Applied Surface Science 179 (2001) 222±229

Surface states and reactivity of pyrite and marcasite

I. Uhliga, R. Szargana,*, H.W. Nesbittb, K. Laajalehtoc
a
Wilhelm-Ostwald-Institut, UniversitaÈt Leipzig, LinneÂstr. 2, D-04103 Leipzig, Germany
b
Department of Earth Sciences, University of Western Ontario, London, Ont., Canada N6A 5B7
c
Department of Applied Physics, Laboratory of Materials Science, University of Turku, Vesilinnantie 5, FIN-20014 Turku, Finland

Abstract

The sulphur surface sites of pristine marcasite surfaces and their reactivity during initial air oxidation were investigated
for the ®rst time using synchrotron radiation excited photoelectron spectroscopy (SRXPS). Both S 2p and Fe 2p spectra were
collected and compared to those of the polymorph pyrite. A new sulphur surface component is found additionally to the two
known pyrite surface signals. Due to the non-isotropic character of the orthorhombic marcasite structure, its peak intensity
depends strongly on the surface orientation. Its binding energy is close to the region of short-chained polysulphides. The
signal is assigned to sulphur trimers produced by electron transfer between surface near sulphur dimers and S formed after
ruptures of S±S bonds which results in S2 and S32 . In the cubic crystal arrangement of pyrite the stabilisation of S to S2
is realised by an electron transfer from iron to sulphur only. A con®rmation of this assumption is found in the Fe 2p spectra.
The contribution of Fe3‡ species Ð a result of the electron transfer process Ð is considerably less pronounced for
marcasite.
The mechanism of initial oxidation is different for marcasite and pyrite. The sulphur species responsible for the new
spectroscopic component are the most reactive sites at the marcasite surface, while S2 is oxidised ®rst in case of pyrite. The
signals indicating the products of marcasite oxidation in the S 2p spectrum are less intense as expected from the intensity
decreasing of the trimer peak. Therefore, an oxidation of sulphur trimers to elemental sulphur is assumed which could not be
detected because of its volatility in the vacuum of the air lock system. # 2001 Elsevier Science B.V. All rights reserved.

Keywords: Photoemission; Synchrotron radiation; Pyrite; Marcasite; Surface states

1. Introduction sulphur and three cations in a distorted tetrahedral

environment, whereas each cation is surrounded by six
The iron sulphides marcasite and pyrite are poly- nearest-neighboured sulphur dumbbells in a distorted
morphs. They have the same chemistry and identical octahedral arrangement (Fig. 1a).
features of local coordination, but different crystal The cubic pyrite structure can be considered as a
structures [1]. The chemical constitution of both derivative of the face-centred rock salt structure,
minerals is FeS2. The cation and anion sites in the replacing Cl and Na‡ by S22 and Fe2‡, respectively.
crystal lattice are occupied by Fe2‡ cations and di- The diatomic axes of the S±S pairs are oriented in
sulphur anions (S22 ). Each sulphur bonds to another alternate alignments in each crystal layer (Fig. 1b).
The cation octahedra share corners.
*
Corresponding author. Tel.: ‡49-341-9736451;
Marcasite has an orthorhombic structure (Fig. 1c).
fax: ‡49-341-9736399. The symmetry reduction by the shortened metal±
E-mail address: szargan@rz.uni-leipzig.de (R. Szargan). metal distances along c-direction [2] causes a splitting

0169-4332/01/$ ± see front matter # 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 9 - 4 3 3 2 ( 0 1 ) 0 0 2 8 3 - 5
 I. Uhlig et al. / Applied Surface Science 179 (2001) 222±229 223

It is known from mineral collections and shown by

conventional XPS investigations that marcasite sur-
faces are much more reactive than pyrite ones [7]. It
seems clear that the different structure is the reason for
this behaviour. However, while clean and air oxidised
pyrite is very well characterised by surface sensitive
methods [6±13], less is known about the near surface
states of pristine marcasite and the initial oxidation
products on it. It is the objective of this study to
identify the chemical states which are responsible
for the high reactivity of marcasite compared to pyrite
using appropriately tuned monochromatized synchro-
tron radiation for excitation of highly surface sensitive
photoemission.
The S 2p spectrum of pyrite collected under surface
sensitive conditions is composed of contributions
from three distinct chemical states [10]. Besides the
main peak at 162.6 eV, caused by fully coordinated
bulk sulphur dimers, two low-energy surface compo-
nents were observed. The ®rst one represents surface-
most sulphur atoms of sulphur dimers located at the
fractured surface after rupture of Fe±S bonds. The
lowest energy peak is interpreted as S2 produced
after rupture of S±S bonds. It was concluded that an
electron transfer from iron to sulphur stabilises S to
S2 , thereby producing Fe3‡ species at the surface.
Evidence for Fe3‡ surface states should be found in
the Fe 2p spectra. However, the Fe3‡ signal disappears
in a wedge-shaped high-energy tail of the pyrite Fe 2p
spectra. The origin of the tail could not be unambigu-
ously explained so far. It has been considered to be a
Doniach-Sunjic contribution as for metals [14,15], an
Fig. 1. Chematic illustration of pyrite and marcasite crystal energy loss feature due to interband excitations or
structures: (a) common local coordination of anions and cations; vacancies in the crystal structure [16], and Fe3‡ sur-
(b) simple-cubic pyrite structure; (c) orthorhombic marcasite face states [10±15]. Studies by Nesbitt et al. [17]
structure.
proved most recently that surface states contribute
signi®cantly to the intensity of the Fe 2p high-energy
tail comparing Fe 2p spectra collected after conven-
of the t2g orbitals and leads to a band gap decrease tional excitation with such excited more surface sen-
from 0.9 eV in pyrite to 0.4 eV in marcasite. The sitive by synchrotron radiation. Nevertheless, quan-
essential difference in the atomic arrangement is that titative statements on intensities remain very dif®cult.
linear chains of edge-sharing cation octahedra can be Both Fe2‡ and Fe3‡ surface states involve unpaired
formed [3±5]. electrons and therefore produce multiplet structures
Both minerals show conchoidal fracture. The clea- due to ®nal state effects that complicate the peak
vage of pyrite along the (1 0 0) plane is usually poor. ®tting. Comparing the high energy tails of Fe 2p
However, ordered surfaces can be obtained as judged spectra collected from systems containing sulphur
by LEED [6]. Marcasite is described to have a good dimers and iron in different crystal arrangements
prismatic cleavage in two directions. like pyrite and marcasite, however, could supply
224 I. Uhlig et al. / Applied Surface Science 179 (2001) 222±229

additional information and shed more light on the 3. Results

problem described above.
3.1. S 2p spectra

2. Experimental Fig. 2 presents the S 2p spectra of pyrite and of two

different marcasite surfaces measured with the same
Pyrite single crystals from Navajun/Spain and mar- experimental parameters. The spectra were ®tted
casite samples from Komorany/Czech Republic (I) and using doublets with a ®xed peak separation of
Sokolov/Czech Republic (II) were used for the experi- 1.18 eV. The optimum ratio of the 2p3/2 and 2p1/2
ments. Identi®cation of pyrite and marcasite was con- intensity was determined to 0.47. The Gaussian and
®rmed by XRD patterns. Pristine surfaces were Lorentzian line widths were allowed to vary with the
prepared by cutting the mineral specimens with a restriction that signals of the same origin should
diamond saw into cubes and fracturing them at a possess very similar peak shapes. Both the pyrite
pressure of about 5  10 6 Pa in a preparation chamber.
The synchrotron radiation excited photoelectron
spectroscopic experiments were carried out at beam-
line I411 (MAXlab, Lund/Sweden) and at Synchro-
tron Radiation Centre (SRC) in Madison/Wisconsin
(USA) using the HERMON-beamline. The UHV
chambers were equipped with SCIENTA 200 analy-
sers and sample cooling systems. All spectra were
collected at a sample temperature of 75 K (MAXlab)
or 20 K (SRC). In order to study the reactivity of
surface sites one sample of each mineral was oxidised
simultaneously venting the air lock system of the
UHV chamber and exposing them for 2 min to labora-
tory atmosphere.
The spectra were ®tted using a sophisticated soft-
ware for core level peak shape analysis [18]. The peak
®tting employs the Marquardt algorithm in a least
square procedure. The peak shape was described as a
convolution of Gaussian and Lorentzian functions. A
Shirley-type background function, describing the
inelastic loss processes of the photoelectrons within
the sample, was combined with a second order poly-
nomial contribution considering the increased second-
ary electron background at low kinetic energies. The
parameters for the background were included in the ®t
routine.
The binding energies of the obtained components
are referred to binding energies of the bulk signals
obtained from pristine pyrite and marcasite surfaces
using a conventional monochromatised Al Ka source Fig. 2. Fitted S 2p spectra (hn ˆ 250 eV) of a pristine pyrite
from a VG ESCALAB220iXL spectrometer (pyrite: S surface in comparison to marcasite spectra of two different
2p (162.6 eV), Fe 2p (707.2 eV); marcasite: S 2p samples. Component `c' represents the bulk sulphur dimers. The
surface most sulphur atoms of surface dimers after rupture of Fe±S
(162.5 eV), Fe 2p (707.3 eV)). The energy scale of bonds are responsible for component `b'. Component `a' is
the spectrometer was calibrated with Au 4f7/2 assigned to surface S2 formed after rupture of S±S. A new
(84.0 eV) and Cu 2p3/2 (932.6 eV). component `d' is observed for marcasite.
 I. Uhlig et al. / Applied Surface Science 179 (2001) 222±229 225

and the marcasite spectra show beside the bulk state

(Fig. 2, peak `c') at 162.6 and 162.5 eV, respectively,
two signals `a' and `b' at lower binding energies.
According to the model of Nesbitt et al. [10] the
spectral component `b' with an energy shift of
0.6 eV for both minerals is emitted from surface
sulphur dimers possessing higher electron density
than the bulk dimers after rupture of S±Fe bonds.
The component with the lowest kinetic energy `a' was
interpreted as S2 formed after stabilisation of S by
an electron transfer from iron to sulphur when S±S
bonds are ruptured. The energy shift for component
`a' was found to be 1.4 and 1.2 eV for pyrite and
marcasite, respectively. However, the modelling of the Fig. 3. Dependence of relative intensities in the S 2p spectra of
low binding energy tail of marcasite S 2p spectra was marcasite I from the excitation energy.
dif®cult. The smaller energy shift between component
`a' and `b' in connection with the low intensity of `a'
and the strong dependence of its peak shape on the ing the marcasite surface for about 2 min to air,
de®nition of the background function complicate the component `a' is largely unaffected, whereas compo-
description of the experimental spectra in this region. nent `d' is diminished.
Signal `a' is unusually broad suggesting that it consists
of two components. But the present data do not permit 3.2. Fe 2p spectra
their resolution. This uncertainty has to be considered
when interpreting the results. Iron bulk and surface sites contribute to the Fe 2p
In comparison to pyrite, the sulphur spectra of spectra. The conventional Fe 2p3/2 spectrum of both
marcasite contain an additional component `d' with pyrite and marcasite shows a narrow single main peak
a binding energy in the region of short-chained poly- at 707.2 and 707.3 eV, respectively, displaying the
sulphides (163.2 eV). It was observed that the inten- low-spin Fe2‡ bulk state. Additionally both minerals
sity of the component `d' varies from sample to show an high energy tail, which is less pronounced for
sample, but the binding energy does not. marcasite [20].
In general, for complete description of the spectra a Changing the excitation energy from 1486.6 to
broad signal is necessary to ®t the high binding energy 850 eV increases the intensity at the high energy side
tail. This feature is observed also with conventional of the main signal, thereby con®rming that species
excitation and can, therefore, be attributed to bulk located at or near the surface essentially contribute to
characteristics. It can be interpreted as interband this part of the signal [17]. Those species are formed as
excitations [19] or as an effect produced by sul- result of bond rupture during fracture of the minerals.
phur-rich areas in the crystal [16]. If Fe±S bonds are broken, the bonding electron pair
The photon energy dependence of relative peak stays with sulphur due to the difference in electro-
intensities is shown in Fig. 3 for the different S 2p negativity, leaving Fe with a dangling bond. The
components of marcasite (sample I). Components `a', lowered electron density at the affected metal ions
`b', and `d' decrease increasing the excitation energy leads to a higher binding energy of the Fe 2p electrons.
related to the bulk signal `c'. Furthermore, the octahedral (Oh) symmetry of the
The changes in the S 2p spectra of pyrite and cation environment reduces at least to a square pyr-
marcasite by initial oxidation are compared in amidal (C4v) ligand ®eld [21]. This changes the energy
Fig. 4. It is known from previous studies on pyrite separation of the `d' levels and the distribution of the
[11,12] that the S2 surface species `a' react ®rst. electrons. The new states contain unpaired electrons
Component `a' disappears and a small amount of resulting in multiplet peak splitting of their associated
sulphate is detected. In contrast to pyrite, after expos- photoemission peaks [22].
226 I. Uhlig et al. / Applied Surface Science 179 (2001) 222±229

Fig. 4. Changes in S 2p spectra (hn ˆ 250 eV) of pyrite and marcasite (sample II) by initial air oxidation. Both samples were simultaneously
exposed for 2 min to laboratory atmosphere.

Additionally to the multiplet from Fe2‡ surface causing additional multiplet peaks at the high energy
states, Fe3‡ species are represented in the high energy side of the Fe 2p bulk signal. The binding energy of
tail of the Fe 2p spectrum. Following Nesbitt et al. this contribution can be estimated by reference
[10], an electron transfer from iron to sulphur is minerals containing Fe2‡ and Fe3‡. Both for pyr-
responsible for the formation of S2 detected in rhotite [23] and for iron oxides [24] the main com-
the S 2p spectra as surface component `a'. The result ponent of the Fe3‡ multiplet was located at about
are Fe3‡ surface sites. These species necessarily 1.75 eV greater binding energy than observed for
possess unpaired electrons in the 3d valence band, Fe2‡.

Table 1
Binding energy shifts and intensity ratios in Fe 2p spectra of pyrite and marcasite

Component Pyrite Marcasite

DEB Intensity DEB Intensity

Fe2‡
bulk 0.0 (707.2 eV) 0.54 0.0 (707.3 eV) 0.66
Fe2‡
surface I 0.6 0.13, 0.24 0.6 0.12, 0.22
Fe2‡
surface II 1.2 0.07 1.2 0.06
Fe2‡
surface III 2.9 0.03 2.9 0.03
Fe2‡
surface IV 5.1 0.01 5.1 0.01
Fe3‡
surface I 1.9 0.12, 0.22 1.9 0.06, 0.11
Fe3‡
surface II 2.8 0.06 2.8 0.03
Fe3‡
surface III 4.2 0.04 4.2 0.02
 I. Uhlig et al. / Applied Surface Science 179 (2001) 222±229 227

tributions were obtained for Fe2‡ surface states rela-

tive to the intensity of the bulk peak for both pyrite and
marcasite, while the multiplet attributed to Fe3‡ spe-
cies is twice as intense in case of pyrite.

4. Discussion and conclusions

The S 2p spectrum of fractured marcasite collected

under surface sensitive conditions is composed of four
major contributions. Three of them are known from
pristine pyrite surfaces: the doublet of the bulk sulphur
dimers `c' and two signals `a' and `b' shifted to lower
binding energy which represent surface sulphur sites
after ruptures of Fe±S and S±S bonds, respectively.
The fourth peak `d' is detected at the high binding
energy side of the bulk component. The dependence of
the relative intensity from the excitation energy
(Fig. 3) proves that `d' displays an additional surface
state. The signal is not signi®cantly broader than the
other doublets. Therefore, it can be assigned to one
de®ned short-chained polysulphidic species such as
Fig. 5. Fitted Fe 2p spectra (hn ˆ 850 eV) of pyrite and marcasite.
sulphur trimers. Their formation can be explained
The most intense signal is the bulk Fe2‡ peak. The high energy tail
contains contributions of Fe2‡ (dotted line) and Fe3‡ (solid line) assuming that the electron that stabilises S Ð
surface multiplets. Energy shifts and intensity ratios are based on derived from S±S bond rupture Ð to S2 (component
assumptions described in [12]. `a') is not or not only delivered from iron but from
sulphur. The arrangement of sulphur in the marcasite
crystal lattice probably permits a disproportionation
The contributions of Fe2‡ and Fe3‡ surface states reaction including S and surface near sulphur dimers
were considered, when ®tting the Fe 2p3/2 spectra of which results in S2 and S32 :
pyrite and marcasite collected with an excitation
energy of 850 eV (Fig. 5). The marcasite Fe 2p3/2 2S ‡ S2 2 ) S2 ‡ S3 2
signal arises from a surface with a S 2p spectrum
shown in Fig. 2 for sample II. Energies and intensity This process is unlikely for the cubic pyrite structure.
ratios were adapted following Schaufuss et al. [12] and A con®rmation of this postulate can be derived from
are summarised in Table 1. Because the multiplet the Fe 2p spectra: the lower Fe3‡ contribution to the
structure of the surface components is not really high energy tail of the Fe 2p3/2 signal of marcasite
certain, the discussion of the results should be compared to pyrite (Fig. 5; Table 1) implies that
restricted here to a comparison of the individual reactions involving electron transfers from Fe2‡ to
contributions to the spectra of the two minerals. sulphur are less signi®cant at marcasite surfaces.
Therefore, the ®tting of the Fe 2p3/2 spectra was The intensity of `d' relative to the bulk signal differs
carried out with tight constraints: disregarding differ- from sample to sample (Fig. 2). This effect can be
ences of the multiplet intensities from slightly varying explained by the orthorhombic structure of marcasite.
sulphur coordination in pyrite and marcasite, identical Unlike pyrite, marcasite is strongly non-isotropic.
peak shapes for all signals, the same energy shifts Therefore, distinct differences in surface properties
related to the bulk signal, and constant intensity ratios are expected depending on the cleavage direction
within the two multiplet structures were assumed. relative to the c-axis of the crystal. Although it was
Applying the described parameters, comparable con- impossible to determine the crystallographic orienta-
228 I. Uhlig et al. / Applied Surface Science 179 (2001) 222±229

tion of the fractured surfaces, some suggestions may Radiation Centre in Madison/Wisconsin, particularly
be drawn from the appearance of the specimens. H. Hoechst, for technical assistance and helpful dis-
Marcasite I was a single crystal and thus fractured cussions. This work was ®nancially supported by
preferably sub-parallel to one de®ned plane. Sample II National Science and Engineering Research Council
was polycrystalline so that faces from different orien- of Canada. I. Uhlig acknowledges NSERC support in
tations contribute to the spectra. The S 2p/Fe 2p peak form of a Post-doc fellowship. The SRC is funded by
area ratios were estimated keeping in mind the uncer- the National Science Foundation through NSF Grant
tainty when deriving quantitative statements from DMR-95-31009.
different recordings of synchrotron spectra which
are in¯uenced by a set of variables. The calculated
intensity ratios of 13.4 for marcasite I and 11.1 for References
marcasite II con®rm the assumption that different
oriented cleavage planes were investigated. Further- [1] S. Matthes, Mineralogie, 4th Edition, Springer, Berlin, 1993,
more, there seems to be a direct connection between p. 38.
the sulphur content of the investigated region and the [2] J.B. Goodenough, J. Solid State Chem. 5 (1972) 144.
[3] D.W. Bullett, J. Phys. C: Solid state Phys. 15 (1982) 6163.
intensity of component `d', i.e. if sulphur terminated [4] J.A. Tossell, D.J. Vaughan, Theoretical Geochemistry:
cleavage planes are produced, the relaxation process Application of Quantum Mechanics in the Earth and Mineral
including sulphur±sulphur electron transfer is pro- Sciences, Oxford University Press, New York, 1992, p. 291.
moted. However, additional investigations are [5] J.A. Tossell, D.J. Vaughan, J.K. Burdett, Phys. Chem. Min. 7
required in order to assess the in¯uence of the mar- (1992) 177.
[6] K.M. Rosso, U. Becker, M.F. Hochella Jr., Am. Min. 84
casite surface orientation on the formation of the (1999) 1535.
surface states in more detail. [7] M.J. Rinker, H.W. Nesbitt, A.R. Pratt, Am. Min. 82 (1997)
The initial oxidation of marcasite is different from 900.
pyrite. The oxidation mechanism proposed for pyrite [8] C. Pettenkofer, W. Jaegermann, H. Tributsch, H. Kuhlenbeck,
W. Braun, S. Bernstorff, BESSY Annual Report, 1988, p. 249.
includes an Fe3‡/Fe2‡ redox cycling [25] acting as
[9] M. Bronold, Y. Tomm, W. Jaegermann, Surf. Sci. Lett. 314
mediator process for the oxidation of surface sulphur (1994).
species to sulphate. The smaller supply of Fe3‡ spe- [10] H.W. Nesbitt, G.M. Bancroft, A.R. Pratt, M.J. Scaini, Am.
cies at the marcasite surface may prevent the fast Min. 83 (1998) 1067.
oxidation of S22 . Instead, the trimers are oxidised [11] A.G. Schaufuss, H.W. Nesbitt, I. Kartio, K. Laajalehto, G.M.
further and the corresponding signal diminishes. How- Bancroft, R. Szargan, Surf. Sci. 411 (1998) 321.
[12] A.G. Schaufuss, H.W. Nesbitt, I. Kartio, K. Laajalehto, G.M.
ever, the spectrum does not indicate oxidation pro- Bancroft, R. Szargan, J. Electron. Spec. Rel. Phen. 96 (1998)
ducts to the same extent; perhaps the sulphur trimers 69.
were partially oxidised to elemental sulphur. The latter [13] K.M. Rosso, U. Becker, M.F. Hochella Jr., Am. Min. 84
is volatile in vacuum and would not be detected (1999) 1549.
[14] S. Doniach, M. Sunjic, J. Phys. C: Solid States Phys. 3 (1970)
without sample cooling during pumping down the
285.
air lock system [26]. However, this assumption has [15] H.W. Nesbitt, I.J. Muir, Geochim. Cosmochim. Acta 58
to be checked by further experiments. Furthermore, it (1994) 4667.
would be useful to study the oxidation of marcasite [16] K. Laajalehto, I. Kartio, T. Kaurila, T. Laiho, E. Suoninen, in:
systematically with controlled dosing of oxygen to H.J. Mathieu, B. Reihl, D. Briggs (Eds.), Proceedings of the
European Conference on Applications of Surface and Inter-
describe the reactivity of the different surface sites of
face Analysis, ECASIA'95, Wiley, Chichester, 1996, pp. 717±
marcasite on a better-de®ned basis. 720.
[17] H.W. Nesbitt, M. Scaini, H. HoÈchst, G.M. Bancroft, A.G.
Schaufuss, R. Szargan, Am. Min. 85 (2000) 850.
Acknowledgements [18] R. Hesse, T. ChasseÂ, R. Szargan, Fresenius' J. Anal. Chem.
365 (1999) 48.
[19] M. Fujisawa, S. Suga, T. Mizokawa, A. Fujimori, K. Sato,
The authors thank W.M. Skinner (University of Phys. Rev. B 49 (1994) 7155.
South Australia), A.R. Pratt (NRC, Canada) and the [20] A.R. Pratt, N.S. McIntyre, S.J. Splinter, Surf. Sci. 396 (1998)
staff of MAXlab in Lund/Sweden and the Synchrotron 266.
 I. Uhlig et al. / Applied Surface Science 179 (2001) 222±229 229

[21] M. Bronold, Y. Tomm, W. Jaegermann, Surf. Sci. 314 (1994) [24] M.S. McIntyre, D.G. Zetaruk, Anal. Chem. 49 (1977) 1521.
L931. [25] C.M. Eggleston, J.-J. Ehrhardt, W. Stumm, Am. Min. 81
[22] R.P. Gupta, S.K. Sen, Phys. Rev. B 12 (1) (1975) 15. (1996) 1036.
[23] A.R. Pratt, I.J. Muir, H.W. Nesbitt, Geochim. Cosmochim. [26] G. Wittstock, I. Kartio, D. Hirsch, S. Kunze, R. Szargan,
Acta 58 (1994) 827. Langmuir 12 (1996) 5709.