 mA. The ASRP end-station, constructed by OmniVac and PreVac, is equipped with a SPECS Phoibos 150 electron energy analyser and an OmniVac UHV-compatible partial yield detector based on a multichannel plate behind retarding grids. Electron analyser pass energies of 10 or 20 eV and an electron takeoff angle of ~54o were used for narrow range spectra. The pressure during surface characterisation was b3 × 10− 10 torr. Conventional monochromatised Al Kα XPS measurements were carried out on a VG ESCALAB 220-iXL spectrometer with an analyser pass energy of 20 eV for narrow range spectra. An electron take-off angle of either 20o or 90o was used, the former value to enhance surface sensitivity and the latter to maximise the contribution from the mineral substrate underlying the adsorbed collector. Binding energies determined in this work are reported relative to a Cu 2p3/2 value of 932.6 eV for Cu metal, and binding energies from published studies have been quoted relative to a hydrocarbon C 1s binding energy of 285 eV. Only nominal photon energies (±0.4 eV) are usually delivered by synchrotron beam-lines, and the true photon energy for each spectrum should be established by a supporting measurement. The Cu L2,3-edge NEXAFS spectra were determined together with the absorption current monitored concomitantly by a Cu metal mesh in the beam-line. During operation of the NEXAFS detector, negative grid potentials of 0.75 and 1.5 kV were used for Cu L2,3-edge PEY and TFY spectra, respectively. The indirect determination of X-ray absorption by means of PEY provides NEXAFS spectra that are as surface sensitive as allowed by the kinetic energies of the Auger electrons corresponding to the absorption edge concerned. Thus the PEY mode surface sensitivity associated with Cu L2,3-edge NEXAFS spectroscopy is that corresponding to the Cu LMM Auger electrons which have a maximum kinetic energy of ~920 eV. 2.3. Fitting of core photoelectron spectra The CasaXPS (Version 2.3.5) software was used to fit photoelectron spectra. In almost all cases the background in narrow range spectra was linear. In agreement with the conclusion of Hesse et al. (2007), a Gaussian–Lorentzian sum SGL(p) with Lorentzian character ‘p’ of about 20 typically gave better fits on the low binding energy side than the more usual product, GL(p), used to approximate the Voigt profile. 3. Results 3.1. Spectroscopic data sought to establish collector coverage on cuprite The presence of DTP adsorbed at the surface of cuprite would be revealed in conventional XPS by S 2p, P 2p, O 1s and C 1s photoelectron peaks if each was at the expected binding energy and relative intensity, but differentiating monolayer and multilayer coverage would not be straightforward. Previous XPS investigations

A.N. Buckley et al. / Int. J. Miner. Process. 93 (2009) 155–164

of the interaction of DTP collector with chalcocite have established that the only unambiguous indicator of the presence of multilayer CuDTP on that substrate is an X-ray excited Cu L3M4,5M4,5 peak from the CuDTP which is at a kinetic energy (~915 eV) sufficiently low to be resolved from the corresponding Auger peak from the substrate and adsorbed monolayer. The same indicator should also be applicable for a cuprite substrate, but the separation of the CuDTP and cuprite Cu L3M4,5M4,5 peaks would be smaller than for DTP adsorbed on chalcocite. There is also some doubt concerning the precise Cu L3M4,5M4,5 kinetic energy for multilayer CuDTP, as previously determined values have been based on quite thick multilayers and both the photoelectron and Auger peaks at high coverage can be shifted by charging. For chalcocite, the S 2p doublet from the adsorbed collector is resolved from that for the substrate and the intensity ratio for the two doublets can be used to estimate coverage. The analogous approach for DTP on cuprite involving O 1s instead of S 2p peaks would be applicable in principle if the binding energy of the O 1s peak from DTP were significantly higher than that for the cuprite substrate and for any hydroxide initially adsorbed at the mineral surface when exposed to an aqueous environment, and this is expected to be the case. In practice, the approach would be successful only if the O 1s binding energy for DTP were sufficiently lower than that for any physically adsorbed water; this difference is expected to be no more than ~ 0.5 eV, and hence marginal for differentiation purposes. However, physically adsorbed water should be minimal at a surface rendered hydrophobic by the adsorption of DTP. Any difference in the Cu core electron binding energies for the CuI in cuprite, chalcocite, DTP adsorbed on surface Cu atoms, or multilayer CuDTP is expected to be too small to be useful for distinguishing multilayer from monolayer coverage. A thick coverage of multilayer CuDTP, most probably (CuDTP)n, should also be evident from the S 2p:P 2p:Cu 3s or 3p intensity ratio, as for such a multilayer and the adsorbed DTP monolayer, the S:P:Cu atomic ratio should be 2:1:1. A higher Cu concentration would be indicative of uncovered or insufficiently covered cuprite substrate. However, it is not clear whether the onset of multilayer formation could be identified in this way, especially if multilayer coverage was initially in patches. The actual S 2p:P 2p:Cu 3s intensity ratio for adsorbed multilayer CuDTP would be best determined from the photoelectron spectra for DTP adsorbed on Cu metal with at least one of the photon energies of interest. It is to be noted that the S 2p, P 2p and Cu 3s binding energies are all within a range of 40 eV, therefore for photon energies of at least 500 eV, the analysis depths associated with these 3 peaks would be essentially the same. The O 1s:S 2p:P 2p:Cu 3s:Cu 3p intensity ratios are expected to be slightly different at photon energies less than the 1486.6 eV applicable for conventional XPS because of differences in the relative photoionisation cross-sections. On the basis of theoretical sub-shell photoionisation cross-sections, relative to the values at 1486.6 eV and unity for the Cu 3s, the O 1s:S 2p:P 2p:Cu 3s:Cu 3p cross-section ratios are expected to be 1.4:1.6:1.6:1:1.2 at 1040 eV, 1.7:2.2:2.3:1:1.4 at 800 eV and 2.1:3.1:3.2:1:1.5 at 600 eV (Yeh and Lindau, 1985). Thus, relative to the Cu 3s intensity, as the photon energy is reduced from 1486.6 eV to 600 eV, the O 1s, S 2p, P 2p and Cu 3p intensities would all be expected to increase, but by no more than a factor of ~ 3 at most. Within that photon energy range, the S 2p:P 2p intensity ratio would not be expected to change noticeably because of cross-section changes. For a given photon energy, the S 2p:P 2p intensity ratio should remain constant if there is no decomposition of the CuDTP. 3.2. Copper metal exposed briefly to a collector-free aqueous environment Conventional XPS measurements confirmed that a freshly abraded Cu metal surface exposed briefly to air and water under ambient conditions, but untreated with collector solution, was oxidised although not heavily so. The Cu L3M4,5M4,5 peak (Fig. 1d) was at a

157

Fig. 1. Conventional XPS Cu L3M4,5M4,5 spectrum from: (a) Cu metal treated with EDTP at pH 9; (b) cuprite treated with EDTP at pH 9; (c) cuprite before treatment with DTP; (d) Cu metal untreated with DTP. The peak kinetic energy for multilayer CuDTP is indicated.

kinetic energy of 918.4 eV and the Cu 2p3/2 binding energy was 932.6 eV giving an Auger parameter of 1851.0 eV. These values are in good agreement with the generally accepted values for Cu metal. The Cu 2p3/2 peak could be fitted moderately well with a single symmetrical component of width ~1 eV, indicating that only a very low concentration of CuO or Cu(OH)2, which have Cu 2p3/2 binding energies above 933.5 eV (e.g., McIntyre and Cook, 1975), could have been present at the surface. The Cu 2p3/2 binding energy for Cu2O, the expected principal oxidation product in air, is only 0.1 eV greater than that for Cu metal. The background between the 2p3/2 and 2p1/2 peaks displayed some structure, with low intensity bumps near 944 eV and 946.5 eV that cannot be identified positively but might otherwise have been assigned to CuII excited final state satellites if there had been any evidence for CuII from the Cu 2p3/2 peak. A relatively broad and intense excited final state satellite near 943 eV is normally evident in the Cu 2p spectrum from CuO or Cu(OH)2 (McIntyre and Cook, 1975). The O 1s spectrum could be fitted with a relatively narrow peak at ~ 530.1 eV (40%) due to oxide, and broader peaks at ~531 eV (37%) from hydroxide, and 533.6 eV (23%) arising from adsorbed water (Skinner et al., 1996). The observed alteration of the briefly exposed surface was consistent with previous XPS and NEXAFS spectroscopic data for a Cu metal surface exposed to air for several days under ambient conditions which had indicated that a low concentration of Cu2O as well as some CuO were present (Goh et al., 2006a). Cano et al. (2001) had concluded from their XPS investigation that after exposure for 21 days, Cu2O and Cu(OH)2 were formed. They had determined that for Cu(OH)2, the Cu L3M4,5M4,5 peak maximum was at a kinetic energy of ~ 916.2 eV, and the Cu 2p3/2 and O 1s binding energies were 934.6 eV and 531.7 eV, respectively.

158

A.N. Buckley et al. / Int. J. Miner. Process. 93 (2009) 155–164

3.3. Cu metal treated with diethyl DTP at pH 9 Characterisation by conventional XPS of a freshly abraded Cu metal surface treated with 10− 3 mol dm− 3 EDTP solution at pH 9 for 10 min indicated strong adsorption of DTP. Intense S 2p and P 2p peaks were present (Fig. 2a) and the Cu L3M4,5M4,5 region (Fig. 1a) was dominated by a broad peak near 915.1 eV kinetic energy. A small Cu L3M4,5M4,5 component from Cu metal remained visible at 918.4 eV, and there was some evidence for an unresolved peak at a kinetic energy of ~ 917 eV that would have been consistent with monolayer DTP adsorbed at the Cu metal surface. Thus the collector coverage was clearly a multilayer of CuDTP or (CuDTP)n. It is not clear whether the broad O 1s peak centred at 533 eV (Fig. 3a) contained an unresolved component from physically adsorbed water at ~533.5 eV. A significant concentration of physically adsorbed water would not be expected at a surface rendered hydrophobic by adsorbed DTP or CuDTP, but some water adsorption cannot be excluded on the basis of the O 1s spectrum. The O 1s:S 2p:P 2p:Cu 3s:Cu 3p peak intensity ratio was 5.8:2.5:0.9:1.0:2.8. This intensity ratio would have been close to that for multilayer CuDTP, but since some Cu metal substrate remained within the depth analysed, the Cu 3s and 3p peaks would have been slightly more intense than those for the multilayer alone. The principal Cu 2p3/2, O 1s, S 2p3/2, P 2p3/2, Cu 3s and Cu 3p3/2 binding energies were 932.75, 533.0, 162.5, 133.3, 122.55 and 75.25 eV, respectively; values similar to those observed for EDTP adsorbed on chalcocite. Better fits to the individual spectra could be obtained with an additional low intensity (b10%) component shifted ~ 0.9 eV to higher binding energy than with only the principal component, and it is expected that this shifted component would have arisen from some charging of the thickest part of the multilayer. The first XPS investigation of the adsorption of EDTP on chalcocite established that the O 1s, S 2p, P 2p and C 1s (C–O) binding energies for the

Fig. 3. Conventional XPS O 1s spectrum from surfaces treated with EDTP at pH 9: (a) Cu metal; (b) cuprite.

adsorbed collector were 532.9, 162.5, 133.3, and 286.5 eV, respectively, with the Cu 2p3/2 binding energy essentially unchanged from the 932.5 eV value for chalcocite (Mielczarski and Minni, 1984). These values were in good agreement with those obtained in a subsequent study of the same system in which the Cu L3M4,5M4,5 kinetic energy of 915.4 eV for the adsorbed multilayer was also obtained (Buckley and Woods, 1993). The Cu 2p3/2, O 1s, P 2p3/2 and S 2p3/2 binding energies are similar to those measured for precipitated bulk iso-amyl DTP, viz, 932.8, 532.9, 133.3 and 162.5 eV. An observed Auger kinetic energy ~3.4 eV lower than the value for Cu metal was close to the 3.3 eV difference obtained by Chadwick and Hashemi (1979) for MBT adsorbed on Cu. As expected, there was no clear evidence for a resolved component from the CuDTP in the Cu 2p or Cu 3p spectra. The small component in the Cu 2p spectrum attributed to charge-shifted multilayer would not have been consistent with a significantly higher binding energy for most of the CuDTP. The Cu 2p3/2 peak could be fitted with a principal component of width 1.3 eV while the Cu 3p spectrum could be fitted with a doublet of component separation 2.5 eV and linewidth 2.4 eV. The Cu 2p spectrum was devoid of excited final state satellites confirming the absence of a significant concentration of CuII species.

3.4. Cu metal treated with di-n-butyl DTP at near neutral pH

Fig. 2. Conventional XPS S 2p/P 2p/Cu 3s spectrum from surfaces treated with EDTP at pH 9: (a) Cu metal; (b) cuprite.

A freshly abraded Cu metal surface treated with 10− 2 mol dm− 3 BDTP at near neutral pH for 5 min was characterised by SXPS on BL14ID at the AS. The survey spectra obtained at photon energies of 1100 and 1150 eV included strong peaks from S, P, O and C consistent with a substantial coverage of the adsorbed collector. The Cu 2p spectrum was essentially a single doublet with no obvious excited final state satellites that would have indicated the presence of a CuII species. The Cu 2p3/2, O 1s, S 2p3/2, P 2p3/2, Cu 3s and Cu 3p3/2 binding

A.N. Buckley et al. / Int. J. Miner. Process. 93 (2009) 155–164

energies were 932.7, 532.6, 162.5, 133.2, 122.2 and 75.0 eV, respectively. The Cu L3M4,5M4,5 spectrum (Fig. 4b) appeared to have two components, a broad principal peak near a kinetic energy of 915.0 eV and a narrow one at 917.1 eV but with no obvious component near 918.4 eV from Cu metal. The measured Auger kinetic energy should be independent of true photon energy given that the latter was significantly above the Cu 2p binding energy (but still subject to a possible shift to lower kinetic energies if the multilayer were thick enough to become charged). The presence of the 915 eV peak confirmed the formation of multilayer CuDTP, and the component at 917.1 eV would have been consistent with the presence of monolayer DTP adsorbed at the Cu metal surface. It was established some years ago that the Cu L3M4,5M4,5 kinetic energy for chalcocite (917.2 eV) and for Cu atoms at the surface of chalcocite chemisorbed on DTP were not measurably different (Buckley and Woods, 1993). At a photon energy of 700 eV and for the same spot on the surface, the S 2p:P 2p:Cu 3s:Cu 3p intensity ratio remained essentially constant with time, confirming that any X-ray damage was not severe. However, the intensity ratio did vary for different spots across the surface, indicating that the coverage was not laterally uniform. For the thickest coverage, the S 2p:P 2p:Cu 3s:Cu 3p intensity ratio was 11.7:6.4:1:3.8 and the S 2p3/2, P 2p3/2, Cu 3s and Cu 3p3/2 binding energies were 162.5, 133.4, 122.5 and 75.1 eV. For lower coverage, the binding energies were 162.5, 133.4, 122.5 and 75.1 eV. Relative to the photon energy used in conventional XPS (1486.6 eV), a lower photon energy such as 700 eV would increase the surface sensitivity of any core electrons ejected, but to excite the Cu L3M4,5M4,5 Auger electrons, a photon energy greater than the Cu 2p binding energy is required, and the surface sensitivity of the Cu Auger electrons would be determined by their kinetic energies which are

Fig. 4. Cu L3M4,5M4,5 spectrum determined at a photon energy of 1100 eV from surfaces treated with BDTP at pH 7: (a) cuprite; (b) Cu metal. The peak kinetic energy for multilayer CuDTP is indicated.

159

independent of photon energy. Hence, the surface sensitivity of the Cu Auger electrons is the same in SXPS as in conventional XPS.

3.5. Cuprite surfaces prior to exposure to DTP In cuprite at ambient temperature, each Cu atom is in a linear bonding arrangement with two O atoms at 0.185 nm and is surrounded by 12 Cu atoms at 0.302 nm and by 6 O atoms at 0.354 nm. There is only a single Cu and a single O electronic environment, therefore the Cu 2p and O 1s spectra from the Cu and O atoms in the mineral bulk would be expected to each reflect the one electronic environment. For freshly abraded surfaces of cuprite washed with water and subsequently characterised by conventional XPS, the Cu 2p3/2 binding energy was at 932.7 eV and the Cu L3M4,5M4,5 peak maximum was at a kinetic energy of 916.8 eV (Fig. 1c), so the Auger Parameter would be 1849.5 eV. These values are close to those reported for thin film Cu2O (Ghijsen et al., 1988; Galtayries and Bonnelle, 1995). The Cu 2p3/2 peak was a structureless single component of width ~ 1 eV that could be fitted adequately with an SGL lineshape. The O 1s spectrum consisted of a narrow main peak at 530.1 eV arising from oxygen in the cuprite lattice, and broader, lower intensity components near 531 and 533.6 eV assigned to surface hydroxyl groups and adsorbed water, respectively. Cuprite surfaces prepared by fracture under UHV have been characterised previously by means of surface-enhanced SXPS and PEY NEXAFS spectroscopy using the same ASRP end-station (Harmer et al., 2009). For fully-coordinated atoms below the outermost layer of fresh fracture surfaces, the Cu 2p3/2 and O 1s binding energies were 932.7 eV and 530.15 eV, while the Cu L3M4,5M4,5 Auger kinetic energy was ~ 916.6 eV. In the surface-enhanced O 1s spectrum determined at a photon energy of 610 eV, a small concentration of a low binding energy (~529.2 eV) surface species assigned to under-coordinated O atoms in the outermost layer was evident that was no longer present after exposure of the fracture surface to air. A slightly larger concentration of a surface species with a binding energy at ~531.2 eV (i.e., higher than for the O in the cuprite lattice) was attributed to surface OH. The high binding energy component remained after the fresh fracture surface was exposed to humid air, and a new unresolved O 1s component at an even higher binding energy (~ 532.1 eV) tentatively assigned to Cu(OH)2 was apparent even at a photon energy of 800 eV (Fig. 5c). The migration of Cu to the surface to form a thin layer of Cu(OH)2 would be consistent with the expected mobility of Cu in cuprite. A correspondingly low concentration of a CuII species was not evident in the surface-enhanced Cu 2p spectrum, but as noted in Section 3.8 below, it was discernible in the more sensitive Cu L2,3-edge absorption spectrum from the surface exposed to air. The Cu 2p spectrum from the fracture surface exposed to air displayed only subtle differences from the corresponding spectrum from the fresh fracture surface. The energy of the onset of photoemission and of the first peak in the valence band determined at photon energies near 1000 eV confirmed the relatively small (~2 eV) band gap of the mineral. The Cu 3d peak was situated at a binding energy of ~ 3.0 eV. The valence band spectra obtained at photon energies near 1000 eV were consistent with the corresponding spectra determined at energies of 21.2, 40.8 and 1486.6 eV by Ghijsen et al. (1988) who found the predominantly Cu 3d peak to be at 3.1 eV and the predominantly O 2p peaks to be near 6.0 and 7.3 eV. The Cu 2p3/2 binding energy for cuprite untreated with collector was in good agreement with the value of 932.7 eV obtained by Ghijsen et al. (1988) for Cu2O prepared as a thin layer on Cu metal, but their O 1s binding energy of 530.5 eV was slightly higher than the value for the same cuprite as that investigated in the present work and reported by Harmer et al. (2009).

160

A.N. Buckley et al. / Int. J. Miner. Process. 93 (2009) 155–164

Fig. 5. O 1s spectrum determined on BL14ID for a cuprite surface before or after treatment with 10− 3 mol dm− 3 BDTP for 20 min at pH 7: (a) at a photon energy of 1120 eV after treatment; (b) at 610 eV after treatment; (c) at 800 eV prior to treatment but after exposure to humid air for 1 h.

3.6. Cuprite treated with diethyl DTP at pH 9 Freshly abraded cuprite surfaces treated for periods of up to 20 min with EDTP solutions of concentration up to 10− 2 mol dm− 3 at pH 9 were characterised with conventional XPS. On emersion from DTP solution, all cuprite surfaces were clearly hydrophobic. All spectra indicated the adsorption of DTP, but in no case was there clear evidence for the coverage being much thicker than a monolayer. The Cu L3M4,5M4,5 spectrum (Fig. 1b) was similar to that for cuprite before treatment with EDTP (Fig. 1c) apart from a very low-intensity shoulder near 915 eV on the low kinetic energy side of the Cu L3M4,5M4,5 peak near 916.6 eV kinetic energy. Unlike the situation for chalcocite where a chemisorbed thiol collector would substantially restore the S coordination of surface Cu atoms, it is expected that chemisorption of DTP on cuprite would result in some change to the Cu LMM spectrum as the surface Cu atom coordination would be ‘completed’ with S rather than O atoms. However, the Cu L3M4,5M4,5 peak energy for monolayer DTP on cuprite might not necessarily be the same as that for multilayer coverage. It is conceivable that monolayer DTP on cuprite has a Cu L3M4,5M4,5 peak energy only slightly different from that for cuprite and hence essentially unresolved from the corresponding peak from the substrate. The width of the Cu 2p3/2 peak was slightly less than 1.3 eV (compared with ~ 1.0 eV for untreated cuprite under similar measurement conditions). The increased width could have arisen partly from broader components at binding energies close to that for cuprite from monolayer adsorbed DTP and from any CuDTP that might have been formed. There was also slight asymmetry on the high binding energy side of the principal peak, strongly suggesting a contribution from a low intensity (20%) component with a binding energy ~0.9 eV higher than for cuprite, as well as some suggestion of very low intensity

excited final state satellites between the Cu 2p3/2 and 2p1/2 peaks. It is possible that both these Cu 2p features arose from a low concentration of cupric hydroxide at the surface, but a more likely explanation for the asymmetry would be a slight charge shifting of the component from CuDTP. The S 2p:P 2p:Cu 3s intensity ratio for cuprite treated for 20 min with 10− 2 mol dm− 3 EDTP (Fig. 2b) was 0.73:0.26:1, also indicating that the coverage was not much thicker than a monolayer. Cuprite has a Cu atomic concentration almost the same as that for chalcocite, therefore the P 2p:Cu 3s intensity ratio should be similar for the same coverage of DTP on the two materials. Accordingly, a thick multilayer of DTP on cuprite should give rise to a P 2p peak more intense than the Cu 3s, even at a photon energy of 1486.6 eV. The S 2p and P 2p peaks for the DTP adsorbed on cuprite appeared to be at binding energies similar to those for multilayer CuDTP on copper, as was the C 1s component from C–O at 286.7 eV. The S 2p spectrum was a single doublet, indicating that no detectable concentration of (EDTP)2, or more correctly bis(O,O-diethyldithiophosphoryl)disulfide, was present at the surface. The S 2p spectrum from (DTP)2 should consist of two doublets of equal intensity but significantly different binding energy similar to the corresponding spectrum from diethyl dixanthogen. The O 1s spectrum from the treated cuprite surface (Fig. 3b) displayed a broad and structureless component near 533 eV accounting for ~60% of the intensity; i.e., that was more intense than the peak from the cuprite substrate. For an electron take-off angle of 90o from a surface of uniform composition, ~ 63% of the intensity would arise from the outermost layer of thickness equal to the inelastic mean free path, λ. At a photon energy of ~ 1487 eV, the kinetic energy of O 1s photoelectrons would be ~955 eV, and λ for EDTP would be expected to be ~2.3 nm. Therefore, if all of the higher binding energy O 1s component had arisen from adsorbed DTP, then it would indicate a coverage slightly in excess of a monolayer, but no more than a single layer (on average) of CuDTP on the adsorbed DTP monolayer, notwithstanding the greater O:Cu atomic ratio in CuDTP than in cuprite. As noted above, any physically adsorbed water could also have contributed to that broad component, but a significant concentration of physically adsorbed water would not be expected at a surface rendered hydrophobic by the adsorption of DTP. The O 1s component near 531 eV, tentatively attributed to surface OH groups, still appeared to be present; the spectrum could not be fitted adequately with only two components. 3.7. Cuprite treated with di-n-butyl DTP at near neutral pH Abraded cuprite surfaces treated with 10− 3 mol dm− 3 BDTP solution at near neutral pH for several periods of up to 90 min were characterised by SXPS on BL24A at the NSRRC and on BL14ID at the AS using photon energies from 210 eV to 1120 eV. The Cu LMM spectra were determined at photon energies of 1120 (Fig. 4a), 1050 and 1010 eV, and were not noticeably different for increasing treatment times apart from the shape of the background. In particular, even for a treatment time of 90 min, there was no evidence for a resolved or even partially-resolved Cu L3M4,5M4,5 component near 915 eV, suggesting little more than monolayer coverage of CuBDTP. The Cu L3M4,5M4,5 peak maximum remained at a kinetic energy of at least 916.5 eV. Similarly, there was no noticeable change in the Cu 2p spectrum determined at a photon energy of 1010 or 1050 eV for the various treatment times. For no treatment time up to 90 min were there discernible excited final state satellites arising from CuII nor was there an obvious component on the high binding energy side of the 2p3/2 peak, although for relatively long treatment times, a low intensity component shifted by ≤0.9 eV was required to achieve a satisfactory fit. That minor component most probably would have arisen from a low concentration of CuBDTP that would have been located on top of the BDTP adsorbed on the Cu atoms in the cuprite surface and have become positively charged when irradiated with the photon beam

A.N. Buckley et al. / Int. J. Miner. Process. 93 (2009) 155–164

from the synchrotron. Correspondingly, the S 2p spectrum determined at 210 eV for each treatment time could usually be fitted with one doublet at a 2p3/2 binding energy of 162.5 ± 0.1 eV and a linewidth of 1.1–1.3 eV. However, for long treatment times, a very low intensity doublet at higher binding energy was also required for an acceptable fit. After treatment with BDTP for 20 min, the second doublet was shifted ~1.3 eV to higher binding energy, consistent with the photon flux being higher at 210 eV than above 1000 eV, but accounted for only ~ 6% of the total S 2p intensity (Fig. 6). The second doublet would almost certainly have arisen from a very low concentration of CuBDTP on top of the BDTP adsorbed on the Cu atoms in the cuprite surface. As for interaction with EDTP, there was no evidence for the presence of (BDTP)2 at the surface. The P 2p spectrum obtained at the same photon energy as the S 2p (210 eV) could also be fitted with one doublet at a 2p3/2 binding energy of 133.3 eV with relatively broad (~1.5 eV) components separated by ~ 0.85 ± 0.03 eV. The P 2p:S 2p intensity ratio remained essentially constant for different treatment times. Each O 1s spectrum determined at a photon energy of at least 1000 eV was dominated by two resolved peaks (Fig. 5a) and could be fitted with an intense and relatively broad component (linewidth ~1.6 eV) at ~532.6 eV, a similarly intense but narrower component (linewidth ~0.7 eV) at 530.1 eV, and a low intensity component at 531 eV. The component at 532.6 eV would be consistent with adsorbed BDTP, that at 530.1 eV would have arisen from the cuprite underneath the adsorbed BDTP, and the low intensity component at 531 eV most probably would have arisen from hydroxyl groups remaining at the surface. As discussed in Section 4, it is expected that these hydroxyl groups would be protonated surface O atoms rather than hydroxylated surface Cu atoms. The O 1s intensity from adsorbed DTP relative to that from the substrate increased with treatment time, consistent

161

with the high binding energy component arising predominantly from DTP rather than adsorbed water. The peak height for the DTP component was less than that for the substrate even for a treatment time of 90 min, but the intensity was greater (accounting for up to 60% of the total O 1s intensity) because of the considerably greater linewidth. For a similar photon energy and coverage by BDTP, less substrate would be included in the depth analysed than for EDTP because of the greater length of the BDTP alkyl chain. Thus, even after prolonged interaction of BDTP with cuprite, the surface coverage would not have exceeded a single layer of CuDTP on top of the adsorbed DTP monolayer. That deduction, based in each case on the O 1s spectrum alone, was substantiated by the S 2p:P 2p:Cu 3s:Cu 3p intensity ratio; for a treatment time of 20 min, that intensity ratio was 0.85:0.31:1:3.6. The above assignment of the three contributions to each O 1s spectrum was supported by O 1s spectra obtained at 610 eV (Fig. 5b), a photon energy that would provide considerably greater surface sensitivity for those photoelectrons than photon energies of more than 1000 eV. In each case, the O 1s spectrum determined at 610 eV could be fitted on the same basis as that determined at a photon energy above 1000 eV but, as expected, with the relative intensity of the 530.1 eV component from the cuprite substrate significantly lower in the former than in the latter. For a treatment time of 20 min, the component from the substrate accounted for ~39% of the overall O 1s intensity at 1120 eV but only 20% at 610 eV. Concomitantly, the relative intensities of the components at 531 and 532.6 eV increased, confirming that both were associated with surface species. It is pertinent to note that for a given collector concentration, interaction time and pH, the coverage of CuBDTP would be expected to be greater than that of CuEDTP. The solubility of CuBDTP (pK s = 18.48) is significantly lower than that of CuEDTP (pKs = 14.38) (Stamboliadis and Salman, 1976). 3.8. NEXAFS spectroscopy

Fig. 6. S 2p spectrum determined on BL14ID at a photon energy of 210 eV for a cuprite surface treated with 10− 3 mol dm− 3 BDTP for 20 min at pH 7. The spectrum is fitted with doublets from two components.

The main reason for NEXAFS spectroscopic characterisation of cuprite surfaces before and after treatment with collector solution was to detect the formation of any CuII species. The Cu L2,3-edge absorption spectrum is more sensitive than XPS in detecting the presence of a low concentration of a nominally 3d9 configuration CuII species in the presence of a high concentration of a nominally 3d10 configuration CuI species because of the unfilled CuII d-states (Grioni et al., 1989; Pattrick et al., 1997; Goh et al., 2006a). A CuII–O species is expected to be formed at cuprite surfaces exposed to oxygen and water, and time-of-flight secondary ion mass spectrometry measurements have established that when multilayer CuDTP is formed, minor amounts of CuII(DTP)2 are also produced (Goh et al., 2006b). For cuprite surfaces prepared by fracture under UHV, Harmer et al. (2009) found that the PEY Cu L2,3-edge spectrum consisted of a single L3 peak at an energy of 933.2 eV (0.6 eV above the L3-edge for Cu metal). As expected there was no pre-edge feature indicative of unfilled Cu 3d-states above the Fermi level, i.e., no indication of significant d9 character in the electronic environment of the Cu atoms. This observation is in contrast to some of the Cu L2,3-edge spectra for cuprite reported previously (Grioni et al., 1989; Todd et al., 2003); in those spectra, a peak near 931.5 eV was observed in addition to the principal peak near 933.7 eV. However, the spectrum obtained by scraping cuprite under UHV exhibited no pre-edge feature (Grioni et al., 1992). The PEY and TEY Cu L2,3-edge NEXAFS spectra of cuprite abraded in air and subsequently washed with propan-2-ol revealed a small concentration of CuII at the surface of the mineral by the presence of a low intensity absorption peak at 931.3 eV. There was more CuII evident in the PEY spectrum compared with the TEY spectrum, consistent with CuII only at the surface. It is expected that most of this CuII would be removed from the surface by the collector solution.

162

A.N. Buckley et al. / Int. J. Miner. Process. 93 (2009) 155–164

However, if multilayer CuDTP were formed subsequently, a low concentration of CuII(DTP)2 might also be produced. Also, a CuII–O species (such as CuO initially) could be formed concomitantly: −

Cu2 O þ DTP → CuDTP þ CuO þ e−

ð1Þ

The Cu 2p spectra obtained in both SXPS and XPS had provided no evidence for the presence of a CuII species at the surface of cuprite treated with DTP. Nevertheless, the PEY Cu L2,3-edge spectrum for cuprite treated with BDTP indicated an extremely low concentration of CuII at the surface by a barely detectable absorption peak near 931 eV (Fig. 7b). The main Cu L3-edge peak was at 933.6 eV, an absorption energy 0.4 eV greater than that for cuprite untreated with DTP. The CuII concentration was much lower than that observed for the greater coverage of multilayer CuDTP at chalcocite surfaces under comparable adsorption conditions (Goh et al., 2006b). Thus the Cu L2,3-edge spectrum from cuprite treated with BDTP was consistent with the removal of most if not all surface Cu(OH)2 and the formation of no more than a low coverage of the Cu thiolate. This would indicate that formation of CuDTP from cuprite does not occur by reaction (1) but by: −

−

Cu2 O þ 2DTP þ H2 O → 2CuDTP þ 2OH

ð2Þ

Eh–pH relationships reported by Chander and Fuerstenau (1975) for the copper/water/EDTP system show that CuDTP can be formed by reaction (2) at pH values below 14.8 — p(DTP). Above these pH values, the reverse process is thermodynamically favoured. Notwithstanding the limitations of calculated Eh–pH and Eh–p(DTP) data, it is interesting to note that these diagrams do not show any stability domain in which CuDTP and CuO co-exist and this would suggest that CuDTP is not likely to be formed on a tenorite surface. 4. Discussion For the interaction conditions investigated, the adsorption of DTP on cuprite was clearly evident from the S 2p and P 2p peaks as well as the O 1s component near 532.6 eV. As expected, the S 2p:P 2p intensity ratio remained constant within experimental error as the treatment time was increased. However, the precise extent of coverage was not so obvious. At the onset of multilayer development on cuprite, the CuDTP might be expected to form patches, rather than a layer of uniform thickness, on top of a monolayer of DTP adsorbed on Cu atoms in the mineral surface. Therefore, it is conceivable that even after multilayer patches had been formed, some cuprite substrate peaks might remain observable, albeit with reduced intensity. Notwithstanding that consequence of patch formation, all of the observed spectroscopic behaviour suggested no more than a thin coverage of CuDTP on cuprite.

The main observations suggesting a coverage of little more than a single layer of CuDTP on an adsorbed DTP monolayer, even for the longest cuprite/DTP treatment times, can be summarised as follows. [1] For EDTP adsorbed at pH ~ 9, the Cu LMM peak near 915 eV corresponding to multilayer CuDTP was no more than a lowintensity shoulder on the peak at or above 916.5 eV from cuprite and monolayer adsorbed DTP; for BDTP adsorbed at pH ~ 7, the 915 eV peak was not discernible at all. [2] The cuprite substrate O 1s intensity at photon energies between 1000 and 1487 eV remained comparable with, although slightly less than, that from the adsorbed DTP. [3] There was no evidence for a new component of appreciable intensity on the high binding energy side of the Cu 2p peaks. Although there should be no significant difference between the Cu 2p3/2 binding energies for CuDTP and Cu2O, it is expected that multilayer patches of CuDTP on any substrate would exhibit some charging in SXPS measurements. Similarly, a high binding energy component of appreciable intensity was not required to fit the S 2p or P 2p spectra for the longest treatment times. [4] The P 2p:Cu 3s intensity ratio in conventional XPS was too low to indicate thick multilayer CuDTP or (DTP)2 coverage on cuprite when compared with the corresponding intensity ratio for multilayer CuDTP on chalcocite which has essentially the same Cu atomic concentration as cuprite. The SXPS intensity ratio O 1s(DTP):S 2p:P 2p:Cu 3s:Cu 3p was inconsistent with coverage by a substantial multilayer. [5] Only a very low concentration of CuII was detected by NEXAFS spectroscopy, consistent with no more than a very low concentration of Cu(DTP)2 expected to be formed concomitantly with any multilayer CuDTP. At the onset of multilayer formation, some restructuring of the outermost layers of the mineral surface might occur, possibly to a more tenorite-like structure, and such a structure would be expected to be revealed also as d9-like environments in the Cu L2,3-edge NEXAFS spectrum. The mechanism of the interaction of DTP with the mineral surface is difficult to elucidate, even from surface-enhanced photoelectron spectra. It could be argued that, in principle, in the absence of significant prior oxidation, some information should be available from any change in the Cu:O ratio at the mineral surface following its interaction with the collector. Nevertheless, in practice, that information is sufficient only to eliminate some possibilities. Monolayer coverage of DTP on cuprite could occur by chemisorption to Cu atoms in the mineral surface via an electrochemical process, or by exchange of surface O atoms (which would be bonded to sub-surface Cu atoms). Chemisorption could occur in a way analogous to that for chalcocite (Buckley and Woods, 1993): −

O–Cusurf þ DTP → O–Cu–DTPsurf þ e−

ð3Þ

However, the concentration of Cu atoms at the surface of cuprite that would not be already hydroxylated and available for DTP chemisorption would be expected to be low. Hydroxylation of initially undercoordinated Cu atoms (O–Cusurf) and O atoms (Cu–Osurf) at a cuprite surface could occur by: O–Cusurf þ Cu–Osurf þ H2 O → O–Cu–ðOHÞsurf þ Cu–ðOHÞsurf

ð4Þ

If chemisorption of DTP to some surface Cu atoms occurred, the other surface Cu atoms and surface O atoms would be expected to become hydroxylated. Exchange of oxygen at the mineral surface is expected to occur by exchange of surface hydroxyl groups formed as in reaction (4): Fig. 7. PEY Cu L2,3-edge spectrum determined on BL24A from: (a) Cu metal; (b) a cuprite surface treated for 6 min with 10− 3 mol dm− 3 BDTP solution at pH 7.

−

Cu–ðOHÞsurf þ DTP → Cu–ðDTPÞsurf þ OH

−

ð5Þ

A.N. Buckley et al. / Int. J. Miner. Process. 93 (2009) 155–164

In the exchange mechanism, if it were only the hydroxyl groups bonded to the initially under-coordinated Cu atoms that exchanged with DTP− (but not the hydroxylated initially under-coordinated O atoms) then in both that mechanism and the chemisorption mechanism, the Cu:O ratio in the outermost atomic layers of the substrate would remain the same after interaction with DTP as before. By contrast, if the initially undercoordinated and subsequently hydroxylated O atoms at the cuprite surface, i.e., the outermost layer lattice O atoms, were also or only involved in exchange with DTP−, then the Cu:O ratio in the substrate surface would increase compared with a surface not treated with collector. Determination of the substrate Cu:O ratio for cuprite reacted with DTP would be complicated by any multilayer formed, as even although the O in the mineral can be distinguished from the O in DTP, the Cu in the mineral cannot be differentiated from the Cu in any lowcoverage Cu(DTP) on the basis of the photoelectron spectra. When it is possible for multilayer CuDTP to form, the superposition of a spectrum from any CuDTP, with a Cu:O of 1:2, on a spectrum from cuprite, with a Cu:O of 2:1, could mask any loss of O from the cuprite surface. In other words, any increase in the Cutotal:Ocuprite ratio might be due to an increase in total Cu from the CuDTP rather than from a loss of O from the surface of the cuprite substrate. Thus, the Cu:O ratio would be informative only for no more than monolayer coverage. Under those conditions, the Cu:O ratio did not appear to change significantly following interaction with DTP, but that observation is not sufficient to distinguish between chemisorption and exchange with hydroxyl groups attached to surface Cu atoms. It could also be argued that, in principle, complementary information concerning the cuprite/DTP interaction mechanism should be provided by the OH concentration remaining at the surface after interaction, but again, in practice, the approach is less promising. Any initially under-coordinated Cu or O atoms should remain hydroxylated (with OH or H, respectively) unless DTP had chemisorbed to at least some of the surface Cu atoms or unless the OH− had been exchanged with DTP−. If surface OH remained after adsorption of DTP, the O 1s spectrum (especially if obtained at lower photon energies) should gain a high binding energy component from DTP but retain a distinguishable surface OH component as well as a component from any sub-surface cuprite remaining within the depth analysed. For cuprite surfaces treated with both EDTP at pH ~ 9 and BDTP at pH ~ 7, an OH component is required for an adequate fit of the O 1s spectra. Nonetheless, that observation alone is also unable to differentiate chemisorption from exchange of OH. It is not fully understood why a thick CuDTP multilayer can form on chalcocite or Cu metal but not on cuprite. Although the mobility of Cu in cuprite is reported to be moderately high, and although cuprite is normally Cu-deficient, perhaps the mobility in chalcocite is significantly higher coupled with a greater propensity for chalcocite to form stable phases of lower Cu:S ratio. The resistivity of cuprite would also be slightly higher than for chalcocite; the band gap for cuprite is slightly greater than 2 eV, whereas that for chalcocite is certainly less than 2 eV and possibly as low as 1.2 eV. Rodriguez et al. (1998) established that the reactivity of H2S and S2 at 300 K with metal oxide surfaces, including Cu2O, increased as the band gap decreased. Furthermore, it is possible that the oxygen reduction electrocatalytic activity for chalcocite is greater than that for cuprite. For another thiol collector, MBT, Chadwick and Hashemi (1979) found that at pH ~ 7, no more than monolayer coverage of CuMBT was formed on Cu metal covered with a layer of Cu2O, whereas multilayer coverage formed on unoxidised Cu metal. The slightly greater solubility of cuprite compared with chalcocite within the pH range of interest should not be the reason for the different collector coverage formed, as a chemisorbed DTP monolayer is formed readily and retained at a cuprite surface. Furthermore, it is the less soluble chalcocite on which a multilayer is readily formed rather than the more soluble cuprite, and the solubility of multilayer CuDTP would be the same in each case.

163

The findings for the interaction of DTP with cuprite in near neutral and mildly alkaline solutions are in contrast to the previously reported efficacy for EX as a collector for cuprite, in that low EX concentrations were effective only at pH values above 10 for pulp potentials near zero (Heyes and Trahar, 1979; Senior et al., 2006). Those laboratory studies with EX were carried out with cuprite particles mixed only with high purity quartz. The interaction behaviour of DTP would be expected to be similar to that of EX. It is interesting to compare the behaviour of cuprite in DTP solution with that of chalcopyrite. Güler et al. (2006) used diffuse reflectance infrared Fourier transform spectroscopy to determine the collector species adsorbed during the interaction of EDTP with finely divided chalcopyrite under electrochemical potential control. In slightly acidic solutions, they found chemisorbed DTP but neither CuDTP nor (DTP)2 at moderately oxidising potentials (+150 mV relative to the SHE). The major surface species was found to be (DTP)2 only at high oxidising potentials (+400 mV). They were not able to establish the adsorption products unequivocally in alkaline solutions because of the metal– oxygen species also formed. By contrast, Grano et al. (1997) had observed the onset of multilayer adsorption of dicresyl DTP on chalcopyrite after an interaction time of ~10 min in 10− 4 mol dm- 3 solution at pH 8 in the presence of oxygen. At this pH, the coverage approached 2 monolayers after an interaction time of 60 min. Concerning the assignment of a small contribution from CuII in Cu L2,3edge spectra to the presence of a low concentration of Cu(DTP)2, Goold and Finkelstein (1972) reported that the disproportionation of Cu(DTP)2 to Cu(DTP) and (DTP)2, while rapid in acidic solutions, was slow in alkaline solutions and very slow at pH 10. They concluded that Cu(DTP)2 was formed at the surface of chalcocite, covellite and chalcopyrite in 10− 4 mol dm− 3 DTP solution. Chander and Fuerstenau (1975) subsequently argued that under metastable conditions, Cu(DTP)2 can form only at high electrochemical potentials and high DTP concentrations, and that under those conditions, neither cupric ions nor CuO can form. CuDTP is thermodynamically more stable than Cu(DTP)2, but the latter can be metastable under some conditions. 5. Conclusions Dithiophosphate adsorbs readily on Cu atoms at the surface of cuprite treated with neutral or mildly alkaline solutions of the collector in the presence of air, in contrast to previous studies that have suggested adsorption would occur only at high pH values. The multilayer species CuDTP can also be formed, but even in relatively high concentrations of DTP and for long interaction times, the extent of collector coverage is no more than a thin layer of CuDTP on the DTP monolayer, unlike the situation with Cu metal or chalcocite where a thick multilayer can be formed. This suggests that for the DTP flotation of oxidised Cu sulfide ores containing cuprite, the cuprous oxide should attract no more than monolayer coverage of the collector. Acknowledgements This work was supported by the Australian Synchrotron Research Program, which is funded by the Commonwealth of Australia under the Major National Research Facilities Program. The research was partly undertaken on the soft X-ray beam-line (BL14ID) at the Australian Synchrotron, Victoria Australia. The IWRI author acknowledges funding support for the Australian Mineral Science Research Institute by AMIRA International, the Australian Research Council, and the South Australian Government. The authors are grateful to Bruce Cowie for assistance in the use of BL14D at the AS, Yaw-wen Yang for access to BL24A at the NSRRC, to Allan Pring, South Australian Museum, for cuprite specimen G18365 and to Harold Gallasch, Australian Minerals, for supplying crystalline cuprite from the same mine. The concentration of trace elements in the cuprite specimen

164

A.N. Buckley et al. / Int. J. Miner. Process. 93 (2009) 155–164

investigated was established by inductively-coupled plasma mass spectrometry by the UNSW Analytical Centre. References Attia, Y.A., 1975. Surface chemistry of copper minerals in water. Trans. Inst. Min. Metall. 84, C221–C230. Buckley, A.N., Woods, R., 1993. Underpotential deposition of dithiophosphate on chalcocite. J. Electroanal. Chem. 357, 387–405. Cano, E., López, M.F., Simancas, J., Bastidas, J.M., 2001. X-ray photoelectron spectroscopy study on the chemical composition of copper tarnish products formed at low humidities. J. Electrochem. Soc. 148, E26–E30. Chadwick, D., Hashemi, T., 1979. Electron spectroscopy of corrosion inhibitors: surface films formed by 2-mercaptobenzothiazole and 2-mercaptobenzimidazole on copper. Surf. Sci. 89, 649–659. Chander, S., Fuerstenau, D.W., 1975. Electrochemical reaction control of contact angles on copper and synthetic chalcocite in aqueous potassium diethyldithiophosphate solutions. Int. J. Miner. Process. 2, 333–352. Galtayries, A., Bonnelle, J.-P., 1995. XPS and ISS studies on the interaction of H2S with polycrystalline Cu, Cu2O and CuO surfaces. Surf. Interface Anal. 23, 171–179. Ghijsen, J., Tjeng, L.H., van Elp, J., Eskes, H., Westerink, J., Sawatzky, G.A., Czyzyk, M.T., 1988. Electronic structure of Cu2O and CuO. Phys. Rev., B 38, 11322–11330. 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. Goh, S.W., Buckley, A.N., Lamb, R.N., Woods, R., 2006. The ability of static secondary ion mass spectrometry to discriminate submonolayer from multilayer adsorption of thiol collectors. Miner. Eng. 19, 571–581. Goold, L.A., Finkelstein, N.P., 1972. The reaction of sulphide minerals with thiol compounds. National Institute for Metallurgy (South Africa), Report # 1439. 9 pp. Grano, S.R., Cnossen, H., Skinner, W., Prestidge, C.A., Ralston, J., 1997. Surface modifications in the chalcopyrite–sulphite ion system. II. Dithiophosphate collector adsorption study. Int. J. Miner. Process. 50, 27–45. Grioni, M., Goedkoop, J.B., Schoorl, R., de Groot, F.M.F., Fuggle, J.C., Schäfers, F., Koch, E.E., Rossi, G., Esteva, J.-M., Karnatak, R.C., 1989. Studies of copper valence states with Cu L3 x-ray-absorption spectroscopy. Phys. Rev., B 39, 1541–1545. Grioni, M., van Acker, J.F., Czyzyk, M.T., Fuggle, J.C., 1992. Unoccupied electronic structure and core-hole effects in the X-ray-absorption spectra of Cu2O. Phys. Rev B 45, 3309–3318. Güler, T., Hiçyilmaz, C., Gökagaç, G., Ekmeçi, Z., 2006. Adsorption of dithiophosphate and dithiophosphinate on chalcopyrite. Miner. Eng. 19, 62–71. Hanson, J.S., Fuerstenau, D.W., 1991. The electrochemical and flotation behaviour of chalcocite and mixed oxide/sulfide ores. Int J Miner Process. 33, 33–47. Harmer, S.L., Skinner, W.M., Buckley, A.N., Fan, L.-J., 2009. Species formed at cuprite fracture surfaces; observation of O 1 s surface core level shift. Surf. Sci. 603, 537–545. Harvey, R.D., 1928. Electrical conductivity and polished mineral surfaces. Econ. Geol. 23, 778–803. Hesse, R., Streubel, P., Szargan, R., 2007. Product or sum: comparative tests of Voigt, and product or sum of Gaussian and Lorentzian functions in the fitting of synthetic Voigt-based X-ray photoelectron spectra. Surf. Interface Anal. 39, 381–391.

Heyes, G.W., Trahar, W.J., 1979. Oxidation–reduction effects in the flotation of chalcocite and cuprite. Int. J. Miner. Process. 6, 229–252. Kelsall, D.F., 1961. Application of probability in the assessment of flotation systems. Trans. Inst. Min. Metall. 70, 191–204. Lee, J.S., Nagaraj, D.R., Coe, J.E., 1998. Practical aspects of oxide copper recovery with alkyl hydroxamates. Miner. Eng. 11, 929–939. McIntyre, N.S., Cook, M.G., 1975. X-ray photoelectron studies on some oxides and hydroxides of cobalt, nickel, and copper. Anal. Chem. 47, 2208–2213. Mielczarski, J., Minni, E., 1984. The adsorption of diethyldithiophosphate on cuprous sulphide. Surf. Interface Anal. 6, 221–226. North, R.F., Pryor, M.J., 1970. The influence of corrosion product structure on the corrosion rate of Cu–Ni alloys. Corros. Sci. 10, 297–311. Ohyama, T., Ogawa, T., Nakata, H., 1997. Determination of deformation-potential constants of Cu2O by microwave cyclotron resonance. Phys. Rev., B 56, 3871–3874. Pattrick, R.A.D., Mosselmans, J.F.W., Charnock, J.M., England, K.E.R., Helz, G.R., Garner, C.D., Vaughan, D.J., 1997. The structure of amorphous copper sulfide precipitates: an X-ray absorption study. Geochim. Cosmochim. Acta 61, 2023–2036. Pierson, J.F., Wiederkehr, D., Billard, A., 2005. Reactive magnetron sputtering of copper, silver, and gold. Thin Solid Films 478, 196–205. Rey, M., 1954. Flotation of oxidized ores of lead, copper and zinc. Trans. Inst. Min. Metall. 63, 541–548. Richardson, P.E., Stout III, J.V., Proctor, C.L., Walker, G.W., 1984. Electrochemical flotation of sulfides: chalcocite–ethylxanthate interactions. Int. J. Miner. Process. 12, 73–93. Rodriguez, J.A., Chaturvedi, S., Kuhn, M., Hrbek, J., 1998. Reaction of H2S and S2 with metal/oxide surfaces: band-gap size and chemical reactivity. J. Phys. Chem. B 102, 5511–5519. Senior, G.D., Guy, P.J., Bruckard, W.J., 2006. The selective flotation of enargite from other copper minerals — a single mineral study in relation to beneficiation of the Tampakan deposit in the Philippines. Int. J. Miner. Process. 81, 15–26. Shuey, R.T., 1975. Semiconducting Ore Minerals. Elsevier, Amsterdam. 415 pp. Skinner, W.M., Prestidge, C.A., Smart, R.St.C., 1996. Irradiation effects during XPS studies of Cu(II) activation of zinc sulphide. Surf. Interface Anal. 24, 620–626. Stamboliadis, E., Salman, T., 1976. Solubility product of metal dithiophosphates. Trans. Soc. Min. Eng. 260, 250–253. Todd, E.C., Sherman, D.M., Purton, J.A., 2003. Surface oxidation of chalcopyrite (CuFeS2) under ambient atmospheric and aqueous (pH 2–10) conditions: Cu, Fe L- and O K-edge X-ray spectroscopy. Geochim. Cosmochim. Acta 67, 2137–2146. Weichman, F.L., 1960. Photoconductivity of cuprous oxide in relation to its other semiconducting properties. Phys. Rev. 117, 998–1002. Woods, R., 1996. Chemisorption of thiols on metal and metal sulfides. In: Bockris, J.O'M., Conway, B.E., White, R.E. (Eds.), Modern Aspects of Electrochemistry, No. 29. Plenum Press, New York, pp. 401–453. Woods, R., Kim, D.S., Yoon, R.-H., 1993. The potential dependence of flotation of chalcocite with diethyl dithiophosphate. Int J Miner Process. 39, 101–106. Yeh, J.J., Lindau, I., 1985. Atomic subshell photoionization cross sections and asymmetry parameters. At. Data Nucl. Data Tables 32, 1–155. Yoshimura, M., Revcolevschi, A., Castaing, J., 1976. Thermogravimetric study of the non-stoichiometry of cuprite Cu2O. J. Mater. Sci. 11, 384–386.