This article was published in the above mentioned Springer issue.The material, including all portions thereof, is protected by copyright;all rights are held exclusively by Springer Science + Business Media.

The material is for personal use only;commercial use is not permitted.

Unauthorized reproduction, transfer and/or usemay be a violation of criminal as well as civil law.

ISSN 1618-2642, Volume 396, Number 8

PAPER IN FOREFRONT

Combined use of X-ray photoelectron and Mössbauerspectroscopic techniques in the analytical characterizationof iron oxidation state in amphibole asbestos

Marzia Fantauzzi & Alessandro Pacella & Davide Atzei &Antonio Gianfagna & Giovanni B. Andreozzi &Antonella Rossi

Received: 10 January 2010 /Revised: 10 February 2010 /Accepted: 12 February 2010 /Published online: 10 March 2010# Springer-Verlag 2010

Abstract Asbestos fibers are an important cause of serioushealth problems and respiratory diseases. The presence,structural coordination, and oxidation state of iron at the fibersurface are potentially important for the biological effects ofasbestos because iron can catalyze the Haber–Weiss reaction,generating the reactive oxygen species ⋅OH. Literature resultsindicate that the surface concentration of Fe(III) may play animportant role in fiber-related radical formation. Amphiboleasbestos were analyzed by X-ray photoelectron spectroscopy(XPS) and Mössbauer spectroscopy, with the aim of deter-mining the surface vs. bulk Fe(III)/Fetot ratios. A standardreference asbestos (Union Internationale Contre le Cancercrocidolite from South Africa) and three fibrous tremolitesamples (from Italy and USA) were investigated. In additionto the Mössbauer spectroscopy study of bulk Fe(III)/Fetotratios, much work was dedicated to the interpretation of the

XPS Fe2p signal and to the quantification of surface Fe(III)/Fetot ratios. Results confirmed the importance of surfaceproperties because this showed that fiber surfaces are alwaysmore oxidized than the bulk and that Fe(III) is present asoxide and oxyhydroxide species. Notably, the highestdifference of surface/bulk Fe oxidation was found for SanMango tremolite—the sample that in preliminary cytotox-icity tests (MTT assay) had revealed a cell mortality delayedwith respect to the other samples.

Keywords Asbestos amphiboles .Mössbauerspectroscopy . X-ray photoelectron spectroscopy .

Oxidized iron . Crocidolite . Tremolite

Introduction

Historically, asbestos is a generic commercial industrialterm used to describe a group of hydrate silicate mineralswith fibrous morphology and forming bundles. The hightensile strength, flexibility, and electrical and thermalresistance of asbestos led to its use in several mechanicaland commercial applications such as thermal insulation,building materials, fireproof and bulletproof materials,textile products, and many more. Typically, the mineralsdesignated as asbestos are six: chrysotile, tremolite,actinolite, anthophyllite, grunerite, and riebeckite asbestos.Chrysotile asbestos, the only fibrous member of theserpentine mineral group, has the theoretical chemicalformula Mg3Si2O5(OH)4. The other above-listed mineralsare amphiboles: anthophyllite asbestos, theoretical: (Mg,Fe2+)7Si8O22(OH)2; tremolite-actinolite asbestos, theoreti-cal: Ca2(Mg,Fe2+)5Si8O22(OH)2; grunerite asbestos, theo-retical: (Fe2+,Mg)7Si8O22(OH)2, usually referred to

Electronic supplementary material The online version of this article(doi:10.1007/s00216-010-3576-0) contains supplementary material,which is available to authorized users.

M. Fantauzzi :D. Atzei :A. RossiDipartimento di Chimica Inorganica e Analitica,INSTM Research Unit,Centro Grandi Strumenti Università di Cagliari,09100 Cagliari, Italy

A. Pacella :A. Gianfagna :G. B. Andreozzi (*)Dipartimento di Scienze della Terra, Sapienza Università di Roma,Piazzale Aldo Moro 5,00185 Rome, Italye-mail: gianni.andreozzi@uniroma1.it

A. Rossi (*)Dipartimento di Chimica Inorganica ed Analitica,SS 554 Bivio per Sestu,09100 Cagliari, Italye-mail: rossi@unica.it

Anal Bioanal Chem (2010) 396:2889–2898DOI 10.1007/s00216-010-3576-0

Author's personal copy

colloquially as “amosite” (from the acronym AMOS,representing Asbestos Mines of South Africa); riebeckiteasbestos, theoretical: Na2(Fe

2+,Mg)3Fe23+Si8O22(OH)2,

usually referred to colloquially as “crocidolite.”Epidemiological studies showed that exposure to asbes-

tos have been linked to numerous health problems andrespiratory diseases. On the other hand, low to moderateexposure to chrysotile asbestos causes a very low healthrisk, and this is presumably related to its solubility in thebody fluids. Conversely, the biosolubility was observed tobe very low in the case of amphibole asbestos [1]. Themechanism through which asbestos fibers may causediseases has not been fully elucidated. Several authorsagree in attributing to mineral fibers the formation ofreactive oxygen species (ROS) that determine a strongrelease of HO° free radicals by partially dissolving intobiological fluids participating in Fenton chemistry [2–5].Toxicological studies provided evidence that the interac-tions between fibrous material and biological environmentare strongly dependent on both the geometry and the crystalchemistry of mineral fibers [6–9]. In particular, the presenceand the bioavailability of Fe received considerable attentionfrom the biomedical community. Both the presence and thestructural coordination of iron were proposed as importantfactors in the toxicity of asbestos [10] and, furthermore,only the iron of the fiber surface is important in the ROSproduction [11, 12]. Among the chemical factors, thesurface chemical properties are the most important indetermining the hazardous effects of the fibers. In fact, theinteraction with the biological fluids occurs at the fibers’surface. The reactive sites at the surface include low-coordinated (due to vacancies and edges) surface atoms andexposed ions that may coordinate molecules and play aprimary role in the catalytic reactions [7]. The surface ofasbestos has basic character and therefore can strongly reactwith acidic molecules such as CO2, benzoic acid, andfluorine, and it is capable of forming hydrogen bonds withadsorbing species [7]. The hydrogen bonding capabilitymakes asbestos strongly hydrophilic, and therefore it caneasily react with cell membranes [7]. In addition, biomo-lecular components (phospholipids, proteins) of the cellmembranes constitute numerous sites that can form hydro-gen bonds, and as a result interaction of mineral fibers andcell membranes leads to membranolysis [7]. On the surfaceof these membranes, asbestos fibers can adsorb (both beforeand after inhalation) toxic substances and become morehazardous. However, nontoxic substances such as phos-pholipids and surfactants present in the air can also coat thefibers’ surfaces, thus preventing dangerous surface reac-tions by making the active sites inaccessible [7]. If this istrue, the characterization of the asbestos’ surface isnecessary to better understand possible mechanisms ofinteraction between fibers and biological system.

In this work, amphibole asbestos were characterized byX-ray photoelectron spectroscopy (XPS); the objective ofthe present work was the determination of their surfacecomposition and of the Fe(III)/Fetot ratio on their surface.Few papers have been published in recent years on XPSanalyses of iron-containing amphibole asbestos and most ofthem are reporting the Mg/Si or Fe/Si ratio at fibers’ surfaces[13–17] as an immediate parameter for comparing fibersused in in vitro tests. The only investigation concerning theinterpretation of Fe2p XPS signals is, as far as the presentauthors know, that of Gold et al. [18]. These authorssuggested that surface Fe(III) ions may play an importantrole in fiber-related carboxylate radical formation, but nodetails were provided on Fe2p curve-fitting procedure.

In the present work, the XP-Fe2p signal was processedaccording to two different approaches. The oxidation stateof iron located at the fibers surfaces is then compared tothat of the bulk determined by Mössbauer spectroscopy.

Experimental

Materials

The minerals investigated in this work were: a sample ofcrocidolite from South Africa supplied by the UnionInternationale Contre le Cancer (UICC), two fibrous tremo-lite samples coming from Italian ophiolitic outcrops, Ala diStura (Piedmont, Italy) and San Mango (Calabria, Italy), anda sample of fibrous tremolite from the ophiolitic complex ofMontgomery County, Maryland (USA). In addition tohematite (α-Fe2O3) and maghemite (γ-Fe2O3), a syntheticmagnetite Fe3O4 (kindly provided by L. Stievano, Labora-toire de Réactivité de Surface, University Pierre et MarieCurie, Paris VI) was also investigated as reference material.

X-ray powder diffraction (XRPD) preliminarily ascer-tained the absence of accessory phases in the amphibolesamples.

57Fe Mössbauer spectroscopy

Amphibole asbestos and reference synthetic magnetite wereinvestigated. The amphibole fibers were gently ground inan agate mortar with acetone and mixed with a powderedacrylic resin to avoid (or reduce) preferred orientation.About 0.100 g of tremolites, about 0.010 g of UICCcrocidolite, and 0.007 g of Fe3O4 were investigated, so thatthe absorber was within the limits for the thin absorberthickness described by Long et al. [19]. Data were collectedat room temperature, using a conventional spectrometersystem (Elscint MVT-3 transducer, MD-3 driving unit,LND Kr-gas proportional counter) operated in constantacceleration mode, with a 57Co source of nominal strength

2890 M. Fantauzzi et al. Author's personal copy

of 25 mCi in rhodium matrix, and recorded in amultichannel analyzer (Takes 269) using 512 channels.For magnetite reference material, the velocity rangebetween −10 and 10 mm/s was investigated. For amphibolesamples, since XRPD check excluded the presence ofmagnetic accessory phases, the velocity range between −4and 4 mm/s was investigated. After velocity calibrationagainst a spectrum of high-purity α-iron foil (25 μm thick),the raw data were folded to 256 channels. The spectrumwas fitted using Recoil 1.04 fitting program [20].

For magnetite, pure Lorentzian line-shapes (sextets)were fitted, and a very satisfactory reproduction of theexperimental spectrum was obtained with two sextets(Fig. 1). For amphibole samples, a first cycle of refinementwas performed by fitting pure Lorentzian doublets, but line-widths obtained were too large. Therefore, a second cycleof refinement was carried out by fitting quadrupole-splittingdistributions (QSD), following the approach of Gunter et al.and Gianfagna et al. [21, 22]. A number of fitting modelswith unconstrained parameters [center shift (δ0), couplingparameter (δ1), center of Gaussian component (Δ0),Gaussian width (σ∆), and absorption area (A)] were triedin order to obtain the best fit. A model based on two sites—1Fe(II) + 1Fe(III), with two components each—was finallychosen (Fig. 2). The uncertainties have been calculatedusing the covariance matrix, and errors were estimated noless than ±3% for both Fe(II) and Fe(III) absorption areas.

Measurement by XPS and sample preparation

The surface composition of the tremolite samples wasinvestigated with X-ray photoelectron spectroscopy. Anal-

yses were performed with a Theta Probe (Thermo FisherScientific, Waltham MA, USA). The samples were gentlyground in an agate mortar with analytical-grade acetone andwere mounted on a double-sided tape on a standard platen.The residual pressure during the analysis was always lowerthan 10−7 Pa. Spectra were collected using a 300-μmmonochromatic beam (Al Kα1,2 E=1,486.6 eV) operated at4.7 mA and 15 kV (70 W). The instrument is also equippedwith an ion gun and a combined low-energy-electron/ion-flood gun for charge compensation. The points of intereston the samples were selected and focused using a high-resolution camera. The position, intensity, and the diameterof the X-ray beam were controlled with a phosphor screen.The analyzer was operated in fixed analyzer transmissionmode at 100-eV pass energy for collecting the high-resolution spectra and at 200 eV for the survey spectrawith a step size of 0.05 and 1 eV, respectively. The fullwidth at half maximum (FWHM) height of the Ag3d5/2signal recorded in the same experimental conditions was0.83 eV. High-resolution regions acquired were: Fe2p,Si2p, O1s, Mg1s, Na1s, Al2p, and Ca2p. To verify thelinear response of Theta Probe spectrometer, a periodiccalibration was performed according to [23]. Data wereacquired under computer control (Avantage v3.45). Bindingenergy values were corrected by referencing the aliphaticcarbon from residual oil to 285.0 eV.

Spectral processing

The spectra were processed using CASAXPS software [24].An iterated Shirley–Sherwood background subtraction wasapplied before curve fitting [25]. Surface quantitative

Fig. 1 Room temperature (RT) 57Fe Mössbauer absorption spectrumof reference Fe3O4 (synthetic magnetite), Lorentzian fit. Crosses =experimental data; thick red line = calculated spectrum; thin blueline = Fe(III) absorption; thin green line = Fe(II)–Fe(III) intervalencecharge transfer absorption

Fig. 2 Room temperature (RT) 57Fe Mössbauer spectra of fibrousamphiboles, QSD fit. Crosses = experimental data; thick red line =calculated spectrum; thin blue line = Fe(III) absorption; thin greenline = Fe(II) absorption

Iron oxidation state in amphibole asbestos by XPS and Mössbauer techniques 2891 Author's personal copy

analysis of the amphibole fibers was determined accordingto [26], assuming the sample homogeneity. Experimentalareas were corrected for the photoelectric cross sections, σ[27], for the asymmetry parameter describing the anisotropyof photoemission [28] and for the intensity analyzerresponse [29]. Inelastic mean free path was calculatedusing G 1 equation [30]. The accuracies for the calculatedatomic concentrations are estimated to be ±10%.

Results and discussion

Mössbauer spectroscopy

The Mössbauer spectrum of reference magnetite (Fig. 1)was reproduced by two sextets with center shift δ=0.29 and0.67 mm/s and magnetic hyperfine field H=49 and 46 T,respectively. In agreement with literature [31], the first onewas assigned to Fe(III) and the second to an intervalencecharge transfer between Fe(II) and Fe(III), conventionallyreferred to as Fe2.5+. The relative proportions of these twospecies (with respect to Fetot) are 34% for Fe(III) and 66%for Fe2.5+ (Table 1). In order to obtain the real Fe(III)proportion, the first value must be summed to the half ofthe second one, and the result is in perfect agreement withmagnetite stoichiometry (Fe3+/Fetot=0.67).

The Mössbauer spectra of the investigated amphiboleasbestos show main absorptions in the central right region(that is, between 0 and 3 mm/s, Fig. 2). This is commonlyexpected for silicates, since they are paramagnetic materi-als, and is in agreement with spectra previously collectedfrom amphiboles [21, 22, 32, 33]. The magnetic contribu-tion from Fe(II)–Fe(III) oxides/hydroxides, due to accesso-ry phases and/or surface alteration, was not observed. Forall samples, the experimental absorption spectra werereproduced by contributions of isolated Fe(II) and Fe(III)species in octahedral coordination. In particular, fortremolite samples, the contribution of Fe(II) was modeledby two components, with δ0 approximately 1.1 mm/s and aroughly bimodal QSD, with probability density distributioncentered at Δ0 values of 1.9 and 2.9 mm/s (Fig. 3a andTable 2). Such values are in very close agreement withprevious studies on tremolite ([21, 34] and referencestherein). For UICC crocidolite, Fe(II) contribution was

modeled by two components very close to each other (withprobability density distribution centered at Δ0 values of 2.6and 2.9 mm/s, Table 2), in good agreement with a previousstudy on UICC crocidolite [32].

For all samples, the contribution of Fe(III) was modeledby two components, both centered at δ0 approximately0.3 mm/s. However, the QSD fit revealed probabilitydensity distributions with Δ0 values very different fromsample to sample (Fig. 3b and Table 2), as commonlyobserved in previous studies (e.g., [22, 33]). The Fe(III)/

Table 1 Room temperature Mössbauer parameters for the reference synthetic magnetite

Sample δ (mm/s) H (T) Γ (mm/s) Assignment Area (%) χ2

Magnetite 0.29 49 0.37 Fe(III) 34 0.64

0.67 46 0.48 Fe2.5+ 66

Lorentzian line-shape (sextet) fitting model. Errors were estimated at about ±0.2 mm/s for center shift (δ) and peak width (Γ), ±0.5 T for magnetichyperfine field (H) and no less than ±2% for sextet areas. EFG perturbation parameter (ε) was very close to zero

Fig. 3 Details of the QSD fit adopted for the fibrous amphiboles: aprobability density distribution for Fe(II); b probability densitydistribution for Fe(III). Data shown are from Ala di Stura tremolite

2892 M. Fantauzzi et al. Author's personal copy

Fetot ratios were quantified on the basis of spectral areameasurement, Fe(III) raw, and corrected for the temperatureeffect applying the empirical factor of [36]. The correctedFe(III) ratios, Fe(III)corr, are reported in Table 2.

X-ray photoelectron spectroscopy

Figure 4 shows the survey spectra of all fibrous samples:oxygen, silicon, iron, and magnesium were identifiedtogether with small amounts of carbon as contaminant.The UICC crocidolite spectrum showed also a small signalat 1,072 eV that supports the presence of small amount ofNa. Al was only detected in San Mango tremolite whereas

Ca was found in all the tremolites samples. The high-resolution spectra of the most intense signals were collectedand curve-fitted with CasaXPS software. The bindingenergy values obtained after the curve fitting are summa-rized in Table 3.

Si2p signal was found at 102.5±0.2 eV for all amphibolefibers, and no significant shifts in binding energy valuesamong different amphiboles fibers might be observedwithin the experimental uncertainty. Mg2p binding energieswere in the range between 49.3±0.2 eV (Ala di Stura andMaryland samples) and 49.8±0.2 eV (San Mango tremo-lite). The agreement with the tremolite binding energyvalues reported in [13] is excellent, and it also holds in thecase of an unknown asbestos sample [14]. The slightlylower binding energy values of Si2p and Mg2p signals

Table 2 Room temperature Mössbauer parameters for the amphibole asbestos

Sample χ2 Fe(II) Fe(III) Fe(III)raw(% Fetot)

Fe(III)corr(% Fetot)

Δ0 (mm/s) sΔ (mm/s) δ0 (mm/s) Δ0 (mm/s) sΔ (mm/s) δ0 (mm/s)

UICC crocidolite 3.16 2.92 0.03 1.07 1.00 1.86 0.37 57 522.64 0.38 1.07 0.43 0.18 0.37

San Mangotremolite

0.83 1.89 0.17 1.19 1.32 0.97 0.30a 7 62.91 0.15 1.19 6.06 0.00 0.30a

Ala di Sturatremolite

0.79 1.89 0.20 1.19 1.02 0.32 0.25 15 122.92 0.17 1.19 2.16 0.62 0.25

Maryland tremolite 1.46 1.90 0.18 1.19 1.30 1.46 0.30a 18 162.90 0.17 1.19 6.06 0.61 0.30a

Fitting approach: quadrupole-splitting distributions (QSD). Symbols according to Rancourt and Ping [35]. Center shift (δ0) with respect toα-iron; δ1 always lower than 0.05; γ=0.194 mm/s; h+/h− = 1. The Fe(III)raw is obtained from the area of absorption peaks assigned to Fe(III). TheFe(III)corr is obtained from the raw value by applying the correction factor C=1.22 [35]. Estimated uncertainties are about 0.02 mm/s for hyperfineparameters and no less than 3% for absorption areasa Kept fixed throughout the refinement

Fig. 4 XPS survey spectra of: (a) UICC crocidolite; (b) San Mangotremolite; (c) Ala Stura tremolite; (d) Maryland tremolite. Al KαX-ray source

Table 3 Binding energy (eV) values of the most intense photoelec-tron signals of UICC crocidolite, San Mango tremolite, Ala di Sturatremolite, and Maryland tremolite

Binding Energy (eV)

UICCcrocidolite

San Mangotremolite

Ala di Sturatremolite

Marylandtremolite

Si 2p 102.4 102.6 102.5 102.1

Mg 2p3/2 49.4 49.8 49.3 49.3

Al 2p nd 74.3 nd nd

Ca 2p3/2 nd 347.6 347.8 347.1

Na 1 s 1072.4 nd nd nd

Binding energy values were corrected by referencing the aliphaticcarbon to 285.0 eV; uncertainty is ± 0.2 eV

nd not detected

Iron oxidation state in amphibole asbestos by XPS and Mössbauer techniques 2893 Author's personal copy

measured on the Maryland sample may be attributed tothe lower Mg/Ca atomic ratio (Table 4). It was reportedthat an increase of the Mg/Ca ratio shifts the signals towardhigher binding energy values due to the different covalencyof Si–O bonds and Mg–O bonds when in the presence ofCa2+ [13].

As far as UICC crocidolite sample is concerned, Si2pand Mg2p binding energy values are similar to those oftremolites despite the absence of calcium ions. This mightbe due to a similar effect exerted by sodium ions in thelattice. The influence of the presence of considerableamounts of iron in the lattice on binding energy valuescannot be ruled out: the effect of this element is still notwell understood [13]. The oxygen signals are multicompo-nent too, and they could be fitted with three Gauss–Lorentzcurves found at 530.0±0.2, 531.2±0.2, and 532.3±0.2 eV.The first one could be attributed to oxygen in oxides (O2−)[37, 38]; the second component might be ascribed to bothnonbridging oxygen atoms in silicates [39, 40] and oxygenin –OH groups of oxyhydroxides species [38]; the thirdcomponent is characteristic of bridging oxygen atoms insilicates [39, 40].

The surface compositions, as determined by XPSanalyses (Table 3), were in good agreement with the bulkcompositions for all samples except UICC crocidolite,which showed an Fe surface content half to that of the

bulk (Table 5 and Fig. 5). The presence of carboncontamination is proven not to affect quantitative analysesresults. Contamination layer thickness, estimated accordingto [41], ranged between 0.1 nm (San Mango tremolite) and0.4 nm (Ala di Stura Tremolite). Their effect on quantitativecomposition is proven to be negligible (see OnlineResources 1).

Iron oxidation state in tremolite asbestos: surface vs bulk

Much work was devoted to determine the oxidation state ofiron in amphiboles fibers. The Fe2p spectrum of Fe(II) andFe(III) compounds is known to be difficult to processbecause of the broadened line-shape if compared with Fe(0). The high-resolution Fe 2p3/2 signal was resolved intoits components by applying two different approaches. Thefirst one is based on the results of the calculation ofmultiplet structures proposed by [42, 43]. They calculatedthe multiplet structure expected for core p-levels Fe(II) andFe(III) ions, taking into account the electrostatic andspin-orbit interactions and using the Hartree–Fock approx-imation for free ion. These models are particularly usefulfor processing the spectra of highly ionic solids. This curve-fitting procedure was first tested on hematite (α-Fe2O3),maghemite (γ-Fe2O3) (see Supplementary materials), andthen on the same Fe3O4 sample used for the Mössbauerspectroscopy. The latter was subjected to the same grindingand mounting procedure adopted for asbestos. The multi-plet peak parameters used for fitting the Fe3O4 spectrum arelisted in Table 6. The relative positions, relative heights ofthe subpeaks, and FWHM were kept constant; only thepeak height of the first component was varied to optimizethe fit. While the fits of hematite and maghemite (notshown here) are in agreement with the data reported in theliterature, the Fe(III)/Fetot ratio of the magnetite sample wasfound to be 0.80. This value is far from the 0.67 valueobtained for this ratio for the bulk by Mössbauer spectros-copy (Table 1). This result may be attributed to theoxidation of the sample surface in agreement with the

Table 4 Quantitative analysis in atomic percent of fibrous amphibolescalculated by applying the first principle approach [26] to X-rayphotoelectron spectra (uncertainty ± 10%)

Fe O Si Na Mg Al Ca

UICC crocidolite 5.2 64.5 23.2 3.3 3.8 – –

San Mangotremolite

0.6 57.7 21.2 – 15.4 0.8 4.3

Ala Stura tremolite 0.5 58.3 23.9 – 12.6 – 4.6

Maryland tremolite 1.4 62.6 24.1 – 7.9 – 4.0

Table 5 Bulk compositions in atomic percent of the investigated fibrous amphiboles

Fe O Si Na Mg Al Ca

UICC crocidolitea 10.7 61.5 20.7 4.1 2.2 0.1 0.7

San Mango tremoliteb 0.9 61.5 20.5 0.1 12.0 – 5.0

Ala di Stura tremolitec 0.7 61.5 20.5 0.1 12.2 – 5.0

Maryland tremoliteb 1.4 61.5 20.5 0.1 11.5 – 5.0

Uncertainty ± 1% for major elements and up to 10% for minor elementsa Obtained by ICPb Obtained by electron microprobe analysis (work in progress)c Obtained by electron microprobe analysis [44]

2894 M. Fantauzzi et al. Author's personal copy

results of McIntyre et al. [45] that investigated a sampleobtained using iron metal as starting material for itssynthesis. This approach was also applied to the Fe 2p3/2spectrum of the amphibole fibers (Online Resources 2): theonly changes needed to optimize the fits were the variation

of the heights of the peaks. The Fe(III)/Fetot ratios obtainedfor the amphibole fibers are shown in Table 7. These valuesare higher than the ratios calculated from Mössbauer data(Table 2), and the lack of agreement is attributable todifferent factors: the separation among peaks of multipletstructure and their relative intensities depend on thecompound and that the fit of a signal, which includes thecontributions of iron in different chemical states, is verycomplicated when using so many peaks. Different multipletpeak parameters for the Fe2p signal in α-Fe2O3, γ-Fe2O3,FeO, or Fe3O4 have been reported in [45]. It has beenpreferred here to resort to a different approach, referred inthe following as II approach. The peak-fitting parameters(Table 8) are those obtained by summing the multipletsplitting components in a single model function ([37] and

Table 6 Curve fit parameters obtained on Fe3O4 using the I approach[41, 42]

Peak Binding energy(eV), s ± 0.2 eV

FWHM (eV),s ± 0.1 eV

Area

Fe(III) peak 1 710.1 Unconstrained

Fe(III) peak 2 BE peak1+1.1

1.8 Area peak1×0.95

Fe(III) peak 3 BE peak1+2.2

1.8 Area peak1×0.57

Fe(III) peak 4 BE peak1+3.3

1.8 Area peak1×0.31

Fe(II) peak 1 708.0±0.2 1.8 Unconstrained

Fe(II) peak 2 BE peak1+1.1

1.53 (peak1×0.85)

Area peak1×1.04

Fe(II) peak 3 BE peak1+2.3

1.8 Area peak1×0.37

Fe(II) peak 4 BE peak1+7.0

3.3 Area peak1×0.25

[Fe(II)shakeupsatellite]

BE binding energy

Table 7 Fe(III)/Fetot intensity ratio calculated applying two differentcurve-fitting approaches

Sample Fe(III)/FetotI approach

Fe(III)/FetotII approach

Fe3O4 0.80 0.68a

UICC Crocidolite 0.61 0.80

San Mangotremolite

0.63 0.65

Maryland tremolite 0.63 0.70

a Fe(III)/Fetot calculated excluding FeOOH contribution

Fig. 5 Comparison of Fe/Si, Mg/Si, and Ca/Si atomic ratios betweenbulk and surface of asbestos samples

Iron oxidation state in amphibole asbestos by XPS and Mössbauer techniques 2895 Author's personal copy

references cited therein). This approach allows fitting thespectra with only three or four curves. The XP high-resolution fitted spectra of Fe2p3/2 of crocidolite andtremolites are shown in Fig. 6. Each signal showed fourcontributions at 709.0, 710.5, 711.6, and 714.5 eV. Thesignals at 709.0 eV (with its shakeup satellite at 714.5 eV)and 710.5 eV were assigned to Fe(II) and Fe(III) oxide,respectively. The signal at 711.6 eV was attributed to ironoxyhydroxide (FeOOH; Table 8). These assignments are inagreement with the literature [37, 38, 45–47]. Curve-fittingdetails of the Fe2p3/2 and 1/2 doublet are reported in OnlineResources 3. It is worthwhile to underline that in the case ofthe Ala di Stura tremolite sample it was not possible to resolvethe peak in its Fe(II) and Fe(III) components; this was dueto the small amount of iron in the sample and the consequentsmall signal-to-noise ratio of the Fe 2p3/2 peak. The Fe(III)/

Fetot ratio obtained for Fe3O4 using the II approach is 0.89when the contribution of the iron oxyhydroxide signal wasincluded in the calculation . Iron oxyhydroxide is very likelylocated in the outer part of the sample surface and may beconsidered as the product of the interaction of the magnetitewith the ambient atmosphere. The Fe(III)/Fetot ratio wasrecalculated, excluding the area of the signal assigned to thehydroxides, and a value of 0.68 was obtained. This is in verygood agreement with the ratio of 0.67 measured byMössbauer spectroscopy, and thus the II approach turns outto be more accurate. Results obtained based on the IIapproach for the fibrous amphiboles are shown in Table 7.For all samples, a difference between surface and bulk Fe(III)/Fetot ratios was observed (Fig. 7). In particular, SanMango tremolite at the surfaces is ten times more oxidized.In addition, a significant amount of the Fe(III) is present atthe surface as FeOOH oxyhydroxide (Table 9). It has to beunderlined that asbestos surfaces are not modified during thesample preparation [13]; therefore, the presence of ironoxyhydroxide is peculiar in the studied fibers. Moreover,such presence is in perfect agreement with the products ofweathering of silicate minerals [48, 49].

Fig. 6 Shirley background subtracted Fe 2p3/2 spectra curve fit for:(a) UICC crocidolite; (b) San Mango tremolite; (c) Marylandtremolite. Fit based on II approach

Fig. 7 Comparison of Fe (III)/Fetot atomic ratios between bulk(Mössbauer results) and surfaces (XPS results) of asbestos samples

Table 8 Curve-fitting parameters (CASA XPS software, v 2.3.15) ofthe Fe2p3/2 signal using the II approach

Component Binding energy (eV),s ± 0.2eV

FWHM (eV),s ± 0.1eV

% G/La

Fe(II) oxide 709.0 3.0 55

Fe(III) oxide 710.5 3.0 55

FeOOH 711.6 4.1[Fe(II)×1.35]

55

Fe(II) oxide satellite 714.5[BE Fe(II) + 5.5]

3.0 55

a A Gauss–Lorentz product function was used for curve fitting; the G/Lratio was kept fixed

Table 9 Fe(II) oxide/Fetot, Fe(III) oxide/Fetot, and Fe(III) hydr/Fetotratios on the surface of the investigated fibrous amphiboles

Sample Fe(II) ox/Fetot

Fe(III) ox/Fetot

Fe(III) hydr/Fetot

UICC crocidolite 0.20 0.18 0.62

San Mangotremolite

0.35 0 0.65

Marylandtremolite

0.30 0.10 0.60

2896 M. Fantauzzi et al. Author's personal copy

Iron content is proven to be closely related to asbestostoxicity: as an example, tests performed on isolatedmitochondria in contact with chrysotile and crocidolite(characterized by very low and very high iron contents,respectively) showed that crocidolite induced more damagesto cells and enhanced the mitochondrial production of ROS[50]. An investigation on the mortality of lung cells incontact with asbestos-like fluoro-edenite fibers has beenrecently carried out by [51], and similar experiments areongoing on exactly the asbestos studied here. The prelimi-nary results of cytotoxicity tests with 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT assay) revealeda significant mortality of the cells kept in contact with oursamples and in addition evidenced a delay of mortality forthe cells in contact with San Mango tremolite with respect tothose tested with other amphibole asbestos [52]. This isparticularly important since the XPS investigation on SanMango tremolite showed that most of the Fe at the surface isoxidized and present as FeOOH. On such basis, the observeddelay of cell mortality may be explained with the first step ofthe iron-catalyzed Haber–Weiss cycle [3], for which the Fe(III) must be previously reduced to Fe(II) to produce a strongrelease of ROS:

1. Fe3þ þ reductant ! Fe2þ þ oxidized reductant2. Fe2þ þ O2 ! Fe3þ þ O��

2

3. O��2 þ 2Hþ þ e� ! H2O2

4. Fe2þ þ H2O2 ! Fe3þ þ OH�þ�OH

Conclusions

Amphibole asbestos were analyzed by XPS and Mössbauerspectroscopy, with the aim to determine the surface vs. bulkFe(III)/Fetot ratios. Results obtained showed that:

1. Mössbauer spectroscopy revealed for all fibrous sam-ples the presence in the bulk of isolated Fe(II) and Fe(III) octahedrally coordinated. In particular, tremolitesamples are Fe(II) dominant, with Fe(III)/Fetot ratiosspanning from 6% to 16%.

2. XPS is demonstrated to be a very powerful analyticaltool for the physicochemical characterization of fibrousamphiboles.

3. XPS investigation pointed out that the curve fitting ofFe2p3/2 signal of complex minerals such as asbestoscan be successfully performed using convolutionfunctions based on multiplet splitting approach.

4. For the first time, an evidence was brought that thesurface of the studied samples is enriched in Fe(III),which is mainly present as FeOOH. The differencebetween surface and bulk Fe oxidation state isparticularly marked in tremolite samples.

5. The presence of oxidized Fe on the San Mangotremolite surface allowed us to justify the delay of cellmortality observed in the cytotoxicity tests.

Acknowledgements The Universities of Cagliari and of Roma “LaSapienza” are acknowledged for the financial support.

References

1. Van Oss CJ, Naim JO, Costanzo PM, Gieise RF Jr, Wu W,Sorlung AF (1999) Impact of different asbestos species and othermineral particles on pulmonary pathogenesis. Clays Clay Miner47:697–707

2. Kane AB (1996) Mechanisms of mineral fibre carcinogenesis. In:Kane AB, Boffetta P, Saracci R, Wilbourn J (eds) IARC scientificpublication 140. International Agency for Research on Cancer,Lyon

3. Fubini B, Otero Aréan C (1999) Chemical aspects of the toxicityof invale mineral dusts. Chem Soc Rev 28:373–381

4. Kamp DW, Weitzman SA (1999) The molecular basis of asbestosinduced lung injury. Thorax 54:638–652

5. Robledo R, Mossman R (1999) Cellular and molecularmechanisms of asbestos-induced fibrosis. J Cell Physiol180:158–166

6. Stanton MF, Layard M, Tegeris A, Miller E, May M, Morgan E,Smith A (1981) Relation of particle dimension to carcinogenicityin amphibole asbestoses and other fibrous minerals. J Natl CancInst 67:965–975

7. Fubini B (1993) In: Wahreit DB (ed) Fiber toxicology. Academic,San Diego

8. Fubini B (1996) Use of physico-chemical and cell free assays toevaluate the potential carcinogenicity of fibres. In: Kane AB,Boffetta P, Saracci R, Wilbourn J (eds) IARC scientific publica-tion 140. International Agency for Research on Cancer, Lyon

9. Gilmour PS, Brown DM, Beswik PH, Macnee W, Rahman I,Donaldson K (1997) Free radical activity of industrial fibers: roleof iron in oxidative stress and activation of transcription factors.Environ Health Persp 105(Suppl 5):1313–1317

10. Fubini B, Fenoglio I, Elias Z, Poirot O (2001) On the variabilityof the biological responses to silicas: effect of origin, crystallinityand state of the surface on the generation of reactive oxygenspecies and consequent morphological transformations in cells. JEnviron Pathol Toxicol Oncol 20:87–100

11. Gazzano E, Riganti C, Tomatis M, Turci F, Bosia A, Fubini B,Ghigo D (2005) Potential toxicity of nonregulated asbestiformminerals: balangeroite from the western Alps. Part 3: depletion ofantioxidant defenses. J Toxicol Environ Health 68:41–49

12. Favero-Longo SE, Castelli D, Salvadori O, Belluso E, PiervittoriR (2005) Pedogenetic action of the lichens Lecidea atrobrunnea,Rhizocarpon geographicum gr. and Sporastatia testudinea onserpentinized ultramafic rocks in an alpine environment. IntBiodeterior Biodegrad 56:17–27

13. Seal S, Krezosky S, Barr TL, Petering DH, Klinowsky J, EvansPH (1996) Surface chemistry and biological pathogenicity ofsilicates: an X-ray photoelectron spectroscopic study. Proc R SocLond B 263:943–951

14. Seal S, Krezosky S, Petering D, Barr TL, Klinowsky J, Evans P(1996) X-ray photoelectron spectroscopy investigations of theinteraction of cells with pathogenic asbestoses. J Vac Sci TechnolA 14:1770–1778

Iron oxidation state in amphibole asbestos by XPS and Mössbauer techniques 2897 Author's personal copy

15. Seal S, Krezosky S, Barr TL, Petering D, Evans PH, Klinowsky J(1997) Surface chemical interaction of fibrous asbestos withbiocells: an ESCA study. J Hazard Mate 53:57–74

16. Keane MJ, Stephens JW, Zhong BZ, Miller WE, Ong TM,Wallace WE (1999) A study of the effect of chrysotile fibresurface composition on genotoxicity in vitro. J Toxicol EnvironHealth 57:529–541

17. Shen Z, Bosbach D, MF H Jr, Bish DL, Williams MG, DodsonRF, Aust AE (2000) Using in vitro iron deposition on asbestos tomodel body formation in human lung. Chem Res Toxicol 13:913–921

18. Gold J, Amandusson H, Krozer A, Kasemo B, Ericsson T, ZanettiG, Fubini B (1997) Chemical characterization and reactivity ofiron chelator-treated amphibole asbestos. Health Persp 105(suppl5):1021–1030

19. Long GJ, Cranshaw TE, Longworth G (1983) The ideal Mössbauereffect absorber thickness. Möss Eff Ref Data J 6:42–49

20. Lagarec K, Rancourt DG (1998) RECOIL. Mössbauer spectralanalysis software for Windows, version 1.0. Department ofPhysics, University of Ottawa, Canada

21. Gunter ME, Dyar MD, Twamley B, FF F Jr, Cornelius C (2003)Composition, Fe3+/ΣFe, and crystal structure of non-asbestiformand asbestiform amphiboles from Libby, Montana, USA. AmMineral 88:1970–1978

22. Gianfagna A, Andreozzi GB, Ballirano P, Mazziotti-Tagliani S,Bruni BM (2007) Structural and chemical contrasts betweenprismatic and fibrous fluoro-edenite from Biancavilla, Sicily, Italy.Can Mineral 45:249–262

23. Seah MP (2001) ISO 15472:2001—surface chemical analysis. X-ray photoelectron spectrometers—calibration of energy scales.Surf Interface Anal 31:721–723

24. Fairley N (1999–2003) CasaXPS version 2.3.1525. Shirley DA (1972) High resolution X-ray photoemission spectrum

of the valence bands of gold. Phys Rev B 5:4709–471426. Seah MP (2003) Quantification in AES and XPS. In: Briggs D,

Grant JT (eds) Surface analysis by auger and X-ray photoelec-tron spectroscopy. IM Publication Surface Science Spectra,Manchester

27. Scofield JH (1976) Hartree-Slater subshell photoionization cross-sections at 1254 and 1487 eV. J Electron Spectrosc Relat Phenom8:129–137

28. Reilman RF, Msezane A, Manson STJ (1976) Relative intensitiesin photoelectron spectroscopy of atoms and molecules. J ElectronSpectrosc Relat Phenom 8:389–394

29. Scorciapino A (2007) Caratterizzazione di leghe NiP nanocristal-line mediante Spettroscopia di Fotoelettronica a Raggi X. PhDThesis, University of Cagliari

30. Gries WH (1996) A universal predictive equation for the inelasticmean free path lengths of X-ray photoelectrons and augerelectrons. Surf Interface Anal 24:38–50

31. Stevens JG, Khasanov AM, Miller JW, Pollak H, Li Z (1998)Mössbauer mineral handbook. Mössbauer effect data center.Biltmore, Asheville

32. Stroink G, Blaauw C, White CG, Leiper W (1980) Mössbauercharacteristics of UICC standard reference asbestos samples. CanMineral 18:285–290

33. Andreozzi GB, Ballirano P, Gianfagna A, Mazziotti-Tagliani S,Pacella A (2009) Structural and spectroscopic characterization of asuite of fibrous amphiboles with high environmental and healthrelevance from Biancavilla (Sicily, Italy). Am Mineral 94:1333–1340

34. Ballirano P, Andreozzi GB, Belardi G (2008) Crystal chemicaland structural characterization of fibrous tremolite from Susa

Valley, Italy, with comments on potential harmful effects onhuman health. Am Mineral 93:1349–1355

35. Rancourt DG, Ping JY (1991) Voigt-based methods for arbitrary-shape static hyperfine parameter distributions in Mossbauerspectroscopy. Nucl Instrum Meth B 58:85–87

36. Dyar MD, Mackwell SM, McGuire AV, Cross LR, RobertsonJD (1993) Crystal chemistry of Fe3+ and H+ in mantlekaersutite: implications for mantle metasomatism. Am Mineral78:968–979

37. Olla M, Navarra G, Elsener B, Rossi A (2006) Nondestructive in-depth composition profile of oxy-hydroxide nanolayers on ironsurfaces from ARXPS measurement. Surf Interface Anal 38:964–974

38. Cutting RS, Coker VS, Fellowes JW, Lloyd JR, Vaughan DJ(2009) Mineralogical and morphological on the reduction of Fe(III) minerals by Geobacter sulfurreducens. Geochim CosmochimActa 73:4004–4022

39. Zakaznova-Herzog VP, Nesbitt HW, Bancroft GM, Tse JS (2006)High resolution core and valence band XPS of non-conductorpyroxenes. Surf Sci 72:69–86

40. Zakaznova-Herzog VP, Nesbitt HW, Bancroft GM, Tse JS (2008)Characterization of leached layers on olivine and pyroxenes usinghigh resolution XPS and density functional calculations. GeochimCosmochim Acta 72:69–86

41. Smith GC (2005) Evaluation of a simple correction for thehydrocarbon contamination layer in quantitative surface analysisby XPS. J Electron Spectrosc Relat Phenom 148:21–28

42. Gupta RP, Sen SK (1974) Calculation of multiple structure of corep-vacancy levels. Phys Rev 10:71–77

43. Gupta RP, Sen SK (1975) Calculation of multiple structure of corep-vacancy levels II. Phys Rev 12:15–19

44. Pacella A, Andreozzi GB, Ballirano P, Gianfagna A (2008) Crystalchemical and structural characterization of fibrous tremolite fromAla di Stura (Lanzo Valley, Italy). Period Miner 77:51–62

45. McIntyre NS, Zetaruk DG (1977) X-ray photoelectron spectro-scopic studies of iron oxides. Anal Chem 49:1521–1529

46. Grosvenor AP, Kobe BA, Biesinger MC, McIntyre NS (2004)Investigation of multiplet splitting of Fe 2p XPS spectra andbonding in iron compounds. Surf Interface Anal 36:1564–1574

47. Mathieu HJ, Landolt D (1986) Investigation of thin oxide filmsthermally grown in situ on Fe-24Cr and Fe-24Cr-11Mo by Augerelectron spectroscopy and X-ray photoelectron spectroscopy.Corros Sci 26:547–559

48. Schott J, Berner RA (1983) X-ray photoelectron studies of themechanisms of iron silicate dissolution during weathering. Geo-chim Cosmochim Acta 47:2233–2240

49. Velbel MA (1993) Formation of protective surface layers duringsilicate-mineral weathering under well-leached, oxidizing condi-tions. Am Mineral 78:405–414

50. Bergamini C, Fato R, Biagini G, Pugnaloni A, Giantomassi F,Foresti E, Lesci GI, Roveri N, Lenaz G (2006) Mitochondrialchanges induced by natural and synthetic asbestos fibers: studieson isolated mitochondria. Cell Mol Biol 52:905–913

51. Pugnaloni A, Lucarini G, Giantomassi F, Lombardo L, Capella S,Belluso E, Zizzi A, Panico AM, Biagini G, Cardile V (2007) Invitro study of biofunctional indicators after exposure to asbestos-like fluoro-edenite fibres. Cell Mol Biol 53:965–980

52. Gianfagna A, Andreozzi GB, Ballirano P, Pacella A, Mazziotti-Tagliani S, Bruni BM, Paoletti L, Cardile V, Pugnaloni A,Giantomassi F, Fournier J, Stievano L (2008) Characterization offibrous tremolites of environmental and health interest. 33rdInternational Geological Congress (Session Earth and Health–Medical Geology), Oslo, Norway. # MGH-01835P (abstr)

2898 M. Fantauzzi et al. Author's personal copy