2554
Langmuir 1996,11, 2554-2562
Scanning Tunneling Microscopy Studies of Galena: The Mechanisms of Oxidation in Aqueous Solution B. S. Kim, R. A. Hayes, C. A. Prestidge, J. Ralston,* and R. St. C. Smart Ian Wark Research Institute, University of South Australia, The Levels, South Australia 5095, Australia Received December 12, 1994. I n Final Form: March 13, 1995@ A scanning tunneling microscope (STM) was used to study the surfaces of galena (PbS) in aqueous solution. The influences of pH and type of purging gas on the dissolution and oxidation processes at galena surfaces were investigated. STM topographical imaging of the galena surfaces showed the development of sub-nanometer pits with increasing reaction time. Complementary atomic force microscopy (AFM) imaging confirmed the behavior. The depths (z-dimensions)of these pits correspond directly to the unit cell dimensions of galena (0.3 or 0.6 nm) and suggest that the main surface process occurring is congruent dissolution; this has been confirmed by X-ray photoelectron spectroscopy studies. The xy-dimensions of these pits and the rate of their formation depended strongly on the pH and the type of purging gas used. Introductionof Pb2+ions resulted in the formation of lead hydroxide colloids at the galena surface; distinct directionalityis evident. Mechanisms for the initial stages of the surface chemical reactions of galena in aqueous solution are proposed and discussed in the context of flotation separation.
Introduction Galena is generally beneficiated from sulfide mineral ores by the flotation process, where its flotation grade and recovery are controlled by surface chemical reactions. The surface dissolution and oxidation of galena are reported to have a critical influence on both the collectorless1,2and collector-induced flotation2s3behavior. Techniques such as XPS (e.g., refs 4-9) and FTIR (e.g., refs 5 and 9- 11)spectroscopies have been extensively applied to the galena surface, determining the chemical forms of the oxidation products and adsorbed collectors. With respect to surface oxidation, detailed models12have been proposed for the solution phase precipitation of lead hydrolysis products onto galena surfaces and both a qualitative and quantitative knowledge of the surface oxidation products have been gained. Until recently however, little was known about the positional arrangement of surface oxidation species and collector complexes or the precise mechanisms for their formation. Scanning Auger microscopy (SAM)13J4and time-of-flight secondary ion mass spectroscopy (TOFSIMSP5are starting to answer the questions of where and when oxidation products and collector complexes form and the nature of their physical
* To whom correspondence should be addressed. Abstract published in Advance A C S Abstracts, May 15, 1995. (1)Hayes, R. A,; Ralston, J. Int. J. Miner. Process. 1988,23,55. (2) Guy, P. J.;Trahar, W. J. Int. J . Miner. Process. 1984,12,15. (3) Granville, A,; Finkelstein, N. P.; Allison, S. A. Inst. Min. Metall. 1972,C1, 81,784. (4)Buckley, A. N.; Woods, R. Appl. Surf: Sei. 1984,17,401. (5) Laajalehto, K.; Nowak, P.;Pomianowski, A,;Suoninen, E. Colloids Surf: 1991,57,319. (6)Fornasiero, D.; Li, F.; Ralston, J.; Smart, R. St. C. J. Colloid Interface Sei. 1994,164,333. (7) Laajalehto, K.; Smart, R. St. C.; Ralston, J.;Suoninen, E. Appl. S u r f Sei. 1993.64.29. (8) Sub-Kim: B.;'Hayes, R. A.; Prestidge, C. A,; Ralston, J.; Smart, R. St. C. Appl. S u r f Scz. 1994,78, 385. (9) Prestidge, C. A.; Ralston, J.;Smart, R. St. C. Colloids Surf: 1993, @
81 _ _ 10.1 I
(10)Leppinen, J. 0. Int. J. Miner. Process. 1990,30,245. (11) Cases, J. M.; Konogolo, M.; de Donato, D.; Michol, L.;Erre, R. Int. J. Miner. Process. 1990,30,35. (12) Fornasiero, D.; Li, F.; Ralston, J. J . Colloid Interface Sci. 1993, 164,345. (13) Smart, R. St. C. Miner. Eng. 1991,4,891. (14) Bandini, P.; Prestidge, C. A.; Ralston, J.; Smart, R. St. C. In preparation. (15)Brinen, J. S.; Greenhouse, S.; Nagaraj, D. R.; Lee, J. Int. J. Miner. Process. 1993,38,93.
form, but these techniques lack the resolution required to image the initial stages of the processes. Furthermore they are high-vacuum techniques and therefore operate only in the exsitu mode, with its attendant disadvantages. The advent of scanning probe microscopes (e.g., atomic force microscopy (AFM) and scanning tunneling microscopy (STM)), which can topographically image surfaces with atomic scale resolution, enables the initial stages of the various surface reactions to be probed. Galena is an ideal sulfide mineral for scanned probe microscope investigation since it has a face centered cubic structure and occurs naturally in large single crystals which are easily cleaved along (100) planes, forming atomically flat regions. The semiconducting nature of galena (band gap 0.4eV) lends itselfto STM investigation; quantitative real-space and atom-resolved images on a subnanometer scale have been reported by Hochella et a1.l6 and Eggleston and Hochella17 for natural galena samples. STM and XPS have been used in several studies17-19of the air oxidation of galena surfaces. Most recently, Kim et a1.8imaged oxidation production on a pure (99.999%) synthetic galena occurring preferentially at corners, steps, and edges rather than on the (100) faces of the galena surfaces. This result is in accord with the expected increased reactivity of the low coordination sites at corners and edges. XPS has identified the growing products as lead hydroxide and, at longer times, lead sulfate. The rate of formation of the oxidation products is, however, very slow. Comparison with natural galena (>99.8% purity) containing impurity atoms at the 1000 ppm level8 gave a much greater rate ( > 10 times) of oxidation for the natural galena. Oxidation product growth also showed no preference for step, edge, or corner sites with instead indiscriminate coverage of edges and (100) faces. XPS again confirmed lead hydroxide and sulfate as the oxidation products. Impurity sites, probably together with defects (e.g., vacancies, emerging dislocations), are ap(16) Hochella, M. F.; Eggleston, C. M.; Elings, V. B.; Parks, G. A,; Brown, G. E.; Wu, C. M.; Kjoller, K. Am. Mineral. 1989,74,1233. (17) Eggleston, C. M.; Hochella, M. F. Geochim. Cosmochim. Acta 1990,54,1511. (18) Cotterill, G. F.; Bartlett, R.; Hughes, A. E.; Sexton, B. A. Surf: Sci. Lett. 1990,232,L211. (19) Eggleston, C. M.; Hochella, M. F. Science 1991,254,983.
0743-7463/95/2411-2554$09.00/0 0 1995 American Chemical Society
STM Studies of Galena parently responsible for the significantly different oxidation behavior. A previous study by Laajalehto et al.' also reported oxidation products formingon the (100)planes of a natural galena oxidizing in air. In this case, XPS identified lead hydroxide and lead hydroxycarbonate without significant sulfate formation. The growth of oxidation products at different sites was found to occur, in most cases, apparently randomly across the (100)face but, in some cases, ordered directional structures were observed. Other STM studies of galena surfaces extensively oxidized in airI7-l9have shown the formation of discontinuous and nonhomogeneous films. These films are apparently porous. Their growth has been explained in terms of'heighboring atom effects"17in which the initiation of reaction a t a particular site (Le., impurity o r defect) activates adjacent sites for further reaction leading to the formation of small, localized regions of oxidation products as in a nanoscale electrochemical cell. Overlap between adjacent regions has been observed'f' as oxidation continues but diffusion pathways are apparently retained as pores in the final structure of the fully oxidized surface. STM studies have clearly enhanced our understanding of the galena surface but, to-date, have been restricted to studies in air. The validity of these findings with respect to aqueous-based systems can obviously be questioned and there is a clear need for in situ work. Early attempts at in situ STM with conventional STM tips proved difficult because of Faradaic currents which grossly affect image quality and complicate data analysis.20 However, glasscoated tips, which are now commercially available, alleviate this problem, enabling studies to be undertaken in aqueous solution. The initial aim of this work was to develop STM for operation with galena under aqueous solution. Investigations ofthe initial surface chemical processes occurring at the galena aqueous solution interface were made and the role of pH and type of purging gas determined. Findings from the STM investigation were correlated with X P S data.
Experimental Section Reagents. High-punty water was produced by reverse osmosis, two stages of mixed bed ion exchange, two stages of activated carbon, and a final stage involving 0.22-pm filtration. S m-l with a surface The conductivity was less than 0.5 x N a t 20 "C. The pH of the aqueous tension of 72.8 x M sodium solutions was controlled by small additions of hydroxide and nitric acid. These, and any other reagents used, were analytical grade, unless otherwise stated. High-purity nitrogen and oxygen were used to purge the pH controlled water and therefore control the electrochemical potential, &. Natural galena crystals, supplied by Wards Natural Science Establishment (NewYork), originated from the Sweetwater mine (Missouri, USA). Microanalysis (Australian Mineral Development Laboratories) showed this natural galena to be '99.7% pure, with minor impurities which were 1550,360,80, and 900 ppm of zinc, iron, copper, and silica, respectively.
Experimental Techniques STM. The STM instrument was an inverted scanner type designed and constructed by Sexton and Cotterilll' at the CSIRO Division of Materials Science and Technology, Melbourne,Australia. Pt-Ir alloy tips were preferred for imaging mineral galena samples in airz1 and were simply cut at an angle of 45" with wire cutters as recommended by the manufacturer; no electrochemical (20)Nagahara,L. A.; Thundat, T.; Lindsay, S. M. Rev. Sci. Instrum. 1989,60,3128. (21)Sub-Kim, B. Ph.D. Thesis, University of South Australia, 1995.
Langmuir, Vol. 11, No. 7, 1995 2555 etching was required. For use in aqueous solution glasscoated platinum tips (Longreach Sci. Instruments, USA) were used, providing atomic resolution images of HOPG (highly oriented pyrolytic graphite) both in air and in water. Galena samples were mounted on an aluminum stub with conducting carbon cement (Neubauer, Germany), then cleaved using a razor blade under argon. The sample was then flushed with argon to remove any loose material and quickly moved into the STM measuring chamber, which was filled with argon to retard oxidation. After ca. 10 min thermal equilibration under an argon gas atmosphere the sample was advanced toward the STM tip on a quartz tube. Sample movement was carried out with xy-range andz-stage potentials set to 10V, this minimized tip drift during the "advance" or "backoff stages of operation. Advancement was terminated automatically upon detection of a tunneling current. Images were recorded in constant current mode with a tip bias of ca. 0.35 V and a tunneling current of 0.2-0.25 nA. The polarization of the bias has been shown to have negligible influence on the features observed or their variation with time. After tip engagement the sample was moved in the x-y plane until an acceptably flat galena surface could be imaged, the chamber was then opened and a drop of pH controlled aqueous solution, purged with Nz, 02,or air, was placed between the galena surface and STM tip. Consecutive images were taken at appropriate time intervals. Further details of the procedure are reported e l s e ~ h e r e . ~Image , ~ ? ~ ~drifting (e.g., thermal) was experienced and although this occurred at less than 5 n d m i n , the location and relocation of suitable areas for analysis presented considerable difficulties. Several cleavages followed by trials on many areas were often necessary to obtain reliable and reproducible images of specificfeatures. It is noted, however, that the images reported in this paper are fully representative of the particular conditions employed. AFM. Atomic force microscope images were obtained with a Nanoscope I11 (DigitalInstruments, Santa Barbara, CAI in contact mode. A fluid cell enabled the acquisition of in situ images. A cantilever of nominal spring constant 0.58 Nlm was employed. The procedure for setting up galena samples prior to imaging was identical to that employed in STM work. Image Analysis. A Cue 3 (Galai) facility allowed the quantitative analysis of STM images. Images were projected onto a high-resolution monitor from a CCD video camera and then captured by frame grabber for subsequent image analysis. XPS. XPS spectra were recorded using a Perkin-Elmer PHI Model 5100 spectrometer, with a Mg KaX-ray source operating at 300 W. The vacuum pressure in the analyzer chamber was ca. Torr during analysis. The energy scale was calibrated using the Fermi edge and the 3 d ~ 2 line (binding energy = 367.9 eV) for silver, while the retardation voltage was calibrated noting the positions of the Cu 2~312peak (binding energy = 932.67 eV) and the Cu 3 p ~ 2peak (binding energy = 75.13 eV). All measurements were performed at a take-off angle of 45", corresponding to analysis depths with '80% of the signal from < 3 nm (based on attenuation lengths for sulfur 2p photoelectrons as reported by Buckley and Woods4). The initial surface was first examined in survey mode (pass energy of 80 eV) to identify all the elements present and then the various elemental regions were scanned (pass energy of 18eV)in order t o extract informationon chemical bonding and oxidation states. Atomic concentrations for each element were determined from XPS peak areas and
Kim et al.
2556 Langmuir, Vol. 11, No. 7, 1995
Figure 2. Schematic of galena unit cell: large open circles, sulfide; small dark circles, lead.
b
a
I
4
'1 b
Figure 1. STM image of freshly cleaved galena: (a) threedimension plot and (b)top view. (xy scale = 1000 nm a n d z scale = 5.9 nm.)
the relevant sensitivity factor using the method reported by Briggs and Seah.22 Sensitivity factors were, where applicable, derived directly from the XPS spectra of the freshly cleaved galena samples or used as reported elsewhere.23 To preserve the surface chemistry of the galenaaqueous solution interface, wet galena samples were transferred directly to the introduction chamber of the XPS instrument as discussed elsewhere by Smart.I3 Estimates of maximum surface coverage by dissolved galena species (e.g., lead and sulfoxy ions) from the small volume of remaining solution ( 1 nm). STM in Aqueous Solution. By way of an example, the STM image of a galena surface after 60 min of exposure to air-purged water controlled initially to pH 7 is reproduced as a three-dimension plot in Figure 3a; the respective top view plot is given in Figure 3b. The general topographical features observed are very different from those observed for air-oxidized galena, where growth of roughly conical oxidation products has been ~ b s e r v e d . ~ ? ~
STM Studies of Galena
Langmuir, Vol. 11, No. 7, 1995 2557
b:
a:
d:
Figure 4. AFM top view images of galena in air-purged water at pH 7 (xy scale = 500 nm): (a) 20 min (z scale = 1.7 nm); (b) 40 min (z scale = 2.1 nm); ( c ) 60 min (z scale = 1.8 nm); (d) 80 min (z scale = 2.5 nm).
Under aqueous solution the main features appear to be the formation of pits or cavities in the (100) plane of the galena surface and the cleavage lines or dislocations, that were sharp in a freshly cleaved galena sample, become diffuse. These features on the STM images suggest that the process being observed in aqueous solution is galena dissolution, rather than the growth of surface oxidation products. It should be noted, however, that such variables as pH, Eh, and the ratio of the exposed mineral surface area-to-volume of aqueous solution significantlyinfluence the topography of galena surfaces; these factors are further discussed in the subsequent sections of this paper and elsewhere.21 Some comments on the physical significance of an STM image are necessary before embarking on a detailed discussion of the influences of surface preparation conditions on the surface topography of galena. It is important to note that STM images represent a superposition of topographical changes along with changes in tunneling current induced by the work function, 4, of the surface phase. For surfaces with homogeneous surface conductivity, the measured tunneling current (J)is directly related to the surface topography7
J=
e2koVexp(-2k0d) .
9-
-
4fhd
(1)
where V is the applied bias (volts), d is the tunneling distance, and IZo = 0.102544. Changes in J can be due to changes in the work function of different surface regions or phases (e.g., lead hydroxide or lead sulfide) or, at the atomic level, to changes in the work function of different atomic species(e.g., Pb or S).16 When operatingin constant current mode at lower magnification, as in this study, the image relates most directly to topographical features (d). However, it must be noted that for heterogeneous surfaces, 4 cannot be expressed as a function of the xy-direction and there is no simple relationship between tunneling current and surface topography. This was clearly seen in our previous s t u d y which reported the growth of oxidation products (mainly lead hydroxide) on the galena surface in air. The heights of the growths could not be accurately obtained from the STM images. With respect to the present study, it needs to be determined whether the depths of the observed pits in the STM images of galena under aqueous solution relate directly to surface topography or are controlled by surface chemical changes and whether the apparent depths of these pits are real. In an attempt to confirm the validity of the topographical STM images, AFM images of galena surfaces under aqueous solution were obtained. AFM images of a galena surface after exposure to air-purged water at pH 7 are reproduced in Figure 4. In agreement with STM images the AFM images show the development
2558 Langmuir, Vol. 11, No. 7, 1995
a:
Kim et al. 3 relate directly to topography (i.e., the z-scale is correct) and any surface chemical changes have little effect on surface conductivity and hence tunneling current. The formation of these pits or cavities therefore represents dissolution on an atomic scale, with subsequent lead ion solution concentrations undetectable by conventional techniques (i.e., M). These findings suggest that under the conditions employed the only significant process occurring is congruent dissolution, where the solution concentrations of lead and sulfur species would be equal, e.g. n(PbS)
+ 202-(n
- l)(PbS)
+ Pb2++ SO-:
There is no evidence for incongruent dissolution, e.g. n(PbS)
b:
Figure 5. Top view STM images of galena as a function of time in air-purged water a t pH 3 (xy scale = 500 nm): (a) 20 min (z scale = 4.3nm); (b) 60 min (z scale = 3.7 nm); ( c )90 min (z scale = 4.2 nm).
of pits or depressions, the depths of which correspond to the unit cell dimensions ofgalena (0.3or 0.6 nm, see Figure 2). In atomic force microscopy using the contact mode, the surface is scanned mechanically at constant tipsurface pressure. The image obtained a t magnifications used here represents the surface topography and is unrelated to the conductivity of the surface phases. The observed correlation between the STM and AFM images (particularly the quantitative agreement in the depths of the pits) leads us to believe that the STM images in Figure
+ 1/202 + H20- Pbn-,Sn + Pb2++ 20H-
(3)
where sulfur-rich, lead-deficient phases are produced; these would be of a lower surface conductivitythan galena and should therefore be observable in the STM images. Such phases have, however, been reported in recent SAM14 and STM21studies where significantly different ratios of the exposed mineral surface area to volume of aqueous solution are employed. In sulfide mineral flotation practice, where the ratio of the exposed mineral surface area to volume of aqueous solution is approximately IO3 times greater than in the present work, such hydrophobic phases are thought to be responsible for collectorless fl0tation.l Influence of pH. STM images determined as a function of time were taken a t pH 3, 6, and 9; these are reproduced in Figures 5,6, and 7, respectively. The only topographical features observed were the formation of dissolution pits, again with z-dimensions corresponding to the unit cell of galena, or multiples of this. The x j dimensions of these pits, the number of pits per unit area, and the rate of pit growth are dependent on pH. The overall dissolution rate appears to decrease with increasing pH, a finding that is in agreement with reported dissolution studies on galena particles (e.g., ref 24). Interestingly, for all pH values studied, the growth of the dissolution pits is significantly greater in the xpdirection than in the z-direction, suggesting that the edges or dislocations are more active toward dissolution than the faces. Step edges and corners have atoms with incomplete coordination are therefore more reactive toward oxygen adsorption and, hence, dissolve preferentially. However, impurity sites may also have enhanced reactivity and affect the position and rate of dissolution. Image analysis of the STM images has been employed in an attempt to quantify the observed dissolution process. For any particular STM image of a galena surface, the number of dissolution pits, their area, and depth were determined by image analysis. The number of equivalent galena monolayers dissolved was then determined from the following algorithm number of lead sulfide (Pb-S) layers dissolved = i=n
where Ai is the area of the ith dissolution pit of depth, di, At is the total area analyzed, and dpbs is the Pb-S layer thickness, i.e., 0.3 nm. Figure 8 shows a plot of the number (24) Hsieh, Y.H.; Huang, C. P. J. Colloid Interface Sci. 1989, 131,
537. \
STM Studies of Galena
Langmuir, Vol. 11, No. 7, 1995 2559
a:
a:
b:
h*
C:
Figure 6. Top view STM images of galena as a function of time in air-purged water a t pH 6 (=cyscale = 500 nm): (a) 40 min (z scale = 3.8 nm); (b) 60 min (z scale = 2.6 nm); (c) 120 min (z scale = 2.8 nm).
of equivalent galena monolayers removed as a function of time at pH values 3, 6, and 9. To rationalize the pH dependency of galena dissolution, Hsieh and H ~ a n proposed g ~ ~ a mechanism where surface protonation occurs initially as PbS
+ 2H+ - PbSH?+
Figure 7. Top view STM images of galena as a function of time in air-purged water at pH 9 (=cy scale = 500 nm): (a) 30 min (z scale = 3.6 nm); (b) 60 min (z scale = 2.9 nm); (c) 120 min (z scale = 2.6 nm).
PbSH:+
+ 20, - PbSH;+*(O2),
then the dissolution step
-
+
PbSH,2+*(02), Pb2+ SO-: (5)
Under aerobic conditions, this is followed by oxygen adsorption
+ 2H+
(7)
Under air-purged conditions the oxygen concentration in the aqueous solution will be relatively high, the rates of reactions 6 and 7 are therefore controlled by the
2560 Langmuir, Vol. 11, No. 7, 1995
Kim et al.
a:
P
0
20
40
60
80
100
120
Time/minu tes Figure 8. Number of equivalent lead sulfide monolayers removed from the galena surface as a function of time in airpurged water, at different pH values.
hr
e
E
Oxygen purged Airpurged Nitrogenpurged
*O 20 a .o
0
20
40
60
80
100
120
Time/minutes Figure 9. Number of equivalent lead sulfide monolayers removed from the galena surface as a function of time in water at pH 7, with different purging gases. C:
concentration of PbSH22+,which is dependent on the protonation step (reaction 5). Hence the rate of dissolution will be dependent on pH, as shown semiquantitatively in Figure 8. The above mechanistic sequence is only fully applicable to a pure galena sample. For a natural galena sample, trace impurities may affect the surface chemical reaction pathways.8 The galena studied in this work contains iron, zinc, copper, and silica impurities. A simple geometric calculation, based on a three-dimensional random distribution of the impurities, shows that a 500 nm by 500 nm area of the natural galena would include ca. 1250 zinc, 800 silica, 300 iron, and 60 copper species. Furthermore, it can be shown21that the oxidative dissolution of the sulfides of these metal impurities is thermodynamically more likely than the equivalent process for galena. For example, the molar free energy change for
is more negative than for reaction 2. Therefore, oxidation and dissolution reactions at these impurity sites will be thermodynamically more favorable than those a t pure galena sites and may activate adjacent lead sites for subsequent oxidation. A comparison of the dissolution behavior of synthetic and natural galena samples may increase our knowledge in this area and is a potential area for further work. It may be hypothesized that the edges are the most energetic sites on the surface of a pure galena sample, having lead and sulfur species with incomplete coordination spheres which should therefore be more reactive toward oxygen adsorption and, hence,
Figure 10. Top view STM images of galena treated with M lead(I1) ions at pH 7, as a function of time (xy scale = 500 nm): (a) 10 min (z scale = 11.2 nm);(b) 30 min (z scale = 13.7 nm); (c) 50 min (z scale = 15.4 nm).
oxidative dissolution. However, on natural galena samples impurity sites are also likely points for the initiation of oxidation. Influence of Purging Gas. The STM images presented so far have been determined in air-purged aqueous solution. A further set of STM images was determined at pH 7 and as a function of the type of purging gas. Under nitrogen purging considerablyfewer dissolution pits were observed compared with air or oxygen purging, i.e., the rate of the dissolution is reduced by nitrogen purging.
STM Studies of Galena
Langmuir, Vol. 11, No. 7, 1995 2561
Table 1. Surface Atomic Concentrations of Galena Surfaces As Determined by XPS” surface atomic concentration, % galena sample freshly fractured under aqueous solution at pH 7 30 min conditioning at pH 7 60 min conditioning at pH 7 120 min conditioning at pH 7 treated with M Pb2+for 30 min at pH 76
C (1s)as carbonate
Pb (40
s (2P)
0 (1s)
‘0.1
44.1 44.8 45.3 44.5 37.8
45.3 42.6 41.6 41.3 28.8
10.6 12.6 13.1 14.2 31.5
