J - Mineralogical Society of America

Document Sample
J - Mineralogical Society of America Powered By Docstoc
					                                 American Mineralogist, Volume 80, pages 1093-1103, 1995

       Asbestiform riebeckite (crocidolite) dissolutionin the presence Fe chelators:
                         Implications for mineral-induceddisease

                            Axonrw J. WrnNnno* Mrcrurr, F. HocHrr,r,a, Jn.
                    Department of Geological Sciences,
                                                     Virginia Polytechnic Institute and State University,
                                           Blacksburg,Virginia 2406l, U.S.A.
                                             Groncr D. Gurnmno Jn.
                                      Jr.lNNB A. H,mnv,** ANx E. Ausr
                             Department of Chemistry and Biochemistry, Utah State University,
                                              Logan,Utah 84332,U.S.A.
                                                J. DoNar,o Rrusrror
                    Department of Geological Sciences,
                                                     Viryinia Polytechnic Institute and StateUniversity,
                                           Blacksburg,Viryinia 24061, U.S.A.

                X-ray photoelectron spectroscopy(XPS) and solution chemistry were used to monitor
              the changesin surfacecomposition of crocidolite fibers in a 50 mM NaCl solution at pH
              : 7.5 and 22"Cinthe presence severalFe chelators(citrate, EDTA, ordesferrioxamine)
              for up to 30 d. The data show that the introduction of Fe chelatorsdramatically increases
              the rate at which Fe is releasedfrom the surfacein comparison with a control goup to
              which no chelators were added. In particular, XPS shows that Fe3* is more efectively
              removed in the presenceof the chelators even though it is highly insoluble in aqueous
              solutions at near neutral pH. This suggests    that Fe chelators can alter the dissolution
              mechanism of amphiboles from the processthat dominatesin NaCl solutions.This change
              in dissolution mechanism (particularly the enhancedrate of Fe release)is an important
              consideration for models of mineral-induced pathogenesis   that rely on oxidation and re-
              duction processes.
                Efforts were made to estimate the Fe-releaselifetimes of crocidolite fibers under the
              conditions of our experiments to guide the assessment the biodurability of these fibers
              in human lung tissue. Our results suggestthat crocidolite fibers may persist for several
              years, releasingFe to lung fluids during this time. This estimated lifetime is longer than
              that previously estimated for chrysotile fibers and is consistent with the lifetimes previ-
              ously observedin asbestosmineral lung-burden studies.

                     INrnoougrroN                               (scarring oflung tissue, which results in a significant de-
   Asbestiform minerals were used widely in the past in         creasein blood oxygenation efficiency),mesothelioma (a
various industrial and nonindustrial applicationsbecause        malignant tumor of the pleura, the membrane lining the
of their fibrous shape, high tensile strength, flexibility,     cavities containing the lungs), and lung cancer,although
long-term durability, and fire resistance.The most com-         exactly which mineral is responsiblefor inducing a par-
monly used asbestosminerals were the l: I sheet silicate        ticular diseaseis often debated (e.g.,Wagner et al., 1960;
chrysotile [Mg,SirO,(OH)o] and the amphibole crocido-           Craigheadetal.,1982; Wagner,l99l; Guthrie and Moss-
lite [NarFei+(Fe2+,Mg2+                                         man, 1993). Unfortunately, the processes which cer-
                           )3Si8Orr(OH)r], fibrous vari-
ety of riebeckite. It is now known that asbestos minerals       tain fibrous minerals induce diseases not yet known,
have the ability to induce lung diseasessuch as asbestosis      although many studieshave investigatedthis problem (see
                                                                Guthrie and Mossman, 1993, and referencestherein).
                                                                Stanton et al. (1981) reported a positive correlationbe-
  t Presentaddress:Robinson and Noble, 5915 Orchard Street
                                                                tween the number of fibers, as well as their dimension,
West, Tacoma,Washington98467,U.S.A.
 ** Presentaddress:Department of Biochemistry and Molecular and the sample'sability to induce cancerfollowing direct
Biology, University of California at Berkeley,Berkeley,Califor- application to the pleural surfaceofrats. They concluded
nia9472O-3202,   U.S.A.                                         that fibers with a diameter <0.25 pm and a length >8
0003-004x/95ll I 2-1093$02.00
             I                                              1093

pm were the most carcinogenic.This is referred to as the          sessing how an agent interacts with DNA, and SSBs
Stanton hypothesis. The Stanton hypothesis contributed            themselvesmay be an important component of the dis-
significantly to the current restrictive regulations and          easeprocess.
guidelines on asbestos    use and exposurein the U.S. (Vu,           One mechanism for the generation of         createdby
1993). In a recent paper, Nolan and Langer (1993) re-             reactionsinvolving Fe either on the surfaceofor released
viewed what are now recognizedas major limitations in             from amphibole asbestosminerals, is shown in the fol-
the Stanton studies; they stated that, although the fiber         lowing seriesof reactions referred to as the modifled, Fe-
dimension vs. pathogenicity(disease       generation)relation-    catalyzedHaber-Weissreactions(e.9.,Zalma et al., 1987   a,
ship may hold true in some instances,the overall findings          1987b;Hardy and Aust, 1995):
ofStanton et al. (1981) are inconclusiveinsofar as their
relevanceto human disease.       Besidesthe fibrous nature of           Reductant('t* Fe3++ reductant("+r) Fe2+
                                                                                                         +              (l)
certain minerals, there are several other mineralogical                          Fe2+* O, -     Fe3* * Oz-              Q)
properties that can potentially control their deleterious
effectson our respiratory systems.For example, the abil-                 HO; + O, * H+ + 02 + H2O2                      (3)
ity of physiologicalfluids to leach constituentmetals from
                                                                              Fe2+ HrOr-Fe3+ * OH- +'OH
mineral fibers has been studied in test animals (e.g.,Mor-
gan et al., l97l), in human lungs (e.g.,Jaurand et al.,                                           (Fenton reaction).    (4)
 1977), and in vitro (e.g.,Hart et al., 1980).Related to
this, fiber durability (i.e., the lifetime of a fiber on the         BecauseFe3+is so immobile under near neutral con-
basis ofits dissolution rate) in lung tissueappearsto play        ditions in aqueous solutions (e.g., Baes and Mesmer,
an important role in pathogenesis(e.g., McDonald and              1976),leachingexperimentshave been conductedin the
McDonald, 1986a, 1986b).Surfacecharacteristics as-      of        presenceof Fe chelators(Aust and Lund, 1990, l99l;
bestosfibers, such as charge,composition, structure, and          Lund and Aust, 1992).It is well establishedthat chelators
microtopography may also influence fiber reaction in and          can promote mineral dissolution by interacting directly
perturbation of respiratory systems(e.g.,Hochella, 1993).         with metals on mineral surfaces  (e.g.,Hering and Stumm,
Even mechanical properties (stiffness,fragmenting char-            1990; Schindler, 1990; Stumm, 1992;'and many refer-
acteristics, etc.) may be important as fibers translocate         encestherein). Becausechelators are present in physio-
from the respiratory tract to the pleura.                         logical fluids, they also may be important in the disease
   A major implication of most of the studies mentioned           processin their ability to extract Fe efrciently from fiber
above is that there is an important need for an under-            surfaces  (Lund and Aust, 1990).
standing of how minerals chemically interact with phys-              There is much interest in Fe in studies dealing with
iological fluids. This is likely to be a necessarycomponent       mineral-induced diseasedespite the fact that hematite,
of any integratedmodel of mineral-induced pathogenesis.           for example, has negligible biochemical reactivity and er-
In this study, we investigated the surface chemistry of           ionite, an Fe-free zeolite, has high reactivity. This may
crocidolite fibers and particularly the surface chemistry         suggest  that there are severalpathways by which various
of Fe on thesefibers. We specificallychosecrocidolite for         minerals induce the same diseasein humans. It is also a
our work becauseit is clear from epidemiological studies          reminder that there are probably several factors that
that crocidolite is far more pathogenic in humans than            combine to result in the pathogenicity of minerals. He-
chrysotile for the induction of mesothelioma (e.g., Wag-          matite, as a nonfibrous mineral but one with very low
ner et al., 1960;McDonald and McDonald, 1986a).          Cro-     solubility, is much more likely to be physically cleared
cidolite can also induce lung cancer and asbestosis      (e.g.,   from the lung by means of the mucociliary escalatorthan
McDonald and McDonald, 1986b; McDonald, 1990).                    a fibrous mineral (e.g.,Lehnert, 1993).On the other hand,
Furthermore, the surfacechemistry of Fe on these fibers           Fe sorbed on the surface of erionite in vitro has been
is of special interest to us becauseof previous work on           shown to produce DNA SSBs,whereaserionite without
the generationofO radicals by asbestos      (seethe extensive     sorbed Fe does not (Eborn and Aust, 1995; Hardy and
review by Kamp er al., 1992). In particular, Weitzman             Aust, 1995).
and Graceffa(1984), Goodglick and Kane (1986), Ken-                  Severalprevious studies on the pathogenicrole ofcro-
nedy et al. (1989),Aust and Lund (1990, l99l), Lund               cidolite have focused primarily on Fe-releaserate, OH
and Aust (1990, 1992),Gulumian et al. (1993),Ghio et              radical generation, and DNA damage. As alluded to
 al. (1994),Lund et al. (1994),Chao and Aust (1993),and                                                      to
                                                                  above,our work was specificallydesigned begin to study
Hardy and Aust (1995) have suggested         that OH radicals     the surface chemistry of crocidolite, especially that per-
('OH) may be important in mineral-induced disease.                taining to both Fe2+and Fe3+,as it dissolvesunder con-
Although extremely short-lived, they have the ability to          ditions similar to those of a human lung. We used X-ray
fragment DNA. For example, Aust and coworkers stud-               photoelectron spectroscopy(XPS) and solution analysis
ied the release of Fe from crocidolite in vitro and re-           to track the changesin surface chemistry and the con-
ported a link betweenthis and DNA single-strandbreaks             tacting solutions over a 30 d period. Scanning electron
(SSBs)(Aust and Lund, 1990, 1991; Lund and Aust,                  microscopy (SEM) was also used to look for any disso-
 1992). Measurements DNA SSBsis one way of as-
                           of                                     lution or depositional featureson these fibers after treat-
                             WERNER ET AL.: SURFACE CHEMISTRY OF CROCIDOLITE                                       1095

ment. In addition, because the potential importance of
                             of                                control. The pH adjustment was performed only during
chelatorsin the diseaseprocessas describedabove, three         resuspensionto keep the solutions sterile more easily.
well-known Fe chelators were chosen for this study, in-        Buffers were not used becausethey could bind to fibers
cluding ethylenediaminetetraacetic    acid (EDTA), desfer-     or chelateFe themselves(Lund and Aust, 1990). Each
rioxamine, and citrate. Although citrate is the only one       sample was placed on a wrist-action shaker for the first
of the three found in the body, EDTA and desferriox-           hour of reaction at each stage,and each was kept in the
amine were also used becausethey are very effective Fe         dark to avoid photochemical reduction of Fe3+ by the
chelatorsand may have application in fiber-surfacemod-         chelators (Chao and Aust, 1993). After the reaction pe-
ification relevant to reducingtoxicity of fibers (e.9.,Brown   riod, the suspensionwas transferred to 50 mL conical
etal., 1990;Klockarsetal.,1990;    Gulumianetal., 1993).       bottom tubes and centrifuged to separatethe fibers from
Ultimately, the results of this study combined with pre-       the supernatant.Total Fe mobilized into the supernatant
vious work may help lead to a model for mineral-induced        was measured by the ferrozine and spectrophotometer
pathogenesis   that more specificallyincludes the involve-     method as described Lund and Aust (1990) exceptin
ment of reactions at the mineral-fluid interface.              the caseof desferrioxamine,in which the absorbanceof
                                                               the supernatantwas measureddirectly in the spectropho-
                                                               tometer at 428 nm. The fibers were washedfive times in
              Mlrrnurs      AND METHoDs
                                                               deionized water to remove any remaining chelator solu-
Samplesand reagents                                            tion, dried on an acid-washedwatch glassat room con-
   Crocidolite sampleswere obtained from Richard Grie-         ditions, and stored in glassvials with screw tops.
semer (National Institute of Environmental Health Sci-
ences,National Toxicology Program, ResearchTriangle            Scanning electron microscopy
Park, North Carolina). Campbell et al. (1980) report me-         The untreated fibers and those treated in the chelator-
dian fiber sizeof roughly 15.0 x 0.25 pm. The bulk com-        containing solutions for 30 d were examined with a Nor-
position was determined by wet chemical methods                an-Tracor Adem SEM equippedwith a LaB6electrongun.
(Campbell et al., 1980) and is as follows (convertedto         Sampleswere dispersion mounted on C rounds and coat-
atomic percentfor usewith XPS data; seeEq. 5): Si 18.60/o,     ed with approximately 100 A of gold.
Fe2+5.9o/o,Fe3*   5.4o/o,Mg2.lo/o, Na 3.3010.
   Two of the Fe chelators (sodium citrate and disodium        X-ray photoelectronspectroscopy
salt of EDTA) were obtained from Mallinckrodt, Inc.          Near-surface composition and Fe oxidation state on
(Paris, Kentucky). Deferoxamine mesylate USP (desfer-      the crocidolite sampleswere determined before and after
rioxamine-B) was obtained from CIBA (Summit, New           the various solution treatmentswith a Perkin Elmer 5400
Jersey).Stability constants for these chelators are avail- XPS (seeHochella, 1988, for a detailed description of
able in Martell and Smith (1974, 1977). In every case,     XPS). Becauseall experiments were performed in tripli-
the stability constant for chelatedFe3+is higher than that cate, three independent sets of XPS data were collected
for chelated Fe2+.However, it is not clear how this may    for each experimental condition reported. Therefore, the
affect crocidolite dissolution.                            XPS data reported in the Results section below are the
                                                           averagesof three measurements.
Sample treatment                                             Each sample of crocidolite, in a matlike form, was
   Crocidolite (l mg/ml) was reacted in a 50 mM NaCl mounted on a I in. diameter aluminum stub for exami-
solution in glass vials in air at a pH of 7.5 and room nation. A colloidal C suspension(in isopropyl alcohol)
temperature (-22 "C). Although physiological saline is was used as a mounting cement. Nonmonochromatic
approximately three times this concentration, we chose AlKa X-rays (1486.6eV) were used to analyzea 3 mm2
50 mM for direct comparison with previous studies (see area on the mat of fibers. The Si 2p, Na 2s, Mg 2s, Fe
Introduction). The pH was initially adjusted to 7.5 using 3p, and C ls photolines were used for all analyses.These
NaOH. The volume of the solution was approximately particular Si, Na, Mg, and Fe photolines were chosen
50 mL. The solutions were reactedfor time periods of l,    because their similar binding energies
                                                                    of                              (and thus similar
24, and 720 h (30 d) in I mM citrate, EDTA, or desfer- kinetic eneryies between 1370 and 1440 eV for AlKa
rioxamine-B. Millimolar chelator concentrations were X-rays). Using peaks with similar kinetic eneryiesis im-
used to approximate the concentrations of citrate and portant becausethe electrons that compose these peaks
other organic acids found under actual physiological con- originate from a similar depth in the sample. The C ls
ditions. All experimentswere performed in triplicate. One peak (kinetic energy= 1176eV) is the only coreJevelC
control group was analyzed, each time period without peak that is attainable for analysis by XPS. The sources
a chelator added.The 30 d experimentswere resuspended of C on the surfacewere the mounting cement and con-
in fresh chelator solutions on the first and seventh days, tamination from exposureto the solution and air.
and the pH was readjustedto 7.5. The resuspension     pro-   A curve-fitting program (Labcalc) was used to deter-
tocol was used to refresh the chelators,which are subject mine the absolute area of the Na 2s and Fe 3p peaks
to unwanted polymerization and potential OH radical de- (which have a slight overlap) and the Fe2+'Fe3+    ratio (Fig.
struction. This also aforded a reasonablemethod of pH      1). The Fe 3p line is best fit using three peaks, one rep-

                                                                  € t.n
.=                                                                     1.2
 o                                                                .9
 c                                                                E 1.0
 o                                                                o
                                                                  6 o.s
 o                                                                fl 0..
.=                                                                6    o.n
G                                                                      o.2
E                                                                      0.0
                                                                             o   20        40       60        80       100
                                                                     Fig.2. Theoretical change Si:Fe
                                                                                               in      ratio(asmeasured XPS)
                                                                  asa functionof Si leaching                    For
                                                                                              depthin angstroms. example,  at
                                                                  a hypothetical  leaching
                                                                                         depth  of0 A (noleaching),     :
                                                                                                                  Si:Fe 1.65,
                                                                  theratio ofthe untreated surface. curvefrom this fixedpoint
                 Binding Energy(eV)                               is derivedfrom Eq. 7 usingI : 26 A. As an example, Si:Fe
                                                                  atomicratioof 1.0,asmeasured XPS,couldresult leaching
                                                                                                  by                by
   Fig. 1. Example a curve-fitXPS spectrum Na 2s,Fe
                    of                           for              all Si in the nearsurface a depthof approximately A.
                                                                                           to                        13
3p, and Mg 2s photolines   usedin this study.The sum of all
curves superimposed theactual
                       on          spectrum.   Whendetermin-
ing the Fe2+:Fe3+ratio, the Fe 3p line was fit usingthe threeUsing the calculated o values and l from the XPS data,
peaks  shownfollowingMclntyre and Zetaruk(1977).     The rela-
                                                             the actual values for n were determined for each cation
tive positions
             ofthe threepeaks wereheldconstant, weretheir
                                                  as         in every sample.The r valuesfor Fe,o,,    Fe2+,Fe3+,Mg,
widths and the relativeintensities the two Fe3*lines.Only
                                 of                          and Na were then consideredproportionately to Si. This
therelativeintensitiesofthe oneFe2*line vs.the two Fe3+   lines
                                                             cation to Si ratio was needed to compare the XPS data
wereallowedto vary. The Fe2+'Fer+    ratio is equalto the peak
                                                             from sample to sample because sampling volume an-
                                                              alyzed, each experiment varies owing to exact sample
                                                             mounting and instrument conditions. Although l for each
resenting Fe2+and two representingFe3+,following the photoline varies proportionally to the amount of sample
often-cited example of Mclntyre andZntaruk(I977). The analyzed, the cation to Si ratios for a given sample are
smaller of the two Fe3+peaks is needed to account for independent of the analytic volume, allowing ratios to be
the numerous multiplet-split peaks that appear on the compared between samples.The changesin cation to Si
high binding-energy side of the Fe 3p peak (Gupta and ratios were then usedto follow the near surfacecomposition
Sen, 1974).The arearatio ofthe Fer+ fitted line relative of the mineral €used by reaction with the solutions.
to the two Fe3+fitted lines is directly proportional to the     To test quantitatively for the development of leached
Fe2+'Fe3+  ratio (Fig. l).                                   layers on the crocidolite surfacewithin the 30 d length of
  The Fe2*, Fe3+,Na, Mg, and Si photolineswere con- the treatments, we evaluated the Na:Si, Mg:Si, and Fe:Si
verted into semiquantitativecompositional data using the ratios as follows. The depth from which photoelectrons
following method. The number of atoms presentin a giv- are derived depends on the attenuation length, \, of the
en volume is related to the area under a photoline by the solid for an electron with the kinetic energy of interest
equation (Hochella, 1988)                                    (seeHochella, 1988). Attenuation length is formally de-
                                                         (5) fined as the distance at which the probability ofan elec-
                                                             tron escapingfrom a solid without the occurrenceofan
where 1 is the area under the peak, n is the atomic per- inelastic scatteringevent drops to e-'. Therefore, there is
centagein the measuredvolume of the sample, and o is an exponential reduction in signal as a function ofdepth,
the photoionization cross section for the sampled orbital which is representedby the equation
for the excitation energy used. The value of o for each
photoline was determined from the above equation using
                                                                            o/osignal : (l - e ') x 100             (6)
a value for r given by the bulk analysis reported above where c is the number of attenuation lengths.Therefore,
and a value for 1 from the XPS data of the untreated 630/o the signal comes from above ltr, 860/o
                                                                  of                                       from above
sample. Becausethe n values for Fe2+and Fe3+were de- 2tr,and 950/o        from above 3),(a: 1,2, and3, respectively).
termined from the samephotoline, their respectiveo val- By convention, the 3\ level in the sample is called the
ueswere setequal on the basisofthe crosssectionderived analysis depth. For example, in amorphous SiOr, tr for
using the total Fe 3p peak area and the Fe,o,content. the Si 2p photoelectronsejectedby AlKa X-rays has been
                             WERNER ET AL.: SURFACE CHEMISTRY OF CROCIDOLITE                                            t097


                                                                                  *     Desferrioxamine
                                                                o                 +     EDTA
                                                               =                  #     Citrate
                                                                o                 *     Gontrol
                                                               .=      200
                                                               E 1oo

                                                                                      200         4{t0       600         800
                                                                    Fig. 4. Solution data showing nanomolesof Fe releasedinto
  Fig. 3. SEM photomicrograph crocidolitereacted the the supernatant per milligram of crocidolite. The durations of
                            of                  in
presence citrate 30 d.
         of     for                                  the experiments were l, 24, and 720 h. Each point represents
                                                               the averageof three experiments.At I and 24 h, the sizesof the
                                                               actual data points on the plot are larger than two standard de-
measuredto be 26 A (Hochellaand Carim, 1988),mak-              viations about the mean. At 720 h, the sizesof the data points
ing the 3), analysisdepth 78 A. Because the photolines
                                         all                   are approximately one standard deviation about the mean.
for the electronsbeing analyzedwere chosento have ki-
netic eneryies similar to Si 2p, the tr for these cations
should be within a few angstroms of that for Si. There-        depth patterns are consistent for all crocidolite samples
fore, in this study one can calculatethe depth ofleaching      measured in this study, we can make best use of the cal-
of one elementwith respectto another in the near surface.      culated leach depths in a relative sense.
For example, if the depth of leaching for a given cation
with respect to Fe is equal to ltr, one would expect the
signal for that cation to be reducedby 630/o relative to Fe                               Rrstnrs
(Fig. 2). Equation 6 is generalizedto yield information        SEM inaging
about the depth of leaching in the following way:                Figure 3 is an SEM image of fibers that reactedin the
          o/osignalreduction: (l - .-vr; x 100           (7)   presenceof citrate for 30 d. This and other SEM images
                                                               at higher magnification show no evidence of dissolution
where x is equal to the depth of cation leaching with          features or of a surface precipitate. Similar results were
respectto some other element.Thus, knowing the amount          obtained for all fibers examined after reaction.
of cation signal reduction with respectto another element
allows the depth of leaching relevant to that element pair
to be determined as demonstrated graphically in Figure         Solution analysis
2. We used this schemeto look for the possible leaching           The amount of Fe found in the posttreatmentsolutions
of Na, Mg, and Fe relative to Si, and Si relative to Fe.       is shown in Figure 4 as nanomoles of Fe in solution per
   The approach presentedabove assumesthat the atten-          milligrarn of treated sample. Even after 30 d, the Fe,.,
uation length is similar in the fresh material and the         releasedto solution in the control samplesaveragedonly
leachedoverlayer. This is probably a very good assump-          I nmol/mg. This amount of Fe, equivalent to I nmoV
tion becauseattenuation lengths are similar for oxides         mL of solution in these experiments, is right at the de-
and silicates (Hochella, 1988). However, calculating ap-       tection limit of the ferrozine and spectrophotometer
parent leaching depths in this way (or any other way)          method used in this study. Therefore, Fe in solution for
should still be considered only a qualitative measure-         theseexperiments may actually be less.The introduction
ment. As demonstratedby Hochella (1990), leaching              of chelators greatly increasedthe releaseof Fe from the
depths are probably not uniform over a surface.Even if         crocidolite fibers. After 30 d, the amount of Fe released
leachingdepthswere uniform, the analyzedsurfacewould           was roughly 200 nmol/mg from the EDTA- and desfer-
have to be flat for the approach presentedabove to be          rioxamine-treated fibers and approximately 100 nmol/
fully valid. Assuming that surfaceroughness   and leaching     mg from the citrate-treated fibers.
r098                            WERNER ET AL.: SURFACE CHEMISTRY OF CROCIDOLITE

       0.68                                                             0.s5
              --r-   Citrate                                                       --r-citrate Fe3*:si
                                                                        0.50       -€-Control                        T
       0.66                                                                                                          J-
              +-     Control
       0.68                                                             0.1
              --r- Desferiloxamine                                      0.55
                                          /_/-           I         o               -r- Desferrioxamine Fe3*:Si
o      0.66   +      contot        /                                    0.50       --E-Control                       I
6                                                        I         o
tr     0.64                                                        c     J
o                                                                  o    0.40
(,     0.62                                                       o     0.3s
o                                                                 a     0.30
it                                                                +.'              +- DesferrioxamineFe2+:si
                                                                  x     0.25       --+Control
                                                                   o                                 ----------a
                                                                  lt.   0.20
       0.58                                                                    =--

       0.66                                                                        --r-EDTA Fe3*:sa

       0.62                                                             0.35
       0.60                                                             0.30
                                                                                     r EDTAFe2*:Si
                                                                                   --E- Control
                               300 400           600 700 800
                                                                               0     100 200 300 400 500 600 700 800
                               Time (h)
  Fig. 5. Fe:Sicationratiosas determined XPSfor control
andchelator              Eachpoint is theaverage
              experiments.                         ofXPS data                  cationratiosvs.time showing Fe2+ Fe3+
                                                                    Fig.6. Fe:Si                         the     and
obtained  from threeidenticalexperiments. pointsareplot-
                                          Data                    components. the captionto Fig. 5 for moredetails.
ted at I, 24, and720h on the horizontal axis.The arrowon the
left of eachplot represents startingelemental
                           the                     ratio of the
untreated  sample. The error bar on the right sideof eachplot  for 30 d, but thesechangeswere not more than one stan-
represents standard
            one        deviationasdetermined    from triplicatedard deviation. The Fe:Si ratio for the control was sig-
experiments.                                                   nificantly higher than those ratios in the chelator groups
                                                               after 30 d.
                                                                  Data were also collected to determine how the surface
XPS analysis                                                   Fe2+:Si and Fe3+:Si  ratios changed   with time (Fig. 6). The
   Fe vs. Si surface analysis. The XPS data in Figure 5 Fe3*:Si ratio in the control increasedinitially in the first
show how the Fe:Si ratios for both the control and treated 24 h and then,remained constant at about one standard
fibers changed over the 30 d period. All the chelator- deviation greaterthan the untreated sample. The citrate-
exposedsamplesand the control group exhibited an ini- and desferrioxamine-treatedsamples showed a decrease
tial rise of the Fe:Si ratio in the first hour and then a in the ratio of about one standard deviation after 30 d,
decreaseof this ratio after 24 h. In the control experi- and the EDTA-treated sample changed less than one
ment, this initial action was followed by an increasein standard deviation. The Fe3+:Siratio in the control was
the Fe:Si ratio after 30 d. The differencein the Fe:Siratios greater than that for the chelator groups by 1.5-2.5 stan-
between the untreated sample and the averageof the 30 dard deviations after 30 d.
d control samples was sigrrificant (about three standard         The Fe'z+:Si ratio for the control rose slightly more than
deviations). The desferrioxamine-, EDTA-, and citrate- one standarddeviation after 30 d. The Fe2+:Siratios also
treated samplesshowedslight variations in the Fe:Si ratio rose in the citrate- and desferrioxamine-treatedsamples
relative to the starting value after reacting in the solutions by more than two standard deviations, whereas the
                               WERNER ET AL.: SURFACE CHEMISTRY OF CROCIDOLITE                                      1099

     o.12                                                             0.19
                    +     Citrate                                                    +    Citrate
     0.11                                                             0.18
                    +-    control                                                    +    Control
     0.08                                                             0.14

     0.07                                                             0.13
     o.12                                                             0.19
                   -f-   Desferrioxamine                                         -l-     Deslerrioramine
.9          G                                                   .9 0.
o    0.tl          +     Control                                6                +       Gontrol
g                                                                     0.17
o    0.10                                                       o
6                                                               G     0.16
(,                                                               (,
a                                                               (t    0.15
o)                                                              id
= 0.08                                                          z     0.14

     0.07                                                             0.13
     o.12                                                             0.19

            +-      +     EDrA                                        0.18
     0.11                                            l
                    +     Control                    I
     0.09                                                             0.1

     0.07                                                             0.13
                 100 200 300 400 500 600 700 800                          01               300 400 500 600 700 800

                            Time(h)                                                         Time (h)
                                                                  Fig. 8. Na:Sicationratiosvs. time. See captionto Fig. 5
  Fig. 7. Mg:Sicationratiosvs. time. See captionto Fig.
                                        the                     for moredetails.
5 for moredetails.

                                                                whereas the control (ust over one standard deviation),
EDTA-treated samplesexperiencedessentiallyno appar-             citrate, and EDTA (less than one standard deviation)
ent changein the ratio after 30 d. All three chelator groups    groups did not. In the case of all three chelator groups,
had a statistically similar Fe2*:Si ratio after 30 d relative   the Na:Si ratio did not vary by more than one standard
to the control (within one standard deviation for desfer-       deviation from the control after 30 d.
rioxamine and citrate, less than two standard deviations           C surface analysis. The C ls spectra for all samples
for EDTA).                                                      showeda relatively narrow, fully resolvedpeak to the low
   Mg and Na vs. Si surface analysis. The Mg:Si ratios          binding-energy side of an often complex, much broader
are shown in Figure 7. After 30 d, the control, desfer-         band. The low binding-energy peak is clearly due to the
rioxamine, and EDTA groups had a large decrease         (four   colloidal C suspensionused to cement the fibers to the
standard deviations) in the Mg:Si ratio in comparison           sample stub (Hochella, 1988). The other portion of the
with the initial ratio. The citrate group also showed a         C ls spectra originated from the fibers. This band gen-
reduction in the Mg:Si ratio but less than half as much         erally increasedin width and complexity with the length
as the control, desferrioxamine,and EDTA groups. The            of time of the treatment, irrespective of whether chelator
citrate group was the only chelator group with a signifi-       was present in the experiment or not. This suggests   that
cantly different Mg:Si ratio relative to the control after      the C ls signal from the fibers was not due to residual
30 d.                                                           chelators on the surfaceafter rinsing but to the progres-
   Figure 8 shows the changesin the Na:Si ratios. Only          sive accumulation of adventitious C of various forms in
the desferrioxamine group showed a significant decrease         aqueoussolution, which has been seenbefore (Stipp and
in the Na:Si ratios after 30 d (two standard deviations),       Hochella,l99l).

                      DrscussroN                              croscopy (HRTEM). Groups that have used HRTEM to
                                                              study dissolution reactions and weathering processes     in-
Control experiments                                           ternal to mineral grains include Ahn and Peacor(1987),
   In the 30 d control experiments,the XPS data show a Banfield and Eggleton(1988, 1990),and Banfield et al.
slight but significant increasein the Fe:Si ratio (equiva- (1991). Specifically crocidolite,each fiber consistsof
lent to a Si:Fe ratio decreasefor direct comparison with exceptionally narrow (lessthan a few tenths of a micro-
Fig. 2). This suggeststhe formation of a surface layer meter in diameter) interlocking fibrils, as has been pre-
enriched in Fe with respectto Si. Using the leachedlayer viously shown in TEM studies (e.g., Ahn and Buseck,
thickness model and assumptions described in the Ma- l99l). Interfibril regionsprobably contain sheetsilicates.
terials and Methods section, the thickness of the Si-de- The sheet silicateswould be only minor to trace mineral
pleted layer is calculated at about 5 A, assuming the components in a bulk sense,            but they could be very influ-
leachedlayer is completely depleted in Si but unchanged ential in reactions with solutions (Veblen and Wvlie.
with respectto Fe (i.e., using Eq. 7; seeFig. 2). Although r993).
the data are consistent with such a depleted layer, other
scenariosare possible.Si could be partially removed from Chelator experiments
the near surface,implying that the leachedlayer extends         As noted in the Resultssection,the Fe2+:Si      and Fe3+:
deeperinto the mineral. Also, it is likely that the leaching Si ratios for the 30 d EDTA. desferrioxamine.and citrate
depth is quite variable (Hochella, 1990). Nevertheless, groups did not vary considerably from that of the un-
the leached layer thickness model gives a relative sense treatedsample,but trends in the Fe2+:Si           and Fe3*:Sira-
of chemical variations at a surface and provides a conve- tios of at least the desferrioxamine and citrate g.roups
nient benchmark with which to compnre other surfaces.         appearedas though they would convergewith time. This
   The Si-deficient layer suggested  here could also reflect probably marks the beginning of the removal of the ox-
the precipitation of an iron oxide on the surface.Work idized layer at the surfacethat was describedin the pre-
by White and Yee (1985) on Fe oxidation and reduction vious section.As the outer oxidized layer is removed, the
in iron silicatesshowsthat at near neutral pH, the rate of surfacecomposition begins to approach that of the bulk,
Fe3+releaseis greatly decreased    becauseofthe initiation where Fe2*:Fe'* - l, causingthe Fe'z+:Si           and Fe3*:Si
of ferric hydroxide or oxyhydroxide precipitation. How- ratios to converge.
ever, we observed no increasein Fe3+relative to either           After 30 d in the presenceof chelators,the total Fe:Si
Si or Fe2+according to the XPS data after 30 d (Fig. 6). ratios did not significantly change.From this, one might
Therefore, we conclude that precipitation of Fe3+-con- infer that the crocidolite fibers dissolved congruently,i.e.,
taining phaseswas not responsiblefor the increasein the Si was being releasedat a rate in stoichiometric propor-
Fe:Si ratio in the control experiments.Also, although the tion to Fe. Assuming that the complete dissolution of a
bulk chemistry data show that the Fe2+'Fe3+     ratio = I in fiber is a function of the releaseof Fe, Si, or both (i.e.,
the bulk mineral, the XPS data show an Fe2+:Fe3+        ratio this is the rate-limiting step), one can calculatean appar-
of approximately 0.4 at the surface.This is probably the ent fiber lifetime (seeestimatesbelow and accompanying
result of oxidation of the Fe at the surface long before assumptions).But it has been shown or inferred in other
any sample treatment. This surfaceabundanceof nearly studies(e.g.,Berneret al., 1985;Mogk and Locke, 1988;
insoluble Fe3+under theseconditions (Baesand Mesmer, Hochellaet al., 1988;Hellmann et al., 1990)that release-
 1976) probably hinders Fe releaseat neutral pH but al- to-solution data and XPS observations generally do not
lows Si to be preferentially removed. Taken together,the have a simple correlation becausethe solution and XPS
above information suggests   that the dissolution of crocid- data measuretwo diferent things. The release-to-solution
olite is dependenton the releaseofFe, particularly Fe3+, data give the total number of cations that have been lib-
from the surface.                                             erated from the solid and have not participated in any
   The XPS data for the control experimentsalso show a kind of reprecipitation reaction. These data provide no
decreasein the Mg:Si and Na:Si ratios at 30 d that is information about where the cations originated. On the
equivalent to a 5 A Na- and a l0 A Mg-depleted layer other hand, XPS data give the composition of the top
with respect to Si. If Si were releasedfrom the surface, several nanometersover a large area of the sample. XPS
Na and Mg must have been leachedas well but to a great- cannot detect what is happeningon internal surfaces,          and
er depth.                                                     XPS data do not provide the distribution of cations with-
   Because the nature of a crocidolite fiber bundle, sur- in the depth of analysis (excluding angle-resolvedXPS,
face chemistry internal to the bundles (i.e., on the surfaces which is not applicable on rough surfacesor fibers). Fi-
of fibrils accessibleto solutions along grain boundaries) nally, the data cannot be used to determine leaching depth
is probably an important component of the cell-mineral- heterogeneity.        Taken together,theselimitations mean that
fluid interaction. These surfaceswould not be observable release-to-solutiondata cannot be used to predict XPS
by classicsurface-sensitive  spectroscopies would cer- results and vice versa. In fact, the same conclusion can
tainly contribute to changesin the solution chemistry, be drawn from the data of a previous crocidolite disso-
and reactions at these surfacescould possibly be studied lution study (Gronow, 1987).Resultsfrom this study show
directly with high-resolution transmission electron mi- crocidolite dissolution to be incongruent accordingto so-
                            WERNER ET AL.: SURFACE CHEMISTRY OF CROCIDOLITE                                           1l0l

lution data after 1024 h betweenpHs of 4 and 9, yet XPS        by the crocidolite dissolution study of Gronow (1987);
data show no changein the averagesurfacechemistry at           she showed that dissolution reachesan apparent steady
least over the first 50 h ofreaction at pH 4. Further ev-      state after this period of time. However, Chao and Aust
idence of incongruent dissolution comes from Crawford          (1994) showed that the Fe-release     rate continues to slow
(1980),an HRTEM study of crocidolitefibersbeforeand            after 30 d in desferrioxamine solutions for samplesthat
after reaction with human blood serum in vitro or rodent       were identical to those used in this study, but that the
lung tissuein vivo. About one-half of the fibers examined      releaserate between 30 and 90 d is relatively stable.Dur-
from both environments showed patchy (i.e., covering           ing this 60 d period, approximately 140 nmol of Fe was
only portions of any one fiber), amorphous, presumably         released per milligram of fiber, equivalent to a loss of
leachedlayerson the surfaceup to 100 A thick. The other        2.7o/o the total Fe present.At that rate, the fiber would
one-halfof the fibersobservedshowedno dissolutionrinds         be depleted in Fe (probably fully dissolved) in approxi-
(i.e., the amphibole structure was seenextending all the       mately 6 yr. Choa and Aust (1994) did not use citrate
way to the surface).This clearly demonstratesonce again        under the conditions of our experiments, but if the Fe
the heterogeneous    nature of surfacereactions (Hochella,     releasecausedby citrate slowed proportionally with what
 1990, 1993). Although the dissolution rinds were not          was observed for desferrioxamine in this study, the Fe-
 chemically analyzed in the Crawford study, their com-         releaselifetime in a citrate solution under the conditions
 position probably varies from that ofcrocidolite (seeCas-     of our experiments would be approximately l3 yr.
 ey and Bunker, 1990, for an example of similar leached            Our estimates of Fe-releaselifetimes of crocidolite fi-
 layers on feldspar grains that were chemically analyzed).     bers should be used in biodurability assessments         with
    On the basis of these studies, we cannot say with full     caution. In the lung, fibers may be engulfedin scavenging
 confidence from our experiments that Si is releasedin         cells (phagocytes)and subjected to diferent chemistries
 stoichiometricproportion to Fe even thouglt the XPS data       (e.g.,pH = 4-4.5) comparedwith thoseof the fluids we
 show that the Fe:Si ratios do not change from the un-         used. This would almost certainly result in shorter life-
 treated to the 30 d treated samples.However, we can say        times, as would increasingthe ambient temperature (our
                                                                                                       oClower than typical
 that the averagecomposition of the top several tens of         experiment temperatureswere - l5
 angstromsof thesefibers maintains similar Fe:Si and Na:        body temperatures). Further,       Lund and Aust (1990)
 Si ratios over 30 d, that the surfacebecomesdepleted in        showed that the presenceofascorbate, which is found in
 Mg, and that, at least for the desferrioxamineand citrate      physiological fluids, can increasethe releaserate ofFe2+
 experiments, the average surface Fe oxidation state is         from crocidolite under conditions similar to those of the
  reduced.                                                      control and chelator experiments of this study. On the
                                                                other hand, fibers in lungs also can be partially covered
Fe:Si fluctuations in the first 24 h                            with crystalline Fe-rich coatings believed to be derived
   The Fe:Si ratios determined from the XPS data for all        from proteins such as hemosiderin and ferritin (Pooley,
the data groups showed an increasein the first hour fol-         1972 Churg and Warnock, 1977). This could signifi-
lowed by a decrease the following hours up to the 24
                       in                                       cantly slow the overall rate of Fe releasefrom the fiber'
h point. The reason for this consistent fluctuation is not      Finally, many other agents in lung fluids that have not
known, but it may relate to highly reactive sites on the        been tested may have significant effectson fiber surface
untreated fiber surface. The chelators probably do not          chemistry and dissolution. Nevertheless,a recent study
play a role in this fluctuation becausethe same behavior        by Chao et al. (1994) has demonstrated that crocidolite
is seenin the control group. We assumebelow that this            fibers phagocytized by cultured human lung carcinoma
fluctuation is not important in determining the long-term       cells show a similar Fe-release   rate over the first 24 h as
effectsof chelators on crocidolite dissolution.                  crocidolite incubated  in citrate-bearing solutions similar
                                                                 to the ones used in this study. Unfortunately, longer du-
Fe-releaselifetimes and implications for mineral-induced         ration studies of this type were not possible becauseof
pathogenesis                                                     the simultaneousoccurrenceof cell death and replication'
   Our goal in this part of the study was to estimate how        making unique interpretation of the data impossible.
long a crocidolite fiber can releaseFe to surrounding so-           The Fe-releaselifetimes of crocidolite fibers discussed
lutions under the conditions used in this study. As dis-         above correspondwell with what is already known about
cussedbelow, this has implications in understandingthe           the biodurability of asbestosminerals from lung-burden
biodurability of crocidolite fibers as well as their ability     studies.Wagner et al. (1974) noted that rats continuously
to induce lung disease.                                          exposed to amphibole fibers sufered an ever increasing
   The amount of Fe releasedinto solution in the first day       lung burden, whereasrats continuously exposedto chrys-
is not a good estimate of long-term Fe releasebecauseof          otile fibers showed only a moderate lung burden' which
the initial chemical fluctuations discussedabove. To ob-         leveled off with time. Joneset al. (1989) more recently
tain a more reasonableFe-release    rate, the concentration      came to a similar conclusion. This fiber accumulation
changebetween the day I and day 30 data points for the           pattern in lung tissuehas also been shown to be the same
 control and each chelator group could be used. The use          in humans (Churg, 1993, and referencestherein). These
 of I d as an appropriate starting point might be justified      observations are consistent with Hume and Rimstidt
 tto2                              WERNER ET AL.: SURFACE CHEMISTRY OF CROCIDOLITE

 (1992), who studied the dissolution behavior of chrysotile               Ahn, J.H., and Peacor,D.R. (1987) Kaolinitization of biotite: TEM data
 fibers at a temperature and pH range consistentwith hu-                       and implications for an alteration mechanism.American Mineralogist,
 man lung conditions. For theseconditions, they estimat-                       72,353-356
                                                                          Aust, A.E., and Lund, L.G. (1990) The role ofiron in asbestos-catalyzed
 ed that a I pm diameter fiber of chrysotile would com-
                                                                              damageto lipids and DNA. Biological Oxidation Systems,      2,597-605.
 pletely dissolve in < I yr. Considering these studies, it is             -         (199l) Iron mobilization from crocidolite resultsin enhanced  iron-
 no surprise that the crocidolite burden recovered from                        calalyzed oxygen consumption and hydroxyl radical generation in the
 autopsiedhuman lungsis alwaysconsiderablygreaterthan                         presence cysteine.In R.C. Brown, J.A. Hoskins, and N.F. Johnson,
                                                                              Eds., Mechanismsin fibre carcinogenesis,   p.397-405. NATO ASI Se-
 the proportion ofcrocidolite fiber dust in the air source
                                                                              ries. Plenum. New York.
 responsible for the lung contamination (e.g., Wagner et                  Baes,C.F., Jr., and Mesmer, R.E. (1976) The hydrolysis of cations, 489
 al.,1982;.  Gardneret al., 1986;Churg, 1988,1993).per-                       p. Wiley, New York.
 haps either chrysotile fibers are not making it into the                 Banfield, J.F., and Fggleton, R.A. (1988) Transmission electron micro-
 lung through the upper bronchial tubes relative to the                       scopestudy of biotite weathering.Clays and Clay Minerals, 36, 47-60.
                                                                         -(1990)           Analytical transmissionelectron microscopestudiesof pla-
 crocidolite fibers for mechanicalor aerodynamic reasons,                     gioclase,muscowite,and K-feldspar weathering.Clays and Clay Min-
 or crocidolite fibers are much more biodurable. However,                     erals.38. 77-89.
 Churg et al. (198a) showedunequivocallythat chrysotile                  Banfield,J.F.,Jones,      B.F.,and Veblen,D.R. (1991)An AEM-TEM srudy
 fibers can be found in both the peripheral and central                       of weathering and diagenesis, Abert Lake, Oregon: I. Weathering re-
 portions of the human lung, indicating that the probable                     actions in the volcanics.Geochimica et CosmochirnicaActa. 55. 27I l-
 causeofthese observationsis greaterbiodurability ofcro-                 Berner, R.A., Holdren, G.R., Jr., and Schott, J. (1985) Surfacelayers on
 cidolite fibers.                                                             dissolving silicates. Geochimica et Cosmochimica Acta, 43, Il7rl     186.
                                                                         Brown, R.C., Carthew, P., Hoskins, J.A., Sara, E., and Simpson, C.F.
                Sulrtvr,lny AND coNclusroNs                                   (1990) Surfacemodification can affect the carcinogenicityofasbestos.
                                                                              Carcinogenesis, 1883-1885.
    We observed with X-ray photoelectron spectroscopy,                   Campbell, W.J., Huggins, C.W., and Wylie, A.G. (1980) Bureau of mines
 for the first time, the changein crocidolite surfacechem-                   report ofinvestigations no. 8452.
 istry in the presenceof saline solutions containing che-                Casey,W.H., and Bunker, B. (1990) I*aching of mineral and glasssur-
 lators that partially mimic physiological fluids. Over 30                   facesduring dissolution. In Mineralogical Societyof America Reviews
                                                                             in Mineralogy, 23, 397-426.
 d, crocidolite fiber surfacesbecome depleted in Mg rela-
                                                                         Chao, C., and Aust, A.E. (1993) Photochemicalreduction offerric iron
 tive to Fe and Si. In addition, Fe is removed at a highly                   by chelaton results in DNA strand breaks. Archives of Biochemistry
acceleratedrate relative to Fe removal in the absenceof                      and Biophysics,    300, 544-550.
the chelators, yet the ratio of Fe to Si on the surface of              -          (1994) Effect of long-term removal of iron from asbestos des-
these fibers remains approximately the same. Although                        ferrioxamine B on subsequentmobilization by other chelatorsand in-
                                                                             duction of DNA single-strand breaks. Archives of Biochemistry and
we cannot pronouncethe dissolution mechanismsof these
                                                                             Biophysics,   308, 64-69.
fibers from thesedata without more solution composition                 Chao, C., Lund, L.G., Zinn, K.R., and Aust, A.E. (1994) Iron mobiliza-
information, the data do suggestthat the fibers dissolve                     tion from crocidolite asbestos human lung carcinomacells.Archives
relatively quickly under the influence of chelators, con-                    of Biochemistry and Biophysics, 314, 384-391.
                                                                        Churg, A. (1988) Chrysotile, tremolite, and mesotheliomain man. Chest,
 sidering the insolubility of Fe in pH-neutral aqueousso-
lutions. After rinsing the fibers briefly in deionized water            -          (1993) Asbestoslung burden and disease   patternsin man. In Min-
after 30 d in the presenceof chelators, no evidence was                      eralogicalSocietyof America Reviews in Mineralogy, 28,410-426.
found of chelators remaining on the fiber surfaces,sug-                 Churg, A., and Warnock, M.L. (1977) Andysis of the coresof fermginous
gestingthat they are only weakly bound to the dissolving                     (asbestos)  bodies from the general population. Iaboratory Investiga-
                                                                            tions. 37. 280-286.
surface, although their influence in fiber dissolution is               Churg, A., DePaoli, L., Kempe, 8., and Stevens,B. (1984) Lung asbestos
dramatic. Despite this, estimates of Fe-releaselifetimes                    content in chrysotile worken with mesothelioma.American Reviews
under the conditions of theseexperimentsis on the order                     on Respiratory Disease,130, 1042-1045.
of l0 yr. It is possible that crocidolite dissolution may               Craighead,J.E., Abraham, J.L., Churg, A., Green, F.H.Y., Kleinerman,
provide a considerableand long-lasting source ofFe po-                      J., Pratt, P.C., Seemayer,T.A., Vallaythan, V., and Weill, H. (1982)
                                                                            The pathology of asbestos-associated    diseases the lungs and pleural
tentially availableto promote DNA damagein lung tissue.                     cayities: Diagnostic criteria and proposed grading scema. Archives of
                                                                            Pathology and Laboratory Medicine, 106, 544-596.
                    Acrxowr,tocMENTs                                    Crawford, D. (1980) Electron microscopy applied to studies ofthe bio-
                                                                            logical significanceof defects in crocidolite asbestos.   Joumal of Mi-
   We thank Jodi Junta Rosso, Cam Weaver, and Udo Becker for their          croscopy, 120, l8l-192.
many suggestions   and discussions,Kevin Russo for compurcr suppon,     Eborn, S.K., and Aust, A.E. (1995) Effect ofiron aquisition on induction
and Robert Raymond for help with the SEM. We also apprecrate con-           ofDNA single-strandbreaks by erionite, a carcinogenicmineral fiber.
structive reviews frorn Robert Nolan and an anonymous referee, which        Archives of Biochemistry and Biophysics, 316, 507-514.
aided in improving the manuscript. This research was supported with     Gardner, M.J., Winter, P.E, Pannett,8., and Powell, C.A. (1986) Follow
funding from the following sources: DOE Laboratory Direcred Research        up study of workers manufacturing chrysotile asbestos cement prod-
and Development grant (to G.D.G.), NIEHS (gant ES05782to A.E.A.),           ucts. British Journal oflndustrial Medicine, 43, 726.
and NSF Grant EAR-9305031 to M.F.H.).                                   Ghio, A.J., Kennedy, T.P., Stonehuerner,     J.G., Crumbliss,A.L., and Hoi-
                                                                            dal, J.R. (1994) DNA strand breaks following in vitro exposuresto
                     RrrnnrNcps crrED                                       asbestosincreases\vith surface-complexed[Fe3+]. Archives of Bio-
                                                                            chemistryand Biophysics, l, l3-18.
Ahn, J.H., and Buseck,P.R. (1991) Microstructures and fiber-formation   Goodglick, L.A., and Kane, A.B. (1986) Role of reactive oxygen metab-
  mechanismsof crocidolite asbestos.
                                   American Mineralogist, 76, 1467-         olites in crocidolite asbestostoxicity to mouse macrophages.Cancer
  1478.                                                                     Research, 5558-5566.
                                        WERNER ET AL.: SURFACE CHEMISTRY OF CROCIDOLITE                                                                     r 103

Gronow, J.G. (1987) The dissolution of asbestosfibers in water. Clay                     gle strand breaks in 6Xl7 4 RFI DNA. Occupational and Environmen-
   Minerals.22.21-35.                                                                    tal Medicine, 51, 200-204.
Gulumian, M., van Wyk, J.A., Hearne, G.R., Kolk, B., and Pollak, H.                   Martell, A.E., and Smith, R.M. (1974) Critical stability constants,vol. l,
   (1993) ESR and Mdssbauer studies on detoxified crocidolite: Mecha-                    470 p. Plenum, New York.
   nism of rcduced toxicity. Joumal of lnorganic Biochemistry, 50, 133-143.           -       (1977) Critical stability constants,vol. 3, 496 p. Plenum, New York.
Gupta, R.P., and Sen, S.K. (1974) Calculation of multiplet structure of               McDonald, J.C. (1990) Cancerrisks due to asbestos       and man-madefibres.
   corep-vacancy levels. Physical Review B, 10,71-77.                                    Recent Resultsin Cancer Research,120,122-133.
Guthrie, G.D., Jr., and Mossman, B.T. (1993) Merging the geologicaland                McDonald. J.C., and McDonald, A.D. (1986a) Epidemiologv of malig-
   biological sciences: An intergrated approach to the study of mineral-                 nant mesothelioma.In K. Antman and J. Aisner, Eds.,Asbestos        related
   induced pulmonary diseases. Mineralogical Societyof America Re-
                                         In                                              malignancy,p. 57-79. Grune and Stratton, New York.
   views in Mineralogy, 28, l-5.                                                      -(1986b)         Epidemiology of asb€stos-related     lung cancer.In K. Ant-
Hardy, J.A., and Aust, A.E. (1995) Iron in asbestoschemistry and car-                    man and J. Aisner, Eds.,Asbestosrelated malignancy,p. 3 l-56. Grune
   cinogenicity. Chemical Reviews, 95, 97-l 18.                                          and Stratton, New York.
Hart, R.W., Kendig, O., Blakeslee,J., and Mizuhira, V. (1980) Effect of               Mclntyre, N.S., and Zfjtaruk, D.G. (1977) X-ray photoelectron spectro-
   cellular ingestion on the elemental ratio of asbestos. R.C. Brown,
                                                                  In                     scopic studiesofiron oxides. Analytical Chemistry, 49' 1521-1529.
   M. Chamberlain,R. Davies, and I.P. Gormley, Eds.,The in vitro effects              Mogk, D W., and Locke, W.W., ilI (1988) Application of augerelectron
   ofmineral dusts,p. l9l-199. Academic,London.                                          spectroscopy (AES) to naturally weathered homblende. Geochimica et
Hellmann, R., Eggleston,C.M., Hochella, M.F., Jr., and Crerar, D.A.                      Cosmochimica Acta, 52, 2537-2542.
   (1990) The formation ofleached layers on albite surfacesduring dis-                Morgan, A., Holmes, A., and Gold, C. (1971) Studiesof the solubility of
   solution under hydrothermal conditions. Geochimicaet Cosmochimica                      constituentsofchrysotile asbestos vlvo using radioactive tracer tech:
   Acta. 54. 1267-128L                                                                    niques. Environmental Research,4, 558-570.
Hering, J.G., and Stumm, W. (1990) Oxidative and reductive dissolution                Nolan, R.P., and Langer, A.M. (1993) Limitations of the Stanton hy-
   of minerals. In Mineralogical Society of America Reviews in Miner-                     pothesis.In Mineralogical Societyof America Reviews in Mineralogy'
    alogy,23, 427-465.                                                                  28,308-326
Hochella, M.F., Jr. (1988) Auger electron and X-ray photoelectronspec-                Pooley, F.D. (1972) Asbestosbodies, their formation, composition and
    troscopies.In Mineralogical Societyof America Reviews in Mineralo-                  character.Environmental Research,5, 363-37 9-
    g y , 1 8 ,5 7 3 - 6 3 8 .                                                        Schindler, P.W. (1990) Co-adsorption ofmetal ions and organic ligands:
-(1990)               Atomic structure, microtopography, composition, and               Formation of ternary surface complexes. In Mineralogical Society of
    reactivity of mineral surfaces. Mineralogical Societyof America Re-
                                          In                                            America Reviews in Mineralogy, 23, 281-307.
    views in Mineralogy,23,87-132.                                                    Stanton, M.F., Layard, M., Tegeris,A., Miller, E., May, M., Morgan, E.,
-(1993)              Surfacechemistry, structure, and reactivity of hazardous           and Smith, A. (1981) Relation of particle dimension to carcinogenicity
    mineral dust. In Mineralogical Societyof America Reviews in Miner-                  of amphibole asbestoses     and other fibrous minerals. Journal of the
    alogy,28, 275-308.                                                                  National Cancer Institute, 67, 965-97 5.
Hochella, M.F., Jr., and Carim, A.H. (1988) A reassessment electron    of             Stipp, S.L., and Hochella, M.F., Jr. (1991) Structure and bonding envi-
    escape     depths in silicon and thermally grown silicon dioxide thin films.         ronment at the calcite surface as observed with X-ray photoelectron
    SurfaceScience.197, L260-L268.                                                       spectroscopy (XPS) and low energy electron diffraction (LEED). Geo-
Hochella,M.F., Jr., Ponader,H.B., Turner, A.M., and Harris, D.W. (1988)                  chimica et CosmochimicaActa, 55, 1723-17 36-
    The complexity of mineral dissolution as viewed by high resolution                Stumm, W (1992) Chemistry of the solid-water interface, 428 p' Wiley,
    scanning Auger microscopy: Labradorite under hydrothermal condi-                     New York.
    tions. Geochimica et Cosmochimica Acta. 52,385-394.                               Veblen, D.R., and Wylie, A.G. (1993) Mineralogy of amphibolesand l:l
Hume, L.A., and Rimstidt, J.D. (1992) The biodurability of chrysotile                    layer silicates. In Mineralogical Society of America Reviews in Min-
     asbestos.    American Mineralogist, 77, 1125-1128.                                 eralogy,28, 6l-137
                                                                                      Vu, V.T. (1993) Regulatory approachesto reduce human health risks
Jaurand, M.C., Bignon, J., Sebastien, and Goni, J. (1977) Leachingof
                                                                                        associatedwith exposuresto mineral fibers. In Mineralogical Society
     chrysotile asb€stos in human lungs. Environmental Research, 14,
                                                                                        of America Reviews in Mineralogy, 28,545-554.
                                                                                      Wagner, J.C. (1991) The discovery ofthe associationbetween blue as-
Jones,A.D., vincent, J.H., Mclntosh, C., McMillan, C.H., and Addison,
                                                                                        bestosand mesotheliomasand the aftermath. Brilish Journal oflndus-
     J. (1989)The effectoffiber durability on the hazardpotential ofinhaled
                                                                                        trial Medicine. 48, 399-403.
     chrysotile asbestos       fibers. Experimental Pathology, 37, 98-102.
                                                                                      Wagner,J.C., Sleggs,  C.A., and Marchand, P' (1960) Diffuse pleural mes-
 Kamp, D.W., Graceffa,P., Pryor, W.A., and Weitzman, S.A. (1992) The                    otheliornasand asbestos   exposurein the north westernCape Province.
     role of free radicalsin asbestos-induced       diseases.Free Radical Biology
                                                                                        British Journal of Industrial Medicine, 17, 260-27 I.
     and Medicine, 12, 293-315.                                                       Wagner, J.C., Berry, G., Skidmore, J.W., and Timbrell, V. (1974) The
 Kennedy,T.P., Dodson,R., Rao, N.V., Ky, H., Hopkins, C., Baser,M.,                     effectsof the inhalation of asbestos rats. British Journal of Cancer,
     Tolley, E., and Hoidal, J.R. (1989) Dusts causingpneumoconiosisgen-                 29,252-269.
     erate OH and produce hemolysis by acting as Fenton catalysts. Ar-                 Wagner, J.C., Berry, G., and Pooley, F.D. (1982) Mesotheliomas and
     chives of Biochemistry and Biophysics, 269, 359-364.                                asbestostype in asbestostextile workers: A study of lung contents.
 Klockars, M., Hedenborg,M., and Vanhala, E. (1990) Efect of two par-                    British Medical Journal, 285, 603.
     ticle surface-modifuing agents, polyvinylpyridine-N-oxide and carbo-              Weitzman, S.A., and Graceffa,P. (1984) Asbestoscatallzes hydroxyl and
     methylcellulose,on the quartz and asbestos         mineral fiber-inducedpro-        superoxide radical generation from hydrogen peroxide. Archives of
     duction of reactive oxygen metabolites by human polymorphonuclear                   Biochemistry and Biophysics, 228, 373-37 6.
     leukocytes.Archives of Environmental Health, 45, 8-14.                            White, A.F., and Yee, A. (1985) Aqueous oxidatron-reduction kinetics
 Lehnert, B.E. (1993) Defense mechanismsagainst inhaled particles and                    associated with coupled electron-cation transfer from iron-containing
     associated     particle-cellinteractions.In Mineralogical Societyof Amer-           silicatesat 25"C. Geochimica et Cosmochimica Acta, 49' 1263-1275'
     ica Reviews in Mineralogy, 28,427-469.                                            Zalma, R., Bonneau,L., Guignard, J., and Pezerat,H (l 987a)Formation
 Lund, L.G., and Aust, A.E. (1990) Iron mobilization from asbestosby                     of oxy radicals by oxygen reduction arising from the surface activity of
     chelatorsand ascorbicacid. Archives of Biochemistry and Biophysics,                 asbestos. Canadian Journal of Chemistry, 65, 2338-2341.
      278.60-64.                                                                       -      (l 987b) I'roduction ofhydroxyl radicalsby iron solid compounds.
 -          (l 992) Iron mobilization from crocidolite asbestos     geatly enhanc-       Toxicology and Environmental Chemistry, 13' l7l-187.
      es crocidolite-dependent formation of DNA single-strand breaks in
      dXl74 RFI DNA. Carcinogenesis, 637-642.   13,
 Lund, L.G., Williams, M.G., Dodson, R.F., and Aust, A.E. (1994)Iron                                     Srprerurssn26, 1994
                                                                                       Mewuscnrrr REcETvED
      associated     with asbestos   bodies is responsiblefor the formation of sin-    MnNuscnpr AccEFTEDJurv 13, 1995