, THE GANADIAN MINERALOGISTJournal of the Mineralogical Association of Ganada

Yolume 25 SEPTEMBER 1987 Part 3

Canadian MineralogistVol. 25, pp. 381-392 (1987)

ABSTRACT

The geometry of the galena surface and the elechonicstructure of the surface region as compared to the bulk inPbS are discussed using literature data, along with new cal-culations and spectroscopic measurements. The latterinclude calculations on a Pbs6lr cluster using the MS-Xa method, valence-region X-ray photoelectron spectra andsulfur LMM Auger spectra of galena crystals, The avail-able evidence suggests a model involving little or no relax-ation,of surface atoms but a reduction in the band gap atthe surface (from -0.4 eV in the bulk to -0.1 eV). Pbor S vacant sites may produce holes in the valence bandand bound states below the conduction-band edge, respec-tively. The considerable number of studies of the reactivityof the galena surface (oxidation in air and inaqueous media)

-are reviewed in the light of the above models. Orygenchemisorption at the galena surface, in which the oxygenbehaves very similarly to sulfur, appears to be an impor-tanl mechanism in galena oxidation, modifying the elec-tronic structure of the valence band at the surface andaffecting conductivities and reaction rates. Although a con-sistent picture of the geometry and electronic structure ofthe (100) surface of galena is emerging, and of the impor-tant effects caused by nomtoichiometry, further calcula-tions involving adsorbed species will be needed for a morecomplete understanding.

Keywords: galena, surface, electronic structure, reactivity,nonstoichiometry, conduction-band edge, chemisorp-tion of oxygen, mechanism of oxidation.

Som,rann

Les aspects g6om6triques de la surface de la galdne etla structure dlectronique de cette r6gion, en comparaisondes propri6t6s globales du PbS, font I'objet d'une discus-

381

ELECTRONIC STRUCTURE AND THE CHEMICAL REACTIVITYOF THE SURFAGE OF GALENA

JOHN A. TOSSELLDepartment of Chentistry, (.Iniversby of Marytand, College park, Maryland 20242, U,S.A.

DAVID J. VAUGHANDepartment of Geological Sciences, Aston University, Birmingham, England, M 7ET

sion critique, i la lumidre de calculs et de mesures spec-troscopiques nouveaux. On pr6sente les rdsultats de calculspour le groupement PbS6l0- par la m6thode MS-Xa, parspectroscopie photo6lectronique des rayons X pour lar€giondes valenbes, et par spectroscopie Auger pour le soufre dansdes cristaux de galbne. Les donndes disponibles indiquentun modEle qui implique trb peu ou pas de relaxation desatomes a la surface, mais par contre une diminution dansI'6cart entre les bandes, d'environ 0.4 eV dans le volumede PbS d environ 0'l eV. Les lacunes dans les sites Pb ouS pourraieqt indufue des trous dans la bande de valence etdans les €tats en liaison en-dessous de la limite de la bandede conduction, respectivement. On passe en revue les 6tu-des nombreuses de la r6activitd de la surface de la galbne(orydation dans I'air a dans les milieux aqueux) i lalumi&redes modbles cit8. La chimiosorption i la surface de l'ory'gbne, qui se comporte de fagon tr&s semblableau soufre,semble Otre un m6canisme important de foxydation de lagallne, en modifiant la structure €lectronique de la bandedes valences et en affectant i la fois conductivitd et tauxde r6action. Quoiqu'un corutersus i propos de la g6om6-trie et de la structure 6lectrodque de la surface (l@) et deseffets importants de la non-stoechiom6trie est en train dese d6gager, des calculs plus pouss6s seront n€cessaires pourmieux comprendre ces ph6nombnes.

Ctraduit par la Rddaction)

Mots-cVs: galbne, surface, structure €lectronique, non-stoechiomdtne, 6cart des bandes 6lectroniques, chimio-sorption d'oxygbne, m6canisme d'oxydation'

INTRoDUCTIoN

Galena, the major ore mineral of lead, has beenextensively itudied with regards to electronic struc-ture because of its importance €rs a detector of

382 THE CANADIAN MINERALOGIST

infrared radiation. There have also been extensivestudies of its solution chemistry because of theimportance of hydrometallurgical processing, and ofsurface alteration effects because of relevance to flo-tation methods of mineral extraction. However, lit-tle attempt has been made to examine the relation-ship between electronic structure and the surfaceproperties of galena. It is, therefore, timely to con-sider the properties of galena that depend on sur-face behavior in the light of the most recent modelsof its electronic stnrcture.

In this paper, experimental and theoretical evi-dence for the arrangement of atoms at the surfaceof galena, and for the electronic structure of PbSin bulk and at the surface, is presented and drscusse.d.Experimental data on the reactivify of the galena sur-face exposed to oxidation in air and aqueous mediaare briefly reviewed and discussed in the light of cur-rent views on electronic structure. The aim is topresent preliminary observations on this complextopic and to emphasize the importance of a unifiedapproach.

GEoMETRY oF TlrE Suuace or Gatnue

Galena has the cubic rocksalt structure, with Pbin perfect octahedral co-ordination to S. It has alsoperfect cubic cleavage {100}, along the planes thatbreak the smallest number of cation-anion bonds.It can, therefore, be assumed (as a first approxima-tion) that the crystallographic surface of galena mostlikely to be exposed to attack is this (100) surface(or the equivalents). A simple model of this surfaceis shown in Figure l, based on the work of Henrich(1983). Here the surface cations have S-fold co-ordination [whereas for (1 10) and (1 I l) surfaces thecation co-ordination is 4- and 3-fold, respectivelyl.In presenting this model of the perfect surface ofgalena, we have assumed no relaxation of the sur-face atoms. On the basis of experimental evidence(Kahn 1983), this appears reasonable for the more"ionic" sulfides, but not for more covalent species,such as ZnS (Duke 1983a).

We can also explain the lack of relaxation in PbScompared to the substantial relaxation observed in

[oro]

FIc' l. Model of the galena (100) surface (after Henrich 1983). Large circles are S anions, and small circles are Pbcations. A { 100} step to another (100) terrace is shown, as are both missing anion and missing cation point-defects.

ZnS @uke 1983b) usine qualitative molecular-orbital(MO) theory. In a covalent II-VI compound suchas ZnS, the (110) surface is reconstructed with amovement of the Zn atoms inward (toward the bulk)and the S atoms outward. Harrison (1976, 1980) hasexplained this in terms of a conversion of half-occupied dangling-bond hybrid orbitals atboth Znand S on the reconstructed surface to a combinationof a fully occupied bond-orbital on S and an unoc-cupied orbital on Zn for the reconstructed surface.The reconstruction stabilizs theorbital at S and con-verts it from sl to s in type whereas the Zn orbitalis destabilized and converted from sp: to spz. 4nequivalent description within the delocalized MOapproach would relate the Zn site to a Gvalence elec-tron AB3 species, expected to be planar as in BH3,and the S site to a 8-valence electron AB, species,expected to be pyramidal as in NH3 (Burdett 1980,Tossell 1984). The deeree ofdistortion will be depen-

REACTIVITY OF THE SURFACE OF GALENA 383

dent upon the change in energy of the surface orbi-tal upon distortion in the angle B-A-8. This willdepend partly upon the difference in electronegativityof the group-Il and group-Vl atoms, with a smallerelectronegativity difference giving a larger degree ofdistortion from the urueconstructed geometry. ThusZnS suffers a larger reconstruction than ZnO. Thissimple qualitative model is consistent with the resultsof accurate quantum-mechanical calculations (Swartset al. 1980), which show for the GaAs surface areconstruction that gives a geometry much like thatof the corresponding trihydrides GaH, and AsHr.An important aspect of this distortion is the stabili-zation of the A-centred highest occupied molecular-orbital (HOMO) of the electron-rich 8-electron AB,species by mixing with orbitals on B. If a geometri-cal reconstruction ofthe surface is unable to give thisstabilization, then reconstruction will not be favored.For a sulfide with the rocksalt structure. we can con-

20Binding Energy (eV)

FIc. 2, X-ray photoelectron (XPS) spectrum of PbS obtained in the present study'Data were obtained from a natural crystal using a Kratos XSAM 800 instrumentand Mgl(a as the exciting radiatron.

oro

384 THE CANADIAN MINERALOGIST

sider the geometry of S on the unreconstructed sur-face to be Doo square pyramidal and consider theeffect of moving the S inward to produce a Duotrigonal bipyramid. Assuming a charge separationof M+S-, f.e., converting the dangling bond-electrons on M and S into an empty orbital on Mand a bond pair on S (as for ZnS above), we obtainthe analogue of a l2-electron AH, species at S.However, such l2-electron AH5 species are expectedon qualitative MO grounds to be most stable insquare-pyramidal geometries (Burdett 1980, p.lGl02). Thus covalent orbital mixing-factors do notfavor reconstruction of the (100) surfaces of com-pounds having the NaCl structure, consistent withthe lack of surface reconstruction in PbS.

The surface of the real material would containnumerous det'ect-sites and could never be modelledin detail. However, fts simplest defects, namelythose arising from removal of S or Pb atoms, canbe considered. Removal of such surface atomst'roduces changes, for example in co-orr{ination andin the screening between cations, and in transfer ofelectrons to retain charge neutrality. A surfacevacancy in an S site creates a symmetrical cluster offour four-fold co-ordinated cations in the surfaceplane below. A surface vacancy in a Pb site createsa symmetrical cluster of four sulfur anions in four-fold co-ordination to Pb and one again in five-foldco-ordination in the plane below (both kinds ofdefects are shown in Fig. l).

Surface steps would be expected on cleaved sur-faces of galena, and the most stable type would bea { 100} step on the (l@) surface as shown in Figure1. The geometry of cations on the step edges is differ-ent from that on the perfect surface or at pointdefects. All step-edge cations have four-fold S co-ordination, but unlike the four-fold cations sur-rounding an S-vacancy, they are still well-screenedfrom other cations. Similarly, of course, all step-edgeanions are 4-co-ordinate.

ELEcTRoNIc STRUC"TURE oF BULK PbS

The electronic structure of PbS in the bulk hasbeen studied using a wide range of experimental tech-niques and theoretical models. Optical spectra werereported by Cardona & Greenaway (1964) andSchoolar & Dixon (1965), and high-resolution X-rayphotoelectron spectra (XPS, which involves meas-uring the kinetic energies of electrons ejected by bom-barding with monoenergetic X-rays), by McFeely e/al. (1973). The theory and experimental methods ofXPS have been extensively reviewed elsewhere (e.9.,Feuerbacher et al.1978, Urch 1985). The XPS dataobtained in the present study are shown in Figure2. A Kratos XSAM 800 ESCA-Auger spectrometerwas employed with Mg/fd as the exciting radiation.The pressure inside the experimental chamber was

- l0-s Torr. In line with the work of previousauthors, the peak labeled I at the top of the valenceband is interpreted to arise essentially from sulfurp orbital nonbonding molecular-orbitals. Below theselie the main 6eading-orbitals of the system, whichare chiefly lead 6slsulfur 3p in character (peak 2);peak 3 represents sulfur 3s orbitals not involved inbonding. The intense (double) peak below this arisesfrom the l*d 5d orbitals.

In the present study, a calculation has been per-formed on a PbS6-10 cluster unit using the MS-Xamethod. This method, which has been applied to avariety of metal sulfide systems, has been welldescribed with regards to general principles (John-son 1973, Slater 1974) and applications to mineralsystems (Vaughan et al. 1974, Sherman 1985). Itinvolves selecting a "molecular" cluster unit (in thiscase Pb2+ surrounded by six Sz- at the sorners ofan octahedron); the space within and around thecluster then is geometrically partitioned into atomicsphere regions centred on tJre metal and anion nuclei,enclosed within an outer sphere, beyond which is anextramolecular region. Within the outer sphere butbetween atomic sphere regions is the interatomicregion. The electrostatic potential of the remainingatoms in the crystal lattice is approximated by aspherical shell of positive charge (the Watson Sphere)passing through the anion nuclei. The potentialenergy in the various regions is evaluated electrostat-ically and then used to solve numerically the one-electron Schrddinger equation in each region. Theresulting wave-functions and their first derivativesare joined continuously through the various regionsusing multiple-scattered-wave theory. From thisresult, the spatial distribution of electron density iscalculated and used to generate a new potential forthe next iteration. The entire procedure is repeateduntil self-consistenry is attained; the result is a setof one-electron molecular orbitals characterized bytheir energies and distributions of electron density.The results of the calculation are shown in Figure3 as an energy-level diagram. Here, molecular orbi-tals of the PbSrto- cluster are labeled 4ssslding tothe irreducible representations of the 01 symmetrygroup under which they transform. The (completelyfilled) lead 5d orbitals are calculated to lie -21 eVbelow the valence band, in good agreement with theresults of XPS studies. The orbitals with eigenvaluesaround -12 eV are essentially S 3s nonbonding orbi-tals. The main bonding-orbitals of the system withappreciable lead and sulfur character are thev1g,t6,t2g and eu orbitals, which lie direclly abovethese in energy. At the top of the valence band liea group of orbitals of t2u, t1o and tk symmetry,closely spaced in energy, that are essentially S 3p non-bonding orbitals, with relatively little metallic charac-ter. However, the topmost filled orbital is antibond-ing and does have appreciable lead (Pb 6s) character.

REACTIVITY OF THE SURFACE OF GALENA 385

Ihe empty orbitals above this in energy (of th, t2sand a6 symmetry) are antibonding orbitals witliappreciable lead and sulfur character. The neglectof relativistic effects in this calculation leads to exag-geration of the separation between the topmost filledorbital and remaining filled orbitals, and of thestabilization of the Pb 5d levels. Relativistic effectscan be estimated from calculations on the free leadatom, data for which are shown in Table l. Thesewill stabilize the a1n highest occupied MO(predominantly Pb 6s in character) by about 2.86 eV,bringing it close to the es and t" occupied levels.The t2* and eu levels of Pb 5d

-character will be

s1alilized by about 4.1 eV.Further improvement in the correspondence

between calculated and measured energy-levels isachieved by making use of the transition-state proce-dure (Slater 1974'), n which an electronic configu-ration halfway between the initial and final states isiterated to self-consistency. When this procedure,which takes into account any orbital relaxation thatoccurs during the transition, is utilized, the effect isto insrease the relative ionization-potentials of highlycontracted levels, such as the metal d levels, com-pared to less localized levels, such as the ligand plevels. Takingtlese considerations into account, theresults of this calculation are generally in good agree-ment \yith the data from XPS studies outlined above.

Amongst other calculations of tle electronic struc-ture of PbS in bulk can be mentioned the band-structure calculations of Herman et al. (1968) andof Tung & Cohen (1969), which are based on theOPW method, and the larger MS-Xcl cluster calcu-lations of Hemstreet (1975). The overall electronicstructure of bulk PbS based on these calculations(and on experimental data) is illustrated in Ficure4. It again shows that in PbS, a topmost (filled) bandin the valence region is dominantly S 3p nonbond-ing in character and lies above a band comprised ofthe main bonding-orbitals of the system, with Pb 6s-S 3p character. More deeply buried are the nonbond-ing S 3s levels and (at -20 eV binding energy) thePb 5d orbitals. A gap of around 0.4 eV separatesthe top of the valence band from the conductionband of largely Pb 6s - S 3p antibonding character.

Er,BcrnoNrc Srnucn-lns op PbS Sunracss

Several spectroscopic studies have been undertakenthat provide information on the electronic structureof a "clean" cleavage-surface of FbS. Grandke &Cardona (1980) studied the UV photoemission of a(lfr)) surface of both n- and p-type PbS, andGrandke et al. (1979) used angle-resolved UV pho-toemission to study temperature effects on thevalence-band levels in PbS. Both UV and soft X-rayphotoemission were used 1o examine core andvalence-band spectra by Hagstrdm & Fahlrnan(l978).

Frc. 3, Molecular orbital energyJevels calculated for aPbS6l0- cluster using the MS-Xa method. MolecularorbiUls are labeled 4ssolding to the irreducible represen-tations of the 01 symmetry group with filled orbitalenergy-levels shown as solid lines, empty as dashed lines.

Auger electron $pectroscopy is one of the mostwidely employed spectroscopic methods in surfacestudies (e.g., Carlson 1975, Lichtman et 41. l98l).It involves the decay of an initial core-orbital hole(formed by X-ray bombardment) by a process inwhich a higher-energy electron drops into the corehole, and tle resultant energy, rather than beingemitted as X-rays (as in X-ray emission spec-troscopy), is transferred to a second higher-energyelectron, the Auger electron, which is ejected. In thepresent work, the S"ry Auger spectra of clean sur-faces of galena have been examined. In Figure 5 areshown the differentiated $pectrum and a computerfit of three peaks of the undifferentiated spectrum,at kinetic energies of 140, 145,2 and 145.9 eV'respectively. The widths and normalized intensitiesof these features are also given in Figure 5. Com-parison with the X-ray photoelectron spectrum @ig.2) and the rqults of calculations described and illus-trated above (Figs. 3, 4) suggest that the peaks atL49.5 , 145 .2 arLd 140 eV arise from the sulfur 3p non-bonding orbitals, lead 6s - sulfur 3p bonding orbitals and the dominantly sulfur 3s nonbonding orbi-tals, respectively.

In his review of transition-metal-oxide surfaces,Henrich (1983) noted, on the basis of experimentalevidence and calculations, that the electronic struc-ture of five-fo16 se-qldinnted surface cations is verysimilar to that of bulk cations in SrTiO, and TiOr.Comparison of the spectroscopic evidence notedabove with calculations of the electronic structureof bulk PbS also suggests that the changes are small.

386 THE CANADIAN MINERALOGIST

TABLE I. RSLATMSTIC EFFECTS ON THE ATOMIC ORBITAL ENERCIES

OF THE FREE LEAD ATOM

Pb Atomlc orbltd NoreLativlstic€ Relatlvbtlc(av*age)b

llu, level is considerably depressed, and that of12u,, raised. The effect is, therefore, to substantiallyreduce (by -2 eV) the gap between highest-energyfilled and lowest-energy empty levels. This predic-tion for the electronic structure of a clean MgO (100)surface is supported by experimental evidence fromelectron-energy-loss spectroscopy (see Tsukada et al.1983).

By analogy of the work done on MgO (100) sur-faces, the electronic structure ofthe perfect PbS (100)surface might be expected to show similar differencesto that of the bulk. Hence the dominantly Pb 6s, 6pconduction levels would be expected to be loweredin energy relative to the top ofthe valence band, lead-ing to a reduction of the band gap. This is also inaccordance with arguments put forward by Levine& Mark (1960 on electrostatic grounds, whichpredict a narrowing of the gap at the surface in suchcompounds.

Experimental evidence bearing on differences inthe electronic structure of bulk and (100) surfacesof PbS is provided by Grandke & Cardona (1980).From the study of the binding energies of core levelsusing UV photoemission spectroscopy, they havesuggested a reduction in the band gap from 0.4 eVin the bulk to less than 0.1 eV at the surface forfreshly cleaved specimens examined in vacuum.These authors also observed changes in the spectraof PbS held under vacuum for 3 hours; theyattributed these changes to the development of a sur-

5.17Jr.0t

2.46

5p!6s

98.33tt32

103.504.24t5.37

a Daia lrom Men (1967).

b. Datslrom Dsdaux(1973).

However, calculations byTstkadaet oL (1983) usingthe DV-Xa cluster method to model the electronicsxructure of the surface of MgO (a phase isostruc-tural with PbS) did show some significant differences(see Fig. 6). Compared with an (MgOo)ttr clusterused to model the electronic structure of bulk MgO,the five-co-ordinate surface-species cluster (MgOr)s-showed the (full) valence-region levels Qareely O 2pin character) to be shifted to higher energy and sub-stantial shifts to occur in the (argely Mg 3s, 3p)empty levels, which overlie them. The remarkablelowering of the llu, level can be explained by thedistortion due to the surface field. Because of thefield normal to the surface, the Mg 3s and 3p orbi-tals are admixed to form the (3s + 3p)-like llu,level and the (3s-3pJJike l2ut levels. Since thepotential decreases toward vacuum, the energy of tJre

(c)la)

1$TF;SFS\ - ,'l

NSplt".beqal'p /r (

[F$dfi5lTqTi.g\ '-

\ \ -\ -\-\ \f f i \ \ \

[S3{ngqb9[d!ng \ \

\ \ \ 2

r"5r \\>(\ \ \l_nonbondlng-

\ \\\ , , ,

3 t

N O L A T ARaduced WqveVeclor

FIG, 4. Electronic structure models for galena: (a) simplistic band-structure representation based on spectroscopic data;(b) molecular orbital energy-levels calculated for a SPb6 S12Pbs cluster using the MS-Xa method (after Hemstreet1975) and with features from the X-ray photoelectron spectrum of PbS superimposed; (c) band-structure model forPbS calculated by the OPW method (after Tung & Cohen 1969).

REACTIVITY OF THE SURFACE OF GALENA 387

l l r l l

130 140 r50 160 170

Kinetic Energy (eV)

120 Kinetic Energy (eV)

Frc. 5, Sulfur LMM Auger electron spectra of PbS obtained from a natural specimen of high purity: (a) differentiatedspectmm; (b) undifferentiated spectrum showing computer fit of the data points (dashes) to three-component peaks(dots) at 140, 145.2 and 149.5 eV, respectively. Data were obtained using a Kratos XSAM 800 instrument.

170

388

Ebu,tllztll

THE CANADIAN MINERALOGIST

(at lower temperatures) and in aqueous media isPbSO4 (Steger & Desjardins 1980, Manocha & Park197). There is evidence, however, that the initial stepof oxidation may differ from the rrajor step andinvolve the formation of PbO and S. Manocha &Park (1977) concluded that lead-rich sulfides initiallyundergo oxidation of Pb and PbO throughchemisorption of oxygen (followed by slow adsorp-tion of hydroryl ions from the atmospheric environ-ment). As also found in the work of Eadington &Prosser (1969), oxidation was initially much slowerin lead-rich samples than in sulfur-rich samples ofgalena.

In the X-ray and UV photoelectron spectroscopicstudies of Evans & Raftery (1982), involving bothreactive oxygen species and normal air as media, itwas concluded that oxygen atoms attack (100) sur-faces of galena at Pb and probably also at S sites(and not only at defects). PbO species then form,whilst substantial quantities of S are lost from thesurface, but some elemental S remains trapped at theinterface between the PbS and its oxidation products.Sulfate species are also formed, and the sulfate andoxide coexist in an unstructured layer (which mayalso contain other species such as hydroxide). Oxi-dation in air was found to give similar products, butat much slower rates of reaction. There was no evi-{ence for thiosulfate formation under any ofthe con-,ditions studied.

Buckley & Woods (1984a, b) used X-ray photo-electron spectroscopy to study the oxidation ofgalena and also reviewed some of the earlier spec-troscopic studies of galena oxidation. As well as con-firming that lead sulfate is the product of exposureof galena to air for extended periods, they notedother changes during the first few days of exposure.It was proposed that air-oxidation results initially inlead becoming concentrated at the surface, where itis present as lead oxide and hydroxide (and subse-quently as carbonate, basic sulfate and sulfate). For-mation of the oxides, it was concluded, must beassociated with the development of a metal-deficientlayer of lead sulfide.

In their detailed study of the oxidation of PbS inaqueous media, Eadington & Prosser (1969) pro-posed an electrochernical model for the oxidation ofgalena in aqueous suspensions, witl anodic reactionof the lead sulfide:

P b S - P t / * + S + 2 e -

and a corresponding catlodic reaction

o ' 2 + 2 H 2 O + 4 r - 4 O H -

This pair of reactions was determined as the majorstep in the reaction and was preceded by an induc-tion period, during which oxygen is rapidlychemisorbed from solution, with transfer of electrons

ofiilJr''- iffiFFtc. 6, Energy levels for MgO6l0- and MgOrs- clusters

from DV-Xa calculations (after Tsukada et al. 1983).

face showing z-type conduction, in turn attributedto loss of sulfur from the surface and a resultingexcess of lead atoms.

With regards to the more complex questions of theelectronic structure associated with defects, and inparticular with Pb or S vacant sites, the calculationsof Hemstreet (1975) provide some information ofrelevance. In the case of a lead vacancy, each vacancyproduces two holes in the valence band, producingap-type,sample; there appears to be no localizationof charge about the vacancy site. A sulfur vacancy,according to the work of Hemstreet (1975), producesa bound state below the conduction-band edge, thetwo electrons associated with the sulfur vacancybeing localized to within a lattice constant or so ofthe defect site.

REACTWITY oF THE GersNa Sunracs

A large number of studies of reactions at the sur-face of galena has been undertaken. most notabryof oxidation in air and in aqueous media, becauseof its relevance to extraction methods involving flo-tation and leaching. The results ofthese studies aresummarized in Table2; the discrepancies arise partlyfrom the problems of accurately characterizing smallquantities of surface species (particularly in the earlywork) and possible variability of the stoichiometryof the galena.

However,.particularly bearing in mind the greatersensitivity of the modern methods of surface analy-sis, certain conclusions can be drawn. The first is thatthe major product of the oxidation of galena in air

REACTIVITY OF TIIE SURFACE OF GALENA

TABLE 2. STUDIES ON THE OXIDATION OF Pbs

389

Medium ild

ConditioE

Oxidation

Products

Method of

Cheacteriation

Relerence

Air;200"C

Air

Moist Air

Air; 100-400eC

Air;400nc

Air

Air

Acid aquru

Neutral aqu@6

Alkaline aqus6

Air ad aqusu

media (varying pH)

Neutral or Alkaline

aqu@u

Moist air

Air

Reactivs oxygen(alo air)

Air (ambient temp.

ed humidity)

Air-sturated

actic acld

Aqu6us

[+H2o)

Arcdic oxidation

Pbso4

PbO+S

Pbs2o3.Pbo

Pb5O4

Pbso4.Pbo

Pb5O4 + Pb5O3

PbS2o3

Elctron

Ditfraction

Chemical + X-ray

Infr&ed

lnfrded

Vet chemi€l

analysis

xPs

E!<trchemicl

Vet chemi@l

ilalysis

xPsxP5

xP5

xP5

xPs

Electrochemical

ed XPs

Haglhaa (1952)

Reuter &

Stein (19J7)

crenler (1962)

Pot.ing &

Leja (1963)

Eadinglon &

Prosr (1969)

Milocha &

Pak (1977)

G{dner d(

V@ds (1979)

Steger &

Desjudins (1980)

Brion (1980)

Eves d(

Raltery (19E2)

Buckley a(

V@ds (1984a'b)

Buckley &

W@ds (i-984a.b)

Buckley &

wods (1984a,b)

Buckley et al.(198r)

+ Pb54O6

S + PbSO,. Ii

S + PbSOr III

PbS:Oe + PbSorr + Pb5+o6l-Pbso,, + I- : - - - . . . . .

Imlrcr ) rrurrary; I

Pb-rlch Pbs glves some Pbo.l

Ll"o"l"tr" .n,*"o"r.from the 5)

Pbso4

Pbso4P b o + S

1+ Pbso4 IInitially Pbo, Pb(oH)2

ithen PbCo3r PbSo4r PbsQ4

|Ultimately Pbso4 I

:'ifffi;ffi;::"'Initiauy Pbt_xs

1[+H2O2 decomposition -

Ielemental Sl PbSo4, Minor

Ielements conc. at suria@ .l

Initialy Pbl-xs 1

At low pH lead hydroxide i

- elemental S od S{xygen Ispeqs

from the PbS. In the overall reaction describedabove, transfer of electrons from the PbS leaves(positive) holes; this also corresponds to the break-ing of a Pb-S bond and results in S remaining at thesurface, with Pb2+ going into $olution. It can alsobe regarded as development of ap-type surface onthe galena. The subsequent reactions leading to thefinal products of oxidation Clable 2) depend on thepH of the solution, with major sulfur plus sulfateunder acid conditions, sulfate and some sulfur atneutral pH, and formation of thiosulfates underalkaline condition$.

Eadington &Prosser (1969) also studied the effectsof varying extent of nonstoichiometry on the aque-ous oxidation of PbS and the action of light on reas-

tion rates. The oxidation of sulfur-rich galena wasfound to take place more rapidly than that of lead-rich samples, which was in turn a more rapid reac-tion than for stoichiometric PbS. Visible light hadthe effect of reducing the steady rate of oxidationbut had no effect on the induction period or productsof reaction. The leaching of galena by nitric acid,on the other hand (Eadington 1973), was enhancedduring the initial stages by the action of UV-visiblelight.

Studies of chemisorption, particularly of oxygen,on the surface of PbS are clearly of interest withregards to an understanding of surface reactivity.Hagstrdm & Fahlman (1978) studied the interactionof oxygen with the lead chalcogenides using UV and

390 THE CANADIAN MINERALOGIST

X-ray photoelectron spectroscopy. They proposedthat oxygen exposed in moderate amounts (as 02gas) to the PbS surface interacted only weakly withthat surface, being "physisorbed" rather thanchemisorbed. They further suggested a relationshipbetween surface reactivity and the "ionicity" of thematerial, with the most ionic phase @bS) showingthe least reactivity, and PbSe and PbTe surfacesbeing successively more reactive.

Grandke & Cardona (1980), in their UV-photoelectron studies of the position of the Fermilevel at the (100) surface of z- and p-type PbS,claimed to have a more sensitive technique for deter-mining changes in surface chemistry than that usedby Hagstrdm & Fahlman (1978). They interpretedmeasured shifts in the Fermi level as due to oxygenatoms occupying vacancies in the surface layer leftby evaporation of sulfur atoms, a process that occursunder vacuum and leads to a\ n-tpe surface. Theseoxygen atoms, therefore, restore the surface to aquasi-stoichiometric state and appear to be about astightly bound as the sulfur atoms at the surface.

DtscussroN

It has been suggested here that the surface geom-etry of a cleavage face of PbS can, for an ideal crys-tal, be modeled in a simple way without recourse todeformation of the surface atoms due to relaxation.The predominantly occuring S-fold co-ordination sitewould nevertheless lead to a reduction in the bandgap at the surface of the galena; this is supportedby experimental evidence (Grandke & Cardona1980), evidence that also lends support to this sim-ple general model of surface geometry.

Oxidation studies, both in air and in aqueous solu-tion, provide evidence for the ocsurrence of oxygenchemisorption onto the surface of galena. Spec-troscopic evidence suggests that oxygen at the sur-face behaves very similarly to sulfur (Grandke &Cardona 1980). The transfer of electrons, whichwould be associated with oxygen chemisorption,should generate positive holes in the valence band(i.e., plrpe conduction and, hence, ap-type surface).Calculations (Hemstreet 1975) suggest that theseholes are delocalized and would enhance conductivi-ties, and therefore reaction rates, in the essentiallyelectrochemical process of oxidation in aqueous solu-tion. This delocalization also accounts for the morerapid rate ofreaction for S-rich galena observed inexperimental systems @adington & Prosser 1969).The sulfur vacancies present at the surface of Pb-rich galena apparently produce localized electron-states (Hemstreet 1975) not conducive to rapid ratesof reaction. Presumably the vacant sites would alsobecome occupied by oxygen chemisorbed into themand thereby approach the more stable quasi-

stoichiometric condition (and also the least reactivestate).

The action of visible light on the system wouldcause a change in the electron-hole distribution,which may also be akin to producing a quasi-stoichiometric condition (Eadineton & Prosser 1969),and hence its effect in slowing down the oxidationreaction. The opposite effect, that of enhancing reac-tivities, which was observed in the leaching of illu-minated lead sulfide with nitric acid @adington1973), must again arise from changes in the electron-hole distribution in the valence-band region.

Thus a consistent picture is emerging of the geom-etry and electronic structure of the (100) surface ofgalena and of the overall effects produced by non-stoichiometry. Certain of the controls influencing thenature and rate of surface oxidatiron are evident. Amore complete picture will depend upon further cal-culations involving distorted clusters and absorbedspecies.

AcKNowLEDGEMENTS

This work was undertaken with the support ofNATO (Research Grant No. 833/83). Mr. J. Sullivanand Mr. A. Abbott (Dept. of Mathematics andPhysics, Aston University) are thanked for help andadvice in obtaining XPS and Auger spectra. Mrs. C.Gee and Ms. S. Knox are thanked for help in prepa-ration of the manuscript and illustrations.

REFERENcES

Bnrox, D. (1980): Photoelectron spectroscopic study ofthe surface degradation of pyrite (FeS), chal-copyrite (CuFeS), sphalerite (ZnS) and galena@bS) in air andwater. Applic. Surf. Sci. 5,133-152.

Bucrler, A.N., HevrrroN, I.C. & Wooos, R. (1985):Investigation of the surface oxidation of sulfideminerals by linear potential sweep voltammetry andX-ray photoelectron spectroscopy. Dev. MineralProcessing 6, 4l-60,

& Wooos, R. (1984a): An X-ray photoelectronspectroscopic study of the oxidation of galena.Applic, Surf. Sci. 17,401-414.

- & - (1984b): An X-ray photoelectron spec-troscopic investigation of the surface oxidation ofsulfide minerals, Proc. Electrochem. Soc. 84.286-302.

BunDErr, J.K. (1980): Molecular Shapes: TheoreticalModels of Inorganic Stereochemistry. Wiley-Interscience, New York.

CanooNa, M. & GnmNewev, D.L. (1964): Opticalproperties and band structure of Group IV-VI andGroup V materials. Phys. Rev. ,{133, 1685-1697.

REACTIVITY OF THE SUMACE OF GALENA 391

CARSLoN, T.A. (1975): Photoelectron and Auger Spec-troscopy. Plenum Press, New York.

Descleux, J.P. (1973): Relativistic Dirac-Fock expec-tation values for atoms with Z = | to Z = 120,Atomic Data and Nuclear Tables 12, 3l l-406.

Durn, C.B. (1983a): Structural chemistry of thecleavage faces of compound serniconductors , J, Vac.Sci. Tech. Bl,732-735.

(1983b): Atomic geometries of semiconductorsurfaces: determination and relation to chemicalbonding. Adv. Ceram. 6, l-14,

EaoncroN, P. (1973): Leaching of illuminated leadsulphide with nitric acid as a function of the solidstate electronic properties. Traw. Inst. MiningMetall. C82, 158-161.

& Pnossen, A,P. (1969): Oxidation of lead sul-phide in aqueous suspensions. Trans. Inst. MiningMetall. C18,74-82.

EvaNs, S. & Rerrenv, E. (1982): Electron spec-troscopic studies of galena and its oxidation bymicrowave-generated oxygen species and by air. "LChem, Soc., Faraday Trans. | 78, 3545-3560.

FeuERBacHsn, B., FrrroN, B. & Wrlus, R.F. (1978):Photoemission and the Electronic Propertiu of Sur-faces. Wiley-Interscience, Chichester, England.

Ganor.En, J.R. & Wooos, R. (1979): A study of thesurface oxidation of galena using cyclic voltam-metry. "L Electroanql. Chem. Interfacial Elec-trochem. 100, 447-459.

GnaNrrs, T. & Cenpore, M. (1980): Electronicproperties of clean and oxygen covered (100) cleavedsurfaces of PbS. Surf. Sci. 92,385-392.

& Lrv, L. (1979): Temperature effectson valence bands in semiconducting lead chal-cogenides. Solid State Comm. 32,353-356,

Gnnrmrn, R.G. (1962): An infrared investigation ofxanthate absorption by lead sulfide. J. Phys. Chem.66, 879-883.

Hacruane, J. (1952): Surface oxidation of galena inrelation to its flotation as revealed by electrondiffraction. J. Phys. Chem. 56,610-615.

Hecsrndu, A.L. & FAuLvraN, A. (1978): The interac-tion between oxygen and the lead chalcogenides atroom temperature studied by photoelectron spec-troscopy. Applic. Surf. Sci. 1, 455-470,

HARRrsoN, W.A. (1976): Surface reconstruction onsemiconductors. Surf. Sci. 55, l-19.

- (1980): Electronic Structure and the Propertiesof Solids. W.H. Freeman & Co., San Francisco,California.

Heusrnnrr, L.A., Jn. (1975): Cluster calculations ofthe effects of single vacancies on the electronicproperties of PbS. Phys. Rev. B1l,2260-2270.

Hnxnrcr, V.E. (1983): The nature of transition-metaloxide surfaces. Prog. Surf. Sci.14,175-200.

Hrnvarl, F., KoRtuu, R.L., OureNnuncrn, I.B. & VaNDvre, J.P. (1968): Relativistic band structures ofGeTe, SnTe, PbTe, PbSe and PbS. J. Phys. (Paris)29, C4, 62-77.

JouNsoN, K.H. (1973): Scattered wave theory of thechemical bond. Adv. Quantum Chem.7,143-185.

KAHN, A. (1983): Semiconductor surface structures.Surf. Sci. Rep. 3, 193-300.

Lrums, J.D. & Menr, P. (1966): Theory and observa-tion of intrinsic surface states on ionic crystals. P&/s.Rev, 14/,751-763.

LtctrveN, D., Cnetc, J.H., Jn., SAILen, V. & DruNr-wnve, M. (1981): AES and XPS spectra of sulfur insulfur compounds. Applic. Surf. Sci. 7,325-331.

MANN, J.B. (1967): Atomic slructure calculations. I.Hartree-Fock energy results for the elements hydro-gen to lawrencium. Zos Alamos Sci. Lab., Rep.LA-3690.

Merocua, A.S. & Psm, R.L. (1977): Flotation relatedESCA studies on PbS surfaces. Applic. Surf. Sci.t, t29-t4t.

McFsrlv, F.R., KowerczyK, S., Lev, L., Poltar,R.A. & SHtnLrv, D.A. (1973): High-resolution X-ray-photoemission spectra of PbS, PbSe, and PbTevalence bands. P&ys. Rev. 87, 5228'5237.

Por-rNc, G.W. & LrJa, J. (1963): Infrared study ofxanthate adsorption on vacuum deposited films oflead sulfide and metallic copper under conditionsof controlled oxidation. J. Phys. Chem. 67,2t2l-2t26.

Reu,en, B. & Srerrs, R. (1957): Die Oxydation vonBleisulfid bei niederen Temperaturen. Z. Elek-trcchenx. 61,440-454.

ScuoolaR, R.B. & DrxoN, J.R. (1965): Optical cons-tants of lead sulfide in the fundamental absorptionedge region. Phys, Rev. 4137,667-670.

SHERMAN, D.M. (1985): The electronic structures ofFe3+ coordination sites in iron oxides; applicationsto spectra, bonding and magnetism. Phys. Chem.Minerals 12, 16l-175.

Sr-ersn, J.C. (1974\: The Self Consistent Field Methodfor Atoms, Molecules and Solids. McGraw-Hill,New York.

Srecrn, H.F. & Dssr,q.xons, L.E. (1980): Oxidation ofsulfide minerals. V. Galena, sphalerite and chalco-cite, Con. Mineral. 18,365-372.

392 THB CANADIAN MINERALOGIST

SwAr.ls, C.A., Gooonm, W.A. III & McGu,r, T.C.(1980): Theoretical studies ofthe reconstruction ofthe (1 l0) surface of III-N and II-VI semiconduc-tor compounds. J. Vac. Sci. Tech.17,982-986,

Tossmr., J.A. (1984): A qualitative MO model forbridging bond angle variations in minerals. P&ys.Chenl Minerals 11, 8l-84.

Tstxeoe, M., Aoecru, H. & Seroro, C. (1983):Theory of electronic structure of oxide surfaces.Prog. Surf. Sci. l4Q), ll3-173.

TuNc, Y.W. & CourN, M.L. (1969): Relativistic bandstruclure and electronic properties of SnTe, GeTe,and PbTe. Phts. Rev.180, 823-826.

Uncn, D.S. (1985): X-ray spectroscopy and chemicalbonding in minerals. In Chemical Bonding and Spec-troscopy in Mineral Chemistry (F.J. Berry & D.J.Vaughan, eds.). Chapman and Hall, London.

VaucHAN, D.J., Tossnr, J.A. & JonrsoN, K.H.(1974): The bonding of ferrous iron to sulfur andoxygen in tetrahedral coordination: a comparativestudy using SCF-Xa scattcred wave molecular orbital calculations, Geochim. Cosmochim. Acta 3E,993-1005.

Received October 23, 1985, revised manuscipt qcceptedSeptember 22, 1986.